Extend worked example and make several refinements

* Verbosity control
* Automatically update working tree
* Identify orphan files in filter-clean and pass through so they don't
  show up as suprious diffs.

1 files changed, 82 insertions, 34 deletions

diff --git a/git-fat b/git-fat
index ab36470..84be082 100755
--- a/git-fat
+++ b/git-fat
@@ -1,6 +1,6 @@
 #!/usr/bin/env python
 
-from __future__ import print_function
+from __future__ import print_function, with_statement
 
 import sys
 import hashlib
@@ -8,6 +8,7 @@ import tempfile
 import os
 import subprocess
 import shlex
+import shutil
 
 BLOCK_SIZE = 4096
 
@@ -15,10 +16,7 @@ def verbose_stderr(*args, **kwargs):
     return print(*args, file=sys.stderr, **kwargs)
 def verbose_ignore(*args, **kwargs):
     pass
-verbose = verbose_ignore
 
-def gitroot():
-    return subprocess.check_output('git rev-parse --show-toplevel'.split()).strip()
 def mkdir_p(path):
     import errno
     try:
@@ -45,7 +43,8 @@ def cat(instream, outstream):
 class GitFat(object):
     DecodeError = RuntimeError
     def __init__(self):
-        self.gitroot = gitroot()
+        self.verbose = verbose_stderr if os.environ.get('GIT_FAT_VERBOSE') else verbose_ignore
+        self.gitroot = subprocess.check_output('git rev-parse --show-toplevel'.split()).strip()
         self.objdir = os.path.join(self.gitroot, '.git', 'fat', 'objects')
         self.magiclen = len(self.encode(hashlib.sha1('dummy').hexdigest()))
     def setup(self):
@@ -62,12 +61,16 @@ class GitFat(object):
             return remote
         except ConfigParser.NoSectionError:
             raise RuntimeError('No rsync.remote in %s' % cfgpath)
+    def revparse(self, revname):
+        return subprocess.check_output(['git', 'rev-parse', revname]).strip()
     def encode(self, digest):
         return '#$# git-fat %s\n' % digest
-    def decode(self, string):
+    def decode(self, string, noraise=False):
         cookie = '#$# git-fat '
         if string.startswith(cookie):
             return string[len(cookie):].split()[0]
+        elif noraise:
+            return None
         else:
             raise GitFat.DecodeError('Could not decode %s' % (string))
     def decode_stream(self, stream):
@@ -94,38 +97,54 @@ class GitFat(object):
         h = hashlib.new('sha1')
         fd, tmpname = tempfile.mkstemp(dir=self.objdir)
         try:
+            ishanging = False
+            cached = False                # changes to True when file is cached, means we
             with os.fdopen(fd, 'w') as cache:
-                for block in readblocks(sys.stdin):
+                outstream = cache
+                blockiter = readblocks(sys.stdin)
+                # Check whether this file is hanging
+                block = next(blockiter)
+                if self.decode(block[0:self.magiclen], noraise=True):
+                    ishanging = True
+                    outstream = sys.stdout
+                h.update(block)
+                outstream.write(block)
+                for block in blockiter:
                     h.update(block)
-                    cache.write(block)
+                    outstream.write(block)
+                outstream.flush()
             digest = h.hexdigest()
             objfile = os.path.join(self.objdir, digest)
-            os.rename(tmpname, objfile)
-            verbose('git-fat filter-clean: caching to %s' % objfile)
-            sys.stdout.write(self.encode(digest))
-            sys.stdout.flush()
-        except:
-            raise
-        #os.remove(tmpname)
+            if not ishanging:
+                os.rename(tmpname, objfile)
+                cached = True
+                self.verbose('git-fat filter-clean: caching to %s' % objfile)
+                sys.stdout.write(self.encode(digest))
+        finally:
+            if not cached:
+                os.remove(tmpname)
+
     def cmd_smudge(self):
         self.setup()
         result = self.decode_stream(sys.stdin)
         if isinstance(result, str):       # We got a digest
             objfile = os.path.join(self.objdir, result)
-            verbose('git-fat filter-smudge: restoring from %s' % objfile)
+            self.verbose('git-fat filter-smudge: restoring from %s' % objfile)
             try:
                 cat(open(objfile), sys.stdout)
             except:
                 sys.stdout.write(self.encode(result))   # could leave a better notice about how to recover this file
         else:                             # We have an iterable over the original input.
-            verbose('git-fat filter-smudge: not a managed file')
+            self.verbose('git-fat filter-smudge: not a managed file')
             cat(result, sys.stdout)
     def catalog_objects(self):
         return set(os.listdir(self.objdir))
-    def referenced_objects(self, rev=None):
+    def referenced_objects(self, rev=None, all=False):
         referenced = set()
-        if rev is None:
+        if all:
             rev = '--all'
+        elif rev is None:
+            rev = self.revparse('HEAD')
         p1 = subprocess.Popen(['git','rev-list','--objects',rev], stdout=subprocess.PIPE)
         p2 = subprocess.Popen(['git','cat-file','--batch-check'], stdin=subprocess.PIPE, stdout=subprocess.PIPE)
         for line in p1.communicate()[0].splitlines():
@@ -136,17 +155,26 @@ class GitFat(object):
                 fathash = self.decode(subprocess.check_output(['git', 'cat-file', '-p', objhash]))
                 referenced.add(fathash)
         return referenced
+    def orphan_files(self):
+        'generator for all orphan placeholders in the working tree'
+        for fname in subprocess.check_output(['git', 'ls-files']).splitlines():
+            digest = self.decode_file(fname)
+            if digest:
+                yield (digest, fname)
     def cmd_status(self, args):
         self.setup()
         catalog = self.catalog_objects()
-        referenced = self.referenced_objects()
+        refargs = dict()
+        if '--all' in args:
+            refargs['all'] = True
+        referenced = self.referenced_objects(**refargs)
         garbage = catalog - referenced
         orphans = referenced - catalog
         if '--all' in args:
             for obj in referenced:
                 print(obj)
         if orphans:
-            print('Orphan files:')
+            print('Orphan objects:')
             for orph in orphans:
                 print('    ' + orph)
         if garbage:
@@ -156,23 +184,44 @@ class GitFat(object):
     def cmd_push(self):
         'Push anything that I have stored and referenced'
         self.setup()
-        files = self.referenced_objects() & self.catalog_objects()
+        # Pushing *all* objects because it's safer. Could implement partial push.
+        files = self.referenced_objects(all=True) & self.catalog_objects()
         remote = self.get_rsync()
-        verbose('Pushing to %s' % (remote))
+        self.verbose('Pushing to %s' % (remote))
         cmd = ['rsync', '--progress', '--ignore-existing', '--from0', '--files-from=-', self.objdir + '/', remote]
         p = subprocess.Popen(cmd, stdin=subprocess.PIPE)
         p.communicate(input='\x00'.join(files))
-    def checkout(self):
-        files = subprocess.check_output(['git', 'ls-files', ''])
-    def cmd_pull(self):
+    def checkout(self, show_orphans=False):
+        'Update any stale files in the present working tree'
+        orphans = []
+        for digest, fname in self.orphan_files():
+            objpath = os.path.join(self.objdir, digest)
+            if os.access(objpath, os.R_OK):
+                print('Restoring %s -> %s' % (digest, fname))
+                shutil.copy(objpath, fname)
+            elif show_orphans:
+                print('Data unavailable: %s %s' % (digest,fname))
+        subprocess.call(['git', 'checkout', '.'])
+    def cmd_pull(self, args):
         'Pull anything that I have referenced, but not stored'
         self.setup()
-        files = self.referenced_objects() - self.catalog_objects()
+        refargs = dict()
+        if '--all' in args:
+            refargs['all'] = True
+        for arg in args:
+            if arg.startswith('-') or len(arg) != 40:
+                continue
+            rev = self.revparse(arg)
+            if rev:
+                refargs['rev'] = rev
+        files = self.referenced_objects(**refargs) - self.catalog_objects()
         remote = self.get_rsync()
         cmd = ['rsync', '--progress', '--ignore-existing', '--from0', '--files-from=-', remote + '/', self.objdir]
         p = subprocess.Popen(cmd, stdin=subprocess.PIPE)
         p.communicate(input='\x00'.join(files))
         self.checkout()
+    def cmd_checkout(self, args):
+        self.checkout(show_orphans=True)
     def cmd_gc(self):
         garbage = self.catalog_objects() - self.referenced_objects()
         print('Unreferenced objects to remove: %d' % len(garbage))
@@ -189,24 +238,23 @@ class GitFat(object):
             ])
 
 if __name__ == '__main__':
-    if os.environ.get('GIT_FAT_VERBOSE'):
-        global verbose
-        verbose = verbose_stderr
     fat = GitFat()
     cmd = sys.argv[1] if len(sys.argv) > 1 else ''
     if cmd == 'filter-clean':
         fat.cmd_clean()
     elif cmd == 'filter-smudge':
         fat.cmd_smudge()
+    elif cmd == 'init':
+        fat.cmd_init()
     elif cmd == 'status':
         fat.cmd_status(sys.argv[2:])
     elif cmd == 'push':
         fat.cmd_push()
     elif cmd == 'pull':
-        fat.cmd_pull()
+        fat.cmd_pull(sys.argv[2:])
     elif cmd == 'gc':
         fat.cmd_gc()
-    elif cmd == 'init':
-        fat.cmd_init()
+    elif cmd == 'checkout':
+        fat.cmd_checkout(sys.argv[2:])
     else:
-        print('Usage: git fat [status|push|pull|gc|init]', file=sys.stderr)
+        print('Usage: git fat [init|status|push|pull|gc|checkout]', file=sys.stderr)