godoc: better support for directory trees for user-defined

file systems provided via -path

R=rsc
CC=golang-dev
https://golang.org/cl/2182041

src/cmd/godoc/Makefile

@@ -14,5 +14,6 @@ GOFILES=\
 	mapping.go\
 	snippet.go\
 	spec.go\
+	utils.go\
 
 include ../../Make.cmd

src/cmd/godoc/dirtrees.go

@@ -7,6 +7,7 @@
 package main
 
 import (
+	"bytes"
 	"go/doc"
 	"go/parser"
 	"io/ioutil"
@@ -80,8 +81,18 @@ func firstSentence(s string) string {
 }
 
 
-func newDirTree(path, name string, depth, maxDepth int) *Directory {
-	if depth >= maxDepth {
+type treeBuilder struct {
+	pathFilter func(string) bool
+	maxDepth   int
+}
+
+
+func (b *treeBuilder) newDirTree(path, name string, depth int) *Directory {
+	if b.pathFilter != nil && !b.pathFilter(path) {
+		return nil
+	}
+
+	if depth >= b.maxDepth {
 		// return a dummy directory so that the parent directory
 		// doesn't get discarded just because we reached the max
 		// directory depth
@@ -132,7 +143,7 @@ func newDirTree(path, name string, depth, maxDepth int) *Directory {
 		i := 0
 		for _, d := range list {
 			if isPkgDir(d) {
-				dd := newDirTree(pathutil.Join(path, d.Name), d.Name, depth+1, maxDepth)
+				dd := b.newDirTree(pathutil.Join(path, d.Name), d.Name, depth+1)
 				if dd != nil {
 					dirs[i] = dd
 					i++
@@ -160,20 +171,41 @@ func newDirTree(path, name string, depth, maxDepth int) *Directory {
 }
 
 
+// Maximum directory depth, adjust as needed.
+const maxDirDepth = 24
+
 // newDirectory creates a new package directory tree with at most maxDepth
 // levels, anchored at root. The result tree is pruned such that it only
 // contains directories that contain package files or that contain
-// subdirectories containing package files (transitively). If maxDepth is
+// subdirectories containing package files (transitively). If a non-nil
+// pathFilter is provided, directory paths additionally must be accepted
+// by the filter (i.e., pathFilter(path) must be true). If maxDepth is
 // too shallow, the leaf nodes are assumed to contain package files even if
 // their contents are not known (i.e., in this case the tree may contain
 // directories w/o any package files).
 //
-func newDirectory(root string, maxDepth int) *Directory {
+func newDirectory(root string, pathFilter func(string) bool, maxDepth int) *Directory {
 	d, err := os.Lstat(root)
 	if err != nil || !isPkgDir(d) {
 		return nil
 	}
-	return newDirTree(root, d.Name, 0, maxDepth)
+	b := treeBuilder{pathFilter, maxDepth}
+	return b.newDirTree(root, d.Name, 0)
+}
+
+
+func (dir *Directory) writeLeafs(buf *bytes.Buffer) {
+	if dir != nil {
+		if len(dir.Dirs) == 0 {
+			buf.WriteString(dir.Path)
+			buf.WriteByte('\n')
+			return
+		}
+
+		for _, d := range dir.Dirs {
+			d.writeLeafs(buf)
+		}
+	}
 }
 
 

src/cmd/godoc/doc.go

@@ -61,6 +61,10 @@ The flags are:
 		repository holding the source files.
 	-sync_minutes=0
 		sync interval in minutes; sync is disabled if <= 0
+	-filter=""
+		file containing permitted permitted directory paths
+	-filter_minutes=0
+		filter update interval in minutes; update is disabled if <= 0
 
 The -path flag accepts a list of colon-separated paths; unrooted paths are relative
 to the current working directory. Each path is considered as an additional root for
@@ -76,6 +80,13 @@ as follows:
 	/home/bar/x        -> bar/x
 	/public/x          -> public/x
 
+Paths provided via -path may point to very large file systems that contain
+non-Go files. Creating the subtree of directories with Go packages may take
+a long amount of time. A file containing newline-separated directory paths
+may be provided with the -filter flag; if it exists, only directories
+on those paths are considered. If -filter_minutes is set, the filter_file is
+updated regularly by walking the entire directory tree.
+
 When godoc runs as a web server, it creates a search index from all .go files
 under -goroot (excluding files starting with .). The index is created at startup
 and is automatically updated every time the -sync command terminates with exit

src/cmd/godoc/godoc.go

@@ -22,40 +22,12 @@ import (
 	"regexp"
 	"runtime"
 	"strings"
-	"sync"
 	"template"
 	"time"
 	"utf8"
 )
 
 
-// ----------------------------------------------------------------------------
-// Support types
-
-// An RWValue wraps a value and permits mutually exclusive
-// access to it and records the time the value was last set.
-type RWValue struct {
-	mutex     sync.RWMutex
-	value     interface{}
-	timestamp int64 // time of last set(), in seconds since epoch
-}
-
-
-func (v *RWValue) set(value interface{}) {
-	v.mutex.Lock()
-	v.value = value
-	v.timestamp = time.Seconds()
-	v.mutex.Unlock()
-}
-
-
-func (v *RWValue) get() (interface{}, int64) {
-	v.mutex.RLock()
-	defer v.mutex.RUnlock()
-	return v.value, v.timestamp
-}
-
-
 // ----------------------------------------------------------------------------
 // Globals
 
@@ -81,6 +53,9 @@ var (
 	// file system roots
 	goroot      = flag.String("goroot", runtime.GOROOT(), "Go root directory")
 	path        = flag.String("path", "", "additional package directories (colon-separated)")
+	filter      = flag.String("filter", "godoc.dirlist", "file containing permitted package directory paths")
+	filterMin   = flag.Int("filter_minutes", 0, "filter update interval in minutes; disabled if <= 0")
+	filterDelay delayTime // actual filter update interval in minutes; usually filterDelay == filterMin, but filterDelay may back off exponentially
 
 	// layout control
 	tabwidth = flag.Int("tabwidth", 4, "tab width")
@@ -88,6 +63,7 @@ var (
 	// file system mapping
 	fsMap      Mapping // user-defined mapping
 	fsTree     RWValue // *Directory tree of packages, updated with each sync
+	pathFilter RWValue // filter used when building fsMap directory trees
 
 	// http handlers
 	fileServer http.Handler // default file server
@@ -113,6 +89,134 @@ func registerPublicHandlers(mux *http.ServeMux) {
 }
 
 
+// ----------------------------------------------------------------------------
+// Directory filters
+
+// isParentOf returns true if p is a parent of (or the same as) q
+// where p and q are directory paths.
+func isParentOf(p, q string) bool {
+	n := len(p)
+	return strings.HasPrefix(q, p) && (len(q) <= n || q[n] == '/')
+}
+
+
+// isRelated returns true if p is a parent or child of (or the same as) q
+// where p and q are directory paths.
+func isRelated(p, q string) bool {
+	return isParentOf(p, q) || isParentOf(q, p)
+}
+
+
+func setPathFilter(list []string) {
+	if len(list) == 0 {
+		pathFilter.set(nil)
+		return
+	}
+
+	// TODO(gri) This leads to quadratic behavior.
+	//           Need to find a better filter solution.
+	pathFilter.set(func(path string) bool {
+		for _, p := range list {
+			if isRelated(path, p) {
+				return true
+			}
+		}
+		return false
+	})
+}
+
+
+func getPathFilter() func(string) bool {
+	f, _ := pathFilter.get()
+	if f != nil {
+		return f.(func(string) bool)
+	}
+	return nil
+}
+
+
+// readDirList reads a file containing newline-separated list
+// of directory paths and returns the list of paths.
+func readDirList(filename string) ([]string, os.Error) {
+	contents, err := ioutil.ReadFile(filename)
+	if err != nil {
+		return nil, err
+	}
+	// create list of valid directory names
+	filter := func(path string) bool {
+		d, err := os.Lstat(path)
+		return err == nil && isPkgDir(d)
+	}
+	return canonicalizePaths(strings.Split(string(contents), "\n", -1), filter), nil
+}
+
+
+func updateFilterFile() {
+	// for each user-defined file system mapping, compute
+	// respective directory tree w/o filter for accuracy
+	fsMap.Iterate(func(path string, value *RWValue) bool {
+		value.set(newDirectory(path, nil, maxDirDepth))
+		return true
+	})
+
+	// collect directory tree leaf node paths
+	var buf bytes.Buffer
+	fsMap.Iterate(func(_ string, value *RWValue) bool {
+		v, _ := value.get()
+		if v != nil && v.(*Directory) != nil {
+			v.(*Directory).writeLeafs(&buf)
+		}
+		return true
+	})
+
+	// update filter file
+	// TODO(gri) should write a tmp file and atomically rename
+	err := ioutil.WriteFile(*filter, buf.Bytes(), 0666)
+	if err != nil {
+		log.Stderrf("ioutil.Writefile(%s): %s", *filter, err)
+		filterDelay.backoff(24 * 60) // back off exponentially, but try at least once a day
+	} else {
+		filterDelay.set(*filterMin) // revert to regular filter update schedule
+	}
+}
+
+
+func initDirTrees() {
+	// setup initial path filter
+	if *filter != "" {
+		list, err := readDirList(*filter)
+		if err != nil {
+			log.Stderrf("%s", err)
+		} else if len(list) == 0 {
+			log.Stderrf("no directory paths in file %s", *filter)
+		}
+		setPathFilter(list)
+	}
+
+	// for each user-defined file system mapping, compute
+	// respective directory tree quickly using pathFilter
+	go fsMap.Iterate(func(path string, value *RWValue) bool {
+		value.set(newDirectory(path, getPathFilter(), maxDirDepth))
+		return true
+	})
+
+	// start filter update goroutine, if enabled.
+	if *filter != "" && *filterMin > 0 {
+		filterDelay.set(*filterMin) // initial filter update delay
+		go func() {
+			for {
+				updateFilterFile()
+				delay, _ := syncDelay.get()
+				if *verbose {
+					log.Stderrf("next filter update in %dmin", delay.(int))
+				}
+				time.Sleep(int64(delay.(int)) * 60e9)
+			}
+		}()
+	}
+}
+
+
 // ----------------------------------------------------------------------------
 // Path mapping
 
@@ -1073,14 +1177,35 @@ func (h *httpHandler) getPageInfo(abspath, relpath, pkgname string, mode PageInf
 		// directory tree is present; lookup respective directory
 		// (may still fail if the file system was updated and the
 		// new directory tree has not yet been computed)
-		// TODO(gri) Need to build directory tree for fsMap entries
 		dir = tree.(*Directory).lookup(abspath)
 	}
+	if dir == nil {
+		// the path may refer to a user-specified file system mapped
+		// via fsMap; lookup that mapping and corresponding RWValue
+		// if any
+		var v *RWValue
+		fsMap.Iterate(func(path string, value *RWValue) bool {
+			if isParentOf(path, abspath) {
+				// mapping found
+				v = value
+				return false
+			}
+			return true
+		})
+		if v != nil {
+			// found a RWValue associated with a user-specified file
+			// system; a non-nil RWValue stores a (possibly out-of-date)
+			// directory tree for that file system
+			if tree, _ := v.get(); tree != nil && tree.(*Directory) != nil {
+				dir = tree.(*Directory).lookup(abspath)
+			}
+		}
+	}
 	if dir == nil {
 		// no directory tree present (either early after startup
-		// or command-line mode, or we don't build a tree for the
-		// directory; e.g. google3); compute one level for this page
-		dir = newDirectory(abspath, 1)
+		// or command-line mode, or we don't have a tree for the
+		// directory yet; e.g. google3); compute one level for this page
+		dir = newDirectory(abspath, getPathFilter(), 1)
 	}
 
 	return PageInfo{abspath, plist, past, pdoc, dir.listing(true), h.isPkg, nil}

src/cmd/godoc/main.go

@@ -49,7 +49,7 @@ var (
 	// periodic sync
 	syncCmd   = flag.String("sync", "", "sync command; disabled if empty")
 	syncMin   = flag.Int("sync_minutes", 0, "sync interval in minutes; disabled if <= 0")
-	syncDelay delayTime // actual sync delay in minutes; usually syncDelay == syncMin, but delay may back off exponentially
+	syncDelay delayTime // actual sync interval in minutes; usually syncDelay == syncMin, but syncDelay may back off exponentially
 
 	// network
 	httpAddr   = flag.String("http", "", "HTTP service address (e.g., '"+defaultAddr+"')")
@@ -118,9 +118,6 @@ func exec(c *http.Conn, args []string) (status int) {
 }
 
 
-// Maximum directory depth, adjust as needed.
-const maxDirDepth = 24
-
 func dosync(c *http.Conn, r *http.Request) {
 	args := []string{"/bin/sh", "-c", *syncCmd}
 	switch exec(c, args) {
@@ -130,7 +127,7 @@ func dosync(c *http.Conn, r *http.Request) {
 		// TODO(gri): The directory tree may be temporarily out-of-sync.
 		//            Consider keeping separate time stamps so the web-
 		//            page can indicate this discrepancy.
-		fsTree.set(newDirectory(*goroot, maxDirDepth))
+		fsTree.set(newDirectory(*goroot, nil, maxDirDepth))
 		fallthrough
 	case 1:
 		// sync failed because no files changed;
@@ -257,12 +254,15 @@ func main() {
 			http.Handle("/debug/sync", http.HandlerFunc(dosync))
 		}
 
-		// Initialize directory tree with corresponding timestamp.
+		// Initialize default directory tree with corresponding timestamp.
 		// Do it in two steps:
 		// 1) set timestamp right away so that the indexer is kicked on
 		fsTree.set(nil)
 		// 2) compute initial directory tree in a goroutine so that launch is quick
-		go func() { fsTree.set(newDirectory(*goroot, maxDirDepth)) }()
+		go func() { fsTree.set(newDirectory(*goroot, nil, maxDirDepth)) }()
+
+		// Initialize directory trees for user-defined file systems (-path flag).
+		initDirTrees()
 
 		// Start sync goroutine, if enabled.
 		if *syncCmd != "" && *syncMin > 0 {

src/cmd/godoc/mapping.go

@@ -11,6 +11,7 @@ import (
 	"io"
 	"os"
 	pathutil "path"
+	"sort"
 	"strings"
 )
 
@@ -42,14 +43,19 @@ import (
 //
 // (assuming that file exists).
 //
+// Each individual mapping also has a RWValue associated with it that
+// may be used to store mapping-specific information. See the Iterate
+// method. 
+//
 type Mapping struct {
 	list     []mapping
-	prefixes []string
+	prefixes []string // lazily computed from list
 }
 
 
 type mapping struct {
 	prefix, path string
+	value        *RWValue
 }
 
 
@@ -75,43 +81,16 @@ type mapping struct {
 //	public -> /home/build/public
 //
 func (m *Mapping) Init(paths string) {
-	cwd, _ := os.Getwd() // ignore errors
-
-	pathlist := strings.Split(paths, ":", -1)
-
+	pathlist := canonicalizePaths(strings.Split(paths, ":", -1), nil)
 	list := make([]mapping, len(pathlist))
-	n := 0 // number of mappings
 
-	for _, path := range pathlist {
-		if len(path) == 0 {
-			// ignore empty paths (don't assume ".")
-			continue
-		}
-
-		// len(path) > 0: normalize path
-		if path[0] != '/' {
-			path = pathutil.Join(cwd, path)
-		} else {
-			path = pathutil.Clean(path)
-		}
-
-		// check if mapping exists already
-		var i int
-		for i = 0; i < n; i++ {
-			if path == list[i].path {
-				break
-			}
-		}
-
-		// add mapping if it is new
-		if i >= n {
+	// create mapping list
+	for i, path := range pathlist {
 		_, prefix := pathutil.Split(path)
-			list[n] = mapping{prefix, path}
-			n++
-		}
+		list[i] = mapping{prefix, path, new(RWValue)}
 	}
 
-	m.list = list[0:n]
+	m.list = list
 }
 
 
@@ -134,24 +113,25 @@ func (m *Mapping) PrefixList() []string {
 	// compute the list lazily
 	if m.prefixes == nil {
 		list := make([]string, len(m.list))
-		n := 0 // nuber of prefixes
 
-		for _, e := range m.list {
-			// check if prefix exists already
-			var i int
-			for i = 0; i < n; i++ {
-				if e.prefix == list[i] {
-					break
+		// populate list
+		for i, e := range m.list {
+			list[i] = e.prefix
+		}
+
+		// sort the list and remove duplicate entries
+		sort.SortStrings(list)
+		i := 0
+		prev := ""
+		for _, path := range list {
+			if path != prev {
+				list[i] = path
+				i++
+				prev = path
 			}
 		}
 
-			// add prefix if it is new
-			if i >= n {
-				list[n] = e.prefix
-				n++
-			}
-		}
-		m.prefixes = list[0:n]
+		m.prefixes = list[0:i]
 	}
 
 	return m.prefixes
@@ -166,7 +146,7 @@ func (m *Mapping) Fprint(w io.Writer) {
 }
 
 
-func split(path string) (head, tail string) {
+func splitFirst(path string) (head, tail string) {
 	i := strings.Index(path, "/")
 	if i > 0 {
 		// 0 < i < len(path)
@@ -181,7 +161,7 @@ func split(path string) (head, tail string) {
 // string is returned.
 //
 func (m *Mapping) ToAbsolute(path string) string {
-	prefix, tail := split(path)
+	prefix, tail := splitFirst(path)
 	for _, e := range m.list {
 		switch {
 		case e.prefix == prefix:
@@ -214,3 +194,15 @@ func (m *Mapping) ToRelative(path string) string {
 	}
 	return "" // no match
 }
+
+
+// Iterate calls f for each path and RWValue in the mapping (in uspecified order)
+// until f returns false.
+//
+func (m *Mapping) Iterate(f func(path string, value *RWValue) bool) {
+	for _, e := range m.list {
+		if !f(e.path, e.value) {
+			return
+		}
+	}
+}

src/cmd/godoc/utils.go

@@ -0,0 +1,87 @@
+// Copyright 2010 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+// This file contains support functionality for godoc.
+
+package main
+
+import (
+	"os"
+	pathutil "path"
+	"sort"
+	"strings"
+	"sync"
+	"time"
+)
+
+
+// An RWValue wraps a value and permits mutually exclusive
+// access to it and records the time the value was last set.
+type RWValue struct {
+	mutex     sync.RWMutex
+	value     interface{}
+	timestamp int64 // time of last set(), in seconds since epoch
+}
+
+
+func (v *RWValue) set(value interface{}) {
+	v.mutex.Lock()
+	v.value = value
+	v.timestamp = time.Seconds()
+	v.mutex.Unlock()
+}
+
+
+func (v *RWValue) get() (interface{}, int64) {
+	v.mutex.RLock()
+	defer v.mutex.RUnlock()
+	return v.value, v.timestamp
+}
+
+
+var cwd, _ = os.Getwd() // ignore errors
+
+// canonicalizePaths takes a list of (directory/file) paths and returns
+// the list of corresponding absolute paths in sorted (increasing) order.
+// Relative paths are assumed to be relative to the current directory,
+// empty and duplicate paths as well as paths for which filter(path) is
+// false are discarded. filter may be nil in which case it is not used.
+//
+func canonicalizePaths(list []string, filter func(path string) bool) []string {
+	i := 0
+	for _, path := range list {
+		path = strings.TrimSpace(path)
+		if len(path) == 0 {
+			continue // ignore empty paths (don't assume ".")
+		}
+		// len(path) > 0: normalize path
+		if path[0] != '/' {
+			path = pathutil.Join(cwd, path)
+		} else {
+			path = pathutil.Clean(path)
+		}
+		// we have a non-empty absolute path
+		if filter != nil && !filter(path) {
+			continue
+		}
+		// keep the path
+		list[i] = path
+		i++
+	}
+	list = list[0:i]
+
+	// sort the list and remove duplicate entries
+	sort.SortStrings(list)
+	i = 0
+	prev := ""
+	for _, path := range list {
+		if path != prev {
+			list[i] = path
+			i++
+			prev = path
+		}
+	}
+
+	return list[0:i]
+}