Source Code for Module lxml.html.tests.test_feedparser_data

 1  import sys 
 2  import os 
 3  import re 
 4  import rfc822 
 5  import unittest 
 6  from lxml.tests.common_imports import doctest 
 7  if sys.version_info >= (2,4): 
 8      from lxml.doctestcompare import LHTMLOutputChecker 
 9   
10  from lxml.html.clean import clean, Cleaner 
11   
12  feed_dirs = [ 
13      os.path.join(os.path.dirname(__file__), 'feedparser-data'), 
14      os.path.join(os.path.dirname(__file__), 'hackers-org-data'), 
15      ] 
16  bar_re = re.compile(r"-----+") 
17   
18 -class DummyInput: 
19 -    def __init__(self, **kw): 
20          for name, value in kw.items(): 
21              setattr(self, name, value) 
22   
23 -class FeedTestCase(unittest.TestCase): 
24   
25 -    def __init__(self, filename): 
26          self.filename = filename 
27          unittest.TestCase.__init__(self) 
28   
29 -    def parse(self): 
30          f = open(self.filename, 'rb') 
31          headers = rfc822.Message(f) 
32          c = f.read() 
33          f.close() 
34          if not headers.keys(): 
35              raise Exception( 
36                  "File %s has no headers" % self.filename) 
37          self.description = headers['Description'] 
38          self.expect = headers.get('Expect', '') 
39          self.ignore = headers.get('Ignore') 
40          self.options = [ 
41              o.strip() for o in headers.get('Options', '').split(',') 
42              if o.strip()] 
43          parts = bar_re.split(c) 
44          self.input = parts[0].rstrip() + '\n' 
45          if parts[1:]: 
46              self.expect = parts[1].rstrip() + '\n' 
47          else: 
48              self.expect = None 
49   
50 -    def runTest(self): 
51          self.parse() 
52          if self.ignore: 
53              # We've marked this test to be ignored. 
54              return 
55          kw = {} 
56          for name in self.options: 
57              if name.startswith('-'): 
58                  kw[name[1:]] = False 
59              else: 
60                  kw[name] = True 
61          if kw.get('clean', True): 
62              transformed = Cleaner(**kw).clean_html(self.input) 
63          else: 
64              transformed = self.input 
65          assert self.expect is not None, ( 
66              "No expected output in %s" % self.filename) 
67          checker = LHTMLOutputChecker() 
68          if not checker.check_output(self.expect, transformed, 0): 
69              result = checker.output_difference( 
70                  DummyInput(want=self.expect), transformed, 0) 
71              #result += '\noptions: %s %r' % (', '.join(self.options), kw) 
72              #result += repr(transformed) 
73              raise Exception("\n"+result) 
74   
75 -    def shortDescription(self): 
76          return self.filename 
77   
78 -def test_suite(): 
79      suite = unittest.TestSuite() 
80      if sys.version_info >= (2,4): 
81          for dir in feed_dirs: 
82              for fn in os.listdir(dir): 
83                  fn = os.path.join(dir, fn) 
84                  if fn.endswith('.data'): 
85                      case = FeedTestCase(fn) 
86                      suite.addTests([case]) 
87                      # This is my lazy way of stopping on first error: 
88                      try: 
89                          case.runTest() 
90                      except: 
91                          break 
92      return suite 
93