pdf-comparison Namespace Reference

Functions
	test_exe (cmd)
	print_cmd (cmd, o)
	check_call (cmd, **kwargs)
	check_output (cmd, **kwargs)
	remove (*paths)
	timeout (deadline, cmd)
	is_same (path1, path2)
	getfilesoftype (directory, ending)
	get_common_paths (dirs, ext)
	printable_path (d)
	spawn (cmd)
	sysopen (arg)
	shard (fn, arglist)
	shardsum (fn, arglist)
	checkout_worktree (checkoutable)
	build_skia (directory, executable)
	build_and_run_dm (directory, data_dir)
	rasterize (path)
	main (control_commitish)

Variables
	EXTRA_GN_ARGS = os.environ.get('PDF_COMPARISON_GN_ARGS', '')
str	REFERENCE_BACKEND = 'gl' if 'PDF_COMPARISON_NOGPU' not in os.environ else '8888'
	DPI = float(os.environ.get('PDF_COMPARISON_DPI', 72))
str	PDF_CONFIG = 'pdf' if 'PDF_COMPARISON_300DPI' not in os.environ else 'pdf300'
list	BAD_TESTS
str	NINJA = 'ninja'
str	PDFIUM_TEST = 'pdfium_test'
	NUM_THREADS = int(os.environ.get('PDF_COMPARISON_THREADS', 40))
list	SOURCES = ['gm']
str	HTML_HEAD
str	HTML_TAIL
tuple	USAGE

Function Documentation

build_and_run_dm()

pdf-comparison.build_and_run_dm	(		directory,
			data_dir
	)

Definition at line 227 of file pdf-comparison.py.

def build_and_run_dm(directory, data_dir):
  dm = build_skia(directory, 'dm')
  for source in SOURCES:
    os.makedirs(os.path.join(data_dir, PDF_CONFIG, source))
  dm_args = [dm, '--src'] + SOURCES + ['--config', PDF_CONFIG, '-w', data_dir]
  if BAD_TESTS:
    dm_args += ['-m'] + ['~^%s$' % x for x in BAD_TESTS]
  check_call(dm_args, cwd=directory)
  return dm
 

build_skia()

pdf-comparison.build_skia	(		directory,
			executable
	)

Definition at line 213 of file pdf-comparison.py.

def build_skia(directory, executable):
  args = ('--args=is_debug=false'
          ' extra_cflags=["-DSK_PDF_LESS_COMPRESSION",'
          ' "-DSK_PDF_BASE85_BINARY"] ')
  if test_exe('ccache'):
    args += ' cc_wrapper="ccache"'
  args += EXTRA_GN_ARGS
  build_dir = directory + '/out/pdftest'
  check_call([sys.executable, 'bin/sync'], cwd=directory)
  check_call([directory + '/bin/gn', 'gen', 'out/pdftest', args],
             cwd=directory)
  check_call([NINJA, executable], cwd=build_dir)
  return os.path.join(build_dir, executable)
 

check_call()

pdf-comparison.check_call	(		cmd,
		**	kwargs
	)

Definition at line 72 of file pdf-comparison.py.

def check_call(cmd, **kwargs):
  print_cmd(cmd, sys.stdout)
  return subprocess.check_call(cmd, **kwargs)
 

check_output()

pdf-comparison.check_output	(		cmd,
		**	kwargs
	)

Definition at line 76 of file pdf-comparison.py.

def check_output(cmd, **kwargs):
  print_cmd(cmd, sys.stdout)
  return subprocess.check_output(cmd, **kwargs)
 

checkout_worktree()

pdf-comparison.checkout_worktree

(

checkoutable

)

Definition at line 201 of file pdf-comparison.py.

def checkout_worktree(checkoutable):
  directory = os.path.join(tempfile.gettempdir(), 'skpdf_control_tree')
  commit = check_output(['git', 'rev-parse', checkoutable]).strip()
  if os.path.isdir(directory):
    try:
      check_call(['git', 'checkout', commit], cwd=directory)
      return directory
    except subprocess.CalledProcessError:
      shutil.rmtree(directory)
  check_call(['git', 'worktree', 'add', '-f', directory, commit])
  return directory
 

get_common_paths()

pdf-comparison.get_common_paths	(		dirs,
			ext
	)

Definition at line 114 of file pdf-comparison.py.

def get_common_paths(dirs, ext):
  return sorted(list(
    set.intersection(*(set(getfilesoftype(d, ext)) for d in dirs))))
 

getfilesoftype()

pdf-comparison.getfilesoftype	(		directory,
			ending
	)

Definition at line 107 of file pdf-comparison.py.

def getfilesoftype(directory, ending):
  for dirpath, _, filenames in os.walk(directory):
    rp = os.path.normpath(os.path.relpath(dirpath, directory))
    for f in filenames:
      if f.endswith(ending):
        yield os.path.join(rp, f)
 

is_same()

pdf-comparison.is_same	(		path1,
			path2
	)

Definition at line 94 of file pdf-comparison.py.

def is_same(path1, path2):
  if not os.path.isfile(path1) or not os.path.isfile(path2):
    return os.path.isfile(path1) == os.path.isfile(path2)
  with open(path1, 'rb') as f1:
    with open(path2, 'rb') as f2:
      while True:
        c1, c2 = f1.read(4096), f2.read(4096)
        if c1 != c2:
          return False
        if not c1:
          return True
 
 

main()

pdf-comparison.main

(

control_commitish

)

Definition at line 245 of file pdf-comparison.py.

def main(control_commitish):
  assert os.pardir == '..'  and '/' in [os.sep, os.altsep]
  assert test_exe(NINJA)
  assert test_exe(PDFIUM_TEST)
  os.chdir(os.path.dirname(__file__) + '/../..')
  control_worktree = checkout_worktree(control_commitish)
  tmpdir = tempfile.mkdtemp(prefix='skpdf_')
  exp = tmpdir + '/experim'
  con = tmpdir + '/control'
  build_and_run_dm(os.curdir, exp)
  dm = build_and_run_dm(control_worktree, con)
  image_diff_metric = build_skia(control_worktree, 'image_diff_metric')
 
  out = sys.stdout
  common_paths = get_common_paths([con, exp], '.pdf')
  out.write('\nNumber of PDFs: %d\n\n' % len(common_paths))
  def compare_identical(path):
    cpath, epath = (os.path.join(x, path) for x in (con, exp))
    if is_same(cpath, epath):
      remove(cpath, epath)
      return True
    return False
  identical_count = shardsum(compare_identical, common_paths)
  out.write('Number of identical PDFs: %d\n\n' % identical_count)
 
  differing_paths = get_common_paths([con, exp], '.pdf')
  if not differing_paths:
    out.write('All PDFs are the same!\n')
    sys.exit(0)
  out.write('Number of differing PDFs: %d\n' % len(differing_paths))
  for p in differing_paths:
    out.write('  %s\n' % printable_path(tmpdir + '/*/' + p))
  out.write('\n')
  shard(rasterize,
        [os.path.join(x, p) for p in differing_paths for x in [con, exp]])
 
  common_pngs = get_common_paths([con, exp], '.pdf.0.png')
  identical_count = shardsum(compare_identical, common_pngs)
  out.write('Number of PDFs that rasterize the same: %d\n\n'
            % identical_count)
 
  differing_pngs = get_common_paths([con, exp], '.pdf.0.png')
  if not differing_pngs:
    out.write('All PDFs rasterize the same!\n')
    sys.exit(0)
  out.write('Number of PDFs that rasterize differently: %d\n'
            % len(differing_pngs))
  for p in differing_pngs:
    out.write('  %s\n' % printable_path(tmpdir + '/*/' + p))
  out.write('\n')
 
  scores = dict()
  def compare_differing_pngs(path):
    cpath, epath = (os.path.join(x, path) for x in (con, exp))
    s = float(subprocess.check_output([image_diff_metric, cpath, epath]))
    indicator = '.' if s < 0.001 else ':' if s < 0.01 else '!'
    sys.stdout.write(indicator)
    sys.stdout.flush()
    scores[path] = s
  shard(compare_differing_pngs, differing_pngs)
  paths = sorted(scores.iterkeys(), key=lambda p: -scores[p])
  out.write('\n\n')
  for p in paths:
    pdfpath = printable_path(tmpdir + '/*/' + p.replace('.0.png', ''))
    out.write('  %6.4f  %s\n' % (scores[p], pdfpath))
  out.write('\n')
 
  errors = []
  rc = re.compile('^' + PDF_CONFIG + r'/([^/]*)/([^/]*)\.pdf\.0\.png$')
  for p in paths:
    m = rc.match(p)
    assert(m)
    source, name = m.groups()
    errors.append((source, name, scores[p]))
 
  for source in SOURCES:
    os.makedirs(os.path.join(con, REFERENCE_BACKEND, source))
  dm_args = [dm, '--src'] + SOURCES + [
             '--config', REFERENCE_BACKEND, '-w', con, '-m'] + [
             '^%s$' % name for _, name, _ in errors]
  check_call(dm_args, cwd=control_worktree)
 
  report = tmpdir + '/report.html'
  with open(report, 'w') as o:
    o.write(HTML_HEAD)
    o.write('c="%s/";\n' % os.path.relpath(con, tmpdir))
    o.write('e="%s/";\n' % os.path.relpath(exp, tmpdir))
    o.write('z=[\n')
    for source, name, score in errors:
      gt = REFERENCE_BACKEND + '/' + source + '/' + name + '.png'
      p = '%s/%s/%s.pdf.0.png' % (PDF_CONFIG, source, name)
      desc = '%s | %s | %g' % (source, name, score)
      o.write('["%s","%s","%s"],\n' % (p, gt, desc))
    o.write(HTML_TAIL)
  out.write(printable_path(report) + '\n')
  sysopen(report)
 

print_cmd()

pdf-comparison.print_cmd	(		cmd,
			o
	)

Definition at line 61 of file pdf-comparison.py.

def print_cmd(cmd, o):
  m = re.compile('[^A-Za-z0-9_./-]')
  o.write('+ ')
  for c in cmd:
    if m.search(c) is not None:
      o.write(repr(c) + ' ')
    else:
      o.write(c + ' ')
  o.write('\n')
  o.flush()
 

printable_path()

pdf-comparison.printable_path

(

d

)

Definition at line 118 of file pdf-comparison.py.

def printable_path(d):
  if 'TMPDIR' in os.environ:
    return d.replace(os.path.normpath(os.environ['TMPDIR']) + '/', '$TMPDIR/')
  return d
 

rasterize()

pdf-comparison.rasterize

(

path

)

Definition at line 237 of file pdf-comparison.py.

def rasterize(path):
  ret = timeout(30, [PDFIUM_TEST, '--png', '--scale=%g' % (DPI / 72.0), path])
  if ret != 0:
    sys.stdout.write(
      '\nTIMEOUT OR ERROR [%d] "%s"\n' % (ret, printable_path(path)))
    return
  assert os.path.isfile(path + '.0.png')
 

remove()

pdf-comparison.remove

(

*

paths

)

Definition at line 80 of file pdf-comparison.py.

def remove(*paths):
  for path in paths:
    os.remove(path)
 

shard()

pdf-comparison.shard	(		fn,
			arglist
	)

Definition at line 181 of file pdf-comparison.py.

def shard(fn, arglist):
  jobs = [[arg for j, arg in enumerate(arglist) if j % NUM_THREADS == i]
          for i in range(NUM_THREADS)]
  results = []
  def do_shard(*args):
    for arg in args:
      results.append(fn(arg))
  thread_list = []
  for job in jobs:
    t = threading.Thread(target=do_shard, args=job)
    t.start()
    thread_list += [t]
  for t in thread_list:
    t.join()
  return results
 

shardsum()

pdf-comparison.shardsum	(		fn,
			arglist
	)

Definition at line 197 of file pdf-comparison.py.

def shardsum(fn, arglist):
  'return the number of True results returned by fn(arg) for arg in arglist.'
  return sum(1 for result in shard(fn, arglist) if result)
 

spawn()

pdf-comparison.spawn

(

cmd

)

Definition at line 123 of file pdf-comparison.py.

def spawn(cmd):
  with open(os.devnull, 'w') as o:
    subprocess.Popen(cmd, stdout=o, stderr=o)
 

sysopen()

pdf-comparison.sysopen

(

arg

)

Definition at line 127 of file pdf-comparison.py.

def sysopen(arg):
  plat = sys.platform
  if plat.startswith('darwin'):
    spawn(["open", arg])
  elif plat.startswith('win'):
    # pylint: disable=no-member
    os.startfile(arg)
  else:
    spawn(["xdg-open", arg])
 

test_exe()

pdf-comparison.test_exe

(

cmd

)

Definition at line 53 of file pdf-comparison.py.

def test_exe(cmd):
  with open(os.devnull, 'w') as o:
    try:
      subprocess.call([cmd], stdout=o, stderr=o)
    except OSError:
      return False
  return True
 

timeout()

pdf-comparison.timeout	(		deadline,
			cmd
	)

Definition at line 84 of file pdf-comparison.py.

def timeout(deadline, cmd):
  #print_cmd(cmd, sys.stdout)
  with open(os.devnull, 'w') as o:
    proc = subprocess.Popen(cmd, stdout=o, stderr=subprocess.STDOUT)
    timer = threading.Timer(deadline, proc.terminate)
    timer.start()
    proc.wait()
    timer.cancel()
    return proc.returncode
 

Variable Documentation

BAD_TESTS

list pdf-comparison.BAD_TESTS

Initial value:

=  [
  'image-cacherator-from-picture',
  'image-cacherator-from-raster',
  'mixershader',
  'shadermaskfilter_image',
  'tilemode_decal',
]

Definition at line 37 of file pdf-comparison.py.

DPI

pdf-comparison.DPI = float(os.environ.get('PDF_COMPARISON_DPI', 72))

Definition at line 33 of file pdf-comparison.py.

EXTRA_GN_ARGS

pdf-comparison.EXTRA_GN_ARGS = os.environ.get('PDF_COMPARISON_GN_ARGS', '')

Definition at line 29 of file pdf-comparison.py.

HTML_HEAD

str pdf-comparison.HTML_HEAD

Definition at line 137 of file pdf-comparison.py.

HTML_TAIL

str pdf-comparison.HTML_TAIL

Initial value:

=  '''];
for(i=0;i<z.length;i++){
r(c+z[i][0],e+z[i][0],z[i][2],c+z[i][1]);}},false);
</script></head><body><table id="t">
<tr><th>BEFORE-AFTER DIFF</th>
<th>BEFORE</th><th>AFTER</th>
<th>REFERENCE</th></tr>
</table></body></html>'''

Definition at line 172 of file pdf-comparison.py.

NINJA

str pdf-comparison.NINJA = 'ninja'

Definition at line 45 of file pdf-comparison.py.

NUM_THREADS

pdf-comparison.NUM_THREADS = int(os.environ.get('PDF_COMPARISON_THREADS', 40))

Definition at line 49 of file pdf-comparison.py.

PDF_CONFIG

str pdf-comparison.PDF_CONFIG = 'pdf' if 'PDF_COMPARISON_300DPI' not in os.environ else 'pdf300'

Definition at line 35 of file pdf-comparison.py.

PDFIUM_TEST

str pdf-comparison.PDFIUM_TEST = 'pdfium_test'

Definition at line 47 of file pdf-comparison.py.

REFERENCE_BACKEND

str pdf-comparison.REFERENCE_BACKEND = 'gl' if 'PDF_COMPARISON_NOGPU' not in os.environ else '8888'

Definition at line 31 of file pdf-comparison.py.

SOURCES

list pdf-comparison.SOURCES = ['gm']

Definition at line 51 of file pdf-comparison.py.

USAGE

tuple pdf-comparison.USAGE

Initial value:

=  ('\nusage:\n  {0} COMMIT_OR_BRANCH_TO_COMPARE_TO\n\n'
             'e.g.:\n  {0} HEAD\nor\n  {0} HEAD~1\n\n')

Definition at line 344 of file pdf-comparison.py.