kumarashishg | 826308d | 2023-06-23 13:21:22 +0000 | [diff] [blame] | 1 | # Copyright 2017 The PDFium Authors |
Haibo Huang | 49cc930 | 2020-04-27 16:14:24 -0700 | [diff] [blame] | 2 | # Use of this source code is governed by a BSD-style license that can be |
| 3 | # found in the LICENSE file. |
| 4 | """Compares pairs of page images and generates an HTML to look at differences. |
| 5 | """ |
| 6 | |
| 7 | import functools |
| 8 | import glob |
| 9 | import multiprocessing |
| 10 | import os |
| 11 | import re |
| 12 | import subprocess |
| 13 | import sys |
| 14 | import webbrowser |
| 15 | |
Haibo Huang | 49cc930 | 2020-04-27 16:14:24 -0700 | [diff] [blame] | 16 | from common import DirectoryFinder |
| 17 | |
| 18 | |
| 19 | def GenerateOneDiffParallel(image_comparison, image): |
| 20 | return image_comparison.GenerateOneDiff(image) |
| 21 | |
| 22 | |
kumarashishg | 826308d | 2023-06-23 13:21:22 +0000 | [diff] [blame] | 23 | class ImageComparison: |
Haibo Huang | 49cc930 | 2020-04-27 16:14:24 -0700 | [diff] [blame] | 24 | """Compares pairs of page images and generates an HTML to look at differences. |
| 25 | |
| 26 | The images are all assumed to have the same name and be in two directories: |
| 27 | [output_path]/[two_labels[0]] and [output_path]/[two_labels[1]]. For example, |
| 28 | if output_path is "/tmp/images" and two_labels is ("before", "after"), |
| 29 | images in /tmp/images/before will be compared to /tmp/images/after. The HTML |
| 30 | produced will be in /tmp/images/compare.html and have relative links to these |
| 31 | images, so /tmp/images is self-contained and can be moved around or shared. |
| 32 | """ |
| 33 | |
| 34 | def __init__(self, build_dir, output_path, two_labels, num_workers, |
| 35 | threshold_fraction): |
| 36 | """Constructor. |
| 37 | |
| 38 | Args: |
| 39 | build_dir: Path to the build directory. |
| 40 | output_path: Path with the pngs and where the html will be created. |
| 41 | two_labels: Tuple of two strings that name the subdirectories in |
| 42 | output_path containing the images. |
| 43 | num_workers: Number of worker threads to start. |
| 44 | threshold_fraction: Minimum percentage (0.0 to 1.0) of pixels below which |
| 45 | an image is considered to have only small changes. They will not be |
| 46 | displayed on the HTML, only listed. |
| 47 | """ |
| 48 | self.build_dir = build_dir |
| 49 | self.output_path = output_path |
| 50 | self.two_labels = two_labels |
| 51 | self.num_workers = num_workers |
| 52 | self.threshold = threshold_fraction * 100 |
| 53 | |
| 54 | def Run(self, open_in_browser): |
| 55 | """Runs the comparison and generates an HTML with the results. |
| 56 | |
| 57 | Returns: |
| 58 | Exit status. |
| 59 | """ |
| 60 | |
| 61 | # Running a test defines a number of attributes on the fly. |
| 62 | # pylint: disable=attribute-defined-outside-init |
| 63 | |
| 64 | if len(self.two_labels) != 2: |
kumarashishg | 826308d | 2023-06-23 13:21:22 +0000 | [diff] [blame] | 65 | print('two_labels must be a tuple of length 2', file=sys.stderr) |
Haibo Huang | 49cc930 | 2020-04-27 16:14:24 -0700 | [diff] [blame] | 66 | return 1 |
| 67 | |
| 68 | finder = DirectoryFinder(self.build_dir) |
| 69 | self.img_diff_bin = finder.ExecutablePath('pdfium_diff') |
| 70 | |
| 71 | html_path = os.path.join(self.output_path, 'compare.html') |
| 72 | |
| 73 | self.diff_path = os.path.join(self.output_path, 'diff') |
| 74 | if not os.path.exists(self.diff_path): |
| 75 | os.makedirs(self.diff_path) |
| 76 | |
| 77 | self.image_locations = ImageLocations(self.output_path, self.diff_path, |
| 78 | self.two_labels) |
| 79 | |
| 80 | difference = self._GenerateDiffs() |
| 81 | |
| 82 | small_changes = [] |
| 83 | |
| 84 | with open(html_path, 'w') as f: |
| 85 | f.write('<html><body>') |
| 86 | f.write('<table>') |
| 87 | for image in self.image_locations.Images(): |
| 88 | diff = difference[image] |
| 89 | if diff is None: |
kumarashishg | 826308d | 2023-06-23 13:21:22 +0000 | [diff] [blame] | 90 | print('Failed to compare image %s' % image, file=sys.stderr) |
Haibo Huang | 49cc930 | 2020-04-27 16:14:24 -0700 | [diff] [blame] | 91 | elif diff > self.threshold: |
| 92 | self._WriteImageRows(f, image, diff) |
| 93 | else: |
| 94 | small_changes.append((image, diff)) |
| 95 | self._WriteSmallChanges(f, small_changes) |
| 96 | f.write('</table>') |
| 97 | f.write('</body></html>') |
| 98 | |
| 99 | if open_in_browser: |
| 100 | webbrowser.open(html_path) |
| 101 | |
| 102 | return 0 |
| 103 | |
| 104 | def _GenerateDiffs(self): |
| 105 | """Runs a diff over all pairs of page images, producing diff images. |
| 106 | |
| 107 | As a side effect, the diff images will be saved to [output_path]/diff |
| 108 | with the same image name. |
| 109 | |
| 110 | Returns: |
| 111 | A dict mapping image names to percentage of pixels changes. |
| 112 | """ |
| 113 | difference = {} |
| 114 | pool = multiprocessing.Pool(self.num_workers) |
| 115 | worker_func = functools.partial(GenerateOneDiffParallel, self) |
| 116 | |
| 117 | try: |
| 118 | # The timeout is a workaround for http://bugs.python.org/issue8296 |
| 119 | # which prevents KeyboardInterrupt from working. |
| 120 | one_year_in_seconds = 3600 * 24 * 365 |
| 121 | worker_results = ( |
| 122 | pool.map_async( |
| 123 | worker_func, |
| 124 | self.image_locations.Images()).get(one_year_in_seconds)) |
| 125 | for worker_result in worker_results: |
| 126 | image, result = worker_result |
| 127 | difference[image] = result |
| 128 | except KeyboardInterrupt: |
| 129 | pool.terminate() |
| 130 | sys.exit(1) |
| 131 | else: |
| 132 | pool.close() |
| 133 | |
| 134 | pool.join() |
| 135 | |
| 136 | return difference |
| 137 | |
| 138 | def GenerateOneDiff(self, image): |
| 139 | """Runs a diff over one pair of images, producing a diff image. |
| 140 | |
| 141 | As a side effect, the diff image will be saved to [output_path]/diff |
| 142 | with the same image name. |
| 143 | |
| 144 | Args: |
| 145 | image: Page image to compare. |
| 146 | |
| 147 | Returns: |
| 148 | A tuple (image, diff), where image is the parameter and diff is the |
| 149 | percentage of pixels changed. |
| 150 | """ |
| 151 | try: |
| 152 | subprocess.check_output([ |
| 153 | self.img_diff_bin, |
| 154 | self.image_locations.Left(image), |
| 155 | self.image_locations.Right(image) |
| 156 | ]) |
| 157 | except subprocess.CalledProcessError as e: |
| 158 | percentage_change = float(re.findall(r'\d+\.\d+', e.output)[0]) |
| 159 | else: |
| 160 | return image, 0 |
| 161 | |
| 162 | try: |
| 163 | subprocess.check_output([ |
| 164 | self.img_diff_bin, '--diff', |
| 165 | self.image_locations.Left(image), |
| 166 | self.image_locations.Right(image), |
| 167 | self.image_locations.Diff(image) |
| 168 | ]) |
| 169 | except subprocess.CalledProcessError as e: |
| 170 | return image, percentage_change |
| 171 | else: |
kumarashishg | 826308d | 2023-06-23 13:21:22 +0000 | [diff] [blame] | 172 | print('Warning: Should have failed the previous diff.', file=sys.stderr) |
Haibo Huang | 49cc930 | 2020-04-27 16:14:24 -0700 | [diff] [blame] | 173 | return image, 0 |
| 174 | |
| 175 | def _GetRelativePath(self, absolute_path): |
| 176 | return os.path.relpath(absolute_path, start=self.output_path) |
| 177 | |
| 178 | def _WriteImageRows(self, f, image, diff): |
| 179 | """Write table rows for a page image comparing its two versions. |
| 180 | |
| 181 | Args: |
| 182 | f: Open HTML file to write to. |
| 183 | image: Image file name. |
| 184 | diff: Percentage of different pixels. |
| 185 | """ |
| 186 | f.write('<tr><td colspan="2">') |
| 187 | f.write('%s (%.4f%% changed)' % (image, diff)) |
| 188 | f.write('</td></tr>') |
| 189 | |
| 190 | f.write('<tr>') |
| 191 | self._WritePageCompareTd( |
| 192 | f, self._GetRelativePath(self.image_locations.Left(image)), |
| 193 | self._GetRelativePath(self.image_locations.Right(image))) |
| 194 | self._WritePageTd(f, self._GetRelativePath( |
| 195 | self.image_locations.Diff(image))) |
| 196 | f.write('</tr>') |
| 197 | |
| 198 | def _WritePageTd(self, f, image_path): |
| 199 | """Write table column with a single image. |
| 200 | |
| 201 | Args: |
| 202 | f: Open HTML file to write to. |
| 203 | image_path: Path to image file. |
| 204 | """ |
| 205 | f.write('<td>') |
| 206 | f.write('<img src="%s">' % image_path) |
| 207 | f.write('</td>') |
| 208 | |
| 209 | def _WritePageCompareTd(self, f, normal_image_path, hover_image_path): |
| 210 | """Write table column for an image comparing its two versions. |
| 211 | |
| 212 | Args: |
| 213 | f: Open HTML file to write to. |
| 214 | normal_image_path: Path to image to be used in the "normal" state. |
| 215 | hover_image_path: Path to image to be used in the "hover" state. |
| 216 | """ |
| 217 | f.write('<td>') |
| 218 | f.write('<img src="%s" ' |
| 219 | 'onmouseover="this.src=\'%s\';" ' |
| 220 | 'onmouseout="this.src=\'%s\';">' % |
| 221 | (normal_image_path, hover_image_path, normal_image_path)) |
| 222 | f.write('</td>') |
| 223 | |
| 224 | def _WriteSmallChanges(self, f, small_changes): |
| 225 | """Write table rows for all images considered to have only small changes. |
| 226 | |
| 227 | Args: |
| 228 | f: Open HTML file to write to. |
| 229 | small_changes: List of (image, change) tuples, where image is the page |
| 230 | image and change is the percentage of pixels changed. |
| 231 | """ |
| 232 | for image, change in small_changes: |
| 233 | f.write('<tr><td colspan="2">') |
| 234 | if not change: |
| 235 | f.write('No change for: %s' % image) |
| 236 | else: |
| 237 | f.write('Small change of %.4f%% for: %s' % (change, image)) |
| 238 | f.write('</td></tr>') |
| 239 | |
| 240 | |
kumarashishg | 826308d | 2023-06-23 13:21:22 +0000 | [diff] [blame] | 241 | class ImageLocations: |
Haibo Huang | 49cc930 | 2020-04-27 16:14:24 -0700 | [diff] [blame] | 242 | """Contains the locations of input and output image files. |
| 243 | """ |
| 244 | |
| 245 | def __init__(self, output_path, diff_path, two_labels): |
| 246 | """Constructor. |
| 247 | |
| 248 | Args: |
| 249 | output_path: Path to directory with the pngs. |
| 250 | diff_path: Path to directory where the diffs will be generated. |
| 251 | two_labels: Tuple of two strings that name the subdirectories in |
| 252 | output_path containing the images. |
| 253 | """ |
| 254 | self.output_path = output_path |
| 255 | self.diff_path = diff_path |
| 256 | self.two_labels = two_labels |
| 257 | |
| 258 | self.left = self._FindImages(self.two_labels[0]) |
| 259 | self.right = self._FindImages(self.two_labels[1]) |
| 260 | |
| 261 | self.images = list(self.left.viewkeys() & self.right.viewkeys()) |
| 262 | |
| 263 | # Sort by pdf filename, then page number |
| 264 | def KeyFn(s): |
| 265 | pieces = s.rsplit('.', 2) |
| 266 | return (pieces[0], int(pieces[1])) |
| 267 | |
| 268 | self.images.sort(key=KeyFn) |
| 269 | self.diff = { |
| 270 | image: os.path.join(self.diff_path, image) for image in self.images |
| 271 | } |
| 272 | |
| 273 | def _FindImages(self, label): |
| 274 | """Traverses a dir and builds a dict of all page images to compare in it. |
| 275 | |
| 276 | Args: |
| 277 | label: name of subdirectory of output_path to traverse. |
| 278 | |
| 279 | Returns: |
| 280 | Dict mapping page image names to the path of the image file. |
| 281 | """ |
| 282 | image_path_matcher = os.path.join(self.output_path, label, '*.*.png') |
| 283 | image_paths = glob.glob(image_path_matcher) |
| 284 | |
| 285 | image_dict = { |
| 286 | os.path.split(image_path)[1]: image_path for image_path in image_paths |
| 287 | } |
| 288 | |
| 289 | return image_dict |
| 290 | |
| 291 | def Images(self): |
| 292 | """Returns a list of all page images present in both directories.""" |
| 293 | return self.images |
| 294 | |
| 295 | def Left(self, test_case): |
| 296 | """Returns the path for a page image in the first subdirectory.""" |
| 297 | return self.left[test_case] |
| 298 | |
| 299 | def Right(self, test_case): |
| 300 | """Returns the path for a page image in the second subdirectory.""" |
| 301 | return self.right[test_case] |
| 302 | |
| 303 | def Diff(self, test_case): |
| 304 | """Returns the path for a page diff image.""" |
| 305 | return self.diff[test_case] |