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