md4c

C Markdown parser. Fast. SAX-like interface. Compliant to CommonMark specification.
git clone https://noulin.net/git/md4c.git
Log | Files | Refs | README | LICENSE

pathological_tests.py (3479B)


      1 #!/usr/bin/env python3
      2 # -*- coding: utf-8 -*-
      3 
      4 import re
      5 import argparse
      6 import sys
      7 import platform
      8 from cmark import CMark
      9 
     10 if __name__ == "__main__":
     11     parser = argparse.ArgumentParser(description='Run cmark tests.')
     12     parser.add_argument('-p', '--program', dest='program', nargs='?', default=None,
     13             help='program to test')
     14     parser.add_argument('--library-dir', dest='library_dir', nargs='?',
     15             default=None, help='directory containing dynamic library')
     16     args = parser.parse_args(sys.argv[1:])
     17 
     18 cmark = CMark(prog=args.program, library_dir=args.library_dir)
     19 
     20 # list of pairs consisting of input and a regex that must match the output.
     21 pathological = {
     22     # note - some pythons have limit of 65535 for {num-matches} in re.
     23     "nested strong emph":
     24                 (("*a **a " * 65000) + "b" + (" a** a*" * 65000),
     25                  re.compile("(<em>a <strong>a ){65000}b( a</strong> a</em>){65000}")),
     26     "many emph closers with no openers":
     27                  (("a_ " * 65000),
     28                   re.compile("(a[_] ){64999}a_")),
     29     "many emph openers with no closers":
     30                  (("_a " * 65000),
     31                   re.compile("(_a ){64999}_a")),
     32     "many link closers with no openers":
     33                  (("a]" * 65000),
     34                   re.compile("(a\]){65000}")),
     35     "many link openers with no closers":
     36                  (("[a" * 65000),
     37                   re.compile("(\[a){65000}")),
     38     "mismatched openers and closers":
     39                  (("*a_ " * 50000),
     40                   re.compile("([*]a[_] ){49999}[*]a_")),
     41     "openers and closers multiple of 3":
     42                  (("a**b" + ("c* " * 50000)),
     43                   re.compile("a[*][*]b(c[*] ){49999}c[*]")),
     44     "link openers and emph closers":
     45                  (("[ a_" * 50000),
     46                   re.compile("(\[ a_){50000}")),
     47     "hard link/emph case":
     48                  ("**x [a*b**c*](d)",
     49                   re.compile("\\*\\*x <a href=\"d\">a<em>b\\*\\*c</em></a>")),
     50     "nested brackets":
     51                  (("[" * 50000) + "a" + ("]" * 50000),
     52                   re.compile("\[{50000}a\]{50000}")),
     53     "nested block quotes":
     54                  ((("> " * 50000) + "a"),
     55                   re.compile("(<blockquote>\r?\n){50000}")),
     56     "U+0000 in input":
     57                  ("abc\u0000de\u0000",
     58                   re.compile("abc\ufffd?de\ufffd?")),
     59     "backticks":
     60                  ("".join(map(lambda x: ("e" + "`" * x), range(1,10000))),
     61                   re.compile("^<p>[e`]*</p>\r?\n$")),
     62     "many links":
     63                  ("[t](/u) " * 50000,
     64                   re.compile("(<a href=\"/u\">t</a> ?){50000}")),
     65     "many references":
     66                  ("".join(map(lambda x: ("[" + str(x) + "]: u\n"), range(1,50000 * 16))) + "[0] " * 50000,
     67                   re.compile("(\[0\] ){49999}"))
     68     }
     69 
     70 whitespace_re = re.compile('/s+/')
     71 passed = 0
     72 errored = 0
     73 failed = 0
     74 
     75 #print("Testing pathological cases:")
     76 for description in pathological:
     77     (inp, regex) = pathological[description]
     78     [rc, actual, err] = cmark.to_html(inp)
     79     if rc != 0:
     80         errored += 1
     81         print(description, '[ERRORED (return code %d)]' %rc)
     82         print(err)
     83     elif regex.search(actual):
     84         #print(description, '[PASSED]')
     85         passed += 1
     86     else:
     87         print(description, '[FAILED]')
     88         print(repr(actual))
     89         failed += 1
     90 
     91 print("%d passed, %d failed, %d errored" % (passed, failed, errored))
     92 if (failed == 0 and errored == 0):
     93     exit(0)
     94 else:
     95     exit(1)