Merge lp:~garyvdm/bzr-search/authors into lp:bzr-search

[Gary van der Merwe]

[Gary van der Merwe (garyvdm)]

This patch enables bzr-search to index revision committers and authors.

[Robert Collins (lifeless)]

 review: needsfixing

You've left a print statement in the code.

Here, getting rid of the temp variable makes the line long and require
wrapping - its less clean. You can also initialise commit terms to a set
directly rather than updating it with the initial set.

Other than that it looks ok.

- message_utf8 = revision.message.encode('utf8')
- commit_terms = _tokeniser_re.split(message_utf8)
+ commit_terms = set()
+ commit_terms.update(
+
set(_tokeniser_re.split(revision.message.encode('utf8'))))
+
+ for author in
[revision.committer]+revision.get_apparent_authors():
+ name, email = bzrlib.config.parse_username(author)
+ commit_terms.add(email.encode('utf8'))
+ commit_terms.update(
+ set(_tokeniser_re.split(name.encode('utf8'))))
+

review: Needs Fixing

=== modified file 'NEWS'
--- NEWS	2008-12-02 22:36:33 +0000
+++ NEWS	2009-10-28 04:45:20 +0000
@@ -35,6 +35,9 @@
     * Compatibility with split-inventory repositories (requires a bzrlib that
       supports them). (Robert Collins)
 
+    * Will now index revision committers and authors. (Bug 320236, Gary van
+      der Merwe)
+
   BUGFIXES:
 
     * Bug 293906 caused by changes in bzrlib has been fixed. This bug caused

=== modified file 'index.py'
--- index.py	2009-06-03 21:44:06 +0000
+++ index.py	2009-10-28 04:45:20 +0000
@@ -745,13 +745,20 @@
             # components of a revision:
             # parents - not indexed (but we could)
             # commit message (done)
-            # author (todo)
-            # committer (todo)
+            # author (done)
+            # committer (done)
             # properties (todo - names only?)
             # bugfixes (a property we know how to read)
             # other filters?
             message_utf8 = revision.message.encode('utf8')
-            commit_terms = _tokeniser_re.split(message_utf8)
+            commit_terms = set(_tokeniser_re.split(message_utf8))
+            
+            for author in [revision.committer]+revision.get_apparent_authors():
+                name, email = bzrlib.config.parse_username(author)
+                commit_terms.add(email.encode('utf8'))
+                name_utf8 = name.encode('utf8')
+                commit_terms.update(set(_tokeniser_re.split(name_utf8)))
+            
             for term in commit_terms:
                 if not term:
                     continue

=== modified file 'tests/test_index.py'
--- tests/test_index.py	2009-03-10 23:10:42 +0000
+++ tests/test_index.py	2009-10-28 04:45:20 +0000
@@ -237,7 +237,8 @@
             )
         rev_index = index.init_index(tree.branch)
         # The double-space is a cheap smoke test for the tokeniser.
-        revid = tree.commit('first  post')
+        revid = tree.commit('first  post', committer="Joe Soap <[e-mail]>",
+                            authors=["Foo Baa <[e-mail]>"])
         rev_index.index_revisions(tree.branch, [revid])
         self.assertEqual(set([(revid,)]), set(rev_index.indexed_revisions()))
         # reopen - it should retain the indexed revisions.
@@ -260,6 +261,12 @@
             ("working",): set([('f', 'an-id', revid)]),
             ("tree",): set([('f', 'an-id', revid)]),
             ('an-id', revid): set([('p', '', 'README.txt')]),
+            ('Baa',): set([('r', '', revid)]),
+            ('Foo',): set([('r', '', revid)]),
+            ('Joe',): set([('r', '', revid)]),
+            ('Soap',): set([('r', '', revid)]),
+            ('[e-mail]',): set([('r', '', revid)]),
+            ('[e-mail]',): set([('r', '', revid)]),
             }
         all_terms = {}
         for term, posting_list in rev_index.all_terms():