@@ -147,24 +147,25 @@ def invoke_code_swarm(code_swarm_jar, proj_cfg, options): def code_swarm_dir(repo): """Returns a directory within the current working directory that we can use for storage within the given repository directory.""" - if repo[0] == '.': - repo = repo[1:] - return os.path.join(os.getcwd(), "." + repo, ".code_swarm") + return os.path.join(os.getcwd(), repo, ".code_swarm") def autogenerate_files(options): """docstring for autogenerate_files""" if os.path.exists(".git"): - dir = code_swarm_dir("git") + dir = code_swarm_dir(".git") generate_log = do_git elif os.path.exists(".svn"): - dir = code_swarm_dir("svn") + dir = code_swarm_dir(".svn") generate_log = do_svn elif os.path.exists(".hg"): - dir = code_swarm_dir("hg") + dir = code_swarm_dir(".hg") generate_log = do_hg + elif os.path.exists("_darcs"): + dir = code_swarm_dir("_darcs") + generate_log = do_darcs else: - msg = "This directory isn't an svn, git, or hg project. " + msg = "This directory isn't an svn, git, hg, or darcs project. " msg += "Run in the base directory of a source-controlled project." print >>sys.stderr, msg sys.exit(2) @@ -198,7 +199,7 @@ def do_git(dir): return do_cmds('git log --name-status --pretty=format:"%s" > "%s"'%(fmt, tmp) ,"convert_logs.py -g '%s' -o '%s'" % (tmp, xml) - ,"rm %s" % tmp) + ,"rm -f '%s'" % tmp) def do_svn(dir): @@ -207,7 +208,7 @@ def do_svn(dir): return do_cmds("svn log -v > '%s'" % tmp ,"convert_logs.py -s '%s' -o '%s'" % (tmp, xml) - ,"rm '%s'" % tmp) + ,"rm -f '%s'" % tmp) def do_hg(dir): @@ -218,6 +219,14 @@ def do_hg(dir): ,"sort_code_swarm_input.py < '%s' > '%s'" % (tmp, xml) ,"rm '%s'" % tmp) +def do_darcs(dir): + tmp = os.path.join(dir, "darcs.xml") + xml = os.path.join(dir, "log.xml") + + return do_cmds("darcs changes -s --xml-output > '%s'" % tmp + ,"convert_logs.py -d '%s' -o '%s'" % (tmp, xml) + ,"rm -f '%s'" % tmp) + def do_freebase(domain): dir = os.path.abspath(tempfile.gettempdir()) xml = os.path.join(dir, md5.new(str(random())).hexdigest() + ".xml") bin/code_swarm @@ -6,6 +6,9 @@ import os import sys import time from xml.sax.saxutils import escape as h +from xml.sax import make_parser +from xml.sax.handler import ContentHandler +from xml.sax._exceptions import SAXParseException from sys import stderr import re import sre_constants @@ -51,6 +54,10 @@ def parse_args(argv): metavar="<log file>", help="input mercurial log to convert to standard event xml") + p.add_option("-d", "--darcs-log", dest="darcs_xml", + metavar="<log file>", + help="output of 'darcs changes -s --xml-output'") + p.add_option("-l", "--gnu-changelog", dest="gnu_log", metavar="<log file>", help="input GNU Changelog to convert to standard event xml") @@ -108,6 +115,9 @@ def main(argv): elif opts.mercurial_log: log_file = opts.mercurial_log parser = parse_mercurial_log + elif opts.darcs_xml: + log_file = opts.darcs_xml + parser = parse_darcs_xml elif opts.gnu_log: log_file = opts.gnu_log parser = parse_gnu_changelog @@ -399,6 +409,57 @@ def parse_gnu_changelog(file_handle, opts): line = file_handle.next() continue +def parse_darcs_xml(file_handle, opts): + class DarcsParser(ContentHandler): + def __init__(self): + self.events = [] + self.author = None + self.date = None + self.getFilename = False + self.filename = "" + self.capture_events = ["add_file", "modify_file", "remove_file"] + + def startElement(self, name, attrs): + if name in self.capture_events: + self.getFilename = True + self.filename = "" + return + if name == "patch": + self.author = attrs["author"].strip() + date = re.sub("[A-Za-z ]+ (\d\d\d\d)$", " \\1", attrs["date"]) + date = re.sub("\s+"," ",date).strip() # normalize whitespace + try: + date = time.strptime(date,"%Y%m%d%H%M%S") + except ValueError, e: + date = time.strptime(date,"%a %b %d %H:%M:%S %Y") + + self.date = int(time.mktime(date))*1000 + elif name == "move": + self.newEvent(attrs["from"]) + self.newEvent(attrs["to"]) + elif name not in ["summary","name","changelog","added_lines" + ,"removed_lines", "add_directory", "remove_directory" + ,"comment", "replaced_tokens"]: + print >> stderr, "warning: unknown tag '%s'" % name + + def characters(self,str): + if self.getFilename: + self.filename += str + + def endElement(self, name): + if name in self.capture_events: + self.newEvent(self.filename) + self.getFilename = False + + def newEvent(self, name): + self.events.append(Event(name.strip(),self.date,self.author)) + + parser = make_parser() + dp = DarcsParser() + parser.setContentHandler(dp) + parser.parse(file_handle) + return dp.events + def parse_perforce_path(file_handle, opts): changelists = run_marshal('p4 -G changelists "' + opts.perforce_path + '"') file_key_re = re.compile("^depotFile") bin/convert_logs.py e491173bdc417b84c39b2c37d42d6c4680bb97b9 Peter Burns peter@metaweb.com http://github.com/rictic/code_swarm/commit/fc54cc6d2de5480b71543c40c7b62bad7f16f01f fc54cc6d2de5480b71543c40c7b62bad7f16f01f 2009-03-29T17:19:44-07:00 2009-03-29T17:19:44-07:00 Added full support for darcs, including autodiscovery Unfortunately darcs sometimes will produce invalid XML when asked. This is a known bug, and related to character encoding issues that they seem unwilling to fix. I might take the performance hit and rewrite this using BeautifulSoup at some point. I've only seen the bug in the darcs repo itself, it comes up when there are authors with non-ascii names. 751b35132395a4e6123f2574aed01ed6ca364c96 Peter Burns peter@metaweb.com