Fred Drake | 45cd9de | 2000-06-29 19:34:54 +0000 | [diff] [blame] | 1 | """ |
| 2 | SAX driver for the Pyexpat C module. This driver works with |
| 3 | pyexpat.__version__ == '1.5'. |
| 4 | |
| 5 | $Id$ |
| 6 | """ |
| 7 | |
| 8 | # Todo on driver: |
| 9 | # - make it support external entities (wait for pyexpat.c) |
| 10 | # - enable configuration between reset() and feed() calls |
| 11 | # - support lexical events? |
| 12 | # - proper inputsource handling |
| 13 | # - properties and features |
| 14 | |
| 15 | # Todo on pyexpat.c: |
| 16 | # - support XML_ExternalEntityParserCreate |
| 17 | # - exceptions in callouts from pyexpat to python code lose position info |
| 18 | |
| 19 | version = "0.20" |
| 20 | |
| 21 | from string import split |
| 22 | |
| 23 | from xml.sax import xmlreader |
| 24 | import pyexpat |
| 25 | import xml.sax |
| 26 | |
| 27 | # --- ExpatParser |
| 28 | |
| 29 | class ExpatParser( xmlreader.IncrementalParser, xmlreader.Locator ): |
| 30 | "SAX driver for the Pyexpat C module." |
| 31 | |
| 32 | def __init__(self, namespaceHandling=0, bufsize=2**16-20): |
| 33 | xmlreader.IncrementalParser.__init__(self, bufsize) |
| 34 | self._source = None |
| 35 | self._parser = None |
| 36 | self._namespaces = namespaceHandling |
| 37 | self._parsing = 0 |
| 38 | |
| 39 | # XMLReader methods |
| 40 | |
| 41 | def parse(self, stream_or_string ): |
Skip Montanaro | 26a7983 | 2000-07-06 02:56:36 +0000 | [diff] [blame] | 42 | "Parse an XML document from a URL." |
Fred Drake | 45cd9de | 2000-06-29 19:34:54 +0000 | [diff] [blame] | 43 | if type( stream_or_string ) == type( "" ): |
| 44 | stream=open( stream_or_string ) |
| 45 | else: |
| 46 | stream=stream_or_string |
| 47 | |
| 48 | self.reset() |
| 49 | self._cont_handler.setDocumentLocator(self) |
| 50 | try: |
| 51 | xmlreader.IncrementalParser.parse(self, stream) |
| 52 | except pyexpat.error: |
| 53 | error_code = self._parser.ErrorCode |
| 54 | raise xml.sax.SAXParseException(pyexpat.ErrorString(error_code), |
| 55 | None, self) |
| 56 | |
| 57 | self._cont_handler.endDocument() |
| 58 | |
| 59 | def prepareParser(self, filename=None): |
| 60 | self._source = filename |
| 61 | |
| 62 | if self._source != None: |
| 63 | self._parser.SetBase(self._source) |
| 64 | |
| 65 | def getFeature(self, name): |
| 66 | "Looks up and returns the state of a SAX2 feature." |
| 67 | raise SAXNotRecognizedException("Feature '%s' not recognized" % name) |
| 68 | |
| 69 | def setFeature(self, name, state): |
| 70 | "Sets the state of a SAX2 feature." |
| 71 | raise SAXNotRecognizedException("Feature '%s' not recognized" % name) |
| 72 | |
| 73 | def getProperty(self, name): |
| 74 | "Looks up and returns the value of a SAX2 property." |
| 75 | raise SAXNotRecognizedException("Property '%s' not recognized" % name) |
| 76 | |
| 77 | def setProperty(self, name, value): |
| 78 | "Sets the value of a SAX2 property." |
| 79 | raise SAXNotRecognizedException("Property '%s' not recognized" % name) |
| 80 | |
| 81 | # IncrementalParser methods |
| 82 | |
| 83 | def feed(self, data): |
| 84 | if not self._parsing: |
| 85 | self._parsing=1 |
| 86 | self.reset() |
| 87 | self._cont_handler.startDocument() |
| 88 | # FIXME: error checking and endDocument() |
| 89 | self._parser.Parse(data, 0) |
| 90 | |
| 91 | def close(self): |
| 92 | if self._parsing: |
| 93 | self._cont_handler.endDocument() |
| 94 | self._parsing=0 |
| 95 | self._parser.Parse("", 1) |
| 96 | |
| 97 | def reset(self): |
| 98 | if self._namespaces: |
| 99 | self._parser = pyexpat.ParserCreate(None, " ") |
| 100 | self._parser.StartElementHandler = self.start_element_ns |
| 101 | self._parser.EndElementHandler = self.end_element_ns |
| 102 | else: |
| 103 | self._parser = pyexpat.ParserCreate() |
Paul Prescod | 6c4753f | 2000-07-04 03:39:33 +0000 | [diff] [blame] | 104 | self._parser.StartElementHandler = self.start_element |
| 105 | self._parser.EndElementHandler = self.end_element |
Fred Drake | 45cd9de | 2000-06-29 19:34:54 +0000 | [diff] [blame] | 106 | |
| 107 | self._parser.ProcessingInstructionHandler = \ |
| 108 | self._cont_handler.processingInstruction |
| 109 | self._parser.CharacterDataHandler = self._cont_handler.characters |
| 110 | self._parser.UnparsedEntityDeclHandler = self.unparsed_entity_decl |
| 111 | self._parser.NotationDeclHandler = self.notation_decl |
| 112 | self._parser.StartNamespaceDeclHandler = self.start_namespace_decl |
| 113 | self._parser.EndNamespaceDeclHandler = self.end_namespace_decl |
| 114 | # self._parser.CommentHandler = |
| 115 | # self._parser.StartCdataSectionHandler = |
| 116 | # self._parser.EndCdataSectionHandler = |
| 117 | # self._parser.DefaultHandler = |
| 118 | # self._parser.DefaultHandlerExpand = |
| 119 | # self._parser.NotStandaloneHandler = |
| 120 | self._parser.ExternalEntityRefHandler = self.external_entity_ref |
| 121 | |
| 122 | # Locator methods |
| 123 | |
| 124 | def getColumnNumber(self): |
| 125 | return self._parser.ErrorColumnNumber |
| 126 | |
| 127 | def getLineNumber(self): |
| 128 | return self._parser.ErrorLineNumber |
| 129 | |
| 130 | def getPublicId(self): |
| 131 | return self._source.getPublicId() |
| 132 | |
| 133 | def getSystemId(self): |
| 134 | return self._parser.GetBase() |
| 135 | |
Fred Drake | 45cd9de | 2000-06-29 19:34:54 +0000 | [diff] [blame] | 136 | # event handlers |
Fred Drake | 45cd9de | 2000-06-29 19:34:54 +0000 | [diff] [blame] | 137 | def start_element(self, name, attrs): |
Paul Prescod | 6c4753f | 2000-07-04 03:39:33 +0000 | [diff] [blame] | 138 | self._cont_handler.startElement(name, name, |
Fred Drake | 45cd9de | 2000-06-29 19:34:54 +0000 | [diff] [blame] | 139 | xmlreader.AttributesImpl(attrs, attrs)) |
| 140 | |
| 141 | def end_element(self, name): |
Paul Prescod | 6c4753f | 2000-07-04 03:39:33 +0000 | [diff] [blame] | 142 | self._cont_handler.endElement( name, name ) |
Fred Drake | 45cd9de | 2000-06-29 19:34:54 +0000 | [diff] [blame] | 143 | |
| 144 | def start_element_ns(self, name, attrs): |
| 145 | pair = split(name) |
| 146 | if len(pair) == 1: |
Paul Prescod | 6c4753f | 2000-07-04 03:39:33 +0000 | [diff] [blame] | 147 | tup = (None, name ) |
Fred Drake | 45cd9de | 2000-06-29 19:34:54 +0000 | [diff] [blame] | 148 | else: |
Paul Prescod | 6c4753f | 2000-07-04 03:39:33 +0000 | [diff] [blame] | 149 | tup = pair |
Fred Drake | 45cd9de | 2000-06-29 19:34:54 +0000 | [diff] [blame] | 150 | |
Paul Prescod | 6c4753f | 2000-07-04 03:39:33 +0000 | [diff] [blame] | 151 | self._cont_handler.startElement(tup, None, |
Fred Drake | 45cd9de | 2000-06-29 19:34:54 +0000 | [diff] [blame] | 152 | xmlreader.AttributesImpl(attrs, None)) |
| 153 | |
| 154 | def end_element_ns(self, name): |
| 155 | pair = split(name) |
| 156 | if len(pair) == 1: |
| 157 | name = (None, name, None) |
| 158 | else: |
| 159 | name = pair+[None] # prefix is not implemented yet! |
| 160 | |
Paul Prescod | 6c4753f | 2000-07-04 03:39:33 +0000 | [diff] [blame] | 161 | self._cont_handler.endElement(name, None) |
Fred Drake | 45cd9de | 2000-06-29 19:34:54 +0000 | [diff] [blame] | 162 | |
Paul Prescod | 6c4753f | 2000-07-04 03:39:33 +0000 | [diff] [blame] | 163 | # this is not used |
Fred Drake | 45cd9de | 2000-06-29 19:34:54 +0000 | [diff] [blame] | 164 | def processing_instruction(self, target, data): |
| 165 | self._cont_handler.processingInstruction(target, data) |
| 166 | |
Paul Prescod | 6c4753f | 2000-07-04 03:39:33 +0000 | [diff] [blame] | 167 | # this is not used |
Fred Drake | 45cd9de | 2000-06-29 19:34:54 +0000 | [diff] [blame] | 168 | def character_data(self, data): |
| 169 | self._cont_handler.characters(data) |
| 170 | |
| 171 | def start_namespace_decl(self, prefix, uri): |
| 172 | self._cont_handler.startPrefixMapping(prefix, uri) |
| 173 | |
| 174 | def end_namespace_decl(self, prefix): |
| 175 | self._cont_handler.endPrefixMapping(prefix) |
| 176 | |
| 177 | def unparsed_entity_decl(self, name, base, sysid, pubid, notation_name): |
| 178 | self._dtd_handler.unparsedEntityDecl(name, pubid, sysid, notation_name) |
| 179 | |
| 180 | def notation_decl(self, name, base, sysid, pubid): |
| 181 | self._dtd_handler.notationDecl(name, pubid, sysid) |
| 182 | |
| 183 | def external_entity_ref(self, context, base, sysid, pubid): |
| 184 | assert 0 # not implemented |
| 185 | source = self._ent_handler.resolveEntity(pubid, sysid) |
| 186 | source = saxutils.prepare_input_source(source) |
| 187 | # FIXME: create new parser, stack self._source and self._parser |
| 188 | # FIXME: reuse code from self.parse(...) |
| 189 | return 1 |
| 190 | |
| 191 | # --- |
| 192 | |
| 193 | def create_parser(*args, **kwargs): |
| 194 | return apply( ExpatParser, args, kwargs ) |
| 195 | |
| 196 | # --- |
| 197 | |
| 198 | if __name__ == "__main__": |
| 199 | import xml.sax |
| 200 | p = create_parser() |
| 201 | p.setContentHandler(xml.sax.XMLGenerator()) |
| 202 | p.setErrorHandler(xml.sax.ErrorHandler()) |
| 203 | p.parse("../../../hamlet.xml") |