gvsig-scripting / org.gvsig.scripting / trunk / org.gvsig.scripting / org.gvsig.scripting.app / org.gvsig.scripting.app.mainplugin / src / main / resources-plugin / scripting / lib / requests / packages / chardet / hebrewprober.py @ 564
History | View | Annotate | Download (13 KB)
1 |
######################## BEGIN LICENSE BLOCK ########################
|
---|---|
2 |
# The Original Code is Mozilla Universal charset detector code.
|
3 |
#
|
4 |
# The Initial Developer of the Original Code is
|
5 |
# Shy Shalom
|
6 |
# Portions created by the Initial Developer are Copyright (C) 2005
|
7 |
# the Initial Developer. All Rights Reserved.
|
8 |
#
|
9 |
# Contributor(s):
|
10 |
# Mark Pilgrim - port to Python
|
11 |
#
|
12 |
# This library is free software; you can redistribute it and/or
|
13 |
# modify it under the terms of the GNU Lesser General Public
|
14 |
# License as published by the Free Software Foundation; either
|
15 |
# version 2.1 of the License, or (at your option) any later version.
|
16 |
#
|
17 |
# This library is distributed in the hope that it will be useful,
|
18 |
# but WITHOUT ANY WARRANTY; without even the implied warranty of
|
19 |
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
|
20 |
# Lesser General Public License for more details.
|
21 |
#
|
22 |
# You should have received a copy of the GNU Lesser General Public
|
23 |
# License along with this library; if not, write to the Free Software
|
24 |
# Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA
|
25 |
# 02110-1301 USA
|
26 |
######################### END LICENSE BLOCK #########################
|
27 |
|
28 |
from .charsetprober import CharSetProber |
29 |
from .constants import eNotMe, eDetecting |
30 |
from .compat import wrap_ord |
31 |
|
32 |
# This prober doesn't actually recognize a language or a charset.
|
33 |
# It is a helper prober for the use of the Hebrew model probers
|
34 |
|
35 |
### General ideas of the Hebrew charset recognition ###
|
36 |
#
|
37 |
# Four main charsets exist in Hebrew:
|
38 |
# "ISO-8859-8" - Visual Hebrew
|
39 |
# "windows-1255" - Logical Hebrew
|
40 |
# "ISO-8859-8-I" - Logical Hebrew
|
41 |
# "x-mac-hebrew" - ?? Logical Hebrew ??
|
42 |
#
|
43 |
# Both "ISO" charsets use a completely identical set of code points, whereas
|
44 |
# "windows-1255" and "x-mac-hebrew" are two different proper supersets of
|
45 |
# these code points. windows-1255 defines additional characters in the range
|
46 |
# 0x80-0x9F as some misc punctuation marks as well as some Hebrew-specific
|
47 |
# diacritics and additional 'Yiddish' ligature letters in the range 0xc0-0xd6.
|
48 |
# x-mac-hebrew defines similar additional code points but with a different
|
49 |
# mapping.
|
50 |
#
|
51 |
# As far as an average Hebrew text with no diacritics is concerned, all four
|
52 |
# charsets are identical with respect to code points. Meaning that for the
|
53 |
# main Hebrew alphabet, all four map the same values to all 27 Hebrew letters
|
54 |
# (including final letters).
|
55 |
#
|
56 |
# The dominant difference between these charsets is their directionality.
|
57 |
# "Visual" directionality means that the text is ordered as if the renderer is
|
58 |
# not aware of a BIDI rendering algorithm. The renderer sees the text and
|
59 |
# draws it from left to right. The text itself when ordered naturally is read
|
60 |
# backwards. A buffer of Visual Hebrew generally looks like so:
|
61 |
# "[last word of first line spelled backwards] [whole line ordered backwards
|
62 |
# and spelled backwards] [first word of first line spelled backwards]
|
63 |
# [end of line] [last word of second line] ... etc' "
|
64 |
# adding punctuation marks, numbers and English text to visual text is
|
65 |
# naturally also "visual" and from left to right.
|
66 |
#
|
67 |
# "Logical" directionality means the text is ordered "naturally" according to
|
68 |
# the order it is read. It is the responsibility of the renderer to display
|
69 |
# the text from right to left. A BIDI algorithm is used to place general
|
70 |
# punctuation marks, numbers and English text in the text.
|
71 |
#
|
72 |
# Texts in x-mac-hebrew are almost impossible to find on the Internet. From
|
73 |
# what little evidence I could find, it seems that its general directionality
|
74 |
# is Logical.
|
75 |
#
|
76 |
# To sum up all of the above, the Hebrew probing mechanism knows about two
|
77 |
# charsets:
|
78 |
# Visual Hebrew - "ISO-8859-8" - backwards text - Words and sentences are
|
79 |
# backwards while line order is natural. For charset recognition purposes
|
80 |
# the line order is unimportant (In fact, for this implementation, even
|
81 |
# word order is unimportant).
|
82 |
# Logical Hebrew - "windows-1255" - normal, naturally ordered text.
|
83 |
#
|
84 |
# "ISO-8859-8-I" is a subset of windows-1255 and doesn't need to be
|
85 |
# specifically identified.
|
86 |
# "x-mac-hebrew" is also identified as windows-1255. A text in x-mac-hebrew
|
87 |
# that contain special punctuation marks or diacritics is displayed with
|
88 |
# some unconverted characters showing as question marks. This problem might
|
89 |
# be corrected using another model prober for x-mac-hebrew. Due to the fact
|
90 |
# that x-mac-hebrew texts are so rare, writing another model prober isn't
|
91 |
# worth the effort and performance hit.
|
92 |
#
|
93 |
#### The Prober ####
|
94 |
#
|
95 |
# The prober is divided between two SBCharSetProbers and a HebrewProber,
|
96 |
# all of which are managed, created, fed data, inquired and deleted by the
|
97 |
# SBCSGroupProber. The two SBCharSetProbers identify that the text is in
|
98 |
# fact some kind of Hebrew, Logical or Visual. The final decision about which
|
99 |
# one is it is made by the HebrewProber by combining final-letter scores
|
100 |
# with the scores of the two SBCharSetProbers to produce a final answer.
|
101 |
#
|
102 |
# The SBCSGroupProber is responsible for stripping the original text of HTML
|
103 |
# tags, English characters, numbers, low-ASCII punctuation characters, spaces
|
104 |
# and new lines. It reduces any sequence of such characters to a single space.
|
105 |
# The buffer fed to each prober in the SBCS group prober is pure text in
|
106 |
# high-ASCII.
|
107 |
# The two SBCharSetProbers (model probers) share the same language model:
|
108 |
# Win1255Model.
|
109 |
# The first SBCharSetProber uses the model normally as any other
|
110 |
# SBCharSetProber does, to recognize windows-1255, upon which this model was
|
111 |
# built. The second SBCharSetProber is told to make the pair-of-letter
|
112 |
# lookup in the language model backwards. This in practice exactly simulates
|
113 |
# a visual Hebrew model using the windows-1255 logical Hebrew model.
|
114 |
#
|
115 |
# The HebrewProber is not using any language model. All it does is look for
|
116 |
# final-letter evidence suggesting the text is either logical Hebrew or visual
|
117 |
# Hebrew. Disjointed from the model probers, the results of the HebrewProber
|
118 |
# alone are meaningless. HebrewProber always returns 0.00 as confidence
|
119 |
# since it never identifies a charset by itself. Instead, the pointer to the
|
120 |
# HebrewProber is passed to the model probers as a helper "Name Prober".
|
121 |
# When the Group prober receives a positive identification from any prober,
|
122 |
# it asks for the name of the charset identified. If the prober queried is a
|
123 |
# Hebrew model prober, the model prober forwards the call to the
|
124 |
# HebrewProber to make the final decision. In the HebrewProber, the
|
125 |
# decision is made according to the final-letters scores maintained and Both
|
126 |
# model probers scores. The answer is returned in the form of the name of the
|
127 |
# charset identified, either "windows-1255" or "ISO-8859-8".
|
128 |
|
129 |
# windows-1255 / ISO-8859-8 code points of interest
|
130 |
FINAL_KAF = 0xea
|
131 |
NORMAL_KAF = 0xeb
|
132 |
FINAL_MEM = 0xed
|
133 |
NORMAL_MEM = 0xee
|
134 |
FINAL_NUN = 0xef
|
135 |
NORMAL_NUN = 0xf0
|
136 |
FINAL_PE = 0xf3
|
137 |
NORMAL_PE = 0xf4
|
138 |
FINAL_TSADI = 0xf5
|
139 |
NORMAL_TSADI = 0xf6
|
140 |
|
141 |
# Minimum Visual vs Logical final letter score difference.
|
142 |
# If the difference is below this, don't rely solely on the final letter score
|
143 |
# distance.
|
144 |
MIN_FINAL_CHAR_DISTANCE = 5
|
145 |
|
146 |
# Minimum Visual vs Logical model score difference.
|
147 |
# If the difference is below this, don't rely at all on the model score
|
148 |
# distance.
|
149 |
MIN_MODEL_DISTANCE = 0.01
|
150 |
|
151 |
VISUAL_HEBREW_NAME = "ISO-8859-8"
|
152 |
LOGICAL_HEBREW_NAME = "windows-1255"
|
153 |
|
154 |
|
155 |
class HebrewProber(CharSetProber): |
156 |
def __init__(self): |
157 |
CharSetProber.__init__(self)
|
158 |
self._mLogicalProber = None |
159 |
self._mVisualProber = None |
160 |
self.reset()
|
161 |
|
162 |
def reset(self): |
163 |
self._mFinalCharLogicalScore = 0 |
164 |
self._mFinalCharVisualScore = 0 |
165 |
# The two last characters seen in the previous buffer,
|
166 |
# mPrev and mBeforePrev are initialized to space in order to simulate
|
167 |
# a word delimiter at the beginning of the data
|
168 |
self._mPrev = ' ' |
169 |
self._mBeforePrev = ' ' |
170 |
# These probers are owned by the group prober.
|
171 |
|
172 |
def set_model_probers(self, logicalProber, visualProber): |
173 |
self._mLogicalProber = logicalProber
|
174 |
self._mVisualProber = visualProber
|
175 |
|
176 |
def is_final(self, c): |
177 |
return wrap_ord(c) in [FINAL_KAF, FINAL_MEM, FINAL_NUN, FINAL_PE, |
178 |
FINAL_TSADI] |
179 |
|
180 |
def is_non_final(self, c): |
181 |
# The normal Tsadi is not a good Non-Final letter due to words like
|
182 |
# 'lechotet' (to chat) containing an apostrophe after the tsadi. This
|
183 |
# apostrophe is converted to a space in FilterWithoutEnglishLetters
|
184 |
# causing the Non-Final tsadi to appear at an end of a word even
|
185 |
# though this is not the case in the original text.
|
186 |
# The letters Pe and Kaf rarely display a related behavior of not being
|
187 |
# a good Non-Final letter. Words like 'Pop', 'Winamp' and 'Mubarak'
|
188 |
# for example legally end with a Non-Final Pe or Kaf. However, the
|
189 |
# benefit of these letters as Non-Final letters outweighs the damage
|
190 |
# since these words are quite rare.
|
191 |
return wrap_ord(c) in [NORMAL_KAF, NORMAL_MEM, NORMAL_NUN, NORMAL_PE] |
192 |
|
193 |
def feed(self, aBuf): |
194 |
# Final letter analysis for logical-visual decision.
|
195 |
# Look for evidence that the received buffer is either logical Hebrew
|
196 |
# or visual Hebrew.
|
197 |
# The following cases are checked:
|
198 |
# 1) A word longer than 1 letter, ending with a final letter. This is
|
199 |
# an indication that the text is laid out "naturally" since the
|
200 |
# final letter really appears at the end. +1 for logical score.
|
201 |
# 2) A word longer than 1 letter, ending with a Non-Final letter. In
|
202 |
# normal Hebrew, words ending with Kaf, Mem, Nun, Pe or Tsadi,
|
203 |
# should not end with the Non-Final form of that letter. Exceptions
|
204 |
# to this rule are mentioned above in isNonFinal(). This is an
|
205 |
# indication that the text is laid out backwards. +1 for visual
|
206 |
# score
|
207 |
# 3) A word longer than 1 letter, starting with a final letter. Final
|
208 |
# letters should not appear at the beginning of a word. This is an
|
209 |
# indication that the text is laid out backwards. +1 for visual
|
210 |
# score.
|
211 |
#
|
212 |
# The visual score and logical score are accumulated throughout the
|
213 |
# text and are finally checked against each other in GetCharSetName().
|
214 |
# No checking for final letters in the middle of words is done since
|
215 |
# that case is not an indication for either Logical or Visual text.
|
216 |
#
|
217 |
# We automatically filter out all 7-bit characters (replace them with
|
218 |
# spaces) so the word boundary detection works properly. [MAP]
|
219 |
|
220 |
if self.get_state() == eNotMe: |
221 |
# Both model probers say it's not them. No reason to continue.
|
222 |
return eNotMe
|
223 |
|
224 |
aBuf = self.filter_high_bit_only(aBuf)
|
225 |
|
226 |
for cur in aBuf: |
227 |
if cur == ' ': |
228 |
# We stand on a space - a word just ended
|
229 |
if self._mBeforePrev != ' ': |
230 |
# next-to-last char was not a space so self._mPrev is not a
|
231 |
# 1 letter word
|
232 |
if self.is_final(self._mPrev): |
233 |
# case (1) [-2:not space][-1:final letter][cur:space]
|
234 |
self._mFinalCharLogicalScore += 1 |
235 |
elif self.is_non_final(self._mPrev): |
236 |
# case (2) [-2:not space][-1:Non-Final letter][
|
237 |
# cur:space]
|
238 |
self._mFinalCharVisualScore += 1 |
239 |
else:
|
240 |
# Not standing on a space
|
241 |
if ((self._mBeforePrev == ' ') and |
242 |
(self.is_final(self._mPrev)) and (cur != ' ')): |
243 |
# case (3) [-2:space][-1:final letter][cur:not space]
|
244 |
self._mFinalCharVisualScore += 1 |
245 |
self._mBeforePrev = self._mPrev |
246 |
self._mPrev = cur
|
247 |
|
248 |
# Forever detecting, till the end or until both model probers return
|
249 |
# eNotMe (handled above)
|
250 |
return eDetecting
|
251 |
|
252 |
def get_charset_name(self): |
253 |
# Make the decision: is it Logical or Visual?
|
254 |
# If the final letter score distance is dominant enough, rely on it.
|
255 |
finalsub = self._mFinalCharLogicalScore - self._mFinalCharVisualScore |
256 |
if finalsub >= MIN_FINAL_CHAR_DISTANCE:
|
257 |
return LOGICAL_HEBREW_NAME
|
258 |
if finalsub <= -MIN_FINAL_CHAR_DISTANCE:
|
259 |
return VISUAL_HEBREW_NAME
|
260 |
|
261 |
# It's not dominant enough, try to rely on the model scores instead.
|
262 |
modelsub = (self._mLogicalProber.get_confidence()
|
263 |
- self._mVisualProber.get_confidence())
|
264 |
if modelsub > MIN_MODEL_DISTANCE:
|
265 |
return LOGICAL_HEBREW_NAME
|
266 |
if modelsub < -MIN_MODEL_DISTANCE:
|
267 |
return VISUAL_HEBREW_NAME
|
268 |
|
269 |
# Still no good, back to final letter distance, maybe it'll save the
|
270 |
# day.
|
271 |
if finalsub < 0.0: |
272 |
return VISUAL_HEBREW_NAME
|
273 |
|
274 |
# (finalsub > 0 - Logical) or (don't know what to do) default to
|
275 |
# Logical.
|
276 |
return LOGICAL_HEBREW_NAME
|
277 |
|
278 |
def get_state(self): |
279 |
# Remain active as long as any of the model probers are active.
|
280 |
if (self._mLogicalProber.get_state() == eNotMe) and \ |
281 |
(self._mVisualProber.get_state() == eNotMe):
|
282 |
return eNotMe
|
283 |
return eDetecting
|