implement optional conversion of svn author to full name & email

4 files changed, 72 insertions, 10 deletions

diff --git a/MgaRepo/commands/clone.py b/MgaRepo/commands/clone.py
index 589ec75..86793c9 100644
--- a/MgaRepo/commands/clone.py
+++ b/MgaRepo/commands/clone.py
@@ -20,6 +20,7 @@ Options:
     -b      The package branch
     -M      Do not use the mirror (use the main repository)
     -h      Show this message
+    -F      Do not convert svn usernames to full name & email
 
 Examples:
     repsys clone pkgname
@@ -33,6 +34,8 @@ def parse_options():
     parser = OptionParser(help=HELP)
     parser.add_option("--distribution", "-d", dest="distro", default=None)
     parser.add_option("--branch", "-b", dest="branch", default=None)
+    parser.add_option("-F", dest="fullnames", default=True,
+            action="store_false")
     opts, args = parser.parse_args()
     if len(args) not in (1, 2):
         raise Error("invalid arguments")
diff --git a/MgaRepo/git.py b/MgaRepo/git.py
index 45c628d..b91dcb2 100644
--- a/MgaRepo/git.py
+++ b/MgaRepo/git.py
@@ -2,8 +2,10 @@ from MgaRepo import Error, config
 from MgaRepo.util import execcmd
 from MgaRepo.VCS import *
 from MgaRepo.svn import SVN
+from MgaRepo.log import UserTagParser
 from os.path import basename, dirname, abspath, lexists, join
 from os import chdir, getcwd
+from tempfile import mkstemp
 import sys
 import re
 import time
@@ -23,7 +25,23 @@ class GIT(VCS):
         self.vcs_supports['clone'] = True
         self.env_defaults = {"GIT_SSH": self.vcs_wrapper}
 
-    def clone(self, url, targetpath, **kwargs):
+    def configget(self, key="", location="--local"):
+        cmd = ["config", location, "--get-regexp", key]
+        config = None
+        status, output = self._execVcs(*cmd, noerror=True)
+        if not status and output:
+            config = eval("{'" + output.replace("\n", "',\n'").replace(" ", "' : '") + "'}")
+        return config
+
+    def configset(self, config, location="--local"):
+        cmd = ("config", location)
+        for pair in config.items():
+            status, output = self._execVcs(*cmd + pair)
+            if status:
+                return False
+        return True
+
+    def clone(self, url, targetpath, fullnames=True, **kwargs):
         if lexists(join(targetpath, SVN.vcs_dirname)):
             raise Error("Target path %s already contains svn checkout, aborting..." % targetpath)
         if url.split(':')[0].find("svn") < 0:
@@ -44,15 +62,31 @@ class GIT(VCS):
             tags = "releases"
             # cloning svn braches as well should rather be optionalif reenabled..
             #cmd = ["svn", "init", topurl, "--trunk="+trunk, "--tags="+tags", targetpath]
+
+
             cmd = ["svn", "init", url, abspath(targetpath)]
             self._execVcs(*cmd, **kwargs)
             os.environ.update({"GIT_WORK_TREE" : abspath(targetpath), "GIT_DIR" : join(abspath(targetpath),".git")})
+
             for entry in logentries:
                 revisions.append(int(entry.attrib["revision"]))
             revisions.sort()
+
+            fetchcmd = ["svn", "fetch", "--log-window-size=1000"]
+            if fullnames:
+                usermap = UserTagParser()
+                # store configuration in local git config so that'll be reused later when ie. updating
+                gitconfig = {"svn-remote.authorlog.url" : usermap.url,
+                        "svn-remote.authorlog.defaultmail": usermap.defaultmail}
+                self.configset(gitconfig)
+                usermapfile = usermap.get_user_map_file()
+                fetchcmd.extend(("--authors-file", usermapfile))
+
             while revisions:
-                cmd = ["svn", "fetch", "--log-window-size=1000", "-r%d" % revisions.pop(0)]
-                self._execVcs(*cmd, **kwargs)
+                self._execVcs(*fetchcmd + ["-r%d"%revisions.pop(0)], **kwargs)
+            if fullnames:
+                usermap.cleanup()
+
             cmd = ["svn", "rebase", "--log-window-size=1000", "--local", "--fetch-all", "git-svn"]
             return self._execVcs_success(*cmd, **kwargs)
 
@@ -114,15 +148,25 @@ class GIT(VCS):
         retval, result = execcmd(*cmd)
         if retval:
             return retval
+
         xmllog = ElementTree.fromstring(result)
         logentries = xmllog.getiterator("logentry")
         revisions = []
         for entry in logentries:
             revisions.append(int(entry.attrib["revision"]))
         revisions.sort()
+
+        fetchcmd = ["svn", "fetch", "--log-window-size=1000"]
+        gitconfig = self.configget("svn-remote.authorlog")
+        if gitconfig:
+            usermap = UserTagParser(url=gitconfig.get("svn-remote.authorlog.url"),defaultmail=gitconfig.get("svn-remote.authorlog.defaultmail"))
+            usermapfile = usermap.get_user_map_file()
+            fetchcmd.extend(("--authors-file", usermapfile))
+
         while revisions:
-            cmd = ["svn", "fetch", "--log-window-size=1000", "-r%d" % revisions.pop(0)]
-            self._execVcs(*cmd, **kwargs)
+            self._execVcs(*fetchcmd + ["-r%d"%revisions.pop(0)], **kwargs)
+        if gitconfig:
+            usermap.cleanup()
 
         cmd = ["svn", "rebase", "--log-window-size=1000", "--local", "--fetch-all", "git-svn"]
         status, output = self._execVcs(*cmd, **kwargs)
diff --git a/MgaRepo/log.py b/MgaRepo/log.py
index ffa5a4d..3fb3cc5 100644
--- a/MgaRepo/log.py
+++ b/MgaRepo/log.py
@@ -8,6 +8,7 @@ from io import StringIO
 
 import sys
 import os
+import os.path
 import re
 import time
 import locale
@@ -569,11 +570,12 @@ class UserTagParser(HTMLParser):
     userpage = None
     namepat = re.compile("(?P<name>.*?)\s*\((?P<user>.*?)\)")
     usermap = {}
+    usermapfile = None
 
-    def __init__(self, url="https://people.mageia.org/u/", defaultmail="mageia.org", *cmd, **kwargs):
+    def __init__(self, url=None, defaultmail=None, *cmd, **kwargs):
         HTMLParser.__init__(self, *cmd, **kwargs)
-        self.url = url
-        self.defaultmail = defaultmail
+        self.url = url or "http://people.mageia.org/u/"
+        self.defaultmail = defaultmail or "mageia.org"
 
     def handle_starttag(self, tag, attrs):
         if tag == "li":
@@ -607,6 +609,19 @@ class UserTagParser(HTMLParser):
         self.feed(userhtml)
         return self.usermap
 
+    def get_user_map_file(self):
+        if not self.usermap:
+            self.get_user_map()
+        self.usermapfile = tempfile.mkstemp(suffix=".txt", prefix="usermap")
+        f = open(self.usermapfile[0], "w", encoding="UTF-8")
+        f.writelines("%s = %s\n" % user for user in sorted(self.usermap.items()))
+        f.close()
+        return self.usermapfile[1]
+
+    def cleanup(self):
+        if os.path.exists(self.usermapfile[1]):
+            os.unlink(self.usermapfile[1])
+
 def _map_user_names():
     if not usermap:
         parser = UserTagParser()
diff --git a/MgaRepo/rpmutil.py b/MgaRepo/rpmutil.py
index a9dfa7b..50313f2 100644
--- a/MgaRepo/rpmutil.py
+++ b/MgaRepo/rpmutil.py
@@ -584,7 +584,7 @@ def checkout(pkgdirurl, path=None, revision=None, branch=None, distro=None, back
         binrepo.download_binaries(path)
     
 def clone(pkgdirurl, path=None, revision=None, branch=None, distro=None, backports=None,
-        spec=False):
+        spec=False, fullnames = True):
     o_pkgdirurl = pkgdirurl
     pkgdirurl = layout.package_url(o_pkgdirurl, distro=distro, backports=backports)
     append = None
@@ -596,7 +596,7 @@ def clone(pkgdirurl, path=None, revision=None, branch=None, distro=None, backpor
         path = layout.package_name(pkgdirurl)
     mirror.info(current, write=True)
     git = GIT()
-    git.clone(current, path, show=1)
+    git.clone(current, path, fullnames=fullnames, show=1)
     if not spec:
         binrepo.download_binaries(path)