]>
Commit | Line | Data |
---|---|---|
97fa0ee9 YL |
1 | ## @file\r |
2 | # This file is used to collect all defined strings in multiple uni files\r | |
3 | #\r | |
71f02911 CS |
4 | #\r |
5 | # Copyright (c) 2014 Hewlett-Packard Development Company, L.P.<BR>\r | |
6 | #\r | |
018f7b82 | 7 | # Copyright (c) 2007 - 2018, Intel Corporation. All rights reserved.<BR>\r |
2e351cbe | 8 | # SPDX-License-Identifier: BSD-2-Clause-Patent\r |
30fdf114 | 9 | \r |
30fdf114 LG |
10 | ##\r |
11 | # Import Modules\r | |
12 | #\r | |
1ccc4d89 | 13 | from __future__ import print_function\r |
1be2ed90 | 14 | import Common.LongFilePathOs as os, codecs, re\r |
40d841f6 | 15 | import distutils.util\r |
30fdf114 | 16 | import Common.EdkLogger as EdkLogger\r |
86379ac4 | 17 | from io import BytesIO\r |
30fdf114 | 18 | from Common.BuildToolError import *\r |
5a57246e | 19 | from Common.StringUtils import GetLineNo\r |
30fdf114 | 20 | from Common.Misc import PathClass\r |
d943b0c3 | 21 | from Common.LongFilePathSupport import LongFilePath\r |
018f7b82 | 22 | from Common.GlobalData import *\r |
30fdf114 LG |
23 | ##\r |
24 | # Static definitions\r | |
25 | #\r | |
26 | UNICODE_WIDE_CHAR = u'\\wide'\r | |
27 | UNICODE_NARROW_CHAR = u'\\narrow'\r | |
28 | UNICODE_NON_BREAKING_CHAR = u'\\nbr'\r | |
29 | UNICODE_UNICODE_CR = '\r'\r | |
30 | UNICODE_UNICODE_LF = '\n'\r | |
31 | \r | |
32 | NARROW_CHAR = u'\uFFF0'\r | |
33 | WIDE_CHAR = u'\uFFF1'\r | |
34 | NON_BREAKING_CHAR = u'\uFFF2'\r | |
35 | CR = u'\u000D'\r | |
36 | LF = u'\u000A'\r | |
37 | NULL = u'\u0000'\r | |
38 | TAB = u'\t'\r | |
71f02911 | 39 | BACK_SLASH_PLACEHOLDER = u'\u0006'\r |
30fdf114 LG |
40 | \r |
41 | gIncludePattern = re.compile("^#include +[\"<]+([^\"< >]+)[>\"]+$", re.MULTILINE | re.UNICODE)\r | |
42 | \r | |
30fdf114 LG |
43 | ## Convert a unicode string to a Hex list\r |
44 | #\r | |
45 | # Convert a unicode string to a Hex list\r | |
46 | # UniToHexList('ABC') is ['0x41', '0x00', '0x42', '0x00', '0x43', '0x00']\r | |
47 | #\r | |
48 | # @param Uni: The python unicode string\r | |
49 | #\r | |
50 | # @retval List: The formatted hex list\r | |
51 | #\r | |
52 | def UniToHexList(Uni):\r | |
53 | List = []\r | |
54 | for Item in Uni:\r | |
55 | Temp = '%04X' % ord(Item)\r | |
56 | List.append('0x' + Temp[2:4])\r | |
57 | List.append('0x' + Temp[0:2])\r | |
58 | return List\r | |
59 | \r | |
60 | LangConvTable = {'eng':'en', 'fra':'fr', \\r | |
61 | 'aar':'aa', 'abk':'ab', 'ave':'ae', 'afr':'af', 'aka':'ak', 'amh':'am', \\r | |
62 | 'arg':'an', 'ara':'ar', 'asm':'as', 'ava':'av', 'aym':'ay', 'aze':'az', \\r | |
63 | 'bak':'ba', 'bel':'be', 'bul':'bg', 'bih':'bh', 'bis':'bi', 'bam':'bm', \\r | |
64 | 'ben':'bn', 'bod':'bo', 'bre':'br', 'bos':'bs', 'cat':'ca', 'che':'ce', \\r | |
65 | 'cha':'ch', 'cos':'co', 'cre':'cr', 'ces':'cs', 'chu':'cu', 'chv':'cv', \\r | |
66 | 'cym':'cy', 'dan':'da', 'deu':'de', 'div':'dv', 'dzo':'dz', 'ewe':'ee', \\r | |
67 | 'ell':'el', 'epo':'eo', 'spa':'es', 'est':'et', 'eus':'eu', 'fas':'fa', \\r | |
68 | 'ful':'ff', 'fin':'fi', 'fij':'fj', 'fao':'fo', 'fry':'fy', 'gle':'ga', \\r | |
69 | 'gla':'gd', 'glg':'gl', 'grn':'gn', 'guj':'gu', 'glv':'gv', 'hau':'ha', \\r | |
70 | 'heb':'he', 'hin':'hi', 'hmo':'ho', 'hrv':'hr', 'hat':'ht', 'hun':'hu', \\r | |
71 | 'hye':'hy', 'her':'hz', 'ina':'ia', 'ind':'id', 'ile':'ie', 'ibo':'ig', \\r | |
72 | 'iii':'ii', 'ipk':'ik', 'ido':'io', 'isl':'is', 'ita':'it', 'iku':'iu', \\r | |
73 | 'jpn':'ja', 'jav':'jv', 'kat':'ka', 'kon':'kg', 'kik':'ki', 'kua':'kj', \\r | |
74 | 'kaz':'kk', 'kal':'kl', 'khm':'km', 'kan':'kn', 'kor':'ko', 'kau':'kr', \\r | |
75 | 'kas':'ks', 'kur':'ku', 'kom':'kv', 'cor':'kw', 'kir':'ky', 'lat':'la', \\r | |
76 | 'ltz':'lb', 'lug':'lg', 'lim':'li', 'lin':'ln', 'lao':'lo', 'lit':'lt', \\r | |
77 | 'lub':'lu', 'lav':'lv', 'mlg':'mg', 'mah':'mh', 'mri':'mi', 'mkd':'mk', \\r | |
78 | 'mal':'ml', 'mon':'mn', 'mar':'mr', 'msa':'ms', 'mlt':'mt', 'mya':'my', \\r | |
79 | 'nau':'na', 'nob':'nb', 'nde':'nd', 'nep':'ne', 'ndo':'ng', 'nld':'nl', \\r | |
80 | 'nno':'nn', 'nor':'no', 'nbl':'nr', 'nav':'nv', 'nya':'ny', 'oci':'oc', \\r | |
81 | 'oji':'oj', 'orm':'om', 'ori':'or', 'oss':'os', 'pan':'pa', 'pli':'pi', \\r | |
82 | 'pol':'pl', 'pus':'ps', 'por':'pt', 'que':'qu', 'roh':'rm', 'run':'rn', \\r | |
83 | 'ron':'ro', 'rus':'ru', 'kin':'rw', 'san':'sa', 'srd':'sc', 'snd':'sd', \\r | |
84 | 'sme':'se', 'sag':'sg', 'sin':'si', 'slk':'sk', 'slv':'sl', 'smo':'sm', \\r | |
85 | 'sna':'sn', 'som':'so', 'sqi':'sq', 'srp':'sr', 'ssw':'ss', 'sot':'st', \\r | |
86 | 'sun':'su', 'swe':'sv', 'swa':'sw', 'tam':'ta', 'tel':'te', 'tgk':'tg', \\r | |
87 | 'tha':'th', 'tir':'ti', 'tuk':'tk', 'tgl':'tl', 'tsn':'tn', 'ton':'to', \\r | |
88 | 'tur':'tr', 'tso':'ts', 'tat':'tt', 'twi':'tw', 'tah':'ty', 'uig':'ug', \\r | |
89 | 'ukr':'uk', 'urd':'ur', 'uzb':'uz', 'ven':'ve', 'vie':'vi', 'vol':'vo', \\r | |
90 | 'wln':'wa', 'wol':'wo', 'xho':'xh', 'yid':'yi', 'yor':'yo', 'zha':'za', \\r | |
91 | 'zho':'zh', 'zul':'zu'}\r | |
92 | \r | |
93 | ## GetLanguageCode\r | |
94 | #\r | |
95 | # Check the language code read from .UNI file and convert ISO 639-2 codes to RFC 4646 codes if appropriate\r | |
fb0b35e0 | 96 | # ISO 639-2 language codes supported in compatibility mode\r |
30fdf114 LG |
97 | # RFC 4646 language codes supported in native mode\r |
98 | #\r | |
99 | # @param LangName: Language codes read from .UNI file\r | |
100 | #\r | |
fb0b35e0 | 101 | # @retval LangName: Valid language code in RFC 4646 format or None\r |
30fdf114 LG |
102 | #\r |
103 | def GetLanguageCode(LangName, IsCompatibleMode, File):\r | |
30fdf114 LG |
104 | length = len(LangName)\r |
105 | if IsCompatibleMode:\r | |
106 | if length == 3 and LangName.isalpha():\r | |
107 | TempLangName = LangConvTable.get(LangName.lower())\r | |
4231a819 | 108 | if TempLangName is not None:\r |
da92f276 | 109 | return TempLangName\r |
30fdf114 LG |
110 | return LangName\r |
111 | else:\r | |
112 | EdkLogger.error("Unicode File Parser", FORMAT_INVALID, "Invalid ISO 639-2 language code : %s" % LangName, File)\r | |
113 | \r | |
40d841f6 LG |
114 | if (LangName[0] == 'X' or LangName[0] == 'x') and LangName[1] == '-':\r |
115 | return LangName\r | |
30fdf114 LG |
116 | if length == 2:\r |
117 | if LangName.isalpha():\r | |
118 | return LangName\r | |
119 | elif length == 3:\r | |
4231a819 | 120 | if LangName.isalpha() and LangConvTable.get(LangName.lower()) is None:\r |
30fdf114 LG |
121 | return LangName\r |
122 | elif length == 5:\r | |
123 | if LangName[0:2].isalpha() and LangName[2] == '-':\r | |
124 | return LangName\r | |
125 | elif length >= 6:\r | |
126 | if LangName[0:2].isalpha() and LangName[2] == '-':\r | |
127 | return LangName\r | |
4231a819 | 128 | if LangName[0:3].isalpha() and LangConvTable.get(LangName.lower()) is None and LangName[3] == '-':\r |
30fdf114 LG |
129 | return LangName\r |
130 | \r | |
131 | EdkLogger.error("Unicode File Parser", FORMAT_INVALID, "Invalid RFC 4646 language code : %s" % LangName, File)\r | |
132 | \r | |
d80e451b JJ |
133 | ## Ucs2Codec\r |
134 | #\r | |
135 | # This is only a partial codec implementation. It only supports\r | |
136 | # encoding, and is primarily used to check that all the characters are\r | |
137 | # valid for UCS-2.\r | |
138 | #\r | |
139 | class Ucs2Codec(codecs.Codec):\r | |
140 | def __init__(self):\r | |
141 | self.__utf16 = codecs.lookup('utf-16')\r | |
142 | \r | |
143 | def encode(self, input, errors='strict'):\r | |
144 | for Char in input:\r | |
145 | CodePoint = ord(Char)\r | |
146 | if CodePoint >= 0xd800 and CodePoint <= 0xdfff:\r | |
147 | raise ValueError("Code Point is in range reserved for " +\r | |
148 | "UTF-16 surrogate pairs")\r | |
149 | elif CodePoint > 0xffff:\r | |
150 | raise ValueError("Code Point too large to encode in UCS-2")\r | |
151 | return self.__utf16.encode(input)\r | |
152 | \r | |
153 | TheUcs2Codec = Ucs2Codec()\r | |
154 | def Ucs2Search(name):\r | |
5d864834 | 155 | if name in ['ucs-2', 'ucs_2']:\r |
d80e451b JJ |
156 | return codecs.CodecInfo(\r |
157 | name=name,\r | |
158 | encode=TheUcs2Codec.encode,\r | |
159 | decode=TheUcs2Codec.decode)\r | |
160 | else:\r | |
161 | return None\r | |
162 | codecs.register(Ucs2Search)\r | |
163 | \r | |
30fdf114 LG |
164 | ## StringDefClassObject\r |
165 | #\r | |
166 | # A structure for language definition\r | |
167 | #\r | |
168 | class StringDefClassObject(object):\r | |
169 | def __init__(self, Name = None, Value = None, Referenced = False, Token = None, UseOtherLangDef = ''):\r | |
170 | self.StringName = ''\r | |
171 | self.StringNameByteList = []\r | |
172 | self.StringValue = ''\r | |
173 | self.StringValueByteList = ''\r | |
174 | self.Token = 0\r | |
175 | self.Referenced = Referenced\r | |
176 | self.UseOtherLangDef = UseOtherLangDef\r | |
177 | self.Length = 0\r | |
178 | \r | |
4231a819 | 179 | if Name is not None:\r |
30fdf114 LG |
180 | self.StringName = Name\r |
181 | self.StringNameByteList = UniToHexList(Name)\r | |
4231a819 | 182 | if Value is not None:\r |
30fdf114 LG |
183 | self.StringValue = Value + u'\x00' # Add a NULL at string tail\r |
184 | self.StringValueByteList = UniToHexList(self.StringValue)\r | |
185 | self.Length = len(self.StringValueByteList)\r | |
4231a819 | 186 | if Token is not None:\r |
30fdf114 LG |
187 | self.Token = Token\r |
188 | \r | |
189 | def __str__(self):\r | |
190 | return repr(self.StringName) + ' ' + \\r | |
191 | repr(self.Token) + ' ' + \\r | |
192 | repr(self.Referenced) + ' ' + \\r | |
193 | repr(self.StringValue) + ' ' + \\r | |
194 | repr(self.UseOtherLangDef)\r | |
195 | \r | |
756ad8f8 | 196 | def UpdateValue(self, Value = None):\r |
4231a819 | 197 | if Value is not None:\r |
756ad8f8 LG |
198 | self.StringValue = Value + u'\x00' # Add a NULL at string tail\r |
199 | self.StringValueByteList = UniToHexList(self.StringValue)\r | |
200 | self.Length = len(self.StringValueByteList)\r | |
201 | \r | |
c23ef28c CJ |
202 | def StripComments(Line):\r |
203 | Comment = u'//'\r | |
204 | CommentPos = Line.find(Comment)\r | |
205 | while CommentPos >= 0:\r | |
206 | # if there are non matched quotes before the comment header\r | |
207 | # then we are in the middle of a string\r | |
208 | # but we need to ignore the escaped quotes and backslashes.\r | |
209 | if ((Line.count(u'"', 0, CommentPos) - Line.count(u'\\"', 0, CommentPos)) & 1) == 1:\r | |
210 | CommentPos = Line.find (Comment, CommentPos + 1)\r | |
211 | else:\r | |
212 | return Line[:CommentPos].strip()\r | |
213 | return Line.strip()\r | |
214 | \r | |
30fdf114 LG |
215 | ## UniFileClassObject\r |
216 | #\r | |
217 | # A structure for .uni file definition\r | |
218 | #\r | |
219 | class UniFileClassObject(object):\r | |
756ad8f8 | 220 | def __init__(self, FileList = [], IsCompatibleMode = False, IncludePathList = []):\r |
30fdf114 LG |
221 | self.FileList = FileList\r |
222 | self.Token = 2\r | |
223 | self.LanguageDef = [] #[ [u'LanguageIdentifier', u'PrintableName'], ... ]\r | |
224 | self.OrderedStringList = {} #{ u'LanguageIdentifier' : [StringDefClassObject] }\r | |
79b74a03 LG |
225 | self.OrderedStringDict = {} #{ u'LanguageIdentifier' : {StringName:(IndexInList)} }\r |
226 | self.OrderedStringListByToken = {} #{ u'LanguageIdentifier' : {Token: StringDefClassObject} }\r | |
30fdf114 | 227 | self.IsCompatibleMode = IsCompatibleMode\r |
756ad8f8 | 228 | self.IncludePathList = IncludePathList\r |
30fdf114 LG |
229 | if len(self.FileList) > 0:\r |
230 | self.LoadUniFiles(FileList)\r | |
231 | \r | |
232 | #\r | |
233 | # Get Language definition\r | |
234 | #\r | |
235 | def GetLangDef(self, File, Line):\r | |
40d841f6 | 236 | Lang = distutils.util.split_quoted((Line.split(u"//")[0]))\r |
30fdf114 LG |
237 | if len(Lang) != 3:\r |
238 | try:\r | |
3b743b3b | 239 | FileIn = UniFileClassObject.OpenUniFile(LongFilePath(File.Path))\r |
5b0671c1 | 240 | except UnicodeError as X:\r |
30fdf114 LG |
241 | EdkLogger.error("build", FILE_READ_FAILURE, "File read failure: %s" % str(X), ExtraData=File);\r |
242 | except:\r | |
243 | EdkLogger.error("build", FILE_OPEN_FAILURE, ExtraData=File);\r | |
244 | LineNo = GetLineNo(FileIn, Line, False)\r | |
245 | EdkLogger.error("Unicode File Parser", PARSER_ERROR, "Wrong language definition",\r | |
47fea6af | 246 | ExtraData="""%s\n\t*Correct format is like '#langdef en-US "English"'""" % Line, File=File, Line=LineNo)\r |
30fdf114 LG |
247 | else:\r |
248 | LangName = GetLanguageCode(Lang[1], self.IsCompatibleMode, self.File)\r | |
40d841f6 | 249 | LangPrintName = Lang[2]\r |
30fdf114 LG |
250 | \r |
251 | IsLangInDef = False\r | |
252 | for Item in self.LanguageDef:\r | |
253 | if Item[0] == LangName:\r | |
254 | IsLangInDef = True\r | |
255 | break;\r | |
256 | \r | |
257 | if not IsLangInDef:\r | |
258 | self.LanguageDef.append([LangName, LangPrintName])\r | |
259 | \r | |
260 | #\r | |
261 | # Add language string\r | |
262 | #\r | |
263 | self.AddStringToList(u'$LANGUAGE_NAME', LangName, LangName, 0, True, Index=0)\r | |
264 | self.AddStringToList(u'$PRINTABLE_LANGUAGE_NAME', LangName, LangPrintName, 1, True, Index=1)\r | |
265 | \r | |
756ad8f8 LG |
266 | if not IsLangInDef:\r |
267 | #\r | |
268 | # The found STRING tokens will be added into new language string list\r | |
f7496d71 | 269 | # so that the unique STRING identifier is reserved for all languages in the package list.\r |
756ad8f8 LG |
270 | #\r |
271 | FirstLangName = self.LanguageDef[0][0]\r | |
272 | if LangName != FirstLangName:\r | |
273 | for Index in range (2, len (self.OrderedStringList[FirstLangName])):\r | |
274 | Item = self.OrderedStringList[FirstLangName][Index]\r | |
275 | if Item.UseOtherLangDef != '':\r | |
276 | OtherLang = Item.UseOtherLangDef\r | |
277 | else:\r | |
278 | OtherLang = FirstLangName\r | |
279 | self.OrderedStringList[LangName].append (StringDefClassObject(Item.StringName, '', Item.Referenced, Item.Token, OtherLang))\r | |
79b74a03 | 280 | self.OrderedStringDict[LangName][Item.StringName] = len(self.OrderedStringList[LangName]) - 1\r |
30fdf114 LG |
281 | return True\r |
282 | \r | |
3b743b3b CJ |
283 | @staticmethod\r |
284 | def OpenUniFile(FileName):\r | |
d80e451b JJ |
285 | #\r |
286 | # Read file\r | |
287 | #\r | |
288 | try:\r | |
289 | UniFile = open(FileName, mode='rb')\r | |
290 | FileIn = UniFile.read()\r | |
291 | UniFile.close()\r | |
292 | except:\r | |
293 | EdkLogger.Error("build", FILE_OPEN_FAILURE, ExtraData=File)\r | |
294 | \r | |
295 | #\r | |
be264422 | 296 | # Detect Byte Order Mark at beginning of file. Default to UTF-8\r |
d80e451b | 297 | #\r |
be264422 JJ |
298 | Encoding = 'utf-8'\r |
299 | if (FileIn.startswith(codecs.BOM_UTF16_BE) or\r | |
300 | FileIn.startswith(codecs.BOM_UTF16_LE)):\r | |
301 | Encoding = 'utf-16'\r | |
d80e451b | 302 | \r |
3b743b3b | 303 | UniFileClassObject.VerifyUcs2Data(FileIn, FileName, Encoding)\r |
d80e451b | 304 | \r |
86379ac4 | 305 | UniFile = BytesIO(FileIn)\r |
d80e451b JJ |
306 | Info = codecs.lookup(Encoding)\r |
307 | (Reader, Writer) = (Info.streamreader, Info.streamwriter)\r | |
308 | return codecs.StreamReaderWriter(UniFile, Reader, Writer)\r | |
309 | \r | |
3b743b3b CJ |
310 | @staticmethod\r |
311 | def VerifyUcs2Data(FileIn, FileName, Encoding):\r | |
d80e451b JJ |
312 | Ucs2Info = codecs.lookup('ucs-2')\r |
313 | #\r | |
314 | # Convert to unicode\r | |
315 | #\r | |
316 | try:\r | |
317 | FileDecoded = codecs.decode(FileIn, Encoding)\r | |
318 | Ucs2Info.encode(FileDecoded)\r | |
319 | except:\r | |
86379ac4 | 320 | UniFile = BytesIO(FileIn)\r |
d80e451b JJ |
321 | Info = codecs.lookup(Encoding)\r |
322 | (Reader, Writer) = (Info.streamreader, Info.streamwriter)\r | |
323 | File = codecs.StreamReaderWriter(UniFile, Reader, Writer)\r | |
324 | LineNumber = 0\r | |
325 | ErrMsg = lambda Encoding, LineNumber: \\r | |
326 | '%s contains invalid %s characters on line %d.' % \\r | |
327 | (FileName, Encoding, LineNumber)\r | |
328 | while True:\r | |
329 | LineNumber = LineNumber + 1\r | |
330 | try:\r | |
331 | Line = File.readline()\r | |
332 | if Line == '':\r | |
333 | EdkLogger.error('Unicode File Parser', PARSER_ERROR,\r | |
334 | ErrMsg(Encoding, LineNumber))\r | |
335 | Ucs2Info.encode(Line)\r | |
336 | except:\r | |
337 | EdkLogger.error('Unicode File Parser', PARSER_ERROR,\r | |
338 | ErrMsg('UCS-2', LineNumber))\r | |
339 | \r | |
30fdf114 LG |
340 | #\r |
341 | # Get String name and value\r | |
342 | #\r | |
343 | def GetStringObject(self, Item):\r | |
30fdf114 LG |
344 | Language = ''\r |
345 | Value = ''\r | |
346 | \r | |
347 | Name = Item.split()[1]\r | |
c3915fa5 | 348 | # Check the string name\r |
2bcc713e | 349 | if Name != '':\r |
1f26f5fd | 350 | MatchString = gIdentifierPattern.match(Name)\r |
bfc8f566 | 351 | if MatchString is None:\r |
c3915fa5 | 352 | EdkLogger.error('Unicode File Parser', FORMAT_INVALID, 'The string token name %s defined in UNI file %s contains the invalid character.' % (Name, self.File))\r |
30fdf114 LG |
353 | LanguageList = Item.split(u'#language ')\r |
354 | for IndexI in range(len(LanguageList)):\r | |
355 | if IndexI == 0:\r | |
356 | continue\r | |
357 | else:\r | |
358 | Language = LanguageList[IndexI].split()[0]\r | |
359 | Value = LanguageList[IndexI][LanguageList[IndexI].find(u'\"') + len(u'\"') : LanguageList[IndexI].rfind(u'\"')] #.replace(u'\r\n', u'')\r | |
360 | Language = GetLanguageCode(Language, self.IsCompatibleMode, self.File)\r | |
361 | self.AddStringToList(Name, Language, Value)\r | |
362 | \r | |
363 | #\r | |
364 | # Get include file list and load them\r | |
365 | #\r | |
366 | def GetIncludeFile(self, Item, Dir):\r | |
367 | FileName = Item[Item.find(u'#include ') + len(u'#include ') :Item.find(u' ', len(u'#include '))][1:-1]\r | |
368 | self.LoadUniFile(FileName)\r | |
369 | \r | |
370 | #\r | |
371 | # Pre-process before parse .uni file\r | |
372 | #\r | |
373 | def PreProcess(self, File):\r | |
30fdf114 | 374 | try:\r |
3b743b3b | 375 | FileIn = UniFileClassObject.OpenUniFile(LongFilePath(File.Path))\r |
5b0671c1 | 376 | except UnicodeError as X:\r |
30fdf114 | 377 | EdkLogger.error("build", FILE_READ_FAILURE, "File read failure: %s" % str(X), ExtraData=File.Path);\r |
0075ab2c FB |
378 | except OSError:\r |
379 | EdkLogger.error("Unicode File Parser", FILE_NOT_FOUND, ExtraData=File.Path)\r | |
30fdf114 LG |
380 | except:\r |
381 | EdkLogger.error("build", FILE_OPEN_FAILURE, ExtraData=File.Path);\r | |
382 | \r | |
383 | Lines = []\r | |
384 | #\r | |
385 | # Use unique identifier\r | |
386 | #\r | |
387 | for Line in FileIn:\r | |
388 | Line = Line.strip()\r | |
71f02911 | 389 | Line = Line.replace(u'\\\\', BACK_SLASH_PLACEHOLDER)\r |
c23ef28c | 390 | Line = StripComments(Line)\r |
71f02911 | 391 | \r |
30fdf114 | 392 | #\r |
71f02911 | 393 | # Ignore empty line\r |
30fdf114 | 394 | #\r |
f7496d71 LG |
395 | if len(Line) == 0:\r |
396 | continue\r | |
397 | \r | |
398 | \r | |
30fdf114 LG |
399 | Line = Line.replace(u'/langdef', u'#langdef')\r |
400 | Line = Line.replace(u'/string', u'#string')\r | |
401 | Line = Line.replace(u'/language', u'#language')\r | |
402 | Line = Line.replace(u'/include', u'#include')\r | |
403 | \r | |
404 | Line = Line.replace(UNICODE_WIDE_CHAR, WIDE_CHAR)\r | |
405 | Line = Line.replace(UNICODE_NARROW_CHAR, NARROW_CHAR)\r | |
406 | Line = Line.replace(UNICODE_NON_BREAKING_CHAR, NON_BREAKING_CHAR)\r | |
407 | \r | |
30fdf114 LG |
408 | Line = Line.replace(u'\\r\\n', CR + LF)\r |
409 | Line = Line.replace(u'\\n', CR + LF)\r | |
410 | Line = Line.replace(u'\\r', CR)\r | |
71f02911 | 411 | Line = Line.replace(u'\\t', u' ')\r |
8546dfea | 412 | Line = Line.replace(u'\t', u' ')\r |
f7496d71 LG |
413 | Line = Line.replace(u'\\"', u'"')\r |
414 | Line = Line.replace(u"\\'", u"'")\r | |
71f02911 | 415 | Line = Line.replace(BACK_SLASH_PLACEHOLDER, u'\\')\r |
4afd3d04 | 416 | \r |
314e2fb1 YZ |
417 | StartPos = Line.find(u'\\x')\r |
418 | while (StartPos != -1):\r | |
419 | EndPos = Line.find(u'\\', StartPos + 1, StartPos + 7)\r | |
420 | if EndPos != -1 and EndPos - StartPos == 6 :\r | |
018f7b82 | 421 | if g4HexChar.match(Line[StartPos + 2 : EndPos], re.UNICODE):\r |
314e2fb1 | 422 | EndStr = Line[EndPos: ]\r |
d943b0c3 | 423 | UniStr = Line[StartPos + 2: EndPos]\r |
314e2fb1 | 424 | if EndStr.startswith(u'\\x') and len(EndStr) >= 7:\r |
018f7b82 | 425 | if EndStr[6] == u'\\' and g4HexChar.match(EndStr[2 : 6], re.UNICODE):\r |
314e2fb1 YZ |
426 | Line = Line[0 : StartPos] + UniStr + EndStr\r |
427 | else:\r | |
428 | Line = Line[0 : StartPos] + UniStr + EndStr[1:]\r | |
ad319b93 | 429 | StartPos = Line.find(u'\\x', StartPos + 1)\r |
30fdf114 LG |
430 | \r |
431 | IncList = gIncludePattern.findall(Line)\r | |
432 | if len(IncList) == 1:\r | |
756ad8f8 LG |
433 | for Dir in [File.Dir] + self.IncludePathList:\r |
434 | IncFile = PathClass(str(IncList[0]), Dir)\r | |
435 | if os.path.isfile(IncFile.Path):\r | |
436 | Lines.extend(self.PreProcess(IncFile))\r | |
437 | break\r | |
438 | else:\r | |
439 | EdkLogger.error("Unicode File Parser", FILE_NOT_FOUND, Message="Cannot find include file", ExtraData=str(IncList[0]))\r | |
30fdf114 LG |
440 | continue\r |
441 | \r | |
442 | Lines.append(Line)\r | |
443 | \r | |
444 | return Lines\r | |
445 | \r | |
446 | #\r | |
447 | # Load a .uni file\r | |
448 | #\r | |
449 | def LoadUniFile(self, File = None):\r | |
4231a819 | 450 | if File is None:\r |
30fdf114 LG |
451 | EdkLogger.error("Unicode File Parser", PARSER_ERROR, 'No unicode file is given')\r |
452 | self.File = File\r | |
453 | #\r | |
454 | # Process special char in file\r | |
455 | #\r | |
456 | Lines = self.PreProcess(File)\r | |
457 | \r | |
458 | #\r | |
459 | # Get Unicode Information\r | |
460 | #\r | |
461 | for IndexI in range(len(Lines)):\r | |
462 | Line = Lines[IndexI]\r | |
463 | if (IndexI + 1) < len(Lines):\r | |
464 | SecondLine = Lines[IndexI + 1]\r | |
465 | if (IndexI + 2) < len(Lines):\r | |
466 | ThirdLine = Lines[IndexI + 2]\r | |
467 | \r | |
468 | #\r | |
469 | # Get Language def information\r | |
470 | #\r | |
471 | if Line.find(u'#langdef ') >= 0:\r | |
472 | self.GetLangDef(File, Line)\r | |
473 | continue\r | |
474 | \r | |
475 | Name = ''\r | |
476 | Language = ''\r | |
477 | Value = ''\r | |
478 | #\r | |
479 | # Get string def information format 1 as below\r | |
480 | #\r | |
481 | # #string MY_STRING_1\r | |
482 | # #language eng\r | |
483 | # My first English string line 1\r | |
484 | # My first English string line 2\r | |
485 | # #string MY_STRING_1\r | |
486 | # #language spa\r | |
487 | # Mi segunda secuencia 1\r | |
488 | # Mi segunda secuencia 2\r | |
489 | #\r | |
490 | if Line.find(u'#string ') >= 0 and Line.find(u'#language ') < 0 and \\r | |
491 | SecondLine.find(u'#string ') < 0 and SecondLine.find(u'#language ') >= 0 and \\r | |
492 | ThirdLine.find(u'#string ') < 0 and ThirdLine.find(u'#language ') < 0:\r | |
493 | Name = Line[Line.find(u'#string ') + len(u'#string ') : ].strip(' ')\r | |
494 | Language = SecondLine[SecondLine.find(u'#language ') + len(u'#language ') : ].strip(' ')\r | |
495 | for IndexJ in range(IndexI + 2, len(Lines)):\r | |
496 | if Lines[IndexJ].find(u'#string ') < 0 and Lines[IndexJ].find(u'#language ') < 0:\r | |
497 | Value = Value + Lines[IndexJ]\r | |
498 | else:\r | |
499 | IndexI = IndexJ\r | |
500 | break\r | |
501 | # Value = Value.replace(u'\r\n', u'')\r | |
502 | Language = GetLanguageCode(Language, self.IsCompatibleMode, self.File)\r | |
c3915fa5 | 503 | # Check the string name\r |
b303ea72 | 504 | if not self.IsCompatibleMode and Name != '':\r |
1f26f5fd | 505 | MatchString = gIdentifierPattern.match(Name)\r |
bfc8f566 | 506 | if MatchString is None:\r |
c3915fa5 | 507 | EdkLogger.error('Unicode File Parser', FORMAT_INVALID, 'The string token name %s defined in UNI file %s contains the invalid character.' % (Name, self.File))\r |
30fdf114 LG |
508 | self.AddStringToList(Name, Language, Value)\r |
509 | continue\r | |
510 | \r | |
511 | #\r | |
512 | # Get string def information format 2 as below\r | |
513 | #\r | |
514 | # #string MY_STRING_1 #language eng "My first English string line 1"\r | |
515 | # "My first English string line 2"\r | |
516 | # #language spa "Mi segunda secuencia 1"\r | |
517 | # "Mi segunda secuencia 2"\r | |
518 | # #string MY_STRING_2 #language eng "My first English string line 1"\r | |
519 | # "My first English string line 2"\r | |
520 | # #string MY_STRING_2 #language spa "Mi segunda secuencia 1"\r | |
521 | # "Mi segunda secuencia 2"\r | |
522 | #\r | |
523 | if Line.find(u'#string ') >= 0 and Line.find(u'#language ') >= 0:\r | |
524 | StringItem = Line\r | |
525 | for IndexJ in range(IndexI + 1, len(Lines)):\r | |
526 | if Lines[IndexJ].find(u'#string ') >= 0 and Lines[IndexJ].find(u'#language ') >= 0:\r | |
527 | IndexI = IndexJ\r | |
528 | break\r | |
529 | elif Lines[IndexJ].find(u'#string ') < 0 and Lines[IndexJ].find(u'#language ') >= 0:\r | |
530 | StringItem = StringItem + Lines[IndexJ]\r | |
531 | elif Lines[IndexJ].count(u'\"') >= 2:\r | |
532 | StringItem = StringItem[ : StringItem.rfind(u'\"')] + Lines[IndexJ][Lines[IndexJ].find(u'\"') + len(u'\"') : ]\r | |
533 | self.GetStringObject(StringItem)\r | |
534 | continue\r | |
535 | \r | |
536 | #\r | |
537 | # Load multiple .uni files\r | |
538 | #\r | |
52302d4d | 539 | def LoadUniFiles(self, FileList):\r |
30fdf114 | 540 | if len(FileList) > 0:\r |
52302d4d LG |
541 | for File in FileList:\r |
542 | self.LoadUniFile(File)\r | |
30fdf114 LG |
543 | \r |
544 | #\r | |
545 | # Add a string to list\r | |
546 | #\r | |
547 | def AddStringToList(self, Name, Language, Value, Token = None, Referenced = False, UseOtherLangDef = '', Index = -1):\r | |
08dd311f LG |
548 | for LangNameItem in self.LanguageDef:\r |
549 | if Language == LangNameItem[0]:\r | |
550 | break\r | |
551 | else:\r | |
552 | EdkLogger.error('Unicode File Parser', FORMAT_NOT_SUPPORTED, "The language '%s' for %s is not defined in Unicode file %s." \\r | |
553 | % (Language, Name, self.File))\r | |
f7496d71 | 554 | \r |
30fdf114 LG |
555 | if Language not in self.OrderedStringList:\r |
556 | self.OrderedStringList[Language] = []\r | |
79b74a03 | 557 | self.OrderedStringDict[Language] = {}\r |
30fdf114 | 558 | \r |
756ad8f8 | 559 | IsAdded = True\r |
79b74a03 LG |
560 | if Name in self.OrderedStringDict[Language]:\r |
561 | IsAdded = False\r | |
4231a819 | 562 | if Value is not None:\r |
79b74a03 LG |
563 | ItemIndexInList = self.OrderedStringDict[Language][Name]\r |
564 | Item = self.OrderedStringList[Language][ItemIndexInList]\r | |
565 | Item.UpdateValue(Value)\r | |
47fea6af | 566 | Item.UseOtherLangDef = ''\r |
756ad8f8 LG |
567 | \r |
568 | if IsAdded:\r | |
30fdf114 LG |
569 | Token = len(self.OrderedStringList[Language])\r |
570 | if Index == -1:\r | |
571 | self.OrderedStringList[Language].append(StringDefClassObject(Name, Value, Referenced, Token, UseOtherLangDef))\r | |
79b74a03 | 572 | self.OrderedStringDict[Language][Name] = Token\r |
756ad8f8 LG |
573 | for LangName in self.LanguageDef:\r |
574 | #\r | |
575 | # New STRING token will be added into all language string lists.\r | |
f7496d71 | 576 | # so that the unique STRING identifier is reserved for all languages in the package list.\r |
756ad8f8 LG |
577 | #\r |
578 | if LangName[0] != Language:\r | |
579 | if UseOtherLangDef != '':\r | |
580 | OtherLangDef = UseOtherLangDef\r | |
581 | else:\r | |
582 | OtherLangDef = Language\r | |
583 | self.OrderedStringList[LangName[0]].append(StringDefClassObject(Name, '', Referenced, Token, OtherLangDef))\r | |
79b74a03 | 584 | self.OrderedStringDict[LangName[0]][Name] = len(self.OrderedStringList[LangName[0]]) - 1\r |
30fdf114 LG |
585 | else:\r |
586 | self.OrderedStringList[Language].insert(Index, StringDefClassObject(Name, Value, Referenced, Token, UseOtherLangDef))\r | |
79b74a03 | 587 | self.OrderedStringDict[Language][Name] = Index\r |
30fdf114 LG |
588 | \r |
589 | #\r | |
590 | # Set the string as referenced\r | |
591 | #\r | |
592 | def SetStringReferenced(self, Name):\r | |
756ad8f8 LG |
593 | #\r |
594 | # String stoken are added in the same order in all language string lists.\r | |
595 | # So, only update the status of string stoken in first language string list.\r | |
596 | #\r | |
597 | Lang = self.LanguageDef[0][0]\r | |
79b74a03 LG |
598 | if Name in self.OrderedStringDict[Lang]:\r |
599 | ItemIndexInList = self.OrderedStringDict[Lang][Name]\r | |
600 | Item = self.OrderedStringList[Lang][ItemIndexInList]\r | |
601 | Item.Referenced = True\r | |
602 | \r | |
30fdf114 LG |
603 | #\r |
604 | # Search the string in language definition by Name\r | |
605 | #\r | |
606 | def FindStringValue(self, Name, Lang):\r | |
79b74a03 LG |
607 | if Name in self.OrderedStringDict[Lang]:\r |
608 | ItemIndexInList = self.OrderedStringDict[Lang][Name]\r | |
609 | return self.OrderedStringList[Lang][ItemIndexInList]\r | |
30fdf114 LG |
610 | \r |
611 | return None\r | |
612 | \r | |
613 | #\r | |
614 | # Search the string in language definition by Token\r | |
615 | #\r | |
616 | def FindByToken(self, Token, Lang):\r | |
617 | for Item in self.OrderedStringList[Lang]:\r | |
618 | if Item.Token == Token:\r | |
619 | return Item\r | |
620 | \r | |
621 | return None\r | |
622 | \r | |
623 | #\r | |
624 | # Re-order strings and re-generate tokens\r | |
625 | #\r | |
626 | def ReToken(self):\r | |
627 | #\r | |
756ad8f8 | 628 | # Retoken all language strings according to the status of string stoken in the first language string.\r |
30fdf114 | 629 | #\r |
756ad8f8 | 630 | FirstLangName = self.LanguageDef[0][0]\r |
52302d4d | 631 | \r |
79b74a03 LG |
632 | # Convert the OrderedStringList to be OrderedStringListByToken in order to faciliate future search by token\r |
633 | for LangNameItem in self.LanguageDef:\r | |
634 | self.OrderedStringListByToken[LangNameItem[0]] = {}\r | |
635 | \r | |
52302d4d | 636 | #\r |
756ad8f8 | 637 | # Use small token for all referred string stoken.\r |
52302d4d | 638 | #\r |
756ad8f8 LG |
639 | RefToken = 0\r |
640 | for Index in range (0, len (self.OrderedStringList[FirstLangName])):\r | |
641 | FirstLangItem = self.OrderedStringList[FirstLangName][Index]\r | |
642 | if FirstLangItem.Referenced == True:\r | |
643 | for LangNameItem in self.LanguageDef:\r | |
644 | LangName = LangNameItem[0]\r | |
645 | OtherLangItem = self.OrderedStringList[LangName][Index]\r | |
646 | OtherLangItem.Referenced = True\r | |
647 | OtherLangItem.Token = RefToken\r | |
79b74a03 | 648 | self.OrderedStringListByToken[LangName][OtherLangItem.Token] = OtherLangItem\r |
756ad8f8 | 649 | RefToken = RefToken + 1\r |
30fdf114 LG |
650 | \r |
651 | #\r | |
756ad8f8 | 652 | # Use big token for all unreferred string stoken.\r |
30fdf114 | 653 | #\r |
756ad8f8 LG |
654 | UnRefToken = 0\r |
655 | for Index in range (0, len (self.OrderedStringList[FirstLangName])):\r | |
656 | FirstLangItem = self.OrderedStringList[FirstLangName][Index]\r | |
657 | if FirstLangItem.Referenced == False:\r | |
658 | for LangNameItem in self.LanguageDef:\r | |
659 | LangName = LangNameItem[0]\r | |
660 | OtherLangItem = self.OrderedStringList[LangName][Index]\r | |
661 | OtherLangItem.Token = RefToken + UnRefToken\r | |
79b74a03 | 662 | self.OrderedStringListByToken[LangName][OtherLangItem.Token] = OtherLangItem\r |
756ad8f8 | 663 | UnRefToken = UnRefToken + 1\r |
30fdf114 LG |
664 | \r |
665 | #\r | |
666 | # Show the instance itself\r | |
667 | #\r | |
668 | def ShowMe(self):\r | |
72443dd2 | 669 | print(self.LanguageDef)\r |
30fdf114 LG |
670 | #print self.OrderedStringList\r |
671 | for Item in self.OrderedStringList:\r | |
72443dd2 | 672 | print(Item)\r |
30fdf114 | 673 | for Member in self.OrderedStringList[Item]:\r |
72443dd2 | 674 | print(str(Member))\r |
30fdf114 LG |
675 | \r |
676 | # This acts like the main() function for the script, unless it is 'import'ed into another\r | |
677 | # script.\r | |
678 | if __name__ == '__main__':\r | |
679 | EdkLogger.Initialize()\r | |
680 | EdkLogger.SetLevel(EdkLogger.DEBUG_0)\r | |
da92f276 | 681 | a = UniFileClassObject([PathClass("C:\\Edk\\Strings.uni"), PathClass("C:\\Edk\\Strings2.uni")])\r |
30fdf114 LG |
682 | a.ReToken()\r |
683 | a.ShowMe()\r |