improve test runner

share/manifest

@@ -31,9 +31,7 @@
 @base.min.css
 
 # fancy CSS styles - Optional
-#TEST_IGNORE_BEGIN
 @fancy.min.css
-#TEST_IGNORE_END
 
 # PDF specific CSS styles - Do not modify
 $css

test/README.md

@@ -0,0 +1,26 @@
+### Dependencies
+
+- wkhtmltoimage
+- python2
+- Python Imaging Library
+
+### Usage
+- Run all tests:
+  - `./test.py`
+- Run selected tests:
+  - `./test.py test_A test_B ...`
+- Environment variables:
+  - set `P2H_TEST_SAVE_TMP=1` to keep the temporary files
+  - set `P2H_TEST_GEN=1` to generate new reference images instead of comparing with old ones
+
+### Guidelines for test cases
+
+- Make sure you have the proper copyrights.
+- Using meaningful file names, a description of the file, or issueXXX.pdf.
+- Make each test case minimal:
+  - One page only, unless the test case is about multiple pages.
+  - Grayscale only, unless the test case is about colors.
+  - Remove unnecessary elements.
+  - Set proper parameters for cropping in `wkhtml2image_args`.
+- [Optional] Include the source files that the PDF file is generated from.
+

test/fancy.min.css

@@ -0,0 +1,4 @@
+/* CSS for test cases */
+#page-container {
+    overflow:hidden;
+}

test/test.py

@@ -13,8 +13,17 @@ class Common(object):
     DATA_DIR = os.path.join(SRC_DIR, 'share')
     PDF2HTMLEX_PATH = os.path.join(SRC_DIR, 'pdf2htmlEX')
 
+    SAVE_TMP = os.environ.get('P2H_TEST_SAVE_TMP')
+    CANONICAL_TEMPDIR = '/tmp/pdf2htmlEX_test'
+    
     def setUp(self):
+        if not self.SAVE_TMP:
             self.cur_temp_dir = tempfile.mkdtemp(prefix='pdf2htmlEX_test')
+        else:
+            shutil.rmtree(self.CANONICAL_TEMPDIR, True)
+            os.mkdir(self.CANONICAL_TEMPDIR)
+            self.cur_temp_dir = self.CANONICAL_TEMPDIR
+
         self.cur_data_dir = os.path.join(self.cur_temp_dir, 'share')
         self.cur_output_dir = os.path.join(self.cur_temp_dir, 'out')
         os.mkdir(self.cur_data_dir)
@@ -36,8 +45,11 @@ class Common(object):
         # copy files
         shutil.copy(os.path.join(self.DATA_DIR, 'base.min.css'),
                 os.path.join(self.cur_data_dir, 'base.min.css'))
+        shutil.copy(os.path.join(self.TEST_DIR, 'fancy.min.css'),
+                os.path.join(self.cur_data_dir, 'fancy.min.css'))
 
     def tearDown(self):
+        if not self.SAVE_TMP:
             shutil.rmtree(self.cur_temp_dir, True)
 
     def run_pdf2htmlEX(self, args):
@@ -49,19 +61,15 @@ class Common(object):
         :return: an object of relevant info
         """
 
-        cmd = [self.PDF2HTMLEX_PATH,
+        args = [self.PDF2HTMLEX_PATH,
                '--data-dir', self.cur_data_dir,
                '--dest-dir', self.cur_output_dir
-              ]
+              ] + args
 
-        for val in args:
+        return_code = subprocess.call(list(map(str, args)))
-            cmd.append(str(val))
+        self.assertEquals(return_code, 0, 'cannot execute pdf2htmlEX')
-
-        return_code = subprocess.call(cmd)
-        self.assertEquals(return_code, 0)
 
         files = os.listdir(self.cur_output_dir)
-        files.sort()
 
         return { 
             'return_code' : return_code,
@@ -75,10 +83,21 @@ if __name__ == '__main__':
         exit(1)
     suites = []
     loader = unittest.TestLoader()
-    for module_name in ['test_naming']:
+    test_names = list(map(lambda x: 'T.'+x, sys.argv[1:]))
+    for module_name in ['test_naming', 'test_conversion']:
         __import__(module_name)
+        if len(test_names) > 0:
+            try:
+                suites.append(loader.loadTestsFromNames(test_names, sys.modules[module_name]))
+            except:
+                pass
+        else:
             suites.append(loader.loadTestsFromModule(sys.modules[module_name]))
 
+    if len(suites) == 0:
+        print >>sys.stderr, 'No test found'
+        exit(1)
+
     failure_count = 0
     runner = unittest.TextTestRunner(verbosity=2)
     for suite in suites:

test/test_conversion.py

@@ -0,0 +1,98 @@
+#!/usr/bin/env python
+
+import unittest
+import os
+import subprocess
+
+from PIL import Image, ImageChops
+from test import Common
+
+class T(Common, unittest.TestCase):
+    GENERATING_MODE = os.environ.get('P2H_TEST_GEN')
+
+    WKHTML2IMAGE = 'wkhtmltoimage'
+    TTFAUTOHINT = 'ttfautohint'
+    TEST_DATA_DIR = os.path.join(Common.TEST_DIR, 'test_conversion')
+
+    DEFAULT_PDF2HTMLEX_ARGS = [
+        '--external-hint-tool', 'ttfautohint',
+        '--fit-width', 800,
+        '--last-page', 1,
+        '--correct-text-visibility', 1,
+    ]
+
+    DEFAULT_WKHTML2IMAGE_ARGS = [
+        '-f', 'png',
+        '--height', 600,
+        '--width', 800,
+        '--quality', 0,
+        '--quiet'
+    ]  
+
+    @classmethod
+    def setUpClass(cls):
+        subprocess.check_call([cls.WKHTML2IMAGE, '--version'])
+        subprocess.check_call([cls.TTFAUTOHINT, '--version'])
+
+    def run_test_case(self, filename, pdf2htmlEX_args=[], wkhtml2image_args=[]):
+        basefilename, extension = os.path.splitext(filename)
+        htmlfilename = basefilename + '.html'
+        pngfilename = basefilename + '.png'
+
+        self.assertEquals(extension.lower(), '.pdf', 'Input file is not PDF')
+
+        pdf2htmlEX_args = self.DEFAULT_PDF2HTMLEX_ARGS \
+            + list(pdf2htmlEX_args) + [
+                os.path.join(self.TEST_DATA_DIR, filename),
+                htmlfilename
+            ]
+        
+        result = self.run_pdf2htmlEX(pdf2htmlEX_args)
+        self.assertIn(htmlfilename, result['output_files'], 'HTML file is not generated')
+
+        png_out_dir = os.path.join(self.cur_temp_dir, 'png_out')
+        os.mkdir(png_out_dir)
+
+        pngfilename_out_fullpath = os.path.join(png_out_dir, pngfilename)
+        pngfilename_raw_fullpath = os.path.join(self.TEST_DATA_DIR, pngfilename)
+
+        wkhtml2image_args = [self.WKHTML2IMAGE] \
+            + self.DEFAULT_WKHTML2IMAGE_ARGS \
+            + list(wkhtml2image_args) + [
+                os.path.join(self.cur_output_dir, htmlfilename),
+                pngfilename_out_fullpath
+            ]
+
+        return_code = subprocess.call(list(map(str, wkhtml2image_args)))
+        self.assertEquals(return_code, 0, 'cannot execute ' + self.WKHTML2IMAGE)
+
+        if self.GENERATING_MODE:
+            shutil.copy(pngfilename_out_fullpath, pngfilename_raw_fullpath)
+        else:
+            original_img = Image.open(pngfilename_raw_fullpath)
+            new_img = Image.open(pngfilename_out_fullpath)
+
+            diff_img = ImageChops.difference(original_img, new_img);
+
+            if diff_img.getbbox() is not None:
+                if self.SAVE_TMP:
+                    # save the diff image
+                    # http://stackoverflow.com/questions/15721484/saving-in-png-using-pil-library-after-taking-imagechops-difference-of-two-png
+                    diff_img.convert('RGB').save(os.path.join(png_out_dir, basefilename + '.diff.png'))
+                self.fail('PNG files differ')
+
+    def test_basic_text(self):
+        self.run_test_case('basic_text.pdf', 
+            wkhtml2image_args=[
+                '--crop-x', 180,
+                '--crop-y', 150,
+                '--crop-w', 220,
+                '--crop-h', 260
+            ])
+
+    def test_geneve_1564(self):
+        self.run_test_case('geneve_1564.pdf', wkhtml2image_args=['--height', 1100])
+
+    def test_text_visibility(self):
+        self.run_test_case('text_visibility.pdf', wkhtml2image_args=['--height', 1200])
+

test/test_conversion/basic_text.tex

@@ -0,0 +1,36 @@
+\documentclass{article}
+\begin{document}
+Normal\hspace{10pt}{\tiny tiny}\hspace{10pt}Text
+\pdfliteral{5 Ts}
+Rise \\
+\pdfliteral{0 Ts}
+
+\pdfliteral{5 Tc}
+CharSpace \\
+\pdfliteral{0 Tc}
+
+\pdfliteral{200 Tz}
+Horizontal\hspace{10pt}Scale \\
+\pdfliteral{100 Tz}
+
+\vspace{3cm}
+
+\pdfliteral{q}
+\pdfliteral{0.71 0.71 -0.71 0.71 0 0 cm}
+Rotated
+\pdfliteral{5 Ts}
+Rise \\
+\pdfliteral{0 Ts}
+
+\pdfliteral{5 Tc}
+CharSpace \\
+\pdfliteral{0 Tc}
+
+\pdfliteral{200 Tz}
+Horizontal\hspace{10pt}Scale \\
+\pdfliteral{100 Tz}
+
+\pdfliteral{Q}
+
+
+\end{document}

test/test_naming.py

@@ -7,11 +7,11 @@ import os
 
 from test import Common
 
-class OutputNamingTests(Common, unittest.TestCase):
+class T(Common, unittest.TestCase):
     def run_test_case(self, input_file, expected_output_files, args=[]):
         args = list(args)
         args.insert(0, os.path.join(self.TEST_DIR, 'test_naming', input_file))
-        self.assertEquals(self.run_pdf2htmlEX(args)['output_files'], sorted(expected_output_files))
+        self.assertItemsEquals(self.run_pdf2htmlEX(args)['output_files'], expected_output_files)
 
     def test_generate_single_html_default_name_single_page_pdf(self):
         self.run_test_case('1-page.pdf', ['1-page.html'])