Rewrite committags

I'm tired of trying to be fast and clever.  This version is dumb and
heartbreakingly slow - but it works.

1 files changed, 76 insertions, 21 deletions

diff --git a/committags b/committags
index 39a532d..0ddd36f 100755
--- a/committags
+++ b/committags
@@ -1,13 +1,16 @@
-#!/usr/bin/python
+#!/usr/bin/pypy
 #
 # Generate a database of commits and major versions they went into.
 #
+# This is the painfully slow reworked brute-force version that
+# takes forever to run, but which hopefully gets the right results
+#
 # committags [git-args]
 #
 # This code is part of the LWN git data miner.
 #
-# Copyright 2007-11 Eklektix, Inc.
-# Copyright 2007-11 Jonathan Corbet <corbet@lwn.net>
+# Copyright 2007-13 Eklektix, Inc.
+# Copyright 2007-13 Jonathan Corbet <corbet@lwn.net>
 #
 # This file may be distributed under the terms of the GNU General
 # Public License, version 2.
@@ -16,30 +19,82 @@ import sys
 import re
 import os
 import pickle
+import argparse
+
+#
+# Arg parsing stuff.
+#
+def setupargs():
+    p = argparse.ArgumentParser()
+    #
+    # -d for the database
+    # -l to load it before running
+    #
+    p.add_argument('-d', '--database', help = 'Database name',
+                   required = False, default = 'committags.db')
+    p.add_argument('-l', '--load', help = 'Load database at startup',
+                   default = False, action = 'store_true')
+    #
+    # Args for git?
+    #
+    p.add_argument('-g', '--git', help = 'Arguments to git',
+                   default = '')
+    #
+    # Where's the repo?
+    #
+    p.add_argument('-r', '--repository', help = 'Repository location',
+                   default = '')
+    return p
 
-git = 'git log --decorate '
-if len(sys.argv) > 1:
-    git += ' '.join(sys.argv[1:])
-input = os.popen(git, 'r')
 
-DB = { }
-Tag = 'None'
+p = setupargs()
+args = p.parse_args()
 
-tagline = re.compile(r'^commit ([\da-f]+) .*tag: (v2\.6\.\d\d)')
-commit = re.compile(r'^commit ([\da-f]+)')
+#
+# Pull in an existing database if requested.
+#
+if args.load:
+    DB = pickle.load(open(args.database, 'r'))
+else:
+    DB = { }
+out = open(args.database, 'w')
+
+#
+# Time to fire up git.
+#
+git = 'git log --pretty=format:%H ' + args.git
+if args.repository:
+    os.chdir(args.repository)
+input = os.popen(git, 'r')
 
+nc = 0
 for line in input.readlines():
-    if not line.startswith('commit'):
-        continue  # This makes it go faster
-    m = tagline.search(line)
-    if m:
-        DB[m.group(1)] = Tag = m.group(2)
+    commit = line.strip()
+    #
+    # If we loaded a database and this commit is already there, we
+    # can quit.
+    #
+    if args.load and DB.has_key(commit):
+        break
+    #
+    # Figure out which version this one came from.
+    #
+    desc = os.popen('git describe --contains --match v\\* ' + commit, 'r')
+    tag = desc.readline().strip()
+    desc.close()
+    dash = tag.find('-')
+    if dash > 0:
+        DB[commit] = tag[:dash]
     else:
-        m = commit.search(line)
-        if m:
-            DB[m.group(1)] = Tag
+        DB[commit] = tag
+    #
+    # Give them something to watch.
+    #
+    nc += 1
+    if (nc % 25) == 0:
+        print '%6d %s %s\r' % (nc, commit[:8], tag),
+        sys.stdout.flush()
 
-print 'Found %d commits' % (len(DB.keys()))
-out = open('committags.db', 'w')
+print '\nFound %d/%d commits' % (nc, len(DB.keys()))
 pickle.dump(DB, out)
 out.close()