1 # -*- coding: utf-8 -*- |
|
2 |
|
3 # Copyright (c) 2004 - 2010 Detlev Offenbach <detlev@die-offenbachs.de> |
|
4 # |
|
5 |
|
6 """ |
|
7 Module implementing a base class for all of eric5s XML handlers. |
|
8 """ |
|
9 |
|
10 import pickle |
|
11 import base64 |
|
12 |
|
13 from xml.sax.handler import ContentHandler |
|
14 |
|
15 class XMLHandlerBase(ContentHandler): |
|
16 """ |
|
17 Class implementing the base class for al of eric5s XML handlers. |
|
18 """ |
|
19 def __init__(self): |
|
20 """ |
|
21 Constructor |
|
22 """ |
|
23 self.startDocumentSpecific = None |
|
24 |
|
25 self.elements = { |
|
26 'none' : (self.defaultStartElement, self.endNone), |
|
27 'int' : (self.defaultStartElement, self.endInt), |
|
28 'float' : (self.defaultStartElement, self.endFloat), |
|
29 'complex' : (self.defaultStartElement, self.endComplex), |
|
30 'bool' : (self.defaultStartElement, self.endBool), |
|
31 'string' : (self.defaultStartElement, self.endString), |
|
32 'bytes' : (self.defaultStartElement, self.endBytes), |
|
33 'bytearray' : (self.defaultStartElement, self.endBytearray), |
|
34 'tuple' : (self.startTuple, self.endTuple), |
|
35 'list' : (self.startList, self.endList), |
|
36 'dict' : (self.startDictionary, self.endDictionary), |
|
37 'set' : (self.startSet, self.endSet), |
|
38 'frozenset' : (self.startFrozenset, self.endFrozenset), |
|
39 'pickle' : (self.startPickle, self.endPickle), |
|
40 # for backward compatibility |
|
41 'long' : (self.defaultStartElement, self.endInt), |
|
42 'unicode' : (self.defaultStartElement, self.endString), |
|
43 } |
|
44 |
|
45 self.buffer = "" |
|
46 self.stack = [] |
|
47 self._marker = '__MARKER__' |
|
48 |
|
49 self.NEWPARA = chr(0x2029) |
|
50 self.NEWLINE = chr(0x2028) |
|
51 |
|
52 def unescape(self, text, attribute = False): |
|
53 """ |
|
54 Public method used to unescape certain characters. |
|
55 |
|
56 @param text the text to unescape (string) |
|
57 @param attribute flag indicating unescaping is done for an attribute |
|
58 """ |
|
59 if attribute: |
|
60 return text.replace(""",'"').replace(">",">")\ |
|
61 .replace("<","<").replace("&","&") |
|
62 else: |
|
63 return text.replace(">",">").replace("<","<").replace("&","&") |
|
64 |
|
65 def decodedNewLines(self, text): |
|
66 """ |
|
67 Public method to decode newlines and paragraph breaks. |
|
68 |
|
69 @param text text to decode (string) |
|
70 """ |
|
71 return text.replace(self.NEWPARA, "\n\n").replace(self.NEWLINE, "\n") |
|
72 |
|
73 def startDocument(self): |
|
74 """ |
|
75 Handler called, when the document parsing is started. |
|
76 """ |
|
77 self.buffer = "" |
|
78 if self.startDocumentSpecific is not None: |
|
79 self.startDocumentSpecific() |
|
80 |
|
81 def startElement(self, name, attrs): |
|
82 """ |
|
83 Handler called, when a starting tag is found. |
|
84 |
|
85 @param name name of the tag (string) |
|
86 @param attrs list of tag attributes |
|
87 """ |
|
88 try: |
|
89 self.elements[name][0](attrs) |
|
90 except KeyError: |
|
91 pass |
|
92 |
|
93 def endElement(self, name): |
|
94 """ |
|
95 Handler called, when an ending tag is found. |
|
96 |
|
97 @param name name of the tag (string) |
|
98 """ |
|
99 try: |
|
100 self.elements[name][1]() |
|
101 except KeyError: |
|
102 pass |
|
103 |
|
104 def characters(self, chars): |
|
105 """ |
|
106 Handler called for ordinary text. |
|
107 |
|
108 @param chars the scanned text (string) |
|
109 """ |
|
110 self.buffer += chars |
|
111 |
|
112 def defaultStartElement(self, attrs): |
|
113 """ |
|
114 Handler method for common start tags. |
|
115 |
|
116 @param attrs list of tag attributes |
|
117 """ |
|
118 self.buffer = "" |
|
119 |
|
120 def defaultEndElement(self): |
|
121 """ |
|
122 Handler method for the common end tags. |
|
123 """ |
|
124 pass |
|
125 |
|
126 def _prepareBasics(self): |
|
127 """ |
|
128 Protected method to prepare the parsing of XML for basic python types. |
|
129 """ |
|
130 self.stack = [] |
|
131 |
|
132 ############################################################################ |
|
133 ## The various handler methods for basic types |
|
134 ############################################################################ |
|
135 |
|
136 def endNone(self): |
|
137 """ |
|
138 Handler method for the "none" end tag. |
|
139 """ |
|
140 self.stack.append(None) |
|
141 |
|
142 def endInt(self): |
|
143 """ |
|
144 Handler method for the "int" end tag. |
|
145 """ |
|
146 self.stack.append(int(self.buffer.strip())) |
|
147 |
|
148 def endBool(self): |
|
149 """ |
|
150 Handler method for the "bool" end tag. |
|
151 """ |
|
152 if self.buffer.strip() == "True": |
|
153 self.stack.append(True) |
|
154 else: |
|
155 self.stack.append(False) |
|
156 |
|
157 def endFloat(self): |
|
158 """ |
|
159 Handler method for the "float" end tag. |
|
160 """ |
|
161 self.stack.append(float(self.buffer.strip())) |
|
162 |
|
163 def endComplex(self): |
|
164 """ |
|
165 Handler method for the "complex" end tag. |
|
166 """ |
|
167 real, imag = self.buffer.strip().split() |
|
168 self.stack.append(float(real) + float(imag)*1j) |
|
169 |
|
170 def endString(self): |
|
171 """ |
|
172 Handler method for the "string" end tag. |
|
173 """ |
|
174 s = str(self.unescape(self.buffer)) |
|
175 self.stack.append(s) |
|
176 |
|
177 def endBytes(self): |
|
178 """ |
|
179 Handler method for the "bytes" end tag. |
|
180 """ |
|
181 by = bytes([int(b) for b in self.buffer.strip().split(",")]) |
|
182 self.stack.append(by) |
|
183 |
|
184 def endBytearray(self): |
|
185 """ |
|
186 Handler method for the "bytearray" end tag. |
|
187 """ |
|
188 by = bytearray([int(b) for b in self.buffer.strip().split(",")]) |
|
189 self.stack.append(by) |
|
190 |
|
191 def startList(self, attrs): |
|
192 """ |
|
193 Handler method for the "list" start tag. |
|
194 |
|
195 @param attrs list of tag attributes |
|
196 """ |
|
197 self.stack.append(self._marker) |
|
198 |
|
199 def endList(self): |
|
200 """ |
|
201 Handler method for the "list" end tag. |
|
202 """ |
|
203 for i in range(len(self.stack) - 1, -1, -1): |
|
204 if self.stack[i] is self._marker: |
|
205 break |
|
206 assert i != -1 |
|
207 l = self.stack[i + 1:len(self.stack)] |
|
208 self.stack[i:] = [l] |
|
209 |
|
210 def startTuple(self, attrs): |
|
211 """ |
|
212 Handler method for the "tuple" start tag. |
|
213 |
|
214 @param attrs list of tag attributes |
|
215 """ |
|
216 self.stack.append(self._marker) |
|
217 |
|
218 def endTuple(self): |
|
219 """ |
|
220 Handler method for the "tuple" end tag. |
|
221 """ |
|
222 for i in range(len(self.stack) - 1, -1, -1): |
|
223 if self.stack[i] is self._marker: |
|
224 break |
|
225 assert i != -1 |
|
226 t = tuple(self.stack[i + 1:len(self.stack)]) |
|
227 self.stack[i:] = [t] |
|
228 |
|
229 def startDictionary(self, attrs): |
|
230 """ |
|
231 Handler method for the "dictionary" start tag. |
|
232 |
|
233 @param attrs list of tag attributes |
|
234 """ |
|
235 self.stack.append(self._marker) |
|
236 |
|
237 def endDictionary(self): |
|
238 """ |
|
239 Handler method for the "dictionary" end tag. |
|
240 """ |
|
241 for i in range(len(self.stack) - 1, -1, -1): |
|
242 if self.stack[i] is self._marker: |
|
243 break |
|
244 assert i != -1 |
|
245 d = {} |
|
246 for j in range(i + 1, len(self.stack), 2): |
|
247 d[self.stack[j]] = self.stack[j + 1] |
|
248 self.stack[i:] = [d] |
|
249 |
|
250 def startSet(self, attrs): |
|
251 """ |
|
252 Handler method for the "set" start tag. |
|
253 |
|
254 @param attrs list of tag attributes |
|
255 """ |
|
256 self.stack.append(self._marker) |
|
257 |
|
258 def endSet(self): |
|
259 """ |
|
260 Handler method for the "set" end tag. |
|
261 """ |
|
262 for i in range(len(self.stack) - 1, -1, -1): |
|
263 if self.stack[i] is self._marker: |
|
264 break |
|
265 assert i != -1 |
|
266 s = set(self.stack[i + 1:len(self.stack)]) |
|
267 self.stack[i:] = [s] |
|
268 |
|
269 def startFrozenset(self, attrs): |
|
270 """ |
|
271 Handler method for the "frozenset" start tag. |
|
272 |
|
273 @param attrs list of tag attributes |
|
274 """ |
|
275 self.stack.append(self._marker) |
|
276 |
|
277 def endFrozenset(self): |
|
278 """ |
|
279 Handler method for the "frozenset" end tag. |
|
280 """ |
|
281 for i in range(len(self.stack) - 1, -1, -1): |
|
282 if self.stack[i] is self._marker: |
|
283 break |
|
284 assert i != -1 |
|
285 f = frozenset(self.stack[i + 1:len(self.stack)]) |
|
286 self.stack[i:] = [f] |
|
287 |
|
288 def startPickle(self, attrs): |
|
289 """ |
|
290 Handler method for the "pickle" start tag. |
|
291 |
|
292 @param attrs list of tag attributes |
|
293 """ |
|
294 self.pickleEnc = attrs.get("encoding", "base64") |
|
295 |
|
296 def endPickle(self): |
|
297 """ |
|
298 Handler method for the "pickle" end tag. |
|
299 """ |
|
300 pic = base64.b64decode(self.buffer.encode("ASCII")) |
|
301 self.stack.append(pickle.loads(pic)) |
|