qbit/xenocara
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Import a copy of ucpp, lightweight cpp for use by xrdb so that it

Browse Source 
doesn't depend on the 'comp'  set. ok espie@ deraadt@
This commit is contained in:
matthieu 2014-07-12 14:25:39 +00:00
parent c536fefa02
commit a2223c7302
23 changed files with 12687 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

21
app/xrdb-cpp/CHANGELOG Normal file
View File

@ -0,0 +1,21 @@
ucpp-1.3.2
* Fixed Issue 8, Included files missing a "terminating carriage
return character" will interrupt preprocessing in sample.c/LEXER
mode.
(http://code.google.com/p/ucpp/issues/detail?id=8)
ucpp-1.3.1
* Fixed Issue 5, "\r\n" carriage return characters are double
counted.
(http://code.google.com/p/ucpp/issues/detail?id=5)
* Fixed Issue 6, Included files missing a "terminating carriage
return character" will interrupt preprocessing in ucpp
(STAND_ALONE mode).
(http://code.google.com/p/ucpp/issues/detail?id=6)
* Fixed Issue 7, STD_MACROS & STD_ASSERTS undefined when trying to
build ucpp -DSTAND_ALONE.
(http://code.google.com/p/ucpp/issues/detail?id=7)
* Build ucpp & libucpp with 'make'.
ucpp-1.3
* Original import into svn at code.google.com/p/ucpp

115
app/xrdb-cpp/Makefile Normal file
View File

@ -0,0 +1,115 @@
# Makefile for ucpp
#
# (c) Thomas Pornin 1999 - 2002
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions
# are met:
# 1. Redistributions of source code must retain the above copyright
# notice, this list of conditions and the following disclaimer.
# 2. Redistributions in binary form must reproduce the above copyright
# notice, this list of conditions and the following disclaimer in the
# documentation and/or other materials provided with the distribution.
# 4. The name of the authors may not be used to endorse or promote
# products derived from this software without specific prior written
# permission.
#
# THIS SOFTWARE IS PROVIDED ``AS IS'' AND WITHOUT ANY EXPRESS OR
# IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
# WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
# ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHORS OR CONTRIBUTORS BE
# LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
# CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT
# OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
# BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
# WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
# OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
# EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
.POSIX:
# ----- user configurable part -----
# Edit the variables to suit your system.
#
# use -DAUDIT to enable some internal sanity checks
# use -DMEM_CHECK to check the return value of malloc()
# (superseded by AUDIT)
# use -DMEM_DEBUG to enable memory leak research (warning: this
# slows down ucpp a bit, and greatly increases memory consumption)
# use -DINLINE=foobar to enable use of the 'foobar'
# non standard qualifier, as an equivalent to the C99 'inline'
# qualifier. See tune.h for details.
#
# Two FLAGS lines are given for each system type; chose the first one for
# debug, the second one for a fast binary.
# for a generic compiler called cc
#CC = cc
#FLAGS = -DAUDIT
#FLAGS = -O -DMEM_CHECK
# for Minix-86
#CC = cc
#LDFLAGS = -i
#FLAGS = -m -DAUDIT
#FLAGS = -O -m -DMEM_CHECK
# for gcc
CC = gcc
FLAGS = -O3 -W -Wall -ansi
#FLAGS = -g -W -Wall -ansi -DAUDIT -DMEM_DEBUG
#FLAGS = -O3 -mcpu=pentiumpro -fomit-frame-pointer -W -Wall -ansi -DMEM_CHECK
#FLAGS = -O -pg -W -Wall -ansi -DMEM_CHECK
#LDFLAGS = -pg
# for the Compaq C compiler on Alpha/Linux
#CC = ccc
#FLAGS = -w0 -g -DAUDIT
#FLAGS = -w0 -fast -DMEM_CHECK
# for the Sun Workshop C Compiler
#CC = cc
#FLAGS = -g -Xa -DAUDIT
#FLAGS = -Xa -fast -DMEM_CHECK
# flags for the link step
LIBS =
#LIBS = libefence.a
#LIBS = -lgc_dbg
STAND_ALONE = -DSTAND_ALONE
ifdef STAND_ALONE
CSRC = mem.c nhash.c cpp.c lexer.c assert.c macro.c eval.c
FINAL_STEP = $(CC) $(LDFLAGS) -DUCPP_CONFIG $(STAND_ALONE) -o ucpp $(CSRC) $(LIBS)
endif
# ----- nothing should be changed below this line -----
COBJ = mem.o nhash.o cpp.o lexer.o assert.o macro.o eval.o
CFLAGS = $(FLAGS)
all: ucpp
@ar cq libucpp.a *.o
clean:
@rm -f *.o ucpp core *.a
ucpp: $(COBJ)
@$(FINAL_STEP)
assert.o: tune.h ucppi.h cpp.h nhash.h mem.h
@$(CC) $(CFLAGS) -c assert.c
cpp.o: tune.h ucppi.h cpp.h nhash.h mem.h
@$(CC) $(CFLAGS) -c cpp.c
eval.o: tune.h ucppi.h cpp.h nhash.h mem.h arith.c arith.h
@$(CC) $(CFLAGS) -c eval.c
lexer.o: tune.h ucppi.h cpp.h nhash.h mem.h
@$(CC) $(CFLAGS) -c lexer.c
macro.o: tune.h ucppi.h cpp.h nhash.h mem.h
@$(CC) $(CFLAGS) -c macro.c
mem.o: mem.h
@$(CC) $(CFLAGS) -c mem.c
nhash.o: nhash.h mem.h
@$(CC) $(CFLAGS) -c nhash.c

877
app/xrdb-cpp/README Normal file
View File

@ -0,0 +1,877 @@
ucpp-1.3 is a C preprocessor compliant to ISO-C99.
Author: Thomas Pornin <pornin@bolet.org>
Main site: http://pornin.nerim.net/ucpp/
INTRODUCTION
------------
A C preprocessor is a part of a C compiler responsible for macro
replacement, conditional compilation and inclusion of header files.
It is often found as a stand-alone program on Unix systems.
ucpp is such a preprocessor; it is designed to be quick and light,
but anyway fully compliant to the ISO standard 9899:1999, also known
as C99. ucpp can be compiled as a stand-alone program, or linked to
some other code; in the latter case, ucpp will output tokens, one
at a time, on demand, as an integrated lexer.
ucpp operates in two modes:
-- lexer mode: ucpp is linked to some other code and outputs a stream of
tokens (each call to the lex() function will yield one token)
-- non-lexer mode: ucpp preprocesses text and outputs the resulting text
to a file descriptor; if linked to some other code, the cpp() function
must be called repeatedly, otherwise ucpp is a stand-alone binary.
INSTALLATION
------------
1. Uncompress the archive file and extract the source files.
2. Edit tune.h. Here is a short explanation of compile-time options:
LOW_MEM
Enable memory-saving functions; this is for low-end and old systems,
but seems to be good for larger systems too. Keep it.
NO_LIBC_BUF
NO_UCPP_BUF
Two options used to disable the two bufferings inside ucpp. Define
both options for maximum memory savings but you will probably want
to keep libc buffering for decent performance. Define none on large
systems (modern 32 or 64-bit systems).
UCPP_MMAP
With this option, if ucpp internal buffering is active, ucpp will
try to mmap() the input files. This might yield a slight performance
improvement, but will work only on a limited set of architectures.
PRAGMA_TOKENIZE
Make ucpp generate tokenized PRAGMA tokens on #pragma and _Pragma();
tokenization is made this way: tokens are assembled as a null
terminated array of unsigned chars; if a token has a string value
(as defined by the STRING_TOKEN macro), the value follows the token,
terminated by PRAGMA_TOKEN_END (by default, a newline character cast
to unsigned char). Whitespace tokens are skipped. The "name" value
of the PRAGMA token is a pointer to that array. This setting is
irrelevant in non-lexer mode.
PRAGMA_DUMP
In non-lexer mode, keep #pragma in output; non-void _Pragma() are
translated to the equivalent #pragma. Irrelevant in lexer mode.
NO_PRAGMA_IN_DIRECTIVE
Do not evaluate _Pragma() inside #if, #include, #include_next and #line
directives; instead, emit an error (since the remaining _Pragma will
surely imply a syntax error).
DSHARP_TOKEN_MERGE
When two tokens are to be merged with the `##' operator, but fail
because they do not merge into a single valid token, ucpp keeps those
two tokens separate by adding an extra space between them in text
output. With this option on, that extra space is not added, which means
that some tokens may merge partially if the text output is preprocessed
again. See tune.h for details.
INMACRO_FLAG
In lexer mode, set the inmacro flag to 1 if the current token comes
from a macro replacement, 0 otherwise. macro_count maintains an
increasing counter of such replacements. CONTEXT tokens count as
one macro replacement each. #pragma, and _Pragma() that do not come
from a macro replacement, also count as one macro replacement each.
This setting is irrelevant in non-lexer mode.
STD_INCLUDE_PATH
Default include path in stand-alone ucpp.
STD_MACROS
Default predefined macros in stand-alone ucpp.
STD_ASSERT
Default assertions in stand-alone ucpp.
NATIVE_SIGNED
NATIVE_UNSIGNED
NATIVE_UNSIGNED_BITS
NATIVE_SIGNED_MIN
NATIVE_SIGNED_MAX
SIMUL_ARITH_SUBTYPE
SIMUL_SUBTYPE_BITS
SIMUL_NUMBITS
WCHAR_SIGNEDNESS
Those options define how #if expressions are evaluated; see the
cross-compilation section of this file for more info, and the
comments in tune.h. Extra info is found in arith.h and arith.c,
at the possible expense of your mental health.
DEFAULT_LEXER_FLAGS
DEFAULT_CPP_FLAGS
Default flags in respectively lexer and non-lexer modes.
POSIX_JMP
Define this if your architecture defines sigsetjmp() and
siglongjmp(); it is known to (very slightly) improve performance
on AIX systems.
MAX_CHAR_VAL
ucpp will consider characters whose value is equal or above
MAX_CHAR_VAL as outside the C source charset (so they will be
treated just like '@', for instance). For ASCII systems, 128
is fine. 256 is a safer value, but uses more (static) memory.
For performance reasons, use a power of two. If MAX_CHAR_VAL is
correctly adjusted, ucpp should be compatible with any character
set.
UNBREAKABLE_SPACE
If you want an extra-whitespace character, define this macro to that
character. For instance, define this to 160 on an ISO-8859-1 system
if you want the 'unbreakable space' to be considered as whitespace.
SEMPER_FIDELIS
With this option set, ucpp, when used as a lexer, will pass
whitespace tokens to its caller, and those tokens will have their
true content; this is intended for reconstruction of the source
line. Beware that some comments may have embedded newlines.
COPY_LINE_LENGTH
ucpp can maintain a copy of the current source line, up to that
length. Irrelevant to stand-alone version.
*_MEMG
Those settings modify ucpp behaviour, wrt memory allocations. With
higher values, ucpp will perform less malloc() calls and will run
faster, but it will use more memory. Reduce INPUT_BUF_MEMG and
OUTPUT_BUF_MEMG on low-memory systems, if you kept ucpp buffering
(see NO_UCPP_BUF option).
3. Edit the Makefile. You should define the variables CC and FLAGS;
there are the following options:
-DAUDIT
Enable internal sanity checks; this slows down a bit ucpp. Do not
define unless you plan to debug ucpp.
-DMEM_CHECK
With this setting, ucpp will check for the return value of malloc()
and exit with a diagnostic when out of memory. MEM_CHECK is implied
by AUDIT.
-DMEM_DEBUG
Enable memory debug code. This will track memory leaks and several
occurrences of memory management errors; it will also slow down
things and increase memory consumption, so you probably do not
want to use this option.
-DINLINE=foobar
The ucpp code uses "inline" qualifier for some functions; by
default, that qualifier is macro-replaced with nothing. Define
INLINE to the correct replacement for your compiler, if supported.
Note that all "inline" functions in ucpp are also "static". For any
C99-compliant compiler, the GNU compiler (gcc), and the Compaq C
compiler under Linux/Alpha, no -DINLINE is needed (see tune.h for
details).
4. Compile by typing "make". This should produce the ucpp executable
file. You might see some warning messages, especially with gcc:
gcc believes some variables might be used prior to their
initialization; ignore those messages.
5. Install wherever you want the binary and the man page ucpp.1. I
have not provided an install sequence because I didn't bother.
6. If you do not have the make utility, compile each file separately
and link them together. The exact details depend on your compiler.
You must define the macro STAND_ALONE when compiling cpp.c (there
is such a definition, commented out, in cpp.c, line 34).
There is no "configure" script because:
-- I do not like the very idea of a "configure" script.
-- ucpp is written in ANSI-C and should be fairly portable.
-- There is no such thing as "standard" settings for a C preprocessor.
The predefined system macros, standard assertions,... must be tuned
by the sysadmin.
-- The primary goal of ucpp is to be included in compilers. The
stand-alone version is mainly a debugging tool.
Please note that you need an ISO-C90 (formerly ANSI) C compiler suite
(including the standard library) to compile ucpp. If your compiler is
not C99 (or later), read the cross-compilation section in this README
file.
The C90 and C99 standards state that external linkage names might be
considered equal or different based upon only their first 6 characters;
this rule might make ucpp not compile on a conformant C implementation.
I have yet to see such an implementation, however.
If you want to use ucpp as an integrated preprocessor and lexer, see the
section REUSE. Compiling ucpp as a library is an exercise left to the
reader.
With the LOW_MEM code enabled, ucpp can run on a Minix-i86 or Msdos
16-bit small-memory-model machine. It will not be fully compliant
on such an architecture to C99, since C99 states that at least one
source code with 4095 simultaneously defined macros must be processed;
ucpp will be limited to about 1500 macros (at most) due to memory
restrictions. At least ucpp can preprocess its own code in these
conditions. LOW_MEM is on by default because it seems to improve
performance on large systems.
LICENSE
-------
The copyright notice and license is at the beginning of the Makefile and
each source file. It is basically a BSD license, without the advertising
subclause (which BSD dropped recently anyway) and with no reference to
Berkeley (since the code is all mine, written from scratch). Informally,
this means that you can reuse and redistribute the code as you want,
provided that you state in the documentation (or any substantial part of
the software) of redistributed code that I am the original author. (If
you press a cdrom with 200 software packages, I do not insist on having
my name on the cover of the cdrom -- just keep a Readme file somewhere
on the cdrom, with the copyright notice included.)
As a courteous gesture, if you reuse my code, please drop me a mail.
It raises my self-esteem.
REUSE
-----
The code has been thought as part of a bigger project; it might be
used as an integrated lexer, that will read files, process them as a
C preprocessor, and output a stream of C tokens. To include this code
into a project, compile with STAND_ALONE undefined.
To use the preprocessor and lexer, several steps should be performed.
See the file 'sample.c' for an example.
1. call init_cpp(). This function initializes the lexer automaton.
2. set the following global variables:
no_special_macros
non-zero if the special macros (__FILE__ and others)
should not be defined. This is a global flag since
it affects the redefinition of such macros (which are
allowed if the special macros are not defined)
c99_compliant
if non-zero, define __STDC_VERSION__ to 199901L; this
is the default; otherwise, do not define __STDC_VERSION__.
Note that ucpp will accept to undefine __STDC_VERSION__
with a #undef directive.
c99_hosted
if strictly positive, define __STDC_HOSTED__ to 1.
If zero, define __STDC_HOSTED__ to 0. If negative,
do not define __STDC_HOSTED__. The default is 1.
emit_defines and emit_assertions should be set to 0 for
the step 3.
3. call init_tables(). This function initializes the macro table
and other things; it will intialize assertions if it has a non-zero
argument.
4. call init_include_path(). This function will reset the include
path to the list of paths given as argument.
5. set the following global variables
emit_dependencies
set to 1 if dependencies should be emitted during
preprocessing
set to 2 if dependencies should also be emitted for
system include files
emit_defines
set to non-zero if #define macro definitions should be
emitted when macros are defined
emit_assertions
set to non-zero if #define macro definitions should be
emitted when macros are defined
emit_output
the FILE * where the above items are sent if one of the
three emit_ variables is set to non zero
transient_characters
this is for some cross-compilation; see the relevant
part in this README file for details
6. call set_init_filename() with the initial filename as argument;
the second argument indicates whether the filename is real or
conventional ("real" means "an fopen() on it will work").
7. initialize your struct lexer_state:
call init_lexer_state()
call init_lexer_mode() if the preprocessor is supposed to
output a list of tokens, otherwise set the flags field
to DEFAULT_CPP_FLAGS and set the output field to the
FILE * where output should be sent
(init_lexer_mode(), if called at all, must be called after
init_lexer_state())
adjust the flags field; here is the meaning of flags:
WARN_STANDARD
emit the standard warnings
WARN_ANNOYING
emit the useless and annoying warnings
WARN_TRIGRAPHS
count trigraphs encountered; it is up to the caller to emit
a warning if some trigraphs were indeed encountered; the count
is stored in the count_trigraphs field of the struct lexer_state
WARN_TRIGRAPHS_MORE
emit a warning for each trigraph encountered
WARN_PRAGMA
emit a warning for each non-void _Pragma encountered in non-lexer
mode (because these are dumped as #pragma in the output) and for each
#pragma too, if ucpp was compiled without PRAGMA_DUMP
FAIL_SHARP
emit errors on '#' tokens beginning a line and not followed
by a valid cpp directive
CCHARSET
emit errors when non-C characters are encountered; if this flag
is not set, each non-C character will be considered as a BUNCH
token (since C99 states that non-C characters are allowed as
long as they "disappear" during preprocessing [through macro
replacement and stringification for instance], this flag must
not be set, for maximum C99 compliance)
DISCARD_COMMENTS
do not keep comments in output (irrelevant in lexer mode)
CPLUSPLUS_COMMENTS
understand new style comments (//) (mandatory for C99)
LINE_NUM
emit #line directives when entering a file, if not in lexer mode;
emit CONTEXT token in lexer mode for #line and new files
GCC_LINE_NUM
if LINE_NUM is set, emit gcc-like directives instead of #line
HANDLE_ASSERTIONS
understand assertions in #if expressions (and #assert, #unassert)
HANDLE_PRAGMA
make PRAGMA tokens for #pragma; irrelevant in non-lexer mode
(handling of some pragmas is required in C99 but is not of
the competence of the preprocessor; without this flag, ucpp will
ignore the contents of #pragma and _Pragma directives)
MACRO_VAARG
understand macros with a variable number of arguments (mandatory
for C99)
UTF8_SOURCE
understand UTF-8 encoding: multibyte characters are considered
equivalent to letters as far as syntax is concerned (they can
be used in identifiers)
LEXER
act as a lexer, outputting tokens
TEXT_OUTPUT
this flag should be set to 0 if ucpp works as a lexer, 1 otherwise.
It is somehow redundant with the LEXER flag, but the presence of
those two different flags is needed in ucpp.
KEEP_OUTPUT
in non-lexer mode, emit the result of preprocessing
COPY_LINE
maintain a copy of the last read line in the copy_line field of
the struct lexer_state ; see below for how to use this buffer
HANDLE_TRIGRAPHS
understand trigraphs, such as ??/ for \. This option should be
set by default, except for some legacy code.
There are other flags, but they are for private usage of ucpp.
8. adjust the input field in the lexer_state to the FILE * from where
source file is read. If you use the UCPP_MMAP compile-time option,
and your input file is eligible to mmap(), then you can call
fopen_mmap_file() to open it, then set_input_file() to set ls->input
and some other internal options. Do not call set_input_file() unless
you just called fopen_mmap_file() just before on the same file.
9. call add_incpath() to add an include path, define_macro() and
undef_macro() to add or remove macros, make_assertion() and
destroy_assertion() to add or remove assertions.
10. call enter_file() (this is needed only in non-lexer mode, or if
LINE_NUM is set).
Afterwards:
-- if you are in lexer mode, call lex(); each call will make the ctok
field point to the next token. A non-zero return value is an error.
lex() skips whitespace tokens. The memory used by the string value
of some tokens (identifiers, numbers...) is automatically freed,
so copy the contents of each such token if you want to keep it
(tokens with a string content are identified by the STRING_TOKEN
macro applied to their type).
When lex() returned a non-zero value: if it is CPPERR_EOF, then
end-of-input was reached. Otherwise, it is a genuine error and
ls->ctok is an undefined token; skip it and call lex() again to
ignore the error.
-- otherwise, call cpp(); each call will analyze one or more tokens
(one token if it did find neither a cpp directive nor a macro name).
A positive return value is an error.
For both functions, if the return value is CPPERR_EOF (which is a
strictly positive value), then it means that the end of file was
reached. Call check_cpp_errors() after end of file for pending errors
(unfinished #if constructions for instance). In non-lexer mode,
call flush_output().
In the struct lexer_state, the following fields might be read:
line the current input line number
oline the current output line number (in non-lexer mode)
flags the flags described above
count_trigraphs the number of trigraphs encountered
inmacro the current token comes from a macro
macro_count the current macro counter
"flags" is an unsigned long and might be modified; the three others
are of long type.
To perform another preprocessing: use free_lexer_state() to release
memory used by the buffers referenced in lexer_state, and go back to
step 2. The different tables (macros, assertions...) should be reset to
their respective initial contents.
There is also the wipeout() function: when called, it should release
(almost) all memory blocks allocated dynamically. After a wipeout(),
ucpp should be back to its state at step 2 (init_cpp() initializes only
static tables, that are never freed nor modified afterwards).
The COPY_LINE buffer: the struct lexer_state contains two interesting
fields, copy_line[] and cli. If the COPY_LINE flag is on, each read
line is stored in this buffer, up to (at most) COPY_LINE_LENGTH - 1
characters (COPY_LINE_LENGTH is defined in tune.h). The last character
of the buffer is always a zero, and if the line was read entirely, it is
zero terminated; the trailing newline is not included.
The purpose of this buffer is error-reporting. When an error occurs
(cpp() returns a strictly positive value, or lex() returns a non-zero
value), if your struct lexer_state is called ls, use this code:
if (ls.cli != 0) ls.copy_line[ls.cli] = 0;
This will add a trailing 0 if the line was not read entirely.
ucpp may be configured at runtime to accept alternate characters as
possible parts of identifiers. Typical intended usage is for the '$'
and '@' characters. The two relevant functions are set_identifier_char()
and unset_identifier_char(). When this call is issued:
set_identifier_char('$');
then for all the remaining input, the '$' character will be considered
as just another letter, as far as identifier tokenizing is concerned. This
is for identifiers only; numeric constants are not modified by that setting.
This call resets things back:
unset_identifier_char('$');
Those two functions modify the static table which is initialized by
init_cpp(). You may call init_cpp() at any time to restore the table
to its standard state.
When using this feature, take care of the following points:
-- Do NOT use a character whose numeric value (as an `unsigned char'
cast into an `int') is greater than or equal to MAX_CHAR_VAL (in tune.h).
This would lead to unpredictable results, including an abrupt crash of
ucpp. ucpp makes absolutely no check whatsoever on that matter: this is
the programmer's responsibility.
-- If you use a standard character such as '+' or '{', tokens which
begin with those characters cease to exist. This can be troublesome.
If you use set_identifier_char() on the '<' character, the handling of
#include directives will be greatly disturbed. Therefore the use of any
standard C character in set_identifier_char() of unset_identifier_char()
is declared unsupported, forbidden and altogether unwise.
-- Stricto sensu, when an extra character is declared as part of an
identifier, ucpp behaviour cease to conform to C99, which mandates that
characters such as '$' or '@' must be treated as independant tokens of
their own. Therefore, if your purpose is to use ucpp in a conformant
C implementation, the use of set_identifier_char() should be made at
least a runtime option.
-- When enabling a new character in the middle of a macro replacement,
the effect of that replacement may be delayed up to the end of that
macro (but this is a "may" !). If you wish to trigger this feature with
a custom #pragma or _Pragma(), you should remember it (for instance,
usine _Pragma() in a macro replacement, and then the extra character
in the same macro replacement, is not reliable).
COMPATIBILITY NOTES
-------------------
The C language has a lengthening history. Nowadays, C comes in three
flavours:
-- Traditional C, aka "K&R". This is the language first described by
Brian Kernighan and Dennis Ritchie, and implemented in the first C
compiler that was ever coded. There are actually several dialects of
K&R, and all of them are considered deprecated.
-- ISO 9899:1990, aka C90, aka C89, aka ANSI-C. Formalized by ANSI
in 1989 and adopted by ISO the next year, it is the C flavour many C
compilers understand. It is mostly backward compatible with K&R C, but
with enhancements, clarifications and several new features.
-- ISO 9899:1999, aka C99. This is an evolution on C90, almost fully
backward compatible with C90. C99 introduces many new and useful
features, however, including in the preprocessor.
There was also a normative addendum in 1995, that added a few features
to C90 (for instance, digraphs) that are also present in C99. It is
sometimes refered to as "C95" or "AMD 1".
ucpp implements the C99 standard, but can be used in a stricter mode,
to enforce C90 compatibility (it will, however, still recognize some
constructions that are not in plain C90).
ucpp also knows about several extensions to C99:
-- Assertions: this is an extension to the defined() operator, with
its own namespace. Assertions seem to be used in several places,
therefore ucpp knows about them. It is recommended to enable
assertions by default on Solaris systems.
-- Unicode: the C99 norm specifies that extended characters, from
the ISO-10646 charset (aka "unicode") can be used in identifiers
with the notations \u and \U. ucpp also accepts (with the proper
flag) the UTF-8 encoding in the source file for such characters.
-- #include_next directive: it works as a #include, but will look
for files only in the directories specified in the include path
after the one the current file was found. This is a GNU-ism that
is useful for writing transparent wrappers around header files.
Assertions and unicode are activated by specific flags; the #include_next
support is always active.
The ucpp code itself should be compatible with any ISO-C90 compiler.
The cpp.c file is rather big (~ 64kB), it might confuse old 16-bit C
compilers; the macro.c file is somewhat large also (~ 47kB).
The evaluation of #if expressions is subject to some subtleties, see the
section "cross-compilation".
The lexer code makes no assumption about the source character set, but
the following: source characters (those which have a syntactic value in
C; comment and string literal contents are not concerned) must have a
strictly positive value that is strictly lower than MAX_CHAR_VAL. The
strict positivity is already assured by the C standard, so you just need
to adjust MAX_CHAR_VAL.
ucpp has been tested succesfully on ASCII/ISO-8859-1 and EBCDIC systems.
Beware that UTF-8 is NOT compatible with EBCDIC.
Pragma handling: when used in non-lexer mode, ucpp tries to output a
source text that, when read again, will yield the exact same stream of
tokens. This is not completely true with regards to line numbering in
some tricky macro replacements, but it should work correctly otherwise,
especially with pragma directives if the compile-time option PRAGMA_DUMP
was set: #pragma are dumped, non-void _Pragma() are converted to the
corresponding #pragma and dumped also.
ucpp does not macro-replace the contents of #pragma and _Pragma();
If you want a macro-replaced pragma, use this:
#define pragma_(x) _Pragma(#x)
#define pragma(x) pragma_(x)
Anyway, pragmas do not nest (an _Pragma() cannot be evaluated if it is
inside a #pragma or another _Pragma).
I wrote ucpp according to what is found in "The C Programming Language"
from Brian Kernighan and Dennis Ritchie (2nd edition) and the C99
standard; but I could have misinterpreted some points. On some tricky
points I got help from the helpful people from the comp.std.c newsgroup.
For assertions and #include_next, I mimicked the behaviour of GNU cpp,
as is stated in the GNU cpp info documentation. An open question is
related to the following code:
#define undefined !
#define makeun(x) un ## x
#if makeun(defined foo)
qux
#else
bar
#endif
ucpp will replace 'defined foo' with 0 first (since foo is not defined),
then it will replace the macro makeun, and the expression will become
'un0', which is replaced by 0 since this is a remaining identifier. The
expression evaluates to false, and 'bar' is emitted.
However, some other preprocessors will replace makeun first, considering
that it is not part of a 'defined' operator application; this will
produce the macro 'undefined', which is replaced, and the expression
becomes '!foo'. 'foo' is replaced by 0, the expression evaluates to
true, and 'qux' is emitted.
My opinion is that the behaviour is undefined, because use of the
'defined' operator does not match an allowed form prior to macro
replacement (I mean, its syntax matches, but its use is reconverted
to inexistant and therefore is not anymore matching). Other people
think that the behaviour is well-specified, and contrary to what ucpp
does. The only thing clear to me is that the wording of the standard
(paragraph 6.10.1.3) is unclear.
Since the ucpp behaviour makes ucpp code simpler and cleaner, and
that it is unlikely that any real-life code would ever be disturbed
by that interpretation of the standard, ucpp will keep its current
behaviour until convincing evidence of my misinterpretation of the
standard is given to me. The problem can only occur if one uses ## to
make a 'defined' operator disappear from a #if expression (everybody
agrees that the generation of a 'defined' operator triggers undefined
behaviour).
Another point about macro replacement has been discussed at length in
several occasions. It is about the following code:
#define CAT(a, b) CAT_(a, b)
#define CAT_(a, b) a ## b
#define AB(x, y) CAT(x, y)
CAT(A, B)(X, Y)
ucpp will produce `CAT(X,Y)' as replacement for the last line, whereas
some other preprocessors output `XY'. The answer to the question
"which behaviour is correct" seems to be "this is not defined by the
C standard". It is the answer that has been actually given by the C
standardization committee in 1992, to the defect report #017, question
23, which asked that very same question. Since the wording of the
standard has not changed in these parts from the 1990 to the 1999
version, the preprocessor behaviour on the above-stated code should
still be considered as undefined.
It seems, however, that there used to be a time (around 1988) when the
committee members agreed upon a precise macro-replacement algorithm,
which specified quite clearly the preprocessor behaviour in such
situation. ucpp behaviour is occasionnaly claimed as "incorrect" with
regards to that algorithm. Since that macro replacement algorithm has
never been published, and the committee itself backed out from it in
1992, I decided to disregard those feeble claims.
It is possible, however, that at some point in the future I rewrite the
ucpp macro replacement code, since that code is a bit messy and might be
made to use less memory in some occasions. It is then possible that, in
the aftermath of such a rewrite, the ucpp behaviour for the above stated
code become tunable. Don't hold your breath, though.
About _Pragma: the standard is not clear about when this operator is
evaluated, and if it is allowed inside #if directives and such. For
ucpp, I coded _Pragma as a special macro with lazy replacement: it will
be evaluated wherever a macro could be replaced, and only at the end of
the macro replacement (for practical purposes, _Pragma can be considered
as a macro taking one argument, and being replaced by nothing, except
for some tricky uses of the # and ## operators). This means that, by
default, ucpp will evaluate _Pragma inside some directives (mainly, #if,
#include, #include_next and #line), but it can be taught not to do so by
defining NO_PRAGMA_IN_DIRECTIVE in tune.h.
CROSS-COMPILATION
-----------------
If compiled with a C99 development suite, ucpp should be fully
C99-compliant on the host platform (up to my own understanding of the
standard -- remember that this software is distributed as-is, without
any guarantee). However, if a pre-C99 compiler is used, or if the
target machine is not the host machine (for instance when you build a
cross-compiler), the evaluation of #if expressions is subject to some
cross-compiling issues:
-- character constants: when evaluating expressions, character constants
are interpreted in the source character set context; this is allowed
by the standard but this can lead to problems with code that expects
this interpretation to match the one made in the C code. To ease
cross-compilation, you can define a conversion array, and make the
global variable transient_characters point to it. The array should
contain 256 int; transient_characters[x] is the value of the character
whose value is x in the source character set.
This facility is provided for inclusion of ucpp inside another code;
if you want a stand-alone ucpp with that conversion, hard-code the
conversion table into eval.c and make transient_characters[] statically
point to it. Alternatively, you could provide an option syntax to
provide such a table on command-line, if you feel like it.
-- wide character constants signedness: by default, ucpp makes wide
characters as signed as what plain chars are on the build host. To
force wide character constant signedness, define WCHAR_SIGNEDNESS to 0
(for unsigned) or 1 (for signed). Beware, however, that "native" wide
character constants, even signed, are considered positive. Non-wide
character constants are, according to the C99 standard, of type int, and
therefore always signed.
-- evaluation type: C90 states that all constants in #if expressions
are considered as either long or unsigned long, and that the evaluation
is performed with operands of that size. In C99, the situation is
equivalent, except that the types used are intmax_t and uintmax_t, as
defined in <stdint.h>.
ucpp can use two expression evaluators: one uses native integer types
(one signed and one unsigned), the other evaluator emulates big integer
numbers by representing them with two values of some unsigned type. The
emulated type handles signed values in two's complement representation,
and can be any width ranging from 2 bits to twice the size of the
underlying native unsigned type used. An odd width is allowed. When
right shifting an emulated signed negative value, it is left-padded with
bits set to 1 (this is sign extension).
When the ARITHMETIC_CHECKS macro is defined in tune.h, all occurrences
of implementation-defined or undefined behaviour during arithmetic
evaluation are reported as errors or warned upon. This includes all
overflows and underflows on signed quantities, constants too large,
and so on. Errors (which terminate immediately evaluation) are emitted
for division by 0 (on / and % operators) and overflow (on / operator);
otherwise, warnings are emitted and the faulty evaluation takes place.
This prevents ucpp from crashing on typical x86 machines, while still
allowing to use some extensions.
FUTURE EVOLUTIONS
-----------------
ucpp is quite complete now. There was a longstanding project of
"traditional" preprocessing, but I dropped it because it would not
map cleanly on the token-based ucpp structure. Maybe I will code a
string-based preprocessor one day; it would certainly use some of the
code from lexer.c, eval.c, mem.c and nhash.c. However, making such a
tool is almost irrelevant nowadays. If one wants to handle such project,
using ucpp as code base, I would happily provide some help, if needed.
CHANGES
-------
From 1.2 to 1.3:
* brand new integer evaluation code, with precise evaluation and checks
* new hash table implementation, with binary trees
* relaxed attitude on failed `##' operators
* bugfix on macro definition on command-line wrt nesting macros
* support for up to 32766 macro arguments in LOW_MEM code
* support for optional additional "identifier" characters such as '$' or '@'
* bugfix: memory leak on void #assert
From 1.1 to 1.2:
* bugfix: numerous memory leaks
* new function: wipeout(); this should release all malloc() blocks
* bugfix: missing "newline" and trailing "context" tokens
* improved included files name caching
* included memory leak detection code
From 1.0 to 1.1:
* bugfix: missing newline when exiting from a non-newline-terminated file
* bugfix: crash when resetting due to definition of the _Pragma pseudo-macro
* bugfix: handling of additional "optional" whitespace with SEMPER_FIDELIS
* improved handling of unreplaced arg macros wrt output line
* tricky handling of utterly tricky #include
* bugfix: spurious token `~=' eliminated
From 0.9 to 1.0:
* bugfix: crash after erroneous #assert
* changed ERR_SHARP to FAIL_SHARP, EMUL_UINTMAX to SIMUL_UINTMAX
* made "inline" default on gcc and DEC ccc (Linux/Alpha)
* semantic of -I is now Unix-like (added directories are looked first)
* added -J flag (to add include directories after the system ones)
* cleaned up non-ascii issues
* bugfix: missing brace in no-LOW_MEM code
* bugfix: argument number check in variadic macros
* bugfix: crash in non-lexer mode after some cases of unreplaced macro
* bugfix: _Pragma() handling wrt # and ##
* made evaluation of _Pragma() optional in #if, #include and #line
* bugfix: re-dump of multiline #pragma
* added the inmacro and macro_count flags
* added mmap() support
* added option to retain whitespace content in lexer mode
From 0.8 to 0.9:
* added check for division by 0 in #if evaluation
* added check for non-standard line numbers
* added check for trailing garbage in most directives
* corrected signedness of char constants (always int, therefore always signed)
* made LOW_MEM code, so that ucpp runs smoothly on low memory architectures
* multiple bugfixes (using the GNU cpp testsuite)
* added handling of _Pragma (as a macro)
* added tokenization of pragma directives
* added conservation of pragma directives in text output
* produced Msdos 16-bit small memory model executable
* produced Minix-86 executable
From 0.7 to 0.8:
* added some support for Amiga systems
* fixed extra spacing in stringified tokens
* fixed bug related to %:% and tolerated rogue sharps
* namespace cleanup
* bugfix for macro redefinition
* added warning for evaluated comma operators in #if (ISO requirement)
* -Dfoo now defines foo with content 1 (and not void content)
* trigraphs can be disabled (for incorrect but legacy code)
* fixed semantics for #include "file" (local directory)
* fixed detection of protected files
* produced a Msdos 16-bit executable
From 0.6 to 0.7:
* officially changed the goal to full C99 compliance
* added the CONTEXT token and let NEWLINE tokens go
* added report_context() for error reporting
* enforced matching of #if/#endif (file-global nesting level = 0)
* added support of C99 digraphs
* added UTF-8 encoding support
* added universal character names
* rewrote #if expressions (sizes fixed, bignum, signed/unsigned fixed)
* fixed incomplete evaluation of #if expressions
* added transient_characters[]
From 0.5 to 0.6:
* disappearance of error_nonl()
* added extra optional warnings for trigraphs
* some bugfixes, especially in lexer mode
* handled MacIntosh files correctly
From 0.4 to 0.5:
* nicer #pragma handling (a token can be emitted)
* bugfix in lexer mode after #line and #error
* sample.c an example of code linked with ucpp
* made #if expressions conforming to standard signed/unsigned handling
* added the copy_line[] buffer feature
From 0.3 to 0.4:
* relaxed interpretation of '#include foo' when foo ends up, after macro
substitution, with a '<bar>' content
* corrected the 'double-dot' bug
* corrected two bugs related to the treatment of macro aborted calls (due
to lack of arguments)
* some namespaces cleanup, to ease integration into other code
* documented the way to include ucpp into another program
* made newlines embedded into strings illegal (and reported as such)
From 0.2 to 0.3:
* added support for system predefined macros
* made several bugfixes
* checked C99 compliance for most of the features
* ucpp now accepts non-C characters on standard when used stand-alone
* removed many useless spaces in the output
From 0.1 to 0.2:
* added support for assertions
* added support for macros with variable arguments
* split the pharaonic cpp.c file into many
* made several bugfixes
* relaxed the behaviour with regards to the void arguments
* made C++-like comments an option
THANKS TO
---------
Volker Barthelmann, Neil Booth, Stephen Davies, Stéphane Ecolivet,
Marc Espie, Marcus Holland-Moritz, Antoine Leca, Cyrille Lefevre,
Dave Rivers, Loic Tortay and Laurent Wacrenier, for suggestions and
beta-testing.
Paul Eggert, Douglas A. Gwyn, Clive D.W. Feather, and the other guys from
comp.std.c, for explanations about the standard.
Dave Brolley, Jamie Lokier and Neil Booth, for discussion about tricky
points on nesting macros.
Brian Kernighan and Dennis Ritchie, for bringing C to mortal Men.

1462
app/xrdb-cpp/arith.c Normal file
View File

File diff suppressed because it is too large Load Diff

255
app/xrdb-cpp/arith.h Normal file
View File

@ -0,0 +1,255 @@
/*
* Integer arithmetic evaluation, header file.
*
* (c) Thomas Pornin 2002
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 4. The name of the authors may not be used to endorse or promote
* products derived from this software without specific prior written
* permission.
*
* THIS SOFTWARE IS PROVIDED ``AS IS'' AND WITHOUT ANY EXPRESS OR
* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
* WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHORS OR CONTRIBUTORS BE
* LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
* CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT
* OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
* BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
* WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
* EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*
*/
/*
* This arithmetic evaluator uses two files: this header file (arith.h)
* and the source file (arith.c). To use this code, the source file should
* be included from another .c file which defines some macros (see below).
* Then the functions defined in the arith.c file become available to the
* including source file. If those functions are defined with external
* linkage (that is, `ARITH_FUNCTION_HEADER' does not contain `static'),
* it is possible for other source files to use the arithmetic functions
* by including the arith.h header only. The source file which includes
* arith.c should *not* include arith.h.
*
* If the #include is for arith.h, the following macros should be
* defined:
*
* -- If the evaluator is supposed to use a native type:
* NATIVE_SIGNED the native signed integer type
* NATIVE_UNSIGNED the native unsigned integer type
*
* -- If the evaluator is supposed to use an emulated type:
* SIMUL_ARITH_SUBTYPE the native unsigned type used for the simulation
* SIMUL_SUBTYPE_BITS the native unsigned type size
* SIMUL_NUMBITS the emulated type size
*
* -- For both cases:
* ARITH_TYPENAME the central arithmetic type name
* ARITH_FUNCTION_HEADER the qualifiers to add to function definitions
*
* The presence (respectively absence) of the NATIVE_SIGNED macro triggers
* the use of the native type evaluator (respectively simulated type
* evaluator).
*
* If the #include is for arith.c, the macros for arith.h should be defined,
* and the following should be defined as well:
*
* -- If the evaluator is supposed to use a native type:
* NATIVE_UNSIGNED_BITS the native unsigned type size
* NATIVE_SIGNED_MIN the native signed minimum value
* NATIVE_SIGNED_MAX the native signed maximum value
* (the last two macros must evaluate to signed constant expressions)
*
* -- For both cases:
* ARITH_WARNING(type) code to perform on warning
* ARITH_ERROR(type) code to perform on error
*
* The macro ARITH_WARNING() and ARITH_ERROR() are invoked with a
* numerical argument which is one of the enumeration constants
* defined below (ARITH_EXCEP_*) that identifies the specific problem.
*
* If the #include is for arith.c, the macro ARITHMETIC_CHECKS may be
* defined. When this macro is defined, checks are performed so that all
* operation which would lead to undefined or implementation-defined
* behaviour are first reported through ARITH_WARNING(). Code is smaller
* and faster without these checks, of course. Regardless of the status
* of that macro, divisions by 0 and overflows on signed division are
* reported as errors through ARITH_ERROR().
*
*/
#ifndef ARITH_H__
#define ARITH_H__
enum {
/* Warnings */
ARITH_EXCEP_CONV_O, /* overflow on conversion */
ARITH_EXCEP_NEG_O, /* overflow on unary minus */
ARITH_EXCEP_NOT_T, /* trap representation on bitwise inversion */
ARITH_EXCEP_PLUS_O, /* overflow on addition */
ARITH_EXCEP_PLUS_U, /* underflow on addition */
ARITH_EXCEP_MINUS_O, /* overflow on subtraction */
ARITH_EXCEP_MINUS_U, /* underflow on subtraction */
ARITH_EXCEP_AND_T, /* trap representation on bitwise and */
ARITH_EXCEP_XOR_T, /* trap representation on bitwise xor */
ARITH_EXCEP_OR_T, /* trap representation on bitwise or */
ARITH_EXCEP_LSH_W, /* left shift by type width or more */
ARITH_EXCEP_LSH_C, /* left shift by negative count */
ARITH_EXCEP_LSH_O, /* overflow on left shift */
ARITH_EXCEP_LSH_U, /* underflow on left shift */
ARITH_EXCEP_RSH_W, /* right shift by type width or more */
ARITH_EXCEP_RSH_C, /* right shift by negative count */
ARITH_EXCEP_RSH_N, /* right shift of negative value */
ARITH_EXCEP_STAR_O, /* overflow on multiplication */
ARITH_EXCEP_STAR_U, /* underflow on multiplication */
/* Errors */
ARITH_EXCEP_SLASH_D, /* division by 0 */
ARITH_EXCEP_SLASH_O, /* overflow on division */
ARITH_EXCEP_PCT_D, /* division by 0 on modulus operator */
ARITH_EXCEP_CONST_O /* constant too large */
};
#define arith_strc_(x, y) x ## y
#define arith_strc(x, y) arith_strc_(x, y)
#define arith_u arith_strc(u_, ARITH_TYPENAME)
#define arith_s arith_strc(s_, ARITH_TYPENAME)
#define arith_op_u(op) arith_strc(ARITH_TYPENAME, arith_strc(_u_, op))
#define arith_op_s(op) arith_strc(ARITH_TYPENAME, arith_strc(_s_, op))
#define ARITH_DECL_MONO_U_U(op) ARITH_FUNCTION_HEADER arith_u \
arith_op_u(op)(arith_u x)
#define ARITH_DECL_MONO_U_S(op) ARITH_FUNCTION_HEADER arith_s \
arith_op_u(op)(arith_u x)
#define ARITH_DECL_MONO_U_I(op) ARITH_FUNCTION_HEADER int \
arith_op_u(op)(arith_u x)
#define ARITH_DECL_MONO_U_L(op) ARITH_FUNCTION_HEADER unsigned long \
arith_op_u(op)(arith_u x)
#define ARITH_DECL_MONO_S_U(op) ARITH_FUNCTION_HEADER arith_u \
arith_op_s(op)(arith_s x)
#define ARITH_DECL_MONO_S_S(op) ARITH_FUNCTION_HEADER arith_s \
arith_op_s(op)(arith_s x)
#define ARITH_DECL_MONO_S_I(op) ARITH_FUNCTION_HEADER int \
arith_op_s(op)(arith_s x)
#define ARITH_DECL_MONO_S_L(op) ARITH_FUNCTION_HEADER long \
arith_op_s(op)(arith_s x)
#define ARITH_DECL_MONO_I_U(op) ARITH_FUNCTION_HEADER arith_u \
arith_op_u(op)(int x)
#define ARITH_DECL_MONO_L_U(op) ARITH_FUNCTION_HEADER arith_u \
arith_op_u(op)(unsigned long x)
#define ARITH_DECL_MONO_I_S(op) ARITH_FUNCTION_HEADER arith_s \
arith_op_s(op)(int x)
#define ARITH_DECL_MONO_L_S(op) ARITH_FUNCTION_HEADER arith_s \
arith_op_s(op)(long x)
#define ARITH_DECL_MONO_ST_US(op) ARITH_FUNCTION_HEADER char *arith_op_u(op) \
(char *c, arith_u *ru, arith_s *rs, int *sp)
#define ARITH_DECL_BI_UU_U(op) ARITH_FUNCTION_HEADER arith_u \
arith_op_u(op)(arith_u x, arith_u y)
#define ARITH_DECL_BI_UI_U(op) ARITH_FUNCTION_HEADER arith_u \
arith_op_u(op)(arith_u x, int y)
#define ARITH_DECL_BI_UU_I(op) ARITH_FUNCTION_HEADER int \
arith_op_u(op)(arith_u x, arith_u y)
#define ARITH_DECL_BI_SS_S(op) ARITH_FUNCTION_HEADER arith_s \
arith_op_s(op)(arith_s x, arith_s y)
#define ARITH_DECL_BI_SI_S(op) ARITH_FUNCTION_HEADER arith_s \
arith_op_s(op)(arith_s x, int y)
#define ARITH_DECL_BI_SS_I(op) ARITH_FUNCTION_HEADER int \
arith_op_s(op)(arith_s x, arith_s y)
#endif
#ifdef NATIVE_SIGNED
typedef NATIVE_SIGNED arith_s;
typedef NATIVE_UNSIGNED arith_u;
#else
#if SIMUL_NUMBITS > (2 * SIMUL_SUBTYPE_BITS)
#error Native subtype too small for arithmetic simulation.
#endif
#define SIMUL_MSW_WIDTH (SIMUL_NUMBITS / 2)
#define SIMUL_LSW_WIDTH ((SIMUL_NUMBITS + 1) / 2)
typedef struct {
SIMUL_ARITH_SUBTYPE msw, lsw;
} arith_u, arith_s;
#endif
/* functions with the unsigned type */
ARITH_DECL_MONO_S_U(to_u);
ARITH_DECL_MONO_I_U(fromint);
ARITH_DECL_MONO_L_U(fromulong);
ARITH_DECL_MONO_U_I(toint);
ARITH_DECL_MONO_U_L(toulong);
ARITH_DECL_MONO_U_U(neg);
ARITH_DECL_MONO_U_U(not);
ARITH_DECL_MONO_U_I(lnot);
ARITH_DECL_MONO_U_I(lval);
ARITH_DECL_BI_UU_U(plus);
ARITH_DECL_BI_UU_U(minus);
ARITH_DECL_BI_UI_U(lsh);
ARITH_DECL_BI_UI_U(rsh);
ARITH_DECL_BI_UU_I(lt);
ARITH_DECL_BI_UU_I(leq);
ARITH_DECL_BI_UU_I(gt);
ARITH_DECL_BI_UU_I(geq);
ARITH_DECL_BI_UU_I(same);
ARITH_DECL_BI_UU_I(neq);
ARITH_DECL_BI_UU_U(and);
ARITH_DECL_BI_UU_U(xor);
ARITH_DECL_BI_UU_U(or);
ARITH_DECL_BI_UU_U(star);
ARITH_DECL_BI_UU_U(slash);
ARITH_DECL_BI_UU_U(pct);
/* functions with the signed type */
ARITH_DECL_MONO_U_S(to_s);
ARITH_DECL_MONO_I_S(fromint);
ARITH_DECL_MONO_L_S(fromlong);
ARITH_DECL_MONO_S_I(toint);
ARITH_DECL_MONO_S_L(tolong);
ARITH_DECL_MONO_S_S(neg);
ARITH_DECL_MONO_S_S(not);
ARITH_DECL_MONO_S_I(lnot);
ARITH_DECL_MONO_S_I(lval);
ARITH_DECL_BI_SS_S(plus);
ARITH_DECL_BI_SS_S(minus);
ARITH_DECL_BI_SI_S(lsh);
ARITH_DECL_BI_SI_S(rsh);
ARITH_DECL_BI_SS_I(lt);
ARITH_DECL_BI_SS_I(leq);
ARITH_DECL_BI_SS_I(gt);
ARITH_DECL_BI_SS_I(geq);
ARITH_DECL_BI_SS_I(same);
ARITH_DECL_BI_SS_I(neq);
ARITH_DECL_BI_SS_S(and);
ARITH_DECL_BI_SS_S(xor);
ARITH_DECL_BI_SS_S(or);
ARITH_DECL_BI_SS_S(star);
ARITH_DECL_BI_SS_S(slash);
ARITH_DECL_BI_SS_S(pct);
/* conversions from string */
ARITH_DECL_MONO_ST_US(octconst);
ARITH_DECL_MONO_ST_US(hexconst);
ARITH_DECL_MONO_ST_US(decconst);

420
app/xrdb-cpp/assert.c Normal file
View File

@ -0,0 +1,420 @@
/*
* (c) Thomas Pornin 1999 - 2002
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 4. The name of the authors may not be used to endorse or promote
* products derived from this software without specific prior written
* permission.
*
* THIS SOFTWARE IS PROVIDED ``AS IS'' AND WITHOUT ANY EXPRESS OR
* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
* WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHORS OR CONTRIBUTORS BE
* LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
* CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT
* OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
* BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
* WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
* EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*
*/
#include "tune.h"
#include <stdio.h>
#include <string.h>
#include <stddef.h>
#include <limits.h>
#include <time.h>
#include "ucppi.h"
#include "mem.h"
#include "nhash.h"
/*
* Assertion support. Each assertion is indexed by its predicate, and
* the list of 'questions' which yield a true answer.
*/
static HTT assertions;
static int assertions_init_done = 0;
static struct assert *new_assertion(void)
{
struct assert *a = getmem(sizeof(struct assert));
a->nbval = 0;
return a;
}
static void del_token_fifo(struct token_fifo *tf)
{
size_t i;
for (i = 0; i < tf->nt; i ++)
if (S_TOKEN(tf->t[i].type)) freemem(tf->t[i].name);
if (tf->nt) freemem(tf->t);
}
static void del_assertion(void *va)
{
struct assert *a = va;
size_t i;
for (i = 0; i < a->nbval; i ++) del_token_fifo(a->val + i);
if (a->nbval) freemem(a->val);
freemem(a);
}
/*
* print the contents of a token list
*/
static void print_token_fifo(struct token_fifo *tf)
{
size_t i;
for (i = 0; i < tf->nt; i ++)
if (ttMWS(tf->t[i].type)) fputc(' ', emit_output);
else fputs(token_name(tf->t + i), emit_output);
}
/*
* print all assertions related to a given name
*/
static void print_assert(void *va)
{
struct assert *a = va;
size_t i;
for (i = 0; i < a->nbval; i ++) {
fprintf(emit_output, "#assert %s(", HASH_ITEM_NAME(a));
print_token_fifo(a->val + i);
fprintf(emit_output, ")\n");
}
}
/*
* compare two token_fifo, return 0 if they are identical, 1 otherwise.
* All whitespace tokens are considered identical, but sequences of
* whitespace are not shrinked.
*/
int cmp_token_list(struct token_fifo *f1, struct token_fifo *f2)
{
size_t i;
if (f1->nt != f2->nt) return 1;
for (i = 0; i < f1->nt; i ++) {
if (ttMWS(f1->t[i].type) && ttMWS(f2->t[i].type)) continue;
if (f1->t[i].type != f2->t[i].type) return 1;
if (f1->t[i].type == MACROARG
&& f1->t[i].line != f2->t[i].line) return 1;
if (S_TOKEN(f1->t[i].type)
&& strcmp(f1->t[i].name, f2->t[i].name)) return 1;
}
return 0;
}
/*
* for #assert
* Assertions are not part of the ISO-C89 standard, but they are sometimes
* encountered, for instance in Solaris standard include files.
*/
int handle_assert(struct lexer_state *ls)
{
int ina = 0, ltww;
struct token t;
struct token_fifo *atl = 0;
struct assert *a;
char *aname;
int ret = -1;
long l = ls->line;
int nnp;
size_t i;
while (!next_token(ls)) {
if (ls->ctok->type == NEWLINE) break;
if (ttMWS(ls->ctok->type)) continue;
if (ls->ctok->type == NAME) {
if (!(a = HTT_get(&assertions, ls->ctok->name))) {
a = new_assertion();
aname = sdup(ls->ctok->name);
ina = 1;
}
goto handle_assert_next;
}
error(l, "illegal assertion name for #assert");
goto handle_assert_warp_ign;
}
goto handle_assert_trunc;
handle_assert_next:
while (!next_token(ls)) {
if (ls->ctok->type == NEWLINE) break;
if (ttMWS(ls->ctok->type)) continue;
if (ls->ctok->type != LPAR) {
error(l, "syntax error in #assert");
goto handle_assert_warp_ign;
}
goto handle_assert_next2;
}
goto handle_assert_trunc;
handle_assert_next2:
atl = getmem(sizeof(struct token_fifo));
atl->art = atl->nt = 0;
for (nnp = 1, ltww = 1; nnp && !next_token(ls);) {
if (ls->ctok->type == NEWLINE) break;
if (ltww && ttMWS(ls->ctok->type)) continue;
ltww = ttMWS(ls->ctok->type);
if (ls->ctok->type == LPAR) nnp ++;
else if (ls->ctok->type == RPAR) {
if (!(-- nnp)) goto handle_assert_next3;
}
t.type = ls->ctok->type;
if (S_TOKEN(t.type)) t.name = sdup(ls->ctok->name);
aol(atl->t, atl->nt, t, TOKEN_LIST_MEMG);
}
goto handle_assert_trunc;
handle_assert_next3:
while (!next_token(ls) && ls->ctok->type != NEWLINE) {
if (!ttWHI(ls->ctok->type) && (ls->flags & WARN_STANDARD)) {
warning(l, "trailing garbage in #assert");
}
}
if (atl->nt && ttMWS(atl->t[atl->nt - 1].type) && (-- atl->nt) == 0)
freemem(atl->t);
if (atl->nt == 0) {
error(l, "void assertion in #assert");
goto handle_assert_error;
}
for (i = 0; i < a->nbval && cmp_token_list(atl, a->val + i); i ++);
if (i != a->nbval) {
/* we already have it */
ret = 0;
goto handle_assert_error;
}
/* This is a new assertion. Let's keep it. */
aol(a->val, a->nbval, *atl, TOKEN_LIST_MEMG);
if (ina) {
HTT_put(&assertions, a, aname);
freemem(aname);
}
if (emit_assertions) {
fprintf(emit_output, "#assert %s(", HASH_ITEM_NAME(a));
print_token_fifo(atl);
fputs(")\n", emit_output);
}
freemem(atl);
return 0;
handle_assert_trunc:
error(l, "unfinished #assert");
handle_assert_error:
if (atl) {
del_token_fifo(atl);
freemem(atl);
}
if (ina) {
freemem(aname);
freemem(a);
}
return ret;
handle_assert_warp_ign:
while (!next_token(ls) && ls->ctok->type != NEWLINE);
if (ina) {
freemem(aname);
freemem(a);
}
return ret;
}
/*
* for #unassert
*/
int handle_unassert(struct lexer_state *ls)
{
int ltww;
struct token t;
struct token_fifo atl;
struct assert *a;
int ret = -1;
long l = ls->line;
int nnp;
size_t i;
atl.art = atl.nt = 0;
while (!next_token(ls)) {
if (ls->ctok->type == NEWLINE) break;
if (ttMWS(ls->ctok->type)) continue;
if (ls->ctok->type == NAME) {
if (!(a = HTT_get(&assertions, ls->ctok->name))) {
ret = 0;
goto handle_unassert_warp;
}
goto handle_unassert_next;
}
error(l, "illegal assertion name for #unassert");
goto handle_unassert_warp;
}
goto handle_unassert_trunc;
handle_unassert_next:
while (!next_token(ls)) {
if (ls->ctok->type == NEWLINE) break;
if (ttMWS(ls->ctok->type)) continue;
if (ls->ctok->type != LPAR) {
error(l, "syntax error in #unassert");
goto handle_unassert_warp;
}
goto handle_unassert_next2;
}
if (emit_assertions)
fprintf(emit_output, "#unassert %s\n", HASH_ITEM_NAME(a));
HTT_del(&assertions, HASH_ITEM_NAME(a));
return 0;
handle_unassert_next2:
for (nnp = 1, ltww = 1; nnp && !next_token(ls);) {
if (ls->ctok->type == NEWLINE) break;
if (ltww && ttMWS(ls->ctok->type)) continue;
ltww = ttMWS(ls->ctok->type);
if (ls->ctok->type == LPAR) nnp ++;
else if (ls->ctok->type == RPAR) {
if (!(-- nnp)) goto handle_unassert_next3;
}
t.type = ls->ctok->type;
if (S_TOKEN(t.type)) t.name = sdup(ls->ctok->name);
aol(atl.t, atl.nt, t, TOKEN_LIST_MEMG);
}
goto handle_unassert_trunc;
handle_unassert_next3:
while (!next_token(ls) && ls->ctok->type != NEWLINE) {
if (!ttWHI(ls->ctok->type) && (ls->flags & WARN_STANDARD)) {
warning(l, "trailing garbage in #unassert");
}
}
if (atl.nt && ttMWS(atl.t[atl.nt - 1].type) && (-- atl.nt) == 0)
freemem(atl.t);
if (atl.nt == 0) {
error(l, "void assertion in #unassert");
return ret;
}
for (i = 0; i < a->nbval && cmp_token_list(&atl, a->val + i); i ++);
if (i != a->nbval) {
/* we have it, undefine it */
del_token_fifo(a->val + i);
if (i < (a->nbval - 1))
mmvwo(a->val + i, a->val + i + 1, (a->nbval - i - 1)
* sizeof(struct token_fifo));
if ((-- a->nbval) == 0) freemem(a->val);
if (emit_assertions) {
fprintf(emit_output, "#unassert %s(",
HASH_ITEM_NAME(a));
print_token_fifo(&atl);
fputs(")\n", emit_output);
}
}
ret = 0;
goto handle_unassert_finish;
handle_unassert_trunc:
error(l, "unfinished #unassert");
handle_unassert_finish:
if (atl.nt) del_token_fifo(&atl);
return ret;
handle_unassert_warp:
while (!next_token(ls) && ls->ctok->type != NEWLINE);
return ret;
}
/*
* Add the given assertion (as string).
*/
int make_assertion(char *aval)
{
struct lexer_state lls;
size_t n = strlen(aval) + 1;
char *c = sdup(aval);
int ret;
*(c + n - 1) = '\n';
init_buf_lexer_state(&lls, 0);
lls.flags = DEFAULT_LEXER_FLAGS;
lls.input = 0;
lls.input_string = (unsigned char *)c;
lls.pbuf = 0;
lls.ebuf = n;
lls.line = -1;
ret = handle_assert(&lls);
freemem(c);
free_lexer_state(&lls);
return ret;
}
/*
* Remove the given assertion (as string).
*/
int destroy_assertion(char *aval)
{
struct lexer_state lls;
size_t n = strlen(aval) + 1;
char *c = sdup(aval);
int ret;
*(c + n - 1) = '\n';
init_buf_lexer_state(&lls, 0);
lls.flags = DEFAULT_LEXER_FLAGS;
lls.input = 0;
lls.input_string = (unsigned char *)c;
lls.pbuf = 0;
lls.ebuf = n;
lls.line = -1;
ret = handle_unassert(&lls);
freemem(c);
free_lexer_state(&lls);
return ret;
}
/*
* erase the assertion table
*/
void wipe_assertions(void)
{
if (assertions_init_done) HTT_kill(&assertions);
assertions_init_done = 0;
}
/*
* initialize the assertion table
*/
void init_assertions(void)
{
wipe_assertions();
HTT_init(&assertions, del_assertion);
assertions_init_done = 1;
}
/*
* retrieve an assertion from the hash table
*/
struct assert *get_assertion(char *name)
{
return HTT_get(&assertions, name);
}
/*
* print already defined assertions
*/
void print_assertions(void)
{
HTT_scan(&assertions, print_assert);
}

236
app/xrdb-cpp/atest.c Normal file
View File

@ -0,0 +1,236 @@
#include <stdio.h>
#include <stdlib.h>
#include <limits.h>
#include <setjmp.h>
#if defined TEST_NATIVE
#define NATIVE_SIGNED int
#define NATIVE_UNSIGNED unsigned
#define NATIVE_UNSIGNED_BITS 32
#define NATIVE_SIGNED_MIN LONG_MIN
#define NATIVE_SIGNED_MAX LONG_MAX
#elif defined TEST_SIMUL
#define SIMUL_ARITH_SUBTYPE unsigned short
#define SIMUL_SUBTYPE_BITS 16
#define SIMUL_NUMBITS 31
#else
#error ====== Either TEST_NATIVE or TEST_SIMUL must be defined.
#endif
#define ARITH_TYPENAME zoinx
#define ARITH_FUNCTION_HEADER static inline
#define ARITH_WARNING(type) z_warn(type)
#define ARITH_ERROR(type) z_error(type)
void z_warn(int type);
void z_error(int type);
#include "arith.c"
#if defined TEST_NATIVE
static inline u_zoinx unsigned_to_uz(unsigned x)
{
return (u_zoinx)x;
}
static inline s_zoinx int_to_sz(int x)
{
return (s_zoinx)x;
}
static inline void print_uz(u_zoinx x)
{
printf("%u", x);
}
static inline void print_sz(s_zoinx x)
{
printf("%d", x);
}
#else
static inline u_zoinx unsigned_to_uz(unsigned x)
{
u_zoinx v;
v.msw = (x >> 16) & 0x7FFFU;
v.lsw = x & 0xFFFFU;
return v;
}
static inline s_zoinx int_to_sz(int x)
{
return unsigned_to_uz((unsigned)x);
}
static inline void print_uz(u_zoinx x)
{
printf("%u", ((unsigned)(x.msw) << 16) + (unsigned)(x.lsw));
}
static inline void print_sz(s_zoinx x)
{
if (x.msw & 0x4000U) {
putchar('-');
x = zoinx_u_neg(x);
}
print_uz(x);
}
#endif
static inline void print_int(int x)
{
printf("%d", x);
}
static jmp_buf jbuf;
void z_warn(int type)
{
switch (type) {
case ARITH_EXCEP_CONV_O:
fputs("[overflow on conversion] ", stdout); break;
case ARITH_EXCEP_NEG_O:
fputs("[overflow on unary minus] ", stdout); break;
case ARITH_EXCEP_NOT_T:
fputs("[trap representation on bitwise inversion] ", stdout);
break;
case ARITH_EXCEP_PLUS_O:
fputs("[overflow on addition] ", stdout); break;
case ARITH_EXCEP_PLUS_U:
fputs("[underflow on addition] ", stdout); break;
case ARITH_EXCEP_MINUS_O:
fputs("[overflow on subtraction] ", stdout); break;
case ARITH_EXCEP_MINUS_U:
fputs("[underflow on subtraction] ", stdout); break;
case ARITH_EXCEP_AND_T:
fputs("[trap representation on bitwise and] ", stdout); break;
case ARITH_EXCEP_XOR_T:
fputs("[trap representation on bitwise xor] ", stdout); break;
case ARITH_EXCEP_OR_T:
fputs("[trap representation on bitwise or] ", stdout); break;
case ARITH_EXCEP_LSH_W:
fputs("[left shift by type width or more] ", stdout); break;
case ARITH_EXCEP_LSH_C:
fputs("[left shift by negative count] ", stdout); break;
case ARITH_EXCEP_LSH_O:
fputs("[overflow on left shift] ", stdout); break;
case ARITH_EXCEP_LSH_U:
fputs("[underflow on left shift] ", stdout); break;
case ARITH_EXCEP_RSH_W:
fputs("[right shift by type width or more] ", stdout); break;
case ARITH_EXCEP_RSH_C:
fputs("[right shift by negative count] ", stdout); break;
case ARITH_EXCEP_RSH_N:
fputs("[right shift of negative value] ", stdout); break;
case ARITH_EXCEP_STAR_O:
fputs("[overflow on multiplication] ", stdout); break;
case ARITH_EXCEP_STAR_U:
fputs("[underflow on multiplication] ", stdout); break;
default:
fprintf(stdout, "UNKNOWN WARNING TYPE: %d\n", type);
exit(EXIT_FAILURE);
}
}
void z_error(int type)
{
switch (type) {
case ARITH_EXCEP_SLASH_D:
fputs("division by 0\n", stdout);
break;
case ARITH_EXCEP_SLASH_O:
fputs("overflow on division\n", stdout);
break;
case ARITH_EXCEP_PCT_D:
fputs("division by 0 on modulus operator\n", stdout);
break;
default:
fprintf(stdout, "UNKNOWN ERROR TYPE: %d\n", type);
exit(EXIT_FAILURE);
}
longjmp(jbuf, 1);
}
int main(void)
{
#define OPTRY_GEN(op, x, y, convx, convy, printz) do { \
printf("%s %s %s -> ", #x, #op, #y); \
if (!setjmp(jbuf)) { \
printz(zoinx_ ## op (convx(x), convy(y))); \
putchar('\n'); \
} \
} while (0)
#define IDENT(x) x
#define OPTRY_UU_U(op, x, y) \
OPTRY_GEN(op, x, y, unsigned_to_uz, unsigned_to_uz, print_uz)
#define OPTRY_UI_U(op, x, y) \
OPTRY_GEN(op, x, y, unsigned_to_uz, IDENT, print_uz)
#define OPTRY_UU_I(op, x, y) \
OPTRY_GEN(op, x, y, unsigned_to_uz, unsigned_to_uz, print_int)
#define OPTRY_SS_S(op, x, y) \
OPTRY_GEN(op, x, y, int_to_sz, int_to_sz, print_sz)
#define OPTRY_SI_S(op, x, y) \
OPTRY_GEN(op, x, y, int_to_sz, IDENT, print_sz)
#define OPTRY_SS_I(op, x, y) \
OPTRY_GEN(op, x, y, int_to_sz, int_to_sz, print_int)
OPTRY_UU_U(u_plus, 3, 4);
OPTRY_UU_U(u_plus, 1549587182, 1790478233);
OPTRY_UU_U(u_minus, 1549587182, 1790478233);
OPTRY_UU_U(u_minus, 1790478233, 1549587182);
OPTRY_UU_U(u_star, 432429875, 347785487);
OPTRY_UU_U(u_slash, 432429875, 34487);
OPTRY_UU_U(u_pct, 432429875, 34487);
OPTRY_UI_U(u_lsh, 1783, 19);
OPTRY_UI_U(u_lsh, 1783, 20);
OPTRY_UI_U(u_lsh, 1783, 21);
OPTRY_UI_U(u_rsh, 475902857, 7);
OPTRY_UI_U(u_rsh, 475902857, 17);
OPTRY_UI_U(u_rsh, 475902857, 38);
OPTRY_SS_S(s_plus, 3, 4);
OPTRY_SS_S(s_plus, 1549587182, 1790478233);
OPTRY_SS_S(s_plus, -1549587182, -1790478233);
OPTRY_SS_S(s_minus, 1549587182, 1790478233);
OPTRY_SS_S(s_minus, 1790478233, 1549587182);
OPTRY_SS_S(s_minus, -1790478233, -1549587182);
OPTRY_SS_S(s_minus, -1790478233, 1549587182);
OPTRY_SS_S(s_star, 432429875, 347785487);
OPTRY_SS_S(s_star, 432429875, -347785487);
OPTRY_SS_S(s_slash, 432429875, 34487);
OPTRY_SS_S(s_slash, -432429875, 34487);
OPTRY_SS_S(s_slash, 432429875, -34487);
OPTRY_SS_S(s_slash, -432429875, -34487);
OPTRY_SS_S(s_slash, 432429875, 0);
OPTRY_SS_S(s_slash, -2147483647 - 1, -1);
OPTRY_SS_S(s_pct, 432429875, 34487);
OPTRY_SS_S(s_pct, 432429875, 0);
OPTRY_SI_S(s_lsh, -1, 10);
OPTRY_SI_S(s_lsh, 1783, 19);
OPTRY_SI_S(s_lsh, 1783, 20);
OPTRY_SI_S(s_lsh, 1783, 21);
OPTRY_SI_S(s_rsh, -1024, 8);
OPTRY_SI_S(s_rsh, 475902857, 7);
OPTRY_SI_S(s_rsh, 475902857, 17);
return 0;
}

352
app/xrdb-cpp/config.h Normal file
View File

@ -0,0 +1,352 @@
/*
* (c) Thomas Pornin 1999 - 2002
* (c) Louis P. Santillan 2011
* This file is derived from tune.h
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 4. The name of the authors may not be used to endorse or promote
* products derived from this software without specific prior written
* permission.
*
* THIS SOFTWARE IS PROVIDED ``AS IS'' AND WITHOUT ANY EXPRESS OR
* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
* WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHORS OR CONTRIBUTORS BE
* LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
* CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT
* OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
* BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
* WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
* EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*
*/
/* ====================================================================== */
/*
* The LOW_MEM macro triggers the use of macro storage which uses less
* memory. It actually also improves performance on large, modern machines
* (due to less cache pressure). This option implies no limitation (except
* on the number of arguments a macro may, which is then limited to 32766)
* so it is on by default. Non-LOW_MEM code is considered deprecated.
*/
#define LOW_MEM
/* ====================================================================== */
/*
* Define AMIGA for systems using "drive letters" at the beginning of
* some paths; define MSDOS on systems with drive letters and using
* backslashes to seperate directory components.
*/
/* #define AMIGA */
/* #define MSDOS */
/* ====================================================================== */
/*
* Define this if your compiler does not know the strftime() function;
* TurboC 2.01 under Msdos does not know strftime().
*/
/* #define NOSTRFTIME */
/* ====================================================================== */
/*
* Buffering: there are two levels of buffering on input and output streams:
* the standard libc buffering (manageable with setbuf() and setvbuf())
* and some buffering provided by ucpp itself. The ucpp buffering uses
* two buffers, of size respectively INPUT_BUF_MEMG and OUTPUT_BUF_MEMG
* (as defined below).
* You can disable one or both of these bufferings by defining the macros
* NO_LIBC_BUF and NO_UCPP_BUF.
*/
/* #define NO_LIBC_BUF */
/* #define NO_UCPP_BUF */
/*
* On Unix stations, the system call mmap() might be used on input files.
* This option is a subclause of ucpp internal buffering. On one station,
* a 10% speed improvement was observed. Do not define this unless the
* host architecture has the following characteristics:
* -- Posix / Single Unix compliance
* -- Text files correspond one to one with memory representation
* If a file is not seekable or not mmapable, ucpp will revert to the
* standard fread() solution.
*
* This feature is still considered beta quality. On some systems where
* files can be bigger than memory address space (mainly, 32-bit systems
* with files bigger than 4 GB), this option makes ucpp fail to operate
* on those extremely large files.
*/
#define UCPP_MMAP
/*
* Performance issues:
* -- On memory-starved systems, such as Minix-i86, do not use ucpp
* buffering; keep only libc buffering.
* -- If you do not use libc buffering, activate the UCPP_MMAP option.
* Note that the UCPP_MMAP option is ignored if ucpp buffering is not
* activated.
*
* On an Athlon 1200 running FreeBSD 4.7, the best performances are
* achieved when libc buffering is activated and/or UCPP_MMAP is on.
*/
/* ====================================================================== */
/*
* Define this if you want ucpp to generate tokenized PRAGMA tokens;
* otherwise, it will generate raw string contents. This setting is
* irrelevant to the stand-alone version of ucpp.
*/
#define PRAGMA_TOKENIZE
/*
* Define this to the special character that marks the end of tokens with
* a string value inside a tokenized PRAGMA token. The #pragma and _Pragma()
* directives which use this character will be a bit more difficult to
* decode (but ucpp will not mind). 0 cannot be used. '\n' is fine because
* it cannot appear inside a #pragma or _Pragma(), since newlines cannot be
* embedded inside tokens, neither directly nor by macro substitution and
* stringization. Besides, '\n' is portable.
*/
#define PRAGMA_TOKEN_END ((unsigned char)'\n')
/*
* Define this if you want ucpp to include encountered #pragma directives
* in its output in non-lexer mode; _Pragma() are translated to equivalent
* #pragma directives.
*/
#define PRAGMA_DUMP
/*
* According to my interpretation of the C99 standard, _Pragma() are
* evaluated wherever macro expansion could take place. However, Neil Booth,
* whose mother language is English (contrary to me) and who is well aware
* of the C99 standard (and especially the C preprocessor) told me that
* it was unclear whether _Pragma() are evaluated inside directives such
* as #if, #include and #line. If you want to disable the evaluation of
* _Pragma() inside such directives, define the following macro.
*/
/* #define NO_PRAGMA_IN_DIRECTIVE */
/*
* The C99 standard mandates that the operator `##' must yield a single,
* valid token, lest undefined behaviour befall upon thy head. Hence,
* for instance, `+ ## +=' is forbidden, because `++=' is not a valid
* token (although it is a valid list of two tokens, `++' and `=').
* However, ucpp only emits a warning for such sin, and unmerges the
* tokens (thus emitting `+' then `+=' for that example). When ucpp
* produces text output, those two tokens will be separated by a space
* character so that the basic rule of text output is preserved: when
* parsed again, text output yields the exact same stream of tokens.
* That extra space is virtual: it does not count as a true whitespace
* token for stringization.
*
* However, it might be desirable, for some uses other than preprocessing
* C source code, not to emit that extra space at all. To make ucpp behave
* that way, define the DSHARP_TOKEN_MERGE macro. Please note that this
* can trigger spurious token merging. For instance, with that macro
* activated, `+ ## +=' will be output as `++=' which, if preprocessed
* again, will read as `++' followed by `='.
*
* All this is irrelevant to lexer mode; and trying to merge incompatible
* tokens is a shooting offence, anyway.
*/
/* #define DSHARP_TOKEN_MERGE */
/* ====================================================================== */
/*
* Define INMACRO_FLAG to include two flags to the structure lexer_state,
* that tell whether tokens come from a macro-replacement, and count those
* macro-replacements.
*/
/* #define INMACRO_FLAG */
/* ====================================================================== */
/*
* Paths where files are looked for by default, when #include is used.
* Typical path is /usr/local/include and /usr/include, in that order.
* If you want to set up no path, define the macro to 0.
*
* For Linux, get gcc includes too, or you will miss things like stddef.h.
* The exact path varies much, depending on the distribution.
*/
#define STD_INCLUDE_PATH "/usr/local/include", "/usr/include"
/* ====================================================================== */
/*
* Arithmetic code for evaluation of #if expressions. Evaluation
* uses either a native machine type, or an emulated two's complement
* type. Division by 0 and overflow on division are considered as errors
* and reported as such. If ARITHMETIC_CHECKS is defined, all other
* operations that imply undefined or implementation-defined behaviour
* are reported as warnings but otherwise performed nonetheless.
*
* For native type evaluation, the following macros should be defined:
* NATIVE_SIGNED the native signed type
* NATIVE_UNSIGNED the native corresponding unsigned type
* NATIVE_UNSIGNED_BITS the native unsigned type width, in bits
* NATIVE_SIGNED_MIN the native signed type minimum value
* NATIVE_SIGNED_MAX the native signed type maximum value
*
* The code in the arith.c file performs some tricky detection
* operations on the native type representation and possible existence
* of a trap representation. These operations assume a C99-compliant
* compiler; on a C90-only compiler, the operations are valid but may
* yield incorrect results. You may force those settings with some
* more macros: see the comments in arith.c (look for "ARCH_DEFINED").
* Remember that this is mostly a non-issue, unless you are building
* ucpp with a pre-C99 cross-compiler and either the host or target
* architecture uses a non-two's complement representation of signed
* integers. Such a combination is pretty rare nowadays, so the best
* you can do is forgetting completely this paragraph and live in peace.
*
*
* If you do not have a handy native type (for instance, you compile ucpp
* with a C90 compiler which lacks the "long long" type, or you compile
* ucpp for a cross-compiler which should support an evaluation integer
* type of a size that is not available on the host machine), you may use
* a simulated type. The type uses two's complement representation and
* may have any width from 2 bits to twice the underlying native type
* width, inclusive (odd widths are allowed). To use an emulated type,
* make sure that NATIVE_SIGNED is not defined, and define the following
* macros:
* SIMUL_ARITH_SUBTYPE the native underlying type to use
* SIMUL_SUBTYPE_BITS the native underlying type width
* SIMUL_NUMBITS the emulated type width
*
* Undefined and implementation-defined behaviours are warned upon, if
* ARITHMETIC_CHECKS is defined. Results are truncated to the type
* width; shift count for the << and >> operators is reduced modulo the
* emulatd type width; right shifting of a signed negative value performs
* sign extension (the result is left-padded with bits set to 1).
*/
/*
* For native type evaluation with a 64-bit "long long" type.
*/
#define NATIVE_SIGNED long long
#define NATIVE_UNSIGNED unsigned long long
#define NATIVE_UNSIGNED_BITS 64
#define NATIVE_SIGNED_MIN (-9223372036854775807LL - 1)
#define NATIVE_SIGNED_MAX 9223372036854775807LL
/*
* For emulation of a 64-bit type using a native 32-bit "unsigned long"
* type.
#undef NATIVE_SIGNED
#define SIMUL_ARITH_SUBTYPE unsigned long
#define SIMUL_SUBTYPE_BITS 32
#define SIMUL_NUMBITS 64
*/
/*
* Comment out the following line if you want to deactivate arithmetic
* checks (warnings upon undefined and implementation-defined
* behaviour). Arithmetic checks slow down a bit arithmetic operations,
* especially multiplications, but this should not be an issue with
* typical C source code.
*/
#define ARITHMETIC_CHECKS
/* ====================================================================== */
/*
* To force signedness of wide character constants, define WCHAR_SIGNEDNESS
* to 0 for unsigned, 1 for signed. By default, wide character constants
* are signed if the native `char' type is signed, and unsigned otherwise.
#define WCHAR_SIGNEDNESS 0
*/
/*
* Standard assertions. They should include one cpu() assertion, one machine()
* assertion (identical to cpu()), and one or more system() assertions.
*
* for Linux/PC: cpu(i386), machine(i386), system(unix), system(linux)
* for Linux/Alpha: cpu(alpha), machine(alpha), system(unix), system(linux)
* for Sparc/Solaris: cpu(sparc), machine(sparc), system(unix), system(solaris)
*
* These are only suggestions. On Solaris, machine() should be defined
* for i386 or sparc (standard system header use such an assertion). For
* cross-compilation, define assertions related to the target architecture.
*
* If you want no standard assertion, define STD_ASSERT to 0.
*/
#define STD_ASSERT 0
/*
#define STD_ASSERT "cpu(i386)", "machine(i386)", "system(unix)", \
"system(freebsd)"
*/
/* ====================================================================== */
/*
* System predefined macros. Nothing really mandatory, but some programs
* might rely on those.
* Each string must be either "name" or "name=token-list". If you want
* no predefined macro, define STD_MACROS to 0.
*/
#define STD_MACROS 0
/*
#define STD_MACROS "__FreeBSD=4", "__unix", "__i386", \
"__FreeBSD__=4", "__unix__", "__i386__"
*/
/* ====================================================================== */
/*
* Default flags; HANDLE_ASSERTIONS is required for Solaris system headers.
* See cpp.h for the definition of these flags.
*/
#define DEFAULT_CPP_FLAGS (DISCARD_COMMENTS | WARN_STANDARD \
| WARN_PRAGMA | FAIL_SHARP | MACRO_VAARG \
| CPLUSPLUS_COMMENTS | LINE_NUM | TEXT_OUTPUT \
| KEEP_OUTPUT | HANDLE_TRIGRAPHS \
| HANDLE_ASSERTIONS)
#define DEFAULT_LEXER_FLAGS (DISCARD_COMMENTS | WARN_STANDARD | FAIL_SHARP \
| MACRO_VAARG | CPLUSPLUS_COMMENTS | LEXER \
| HANDLE_TRIGRAPHS | HANDLE_ASSERTIONS)
/* ====================================================================== */
/*
* Define this to use sigsetjmp()/siglongjmp() instead of setjmp()/longjmp().
* This is non-ANSI, but it improves performance on some POSIX system.
* On typical C source code, such improvement is completely negligeable.
*/
/* #define POSIX_JMP */
/* ====================================================================== */
/*
* Maximum value (plus one) of a character handled by the lexer; 128 is
* alright for ASCII native source code, but 256 is needed for EBCDIC.
* 256 is safe in both cases; you will have big problems if you set
* this value to INT_MAX or above. On Minix-i86 or Msdos (small memory
* model), define MAX_CHAR_VAL to 128.
*
* Set MAX_CHAR_VAL to a power of two to increase lexing speed. Beware
* that lexer.c defines a static array of size MSTATE * MAX_CHAR_VAL
* values of type int (MSTATE is defined in lexer.c and is about 40).
*/
#define MAX_CHAR_VAL 128
/*
* If you want some extra character to be considered as whitespace,
* define this macro to that space. On ISO-8859-1 machines, 160 is
* the code for the unbreakable space.
*/
/* #define UNBREAKABLE_SPACE 160 */
/*
* If you want whitespace tokens contents to be recorded (making them
* tokens with a string content), define this. The macro STRING_TOKEN
* will be adjusted accordingly.
* Without this option, whitespace tokens are not even returned by the
* lex() function. This is irrelevant for the non-lexer mode (almost --
* it might slow down a bit ucpp, and with this option, comments will be
* kept inside #pragma directives).
*/
/* #define SEMPER_FIDELIS */
/* End of options overridable by UCPP_CONFIG and config.h */

2565
app/xrdb-cpp/cpp.c Normal file
View File

File diff suppressed because it is too large Load Diff

317
app/xrdb-cpp/cpp.h Normal file
View File

@ -0,0 +1,317 @@
/*
* (c) Thomas Pornin 1999 - 2002
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 4. The name of the authors may not be used to endorse or promote
* products derived from this software without specific prior written
* permission.
*
* THIS SOFTWARE IS PROVIDED ``AS IS'' AND WITHOUT ANY EXPRESS OR
* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
* WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHORS OR CONTRIBUTORS BE
* LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
* CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT
* OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
* BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
* WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
* EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*
*/
#ifndef UCPP__CPP__
#define UCPP__CPP__
/*
* Uncomment the following if you want ucpp to use externally provided
* error-reporting functions (ucpp_warning(), ucpp_error() and ucpp_ouch())
*/
/* #define NO_UCPP_ERROR_FUNCTIONS */
/*
* Tokens (do not change the order unless checking operators_name[] in cpp.c)
*
* It is important that the token NONE is 0
* Check the STRING_TOKEN macro
*/
#define CPPERR 512
enum {
NONE, /* whitespace */
NEWLINE, /* newline */
COMMENT, /* comment */
NUMBER, /* number constant */
NAME, /* identifier */
BUNCH, /* non-C characters */
PRAGMA, /* a #pragma directive */
CONTEXT, /* new file or #line */
STRING, /* constant "xxx" */
CHAR, /* constant 'xxx' */
SLASH, /* / */
ASSLASH, /* /= */
MINUS, /* - */
MMINUS, /* -- */
ASMINUS, /* -= */
ARROW, /* -> */
PLUS, /* + */
PPLUS, /* ++ */
ASPLUS, /* += */
LT, /* < */
LEQ, /* <= */
LSH, /* << */
ASLSH, /* <<= */
GT, /* > */
GEQ, /* >= */
RSH, /* >> */
ASRSH, /* >>= */
ASGN, /* = */
SAME, /* == */
#ifdef CAST_OP
CAST, /* => */
#endif
NOT, /* ~ */
NEQ, /* != */
AND, /* & */
LAND, /* && */
ASAND, /* &= */
OR, /* | */
LOR, /* || */
ASOR, /* |= */
PCT, /* % */
ASPCT, /* %= */
STAR, /* * */
ASSTAR, /* *= */
CIRC, /* ^ */
ASCIRC, /* ^= */
LNOT, /* ! */
LBRA, /* { */
RBRA, /* } */
LBRK, /* [ */
RBRK, /* ] */
LPAR, /* ( */
RPAR, /* ) */
COMMA, /* , */
QUEST, /* ? */
SEMIC, /* ; */
COLON, /* : */
DOT, /* . */
MDOTS, /* ... */
SHARP, /* # */
DSHARP, /* ## */
OPT_NONE, /* optional space to separate tokens in text output */
DIGRAPH_TOKENS, /* there begin digraph tokens */
/* for DIG_*, do not change order, unless checking undig() in cpp.c */
DIG_LBRK, /* <: */
DIG_RBRK, /* :> */
DIG_LBRA, /* <% */
DIG_RBRA, /* %> */
DIG_SHARP, /* %: */
DIG_DSHARP, /* %:%: */
DIGRAPH_TOKENS_END, /* digraph tokens end here */
LAST_MEANINGFUL_TOKEN, /* reserved words will go there */
MACROARG, /* special token for representing macro arguments */
UPLUS = CPPERR, /* unary + */
UMINUS /* unary - */
};
#include "tune.h"
#include <stdio.h>
#include <setjmp.h>
struct token {
int type;
long line;
char *name;
};
struct token_fifo {
struct token *t;
size_t nt, art;
};
struct lexer_state {
/* input control */
FILE *input;
#ifndef NO_UCPP_BUF
unsigned char *input_buf;
#ifdef UCPP_MMAP
int from_mmap;
unsigned char *input_buf_sav;
#endif
#endif
unsigned char *input_string;
size_t ebuf;
size_t pbuf;
int lka[2];
int nlka;
int macfile;
int last;
int discard;
unsigned long utf8;
unsigned char copy_line[COPY_LINE_LENGTH];
int cli;
/* output control */
FILE *output;
struct token_fifo *output_fifo, *toplevel_of;
#ifndef NO_UCPP_BUF
unsigned char *output_buf;
#endif
size_t sbuf;
/* token control */
struct token *ctok;
struct token *save_ctok;
size_t tknl;
int ltwnl;
int pending_token;
#ifdef INMACRO_FLAG
int inmacro;
long macro_count;
#endif
/* lexer options */
long line;
long oline;
unsigned long flags;
long count_trigraphs;
struct garbage_fifo *gf;
int ifnest;
int condnest;
int condcomp;
int condmet;
unsigned long condf[2];
};
/*
* Flags for struct lexer_state
*/
/* warning flags */
#define WARN_STANDARD 0x000001UL /* emit standard warnings */
#define WARN_ANNOYING 0x000002UL /* emit annoying warnings */
#define WARN_TRIGRAPHS 0x000004UL /* warn when trigraphs are used */
#define WARN_TRIGRAPHS_MORE 0x000008UL /* extra-warn for trigraphs */
#define WARN_PRAGMA 0x000010UL /* warn for pragmas in non-lexer mode */
/* error flags */
#define FAIL_SHARP 0x000020UL /* emit errors on rogue '#' */
#define CCHARSET 0x000040UL /* emit errors on non-C characters */
/* emission flags */
#define DISCARD_COMMENTS 0x000080UL /* discard comments from text output */
#define CPLUSPLUS_COMMENTS 0x000100UL /* understand C++-like comments */
#define LINE_NUM 0x000200UL /* emit #line directives in output */
#define GCC_LINE_NUM 0x000400UL /* same as #line, with gcc-syntax */
/* language flags */
#define HANDLE_ASSERTIONS 0x000800UL /* understand assertions */
#define HANDLE_PRAGMA 0x001000UL /* emit PRAGMA tokens in lexer mode */
#define MACRO_VAARG 0x002000UL /* understand macros with '...' */
#define UTF8_SOURCE 0x004000UL /* identifiers are in UTF8 encoding */
#define HANDLE_TRIGRAPHS 0x008000UL /* handle trigraphs */
/* global ucpp behaviour */
#define LEXER 0x010000UL /* behave as a lexer */
#define KEEP_OUTPUT 0x020000UL /* emit the result of preprocessing */
#define COPY_LINE 0x040000UL /* make a copy of the parsed line */
/* internal flags */
#define READ_AGAIN 0x080000UL /* emit again the last token */
#define TEXT_OUTPUT 0x100000UL /* output text */
/*
* Public function prototypes
*/
#ifndef NO_UCPP_BUF
void flush_output(struct lexer_state *);
#endif
void init_assertions(void);
int make_assertion(char *);
int destroy_assertion(char *);
void print_assertions(void);
void init_macros(void);
int define_macro(struct lexer_state *, char *);
int undef_macro(struct lexer_state *, char *);
void print_defines(void);
void set_init_filename(char *, int);
void init_cpp(void);
void init_include_path(char *[]);
void init_lexer_state(struct lexer_state *);
void init_lexer_mode(struct lexer_state *);
void free_lexer_state(struct lexer_state *);
void wipeout(void);
int lex(struct lexer_state *);
int check_cpp_errors(struct lexer_state *);
void add_incpath(char *);
void init_tables(int);
int enter_file(struct lexer_state *, unsigned long);
int cpp(struct lexer_state *);
void set_identifier_char(int c);
void unset_identifier_char(int c);
#ifdef UCPP_MMAP
FILE *fopen_mmap_file(char *);
void set_input_file(struct lexer_state *, FILE *);
#endif
struct stack_context {
char *long_name, *name;
long line;
};
struct stack_context *report_context(void);
extern int no_special_macros, system_macros,
emit_dependencies, emit_defines, emit_assertions;
extern int c99_compliant, c99_hosted;
extern FILE *emit_output;
extern char *current_filename, *current_long_filename;
extern char *operators_name[];
extern struct protect {
char *macro;
int state;
struct found_file *ff;
} protect_detect;
void ucpp_ouch(char *, ...);
void ucpp_error(long, char *, ...);
void ucpp_warning(long, char *, ...);
extern int *transient_characters;
/*
* Errors from CPPERR_EOF and above are not real erros, only show-stoppers.
* Errors below CPPERR_EOF are real ones.
*/
#define CPPERR_NEST 900
#define CPPERR_EOF 1000
/*
* This macro tells whether the name field of a given token type is
* relevant, or not. Irrelevant name field means that it might point
* to outerspace.
*/
#ifdef SEMPER_FIDELIS
#define STRING_TOKEN(x) ((x) == NONE || ((x) >= COMMENT && (x) <= CHAR))
#else
#define STRING_TOKEN(x) ((x) >= NUMBER && (x) <= CHAR)
#endif
#endif

699
app/xrdb-cpp/eval.c Normal file
View File

@ -0,0 +1,699 @@
/*
* (c) Thomas Pornin 1999 - 2002
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 4. The name of the authors may not be used to endorse or promote
* products derived from this software without specific prior written
* permission.
*
* THIS SOFTWARE IS PROVIDED ``AS IS'' AND WITHOUT ANY EXPRESS OR
* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
* WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHORS OR CONTRIBUTORS BE
* LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
* CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT
* OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
* BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
* WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
* EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*
*/
#include "tune.h"
#include <stdio.h>
#include <string.h>
#include <setjmp.h>
#include <limits.h>
#include "ucppi.h"
#include "mem.h"
JMP_BUF eval_exception;
long eval_line;
static int emit_eval_warnings;
/*
* If you want to hardcode a conversion table, define a static array
* of 256 int, and make transient_characters point to it.
*/
int *transient_characters = 0;
#define OCTAL(x) ((x) >= '0' && (x) <= '7')
#define DECIM(x) ((x) >= '0' && (x) <= '9')
#define HEXAD(x) (DECIM(x) \
|| (x) == 'a' || (x) == 'b' || (x) == 'c' \
|| (x) == 'd' || (x) == 'e' || (x) == 'f' \
|| (x) == 'A' || (x) == 'B' || (x) == 'C' \
|| (x) == 'D' || (x) == 'E' || (x) == 'F')
#define OVAL(x) ((int)((x) - '0'))
#define DVAL(x) ((int)((x) - '0'))
#define HVAL(x) (DECIM(x) ? DVAL(x) \
: (x) == 'a' || (x) == 'A' ? 10 \
: (x) == 'b' || (x) == 'B' ? 11 \
: (x) == 'c' || (x) == 'C' ? 12 \
: (x) == 'd' || (x) == 'D' ? 13 \
: (x) == 'e' || (x) == 'E' ? 14 : 15)
#define ARITH_TYPENAME big
#define ARITH_FUNCTION_HEADER static inline
#define ARITH_ERROR(type) z_error(type)
static void z_error(int type);
#ifdef ARITHMETIC_CHECKS
#define ARITH_WARNING(type) z_warn(type)
static void z_warn(int type);
#endif
#include "arith.c"
static void z_error(int type)
{
switch (type) {
case ARITH_EXCEP_SLASH_D:
error(eval_line, "division by 0");
break;
case ARITH_EXCEP_SLASH_O:
error(eval_line, "overflow on division");
break;
case ARITH_EXCEP_PCT_D:
error(eval_line, "division by 0 on modulus operator");
break;
case ARITH_EXCEP_CONST_O:
error(eval_line, "constant too large for destination type");
break;
#ifdef AUDIT
default:
ouch("erroneous integer error: %d", type);
#endif
}
throw(eval_exception);
}
#ifdef ARITHMETIC_CHECKS
static void z_warn(int type)
{
switch (type) {
case ARITH_EXCEP_CONV_O:
warning(eval_line, "overflow on integer conversion");
break;
case ARITH_EXCEP_NEG_O:
warning(eval_line, "overflow on unary minus");
break;
case ARITH_EXCEP_NOT_T:
warning(eval_line,
"bitwise inversion yields trap representation");
break;
case ARITH_EXCEP_PLUS_O:
warning(eval_line, "overflow on addition");
break;
case ARITH_EXCEP_PLUS_U:
warning(eval_line, "underflow on addition");
break;
case ARITH_EXCEP_MINUS_O:
warning(eval_line, "overflow on subtraction");
break;
case ARITH_EXCEP_MINUS_U:
warning(eval_line, "underflow on subtraction");
break;
case ARITH_EXCEP_AND_T:
warning(eval_line,
"bitwise AND yields trap representation");
break;
case ARITH_EXCEP_XOR_T:
warning(eval_line,
"bitwise XOR yields trap representation");
break;
case ARITH_EXCEP_OR_T:
warning(eval_line,
"bitwise OR yields trap representation");
break;
case ARITH_EXCEP_LSH_W:
warning(eval_line, "left shift count greater than "
"or equal to type width");
break;
case ARITH_EXCEP_LSH_C:
warning(eval_line, "left shift count negative");
break;
case ARITH_EXCEP_LSH_O:
warning(eval_line, "overflow on left shift");
break;
case ARITH_EXCEP_RSH_W:
warning(eval_line, "right shift count greater than "
"or equal to type width");
break;
case ARITH_EXCEP_RSH_C:
warning(eval_line, "right shift count negative");
break;
case ARITH_EXCEP_RSH_N:
warning(eval_line, "right shift of negative value");
break;
case ARITH_EXCEP_STAR_O:
warning(eval_line, "overflow on multiplication");
break;
case ARITH_EXCEP_STAR_U:
warning(eval_line, "underflow on multiplication");
break;
#ifdef AUDIT
default:
ouch("erroneous integer warning: %d", type);
#endif
}
}
#endif
typedef struct {
int sign;
union {
u_big uv;
s_big sv;
} u;
} ppval;
static int boolval(ppval x)
{
return x.sign ? big_s_lval(x.u.sv) : big_u_lval(x.u.uv);
}
#if !defined(WCHAR_SIGNEDNESS)
# if CHAR_MIN == 0
# define WCHAR_SIGNEDNESS 0
# else
# define WCHAR_SIGNEDNESS 1
# endif
#endif
/*
* Check the suffix, return 1 if it is signed, 0 otherwise. 1 is
* returned for a void suffix. Legal suffixes are:
* unsigned: u U ul uL Ul UL lu Lu lU LU ull uLL Ull ULL llu LLu llU LLU
* signed: l L ll LL
*/
static int pp_suffix(char *d, char *refc)
{
if (!*d) return 1;
if (*d == 'u' || *d == 'U') {
if (!*(++ d)) return 0;
if (*d == 'l' || *d == 'L') {
char *e = d + 1;
if (*e && *e != *d) goto suffix_error;
if (!*e || !*(e + 1)) return 0;
goto suffix_error;
}
goto suffix_error;
}
if (*d == 'l' || *d == 'L') {
if (!*(++ d)) return 1;
if (*d == *(d - 1)) {
d ++;
if (!*d) return 1;
}
if (*d == 'u' || *d == 'U') {
d ++;
if (!*d) return 0;
}
goto suffix_error;
}
suffix_error:
error(eval_line, "invalid integer constant '%s'", refc);
throw(eval_exception);
return 666;
}
static unsigned long pp_char(char *c, char *refc)
{
unsigned long r = 0;
c ++;
if (*c == '\\') {
int i;
c ++;
switch (*c) {
case 'n': r = '\n'; c ++; break;
case 't': r = '\t'; c ++; break;
case 'v': r = '\v'; c ++; break;
case 'b': r = '\b'; c ++; break;
case 'r': r = '\r'; c ++; break;
case 'f': r = '\f'; c ++; break;
case 'a': r = '\a'; c ++; break;
case '\\': r = '\\'; c ++; break;
case '\?': r = '\?'; c ++; break;
case '\'': r = '\''; c ++; break;
case '\"': r = '\"'; c ++; break;
case 'u':
for (i = 0, c ++; i < 4 && HEXAD(*c); i ++, c ++) {
r = (r * 16) + HVAL(*c);
}
if (i != 4) {
error(eval_line, "malformed UCN in %s", refc);
throw(eval_exception);
}
break;
case 'U':
for (i = 0, c ++; i < 8 && HEXAD(*c); i ++, c ++) {
r = (r * 16) + HVAL(*c);
}
if (i != 8) {
error(eval_line, "malformed UCN in %s", refc);
throw(eval_exception);
}
break;
case 'x':
for (c ++; HEXAD(*c); c ++) r = (r * 16) + HVAL(*c);
break;
default:
if (OCTAL(*c)) {
r = OVAL(*(c ++));
if (OCTAL(*c)) r = (r * 8) + OVAL(*(c ++));
if (OCTAL(*c)) r = (r * 8) + OVAL(*(c ++));
} else {
error(eval_line, "invalid escape sequence "
"'\\%c'", *c);
throw(eval_exception);
}
}
} else if (*c == '\'') {
error(eval_line, "empty character constant");
throw(eval_exception);
} else {
r = *((unsigned char *)(c ++));
}
if (transient_characters && r < 256) {
r = transient_characters[(size_t)r];
}
if (*c != '\'' && emit_eval_warnings) {
warning(eval_line, "multicharacter constant");
}
return r;
}
static ppval pp_strtoconst(char *refc)
{
ppval q;
char *c = refc, *d;
u_big ru;
s_big rs;
int sp, dec;
if (*c == '\'' || *c == 'L') {
q.sign = (*c == 'L') ? WCHAR_SIGNEDNESS : 1;
if (*c == 'L' && *(++ c) != '\'') {
error(eval_line,
"invalid wide character constant: %s", refc);
throw(eval_exception);
}
if (q.sign) {
q.u.sv = big_s_fromlong(pp_char(c, refc));
} else {
q.u.uv = big_u_fromulong(pp_char(c, refc));
}
return q;
}
if (*c == '0') {
/* octal or hexadecimal */
dec = 0;
c ++;
if (*c == 'x' || *c == 'X') {
c ++;
d = big_u_hexconst(c, &ru, &rs, &sp);
} else {
d = big_u_octconst(c, &ru, &rs, &sp);
}
} else {
dec = 1;
d = big_u_decconst(c, &ru, &rs, &sp);
}
q.sign = pp_suffix(d, refc);
if (q.sign) {
if (!sp) {
if (dec) {
error(eval_line, "constant too large "
"for destination type");
throw(eval_exception);
} else {
warning(eval_line, "constant is so large "
"that it is unsigned");
}
q.u.uv = ru;
q.sign = 0;
} else {
q.u.sv = rs;
}
} else {
q.u.uv = ru;
}
return q;
}
/*
* Used by #line directives -- anything beyond what can be put in an
* unsigned long, is considered absurd.
*/
unsigned long strtoconst(char *c)
{
ppval q = pp_strtoconst(c);
if (q.sign) q.u.uv = big_s_to_u(q.u.sv);
return big_u_toulong(q.u.uv);
}
#define OP_UN(x) ((x) == LNOT || (x) == NOT || (x) == UPLUS \
|| (x) == UMINUS)
static ppval eval_opun(int op, ppval v)
{
if (op == LNOT) {
v.sign = 1;
v.u.sv = big_s_fromint(big_s_lnot(v.u.sv));
return v;
}
if (v.sign) {
switch (op) {
case NOT: v.u.sv = big_s_not(v.u.sv); break;
case UPLUS: break;
case UMINUS: v.u.sv = big_s_neg(v.u.sv); break;
}
} else {
switch (op) {
case NOT: v.u.uv = big_u_not(v.u.uv); break;
case UPLUS: break;
case UMINUS: v.u.uv = big_u_neg(v.u.uv); break;
}
}
return v;
}
#define OP_BIN(x) ((x) == STAR || (x) == SLASH || (x) == PCT \
|| (x) == PLUS || (x) == MINUS || (x) == LSH \
|| (x) == RSH || (x) == LT || (x) == LEQ \
|| (x) == GT || (x) == GEQ || (x) == SAME \
|| (x) == NEQ || (x) == AND || (x) == CIRC \
|| (x) == OR || (x) == LAND || (x) == LOR \
|| (x) == COMMA)
static ppval eval_opbin(int op, ppval v1, ppval v2)
{
ppval r;
int iv2 = 0;
switch (op) {
case STAR: case SLASH: case PCT:
case PLUS: case MINUS: case AND:
case CIRC: case OR:
/* promote operands, adjust signedness of result */
if (!v1.sign || !v2.sign) {
if (v1.sign) {
v1.u.uv = big_s_to_u(v1.u.sv);
v1.sign = 0;
} else if (v2.sign) {
v2.u.uv = big_s_to_u(v2.u.sv);
v2.sign = 0;
}
r.sign = 0;
} else {
r.sign = 1;
}
break;
case LT: case LEQ: case GT:
case GEQ: case SAME: case NEQ:
/* promote operands */
if (!v1.sign || !v2.sign) {
if (v1.sign) {
v1.u.uv = big_s_to_u(v1.u.sv);
v1.sign = 0;
} else if (v2.sign) {
v2.u.uv = big_s_to_u(v2.u.sv);
v2.sign = 0;
}
}
/* fall through */
case LAND:
case LOR:
/* result is signed anyway */
r.sign = 1;
break;
case LSH:
case RSH:
/* result is as signed as left operand; convert right
operand to int */
r.sign = v1.sign;
if (v2.sign) {
iv2 = big_s_toint(v2.u.sv);
} else {
iv2 = big_u_toint(v2.u.uv);
}
break;
case COMMA:
if (emit_eval_warnings) {
warning(eval_line, "ISO C forbids evaluated comma "
"operators in #if expressions");
}
r.sign = v2.sign;
break;
#ifdef AUDIT
default: ouch("a good operator is a dead operator");
#endif
}
#define SBINOP(x) if (r.sign) r.u.sv = big_s_ ## x (v1.u.sv, v2.u.sv); \
else r.u.uv = big_u_ ## x (v1.u.uv, v2.u.uv);
#define NSSBINOP(x) if (v1.sign) r.u.sv = big_s_fromint(big_s_ ## x \
(v1.u.sv, v2.u.sv)); else r.u.sv = big_s_fromint( \
big_u_ ## x (v1.u.uv, v2.u.uv));
#define LBINOP(x) if (v1.sign) r.u.sv = big_s_fromint( \
big_s_lval(v1.u.sv) x big_s_lval(v2.u.sv)); \
else r.u.sv = big_s_fromint( \
big_u_lval(v1.u.uv) x big_u_lval(v2.u.uv));
#define ABINOP(x) if (r.sign) r.u.sv = big_s_ ## x (v1.u.sv, iv2); \
else r.u.uv = big_u_ ## x (v1.u.uv, iv2);
switch (op) {
case STAR: SBINOP(star); break;
case SLASH: SBINOP(slash); break;
case PCT: SBINOP(pct); break;
case PLUS: SBINOP(plus); break;
case MINUS: SBINOP(minus); break;
case LSH: ABINOP(lsh); break;
case RSH: ABINOP(rsh); break;
case LT: NSSBINOP(lt); break;
case LEQ: NSSBINOP(leq); break;
case GT: NSSBINOP(gt); break;
case GEQ: NSSBINOP(geq); break;
case SAME: NSSBINOP(same); break;
case NEQ: NSSBINOP(neq); break;
case AND: SBINOP(and); break;
case CIRC: SBINOP(xor); break;
case OR: SBINOP(or); break;
case LAND: LBINOP(&&); break;
case LOR: LBINOP(||); break;
case COMMA: r = v2; break;
}
return r;
}
#define ttOP(x) (OP_UN(x) || OP_BIN(x) || (x) == QUEST || (x) == COLON)
static int op_prec(int op)
{
switch (op) {
case LNOT:
case NOT:
case UPLUS:
case UMINUS:
return 13;
case STAR:
case SLASH:
case PCT:
return 12;
case PLUS:
case MINUS:
return 11;
case LSH:
case RSH:
return 10;
case LT:
case LEQ:
case GT:
case GEQ:
return 9;
case SAME:
case NEQ:
return 8;
case AND:
return 7;
case CIRC:
return 6;
case OR:
return 5;
case LAND:
return 4;
case LOR:
return 3;
case QUEST:
return 2;
case COMMA:
return 1;
}
#ifdef AUDIT
ouch("an unknown species should have a higher precedence");
#endif
return 666;
}
/*
* Perform the hard work of evaluation.
*
* This function works because:
* -- all unary operators are right to left associative, and with
* identical precedence
* -- all binary operators are left to right associative
* -- there is only one non-unary and non-binary operator: the quest-colon
*
* If do_eval is 0, the evaluation of operators is not done. This is
* for sequence point operators (&&, || and ?:).
*/
static ppval eval_shrd(struct token_fifo *tf, int minprec, int do_eval)
{
ppval top;
struct token *ct;
top.sign = 1;
if (tf->art == tf->nt) goto trunc_err;
ct = tf->t + (tf->art ++);
if (ct->type == LPAR) {
top = eval_shrd(tf, 0, do_eval);
if (tf->art == tf->nt) goto trunc_err;
ct = tf->t + (tf->art ++);
if (ct->type != RPAR) {
error(eval_line, "a right parenthesis was expected");
throw(eval_exception);
}
} else if (ct->type == NUMBER || ct->type == CHAR) {
top = pp_strtoconst(ct->name);
} else if (OP_UN(ct->type)) {
top = eval_opun(ct->type, eval_shrd(tf,
op_prec(ct->type), do_eval));
goto eval_loop;
} else if (ttOP(ct->type)) goto rogue_op_err;
else {
goto invalid_token_err;
}
eval_loop:
if (tf->art == tf->nt) {
return top;
}
ct = tf->t + (tf->art ++);
if (OP_BIN(ct->type)) {
int bp = op_prec(ct->type);
if (bp > minprec) {
ppval tr;
if ((ct->type == LOR && boolval(top))
|| (ct->type == LAND && !boolval(top))) {
tr = eval_shrd(tf, bp, 0);
if (do_eval) {
top.sign = 1;
if (ct->type == LOR)
top.u.sv = big_s_fromint(1);
if (ct->type == LAND)
top.u.sv = big_s_fromint(0);
}
} else {
tr = eval_shrd(tf, bp, do_eval);
if (do_eval)
top = eval_opbin(ct->type, top, tr);
}
goto eval_loop;
}
} else if (ct->type == QUEST) {
int bp = op_prec(QUEST);
ppval r1, r2;
if (bp >= minprec) {
int qv = boolval(top);
r1 = eval_shrd(tf, bp, qv ? do_eval : 0);
if (tf->art == tf->nt) goto trunc_err;
ct = tf->t + (tf->art ++);
if (ct->type != COLON) {
error(eval_line, "a colon was expected");
throw(eval_exception);
}
r2 = eval_shrd(tf, bp, qv ? 0 : do_eval);
if (do_eval) {
if (qv) top = r1; else top = r2;
}
goto eval_loop;
}
}
tf->art --;
return top;
trunc_err:
error(eval_line, "truncated constant integral expression");
throw(eval_exception);
rogue_op_err:
error(eval_line, "rogue operator '%s' in constant integral "
"expression", operators_name[ct->type]);
throw(eval_exception);
invalid_token_err:
error(eval_line, "invalid token in constant integral expression");
throw(eval_exception);
}
#define UNARY(x) ((x) != NUMBER && (x) != NAME && (x) != CHAR \
&& (x) != RPAR)
/*
* Evaluate the integer expression contained in the given token_fifo.
* Evaluation is made by precedence of operators, as described in the
* Dragon Book. The unary + and - are distinguished from their binary
* counterparts using the Fortran way: a + or a - is considered unary
* if it does not follow a constant, an identifier or a right parenthesis.
*/
unsigned long eval_expr(struct token_fifo *tf, int *ret, int ew)
{
size_t sart;
ppval r;
emit_eval_warnings = ew;
if (catch(eval_exception)) goto eval_err;
/* first, distinguish unary + and - from binary + and - */
for (sart = tf->art; tf->art < tf->nt; tf->art ++) {
if (tf->t[tf->art].type == PLUS) {
if (sart == tf->art || UNARY(tf->t[tf->art - 1].type))
tf->t[tf->art].type = UPLUS;
} else if (tf->t[tf->art].type == MINUS) {
if (sart == tf->art || UNARY(tf->t[tf->art - 1].type))
tf->t[tf->art].type = UMINUS;
}
}
tf->art = sart;
r = eval_shrd(tf, 0, 1);
if (tf->art < tf->nt) {
error(eval_line, "trailing garbage in constant integral "
"expression");
goto eval_err;
}
*ret = 0;
return boolval(r);
eval_err:
*ret = 1;
return 0;
}

329
app/xrdb-cpp/hash.c Normal file
View File

@ -0,0 +1,329 @@
/*
* Generic hash table routines.
* (c) Thomas Pornin 1998, 1999, 2000
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 4. The name of the authors may not be used to endorse or promote
* products derived from this software without specific prior written
* permission.
*
* THIS SOFTWARE IS PROVIDED ``AS IS'' AND WITHOUT ANY EXPRESS OR
* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
* WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHORS OR CONTRIBUTORS BE
* LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
* CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT
* OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
* BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
* WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
* EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*
*/
#include <string.h>
#include "hash.h"
#include "mem.h"
#include "tune.h"
/*
* hash_string() is a sample hash function for strings
*/
int hash_string(char *s)
{
#ifdef FAST_HASH
unsigned h = 0, g;
while (*s) {
h = (h << 4) + *(unsigned char *)(s ++);
if ((g = h & 0xF000U) != 0) h ^= (g >> 12);
h &= ~g;
}
return (h ^ (h >> 9)) & 127U;
#else
unsigned char h = 0;
for (; *s; s ++) h ^= (unsigned char)(*s);
return ((int)h);
#endif
}
/*
* struct hash_item is the basic data type to internally handle hash tables
*/
struct hash_item {
void *data;
struct hash_item *next;
};
/*
* This function adds an entry to the struct hash_item list
*/
static struct hash_item *add_entry(struct hash_item *blist, void *data)
{
struct hash_item *t = getmem(sizeof(struct hash_item));
t->data = data;
t->next = blist;
return t;
}
/*
* This function finds a struct hash_item in a list, using the
* comparison function provided as cmpdata (*cmpdata() returns
* non-zero if the two parameters are to be considered identical).
*
* It returns 0 if the item is not found.
*/
static struct hash_item *get_entry(struct hash_item *blist, void *data,
int (*cmpdata)(void *, void *))
{
while (blist) {
if ((*cmpdata)(data, blist->data)) return blist;
blist = blist->next;
}
return 0;
}
/*
* This function acts like get_entry but deletes the found item, using
* the provided function deldata(); it returns 0 if the given data was
* not found.
*/
static struct hash_item *del_entry(struct hash_item *blist, void *data,
int (*cmpdata)(void *, void *), void (*deldata)(void *))
{
struct hash_item *prev = 0, *save = blist;
while (blist) {
if ((*cmpdata)(data, blist->data)) {
if (deldata) (*deldata)(blist->data);
if (prev) prev->next = blist->next;
if (save == blist) save = blist->next;
freemem(blist);
return save;
}
prev = blist;
blist = blist->next;
}
return 0;
}
/*
* This function creates a new hashtable, with the hashing and comparison
* functions given as parameters
*/
struct HT *newHT(int n, int (*cmpdata)(void *, void *), int (*hash)(void *),
void (*deldata)(void *))
{
struct HT *t = getmem(sizeof(struct HT));
int i;
t->lists = getmem(n * sizeof(struct hash_item *));
for (i = 0; i < n; i ++) t->lists[i] = 0;
t->nb_lists = n;
t->cmpdata = cmpdata;
t->hash = hash;
t->deldata = deldata;
return t;
}
/*
* This function adds a new entry in the hashtable ht; it returns 0
* on success, or a pointer to the already present item otherwise.
*/
void *putHT(struct HT *ht, void *data)
{
int h;
struct hash_item *d;
h = ((*(ht->hash))(data));
#ifndef FAST_HASH
h %= ht->nb_lists;
#endif
if ((d = get_entry(ht->lists[h], data, ht->cmpdata)))
return d->data;
ht->lists[h] = add_entry(ht->lists[h], data);
return 0;
}
/*
* This function adds a new entry in the hashtable ht, even if an equal
* entry is already there. Exercise caution !
* The new entry will "hide" the old one, which means that the new will be
* found upon lookup/delete, not the old one.
*/
void *forceputHT(struct HT *ht, void *data)
{
int h;
h = ((*(ht->hash))(data));
#ifndef FAST_HASH
h %= ht->nb_lists;
#endif
ht->lists[h] = add_entry(ht->lists[h], data);
return 0;
}
/*
* This function finds the entry corresponding to *data in the
* hashtable ht (using the comparison function given as argument
* to newHT)
*/
void *getHT(struct HT *ht, void *data)
{
int h;
struct hash_item *t;
h = ((*(ht->hash))(data));
#ifndef FAST_HASH
h %= ht->nb_lists;
#endif
if ((t = get_entry(ht->lists[h], data, ht->cmpdata)) == 0)
return 0;
return (t->data);
}
/*
* This function finds and delete the entry corresponding to *data
* in the hashtable ht (using the comparison function given as
* argument to newHT).
*/
int delHT(struct HT *ht, void *data)
{
int h;
h = ((*(ht->hash))(data));
#ifndef FAST_HASH
h %= ht->nb_lists;
#endif
ht->lists[h] = del_entry(ht->lists[h], data, ht->cmpdata, ht->deldata);
return 1;
}
/*
* This function completely eradicates from memory a given hash table,
* releasing all objects
*/
void killHT(struct HT *ht)
{
int i;
struct hash_item *t, *n;
void (*dd)(void *) = ht->deldata;
for (i = 0; i < ht->nb_lists; i ++) for (t = ht->lists[i]; t;) {
n = t->next;
if (dd) (*dd)(t->data);
freemem(t);
t = n;
}
freemem(ht->lists);
freemem(ht);
}
/*
* This function stores a backup of the hash table, for context stacking.
*/
void saveHT(struct HT *ht, void **buffer)
{
struct hash_item **b = (struct hash_item **)buffer;
mmv(b, ht->lists, ht->nb_lists * sizeof(struct hash_item *));
}
/*
* This function restores the saved state of the hash table.
* Do NOT use if some of the entries that were present before the backup
* have been removed (even temporarily).
*/
void restoreHT(struct HT *ht, void **buffer)
{
struct hash_item **b = (struct hash_item **)buffer;
int i;
for (i = 0; i < ht->nb_lists; i ++) {
struct hash_item *t = ht->lists[i], *n;
while (t != b[i]) {
n = t->next;
(*(ht->deldata))(t->data);
freemem(t);
t = n;
}
ht->lists[i] = b[i];
}
}
/*
* This function is evil. It inserts a new item in a saved hash table,
* tweaking the save buffer and the hash table in order to keep things
* stable. There are no checks.
*/
void tweakHT(struct HT *ht, void **buffer, void *data)
{
int h;
struct hash_item *d, *e;
h = ((*(ht->hash))(data));
#ifndef FAST_HASH
h %= ht->nb_lists;
#endif
for (d = ht->lists[h]; d != buffer[h]; d = d->next);
d = add_entry(buffer[h], data);
if (buffer[h] == ht->lists[h]) {
buffer[h] = ht->lists[h] = d;
return;
}
for (e = ht->lists[h]; e->next != buffer[h]; e = e->next);
e->next = d;
buffer[h] = d;
}
/*
* This function scans the whole table and calls the given function on
* each entry.
*/
void scanHT(struct HT *ht, void (*action)(void *))
{
int i;
for (i = 0; i < ht->nb_lists; i ++) {
struct hash_item *t = ht->lists[i];
while (t) {
(*action)(t->data);
t = t->next;
}
}
}
/*
* The two following fonctions are generic for storing structures
* uniquely identified by their name, which must be the first
* field of the structure.
*/
int hash_struct(void *m)
{
char *n = *(char **)m;
#ifdef FAST_HASH
return hash_string(n);
#else
return hash_string(n) & 127;
#endif
}
int cmp_struct(void *m1, void *m2)
{
char *n1 = *(char **)m1, *n2 = *(char **)m2;
return !strcmp(n1, n2);
}

58
app/xrdb-cpp/hash.h Normal file
View File

@ -0,0 +1,58 @@
/*
* (c) Thomas Pornin 1998, 1999, 2000
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 4. The name of the authors may not be used to endorse or promote
* products derived from this software without specific prior written
* permission.
*
* THIS SOFTWARE IS PROVIDED ``AS IS'' AND WITHOUT ANY EXPRESS OR
* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
* WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHORS OR CONTRIBUTORS BE
* LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
* CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT
* OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
* BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
* WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
* EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*
*/
#ifndef UCPP__HASH__
#define UCPP__HASH__
struct hash_item;
struct HT {
struct hash_item **lists;
int nb_lists;
int (*cmpdata)(void *, void *);
int (*hash)(void *);
void (*deldata)(void *);
};
int hash_string(char *);
struct HT *newHT(int, int (*)(void *, void *), int (*)(void *),
void (*)(void *));
void *putHT(struct HT *, void *);
void *forceputHT(struct HT *, void *);
void *getHT(struct HT *, void *);
int delHT(struct HT *, void *);
void killHT(struct HT *);
void saveHT(struct HT *, void **);
void restoreHT(struct HT *, void **);
void tweakHT(struct HT *, void **, void *);
void scanHT(struct HT *, void (*)(void *));
int hash_struct(void *);
int cmp_struct(void *, void *);
#endif

1020
app/xrdb-cpp/lexer.c Normal file
View File

File diff suppressed because it is too large Load Diff

1921
app/xrdb-cpp/macro.c Normal file
View File

File diff suppressed because it is too large Load Diff

328
app/xrdb-cpp/mem.c Normal file
View File

@ -0,0 +1,328 @@
/*
* Memory manipulation routines
* (c) Thomas Pornin 1998 - 2002
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 4. The name of the authors may not be used to endorse or promote
* products derived from this software without specific prior written
* permission.
*
* THIS SOFTWARE IS PROVIDED ``AS IS'' AND WITHOUT ANY EXPRESS OR
* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
* WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHORS OR CONTRIBUTORS BE
* LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
* CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT
* OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
* BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
* WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
* EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*
*/
#include "mem.h"
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
/*
* Shifting a pointer of that some bytes is supposed to satisfy
* alignment requirements. This is *not* guaranteed by the standard
* but should work everywhere anyway.
*/
#define ALIGNSHIFT (sizeof(long) > sizeof(long double) \
? sizeof(long) : sizeof(long double))
#ifdef AUDIT
void die(void)
{
abort();
}
static void suicide(unsigned long e)
{
fprintf(stderr, "ouch: Schrodinger's beef is not dead ! %lx\n", e);
die();
}
#else
void die(void)
{
exit(EXIT_FAILURE);
}
#endif
#if defined AUDIT || defined MEM_CHECK || defined MEM_DEBUG
/*
* This function is equivalent to a malloc(), but will display an error
* message and exit if the wanted memory is not available
*/
#ifdef MEM_DEBUG
static void *getmem_raw(size_t x)
#else
void *(getmem)(size_t x)
#endif
{
void *m;
#ifdef AUDIT
m = malloc(x + ALIGNSHIFT);
#else
m = malloc(x);
#endif
if (m == 0) {
fprintf(stderr, "ouch: malloc() failed\n");
die();
}
#ifdef AUDIT
*((unsigned long *)m) = 0xdeadbeefUL;
return (void *)(((char *)m) + ALIGNSHIFT);
#else
return m;
#endif
}
#endif
#ifndef MEM_DEBUG
/*
* This function is equivalent to a realloc(); if the realloc() call
* fails, it will try a malloc() and a memcpy(). If not enough memory is
* available, the program exits with an error message
*/
void *(incmem)(void *m, size_t x, size_t nx)
{
void *nm;
#ifdef AUDIT
m = (void *)(((char *)m) - ALIGNSHIFT);
if (*((unsigned long *)m) != 0xdeadbeefUL)
suicide(*((unsigned long *)m));
x += ALIGNSHIFT; nx += ALIGNSHIFT;
#endif
if (!(nm = realloc(m, nx))) {
if (x > nx) x = nx;
nm = (getmem)(nx);
memcpy(nm, m, x);
/* free() and not freemem(), because of the Schrodinger beef */
free(m);
}
#ifdef AUDIT
return (void *)(((char *)nm) + ALIGNSHIFT);
#else
return nm;
#endif
}
#endif
#if defined AUDIT || defined MEM_DEBUG
/*
* This function frees the given block
*/
#ifdef MEM_DEBUG
static void freemem_raw(void *x)
#else
void (freemem)(void *x)
#endif
{
#ifdef AUDIT
void *y = (void *)(((char *)x) - ALIGNSHIFT);
if ((*((unsigned long *)y)) != 0xdeadbeefUL)
suicide(*((unsigned long *)y));
*((unsigned long *)y) = 0xfeedbabeUL;
free(y);
#else
free(x);
#endif
}
#endif
#ifdef AUDIT
/*
* This function copies n bytes from src to dest
*/
void *mmv(void *dest, void *src, size_t n)
{
return memcpy(dest, src, n);
}
/*
* This function copies n bytes from src to dest
*/
void *mmvwo(void *dest, void *src, size_t n)
{
return memmove(dest, src, n);
}
#endif
#ifndef MEM_DEBUG
/*
* This function creates a new char * and fills it with a copy of src
*/
char *(sdup)(char *src)
{
size_t n = 1 + strlen(src);
char *x = getmem(n);
mmv(x, src, n);
return x;
}
#endif
#ifdef MEM_DEBUG
/*
* We include here special versions of getmem(), freemem() and incmem()
* that track allocations and are used to detect memory leaks.
*
* Each allocation is referenced in a list, with a serial number.
*/
/*
* Define "true" functions for applications that need pointers
* to such functions.
*/
void *(getmem)(size_t n)
{
return getmem(n);
}
void (freemem)(void *x)
{
freemem(x);
}
void *(incmem)(void *x, size_t s, size_t ns)
{
return incmem(x, s, ns);
}
char *(sdup)(char *s)
{
return sdup(s);
}
static long current_serial = 0L;
/* must be a power of two */
#define MEMDEBUG_MEMG 128U
static struct mem_track {
void *block;
long serial;
char *file;
int line;
} *mem = 0;
static size_t meml = 0;
static unsigned int current_ptr = 0;
static void *true_incmem(void *x, size_t old_size, size_t new_size)
{
void * y = realloc(x, new_size);
if (y == 0) {
y = malloc(new_size);
if (y == 0) {
fprintf(stderr, "ouch: malloc() failed\n");
die();
}
mmv(y, x, old_size < new_size ? old_size : new_size);
free(x);
}
return y;
}
static long find_free_block(void)
{
unsigned int n;
size_t i;
for (i = 0, n = current_ptr; i < meml; i ++) {
if (mem[n].block == 0) {
current_ptr = n;
return n;
}
n = (n + 1) & (meml - 1U);
}
if (meml == 0) {
size_t j;
meml = MEMDEBUG_MEMG;
mem = malloc(meml * sizeof(struct mem_track));
current_ptr = 0;
for (j = 0; j < meml ; j ++) mem[j].block = 0;
} else {
size_t j;
mem = true_incmem(mem, meml * sizeof(struct mem_track),
2 * meml * sizeof(struct mem_track));
current_ptr = meml;
for (j = meml; j < 2 * meml ; j ++) mem[j].block = 0;
meml *= 2;
}
return current_ptr;
}
void *getmem_debug(size_t n, char *file, int line)
{
void *x = getmem_raw(n + ALIGNSHIFT);
long i = find_free_block();
*(long *)x = i;
mem[i].block = x;
mem[i].serial = current_serial ++;
mem[i].file = file;
mem[i].line = line;
return (void *)((unsigned char *)x + ALIGNSHIFT);
}
void freemem_debug(void *x, char *file, int line)
{
void *y = (unsigned char *)x - ALIGNSHIFT;
long i = *(long *)y;
if (i < 0 || (size_t)i >= meml || mem[i].block != y) {
fprintf(stderr, "ouch: freeing free people (from %s:%d)\n",
file, line);
die();
}
mem[i].block = 0;
freemem_raw(y);
}
void *incmem_debug(void *x, size_t ol, size_t nl, char *file, int line)
{
void *y = getmem_debug(nl, file, line);
mmv(y, x, ol < nl ? ol : nl);
freemem_debug(x, file, line);
return y;
}
char *sdup_debug(char *src, char *file, int line)
{
size_t n = 1 + strlen(src);
char *x = getmem_debug(n, file, line);
mmv(x, src, n);
return x;
}
void report_leaks(void)
{
size_t i;
for (i = 0; i < meml; i ++) {
if (mem[i].block) fprintf(stderr, "leak: serial %ld, %s:%d\n",
mem[i].serial, mem[i].file, mem[i].line);
}
}
#endif

155
app/xrdb-cpp/mem.h Normal file
View File

@ -0,0 +1,155 @@
/*
* (c) Thomas Pornin 1998 - 2002
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 4. The name of the authors may not be used to endorse or promote
* products derived from this software without specific prior written
* permission.
*
* THIS SOFTWARE IS PROVIDED ``AS IS'' AND WITHOUT ANY EXPRESS OR
* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
* WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHORS OR CONTRIBUTORS BE
* LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
* CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT
* OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
* BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
* WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
* EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*
*/
#ifndef UCPP__MEM__
#define UCPP__MEM__
#include <stdlib.h>
void die(void);
#if defined AUDIT || defined MEM_CHECK || defined MEM_DEBUG
void *getmem(size_t);
#else
#define getmem malloc
#endif
#if defined MEM_DEBUG
void *getmem_debug(size_t, char *, int);
#undef getmem
#define getmem(x) getmem_debug(x, __FILE__, __LINE__)
#endif
#if defined AUDIT || defined MEM_DEBUG
void freemem(void *);
#else
#define freemem free
#endif
#if defined MEM_DEBUG
void freemem_debug(void *, char *, int);
#undef freemem
#define freemem(x) freemem_debug(x, __FILE__, __LINE__)
#endif
void *incmem(void *, size_t, size_t);
char *sdup(char *);
#if defined MEM_DEBUG
void *incmem_debug(void *, size_t, size_t, char *, int);
#undef incmem
#define incmem(x, y, z) incmem_debug(x, y, z, __FILE__, __LINE__)
void report_leaks(void);
char *sdup_debug(char *, char *, int);
#define sdup(x) sdup_debug(x, __FILE__, __LINE__)
#endif
#ifdef AUDIT
void *mmv(void *, void *, size_t);
void *mmvwo(void *, void *, size_t);
#else
#define mmv memcpy
#define mmvwo memmove
#endif
/*
* this macro adds the object obj at the end of the array list, handling
* memory allocation when needed; ptr contains the number of elements in
* the array, and memg is the granularity of memory allocations (a power
* of 2 is recommanded, for optimization reasons).
*
* list and ptr may be updated, and thus need to be lvalues.
*/
#define aol(list, ptr, obj, memg) do { \
if (((ptr) % (memg)) == 0) { \
if ((ptr) != 0) { \
(list) = incmem((list), (ptr) * sizeof(obj), \
((ptr) + (memg)) * sizeof(obj)); \
} else { \
(list) = getmem((memg) * sizeof(obj)); \
} \
} \
(list)[(ptr) ++] = (obj); \
} while (0)
/*
* bol() does the same as aol(), but adds the new item at the beginning
* of the list; beware, the computational cost is greater.
*/
#define bol(list, ptr, obj, memg) do { \
if (((ptr) % (memg)) == 0) { \
if ((ptr) != 0) { \
(list) = incmem((list), (ptr) * sizeof(obj), \
((ptr) + (memg)) * sizeof(obj)); \
} else { \
(list) = getmem((memg) * sizeof(obj)); \
} \
} \
if ((ptr) != 0) \
mmvwo((list) + 1, (list), (ptr) * sizeof(obj)); \
(ptr) ++; \
(list)[0] = (obj); \
} while (0)
/*
* mbol() does the same as bol(), but adds the new item at the given
* emplacement; bol() is equivalent to mbol with 0 as last argument.
*/
#define mbol(list, ptr, obj, memg, n) do { \
if (((ptr) % (memg)) == 0) { \
if ((ptr) != 0) { \
(list) = incmem((list), (ptr) * sizeof(obj), \
((ptr) + (memg)) * sizeof(obj)); \
} else { \
(list) = getmem((memg) * sizeof(obj)); \
} \
} \
if ((ptr) > n) \
mmvwo((list) + n + 1, (list) + n, \
((ptr) - n) * sizeof(obj)); \
(ptr) ++; \
(list)[n] = (obj); \
} while (0)
/*
* this macro adds the object obj at the end of the array list, doubling
* the size of list when needed; as for aol(), ptr and list must be
* lvalues, and so must be llng
*/
#define wan(list, ptr, obj, llng) do { \
if ((ptr) == (llng)) { \
(llng) += (llng); \
(list) = incmem((list), (ptr) * sizeof(obj), \
(llng) * sizeof(obj)); \
} \
(list)[(ptr) ++] = (obj); \
} while (0)
#endif

481
app/xrdb-cpp/nhash.c Normal file
View File

@ -0,0 +1,481 @@
/*
* Mixed hash table / binary tree code.
* (c) Thomas Pornin 2002
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 4. The name of the authors may not be used to endorse or promote
* products derived from this software without specific prior written
* permission.
*
* THIS SOFTWARE IS PROVIDED ``AS IS'' AND WITHOUT ANY EXPRESS OR
* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
* WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHORS OR CONTRIBUTORS BE
* LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
* CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT
* OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
* BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
* WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
* EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*
*/
#include <stddef.h>
#include <string.h>
#include <limits.h>
#include "nhash.h"
#include "mem.h"
/*
* Hash a string into an `unsigned' value. This function is derived
* from the hash function used in the ELF binary object file format
* hash tables. The result size is a 32-bit number if the `unsigned'
* type is big enough to hold 32-bit arbitrary numbers, a 16-bit number
* otherwise.
*/
static unsigned hash_string(char *name)
{
unsigned h = 0;
for (h = 0; *name; name ++) {
unsigned g;
h = (h << 4) + *(unsigned char *)name;
#if UINT_MAX >= 0xffffffffU
g = h & 0xF0000000U;
h ^= (g >> 24);
#else
g = h & 0xF000U;
h ^= (g >> 12);
#endif
h &= ~g;
}
return h;
}
/*
* Each item in the table is a structure beginning with a `hash_item_header'
* structure. Those headers define binary trees such that all left-descendants
* (respectively right-descendants) of a given tree node have an associated
* hash value strictly smaller (respectively greater) than the hash value
* associated with this node.
*
* The `ident' field points to an array of char. The `sizeof(unsigned)'
* first `char' contain a copy of an `unsigned' value which is the hashed
* string, except the least significant bit. When this bit is set to 0,
* the node contains the unique item using that hash value. If the bit
* is set to 1, then there are several items with that hash value.
*
* When several items share the same hash value, they are linked together
* in a linked list by their `left' field. The node contains no data;
* it is a "fake item".
*
* The `char' following the hash value encode the item name for true items.
* For fake items, they contain the pointer to the first true item of the
* corresponding link list (suitably aligned).
*
* There are HTT_NUM_TREES trees; the items are sorted among trees by the
* lest significant bits of their hash value.
*/
static void internal_init(HTT *htt, void (*deldata)(void *), int reduced)
{
htt->deldata = deldata;
if (reduced) {
HTT2 *htt2 = (HTT2 *)htt;
htt2->tree[0] = htt2->tree[1] = NULL;
} else {
unsigned u;
for (u = 0; u < HTT_NUM_TREES; u ++) htt->tree[u] = NULL;
}
}
/* see nhash.h */
void HTT_init(HTT *htt, void (*deldata)(void *))
{
internal_init(htt, deldata, 0);
}
/* see nhash.h */
void HTT2_init(HTT2 *htt, void (*deldata)(void *))
{
internal_init((HTT *)htt, deldata, 1);
}
#define PTR_SHIFT (sizeof(hash_item_header *) * \
((sizeof(unsigned) + sizeof(hash_item_header *) - 1) / \
sizeof(hash_item_header *)))
#define TREE(u) (*(reduced ? ((HTT2 *)htt)->tree + ((u) & 1) \
: htt->tree + ((u) & (HTT_NUM_TREES - 1))))
/*
* Find a node for the given hash value. If `father' is not NULL, fill
* `*father' with a pointer to the node's father.
* If the return value is NULL, then no existing node was found; if `*father'
* is also NULL, the tree is empty. If the return value is not NULL but
* `*father' is NULL, then the found node is the tree root.
*
* If `father' is not NULL, then `*leftson' is filled with 1 if the node
* was looked for as the father left son, 0 otherwise.
*/
static hash_item_header *find_node(HTT *htt, unsigned u,
hash_item_header **father, int *leftson, int reduced)
{
hash_item_header *node = TREE(u);
hash_item_header *nodef = NULL;
int ls;
u &= ~1U;
while (node != NULL) {
unsigned v = *(unsigned *)(node->ident);
unsigned w = v & ~1U;
if (u == w) break;
nodef = node;
if (u < w) {
node = node->left;
ls = 1;
} else {
node = node->right;
ls = 0;
}
}
if (father != NULL) {
*father = nodef;
*leftson = ls;
}
return node;
}
static void *internal_get(HTT *htt, char *name, int reduced)
{
unsigned u = hash_string(name), v;
hash_item_header *node = find_node(htt, u, NULL, NULL, reduced);
if (node == NULL) return NULL;
v = *(unsigned *)(node->ident);
if ((v & 1U) == 0) {
return (strcmp(HASH_ITEM_NAME(node), name) == 0) ? node : NULL;
}
node = *(hash_item_header **)(node->ident + PTR_SHIFT);
while (node != NULL) {
if (strcmp(HASH_ITEM_NAME(node), name) == 0) return node;
node = node->left;
}
return NULL;
}
/* see nhash.h */
void *HTT_get(HTT *htt, char *name)
{
return internal_get(htt, name, 0);
}
/* see nhash.h */
void *HTT2_get(HTT2 *htt, char *name)
{
return internal_get((HTT *)htt, name, 1);
}
/*
* Make an item identifier from its name and its hash value.
*/
static char *make_ident(char *name, unsigned u)
{
size_t n = strlen(name) + 1;
char *ident = getmem(n + sizeof(unsigned));
*(unsigned *)ident = u & ~1U;
memcpy(ident + sizeof(unsigned), name, n);
return ident;
}
/*
* Make an identifier for a fake item, pointing to a true item.
*/
static char *make_fake_ident(unsigned u, hash_item_header *next)
{
char *ident = getmem(PTR_SHIFT + sizeof(hash_item_header *));
*(unsigned *)ident = u | 1U;
*(hash_item_header **)(ident + PTR_SHIFT) = next;
return ident;
}
/*
* Adding an item is straightforward:
* 1. look for its emplacement
* 2. if no node is found, use the item as a new node and link it to the tree
* 3. if a node is found:
* 3.1. if the node is real, check for name inequality, then create a
* fake node and assemble the two-element linked list
* 3.2. if the node is fake, look for the name in the list; if not found,
* add the node at the list end
*/
static void *internal_put(HTT *htt, void *item, char *name, int reduced)
{
unsigned u = hash_string(name), v;
int ls;
hash_item_header *father;
hash_item_header *node = find_node(htt, u, &father, &ls, reduced);
hash_item_header *itemg = item, *pnode;
if (node == NULL) {
itemg->left = itemg->right = NULL;
itemg->ident = make_ident(name, u);
if (father == NULL) {
TREE(u) = itemg;
} else if (ls) {
father->left = itemg;
} else {
father->right = itemg;
}
return NULL;
}
v = *(unsigned *)(node->ident);
if ((v & 1U) == 0) {
if (strcmp(HASH_ITEM_NAME(node), name) == 0)
return node;
pnode = getmem(sizeof *pnode);
pnode->left = node->left;
pnode->right = node->right;
pnode->ident = make_fake_ident(u, node);
node->left = itemg;
node->right = NULL;
itemg->left = itemg->right = NULL;
itemg->ident = make_ident(name, u);
if (father == NULL) {
TREE(u) = pnode;
} else if (ls) {
father->left = pnode;
} else {
father->right = pnode;
}
return NULL;
}
node = *(hash_item_header **)(node->ident + PTR_SHIFT);
while (node != NULL) {
if (strcmp(HASH_ITEM_NAME(node), name) == 0) return node;
pnode = node;
node = node->left;
}
itemg->left = itemg->right = NULL;
itemg->ident = make_ident(name, u);
pnode->left = itemg;
return NULL;
}
/* see nhash.h */
void *HTT_put(HTT *htt, void *item, char *name)
{
return internal_put(htt, item, name, 0);
}
/* see nhash.h */
void *HTT2_put(HTT2 *htt, void *item, char *name)
{
return internal_put((HTT *)htt, item, name, 1);
}
/*
* A fake node subnode list has shrunk to one item only; make the
* node real again.
* fnode the fake node
* node the last remaining node
* father the fake node father (NULL if the fake node is root)
* leftson 1 if the fake node is a left son, 0 otehrwise
* u the hash value for this node
*/
static void shrink_node(HTT *htt, hash_item_header *fnode,
hash_item_header *node, hash_item_header *father, int leftson,
unsigned u, int reduced)
{
node->left = fnode->left;
node->right = fnode->right;
if (father == NULL) {
TREE(u) = node;
} else if (leftson) {
father->left = node;
} else {
father->right = node;
}
freemem(fnode->ident);
freemem(fnode);
}
/*
* Deletion algorithm:
* 1. look for the node; if not found, exit
* 2. if the node is real:
* 2.1. check for equality; exit otherwise
* 2.2. delete the node
* 2.3. promote the leftest of right descendants or rightest of left
* descendants
* 3. if the node is fake:
* 3.1. check the list items for equality; exit otherwise
* 3.2. delete the correct item
* 3.3. if there remains only one item, supress the fake node
*/
static int internal_del(HTT *htt, char *name, int reduced)
{
unsigned u = hash_string(name), v;
int ls;
hash_item_header *father;
hash_item_header *node = find_node(htt, u, &father, &ls, reduced);
hash_item_header *pnode, *fnode, *znode;
char *tmp;
if (node == NULL) return 0;
v = *(unsigned *)(node->ident);
if ((v & 1U) != 0) {
fnode = node;
node = znode = *(hash_item_header **)(node->ident + PTR_SHIFT);
pnode = NULL;
while (node != NULL) {
if (strcmp(HASH_ITEM_NAME(node), name) == 0) break;
pnode = node;
node = node->left;
}
if (node == NULL) return 0;
if (pnode == NULL) {
/*
* We supress the first item in the list.
*/
*(hash_item_header **)(fnode->ident + PTR_SHIFT) =
node->left;
if (node->left->left == NULL) {
shrink_node(htt, fnode, node->left,
father, ls, u, reduced);
}
} else {
pnode->left = node->left;
if (pnode->left == NULL && znode == pnode) {
shrink_node(htt, fnode, pnode,
father, ls, u, reduced);
}
}
} else {
if (strcmp(HASH_ITEM_NAME(node), name) != 0) return 0;
if (node->left != NULL) {
for (znode = node, pnode = node->left; pnode->right;
znode = pnode, pnode = pnode->right);
if (znode != node) {
znode->right = pnode->left;
pnode->left = node->left;
}
pnode->right = node->right;
} else if (node->right != NULL) {
for (znode = node, pnode = node->right; pnode->left;
znode = pnode, pnode = pnode->left);
if (znode != node) {
znode->left = pnode->right;
pnode->right = node->right;
}
pnode->left = node->left;
} else pnode = NULL;
if (father == NULL) {
TREE(u) = pnode;
} else if (ls) {
father->left = pnode;
} else {
father->right = pnode;
}
}
tmp = node->ident;
htt->deldata(node);
freemem(tmp);
return 1;
}
/* see nhash.h */
int HTT_del(HTT *htt, char *name)
{
return internal_del(htt, name, 0);
}
/* see nhash.h */
int HTT2_del(HTT2 *htt, char *name)
{
return internal_del((HTT *)htt, name, 1);
}
/*
* Apply `action()' on all nodes of the tree whose root is given as
* parameter `node'. If `wipe' is non-zero, the nodes are removed
* from memory.
*/
static void scan_node(hash_item_header *node, void (*action)(void *), int wipe)
{
unsigned v;
if (node == NULL) return;
scan_node(node->left, action, wipe);
scan_node(node->right, action, wipe);
v = *(unsigned *)(node->ident);
if ((v & 1U) != 0) {
hash_item_header *pnode, *nnode;
for (pnode = *(hash_item_header **)(node->ident + PTR_SHIFT);
pnode != NULL; pnode = nnode) {
char *tmp = pnode->ident;
nnode = pnode->left;
action(pnode);
if (wipe) freemem(tmp);
}
if (wipe) {
freemem(node->ident);
freemem(node);
}
} else {
char *tmp = node->ident;
action(node);
if (wipe) freemem(tmp);
}
}
/* see nhash.h */
void HTT_scan(HTT *htt, void (*action)(void *))
{
unsigned u;
for (u = 0; u < HTT_NUM_TREES; u ++) {
scan_node(htt->tree[u], action, 0);
}
}
/* see nhash.h */
void HTT2_scan(HTT2 *htt, void (*action)(void *))
{
scan_node(htt->tree[0], action, 0);
scan_node(htt->tree[1], action, 0);
}
/* see nhash.h */
void HTT_kill(HTT *htt)
{
unsigned u;
for (u = 0; u < HTT_NUM_TREES; u ++) {
scan_node(htt->tree[u], htt->deldata, 1);
}
}
/* see nhash.h */
void HTT2_kill(HTT2 *htt)
{
scan_node(htt->tree[0], htt->deldata, 1);
scan_node(htt->tree[1], htt->deldata, 1);
}

132
app/xrdb-cpp/nhash.h Normal file
View File

@ -0,0 +1,132 @@
/*
* (c) Thomas Pornin 2002
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 4. The name of the authors may not be used to endorse or promote
* products derived from this software without specific prior written
* permission.
*
* THIS SOFTWARE IS PROVIDED ``AS IS'' AND WITHOUT ANY EXPRESS OR
* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
* WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHORS OR CONTRIBUTORS BE
* LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
* CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT
* OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
* BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
* WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
* EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*
*/
#ifndef UCPP__NHASH__
#define UCPP__NHASH__
/*
* Each item stored in the hash table should be a structure beginning
* with the following header.
*/
typedef struct hash_item_header_ {
char *ident;
struct hash_item_header_ *left, *right;
} hash_item_header;
/*
* This macro takes as argument a pointer to a hash table item (a
* structure beginning with `hash_item_header') and returns a pointer to
* the item name. This name should be considered as read-only. The
* retrieved pointer can become invalid whenever a new item is inserted
* in or removed from the table.
*/
#define HASH_ITEM_NAME(s) (((hash_item_header *)(s))->ident + sizeof(unsigned))
/*
* Number of lists for the primary hash step. Can be reduced to save more
* memory, or increased to speed things up. It should be a power of 2
* greater or equal than 2 and smaller than UINT_MAX.
*/
#define HTT_NUM_TREES 128
/*
* Type for a hash table.
*/
typedef struct {
void (*deldata)(void *);
hash_item_header *tree[HTT_NUM_TREES];
} HTT;
/*
* Type for a reduced version of HTT with only two binary trees. That
* version has a lower initialization time and is suitable for situation
* where only a limited number of elements will be stored, but new tables
* need frequent initializations.
*/
typedef struct {
void (*deldata)(void *);
hash_item_header *tree[2];
} HTT2;
/*
* Initialize a hash table. The `deldata' parameter should point to a
* function which will be invoked on any item removed from the table;
* that function should take care of the release of memory allocated for
* that item (except the hash_item_header contents, which are handled
* internally).
*/
void HTT_init(HTT *htt, void (*deldata)(void *));
/*
* Link an item into the hash table under the given name. If another
* item of identical name is already present in the table, a pointer to
* that item is returned; otherwise, the new item is linked into the
* table and NULL is returned. The object pointed to by `item' is
* linked from the table, but not the string pointed to by `name'.
*/
void *HTT_put(HTT *htt, void *item, char *name);
/*
* Retrieve an item by name from the hash table. NULL is returned if
* the object is not found.
*/
void *HTT_get(HTT *htt, char *name);
/*
* Remove an item from the hash table. 1 is returned if the item was
* removed, 0 if it was not found.
*/
int HTT_del(HTT *htt, char *name);
/*
* For all items stored within the hash table, invoke the provided
* function with the item as parameter. The function may abort the
* scan by performing a longjmp() to a context encapsulating the
* call to that function.
*/
void HTT_scan(HTT *htt, void (*action)(void *));
/*
* Release the whole table contents. After a call to this function,
* the table is ready to accept new items.
*/
void HTT_kill(HTT *htt);
/*
* The following functions are identical to the HTT_*() functions, except
* that they operate on the reduced HTT2 tables.
*/
void HTT2_init(HTT2 *htt, void (*deldata)(void *));
void *HTT2_put(HTT2 *htt, void *item, char *name);
void *HTT2_get(HTT2 *htt, char *name);
int HTT2_del(HTT2 *htt, char *name);
void HTT2_scan(HTT2 *htt, void (*action)(void *));
void HTT2_kill(HTT2 *htt);
#endif

114
app/xrdb-cpp/sample.c Normal file
View File

@ -0,0 +1,114 @@
/*
* Sample code showing how to use ucpp as an integrated lexer.
* This file is public domain.
*/
/*
* This is an example of how to use ucpp as a preprocessor and lexer
* into another project. The steps are those described in ucpp README
* file. To use this code, compile the ucpp source files with
* STAND_ALONE not defined, and link them with this code. The resulting
* binary will take a C source file as standard input, preprocess it,
* and output each non-whitespace token on stdout, with its numerical
* value (defined as an enum in cpp.h) and its contents. This code
* defines no system include path.
*
* This code supposes that the ucpp files are compiled with PRAGMA_TOKENIZE
* enabled (see the tune.h file).
*/
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include "mem.h"
#include "cpp.h"
int main(int argc, char *argv[])
{
int i, r;
struct lexer_state ls;
/* step 1 */
init_cpp();
/* step 2 */
no_special_macros = 0;
emit_defines = emit_assertions = 0;
/* step 3 -- with assertions */
init_tables(1);
/* step 4 -- no default include path */
init_include_path(0);
/* step 5 -- no need to reset the two emit_* variables set in 2 */
emit_dependencies = 0;
/* step 6 -- we work with stdin, this is not a real filename */
set_init_filename("[stdin]", 0);
/* step 7 -- we make sure that assertions are on, and pragma are
handled */
init_lexer_state(&ls);
init_lexer_mode(&ls);
ls.flags |= HANDLE_ASSERTIONS | HANDLE_PRAGMA | LINE_NUM;
/* step 8 -- input is from stdin */
ls.input = stdin;
/* step 9 -- we do not have any macro to define, but we add any
argument as an include path */
for (i = 1; i < argc; i ++) add_incpath(argv[i]);
/* step 10 -- we are a lexer and we want CONTEXT tokens */
enter_file(&ls, ls.flags);
/* read tokens until end-of-input is reached -- errors (non-zero
return values different from CPPERR_EOF) are ignored */
while ((r = lex(&ls)) < CPPERR_EOF) {
if (r) {
/* error condition -- no token was retrieved */
continue;
}
/* we print each token: its numerical value, and its
string content; if this is a PRAGMA token, the
string content is in fact a compressed token list,
that we uncompress and print. */
if (ls.ctok->type == PRAGMA) {
unsigned char *c = (unsigned char *)(ls.ctok->name);
printf("line %ld: <#pragma>\n", ls.line);
for (; *c; c ++) {
int t = *c;
if (STRING_TOKEN(t)) {
printf(" <%2d> ", t);
for (c ++; *c != PRAGMA_TOKEN_END;
c ++) putchar(*c);
putchar('\n');
} else {
printf(" <%2d> `%s'\n", t,
operators_name[t]);
}
}
} else if (ls.ctok->type == CONTEXT) {
printf("new context: file '%s', line %ld\n",
ls.ctok->name, ls.ctok->line);
} else if (ls.ctok->type == NEWLINE) {
printf("[newline]\n");
} else {
printf("line %ld: <%2d> `%s'\n", ls.ctok->line,
ls.ctok->type,
STRING_TOKEN(ls.ctok->type) ? ls.ctok->name
: operators_name[ls.ctok->type]);
}
}
/* give back memory and exit */
wipeout();
free_lexer_state(&ls);
#ifdef MEM_DEBUG
report_leaks();
#endif
return 0;
}

422
app/xrdb-cpp/tune.h Normal file
View File

@ -0,0 +1,422 @@
/*
* (c) Thomas Pornin 1999 - 2002
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 4. The name of the authors may not be used to endorse or promote
* products derived from this software without specific prior written
* permission.
*
* THIS SOFTWARE IS PROVIDED ``AS IS'' AND WITHOUT ANY EXPRESS OR
* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
* WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHORS OR CONTRIBUTORS BE
* LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
* CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT
* OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
* BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
* WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
* EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*
*/
#ifndef UCPP__TUNE__
#define UCPP__TUNE__
#ifdef UCPP_CONFIG
#include "config.h"
#else
/* ====================================================================== */
/*
* The LOW_MEM macro triggers the use of macro storage which uses less
* memory. It actually also improves performance on large, modern machines
* (due to less cache pressure). This option implies no limitation (except
* on the number of arguments a macro may, which is then limited to 32766)
* so it is on by default. Non-LOW_MEM code is considered deprecated.
*/
#define LOW_MEM
/* ====================================================================== */
/*
* Define AMIGA for systems using "drive letters" at the beginning of
* some paths; define MSDOS on systems with drive letters and using
* backslashes to seperate directory components.
*/
/* #define AMIGA */
/* #define MSDOS */
/* ====================================================================== */
/*
* Define this if your compiler does not know the strftime() function;
* TurboC 2.01 under Msdos does not know strftime().
*/
/* #define NOSTRFTIME */
/* ====================================================================== */
/*
* Buffering: there are two levels of buffering on input and output streams:
* the standard libc buffering (manageable with setbuf() and setvbuf())
* and some buffering provided by ucpp itself. The ucpp buffering uses
* two buffers, of size respectively INPUT_BUF_MEMG and OUTPUT_BUF_MEMG
* (as defined below).
* You can disable one or both of these bufferings by defining the macros
* NO_LIBC_BUF and NO_UCPP_BUF.
*/
/* #define NO_LIBC_BUF */
/* #define NO_UCPP_BUF */
/*
* On Unix stations, the system call mmap() might be used on input files.
* This option is a subclause of ucpp internal buffering. On one station,
* a 10% speed improvement was observed. Do not define this unless the
* host architecture has the following characteristics:
* -- Posix / Single Unix compliance
* -- Text files correspond one to one with memory representation
* If a file is not seekable or not mmapable, ucpp will revert to the
* standard fread() solution.
*
* This feature is still considered beta quality. On some systems where
* files can be bigger than memory address space (mainly, 32-bit systems
* with files bigger than 4 GB), this option makes ucpp fail to operate
* on those extremely large files.
*/
#define UCPP_MMAP
/*
* Performance issues:
* -- On memory-starved systems, such as Minix-i86, do not use ucpp
* buffering; keep only libc buffering.
* -- If you do not use libc buffering, activate the UCPP_MMAP option.
* Note that the UCPP_MMAP option is ignored if ucpp buffering is not
* activated.
*
* On an Athlon 1200 running FreeBSD 4.7, the best performances are
* achieved when libc buffering is activated and/or UCPP_MMAP is on.
*/
/* ====================================================================== */
/*
* Define this if you want ucpp to generate tokenized PRAGMA tokens;
* otherwise, it will generate raw string contents. This setting is
* irrelevant to the stand-alone version of ucpp.
*/
#define PRAGMA_TOKENIZE
/*
* Define this to the special character that marks the end of tokens with
* a string value inside a tokenized PRAGMA token. The #pragma and _Pragma()
* directives which use this character will be a bit more difficult to
* decode (but ucpp will not mind). 0 cannot be used. '\n' is fine because
* it cannot appear inside a #pragma or _Pragma(), since newlines cannot be
* embedded inside tokens, neither directly nor by macro substitution and
* stringization. Besides, '\n' is portable.
*/
#define PRAGMA_TOKEN_END ((unsigned char)'\n')
/*
* Define this if you want ucpp to include encountered #pragma directives
* in its output in non-lexer mode; _Pragma() are translated to equivalent
* #pragma directives.
*/
#define PRAGMA_DUMP
/*
* According to my interpretation of the C99 standard, _Pragma() are
* evaluated wherever macro expansion could take place. However, Neil Booth,
* whose mother language is English (contrary to me) and who is well aware
* of the C99 standard (and especially the C preprocessor) told me that
* it was unclear whether _Pragma() are evaluated inside directives such
* as #if, #include and #line. If you want to disable the evaluation of
* _Pragma() inside such directives, define the following macro.
*/
/* #define NO_PRAGMA_IN_DIRECTIVE */
/*
* The C99 standard mandates that the operator `##' must yield a single,
* valid token, lest undefined behaviour befall upon thy head. Hence,
* for instance, `+ ## +=' is forbidden, because `++=' is not a valid
* token (although it is a valid list of two tokens, `++' and `=').
* However, ucpp only emits a warning for such sin, and unmerges the
* tokens (thus emitting `+' then `+=' for that example). When ucpp
* produces text output, those two tokens will be separated by a space
* character so that the basic rule of text output is preserved: when
* parsed again, text output yields the exact same stream of tokens.
* That extra space is virtual: it does not count as a true whitespace
* token for stringization.
*
* However, it might be desirable, for some uses other than preprocessing
* C source code, not to emit that extra space at all. To make ucpp behave
* that way, define the DSHARP_TOKEN_MERGE macro. Please note that this
* can trigger spurious token merging. For instance, with that macro
* activated, `+ ## +=' will be output as `++=' which, if preprocessed
* again, will read as `++' followed by `='.
*
* All this is irrelevant to lexer mode; and trying to merge incompatible
* tokens is a shooting offence, anyway.
*/
/* #define DSHARP_TOKEN_MERGE */
/* ====================================================================== */
/*
* Define INMACRO_FLAG to include two flags to the structure lexer_state,
* that tell whether tokens come from a macro-replacement, and count those
* macro-replacements.
*/
/* #define INMACRO_FLAG */
/* ====================================================================== */
/*
* Paths where files are looked for by default, when #include is used.
* Typical path is /usr/local/include and /usr/include, in that order.
* If you want to set up no path, define the macro to 0.
*
* For Linux, get gcc includes too, or you will miss things like stddef.h.
* The exact path varies much, depending on the distribution.
*/
#define STD_INCLUDE_PATH "/usr/local/include", "/usr/include"
/* ====================================================================== */
/*
* Arithmetic code for evaluation of #if expressions. Evaluation
* uses either a native machine type, or an emulated two's complement
* type. Division by 0 and overflow on division are considered as errors
* and reported as such. If ARITHMETIC_CHECKS is defined, all other
* operations that imply undefined or implementation-defined behaviour
* are reported as warnings but otherwise performed nonetheless.
*
* For native type evaluation, the following macros should be defined:
* NATIVE_SIGNED the native signed type
* NATIVE_UNSIGNED the native corresponding unsigned type
* NATIVE_UNSIGNED_BITS the native unsigned type width, in bits
* NATIVE_SIGNED_MIN the native signed type minimum value
* NATIVE_SIGNED_MAX the native signed type maximum value
*
* The code in the arith.c file performs some tricky detection
* operations on the native type representation and possible existence
* of a trap representation. These operations assume a C99-compliant
* compiler; on a C90-only compiler, the operations are valid but may
* yield incorrect results. You may force those settings with some
* more macros: see the comments in arith.c (look for "ARCH_DEFINED").
* Remember that this is mostly a non-issue, unless you are building
* ucpp with a pre-C99 cross-compiler and either the host or target
* architecture uses a non-two's complement representation of signed
* integers. Such a combination is pretty rare nowadays, so the best
* you can do is forgetting completely this paragraph and live in peace.
*
*
* If you do not have a handy native type (for instance, you compile ucpp
* with a C90 compiler which lacks the "long long" type, or you compile
* ucpp for a cross-compiler which should support an evaluation integer
* type of a size that is not available on the host machine), you may use
* a simulated type. The type uses two's complement representation and
* may have any width from 2 bits to twice the underlying native type
* width, inclusive (odd widths are allowed). To use an emulated type,
* make sure that NATIVE_SIGNED is not defined, and define the following
* macros:
* SIMUL_ARITH_SUBTYPE the native underlying type to use
* SIMUL_SUBTYPE_BITS the native underlying type width
* SIMUL_NUMBITS the emulated type width
*
* Undefined and implementation-defined behaviours are warned upon, if
* ARITHMETIC_CHECKS is defined. Results are truncated to the type
* width; shift count for the << and >> operators is reduced modulo the
* emulatd type width; right shifting of a signed negative value performs
* sign extension (the result is left-padded with bits set to 1).
*/
/*
* For native type evaluation with a 64-bit "long long" type.
*/
#define NATIVE_SIGNED long long
#define NATIVE_UNSIGNED unsigned long long
#define NATIVE_UNSIGNED_BITS 64
#define NATIVE_SIGNED_MIN (-9223372036854775807LL - 1)
#define NATIVE_SIGNED_MAX 9223372036854775807LL
/*
* For emulation of a 64-bit type using a native 32-bit "unsigned long"
* type.
#undef NATIVE_SIGNED
#define SIMUL_ARITH_SUBTYPE unsigned long
#define SIMUL_SUBTYPE_BITS 32
#define SIMUL_NUMBITS 64
*/
/*
* Comment out the following line if you want to deactivate arithmetic
* checks (warnings upon undefined and implementation-defined
* behaviour). Arithmetic checks slow down a bit arithmetic operations,
* especially multiplications, but this should not be an issue with
* typical C source code.
*/
#define ARITHMETIC_CHECKS
/* ====================================================================== */
/*
* To force signedness of wide character constants, define WCHAR_SIGNEDNESS
* to 0 for unsigned, 1 for signed. By default, wide character constants
* are signed if the native `char' type is signed, and unsigned otherwise.
#define WCHAR_SIGNEDNESS 0
*/
/*
* Standard assertions. They should include one cpu() assertion, one machine()
* assertion (identical to cpu()), and one or more system() assertions.
*
* for Linux/PC: cpu(i386), machine(i386), system(unix), system(linux)
* for Linux/Alpha: cpu(alpha), machine(alpha), system(unix), system(linux)
* for Sparc/Solaris: cpu(sparc), machine(sparc), system(unix), system(solaris)
*
* These are only suggestions. On Solaris, machine() should be defined
* for i386 or sparc (standard system header use such an assertion). For
* cross-compilation, define assertions related to the target architecture.
*
* If you want no standard assertion, define STD_ASSERT to 0.
*/
/*
#define STD_ASSERT "cpu(i386)", "machine(i386)", "system(unix)", \
"system(freebsd)"
*/
/* ====================================================================== */
/*
* System predefined macros. Nothing really mandatory, but some programs
* might rely on those.
* Each string must be either "name" or "name=token-list". If you want
* no predefined macro, define STD_MACROS to 0.
*/
/*
#define STD_MACROS "__FreeBSD=4", "__unix", "__i386", \
"__FreeBSD__=4", "__unix__", "__i386__"
*/
/* ====================================================================== */
/*
* Default flags; HANDLE_ASSERTIONS is required for Solaris system headers.
* See cpp.h for the definition of these flags.
*/
#define DEFAULT_CPP_FLAGS (DISCARD_COMMENTS | WARN_STANDARD \
| WARN_PRAGMA | FAIL_SHARP | MACRO_VAARG \
| CPLUSPLUS_COMMENTS | LINE_NUM | TEXT_OUTPUT \
| KEEP_OUTPUT | HANDLE_TRIGRAPHS \
| HANDLE_ASSERTIONS)
#define DEFAULT_LEXER_FLAGS (DISCARD_COMMENTS | WARN_STANDARD | FAIL_SHARP \
| MACRO_VAARG | CPLUSPLUS_COMMENTS | LEXER \
| HANDLE_TRIGRAPHS | HANDLE_ASSERTIONS)
/* ====================================================================== */
/*
* Define this to use sigsetjmp()/siglongjmp() instead of setjmp()/longjmp().
* This is non-ANSI, but it improves performance on some POSIX system.
* On typical C source code, such improvement is completely negligeable.
*/
/* #define POSIX_JMP */
/* ====================================================================== */
/*
* Maximum value (plus one) of a character handled by the lexer; 128 is
* alright for ASCII native source code, but 256 is needed for EBCDIC.
* 256 is safe in both cases; you will have big problems if you set
* this value to INT_MAX or above. On Minix-i86 or Msdos (small memory
* model), define MAX_CHAR_VAL to 128.
*
* Set MAX_CHAR_VAL to a power of two to increase lexing speed. Beware
* that lexer.c defines a static array of size MSTATE * MAX_CHAR_VAL
* values of type int (MSTATE is defined in lexer.c and is about 40).
*/
#define MAX_CHAR_VAL 128
/*
* If you want some extra character to be considered as whitespace,
* define this macro to that space. On ISO-8859-1 machines, 160 is
* the code for the unbreakable space.
*/
/* #define UNBREAKABLE_SPACE 160 */
/*
* If you want whitespace tokens contents to be recorded (making them
* tokens with a string content), define this. The macro STRING_TOKEN
* will be adjusted accordingly.
* Without this option, whitespace tokens are not even returned by the
* lex() function. This is irrelevant for the non-lexer mode (almost --
* it might slow down a bit ucpp, and with this option, comments will be
* kept inside #pragma directives).
*/
/* #define SEMPER_FIDELIS */
#endif
/* End of options overridable by UCPP_CONFIG and config.h */
/* ====================================================================== */
/*
* Some constants used for memory increment granularity. Increasing these
* values reduces the number of calls to malloc() but increases memory
* consumption.
*
* Values should be powers of 2.
*/
/* for cpp.c */
#define COPY_LINE_LENGTH 80
#define INPUT_BUF_MEMG 8192
#define OUTPUT_BUF_MEMG 8192
#define TOKEN_NAME_MEMG 64 /* must be at least 4 */
#define TOKEN_LIST_MEMG 32
#define INCPATH_MEMG 16
#define GARBAGE_LIST_MEMG 32
#define LS_STACK_MEMG 4
#define FNAME_MEMG 32
/* ====================================================================== */
/* To protect the innocent. */
#if defined(NO_UCPP_BUF) && defined(UCPP_MMAP)
#undef UCPP_MMAP
#endif
#if defined(UCPP_MMAP) || defined(POSIX_JMP)
#ifndef _POSIX_SOURCE
#define _POSIX_SOURCE 1
#endif
#endif
/*
* C90 does not know about the "inline" keyword, but C99 does know,
* and some C90 compilers know it as an extension. This part detects
* these occurrences.
*/
#ifndef INLINE
#if __STDC__ && __STDC_VERSION__ >= 199901L
/* this is a C99 compiler, keep inline unchanged */
#elif defined(__GNUC__)
/* this is GNU gcc; modify inline. The semantics is not identical to C99
but the differences are irrelevant as long as inline functions are static */
#undef inline
#define inline __inline__
#elif defined(__DECC) && defined(__linux__)
/* this is Compaq C under Linux, use __inline__ */
#undef inline
#define inline __inline__
#else
/* unknown compiler -> deactivate inline */
#undef inline
#define inline
#endif
#else
/* INLINE has been set, use its value */
#undef inline
#define inline INLINE
#endif
#endif

212
app/xrdb-cpp/ucpp.1 Normal file
View File

@ -0,0 +1,212 @@
.TH UCPP 1 "Oct 21 2000"
.SH NAME
ucpp \- C preprocessor
.SH SYNOPSIS
.B ucpp
[
.I options
]
[
.I file
]
.SH DESCRIPTION
.LP
.B ucpp
is a C preprocessor mostly compatible with ISO-C99.
It is rather strict and uses only a small amount of memory. It uses
standard input as primary input if no file argument is given.
.SH OPTIONS
There are several classes of options.
.TP
.B Language Options
.TP
.BI \-C
keep comments in the output.
.TP
.BI \-s
if a rogue '#' is encountered, do not emit an error and keep it in
the output.
.TP
.BI \-l
supress the emission of '#line' directives in the output.
.TP
.BI \-lg
convert the '#line' to the gcc-style equivalent.
.TP
.BI \-CC
disable C++-like comments (a '//' begins a comment, up to the end
of the line). Use this option to get closer to C90 behaviour.
.TP
.B \-a, \-na
handle assertions (defined with #assert);
.B \-a
also defines the standard assertions
.I #machine
,
.I #cpu
and
.I #system
(see
.B \-e
to get the local definition of such assertions).
.TP
.BI \-a0
disable assertion support.
.TP
.BI \-V
disable support for macros with a variable number of arguments: in C99,
a macro may be declared with
.I ...
as the last argument; inside the replacement list,
.I __VA_ARGS__
is replaced with the optional extra arguments given in the call to the macro.
Use this option to get closer to C90 behaviour.
.TP
.BI \-u
enable UTF-8 support: with this option, the source is considered as
an ISO/10646 source, encoded in UTF-8. Characters represented as two bytes
or more are considered as alphabetic characters, like letters, and
therefore usable in identifiers. These characters hold the same
syntactic value than the corresponding Universal Character Names.
.TP
.BI \-X
enable
.B \-a, \-u
and
.B \-Y.
This should make
.B ucpp
behave closer to what is requested from a "modern" C preprocessor.
.TP
.BI \-c90
enable
.B \-V
and
.B \-CC,
and do not define
.B __STDC_VERSION__.
This should make
.B ucpp
mimic older C90 behaviour.
.TP
.BI \-t
disable trigraph support; this seems to be required for some legacy code.
.TP
.B Warning Options
.TP
.BI \-wt
emit a final warning when trigraphs are encountered.
.TP
.BI \-wtt
emit warnings for each trigraph encountered.
.TP
.BI \-wa
emit annoying warnings (these are usually useless).
.TP
.BI \-w0
supress standard warnings.
.TP
.B Directory Options
.TP
.BI \-I directory
.TP
.BI "\-I " directory
add
.I directory
to the include path, before the standard include path.
.TP
.BI \-J directory
.TP
.BI "\-J " directory
add
.I directory
to the include path, after the standard include path.
.TP
.BI \-zI
do not use the standard (compile-time) include path.
.TP
.BI \-M
emit only the names of encountered files, separated by spaces; this is
intended for automatic generation of Makefile dependencies.
.TP
.BI \-Ma
do the same as
.B \-M
but also for system files.
.TP
.BI "\-o " file
direct the ouput to
.I file
instead of standard output.
.TP
.B Macro Options
.TP
.BI \-D macro
predefine
.I macro
with content
.B 1.
.TP
.BI \-D macro=def
predefine
.I macro
with the content
.I def.
.TP
.BI \-U macro
undefine
.I macro.
.TP
.BI \-Y
predefine system-dependant macros.
.TP
.BI \-Z
do not predefine special macros such as
.B __TIME__.
.TP
.BI \-A foo(bar)
add
.I foo(bar)
to the list of assertions.
.TP
.BI \-B foo(bar)
remove
.I foo(bar)
of the list of assertions; you may also use
.BI \-B foo
to remove all
.BI \-B foo(xxx)
from the list of assertions.
.TP
.BI \-d
instead of normal output, emit '#define' directives representing all
macros defined during processing.
.TP
.BI \-e
instead of normal output, emit '#assert' directives representing all
assertions defined during processing.
.TP
.B Miscellaneous Options
.TP
.BI \-v
print version number, include path and (optionaly) defined assertions.
.TP
.BI \-h
print some help.
.SH ENVIRONMENT
.PP
.B ucpp
is not itself affected by environment variables. However, it uses
library functions that might be affected, depending on the system.
.SH AUTHOR
Thomas Pornin <pornin@bolet.org>
.SH BUGS
.PP
.B ucpp
is considered stable software. However improbable it is, please report
bugs to the author (possibly with a file that exhibits the problem) if
the latest version, available from this site:
.TP
http://pornin.nerim.net/ucpp/
.PP
has the bug.

196
app/xrdb-cpp/ucppi.h Normal file
View File

@ -0,0 +1,196 @@
/*
* (c) Thomas Pornin 1999 - 2002
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 4. The name of the authors may not be used to endorse or promote
* products derived from this software without specific prior written
* permission.
*
* THIS SOFTWARE IS PROVIDED ``AS IS'' AND WITHOUT ANY EXPRESS OR
* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
* WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHORS OR CONTRIBUTORS BE
* LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
* CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT
* OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
* BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
* WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
* EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*
*/
#ifndef UCPP__UCPPI__
#define UCPP__UCPPI__
#include "tune.h"
#include "cpp.h"
#include "nhash.h"
/*
* A macro represented in a compact form; simple tokens are represented
* by one byte, containing their number. Tokens with a string value are
* followed by the value (string finished by a 0). Macro arguments are
* followed by the argument number (in one byte -- thus implying a hard
* limit of 254 arguments (number 255 is for __VA_ARGS__).
*/
struct comp_token_fifo {
size_t length;
size_t rp;
unsigned char *t;
};
/* These declarations are used only internally by ucpp */
/*
* S_TOKEN(x) checks whether x is a token type with an embedded string
* ttMWS(x) checks whether x is macro whitespace (space, comment...)
* ttWHI(x) checks whether x is whitespace (MWS or newline)
*/
#define S_TOKEN(x) STRING_TOKEN(x)
#define ttMWS(x) ((x) == NONE || (x) == COMMENT || (x) == OPT_NONE)
#define ttWHI(x) (ttMWS(x) || (x) == NEWLINE)
/*
* Function prototypes
*/
/*
* from lexer.c
*/
#define init_cppm ucpp_init_cppm
#define put_char ucpp_put_char
#define discard_char ucpp_discard_char
#define next_token ucpp_next_token
#define grap_char ucpp_grap_char
#define space_char ucpp_space_char
void init_cppm(void);
void put_char(struct lexer_state *, unsigned char);
void discard_char(struct lexer_state *);
int next_token(struct lexer_state *);
int grap_char(struct lexer_state *);
int space_char(int);
/*
* from assert.c
*/
struct assert {
hash_item_header head; /* first field */
size_t nbval;
struct token_fifo *val;
};
#define cmp_token_list ucpp_cmp_token_list
#define handle_assert ucpp_handle_assert
#define handle_unassert ucpp_handle_unassert
#define get_assertion ucpp_get_assertion
#define wipe_assertions ucpp_wipe_assertions
int cmp_token_list(struct token_fifo *, struct token_fifo *);
int handle_assert(struct lexer_state *);
int handle_unassert(struct lexer_state *);
struct assert *get_assertion(char *);
void wipe_assertions(void);
/*
* from macro.c
*/
struct macro {
hash_item_header head; /* first field */
int narg;
char **arg;
int nest;
int vaarg;
#ifdef LOW_MEM
struct comp_token_fifo cval;
#else
struct token_fifo val;
#endif
};
#define print_token ucpp_print_token
#define handle_define ucpp_handle_define
#define handle_undef ucpp_handle_undef
#define handle_ifdef ucpp_handle_ifdef
#define handle_ifndef ucpp_handle_ifndef
#define substitute_macro ucpp_substitute_macro
#define get_macro ucpp_get_macro
#define wipe_macros ucpp_wipe_macros
#define dsharp_lexer ucpp_dsharp_lexer
#define compile_time ucpp_compile_time
#define compile_date ucpp_compile_date
#ifdef PRAGMA_TOKENIZE
#define tokenize_lexer ucpp_tokenize_lexer
#endif
void print_token(struct lexer_state *, struct token *, long);
int handle_define(struct lexer_state *);
int handle_undef(struct lexer_state *);
int handle_ifdef(struct lexer_state *);
int handle_ifndef(struct lexer_state *);
int substitute_macro(struct lexer_state *, struct macro *,
struct token_fifo *, int, int, long);
struct macro *get_macro(char *);
void wipe_macros(void);
extern struct lexer_state dsharp_lexer;
extern char compile_time[], compile_date[];
#ifdef PRAGMA_TOKENIZE
extern struct lexer_state tokenize_lexer;
#endif
/*
* from eval.c
*/
#define strtoconst ucpp_strtoconst
#define eval_expr ucpp_eval_expr
#define eval_line ucpp_eval_line
unsigned long strtoconst(char *);
unsigned long eval_expr(struct token_fifo *, int *, int);
extern long eval_line;
#define eval_exception ucpp_eval_exception
#ifdef POSIX_JMP
#define JMP_BUF sigjmp_buf
#define catch(x) sigsetjmp((x), 0)
#define throw(x) siglongjmp((x), 1)
#else
#define JMP_BUF jmp_buf
#define catch(x) setjmp((x))
#define throw(x) longjmp((x), 1)
#endif
extern JMP_BUF eval_exception;
/*
* from cpp.c
*/
#define token_name ucpp_token_name
#define throw_away ucpp_throw_away
#define garbage_collect ucpp_garbage_collect
#define init_buf_lexer_state ucpp_init_buf_lexer_state
#ifdef PRAGMA_TOKENIZE
#define compress_token_list ucpp_compress_token_list
#endif
char *token_name(struct token *);
void throw_away(struct garbage_fifo *, char *);
void garbage_collect(struct garbage_fifo *);
void init_buf_lexer_state(struct lexer_state *, int);
#ifdef PRAGMA_TOKENIZE
struct comp_token_fifo compress_token_list(struct token_fifo *);
#endif
#define ouch ucpp_ouch
#define error ucpp_error
#define warning ucpp_warning
#endif