more changes, and cleanup

diff --git a/git-svn-keywords.py b/git-svn-keywords.py
index 765fc625f1eeacc3c41fa1ecaf6903d7914081ac..102e3e7b9ddc4a329d818c9aa8fac77fe00f9749 100755 (executable)
--- a/git-svn-keywords.py
+++ b/git-svn-keywords.py
@@ -23,11 +23,83 @@
 # CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
 # SOFTWARE.
 
-# git svn keyword parsing and populating
+# git svn keyword parsing, populating, and clearing.
 
-import os
+from __future__ import with_statement
+import ConfigParser, errno, os, re, urllib
+import git
 
-if 'GIT_DIR' not in os.environ:
-    print "This is a git post-checkout hook and must be run in the context of git."
-    exit(1)
+VERSION = 0
 
+def parse_svn_unhandled(g):
+    gsk = g.path + '/svn_keywords'
+    try:
+        os.mkdir(gsk)
+    except os.error, e:
+        if e.errno != errno.EEXIST:
+            raise
+
+    config = ConfigParser.ConfigParser()
+    config_path = gsk + '/conf.ini'
+    config.read(config_path)
+
+    files = ConfigParser.ConfigParser()
+    files_path = gsk + '/files.ini'
+
+    if not config.has_section('core'):
+        config.add_section('core')
+    if config.has_option('core', 'version'):
+        ver = config.getint('core', 'ver')
+
+    lastrev = None
+    if ver == VERSION:
+        files.read(files_path)
+        if config.has_option('core', 'lastrev'):
+            lastrev = config.getint('core', 'lastrev')
+
+    with open(g.path + '/svn/git-svn/unhandled.log', 'r') as f:
+        # Compile the regular expressions we'll be using here.
+        re_rev = re.compile("^r(\d+)$")
+        re_keywords = re.compile("^\s+[-+]file_prop: (\S+) svn:keywords ?(\S*)$")
+
+        rev = None
+        for line in f:
+            m = re_rev.match(line)
+            if m:
+                rev = m.group(1)
+                continue
+
+            if (lastrev >= int(rev)):
+                continue
+
+            m = re_keywords.match(line)
+            if m:
+                path = urllib.unquote(m.group(1))
+                keywords = set(urllib.unquote(m.group(2)).split(' '))
+                if not files.has_section(path):
+                    files.add_section(path)
+                files.set(path, rev, keywords)
+
+        lastrev = max(int(rev), lastrev)
+        config.set('core', 'lastrev', lastrev)
+        config.set('core', 'version', VERSION)
+
+    with open(files_path, 'wb') as f:
+        files.write(f)
+
+    with open(config_path, 'wb') as f:
+        config.write(f)
+
+def smudge(g):
+    return 0
+
+def clean(g):
+    return 0
+
+if __name__ == '__main__':
+    try:
+        g = git.Repo()
+    except git.errors.InvalidGitRepositoryError:
+        print "You are not in a git repository or working directory."
+        exit(1)
+    parse_svn_unhandled(g)