• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 /*
2  **********************************************************************
3  *   Copyright (C) 1999-2008, International Business Machines
4  *   Corporation and others.  All Rights Reserved.
5  **********************************************************************
6  *   Date        Name        Description
7  *   11/17/99    aliu        Creation.
8  **********************************************************************
9  */
10 
11 #include "unicode/utypes.h"
12 
13 #if !UCONFIG_NO_TRANSLITERATION
14 
15 #include "unicode/unistr.h"
16 #include "unicode/uniset.h"
17 #include "rbt_set.h"
18 #include "rbt_rule.h"
19 #include "cmemory.h"
20 #include "putilimp.h"
21 
22 U_CDECL_BEGIN
_deleteRule(void * rule)23 static void U_CALLCONV _deleteRule(void *rule) {
24     delete (U_NAMESPACE_QUALIFIER TransliterationRule *)rule;
25 }
26 U_CDECL_END
27 
28 //----------------------------------------------------------------------
29 // BEGIN Debugging support
30 //----------------------------------------------------------------------
31 
32 // #define DEBUG_RBT
33 
34 #ifdef DEBUG_RBT
35 #include <stdio.h>
36 #include "charstr.h"
37 
38 /**
39  * @param appendTo result is appended to this param.
40  * @param input the string being transliterated
41  * @param pos the index struct
42  */
_formatInput(UnicodeString & appendTo,const UnicodeString & input,const UTransPosition & pos)43 static UnicodeString& _formatInput(UnicodeString &appendTo,
44                                    const UnicodeString& input,
45                                    const UTransPosition& pos) {
46     // Output a string of the form aaa{bbb|ccc|ddd}eee, where
47     // the {} indicate the context start and limit, and the ||
48     // indicate the start and limit.
49     if (0 <= pos.contextStart &&
50         pos.contextStart <= pos.start &&
51         pos.start <= pos.limit &&
52         pos.limit <= pos.contextLimit &&
53         pos.contextLimit <= input.length()) {
54 
55         UnicodeString a, b, c, d, e;
56         input.extractBetween(0, pos.contextStart, a);
57         input.extractBetween(pos.contextStart, pos.start, b);
58         input.extractBetween(pos.start, pos.limit, c);
59         input.extractBetween(pos.limit, pos.contextLimit, d);
60         input.extractBetween(pos.contextLimit, input.length(), e);
61         appendTo.append(a).append((UChar)123/*{*/).append(b).
62             append((UChar)124/*|*/).append(c).append((UChar)124/*|*/).append(d).
63             append((UChar)125/*}*/).append(e);
64     } else {
65         appendTo.append("INVALID UTransPosition");
66         //appendTo.append((UnicodeString)"INVALID UTransPosition {cs=" +
67         //                pos.contextStart + ", s=" + pos.start + ", l=" +
68         //                pos.limit + ", cl=" + pos.contextLimit + "} on " +
69         //                input);
70     }
71     return appendTo;
72 }
73 
74 // Append a hex string to the target
_appendHex(uint32_t number,int32_t digits,UnicodeString & target)75 UnicodeString& _appendHex(uint32_t number,
76                           int32_t digits,
77                           UnicodeString& target) {
78     static const UChar digitString[] = {
79         0x30, 0x31, 0x32, 0x33, 0x34, 0x35, 0x36, 0x37, 0x38, 0x39,
80         0x41, 0x42, 0x43, 0x44, 0x45, 0x46, 0
81     };
82     while (digits--) {
83         target += digitString[(number >> (digits*4)) & 0xF];
84     }
85     return target;
86 }
87 
88 // Replace nonprintable characters with unicode escapes
_escape(const UnicodeString & source,UnicodeString & target)89 UnicodeString& _escape(const UnicodeString &source,
90                        UnicodeString &target) {
91     for (int32_t i = 0; i < source.length(); ) {
92         UChar32 ch = source.char32At(i);
93         i += UTF_CHAR_LENGTH(ch);
94         if (ch < 0x09 || (ch > 0x0A && ch < 0x20)|| ch > 0x7E) {
95             if (ch <= 0xFFFF) {
96                 target += "\\u";
97                 _appendHex(ch, 4, target);
98             } else {
99                 target += "\\U";
100                 _appendHex(ch, 8, target);
101             }
102         } else {
103             target += ch;
104         }
105     }
106     return target;
107 }
108 
_debugOut(const char * msg,TransliterationRule * rule,const Replaceable & theText,UTransPosition & pos)109 inline void _debugOut(const char* msg, TransliterationRule* rule,
110                       const Replaceable& theText, UTransPosition& pos) {
111     UnicodeString buf(msg, "");
112     if (rule) {
113         UnicodeString r;
114         rule->toRule(r, TRUE);
115         buf.append((UChar)32).append(r);
116     }
117     buf.append(UnicodeString(" => ", ""));
118     UnicodeString* text = (UnicodeString*)&theText;
119     _formatInput(buf, *text, pos);
120     UnicodeString esc;
121     _escape(buf, esc);
122     CharString cbuf(esc);
123     printf("%s\n", (const char*) cbuf);
124 }
125 
126 #else
127 #define _debugOut(msg, rule, theText, pos)
128 #endif
129 
130 //----------------------------------------------------------------------
131 // END Debugging support
132 //----------------------------------------------------------------------
133 
134 // Fill the precontext and postcontext with the patterns of the rules
135 // that are masking one another.
maskingError(const U_NAMESPACE_QUALIFIER TransliterationRule & rule1,const U_NAMESPACE_QUALIFIER TransliterationRule & rule2,UParseError & parseError)136 static void maskingError(const U_NAMESPACE_QUALIFIER TransliterationRule& rule1,
137                          const U_NAMESPACE_QUALIFIER TransliterationRule& rule2,
138                          UParseError& parseError) {
139     U_NAMESPACE_QUALIFIER UnicodeString r;
140     int32_t len;
141 
142     parseError.line = parseError.offset = -1;
143 
144     // for pre-context
145     rule1.toRule(r, FALSE);
146     len = uprv_min(r.length(), U_PARSE_CONTEXT_LEN-1);
147     r.extract(0, len, parseError.preContext);
148     parseError.preContext[len] = 0;
149 
150     //for post-context
151     r.truncate(0);
152     rule2.toRule(r, FALSE);
153     len = uprv_min(r.length(), U_PARSE_CONTEXT_LEN-1);
154     r.extract(0, len, parseError.postContext);
155     parseError.postContext[len] = 0;
156 }
157 
158 U_NAMESPACE_BEGIN
159 
160 /**
161  * Construct a new empty rule set.
162  */
TransliterationRuleSet(UErrorCode & status)163 TransliterationRuleSet::TransliterationRuleSet(UErrorCode& status) : UMemory() {
164     ruleVector = new UVector(&_deleteRule, NULL, status);
165     if (U_FAILURE(status)) {
166         return;
167     }
168     if (ruleVector == NULL) {
169         status = U_MEMORY_ALLOCATION_ERROR;
170     }
171     rules = NULL;
172     maxContextLength = 0;
173 }
174 
175 /**
176  * Copy constructor.
177  */
TransliterationRuleSet(const TransliterationRuleSet & other)178 TransliterationRuleSet::TransliterationRuleSet(const TransliterationRuleSet& other) :
179     UMemory(other),
180     ruleVector(0),
181     rules(0),
182     maxContextLength(other.maxContextLength) {
183 
184     int32_t i, len;
185     uprv_memcpy(index, other.index, sizeof(index));
186     UErrorCode status = U_ZERO_ERROR;
187     ruleVector = new UVector(&_deleteRule, NULL, status);
188     if (other.ruleVector != 0 && ruleVector != 0 && U_SUCCESS(status)) {
189         len = other.ruleVector->size();
190         for (i=0; i<len && U_SUCCESS(status); ++i) {
191             TransliterationRule *tempTranslitRule = new TransliterationRule(*(TransliterationRule*)other.ruleVector->elementAt(i));
192             // Null pointer test
193             if (tempTranslitRule == NULL) {
194                 status = U_MEMORY_ALLOCATION_ERROR;
195                 break;
196             }
197             ruleVector->addElement(tempTranslitRule, status);
198             if (U_FAILURE(status)) {
199                 break;
200             }
201         }
202     }
203     if (other.rules != 0 && U_SUCCESS(status)) {
204         UParseError p;
205         freeze(p, status);
206     }
207 }
208 
209 /**
210  * Destructor.
211  */
~TransliterationRuleSet()212 TransliterationRuleSet::~TransliterationRuleSet() {
213     delete ruleVector; // This deletes the contained rules
214     uprv_free(rules);
215 }
216 
setData(const TransliterationRuleData * d)217 void TransliterationRuleSet::setData(const TransliterationRuleData* d) {
218     /**
219      * We assume that the ruleset has already been frozen.
220      */
221     int32_t len = index[256]; // see freeze()
222     for (int32_t i=0; i<len; ++i) {
223         rules[i]->setData(d);
224     }
225 }
226 
227 /**
228  * Return the maximum context length.
229  * @return the length of the longest preceding context.
230  */
getMaximumContextLength(void) const231 int32_t TransliterationRuleSet::getMaximumContextLength(void) const {
232     return maxContextLength;
233 }
234 
235 /**
236  * Add a rule to this set.  Rules are added in order, and order is
237  * significant.  The last call to this method must be followed by
238  * a call to <code>freeze()</code> before the rule set is used.
239  *
240  * <p>If freeze() has already been called, calling addRule()
241  * unfreezes the rules, and freeze() must be called again.
242  *
243  * @param adoptedRule the rule to add
244  */
addRule(TransliterationRule * adoptedRule,UErrorCode & status)245 void TransliterationRuleSet::addRule(TransliterationRule* adoptedRule,
246                                      UErrorCode& status) {
247     if (U_FAILURE(status)) {
248         delete adoptedRule;
249         return;
250     }
251     ruleVector->addElement(adoptedRule, status);
252 
253     int32_t len;
254     if ((len = adoptedRule->getContextLength()) > maxContextLength) {
255         maxContextLength = len;
256     }
257 
258     uprv_free(rules);
259     rules = 0;
260 }
261 
262 /**
263  * Check this for masked rules and index it to optimize performance.
264  * The sequence of operations is: (1) add rules to a set using
265  * <code>addRule()</code>; (2) freeze the set using
266  * <code>freeze()</code>; (3) use the rule set.  If
267  * <code>addRule()</code> is called after calling this method, it
268  * invalidates this object, and this method must be called again.
269  * That is, <code>freeze()</code> may be called multiple times,
270  * although for optimal performance it shouldn't be.
271  */
freeze(UParseError & parseError,UErrorCode & status)272 void TransliterationRuleSet::freeze(UParseError& parseError,UErrorCode& status) {
273     /* Construct the rule array and index table.  We reorder the
274      * rules by sorting them into 256 bins.  Each bin contains all
275      * rules matching the index value for that bin.  A rule
276      * matches an index value if string whose first key character
277      * has a low byte equal to the index value can match the rule.
278      *
279      * Each bin contains zero or more rules, in the same order
280      * they were found originally.  However, the total rules in
281      * the bins may exceed the number in the original vector,
282      * since rules that have a variable as their first key
283      * character will generally fall into more than one bin.
284      *
285      * That is, each bin contains all rules that either have that
286      * first index value as their first key character, or have
287      * a set containing the index value as their first character.
288      */
289     int32_t n = ruleVector->size();
290     int32_t j;
291     int16_t x;
292     UVector v(2*n, status); // heuristic; adjust as needed
293 
294     if (U_FAILURE(status)) {
295         return;
296     }
297 
298     /* Precompute the index values.  This saves a LOT of time.
299      * Be careful not to call malloc(0).
300      */
301     int16_t* indexValue = (int16_t*) uprv_malloc( sizeof(int16_t) * (n > 0 ? n : 1) );
302     /* test for NULL */
303     if (indexValue == 0) {
304         status = U_MEMORY_ALLOCATION_ERROR;
305         return;
306     }
307     for (j=0; j<n; ++j) {
308         TransliterationRule* r = (TransliterationRule*) ruleVector->elementAt(j);
309         indexValue[j] = r->getIndexValue();
310     }
311     for (x=0; x<256; ++x) {
312         index[x] = v.size();
313         for (j=0; j<n; ++j) {
314             if (indexValue[j] >= 0) {
315                 if (indexValue[j] == x) {
316                     v.addElement(ruleVector->elementAt(j), status);
317                 }
318             } else {
319                 // If the indexValue is < 0, then the first key character is
320                 // a set, and we must use the more time-consuming
321                 // matchesIndexValue check.  In practice this happens
322                 // rarely, so we seldom tread this code path.
323                 TransliterationRule* r = (TransliterationRule*) ruleVector->elementAt(j);
324                 if (r->matchesIndexValue((uint8_t)x)) {
325                     v.addElement(r, status);
326                 }
327             }
328         }
329     }
330     uprv_free(indexValue);
331     index[256] = v.size();
332 
333     /* Freeze things into an array.
334      */
335     uprv_free(rules); // Contains alias pointers
336 
337     /* You can't do malloc(0)! */
338     if (v.size() == 0) {
339         rules = NULL;
340         return;
341     }
342     rules = (TransliterationRule **)uprv_malloc(v.size() * sizeof(TransliterationRule *));
343     /* test for NULL */
344     if (rules == 0) {
345         status = U_MEMORY_ALLOCATION_ERROR;
346         return;
347     }
348     for (j=0; j<v.size(); ++j) {
349         rules[j] = (TransliterationRule*) v.elementAt(j);
350     }
351 
352     // TODO Add error reporting that indicates the rules that
353     //      are being masked.
354     //UnicodeString errors;
355 
356     /* Check for masking.  This is MUCH faster than our old check,
357      * which was each rule against each following rule, since we
358      * only have to check for masking within each bin now.  It's
359      * 256*O(n2^2) instead of O(n1^2), where n1 is the total rule
360      * count, and n2 is the per-bin rule count.  But n2<<n1, so
361      * it's a big win.
362      */
363     for (x=0; x<256; ++x) {
364         for (j=index[x]; j<index[x+1]-1; ++j) {
365             TransliterationRule* r1 = rules[j];
366             for (int32_t k=j+1; k<index[x+1]; ++k) {
367                 TransliterationRule* r2 = rules[k];
368                 if (r1->masks(*r2)) {
369 //|                 if (errors == null) {
370 //|                     errors = new StringBuffer();
371 //|                 } else {
372 //|                     errors.append("\n");
373 //|                 }
374 //|                 errors.append("Rule " + r1 + " masks " + r2);
375                     status = U_RULE_MASK_ERROR;
376                     maskingError(*r1, *r2, parseError);
377                     return;
378                 }
379             }
380         }
381     }
382 
383     //if (errors != null) {
384     //    throw new IllegalArgumentException(errors.toString());
385     //}
386 }
387 
388 /**
389  * Transliterate the given text with the given UTransPosition
390  * indices.  Return TRUE if the transliteration should continue
391  * or FALSE if it should halt (because of a U_PARTIAL_MATCH match).
392  * Note that FALSE is only ever returned if isIncremental is TRUE.
393  * @param text the text to be transliterated
394  * @param pos the position indices, which will be updated
395  * @param incremental if TRUE, assume new text may be inserted
396  * at index.limit, and return FALSE if thre is a partial match.
397  * @return TRUE unless a U_PARTIAL_MATCH has been obtained,
398  * indicating that transliteration should stop until more text
399  * arrives.
400  */
transliterate(Replaceable & text,UTransPosition & pos,UBool incremental)401 UBool TransliterationRuleSet::transliterate(Replaceable& text,
402                                             UTransPosition& pos,
403                                             UBool incremental) {
404     int16_t indexByte = (int16_t) (text.char32At(pos.start) & 0xFF);
405     for (int32_t i=index[indexByte]; i<index[indexByte+1]; ++i) {
406         UMatchDegree m = rules[i]->matchAndReplace(text, pos, incremental);
407         switch (m) {
408         case U_MATCH:
409             _debugOut("match", rules[i], text, pos);
410             return TRUE;
411         case U_PARTIAL_MATCH:
412             _debugOut("partial match", rules[i], text, pos);
413             return FALSE;
414         default: /* Ram: added default to make GCC happy */
415             break;
416         }
417     }
418     // No match or partial match from any rule
419     pos.start += UTF_CHAR_LENGTH(text.char32At(pos.start));
420     _debugOut("no match", NULL, text, pos);
421     return TRUE;
422 }
423 
424 /**
425  * Create rule strings that represents this rule set.
426  */
toRules(UnicodeString & ruleSource,UBool escapeUnprintable) const427 UnicodeString& TransliterationRuleSet::toRules(UnicodeString& ruleSource,
428                                                UBool escapeUnprintable) const {
429     int32_t i;
430     int32_t count = ruleVector->size();
431     ruleSource.truncate(0);
432     for (i=0; i<count; ++i) {
433         if (i != 0) {
434             ruleSource.append((UChar) 0x000A /*\n*/);
435         }
436         TransliterationRule *r =
437             (TransliterationRule*) ruleVector->elementAt(i);
438         r->toRule(ruleSource, escapeUnprintable);
439     }
440     return ruleSource;
441 }
442 
443 /**
444  * Return the set of all characters that may be modified
445  * (getTarget=false) or emitted (getTarget=true) by this set.
446  */
getSourceTargetSet(UnicodeSet & result,UBool getTarget) const447 UnicodeSet& TransliterationRuleSet::getSourceTargetSet(UnicodeSet& result,
448                                UBool getTarget) const
449 {
450     result.clear();
451     int32_t count = ruleVector->size();
452     for (int32_t i=0; i<count; ++i) {
453         TransliterationRule* r =
454             (TransliterationRule*) ruleVector->elementAt(i);
455         if (getTarget) {
456             r->addTargetSetTo(result);
457         } else {
458             r->addSourceSetTo(result);
459         }
460     }
461     return result;
462 }
463 
464 U_NAMESPACE_END
465 
466 #endif /* #if !UCONFIG_NO_TRANSLITERATION */
467