Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 1 | # A parser for XML, using the derived class as static DTD. |
Guido van Rossum | 5d68e8e | 1997-11-18 15:27:20 +0000 | [diff] [blame] | 2 | # Author: Sjoerd Mullender. |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 3 | |
| 4 | import re |
| 5 | import string |
| 6 | |
| 7 | |
Guido van Rossum | b083a9f | 1998-12-18 20:17:13 +0000 | [diff] [blame] | 8 | version = '0.2' |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 9 | |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 10 | # Regular expressions used for parsing |
| 11 | |
Guido van Rossum | eeb2f32 | 1998-10-19 13:28:26 +0000 | [diff] [blame] | 12 | _S = '[ \t\r\n]+' # white space |
| 13 | _opS = '[ \t\r\n]*' # optional white space |
| 14 | _Name = '[a-zA-Z_:][-a-zA-Z0-9._:]*' # valid XML name |
| 15 | _QStr = "(?:'[^']*'|\"[^\"]*\")" # quoted XML string |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 16 | illegal = re.compile('[^\t\r\n -\176\240-\377]') # illegal chars in content |
| 17 | interesting = re.compile('[]&<]') |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 18 | |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 19 | amp = re.compile('&') |
| 20 | ref = re.compile('&(' + _Name + '|#[0-9]+|#x[0-9a-fA-F]+)[^-a-zA-Z0-9._:]') |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 21 | entityref = re.compile('&(?P<name>' + _Name + ')[^-a-zA-Z0-9._:]') |
| 22 | charref = re.compile('&#(?P<char>[0-9]+[^0-9]|x[0-9a-fA-F]+[^0-9a-fA-F])') |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 23 | space = re.compile(_S + '$') |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 24 | newline = re.compile('\n') |
| 25 | |
Guido van Rossum | eeb2f32 | 1998-10-19 13:28:26 +0000 | [diff] [blame] | 26 | attrfind = re.compile( |
| 27 | _S + '(?P<name>' + _Name + ')' |
| 28 | '(' + _opS + '=' + _opS + |
| 29 | '(?P<value>'+_QStr+'|[-a-zA-Z0-9.:+*%?!()_#=~]+))?') |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 30 | starttagopen = re.compile('<' + _Name) |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 31 | starttagend = re.compile(_opS + '(?P<slash>/?)>') |
Guido van Rossum | eeb2f32 | 1998-10-19 13:28:26 +0000 | [diff] [blame] | 32 | starttagmatch = re.compile('<(?P<tagname>'+_Name+')' |
| 33 | '(?P<attrs>(?:'+attrfind.pattern+')*)'+ |
| 34 | starttagend.pattern) |
| 35 | endtagopen = re.compile('</') |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 36 | endbracket = re.compile(_opS + '>') |
Guido van Rossum | eeb2f32 | 1998-10-19 13:28:26 +0000 | [diff] [blame] | 37 | endbracketfind = re.compile('(?:[^>\'"]|'+_QStr+')*>') |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 38 | tagfind = re.compile(_Name) |
Guido van Rossum | 02505e4 | 1998-01-29 14:55:24 +0000 | [diff] [blame] | 39 | cdataopen = re.compile(r'<!\[CDATA\[') |
| 40 | cdataclose = re.compile(r'\]\]>') |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 41 | # this matches one of the following: |
| 42 | # SYSTEM SystemLiteral |
| 43 | # PUBLIC PubidLiteral SystemLiteral |
Guido van Rossum | eeb2f32 | 1998-10-19 13:28:26 +0000 | [diff] [blame] | 44 | _SystemLiteral = '(?P<%s>'+_QStr+')' |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 45 | _PublicLiteral = '(?P<%s>"[-\'()+,./:=?;!*#@$_%% \n\ra-zA-Z0-9]*"|' \ |
| 46 | "'[-()+,./:=?;!*#@$_%% \n\ra-zA-Z0-9]*')" |
| 47 | _ExternalId = '(?:SYSTEM|' \ |
| 48 | 'PUBLIC'+_S+_PublicLiteral%'pubid'+ \ |
| 49 | ')'+_S+_SystemLiteral%'syslit' |
| 50 | doctype = re.compile('<!DOCTYPE'+_S+'(?P<name>'+_Name+')' |
| 51 | '(?:'+_S+_ExternalId+')?'+_opS) |
| 52 | xmldecl = re.compile('<\?xml'+_S+ |
Guido van Rossum | eeb2f32 | 1998-10-19 13:28:26 +0000 | [diff] [blame] | 53 | 'version'+_opS+'='+_opS+'(?P<version>'+_QStr+')'+ |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 54 | '(?:'+_S+'encoding'+_opS+'='+_opS+ |
| 55 | "(?P<encoding>'[A-Za-z][-A-Za-z0-9._]*'|" |
| 56 | '"[A-Za-z][-A-Za-z0-9._]*"))?' |
| 57 | '(?:'+_S+'standalone'+_opS+'='+_opS+ |
| 58 | '(?P<standalone>\'(?:yes|no)\'|"(?:yes|no)"))?'+ |
| 59 | _opS+'\?>') |
| 60 | procopen = re.compile(r'<\?(?P<proc>' + _Name + ')' + _opS) |
Guido van Rossum | 02505e4 | 1998-01-29 14:55:24 +0000 | [diff] [blame] | 61 | procclose = re.compile(_opS + r'\?>') |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 62 | commentopen = re.compile('<!--') |
| 63 | commentclose = re.compile('-->') |
| 64 | doubledash = re.compile('--') |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 65 | attrtrans = string.maketrans(' \r\n\t', ' ') |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 66 | |
Guido van Rossum | b083a9f | 1998-12-18 20:17:13 +0000 | [diff] [blame] | 67 | # definitions for XML namespaces |
| 68 | _NCName = '[a-zA-Z_][-a-zA-Z0-9._]*' # XML Name, minus the ":" |
| 69 | ncname = re.compile(_NCName + '$') |
| 70 | qname = re.compile('(?:(?P<prefix>' + _NCName + '):)?' # optional prefix |
| 71 | '(?P<local>' + _NCName + ')$') |
| 72 | |
| 73 | xmlns = re.compile('xmlns(?::(?P<ncname>'+_NCName+'))?$') |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 74 | |
| 75 | # XML parser base class -- find tags and call handler functions. |
| 76 | # Usage: p = XMLParser(); p.feed(data); ...; p.close(). |
Guido van Rossum | 5d68e8e | 1997-11-18 15:27:20 +0000 | [diff] [blame] | 77 | # The dtd is defined by deriving a class which defines methods with |
| 78 | # special names to handle tags: start_foo and end_foo to handle <foo> |
| 79 | # and </foo>, respectively. The data between tags is passed to the |
| 80 | # parser by calling self.handle_data() with some data as argument (the |
| 81 | # data may be split up in arbutrary chunks). Entity references are |
| 82 | # passed by calling self.handle_entityref() with the entity reference |
| 83 | # as argument. |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 84 | |
| 85 | class XMLParser: |
Guido van Rossum | b083a9f | 1998-12-18 20:17:13 +0000 | [diff] [blame] | 86 | attributes = {} # default, to be overridden |
| 87 | elements = {} # default, to be overridden |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 88 | |
| 89 | # Interface -- initialize and reset this instance |
Guido van Rossum | b083a9f | 1998-12-18 20:17:13 +0000 | [diff] [blame] | 90 | def __init__(self): |
Guido van Rossum | cc2c291 | 1999-02-02 17:54:38 +0000 | [diff] [blame] | 91 | self.__fixed = 0 |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 92 | self.reset() |
Guido van Rossum | fe64935 | 1999-02-01 15:35:13 +0000 | [diff] [blame] | 93 | |
| 94 | def __fixelements(self): |
Guido van Rossum | cc2c291 | 1999-02-02 17:54:38 +0000 | [diff] [blame] | 95 | self.__fixed = 1 |
Guido van Rossum | fe64935 | 1999-02-01 15:35:13 +0000 | [diff] [blame] | 96 | self.elements = {} |
| 97 | self.__fixdict(self.__dict__) |
| 98 | self.__fixclass(self.__class__) |
| 99 | |
| 100 | def __fixclass(self, kl): |
| 101 | self.__fixdict(kl.__dict__) |
| 102 | for k in kl.__bases__: |
| 103 | self.__fixclass(k) |
| 104 | |
| 105 | def __fixdict(self, dict): |
Guido van Rossum | 2e3f7be | 1999-02-02 22:15:24 +0000 | [diff] [blame] | 106 | for key in dict.keys(): |
Guido van Rossum | fe64935 | 1999-02-01 15:35:13 +0000 | [diff] [blame] | 107 | if key[:6] == 'start_': |
Guido van Rossum | 2e3f7be | 1999-02-02 22:15:24 +0000 | [diff] [blame] | 108 | tag = key[6:] |
| 109 | start, end = self.elements.get(tag, (None, None)) |
Guido van Rossum | fe64935 | 1999-02-01 15:35:13 +0000 | [diff] [blame] | 110 | if start is None: |
Guido van Rossum | 2e3f7be | 1999-02-02 22:15:24 +0000 | [diff] [blame] | 111 | self.elements[tag] = getattr(self, key), end |
Guido van Rossum | fe64935 | 1999-02-01 15:35:13 +0000 | [diff] [blame] | 112 | elif key[:4] == 'end_': |
Guido van Rossum | 2e3f7be | 1999-02-02 22:15:24 +0000 | [diff] [blame] | 113 | tag = key[4:] |
| 114 | start, end = self.elements.get(tag, (None, None)) |
Guido van Rossum | fe64935 | 1999-02-01 15:35:13 +0000 | [diff] [blame] | 115 | if end is None: |
Guido van Rossum | 2e3f7be | 1999-02-02 22:15:24 +0000 | [diff] [blame] | 116 | self.elements[tag] = start, getattr(self, key) |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 117 | |
| 118 | # Interface -- reset this instance. Loses all unprocessed data |
| 119 | def reset(self): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 120 | self.rawdata = '' |
| 121 | self.stack = [] |
| 122 | self.nomoretags = 0 |
| 123 | self.literal = 0 |
| 124 | self.lineno = 1 |
| 125 | self.__at_start = 1 |
| 126 | self.__seen_doctype = None |
| 127 | self.__seen_starttag = 0 |
Guido van Rossum | 91c8f59 | 1998-12-21 18:15:28 +0000 | [diff] [blame] | 128 | self.__use_namespaces = 0 |
Guido van Rossum | b083a9f | 1998-12-18 20:17:13 +0000 | [diff] [blame] | 129 | self.__namespaces = {'xml':None} # xml is implicitly declared |
Guido van Rossum | cc2c291 | 1999-02-02 17:54:38 +0000 | [diff] [blame] | 130 | # backward compatipibility hack: if elements not overridden, |
| 131 | # fill it in ourselves |
| 132 | if self.elements is XMLParser.elements: |
| 133 | self.__fixelements() |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 134 | |
| 135 | # For derived classes only -- enter literal mode (CDATA) till EOF |
| 136 | def setnomoretags(self): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 137 | self.nomoretags = self.literal = 1 |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 138 | |
| 139 | # For derived classes only -- enter literal mode (CDATA) |
| 140 | def setliteral(self, *args): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 141 | self.literal = 1 |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 142 | |
| 143 | # Interface -- feed some data to the parser. Call this as |
| 144 | # often as you want, with as little or as much text as you |
| 145 | # want (may include '\n'). (This just saves the text, all the |
| 146 | # processing is done by goahead().) |
| 147 | def feed(self, data): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 148 | self.rawdata = self.rawdata + data |
| 149 | self.goahead(0) |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 150 | |
| 151 | # Interface -- handle the remaining data |
| 152 | def close(self): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 153 | self.goahead(1) |
Guido van Rossum | cc2c291 | 1999-02-02 17:54:38 +0000 | [diff] [blame] | 154 | if self.__fixed: |
| 155 | self.__fixed = 0 |
| 156 | # remove self.elements so that we don't leak |
| 157 | del self.elements |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 158 | |
| 159 | # Interface -- translate references |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 160 | def translate_references(self, data, all = 1): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 161 | i = 0 |
| 162 | while 1: |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 163 | res = amp.search(data, i) |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 164 | if res is None: |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 165 | return data |
| 166 | res = ref.match(data, res.start(0)) |
| 167 | if res is None: |
| 168 | self.syntax_error("bogus `&'") |
| 169 | i =i+1 |
| 170 | continue |
| 171 | i = res.end(0) |
| 172 | if data[i - 1] != ';': |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 173 | self.syntax_error("`;' missing after entity/char reference") |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 174 | i = i-1 |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 175 | str = res.group(1) |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 176 | pre = data[:res.start(0)] |
| 177 | post = data[i:] |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 178 | if str[0] == '#': |
| 179 | if str[1] == 'x': |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 180 | str = chr(string.atoi(str[2:], 16)) |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 181 | else: |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 182 | str = chr(string.atoi(str[1:])) |
| 183 | data = pre + str + post |
| 184 | i = res.start(0)+len(str) |
| 185 | elif all: |
| 186 | if self.entitydefs.has_key(str): |
| 187 | data = pre + self.entitydefs[str] + post |
| 188 | i = res.start(0) # rescan substituted text |
| 189 | else: |
Guido van Rossum | cc2c291 | 1999-02-02 17:54:38 +0000 | [diff] [blame] | 190 | self.syntax_error("reference to unknown entity `&%s;'" % str) |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 191 | # can't do it, so keep the entity ref in |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 192 | data = pre + '&' + str + ';' + post |
| 193 | i = res.start(0) + len(str) + 2 |
| 194 | else: |
| 195 | # just translating character references |
| 196 | pass # i is already postioned correctly |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 197 | |
| 198 | # Internal -- handle data as far as reasonable. May leave state |
| 199 | # and data to be processed by a subsequent call. If 'end' is |
| 200 | # true, force handling all data as if followed by EOF marker. |
| 201 | def goahead(self, end): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 202 | rawdata = self.rawdata |
| 203 | i = 0 |
| 204 | n = len(rawdata) |
| 205 | while i < n: |
| 206 | if i > 0: |
| 207 | self.__at_start = 0 |
| 208 | if self.nomoretags: |
| 209 | data = rawdata[i:n] |
| 210 | self.handle_data(data) |
| 211 | self.lineno = self.lineno + string.count(data, '\n') |
| 212 | i = n |
| 213 | break |
| 214 | res = interesting.search(rawdata, i) |
| 215 | if res: |
| 216 | j = res.start(0) |
| 217 | else: |
| 218 | j = n |
| 219 | if i < j: |
Guido van Rossum | 91c8f59 | 1998-12-21 18:15:28 +0000 | [diff] [blame] | 220 | data = rawdata[i:j] |
| 221 | if self.__at_start and space.match(data) is None: |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 222 | self.syntax_error('illegal data at start of file') |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 223 | self.__at_start = 0 |
Guido van Rossum | eeb2f32 | 1998-10-19 13:28:26 +0000 | [diff] [blame] | 224 | if not self.stack and space.match(data) is None: |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 225 | self.syntax_error('data not in content') |
| 226 | if illegal.search(data): |
| 227 | self.syntax_error('illegal character in content') |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 228 | self.handle_data(data) |
| 229 | self.lineno = self.lineno + string.count(data, '\n') |
| 230 | i = j |
| 231 | if i == n: break |
| 232 | if rawdata[i] == '<': |
| 233 | if starttagopen.match(rawdata, i): |
| 234 | if self.literal: |
| 235 | data = rawdata[i] |
| 236 | self.handle_data(data) |
| 237 | self.lineno = self.lineno + string.count(data, '\n') |
| 238 | i = i+1 |
| 239 | continue |
| 240 | k = self.parse_starttag(i) |
| 241 | if k < 0: break |
| 242 | self.__seen_starttag = 1 |
| 243 | self.lineno = self.lineno + string.count(rawdata[i:k], '\n') |
| 244 | i = k |
| 245 | continue |
| 246 | if endtagopen.match(rawdata, i): |
| 247 | k = self.parse_endtag(i) |
| 248 | if k < 0: break |
| 249 | self.lineno = self.lineno + string.count(rawdata[i:k], '\n') |
| 250 | i = k |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 251 | continue |
| 252 | if commentopen.match(rawdata, i): |
| 253 | if self.literal: |
| 254 | data = rawdata[i] |
| 255 | self.handle_data(data) |
| 256 | self.lineno = self.lineno + string.count(data, '\n') |
| 257 | i = i+1 |
| 258 | continue |
| 259 | k = self.parse_comment(i) |
| 260 | if k < 0: break |
| 261 | self.lineno = self.lineno + string.count(rawdata[i:k], '\n') |
| 262 | i = k |
| 263 | continue |
| 264 | if cdataopen.match(rawdata, i): |
| 265 | k = self.parse_cdata(i) |
| 266 | if k < 0: break |
| 267 | self.lineno = self.lineno + string.count(rawdata[i:i], '\n') |
| 268 | i = k |
| 269 | continue |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 270 | res = xmldecl.match(rawdata, i) |
| 271 | if res: |
| 272 | if not self.__at_start: |
| 273 | self.syntax_error("<?xml?> declaration not at start of document") |
| 274 | version, encoding, standalone = res.group('version', |
| 275 | 'encoding', |
| 276 | 'standalone') |
| 277 | if version[1:-1] != '1.0': |
| 278 | raise RuntimeError, 'only XML version 1.0 supported' |
| 279 | if encoding: encoding = encoding[1:-1] |
| 280 | if standalone: standalone = standalone[1:-1] |
| 281 | self.handle_xml(encoding, standalone) |
| 282 | i = res.end(0) |
| 283 | continue |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 284 | res = procopen.match(rawdata, i) |
| 285 | if res: |
| 286 | k = self.parse_proc(i) |
| 287 | if k < 0: break |
| 288 | self.lineno = self.lineno + string.count(rawdata[i:k], '\n') |
| 289 | i = k |
| 290 | continue |
| 291 | res = doctype.match(rawdata, i) |
| 292 | if res: |
| 293 | if self.literal: |
| 294 | data = rawdata[i] |
| 295 | self.handle_data(data) |
| 296 | self.lineno = self.lineno + string.count(data, '\n') |
| 297 | i = i+1 |
| 298 | continue |
| 299 | if self.__seen_doctype: |
| 300 | self.syntax_error('multiple DOCTYPE elements') |
| 301 | if self.__seen_starttag: |
| 302 | self.syntax_error('DOCTYPE not at beginning of document') |
| 303 | k = self.parse_doctype(res) |
| 304 | if k < 0: break |
| 305 | self.__seen_doctype = res.group('name') |
| 306 | self.lineno = self.lineno + string.count(rawdata[i:k], '\n') |
| 307 | i = k |
| 308 | continue |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 309 | elif rawdata[i] == '&': |
Guido van Rossum | 204b65c | 1998-12-08 13:25:00 +0000 | [diff] [blame] | 310 | if self.literal: |
| 311 | data = rawdata[i] |
| 312 | self.handle_data(data) |
| 313 | i = i+1 |
| 314 | continue |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 315 | res = charref.match(rawdata, i) |
| 316 | if res is not None: |
| 317 | i = res.end(0) |
| 318 | if rawdata[i-1] != ';': |
| 319 | self.syntax_error("`;' missing in charref") |
| 320 | i = i-1 |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 321 | if not self.stack: |
| 322 | self.syntax_error('data not in content') |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 323 | self.handle_charref(res.group('char')[:-1]) |
| 324 | self.lineno = self.lineno + string.count(res.group(0), '\n') |
| 325 | continue |
| 326 | res = entityref.match(rawdata, i) |
| 327 | if res is not None: |
| 328 | i = res.end(0) |
| 329 | if rawdata[i-1] != ';': |
| 330 | self.syntax_error("`;' missing in entityref") |
| 331 | i = i-1 |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 332 | name = res.group('name') |
| 333 | if self.entitydefs.has_key(name): |
| 334 | self.rawdata = rawdata = rawdata[:res.start(0)] + self.entitydefs[name] + rawdata[i:] |
| 335 | n = len(rawdata) |
| 336 | i = res.start(0) |
| 337 | else: |
Guido van Rossum | cc2c291 | 1999-02-02 17:54:38 +0000 | [diff] [blame] | 338 | self.syntax_error("reference to unknown entity `&%s;'" % name) |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 339 | self.unknown_entityref(name) |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 340 | self.lineno = self.lineno + string.count(res.group(0), '\n') |
| 341 | continue |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 342 | elif rawdata[i] == ']': |
Guido van Rossum | 204b65c | 1998-12-08 13:25:00 +0000 | [diff] [blame] | 343 | if self.literal: |
| 344 | data = rawdata[i] |
| 345 | self.handle_data(data) |
| 346 | i = i+1 |
| 347 | continue |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 348 | if n-i < 3: |
| 349 | break |
| 350 | if cdataclose.match(rawdata, i): |
| 351 | self.syntax_error("bogus `]]>'") |
| 352 | self.handle_data(rawdata[i]) |
| 353 | i = i+1 |
| 354 | continue |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 355 | else: |
| 356 | raise RuntimeError, 'neither < nor & ??' |
| 357 | # We get here only if incomplete matches but |
| 358 | # nothing else |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 359 | break |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 360 | # end while |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 361 | if i > 0: |
| 362 | self.__at_start = 0 |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 363 | if end and i < n: |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 364 | data = rawdata[i] |
| 365 | self.syntax_error("bogus `%s'" % data) |
| 366 | if illegal.search(data): |
| 367 | self.syntax_error('illegal character in content') |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 368 | self.handle_data(data) |
| 369 | self.lineno = self.lineno + string.count(data, '\n') |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 370 | self.rawdata = rawdata[i+1:] |
| 371 | return self.goahead(end) |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 372 | self.rawdata = rawdata[i:] |
| 373 | if end: |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 374 | if not self.__seen_starttag: |
| 375 | self.syntax_error('no elements in file') |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 376 | if self.stack: |
| 377 | self.syntax_error('missing end tags') |
| 378 | while self.stack: |
Guido van Rossum | b083a9f | 1998-12-18 20:17:13 +0000 | [diff] [blame] | 379 | self.finish_endtag(self.stack[-1][0]) |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 380 | |
| 381 | # Internal -- parse comment, return length or -1 if not terminated |
| 382 | def parse_comment(self, i): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 383 | rawdata = self.rawdata |
| 384 | if rawdata[i:i+4] <> '<!--': |
| 385 | raise RuntimeError, 'unexpected call to handle_comment' |
| 386 | res = commentclose.search(rawdata, i+4) |
Guido van Rossum | eeb2f32 | 1998-10-19 13:28:26 +0000 | [diff] [blame] | 387 | if res is None: |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 388 | return -1 |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 389 | if doubledash.search(rawdata, i+4, res.start(0)): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 390 | self.syntax_error("`--' inside comment") |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 391 | if rawdata[res.start(0)-1] == '-': |
| 392 | self.syntax_error('comment cannot end in three dashes') |
| 393 | if illegal.search(rawdata, i+4, res.start(0)): |
| 394 | self.syntax_error('illegal character in comment') |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 395 | self.handle_comment(rawdata[i+4: res.start(0)]) |
| 396 | return res.end(0) |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 397 | |
Guido van Rossum | 02505e4 | 1998-01-29 14:55:24 +0000 | [diff] [blame] | 398 | # Internal -- handle DOCTYPE tag, return length or -1 if not terminated |
| 399 | def parse_doctype(self, res): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 400 | rawdata = self.rawdata |
| 401 | n = len(rawdata) |
| 402 | name = res.group('name') |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 403 | pubid, syslit = res.group('pubid', 'syslit') |
| 404 | if pubid is not None: |
| 405 | pubid = pubid[1:-1] # remove quotes |
| 406 | pubid = string.join(string.split(pubid)) # normalize |
| 407 | if syslit is not None: syslit = syslit[1:-1] # remove quotes |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 408 | j = k = res.end(0) |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 409 | if k >= n: |
| 410 | return -1 |
| 411 | if rawdata[k] == '[': |
| 412 | level = 0 |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 413 | k = k+1 |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 414 | dq = sq = 0 |
| 415 | while k < n: |
| 416 | c = rawdata[k] |
| 417 | if not sq and c == '"': |
| 418 | dq = not dq |
| 419 | elif not dq and c == "'": |
| 420 | sq = not sq |
| 421 | elif sq or dq: |
| 422 | pass |
| 423 | elif level <= 0 and c == ']': |
| 424 | res = endbracket.match(rawdata, k+1) |
Guido van Rossum | eeb2f32 | 1998-10-19 13:28:26 +0000 | [diff] [blame] | 425 | if res is None: |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 426 | return -1 |
| 427 | self.handle_doctype(name, pubid, syslit, rawdata[j+1:k]) |
| 428 | return res.end(0) |
| 429 | elif c == '<': |
| 430 | level = level + 1 |
| 431 | elif c == '>': |
| 432 | level = level - 1 |
| 433 | if level < 0: |
| 434 | self.syntax_error("bogus `>' in DOCTYPE") |
| 435 | k = k+1 |
Guido van Rossum | eeb2f32 | 1998-10-19 13:28:26 +0000 | [diff] [blame] | 436 | res = endbracketfind.match(rawdata, k) |
| 437 | if res is None: |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 438 | return -1 |
Guido van Rossum | eeb2f32 | 1998-10-19 13:28:26 +0000 | [diff] [blame] | 439 | if endbracket.match(rawdata, k) is None: |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 440 | self.syntax_error('garbage in DOCTYPE') |
| 441 | self.handle_doctype(name, pubid, syslit, None) |
| 442 | return res.end(0) |
Guido van Rossum | 02505e4 | 1998-01-29 14:55:24 +0000 | [diff] [blame] | 443 | |
| 444 | # Internal -- handle CDATA tag, return length or -1 if not terminated |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 445 | def parse_cdata(self, i): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 446 | rawdata = self.rawdata |
| 447 | if rawdata[i:i+9] <> '<![CDATA[': |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 448 | raise RuntimeError, 'unexpected call to parse_cdata' |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 449 | res = cdataclose.search(rawdata, i+9) |
Guido van Rossum | eeb2f32 | 1998-10-19 13:28:26 +0000 | [diff] [blame] | 450 | if res is None: |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 451 | return -1 |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 452 | if illegal.search(rawdata, i+9, res.start(0)): |
| 453 | self.syntax_error('illegal character in CDATA') |
| 454 | if not self.stack: |
| 455 | self.syntax_error('CDATA not in content') |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 456 | self.handle_cdata(rawdata[i+9:res.start(0)]) |
| 457 | return res.end(0) |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 458 | |
Guido van Rossum | b083a9f | 1998-12-18 20:17:13 +0000 | [diff] [blame] | 459 | __xml_namespace_attributes = {'ns':None, 'src':None, 'prefix':None} |
Guido van Rossum | 02505e4 | 1998-01-29 14:55:24 +0000 | [diff] [blame] | 460 | # Internal -- handle a processing instruction tag |
| 461 | def parse_proc(self, i): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 462 | rawdata = self.rawdata |
| 463 | end = procclose.search(rawdata, i) |
Guido van Rossum | eeb2f32 | 1998-10-19 13:28:26 +0000 | [diff] [blame] | 464 | if end is None: |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 465 | return -1 |
| 466 | j = end.start(0) |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 467 | if illegal.search(rawdata, i+2, j): |
| 468 | self.syntax_error('illegal character in processing instruction') |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 469 | res = tagfind.match(rawdata, i+2) |
Guido van Rossum | eeb2f32 | 1998-10-19 13:28:26 +0000 | [diff] [blame] | 470 | if res is None: |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 471 | raise RuntimeError, 'unexpected call to parse_proc' |
| 472 | k = res.end(0) |
| 473 | name = res.group(0) |
Guido van Rossum | b083a9f | 1998-12-18 20:17:13 +0000 | [diff] [blame] | 474 | if name == 'xml:namespace': |
| 475 | self.syntax_error('old-fashioned namespace declaration') |
Guido van Rossum | 91c8f59 | 1998-12-21 18:15:28 +0000 | [diff] [blame] | 476 | self.__use_namespaces = -1 |
Guido van Rossum | b083a9f | 1998-12-18 20:17:13 +0000 | [diff] [blame] | 477 | # namespace declaration |
| 478 | # this must come after the <?xml?> declaration (if any) |
| 479 | # and before the <!DOCTYPE> (if any). |
| 480 | if self.__seen_doctype or self.__seen_starttag: |
| 481 | self.syntax_error('xml:namespace declaration too late in document') |
| 482 | attrdict, namespace, k = self.parse_attributes(name, k, j) |
| 483 | if namespace: |
| 484 | self.syntax_error('namespace declaration inside namespace declaration') |
| 485 | for attrname in attrdict.keys(): |
| 486 | if not self.__xml_namespace_attributes.has_key(attrname): |
| 487 | self.syntax_error("unknown attribute `%s' in xml:namespace tag" % attrname) |
| 488 | if not attrdict.has_key('ns') or not attrdict.has_key('prefix'): |
| 489 | self.syntax_error('xml:namespace without required attributes') |
| 490 | prefix = attrdict.get('prefix') |
| 491 | if ncname.match(prefix) is None: |
| 492 | self.syntax_error('xml:namespace illegal prefix value') |
| 493 | return end.end(0) |
| 494 | if self.__namespaces.has_key(prefix): |
| 495 | self.syntax_error('xml:namespace prefix not unique') |
| 496 | self.__namespaces[prefix] = attrdict['ns'] |
| 497 | else: |
Guido van Rossum | 1f2e09b | 1999-06-08 21:23:26 +0000 | [diff] [blame^] | 498 | if string.lower(name) == 'xml': |
Guido van Rossum | b083a9f | 1998-12-18 20:17:13 +0000 | [diff] [blame] | 499 | self.syntax_error('illegal processing instruction target name') |
| 500 | self.handle_proc(name, rawdata[k:j]) |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 501 | return end.end(0) |
Guido van Rossum | 02505e4 | 1998-01-29 14:55:24 +0000 | [diff] [blame] | 502 | |
| 503 | # Internal -- parse attributes between i and j |
Guido van Rossum | b083a9f | 1998-12-18 20:17:13 +0000 | [diff] [blame] | 504 | def parse_attributes(self, tag, i, j): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 505 | rawdata = self.rawdata |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 506 | attrdict = {} |
Guido van Rossum | b083a9f | 1998-12-18 20:17:13 +0000 | [diff] [blame] | 507 | namespace = {} |
Guido van Rossum | eeb2f32 | 1998-10-19 13:28:26 +0000 | [diff] [blame] | 508 | while i < j: |
| 509 | res = attrfind.match(rawdata, i) |
| 510 | if res is None: |
| 511 | break |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 512 | attrname, attrvalue = res.group('name', 'value') |
Guido van Rossum | b083a9f | 1998-12-18 20:17:13 +0000 | [diff] [blame] | 513 | i = res.end(0) |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 514 | if attrvalue is None: |
Guido van Rossum | eeb2f32 | 1998-10-19 13:28:26 +0000 | [diff] [blame] | 515 | self.syntax_error("no value specified for attribute `%s'" % attrname) |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 516 | attrvalue = attrname |
| 517 | elif attrvalue[:1] == "'" == attrvalue[-1:] or \ |
| 518 | attrvalue[:1] == '"' == attrvalue[-1:]: |
| 519 | attrvalue = attrvalue[1:-1] |
| 520 | else: |
Guido van Rossum | eeb2f32 | 1998-10-19 13:28:26 +0000 | [diff] [blame] | 521 | self.syntax_error("attribute `%s' value not quoted" % attrname) |
Guido van Rossum | b083a9f | 1998-12-18 20:17:13 +0000 | [diff] [blame] | 522 | res = xmlns.match(attrname) |
| 523 | if res is not None: |
| 524 | # namespace declaration |
| 525 | ncname = res.group('ncname') |
| 526 | namespace[ncname or ''] = attrvalue or None |
Guido van Rossum | 91c8f59 | 1998-12-21 18:15:28 +0000 | [diff] [blame] | 527 | if not self.__use_namespaces: |
| 528 | self.__use_namespaces = len(self.stack)+1 |
Guido van Rossum | b083a9f | 1998-12-18 20:17:13 +0000 | [diff] [blame] | 529 | continue |
Guido van Rossum | eeb2f32 | 1998-10-19 13:28:26 +0000 | [diff] [blame] | 530 | if '<' in attrvalue: |
| 531 | self.syntax_error("`<' illegal in attribute value") |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 532 | if attrdict.has_key(attrname): |
Guido van Rossum | eeb2f32 | 1998-10-19 13:28:26 +0000 | [diff] [blame] | 533 | self.syntax_error("attribute `%s' specified twice" % attrname) |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 534 | attrvalue = string.translate(attrvalue, attrtrans) |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 535 | attrdict[attrname] = self.translate_references(attrvalue) |
Guido van Rossum | b083a9f | 1998-12-18 20:17:13 +0000 | [diff] [blame] | 536 | return attrdict, namespace, i |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 537 | |
| 538 | # Internal -- handle starttag, return length or -1 if not terminated |
| 539 | def parse_starttag(self, i): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 540 | rawdata = self.rawdata |
| 541 | # i points to start of tag |
Guido van Rossum | eeb2f32 | 1998-10-19 13:28:26 +0000 | [diff] [blame] | 542 | end = endbracketfind.match(rawdata, i+1) |
| 543 | if end is None: |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 544 | return -1 |
Guido van Rossum | eeb2f32 | 1998-10-19 13:28:26 +0000 | [diff] [blame] | 545 | tag = starttagmatch.match(rawdata, i) |
| 546 | if tag is None or tag.end(0) != end.end(0): |
| 547 | self.syntax_error('garbage in starttag') |
| 548 | return end.end(0) |
Guido van Rossum | b083a9f | 1998-12-18 20:17:13 +0000 | [diff] [blame] | 549 | nstag = tagname = tag.group('tagname') |
Guido van Rossum | eeb2f32 | 1998-10-19 13:28:26 +0000 | [diff] [blame] | 550 | if not self.__seen_starttag and self.__seen_doctype and \ |
| 551 | tagname != self.__seen_doctype: |
| 552 | self.syntax_error('starttag does not match DOCTYPE') |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 553 | if self.__seen_starttag and not self.stack: |
| 554 | self.syntax_error('multiple elements on top level') |
Guido van Rossum | eeb2f32 | 1998-10-19 13:28:26 +0000 | [diff] [blame] | 555 | k, j = tag.span('attrs') |
Guido van Rossum | b083a9f | 1998-12-18 20:17:13 +0000 | [diff] [blame] | 556 | attrdict, nsdict, k = self.parse_attributes(tagname, k, j) |
| 557 | self.stack.append((tagname, nsdict, nstag)) |
Guido van Rossum | 91c8f59 | 1998-12-21 18:15:28 +0000 | [diff] [blame] | 558 | if self.__use_namespaces: |
| 559 | res = qname.match(tagname) |
| 560 | else: |
| 561 | res = None |
Guido van Rossum | b083a9f | 1998-12-18 20:17:13 +0000 | [diff] [blame] | 562 | if res is not None: |
| 563 | prefix, nstag = res.group('prefix', 'local') |
| 564 | if prefix is None: |
| 565 | prefix = '' |
| 566 | ns = None |
| 567 | for t, d, nst in self.stack: |
| 568 | if d.has_key(prefix): |
| 569 | ns = d[prefix] |
| 570 | if ns is None and prefix != '': |
| 571 | ns = self.__namespaces.get(prefix) |
| 572 | if ns is not None: |
| 573 | nstag = ns + ' ' + nstag |
| 574 | elif prefix != '': |
| 575 | nstag = prefix + ':' + nstag # undo split |
| 576 | self.stack[-1] = tagname, nsdict, nstag |
| 577 | # translate namespace of attributes |
Guido van Rossum | 91c8f59 | 1998-12-21 18:15:28 +0000 | [diff] [blame] | 578 | if self.__use_namespaces: |
| 579 | nattrdict = {} |
| 580 | for key, val in attrdict.items(): |
| 581 | res = qname.match(key) |
| 582 | if res is not None: |
| 583 | aprefix, key = res.group('prefix', 'local') |
| 584 | if aprefix is None: |
| 585 | aprefix = '' |
| 586 | ans = None |
| 587 | for t, d, nst in self.stack: |
| 588 | if d.has_key(aprefix): |
| 589 | ans = d[aprefix] |
| 590 | if ans is None and aprefix != '': |
| 591 | ans = self.__namespaces.get(aprefix) |
| 592 | if ans is not None: |
| 593 | key = ans + ' ' + key |
| 594 | elif aprefix != '': |
| 595 | key = aprefix + ':' + key |
| 596 | elif ns is not None: |
| 597 | key = ns + ' ' + key |
| 598 | nattrdict[key] = val |
| 599 | attrdict = nattrdict |
Guido van Rossum | b083a9f | 1998-12-18 20:17:13 +0000 | [diff] [blame] | 600 | attributes = self.attributes.get(nstag) |
| 601 | if attributes is not None: |
| 602 | for key in attrdict.keys(): |
| 603 | if not attributes.has_key(key): |
| 604 | self.syntax_error("unknown attribute `%s' in tag `%s'" % (key, tagname)) |
| 605 | for key, val in attributes.items(): |
| 606 | if val is not None and not attrdict.has_key(key): |
| 607 | attrdict[key] = val |
| 608 | method = self.elements.get(nstag, (None, None))[0] |
| 609 | self.finish_starttag(nstag, attrdict, method) |
Guido van Rossum | eeb2f32 | 1998-10-19 13:28:26 +0000 | [diff] [blame] | 610 | if tag.group('slash') == '/': |
| 611 | self.finish_endtag(tagname) |
| 612 | return tag.end(0) |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 613 | |
| 614 | # Internal -- parse endtag |
| 615 | def parse_endtag(self, i): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 616 | rawdata = self.rawdata |
Guido van Rossum | eeb2f32 | 1998-10-19 13:28:26 +0000 | [diff] [blame] | 617 | end = endbracketfind.match(rawdata, i+1) |
| 618 | if end is None: |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 619 | return -1 |
| 620 | res = tagfind.match(rawdata, i+2) |
Guido van Rossum | eeb2f32 | 1998-10-19 13:28:26 +0000 | [diff] [blame] | 621 | if res is None: |
Guido van Rossum | f484a33 | 1998-12-07 21:59:56 +0000 | [diff] [blame] | 622 | if self.literal: |
| 623 | self.handle_data(rawdata[i]) |
| 624 | return i+1 |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 625 | self.syntax_error('no name specified in end tag') |
| 626 | tag = '' |
| 627 | k = i+2 |
| 628 | else: |
| 629 | tag = res.group(0) |
Guido van Rossum | f484a33 | 1998-12-07 21:59:56 +0000 | [diff] [blame] | 630 | if self.literal: |
Guido van Rossum | b083a9f | 1998-12-18 20:17:13 +0000 | [diff] [blame] | 631 | if not self.stack or tag != self.stack[-1][0]: |
Guido van Rossum | f484a33 | 1998-12-07 21:59:56 +0000 | [diff] [blame] | 632 | self.handle_data(rawdata[i]) |
| 633 | return i+1 |
| 634 | self.literal = 0 |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 635 | k = res.end(0) |
Guido van Rossum | eeb2f32 | 1998-10-19 13:28:26 +0000 | [diff] [blame] | 636 | if endbracket.match(rawdata, k) is None: |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 637 | self.syntax_error('garbage in end tag') |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 638 | self.finish_endtag(tag) |
| 639 | return end.end(0) |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 640 | |
| 641 | # Internal -- finish processing of start tag |
Guido van Rossum | b083a9f | 1998-12-18 20:17:13 +0000 | [diff] [blame] | 642 | def finish_starttag(self, tagname, attrdict, method): |
| 643 | if method is not None: |
| 644 | self.handle_starttag(tagname, method, attrdict) |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 645 | else: |
Guido van Rossum | b083a9f | 1998-12-18 20:17:13 +0000 | [diff] [blame] | 646 | self.unknown_starttag(tagname, attrdict) |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 647 | |
| 648 | # Internal -- finish processing of end tag |
| 649 | def finish_endtag(self, tag): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 650 | if not tag: |
| 651 | self.syntax_error('name-less end tag') |
| 652 | found = len(self.stack) - 1 |
| 653 | if found < 0: |
| 654 | self.unknown_endtag(tag) |
| 655 | return |
| 656 | else: |
Guido van Rossum | b083a9f | 1998-12-18 20:17:13 +0000 | [diff] [blame] | 657 | found = -1 |
| 658 | for i in range(len(self.stack)): |
| 659 | if tag == self.stack[i][0]: |
| 660 | found = i |
| 661 | if found == -1: |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 662 | self.syntax_error('unopened end tag') |
Guido van Rossum | b083a9f | 1998-12-18 20:17:13 +0000 | [diff] [blame] | 663 | method = self.elements.get(tag, (None, None))[1] |
| 664 | if method is not None: |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 665 | self.handle_endtag(tag, method) |
| 666 | else: |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 667 | self.unknown_endtag(tag) |
| 668 | return |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 669 | while len(self.stack) > found: |
| 670 | if found < len(self.stack) - 1: |
Guido van Rossum | b083a9f | 1998-12-18 20:17:13 +0000 | [diff] [blame] | 671 | self.syntax_error('missing close tag for %s' % self.stack[-1][2]) |
| 672 | nstag = self.stack[-1][2] |
| 673 | method = self.elements.get(nstag, (None, None))[1] |
| 674 | if method is not None: |
| 675 | self.handle_endtag(nstag, method) |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 676 | else: |
Guido van Rossum | b083a9f | 1998-12-18 20:17:13 +0000 | [diff] [blame] | 677 | self.unknown_endtag(nstag) |
Guido van Rossum | 91c8f59 | 1998-12-21 18:15:28 +0000 | [diff] [blame] | 678 | if self.__use_namespaces == len(self.stack): |
| 679 | self.__use_namespaces = 0 |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 680 | del self.stack[-1] |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 681 | |
Guido van Rossum | 02505e4 | 1998-01-29 14:55:24 +0000 | [diff] [blame] | 682 | # Overridable -- handle xml processing instruction |
| 683 | def handle_xml(self, encoding, standalone): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 684 | pass |
Guido van Rossum | 02505e4 | 1998-01-29 14:55:24 +0000 | [diff] [blame] | 685 | |
| 686 | # Overridable -- handle DOCTYPE |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 687 | def handle_doctype(self, tag, pubid, syslit, data): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 688 | pass |
Guido van Rossum | 02505e4 | 1998-01-29 14:55:24 +0000 | [diff] [blame] | 689 | |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 690 | # Overridable -- handle start tag |
| 691 | def handle_starttag(self, tag, method, attrs): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 692 | method(attrs) |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 693 | |
| 694 | # Overridable -- handle end tag |
| 695 | def handle_endtag(self, tag, method): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 696 | method() |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 697 | |
| 698 | # Example -- handle character reference, no need to override |
| 699 | def handle_charref(self, name): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 700 | try: |
| 701 | if name[0] == 'x': |
| 702 | n = string.atoi(name[1:], 16) |
| 703 | else: |
| 704 | n = string.atoi(name) |
| 705 | except string.atoi_error: |
| 706 | self.unknown_charref(name) |
| 707 | return |
| 708 | if not 0 <= n <= 255: |
| 709 | self.unknown_charref(name) |
| 710 | return |
| 711 | self.handle_data(chr(n)) |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 712 | |
| 713 | # Definition of entities -- derived classes may override |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 714 | entitydefs = {'lt': '<', # must use charref |
| 715 | 'gt': '>', |
| 716 | 'amp': '&', # must use charref |
| 717 | 'quot': '"', |
| 718 | 'apos': ''', |
| 719 | } |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 720 | |
| 721 | # Example -- handle entity reference, no need to override |
| 722 | def handle_entityref(self, name): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 723 | table = self.entitydefs |
| 724 | if table.has_key(name): |
| 725 | self.handle_data(table[name]) |
| 726 | else: |
| 727 | self.unknown_entityref(name) |
| 728 | return |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 729 | |
| 730 | # Example -- handle data, should be overridden |
| 731 | def handle_data(self, data): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 732 | pass |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 733 | |
| 734 | # Example -- handle cdata, could be overridden |
| 735 | def handle_cdata(self, data): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 736 | pass |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 737 | |
| 738 | # Example -- handle comment, could be overridden |
| 739 | def handle_comment(self, data): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 740 | pass |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 741 | |
| 742 | # Example -- handle processing instructions, could be overridden |
| 743 | def handle_proc(self, name, data): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 744 | pass |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 745 | |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 746 | # Example -- handle relatively harmless syntax errors, could be overridden |
Guido van Rossum | 02505e4 | 1998-01-29 14:55:24 +0000 | [diff] [blame] | 747 | def syntax_error(self, message): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 748 | raise RuntimeError, 'Syntax error at line %d: %s' % (self.lineno, message) |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 749 | |
| 750 | # To be overridden -- handlers for unknown objects |
| 751 | def unknown_starttag(self, tag, attrs): pass |
| 752 | def unknown_endtag(self, tag): pass |
| 753 | def unknown_charref(self, ref): pass |
| 754 | def unknown_entityref(self, ref): pass |
| 755 | |
| 756 | |
| 757 | class TestXMLParser(XMLParser): |
| 758 | |
Guido van Rossum | b083a9f | 1998-12-18 20:17:13 +0000 | [diff] [blame] | 759 | def __init__(self): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 760 | self.testdata = "" |
Guido van Rossum | b083a9f | 1998-12-18 20:17:13 +0000 | [diff] [blame] | 761 | XMLParser.__init__(self) |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 762 | |
Guido van Rossum | 02505e4 | 1998-01-29 14:55:24 +0000 | [diff] [blame] | 763 | def handle_xml(self, encoding, standalone): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 764 | self.flush() |
| 765 | print 'xml: encoding =',encoding,'standalone =',standalone |
Guido van Rossum | 02505e4 | 1998-01-29 14:55:24 +0000 | [diff] [blame] | 766 | |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 767 | def handle_doctype(self, tag, pubid, syslit, data): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 768 | self.flush() |
| 769 | print 'DOCTYPE:',tag, `data` |
Guido van Rossum | 02505e4 | 1998-01-29 14:55:24 +0000 | [diff] [blame] | 770 | |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 771 | def handle_entity(self, name, strval, pubid, syslit, ndata): |
| 772 | self.flush() |
| 773 | print 'ENTITY:',`data` |
| 774 | |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 775 | def handle_data(self, data): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 776 | self.testdata = self.testdata + data |
| 777 | if len(`self.testdata`) >= 70: |
| 778 | self.flush() |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 779 | |
| 780 | def flush(self): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 781 | data = self.testdata |
| 782 | if data: |
| 783 | self.testdata = "" |
| 784 | print 'data:', `data` |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 785 | |
| 786 | def handle_cdata(self, data): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 787 | self.flush() |
| 788 | print 'cdata:', `data` |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 789 | |
| 790 | def handle_proc(self, name, data): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 791 | self.flush() |
| 792 | print 'processing:',name,`data` |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 793 | |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 794 | def handle_comment(self, data): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 795 | self.flush() |
| 796 | r = `data` |
| 797 | if len(r) > 68: |
| 798 | r = r[:32] + '...' + r[-32:] |
| 799 | print 'comment:', r |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 800 | |
Guido van Rossum | 02505e4 | 1998-01-29 14:55:24 +0000 | [diff] [blame] | 801 | def syntax_error(self, message): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 802 | print 'error at line %d:' % self.lineno, message |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 803 | |
| 804 | def unknown_starttag(self, tag, attrs): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 805 | self.flush() |
| 806 | if not attrs: |
| 807 | print 'start tag: <' + tag + '>' |
| 808 | else: |
| 809 | print 'start tag: <' + tag, |
| 810 | for name, value in attrs.items(): |
| 811 | print name + '=' + '"' + value + '"', |
| 812 | print '>' |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 813 | |
| 814 | def unknown_endtag(self, tag): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 815 | self.flush() |
| 816 | print 'end tag: </' + tag + '>' |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 817 | |
| 818 | def unknown_entityref(self, ref): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 819 | self.flush() |
| 820 | print '*** unknown entity ref: &' + ref + ';' |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 821 | |
| 822 | def unknown_charref(self, ref): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 823 | self.flush() |
| 824 | print '*** unknown char ref: &#' + ref + ';' |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 825 | |
| 826 | def close(self): |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 827 | XMLParser.close(self) |
| 828 | self.flush() |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 829 | |
| 830 | def test(args = None): |
Guido van Rossum | af5add4 | 1998-12-22 14:23:50 +0000 | [diff] [blame] | 831 | import sys, getopt |
| 832 | from time import time |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 833 | |
| 834 | if not args: |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 835 | args = sys.argv[1:] |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 836 | |
Guido van Rossum | af5add4 | 1998-12-22 14:23:50 +0000 | [diff] [blame] | 837 | opts, args = getopt.getopt(args, 'st') |
| 838 | klass = TestXMLParser |
| 839 | do_time = 0 |
| 840 | for o, a in opts: |
| 841 | if o == '-s': |
| 842 | klass = XMLParser |
| 843 | elif o == '-t': |
| 844 | do_time = 1 |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 845 | |
| 846 | if args: |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 847 | file = args[0] |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 848 | else: |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 849 | file = 'test.xml' |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 850 | |
| 851 | if file == '-': |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 852 | f = sys.stdin |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 853 | else: |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 854 | try: |
| 855 | f = open(file, 'r') |
| 856 | except IOError, msg: |
| 857 | print file, ":", msg |
| 858 | sys.exit(1) |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 859 | |
| 860 | data = f.read() |
| 861 | if f is not sys.stdin: |
Guido van Rossum | 45e2fbc | 1998-03-26 21:13:24 +0000 | [diff] [blame] | 862 | f.close() |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 863 | |
| 864 | x = klass() |
Guido van Rossum | af5add4 | 1998-12-22 14:23:50 +0000 | [diff] [blame] | 865 | t0 = time() |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 866 | try: |
Guido van Rossum | af5add4 | 1998-12-22 14:23:50 +0000 | [diff] [blame] | 867 | if do_time: |
| 868 | x.feed(data) |
| 869 | x.close() |
| 870 | else: |
| 871 | for c in data: |
| 872 | x.feed(c) |
| 873 | x.close() |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 874 | except RuntimeError, msg: |
Guido van Rossum | af5add4 | 1998-12-22 14:23:50 +0000 | [diff] [blame] | 875 | t1 = time() |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 876 | print msg |
Guido van Rossum | af5add4 | 1998-12-22 14:23:50 +0000 | [diff] [blame] | 877 | if do_time: |
| 878 | print 'total time: %g' % (t1-t0) |
Guido van Rossum | 7e07b38 | 1998-04-03 16:02:39 +0000 | [diff] [blame] | 879 | sys.exit(1) |
Guido van Rossum | af5add4 | 1998-12-22 14:23:50 +0000 | [diff] [blame] | 880 | t1 = time() |
| 881 | if do_time: |
| 882 | print 'total time: %g' % (t1-t0) |
Guido van Rossum | a219efa | 1997-11-18 15:09:54 +0000 | [diff] [blame] | 883 | |
| 884 | |
| 885 | if __name__ == '__main__': |
| 886 | test() |