sl@0: /*
sl@0: **********************************************************************
sl@0: *   Copyright (C) 1999-2002, International Business Machines
sl@0: *   Corporation and others.  All Rights Reserved.
sl@0: **********************************************************************
sl@0: *
sl@0: * File USC_IMPL.H
sl@0: *
sl@0: * Modification History:
sl@0: *
sl@0: *   Date        Name        Description
sl@0: *   07/08/2002  Eric Mader  Creation.
sl@0: ******************************************************************************
sl@0: */
sl@0: 
sl@0: #ifndef USC_IMPL_H
sl@0: #define USC_IMPL_H
sl@0: #include "unicode/utypes.h"
sl@0: #include "unicode/uscript.h"
sl@0: 
sl@0: /**
sl@0:  * <code>UScriptRun</code> is used to find runs of characters in
sl@0:  * the same script. It implements a simple iterator over an array
sl@0:  * of characters. The iterator will resolve script-neutral characters
sl@0:  * like punctuation into the script of the surrounding characters.
sl@0:  *
sl@0:  * The iterator will try to match paired punctuation. If it sees an
sl@0:  * opening punctuation character, it will remember the script that
sl@0:  * was assigned to that character, and assign the same script to the
sl@0:  * matching closing punctuation.
sl@0:  *
sl@0:  * Scripts are chosen based on the <code>UScriptCode</code> enumeration.
sl@0:  * No attempt is made to combine related scripts into a single run. In
sl@0:  * particular, Hiragana, Katakana, and Han characters will appear in seperate
sl@0:  * runs.
sl@0: 
sl@0:  * Here is an example of how to iterate over script runs:
sl@0:  * <pre>
sl@0:  * \code
sl@0:  * void printScriptRuns(const UChar *text, int32_t length)
sl@0:  * {
sl@0:  *     UErrorCode error = U_ZERO_ERROR;
sl@0:  *     UScriptRun *scriptRun = uscript_openRun(text, testLength, &error);
sl@0:  *     int32_t start = 0, limit = 0;
sl@0:  *     UScriptCode code = USCRIPT_INVALID_CODE;
sl@0:  *
sl@0:  *     while (uscript_nextRun(&start, &limit, &code)) {
sl@0:  *         printf("Script '%s' from %d to %d.\n", uscript_getName(code), start, limit);
sl@0:  *     }
sl@0:  *
sl@0:  *     uscript_closeRun(scriptRun);
sl@0:  *  }
sl@0:  * </pre>
sl@0:  *
sl@0:  * @draft ICU 2.2
sl@0:  */
sl@0: struct UScriptRun;
sl@0: 
sl@0: typedef struct UScriptRun UScriptRun;
sl@0: 
sl@0: /**
sl@0:  * Create a <code>UScriptRun</code> object for iterating over the given text. This object must
sl@0:  * be freed using <code>uscript_closeRun()</code>. Note that this object does not copy the source text,
sl@0:  * only the pointer to it. You must make sure that the pointer remains valid until you call
sl@0:  * <code>uscript_closeRun()</code> or <code>uscript_setRunText()</code>.
sl@0:  *
sl@0:  * @param src is the address of the array of characters over which to iterate.
sl@0:  *        if <code>src == NULL</code> and <code>length == 0</code>,
sl@0:  *        an empty <code>UScriptRun</code> object will be returned.
sl@0:  *
sl@0:  * @param length is the number of characters over which to iterate.
sl@0:  *
sl@0:  * @param pErrorCode is a pointer to a valid <code>UErrorCode</code> value. If this value
sl@0:  *        indicates a failure on entry, the function will immediately return.
sl@0:  *        On exit the value will indicate the success of the operation.
sl@0:  *
sl@0:  * @return the address of <code>UScriptRun</code> object which will iterate over the text,
sl@0:  *         or <code>NULL</code> if the operation failed.
sl@0:  *
sl@0:  * @draft ICU 2.2
sl@0:  */
sl@0: U_CAPI UScriptRun * U_EXPORT2
sl@0: uscript_openRun(const UChar *src, int32_t length, UErrorCode *pErrorCode);
sl@0: 
sl@0: /**
sl@0:  * Frees the given <code>UScriptRun</code> object and any storage associated with it.
sl@0:  * On return, scriptRun no longer points to a valid <code>UScriptRun</code> object.
sl@0:  *
sl@0:  * @param scriptRun is the <code>UScriptRun</code> object which will be freed.
sl@0:  *
sl@0:  * @draft ICU 2.2
sl@0:  */
sl@0: U_CAPI void U_EXPORT2
sl@0: uscript_closeRun(UScriptRun *scriptRun);
sl@0: 
sl@0: /**
sl@0:  * Reset the <code>UScriptRun</code> object so that it will start iterating from
sl@0:  * the beginning.
sl@0:  *
sl@0:  * @param scriptRun is the address of the <code>UScriptRun</code> object to be reset.
sl@0:  *
sl@0:  * @draft ICU 2.2
sl@0:  */
sl@0: U_CAPI void U_EXPORT2
sl@0: uscript_resetRun(UScriptRun *scriptRun);
sl@0: 
sl@0: /**
sl@0:  * Change the text over which the given <code>UScriptRun</code> object iterates.
sl@0:  *
sl@0:  * @param scriptRun is the <code>UScriptRun</code> object which will be changed.
sl@0:  *
sl@0:  * @param src is the address of the new array of characters over which to iterate.
sl@0:  *        If <code>src == NULL</code> and <code>length == 0</code>,
sl@0:  *        the <code>UScriptRun</code> object will become empty.
sl@0:  *
sl@0:  * @param length is the new number of characters over which to iterate
sl@0:  *
sl@0:  * @param pErrorCode is a pointer to a valid <code>UErrorCode</code> value. If this value
sl@0:  *        indicates a failure on entry, the function will immediately return.
sl@0:  *        On exit the value will indicate the success of the operation.
sl@0:  *
sl@0:  * @draft ICU 2.2
sl@0:  */
sl@0: U_CAPI void U_EXPORT2
sl@0: uscript_setRunText(UScriptRun *scriptRun, const UChar *src, int32_t length, UErrorCode *pErrorCode);
sl@0: 
sl@0: /**
sl@0:  * Advance the <code>UScriptRun</code> object to the next script run, return the start and limit
sl@0:  * offsets, and the script of the run.
sl@0:  *
sl@0:  * @param scriptRun is the address of the <code>UScriptRun</code> object.
sl@0:  *
sl@0:  * @param pRunStart is a pointer to the variable to receive the starting offset of the next run.
sl@0:  *        This pointer can be <code>NULL</code> if the value is not needed.
sl@0:  *
sl@0:  * @param pRunLimit is a pointer to the variable to receive the limit offset of the next run.
sl@0:  *        This pointer can be <code>NULL</code> if the value is not needed.
sl@0:  *
sl@0:  * @param pRunScript is a pointer to the variable to receive the UScriptCode for the
sl@0:  *        script of the current run. This pointer can be <code>NULL</code> if the value is not needed.
sl@0:  *
sl@0:  * @return true if there was another script run.
sl@0:  *
sl@0:  * @draft ICU 2.2
sl@0:  */
sl@0: U_CAPI UBool U_EXPORT2
sl@0: uscript_nextRun(UScriptRun *scriptRun, int32_t *pRunStart, int32_t *pRunLimit, UScriptCode *pRunScript);
sl@0: 
sl@0: #endif