Gold support in PDFium

testing/tools/gold.py

+# Copyright 2015 The PDFium Authors. All rights reserved.
+# Use of this source code is governed by a BSD-style license that can be
+# found in the LICENSE file.
+

[borenet, 2016/12/21 19:47:28]

Style nit: 2 spaces between anything at the top level.

[stephana, 2016/12/21 21:47:57]

Done.

+import json
+import os
+import shutil
+
+# This module collects and writes output in a format expected by the
+# Gold baseline tool. Based on meta data provided explicitly and by
+# adding a series of test results it can be used to produce
+# a JSON file that is uploaded to Google Storage and ingested by Gold.
+#
+# The output will look similar this:
+#
+# {
+#    "build_number" : "2",
+#    "gitHash" : "a4a338179013b029d6dd55e737b5bd648a9fb68c",
+#    "key" : {
+#       "arch" : "arm64",
+#       "compiler" : "Clang",
+#    },
+#    "results" : [
+#       {
+#          "key" : {
+#             "config" : "vk",
+#             "name" : "yuv_nv12_to_rgb_effect",
+#             "source_type" : "gm"
+#          },
+#          "md5" : "7db34da246868d50ab9ddd776ce6d779",
+#          "options" : {
+#             "ext" : "png",
+#             "gamma_correct" : "no"
+#          }
+#       },
+#       {
+#          "key" : {
+#             "config" : "vk",
+#             "name" : "yuv_to_rgb_effect",
+#             "source_type" : "gm"
+#          },
+#          "md5" : "0b955f387740c66eb23bf0e253c80d64",
+#          "options" : {
+#             "ext" : "png",
+#             "gamma_correct" : "no"
+#          }
+#       }
+#    ],
+# }
+#
+class GoldResults(object):
+  def __init__(self, source_type, outputDir, propertiesStr, keyStr):
+    """
+    source_type is the source_type (=corpus) field used for all results.
+    output_dir is the directory where the resulting images are copied and
+               the dm.json file is written.
+    propertiesStr is a string with space separated key/value pairs that
+               is used to set the top level fields in the output JSON file.
+    keyStr is a string with space separated key/value pairs that
+               is used to set the 'key' field in the output JSON file.
+    """
+    self._source_type = source_type
+    self._properties = self._parseKeyValuePairs(propertiesStr)
+    self._properties["key"] = self._parseKeyValuePairs(keyStr)
+    self._results =  []
+    self._outputDir = outputDir
+
+  def AddTestResult(self, testName, md5Hash, outputImagePath):
+    # Copy the image to <output_dir>/<md5Hash>.<image_extension>
+    imgExt = os.path.splitext(outputImagePath)[1].lstrip(".")
+    if not imgExt:
+      raise ValueError("File %s does not have an extension" % outputImagePath)
+    newFilePath = os.path.join(self._outputDir, md5Hash + '.' + imgExt)
+    shutil.copy2(outputImagePath, newFilePath)
+
+    # Add an entry to the list of test results
+    self._results.append({
+      "key": {
+        "name": testName,
+        "source_type": self._source_type,
+      },
+      "md5": md5Hash,
+      "options": {
+        "ext": imgExt,
+        "gamma_correct": "no"
+      }
+    })
+
+  def _parseKeyValuePairs(self, kvStr):
+    kvPairs = kvStr.split()

[borenet, 2016/12/21 19:47:28]

I recommend shlex.split, if kvPairs might include quoted strings with spaces in
them.

[stephana, 2016/12/21 21:47:57]

Done.

+    if len(kvPairs) % 2:
+      raise ValueError("Uneven number of key/value pairs. Got %s" % kvStr)
+    return { kvPairs[i]:kvPairs[i+1] for i in range(0, len(kvPairs), 2) }
+
+  def WriteResults(self):
+    self._properties.update({
+      "results": self._results
+    })
+
+    outputFileName = os.path.join(self._outputDir, "dm.json")
+    with open(outputFileName, 'wb') as outfile:
+      json.dump(self._properties, outfile, indent=1)
+      outfile.write("\n")
+
+# Produce example output for manual testing.
+if __name__ == "__main__":

[borenet, 2016/12/21 19:47:28]

I prefer to put the below lines into a function since any variables declared in
this block become file-global.

[stephana, 2016/12/21 21:47:57]

I don't think so. If I run this from another file:

import gold 
gold.testDir 

The second line will cause an error.
I might misunderstand what you mean though.  

+  # Create a test directory with three empty 'image' files.
+  testDir = "./testdirectory"
+  if not os.path.exists(testDir):
+    os.makedirs(testDir)
+  open(os.path.join(testDir, "image1.png"), 'wb').close()
+  open(os.path.join(testDir, "image2.png"), 'wb').close()
+  open(os.path.join(testDir, "image3.png"), 'wb').close()
+
+  # Create an instance and add results.
+  propStr = "build_number 2 builder Builder-Name gitHash a4a338179013b029d6dd55e737b5bd648a9fb68c"

[borenet, 2016/12/21 19:47:28]

Nit: line break

[stephana, 2016/12/21 21:47:57]

Done.

+  keyStr = "arch arm64 compiler Clang configuration Debug"
+
+  gr = GoldResults("pdfium", testDir, propStr, keyStr)
+  gr.AddTestResult("test-1", "hash-1", os.path.join(testDir, "image1.png"))
+  gr.AddTestResult("test-2", "hash-2", os.path.join(testDir, "image2.png"))
+  gr.AddTestResult("test-3", "hash-3", os.path.join(testDir, "image3.png"))
+  gr.WriteResults()