Move gitosis.conf handling script to printing "patches"

2 files changed, 180 insertions, 109 deletions

diff --git a/diff-gitosis-conf-against-repositories-xml.py b/diff-gitosis-conf-against-repositories-xml.py
new file mode 100755
index 0000000..580f2ad
--- /dev/null
+++ b/diff-gitosis-conf-against-repositories-xml.py
@@ -0,0 +1,180 @@
+#!/usr/bin/env python
+# Copyright (C) 2009 Sebastian Pipping <sebastian@pipping.org>
+# Licensed under GPL 2 or later
+
+import sys
+import os
+if len(sys.argv) not in (3, 4):
+    print "USAGE:\n  python  %s  foo/gitosis.conf bar/repositories.xml [baz/extended.xml]" % \
+            os.path.basename(sys.argv[0])
+    sys.exit(1)
+gitosis_conf_location = sys.argv[1]
+repositories_xml_location = sys.argv[2]
+extended_xml_location = (len(sys.argv) == 4) and sys.argv[3] or None
+
+
+import xml.etree.ElementTree as ET
+from ConfigParser import ConfigParser
+import re
+
+from sharedutils import * # local
+
+
+def sort_as_in(elements, tag_order):
+    order_map = dict((v, i) for i, v in enumerate(tag_order))
+    deco = (t for t in enumerate(elements))
+    deco_sorted = sorted(deco, key=lambda (i, v): (order_map[v.tag], i))
+    return list(v for _, v in deco_sorted)
+
+class ChangeLog:
+    def __init__(self, repo_name, is_new):
+        self.empty = True
+        self.repo_name = repo_name
+        self.is_new = is_new
+
+    def log(self, kind, details):
+        if self.empty:
+            if self.is_new:
+                print 'Repo "%s" missing completely' % self.repo_name
+            else:
+                print 'Analyzing repo "%s":' % self.repo_name
+            self.empty = False
+        elif not self.is_new:
+            print '- Missing %s "%s"' % (kind, details)
+
+
+OWNER_REGEX = re.compile('^([^<]+) (?:\([^)]+\) )?<([^ ]+@[^ ]+)>$')
+NOT_AN_OVERLAY_MESSAGE = 'Skipping %s (not an overlay)'
+
+
+gitosis_conf = ConfigParser()
+gitosis_conf.read(gitosis_conf_location)
+
+a = ET.parse(open(repositories_xml_location))
+repositories = a.getroot()
+overlays_gentoo_org_dict = dict([[e.find('name').text, e] for e in repositories])
+
+for section_name in gitosis_conf.sections():
+    if section_name.startswith('repo '):
+        _repo_base = section_name[len('repo '):].strip()
+        if _repo_base.startswith('dev/'):
+            repo_name = _repo_base[len('dev/'):].strip()
+            owner_type = "person"
+        elif _repo_base.startswith('proj/'):
+            repo_name = _repo_base[len('proj/'):].strip()
+            owner_type = "project"
+        else:
+            # TODO print NOT_AN_OVERLAY_MESSAGE % repo_name
+            continue
+
+        _description = gitosis_conf.get(section_name, 'description')
+        if not gitosis_conf.has_option(section_name, 'gentoo-is-overlay') or \
+                not gitosis_conf.getboolean(section_name, 'gentoo-is-overlay'):
+            if not _description.lower().endswith('overlay'):
+                # TODO print NOT_AN_OVERLAY_MESSAGE % repo_name
+                continue
+
+        is_new = repo_name not in overlays_gentoo_org_dict
+        if is_new:
+            repo = ET.Element('repo')
+            repositories.append(repo)
+            name = ET.Element('name')
+            name.text = repo_name
+            repo.append(name)
+        else:
+            repo = overlays_gentoo_org_dict[repo_name]
+        log = ChangeLog(repo_name, is_new)
+
+        if 'status' not in repo.attrib:
+            repo.attrib['status'] = 'official'
+            log.log('attribute', 'status')
+
+        if 'quality' not in repo.attrib:
+            repo.attrib['quality'] = 'experiental'
+            log.log('attribute', 'quality')
+
+        homepage = repo.find('homepage')
+        if homepage == None:
+            homepage = ET.Element('homepage')
+            homepage.text = 'http://git.overlays.gentoo.org/gitweb/?p=%s.git;a=summary' % _repo_base
+            repo.append(homepage)
+            log.log('homepage', homepage.text)
+
+        description = repo.find('description')
+        if description == None:
+            description = ET.Element('description', lang='en')
+            description.text = _description
+            repo.append(description)
+            log.log('description', _description)
+
+
+        _owner = gitosis_conf.get(section_name, 'owner')
+        _owner_match = OWNER_REGEX.match(_owner)
+
+        owner = repo.find('owner')
+        if owner == None:
+            owner = ET.Element('owner', type=owner_type)
+            repo.append(owner)
+            log.log('owner', 'TODO')
+
+        owner_name = owner.find('name')
+        if owner_name == None:
+            owner_name = ET.Element('name')
+            owner_name.text = _owner_match.group(1)
+            log.log('owner name', owner_name.text)
+
+        owner_email = owner.find('email')
+        if owner_email == None:
+            owner_email = ET.Element('email')
+            owner_email.text = _owner_match.group(2)
+            log.log('owner email', owner_email.text)
+
+        owner[:] = [owner_email, owner_name]
+
+
+        _sources = set((source.attrib['type'], source.text) for source in repo.findall('source'))
+        source_uris = (
+            'git://git.overlays.gentoo.org/%s.git' % _repo_base,
+            'http://git.overlays.gentoo.org/gitroot/%s.git' % _repo_base,
+            'git+ssh://git@git.overlays.gentoo.org/%s.git' % _repo_base,
+        )
+        for uri in source_uris:
+            if ('git', uri) not in _sources and \
+                    ('git', uri[:-len('.git')]) not in _sources:
+                source = ET.Element('source', type='git')
+                source.text = uri
+                repo.append(source)
+                log.log('git source', uri)
+
+
+        _feeds = set(feed.text for feed in repo.findall('feed'))
+        feed_uris = (
+            'http://git.overlays.gentoo.org/gitweb/?p=%s.git;a=atom' % _repo_base,
+            'http://git.overlays.gentoo.org/gitweb/?p=%s.git;a=rss' % _repo_base,
+        )
+        for uri in feed_uris:
+            if uri not in _feeds:
+                feed = ET.Element('feed')
+                feed.text = uri
+                repo.append(feed)
+                log.log('feed', uri)
+
+        repo[:] = sort_as_in(repo[:], (
+                'name', 'description', 'longdescription',
+                'homepage', 'owner', 'source', 'feed'))
+
+        if is_new or not log.empty:
+            if extended_xml_location == None:
+                TERM_WIDTH = 67
+                print '-'*TERM_WIDTH
+                sys.stdout.write('  ')
+                indent(repo, 1)
+                repo.tail = '\n'
+                ET.ElementTree(repo).write(sys.stdout)
+                print '-'*TERM_WIDTH
+            print
+
+
+if extended_xml_location != None:
+    indent(repositories)
+    a.write(extended_xml_location)
diff --git a/merge-gitosis-conf-into-repositories-xml.py b/merge-gitosis-conf-into-repositories-xml.py
deleted file mode 100755
index fd26102..0000000
--- a/merge-gitosis-conf-into-repositories-xml.py
+++ /dev/null
@@ -1,109 +0,0 @@
-#!/usr/bin/env python
-# Copyright (C) 2009 Sebastian Pipping <sebastian@pipping.org>
-# Licensed under GPL 2 or later
-
-import sys
-import os
-if len(sys.argv) != 1 + 2:
-    print "USAGE:\n  python  %s  foo/gitosis.conf bar/repositories.xml" % \
-            os.path.basename(sys.argv[0])
-    sys.exit(1)
-gitosis_conf_location = sys.argv[1]
-repositories_xml_location = sys.argv[2]
-
-
-import xml.etree.ElementTree as ET
-from ConfigParser import ConfigParser
-import re
-
-from sharedutils import * # local
-
-OWNER_REGEX = re.compile('^([^<]+) (?:\([^)]+\) )?<([^ ]+@[^ ]+)>$')
-NOT_AN_OVERLAY_MESSAGE = 'Skipping %s (not an overlay)'
-
-
-gitosis_conf = ConfigParser()
-gitosis_conf.read(gitosis_conf_location)
-
-a = ET.parse(open(repositories_xml_location))
-repositories = a.getroot()
-overlays_gentoo_org_dict = dict([[e.find('name').text, e] for e in repositories])
-
-for section_name in gitosis_conf.sections():
-    if section_name.startswith('repo '):
-        _repo_base = section_name[len('repo '):].strip()
-        if _repo_base.startswith('dev/'):
-            repo_name = _repo_base[len('dev/'):].strip()
-            owner_type = "person"
-        elif _repo_base.startswith('proj/'):
-            repo_name = _repo_base[len('proj/'):].strip()
-            owner_type = "project"
-        else:
-            print NOT_AN_OVERLAY_MESSAGE % repo_name
-            continue
-
-        if not gitosis_conf.has_option(section_name, 'gentoo-is-overlay') or \
-                not gitosis_conf.getboolean(section_name, 'gentoo-is-overlay'):
-            print NOT_AN_OVERLAY_MESSAGE % repo_name
-            continue
-
-        if repo_name in overlays_gentoo_org_dict:
-            print 'Updating %s' % repo_name
-            repo = overlays_gentoo_org_dict[repo_name]
-            # Keep due to poor descriptions in gitosis.conf atm
-            # Assume first <description> is english
-            _description = repo.find('description').text
-
-            # Clear
-            repo.attrib['status'] = 'official'
-            repo[:] = []
-        else:
-            print 'Adding   %s' % repo_name
-            _description = gitosis_conf.get(section_name, 'description')
-            repo = ET.Element('repo', status='official')
-            repositories.append(repo)
-
-        name = ET.Element('name')
-        name.text = repo_name
-
-        homepage = ET.Element('homepage')
-        homepage.text = 'http://git.overlays.gentoo.org/gitweb/?p=%s.git;a=summary' % _repo_base
-
-        description = ET.Element('description', lang='en')
-        description.text = _description
-
-        _owner = gitosis_conf.get(section_name, 'owner')
-        _owner_match = OWNER_REGEX.match(_owner)
-
-        owner = ET.Element('owner', type=owner_type)
-        owner_name = ET.Element('name')
-        owner_name.text = _owner_match.group(1)
-        owner_email = ET.Element('email')
-        owner_email.text = _owner_match.group(2)
-        owner[:] = [owner_email, owner_name]
-
-        repo[:] = [name, description, homepage, owner]
-
-        source_uris = (
-            'git://git.overlays.gentoo.org/%s.git' % _repo_base,
-            'http://git.overlays.gentoo.org/gitroot/%s.git' % _repo_base,
-            'git+ssh://git@git.overlays.gentoo.org/%s.git' % _repo_base,
-        )
-        for uri in source_uris:
-            source = ET.Element('source', type='git')
-            source.text = uri
-            repo.append(source)
-
-        feed_uris = (
-            'http://git.overlays.gentoo.org/gitweb/?p=%s.git;a=atom' % _repo_base,
-            'http://git.overlays.gentoo.org/gitweb/?p=%s.git;a=rss' % _repo_base,
-        )
-        for uri in feed_uris:
-            feed = ET.Element('feed')
-            feed.text = uri
-            repo.append(feed)
-
-        # recurse_print(repo, 1)
-
-indent(repositories)
-a.write(repositories_xml_location)