Add --regenerate_expected option to test_runner.py.chromium/3250

After a change in rendering, corpus and pixel test expected pngs
need to be regenerated manually. This option reduces the toil by
automating renaming and moving the new expected files over the old
ones.

Change-Id: I9d490369ccf946d4d4567e440b7b5252469eec46
Reviewed-on: https://pdfium-review.googlesource.com/16451
Commit-Queue: Henrique Nakashima <hnakashima@chromium.org>
Reviewed-by: Nicolás Peña Moreno <npm@chromium.org>

2 files changed, 83 insertions, 41 deletions

diff --git a/testing/tools/pngdiffer.py b/testing/tools/pngdiffer.py
index a1533b8329..95acb198a6 100755
--- a/testing/tools/pngdiffer.py
+++ b/testing/tools/pngdiffer.py
@@ -4,32 +4,26 @@
 # found in the LICENSE file.
 
 import os
+import shutil
 import sys
 
 import common
 
 class PNGDiffer():
-  ACTUAL_TEMPLATE = '.pdf.%d.png'
-  EXPECTED_TEMPLATE = '_expected' + ACTUAL_TEMPLATE
-  PLATFORM_EXPECTED_TEMPLATE = '_expected_%s' + ACTUAL_TEMPLATE
-
   def __init__(self, finder):
     self.pdfium_diff_path = finder.ExecutablePath('pdfium_diff')
     self.os_name = finder.os_name
 
   def GetActualFiles(self, input_filename, source_dir, working_dir):
     actual_paths = []
-    template_paths = self._GetTemplatePaths(
-        input_filename, source_dir, working_dir)
-    actual_path_template = template_paths[0];
-    expected_path_template = template_paths[1]
-    platform_expected_path_template = template_paths[2]
+    path_templates = PathTemplates(input_filename, source_dir, working_dir)
+
     i = 0
     while True:
-      actual_path = actual_path_template % i
-      expected_path = expected_path_template % i
-      platform_expected_path = (
-          platform_expected_path_template % (self.os_name, i))
+      actual_path = path_templates.GetActualPath(i)
+      expected_path = path_templates.GetExpectedPath(i)
+      platform_expected_path = path_templates.GetPlatformExpectedPath(
+          self.os_name, i)
       if os.path.exists(platform_expected_path):
         expected_path = platform_expected_path
       elif not os.path.exists(expected_path):
@@ -39,19 +33,16 @@ class PNGDiffer():
     return actual_paths
 
   def HasDifferences(self, input_filename, source_dir, working_dir):
-    template_paths = self._GetTemplatePaths(
-        input_filename, source_dir, working_dir)
-    actual_path_template = template_paths[0];
-    expected_path_template = template_paths[1]
-    platform_expected_path_template = template_paths[2]
+    path_templates = PathTemplates(input_filename, source_dir, working_dir)
+
     i = 0
     while True:
-      actual_path = actual_path_template % i
-      expected_path = expected_path_template % i
+      actual_path = path_templates.GetActualPath(i)
+      expected_path = path_templates.GetExpectedPath(i)
       # PDFium tests should be platform independent. Platform based results are
       # used to capture platform dependent implementations.
-      platform_expected_path = (
-          platform_expected_path_template % (self.os_name, i))
+      platform_expected_path = path_templates.GetPlatformExpectedPath(
+          self.os_name, i)
       if (not os.path.exists(expected_path) and
           not os.path.exists(platform_expected_path)):
         if i == 0:
@@ -75,11 +66,54 @@ class PNGDiffer():
       i += 1
     return False
 
-  def _GetTemplatePaths(self, input_filename, source_dir, working_dir):
+  def Regenerate(self, input_filename, source_dir, working_dir):
+    path_templates = PathTemplates(input_filename, source_dir, working_dir)
+
+    page = 0
+    while True:
+      # Loop through the generated page images. Stop when there is a page
+      # missing a png, which means the document ended.
+      actual_path = path_templates.GetActualPath(page)
+      if not os.path.isfile(actual_path):
+        break
+
+      platform_expected_path = path_templates.GetPlatformExpectedPath(
+          self.os_name, page)
+
+      # If there is a platform expected png, we will overwrite it. Otherwise,
+      # overwrite the generic png.
+      if os.path.exists(platform_expected_path):
+        expected_path = platform_expected_path
+      else:
+        expected_path = path_templates.GetExpectedPath(page)
+
+      if os.path.exists(expected_path):
+        shutil.copyfile(actual_path, expected_path)
+
+      page += 1
+
+
+ACTUAL_TEMPLATE = '.pdf.%d.png'
+EXPECTED_TEMPLATE = '_expected' + ACTUAL_TEMPLATE
+PLATFORM_EXPECTED_TEMPLATE = '_expected_%s' + ACTUAL_TEMPLATE
+
+
+class PathTemplates(object):
+
+  def __init__(self, input_filename, source_dir, working_dir):
     input_root, _ = os.path.splitext(input_filename)
-    actual_path = os.path.join(working_dir, input_root + self.ACTUAL_TEMPLATE)
-    expected_path = os.path.join(
-        source_dir, input_root + self.EXPECTED_TEMPLATE)
-    platform_expected_path = os.path.join(
-        source_dir, input_root + self.PLATFORM_EXPECTED_TEMPLATE)
-    return (actual_path, expected_path, platform_expected_path)
+    self.actual_path_template = os.path.join(working_dir,
+                                             input_root + ACTUAL_TEMPLATE)
+    self.expected_path = os.path.join(
+        source_dir, input_root + EXPECTED_TEMPLATE)
+    self.platform_expected_path = os.path.join(
+        source_dir, input_root + PLATFORM_EXPECTED_TEMPLATE)
+
+  def GetActualPath(self, page):
+    return self.actual_path_template % page
+
+  def GetExpectedPath(self, page):
+    return self.expected_path % page
+
+  def GetPlatformExpectedPath(self, platform, page):
+    return self.platform_expected_path % (platform, page)
diff --git a/testing/tools/test_runner.py b/testing/tools/test_runner.py
index 86e05636ab..ea8f2b8eaf 100644
--- a/testing/tools/test_runner.py
+++ b/testing/tools/test_runner.py
@@ -81,6 +81,11 @@ class TestRunner:
     if actual_images:
       if self.image_differ.HasDifferences(input_filename, source_dir,
                                           self.working_dir):
+        if (self.options.regenerate_expected
+            and not self.test_suppressor.IsResultSuppressed(input_filename)
+            and not self.test_suppressor.IsImageDiffSuppressed(input_filename)):
+          self.image_differ.Regenerate(input_filename, source_dir,
+                                       self.working_dir)
         return False, results
     else:
       if (self.enforce_expected_images
@@ -171,12 +176,15 @@ class TestRunner:
     parser.add_option('--gold_ignore_hashes', default='', dest="gold_ignore_hashes",
                       help='Path to a file with MD5 hashes we wish to ignore.')
 
+    parser.add_option('--regenerate_expected', action="store_true", dest="regenerate_expected",
+                      help='Regenerates expected images.')
+
     parser.add_option('--ignore_errors', action="store_true", dest="ignore_errors",
                       help='Prevents the return value from being non-zero when image comparison fails.')
 
-    options, args = parser.parse_args()
+    self.options, self.args = parser.parse_args()
 
-    finder = common.DirectoryFinder(options.build_dir)
+    finder = common.DirectoryFinder(self.options.build_dir)
     self.fixup_path = finder.ScriptPath('fixup_pdf_template.py')
     self.text_diff_path = finder.ScriptPath('text_diff.py')
 
@@ -206,8 +214,8 @@ class TestRunner:
     self.test_cases = []
     self.execution_suppressed_cases = []
     input_file_re = re.compile('^.+[.](in|pdf)$')
-    if args:
-      for file_name in args:
+    if self.args:
+      for file_name in self.args:
         file_name.replace('.pdf', '.in')
         input_path = os.path.join(walk_from_dir, file_name)
         if not os.path.isfile(input_path):
@@ -233,16 +241,16 @@ class TestRunner:
 
     # Collect Gold results if an output directory was named.
     self.gold_results = None
-    if options.gold_output_dir:
+    if self.options.gold_output_dir:
       self.gold_results = gold.GoldResults('pdfium',
-                                           options.gold_output_dir,
-                                           options.gold_properties,
-                                           options.gold_key,
-                                           options.gold_ignore_hashes)
+                                           self.options.gold_output_dir,
+                                           self.options.gold_properties,
+                                           self.options.gold_key,
+                                           self.options.gold_ignore_hashes)
 
-    if options.num_workers > 1 and len(self.test_cases) > 1:
+    if self.options.num_workers > 1 and len(self.test_cases) > 1:
       try:
-        pool = multiprocessing.Pool(options.num_workers)
+        pool = multiprocessing.Pool(self.options.num_workers)
         worker_func = functools.partial(TestOneFileParallel, self)
 
         worker_results = pool.imap(worker_func, self.test_cases)
@@ -282,7 +290,7 @@ class TestRunner:
     self._PrintSummary()
 
     if self.failures:
-      if not options.ignore_errors:
+      if not self.options.ignore_errors:
         return 1
 
     return 0