buildscripts/output-distance.py

   1 #!@TARGET_PYTHON@
   2 import sys
   3 import optparse
   4 import os
   5 import math
   6
   7 ## so we can call directly as buildscripts/output-distance.py
   8 me_path = os.path.abspath (os.path.split (sys.argv[0])[0])
   9 sys.path.insert (0, me_path + '/../python/')
  10 sys.path.insert (0, me_path + '/../python/out/')
  11
  12
  13 X_AXIS = 0
  14 Y_AXIS = 1
  15 INFTY = 1e6
  16
  17 OUTPUT_EXPRESSION_PENALTY = 1
  18 ORPHAN_GROB_PENALTY = 1
  19 options = None
  20
  21 ################################################################
  22 # system interface.
  23 temp_dir = None
  24 class TempDirectory:
  25     def __init__ (self):
  26         import tempfile
  27         self.dir = tempfile.mkdtemp ()
  28         print 'dir is', self.dir
  29     def __del__ (self):
  30         print 'rm -rf %s' % self.dir
  31         os.system ('rm -rf %s' % self.dir)
  32     def __call__ (self):
  33         return self.dir
  34
  35
  36 def get_temp_dir  ():
  37     global temp_dir
  38     if not temp_dir:
  39         temp_dir = TempDirectory ()
  40     return temp_dir ()
  41
  42 def read_pipe (c):
  43     print 'pipe' , c
  44     return os.popen (c).read ()
  45
  46 def system (c):
  47     print 'system' , c
  48     s = os.system (c)
  49     if s :
  50         raise Exception ("failed")
  51     return
  52
  53 def shorten_string (s):
  54     threshold = 15
  55     if len (s) > 2*threshold:
  56         s = s[:threshold] + '..' + s[-threshold:]
  57     return s
  58
  59 def max_distance (x1, x2):
  60     dist = 0.0
  61
  62     for (p,q) in zip (x1, x2):
  63         dist = max (abs (p-q), dist)
  64
  65     return dist
  66
  67
  68 def compare_png_images (old, new, dest_dir):
  69     def png_dims (f):
  70         m = re.search ('([0-9]+) x ([0-9]+)', read_pipe ('file %s' % f))
  71
  72         return tuple (map (int, m.groups ()))
  73
  74     dest = os.path.join (dest_dir, new.replace ('.png', '.compare.jpeg'))
  75     try:
  76         dims1 = png_dims (old)
  77         dims2 = png_dims (new)
  78     except AttributeError:
  79         ## hmmm. what to do?
  80         system ('touch %(dest)s' % locals ())
  81         return
  82
  83     dims = (min (dims1[0], dims2[0]),
  84             min (dims1[1], dims2[1]))
  85
  86     dir = get_temp_dir ()
  87     system ('convert -depth 8 -crop %dx%d+0+0 %s %s/crop1.png' % (dims + (old, dir)))
  88     system ('convert -depth 8 -crop %dx%d+0+0 %s %s/crop2.png' % (dims + (new, dir)))
  89
  90     system ('compare -depth 8 %(dir)s/crop1.png %(dir)s/crop2.png %(dir)s/diff.png' % locals ())
  91
  92     system ("convert  -depth 8 %(dir)s/diff.png -blur 0x3 -negate -channel alpha,blue -type TrueColorMatte -fx 'intensity'    %(dir)s/matte.png" % locals ())
  93
  94     system ("composite -compose atop -quality 65 %(dir)s/matte.png %(new)s %(dest)s" % locals ())
  95
  96
  97 ################################################################
  98 # interval/bbox arithmetic.
  99
 100 empty_interval = (INFTY, -INFTY)
 101 empty_bbox = (empty_interval, empty_interval)
 102
 103 def interval_is_empty (i):
 104     return i[0] > i[1]
 105
 106 def interval_length (i):
 107     return max (i[1]-i[0], 0)
 108
 109 def interval_union (i1, i2):
 110     return (min (i1[0], i2[0]),
 111             max (i1[1], i2[1]))
 112
 113 def interval_intersect (i1, i2):
 114     return (max (i1[0], i2[0]),
 115             min (i1[1], i2[1]))
 116
 117 def bbox_is_empty (b):
 118     return (interval_is_empty (b[0])
 119             or interval_is_empty (b[1]))
 120
 121 def bbox_union (b1, b2):
 122     return (interval_union (b1[X_AXIS], b2[X_AXIS]),
 123             interval_union (b2[Y_AXIS], b2[Y_AXIS]))
 124
 125 def bbox_intersection (b1, b2):
 126     return (interval_intersect (b1[X_AXIS], b2[X_AXIS]),
 127             interval_intersect (b2[Y_AXIS], b2[Y_AXIS]))
 128
 129 def bbox_area (b):
 130     return interval_length (b[X_AXIS]) * interval_length (b[Y_AXIS])
 131
 132 def bbox_diameter (b):
 133     return max (interval_length (b[X_AXIS]),
 134                 interval_length (b[Y_AXIS]))
 135
 136
 137 def difference_area (a, b):
 138     return bbox_area (a) - bbox_area (bbox_intersection (a,b))
 139
 140 class GrobSignature:
 141     def __init__ (self, exp_list):
 142         (self.name, self.origin, bbox_x,
 143          bbox_y, self.output_expression) = tuple (exp_list)
 144
 145         self.bbox = (bbox_x, bbox_y)
 146         self.centroid = (bbox_x[0] + bbox_x[1], bbox_y[0] + bbox_y[1])
 147
 148     def __repr__ (self):
 149         return '%s: (%.2f,%.2f), (%.2f,%.2f)\n' % (self.name,
 150                                                    self.bbox[0][0],
 151                                                    self.bbox[0][1],
 152                                                    self.bbox[1][0],
 153                                                    self.bbox[1][1])
 154
 155     def axis_centroid (self, axis):
 156         return apply (sum, self.bbox[axis])  / 2
 157
 158     def centroid_distance (self, other, scale):
 159         return max_distance (self.centroid, other.centroid) / scale
 160
 161     def bbox_distance (self, other):
 162         divisor = bbox_area (self.bbox) + bbox_area (other.bbox)
 163
 164         if divisor:
 165             return (difference_area (self.bbox, other.bbox) +
 166                     difference_area (other.bbox, self.bbox)) / divisor
 167         else:
 168             return 0.0
 169
 170     def expression_distance (self, other):
 171         if self.output_expression == other.output_expression:
 172             return 0
 173         else:
 174             return 1
 175
 176 ################################################################
 177 # single System.
 178
 179 class SystemSignature:
 180     def __init__ (self, grob_sigs):
 181         d = {}
 182         for g in grob_sigs:
 183             val = d.setdefault (g.name, [])
 184             val += [g]
 185
 186         self.grob_dict = d
 187         self.set_all_bbox (grob_sigs)
 188
 189     def set_all_bbox (self, grobs):
 190         self.bbox = empty_bbox
 191         for g in grobs:
 192             self.bbox = bbox_union (g.bbox, self.bbox)
 193
 194     def closest (self, grob_name, centroid):
 195         min_d = INFTY
 196         min_g = None
 197         try:
 198             grobs = self.grob_dict[grob_name]
 199
 200             for g in grobs:
 201                 d = max_distance (g.centroid, centroid)
 202                 if d < min_d:
 203                     min_d = d
 204                     min_g = g
 205
 206
 207             return min_g
 208
 209         except KeyError:
 210             return None
 211     def grobs (self):
 212         return reduce (lambda x,y: x+y, self.grob_dict.values(), [])
 213
 214 ################################################################
 215 ## comparison of systems.
 216
 217 class SystemLink:
 218     def __init__ (self, system1, system2):
 219         self.system1 = system1
 220         self.system2 = system2
 221
 222         self.link_list_dict = {}
 223         self.back_link_dict = {}
 224
 225
 226         ## pairs
 227         self.orphans = []
 228
 229         ## pair -> distance
 230         self.geo_distances = {}
 231
 232         ## pairs
 233         self.expression_changed = []
 234
 235         self._geometric_distance = None
 236         self._expression_change_count = None
 237         self._orphan_count = None
 238
 239         for g in system1.grobs ():
 240
 241             ## skip empty bboxes.
 242             if bbox_is_empty (g.bbox):
 243                 continue
 244
 245             closest = system2.closest (g.name, g.centroid)
 246
 247             self.link_list_dict.setdefault (closest, [])
 248             self.link_list_dict[closest].append (g)
 249             self.back_link_dict[g] = closest
 250
 251
 252     def calc_geometric_distance (self):
 253         total = 0.0
 254         for (g1,g2) in self.back_link_dict.items ():
 255             if g2:
 256                 d = g1.bbox_distance (g2)
 257                 if d:
 258                     self.geo_distances[(g1,g2)] = d
 259
 260                 total += d
 261
 262         self._geometric_distance = total
 263
 264     def calc_orphan_count (self):
 265         count = 0
 266         for (g1, g2) in self.back_link_dict.items ():
 267             if g2 == None:
 268                 self.orphans.append ((g1, None))
 269
 270                 count += 1
 271
 272         self._orphan_count = count
 273
 274     def calc_output_exp_distance (self):
 275         d = 0
 276         for (g1,g2) in self.back_link_dict.items ():
 277             if g2:
 278                 d += g1.expression_distance (g2)
 279
 280         self._expression_change_count = d
 281
 282     def output_expression_details_string (self):
 283         return ', '.join ([g1.name for g1 in self.expression_changed])
 284
 285     def geo_details_string (self):
 286         results = [(d, g1,g2) for ((g1, g2), d) in self.geo_distances.items()]
 287         results.sort ()
 288         results.reverse ()
 289
 290         return ', '.join (['%s: %f' % (g1.name, d) for (d, g1, g2) in results])
 291
 292     def orphan_details_string (self):
 293         return ', '.join (['%s-None' % g1.name for (g1,g2) in self.orphans if g2==None])
 294
 295     def geometric_distance (self):
 296         if self._geometric_distance == None:
 297             self.calc_geometric_distance ()
 298         return self._geometric_distance
 299
 300     def orphan_count (self):
 301         if self._orphan_count == None:
 302             self.calc_orphan_count ()
 303
 304         return self._orphan_count
 305
 306     def output_expression_change_count (self):
 307         if self._expression_change_count == None:
 308             self.calc_output_exp_distance ()
 309         return self._expression_change_count
 310
 311     def distance (self):
 312         return (self.output_expression_change_count (),
 313                 self.orphan_count (),
 314                 self.geometric_distance ())
 315
 316 def read_signature_file (name):
 317     print 'reading', name
 318
 319     entries = open (name).read ().split ('\n')
 320     def string_to_tup (s):
 321         return tuple (map (float, s.split (' ')))
 322
 323     def string_to_entry (s):
 324         fields = s.split('@')
 325         fields[2] = string_to_tup (fields[2])
 326         fields[3] = string_to_tup (fields[3])
 327
 328         return tuple (fields)
 329
 330     entries = [string_to_entry (e) for e in entries
 331                if e and not e.startswith ('#')]
 332
 333     grob_sigs = [GrobSignature (e) for e in entries]
 334     sig = SystemSignature (grob_sigs)
 335     return sig
 336
 337
 338 ################################################################
 339 # different systems of a .ly file.
 340
 341 hash_to_original_name = {}
 342
 343 class FileLink:
 344     def __init__ (self, f1, f2):
 345         self._distance = None
 346         self.file_names = (f1, f2)
 347
 348     def text_record_string (self):
 349         return '%-30f %-20s\n' % (self.distance (),
 350                                   self.name ()
 351                                   + os.path.splitext (self.file_names[1])[1]
 352                                   )
 353
 354     def calc_distance (self):
 355         return 0.0
 356
 357     def distance (self):
 358         if self._distance == None:
 359            self._distance = self.calc_distance ()
 360
 361         return self._distance
 362
 363     def source_file (self):
 364         for ext in ('.ly', '.ly.txt'):
 365             base = os.path.splitext (self.file_names[1])[0]
 366             f = base + ext
 367             if os.path.exists (f):
 368                 return f
 369
 370         return ''
 371
 372     def name (self):
 373         base = os.path.basename (self.file_names[1])
 374         base = os.path.splitext (base)[0]
 375         base = hash_to_original_name.get (base, base)
 376         base = os.path.splitext (base)[0]
 377         return base
 378
 379     def extension (self):
 380         return os.path.splitext (self.file_names[1])[1]
 381
 382     def link_files_for_html (self, dest_dir):
 383         for f in self.file_names:
 384             link_file (f, os.path.join (dest_dir, f))
 385
 386     def get_distance_details (self):
 387         return ''
 388
 389     def get_cell (self, oldnew):
 390         return ''
 391
 392     def get_file (self, oldnew):
 393         return self.file_names[oldnew]
 394
 395     def html_record_string (self, dest_dir):
 396         dist = self.distance()
 397
 398         details = self.get_distance_details ()
 399         if details:
 400             details_base = os.path.splitext (self.file_names[1])[0]
 401             details_base += '.details.html'
 402             fn = dest_dir + '/'  + details_base
 403             open_write_file (fn).write (details)
 404
 405             details = '<br>(<a href="%(details_base)s">details</a>)' % locals ()
 406
 407         cell1 = self.get_cell (0)
 408         cell2 = self.get_cell (1)
 409
 410         name = self.name () + self.extension ()
 411         file1 = self.get_file (0)
 412         file2 = self.get_file (1)
 413
 414         return '''<tr>
 415 <td>
 416 %(dist)f
 417 %(details)s
 418 </td>
 419 <td>%(cell1)s<br><font size=-2><a href="%(file1)s"><tt>%(name)s</tt></font></td>
 420 <td>%(cell2)s<br><font size=-2><a href="%(file2)s"><tt>%(name)s</tt></font></td>
 421 </tr>''' % locals ()
 422
 423
 424 class FileCompareLink (FileLink):
 425     def __init__ (self, f1, f2):
 426         FileLink.__init__ (self, f1, f2)
 427         self.contents = (self.get_content (self.file_names[0]),
 428                          self.get_content (self.file_names[1]))
 429
 430
 431     def calc_distance (self):
 432         ## todo: could use import MIDI to pinpoint
 433         ## what & where changed.
 434
 435         if self.contents[0] == self.contents[1]:
 436             return 0.0
 437         else:
 438             return 100.0;
 439
 440     def get_content (self, f):
 441         print 'reading', f
 442         s = open (f).read ()
 443         return s
 444
 445
 446 class GitFileCompareLink (FileCompareLink):
 447     def get_cell (self, oldnew):
 448         str = self.contents[oldnew]
 449
 450         # truncate long lines
 451         str = '\n'.join ([l[:80] for l in str.split ('\n')])
 452
 453
 454         str = '<font size="-2"><pre>%s</pre></font>' % str
 455         return str
 456
 457     def calc_distance (self):
 458         if self.contents[0] == self.contents[1]:
 459             d = 0.0
 460         else:
 461             d = 1.0001 *options.threshold
 462
 463         return d
 464
 465
 466 class TextFileCompareLink (FileCompareLink):
 467     def calc_distance (self):
 468         import difflib
 469         diff = difflib.unified_diff (self.contents[0].strip().split ('\n'),
 470                                      self.contents[1].strip().split ('\n'),
 471                                      fromfiledate = self.file_names[0],
 472                                      tofiledate = self.file_names[1]
 473                                      )
 474
 475         self.diff_lines =  [l for l in diff]
 476         self.diff_lines = self.diff_lines[2:]
 477
 478         return math.sqrt (float (len ([l for l in self.diff_lines if l[0] in '-+'])))
 479
 480     def get_cell (self, oldnew):
 481         str = ''
 482         if oldnew == 1:
 483             str = '\n'.join ([d.replace ('\n','') for d in self.diff_lines])
 484         str = '<font size="-2"><pre>%s</pre></font>' % str
 485         return str
 486
 487 class LogFileCompareLink (TextFileCompareLink):
 488   def get_content (self, f):
 489       c = TextFileCompareLink.get_content (self, f)
 490       c = re.sub ("\nProcessing `[^\n]+'\n", '', c)
 491       return c
 492
 493 class ProfileFileLink (FileCompareLink):
 494     def __init__ (self, f1, f2):
 495         FileCompareLink.__init__ (self, f1, f2)
 496         self.results = [{}, {}]
 497
 498     def get_cell (self, oldnew):
 499         str = ''
 500         for k in ('time', 'cells'):
 501             if oldnew==0:
 502                 str += '%-8s: %d\n' %  (k, int (self.results[oldnew][k]))
 503             else:
 504                 str += '%-8s: %8d (%5.3f)\n' % (k, int (self.results[oldnew][k]),
 505                                                 self.get_ratio (k))
 506
 507         return '<pre>%s</pre>' % str
 508
 509     def get_ratio (self, key):
 510         (v1,v2) = (self.results[0].get (key, -1),
 511                    self.results[1].get (key, -1))
 512
 513         if v1 <= 0 or v2 <= 0:
 514             return 0.0
 515
 516         return (v1 - v2) / float (v1+v2)
 517
 518     def calc_distance (self):
 519         for oldnew in (0,1):
 520             def note_info (m):
 521                 self.results[oldnew][m.group(1)] = float (m.group (2))
 522
 523             re.sub ('([a-z]+): ([-0-9.]+)\n',
 524                     note_info, self.contents[oldnew])
 525
 526         dist = 0.0
 527         factor = {
 528             'time': 0.1,
 529             'cells': 5.0,
 530             }
 531
 532         for k in ('time', 'cells'):
 533             real_val = math.tan (self.get_ratio (k) * 0.5 * math.pi)
 534             dist += math.exp (math.fabs (real_val) * factor[k])  - 1
 535
 536         dist = min (dist, 100)
 537         return dist
 538
 539
 540 class MidiFileLink (TextFileCompareLink):
 541     def get_content (self, oldnew):
 542         import midi
 543
 544         data = FileCompareLink.get_content (self, oldnew)
 545         midi = midi.parse (data)
 546         tracks = midi[1]
 547
 548         str = ''
 549         j = 0
 550         for t in tracks:
 551             str += 'track %d' % j
 552             j += 1
 553
 554             for e in t:
 555                 ev_str = repr (e)
 556                 if re.search ('LilyPond [0-9.]+', ev_str):
 557                     continue
 558
 559                 str += '  ev %s\n' % `e`
 560         return str
 561
 562
 563
 564 class SignatureFileLink (FileLink):
 565     def __init__ (self, f1, f2 ):
 566         FileLink.__init__ (self, f1, f2)
 567         self.system_links = {}
 568
 569     def add_system_link (self, link, number):
 570         self.system_links[number] = link
 571
 572     def calc_distance (self):
 573         d = 0.0
 574
 575         orphan_distance = 0.0
 576         for l in self.system_links.values ():
 577             d = max (d, l.geometric_distance ())
 578             orphan_distance += l.orphan_count ()
 579
 580         return d + orphan_distance
 581
 582     def add_file_compare (self, f1, f2):
 583         system_index = []
 584
 585         def note_system_index (m):
 586             system_index.append (int (m.group (1)))
 587             return ''
 588
 589         base1 = re.sub ("-([0-9]+).signature", note_system_index, f1)
 590         base2 = re.sub ("-([0-9]+).signature", note_system_index, f2)
 591
 592         self.base_names = (os.path.normpath (base1),
 593                            os.path.normpath (base2))
 594
 595         s1 = read_signature_file (f1)
 596         s2 = read_signature_file (f2)
 597
 598         link = SystemLink (s1, s2)
 599
 600         self.add_system_link (link, system_index[0])
 601
 602
 603     def create_images (self, dest_dir):
 604
 605         files_created = [[], []]
 606         for oldnew in (0, 1):
 607             pat = self.base_names[oldnew] + '.eps'
 608
 609             for f in glob.glob (pat):
 610                 infile = f
 611                 outfile = (dest_dir + '/' + f).replace ('.eps', '.png')
 612                 data_option = ''
 613                 if options.local_data_dir:
 614                     data_option = ('-slilypond-datadir=%s/../share/lilypond/current '
 615                                    % os.path.dirname(infile))
 616
 617                 mkdir (os.path.split (outfile)[0])
 618                 cmd = ('gs -sDEVICE=png16m -dGraphicsAlphaBits=4 -dTextAlphaBits=4 '
 619                        ' %(data_option)s '
 620                        ' -r101 '
 621                        ' -sOutputFile=%(outfile)s -dNOSAFER -dEPSCrop -q -dNOPAUSE '
 622                        ' %(infile)s  -c quit ') % locals ()
 623
 624                 files_created[oldnew].append (outfile)
 625                 system (cmd)
 626
 627         return files_created
 628
 629     def link_files_for_html (self, dest_dir):
 630         FileLink.link_files_for_html (self, dest_dir)
 631         to_compare = [[], []]
 632
 633         exts = []
 634         if options.create_images:
 635             to_compare = self.create_images (dest_dir)
 636         else:
 637             exts += ['.png', '-page*png']
 638
 639         for ext in exts:
 640             for oldnew in (0,1):
 641                 for f in glob.glob (self.base_names[oldnew] + ext):
 642                     dst = dest_dir + '/' + f
 643                     link_file (f, dst)
 644
 645                     if f.endswith ('.png'):
 646                         to_compare[oldnew].append (f)
 647
 648         if options.compare_images:
 649             for (old, new) in zip (to_compare[0], to_compare[1]):
 650                 compare_png_images (old, new, dest_dir)
 651
 652
 653     def get_cell (self, oldnew):
 654         def img_cell (ly, img, name):
 655             if not name:
 656                 name = 'source'
 657             else:
 658                 name = '<tt>%s</tt>' % name
 659
 660             return '''
 661 <a href="%(img)s">
 662 <img src="%(img)s" style="border-style: none; max-width: 500px;">
 663 </a><br>
 664 ''' % locals ()
 665         def multi_img_cell (ly, imgs, name):
 666             if not name:
 667                 name = 'source'
 668             else:
 669                 name = '<tt>%s</tt>' % name
 670
 671             imgs_str = '\n'.join (['''<a href="%s">
 672 <img src="%s" style="border-style: none; max-width: 500px;">
 673 </a><br>''' % (img, img)
 674                                   for img in imgs])
 675
 676
 677             return '''
 678 %(imgs_str)s
 679 ''' % locals ()
 680
 681
 682
 683         def cell (base, name):
 684             pat = base + '-page*.png'
 685             pages = glob.glob (pat)
 686
 687             if pages:
 688                 return multi_img_cell (base + '.ly', sorted (pages), name)
 689             else:
 690                 return img_cell (base + '.ly', base + '.png', name)
 691
 692
 693
 694         str = cell (os.path.splitext (self.file_names[oldnew])[0], self.name ())
 695         if options.compare_images and oldnew == 1:
 696             str = str.replace ('.png', '.compare.jpeg')
 697
 698         return str
 699
 700
 701     def get_distance_details (self):
 702         systems = self.system_links.items ()
 703         systems.sort ()
 704
 705         html = ""
 706         for (c, link) in systems:
 707             e = '<td>%d</td>' % c
 708             for d in link.distance ():
 709                 e += '<td>%f</td>' % d
 710
 711             e = '<tr>%s</tr>' % e
 712
 713             html += e
 714
 715             e = '<td>%d</td>' % c
 716             for s in (link.output_expression_details_string (),
 717                       link.orphan_details_string (),
 718                       link.geo_details_string ()):
 719                 e += "<td>%s</td>" % s
 720
 721
 722             e = '<tr>%s</tr>' % e
 723             html += e
 724
 725         original = self.name ()
 726         html = '''<html>
 727 <head>
 728 <title>comparison details for %(original)s</title>
 729 </head>
 730 <body>
 731 <table border=1>
 732 <tr>
 733 <th>system</th>
 734 <th>output</th>
 735 <th>orphan</th>
 736 <th>geo</th>
 737 </tr>
 738
 739 %(html)s
 740 </table>
 741
 742 </body>
 743 </html>
 744 ''' % locals ()
 745         return html
 746
 747
 748 ################################################################
 749 # Files/directories
 750
 751 import glob
 752 import re
 753
 754 def compare_signature_files (f1, f2):
 755     s1 = read_signature_file (f1)
 756     s2 = read_signature_file (f2)
 757
 758     return SystemLink (s1, s2).distance ()
 759
 760 def paired_files (dir1, dir2, pattern):
 761     """
 762     Search DIR1 and DIR2 for PATTERN.
 763
 764     Return (PAIRED, MISSING-FROM-2, MISSING-FROM-1)
 765
 766     """
 767
 768     files = []
 769     for d in (dir1,dir2):
 770         found = [os.path.split (f)[1] for f in glob.glob (d + '/' + pattern)]
 771         found = dict ((f, 1) for f in found)
 772         files.append (found)
 773
 774     pairs = []
 775     missing = []
 776     for f in files[0]:
 777         try:
 778             files[1].pop (f)
 779             pairs.append (f)
 780         except KeyError:
 781             missing.append (f)
 782
 783     return (pairs, files[1].keys (), missing)
 784
 785 class ComparisonData:
 786     def __init__ (self):
 787         self.result_dict = {}
 788         self.missing = []
 789         self.added = []
 790         self.file_links = {}
 791
 792     def read_sources (self):
 793
 794         ## ugh: drop the .ly.txt
 795         for (key, val) in self.file_links.items ():
 796
 797             def note_original (match, ln=val):
 798                 key = ln.name ()
 799                 hash_to_original_name[key] = match.group (1)
 800                 return ''
 801
 802             sf = val.source_file ()
 803             if sf:
 804                 re.sub (r'\\sourcefilename "([^"]+)"',
 805                         note_original, open (sf).read ())
 806             else:
 807                 print 'no source for', val
 808
 809     def compare_trees (self, dir1, dir2):
 810         self.compare_directories (dir1, dir2)
 811
 812         (root, dirs, files) = os.walk (dir1).next ()
 813         for d in dirs:
 814             d1 = os.path.join (dir1, d)
 815             d2 = os.path.join (dir2, d)
 816
 817             if os.path.islink (d1) or os.path.islink (d2):
 818                 continue
 819
 820             if os.path.isdir (d2):
 821                 self.compare_trees (d1, d2)
 822
 823     def compare_directories (self, dir1, dir2):
 824         for ext in ['signature',
 825                     'midi',
 826                     'log',
 827                     'profile',
 828                     'gittxt']:
 829             (paired, m1, m2) = paired_files (dir1, dir2, '*.' + ext)
 830
 831             self.missing += [(dir1, m) for m in m1]
 832             self.added += [(dir2, m) for m in m2]
 833
 834             for p in paired:
 835                 if (options.max_count
 836                     and len (self.file_links) > options.max_count):
 837                     continue
 838
 839                 f2 = dir2 +  '/' + p
 840                 f1 = dir1 +  '/' + p
 841                 self.compare_files (f1, f2)
 842
 843     def compare_files (self, f1, f2):
 844         if f1.endswith ('signature'):
 845             self.compare_signature_files (f1, f2)
 846         else:
 847             ext = os.path.splitext (f1)[1]
 848             klasses = {
 849                 '.midi': MidiFileLink,
 850                 '.log' : LogFileCompareLink,
 851                 '.profile': ProfileFileLink,
 852                 '.gittxt': GitFileCompareLink,
 853                 }
 854
 855             if klasses.has_key (ext):
 856                 self.compare_general_files (klasses[ext], f1, f2)
 857
 858     def compare_general_files (self, klass, f1, f2):
 859         name = os.path.split (f1)[1]
 860
 861         file_link = klass (f1, f2)
 862         self.file_links[name] = file_link
 863
 864     def compare_signature_files (self, f1, f2):
 865         name = os.path.split (f1)[1]
 866         name = re.sub ('-[0-9]+.signature', '', name)
 867
 868         file_link = None
 869         try:
 870             file_link = self.file_links[name]
 871         except KeyError:
 872             generic_f1 = re.sub ('-[0-9]+.signature', '.ly', f1)
 873             generic_f2 = re.sub ('-[0-9]+.signature', '.ly', f2)
 874             file_link = SignatureFileLink (generic_f1, generic_f2)
 875             self.file_links[name] = file_link
 876
 877         file_link.add_file_compare (f1, f2)
 878
 879     def write_changed (self, dest_dir, threshold):
 880         (changed, below, unchanged) = self.thresholded_results (threshold)
 881
 882         str = '\n'.join ([os.path.splitext (link.file_names[1])[0]
 883                         for link in changed])
 884         fn = dest_dir + '/changed.txt'
 885
 886         open_write_file (fn).write (str)
 887
 888     def thresholded_results (self, threshold):
 889         ## todo: support more scores.
 890         results = [(link.distance(), link)
 891                    for link in self.file_links.values ()]
 892         results.sort ()
 893         results.reverse ()
 894
 895         unchanged = [r for (d,r) in results if d == 0.0]
 896         below = [r for (d,r) in results if threshold >= d > 0.0]
 897         changed = [r for (d,r) in results if d > threshold]
 898
 899         return (changed, below, unchanged)
 900
 901     def write_text_result_page (self, filename, threshold):
 902         out = None
 903         if filename == '':
 904             out = sys.stdout
 905         else:
 906             print 'writing "%s"' % filename
 907             out = open_write_file (filename)
 908
 909         (changed, below, unchanged) = self.thresholded_results (threshold)
 910
 911
 912         for link in changed:
 913             out.write (link.text_record_string ())
 914
 915         out.write ('\n\n')
 916         out.write ('%d below threshold\n' % len (below))
 917         out.write ('%d unchanged\n' % len (unchanged))
 918
 919     def create_text_result_page (self, dir1, dir2, dest_dir, threshold):
 920         self.write_text_result_page (dest_dir + '/index.txt', threshold)
 921
 922     def create_html_result_page (self, dir1, dir2, dest_dir, threshold):
 923         dir1 = dir1.replace ('//', '/')
 924         dir2 = dir2.replace ('//', '/')
 925
 926         (changed, below, unchanged) = self.thresholded_results (threshold)
 927
 928
 929         html = ''
 930         old_prefix = os.path.split (dir1)[1]
 931         for link in changed:
 932             html += link.html_record_string (dest_dir)
 933
 934
 935         short_dir1 = shorten_string (dir1)
 936         short_dir2 = shorten_string (dir2)
 937         html = '''<html>
 938 <table rules="rows" border bordercolor="blue">
 939 <tr>
 940 <th>distance</th>
 941 <th>%(short_dir1)s</th>
 942 <th>%(short_dir2)s</th>
 943 </tr>
 944 %(html)s
 945 </table>
 946 </html>''' % locals()
 947
 948         html += ('<p>')
 949         below_count = len (below)
 950
 951         if below_count:
 952             html += ('<p>%d below threshold</p>' % below_count)
 953
 954         html += ('<p>%d unchanged</p>' % len (unchanged))
 955
 956         dest_file = dest_dir + '/index.html'
 957         open_write_file (dest_file).write (html)
 958
 959
 960         for link in changed:
 961             link.link_files_for_html (dest_dir)
 962
 963
 964     def print_results (self, threshold):
 965         self.write_text_result_page ('', threshold)
 966
 967 def compare_trees (dir1, dir2, dest_dir, threshold):
 968     data = ComparisonData ()
 969     data.compare_trees (dir1, dir2)
 970     data.read_sources ()
 971
 972
 973     data.print_results (threshold)
 974
 975     if os.path.isdir (dest_dir):
 976         system ('rm -rf %s '% dest_dir)
 977
 978     data.write_changed (dest_dir, threshold)
 979     data.create_html_result_page (dir1, dir2, dest_dir, threshold)
 980     data.create_text_result_page (dir1, dir2, dest_dir, threshold)
 981
 982 ################################################################
 983 # TESTING
 984
 985 def mkdir (x):
 986     if not os.path.isdir (x):
 987         print 'mkdir', x
 988         os.makedirs (x)
 989
 990 def link_file (x, y):
 991     mkdir (os.path.split (y)[0])
 992     try:
 993         print x, '->', y
 994         os.link (x, y)
 995     except OSError, z:
 996         print 'OSError', x, y, z
 997         raise OSError
 998
 999 def open_write_file (x):
1000     d = os.path.split (x)[0]
1001     mkdir (d)
1002     return open (x, 'w')
1003
1004
1005 def system (x):
1006
1007     print 'invoking', x
1008     stat = os.system (x)
1009     assert stat == 0
1010
1011
1012 def test_paired_files ():
1013     print paired_files (os.environ["HOME"] + "/src/lilypond/scripts/",
1014                         os.environ["HOME"] + "/src/lilypond-stable/buildscripts/", '*.py')
1015
1016
1017 def test_compare_trees ():
1018     system ('rm -rf dir1 dir2')
1019     system ('mkdir dir1 dir2')
1020     system ('cp 20{-*.signature,.ly,.png,.eps,.log,.profile} dir1')
1021     system ('cp 20{-*.signature,.ly,.png,.eps,.log,.profile} dir2')
1022     system ('cp 20expr{-*.signature,.ly,.png,.eps,.log,.profile} dir1')
1023     system ('cp 19{-*.signature,.ly,.png,.eps,.log,.profile} dir2/')
1024     system ('cp 19{-*.signature,.ly,.png,.eps,.log,.profile} dir1/')
1025     system ('cp 19-1.signature 19.sub-1.signature')
1026     system ('cp 19.ly 19.sub.ly')
1027     system ('cp 19.profile 19.sub.profile')
1028     system ('cp 19.log 19.sub.log')
1029     system ('cp 19.png 19.sub.png')
1030     system ('cp 19.eps 19.sub.eps')
1031
1032     system ('cp 20multipage* dir1')
1033     system ('cp 20multipage* dir2')
1034     system ('cp 19multipage-1.signature dir2/20multipage-1.signature')
1035
1036
1037     system ('mkdir -p dir1/subdir/ dir2/subdir/')
1038     system ('cp 19.sub{-*.signature,.ly,.png,.eps,.log,.profile} dir1/subdir/')
1039     system ('cp 19.sub{-*.signature,.ly,.png,.eps,.log,.profile} dir2/subdir/')
1040     system ('cp 20grob{-*.signature,.ly,.png,.eps,.log,.profile} dir2/')
1041     system ('cp 20grob{-*.signature,.ly,.png,.eps,.log,.profile} dir1/')
1042     system ('echo HEAD is 1 > dir1/tree.gittxt')
1043     system ('echo HEAD is 2 > dir2/tree.gittxt')
1044
1045     ## introduce differences
1046     system ('cp 19-1.signature dir2/20-1.signature')
1047     system ('cp 19.profile dir2/20.profile')
1048     system ('cp 19.png dir2/20.png')
1049     system ('cp 19multipage-page1.png dir2/20multipage-page1.png')
1050     system ('cp 20-1.signature dir2/subdir/19.sub-1.signature')
1051     system ('cp 20.png dir2/subdir/19.sub.png')
1052     system ("sed 's/: /: 1/g'  20.profile > dir2/subdir/19.sub.profile")
1053
1054     ## radical diffs.
1055     system ('cp 19-1.signature dir2/20grob-1.signature')
1056     system ('cp 19-1.signature dir2/20grob-2.signature')
1057     system ('cp 19multipage.midi dir1/midi-differ.midi')
1058     system ('cp 20multipage.midi dir2/midi-differ.midi')
1059     system ('cp 19multipage.log dir1/log-differ.log')
1060     system ('cp 19multipage.log dir2/log-differ.log &&  echo different >> dir2/log-differ.log &&  echo different >> dir2/log-differ.log')
1061
1062     compare_trees ('dir1', 'dir2', 'compare-dir1dir2', options.threshold)
1063
1064
1065 def test_basic_compare ():
1066     ly_template = r"""
1067
1068 \version "2.10.0"
1069 #(define default-toplevel-book-handler
1070   print-book-with-defaults-as-systems )
1071
1072 #(ly:set-option (quote no-point-and-click))
1073
1074 \sourcefilename "my-source.ly"
1075
1076 %(papermod)s
1077 \header { tagline = ##f }
1078 \score {
1079 <<
1080 \new Staff \relative c {
1081   c4^"%(userstring)s" %(extragrob)s
1082   }
1083 \new Staff \relative c {
1084   c4^"%(userstring)s" %(extragrob)s
1085   }
1086 >>
1087 \layout{}
1088 }
1089
1090 """
1091
1092     dicts = [{ 'papermod' : '',
1093                'name' : '20',
1094                'extragrob': '',
1095                'userstring': 'test' },
1096              { 'papermod' : '#(set-global-staff-size 19.5)',
1097                'name' : '19',
1098                'extragrob': '',
1099                'userstring': 'test' },
1100              { 'papermod' : '',
1101                'name' : '20expr',
1102                'extragrob': '',
1103                'userstring': 'blabla' },
1104              { 'papermod' : '',
1105                'name' : '20grob',
1106                'extragrob': 'r2. \\break c1',
1107                'userstring': 'test' },
1108              ]
1109
1110     for d in dicts:
1111         open (d['name'] + '.ly','w').write (ly_template % d)
1112
1113     names = [d['name'] for d in dicts]
1114
1115     system ('lilypond -ddump-profile -dseparate-log-files -ddump-signatures --png -dbackend=eps ' + ' '.join (names))
1116
1117
1118     multipage_str = r'''
1119     #(set-default-paper-size "a6")
1120     \score {
1121       \relative {c1 \pageBreak c1 }
1122       \layout {}
1123       \midi {}
1124     }
1125     '''
1126
1127     open ('20multipage.ly', 'w').write (multipage_str.replace ('c1', 'd1'))
1128     open ('19multipage.ly', 'w').write ('#(set-global-staff-size 19.5)\n' + multipage_str)
1129     system ('lilypond -dseparate-log-files -ddump-signatures --png 19multipage 20multipage ')
1130
1131     test_compare_signatures (names)
1132
1133 def test_compare_signatures (names, timing=False):
1134     import time
1135
1136     times = 1
1137     if timing:
1138         times = 100
1139
1140     t0 = time.clock ()
1141
1142     count = 0
1143     for t in range (0, times):
1144         sigs = dict ((n, read_signature_file ('%s-1.signature' % n)) for n in names)
1145         count += 1
1146
1147     if timing:
1148         print 'elapsed', (time.clock() - t0)/count
1149
1150
1151     t0 = time.clock ()
1152     count = 0
1153     combinations = {}
1154     for (n1, s1) in sigs.items():
1155         for (n2, s2) in sigs.items():
1156             combinations['%s-%s' % (n1, n2)] = SystemLink (s1,s2).distance ()
1157             count += 1
1158
1159     if timing:
1160         print 'elapsed', (time.clock() - t0)/count
1161
1162     results = combinations.items ()
1163     results.sort ()
1164     for k,v in results:
1165         print '%-20s' % k, v
1166
1167     assert combinations['20-20'] == (0.0,0.0,0.0)
1168     assert combinations['20-20expr'][0] > 0.0
1169     assert combinations['20-19'][2] < 10.0
1170     assert combinations['20-19'][2] > 0.0
1171
1172
1173 def run_tests ():
1174     dir = 'test-output-distance'
1175
1176     do_clean = not os.path.exists (dir)
1177
1178     print 'test results in ', dir
1179     if do_clean:
1180         system ('rm -rf ' + dir)
1181         system ('mkdir ' + dir)
1182
1183     os.chdir (dir)
1184     if do_clean:
1185         test_basic_compare ()
1186
1187     test_compare_trees ()
1188
1189 ################################################################
1190 #
1191
1192 def main ():
1193     p = optparse.OptionParser ("output-distance - compare LilyPond formatting runs")
1194     p.usage = 'output-distance.py [options] tree1 tree2'
1195
1196     p.add_option ('', '--test-self',
1197                   dest="run_test",
1198                   action="store_true",
1199                   help='run test method')
1200
1201     p.add_option ('--max-count',
1202                   dest="max_count",
1203                   metavar="COUNT",
1204                   type="int",
1205                   default=0,
1206                   action="store",
1207                   help='only analyze COUNT signature pairs')
1208
1209     p.add_option ('', '--threshold',
1210                   dest="threshold",
1211                   default=0.3,
1212                   action="store",
1213                   type="float",
1214                   help='threshold for geometric distance')
1215
1216     p.add_option ('--no-compare-images',
1217                   dest="compare_images",
1218                   default=True,
1219                   action="store_false",
1220                   help="Don't run graphical comparisons")
1221
1222     p.add_option ('--create-images',
1223                   dest="create_images",
1224                   default=False,
1225                   action="store_true",
1226                   help="Create PNGs from EPSes")
1227
1228
1229     p.add_option ('--local-datadir',
1230                   dest="local_data_dir",
1231                   default=False,
1232                   action="store_true",
1233                   help='whether to use the share/lilypond/ directory in the test directory')
1234
1235     p.add_option ('-o', '--output-dir',
1236                   dest="output_dir",
1237                   default=None,
1238                   action="store",
1239                   type="string",
1240                   help='where to put the test results [tree2/compare-tree1tree2]')
1241
1242     global options
1243     (options, args) = p.parse_args ()
1244
1245     if options.run_test:
1246         run_tests ()
1247         sys.exit (0)
1248
1249     if len (args) != 2:
1250         p.print_usage()
1251         sys.exit (2)
1252
1253     name = options.output_dir
1254     if not name:
1255         name = args[0].replace ('/', '')
1256         name = os.path.join (args[1], 'compare-' + shorten_string (name))
1257
1258     compare_trees (args[0], args[1], name, options.threshold)
1259
1260 if __name__ == '__main__':
1261     main()
1262