development/thorough_export_tests/lyxhtml_validity.py

   1 # Stricter version of the export tests: validate the XHTML code produced by
   2 # LyX' lyxhtml output as HTML5. It also validates the CSS and MathML parts.
   3 # Validation errors usually are mistakes in the generator.
   4 #
   5 # Call:
   6 #     python lyxhtml_validity.py PATH_TO_LYX/lyx
   7 #
   8 # Written with Python 3.8.8.
   9 # Requirements:
  10 # - Python package: html5validator: at least v0.4.2
  11 # - Java runtime engine (JRE): at least v8 (depending on html5validator)
  12 # Run:
  13 #     pip install html5validator>=0.4.2
  14
  15 import collections
  16 import glob
  17 import sys
  18 import tempfile
  19 import os
  20
  21 import html5validator
  22
  23
  24 if len(sys.argv) != 2:
  25     print('Expecting one argument, the path to the LyX binary to test')
  26     sys.exit(-1)
  27 if not os.path.exists(sys.argv[1]):
  28     print('The given path does not point to an existing file')
  29     sys.exit(-1)
  30 if not os.access(sys.argv[1], os.X_OK):
  31     print('The given path does not point to an executable file')
  32     sys.exit(-1)
  33
  34
  35 PATH_SCRIPT = os.path.dirname(os.path.realpath(__file__))
  36 PATH_EXPORT_TESTS = os.path.realpath(PATH_SCRIPT + '/../../autotests/export/')
  37 PATH_LYX = sys.argv[1]
  38
  39
  40 TestFile = collections.namedtuple(
  41     "TestFile", ["lyx_path", "lyx_file_name", "xhtml_path", "xhtml_file_name"]
  42 )
  43
  44 validator = html5validator.Validator(format='text')
  45
  46
  47 with tempfile.TemporaryDirectory() as out_dir_name:
  48     all_lyx_files = glob.glob(PATH_EXPORT_TESTS + '/**/*.lyx', recursive=True)
  49     all_files = [
  50         TestFile(
  51             lyx_path=path,
  52             lyx_file_name=os.path.basename(path),
  53             xhtml_path = os.path.join(out_dir_name, os.path.basename(path)),
  54             xhtml_file_name=os.path.basename(path).replace('.lyx', '.html')
  55         ) for path in all_lyx_files
  56     ]
  57
  58     # Generate XHTML files.
  59     print(
  60         f'Exporting {len(all_lyx_files)} LyX files to LyXHTML format in the ' +
  61         f'directory {out_dir_name}'
  62     )
  63
  64     for file in all_files:
  65         print(f'* Generating {file.lyx_file_name}...')
  66         os.system(f'{PATH_LYX} --export-to xhtml "{file.xhtml_path}" "{file.lyx_path}"')
  67         print(f'> Done generating {file.lyx_path} to {file.xhtml_path}')
  68
  69         # print()
  70         # print(open(file.xhtml_path, 'r').read())
  71
  72     print(f'Exported successfully all {len(all_lyx_files)} files!')
  73
  74     # Validate the XHTML files.
  75     print(f'Validating {len(all_lyx_files)} XHTML files...')
  76
  77     n_valid = 0
  78     n_invalid = 0
  79     for file in all_files:
  80         print(f'* Validating {file.xhtml_file_name}...')
  81         error_count = validator.validate([file.xhtml_file_name])
  82         # Caution: this call outputs all validation errors to stdout!
  83         # This line is equivalent to running vnu on the corresponding file.
  84         # https://github.com/validator/validator
  85
  86         if error_count == 0:
  87             n_valid += 1
  88             print(f'> Found no validation error!')
  89         else:
  90             n_invalid += 1
  91             print(f'> Found {error_count} validation error{"" if error_count == 1 else "s"}!')
  92
  93     print(f'Validated all {len(all_lyx_files)} files! Among them:')
  94     print(f'> {n_valid} were valid ({100.0 * n_valid / len(all_files)}%)')
  95     print(f'> {n_invalid} were invalid ({100.0 * n_invalid / len(all_files)}%)')
  96
  97     if n_invalid == 0:
  98         print("That's excellent! Give yourself a pat on the back!")
  99     elif 100.0 * n_invalid / len(all_files) <= 5.0:
 100         print("That's a pretty good job!")