icuSources/common/usc_impl.h

   1 /*
   2 **********************************************************************
   3 *   Copyright (C) 1999-2011, International Business Machines
   4 *   Corporation and others.  All Rights Reserved.
   5 **********************************************************************
   6 *
   7 * File USC_IMPL.H
   8 *
   9 * Modification History:
  10 *
  11 *   Date        Name        Description
  12 *   07/08/2002  Eric Mader  Creation.
  13 ******************************************************************************
  14 */
  15
  16 #ifndef USC_IMPL_H
  17 #define USC_IMPL_H
  18 #include "unicode/utypes.h"
  19 #include "unicode/uscript.h"
  20
  21 /**
  22  * <code>UScriptRun</code> is used to find runs of characters in
  23  * the same script. It implements a simple iterator over an array
  24  * of characters. The iterator will resolve script-neutral characters
  25  * like punctuation into the script of the surrounding characters.
  26  *
  27  * The iterator will try to match paired punctuation. If it sees an
  28  * opening punctuation character, it will remember the script that
  29  * was assigned to that character, and assign the same script to the
  30  * matching closing punctuation.
  31  *
  32  * Scripts are chosen based on the <code>UScriptCode</code> enumeration.
  33  * No attempt is made to combine related scripts into a single run. In
  34  * particular, Hiragana, Katakana, and Han characters will appear in seperate
  35  * runs.
  36
  37  * Here is an example of how to iterate over script runs:
  38  * <pre>
  39  * \code
  40  * void printScriptRuns(const UChar *text, int32_t length)
  41  * {
  42  *     UErrorCode error = U_ZERO_ERROR;
  43  *     UScriptRun *scriptRun = uscript_openRun(text, testLength, &error);
  44  *     int32_t start = 0, limit = 0;
  45  *     UScriptCode code = USCRIPT_INVALID_CODE;
  46  *
  47  *     while (uscript_nextRun(&start, &limit, &code)) {
  48  *         printf("Script '%s' from %d to %d.\n", uscript_getName(code), start, limit);
  49  *     }
  50  *
  51  *     uscript_closeRun(scriptRun);
  52  *  }
  53  * </pre>
  54  */
  55 struct UScriptRun;
  56
  57 typedef struct UScriptRun UScriptRun;
  58
  59 /**
  60  * Create a <code>UScriptRun</code> object for iterating over the given text. This object must
  61  * be freed using <code>uscript_closeRun()</code>. Note that this object does not copy the source text,
  62  * only the pointer to it. You must make sure that the pointer remains valid until you call
  63  * <code>uscript_closeRun()</code> or <code>uscript_setRunText()</code>.
  64  *
  65  * @param src is the address of the array of characters over which to iterate.
  66  *        if <code>src == NULL</code> and <code>length == 0</code>,
  67  *        an empty <code>UScriptRun</code> object will be returned.
  68  *
  69  * @param length is the number of characters over which to iterate.
  70  *
  71  * @param pErrorCode is a pointer to a valid <code>UErrorCode</code> value. If this value
  72  *        indicates a failure on entry, the function will immediately return.
  73  *        On exit the value will indicate the success of the operation.
  74  *
  75  * @return the address of <code>UScriptRun</code> object which will iterate over the text,
  76  *         or <code>NULL</code> if the operation failed.
  77  */
  78 U_CAPI UScriptRun * U_EXPORT2
  79 uscript_openRun(const UChar *src, int32_t length, UErrorCode *pErrorCode);
  80
  81 /**
  82  * Frees the given <code>UScriptRun</code> object and any storage associated with it.
  83  * On return, scriptRun no longer points to a valid <code>UScriptRun</code> object.
  84  *
  85  * @param scriptRun is the <code>UScriptRun</code> object which will be freed.
  86  */
  87 U_CAPI void U_EXPORT2
  88 uscript_closeRun(UScriptRun *scriptRun);
  89
  90 /**
  91  * Reset the <code>UScriptRun</code> object so that it will start iterating from
  92  * the beginning.
  93  *
  94  * @param scriptRun is the address of the <code>UScriptRun</code> object to be reset.
  95  */
  96 U_CAPI void U_EXPORT2
  97 uscript_resetRun(UScriptRun *scriptRun);
  98
  99 /**
 100  * Change the text over which the given <code>UScriptRun</code> object iterates.
 101  *
 102  * @param scriptRun is the <code>UScriptRun</code> object which will be changed.
 103  *
 104  * @param src is the address of the new array of characters over which to iterate.
 105  *        If <code>src == NULL</code> and <code>length == 0</code>,
 106  *        the <code>UScriptRun</code> object will become empty.
 107  *
 108  * @param length is the new number of characters over which to iterate
 109  *
 110  * @param pErrorCode is a pointer to a valid <code>UErrorCode</code> value. If this value
 111  *        indicates a failure on entry, the function will immediately return.
 112  *        On exit the value will indicate the success of the operation.
 113  */
 114 U_CAPI void U_EXPORT2
 115 uscript_setRunText(UScriptRun *scriptRun, const UChar *src, int32_t length, UErrorCode *pErrorCode);
 116
 117 /**
 118  * Advance the <code>UScriptRun</code> object to the next script run, return the start and limit
 119  * offsets, and the script of the run.
 120  *
 121  * @param scriptRun is the address of the <code>UScriptRun</code> object.
 122  *
 123  * @param pRunStart is a pointer to the variable to receive the starting offset of the next run.
 124  *        This pointer can be <code>NULL</code> if the value is not needed.
 125  *
 126  * @param pRunLimit is a pointer to the variable to receive the limit offset of the next run.
 127  *        This pointer can be <code>NULL</code> if the value is not needed.
 128  *
 129  * @param pRunScript is a pointer to the variable to receive the UScriptCode for the
 130  *        script of the current run. This pointer can be <code>NULL</code> if the value is not needed.
 131  *
 132  * @return true if there was another script run.
 133  */
 134 U_CAPI UBool U_EXPORT2
 135 uscript_nextRun(UScriptRun *scriptRun, int32_t *pRunStart, int32_t *pRunLimit, UScriptCode *pRunScript);
 136
 137 #endif