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(h1, h2):
  73     return len(h2.patches) - len(h1.patches)
  74
  75 def ReportByPCount(hlist, cscount):
  76     hlist.sort(ComparePCount)
  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(h1, h2):
  91     return h2.changed - h1.changed
  92
  93 def ReportByLChanged(hlist, totalchanged):
  94     hlist.sort(CompareLChanged)
  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(h1, h2):
 107     return (h2.removed - h2.added) - (h1.removed - h1.added)
 108
 109 def ReportByLRemoved(hlist, totalremoved):
 110     hlist.sort(CompareLRemoved)
 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(e1, e2):
 125     return e2.count - e1.count
 126
 127 def ReportByPCEmpl(elist, cscount):
 128     elist.sort(CompareEPCount)
 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(e1, e2):
 141     return e2.changed - e1.changed
 142
 143 def ReportByELChanged(elist, totalchanged):
 144     elist.sort(CompareELChanged)
 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(h1, h2):
 158     return len(h2.signoffs) - len(h1.signoffs)
 159
 160 def ReportBySOBs(hlist):
 161     hlist.sort(CompareSOBs)
 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(h1, h2):
 180     return len(h2.reviews) - len(h1.reviews)
 181
 182 def ReportByRevs(hlist):
 183     hlist.sort(CompareRevs)
 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(h1, h2):
 202     return len(h2.tested) - len(h1.tested)
 203
 204 def ReportByTests(hlist):
 205     hlist.sort(CompareTests)
 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(h1, h2):
 221     return h2.testcred - h1.testcred
 222
 223 def ReportByTestCreds(hlist):
 224     hlist.sort(CompareTestCred)
 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(h1, h2):
 244     return len(h2.reports) - len(h1.reports)
 245
 246 def ReportByReports(hlist):
 247     hlist.sort(CompareReports)
 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(h1, h2):
 263     return h2.repcred - h1.repcred
 264
 265 def ReportByRepCreds(hlist):
 266     hlist.sort(CompareRepCred)
 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(h1, h2):
 284     if h1.versions and h2.versions:
 285         return len(h2.versions) - len(h1.versions)
 286     if h2.versions:
 287         return 1
 288     if h1.versions:
 289         return -1
 290     return 0
 291
 292 def MissedVersions(hv, allv):
 293     missed = [v for v in allv if v not in hv]
 294     missed.reverse()
 295     return ' '.join(missed)
 296
 297 def ReportVersions(hlist):
 298     hlist.sort(CompareVersionCounts)
 299     BeginReport('Developers represented in the most kernel versions')
 300     count = 0
 301     allversions = hlist[0].versions
 302     for h in hlist:
 303         ReportLineStr(h.name, len(h.versions), MissedVersions(h.versions, allversions))
 304         count += 1
 305         if count >= ListCount:
 306             break
 307     EndReport()
 308
 309
 310 def CompareESOBs(e1, e2):
 311     return e2.sobs - e1.sobs
 312
 313 def ReportByESOBs(elist):
 314     elist.sort(CompareESOBs)
 315     totalsobs = 0
 316     for e in elist:
 317         totalsobs += e.sobs
 318     count = 0
 319     BeginReport('Employers with the most signoffs (total %d)' % totalsobs)
 320     for e in elist:
 321         if e.sobs > 0:
 322             ReportLine(e.name, e.sobs, (e.sobs*100.0)/totalsobs)
 323         count += 1
 324         if count >= ListCount:
 325             break
 326     EndReport()
 327
 328 def CompareHackers(e1, e2):
 329     return len(e2.hackers) - len(e1.hackers)
 330
 331 def ReportByEHackers(elist):
 332     elist.sort(CompareHackers)
 333     totalhackers = 0
 334     for e in elist:
 335         totalhackers += len(e.hackers)
 336     count = 0
 337     BeginReport('Employers with the most hackers (total %d)' % totalhackers)
 338     for e in elist:
 339         nhackers = len(e.hackers)
 340         if nhackers > 0:
 341             ReportLine(e.name, nhackers, (nhackers*100.0)/totalhackers)
 342         count += 1
 343         if count >= ListCount:
 344             break
 345     EndReport()
 346
 347
 348 def DevReports(hlist, totalchanged, cscount, totalremoved):
 349     ReportByPCount(hlist, cscount)
 350     ReportByLChanged(hlist, totalchanged)
 351     ReportByLRemoved(hlist, totalremoved)
 352     ReportBySOBs(hlist)
 353     ReportByRevs(hlist)
 354     ReportByTests(hlist)
 355     ReportByTestCreds(hlist)
 356     ReportByReports(hlist)
 357     ReportByRepCreds(hlist)
 358
 359 def EmplReports(elist, totalchanged, cscount):
 360     ReportByPCEmpl(elist, cscount)
 361     ReportByELChanged(elist, totalchanged)
 362     ReportByESOBs(elist)
 363     ReportByEHackers(elist)
 364
 365 #
 366 # Who are the unknown hackers?
 367 #
 368 def IsUnknown(h):
 369     empl = h.employer[0][0][1].name
 370     return h.email[0] == empl or empl == '(Unknown)'
 371
 372 def ReportUnknowns(hlist, cscount):
 373     #
 374     # Trim the list to just the unknowns; try to work properly whether
 375     # mapping to (Unknown) is happening or not.
 376     #
 377     ulist = [ h for h in hlist if IsUnknown(h) ]
 378     ulist.sort(ComparePCount)
 379     count = 0
 380     BeginReport('Developers with unknown affiliation')
 381     for h in ulist:
 382         pcount = len(h.patches)
 383         if pcount > 0:
 384             ReportLine(h.name, pcount, (pcount*100.0)/cscount)
 385             count += 1
 386         if count >= ListCount:
 387             break
 388     EndReport()
 389
 390
 391
 392 def ReportByFileType(hacker_list):
 393     total = {}
 394     total_by_hacker = {}
 395
 396     BeginReport('Developer contributions by type')
 397     for h in hacker_list:
 398         by_hacker = {}
 399         for patch in h.patches:
 400             # Get a summary by hacker
 401             for (filetype, (added, removed)) in patch.filetypes.iteritems():
 402                 if by_hacker.has_key(filetype):
 403                     by_hacker[filetype][patch.ADDED] += added
 404                     by_hacker[filetype][patch.REMOVED] += removed
 405                 else:
 406                     by_hacker[filetype] = [added, removed]
 407
 408                 # Update the totals
 409                 if total.has_key(filetype):
 410                     total[filetype][patch.ADDED] += added
 411                     total[filetype][patch.REMOVED] += removed
 412                 else:
 413                     total[filetype] = [added, removed, []]
 414
 415         # Print a summary by hacker
 416         print h.name
 417         for filetype, counters in by_hacker.iteritems():
 418             print '\t', filetype, counters
 419             h_added = by_hacker[filetype][patch.ADDED]
 420             h_removed = by_hacker[filetype][patch.REMOVED]
 421             total[filetype][2].append([h.name, h_added, h_removed])
 422
 423     # Print the global summary
 424     BeginReport('Contributions by type and developers')
 425     for filetype, (added, removed, hackers) in total.iteritems():
 426         print filetype, added, removed
 427         for h, h_added, h_removed in hackers:
 428             print '\t%s: [%d, %d]' % (h, h_added, h_removed)
 429
 430     # Print the very global summary
 431     BeginReport('General contributions by type')
 432     for filetype, (added, removed, hackers) in total.iteritems():
 433         print filetype, added, removed
 434
 435 #
 436 # The file access report is a special beast.
 437 #
 438 def FileAccessReport(name, accesses, total):
 439     outf = open(name, 'w')
 440     files = accesses.keys()
 441     files.sort()
 442     for file in files:
 443         a = accesses[file]
 444         outf.write('%6d %6.1f%% %s\n' % (a, (100.0*a)/total, file))
 445     outf.close()