![[tokkee]](http://tokkee.org/images/avatar.png){kind=avatar}
e82df0bb608d783cdec364ba8de35ee369246371
1 #!/usr/bin/env python
2 # -*- coding: utf-8 -*-
4 # Scour
5 #
6 # Copyright 2009 Jeff Schiller
7 #
8 # This file is part of Scour, http://www.codedread.com/scour/
9 #
10 # Licensed under the Apache License, Version 2.0 (the "License");
11 # you may not use this file except in compliance with the License.
12 # You may obtain a copy of the License at
13 #
14 # http://www.apache.org/licenses/LICENSE-2.0
15 #
16 # Unless required by applicable law or agreed to in writing, software
17 # distributed under the License is distributed on an "AS IS" BASIS,
18 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
19 # See the License for the specific language governing permissions and
20 # limitations under the License.
22 # Notes:
24 # rubys' path-crunching ideas here: http://intertwingly.net/code/svgtidy/spec.rb
25 # (and implemented here: http://intertwingly.net/code/svgtidy/svgtidy.rb )
27 # Yet more ideas here: http://wiki.inkscape.org/wiki/index.php/Save_Cleaned_SVG
28 #
29 # * Process Transformations
30 # * Collapse all group based transformations
32 # Even more ideas here: http://esw.w3.org/topic/SvgTidy
33 # * analysis of path elements to see if rect can be used instead? (must also need to look
34 # at rounded corners)
35 # * removal of unused attributes in groups:
36 # <g fill="blue" ...>
37 # <rect fill="red" ... />
38 # <rect fill="red" ... />
39 # <rect fill="red" ... />
40 # </g>
41 # in this case, fill="blue" should be removed
43 # Next Up:
44 # + analyze all children of a group, if they have common inheritable attributes, then move them to the group
45 # + scour lengths for *opacity, svg:x,y,width,height, stroke-miterlimit, stroke-width
46 # - analyze a group and its children, if a group's attribute is not being used by any children
47 # (or descendants?) then remove it
48 # - add an option to remove ids if they match the Inkscape-style of IDs
49 # - investigate point-reducing algorithms
50 # - parse transform attribute
51 # - if a <g> has only one element in it, collapse the <g> (ensure transform, etc are carried down)
52 # - option to remove metadata
53 # - prevent elements from being stripped if they are referenced in a <style> element
54 # (for instance, filter, marker, pattern) - need a crude CSS parser
56 # necessary to get true division
57 from __future__ import division
59 import os
60 import sys
61 import xml.dom.minidom
62 import re
63 import math
64 import base64
65 import urllib
66 from svg_regex import svg_parser
67 import gzip
68 import optparse
70 # Python 2.3- did not have Decimal
71 try:
72 from decimal import *
73 except ImportError:
74 from fixedpoint import *
75 Decimal = FixedPoint
77 APP = 'scour'
78 VER = '0.19'
79 COPYRIGHT = 'Copyright Jeff Schiller, 2009'
81 NS = { 'SVG': 'http://www.w3.org/2000/svg',
82 'XLINK': 'http://www.w3.org/1999/xlink',
83 'SODIPODI': 'http://sodipodi.sourceforge.net/DTD/sodipodi-0.dtd',
84 'INKSCAPE': 'http://www.inkscape.org/namespaces/inkscape',
85 'ADOBE_ILLUSTRATOR': 'http://ns.adobe.com/AdobeIllustrator/10.0/',
86 'ADOBE_GRAPHS': 'http://ns.adobe.com/Graphs/1.0/',
87 'ADOBE_SVG_VIEWER': 'http://ns.adobe.com/AdobeSVGViewerExtensions/3.0/',
88 'ADOBE_VARIABLES': 'http://ns.adobe.com/Variables/1.0/',
89 'ADOBE_SFW': 'http://ns.adobe.com/SaveForWeb/1.0/',
90 'ADOBE_EXTENSIBILITY': 'http://ns.adobe.com/Extensibility/1.0/',
91 'ADOBE_FLOWS': 'http://ns.adobe.com/Flows/1.0/',
92 'ADOBE_IMAGE_REPLACEMENT': 'http://ns.adobe.com/ImageReplacement/1.0/',
93 'ADOBE_CUSTOM': 'http://ns.adobe.com/GenericCustomNamespace/1.0/',
94 'ADOBE_XPATH': 'http://ns.adobe.com/XPath/1.0/'
95 }
97 unwanted_ns = [ NS['SODIPODI'], NS['INKSCAPE'], NS['ADOBE_ILLUSTRATOR'],
98 NS['ADOBE_GRAPHS'], NS['ADOBE_SVG_VIEWER'], NS['ADOBE_VARIABLES'],
99 NS['ADOBE_SFW'], NS['ADOBE_EXTENSIBILITY'], NS['ADOBE_FLOWS'],
100 NS['ADOBE_IMAGE_REPLACEMENT'], NS['ADOBE_CUSTOM'], NS['ADOBE_XPATH'] ]
102 svgAttributes = [
103 'clip-rule',
104 'display',
105 'fill',
106 'fill-opacity',
107 'fill-rule',
108 'filter',
109 'font-family',
110 'font-size',
111 'font-stretch',
112 'font-style',
113 'font-variant',
114 'font-weight',
115 'line-height',
116 'marker',
117 'opacity',
118 'overflow',
119 'stop-color',
120 'stop-opacity',
121 'stroke',
122 'stroke-dashoffset',
123 'stroke-linecap',
124 'stroke-linejoin',
125 'stroke-miterlimit',
126 'stroke-opacity',
127 'stroke-width',
128 'visibility'
129 ]
131 colors = {
132 'aliceblue': 'rgb(240, 248, 255)',
133 'antiquewhite': 'rgb(250, 235, 215)',
134 'aqua': 'rgb( 0, 255, 255)',
135 'aquamarine': 'rgb(127, 255, 212)',
136 'azure': 'rgb(240, 255, 255)',
137 'beige': 'rgb(245, 245, 220)',
138 'bisque': 'rgb(255, 228, 196)',
139 'black': 'rgb( 0, 0, 0)',
140 'blanchedalmond': 'rgb(255, 235, 205)',
141 'blue': 'rgb( 0, 0, 255)',
142 'blueviolet': 'rgb(138, 43, 226)',
143 'brown': 'rgb(165, 42, 42)',
144 'burlywood': 'rgb(222, 184, 135)',
145 'cadetblue': 'rgb( 95, 158, 160)',
146 'chartreuse': 'rgb(127, 255, 0)',
147 'chocolate': 'rgb(210, 105, 30)',
148 'coral': 'rgb(255, 127, 80)',
149 'cornflowerblue': 'rgb(100, 149, 237)',
150 'cornsilk': 'rgb(255, 248, 220)',
151 'crimson': 'rgb(220, 20, 60)',
152 'cyan': 'rgb( 0, 255, 255)',
153 'darkblue': 'rgb( 0, 0, 139)',
154 'darkcyan': 'rgb( 0, 139, 139)',
155 'darkgoldenrod': 'rgb(184, 134, 11)',
156 'darkgray': 'rgb(169, 169, 169)',
157 'darkgreen': 'rgb( 0, 100, 0)',
158 'darkgrey': 'rgb(169, 169, 169)',
159 'darkkhaki': 'rgb(189, 183, 107)',
160 'darkmagenta': 'rgb(139, 0, 139)',
161 'darkolivegreen': 'rgb( 85, 107, 47)',
162 'darkorange': 'rgb(255, 140, 0)',
163 'darkorchid': 'rgb(153, 50, 204)',
164 'darkred': 'rgb(139, 0, 0)',
165 'darksalmon': 'rgb(233, 150, 122)',
166 'darkseagreen': 'rgb(143, 188, 143)',
167 'darkslateblue': 'rgb( 72, 61, 139)',
168 'darkslategray': 'rgb( 47, 79, 79)',
169 'darkslategrey': 'rgb( 47, 79, 79)',
170 'darkturquoise': 'rgb( 0, 206, 209)',
171 'darkviolet': 'rgb(148, 0, 211)',
172 'deeppink': 'rgb(255, 20, 147)',
173 'deepskyblue': 'rgb( 0, 191, 255)',
174 'dimgray': 'rgb(105, 105, 105)',
175 'dimgrey': 'rgb(105, 105, 105)',
176 'dodgerblue': 'rgb( 30, 144, 255)',
177 'firebrick': 'rgb(178, 34, 34)',
178 'floralwhite': 'rgb(255, 250, 240)',
179 'forestgreen': 'rgb( 34, 139, 34)',
180 'fuchsia': 'rgb(255, 0, 255)',
181 'gainsboro': 'rgb(220, 220, 220)',
182 'ghostwhite': 'rgb(248, 248, 255)',
183 'gold': 'rgb(255, 215, 0)',
184 'goldenrod': 'rgb(218, 165, 32)',
185 'gray': 'rgb(128, 128, 128)',
186 'grey': 'rgb(128, 128, 128)',
187 'green': 'rgb( 0, 128, 0)',
188 'greenyellow': 'rgb(173, 255, 47)',
189 'honeydew': 'rgb(240, 255, 240)',
190 'hotpink': 'rgb(255, 105, 180)',
191 'indianred': 'rgb(205, 92, 92)',
192 'indigo': 'rgb( 75, 0, 130)',
193 'ivory': 'rgb(255, 255, 240)',
194 'khaki': 'rgb(240, 230, 140)',
195 'lavender': 'rgb(230, 230, 250)',
196 'lavenderblush': 'rgb(255, 240, 245)',
197 'lawngreen': 'rgb(124, 252, 0)',
198 'lemonchiffon': 'rgb(255, 250, 205)',
199 'lightblue': 'rgb(173, 216, 230)',
200 'lightcoral': 'rgb(240, 128, 128)',
201 'lightcyan': 'rgb(224, 255, 255)',
202 'lightgoldenrodyellow': 'rgb(250, 250, 210)',
203 'lightgray': 'rgb(211, 211, 211)',
204 'lightgreen': 'rgb(144, 238, 144)',
205 'lightgrey': 'rgb(211, 211, 211)',
206 'lightpink': 'rgb(255, 182, 193)',
207 'lightsalmon': 'rgb(255, 160, 122)',
208 'lightseagreen': 'rgb( 32, 178, 170)',
209 'lightskyblue': 'rgb(135, 206, 250)',
210 'lightslategray': 'rgb(119, 136, 153)',
211 'lightslategrey': 'rgb(119, 136, 153)',
212 'lightsteelblue': 'rgb(176, 196, 222)',
213 'lightyellow': 'rgb(255, 255, 224)',
214 'lime': 'rgb( 0, 255, 0)',
215 'limegreen': 'rgb( 50, 205, 50)',
216 'linen': 'rgb(250, 240, 230)',
217 'magenta': 'rgb(255, 0, 255)',
218 'maroon': 'rgb(128, 0, 0)',
219 'mediumaquamarine': 'rgb(102, 205, 170)',
220 'mediumblue': 'rgb( 0, 0, 205)',
221 'mediumorchid': 'rgb(186, 85, 211)',
222 'mediumpurple': 'rgb(147, 112, 219)',
223 'mediumseagreen': 'rgb( 60, 179, 113)',
224 'mediumslateblue': 'rgb(123, 104, 238)',
225 'mediumspringgreen': 'rgb( 0, 250, 154)',
226 'mediumturquoise': 'rgb( 72, 209, 204)',
227 'mediumvioletred': 'rgb(199, 21, 133)',
228 'midnightblue': 'rgb( 25, 25, 112)',
229 'mintcream': 'rgb(245, 255, 250)',
230 'mistyrose': 'rgb(255, 228, 225)',
231 'moccasin': 'rgb(255, 228, 181)',
232 'navajowhite': 'rgb(255, 222, 173)',
233 'navy': 'rgb( 0, 0, 128)',
234 'oldlace': 'rgb(253, 245, 230)',
235 'olive': 'rgb(128, 128, 0)',
236 'olivedrab': 'rgb(107, 142, 35)',
237 'orange': 'rgb(255, 165, 0)',
238 'orangered': 'rgb(255, 69, 0)',
239 'orchid': 'rgb(218, 112, 214)',
240 'palegoldenrod': 'rgb(238, 232, 170)',
241 'palegreen': 'rgb(152, 251, 152)',
242 'paleturquoise': 'rgb(175, 238, 238)',
243 'palevioletred': 'rgb(219, 112, 147)',
244 'papayawhip': 'rgb(255, 239, 213)',
245 'peachpuff': 'rgb(255, 218, 185)',
246 'peru': 'rgb(205, 133, 63)',
247 'pink': 'rgb(255, 192, 203)',
248 'plum': 'rgb(221, 160, 221)',
249 'powderblue': 'rgb(176, 224, 230)',
250 'purple': 'rgb(128, 0, 128)',
251 'red': 'rgb(255, 0, 0)',
252 'rosybrown': 'rgb(188, 143, 143)',
253 'royalblue': 'rgb( 65, 105, 225)',
254 'saddlebrown': 'rgb(139, 69, 19)',
255 'salmon': 'rgb(250, 128, 114)',
256 'sandybrown': 'rgb(244, 164, 96)',
257 'seagreen': 'rgb( 46, 139, 87)',
258 'seashell': 'rgb(255, 245, 238)',
259 'sienna': 'rgb(160, 82, 45)',
260 'silver': 'rgb(192, 192, 192)',
261 'skyblue': 'rgb(135, 206, 235)',
262 'slateblue': 'rgb(106, 90, 205)',
263 'slategray': 'rgb(112, 128, 144)',
264 'slategrey': 'rgb(112, 128, 144)',
265 'snow': 'rgb(255, 250, 250)',
266 'springgreen': 'rgb( 0, 255, 127)',
267 'steelblue': 'rgb( 70, 130, 180)',
268 'tan': 'rgb(210, 180, 140)',
269 'teal': 'rgb( 0, 128, 128)',
270 'thistle': 'rgb(216, 191, 216)',
271 'tomato': 'rgb(255, 99, 71)',
272 'turquoise': 'rgb( 64, 224, 208)',
273 'violet': 'rgb(238, 130, 238)',
274 'wheat': 'rgb(245, 222, 179)',
275 'white': 'rgb(255, 255, 255)',
276 'whitesmoke': 'rgb(245, 245, 245)',
277 'yellow': 'rgb(255, 255, 0)',
278 'yellowgreen': 'rgb(154, 205, 50)',
279 }
281 def isSameSign(a,b): return (a <= 0 and b <= 0) or (a >= 0 and b >= 0)
283 coord = re.compile("\\-?\\d+\\.?\\d*")
284 scinumber = re.compile("[\\-\\+]?(\\d*\\.?)?\\d+[eE][\\-\\+]?\\d+")
285 number = re.compile("[\\-\\+]?(\\d*\\.?)?\\d+")
286 sciExponent = re.compile("[eE]([\\-\\+]?\\d+)")
287 unit = re.compile("(em|ex|px|pt|pc|cm|mm|in|\\%){1,1}$")
289 class Unit(object):
290 INVALID = -1
291 NONE = 0
292 PCT = 1
293 PX = 2
294 PT = 3
295 PC = 4
296 EM = 5
297 EX = 6
298 CM = 7
299 MM = 8
300 IN = 9
302 # @staticmethod
303 def get(str):
304 # GZ: shadowing builtins like 'str' is generally bad form
305 # GZ: encoding stuff like this in a dict makes for nicer code
306 if str == None or str == '': return Unit.NONE
307 elif str == '%': return Unit.PCT
308 elif str == 'px': return Unit.PX
309 elif str == 'pt': return Unit.PT
310 elif str == 'pc': return Unit.PC
311 elif str == 'em': return Unit.EM
312 elif str == 'ex': return Unit.EX
313 elif str == 'cm': return Unit.CM
314 elif str == 'mm': return Unit.MM
315 elif str == 'in': return Unit.IN
316 return Unit.INVALID
318 # @staticmethod
319 def str(u):
320 if u == Unit.NONE: return ''
321 elif u == Unit.PCT: return '%'
322 elif u == Unit.PX: return 'px'
323 elif u == Unit.PT: return 'pt'
324 elif u == Unit.PC: return 'pc'
325 elif u == Unit.EM: return 'em'
326 elif u == Unit.EX: return 'ex'
327 elif u == Unit.CM: return 'cm'
328 elif u == Unit.MM: return 'mm'
329 elif u == Unit.IN: return 'in'
330 return 'INVALID'
332 get = staticmethod(get)
333 str = staticmethod(str)
335 class SVGLength(object):
336 def __init__(self, str):
337 try: # simple unitless and no scientific notation
338 self.value = float(str)
339 if int(self.value) == self.value:
340 self.value = int(self.value)
341 self.units = Unit.NONE
342 except ValueError:
343 # we know that the length string has an exponent, a unit, both or is invalid
345 # parse out number, exponent and unit
346 self.value = 0
347 unitBegin = 0
348 scinum = scinumber.match(str)
349 if scinum != None:
350 # this will always match, no need to check it
351 numMatch = number.match(str)
352 expMatch = sciExponent.search(str, numMatch.start(0))
353 self.value = (float(numMatch.group(0)) *
354 10 ** float(expMatch.group(1)))
355 unitBegin = expMatch.end(1)
356 else:
357 # unit or invalid
358 numMatch = number.match(str)
359 if numMatch != None:
360 self.value = float(numMatch.group(0))
361 unitBegin = numMatch.end(0)
363 if int(self.value) == self.value:
364 self.value = int(self.value)
366 if unitBegin != 0 :
367 unitMatch = unit.search(str, unitBegin)
368 if unitMatch != None :
369 self.units = Unit.get(unitMatch.group(0))
371 # invalid
372 else:
373 # TODO: this needs to set the default for the given attribute (how?)
374 self.value = 0
375 self.units = Unit.INVALID
377 # returns the length of a property
378 # TODO: eventually use the above class once it is complete
379 def getSVGLength(value):
380 try:
381 v = float(value)
382 except ValueError:
383 coordMatch = coord.match(value)
384 if coordMatch != None:
385 unitMatch = unit.search(value, coordMatch.start(0))
386 v = value
387 return v
389 def findElementById(node, id):
390 if node == None or node.nodeType != 1: return None
391 if node.getAttribute('id') == id: return node
392 for child in node.childNodes :
393 e = findElementById(child,id)
394 if e != None: return e
395 return None
397 def findElementsWithId(node, elems=None):
398 """
399 Returns all elements with id attributes
400 """
401 if elems is None:
402 elems = {}
403 id = node.getAttribute('id')
404 if id != '' :
405 elems[id] = node
406 if node.hasChildNodes() :
407 for child in node.childNodes:
408 # from http://www.w3.org/TR/DOM-Level-2-Core/idl-definitions.html
409 # we are only really interested in nodes of type Element (1)
410 if child.nodeType == 1 :
411 findElementsWithId(child, elems)
412 return elems
414 def findReferencedElements(node, ids=None):
415 """
416 Returns the number of times an ID is referenced as well as all elements
417 that reference it.
419 Currently looks at fill, stroke, clip-path, mask, marker, and
420 xlink:href attributes.
421 """
422 if ids is None:
423 ids = {}
424 # TODO: input argument ids is clunky here (see below how it is called)
425 # GZ: alternative to passing dict, use **kwargs
426 href = node.getAttributeNS(NS['XLINK'],'href')
428 # if xlink:href is set, then grab the id
429 if href != '' and len(href) > 1 and href[0] == '#':
430 # we remove the hash mark from the beginning of the id
431 id = href[1:]
432 if id in ids:
433 ids[id][0] += 1
434 ids[id][1].append(node)
435 else:
436 ids[id] = [1,[node]]
438 # now get all style properties and the fill, stroke, filter attributes
439 styles = node.getAttribute('style').split(';')
440 referencingProps = ['fill', 'stroke', 'filter', 'clip-path', 'mask', 'marker-start',
441 'marker-end', 'marker-mid']
442 for attr in referencingProps:
443 styles.append(':'.join([attr, node.getAttribute(attr)]))
445 for style in styles:
446 propval = style.split(':')
447 if len(propval) == 2 :
448 prop = propval[0].strip()
449 val = propval[1].strip()
450 if prop in referencingProps and val != '' :
451 if len(val) >= 7 and val[0:5] == 'url(#' :
452 id = val[5:val.find(')')]
453 if ids.has_key(id) :
454 ids[id][0] += 1
455 ids[id][1].append(node)
456 else:
457 ids[id] = [1,[node]]
458 # if the url has a quote in it, we need to compensate
459 elif len(val) >= 8 :
460 id = None
461 # double-quote
462 if val[0:6] == 'url("#' :
463 id = val[6:val.find('")')]
464 # single-quote
465 elif val[0:6] == "url('#" :
466 id = val[6:val.find("')")]
467 if id != None:
468 if ids.has_key(id) :
469 ids[id][0] += 1
470 ids[id][1].append(node)
471 else:
472 ids[id] = [1,[node]]
474 if node.hasChildNodes() :
475 for child in node.childNodes:
476 if child.nodeType == 1 :
477 findReferencedElements(child, ids)
478 return ids
480 numIDsRemoved = 0
481 numElemsRemoved = 0
482 numAttrsRemoved = 0
483 numRastersEmbedded = 0
484 numPathSegmentsReduced = 0
485 numCurvesStraightened = 0
486 numBytesSavedInPathData = 0
487 numBytesSavedInColors = 0
488 numPointsRemovedFromPolygon = 0
490 def removeUnusedDefs(doc, defElem, elemsToRemove=None):
491 if elemsToRemove is None:
492 elemsToRemove = []
494 identifiedElements = findElementsWithId(doc.documentElement)
495 referencedIDs = findReferencedElements(doc.documentElement)
497 keepTags = ['font', 'style', 'metadata', 'script', 'title', 'desc']
498 for elem in defElem.childNodes:
499 if elem.nodeName == 'g' and elem.namespaceURI == NS['SVG']:
500 elemsToRemove = removeUnusedDefs(doc, elem, elemsToRemove)
501 continue
502 if elem.nodeType == 1 and (elem.getAttribute('id') == '' or \
503 (not elem.getAttribute('id') in referencedIDs)) and \
504 not elem.nodeName in keepTags:
505 elemsToRemove.append(elem)
506 return elemsToRemove
508 def removeUnreferencedElements(doc):
509 """
510 Removes all unreferenced elements except for <svg>, <font>, <metadata>, <title>, and <desc>.
511 Also vacuums the defs of any non-referenced renderable elements.
513 Returns the number of unreferenced elements removed from the document.
514 """
515 global numElemsRemoved
516 num = 0
517 removeTags = ['linearGradient', 'radialGradient', 'pattern']
519 identifiedElements = findElementsWithId(doc.documentElement)
520 referencedIDs = findReferencedElements(doc.documentElement)
522 for id in identifiedElements:
523 if not id in referencedIDs:
524 goner = findElementById(doc.documentElement, id)
525 if goner != None and goner.parentNode != None and goner.nodeName in removeTags:
526 goner.parentNode.removeChild(goner)
527 num += 1
528 numElemsRemoved += 1
530 # TODO: should also go through defs and vacuum it
531 num = 0
532 defs = doc.documentElement.getElementsByTagName('defs')
533 for aDef in defs:
534 elemsToRemove = removeUnusedDefs(doc, aDef)
535 for elem in elemsToRemove:
536 elem.parentNode.removeChild(elem)
537 numElemsRemoved += 1
538 num += 1
539 return num
541 def removeUnreferencedIDs(referencedIDs, identifiedElements):
542 """
543 Removes the unreferenced ID attributes.
545 Returns the number of ID attributes removed
546 """
547 global numIDsRemoved
548 keepTags = ['font']
549 num = 0;
550 for id in identifiedElements.keys():
551 node = identifiedElements[id]
552 if referencedIDs.has_key(id) == False and not node.nodeName in keepTags:
553 node.removeAttribute('id')
554 numIDsRemoved += 1
555 num += 1
556 return num
558 def removeNamespacedAttributes(node, namespaces):
559 global numAttrsRemoved
560 num = 0
561 if node.nodeType == 1 :
562 # remove all namespace'd attributes from this element
563 attrList = node.attributes
564 attrsToRemove = []
565 for attrNum in range(attrList.length):
566 attr = attrList.item(attrNum)
567 if attr != None and attr.namespaceURI in namespaces:
568 attrsToRemove.append(attr.nodeName)
569 for attrName in attrsToRemove :
570 num += 1
571 numAttrsRemoved += 1
572 node.removeAttribute(attrName)
574 # now recurse for children
575 for child in node.childNodes:
576 num += removeNamespacedAttributes(child, namespaces)
577 return num
579 def removeNamespacedElements(node, namespaces):
580 global numElemsRemoved
581 num = 0
582 if node.nodeType == 1 :
583 # remove all namespace'd child nodes from this element
584 childList = node.childNodes
585 childrenToRemove = []
586 for child in childList:
587 if child != None and child.namespaceURI in namespaces:
588 childrenToRemove.append(child)
589 for child in childrenToRemove :
590 num += 1
591 numElemsRemoved += 1
592 node.removeChild(child)
594 # now recurse for children
595 for child in node.childNodes:
596 num += removeNamespacedElements(child, namespaces)
597 return num
599 def removeNestedGroups(node):
600 """
601 This walks further and further down the tree, removing groups
602 which do not have any attributes or a title/desc child and
603 promoting their children up one level
604 """
605 global numElemsRemoved
606 num = 0
608 groupsToRemove = []
609 for child in node.childNodes:
610 if child.nodeName == 'g' and child.namespaceURI == NS['SVG'] and len(child.attributes) == 0:
611 # only collapse group if it does not have a title or desc as a direct descendant
612 for grandchild in child.childNodes:
613 if grandchild.nodeType == 1 and grandchild.namespaceURI == NS['SVG'] and \
614 grandchild.nodeName in ['title','desc']:
615 break
616 else:
617 groupsToRemove.append(child)
619 for g in groupsToRemove:
620 while g.childNodes.length > 0:
621 g.parentNode.insertBefore(g.firstChild, g)
622 g.parentNode.removeChild(g)
623 numElemsRemoved += 1
624 num += 1
626 # now recurse for children
627 for child in node.childNodes:
628 if child.nodeType == 1:
629 num += removeNestedGroups(child)
630 return num
632 def moveCommonAttributesToParentGroup(elem):
633 """
634 This recursively calls this function on all children of the passed in element
635 and then iterates over all child elements and removes common inheritable attributes
636 from the children and places them in the parent group.
637 """
638 num = 0
640 childElements = []
641 # recurse first into the children (depth-first)
642 for child in elem.childNodes:
643 if child.nodeType == 1:
644 childElements.append(child)
645 num += moveCommonAttributesToParentGroup(child)
647 # only process the children if there are more than one element
648 if len(childElements) <= 1: return num
650 commonAttrs = {}
651 # add all inheritable properties of the first child element
652 # FIXME: Note there is a chance that the first child is a set/animate in which case
653 # its fill attribute is not what we want to look at, we should look for the first
654 # non-animate/set element
655 attrList = childElements[0].attributes
656 for num in range(attrList.length):
657 attr = attrList.item(num)
658 # this is most of the inheritable properties from http://www.w3.org/TR/SVG11/propidx.html
659 # and http://www.w3.org/TR/SVGTiny12/attributeTable.html
660 if attr.nodeName in ['clip-rule',
661 'display-align',
662 'fill', 'fill-opacity', 'fill-rule',
663 'font', 'font-family', 'font-size', 'font-size-adjust', 'font-stretch',
664 'font-style', 'font-variant', 'font-weight',
665 'letter-spacing',
666 'pointer-events', 'shape-rendering',
667 'stroke', 'stroke-dasharray', 'stroke-dashoffset', 'stroke-linecap', 'stroke-linejoin',
668 'stroke-miterlimit', 'stroke-opacity', 'stroke-width',
669 'text-anchor', 'text-decoration', 'text-rendering', 'visibility',
670 'word-spacing', 'writing-mode']:
671 # we just add all the attributes from the first child
672 commonAttrs[attr.nodeName] = attr.nodeValue
674 # for each subsequent child element
675 for childNum in range(len(childElements)):
676 # skip first child
677 if childNum == 0:
678 continue
680 child = childElements[childNum]
681 # if we are on an animateXXX/set element, ignore it (due to the 'fill' attribute)
682 if child.localName in ['set', 'animate', 'animateColor', 'animateTransform', 'animateMotion']:
683 continue
685 distinctAttrs = []
686 # loop through all current 'common' attributes
687 for name in commonAttrs.keys():
688 # if this child doesn't match that attribute, schedule it for removal
689 if child.getAttribute(name) != commonAttrs[name]:
690 distinctAttrs.append(name)
691 # remove those attributes which are not common
692 for name in distinctAttrs:
693 del commonAttrs[name]
695 # commonAttrs now has all the inheritable attributes which are common among all child elements
696 for name in commonAttrs.keys():
697 for child in childElements:
698 child.removeAttribute(name)
699 elem.setAttribute(name, commonAttrs[name])
701 # update our statistic (we remove N*M attributes and add back in M attributes)
702 num += (len(childElements)-1) * len(commonAttrs)
703 return num
705 def removeDuplicateGradientStops(doc):
706 global numElemsRemoved
707 num = 0
709 for gradType in ['linearGradient', 'radialGradient']:
710 for grad in doc.getElementsByTagName(gradType):
711 stops = {}
712 stopsToRemove = []
713 for stop in grad.getElementsByTagName('stop'):
714 # convert percentages into a floating point number
715 offsetU = SVGLength(stop.getAttribute('offset'))
716 if offsetU.units == Unit.PCT:
717 offset = offsetU.value / 100.0
718 elif offsetU.units == Unit.NONE:
719 offset = offsetU.value
720 else:
721 offset = 0
722 # set the stop offset value to the integer or floating point equivalent
723 if int(offset) == offset: stop.setAttribute('offset', str(int(offset)))
724 else: stop.setAttribute('offset', str(offset))
726 color = stop.getAttribute('stop-color')
727 opacity = stop.getAttribute('stop-opacity')
728 if stops.has_key(offset) :
729 oldStop = stops[offset]
730 if oldStop[0] == color and oldStop[1] == opacity:
731 stopsToRemove.append(stop)
732 stops[offset] = [color, opacity]
734 for stop in stopsToRemove:
735 stop.parentNode.removeChild(stop)
736 num += 1
737 numElemsRemoved += 1
739 # linear gradients
740 return num
742 def collapseSinglyReferencedGradients(doc):
743 global numElemsRemoved
744 num = 0
746 # make sure to reset the ref'ed ids for when we are running this in testscour
747 for rid,nodeCount in findReferencedElements(doc.documentElement).iteritems():
748 count = nodeCount[0]
749 nodes = nodeCount[1]
750 if count == 1:
751 elem = findElementById(doc.documentElement,rid)
752 if elem != None and elem.nodeType == 1 and elem.nodeName in ['linearGradient', 'radialGradient'] \
753 and elem.namespaceURI == NS['SVG']:
754 # found a gradient that is referenced by only 1 other element
755 refElem = nodes[0]
756 if refElem.nodeType == 1 and refElem.nodeName in ['linearGradient', 'radialGradient'] \
757 and refElem.namespaceURI == NS['SVG']:
758 # elem is a gradient referenced by only one other gradient (refElem)
760 # add the stops to the referencing gradient (this removes them from elem)
761 if len(refElem.getElementsByTagName('stop')) == 0:
762 stopsToAdd = elem.getElementsByTagName('stop')
763 for stop in stopsToAdd:
764 refElem.appendChild(stop)
766 # adopt the gradientUnits, spreadMethod, gradientTransform attributes if
767 # they are unspecified on refElem
768 for attr in ['gradientUnits','spreadMethod','gradientTransform']:
769 if refElem.getAttribute(attr) == '' and not elem.getAttribute(attr) == '':
770 refElem.setAttributeNS(None, attr, elem.getAttribute(attr))
772 # if both are radialGradients, adopt elem's fx,fy,cx,cy,r attributes if
773 # they are unspecified on refElem
774 if elem.nodeName == 'radialGradient' and refElem.nodeName == 'radialGradient':
775 for attr in ['fx','fy','cx','cy','r']:
776 if refElem.getAttribute(attr) == '' and not elem.getAttribute(attr) == '':
777 refElem.setAttributeNS(None, attr, elem.getAttribute(attr))
779 # if both are linearGradients, adopt elem's x1,y1,x2,y2 attributes if
780 # they are unspecified on refElem
781 if elem.nodeName == 'linearGradient' and refElem.nodeName == 'linearGradient':
782 for attr in ['x1','y1','x2','y2']:
783 if refElem.getAttribute(attr) == '' and not elem.getAttribute(attr) == '':
784 refElem.setAttributeNS(None, attr, elem.getAttribute(attr))
786 # now remove the xlink:href from refElem
787 refElem.removeAttributeNS(NS['XLINK'], 'href')
789 # now delete elem
790 elem.parentNode.removeChild(elem)
791 numElemsRemoved += 1
792 num += 1
793 return num
795 def removeDuplicateGradients(doc):
796 global numElemsRemoved
797 num = 0
799 gradientsToRemove = {}
800 duplicateToMaster = {}
802 for gradType in ['linearGradient', 'radialGradient']:
803 grads = doc.getElementsByTagName(gradType)
804 for grad in grads:
805 # TODO: should slice grads from 'grad' here to optimize
806 for ograd in grads:
807 # do not compare gradient to itself
808 if grad == ograd: continue
810 # compare grad to ograd (all properties, then all stops)
811 # if attributes do not match, go to next gradient
812 someGradAttrsDoNotMatch = False
813 for attr in ['gradientUnits','spreadMethod','gradientTransform','x1','y1','x2','y2','cx','cy','fx','fy','r']:
814 if grad.getAttribute(attr) != ograd.getAttribute(attr):
815 someGradAttrsDoNotMatch = True
816 break;
818 if someGradAttrsDoNotMatch: continue
820 # compare xlink:href values too
821 if grad.getAttributeNS(NS['XLINK'], 'href') != ograd.getAttributeNS(NS['XLINK'], 'href'):
822 continue
824 # all gradient properties match, now time to compare stops
825 stops = grad.getElementsByTagName('stop')
826 ostops = ograd.getElementsByTagName('stop')
828 if stops.length != ostops.length: continue
830 # now compare stops
831 stopsNotEqual = False
832 for i in range(stops.length):
833 if stopsNotEqual: break
834 stop = stops.item(i)
835 ostop = ostops.item(i)
836 for attr in ['offset', 'stop-color', 'stop-opacity']:
837 if stop.getAttribute(attr) != ostop.getAttribute(attr):
838 stopsNotEqual = True
839 break
840 if stopsNotEqual: continue
842 # ograd is a duplicate of grad, we schedule it to be removed UNLESS
843 # ograd is ALREADY considered a 'master' element
844 if not gradientsToRemove.has_key(ograd):
845 if not duplicateToMaster.has_key(ograd):
846 if not gradientsToRemove.has_key(grad):
847 gradientsToRemove[grad] = []
848 gradientsToRemove[grad].append( ograd )
849 duplicateToMaster[ograd] = grad
851 # get a collection of all elements that are referenced and their referencing elements
852 referencedIDs = findReferencedElements(doc.documentElement)
853 for masterGrad in gradientsToRemove.keys():
854 master_id = masterGrad.getAttribute('id')
855 for dupGrad in gradientsToRemove[masterGrad]:
856 # if the duplicate gradient no longer has a parent that means it was
857 # already re-mapped to another master gradient
858 if not dupGrad.parentNode: continue
859 dup_id = dupGrad.getAttribute('id')
860 # for each element that referenced the gradient we are going to remove
861 for elem in referencedIDs[dup_id][1]:
862 # find out which attribute referenced the duplicate gradient
863 for attr in ['fill', 'stroke']:
864 v = elem.getAttribute(attr)
865 if v == 'url(#'+dup_id+')' or v == 'url("#'+dup_id+'")' or v == "url('#"+dup_id+"')":
866 elem.setAttribute(attr, 'url(#'+master_id+')')
867 if elem.getAttributeNS(NS['XLINK'], 'href') == '#'+dup_id:
868 elem.setAttributeNS(NS['XLINK'], 'href', '#'+master_id)
870 # now that all referencing elements have been re-mapped to the master
871 # it is safe to remove this gradient from the document
872 dupGrad.parentNode.removeChild(dupGrad)
873 numElemsRemoved += 1
874 num += 1
875 return num
877 def repairStyle(node, options):
878 num = 0
879 if node.nodeType == 1 and len(node.getAttribute('style')) > 0 :
880 # get all style properties and stuff them into a dictionary
881 styleMap = { }
882 rawStyles = node.getAttribute('style').split(';')
883 for style in rawStyles:
884 propval = style.split(':')
885 if len(propval) == 2 :
886 styleMap[propval[0].strip()] = propval[1].strip()
888 # I've seen this enough to know that I need to correct it:
889 # fill: url(#linearGradient4918) rgb(0, 0, 0);
890 for prop in ['fill', 'stroke'] :
891 if styleMap.has_key(prop) :
892 chunk = styleMap[prop].split(') ')
893 if len(chunk) == 2 and (chunk[0][:5] == 'url(#' or chunk[0][:6] == 'url("#' or chunk[0][:6] == "url('#") and chunk[1] == 'rgb(0, 0, 0)' :
894 styleMap[prop] = chunk[0] + ')'
895 num += 1
897 # Here is where we can weed out unnecessary styles like:
898 # opacity:1
899 if styleMap.has_key('opacity') :
900 opacity = float(styleMap['opacity'])
901 # opacity='1.0' is useless, remove it
902 if opacity == 1.0 :
903 del styleMap['opacity']
904 num += 1
906 # if opacity='0' then all fill and stroke properties are useless, remove them
907 elif opacity == 0.0 :
908 for uselessStyle in ['fill', 'fill-opacity', 'fill-rule', 'stroke', 'stroke-linejoin',
909 'stroke-opacity', 'stroke-miterlimit', 'stroke-linecap', 'stroke-dasharray',
910 'stroke-dashoffset', 'stroke-opacity'] :
911 if styleMap.has_key(uselessStyle):
912 del styleMap[uselessStyle]
913 num += 1
915 # if stroke:none, then remove all stroke-related properties (stroke-width, etc)
916 # TODO: should also detect if the computed value of this element is stroke="none"
917 if styleMap.has_key('stroke') and styleMap['stroke'] == 'none' :
918 for strokestyle in [ 'stroke-width', 'stroke-linejoin', 'stroke-miterlimit',
919 'stroke-linecap', 'stroke-dasharray', 'stroke-dashoffset', 'stroke-opacity'] :
920 if styleMap.has_key(strokestyle) :
921 del styleMap[strokestyle]
922 num += 1
923 # TODO: This is actually a problem if a parent element has a specified stroke
924 # we need to properly calculate computed values
925 del styleMap['stroke']
927 # if fill:none, then remove all fill-related properties (fill-rule, etc)
928 if styleMap.has_key('fill') and styleMap['fill'] == 'none' :
929 for fillstyle in [ 'fill-rule', 'fill-opacity' ] :
930 if styleMap.has_key(fillstyle) :
931 del styleMap[fillstyle]
932 num += 1
934 # stop-opacity: 1
935 if styleMap.has_key('stop-opacity') :
936 if float(styleMap['stop-opacity']) == 1.0 :
937 del styleMap['stop-opacity']
938 num += 1
940 # fill-opacity: 1 or 0
941 if styleMap.has_key('fill-opacity') :
942 fillOpacity = float(styleMap['fill-opacity'])
943 # TODO: This is actually a problem if the parent element does not have fill-opacity=1
944 if fillOpacity == 1.0 :
945 del styleMap['fill-opacity']
946 num += 1
947 elif fillOpacity == 0.0 :
948 for uselessFillStyle in [ 'fill', 'fill-rule' ] :
949 if styleMap.has_key(uselessFillStyle):
950 del styleMap[uselessFillStyle]
951 num += 1
953 # stroke-opacity: 1 or 0
954 if styleMap.has_key('stroke-opacity') :
955 strokeOpacity = float(styleMap['stroke-opacity'])
956 # TODO: This is actually a problem if the parent element does not have stroke-opacity=1
957 if strokeOpacity == 1.0 :
958 del styleMap['stroke-opacity']
959 num += 1
960 elif strokeOpacity == 0.0 :
961 for uselessStrokeStyle in [ 'stroke', 'stroke-width', 'stroke-linejoin', 'stroke-linecap',
962 'stroke-dasharray', 'stroke-dashoffset' ] :
963 if styleMap.has_key(uselessStrokeStyle):
964 del styleMap[uselessStrokeStyle]
965 num += 1
967 # stroke-width: 0
968 if styleMap.has_key('stroke-width') :
969 strokeWidth = getSVGLength(styleMap['stroke-width'])
970 if strokeWidth == 0.0 :
971 for uselessStrokeStyle in [ 'stroke', 'stroke-linejoin', 'stroke-linecap',
972 'stroke-dasharray', 'stroke-dashoffset', 'stroke-opacity' ] :
973 if styleMap.has_key(uselessStrokeStyle):
974 del styleMap[uselessStrokeStyle]
975 num += 1
977 # remove font properties for non-text elements
978 # I've actually observed this in real SVG content
979 if node.nodeName in ['rect', 'circle', 'ellipse', 'line', 'polyline', 'polygon', 'path']:
980 for fontstyle in [ 'font-family', 'font-size', 'font-stretch', 'font-size-adjust',
981 'font-style', 'font-variant', 'font-weight',
982 'letter-spacing', 'line-height', 'kerning',
983 'text-anchor', 'text-decoration', 'text-rendering',
984 'unicode-bidi', 'word-spacing', 'writing-mode'] :
985 if styleMap.has_key(fontstyle) :
986 del styleMap[fontstyle]
987 num += 1
989 # remove inkscape-specific styles
990 # TODO: need to get a full list of these
991 for inkscapeStyle in ['-inkscape-font-specification']:
992 if styleMap.has_key(inkscapeStyle):
993 del styleMap[inkscapeStyle]
994 num += 1
996 # visibility: visible
997 if styleMap.has_key('visibility') :
998 if styleMap['visibility'] == 'visible':
999 del styleMap['visibility']
1000 num += 1
1002 # display: inline
1003 if styleMap.has_key('display') :
1004 if styleMap['display'] == 'inline':
1005 del styleMap['display']
1006 num += 1
1008 # overflow: visible or overflow specified on element other than svg, marker, pattern
1009 if styleMap.has_key('overflow') :
1010 if styleMap['overflow'] == 'visible' or node.nodeName in ['svg','marker','pattern']:
1011 del styleMap['overflow']
1012 num += 1
1014 # marker: none
1015 if styleMap.has_key('marker') :
1016 if styleMap['marker'] == 'none':
1017 del styleMap['marker']
1018 num += 1
1020 # now if any of the properties match known SVG attributes we prefer attributes
1021 # over style so emit them and remove them from the style map
1022 if options.style_to_xml:
1023 for propName in styleMap.keys() :
1024 if propName in svgAttributes :
1025 node.setAttribute(propName, styleMap[propName])
1026 del styleMap[propName]
1028 # sew our remaining style properties back together into a style attribute
1029 fixedStyle = ''
1030 for prop in styleMap.keys() :
1031 fixedStyle += prop + ':' + styleMap[prop] + ';'
1033 if fixedStyle != '' :
1034 node.setAttribute('style', fixedStyle)
1035 else:
1036 node.removeAttribute('style')
1038 # recurse for our child elements
1039 for child in node.childNodes :
1040 num += repairStyle(child,options)
1042 return num
1044 def removeDefaultAttributeValues(node, options):
1045 num = 0
1046 if node.nodeType != 1: return 0
1048 # gradientUnits: objectBoundingBox
1049 if node.getAttribute('gradientUnits') == 'objectBoundingBox':
1050 node.removeAttribute('gradientUnits')
1051 num += 1
1053 # spreadMethod: pad
1054 if node.getAttribute('spreadMethod') == 'pad':
1055 node.removeAttribute('spreadMethod')
1056 num += 1
1058 # x1: 0%
1059 if node.getAttribute('x1') != '':
1060 x1 = SVGLength(node.getAttribute('x1'))
1061 if x1.value == 0:
1062 node.removeAttribute('x1')
1063 num += 1
1065 # y1: 0%
1066 if node.getAttribute('y1') != '':
1067 y1 = SVGLength(node.getAttribute('y1'))
1068 if y1.value == 0:
1069 node.removeAttribute('y1')
1070 num += 1
1072 # x2: 100%
1073 if node.getAttribute('x2') != '':
1074 x2 = SVGLength(node.getAttribute('x2'))
1075 if (x2.value == 100 and x2.units == Unit.PCT) or (x2.value == 1 and x2.units == Unit.NONE):
1076 node.removeAttribute('x2')
1077 num += 1
1079 # y2: 0%
1080 if node.getAttribute('y2') != '':
1081 y2 = SVGLength(node.getAttribute('y2'))
1082 if y2.value == 0:
1083 node.removeAttribute('y2')
1084 num += 1
1086 # fx: equal to rx
1087 if node.getAttribute('fx') != '':
1088 if node.getAttribute('fx') == node.getAttribute('cx'):
1089 node.removeAttribute('fx')
1090 num += 1
1092 # fy: equal to ry
1093 if node.getAttribute('fy') != '':
1094 if node.getAttribute('fy') == node.getAttribute('cy'):
1095 node.removeAttribute('fy')
1096 num += 1
1098 # cx: 50%
1099 if node.getAttribute('cx') != '':
1100 cx = SVGLength(node.getAttribute('cx'))
1101 if (cx.value == 50 and cx.units == Unit.PCT) or (cx.value == 0.5 and cx.units == Unit.NONE):
1102 node.removeAttribute('cx')
1103 num += 1
1105 # cy: 50%
1106 if node.getAttribute('cy') != '':
1107 cy = SVGLength(node.getAttribute('cy'))
1108 if (cy.value == 50 and cy.units == Unit.PCT) or (cy.value == 0.5 and cy.units == Unit.NONE):
1109 node.removeAttribute('cy')
1110 num += 1
1112 # r: 50%
1113 if node.getAttribute('r') != '':
1114 r = SVGLength(node.getAttribute('r'))
1115 if (r.value == 50 and r.units == Unit.PCT) or (r.value == 0.5 and r.units == Unit.NONE):
1116 node.removeAttribute('r')
1117 num += 1
1119 # recurse for our child elements
1120 for child in node.childNodes :
1121 num += removeDefaultAttributeValues(child,options)
1123 return num
1125 rgb = re.compile("\\s*rgb\\(\\s*(\\d+)\\s*\\,\\s*(\\d+)\\s*\\,\\s*(\\d+)\\s*\\)\\s*")
1126 rgbp = re.compile("\\s*rgb\\(\\s*(\\d*\\.?\\d+)\\%\\s*\\,\\s*(\\d*\\.?\\d+)\\%\\s*\\,\\s*(\\d*\\.?\\d+)\\%\\s*\\)\\s*")
1127 def convertColor(value):
1128 """
1129 Converts the input color string and returns a #RRGGBB (or #RGB if possible) string
1130 """
1131 s = value
1133 if s in colors.keys():
1134 s = colors[s]
1136 rgbpMatch = rgbp.match(s)
1137 if rgbpMatch != None :
1138 r = int(float(rgbpMatch.group(1)) * 255.0 / 100.0)
1139 g = int(float(rgbpMatch.group(2)) * 255.0 / 100.0)
1140 b = int(float(rgbpMatch.group(3)) * 255.0 / 100.0)
1141 s = 'rgb(%d,%d,%d)' % (r,g,b)
1143 rgbMatch = rgb.match(s)
1144 if rgbMatch != None :
1145 r = hex( int( rgbMatch.group(1) ) )[2:].upper()
1146 g = hex( int( rgbMatch.group(2) ) )[2:].upper()
1147 b = hex( int( rgbMatch.group(3) ) )[2:].upper()
1148 if len(r) == 1: r='0'+r
1149 if len(g) == 1: g='0'+g
1150 if len(b) == 1: b='0'+b
1151 s = '#'+r+g+b
1153 if s[0] == '#' and len(s)==7 and s[1]==s[2] and s[3]==s[4] and s[5]==s[6]:
1154 s = s.upper()
1155 s = '#'+s[1]+s[3]+s[5]
1157 return s
1159 def convertColors(element) :
1160 """
1161 Recursively converts all color properties into #RRGGBB format if shorter
1162 """
1163 numBytes = 0
1165 if element.nodeType != 1: return 0
1167 # set up list of color attributes for each element type
1168 attrsToConvert = []
1169 if element.nodeName in ['rect', 'circle', 'ellipse', 'polygon', \
1170 'line', 'polyline', 'path', 'g', 'a']:
1171 attrsToConvert = ['fill', 'stroke']
1172 elif element.nodeName in ['stop']:
1173 attrsToConvert = ['stop-color']
1174 elif element.nodeName in ['solidColor']:
1175 attrsToConvert = ['solid-color']
1177 # now convert all the color formats
1178 for attr in attrsToConvert:
1179 oldColorValue = element.getAttribute(attr)
1180 if oldColorValue != '':
1181 newColorValue = convertColor(oldColorValue)
1182 oldBytes = len(oldColorValue)
1183 newBytes = len(newColorValue)
1184 if oldBytes > newBytes:
1185 element.setAttribute(attr, newColorValue)
1186 numBytes += (oldBytes - len(element.getAttribute(attr)))
1188 # now recurse for our child elements
1189 for child in element.childNodes :
1190 numBytes += convertColors(child)
1192 return numBytes
1194 # TODO: go over what this method does and see if there is a way to optimize it
1195 # TODO: go over the performance of this method and see if I can save memory/speed by
1196 # reusing data structures, etc
1197 def cleanPath(element) :
1198 """
1199 Cleans the path string (d attribute) of the element
1200 """
1201 global numBytesSavedInPathData
1202 global numPathSegmentsReduced
1203 global numCurvesStraightened
1205 # this gets the parser object from svg_regex.py
1206 oldPathStr = element.getAttribute('d')
1207 pathObj = svg_parser.parse(oldPathStr)
1209 # however, this parser object has some ugliness in it (lists of tuples of tuples of
1210 # numbers and booleans). we just need a list of (cmd,[numbers]):
1211 path = []
1212 for (cmd,dataset) in pathObj:
1213 if cmd in ['M','m','L','l','T','t']:
1214 # one or more tuples, each containing two numbers
1215 nums = []
1216 for t in dataset:
1217 # convert to a Decimal
1218 nums.append(Decimal(str(t[0])) * Decimal(1))
1219 nums.append(Decimal(str(t[1])) * Decimal(1))
1221 # only create this segment if it is not empty
1222 if nums:
1223 path.append( (cmd, nums) )
1225 elif cmd in ['V','v','H','h']:
1226 # one or more numbers
1227 nums = []
1228 for n in dataset:
1229 nums.append(Decimal(str(n)))
1230 if nums:
1231 path.append( (cmd, nums) )
1233 elif cmd in ['C','c']:
1234 # one or more tuples, each containing three tuples of two numbers each
1235 nums = []
1236 for t in dataset:
1237 for pair in t:
1238 nums.append(Decimal(str(pair[0])) * Decimal(1))
1239 nums.append(Decimal(str(pair[1])) * Decimal(1))
1240 path.append( (cmd, nums) )
1242 elif cmd in ['S','s','Q','q']:
1243 # one or more tuples, each containing two tuples of two numbers each
1244 nums = []
1245 for t in dataset:
1246 for pair in t:
1247 nums.append(Decimal(str(pair[0])) * Decimal(1))
1248 nums.append(Decimal(str(pair[1])) * Decimal(1))
1249 path.append( (cmd, nums) )
1251 elif cmd in ['A','a']:
1252 # one or more tuples, each containing a tuple of two numbers, a number, a boolean,
1253 # another boolean, and a tuple of two numbers
1254 nums = []
1255 for t in dataset:
1256 nums.append( Decimal(str(t[0][0])) * Decimal(1) )
1257 nums.append( Decimal(str(t[0][1])) * Decimal(1) )
1258 nums.append( Decimal(str(t[1])) * Decimal(1))
1260 if t[2]: nums.append( Decimal(1) )
1261 else: nums.append( Decimal(0) )
1263 if t[3]: nums.append( Decimal(1) )
1264 else: nums.append( Decimal(0) )
1266 nums.append( Decimal(str(t[4][0])) * Decimal(1) )
1267 nums.append( Decimal(str(t[4][1])) * Decimal(1) )
1268 path.append( (cmd, nums) )
1270 elif cmd in ['Z','z']:
1271 path.append( (cmd, []) )
1273 # calculate the starting x,y coord for the second path command
1274 if len(path[0][1]) == 2:
1275 (x,y) = path[0][1]
1276 else:
1277 # we have a move and then 1 or more coords for lines
1278 N = len(path[0][1])
1279 if path[0] == 'M':
1280 # take the last pair of coordinates for the starting point
1281 x = path[0][1][N-2]
1282 y = path[0][1][N-1]
1283 else: # relative move, accumulate coordinates for the starting point
1284 (x,y) = path[0][1][0],path[0][1][1]
1285 n = 2
1286 while n < N:
1287 x += path[0][1][n]
1288 y += path[0][1][n+1]
1289 n += 2
1291 # now we have the starting point at x,y so let's save it
1292 (startx,starty) = (x,y)
1294 # convert absolute coordinates into relative ones (start with the second subcommand
1295 # and leave the first M as absolute)
1296 newPath = [path[0]]
1297 for (cmd,data) in path[1:]:
1298 i = 0
1299 newCmd = cmd
1300 newData = data
1301 # adjust abs to rel
1302 # only the A command has some values that we don't want to adjust (radii, rotation, flags)
1303 if cmd == 'A':
1304 newCmd = 'a'
1305 newData = []
1306 while i < len(data):
1307 newData.append(data[i])
1308 newData.append(data[i+1])
1309 newData.append(data[i+2])
1310 newData.append(data[i+3])
1311 newData.append(data[i+4])
1312 newData.append(data[i+5]-x)
1313 newData.append(data[i+6]-y)
1314 x = data[i+5]
1315 y = data[i+6]
1316 i += 7
1317 elif cmd == 'a':
1318 while i < len(data):
1319 x += data[i+5]
1320 y += data[i+6]
1321 i += 7
1322 elif cmd == 'H':
1323 newCmd = 'h'
1324 newData = []
1325 while i < len(data):
1326 newData.append(data[i]-x)
1327 x = data[i]
1328 i += 1
1329 elif cmd == 'h':
1330 while i < len(data):
1331 x += data[i]
1332 i += 1
1333 elif cmd == 'V':
1334 newCmd = 'v'
1335 newData = []
1336 while i < len(data):
1337 newData.append(data[i] - y)
1338 y = data[i]
1339 i += 1
1340 elif cmd == 'v':
1341 while i < len(data):
1342 y += data[i]
1343 i += 1
1344 elif cmd in ['M']:
1345 newCmd = cmd.lower()
1346 newData = []
1347 startx = data[0]
1348 starty = data[1]
1349 while i < len(data):
1350 newData.append( data[i] - x )
1351 newData.append( data[i+1] - y )
1352 x = data[i]
1353 y = data[i+1]
1354 i += 2
1355 elif cmd in ['L','T']:
1356 newCmd = cmd.lower()
1357 newData = []
1358 while i < len(data):
1359 newData.append( data[i] - x )
1360 newData.append( data[i+1] - y )
1361 x = data[i]
1362 y = data[i+1]
1363 i += 2
1364 elif cmd in ['m']:
1365 startx += data[0]
1366 starty += data[1]
1367 while i < len(data):
1368 x += data[i]
1369 y += data[i+1]
1370 i += 2
1371 elif cmd in ['l','t']:
1372 while i < len(data):
1373 x += data[i]
1374 y += data[i+1]
1375 i += 2
1376 elif cmd in ['S','Q']:
1377 newCmd = cmd.lower()
1378 newData = []
1379 while i < len(data):
1380 newData.append( data[i] - x )
1381 newData.append( data[i+1] - y )
1382 newData.append( data[i+2] - x )
1383 newData.append( data[i+3] - y )
1384 x = data[i+2]
1385 y = data[i+3]
1386 i += 4
1387 elif cmd in ['s','q']:
1388 while i < len(data):
1389 x += data[i+2]
1390 y += data[i+3]
1391 i += 4
1392 elif cmd == 'C':
1393 newCmd = 'c'
1394 newData = []
1395 while i < len(data):
1396 newData.append( data[i] - x )
1397 newData.append( data[i+1] - y )
1398 newData.append( data[i+2] - x )
1399 newData.append( data[i+3] - y )
1400 newData.append( data[i+4] - x )
1401 newData.append( data[i+5] - y )
1402 x = data[i+4]
1403 y = data[i+5]
1404 i += 6
1405 elif cmd == 'c':
1406 while i < len(data):
1407 x += data[i+4]
1408 y += data[i+5]
1409 i += 6
1410 elif cmd in ['z','Z']:
1411 x = startx
1412 y = starty
1413 newCmd = 'z'
1414 newPath.append( (newCmd, newData) )
1415 path = newPath
1417 # remove empty segments
1418 newPath = [path[0]]
1419 for (cmd,data) in path[1:]:
1420 if cmd in ['m','l','t']:
1421 newData = []
1422 i = 0
1423 while i < len(data):
1424 if data[i] != 0 or data[i+1] != 0:
1425 newData.append(data[i])
1426 newData.append(data[i+1])
1427 else:
1428 numPathSegmentsReduced += 1
1429 i += 2
1430 if newData:
1431 newPath.append( (cmd,newData) )
1432 elif cmd == 'c':
1433 newData = []
1434 i = 0
1435 while i < len(data):
1436 if data[i+4] != 0 or data[i+5] != 0:
1437 newData.append(data[i])
1438 newData.append(data[i+1])
1439 newData.append(data[i+2])
1440 newData.append(data[i+3])
1441 newData.append(data[i+4])
1442 newData.append(data[i+5])
1443 else:
1444 numPathSegmentsReduced += 1
1445 i += 6
1446 if newData:
1447 newPath.append( (cmd,newData) )
1448 elif cmd == 'a':
1449 newData = []
1450 i = 0
1451 while i < len(data):
1452 if data[i+5] != 0 or data[i+6] != 0:
1453 newData.append(data[i])
1454 newData.append(data[i+1])
1455 newData.append(data[i+2])
1456 newData.append(data[i+3])
1457 newData.append(data[i+4])
1458 newData.append(data[i+5])
1459 newData.append(data[i+6])
1460 else:
1461 numPathSegmentsReduced += 1
1462 i += 7
1463 if newData:
1464 newPath.append( (cmd,newData) )
1465 elif cmd == 'q':
1466 newData = []
1467 i = 0
1468 while i < len(data):
1469 if data[i+2] != 0 or data[i+3] != 0:
1470 newData.append(data[i])
1471 newData.append(data[i+1])
1472 newData.append(data[i+2])
1473 newData.append(data[i+3])
1474 else:
1475 numPathSegmentsReduced += 1
1476 i += 4
1477 if newData:
1478 newPath.append( (cmd,newData) )
1479 elif cmd in ['h','v']:
1480 newData = []
1481 i = 0
1482 while i < len(data):
1483 if data[i] != 0:
1484 newData.append(data[i])
1485 else:
1486 numPathSegmentsReduced += 1
1487 i += 1
1488 if newData:
1489 newPath.append( (cmd,newData) )
1490 else:
1491 newPath.append( (cmd,data) )
1492 path = newPath
1494 # convert straight curves into lines
1495 newPath = [path[0]]
1496 for (cmd,data) in path[1:]:
1497 i = 0
1498 newData = data
1499 if cmd == 'c':
1500 newData = []
1501 while i < len(data):
1502 # since all commands are now relative, we can think of previous point as (0,0)
1503 # and new point (dx,dy) is (data[i+4],data[i+5])
1504 # eqn of line will be y = (dy/dx)*x or if dx=0 then eqn of line is x=0
1505 (p1x,p1y) = (data[i],data[i+1])
1506 (p2x,p2y) = (data[i+2],data[i+3])
1507 dx = data[i+4]
1508 dy = data[i+5]
1510 foundStraightCurve = False
1512 if dx == 0:
1513 if p1x == 0 and p2x == 0:
1514 foundStraightCurve = True
1515 else:
1516 m = dy/dx
1517 if p1y == m*p1x and p2y == m*p2y:
1518 foundStraightCurve = True
1520 if foundStraightCurve:
1521 # flush any existing curve coords first
1522 if newData:
1523 newPath.append( (cmd,newData) )
1524 newData = []
1525 # now create a straight line segment
1526 newPath.append( ('l', [dx,dy]) )
1527 numCurvesStraightened += 1
1528 else:
1529 newData.append(data[i])
1530 newData.append(data[i+1])
1531 newData.append(data[i+2])
1532 newData.append(data[i+3])
1533 newData.append(data[i+4])
1534 newData.append(data[i+5])
1536 i += 6
1537 if newData or cmd == 'z' or cmd == 'Z':
1538 newPath.append( (cmd,newData) )
1539 path = newPath
1541 # collapse all consecutive commands of the same type into one command
1542 prevCmd = ''
1543 prevData = []
1544 newPath = [path[0]]
1545 for (cmd,data) in path[1:]:
1546 # flush the previous command if it is not the same type as the current command
1547 if prevCmd != '':
1548 if cmd != prevCmd:
1549 newPath.append( (prevCmd, prevData) )
1550 prevCmd = ''
1551 prevData = []
1553 # if the previous and current commands are the same type, collapse
1554 if cmd == prevCmd:
1555 for coord in data:
1556 prevData.append(coord)
1558 # save last command and data
1559 else:
1560 prevCmd = cmd
1561 prevData = data
1562 # flush last command and data
1563 if prevCmd != '':
1564 newPath.append( (prevCmd, prevData) )
1565 path = newPath
1567 # convert to shorthand path segments where possible
1568 newPath = [path[0]]
1569 for (cmd,data) in path[1:]:
1570 # convert line segments into h,v where possible
1571 if cmd == 'l':
1572 i = 0
1573 lineTuples = []
1574 while i < len(data):
1575 if data[i] == 0:
1576 # vertical
1577 if lineTuples:
1578 # flush the existing line command
1579 newPath.append( ('l', lineTuples) )
1580 lineTuples = []
1581 # append the v and then the remaining line coords
1582 newPath.append( ('v', [data[i+1]]) )
1583 numPathSegmentsReduced += 1
1584 elif data[i+1] == 0:
1585 if lineTuples:
1586 # flush the line command, then append the h and then the remaining line coords
1587 newPath.append( ('l', lineTuples) )
1588 lineTuples = []
1589 newPath.append( ('h', [data[i]]) )
1590 numPathSegmentsReduced += 1
1591 else:
1592 lineTuples.append(data[i])
1593 lineTuples.append(data[i+1])
1594 i += 2
1595 if lineTuples:
1596 newPath.append( ('l', lineTuples) )
1597 # convert Bézier curve segments into s where possible
1598 elif cmd == 'c':
1599 bez_ctl_pt = (0,0)
1600 i = 0
1601 curveTuples = []
1602 while i < len(data):
1603 # rotate by 180deg means negate both coordinates
1604 # if the previous control point is equal then we can substitute a
1605 # shorthand bezier command
1606 if bez_ctl_pt[0] == data[i] and bez_ctl_pt[1] == data[i+1]:
1607 if curveTuples:
1608 newPath.append( ('c', curveTuples) )
1609 curveTuples = []
1610 # append the s command
1611 newPath.append( ('s', [data[i+2], data[i+3], data[i+4], data[i+5]]) )
1612 numPathSegmentsReduced += 1
1613 else:
1614 j = 0
1615 while j <= 5:
1616 curveTuples.append(data[i+j])
1617 j += 1
1619 # set up control point for next curve segment
1620 bez_ctl_pt = (data[i+4]-data[i+2], data[i+5]-data[i+3])
1621 i += 6
1623 if curveTuples:
1624 newPath.append( ('c', curveTuples) )
1625 # convert quadratic curve segments into t where possible
1626 elif cmd == 'q':
1627 quad_ctl_pt = (0,0)
1628 i = 0
1629 curveTuples = []
1630 while i < len(data):
1631 if quad_ctl_pt[0] == data[i] and quad_ctl_pt[1] == data[i+1]:
1632 if curveTuples:
1633 newPath.append( ('q', curveTuples) )
1634 curveTuples = []
1635 # append the t command
1636 newPath.append( ('t', [data[i+2], data[i+3]]) )
1637 numPathSegmentsReduced += 1
1638 else:
1639 j = 0;
1640 while j <= 3:
1641 curveTuples.append(data[i+j])
1642 j += 1
1644 quad_ctl_pt = (data[i+2]-data[i], data[i+3]-data[i+1])
1645 i += 4
1647 if curveTuples:
1648 newPath.append( ('q', curveTuples) )
1649 else:
1650 newPath.append( (cmd, data) )
1651 path = newPath
1653 # for each h or v, collapse unnecessary coordinates that run in the same direction
1654 # i.e. "h-100-100" becomes "h-200" but "h300-100" does not change
1655 newPath = [path[0]]
1656 for (cmd,data) in path[1:]:
1657 if cmd in ['h','v'] and len(data) > 1:
1658 newData = []
1659 prevCoord = data[0]
1660 for coord in data[1:]:
1661 if isSameSign(prevCoord, coord):
1662 prevCoord += coord
1663 numPathSegmentsReduced += 1
1664 else:
1665 newData.append(prevCoord)
1666 prevCoord = coord
1667 newData.append(prevCoord)
1668 newPath.append( (cmd, newData) )
1669 else:
1670 newPath.append( (cmd, data) )
1671 path = newPath
1673 # it is possible that we have consecutive h, v, c, t commands now
1674 # so again collapse all consecutive commands of the same type into one command
1675 prevCmd = ''
1676 prevData = []
1677 newPath = [path[0]]
1678 for (cmd,data) in path[1:]:
1679 # flush the previous command if it is not the same type as the current command
1680 if prevCmd != '':
1681 if cmd != prevCmd:
1682 newPath.append( (prevCmd, prevData) )
1683 prevCmd = ''
1684 prevData = []
1686 # if the previous and current commands are the same type, collapse
1687 if cmd == prevCmd:
1688 for coord in data:
1689 prevData.append(coord)
1691 # save last command and data
1692 else:
1693 prevCmd = cmd
1694 prevData = data
1695 # flush last command and data
1696 if prevCmd != '':
1697 newPath.append( (prevCmd, prevData) )
1698 path = newPath
1700 newPathStr = serializePath(path)
1701 numBytesSavedInPathData += ( len(oldPathStr) - len(newPathStr) )
1702 element.setAttribute('d', newPathStr)
1704 def parseListOfPoints(s):
1705 """
1706 Parse string into a list of points.
1708 Returns a list of containing an even number of coordinate strings
1709 """
1711 # (wsp)? comma-or-wsp-separated coordinate pairs (wsp)?
1712 # coordinate-pair = coordinate comma-or-wsp coordinate
1713 # coordinate = sign? integer
1714 nums = re.split("\\s*\\,?\\s*", s)
1715 i = 0
1716 points = []
1717 while i < len(nums):
1718 x = SVGLength(nums[i])
1719 # if we had an odd number of points, return empty
1720 if i == len(nums)-1: return []
1721 else: y = SVGLength(nums[i+1])
1723 # if the coordinates were not unitless, return empty
1724 if x.units != Unit.NONE or y.units != Unit.NONE: return []
1725 points.append( str(x.value) )
1726 points.append( str(y.value) )
1727 i += 2
1729 return points
1731 def cleanPolygon(elem):
1732 """
1733 Remove unnecessary closing point of polygon points attribute
1734 """
1735 global numPointsRemovedFromPolygon
1737 pts = parseListOfPoints(elem.getAttribute('points'))
1738 N = len(pts)/2
1739 if N >= 2:
1740 (startx,starty) = (pts[0],pts[0])
1741 (endx,endy) = (pts[len(pts)-2],pts[len(pts)-1])
1742 if startx == endx and starty == endy:
1743 pts = pts[:-2]
1744 numPointsRemovedFromPolygon += 1
1745 elem.setAttribute('points', scourCoordinates(pts))
1747 def cleanPolyline(elem):
1748 """
1749 Scour the polyline points attribute
1750 """
1751 pts = parseListOfPoints(elem.getAttribute('points'))
1752 elem.setAttribute('points', scourCoordinates(pts))
1754 def serializePath(pathObj):
1755 """
1756 Reserializes the path data with some cleanups.
1757 """
1758 pathStr = ""
1759 for (cmd,data) in pathObj:
1760 pathStr += cmd
1761 # elliptical arc commands must have comma/wsp separating the coordinates
1762 # this fixes an issue outlined in Fix https://bugs.launchpad.net/scour/+bug/412754
1763 pathStr += scourCoordinates(data, (cmd == 'a'))
1764 return pathStr
1766 def scourCoordinates(data, forceCommaWsp = False):
1767 """
1768 Serializes coordinate data with some cleanups:
1769 - removes all trailing zeros after the decimal
1770 - integerize coordinates if possible
1771 - removes extraneous whitespace
1772 - adds commas between values in a subcommand if required (or if forceCommaWsp is True)
1773 """
1774 coordsStr = ""
1775 if data != None:
1776 c = 0
1777 for coord in data:
1778 # add the scoured coordinate to the path string
1779 coordsStr += scourLength(coord)
1781 # only need the comma if the next number is non-negative or if forceCommaWsp is True
1782 if c < len(data)-1 and (forceCommaWsp or Decimal(data[c+1]) >= 0):
1783 coordsStr += ','
1784 c += 1
1785 return coordsStr
1787 def scourLength(str):
1788 length = SVGLength(str)
1789 coord = length.value
1791 # reduce to the proper number of digits
1792 coord = Decimal(unicode(coord)) * Decimal(1)
1794 # integerize if we can
1795 if int(coord) == coord: coord = Decimal(unicode(int(coord)))
1797 # Decimal.trim() is available in Python 2.6+ to trim trailing zeros
1798 try:
1799 coord = coord.trim()
1800 except AttributeError:
1801 # trim it ourselves
1802 s = unicode(coord)
1803 dec = s.find('.')
1804 if dec != -1:
1805 while s[-1] == '0':
1806 s = s[:-1]
1807 coord = Decimal(s)
1809 # Decimal.normalize() will uses scientific notation - if that
1810 # string is smaller, then use it
1811 normd = coord.normalize()
1812 if len(unicode(normd)) < len(unicode(coord)):
1813 coord = normd
1815 return unicode(coord)+Unit.str(length.units)
1817 def embedRasters(element, options) :
1818 """
1819 Converts raster references to inline images.
1820 NOTE: there are size limits to base64-encoding handling in browsers
1821 """
1822 global numRastersEmbedded
1824 href = element.getAttributeNS(NS['XLINK'],'href')
1826 # if xlink:href is set, then grab the id
1827 if href != '' and len(href) > 1:
1828 # find if href value has filename ext
1829 ext = os.path.splitext(os.path.basename(href))[1].lower()[1:]
1831 # look for 'png', 'jpg', and 'gif' extensions
1832 if ext == 'png' or ext == 'jpg' or ext == 'gif':
1834 # check if href resolves to an existing file
1835 if os.path.isfile(href) == False :
1836 if href[:7] != 'http://' and os.path.isfile(href) == False :
1837 # if this is not an absolute path, set path relative
1838 # to script file based on input arg
1839 infilename = '.'
1840 if options.infilename: infilename = options.infilename
1841 href = os.path.join(os.path.dirname(infilename), href)
1843 rasterdata = ''
1844 # test if file exists locally
1845 if os.path.isfile(href) == True :
1846 # open raster file as raw binary
1847 raster = open( href, "rb")
1848 rasterdata = raster.read()
1850 elif href[:7] == 'http://':
1851 # raster = open( href, "rb")
1852 webFile = urllib.urlopen( href )
1853 rasterdata = webFile.read()
1854 webFile.close()
1856 # ... should we remove all images which don't resolve?
1857 if rasterdata != '' :
1858 # base64-encode raster
1859 b64eRaster = base64.b64encode( rasterdata )
1861 # set href attribute to base64-encoded equivalent
1862 if b64eRaster != '':
1863 # PNG and GIF both have MIME Type 'image/[ext]', but
1864 # JPEG has MIME Type 'image/jpeg'
1865 if ext == 'jpg':
1866 ext = 'jpeg'
1868 element.setAttributeNS(NS['XLINK'], 'href', 'data:image/' + ext + ';base64,' + b64eRaster)
1869 numRastersEmbedded += 1
1870 del b64eRaster
1872 def properlySizeDoc(docElement):
1873 # get doc width and height
1874 w = SVGLength(docElement.getAttribute('width'))
1875 h = SVGLength(docElement.getAttribute('height'))
1877 # if width/height are not unitless or px then it is not ok to rewrite them into a viewBox
1878 if ((w.units != Unit.NONE and w.units != Unit.PX) or
1879 (w.units != Unit.NONE and w.units != Unit.PX)):
1880 return
1882 # else we have a statically sized image and we should try to remedy that
1884 # parse viewBox attribute
1885 vbSep = re.split("\\s*\\,?\\s*", docElement.getAttribute('viewBox'), 3)
1886 # if we have a valid viewBox we need to check it
1887 vbWidth,vbHeight = 0,0
1888 if len(vbSep) == 4:
1889 try:
1890 # if x or y are specified and non-zero then it is not ok to overwrite it
1891 vbX = float(vbSep[0])
1892 vbY = float(vbSep[1])
1893 if vbX != 0 or vbY != 0:
1894 return
1896 # if width or height are not equal to doc width/height then it is not ok to overwrite it
1897 vbWidth = float(vbSep[2])
1898 vbHeight = float(vbSep[3])
1899 if vbWidth != w.value or vbHeight != h.value:
1900 return
1901 # if the viewBox did not parse properly it is invalid and ok to overwrite it
1902 except ValueError:
1903 pass
1905 # at this point it's safe to set the viewBox and remove width/height
1906 docElement.setAttribute('viewBox', '0 0 %s %s' % (w.value, h.value))
1907 docElement.removeAttribute('width')
1908 docElement.removeAttribute('height')
1910 def remapNamespacePrefix(node, oldprefix, newprefix):
1911 if node == None or node.nodeType != 1: return
1913 if node.prefix == oldprefix:
1914 localName = node.localName
1915 namespace = node.namespaceURI
1916 doc = node.ownerDocument
1917 parent = node.parentNode
1919 # create a replacement node
1920 newNode = None
1921 if newprefix != '':
1922 newNode = doc.createElementNS(namespace, newprefix+":"+localName)
1923 else:
1924 newNode = doc.createElement(localName);
1926 # add all the attributes
1927 attrList = node.attributes
1928 for i in range(attrList.length):
1929 attr = attrList.item(i)
1930 newNode.setAttributeNS( attr.namespaceURI, attr.localName, attr.nodeValue)
1932 # clone and add all the child nodes
1933 for child in node.childNodes:
1934 newNode.appendChild(child.cloneNode(True))
1936 # replace old node with new node
1937 parent.replaceChild( newNode, node )
1938 # set the node to the new node in the remapped namespace prefix
1939 node = newNode
1941 # now do all child nodes
1942 for child in node.childNodes :
1943 remapNamespacePrefix(child, oldprefix, newprefix)
1945 def makeWellFormed(str):
1946 newstr = str
1948 # encode & as & ( must do this first so that < does not become &lt; )
1949 if str.find('&') != -1:
1950 newstr = str.replace('&', '&')
1952 # encode < as <
1953 if str.find("<") != -1:
1954 newstr = str.replace('<', '<')
1956 # encode > as > (TODO: is this necessary?)
1957 if str.find('>') != -1:
1958 newstr = str.replace('>', '>')
1960 return newstr
1962 # hand-rolled serialization function that has the following benefits:
1963 # - pretty printing
1964 # - somewhat judicious use of whitespace
1965 # - ensure id attributes are first
1966 def serializeXML(element, options, ind = 0):
1967 indent = ind
1968 I=''
1969 if options.indent_type == 'tab': I='\t'
1970 elif options.indent_type == 'space': I=' '
1972 outString = (I * ind) + '<' + element.nodeName
1974 # always serialize the id or xml:id attributes first
1975 if element.getAttribute('id') != '':
1976 id = element.getAttribute('id')
1977 quot = '"'
1978 if id.find('"') != -1:
1979 quot = "'"
1980 outString += ' ' + 'id=' + quot + id + quot
1981 if element.getAttribute('xml:id') != '':
1982 id = element.getAttribute('xml:id')
1983 quot = '"'
1984 if id.find('"') != -1:
1985 quot = "'"
1986 outString += ' ' + 'xml:id=' + quot + id + quot
1988 # now serialize the other attributes
1989 attrList = element.attributes
1990 for num in range(attrList.length) :
1991 attr = attrList.item(num)
1992 if attr.nodeName == 'id' or attr.nodeName == 'xml:id': continue
1993 # if the attribute value contains a double-quote, use single-quotes
1994 quot = '"'
1995 if attr.nodeValue.find('"') != -1:
1996 quot = "'"
1998 attrValue = makeWellFormed( attr.nodeValue )
2000 outString += ' '
2001 # preserve xmlns: if it is a namespace prefix declaration
2002 if attr.namespaceURI == 'http://www.w3.org/2000/xmlns/' and attr.nodeName.find('xmlns') == -1:
2003 outString += 'xmlns:'
2004 outString += attr.nodeName + '=' + quot + attrValue + quot
2006 # if no children, self-close
2007 children = element.childNodes
2008 if children.length > 0:
2009 outString += '>'
2011 onNewLine = False
2012 for child in element.childNodes:
2013 # element node
2014 if child.nodeType == 1:
2015 outString += '\n' + serializeXML(child, options, indent + 1)
2016 onNewLine = True
2017 # text node
2018 elif child.nodeType == 3:
2019 # trim it only in the case of not being a child of an element
2020 # where whitespace might be important
2021 if element.nodeName in ["text", "tspan", "textPath", "tref", "title", "desc", "textArea"]:
2022 outString += makeWellFormed(child.nodeValue)
2023 else:
2024 outString += makeWellFormed(child.nodeValue.strip())
2025 # CDATA node
2026 elif child.nodeType == 4:
2027 outString += '<![CDATA[' + child.nodeValue + ']]>'
2028 # Comment node
2029 elif child.nodeType == 8:
2030 outString += '<!--' + child.nodeValue + '-->'
2031 # TODO: entities, processing instructions, what else?
2032 else: # ignore the rest
2033 pass
2035 if onNewLine: outString += (I * ind)
2036 outString += '</' + element.nodeName + '>'
2037 if indent > 0: outString += '\n'
2038 else:
2039 outString += '/>'
2040 if indent > 0: outString += '\n'
2042 return outString
2044 # this is the main method
2045 # input is a string representation of the input XML
2046 # returns a string representation of the output XML
2047 def scourString(in_string, options=None):
2048 if options is None:
2049 options = _options_parser.get_default_values()
2050 getcontext().prec = options.digits
2051 global numAttrsRemoved
2052 global numStylePropsFixed
2053 global numElemsRemoved
2054 global numBytesSavedInColors
2055 doc = xml.dom.minidom.parseString(in_string)
2057 # for whatever reason this does not always remove all inkscape/sodipodi attributes/elements
2058 # on the first pass, so we do it multiple times
2059 # does it have to do with removal of children affecting the childlist?
2060 if options.keep_editor_data == False:
2061 while removeNamespacedElements( doc.documentElement, unwanted_ns ) > 0 :
2062 pass
2063 while removeNamespacedAttributes( doc.documentElement, unwanted_ns ) > 0 :
2064 pass
2066 # remove the xmlns: declarations now
2067 xmlnsDeclsToRemove = []
2068 attrList = doc.documentElement.attributes
2069 for num in range(attrList.length) :
2070 if attrList.item(num).nodeValue in unwanted_ns :
2071 xmlnsDeclsToRemove.append(attrList.item(num).nodeName)
2073 for attr in xmlnsDeclsToRemove :
2074 doc.documentElement.removeAttribute(attr)
2075 numAttrsRemoved += 1
2077 # ensure namespace for SVG is declared
2078 # TODO: what if the default namespace is something else (i.e. some valid namespace)?
2079 if doc.documentElement.getAttribute('xmlns') != 'http://www.w3.org/2000/svg':
2080 doc.documentElement.setAttribute('xmlns', 'http://www.w3.org/2000/svg')
2081 # TODO: throw error or warning?
2083 # check for redundant SVG namespace declaration
2084 attrList = doc.documentElement.attributes
2085 xmlnsDeclsToRemove = []
2086 redundantPrefixes = []
2087 for i in range(attrList.length):
2088 attr = attrList.item(i)
2089 name = attr.nodeName
2090 val = attr.nodeValue
2091 if name[0:6] == 'xmlns:' and val == 'http://www.w3.org/2000/svg':
2092 redundantPrefixes.append(name[6:])
2093 xmlnsDeclsToRemove.append(name)
2095 for attrName in xmlnsDeclsToRemove:
2096 doc.documentElement.removeAttribute(attrName)
2098 for prefix in redundantPrefixes:
2099 remapNamespacePrefix(doc.documentElement, prefix, '')
2101 # repair style (remove unnecessary style properties and change them into XML attributes)
2102 numStylePropsFixed = repairStyle(doc.documentElement, options)
2104 # convert colors to #RRGGBB format
2105 if options.simple_colors:
2106 numBytesSavedInColors = convertColors(doc.documentElement)
2108 # remove empty defs, metadata, g
2109 # NOTE: these elements will be removed even if they have (invalid) text nodes
2110 elemsToRemove = []
2111 for tag in ['defs', 'metadata', 'g'] :
2112 for elem in doc.documentElement.getElementsByTagName(tag) :
2113 removeElem = not elem.hasChildNodes()
2114 if removeElem == False :
2115 for child in elem.childNodes :
2116 if child.nodeType in [1, 3, 4, 8] :
2117 break
2118 else:
2119 removeElem = True
2120 if removeElem :
2121 elem.parentNode.removeChild(elem)
2122 numElemsRemoved += 1
2124 # remove unreferenced gradients/patterns outside of defs
2125 while removeUnreferencedElements(doc) > 0:
2126 pass
2128 if options.strip_ids:
2129 bContinueLooping = True
2130 while bContinueLooping:
2131 identifiedElements = findElementsWithId(doc.documentElement)
2132 referencedIDs = findReferencedElements(doc.documentElement)
2133 bContinueLooping = (removeUnreferencedIDs(referencedIDs, identifiedElements) > 0)
2135 if options.group_collapse:
2136 while removeNestedGroups(doc.documentElement) > 0:
2137 pass
2139 # move common attributes to parent group
2140 # TODO: should make sure this is called with most-nested groups first
2141 numAttrsRemoved += moveCommonAttributesToParentGroup(doc.documentElement)
2143 while removeDuplicateGradientStops(doc) > 0:
2144 pass
2146 # remove gradients that are only referenced by one other gradient
2147 while collapseSinglyReferencedGradients(doc) > 0:
2148 pass
2150 # remove duplicate gradients
2151 while removeDuplicateGradients(doc) > 0:
2152 pass
2154 # clean path data
2155 for elem in doc.documentElement.getElementsByTagName('path') :
2156 if elem.getAttribute('d') == '':
2157 elem.parentNode.removeChild(elem)
2158 else:
2159 cleanPath(elem)
2161 # remove unnecessary closing point of polygons and scour points
2162 for polygon in doc.documentElement.getElementsByTagName('polygon') :
2163 cleanPolygon(polygon)
2165 # scour points of polyline
2166 for polyline in doc.documentElement.getElementsByTagName('polyline') :
2167 cleanPolygon(polyline)
2169 # scour lengths (including coordinates)
2170 for type in ['svg', 'image', 'rect', 'circle', 'ellipse', 'line', 'linearGradient', 'radialGradient', 'stop']:
2171 for elem in doc.getElementsByTagName(type):
2172 for attr in ['x', 'y', 'width', 'height', 'cx', 'cy', 'r', 'rx', 'ry',
2173 'x1', 'y1', 'x2', 'y2', 'fx', 'fy', 'offset', 'opacity',
2174 'fill-opacity', 'stroke-opacity', 'stroke-width', 'stroke-miterlimit']:
2175 if elem.getAttribute(attr) != '':
2176 elem.setAttribute(attr, scourLength(elem.getAttribute(attr)))
2178 # remove default values of attributes
2179 numAttrsRemoved += removeDefaultAttributeValues(doc.documentElement, options)
2181 # convert rasters references to base64-encoded strings
2182 if options.embed_rasters:
2183 for elem in doc.documentElement.getElementsByTagName('image') :
2184 embedRasters(elem, options)
2186 # properly size the SVG document (ideally width/height should be 100% with a viewBox)
2187 properlySizeDoc(doc.documentElement)
2189 # output the document as a pretty string with a single space for indent
2190 # NOTE: removed pretty printing because of this problem:
2191 # http://ronrothman.com/public/leftbraned/xml-dom-minidom-toprettyxml-and-silly-whitespace/
2192 # rolled our own serialize function here to save on space, put id first, customize indentation, etc
2193 # out_string = doc.documentElement.toprettyxml(' ')
2194 out_string = serializeXML(doc.documentElement, options)
2196 # now strip out empty lines
2197 lines = []
2198 # Get rid of empty lines
2199 for line in out_string.splitlines(True):
2200 if line.strip():
2201 lines.append(line)
2203 # return the string stripped of empty lines
2204 if options.strip_xml_prolog == False:
2205 xmlprolog = '<?xml version="1.0" encoding="UTF-8" standalone="no"?>\n'
2206 else:
2207 xmlprolog = ""
2209 return xmlprolog + "".join(lines)
2211 # used mostly by unit tests
2212 # input is a filename
2213 # returns the minidom doc representation of the SVG
2214 def scourXmlFile(filename, options=None):
2215 in_string = open(filename).read()
2216 out_string = scourString(in_string, options)
2217 return xml.dom.minidom.parseString(out_string.encode('utf-8'))
2219 # GZ: Seems most other commandline tools don't do this, is it really wanted?
2220 class HeaderedFormatter(optparse.IndentedHelpFormatter):
2221 """
2222 Show application name, version number, and copyright statement
2223 above usage information.
2224 """
2225 def format_usage(self, usage):
2226 return "%s %s\n%s\n%s" % (APP, VER, COPYRIGHT,
2227 optparse.IndentedHelpFormatter.format_usage(self, usage))
2229 # GZ: would prefer this to be in a function or class scope, but tests etc need
2230 # access to the defaults anyway
2231 _options_parser = optparse.OptionParser(
2232 usage="%prog [-i input.svg] [-o output.svg] [OPTIONS]",
2233 description=("If the input/output files are specified with a svgz"
2234 " extension, then compressed SVG is assumed. If the input file is not"
2235 " specified, stdin is used. If the output file is not specified, "
2236 " stdout is used."),
2237 formatter=HeaderedFormatter(max_help_position=30),
2238 version=VER)
2240 _options_parser.add_option("--disable-simplify-colors",
2241 action="store_false", dest="simple_colors", default=True,
2242 help="won't convert all colors to #RRGGBB format")
2243 _options_parser.add_option("--disable-style-to-xml",
2244 action="store_false", dest="style_to_xml", default=True,
2245 help="won't convert styles into XML attributes")
2246 _options_parser.add_option("--disable-group-collapsing",
2247 action="store_false", dest="group_collapse", default=True,
2248 help="won't collapse <g> elements")
2249 _options_parser.add_option("--enable-id-stripping",
2250 action="store_true", dest="strip_ids", default=False,
2251 help="remove all un-referenced ID attributes")
2252 _options_parser.add_option("--disable-embed-rasters",
2253 action="store_false", dest="embed_rasters", default=True,
2254 help="won't embed rasters as base64-encoded data")
2255 _options_parser.add_option("--keep-editor-data",
2256 action="store_true", dest="keep_editor_data", default=False,
2257 help="won't remove Inkscape, Sodipodi or Adobe Illustrator elements and attributes")
2258 _options_parser.add_option("--strip-xml-prolog",
2259 action="store_true", dest="strip_xml_prolog", default=False,
2260 help="won't output the <?xml ?> prolog")
2262 # GZ: this is confusing, most people will be thinking in terms of
2263 # decimal places, which is not what decimal precision is doing
2264 _options_parser.add_option("-p", "--set-precision",
2265 action="store", type=int, dest="digits", default=5,
2266 help="set number of significant digits (default: %default)")
2267 _options_parser.add_option("-i",
2268 action="store", dest="infilename", help=optparse.SUPPRESS_HELP)
2269 _options_parser.add_option("-o",
2270 action="store", dest="outfilename", help=optparse.SUPPRESS_HELP)
2271 _options_parser.add_option("--indent",
2272 action="store", type="string", dest="indent_type", default="space",
2273 help="indentation of the output: none, space, tab (default: %default)")
2275 def maybe_gziped_file(filename, mode="r"):
2276 if os.path.splitext(filename)[1].lower() in (".svgz", ".gz"):
2277 return gzip.GzipFile(filename, mode)
2278 return file(filename, mode)
2280 def parse_args(args=None):
2281 options, rargs = _options_parser.parse_args(args)
2283 if rargs:
2284 _options_parser.error("Additional arguments not handled: %r, see --help" % rargs)
2285 if options.digits < 0:
2286 _options_parser.error("Can't have negative significant digits, see --help")
2287 if not options.indent_type in ["tab", "space", "none"]:
2288 _options_parser.error("Invalid value for --indent, see --help")
2290 if options.infilename:
2291 infile = maybe_gziped_file(options.infilename)
2292 # GZ: could catch a raised IOError here and report
2293 else:
2294 # GZ: could sniff for gzip compression here
2295 infile = sys.stdin
2296 if options.outfilename:
2297 outfile = maybe_gziped_file(options.outfilename, "w")
2298 else:
2299 outfile = sys.stdout
2301 return options, [infile, outfile]
2303 def getReport():
2304 return ' Number of elements removed: ' + str(numElemsRemoved) + \
2305 '\n Number of attributes removed: ' + str(numAttrsRemoved) + \
2306 '\n Number of unreferenced id attributes removed: ' + str(numIDsRemoved) + \
2307 '\n Number of style properties fixed: ' + str(numStylePropsFixed) + \
2308 '\n Number of raster images embedded inline: ' + str(numRastersEmbedded) + \
2309 '\n Number of path segments reduced/removed: ' + str(numPathSegmentsReduced) + \
2310 '\n Number of bytes saved in path data: ' + str(numBytesSavedInPathData) + \
2311 '\n Number of bytes saved in colors: ' + str(numBytesSavedInColors) + \
2312 '\n Number of points removed from polygons: ' + str(numPointsRemovedFromPolygon)
2314 if __name__ == '__main__':
2315 if sys.platform == "win32":
2316 from time import clock as get_tick
2317 else:
2318 # GZ: is this different from time.time() in any way?
2319 def get_tick():
2320 return os.times()[0]
2322 start = get_tick()
2324 options, (input, output) = parse_args()
2326 print >>sys.stderr, "%s %s\n%s" % (APP, VER, COPYRIGHT)
2328 # do the work
2329 in_string = input.read()
2330 out_string = scourString(in_string, options).encode("UTF-8")
2331 output.write(out_string)
2333 # Close input and output files
2334 input.close()
2335 output.close()
2337 end = get_tick()
2339 # GZ: unless silenced by -q or something?
2340 # GZ: not using globals would be good too
2341 print >>sys.stderr, ' File:', input.name, \
2342 '\n Time taken:', str(end-start) + 's\n', \
2343 getReport()
2345 oldsize = len(in_string)
2346 newsize = len(out_string)
2347 sizediff = (newsize / oldsize) * 100
2348 print >>sys.stderr, ' Original file size:', oldsize, 'bytes;', \
2349 'new file size:', newsize, 'bytes (' + str(sizediff)[:5] + '%)'