[apple/icu.git] / icuSources / i18n / strrepl.cpp

/*
**********************************************************************
*   Copyright (c) 2002-2004, International Business Machines Corporation
*   and others.  All Rights Reserved.
**********************************************************************
*   Date        Name        Description
*   01/21/2002  aliu        Creation.
**********************************************************************
*/

#include "unicode/utypes.h"

#if !UCONFIG_NO_TRANSLITERATION

#include "strrepl.h"
#include "rbt_data.h"
#include "util.h"
#include "unicode/uniset.h"

U_NAMESPACE_BEGIN

static const UChar EMPTY[] = { 0 }; // empty string: ""

UnicodeReplacer::~UnicodeReplacer() {}
UOBJECT_DEFINE_RTTI_IMPLEMENTATION(StringReplacer)

/**
 * Construct a StringReplacer that sets the emits the given output
 * text and sets the cursor to the given position.
 * @param theOutput text that will replace input text when the
 * replace() method is called.  May contain stand-in characters
 * that represent nested replacers.
 * @param theCursorPos cursor position that will be returned by
 * the replace() method
 * @param theData transliterator context object that translates
 * stand-in characters to UnicodeReplacer objects
 */
StringReplacer::StringReplacer(const UnicodeString& theOutput,
                               int32_t theCursorPos,
                               const TransliterationRuleData* theData) {
    output = theOutput;
    cursorPos = theCursorPos;
    hasCursor = TRUE;
    data = theData;
    isComplex = TRUE;
}

/**
 * Construct a StringReplacer that sets the emits the given output
 * text and does not modify the cursor.
 * @param theOutput text that will replace input text when the
 * replace() method is called.  May contain stand-in characters
 * that represent nested replacers.
 * @param theData transliterator context object that translates
 * stand-in characters to UnicodeReplacer objects
 */
StringReplacer::StringReplacer(const UnicodeString& theOutput,
                               const TransliterationRuleData* theData) {
    output = theOutput;
    cursorPos = 0;
    hasCursor = FALSE;
    data = theData;
    isComplex = TRUE;
}

/**
 * Copy constructor.
 */
StringReplacer::StringReplacer(const StringReplacer& other) :
    UnicodeFunctor(other),
    UnicodeReplacer(other)
{
    output = other.output;
    cursorPos = other.cursorPos;
    hasCursor = other.hasCursor;
    data = other.data;
    isComplex = other.isComplex;
}

/**
 * Destructor
 */
StringReplacer::~StringReplacer() {
}

/**
 * Implement UnicodeFunctor
 */
UnicodeFunctor* StringReplacer::clone() const {
    return new StringReplacer(*this);
}

/**
 * Implement UnicodeFunctor
 */
UnicodeReplacer* StringReplacer::toReplacer() const {
    return (UnicodeReplacer*) this;
}

/**
 * UnicodeReplacer API
 */
int32_t StringReplacer::replace(Replaceable& text,
                                int32_t start,
                                int32_t limit,
                                int32_t& cursor) {
    int32_t outLen;
    int32_t newStart = 0;

    // NOTE: It should be possible to _always_ run the complex
    // processing code; just slower.  If not, then there is a bug
    // in the complex processing code.

    // Simple (no nested replacers) Processing Code :
    if (!isComplex) {
        text.handleReplaceBetween(start, limit, output);
        outLen = output.length();

        // Setup default cursor position (for cursorPos within output)
        newStart = cursorPos;
    }

    // Complex (nested replacers) Processing Code :
    else {
        /* When there are segments to be copied, use the Replaceable.copy()
         * API in order to retain out-of-band data.  Copy everything to the
         * end of the string, then copy them back over the key.  This preserves
         * the integrity of indices into the key and surrounding context while
         * generating the output text.
         */
        UnicodeString buf;
        int32_t oOutput; // offset into 'output'
        isComplex = FALSE;

        // The temporary buffer starts at tempStart, and extends
        // to destLimit.  The start of the buffer has a single
        // character from before the key.  This provides style
        // data when addition characters are filled into the
        // temporary buffer.  If there is nothing to the left, use
        // the non-character U+FFFF, which Replaceable subclasses
        // should treat specially as a "no-style character."
        // destStart points to the point after the style context
        // character, so it is tempStart+1 or tempStart+2.
        int32_t tempStart = text.length(); // start of temp buffer
        int32_t destStart = tempStart; // copy new text to here
        if (start > 0) {
            int32_t len = UTF_CHAR_LENGTH(text.char32At(start-1));
            text.copy(start-len, start, tempStart);
            destStart += len;
        } else {
            UnicodeString str((UChar) 0xFFFF);
            text.handleReplaceBetween(tempStart, tempStart, str);
            destStart++;
        }
        int32_t destLimit = destStart;

        for (oOutput=0; oOutput<output.length(); ) {
            if (oOutput == cursorPos) {
                // Record the position of the cursor
                newStart = destLimit - destStart; // relative to start
            }
            UChar32 c = output.char32At(oOutput);
            UnicodeReplacer* r = data->lookupReplacer(c);
            if (r == NULL) {
                // Accumulate straight (non-segment) text.
                buf.append(c);
            } else {
                isComplex = TRUE;

                // Insert any accumulated straight text.
                if (buf.length() > 0) {
                    text.handleReplaceBetween(destLimit, destLimit, buf);
                    destLimit += buf.length();
                    buf.truncate(0);
                }

                // Delegate output generation to replacer object
                int32_t len = r->replace(text, destLimit, destLimit, cursor);
                destLimit += len;
            }
            oOutput += UTF_CHAR_LENGTH(c);
        }
        // Insert any accumulated straight text.
        if (buf.length() > 0) {
            text.handleReplaceBetween(destLimit, destLimit, buf);
            destLimit += buf.length();
        }
        if (oOutput == cursorPos) {
            // Record the position of the cursor
            newStart = destLimit - destStart; // relative to start
        }

        outLen = destLimit - destStart;

        // Copy new text to start, and delete it
        text.copy(destStart, destLimit, start);
        text.handleReplaceBetween(tempStart + outLen, destLimit + outLen, EMPTY);

        // Delete the old text (the key)
        text.handleReplaceBetween(start + outLen, limit + outLen, EMPTY);
    }        

    if (hasCursor) {
        // Adjust the cursor for positions outside the key.  These
        // refer to code points rather than code units.  If cursorPos
        // is within the output string, then use newStart, which has
        // already been set above.
        if (cursorPos < 0) {
            newStart = start;
            int32_t n = cursorPos;
            // Outside the output string, cursorPos counts code points
            while (n < 0 && newStart > 0) {
                newStart -= UTF_CHAR_LENGTH(text.char32At(newStart-1));
                ++n;
            }
            newStart += n;
        } else if (cursorPos > output.length()) {
            newStart = start + outLen;
            int32_t n = cursorPos - output.length();
            // Outside the output string, cursorPos counts code points
            while (n > 0 && newStart < text.length()) {
                newStart += UTF_CHAR_LENGTH(text.char32At(newStart));
                --n;
            }
            newStart += n;
        } else {
            // Cursor is within output string.  It has been set up above
            // to be relative to start.
            newStart += start;
        }

        cursor = newStart;
    }

    return outLen;
}

/**
 * UnicodeReplacer API
 */
UnicodeString& StringReplacer::toReplacerPattern(UnicodeString& rule,
                                                 UBool escapeUnprintable) const {
    rule.truncate(0);
    UnicodeString quoteBuf;

    int32_t cursor = cursorPos;

    // Handle a cursor preceding the output
    if (hasCursor && cursor < 0) {
        while (cursor++ < 0) {
            ICU_Utility::appendToRule(rule, (UChar)0x0040 /*@*/, TRUE, escapeUnprintable, quoteBuf);
        }
        // Fall through and append '|' below
    }

    for (int32_t i=0; i<output.length(); ++i) {
        if (hasCursor && i == cursor) {
            ICU_Utility::appendToRule(rule, (UChar)0x007C /*|*/, TRUE, escapeUnprintable, quoteBuf);
        }
        UChar c = output.charAt(i); // Ok to use 16-bits here

        UnicodeReplacer* r = data->lookupReplacer(c);
        if (r == NULL) {
            ICU_Utility::appendToRule(rule, c, FALSE, escapeUnprintable, quoteBuf);
        } else {
            UnicodeString buf;
            r->toReplacerPattern(buf, escapeUnprintable);
            buf.insert(0, (UChar)0x20);
            buf.append((UChar)0x20);
            ICU_Utility::appendToRule(rule, buf,
                                      TRUE, escapeUnprintable, quoteBuf);
        }
    }

    // Handle a cursor after the output.  Use > rather than >= because
    // if cursor == output.length() it is at the end of the output,
    // which is the default position, so we need not emit it.
    if (hasCursor && cursor > output.length()) {
        cursor -= output.length();
        while (cursor-- > 0) {
            ICU_Utility::appendToRule(rule, (UChar)0x0040 /*@*/, TRUE, escapeUnprintable, quoteBuf);
        }
        ICU_Utility::appendToRule(rule, (UChar)0x007C /*|*/, TRUE, escapeUnprintable, quoteBuf);
    }
    // Flush quoteBuf out to result
    ICU_Utility::appendToRule(rule, -1,
                              TRUE, escapeUnprintable, quoteBuf);

    return rule;
}

/**
 * Implement UnicodeReplacer
 */
void StringReplacer::addReplacementSetTo(UnicodeSet& toUnionTo) const {
    UChar32 ch;
    for (int32_t i=0; i<output.length(); i+=UTF_CHAR_LENGTH(ch)) {
    ch = output.char32At(i);
    UnicodeReplacer* r = data->lookupReplacer(ch);
    if (r == NULL) {
        toUnionTo.add(ch);
    } else {
        r->addReplacementSetTo(toUnionTo);
    }
    }
}

/**
 * UnicodeFunctor API
 */
void StringReplacer::setData(const TransliterationRuleData* d) {
    data = d;
    int32_t i = 0;
    while (i<output.length()) {
        UChar32 c = output.char32At(i);
        UnicodeFunctor* f = data->lookup(c);
        if (f != NULL) {
            f->setData(data);
        }
        i += UTF_CHAR_LENGTH(c);
    }
}

U_NAMESPACE_END

#endif /* #if !UCONFIG_NO_TRANSLITERATION */

//eof
Commit	Line	Data
b75a7d8f A	1	/*
b75a7d8f A	2	**********************************************************************
374ca955	3	* Copyright (c) 2002-2004, International Business Machines Corporation
b75a7d8f A	4	* and others. All Rights Reserved.
	5	**********************************************************************
	6	* Date Name Description
	7	* 01/21/2002 aliu Creation.
	8	**********************************************************************
	9	*/
	10
	11	#include "unicode/utypes.h"
	12
	13	#if !UCONFIG_NO_TRANSLITERATION
	14
	15	#include "strrepl.h"
	16	#include "rbt_data.h"
	17	#include "util.h"
	18	#include "unicode/uniset.h"
	19
	20	U_NAMESPACE_BEGIN
	21
374ca955	22	static const UChar EMPTY[] = { 0 }; // empty string: ""
b75a7d8f	23
374ca955 A	24	UnicodeReplacer::~UnicodeReplacer() {}
374ca955 A	25	UOBJECT_DEFINE_RTTI_IMPLEMENTATION(StringReplacer)
b75a7d8f A	26
	27	/**
	28	* Construct a StringReplacer that sets the emits the given output
	29	* text and sets the cursor to the given position.
	30	* @param theOutput text that will replace input text when the
	31	* replace() method is called. May contain stand-in characters
	32	* that represent nested replacers.
	33	* @param theCursorPos cursor position that will be returned by
	34	* the replace() method
	35	* @param theData transliterator context object that translates
	36	* stand-in characters to UnicodeReplacer objects
	37	*/
	38	StringReplacer::StringReplacer(const UnicodeString& theOutput,
	39	int32_t theCursorPos,
	40	const TransliterationRuleData* theData) {
	41	output = theOutput;
	42	cursorPos = theCursorPos;
	43	hasCursor = TRUE;
	44	data = theData;
	45	isComplex = TRUE;
	46	}
	47
	48	/**
	49	* Construct a StringReplacer that sets the emits the given output
	50	* text and does not modify the cursor.
	51	* @param theOutput text that will replace input text when the
	52	* replace() method is called. May contain stand-in characters
	53	* that represent nested replacers.
	54	* @param theData transliterator context object that translates
	55	* stand-in characters to UnicodeReplacer objects
	56	*/
	57	StringReplacer::StringReplacer(const UnicodeString& theOutput,
	58	const TransliterationRuleData* theData) {
	59	output = theOutput;
	60	cursorPos = 0;
	61	hasCursor = FALSE;
	62	data = theData;
	63	isComplex = TRUE;
	64	}
	65
	66	/**
	67	* Copy constructor.
	68	*/
374ca955 A	69	StringReplacer::StringReplacer(const StringReplacer& other) :
	70	UnicodeFunctor(other),
	71	UnicodeReplacer(other)
	72	{
b75a7d8f A	73	output = other.output;
	74	cursorPos = other.cursorPos;
	75	hasCursor = other.hasCursor;
	76	data = other.data;
	77	isComplex = other.isComplex;
	78	}
	79
	80	/**
	81	* Destructor
	82	*/
	83	StringReplacer::~StringReplacer() {
	84	}
	85
	86	/**
	87	* Implement UnicodeFunctor
	88	*/
	89	UnicodeFunctor* StringReplacer::clone() const {
	90	return new StringReplacer(*this);
	91	}
	92
	93	/**
	94	* Implement UnicodeFunctor
	95	*/
	96	UnicodeReplacer* StringReplacer::toReplacer() const {
	97	return (UnicodeReplacer*) this;
	98	}
	99
	100	/**
	101	* UnicodeReplacer API
	102	*/
	103	int32_t StringReplacer::replace(Replaceable& text,
	104	int32_t start,
	105	int32_t limit,
	106	int32_t& cursor) {
	107	int32_t outLen;
	108	int32_t newStart = 0;
	109
	110	// NOTE: It should be possible to _always_ run the complex
	111	// processing code; just slower. If not, then there is a bug
	112	// in the complex processing code.
	113
	114	// Simple (no nested replacers) Processing Code :
	115	if (!isComplex) {
	116	text.handleReplaceBetween(start, limit, output);
	117	outLen = output.length();
	118
	119	// Setup default cursor position (for cursorPos within output)
	120	newStart = cursorPos;
	121	}
	122
	123	// Complex (nested replacers) Processing Code :
	124	else {
	125	/* When there are segments to be copied, use the Replaceable.copy()
	126	* API in order to retain out-of-band data. Copy everything to the
	127	* end of the string, then copy them back over the key. This preserves
	128	* the integrity of indices into the key and surrounding context while
	129	* generating the output text.
	130	*/
	131	UnicodeString buf;
	132	int32_t oOutput; // offset into 'output'
	133	isComplex = FALSE;
	134
	135	// The temporary buffer starts at tempStart, and extends
	136	// to destLimit. The start of the buffer has a single
137	// character from before the key. This provides style
138	// data when addition characters are filled into the
139	// temporary buffer. If there is nothing to the left, use
140	// the non-character U+FFFF, which Replaceable subclasses
141	// should treat specially as a "no-style character."
142	// destStart points to the point after the style context
143	// character, so it is tempStart+1 or tempStart+2.
144	int32_t tempStart = text.length(); // start of temp buffer
145	int32_t destStart = tempStart; // copy new text to here
146	if (start > 0) {
147	int32_t len = UTF_CHAR_LENGTH(text.char32At(start-1));
148	text.copy(start-len, start, tempStart);
149	destStart += len;
150	} else {
151	UnicodeString str((UChar) 0xFFFF);
152	text.handleReplaceBetween(tempStart, tempStart, str);
153	destStart++;
154	}
155	int32_t destLimit = destStart;
156
157	for (oOutput=0; oOutput<output.length(); ) {
158	if (oOutput == cursorPos) {
159	// Record the position of the cursor
160	newStart = destLimit - destStart; // relative to start
161	}
162	UChar32 c = output.char32At(oOutput);
163	UnicodeReplacer* r = data->lookupReplacer(c);
164	if (r == NULL) {
165	// Accumulate straight (non-segment) text.
166	buf.append(c);
167	} else {
168	isComplex = TRUE;
169
170	// Insert any accumulated straight text.
171	if (buf.length() > 0) {
172	text.handleReplaceBetween(destLimit, destLimit, buf);
173	destLimit += buf.length();
174	buf.truncate(0);
175	}
176
177	// Delegate output generation to replacer object
178	int32_t len = r->replace(text, destLimit, destLimit, cursor);
179	destLimit += len;
180	}
181	oOutput += UTF_CHAR_LENGTH(c);
182	}
183	// Insert any accumulated straight text.
184	if (buf.length() > 0) {
185	text.handleReplaceBetween(destLimit, destLimit, buf);
186	destLimit += buf.length();
187	}
188	if (oOutput == cursorPos) {
189	// Record the position of the cursor
190	newStart = destLimit - destStart; // relative to start
191	}
192
193	outLen = destLimit - destStart;
194
195	// Copy new text to start, and delete it
196	text.copy(destStart, destLimit, start);
197	text.handleReplaceBetween(tempStart + outLen, destLimit + outLen, EMPTY);
198
199	// Delete the old text (the key)
200	text.handleReplaceBetween(start + outLen, limit + outLen, EMPTY);
201	}
202
203	if (hasCursor) {
204	// Adjust the cursor for positions outside the key. These
205	// refer to code points rather than code units. If cursorPos
206	// is within the output string, then use newStart, which has
207	// already been set above.
208	if (cursorPos < 0) {
209	newStart = start;
210	int32_t n = cursorPos;
211	// Outside the output string, cursorPos counts code points
212	while (n < 0 && newStart > 0) {
213	newStart -= UTF_CHAR_LENGTH(text.char32At(newStart-1));
214	++n;
215	}
216	newStart += n;
217	} else if (cursorPos > output.length()) {
218	newStart = start + outLen;
219	int32_t n = cursorPos - output.length();
220	// Outside the output string, cursorPos counts code points
221	while (n > 0 && newStart < text.length()) {
222	newStart += UTF_CHAR_LENGTH(text.char32At(newStart));
223	--n;
224	}
225	newStart += n;
226	} else {
227	// Cursor is within output string. It has been set up above
228	// to be relative to start.
229	newStart += start;
230	}
231
232	cursor = newStart;
233	}
234
235	return outLen;
236	}
237
238	/**
239	* UnicodeReplacer API
240	*/
241	UnicodeString& StringReplacer::toReplacerPattern(UnicodeString& rule,
242	UBool escapeUnprintable) const {
243	rule.truncate(0);
244	UnicodeString quoteBuf;
245
246	int32_t cursor = cursorPos;
247
248	// Handle a cursor preceding the output
249	if (hasCursor && cursor < 0) {
250	while (cursor++ < 0) {
251	ICU_Utility::appendToRule(rule, (UChar)0x0040 /@/, TRUE, escapeUnprintable, quoteBuf);
252	}
253	// Fall through and append '\|' below
254	}
255
256	for (int32_t i=0; i<output.length(); ++i) {
257	if (hasCursor && i == cursor) {
258	ICU_Utility::appendToRule(rule, (UChar)0x007C /\|/, TRUE, escapeUnprintable, quoteBuf);
259	}
260	UChar c = output.charAt(i); // Ok to use 16-bits here
261
262	UnicodeReplacer* r = data->lookupReplacer(c);
263	if (r == NULL) {
264	ICU_Utility::appendToRule(rule, c, FALSE, escapeUnprintable, quoteBuf);
265	} else {
266	UnicodeString buf;
267	r->toReplacerPattern(buf, escapeUnprintable);
268	buf.insert(0, (UChar)0x20);
269	buf.append((UChar)0x20);
270	ICU_Utility::appendToRule(rule, buf,
271	TRUE, escapeUnprintable, quoteBuf);
272	}
273	}
274
275	// Handle a cursor after the output. Use > rather than >= because
276	// if cursor == output.length() it is at the end of the output,
277	// which is the default position, so we need not emit it.
278	if (hasCursor && cursor > output.length()) {
279	cursor -= output.length();
280	while (cursor-- > 0) {
281	ICU_Utility::appendToRule(rule, (UChar)0x0040 /@/, TRUE, escapeUnprintable, quoteBuf);
282	}
283	ICU_Utility::appendToRule(rule, (UChar)0x007C /\|/, TRUE, escapeUnprintable, quoteBuf);
284	}
285	// Flush quoteBuf out to result
286	ICU_Utility::appendToRule(rule, -1,
287	TRUE, escapeUnprintable, quoteBuf);
288
289	return rule;
290	}
291
292	/**
293	* Implement UnicodeReplacer
294	*/
295	void StringReplacer::addReplacementSetTo(UnicodeSet& toUnionTo) const {
296	UChar32 ch;
297	for (int32_t i=0; i<output.length(); i+=UTF_CHAR_LENGTH(ch)) {
374ca955 A	298	ch = output.char32At(i);
	299	UnicodeReplacer* r = data->lookupReplacer(ch);
	300	if (r == NULL) {
	301	toUnionTo.add(ch);
	302	} else {
	303	r->addReplacementSetTo(toUnionTo);
	304	}
b75a7d8f A	305	}
	306	}
	307
	308	/**
	309	* UnicodeFunctor API
	310	*/
	311	void StringReplacer::setData(const TransliterationRuleData* d) {
	312	data = d;
	313	int32_t i = 0;
	314	while (i<output.length()) {
	315	UChar32 c = output.char32At(i);
	316	UnicodeFunctor* f = data->lookup(c);
	317	if (f != NULL) {
	318	f->setData(data);
	319	}
	320	i += UTF_CHAR_LENGTH(c);
	321	}
	322	}
	323
	324	U_NAMESPACE_END
	325
	326	#endif /* #if !UCONFIG_NO_TRANSLITERATION */
	327
	328	//eof