hswg: use inotify to watch for changed documents

Now we force the glob to be *.adoc, as well as *.asciidoc,
and there can only be one document directory.

The previous single-run mode is no longer supported.

1 files changed, 195 insertions, 49 deletions

diff --git a/hswg/main.go b/hswg/main.go
index 4626472..7854bf5 100644
--- a/hswg/main.go
+++ b/hswg/main.go
@@ -4,6 +4,7 @@ package main
 
 import (
 	"bytes"
+	"encoding/binary"
 	"encoding/xml"
 	"fmt"
 	"html/template"
@@ -11,10 +12,12 @@ import (
 	"io/ioutil"
 	"log"
 	"os"
+	"os/signal"
 	"path/filepath"
 	"regexp"
 	"sort"
 	"strings"
+	"syscall"
 	"time"
 	"unicode"
 	"unicode/utf8"
@@ -157,7 +160,14 @@ func (e *Entry) Published() *time.Time {
 	}
 }
 
-var extRE = regexp.MustCompile(`\.[^/.]*$`)
+var (
+	globs = []string{"*.adoc", "*.asciidoc"}
+	extRE = regexp.MustCompile(`\.[^/.]*$`)
+)
+
+func pathToName(path string) string {
+	return stripExtension(filepath.Base(path))
+}
 
 func stripExtension(path string) string {
 	return extRE.ReplaceAllString(path, "")
@@ -172,7 +182,8 @@ func resultPath(path string) string {
 
 func makeLink(m *map[string]*Entry, name string) string {
 	e := (*m)[name]
-	return fmt.Sprintf("<a href='%s'>%s</a>", e.PathDestination, name)
+	return fmt.Sprintf("<a href='%s'>%s</a>",
+		filepath.Clean(e.PathDestination), name)
 }
 
 var linkWordRE = regexp.MustCompile(`\b\p{Lu}\p{L}*\b`)
@@ -219,31 +230,28 @@ func renderEntry(name string, e *Entry) error {
 	return nil
 }
 
-func loadEntries(globs []string) (map[string]*Entry, error) {
-	// Create a map from document names to their page entries.
+func makeEntry(path string) *Entry {
+	return &Entry{
+		PathSource:      path,
+		PathDestination: resultPath(path),
+	}
+}
+
+// loadEntries creates a map from document names to their page entries.
+func loadEntries(dirname string) (map[string]*Entry, error) {
 	entries := map[string]*Entry{}
 	for _, glob := range globs {
-		matches, err := filepath.Glob(glob)
+		matches, err := filepath.Glob(filepath.Join(dirname, glob))
 		if err != nil {
-			return nil, fmt.Errorf("%s: %s\n", glob, err)
+			return nil, fmt.Errorf("%s: %s", dirname, err)
 		}
 		for _, path := range matches {
-			name := stripExtension(filepath.Base(path))
+			name := pathToName(path)
 			if conflict, ok := entries[name]; ok {
-				return nil, fmt.Errorf("%s: conflicts with %s\n",
+				return nil, fmt.Errorf("%s: conflicts with %s",
 					name, conflict.PathSource)
 			}
-			entries[name] = &Entry{
-				PathSource:      path,
-				PathDestination: resultPath(path),
-				backlinks:       map[string]bool{},
-			}
-		}
-	}
-
-	for name, e := range entries {
-		if err := renderEntry(name, e); err != nil {
-			return nil, err
+			entries[name] = makeEntry(path)
 		}
 	}
 	return entries, nil
@@ -269,21 +277,6 @@ func writeEntry(e *Entry, t *template.Template,
 	return t.Execute(f, e)
 }
 
-func finalizeEntries(entries *map[string]*Entry) {
-	for name, e := range *entries {
-		// Expand LinkWords anywhere between <tags>.
-		// We want something like the inverse of Regexp.ReplaceAllStringFunc.
-		raw, last, expanded := e.raw, 0, bytes.NewBuffer(nil)
-		for _, where := range tagRE.FindAllIndex(raw, -1) {
-			_, _ = expanded.Write(expand(entries, name, raw[last:where[0]]))
-			_, _ = expanded.Write(raw[where[0]:where[1]])
-			last = where[1]
-		}
-		_, _ = expanded.Write(expand(entries, name, raw[last:]))
-		e.Content = template.HTML(expanded.String())
-	}
-}
-
 func writeIndex(path string, t *template.Template,
 	entries *map[string]*Entry) error {
 	// Reorder entries reversely, primarily by date, secondarily by filename.
@@ -315,10 +308,144 @@ func writeIndex(path string, t *template.Template,
 		return err
 	}
 
-	// TODO(p): Splitting content to categories would be nice.
+	// TODO(p): Splitting content to categories would be nice. Or tags.
 	return t.Execute(f, ordered)
 }
 
+func finalizeEntries(entries *map[string]*Entry, t *template.Template,
+	indexPath string, indexT *template.Template) {
+	for name, e := range *entries {
+		e.backlinks = map[string]bool{}
+		if e.raw == nil {
+			if err := renderEntry(name, e); err != nil {
+				log.Printf("%s: %s\n", name, err)
+			}
+		}
+	}
+	for name, e := range *entries {
+		// Expand LinkWords anywhere between <tags>.
+		// We want something like the inverse of Regexp.ReplaceAllStringFunc.
+		raw, last, expanded := e.raw, 0, bytes.NewBuffer(nil)
+		for _, where := range tagRE.FindAllIndex(raw, -1) {
+			_, _ = expanded.Write(expand(entries, name, raw[last:where[0]]))
+			_, _ = expanded.Write(raw[where[0]:where[1]])
+			last = where[1]
+		}
+		_, _ = expanded.Write(expand(entries, name, raw[last:]))
+		e.Content = template.HTML(expanded.String())
+	}
+	for name, e := range *entries {
+		// Don't overwrite failed renders.
+		if e.raw == nil {
+			continue
+		}
+		if err := writeEntry(e, t, entries); err != nil {
+			log.Printf("%s: %s\n", name, err)
+		}
+	}
+	if err := writeIndex(indexPath, indexT, entries); err != nil {
+		log.Printf("%s: %s\n", indexPath, err)
+	}
+}
+
+type watchEvent struct {
+	path    string // the path of the target
+	present bool   // if not, the file has been removed
+}
+
+func dispatchEvents(dirname string, r io.Reader, ch chan<- *watchEvent) error {
+	var e syscall.InotifyEvent
+	for {
+		// FIXME(p): This has to respect the machine's endianness.
+		// Perhaps use the unsafe package.
+		err := binary.Read(r, binary.LittleEndian, &e)
+		if err == io.EOF {
+			return nil
+		}
+		if err != nil {
+			return err
+		}
+
+		switch {
+		case e.Mask&syscall.IN_IGNORED != 0:
+			return fmt.Errorf("watch removed by kernel")
+		case e.Mask&syscall.IN_Q_OVERFLOW != 0:
+			log.Println("inotify: queue overflowed")
+			ch <- nil
+			continue
+		case e.Len == 0:
+			continue
+		}
+
+		base := make([]byte, e.Len)
+		if n, err := r.Read(base); err != nil {
+			return err
+		} else if n < int(e.Len) {
+			return fmt.Errorf("short read")
+		}
+
+		basename, interesting := string(base[:bytes.IndexByte(base, 0)]), false
+		for _, glob := range globs {
+			if matches, _ := filepath.Match(glob, basename); matches {
+				interesting = true
+			}
+		}
+		if !interesting {
+			continue
+		}
+
+		event := &watchEvent{path: filepath.Join(dirname, basename)}
+		if e.Mask&syscall.IN_MODIFY != 0 || e.Mask&syscall.IN_MOVED_TO != 0 ||
+			e.Mask&syscall.IN_CLOSE_WRITE != 0 {
+			event.present = true
+			ch <- event
+		}
+		if e.Mask&syscall.IN_DELETE != 0 || e.Mask&syscall.IN_MOVED_FROM != 0 {
+			event.present = false
+			ch <- event
+		}
+	}
+}
+
+func watchDirectory(dirname string) (<-chan *watchEvent, error) {
+	inotifyFD, err := syscall.InotifyInit1(0)
+	if err != nil {
+		return nil, err
+	}
+
+	// We're ignoring IN_CREATE, as it doesn't seem to be useful,
+	// and we're leaving out IN_MODIFY since VIM always triggers IN_CLOSE_WRITE,
+	// saving us from having to coalesce plentiful similar events.
+	_, err = syscall.InotifyAddWatch(inotifyFD, dirname, syscall.IN_ONLYDIR|
+		syscall.IN_MOVE|syscall.IN_DELETE|syscall.IN_CLOSE_WRITE)
+	if err != nil {
+		return nil, err
+	}
+
+	inotifyFile := os.NewFile(uintptr(inotifyFD), "inotify")
+	buf := make([]byte, syscall.SizeofInotifyEvent+syscall.PathMax+1)
+	ch := make(chan *watchEvent)
+	go func() {
+		// Trigger an initial rendering run.
+		ch <- nil
+
+		defer close(ch)
+		for {
+			n, err := inotifyFile.Read(buf)
+			if err != nil {
+				log.Println(err)
+				return
+			}
+			err = dispatchEvents(dirname, bytes.NewReader(buf[:n]), ch)
+			if err != nil {
+				log.Printf("inotify: %s\n", err)
+				return
+			}
+		}
+	}()
+	return ch, nil
+}
+
 func singleFile() {
 	html, meta, err := Render(os.Stdin, configuration.NewConfiguration())
 	if err != nil {
@@ -336,11 +463,11 @@ func main() {
 		singleFile()
 		return
 	}
-	if len(os.Args) < 4 {
-		log.Fatalf("usage: %s TEMPLATE INDEX GLOB...\n", os.Args[0])
+	if len(os.Args) != 4 {
+		log.Fatalf("usage: %s TEMPLATE INDEX DIRECTORY\n", os.Args[0])
 	}
 
-	argTemplate, argIndex, argGlobs := os.Args[1], os.Args[2], os.Args[3:]
+	argTemplate, argIndex, argDirectory := os.Args[1], os.Args[2], os.Args[3]
 
 	// Read a template for entries.
 	header, err := ioutil.ReadFile(argTemplate)
@@ -362,21 +489,40 @@ func main() {
 		log.Fatalln(err)
 	}
 
-	// Process all entries.
-	entries, err := loadEntries(argGlobs)
+	// Re-render as needed, avoid having to trigger anything manually.
+	var entries map[string]*Entry
+	directoryWatch, err := watchDirectory(argDirectory)
 	if err != nil {
 		log.Fatalln(err)
 	}
 
-	finalizeEntries(&entries)
-	for _, e := range entries {
-		if err := writeEntry(e, tmplEntry, &entries); err != nil {
-			log.Fatalln(err)
-		}
-	}
+	signals := make(chan os.Signal)
+	signal.Notify(signals, syscall.SIGINT, syscall.SIGHUP, syscall.SIGTERM)
+	for {
+		select {
+		case <-signals:
+			os.Exit(0)
+		case event, ok := <-directoryWatch:
+			if !ok {
+				os.Exit(1)
+			}
 
-	// Write an index.
-	if err := writeIndex(argIndex, tmplIndex, &entries); err != nil {
-		log.Fatalln(err)
+			if event == nil {
+				log.Println("reloading all files")
+				if entries, err = loadEntries(argDirectory); err != nil {
+					log.Println(err)
+				}
+			} else if event.present {
+				log.Printf("updating %s\n", event.path)
+				entries[pathToName(event.path)] = makeEntry(event.path)
+			} else {
+				log.Printf("removing %s\n", event.path)
+				delete(entries, pathToName(event.path))
+				os.Remove(resultPath(event.path))
+			}
+
+			finalizeEntries(&entries, tmplEntry, argIndex, tmplIndex)
+			log.Println("done")
+		}
 	}
 }