Tested-by / Reported-by credits and more

Add tracking of tested-by, reported-by, and reviewed-by.  For the first
two, we also track who is *giving* those credits.

While I was in the neighborhood I also:

 - Started turning the "patch" class into something more than a bare
   container; this work has just begin.

 - Moved the report-writing code into its own file (reports.py)

4 files changed, 410 insertions, 194 deletions

diff --git a/database.py b/database.py
index 95334f7..dbada8a 100644
--- a/database.py
+++ b/database.py
@@ -21,6 +21,10 @@ class Hacker:
         self.added = self.removed = 0
         self.patches = [ ]
         self.signoffs = [ ]
+        self.reviews = [ ]
+        self.tested = [ ]
+        self.reports = [ ]
+        self.testcred = self.repcred = 0
 
     def addemail (self, email, elist):
         self.email.append (email)
@@ -41,8 +45,22 @@ class Hacker:
         self.removed += patch.removed
         self.patches.append (patch)
 
+    #
+    # There's got to be a better way.
+    #
     def addsob (self, patch):
         self.signoffs.append (patch)
+    def addreview (self, patch):
+        self.reviews.append (patch)
+    def addtested (self, patch):
+        self.tested.append (patch)
+    def addreport (self, patch):
+        self.reports.append (patch)
+
+    def reportcredit (self, patch):
+        self.repcred += 1
+    def testcredit (self, patch):
+        self.testcred += 1
 
 HackersByName = { }
 HackersByEmail = { }
diff --git a/gitdm b/gitdm
index 1c3adc7..178adf8 100755
--- a/gitdm
+++ b/gitdm
@@ -11,20 +11,15 @@
 # Public License, version 2.
 
 
-import database, csv, ConfigFile
+import database, csv, ConfigFile, reports
 import getopt, datetime
 import os, re, sys, rfc822, string
 from patterns import *
 
-class patch:
-    pass
-
 Today = datetime.date.today()
 #
 # Control options.
 #
-Outfile = sys.stdout
-ListCount = 999999
 MapUnknown = 0
 DevReports = 1
 DateStats = 0
@@ -51,7 +46,7 @@ CFName = 'gitdm.config'
 # -z		Dump out the hacker database at completion
 
 def ParseOpts ():
-    global Outfile, ListCount, MapUnknown, HTMLfile, DevReports
+    global MapUnknown, DevReports
     global DateStats, AuthorSOBs, FileFilter, AkpmOverLt, DumpDB
     global CFName, CSVFile
 
@@ -66,11 +61,11 @@ def ParseOpts ():
         elif opt[0] == '-D':
             DateStats = 1
         elif opt[0] == '-h':
-            HTMLfile = open (opt[1], 'w')
+            reports.SetHTMLOutput (open (opt[1], 'w'))
         elif opt[0] == '-l':
-            ListCount = int (opt[1])
+            reports.SetMaxList (int (opt[1]))
         elif opt[0] == '-o':
-            Outfile = open (opt[1], 'w')
+            reports.SetOutput (open (opt[1], 'w'))
         elif opt[0] == '-r':
             print 'Filter on "%s"' % (opt[1])
             FileFilter = re.compile (opt[1])
@@ -123,6 +118,29 @@ def PrintDateStats():
         datef.write ('%d/%02d/%02d %6d %7d\n' % (date.year, date.month, date.day,
                                     DateMap[date], total))
 
+
+#
+# Let's slowly try to move some smarts into this class.
+#
+class patch:
+    def __init__ (self, commit):
+        self.commit = commit
+        self.merge = self.added = self.removed = 0
+        self.author = LookupStoreHacker('Unknown hacker', 'unknown@hacker.net')
+        self.email = 'unknown@hacker.net'
+        self.sobs = [ ]
+        self.reviews = [ ]
+        self.testers = [ ]
+        self.reports = [ ]
+
+    def addreviewer (self, reviewer):
+        self.reviews.append (reviewer)
+
+    def addtester (self, tester):
+        self.testers.append (tester)
+
+    def addreporter (self, reporter):
+        self.reports.append (reporter)
 #
 # The core hack for grabbing the information about a changeset.
 #
@@ -137,12 +155,7 @@ def grabpatch():
         if not NextLine:
             return
 
-    p = patch()
-    p.commit = m.group (1)
-    p.merge = p.added = p.removed = 0
-    p.author = LookupStoreHacker('Unknown hacker', 'unknown@hacker.net')
-    p.email = 'unknown@hacker.net'
-    p.sobs = [ ]
+    p = patch(m.group (1))
     NextLine = sys.stdin.readline ()
     ignore = (FileFilter is not None)
     while NextLine:
@@ -173,6 +186,35 @@ def grabpatch():
                 p.sobs.append ((email, LookupStoreHacker(m.group (1), m.group (2))))
             continue
         #
+        # Various other tags of interest.
+        #
+        m = Preview.search (Line)  # Reviewed-by:
+        if m:
+            email = database.RemapEmail (m.group (2))
+            p.addreviewer (LookupStoreHacker(m.group (1), email))
+            continue
+        m = Ptest.search (Line)    # Tested-by:
+        if m:
+            email = database.RemapEmail (m.group (2))
+            p.addtester (LookupStoreHacker (m.group (1), email))
+            p.author.testcredit (patch)
+            continue
+        m = Prep.search (Line)     # Reported-by:
+        if m:
+            email = database.RemapEmail (m.group (2))
+            p.addreporter (LookupStoreHacker (m.group (1), email))
+            p.author.reportcredit (patch)
+            continue
+        m = Preptest.search (Line)  # Reported-and-tested-by:
+        if m:
+            email = database.RemapEmail (m.group (2))
+            h = LookupStoreHacker (m.group (1), email)
+            p.addreporter (h)
+            p.addtester (h)
+            p.author.reportcredit (patch)
+            p.author.testcredit (patch)
+            continue
+        #
         # If this one is a merge, make note of the fact.
         #
         m = Pmerge.match (Line)
@@ -253,170 +295,6 @@ def TrimLTSOBs (p):
     if Linus in p.sobs and Akpm in p.sobs:
         p.sobs.remove (Linus)
 
-#
-# HTML output support stuff.
-#
-HTMLfile = None
-HTMLclass = 0
-HClasses = ['Even', 'Odd']
-
-THead = '''<p>
-<table cellspacing=3>
-<tr><th colspan=3>%s</th></tr>
-'''
-
-def BeginReport (title):
-    global HTMLclass
-    
-    Outfile.write ('\n%s\n' % title)
-    if HTMLfile:
-        HTMLfile.write (THead % title)
-        HTMLclass = 0
-
-TRow = '''    <tr class="%s">
-<td>%s</td><td align="right">%d</td><td align="right">%.1f%%</td></tr>
-'''
-
-def ReportLine (text, count, pct):
-    global HTMLclass
-    if count == 0:
-        return
-    Outfile.write ('%-25s %4d (%.1f%%)\n' % (text, count, pct))
-    if HTMLfile:
-        HTMLfile.write (TRow % (HClasses[HTMLclass], text, count, pct))
-        HTMLclass ^= 1
-
-def EndReport ():
-    if HTMLfile:
-        HTMLfile.write ('</table>\n\n')
-        
-#
-# Comparison and report generation functions.
-#
-def ComparePCount (h1, h2):
-    return len (h2.patches) - len (h1.patches)
-
-def ReportByPCount (hlist):
-    hlist.sort (ComparePCount)
-    count = 0
-    BeginReport ('Developers with the most changesets')
-    for h in hlist:
-        pcount = len (h.patches)
-        changed = max(h.added, h.removed)
-        delta = h.added - h.removed
-        if pcount > 0:
-            ReportLine (h.name, pcount, (pcount*100.0)/CSCount)
-        count += 1
-        if count >= ListCount:
-            break
-    EndReport ()
-            
-def CompareLChanged (h1, h2):
-    return max(h2.added, h2.removed) - max(h1.added, h1.removed)
-
-def ReportByLChanged (hlist):
-    hlist.sort (CompareLChanged)
-    count = 0
-    BeginReport ('Developers with the most changed lines')
-    for h in hlist:
-        pcount = len (h.patches)
-        changed = max(h.added, h.removed)
-        delta = h.added - h.removed
-        if (h.added + h.removed) > 0:
-            ReportLine (h.name, changed, (changed*100.0)/TotalChanged)
-        count += 1
-        if count >= ListCount:
-            break
-    EndReport ()
-            
-def CompareLRemoved (h1, h2):
-    return (h2.removed - h2.added) - (h1.removed - h1.added)
-
-def ReportByLRemoved (hlist):
-    hlist.sort (CompareLRemoved)
-    count = 0
-    BeginReport ('Developers with the most lines removed')
-    for h in hlist:
-        pcount = len (h.patches)
-        changed = max(h.added, h.removed)
-        delta = h.added - h.removed
-        if delta < 0:
-            ReportLine (h.name, -delta, (-delta*100.0)/TotalRemoved)
-        count += 1
-        if count >= ListCount:
-            break
-    EndReport ()
-
-def CompareEPCount (e1, e2):
-    return e2.count - e1.count
-
-def ReportByPCEmpl (elist):
-    elist.sort (CompareEPCount)
-    count = 0
-    BeginReport ('Top changeset contributors by employer')
-    for e in elist:
-        if e.count != 0:
-            ReportLine (e.name, e.count, (e.count*100.0)/CSCount)
-        count += 1
-        if count >= ListCount:
-            break
-    EndReport ()
-
-
-
-def CompareELChanged (e1, e2):
-    return e2.changed - e1.changed
-
-def ReportByELChanged (elist):
-    elist.sort (CompareELChanged)
-    count = 0
-    BeginReport ('Top lines changed by employer')
-    for e in elist:
-        if e.changed != 0:
-            ReportLine (e.name, e.changed, (e.changed*100.0)/TotalChanged)
-        count += 1
-        if count >= ListCount:
-            break
-    EndReport ()
-
-
-
-def CompareSOBs (h1, h2):
-    return len (h2.signoffs) - len (h1.signoffs)
-
-def ReportBySOBs (hlist):
-    hlist.sort (CompareSOBs)
-    totalsobs = 0
-    for h in hlist:
-        totalsobs += len (h.signoffs)
-    count = 0
-    BeginReport ('Developers with the most signoffs (total %d)' % totalsobs)
-    for h in hlist:
-        scount = len (h.signoffs)
-        if scount > 0:
-            ReportLine (h.name, scount, (scount*100.0)/totalsobs)
-        count += 1
-        if count >= ListCount:
-            break
-    EndReport ()
-
-def CompareESOBs (e1, e2):
-    return e2.sobs - e1.sobs
-
-def ReportByESOBs (elist):
-    elist.sort (CompareESOBs)
-    totalsobs = 0
-    for e in elist:
-        totalsobs += e.sobs
-    count = 0
-    BeginReport ('Employers with the most signoffs (total %d)' % totalsobs)
-    for e in elist:
-        if e.sobs > 0:
-            ReportLine (e.name, e.sobs, (e.sobs*100.0)/totalsobs)
-        count += 1
-        if count >= ListCount:
-            break
-    EndReport ()
 
 #
 # Here starts the real program.
@@ -453,15 +331,21 @@ while (1):
     p = grabpatch()
     if not p:
         break
-    if p.added > 100000 or p.removed > 100000:
-        print 'Skipping massive add'
-        continue
+#    if p.added > 100000 or p.removed > 100000:
+#        print 'Skipping massive add', p.commit
+#        continue
     if FileFilter and p.added == 0 and p.removed == 0:
         continue
     if not p.merge:
         p.author.addpatch (p)
         for sobemail, sob in p.sobs:
             sob.addsob (p)
+        for hacker in p.reviews:
+            hacker.addreview (p)
+        for hacker in p.testers:
+            hacker.addtested (p)
+        for hacker in p.reports:
+            hacker.addreport (p)
     CSCount += 1
     csv.AccumulatePatch (p)
 print >> sys.stderr, 'Grabbing changesets...done'
@@ -473,10 +357,10 @@ if DumpDB:
 #
 hlist = database.AllHackers ()
 elist = database.AllEmployers ()
-Outfile.write ('Processed %d csets from %d developers\n' % (CSCount,
+reports.Write ('Processed %d csets from %d developers\n' % (CSCount,
                                                             len (hlist)))
-Outfile.write ('%d employers found\n' % len (elist))
-Outfile.write ('A total of %d lines added, %d removed (delta %d)\n' %
+reports.Write ('%d employers found\n' % len (elist))
+reports.Write ('A total of %d lines added, %d removed (delta %d)\n' %
                (TotalAdded, TotalRemoved, TotalAdded - TotalRemoved))
 if TotalChanged == 0:
     TotalChanged = 1 # HACK to avoid div by zero
@@ -489,10 +373,5 @@ if CSVFile is not None:
         CSVFile.close ()
 
 if DevReports:
-    ReportByPCount (hlist)
-    ReportByLChanged (hlist)
-    ReportByLRemoved (hlist)
-    ReportBySOBs (hlist)
-ReportByPCEmpl (elist)
-ReportByELChanged (elist)
-ReportByESOBs (elist)
+    reports.DevReports (hlist, TotalChanged, CSCount, TotalRemoved)
+reports.EmplReports (elist, TotalChanged, CSCount)
diff --git a/patterns.py b/patterns.py
index c833c3e..9a92cf0 100644
--- a/patterns.py
+++ b/patterns.py
@@ -25,7 +25,10 @@ Prem = re.compile (r'^-[^-].*$')
 Pdate = re.compile (r'^(Commit)?Date:\s+(.*)$')
 Pfilea = re.compile (r'^---\s+(.*)$')
 Pfileb = re.compile (r'^\+\+\+\s+(.*)$')
-
+Preview = re.compile (r'Reviewed-by:\s+([^<]+)\s+<([^>]+)>')
+Ptest = re.compile (r' tested-by:\s+([^<]+)\s+<([^>]+)>', re.I)
+Prep = re.compile (r'Reported-by:\s+([^<]+)\s+<([^>]+)>')
+Preptest = re.compile (r'reported-and-tested-by:\s+([^<]+)\s+<([^>]+)>', re.I)
 #
 # Merges are described with a variety of lines.
 #
diff --git a/reports.py b/reports.py
new file mode 100644
index 0000000..08805b8
--- /dev/null
+++ b/reports.py
@@ -0,0 +1,316 @@
+#
+# A new home for the reporting code.
+#
+
+import sys
+
+Outfile = sys.stdout
+HTMLfile = None
+ListCount = 999999
+
+
+def SetOutput (file):
+    global Outfile
+    Outfile = file
+
+def SetHTMLOutput (file):
+    global HTMLfile
+    HTMLfile = file
+
+def SetMaxList (max):
+    global ListCount
+    ListCount = max
+
+
+def Write (stuff):
+    Outfile.write (stuff)
+
+
+
+#
+# HTML output support stuff.
+#
+HTMLclass = 0
+HClasses = ['Even', 'Odd']
+
+THead = '''<p>
+<table cellspacing=3>
+<tr><th colspan=3>%s</th></tr>
+'''
+
+def BeginReport (title):
+    global HTMLclass
+    
+    Outfile.write ('\n%s\n' % title)
+    if HTMLfile:
+        HTMLfile.write (THead % title)
+        HTMLclass = 0
+
+TRow = '''    <tr class="%s">
+<td>%s</td><td align="right">%d</td><td align="right">%.1f%%</td></tr>
+'''
+
+def ReportLine (text, count, pct):
+    global HTMLclass
+    if count == 0:
+        return
+    Outfile.write ('%-25s %4d (%.1f%%)\n' % (text, count, pct))
+    if HTMLfile:
+        HTMLfile.write (TRow % (HClasses[HTMLclass], text, count, pct))
+        HTMLclass ^= 1
+
+def EndReport ():
+    if HTMLfile:
+        HTMLfile.write ('</table>\n\n')
+        
+#
+# Comparison and report generation functions.
+#
+def ComparePCount (h1, h2):
+    return len (h2.patches) - len (h1.patches)
+
+def ReportByPCount (hlist, cscount):
+    hlist.sort (ComparePCount)
+    count = 0
+    BeginReport ('Developers with the most changesets')
+    for h in hlist:
+        pcount = len (h.patches)
+        changed = max(h.added, h.removed)
+        delta = h.added - h.removed
+        if pcount > 0:
+            ReportLine (h.name, pcount, (pcount*100.0)/cscount)
+        count += 1
+        if count >= ListCount:
+            break
+    EndReport ()
+            
+def CompareLChanged (h1, h2):
+    return max(h2.added, h2.removed) - max(h1.added, h1.removed)
+
+def ReportByLChanged (hlist, totalchanged):
+    hlist.sort (CompareLChanged)
+    count = 0
+    BeginReport ('Developers with the most changed lines')
+    for h in hlist:
+        pcount = len (h.patches)
+        changed = max(h.added, h.removed)
+        delta = h.added - h.removed
+        if (h.added + h.removed) > 0:
+            ReportLine (h.name, changed, (changed*100.0)/totalchanged)
+        count += 1
+        if count >= ListCount:
+            break
+    EndReport ()
+            
+def CompareLRemoved (h1, h2):
+    return (h2.removed - h2.added) - (h1.removed - h1.added)
+
+def ReportByLRemoved (hlist, totalremoved):
+    hlist.sort (CompareLRemoved)
+    count = 0
+    BeginReport ('Developers with the most lines removed')
+    for h in hlist:
+        pcount = len (h.patches)
+        changed = max(h.added, h.removed)
+        delta = h.added - h.removed
+        if delta < 0:
+            ReportLine (h.name, -delta, (-delta*100.0)/totalremoved)
+        count += 1
+        if count >= ListCount:
+            break
+    EndReport ()
+
+def CompareEPCount (e1, e2):
+    return e2.count - e1.count
+
+def ReportByPCEmpl (elist, cscount):
+    elist.sort (CompareEPCount)
+    count = 0
+    BeginReport ('Top changeset contributors by employer')
+    for e in elist:
+        if e.count != 0:
+            ReportLine (e.name, e.count, (e.count*100.0)/cscount)
+        count += 1
+        if count >= ListCount:
+            break
+    EndReport ()
+
+
+
+def CompareELChanged (e1, e2):
+    return e2.changed - e1.changed
+
+def ReportByELChanged (elist, totalchanged):
+    elist.sort (CompareELChanged)
+    count = 0
+    BeginReport ('Top lines changed by employer')
+    for e in elist:
+        if e.changed != 0:
+            ReportLine (e.name, e.changed, (e.changed*100.0)/totalchanged)
+        count += 1
+        if count >= ListCount:
+            break
+    EndReport ()
+
+
+
+def CompareSOBs (h1, h2):
+    return len (h2.signoffs) - len (h1.signoffs)
+
+def ReportBySOBs (hlist):
+    hlist.sort (CompareSOBs)
+    totalsobs = 0
+    for h in hlist:
+        totalsobs += len (h.signoffs)
+    count = 0
+    BeginReport ('Developers with the most signoffs (total %d)' % totalsobs)
+    for h in hlist:
+        scount = len (h.signoffs)
+        if scount > 0:
+            ReportLine (h.name, scount, (scount*100.0)/totalsobs)
+        count += 1
+        if count >= ListCount:
+            break
+    EndReport ()
+
+#
+# Reviewer reporting.
+#
+def CompareRevs (h1, h2):
+    return len (h2.reviews) - len (h1.reviews)
+
+def ReportByRevs (hlist):
+    hlist.sort (CompareRevs)
+    totalrevs = 0
+    for h in hlist:
+        totalrevs += len (h.reviews)
+    count = 0
+    BeginReport ('Developers with the most reviews (total %d)' % totalrevs)
+    for h in hlist:
+        scount = len (h.reviews)
+        if scount > 0:
+            ReportLine (h.name, scount, (scount*100.0)/totalrevs)
+        count += 1
+        if count >= ListCount:
+            break
+    EndReport ()
+
+#
+# tester reporting.
+#
+def CompareTests (h1, h2):
+    return len (h2.tested) - len (h1.tested)
+
+def ReportByTests (hlist):
+    hlist.sort (CompareTests)
+    totaltests = 0
+    for h in hlist:
+        totaltests += len (h.tested)
+    count = 0
+    BeginReport ('Developers with the most test credits (total %d)' % totaltests)
+    for h in hlist:
+        scount = len (h.tested)
+        if scount > 0:
+            ReportLine (h.name, scount, (scount*100.0)/totaltests)
+        count += 1
+        if count >= ListCount:
+            break
+    EndReport ()
+
+def CompareTestCred (h1, h2):
+    return h2.testcred - h1.testcred
+
+def ReportByTestCreds (hlist):
+    hlist.sort (CompareTestCred)
+    totaltests = 0
+    for h in hlist:
+        totaltests += h.testcred
+    count = 0
+    BeginReport ('Developers who gave the most tested-by credits (total %d)' % totaltests)
+    for h in hlist:
+        if h.testcred > 0:
+            ReportLine (h.name, h.testcred, (h.testcred*100.0)/totaltests)
+        count += 1
+        if count >= ListCount:
+            break
+    EndReport ()
+
+
+
+#
+# Reporter reporting.
+#
+def CompareReports (h1, h2):
+    return len (h2.reports) - len (h1.reports)
+
+def ReportByReports (hlist):
+    hlist.sort (CompareReports)
+    totalreps = 0
+    for h in hlist:
+        totalreps += len (h.reports)
+    count = 0
+    BeginReport ('Developers with the most report credits (total %d)' % totalreps)
+    for h in hlist:
+        scount = len (h.reports)
+        if scount > 0:
+            ReportLine (h.name, scount, (scount*100.0)/totalreps)
+        count += 1
+        if count >= ListCount:
+            break
+    EndReport ()
+
+def CompareRepCred (h1, h2):
+    return h2.repcred - h1.repcred
+
+def ReportByRepCreds (hlist):
+    hlist.sort (CompareRepCred)
+    totalreps = 0
+    for h in hlist:
+        totalreps += h.repcred
+    count = 0
+    BeginReport ('Developers who gave the most report credits (total %d)' % totalreps)
+    for h in hlist:
+        if h.repcred > 0:
+            ReportLine (h.name, h.repcred, (h.repcred*100.0)/totalreps)
+        count += 1
+        if count >= ListCount:
+            break
+    EndReport ()
+
+
+
+def CompareESOBs (e1, e2):
+    return e2.sobs - e1.sobs
+
+def ReportByESOBs (elist):
+    elist.sort (CompareESOBs)
+    totalsobs = 0
+    for e in elist:
+        totalsobs += e.sobs
+    count = 0
+    BeginReport ('Employers with the most signoffs (total %d)' % totalsobs)
+    for e in elist:
+        if e.sobs > 0:
+            ReportLine (e.name, e.sobs, (e.sobs*100.0)/totalsobs)
+        count += 1
+        if count >= ListCount:
+            break
+    EndReport ()
+
+
+def DevReports (hlist, totalchanged, cscount, totalremoved):
+    ReportByPCount (hlist, cscount)
+    ReportByLChanged (hlist, totalchanged)
+    ReportByLRemoved (hlist, totalremoved)
+    ReportBySOBs (hlist)
+    ReportByRevs (hlist)
+    ReportByTests (hlist)
+    ReportByTestCreds (hlist)
+    ReportByReports (hlist)
+    ReportByRepCreds (hlist)
+
+def EmplReports (elist, totalchanged, cscount):
+    ReportByPCEmpl (elist, cscount)
+    ReportByELChanged (elist, totalchanged)
+    ReportByESOBs (elist)
+