reports.py

   1 #
   2 # A new home for the reporting code.
   3 #
   4 # This code is part of the LWN git data miner.
   5 #
   6 # Copyright 2007-16 Eklektix, Inc.
   7 # Copyright 2007-16 Jonathan Corbet <corbet@lwn.net>
   8 #
   9 # This file may be distributed under the terms of the GNU General
  10 # Public License, version 2.
  11 #
  12
  13 import sys
  14
  15 Outfile = sys.stdout
  16 HTMLfile = None
  17 ListCount = 999999
  18
  19
  20 def SetOutput(file):
  21     global Outfile
  22     Outfile = file
  23
  24 def SetHTMLOutput(file):
  25     global HTMLfile
  26     HTMLfile = file
  27
  28 def SetMaxList(max):
  29     global ListCount
  30     ListCount = max
  31
  32
  33 def Write(stuff):
  34     Outfile.write(stuff)
  35
  36
  37
  38 THead = '''<p>
  39 <table cellspacing=3 class="OddEven">
  40 <tr><th colspan=3>%s</th></tr>
  41 '''
  42
  43 def BeginReport(title):
  44     Outfile.write('\n%s\n' % title)
  45     if HTMLfile:
  46         HTMLfile.write(THead % title)
  47
  48 TRow = ' <tr><td>%s</td><td align="right">%d</td><td align="right">%.1f%%</td></tr>\n'
  49 TRowStr = ' <tr><td>%s</td><td align="right">%d</td><td>%s</td></tr>\n'
  50
  51 def ReportLine(text, count, pct):
  52     if count == 0:
  53         return
  54     Outfile.write ('%-25s %4d (%.1f%%)\n' % (text, count, pct))
  55     if HTMLfile:
  56         HTMLfile.write(TRow % (text, count, pct))
  57
  58 def ReportLineStr(text, count, extra):
  59     if count == 0:
  60         return
  61     Outfile.write ('%-25s %4d %s\n' % (text, count, extra))
  62     if HTMLfile:
  63         HTMLfile.write(TRowStr % (text, count, extra))
  64
  65 def EndReport():
  66     if HTMLfile:
  67         HTMLfile.write('</table>\n\n')
  68
  69 #
  70 # Comparison and report generation functions.
  71 #
  72 def ComparePCount(h):
  73     return len(h.patches)
  74
  75 def ReportByPCount(hlist, cscount):
  76     hlist.sort(key = ComparePCount, reverse = True)
  77     count = 0
  78     BeginReport('Developers with the most changesets')
  79     for h in hlist:
  80         pcount = len(h.patches)
  81         changed = max(h.added, h.removed)
  82         delta = h.added - h.removed
  83         if pcount > 0:
  84             ReportLine(h.name, pcount, (pcount*100.0)/cscount)
  85         count += 1
  86         if count >= ListCount:
  87             break
  88     EndReport()
  89
  90 def CompareLChanged(h):
  91     return h.changed
  92
  93 def ReportByLChanged(hlist, totalchanged):
  94     hlist.sort(key = CompareLChanged, reverse = True)
  95     count = 0
  96     BeginReport('Developers with the most changed lines')
  97     for h in hlist:
  98         pcount = len(h.patches)
  99         if h.changed > 0:
 100             ReportLine(h.name, h.changed, (h.changed*100.0)/totalchanged)
 101         count += 1
 102         if count >= ListCount:
 103             break
 104     EndReport()
 105
 106 def CompareLRemoved(h):
 107     return (h.removed - h.added)
 108
 109 def ReportByLRemoved(hlist, totalremoved):
 110     hlist.sort(key = CompareLRemoved, reverse = True)
 111     count = 0
 112     BeginReport('Developers with the most lines removed')
 113     for h in hlist:
 114         pcount = len(h.patches)
 115         changed = max(h.added, h.removed)
 116         delta = h.added - h.removed
 117         if delta < 0:
 118             ReportLine(h.name, -delta, (-delta*100.0)/totalremoved)
 119         count += 1
 120         if count >= ListCount:
 121             break
 122     EndReport()
 123
 124 def CompareEPCount(e):
 125     return e.count
 126
 127 def ReportByPCEmpl(elist, cscount):
 128     elist.sort(key = CompareEPCount, reverse = True)
 129     count = 0
 130     BeginReport('Top changeset contributors by employer')
 131     for e in elist:
 132         if e.count != 0:
 133             ReportLine(e.name, e.count, (e.count*100.0)/cscount)
 134         count += 1
 135         if count >= ListCount:
 136             break
 137     EndReport()
 138
 139
 140 def CompareELChanged(e):
 141     return e.changed
 142
 143 def ReportByELChanged(elist, totalchanged):
 144     elist.sort(key = CompareELChanged, reverse = True)
 145     count = 0
 146     BeginReport('Top lines changed by employer')
 147     for e in elist:
 148         if e.changed != 0:
 149             ReportLine(e.name, e.changed, (e.changed*100.0)/totalchanged)
 150         count += 1
 151         if count >= ListCount:
 152             break
 153     EndReport()
 154
 155
 156
 157 def CompareSOBs(h):
 158     return len(h.signoffs)
 159
 160 def ReportBySOBs(hlist):
 161     hlist.sort(key = CompareSOBs, reverse = True)
 162     totalsobs = 0
 163     for h in hlist:
 164         totalsobs += len(h.signoffs)
 165     count = 0
 166     BeginReport('Developers with the most signoffs (total %d)' % totalsobs)
 167     for h in hlist:
 168         scount = len(h.signoffs)
 169         if scount > 0:
 170             ReportLine(h.name, scount, (scount*100.0)/totalsobs)
 171         count += 1
 172         if count >= ListCount:
 173             break
 174     EndReport()
 175
 176 #
 177 # Reviewer reporting.
 178 #
 179 def CompareRevs(h):
 180     return len(h.reviews)
 181
 182 def ReportByRevs(hlist):
 183     hlist.sort(key = CompareRevs, reverse = True)
 184     totalrevs = 0
 185     for h in hlist:
 186         totalrevs += len(h.reviews)
 187     count = 0
 188     BeginReport('Developers with the most reviews (total %d)' % totalrevs)
 189     for h in hlist:
 190         scount = len(h.reviews)
 191         if scount > 0:
 192             ReportLine(h.name, scount, (scount*100.0)/totalrevs)
 193         count += 1
 194         if count >= ListCount:
 195             break
 196     EndReport()
 197
 198 #
 199 # tester reporting.
 200 #
 201 def CompareTests(h):
 202     return len(h.tested)
 203
 204 def ReportByTests(hlist):
 205     hlist.sort(key = CompareTests, reverse = True)
 206     totaltests = 0
 207     for h in hlist:
 208         totaltests += len(h.tested)
 209     count = 0
 210     BeginReport('Developers with the most test credits (total %d)' % totaltests)
 211     for h in hlist:
 212         scount = len(h.tested)
 213         if scount > 0:
 214             ReportLine(h.name, scount, (scount*100.0)/totaltests)
 215         count += 1
 216         if count >= ListCount:
 217             break
 218     EndReport()
 219
 220 def CompareTestCred(h):
 221     return h.testcred
 222
 223 def ReportByTestCreds(hlist):
 224     hlist.sort(key = CompareTestCred, reverse = True)
 225     totaltests = 0
 226     for h in hlist:
 227         totaltests += h.testcred
 228     count = 0
 229     BeginReport('Developers who gave the most tested-by credits (total %d)' % totaltests)
 230     for h in hlist:
 231         if h.testcred > 0:
 232             ReportLine(h.name, h.testcred, (h.testcred*100.0)/totaltests)
 233         count += 1
 234         if count >= ListCount:
 235             break
 236     EndReport()
 237
 238
 239
 240 #
 241 # Reporter reporting.
 242 #
 243 def CompareReports(h):
 244     return len(h.reports)
 245
 246 def ReportByReports(hlist):
 247     hlist.sort(key = CompareReports, reverse = True)
 248     totalreps = 0
 249     for h in hlist:
 250         totalreps += len(h.reports)
 251     count = 0
 252     BeginReport('Developers with the most report credits (total %d)' % totalreps)
 253     for h in hlist:
 254         scount = len(h.reports)
 255         if scount > 0:
 256             ReportLine(h.name, scount, (scount*100.0)/totalreps)
 257         count += 1
 258         if count >= ListCount:
 259             break
 260     EndReport()
 261
 262 def CompareRepCred(h):
 263     return h.repcred
 264
 265 def ReportByRepCreds(hlist):
 266     hlist.sort(key = CompareRepCred, reverse = True)
 267     totalreps = 0
 268     for h in hlist:
 269         totalreps += h.repcred
 270     count = 0
 271     BeginReport('Developers who gave the most report credits (total %d)' % totalreps)
 272     for h in hlist:
 273         if h.repcred > 0:
 274             ReportLine(h.name, h.repcred, (h.repcred*100.0)/totalreps)
 275         count += 1
 276         if count >= ListCount:
 277             break
 278     EndReport()
 279
 280 #
 281 # Versions.
 282 #
 283 def CompareVersionCounts(h):
 284     if h.versions:
 285         return len(h.versions)
 286     return -1
 287
 288 def MissedVersions(hv, allv):
 289     missed = [v for v in allv if v not in hv]
 290     missed.reverse()
 291     return ' '.join(missed)
 292
 293 def ReportVersions(hlist):
 294     hlist.sort(key = CompareVersionCounts, reverse = True)
 295     BeginReport('Developers represented in the most kernel versions')
 296     count = 0
 297     allversions = hlist[0].versions
 298     for h in hlist:
 299         ReportLineStr(h.name, len(h.versions), MissedVersions(h.versions, allversions))
 300         count += 1
 301         if count >= ListCount:
 302             break
 303     EndReport()
 304
 305
 306 def CompareESOBs(e):
 307     return e.sobs
 308
 309 def ReportByESOBs(elist):
 310     elist.sort(key = CompareESOBs, reverse = True)
 311     totalsobs = 0
 312     for e in elist:
 313         totalsobs += e.sobs
 314     count = 0
 315     BeginReport('Employers with the most signoffs (total %d)' % totalsobs)
 316     for e in elist:
 317         if e.sobs > 0:
 318             ReportLine(e.name, e.sobs, (e.sobs*100.0)/totalsobs)
 319         count += 1
 320         if count >= ListCount:
 321             break
 322     EndReport()
 323
 324 def CompareHackers(e):
 325     return len(e.hackers)
 326
 327 def ReportByEHackers(elist):
 328     elist.sort(key = CompareHackers, reverse = True)
 329     totalhackers = 0
 330     for e in elist:
 331         totalhackers += len(e.hackers)
 332     count = 0
 333     BeginReport('Employers with the most hackers (total %d)' % totalhackers)
 334     for e in elist:
 335         nhackers = len(e.hackers)
 336         if nhackers > 0:
 337             ReportLine(e.name, nhackers, (nhackers*100.0)/totalhackers)
 338         count += 1
 339         if count >= ListCount:
 340             break
 341     EndReport()
 342
 343
 344 def DevReports(hlist, totalchanged, cscount, totalremoved):
 345     ReportByPCount(hlist, cscount)
 346     ReportByLChanged(hlist, totalchanged)
 347     ReportByLRemoved(hlist, totalremoved)
 348     ReportBySOBs(hlist)
 349     ReportByRevs(hlist)
 350     ReportByTests(hlist)
 351     ReportByTestCreds(hlist)
 352     ReportByReports(hlist)
 353     ReportByRepCreds(hlist)
 354
 355 def EmplReports(elist, totalchanged, cscount):
 356     ReportByPCEmpl(elist, cscount)
 357     ReportByELChanged(elist, totalchanged)
 358     ReportByESOBs(elist)
 359     ReportByEHackers(elist)
 360
 361 #
 362 # Who are the unknown hackers?
 363 #
 364 def IsUnknown(h):
 365     empl = h.employer[0][0][1].name
 366     return h.email[0] == empl or empl == '(Unknown)'
 367
 368 def ReportUnknowns(hlist, cscount):
 369     #
 370     # Trim the list to just the unknowns; try to work properly whether
 371     # mapping to (Unknown) is happening or not.
 372     #
 373     ulist = [ h for h in hlist if IsUnknown(h) ]
 374     ulist.sort(key = ComparePCount, reverse = True)
 375     count = 0
 376     BeginReport('Developers with unknown affiliation')
 377     for h in ulist:
 378         pcount = len(h.patches)
 379         if pcount > 0:
 380             ReportLine(h.name, pcount, (pcount*100.0)/cscount)
 381             count += 1
 382         if count >= ListCount:
 383             break
 384     EndReport()
 385
 386
 387
 388 def ReportByFileType(hacker_list):
 389     total = {}
 390     total_by_hacker = {}
 391
 392     BeginReport('Developer contributions by type')
 393     for h in hacker_list:
 394         by_hacker = {}
 395         for patch in h.patches:
 396             # Get a summary by hacker
 397             for (filetype, (added, removed)) in patch.filetypes.iteritems():
 398                 if by_hacker.has_key(filetype):
 399                     by_hacker[filetype][patch.ADDED] += added
 400                     by_hacker[filetype][patch.REMOVED] += removed
 401                 else:
 402                     by_hacker[filetype] = [added, removed]
 403
 404                 # Update the totals
 405                 if total.has_key(filetype):
 406                     total[filetype][patch.ADDED] += added
 407                     total[filetype][patch.REMOVED] += removed
 408                 else:
 409                     total[filetype] = [added, removed, []]
 410
 411         # Print a summary by hacker
 412         # FIXME why isn't this using Outfile?
 413         print(h.name)
 414         for filetype, counters in by_hacker.iteritems():
 415             print('\t', filetype, counters)
 416             h_added = by_hacker[filetype][patch.ADDED]
 417             h_removed = by_hacker[filetype][patch.REMOVED]
 418             total[filetype][2].append([h.name, h_added, h_removed])
 419
 420     # Print the global summary
 421     BeginReport('Contributions by type and developers')
 422     for filetype, (added, removed, hackers) in total.iteritems():
 423         print(filetype, added, removed)
 424         for h, h_added, h_removed in hackers:
 425             print('\t%s: [%d, %d]' % (h, h_added, h_removed))
 426
 427     # Print the very global summary
 428     BeginReport('General contributions by type')
 429     for filetype, (added, removed, hackers) in total.iteritems():
 430         print(filetype, added, removed)
 431
 432 #
 433 # The file access report is a special beast.
 434 #
 435 def FileAccessReport(name, accesses, total):
 436     outf = open(name, 'w')
 437     files = sorted(accesses)
 438     for file in files:
 439         a = accesses[file]
 440         outf.write('%6d %6.1f%% %s\n' % (a, (100.0*a)/total, file))
 441     outf.close()