1  #!/usr/bin/env python3
       2  #
       3  # this tests the entities substitutions with the XmlTextReader interface
       4  #
       5  import sys
       6  import setup_test
       7  import libxml2
       8  try:
       9      import StringIO
      10      str_io = StringIO.StringIO
      11  except:
      12      import io
      13      str_io = io.StringIO
      14  
      15  docstr="""<?xml version='1.0'?>
      16  <!DOCTYPE doc [
      17  <!ENTITY tst "<p>test</p>">
      18  ]>
      19  <doc>&tst;</doc>"""
      20  
      21  # Memory debug specific
      22  libxml2.debugMemory(1)
      23  
      24  #
      25  # First test, normal don't substitute entities.
      26  #
      27  f = str_io(docstr)
      28  input = libxml2.inputBuffer(f)
      29  reader = input.newTextReader("test_noent")
      30  ret = reader.Read()
      31  if ret != 1:
      32      print("Error reading to root")
      33      sys.exit(1)
      34  if reader.Name() == "doc" or reader.NodeType() == 10:
      35      ret = reader.Read()
      36  if ret != 1:
      37      print("Error reading to root")
      38      sys.exit(1)
      39  if reader.Name() != "doc" or reader.NodeType() != 1:
      40      print("test_normal: Error reading the root element")
      41      sys.exit(1)
      42  ret = reader.Read()
      43  if ret != 1:
      44      print("test_normal: Error reading to the entity")
      45      sys.exit(1)
      46  if reader.Name() != "tst" or reader.NodeType() != 5:
      47      print("test_normal: Error reading the entity")
      48      sys.exit(1)
      49  ret = reader.Read()
      50  if ret != 1:
      51      print("test_normal: Error reading to the end of root")
      52      sys.exit(1)
      53  if reader.Name() != "doc" or reader.NodeType() != 15:
      54      print("test_normal: Error reading the end of the root element")
      55      sys.exit(1)
      56  ret = reader.Read()
      57  if ret != 0:
      58      print("test_normal: Error detecting the end")
      59      sys.exit(1)
      60  
      61  #
      62  # Second test, completely substitute the entities.
      63  #
      64  f = str_io(docstr)
      65  input = libxml2.inputBuffer(f)
      66  reader = input.newTextReader("test_noent")
      67  reader.SetParserProp(libxml2.PARSER_SUBST_ENTITIES, 1)
      68  ret = reader.Read()
      69  if ret != 1:
      70      print("Error reading to root")
      71      sys.exit(1)
      72  if reader.Name() == "doc" or reader.NodeType() == 10:
      73      ret = reader.Read()
      74  if ret != 1:
      75      print("Error reading to root")
      76      sys.exit(1)
      77  if reader.Name() != "doc" or reader.NodeType() != 1:
      78      print("test_noent: Error reading the root element")
      79      sys.exit(1)
      80  ret = reader.Read()
      81  if ret != 1:
      82      print("test_noent: Error reading to the entity content")
      83      sys.exit(1)
      84  if reader.Name() != "p" or reader.NodeType() != 1:
      85      print("test_noent: Error reading the p element from entity")
      86      sys.exit(1)
      87  ret = reader.Read()
      88  if ret != 1:
      89      print("test_noent: Error reading to the text node")
      90      sys.exit(1)
      91  if reader.NodeType() != 3 or reader.Value() != "test":
      92      print("test_noent: Error reading the text node")
      93      sys.exit(1)
      94  ret = reader.Read()
      95  if ret != 1:
      96      print("test_noent: Error reading to the end of p element")
      97      sys.exit(1)
      98  if reader.Name() != "p" or reader.NodeType() != 15:
      99      print("test_noent: Error reading the end of the p element")
     100      sys.exit(1)
     101  ret = reader.Read()
     102  if ret != 1:
     103      print("test_noent: Error reading to the end of root")
     104      sys.exit(1)
     105  if reader.Name() != "doc" or reader.NodeType() != 15:
     106      print("test_noent: Error reading the end of the root element")
     107      sys.exit(1)
     108  ret = reader.Read()
     109  if ret != 0:
     110      print("test_noent: Error detecting the end")
     111      sys.exit(1)
     112  
     113  #
     114  # third test, crazy stuff about empty element in external parsed entities
     115  #
     116  s = """<!DOCTYPE struct [
     117  <!ENTITY simplestruct2.ent SYSTEM "simplestruct2.ent">
     118  ]>
     119  <struct>&simplestruct2.ent;</struct>
     120  """
     121  expect="""10 struct 0 0
     122  1 struct 0 0
     123  1 descr 1 1
     124  15 struct 0 0
     125  """
     126  res=""
     127  simplestruct2_ent="""<descr/>"""
     128  
     129  def myResolver(URL, ID, ctxt):
     130      if URL == "simplestruct2.ent":
     131          return(str_io(simplestruct2_ent))
     132      return None
     133  
     134  libxml2.setEntityLoader(myResolver)
     135  
     136  input = libxml2.inputBuffer(str_io(s))
     137  reader = input.newTextReader("test3")
     138  reader.SetParserProp(libxml2.PARSER_SUBST_ENTITIES,1)
     139  while reader.Read() == 1:
     140      res = res + "%s %s %d %d\n" % (reader.NodeType(),reader.Name(),
     141                                     reader.Depth(),reader.IsEmptyElement())
     142  
     143  if res != expect:
     144      print("test3 failed: unexpected output")
     145      print(res)
     146      sys.exit(1)
     147  
     148  #
     149  # cleanup
     150  #
     151  del f
     152  del input
     153  del reader
     154  
     155  # Memory debug specific
     156  libxml2.cleanupParser()
     157  if libxml2.debugMemory(1) == 0:
     158      print("OK")
     159  else:
     160      print("Memory leak %d bytes" % (libxml2.debugMemory(1)))