Minor refactoring for clarity and approachability

bin/convert_logs.py

@@ -1,5 +1,5 @@
-#!/usr/bin/python
-# Standard library imports
+#!/usr/bin/env python
+
 from datetime import datetime
 from optparse import OptionParser
 import os
@@ -17,27 +17,64 @@
 CVS_SEP = "----------------------------"
 
 
-class Event(object):
-    """ Event to hold all of the separate events as we parse them from the logs. """
-
-    def __init__(self, filename, date, author):
-        self.filename = filename
-        self.date = date
-        self.author = author
+def parse_args(argv):
+    """ Parses command line arguments and returns an options object
+    along with any extra arguments.
+    """
+    p = OptionParser()
+
+    p.add_option("-s", "--svn-log", dest="svn_log", 
+        metavar="<log file>",
+        help="input svn log to convert to standard event xml")
+
+    p.add_option("-c", "--cvs-log", dest="cvs_log", 
+        metavar="<log file>",
+        help="input cvs log to convert to standard event xml")
+
+    p.add_option("-g", "--git-log", dest="git_log", 
+        metavar="<log file>",
+        help="input git log to convert to standard event xml")
+
+    p.add_option("-v", "--vss-log", dest="vss_log",
+        metavar="<log file>",
+        help="input vss report to convert to standard event xml")
+
+    p.add_option("-t", "--starteam-log", dest="starteam_log",
+        metavar="<log file>",
+        help="input starteam log to convert to standard event xml")
+
+    p.add_option("-w", "--wikiswarm-log", dest="wikiswarm_log", 
+        metavar="<log file>",
+        help="input wikiswarm log to convert to standard event xml")
+
+    p.add_option("-m", "--mercurial-log", dest="mercurial_log", 
+        metavar="<log file>",
+        help="input mercurial log to convert to standard event xml")
+
+    p.add_option("-l", "--gnu-changelog", dest="gnu_log",
+        metavar="<log file>",
+        help="input GNU Changelog to convert to standard event xml")
+
+    p.add_option( "-p", "--perforce-path", dest="perforce_path",
+        metavar="<log file>",
+        help="get data from perforce and save it to standard event xml")
 
-    def properties(self):
-        """returns a dict of properties and their names for XML serialization"""
-        return {
-                "date": str(self.date),
-                "filename": self.filename,
-                "author": self.author
-        }
-
-    # Some version control system's logs are not in chronological order, so
-    # this compare method will return a compare of the date attributes.
-    def __cmp__(self, other):
-        return cmp(self.date, other.date)
+    p.add_option( "-o", "--output-log", dest="output_log", 
+        metavar="<log file>",
+        help="specify event log output file")
 
+    p.add_option("--nosort", dest="nosort", default=False, action="store_true",
+        help="use if the input log is already in chronological order for speed")
+
+    p.add_option("--ignore-author", dest="ignore_author", default=None,
+                action="store",
+                help="Ignore authors that match this regular expression.")
+
+
+    (options, args) = p.parse_args(argv)
+
+    return (options, args)
+
 
 def main(argv):
     """ Calls the parse_args function based on the 
@@ -85,15 +122,15 @@ def main(argv):
     # check for valid cmd line arguments before doing anything
     if opts.ignore_author is not None:
         try:
-            re.compile(opts.ignore_author)
+            opts.ignore_author = re.compile(opts.ignore_author)
         except sre_constants.error:
             print >>stderr, "Unable to compile author reg ex: %s" % \
                   opts.ignore_author
             sys.exit(1)
 
     if not os.path.exists(log_file):
         #hacky, but OptionParse doesn't support options that only sometimes
-        # have an extra value
+        # take an argument
         if log_file == "stdin":
             log_file = sys.stdin
         else:
@@ -102,13 +139,15 @@ def main(argv):
     else:
         log_file = open(log_file, 'r')
 
+    #Parse the log_file into an iterable of Events
     events = parser(log_file, opts)
 
     # Remove all authors we wanted to ignore here
     if opts.ignore_author is not None:
-        events = remove_ignored_author(opts.ignore_author, events)
+        events = ifilter(lambda e: opts.ignore_author.match(e.author) is None,
+                        events)
 
-    #its really best if we don't have to sort, but by default most are
+    #its really best if we don't have to sort, but most data sources are
     # in the reverse order that we want, so we sort by default
     if not opts.nosort:
         events= sorted(list(events))
@@ -136,63 +175,6 @@ def create_event_xml(events, output):
     generator.endElement('file_events')
     generator.endDocument()
 
-def parse_args(argv):
-    """ Parses command line arguments and returns an options object
-    along with any extra arguments.
-    """
-    p = OptionParser()
-
-    p.add_option("-s", "--svn-log", dest="svn_log", 
-        metavar="<log file>",
-        help="input svn log to convert to standard event xml")
-
-    p.add_option("-c", "--cvs-log", dest="cvs_log", 
-        metavar="<log file>",
-        help="input cvs log to convert to standard event xml")
-
-    p.add_option("-g", "--git-log", dest="git_log", 
-        metavar="<log file>",
-        help="input git log to convert to standard event xml")
-
-    p.add_option("-v", "--vss-log", dest="vss_log",
-        metavar="<log file>",
-        help="input vss report to convert to standard event xml")
-
-    p.add_option("-t", "--starteam-log", dest="starteam_log",
-        metavar="<log file>",
-        help="input starteam log to convert to standard event xml")
-
-    p.add_option("-w", "--wikiswarm-log", dest="wikiswarm_log", 
-        metavar="<log file>",
-        help="input wikiswarm log to convert to standard event xml")
-
-    p.add_option("-m", "--mercurial-log", dest="mercurial_log", 
-        metavar="<log file>",
-        help="input mercurial log to convert to standard event xml")
-
-    p.add_option("-l", "--gnu-changelog", dest="gnu_log",
-        metavar="<log file>",
-        help="input GNU Changelog to convert to standard event xml")
-
-    p.add_option( "-p", "--perforce-path", dest="perforce_path",
-        metavar="<log file>",
-        help="get data from perforce and save it to standard event xml")
-
-    p.add_option( "-o", "--output-log", dest="output_log", 
-        metavar="<log file>",
-        help="specify event log output file")
-
-    p.add_option("--nosort", dest="nosort", default=False, action="store_true",
-        help="use if the input log is already in chronological order for speed")
-
-    p.add_option("--ignore-author", dest="ignore_author", default=None,
-                action="store",
-                help="Ignore authors that match this regular expression.")
-
-
-    (options, args) = p.parse_args(argv)
-
-    return (options, args)
 
 
 def parse_svn_log(file_handle, opts):
@@ -428,12 +410,6 @@ def parse_perforce_path(file_handle, opts):
                     yield Event(file_name, int(changelist['time'] + '000'), changelist['user'])
 
 
-def remove_ignored_author(ignore, events):
-    """ Remove the events that match the given ignore reg ex. """
-    events = ifilter(lambda evt: re.match(ignore, evt.author) is None,
-                    events)
-    return events
-
 
 def run_marshal(command):
     import marshal
@@ -449,6 +425,27 @@ def run_marshal(command):
     stream.close()
     return results
 
+class Event(object):
+    """ Event to hold all of the separate events as we parse them from the logs. """
+
+    def __init__(self, filename, date, author):
+        self.filename = filename
+        self.date = date
+        self.author = author
+
+    def properties(self):
+        """returns a dict of properties and their names for XML serialization"""
+        return {
+                "date": str(self.date),
+                "filename": self.filename,
+                "author": self.author
+        }
+
+    # Some version control system's logs are not in chronological order, so
+    # this compare method will return a compare of the date attributes.
+    def __cmp__(self, other):
+        return cmp(self.date, other.date)
+
 
 # Main entry point.
 if __name__ == "__main__":