1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

71

72

73

74

75

76

77

78

79

80

81

82

83

84

85

86

87

88

89

90

91

92

93

94

95

96

97

98

99

100

101

102

103

104

105

106

107

108

109

110

111

112

113

114

115

116

117

118

119

120

121

122

123

124

125

126

127

128

129

130

131

132

133

134

135

136

137

138

139

140

141

142

143

144

145

146

147

148

149

150

151

152

153

154

155

156

157

158

159

160

161

162

163

164

165

166

167

168

169

170

171

172

173

174

175

176

177

178

179

180

181

182

183

184

185

186

187

188

189

190

191

192

193

194

195

196

197

198

199

200

201

202

203

204

205

206

207

208

209

210

211

212

213

214

215

216

217

218

219

220

221

222

223

224

225

226

227

228

229

230

231

232

233

234

235

236

237

238

239

240

241

242

243

244

245

246

247

248

249

250

251

252

253

254

255

256

257

258

259

260

261

262

263

264

265

266

267

268

269

270

271

272

273

274

275

276

277

278

279

280

281

282

283

284

285

286

287

288

289

290

291

292

######################## BEGIN LICENSE BLOCK ######################## 

# The Original Code is Mozilla Universal charset detector code. 

# 

# The Initial Developer of the Original Code is 

# Shy Shalom 

# Portions created by the Initial Developer are Copyright (C) 2005 

# the Initial Developer. All Rights Reserved. 

# 

# Contributor(s): 

# Mark Pilgrim - port to Python 

# 

# This library is free software; you can redistribute it and/or 

# modify it under the terms of the GNU Lesser General Public 

# License as published by the Free Software Foundation; either 

# version 2.1 of the License, or (at your option) any later version. 

# 

# This library is distributed in the hope that it will be useful, 

# but WITHOUT ANY WARRANTY; without even the implied warranty of 

# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU 

# Lesser General Public License for more details. 

# 

# You should have received a copy of the GNU Lesser General Public 

# License along with this library; if not, write to the Free Software 

# Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 

# 02110-1301 USA 

######################### END LICENSE BLOCK ######################### 

 

from .charsetprober import CharSetProber 

from .enums import ProbingState 

 

# This prober doesn't actually recognize a language or a charset. 

# It is a helper prober for the use of the Hebrew model probers 

 

### General ideas of the Hebrew charset recognition ### 

# 

# Four main charsets exist in Hebrew: 

# "ISO-8859-8" - Visual Hebrew 

# "windows-1255" - Logical Hebrew 

# "ISO-8859-8-I" - Logical Hebrew 

# "x-mac-hebrew" - ?? Logical Hebrew ?? 

# 

# Both "ISO" charsets use a completely identical set of code points, whereas 

# "windows-1255" and "x-mac-hebrew" are two different proper supersets of 

# these code points. windows-1255 defines additional characters in the range 

# 0x80-0x9F as some misc punctuation marks as well as some Hebrew-specific 

# diacritics and additional 'Yiddish' ligature letters in the range 0xc0-0xd6. 

# x-mac-hebrew defines similar additional code points but with a different 

# mapping. 

# 

# As far as an average Hebrew text with no diacritics is concerned, all four 

# charsets are identical with respect to code points. Meaning that for the 

# main Hebrew alphabet, all four map the same values to all 27 Hebrew letters 

# (including final letters). 

# 

# The dominant difference between these charsets is their directionality. 

# "Visual" directionality means that the text is ordered as if the renderer is 

# not aware of a BIDI rendering algorithm. The renderer sees the text and 

# draws it from left to right. The text itself when ordered naturally is read 

# backwards. A buffer of Visual Hebrew generally looks like so: 

# "[last word of first line spelled backwards] [whole line ordered backwards 

# and spelled backwards] [first word of first line spelled backwards] 

# [end of line] [last word of second line] ... etc' " 

# adding punctuation marks, numbers and English text to visual text is 

# naturally also "visual" and from left to right. 

# 

# "Logical" directionality means the text is ordered "naturally" according to 

# the order it is read. It is the responsibility of the renderer to display 

# the text from right to left. A BIDI algorithm is used to place general 

# punctuation marks, numbers and English text in the text. 

# 

# Texts in x-mac-hebrew are almost impossible to find on the Internet. From 

# what little evidence I could find, it seems that its general directionality 

# is Logical. 

# 

# To sum up all of the above, the Hebrew probing mechanism knows about two 

# charsets: 

# Visual Hebrew - "ISO-8859-8" - backwards text - Words and sentences are 

# backwards while line order is natural. For charset recognition purposes 

# the line order is unimportant (In fact, for this implementation, even 

# word order is unimportant). 

# Logical Hebrew - "windows-1255" - normal, naturally ordered text. 

# 

# "ISO-8859-8-I" is a subset of windows-1255 and doesn't need to be 

# specifically identified. 

# "x-mac-hebrew" is also identified as windows-1255. A text in x-mac-hebrew 

# that contain special punctuation marks or diacritics is displayed with 

# some unconverted characters showing as question marks. This problem might 

# be corrected using another model prober for x-mac-hebrew. Due to the fact 

# that x-mac-hebrew texts are so rare, writing another model prober isn't 

# worth the effort and performance hit. 

# 

#### The Prober #### 

# 

# The prober is divided between two SBCharSetProbers and a HebrewProber, 

# all of which are managed, created, fed data, inquired and deleted by the 

# SBCSGroupProber. The two SBCharSetProbers identify that the text is in 

# fact some kind of Hebrew, Logical or Visual. The final decision about which 

# one is it is made by the HebrewProber by combining final-letter scores 

# with the scores of the two SBCharSetProbers to produce a final answer. 

# 

# The SBCSGroupProber is responsible for stripping the original text of HTML 

# tags, English characters, numbers, low-ASCII punctuation characters, spaces 

# and new lines. It reduces any sequence of such characters to a single space. 

# The buffer fed to each prober in the SBCS group prober is pure text in 

# high-ASCII. 

# The two SBCharSetProbers (model probers) share the same language model: 

# Win1255Model. 

# The first SBCharSetProber uses the model normally as any other 

# SBCharSetProber does, to recognize windows-1255, upon which this model was 

# built. The second SBCharSetProber is told to make the pair-of-letter 

# lookup in the language model backwards. This in practice exactly simulates 

# a visual Hebrew model using the windows-1255 logical Hebrew model. 

# 

# The HebrewProber is not using any language model. All it does is look for 

# final-letter evidence suggesting the text is either logical Hebrew or visual 

# Hebrew. Disjointed from the model probers, the results of the HebrewProber 

# alone are meaningless. HebrewProber always returns 0.00 as confidence 

# since it never identifies a charset by itself. Instead, the pointer to the 

# HebrewProber is passed to the model probers as a helper "Name Prober". 

# When the Group prober receives a positive identification from any prober, 

# it asks for the name of the charset identified. If the prober queried is a 

# Hebrew model prober, the model prober forwards the call to the 

# HebrewProber to make the final decision. In the HebrewProber, the 

# decision is made according to the final-letters scores maintained and Both 

# model probers scores. The answer is returned in the form of the name of the 

# charset identified, either "windows-1255" or "ISO-8859-8". 

 

class HebrewProber(CharSetProber): 

# windows-1255 / ISO-8859-8 code points of interest 

FINAL_KAF = 0xea 

NORMAL_KAF = 0xeb 

FINAL_MEM = 0xed 

NORMAL_MEM = 0xee 

FINAL_NUN = 0xef 

NORMAL_NUN = 0xf0 

FINAL_PE = 0xf3 

NORMAL_PE = 0xf4 

FINAL_TSADI = 0xf5 

NORMAL_TSADI = 0xf6 

 

# Minimum Visual vs Logical final letter score difference. 

# If the difference is below this, don't rely solely on the final letter score 

# distance. 

MIN_FINAL_CHAR_DISTANCE = 5 

 

# Minimum Visual vs Logical model score difference. 

# If the difference is below this, don't rely at all on the model score 

# distance. 

MIN_MODEL_DISTANCE = 0.01 

 

VISUAL_HEBREW_NAME = "ISO-8859-8" 

LOGICAL_HEBREW_NAME = "windows-1255" 

 

def __init__(self): 

super(HebrewProber, self).__init__() 

self._final_char_logical_score = None 

self._final_char_visual_score = None 

self._prev = None 

self._before_prev = None 

self._logical_prober = None 

self._visual_prober = None 

self.reset() 

 

def reset(self): 

self._final_char_logical_score = 0 

self._final_char_visual_score = 0 

# The two last characters seen in the previous buffer, 

# mPrev and mBeforePrev are initialized to space in order to simulate 

# a word delimiter at the beginning of the data 

self._prev = ' ' 

self._before_prev = ' ' 

# These probers are owned by the group prober. 

 

def set_model_probers(self, logicalProber, visualProber): 

self._logical_prober = logicalProber 

self._visual_prober = visualProber 

 

def is_final(self, c): 

return c in [self.FINAL_KAF, self.FINAL_MEM, self.FINAL_NUN, 

self.FINAL_PE, self.FINAL_TSADI] 

 

def is_non_final(self, c): 

# The normal Tsadi is not a good Non-Final letter due to words like 

# 'lechotet' (to chat) containing an apostrophe after the tsadi. This 

# apostrophe is converted to a space in FilterWithoutEnglishLetters 

# causing the Non-Final tsadi to appear at an end of a word even 

# though this is not the case in the original text. 

# The letters Pe and Kaf rarely display a related behavior of not being 

# a good Non-Final letter. Words like 'Pop', 'Winamp' and 'Mubarak' 

# for example legally end with a Non-Final Pe or Kaf. However, the 

# benefit of these letters as Non-Final letters outweighs the damage 

# since these words are quite rare. 

return c in [self.NORMAL_KAF, self.NORMAL_MEM, 

self.NORMAL_NUN, self.NORMAL_PE] 

 

def feed(self, byte_str): 

# Final letter analysis for logical-visual decision. 

# Look for evidence that the received buffer is either logical Hebrew 

# or visual Hebrew. 

# The following cases are checked: 

# 1) A word longer than 1 letter, ending with a final letter. This is 

# an indication that the text is laid out "naturally" since the 

# final letter really appears at the end. +1 for logical score. 

# 2) A word longer than 1 letter, ending with a Non-Final letter. In 

# normal Hebrew, words ending with Kaf, Mem, Nun, Pe or Tsadi, 

# should not end with the Non-Final form of that letter. Exceptions 

# to this rule are mentioned above in isNonFinal(). This is an 

# indication that the text is laid out backwards. +1 for visual 

# score 

# 3) A word longer than 1 letter, starting with a final letter. Final 

# letters should not appear at the beginning of a word. This is an 

# indication that the text is laid out backwards. +1 for visual 

# score. 

# 

# The visual score and logical score are accumulated throughout the 

# text and are finally checked against each other in GetCharSetName(). 

# No checking for final letters in the middle of words is done since 

# that case is not an indication for either Logical or Visual text. 

# 

# We automatically filter out all 7-bit characters (replace them with 

# spaces) so the word boundary detection works properly. [MAP] 

 

if self.state == ProbingState.NOT_ME: 

# Both model probers say it's not them. No reason to continue. 

return ProbingState.NOT_ME 

 

byte_str = self.filter_high_byte_only(byte_str) 

 

for cur in byte_str: 

if cur == ' ': 

# We stand on a space - a word just ended 

if self._before_prev != ' ': 

# next-to-last char was not a space so self._prev is not a 

# 1 letter word 

if self.is_final(self._prev): 

# case (1) [-2:not space][-1:final letter][cur:space] 

self._final_char_logical_score += 1 

elif self.is_non_final(self._prev): 

# case (2) [-2:not space][-1:Non-Final letter][ 

# cur:space] 

self._final_char_visual_score += 1 

else: 

# Not standing on a space 

if ((self._before_prev == ' ') and 

(self.is_final(self._prev)) and (cur != ' ')): 

# case (3) [-2:space][-1:final letter][cur:not space] 

self._final_char_visual_score += 1 

self._before_prev = self._prev 

self._prev = cur 

 

# Forever detecting, till the end or until both model probers return 

# ProbingState.NOT_ME (handled above) 

return ProbingState.DETECTING 

 

@property 

def charset_name(self): 

# Make the decision: is it Logical or Visual? 

# If the final letter score distance is dominant enough, rely on it. 

finalsub = self._final_char_logical_score - self._final_char_visual_score 

if finalsub >= self.MIN_FINAL_CHAR_DISTANCE: 

return self.LOGICAL_HEBREW_NAME 

if finalsub <= -self.MIN_FINAL_CHAR_DISTANCE: 

return self.VISUAL_HEBREW_NAME 

 

# It's not dominant enough, try to rely on the model scores instead. 

modelsub = (self._logical_prober.get_confidence() 

- self._visual_prober.get_confidence()) 

if modelsub > self.MIN_MODEL_DISTANCE: 

return self.LOGICAL_HEBREW_NAME 

if modelsub < -self.MIN_MODEL_DISTANCE: 

return self.VISUAL_HEBREW_NAME 

 

# Still no good, back to final letter distance, maybe it'll save the 

# day. 

if finalsub < 0.0: 

return self.VISUAL_HEBREW_NAME 

 

# (finalsub > 0 - Logical) or (don't know what to do) default to 

# Logical. 

return self.LOGICAL_HEBREW_NAME 

 

@property 

def language(self): 

return 'Hebrew' 

 

@property 

def state(self): 

# Remain active as long as any of the model probers are active. 

if (self._logical_prober.state == ProbingState.NOT_ME) and \ 

(self._visual_prober.state == ProbingState.NOT_ME): 

return ProbingState.NOT_ME 

return ProbingState.DETECTING