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'. |
Fred Drake | 45cd9de | 2000-06-29 19:34:54 +0000 | [diff] [blame] | 4 | """ |
| 5 | |
| 6 | # Todo on driver: |
| 7 | # - make it support external entities (wait for pyexpat.c) |
| 8 | # - enable configuration between reset() and feed() calls |
| 9 | # - support lexical events? |
| 10 | # - proper inputsource handling |
| 11 | # - properties and features |
| 12 | |
| 13 | # Todo on pyexpat.c: |
| 14 | # - support XML_ExternalEntityParserCreate |
| 15 | # - exceptions in callouts from pyexpat to python code lose position info |
| 16 | |
| 17 | version = "0.20" |
| 18 | |
Lars Gustäbel | f43cf31 | 2000-09-24 18:29:24 +0000 | [diff] [blame] | 19 | from xml.sax._exceptions import * |
Fred Drake | 96ea196 | 2000-09-23 04:49:30 +0000 | [diff] [blame] | 20 | from xml.parsers import expat |
Lars Gustäbel | 523b0a6 | 2000-09-24 18:54:49 +0000 | [diff] [blame^] | 21 | from xml.sax import xmlreader, saxutils |
Fred Drake | 45cd9de | 2000-06-29 19:34:54 +0000 | [diff] [blame] | 22 | |
Lars Gustäbel | 32bf12e | 2000-09-24 18:39:23 +0000 | [diff] [blame] | 23 | AttributesImpl = xmlreader.AttributesImpl |
| 24 | AttributesNSImpl = xmlreader.AttributesNSImpl |
| 25 | |
Fred Drake | 45cd9de | 2000-06-29 19:34:54 +0000 | [diff] [blame] | 26 | # --- ExpatParser |
| 27 | |
Fred Drake | ddb4867 | 2000-09-23 05:32:26 +0000 | [diff] [blame] | 28 | class ExpatParser(xmlreader.IncrementalParser, xmlreader.Locator): |
Fred Drake | 45cd9de | 2000-06-29 19:34:54 +0000 | [diff] [blame] | 29 | "SAX driver for the Pyexpat C module." |
| 30 | |
| 31 | def __init__(self, namespaceHandling=0, bufsize=2**16-20): |
| 32 | xmlreader.IncrementalParser.__init__(self, bufsize) |
| 33 | self._source = None |
| 34 | self._parser = None |
| 35 | self._namespaces = namespaceHandling |
| 36 | self._parsing = 0 |
| 37 | |
| 38 | # XMLReader methods |
| 39 | |
Lars Gustäbel | 523b0a6 | 2000-09-24 18:54:49 +0000 | [diff] [blame^] | 40 | def parse(self, source): |
Skip Montanaro | 26a7983 | 2000-07-06 02:56:36 +0000 | [diff] [blame] | 41 | "Parse an XML document from a URL." |
Lars Gustäbel | 523b0a6 | 2000-09-24 18:54:49 +0000 | [diff] [blame^] | 42 | source = saxutils.prepare_input_source(source) |
| 43 | |
| 44 | self._source = source |
Fred Drake | 45cd9de | 2000-06-29 19:34:54 +0000 | [diff] [blame] | 45 | self.reset() |
| 46 | self._cont_handler.setDocumentLocator(self) |
| 47 | try: |
Lars Gustäbel | 523b0a6 | 2000-09-24 18:54:49 +0000 | [diff] [blame^] | 48 | xmlreader.IncrementalParser.parse(self, source) |
Fred Drake | 96ea196 | 2000-09-23 04:49:30 +0000 | [diff] [blame] | 49 | except expat.error: |
Fred Drake | 45cd9de | 2000-06-29 19:34:54 +0000 | [diff] [blame] | 50 | error_code = self._parser.ErrorCode |
Lars Gustäbel | f43cf31 | 2000-09-24 18:29:24 +0000 | [diff] [blame] | 51 | raise SAXParseException(expat.ErrorString(error_code), None, self) |
Fred Drake | 45cd9de | 2000-06-29 19:34:54 +0000 | [diff] [blame] | 52 | |
| 53 | self._cont_handler.endDocument() |
| 54 | |
Lars Gustäbel | 523b0a6 | 2000-09-24 18:54:49 +0000 | [diff] [blame^] | 55 | def prepareParser(self, source): |
| 56 | if source.getSystemId() != None: |
| 57 | self._parser.SetBase(source.getSystemId()) |
Fred Drake | 45cd9de | 2000-06-29 19:34:54 +0000 | [diff] [blame] | 58 | |
| 59 | def getFeature(self, name): |
Lars Gustäbel | f43cf31 | 2000-09-24 18:29:24 +0000 | [diff] [blame] | 60 | if name == feature_namespaces: |
| 61 | return self._namespaces |
Fred Drake | 45cd9de | 2000-06-29 19:34:54 +0000 | [diff] [blame] | 62 | raise SAXNotRecognizedException("Feature '%s' not recognized" % name) |
| 63 | |
| 64 | def setFeature(self, name, state): |
Lars Gustäbel | f43cf31 | 2000-09-24 18:29:24 +0000 | [diff] [blame] | 65 | if self._parsing: |
| 66 | raise SAXNotSupportedException("Cannot set features while parsing") |
| 67 | if name == feature_namespaces: |
| 68 | self._namespaces = state |
| 69 | else: |
| 70 | raise SAXNotRecognizedException("Feature '%s' not recognized" % |
| 71 | name) |
Fred Drake | 45cd9de | 2000-06-29 19:34:54 +0000 | [diff] [blame] | 72 | |
| 73 | def getProperty(self, name): |
Fred Drake | 45cd9de | 2000-06-29 19:34:54 +0000 | [diff] [blame] | 74 | raise SAXNotRecognizedException("Property '%s' not recognized" % name) |
| 75 | |
| 76 | def setProperty(self, name, value): |
Fred Drake | 45cd9de | 2000-06-29 19:34:54 +0000 | [diff] [blame] | 77 | raise SAXNotRecognizedException("Property '%s' not recognized" % name) |
| 78 | |
| 79 | # IncrementalParser methods |
| 80 | |
| 81 | def feed(self, data): |
| 82 | if not self._parsing: |
Fred Drake | ddb4867 | 2000-09-23 05:32:26 +0000 | [diff] [blame] | 83 | self._parsing = 1 |
Fred Drake | 45cd9de | 2000-06-29 19:34:54 +0000 | [diff] [blame] | 84 | self.reset() |
| 85 | self._cont_handler.startDocument() |
Lars Gustäbel | f43cf31 | 2000-09-24 18:29:24 +0000 | [diff] [blame] | 86 | |
| 87 | if not self._parser.Parse(data, 0): |
| 88 | msg = pyexpat.ErrorString(self._parser.ErrorCode) |
| 89 | raise SAXParseException(msg, None, self) |
Fred Drake | 45cd9de | 2000-06-29 19:34:54 +0000 | [diff] [blame] | 90 | |
| 91 | def close(self): |
| 92 | if self._parsing: |
| 93 | self._cont_handler.endDocument() |
Fred Drake | ddb4867 | 2000-09-23 05:32:26 +0000 | [diff] [blame] | 94 | self._parsing = 0 |
Fred Drake | 45cd9de | 2000-06-29 19:34:54 +0000 | [diff] [blame] | 95 | self._parser.Parse("", 1) |
| 96 | |
| 97 | def reset(self): |
| 98 | if self._namespaces: |
Fred Drake | 96ea196 | 2000-09-23 04:49:30 +0000 | [diff] [blame] | 99 | self._parser = expat.ParserCreate(None, " ") |
Fred Drake | 45cd9de | 2000-06-29 19:34:54 +0000 | [diff] [blame] | 100 | self._parser.StartElementHandler = self.start_element_ns |
| 101 | self._parser.EndElementHandler = self.end_element_ns |
| 102 | else: |
Fred Drake | 96ea196 | 2000-09-23 04:49:30 +0000 | [diff] [blame] | 103 | self._parser = expat.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): |
Lars Gustäbel | 32bf12e | 2000-09-24 18:39:23 +0000 | [diff] [blame] | 138 | self._cont_handler.startElement(name, AttributesImpl(attrs)) |
Fred Drake | 45cd9de | 2000-06-29 19:34:54 +0000 | [diff] [blame] | 139 | |
| 140 | def end_element(self, name): |
Lars Gustäbel | f43cf31 | 2000-09-24 18:29:24 +0000 | [diff] [blame] | 141 | self._cont_handler.endElement(name) |
Fred Drake | 45cd9de | 2000-06-29 19:34:54 +0000 | [diff] [blame] | 142 | |
| 143 | def start_element_ns(self, name, attrs): |
Fred Drake | 96ea196 | 2000-09-23 04:49:30 +0000 | [diff] [blame] | 144 | pair = name.split() |
Fred Drake | 45cd9de | 2000-06-29 19:34:54 +0000 | [diff] [blame] | 145 | if len(pair) == 1: |
Lars Gustäbel | f43cf31 | 2000-09-24 18:29:24 +0000 | [diff] [blame] | 146 | pair = (None, name) |
Fred Drake | 45cd9de | 2000-06-29 19:34:54 +0000 | [diff] [blame] | 147 | |
Lars Gustäbel | 32bf12e | 2000-09-24 18:39:23 +0000 | [diff] [blame] | 148 | newattrs = {} |
| 149 | for (aname, value) in attrs.items(): |
| 150 | apair = aname.split() |
| 151 | if len(apair) == 1: |
| 152 | apair = (None, aname) |
| 153 | else: |
| 154 | apair = tuple(apair) |
| 155 | |
| 156 | newattrs[apair] = value |
| 157 | |
| 158 | self._cont_handler.startElementNS(pair, None, |
| 159 | AttributesNSImpl(newattrs, {})) |
Fred Drake | 45cd9de | 2000-06-29 19:34:54 +0000 | [diff] [blame] | 160 | |
| 161 | def end_element_ns(self, name): |
Fred Drake | 96ea196 | 2000-09-23 04:49:30 +0000 | [diff] [blame] | 162 | pair = name.split() |
Fred Drake | 45cd9de | 2000-06-29 19:34:54 +0000 | [diff] [blame] | 163 | if len(pair) == 1: |
Lars Gustäbel | 32bf12e | 2000-09-24 18:39:23 +0000 | [diff] [blame] | 164 | pair = (None, name) |
Fred Drake | 45cd9de | 2000-06-29 19:34:54 +0000 | [diff] [blame] | 165 | |
Lars Gustäbel | f43cf31 | 2000-09-24 18:29:24 +0000 | [diff] [blame] | 166 | self._cont_handler.endElementNS(pair, None) |
Fred Drake | 45cd9de | 2000-06-29 19:34:54 +0000 | [diff] [blame] | 167 | |
Lars Gustäbel | f43cf31 | 2000-09-24 18:29:24 +0000 | [diff] [blame] | 168 | # this is not used (call directly to ContentHandler) |
Fred Drake | 45cd9de | 2000-06-29 19:34:54 +0000 | [diff] [blame] | 169 | def processing_instruction(self, target, data): |
| 170 | self._cont_handler.processingInstruction(target, data) |
| 171 | |
Lars Gustäbel | f43cf31 | 2000-09-24 18:29:24 +0000 | [diff] [blame] | 172 | # this is not used (call directly to ContentHandler) |
Fred Drake | 45cd9de | 2000-06-29 19:34:54 +0000 | [diff] [blame] | 173 | def character_data(self, data): |
| 174 | self._cont_handler.characters(data) |
| 175 | |
| 176 | def start_namespace_decl(self, prefix, uri): |
| 177 | self._cont_handler.startPrefixMapping(prefix, uri) |
| 178 | |
| 179 | def end_namespace_decl(self, prefix): |
| 180 | self._cont_handler.endPrefixMapping(prefix) |
| 181 | |
| 182 | def unparsed_entity_decl(self, name, base, sysid, pubid, notation_name): |
| 183 | self._dtd_handler.unparsedEntityDecl(name, pubid, sysid, notation_name) |
| 184 | |
| 185 | def notation_decl(self, name, base, sysid, pubid): |
| 186 | self._dtd_handler.notationDecl(name, pubid, sysid) |
| 187 | |
| 188 | def external_entity_ref(self, context, base, sysid, pubid): |
Lars Gustäbel | f43cf31 | 2000-09-24 18:29:24 +0000 | [diff] [blame] | 189 | raise NotImplementedError() |
Fred Drake | 45cd9de | 2000-06-29 19:34:54 +0000 | [diff] [blame] | 190 | source = self._ent_handler.resolveEntity(pubid, sysid) |
| 191 | source = saxutils.prepare_input_source(source) |
| 192 | # FIXME: create new parser, stack self._source and self._parser |
| 193 | # FIXME: reuse code from self.parse(...) |
| 194 | return 1 |
| 195 | |
| 196 | # --- |
| 197 | |
| 198 | def create_parser(*args, **kwargs): |
Fred Drake | ddb4867 | 2000-09-23 05:32:26 +0000 | [diff] [blame] | 199 | return apply(ExpatParser, args, kwargs) |
Fred Drake | 45cd9de | 2000-06-29 19:34:54 +0000 | [diff] [blame] | 200 | |
| 201 | # --- |
| 202 | |
| 203 | if __name__ == "__main__": |
| 204 | import xml.sax |
| 205 | p = create_parser() |
| 206 | p.setContentHandler(xml.sax.XMLGenerator()) |
| 207 | p.setErrorHandler(xml.sax.ErrorHandler()) |
| 208 | p.parse("../../../hamlet.xml") |