1 | /* -*- C++ -*- |
2 | SPDX-FileCopyrightText: 1998 Netscape Communications Corporation <developer@mozilla.org> |
3 | |
4 | SPDX-License-Identifier: MIT |
5 | */ |
6 | |
7 | #ifndef nsHebrewProber_h__ |
8 | #define nsHebrewProber_h__ |
9 | |
10 | #include "nsSBCharSetProber.h" |
11 | namespace kencodingprober |
12 | { |
13 | // This prober doesn't actually recognize a language or a charset. |
14 | // It is a helper prober for the use of the Hebrew model probers |
15 | class KCODECS_NO_EXPORT nsHebrewProber : public nsCharSetProber |
16 | { |
17 | public: |
18 | nsHebrewProber(void) |
19 | : mLogicalProb(nullptr) |
20 | , mVisualProb(nullptr) |
21 | { |
22 | Reset(); |
23 | } |
24 | |
25 | ~nsHebrewProber(void) override |
26 | { |
27 | } |
28 | nsProbingState HandleData(const char *aBuf, unsigned int aLen) override; |
29 | const char *GetCharSetName() override; |
30 | void Reset(void) override; |
31 | |
32 | nsProbingState GetState(void) override; |
33 | |
34 | float GetConfidence(void) override |
35 | { |
36 | return (float)0.0; |
37 | } |
38 | void SetOpion() override |
39 | { |
40 | } |
41 | |
42 | void SetModelProbers(nsCharSetProber *logicalPrb, nsCharSetProber *visualPrb) |
43 | { |
44 | mLogicalProb = logicalPrb; |
45 | mVisualProb = visualPrb; |
46 | } |
47 | |
48 | #ifdef DEBUG_PROBE |
49 | void DumpStatus() override; |
50 | #endif |
51 | |
52 | protected: |
53 | static bool isFinal(char c); |
54 | static bool isNonFinal(char c); |
55 | |
56 | int mFinalCharLogicalScore, mFinalCharVisualScore; |
57 | |
58 | // The two last characters seen in the previous buffer. |
59 | char mPrev, mBeforePrev; |
60 | |
61 | // These probers are owned by the group prober. |
62 | nsCharSetProber *mLogicalProb, *mVisualProb; |
63 | }; |
64 | } |
65 | |
66 | /** |
67 | * ** General ideas of the Hebrew charset recognition ** |
68 | * |
69 | * Four main charsets exist in Hebrew: |
70 | * "ISO-8859-8" - Visual Hebrew |
71 | * "windows-1255" - Logical Hebrew |
72 | * "ISO-8859-8-I" - Logical Hebrew |
73 | * "x-mac-hebrew" - ?? Logical Hebrew ?? |
74 | * |
75 | * Both "ISO" charsets use a completely identical set of code points, whereas |
76 | * "windows-1255" and "x-mac-hebrew" are two different proper supersets of |
77 | * these code points. windows-1255 defines additional characters in the range |
78 | * 0x80-0x9F as some misc punctuation marks as well as some Hebrew-specific |
79 | * diacritics and additional 'Yiddish' ligature letters in the range 0xc0-0xd6. |
80 | * x-mac-hebrew defines similar additional code points but with a different |
81 | * mapping. |
82 | * |
83 | * As far as an average Hebrew text with no diacritics is concerned, all four |
84 | * charsets are identical with respect to code points. Meaning that for the |
85 | * main Hebrew alphabet, all four map the same values to all 27 Hebrew letters |
86 | * (including final letters). |
87 | * |
88 | * The dominant difference between these charsets is their directionality. |
89 | * "Visual" directionality means that the text is ordered as if the renderer is |
90 | * not aware of a BIDI rendering algorithm. The renderer sees the text and |
91 | * draws it from left to right. The text itself when ordered naturally is read |
92 | * backwards. A buffer of Visual Hebrew generally looks like so: |
93 | * "[last word of first line spelled backwards] [whole line ordered backwards |
94 | * and spelled backwards] [first word of first line spelled backwards] |
95 | * [end of line] [last word of second line] ... etc' " |
96 | * adding punctuation marks, numbers and English text to visual text is |
97 | * naturally also "visual" and from left to right. |
98 | * |
99 | * "Logical" directionality means the text is ordered "naturally" according to |
100 | * the order it is read. It is the responsibility of the renderer to display |
101 | * the text from right to left. A BIDI algorithm is used to place general |
102 | * punctuation marks, numbers and English text in the text. |
103 | * |
104 | * Texts in x-mac-hebrew are almost impossible to find on the Internet. From |
105 | * what little evidence I could find, it seems that its general directionality |
106 | * is Logical. |
107 | * |
108 | * To sum up all of the above, the Hebrew probing mechanism knows about two |
109 | * charsets: |
110 | * Visual Hebrew - "ISO-8859-8" - backwards text - Words and sentences are |
111 | * backwards while line order is natural. For charset recognition purposes |
112 | * the line order is unimportant (In fact, for this implementation, even |
113 | * word order is unimportant). |
114 | * Logical Hebrew - "windows-1255" - normal, naturally ordered text. |
115 | * |
116 | * "ISO-8859-8-I" is a subset of windows-1255 and doesn't need to be |
117 | * specifically identified. |
118 | * "x-mac-hebrew" is also identified as windows-1255. A text in x-mac-hebrew |
119 | * that contain special punctuation marks or diacritics is displayed with |
120 | * some unconverted characters showing as question marks. This problem might |
121 | * be corrected using another model prober for x-mac-hebrew. Due to the fact |
122 | * that x-mac-hebrew texts are so rare, writing another model prober isn't |
123 | * worth the effort and performance hit. |
124 | * |
125 | * *** The Prober *** |
126 | * |
127 | * The prober is divided between two nsSBCharSetProbers and an nsHebrewProber, |
128 | * all of which are managed, created, fed data, inquired and deleted by the |
129 | * nsSBCSGroupProber. The two nsSBCharSetProbers identify that the text is in |
130 | * fact some kind of Hebrew, Logical or Visual. The final decision about which |
131 | * one is it is made by the nsHebrewProber by combining final-letter scores |
132 | * with the scores of the two nsSBCharSetProbers to produce a final answer. |
133 | * |
134 | * The nsSBCSGroupProber is responsible for stripping the original text of HTML |
135 | * tags, English characters, numbers, low-ASCII punctuation characters, spaces |
136 | * and new lines. It reduces any sequence of such characters to a single space. |
137 | * The buffer fed to each prober in the SBCS group prober is pure text in |
138 | * high-ASCII. |
139 | * The two nsSBCharSetProbers (model probers) share the same language model: |
140 | * Win1255Model. |
141 | * The first nsSBCharSetProber uses the model normally as any other |
142 | * nsSBCharSetProber does, to recognize windows-1255, upon which this model was |
143 | * built. The second nsSBCharSetProber is told to make the pair-of-letter |
144 | * lookup in the language model backwards. This in practice exactly simulates |
145 | * a visual Hebrew model using the windows-1255 logical Hebrew model. |
146 | * |
147 | * The nsHebrewProber is not using any language model. All it does is look for |
148 | * final-letter evidence suggesting the text is either logical Hebrew or visual |
149 | * Hebrew. Disjointed from the model probers, the results of the nsHebrewProber |
150 | * alone are meaningless. nsHebrewProber always returns 0.00 as confidence |
151 | * since it never identifies a charset by itself. Instead, the pointer to the |
152 | * nsHebrewProber is passed to the model probers as a helper "Name Prober". |
153 | * When the Group prober receives a positive identification from any prober, |
154 | * it asks for the name of the charset identified. If the prober queried is a |
155 | * Hebrew model prober, the model prober forwards the call to the |
156 | * nsHebrewProber to make the final decision. In the nsHebrewProber, the |
157 | * decision is made according to the final-letters scores maintained and Both |
158 | * model probers scores. The answer is returned in the form of the name of the |
159 | * charset identified, either "windows-1255" or "ISO-8859-8". |
160 | * |
161 | */ |
162 | #endif /* nsHebrewProber_h__ */ |
163 | |