Add version tracking support and an "unknown hackers" report

Version tracking was used to see who had contributed to the most kernel
releases; not sure it's a long-term-useful feature.  The unknown hackers
report helps when trying to improve the database.

Signed-off-by: Jonathan Corbet <corbet@lwn.net>

3 files changed, 82 insertions, 7 deletions

diff --git a/gitdm b/gitdm
index 1367ebd..25ffafb 100755
--- a/gitdm
+++ b/gitdm
@@ -5,8 +5,8 @@
 #
 # This code is part of the LWN git data miner.
 #
-# Copyright 2007-11 Eklektix, Inc.
-# Copyright 2007-11 Jonathan Corbet <corbet@lwn.net>
+# Copyright 2007-12 Eklektix, Inc.
+# Copyright 2007-12 Jonathan Corbet <corbet@lwn.net>
 # Copyright 2011 Germán Póo-Caamaño <gpoo@gnome.org>
 #
 # This file may be distributed under the terms of the GNU General
@@ -43,6 +43,7 @@ DirName = ''
 Aggregate = 'month'
 Numstat = 0
 ReportByFileType = 0
+ReportUnknowns = False
 
 #
 # Options:
@@ -60,6 +61,7 @@ ReportByFileType = 0
 # -r pattern	Restrict to files matching pattern
 # -s		Ignore author SOB lines
 # -u		Map unknown employers to '(Unknown)'
+# -U 		Dump unknown hackers in report
 # -x file.csv   Export raw statistics as CSV
 # -w        Aggregrate the raw statistics by weeks instead of months
 # -y            Aggregrate the raw statistics by years instead of months
@@ -69,9 +71,9 @@ def ParseOpts ():
     global MapUnknown, DevReports
     global DateStats, AuthorSOBs, FileFilter, AkpmOverLt, DumpDB
     global CFName, CSVFile, CSVPrefix,DirName, Aggregate, Numstat
-    global ReportByFileType
+    global ReportByFileType, ReportUnknowns
 
-    opts, rest = getopt.getopt (sys.argv[1:], 'ab:dc:Dh:l:no:p:r:stuwx:yz')
+    opts, rest = getopt.getopt (sys.argv[1:], 'ab:dc:Dh:l:no:p:r:stUuwx:yz')
     for opt in opts:
         if opt[0] == '-a':
             AkpmOverLt = 1
@@ -102,6 +104,8 @@ def ParseOpts ():
             ReportByFileType = 1
         elif opt[0] == '-u':
             MapUnknown = 1
+        elif opt[0] == '-U':
+            ReportUnknowns = True
         elif opt[0] == '-x':
             CSVFile = open (opt[1], 'w')
             print "open output file " + opt[1] + "\n"
@@ -492,6 +496,8 @@ if CSVFile:
 
 if DevReports:
     reports.DevReports (hlist, TotalChanged, CSCount, TotalRemoved)
+if ReportUnknowns:
+    reports.ReportUnknowns(hlist, CSCount)
 reports.EmplReports (elist, TotalChanged, CSCount)
 
 if ReportByFileType and Numstat:
diff --git a/patterns.py b/patterns.py
index 803e532..4d4a347 100644
--- a/patterns.py
+++ b/patterns.py
@@ -21,7 +21,8 @@ import re
 _pemail = r'\s+"?([^<"]+)"?\s<([^>]+)>' # just email addr + name
 
 patterns = {
-    'commit': re.compile (r'^commit ([0-9a-f ]+)$'),
+    'tagcommit': re.compile (r'^commit ([\da-f]+) .*tag: (v[23]\.\d(\.\d\d?)?)'),
+    'commit': re.compile (r'^commit ([0-9a-f ]+)'),
     'author': re.compile (r'^Author:' + _pemail + '$'),
     'signed-off-by': re.compile (r'^\s+Signed-off-by:' + _pemail + '.*$'),
     'merge': re.compile (r'^Merge:.*$'),
diff --git a/reports.py b/reports.py
index 9b8cce9..bc1e18c 100644
--- a/reports.py
+++ b/reports.py
@@ -3,8 +3,8 @@
 #
 # This code is part of the LWN git data miner.
 #
-# Copyright 2007-11 Eklektix, Inc.
-# Copyright 2007-11 Jonathan Corbet <corbet@lwn.net>
+# Copyright 2007-12 Eklektix, Inc.
+# Copyright 2007-12 Jonathan Corbet <corbet@lwn.net>
 #
 # This file may be distributed under the terms of the GNU General
 # Public License, version 2.
@@ -58,6 +58,10 @@ TRow = '''    <tr class="%s">
 <td>%s</td><td align="right">%d</td><td align="right">%.1f%%</td></tr>
 '''
 
+TRowStr = '''    <tr class="%s">
+<td>%s</td><td align="right">%d</td><td>%s</td></tr>
+'''
+
 def ReportLine (text, count, pct):
     global HTMLclass
     if count == 0:
@@ -67,6 +71,15 @@ def ReportLine (text, count, pct):
         HTMLfile.write (TRow % (HClasses[HTMLclass], text, count, pct))
         HTMLclass ^= 1
 
+def ReportLineStr (text, count, extra):
+    global HTMLclass
+    if count == 0:
+        return
+    Outfile.write ('%-25s %4d %s\n' % (text, count, extra))
+    if HTMLfile:
+        HTMLfile.write (TRowStr % (HClasses[HTMLclass], text, count, extra))
+        HTMLclass ^= 1
+
 def EndReport ():
     if HTMLfile:
         HTMLfile.write ('</table>\n\n')
@@ -284,6 +297,34 @@ def ReportByRepCreds (hlist):
             break
     EndReport ()
 
+#
+# Versions.
+#
+def CompareVersionCounts (h1, h2):
+    if h1.versions and h2.versions:
+        return len (h2.versions) - len (h1.versions)
+    if h2.versions:
+        return 1
+    if h1.versions:
+        return -1
+    return 0
+
+def MissedVersions (hv, allv):
+    missed = [v for v in allv if v not in hv]
+    missed.reverse ()
+    return ' '.join (missed)
+
+def ReportVersions (hlist):
+    hlist.sort (CompareVersionCounts)
+    BeginReport ('Developers represented in the most kernel versions')
+    count = 0
+    allversions = hlist[0].versions
+    for h in hlist:
+        ReportLineStr (h.name, len (h.versions), MissedVersions (h.versions, allversions))
+        count += 1
+        if count >= ListCount:
+            break
+    EndReport ()
 
 
 def CompareESOBs (e1, e2):
@@ -341,6 +382,33 @@ def EmplReports (elist, totalchanged, cscount):
     ReportByESOBs (elist)
     ReportByEHackers (elist)
 
+#
+# Who are the unknown hackers?
+#
+def IsUnknown(h):
+    empl = h.employer[0][0][1].name
+    return h.email[0] == empl or empl == '(Unknown)'
+
+def ReportUnknowns(hlist, cscount):
+    #
+    # Trim the list to just the unknowns; try to work properly whether
+    # mapping to (Unknown) is happening or not.
+    #
+    ulist = [ h for h in hlist if IsUnknown(h) ]
+    ulist.sort(ComparePCount)
+    count = 0
+    BeginReport('Developers with unknown affiliation')
+    for h in ulist:
+        pcount = len(h.patches)
+        if pcount > 0:
+            ReportLine(h.name, pcount, (pcount*100.0)/cscount)
+            count += 1
+        if count >= ListCount:
+            break
+    EndReport()
+
+
+
 def ReportByFileType (hacker_list):
     total = {}
     total_by_hacker = {}