OLD | NEW |
| (Empty) |
1 #!/usr/local/bin/python | |
2 # coding: utf-8 | |
3 | |
4 import sys | |
5 import re | |
6 | |
7 # TODO(edisonn): put processed part of file in a new file | |
8 # put unprocessed part, in a new file, so we see what we miss | |
9 # keep blank lines, and generate a version without the blank lines | |
10 | |
11 #TODO (edisonn): deal manually with tables that don't have "KEY TYPE VALUE' head
er, e.g. | |
12 # TABLE 7.11 Restrictions on the entries in a soft-mask image dictionary | |
13 #KEY RESTRICTION | |
14 | |
15 | |
16 | |
17 lines = 0 | |
18 table = '' | |
19 tableHeaderFound = False | |
20 tableLine = 0 | |
21 tableRow = 0 | |
22 columnWidth = [] | |
23 columnValues = None | |
24 mustFollowTableHeader = False | |
25 emitedDitionaryName = '' | |
26 | |
27 knownTypes = { | |
28 '(any)', | |
29 unicode('undefined', 'utf8'), | |
30 '(undefined)', | |
31 '(various)', | |
32 'array', | |
33 'or', | |
34 'boolean', | |
35 'date', | |
36 'dictionary', | |
37 'function', | |
38 'integer', | |
39 unicode('file', 'utf8'), | |
40 'file', | |
41 unicode('specification', 'utf8'), | |
42 'specification', | |
43 'name', | |
44 'tree', | |
45 'number', | |
46 'rectangle', | |
47 'stream', | |
48 'string', | |
49 'text', | |
50 ',', | |
51 ' ' | |
52 } | |
53 | |
54 # TODO(edisonn): add a third element in the vector, the base class, by default i
t is Dictionary | |
55 # TODO(edisonn): add overrides for types map<field_name, type_name> | |
56 # e.g. ,{'Resources', 'ResourceDictionary'} | |
57 # TODO(edisonn): can be added one by one, or extracted from documentation | |
58 | |
59 tableToClassName = { | |
60 'TABLE 3.4': ['StreamCommonDictionary', 'Entries common to all stream dictionari
es'], | |
61 'TABLE 3.7': ['LzwdecodeAndFlatedecodeFiltersDictionary', 'Optional parameters f
or LZWDecode and FlateDecode filters'], | |
62 'TABLE 3.9': ['CcittfaxdecodeFilterDictionary', 'Optional parameters for the CCI
TTFaxDecode filter'], | |
63 'TABLE 3.10': ['Jbig2DecodeFilterDictionary', 'Optional parameter for the JBIG2D
ecode filter'], | |
64 'TABLE 3.11': ['DctdecodeFilterDictionary', 'Optional parameter for the DCTDecod
e filter'], | |
65 'TABLE 3.12': ['FileTrailerDictionary', 'Entries in the file trailer dictionary'
], | |
66 'TABLE 3.13': ['EncryptionCommonDictionary', 'Entries common to all encryption d
ictionaries'], | |
67 'TABLE 3.14': ['StandardSecurityHandlerDictionary', 'Additional encryption dicti
onary entries for the standard security handler'], | |
68 'TABLE 3.16': ['CatalogDictionary', 'Entries in the catalog dictionary'], | |
69 'TABLE 3.17': ['PageTreeNodeDictionary', 'Required entries in a page tree node']
, | |
70 'TABLE 3.18': ['PageObjectDictionary', 'Entries in a page object'], | |
71 'TABLE 3.19': ['NameDictionary', 'Entries in the name dictionary'], | |
72 'TABLE 3.21': ['ResourceDictionary', 'Entries in a resource dictionary'], | |
73 'TABLE 3.23': ['NameTreeNodeDictionary', 'Entries in a name tree node dictionary
'], | |
74 'TABLE 3.25': ['NumberTreeNodeDictionary', 'Entries in a number tree node dictio
nary'], | |
75 'TABLE 3.26': ['FunctionCommonDictionary', 'Entries common to all function dicti
onaries'], | |
76 'TABLE 3.27': ['Type0FunctionDictionary', 'Additional entries specific to a type
0 function dictionary'], | |
77 'TABLE 3.28': ['Type2FunctionDictionary', 'Additional entries specific to a type
2 function dictionary'], | |
78 'TABLE 3.29': ['Type3FunctionDictionary', 'Additional entries specific to a type
3 function dictionary'], | |
79 'TABLE 3.32': ['FileSpecificationDictionary', 'Entries in a file specification d
ictionary'], | |
80 'TABLE 3.33': ['EmbeddedFileStreamDictionary', 'Additional entries in an embedde
d file stream dictionary'], | |
81 'TABLE 3.34': ['EmbeddedFileParameterDictionary', 'Entries in an embedded file p
arameter dictionary'], | |
82 'TABLE 3.35': ['MacOsFileInformationDictionary', 'Entries in a Mac OS file infor
mation dictionary'], | |
83 'TABLE 4.8': ['GraphicsStateDictionary', 'Entries in a graphics state parameter
dictionary'], | |
84 'TABLE 4.13': ['CalgrayColorSpaceDictionary', 'Entries in a CalGray color space
dictionary'], | |
85 'TABLE 4.14': ['CalrgbColorSpaceDictionary', 'Entries in a CalRGB color space di
ctionary'], | |
86 'TABLE 4.15': ['LabColorSpaceDictionary', 'Entries in a Lab color space dictiona
ry'], | |
87 'TABLE 4.16': ['IccProfileStreamDictionary', 'Additional entries specific to an
ICC profile stream dictionary'], | |
88 'TABLE 4.20': ['DeviceNColorSpaceDictionary', 'Entry in a DeviceN color space at
tributes dictionary'], | |
89 'TABLE 4.22': ['Type1PatternDictionary', 'Additional entries specific to a type
1 pattern dictionary'], | |
90 'TABLE 4.23': ['Type2PatternDictionary', 'Entries in a type 2 pattern dictionary
'], | |
91 'TABLE 4.25': ['ShadingDictionary', 'Entries common to all shading dictionaries'
], | |
92 'TABLE 4.26': ['Type1ShadingDictionary', 'Additional entries specific to a type
1 shading dictionary', 'ShadingDictionary'], | |
93 'TABLE 4.27': ['Type2ShadingDictionary', 'Additional entries specific to a type
2 shading dictionary', 'ShadingDictionary'], | |
94 'TABLE 4.28': ['Type3ShadingDictionary', 'Additional entries specific to a type
3 shading dictionary', 'ShadingDictionary'], | |
95 'TABLE 4.29': ['Type4ShadingDictionary', 'Additional entries specific to a type
4 shading dictionary', 'ShadingDictionary'], | |
96 'TABLE 4.30': ['Type5ShadingDictionary', 'Additional entries specific to a type
5 shading dictionary', 'ShadingDictionary'], | |
97 'TABLE 4.31': ['Type6ShadingDictionary', 'Additional entries specific to a type
6 shading dictionary', 'ShadingDictionary'], | |
98 'TABLE 4.35': ['ImageDictionary', 'Additional entries specific to an image dicti
onary', 'XObjectDictionary', {'Subtype': '[datatypes.PdfName(\'Image\')]'}], | |
99 'TABLE 4.37': ['AlternateImageDictionary', 'Entries in an alternate image dictio
nary'], | |
100 'TABLE 4.41': ['Type1FormDictionary', 'Additional entries specific to a type 1 f
orm dictionary', 'XObjectDictionary', {'Subtype': '[datatypes.PdfName(\'Form\')]
'}], | |
101 'TABLE 4.42': ['GroupAttributesDictionary', 'Entries common to all group attribu
tes dictionaries'], | |
102 'TABLE 4.43': ['ReferenceDictionary', 'Entries in a reference dictionary'], | |
103 'TABLE 4.44': ['PSXobjectDictionary', 'Additional entries specific to a PostScri
pt XObject dictionary'], | |
104 'TABLE 5.8': ['Type1FontDictionary', 'Entries in a Type 1 font dictionary', 'Fon
tDictionary', {'Subtype': '[datatypes.PdfName(\'Type1\')]'}], | |
105 'TABLE 5.9': ['Type3FontDictionary', 'Entries in a Type 3 font dictionary', 'Typ
e0FontDictionary', {'Subtype': '[datatypes.PdfName(\'Type3\')]'}], | |
106 'TABLE 5.11': ['EncodingDictionary', 'Entries in an encoding dictionary'], | |
107 'TABLE 5.12': ['CIDSystemInfoDictionary', 'Entries in a CIDSystemInfo dictionary
'], | |
108 'TABLE 5.13': ['CIDFontDictionary', 'Entries in a CIDFont dictionary', 'FontDict
ionary', {'Subtype': '[datatypes.PdfName(\'CIDFontType0\'), datatypes.PdfName(\'
CIDFontType2\')]'}], | |
109 'TABLE 5.16': ['CMapDictionary', 'Additional entries in a CMap dictionary'], | |
110 'TABLE 5.17': ['Type0FontDictionary', 'Entries in a Type 0 font dictionary', 'Fo
ntDictionary', {'Subtype': '[datatypes.PdfName(\'Type0\')]'}], | |
111 'TABLE 5.18': ['FontDescriptorDictionary', 'Entries common to all font descripto
rs'], | |
112 'TABLE 5.20': ['CIDFontDescriptorDictionary', 'Additional font descriptor entrie
s for CIDFonts'], | |
113 'TABLE 5.23': ['EmbeddedFontStreamDictionary', 'Additional entries in an embedde
d font stream dictionary'], | |
114 'TABLE 6.3': ['Type1HalftoneDictionary', 'Entries in a type 1 halftone dictionar
y'], | |
115 'TABLE 6.4': ['Type6HalftoneDictionary', 'Additional entries specific to a type
6 halftone dictionary'], | |
116 'TABLE 6.5': ['Type10HalftoneDictionary', 'Additional entries specific to a type
10 halftone dictionary'], | |
117 'TABLE 6.6': ['Type16HalftoneDictionary', 'Additional entries specific to a type
16 halftone dictionary'], | |
118 'TABLE 6.7': ['Type5HalftoneDictionary', 'Entries in a type 5 halftone dictionar
y'], | |
119 'TABLE 7.10': ['SoftMaskDictionary', 'Entries in a soft-mask dictionary'], | |
120 'TABLE 7.12': ['SoftMaskImageDictionary', 'Additional entry in a soft-mask image
dictionary'], | |
121 'TABLE 7.13': ['TransparencyGroupDictionary', 'Additional entries specific to a
transparency group attributes dictionary'], | |
122 'TABLE 8.1': ['ViewerPreferencesDictionary', 'Entries in a viewer preferences di
ctionary'], | |
123 'TABLE 8.3': ['OutlineDictionary', 'Entries in the outline dictionary'], | |
124 'TABLE 8.4': ['OutlineItemDictionary', 'Entries in an outline item dictionary'], | |
125 'TABLE 8.6': ['PageLabelDictionary', 'Entries in a page label dictionary'], | |
126 'TABLE 8.7': ['ThreadDictionary', 'Entries in a thread dictionary'], | |
127 'TABLE 8.8': ['BeadDictionary', 'Entries in a bead dictionary'], | |
128 'TABLE 8.9': ['TransitionDictionary', 'Entries in a transition dictionary'], | |
129 'TABLE 8.10': ['AnnotationDictionary', 'Entries common to all annotation diction
aries'], | |
130 'TABLE 8.12': ['BorderStyleDictionary', 'Entries in a border style dictionary'], | |
131 'TABLE 8.13': ['AppearanceDictionary', 'Entries in an appearance dictionary'], | |
132 'TABLE 8.15': ['TextAnnotationDictionary', 'Additional entries specific to a tex
t annotation'], | |
133 'TABLE 8.16': ['ALinkAnnotationDictionary', 'Additional entries specific to a li
nk annotation'], | |
134 'TABLE 8.17': ['FreeTextAnnotationDictionary', 'Additional entries specific to a
free text annotation'], | |
135 'TABLE 8.18': ['LineAnnotationDictionary', 'Additional entries specific to a lin
e annotation'], | |
136 'TABLE 8.20': ['SquareOrCircleAnnotation', 'Additional entries specific to a squ
are or circle annotation'], | |
137 'TABLE 8.21': ['MarkupAnnotationsDictionary', 'Additional entries specific to ma
rkup annotations'], | |
138 'TABLE 8.22': ['RubberStampAnnotationDictionary', 'Additional entries specific t
o a rubber stamp annotation'], | |
139 'TABLE 8.23': ['InkAnnotationDictionary', 'Additional entries specific to an ink
annotation'], | |
140 'TABLE 8.24': ['PopUpAnnotationDictionary', 'Additional entries specific to a po
p-up annotation'], | |
141 'TABLE 8.25': ['FileAttachmentAnnotationDictionary', 'Additional entries specifi
c to a file attachment annotation'], | |
142 'TABLE 8.26': ['SoundAnnotationDictionary', 'Additional entries specific to a so
und annotation'], | |
143 'TABLE 8.27': ['MovieAnnotationDictionary', 'Additional entries specific to a mo
vie annotation'], | |
144 'TABLE 8.28': ['WidgetAnnotationDictionary', 'Additional entries specific to a w
idget annotation'], | |
145 'TABLE 8.29': ['ActionDictionary', 'Entries common to all action dictionaries'], | |
146 'TABLE 8.30': ['AnnotationActionsDictionary', 'Entries in an annotation\'s addit
ional-actions dictionary'], | |
147 'TABLE 8.31': ['PageObjectActionsDictionary', 'Entries in a page object\'s addit
ional-actions dictionary'], | |
148 'TABLE 8.32': ['FormFieldActionsDictionary', 'Entries in a form field\'s additio
nal-actions dictionary'], | |
149 'TABLE 8.33': ['DocumentCatalogActionsDictionary', 'Entries in the document cata
log\'s additional-actions dictionary'], | |
150 'TABLE 8.35': ['GoToActionDictionary', 'Additional entries specific to a go-to a
ction'], | |
151 'TABLE 8.36': ['RemoteGoToActionDictionary', 'Additional entries specific to a r
emote go-to action'], | |
152 'TABLE 8.37': ['LaunchActionDictionary', 'Additional entries specific to a launc
h action'], | |
153 'TABLE 8.38': ['WindowsLaunchActionDictionary', 'Entries in a Windows launch par
ameter dictionary'], | |
154 'TABLE 8.39': ['ThreadActionDictionary', 'Additional entries specific to a threa
d action'], | |
155 'TABLE 8.40': ['URIActionDictionary', 'Additional entries specific to a URI acti
on'], | |
156 'TABLE 8.41': ['URIDictionary', 'Entry in a URI dictionary'], | |
157 'TABLE 8.42': ['SoundActionDictionary', 'Additional entries specific to a sound
action'], | |
158 'TABLE 8.43': ['MovieActionDictionary', 'Additional entries specific to a movie
action'], | |
159 'TABLE 8.44': ['HideActionDictionary', 'Additional entries specific to a hide ac
tion'], | |
160 'TABLE 8.46': ['NamedActionsDictionary', 'Additional entries specific to named a
ctions'], | |
161 'TABLE 8.47': ['InteractiveFormDictionary', 'Entries in the interactive form dic
tionary'], | |
162 'TABLE 8.49': ['FieldDictionary', 'Entries common to all field dictionaries'], | |
163 'TABLE 8.51': ['VariableTextFieldDictionary', 'Additional entries common to all
fields containing variable text'], | |
164 'TABLE 8.52': ['AppearanceCharacteristicsDictionary', 'Entries in an appearance
characteristics dictionary'], | |
165 'TABLE 8.54': ['CheckboxFieldDictionary', 'Additional entry specific to a checkb
ox field'], | |
166 'TABLE 8.55': ['RadioButtonFieldDictionary', 'Additional entry specific to a rad
io button field'], | |
167 'TABLE 8.57': ['TextFieldDictionary', 'Additional entry specific to a text field
'], | |
168 'TABLE 8.59': ['ChoiceFieldDictionary', 'Additional entries specific to a choice
field'], | |
169 'TABLE 8.60': ['SignatureDictionary', 'Entries in a signature dictionary'], | |
170 'TABLE 8.61': ['SubmitFormActionDictionary', 'Additional entries specific to a s
ubmit-form action'], | |
171 'TABLE 8.63': ['ResetFormActionDictionary', 'Additional entries specific to a re
set-form action'], | |
172 'TABLE 8.65': ['ImportDataActionDictionary', 'Additional entries specific to an
import-data action'], | |
173 'TABLE 8.66': ['JavascriptActionDictionary', 'Additional entries specific to a J
avaScript action'], | |
174 'TABLE 8.67': ['FDFTrailerDictionary', 'Entry in the FDF trailer dictionary'], | |
175 'TABLE 8.68': ['FDFCatalogDictionary', 'Entries in the FDF catalog dictionary'], | |
176 'TABLE 8.69': ['FDFDictionary', 'Entries in the FDF dictionary'], | |
177 'TABLE 8.70': ['EncryptedEmbeddedFileStreamDictionary', 'Additional entry in an
embedded file stream dictionary for an encrypted FDF file'], | |
178 'TABLE 8.71': ['JavascriptDictionary', 'Entries in the JavaScript dictionary'], | |
179 'TABLE 8.72': ['FDFFieldDictionary', 'Entries in an FDF field dictionary'], | |
180 'TABLE 8.73': ['IconFitDictionary', 'Entries in an icon fit dictionary'], | |
181 'TABLE 8.74': ['FDFPageDictionary', 'Entries in an FDF page dictionary'], | |
182 'TABLE 8.75': ['FDFTemplateDictionary', 'Entries in an FDF template dictionary']
, | |
183 'TABLE 8.76': ['FDFNamedPageReferenceDictionary', 'Entries in an FDF named page
reference dictionary'], | |
184 'TABLE 8.77': ['FDFFileAnnotationDictionary', 'Additional entry for annotation d
ictionaries in an FDF file'], | |
185 'TABLE 8.78': ['SoundObjectDictionary', 'Additional entries specific to a sound
object'], | |
186 'TABLE 8.79': ['MovieDictionary', 'Entries in a movie dictionary'], | |
187 'TABLE 8.80': ['MovieActivationDictionary', 'Entries in a movie activation dicti
onary'], | |
188 'TABLE 9.2': ['DocumentInformationDictionary', 'Entries in the document informat
ion dictionary'], | |
189 'TABLE 9.3': ['MetadataStreamDictionary', 'Additional entries in a metadata stre
am dictionary'], | |
190 'TABLE 9.4': ['ComponentsWithMetadataDictionary', 'Additional entry for componen
ts having metadata'], | |
191 'TABLE 9.6': ['PagePieceDictionary', 'Entries in a page-piece dictionary'], | |
192 'TABLE 9.7': ['ApplicationDataDictionary', 'Entries in an application data dicti
onary'], | |
193 'TABLE 9.9': ['StructureTreeRootDictionary', 'Entries in the structure tree root
'], | |
194 'TABLE 9.10': ['StructureElementDictionary', 'Entries in a structure element dic
tionary'], | |
195 'TABLE 9.11': ['MarkedContentReferenceDictionary', 'Entries in a marked-content
reference dictionary'], | |
196 'TABLE 9.12': ['ObjectReferenceDictionary', 'Entries in an object reference dict
ionary'], | |
197 'TABLE 9.13': ['StructureElementAccessDictionary', 'Additional dictionary entrie
s for structure element access'], | |
198 'TABLE 9.14': ['AttributeObjectDictionary', 'Entry common to all attribute objec
ts'], | |
199 'TABLE 9.15': ['MarkInformationDictionary', 'Entry in the mark information dicti
onary'], | |
200 'TABLE 9.16': ['ArtifactsDictionary', 'Property list entries for artifacts'], | |
201 'TABLE 9.27': ['StandardStructureDictionary', 'Standard layout attributes common
to all standard structure types'], | |
202 'TABLE 9.28': ['BlockLevelStructureElementsDictionary', 'Additional standard lay
out attributes specific to block-level structure elements'], | |
203 'TABLE 9.29': ['InlineLevelStructureElementsDictionary', 'Standard layout attrib
utes specific to inline-level structure elements'], | |
204 'TABLE 9.30': ['ListAttributeDictionary', 'Standard list attribute'], | |
205 'TABLE 9.31': ['TableAttributesDictionary', 'Standard table attributes'], | |
206 'TABLE 9.32': ['WebCaptureInformationDictionary', 'Entries in the Web Capture in
formation dictionary'], | |
207 'TABLE 9.33': ['WebCaptureDictionary', 'Entries common to all Web Capture conten
t sets'], | |
208 'TABLE 9.34': ['WebCapturePageSetDictionary', 'Additional entries specific to a
Web Capture page set'], | |
209 'TABLE 9.35': ['WebCaptureImageSetDictionary', 'Additional entries specific to a
Web Capture image set'], | |
210 'TABLE 9.36': ['SourceInformationDictionary', 'Entries in a source information d
ictionary'], | |
211 'TABLE 9.37': ['URLAliasDictionary', 'Entries in a URL alias dictionary'], | |
212 'TABLE 9.38': ['WebCaptureCommandDictionary', 'Entries in a Web Capture command
dictionary'], | |
213 'TABLE 9.40': ['WebCaptureCommandSettingsDictionary', 'Entries in a Web Capture
command settings dictionary'], | |
214 'TABLE 9.41': ['BoxColorInformationDictionary', 'Entries in a box color informat
ion dictionary'], | |
215 'TABLE 9.42': ['BoxStyleDictionary', 'Entries in a box style dictionary'], | |
216 'TABLE 9.43': ['PrinterMarkAnnotationDictionary', 'Additional entries specific t
o a printer\'s mark annotation'], | |
217 'TABLE 9.44': ['PrinterMarkFormDictionary', 'Additional entries specific to a pr
inter\'s mark form dictionary'], | |
218 'TABLE 9.45': ['SeparationDictionary', 'Entries in a separation dictionary'], | |
219 'TABLE 9.46': ['PDF_XOutputIntentDictionary', 'Entries in a PDF/X output intent
dictionary'], | |
220 'TABLE 9.47': ['TrapNetworkAnnotationDictionary', 'Additional entries specific t
o a trap network annotation'], | |
221 'TABLE 9.48': ['TrapNetworkAppearanceStreamDictionary', 'Additional entries spec
ific to a trap network appearance stream'], | |
222 'TABLE 9.49': ['OpiVersionDictionary', 'Entry in an OPI version dictionary'], | |
223 } | |
224 | |
225 | |
226 def acceptType(val): | |
227 global knownTypes | |
228 | |
229 ret = val | |
230 | |
231 for item in knownTypes: | |
232 ret = ret.replace(item, '') | |
233 | |
234 return ret == '' | |
235 | |
236 | |
237 def inTable(): | |
238 global tableHeaderFound | |
239 return tableHeaderFound | |
240 | |
241 def tableDescriptionFound(desc): | |
242 global table | |
243 table = desc.strip() | |
244 | |
245 def tableHasHeader(): | |
246 global table | |
247 global tableHeaderFound | |
248 | |
249 tableHeaderFound = True | |
250 #print table | |
251 | |
252 def fix(val): | |
253 ret = val | |
254 | |
255 # fix unicode chars | |
256 ret = ret.replace(unicode('fi', 'utf8'), 'fi') | |
257 ret = ret.replace(u'\u201c', '\"') | |
258 ret = ret.replace(u'\u201d', '\"') | |
259 ret = ret.replace(u'\u2019', '\'') | |
260 ret = ret.replace(u'\ufb02', 'fl') | |
261 ret = ret.replace(u'\xae', '(R)') | |
262 ret = ret.replace(u'\u2026', '...') | |
263 ret = ret.replace(u'\xd7', 'x') | |
264 ret = ret.replace(u'\u2212', '-') | |
265 ret = ret.replace(u'\u2264', '<=') | |
266 ret = ret.replace(u'\u2014', '-') | |
267 ret = ret.replace(u'\u2013', '\'') | |
268 ret = ret.replace(u'\u2022', '*') | |
269 ret = ret.replace(u'\xb5', 'mu') | |
270 ret = ret.replace(u'\xf7', '/') | |
271 | |
272 | |
273 # how enable to emit this a python string | |
274 ret = ret.replace('\'', '\\\'') | |
275 ret = ret.replace('\n', '\\n') | |
276 | |
277 | |
278 return ret | |
279 | |
280 def commitRow(): | |
281 global columnValues | |
282 global emitedDitionaryName | |
283 global table | |
284 global tableToClassName | |
285 | |
286 if columnValues == None: | |
287 return | |
288 | |
289 #print columnValues | |
290 | |
291 lastClosed = columnValues[2].find(')') | |
292 if lastClosed < 0: | |
293 print 'ERRRRRRRRRRRRRRROR' | |
294 print columnValues | |
295 return | |
296 | |
297 spec = columnValues[2][:lastClosed + 1] | |
298 spec = spec.replace('(', ';') | |
299 spec = spec.replace(')', ';') | |
300 spec = spec.strip(';') | |
301 | |
302 specs = spec.split(';') | |
303 | |
304 # clearly required, but it can be required with conditions. don't handle this
ones here, but manually | |
305 required = specs[0] == 'Required' | |
306 | |
307 inheritable = False | |
308 version = '' | |
309 for s in specs: | |
310 if s.strip() == 'inheritable' or s.strip() == 'Inheritable': | |
311 inheritable = True | |
312 elif re.match('^PDF [0-9]*[\.[0-9]*]*', s.strip()): | |
313 version = s.strip() | |
314 elif s != 'Required': | |
315 required = False | |
316 | |
317 #print spec | |
318 #print specs | |
319 #print required | |
320 #print inheritable | |
321 #print version | |
322 #print columnValues | |
323 | |
324 columnValues = [fix(columnValues[0]), fix(columnValues[1]), fix(columnValues[2
])] | |
325 | |
326 tableKey = re.search('(TABLE [0-9].[0-9][0-9]?)', table).group(1) | |
327 | |
328 if emitedDitionaryName == '': | |
329 table = fix(table) | |
330 | |
331 #print table | |
332 emitedDitionaryName = 'foo' | |
333 e = re.search('[Entries|Entry] in [a-z]* (.* dictionary)', table) | |
334 a = re.search('Additional [a-z]* in a[n]? (.* dictionary)', table) | |
335 s = re.search('Additional [a-z]* (.*)', table) | |
336 c = re.search('[Entries|Entry] common to all (.*)', table) | |
337 o1 = re.search('Optional parameter[s]? for the (.*)', table) | |
338 o2 = re.search('Optional parameter[s]? for (.*)', table) | |
339 t = re.search('.*ntries in [a-z]* (.*)', table) | |
340 | |
341 r = re.search('Property list entries for (.*)', table) | |
342 st = re.search('Standard (.*)', table) | |
343 | |
344 if e: | |
345 emitedDitionaryName = e.group(1).title().replace(' ', '') | |
346 #print emitedDitionaryName | |
347 elif a: | |
348 emitedDitionaryName = a.group(1).title().replace(' ', '') | |
349 #print emitedDitionaryName | |
350 elif s: | |
351 emitedDitionaryName = s.group(1).title().replace(' ', '') | |
352 #print emitedDitionaryName | |
353 elif c: | |
354 emitedDitionaryName = c.group(1).title().replace(' ', '') + 'Common' | |
355 #print emitedDitionaryName | |
356 elif o1: | |
357 emitedDitionaryName = o1.group(1).title().replace(' ', '') + 'OptionalPara
meters' | |
358 #print emitedDitionaryName | |
359 elif o2: | |
360 emitedDitionaryName = o2.group(1).title().replace(' ', '') + 'OptionalPara
meters' | |
361 #print emitedDitionaryName | |
362 elif t: | |
363 emitedDitionaryName = t.group(1).title().replace(' ', '') + 'Dictionary' | |
364 #print emitedDitionaryName | |
365 elif r: | |
366 emitedDitionaryName = r.group(1).title().replace(' ', '') + 'Dictionary' | |
367 #print emitedDitionaryName | |
368 elif st: | |
369 emitedDitionaryName = st.group(1).title().replace(' ', '') + 'Dictionary' | |
370 #print emitedDitionaryName | |
371 #else: | |
372 #print table | |
373 | |
374 #print tableKey | |
375 #print('\'' + tableKey + '\': [\'' + emitedDitionaryName + '\', \'' + table[
len(tableKey) + 1:] + '\'],') | |
376 | |
377 emitedDitionaryName = tableToClassName[tableKey][0] | |
378 comment = fix(tableToClassName[tableKey][1]) | |
379 | |
380 if len(tableToClassName[tableKey]) >= 3 and tableToClassName[tableKey][2] !=
'': | |
381 print(' pdfspec.addClass(\'' + emitedDitionaryName + '\', \'' + tableToCl
assName[tableKey][2] + '\', \'' + comment + '\')\\') | |
382 else: | |
383 print(' pdfspec.addClass(\'' + emitedDitionaryName + '\', \'Dictionary\',
\'' + comment + '\')\\') | |
384 | |
385 if len(tableToClassName[tableKey]) >= 4 and columnValues[0] in tableToClassNam
e[tableKey][3]: | |
386 required = True | |
387 | |
388 if required: | |
389 print(' .required(\'NULL\')\\') | |
390 else: | |
391 print(' .optional()\\') | |
392 | |
393 print(' .field(\'' + columnValues[0] + '\')\\') | |
394 print(' .name(\'' + columnValues[0] + '\')\\') | |
395 print(' .type(\'' + columnValues[1] + '\')\\') | |
396 print(' .comment(\'' + columnValues[2] + '\')\\') | |
397 | |
398 if len(tableToClassName[tableKey]) >= 4 and columnValues[0] in tableToClassNam
e[tableKey][3]: | |
399 print(' .must(' + tableToClassName[tableKey][3][columnValues[0]] +
')\\') | |
400 | |
401 print(' .done().done()\\') | |
402 | |
403 | |
404 columnValues = None | |
405 | |
406 def newRow(first, second, third): | |
407 global columnValues | |
408 columnValues = [first.rstrip(), second.rstrip(), third.rstrip()] | |
409 | |
410 def appendRow(second, third): | |
411 global columnValues | |
412 if second.rstrip() != '': | |
413 columnValues[1] = columnValues[1] + ' ' + second.rstrip() | |
414 if third.rstrip() != '': | |
415 columnValues[2] = columnValues[2] + '\n' + third.rstrip() | |
416 | |
417 def rebaseTable(line): | |
418 global knownTypes | |
419 global columnWidth | |
420 | |
421 line2 = line.replace(',', ' , ') | |
422 | |
423 words = line2.split() | |
424 | |
425 if len(words) < 3: | |
426 return False | |
427 | |
428 i = 1 | |
429 while i < len(words) - 1 and words[i] in knownTypes: | |
430 i = i + 1 | |
431 | |
432 if words[i].startswith('(Optional') or words[i].startswith('(Required'): | |
433 commitRow() | |
434 | |
435 columnWidth[0] = line.find(words[1]) | |
436 | |
437 if words[i].startswith('(Optional'): | |
438 columnWidth[1] = line.find('(Optional') - columnWidth[0] | |
439 if words[i].startswith('(Required'): | |
440 columnWidth[1] = line.find('(Required') - columnWidth[0] | |
441 return True | |
442 | |
443 return False | |
444 | |
445 | |
446 def stopTable(): | |
447 global tableHeaderFound | |
448 global emitedDitionaryName | |
449 | |
450 if not inTable(): | |
451 return | |
452 | |
453 commitRow() | |
454 tableHeaderFound = False | |
455 emitedDitionaryName = '' | |
456 print(' .done()') | |
457 print | |
458 | |
459 | |
460 def killTable(): | |
461 return | |
462 | |
463 def processLineCore(line): | |
464 global lines | |
465 global tableLine | |
466 global tableRow | |
467 global columnWidth | |
468 global columnValues | |
469 global mustFollowTableHeader | |
470 | |
471 global fnewspec | |
472 | |
473 lines = lines + 1 | |
474 | |
475 line = unicode(line, 'utf8') | |
476 | |
477 striped = line.rstrip() | |
478 | |
479 words = line.split() | |
480 if len(words) == 0: | |
481 stopTable() | |
482 return False | |
483 | |
484 isTableHeader = re.search('^[\s]*(TABLE [0-9].[0-9][0-9]?)', striped) | |
485 if isTableHeader: | |
486 stopTable() | |
487 tableDescriptionFound(striped) | |
488 mustFollowTableHeader = True | |
489 return False | |
490 | |
491 if mustFollowTableHeader: | |
492 mustFollowTableHeader = False | |
493 if len(words) != 3: | |
494 killTable() | |
495 return False | |
496 | |
497 # TODO(edisonn): support for generic table! | |
498 if words[0] != 'KEY' or words[1] != 'TYPE' or words[2] != 'VALUE': | |
499 killTable() | |
500 return False | |
501 | |
502 tableHasHeader() | |
503 columnWidth = [0, 0, 0] | |
504 columnWidth[0] = striped.index('TYPE') | |
505 columnWidth[1] = striped.index('VALUE') - striped.index('TYPE') | |
506 columnWidth[2] = 0 | |
507 return True | |
508 | |
509 if inTable(): | |
510 tableLine = tableLine + 1 | |
511 first = striped[0 : columnWidth[0]] | |
512 second = striped[columnWidth[0] : columnWidth[0] + columnWidth[1]] | |
513 third = striped[columnWidth[0] + columnWidth[1] :] | |
514 | |
515 if tableLine == 1: | |
516 if third[0] != '(': | |
517 killTable() | |
518 return False | |
519 | |
520 newRow(first, second, third) | |
521 return True | |
522 | |
523 if rebaseTable(striped): | |
524 first = striped[0 : columnWidth[0]] | |
525 second = striped[columnWidth[0] : columnWidth[0] + columnWidth[1]] | |
526 third = striped[columnWidth[0] + columnWidth[1] :] | |
527 | |
528 first = first.rstrip() | |
529 second = second.rstrip() | |
530 third = third.rstrip() | |
531 | |
532 if first == '' and second == '' and third != '': | |
533 appendRow(second, third) | |
534 return True | |
535 | |
536 if len(first.split()) > 1: | |
537 stopTable() | |
538 return False | |
539 | |
540 if first != '' and first[0] == ' ': | |
541 stopTable() | |
542 return False | |
543 | |
544 if first != '' and second != '' and third == '': | |
545 stopTable() | |
546 return False | |
547 | |
548 if first == '' and second != '' and second[0] != ' ': | |
549 if acceptType(second): | |
550 appendRow(second, third) | |
551 return True | |
552 else: | |
553 stopTable() | |
554 return False | |
555 | |
556 if first != '' and second != '' and third[0] != '(': | |
557 stopTable() | |
558 return False | |
559 | |
560 if first == '' and second != '' and second[0] == ' ': | |
561 stopTable() | |
562 return False | |
563 | |
564 if first != '' and second != '' and third[0] == '(': | |
565 commitRow() | |
566 newRow(first, second, third) | |
567 return True | |
568 | |
569 return False | |
570 return False | |
571 | |
572 def processLine(line): | |
573 global fnewspec | |
574 | |
575 inSpec = processLineCore(line) | |
576 | |
577 #just return, use the next lines if you wish to rewrite spec | |
578 #return | |
579 | |
580 if inSpec: | |
581 #resize colum with types | |
582 line = line[:columnWidth[0] + columnWidth[1]] + (' ' * (60 - columnWidth[1])
) + line[columnWidth[0] + columnWidth[1]:] | |
583 line = line[:columnWidth[0]] + (' ' * (40 - columnWidth[0])) + line[columnWi
dth[0]:] | |
584 | |
585 fnewspec.write(line) | |
586 | |
587 | |
588 def generateDef(): | |
589 global lines | |
590 global fnewspec | |
591 | |
592 fnewspec = open('PdfReference-okular-2.txt', 'w') | |
593 | |
594 print 'import datatypes' | |
595 print | |
596 | |
597 print 'def buildPdfSpec(pdfspec):' | |
598 | |
599 for line in sys.stdin: | |
600 processLine(line) | |
601 | |
602 # close last table if it was not closed already | |
603 stopTable() | |
604 | |
605 print | |
606 | |
607 #print lines | |
608 fnewspec.close() | |
609 | |
610 if '__main__' == __name__: | |
611 sys.exit(generateDef()) | |
OLD | NEW |