michael@0: /*
michael@0: **********************************************************************
michael@0: *   Copyright (C) 1999-2011, International Business Machines
michael@0: *   Corporation and others.  All Rights Reserved.
michael@0: **********************************************************************
michael@0: *
michael@0: * File USC_IMPL.H
michael@0: *
michael@0: * Modification History:
michael@0: *
michael@0: *   Date        Name        Description
michael@0: *   07/08/2002  Eric Mader  Creation.
michael@0: ******************************************************************************
michael@0: */
michael@0: 
michael@0: #ifndef USC_IMPL_H
michael@0: #define USC_IMPL_H
michael@0: #include "unicode/utypes.h"
michael@0: #include "unicode/uscript.h"
michael@0: 
michael@0: /**
michael@0:  * <code>UScriptRun</code> is used to find runs of characters in
michael@0:  * the same script. It implements a simple iterator over an array
michael@0:  * of characters. The iterator will resolve script-neutral characters
michael@0:  * like punctuation into the script of the surrounding characters.
michael@0:  *
michael@0:  * The iterator will try to match paired punctuation. If it sees an
michael@0:  * opening punctuation character, it will remember the script that
michael@0:  * was assigned to that character, and assign the same script to the
michael@0:  * matching closing punctuation.
michael@0:  *
michael@0:  * Scripts are chosen based on the <code>UScriptCode</code> enumeration.
michael@0:  * No attempt is made to combine related scripts into a single run. In
michael@0:  * particular, Hiragana, Katakana, and Han characters will appear in seperate
michael@0:  * runs.
michael@0: 
michael@0:  * Here is an example of how to iterate over script runs:
michael@0:  * <pre>
michael@0:  * \code
michael@0:  * void printScriptRuns(const UChar *text, int32_t length)
michael@0:  * {
michael@0:  *     UErrorCode error = U_ZERO_ERROR;
michael@0:  *     UScriptRun *scriptRun = uscript_openRun(text, testLength, &error);
michael@0:  *     int32_t start = 0, limit = 0;
michael@0:  *     UScriptCode code = USCRIPT_INVALID_CODE;
michael@0:  *
michael@0:  *     while (uscript_nextRun(&start, &limit, &code)) {
michael@0:  *         printf("Script '%s' from %d to %d.\n", uscript_getName(code), start, limit);
michael@0:  *     }
michael@0:  *
michael@0:  *     uscript_closeRun(scriptRun);
michael@0:  *  }
michael@0:  * </pre>
michael@0:  */
michael@0: struct UScriptRun;
michael@0: 
michael@0: typedef struct UScriptRun UScriptRun;
michael@0: 
michael@0: /**
michael@0:  * Create a <code>UScriptRun</code> object for iterating over the given text. This object must
michael@0:  * be freed using <code>uscript_closeRun()</code>. Note that this object does not copy the source text,
michael@0:  * only the pointer to it. You must make sure that the pointer remains valid until you call
michael@0:  * <code>uscript_closeRun()</code> or <code>uscript_setRunText()</code>.
michael@0:  *
michael@0:  * @param src is the address of the array of characters over which to iterate.
michael@0:  *        if <code>src == NULL</code> and <code>length == 0</code>,
michael@0:  *        an empty <code>UScriptRun</code> object will be returned.
michael@0:  *
michael@0:  * @param length is the number of characters over which to iterate.
michael@0:  *
michael@0:  * @param pErrorCode is a pointer to a valid <code>UErrorCode</code> value. If this value
michael@0:  *        indicates a failure on entry, the function will immediately return.
michael@0:  *        On exit the value will indicate the success of the operation.
michael@0:  *
michael@0:  * @return the address of <code>UScriptRun</code> object which will iterate over the text,
michael@0:  *         or <code>NULL</code> if the operation failed.
michael@0:  */
michael@0: U_CAPI UScriptRun * U_EXPORT2
michael@0: uscript_openRun(const UChar *src, int32_t length, UErrorCode *pErrorCode);
michael@0: 
michael@0: /**
michael@0:  * Frees the given <code>UScriptRun</code> object and any storage associated with it.
michael@0:  * On return, scriptRun no longer points to a valid <code>UScriptRun</code> object.
michael@0:  *
michael@0:  * @param scriptRun is the <code>UScriptRun</code> object which will be freed.
michael@0:  */
michael@0: U_CAPI void U_EXPORT2
michael@0: uscript_closeRun(UScriptRun *scriptRun);
michael@0: 
michael@0: /**
michael@0:  * Reset the <code>UScriptRun</code> object so that it will start iterating from
michael@0:  * the beginning.
michael@0:  *
michael@0:  * @param scriptRun is the address of the <code>UScriptRun</code> object to be reset.
michael@0:  */
michael@0: U_CAPI void U_EXPORT2
michael@0: uscript_resetRun(UScriptRun *scriptRun);
michael@0: 
michael@0: /**
michael@0:  * Change the text over which the given <code>UScriptRun</code> object iterates.
michael@0:  *
michael@0:  * @param scriptRun is the <code>UScriptRun</code> object which will be changed.
michael@0:  *
michael@0:  * @param src is the address of the new array of characters over which to iterate.
michael@0:  *        If <code>src == NULL</code> and <code>length == 0</code>,
michael@0:  *        the <code>UScriptRun</code> object will become empty.
michael@0:  *
michael@0:  * @param length is the new number of characters over which to iterate
michael@0:  *
michael@0:  * @param pErrorCode is a pointer to a valid <code>UErrorCode</code> value. If this value
michael@0:  *        indicates a failure on entry, the function will immediately return.
michael@0:  *        On exit the value will indicate the success of the operation.
michael@0:  */
michael@0: U_CAPI void U_EXPORT2
michael@0: uscript_setRunText(UScriptRun *scriptRun, const UChar *src, int32_t length, UErrorCode *pErrorCode);
michael@0: 
michael@0: /**
michael@0:  * Advance the <code>UScriptRun</code> object to the next script run, return the start and limit
michael@0:  * offsets, and the script of the run.
michael@0:  *
michael@0:  * @param scriptRun is the address of the <code>UScriptRun</code> object.
michael@0:  *
michael@0:  * @param pRunStart is a pointer to the variable to receive the starting offset of the next run.
michael@0:  *        This pointer can be <code>NULL</code> if the value is not needed.
michael@0:  *
michael@0:  * @param pRunLimit is a pointer to the variable to receive the limit offset of the next run.
michael@0:  *        This pointer can be <code>NULL</code> if the value is not needed.
michael@0:  *
michael@0:  * @param pRunScript is a pointer to the variable to receive the UScriptCode for the
michael@0:  *        script of the current run. This pointer can be <code>NULL</code> if the value is not needed.
michael@0:  *
michael@0:  * @return true if there was another script run.
michael@0:  */
michael@0: U_CAPI UBool U_EXPORT2
michael@0: uscript_nextRun(UScriptRun *scriptRun, int32_t *pRunStart, int32_t *pRunLimit, UScriptCode *pRunScript);
michael@0: 
michael@0: #endif