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

/*
**********************************************************************
*   Copyright (c) 2002-2012, 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 "unicode/uniset.h"
#include "unicode/utf16.h"
#include "strrepl.h"
#include "rbt_data.h"
#include "util.h"

U_NAMESPACE_BEGIN

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 const_cast<StringReplacer *>(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 = U16_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 += U16_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, UnicodeString());

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

    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 -= U16_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 += U16_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+=U16_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 += U16_LENGTH(c);
    }
}

U_NAMESPACE_END

#endif /* #if !UCONFIG_NO_TRANSLITERATION */

//eof
Commit	Line	Data
b75a7d8f A	1	/*
b75a7d8f A	2	**********************************************************************
51004dcb	3	* Copyright (c) 2002-2012, 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
4388f060 A	15	#include "unicode/uniset.h"
4388f060 A	16	#include "unicode/utf16.h"
b75a7d8f A	17	#include "strrepl.h"
	18	#include "rbt_data.h"
	19	#include "util.h"
b75a7d8f A	20
	21	U_NAMESPACE_BEGIN
	22
374ca955 A	23	UnicodeReplacer::~UnicodeReplacer() {}
374ca955 A	24	UOBJECT_DEFINE_RTTI_IMPLEMENTATION(StringReplacer)
b75a7d8f A	25
	26	/**
	27	* Construct a StringReplacer that sets the emits the given output
	28	* text and sets the cursor to the given position.
	29	* @param theOutput text that will replace input text when the
	30	* replace() method is called. May contain stand-in characters
	31	* that represent nested replacers.
	32	* @param theCursorPos cursor position that will be returned by
	33	* the replace() method
	34	* @param theData transliterator context object that translates
	35	* stand-in characters to UnicodeReplacer objects
	36	*/
	37	StringReplacer::StringReplacer(const UnicodeString& theOutput,
	38	int32_t theCursorPos,
	39	const TransliterationRuleData* theData) {
	40	output = theOutput;
	41	cursorPos = theCursorPos;
	42	hasCursor = TRUE;
	43	data = theData;
	44	isComplex = TRUE;
	45	}
	46
	47	/**
	48	* Construct a StringReplacer that sets the emits the given output
	49	* text and does not modify the cursor.
	50	* @param theOutput text that will replace input text when the
	51	* replace() method is called. May contain stand-in characters
	52	* that represent nested replacers.
	53	* @param theData transliterator context object that translates
	54	* stand-in characters to UnicodeReplacer objects
	55	*/
	56	StringReplacer::StringReplacer(const UnicodeString& theOutput,
	57	const TransliterationRuleData* theData) {
	58	output = theOutput;
	59	cursorPos = 0;
	60	hasCursor = FALSE;
	61	data = theData;
	62	isComplex = TRUE;
	63	}
	64
	65	/**
	66	* Copy constructor.
	67	*/
374ca955 A	68	StringReplacer::StringReplacer(const StringReplacer& other) :
	69	UnicodeFunctor(other),
	70	UnicodeReplacer(other)
	71	{
b75a7d8f A	72	output = other.output;
	73	cursorPos = other.cursorPos;
	74	hasCursor = other.hasCursor;
	75	data = other.data;
	76	isComplex = other.isComplex;
	77	}
	78
	79	/**
	80	* Destructor
	81	*/
	82	StringReplacer::~StringReplacer() {
	83	}
	84
	85	/**
	86	* Implement UnicodeFunctor
	87	*/
	88	UnicodeFunctor* StringReplacer::clone() const {
	89	return new StringReplacer(*this);
	90	}
	91
	92	/**
	93	* Implement UnicodeFunctor
	94	*/
	95	UnicodeReplacer* StringReplacer::toReplacer() const {
51004dcb	96	return const_cast<StringReplacer *>(this);
b75a7d8f A	97	}
	98
	99	/**
	100	* UnicodeReplacer API
	101	*/
	102	int32_t StringReplacer::replace(Replaceable& text,
	103	int32_t start,
	104	int32_t limit,
	105	int32_t& cursor) {
	106	int32_t outLen;
	107	int32_t newStart = 0;
	108
	109	// NOTE: It should be possible to _always_ run the complex
	110	// processing code; just slower. If not, then there is a bug
	111	// in the complex processing code.
	112
	113	// Simple (no nested replacers) Processing Code :
	114	if (!isComplex) {
	115	text.handleReplaceBetween(start, limit, output);
	116	outLen = output.length();
	117
	118	// Setup default cursor position (for cursorPos within output)
	119	newStart = cursorPos;
	120	}
	121
	122	// Complex (nested replacers) Processing Code :
	123	else {
	124	/* When there are segments to be copied, use the Replaceable.copy()
	125	* API in order to retain out-of-band data. Copy everything to the
	126	* end of the string, then copy them back over the key. This preserves
	127	* the integrity of indices into the key and surrounding context while
	128	* generating the output text.
	129	*/
	130	UnicodeString buf;
	131	int32_t oOutput; // offset into 'output'
	132	isComplex = FALSE;
	133
	134	// The temporary buffer starts at tempStart, and extends
	135	// to destLimit. The start of the buffer has a single
	136	// character from before the key. This provides style
	137	// data when addition characters are filled into the
	138	// temporary buffer. If there is nothing to the left, use
	139	// the non-character U+FFFF, which Replaceable subclasses
	140	// should treat specially as a "no-style character."
	141	// destStart points to the point after the style context
	142	// character, so it is tempStart+1 or tempStart+2.
	143	int32_t tempStart = text.length(); // start of temp buffer
	144	int32_t destStart = tempStart; // copy new text to here
	145	if (start > 0) {
4388f060	146	int32_t len = U16_LENGTH(text.char32At(start-1));
b75a7d8f A	147	text.copy(start-len, start, tempStart);
	148	destStart += len;
	149	} else {
	150	UnicodeString str((UChar) 0xFFFF);
	151	text.handleReplaceBetween(tempStart, tempStart, str);
	152	destStart++;
	153	}
	154	int32_t destLimit = destStart;
	155
	156	for (oOutput=0; oOutput<output.length(); ) {
	157	if (oOutput == cursorPos) {
	158	// Record the position of the cursor
	159	newStart = destLimit - destStart; // relative to start
	160	}
	161	UChar32 c = output.char32At(oOutput);
	162	UnicodeReplacer* r = data->lookupReplacer(c);
	163	if (r == NULL) {
	164	// Accumulate straight (non-segment) text.
	165	buf.append(c);
	166	} else {
	167	isComplex = TRUE;
	168
	169	// Insert any accumulated straight text.
	170	if (buf.length() > 0) {
	171	text.handleReplaceBetween(destLimit, destLimit, buf);
	172	destLimit += buf.length();
	173	buf.truncate(0);
	174	}
	175
	176	// Delegate output generation to replacer object
	177	int32_t len = r->replace(text, destLimit, destLimit, cursor);
	178	destLimit += len;
	179	}
4388f060	180	oOutput += U16_LENGTH(c);
b75a7d8f A	181	}
	182	// Insert any accumulated straight text.
	183	if (buf.length() > 0) {
	184	text.handleReplaceBetween(destLimit, destLimit, buf);
	185	destLimit += buf.length();
	186	}
	187	if (oOutput == cursorPos) {
	188	// Record the position of the cursor
	189	newStart = destLimit - destStart; // relative to start
	190	}
	191
	192	outLen = destLimit - destStart;
	193
	194	// Copy new text to start, and delete it
	195	text.copy(destStart, destLimit, start);
4388f060	196	text.handleReplaceBetween(tempStart + outLen, destLimit + outLen, UnicodeString());
b75a7d8f A	197
b75a7d8f A	198	// Delete the old text (the key)
4388f060	199	text.handleReplaceBetween(start + outLen, limit + outLen, UnicodeString());
b75a7d8f A	200	}
	201
	202	if (hasCursor) {
	203	// Adjust the cursor for positions outside the key. These
	204	// refer to code points rather than code units. If cursorPos
	205	// is within the output string, then use newStart, which has
	206	// already been set above.
	207	if (cursorPos < 0) {
	208	newStart = start;
	209	int32_t n = cursorPos;
	210	// Outside the output string, cursorPos counts code points
	211	while (n < 0 && newStart > 0) {
4388f060	212	newStart -= U16_LENGTH(text.char32At(newStart-1));
b75a7d8f A	213	++n;
	214	}
	215	newStart += n;
	216	} else if (cursorPos > output.length()) {
	217	newStart = start + outLen;
	218	int32_t n = cursorPos - output.length();
	219	// Outside the output string, cursorPos counts code points
	220	while (n > 0 && newStart < text.length()) {
4388f060	221	newStart += U16_LENGTH(text.char32At(newStart));
b75a7d8f A	222	--n;
	223	}
	224	newStart += n;
	225	} else {
	226	// Cursor is within output string. It has been set up above
	227	// to be relative to start.
	228	newStart += start;
	229	}
	230
	231	cursor = newStart;
	232	}
	233
	234	return outLen;
	235	}
	236
	237	/**
	238	* UnicodeReplacer API
	239	*/
	240	UnicodeString& StringReplacer::toReplacerPattern(UnicodeString& rule,
	241	UBool escapeUnprintable) const {
	242	rule.truncate(0);
	243	UnicodeString quoteBuf;
	244
	245	int32_t cursor = cursorPos;
	246
	247	// Handle a cursor preceding the output
	248	if (hasCursor && cursor < 0) {
	249	while (cursor++ < 0) {
	250	ICU_Utility::appendToRule(rule, (UChar)0x0040 /@/, TRUE, escapeUnprintable, quoteBuf);
	251	}
	252	// Fall through and append '\|' below
	253	}
	254
	255	for (int32_t i=0; i<output.length(); ++i) {
	256	if (hasCursor && i == cursor) {
	257	ICU_Utility::appendToRule(rule, (UChar)0x007C /\|/, TRUE, escapeUnprintable, quoteBuf);
	258	}
	259	UChar c = output.charAt(i); // Ok to use 16-bits here
	260
	261	UnicodeReplacer* r = data->lookupReplacer(c);
	262	if (r == NULL) {
	263	ICU_Utility::appendToRule(rule, c, FALSE, escapeUnprintable, quoteBuf);
	264	} else {
	265	UnicodeString buf;
	266	r->toReplacerPattern(buf, escapeUnprintable);
	267	buf.insert(0, (UChar)0x20);
	268	buf.append((UChar)0x20);
	269	ICU_Utility::appendToRule(rule, buf,
	270	TRUE, escapeUnprintable, quoteBuf);
	271	}
	272	}
	273
	274	// Handle a cursor after the output. Use > rather than >= because
	275	// if cursor == output.length() it is at the end of the output,
	276	// which is the default position, so we need not emit it.
	277	if (hasCursor && cursor > output.length()) {
	278	cursor -= output.length();
	279	while (cursor-- > 0) {
	280	ICU_Utility::appendToRule(rule, (UChar)0x0040 /@/, TRUE, escapeUnprintable, quoteBuf);
	281	}
	282	ICU_Utility::appendToRule(rule, (UChar)0x007C /\|/, TRUE, escapeUnprintable, quoteBuf);
	283	}
	284	// Flush quoteBuf out to result
	285	ICU_Utility::appendToRule(rule, -1,
286	TRUE, escapeUnprintable, quoteBuf);
287
288	return rule;
289	}
290
291	/**
292	* Implement UnicodeReplacer
293	*/
294	void StringReplacer::addReplacementSetTo(UnicodeSet& toUnionTo) const {
295	UChar32 ch;
4388f060	296	for (int32_t i=0; i<output.length(); i+=U16_LENGTH(ch)) {
374ca955 A	297	ch = output.char32At(i);
	298	UnicodeReplacer* r = data->lookupReplacer(ch);
	299	if (r == NULL) {
	300	toUnionTo.add(ch);
	301	} else {
	302	r->addReplacementSetTo(toUnionTo);
	303	}
b75a7d8f A	304	}
	305	}
	306
	307	/**
	308	* UnicodeFunctor API
	309	*/
	310	void StringReplacer::setData(const TransliterationRuleData* d) {
	311	data = d;
	312	int32_t i = 0;
	313	while (i<output.length()) {
	314	UChar32 c = output.char32At(i);
	315	UnicodeFunctor* f = data->lookup(c);
	316	if (f != NULL) {
	317	f->setData(data);
	318	}
4388f060	319	i += U16_LENGTH(c);
b75a7d8f A	320	}
	321	}
	322
	323	U_NAMESPACE_END
	324
	325	#endif /* #if !UCONFIG_NO_TRANSLITERATION */
	326
	327	//eof