fshash.in, fshash.1: Evict cache entries identified by diff files.

A shiny new option which we can use to recover automatically from
cache-staleness problems.

diff --git a/fshash.1 b/fshash.1
index f9d02bf..2ceb175 100644 (file)
--- a/fshash.1
+++ b/fshash.1
@@ -16,6 +16,14 @@
 .IR hash ]
 .RI [ file
 \&...]
+.br
+.B fshash
+.RB \-u
+.B \-c
+.I cache
+.RB [ \-H
+.IR hash ]
+.RI [ dir ]
 .SH DESCRIPTION
 The
 .B fshash
@@ -94,6 +102,14 @@ function, which can be any hash function supported by Python's
 .BR hashlib .
 If this option may be omitted then the hash is read from the cache file;
 if there is no cache file either, then an error is reported.
+.TP
+.B \-u, \-\-udiff
+Rather than produce a manifest, read a unified
+.BR diff (1)
+from standard input, and clear from the cache all files mentioned as
+being different.  Filenames in the diff are considered relative to
+.I dir ,
+defaulting to the current working directory.
 .PP
 Positional arguments are interpreted as files and directories to be
 processed, in order.  A directory name which ends in
@@ -110,7 +126,6 @@ hexadecimal.  For other kinds of filesystem object, a description of the
 object type and any special information about it, in square brackets,
 and padded with spaces so as to take the same width as a hash; see
 below for details.
-as follows.
 .hP 2.
 A `virtual inode identifier': a string which will be the same in two
 lines if and only if they represent hard links to the same underlying

diff --git a/fshash.in b/fshash.in
index 4f8b1c7..525b0f0 100644 (file)
--- a/fshash.in
+++ b/fshash.in
@@ -27,6 +27,7 @@ from sys import argv, exit, stdin, stdout, stderr
 import os as OS
 import re as RX
 import time as T
+import errno as E
 import stat as ST
 import optparse as OP
 import hashlib as H
@@ -331,6 +332,11 @@ class HashCache (object):
     if not me._db:
       die("no cache database")
 
+  def forget(me, ino):
+    me.need_db()
+    c = me._db.cursor()
+    c.execute('DELETE FROM hash WHERE ino = ?', [ino])
+
   def reset(me):
     me.need_db()
     c = me._db.cursor()
@@ -450,6 +456,109 @@ class Reporter (object):
       fmt.mtime(), fmt.size(), fmt.name())
 
 ###--------------------------------------------------------------------------
+### Database clearing from diff files.
+
+R_HUNK = RX.compile(r'^@@ -\d+,(\d+) \+\d+,(\d+) @@$')
+
+def clear_entry(db, lno, line):
+
+  good = True
+
+  if line.startswith('['):
+    pos = line.find(']')
+    if pos < 0:
+      moan("failed to parse file entry (type field; line %d)" % lno)
+      return False
+    ty = line[1:pos].strip()
+    rest = line[pos + 1:]
+    hash = None
+  else:
+    ff = line.split(None, 1)
+    if len(ff) != 2:
+      moan("failed to parse file entry (field split; line %d)" % lno)
+      return False
+    ty = 'regular-file'
+    hash, rest = ff
+
+  ff = rest.split(None, 5)
+  if len(ff) != 6:
+    moan("failed to parse file entry (field split; line %d)" % lno)
+    return False
+  ino, mode, uidgid, mtime, sz, name = ff
+
+  if ty != 'symbolic-link':
+    target = None
+  else:
+    nn = name.split(' -> ', 1)
+    if len(nn) != 2:
+      moan("failed to parse file entry (name split; line %d)" % lno)
+      return False
+    name, target = nn
+    target = target.decode('string_escape')
+  name = name.decode('string_escape')
+
+  try:
+    st = OS.lstat(name)
+  except OSError, e:
+    moan("failed to stat `%s': %s" % (name, e.strerror))
+    if e.errno != E.ENOENT: good = False
+  else:
+    print "Clear cache entry for `%s'" % name
+    db.forget(st.st_ino)
+
+  return good
+
+def clear_cache(db):
+
+  ## Work through the input diff file one line at a time.
+  diffstate = 'gap'
+  lno = 0
+  good = True
+  for line in stdin:
+    if line.endswith('\n'): line = line[:-1]
+    lno += 1
+
+    ## We're in a gap between hunks.  Find a hunk header and extract the line
+    ## counts.
+    if diffstate == 'gap':
+      m = R_HUNK.match(line)
+      if m:
+        oldlines = int(m.group(1))
+        newlines = int(m.group(2))
+        diffstate = 'hunk'
+        hdrlno = lno
+
+    ## We're in a hunk.  Keep track of whether we've reached the end, and
+    ## discard entries from the cache for mismatching lines.
+    elif diffstate == 'hunk':
+      if len(line) == 0:
+        moan("empty line in diff hunk (line %d)" % lno)
+        good = False
+      ty = line[0]
+      if ty == ' ':
+        oldlines -= 1; newlines -= 1
+      elif ty == '+':
+        newlines -= 1
+        if not clear_entry(db, lno, line[1:]): good = False
+      elif ty == '-':
+        oldlines -= 1
+        if not clear_entry(db, lno, line[1:]): good = False
+      else:
+        moan("incomprehensible line in diff hunk (line %d)" % lno)
+        good = false
+      if oldlines < 0 or newlines < 0:
+        moan("inconsistent lengths in diff hunk header (line %d)" % hdrlno)
+        good = False
+      if oldlines == newlines == 0:
+        diffstate = 'gap'
+
+  if diffstate == 'hunk':
+    moan("truncated diff hunk (started at line %d)" % hdrlno)
+    good = False
+
+  return good
+
+###--------------------------------------------------------------------------
 ### Main program.
 
 FMTMAP = {
@@ -457,7 +566,7 @@ FMTMAP = {
   'find0': lambda f: enum_find0(stdin, f)
 }
 op = OP.OptionParser(
-  usage = '%prog [-a] [-c CACHE] [-f FORMAT] [-H HASH] [FILE ...]',
+  usage = '%prog [-au] [-c CACHE] [-f FORMAT] [-H HASH] [FILE ...]',
   version = '%%prog, version %s' % VERSION,
   description = '''\
 Print a digest of a filesystem (or a collection of specified files) to
@@ -475,24 +584,36 @@ for short, long, props in [
   ('-f', '--files', { 'dest': 'files', 'metavar': 'FORMAT',
                       'type': 'choice', 'choices': FMTMAP.keys(),
                       'help': 'read files to report in the given FORMAT' }),
+  ('-u', '--udiff', { 'action': 'store_true', 'dest': 'udiff',
+                      'help': 'read diff from stdin, clear cache entries' }),
   ('-H', '--hash', { 'dest': 'hash', 'metavar': 'HASH',
                      ##'type': 'choice', 'choices': H.algorithms,
                      'help': 'use HASH as the hash function' })]:
   op.add_option(short, long, **props)
 opts, args = op.parse_args(argv)
 
-if not opts.files and len(args) <= 1:
-  die("no filename sources: nothing to do")
-db = HashCache(opts.cache, opts.hash)
-if opts.all:
-  db.reset()
-rep = Reporter(db)
-if opts.files:
-  FMTMAP[opts.files](rep.file)
-for dir in args[1:]:
-  enum_walk(dir, rep.file)
-if opts.all:
-  db.prune()
-db.flush()
+if opts.udiff:
+  if opts.cache is None or opts.all or opts.files or len(args) > 2:
+    die("incompatible options: `-u' requires `-c CACHE', forbids others")
+  db = HashCache(opts.cache, opts.hash)
+  if len(args) == 2: OS.chdir(args[1])
+  good = True
+  if not clear_cache(db): good = False
+  if good: db.flush()
+  else: exit(2)
+else:
+  if not opts.files and len(args) <= 1:
+    die("no filename sources: nothing to do")
+  db = HashCache(opts.cache, opts.hash)
+  if opts.all:
+    db.reset()
+  rep = Reporter(db)
+  if opts.files:
+    FMTMAP[opts.files](rep.file)
+  for dir in args[1:]:
+    enum_walk(dir, rep.file)
+  if opts.all:
+    db.prune()
+  db.flush()
 
 ###----- That's all, folks --------------------------------------------------