fix memory usage bug of UStrGetBreaks; UStrGetBreaks does not return when it encounters a hard break

include/gdi.h

@@ -8730,18 +8730,16 @@ MG_EXPORT void GUIAPI UBidiShape(Uint32 shaping_flags,
     /** @} end of breaking_opportunities */
 
 /**
- * \fn int GUIAPI UStrGetBreaks(Uchar32* ucs, int nr_ucs,
- *          UCharScriptType writing_system, Uint8 ctr, Uint8 wbr, Uint8 lbp,
- *          Uint16** break_oppos);
- * \brief Convert a multi-byte character string to a Unicode character string
- *      and calculate the breaking opportunities under the specified rules and
- *      line breaking policy.
+ * \fn int GUIAPI UStrGetBreaks(UCharScriptType writing_system,
+ *          Uint8 ctr, Uint8 wbr, Uint8 lbp,
+ *          Uchar32* ucs, int nr_ucs, Uint16** break_oppos);
+ * \brief Calculate the breaking opportunities of a Uchar32 string under
+ *      the specified rules and line breaking policy.
  *
- * This function calculates and allocates the Uchar32 string and the breaking
- * opportunities of the characters from a multi-byte string under the specified
- * content language \a content_language, the writing system \a writing_system,
- * the white space rule \a wsr, the text transformation rule
- * \a ctr, the word breaking rule \a wbr, and the line breaking policy \a lbp.
+ * This function calculates the breaking opportunities of the Unicode characters
+ * under the specified the writing system \a writing_system, the word breaking
+ * rule \a wbr, and the line breaking policy \a lbp. This function also
+ * transforms the character according to the text transformation rule \a ctr.
  *
  * The implementation of this function conforms to UNICODE LINE BREAKING
  * ALGORITHM:
@@ -8756,50 +8754,44 @@ MG_EXPORT void GUIAPI UBidiShape(Uint32 shaping_flags,
  *
  *      https://www.w3.org/TR/css-text-3/
  *
- * The function will return if it encounters any hard line break or the null
- * character. The hard line break will be included in the returned Uchar32 if
- * there was one.
+ * The function will return if it encounters the end of the text.
  *
- * Note that you are responsible for freeing the Uchar32 string and the
- * break opportunities array allocated by this function.
+ * Note that you are responsible for freeing the break opportunities array
+ * allocated by this function if it allocates the buffer.
  *
- * \param logfont The logfont used to parse the string.
- * \param mstr The pointer to the multi-byte string.
- * \param mstr_len The length of \a mstr in bytes.
- * \param content_language The content lanuage identifier.
  * \param writing_system The writing system (script) identifier.
- * \param wsr The white space rule; see \a white_space_rules.
- * \param ctr The character transformation rule;
- *        see \a char_transform_rule.
+ * \param ctr The character transformation rule; see \a char_transform_rule.
  * \param wbr The word breaking rule; see \a word_break_rules.
- * \param lbp The line breaking policy;
- *        see \a line_break_policies.
- * \param uchars The pointer to a buffer to store the address of the
- *      Uchar32 array which contains the Unicode character values.
- * \param break_oppos The pointer to a buffer to store the address of the
- *        Uint16 array which contains the break opportunities of the glyphs.
+ * \param lbp The line breaking policy; see \a line_break_policies.
+ * \param ucs The Uchar32 string.
+ * \param nr_ucs The length of the Uchar32 string.
+ * \param break_oppos The pointer to a buffer to store the address of a
+ *        Uint16 array which will return the break opportunities of the uchars.
+ *        If the buffer contains a NULL value, this function will try to
+ *        allocate a new space for the break opportunities.
  *        Note that the length of this array is always one longer than
- *        the glyphs array. The first unit of the array stores the
- *        break opportunity before the first glyph, and the others store
+ *        the Unicode array. The first unit of the array stores the
+ *        break opportunity before the first uchar, and the others store
  *        the break opportunities after other gyphs.
  *        The break opportunity can be one of the following values:
  *          - BOV_LB_MANDATORY\n
  *            The mandatory breaking.
  *          - BOV_LB_NOTALLOWED\n
- *            No breaking allowed after the glyph definitely.
+ *            No breaking allowed after the uchar definitely.
  *          - BOV_LB_ALLOWED\n
- *            Breaking allowed after the glyph.
- * \param nr_glyphs The buffer to store the number of the allocated glyphs.
+ *            Breaking allowed after the uchar.
  *
- * \return The number of the bytes consumed in \a mstr; zero on error.
+ * \return The number of the Uchar32 characters consumed in \a ucs;
+ *      zero on error.
  *
  * \note Only available when support for UNICODE is enabled.
  *
- * \sa DrawGlyphStringEx, white_space_rules, char_transform_rule
+ * \sa DrawGlyphStringEx, word_break_rules, char_transform_rule,
+ *      line_break_policies
  */
-MG_EXPORT int GUIAPI UStrGetBreaks(Uchar32* ucs, int nr_ucs,
-        UCharScriptType writing_system, Uint8 ctr, Uint8 wbr, Uint8 lbp,
-        Uint16** break_oppos);
+MG_EXPORT int GUIAPI UStrGetBreaks(UCharScriptType writing_system,
+        Uint8 ctr, Uint8 wbr, Uint8 lbp,
+        Uchar32* ucs, int nr_ucs, Uint16** break_oppos);
 
 /** The function determines whether a character is alphanumeric. */
 MG_EXPORT BOOL GUIAPI IsUCharAlnum(Uchar32 uc);

src/font/unicode-bidi.c

@@ -52,8 +52,6 @@
 #include <ctype.h>
 #include <assert.h>
 
-#define DEBUG
-
 #include "common.h"
 
 #ifdef _MGCHARSET_UNICODE

src/font/unicode-break.c

@@ -39,8 +39,9 @@
 **
 ** Reference:
 **
-**  [UNICODE LINE BREAKING ALGORITHM](https://www.unicode.org/reports/tr14/)
 **  [UNICODE TEXT SEGMENTATION](https://www.unicode.org/reports/tr29/)
+**  [UNICODE LINE BREAKING ALGORITHM](https://www.unicode.org/reports/tr14/)
+**  [CSS Text Module Level 3](https://www.w3.org/TR/css-text-3/#content-writing-system)
 **
 ** Create by WEI Yongming at 2019/01/16
 */
@@ -1303,7 +1304,8 @@ static void dbg_dump_ctxt(struct break_ctxt* ctxt,
 #define LOCAL_ARRAY_SIZE        256
 
 static int break_init_spaces(struct break_ctxt* ctxt,
-        Uchar32* ucs, Uint8* local_bts, Uint8* local_ods, int nr_ucs)
+        Uchar32* ucs, Uint16* bos, Uint8* local_bts, Uint8* local_ods,
+        int nr_ucs)
 {
     ctxt->ucs = ucs;
     ctxt->nr_ucs = nr_ucs;
@@ -1317,7 +1319,10 @@ static int break_init_spaces(struct break_ctxt* ctxt,
         ctxt->ods = local_ods;
     }
 
-    ctxt->bos = (Uint16*)malloc(sizeof(Uint16) * (ctxt->nr_ucs + 1));
+    if (bos)
+        ctxt->bos = bos;
+    else
+        ctxt->bos = (Uint16*)malloc(sizeof(Uint16) * (ctxt->nr_ucs + 1));
 
     if (ctxt->bts == NULL || ctxt->bos == NULL || ctxt->ods == NULL)
         return 0;
@@ -2046,9 +2051,9 @@ static int check_subsequent_ri(struct break_ctxt* ctxt,
     return cosumed;
 }
 
-int GUIAPI UStrGetBreaks(Uchar32* ucs, int nr_ucs,
-            UCharScriptType writing_system, Uint8 ctr, Uint8 wbr, Uint8 lbp,
-            Uint16** break_oppos)
+int GUIAPI UStrGetBreaks(UCharScriptType writing_system,
+            Uint8 ctr, Uint8 wbr, Uint8 lbp,
+            Uchar32* ucs, int nr_ucs, Uint16** break_oppos)
 {
     struct break_ctxt ctxt;
     Uint8 local_bts [LOCAL_ARRAY_SIZE];
@@ -2079,12 +2084,11 @@ int GUIAPI UStrGetBreaks(Uchar32* ucs, int nr_ucs,
     ctxt.prev_jamo = NO_JAMO;
     ctxt.curr_wt = WordNone;
 
-    *break_oppos = NULL;
-
     if (ucs == NULL || nr_ucs == 0)
         return 0;
 
-    if (break_init_spaces(&ctxt, ucs, local_bts, local_ods, nr_ucs) <= 0) {
+    if (break_init_spaces(&ctxt, ucs, *break_oppos, local_bts, local_ods,
+            nr_ucs) <= 0) {
         goto error;
     }
 
@@ -2938,10 +2942,14 @@ next_uchar:
         ucs_left += cosumed_one_loop;
         cosumed += cosumed_one_loop;
 
-        // Return if we got any BK!
-        if ((ctxt.bos[ctxt.n] & BOV_LB_MASK) == BOV_LB_MANDATORY) {
+        _DBG_PRINTF("%s: nr_ucs: %d, ctxt.n: %d, nr_left_ucs: %d\n",
+                __FUNCTION__, nr_ucs, ctxt.n, nr_left_ucs);
+#if 0
+        // Do not return if we got any BK!
+        if (0 && (ctxt.bos[ctxt.n - 1] & BOV_LB_MASK) == BOV_LB_MANDATORY) {
             break;
         }
+#endif
     }
 
     if (ctxt.n > 0) {
@@ -2968,18 +2976,18 @@ next_uchar:
             }
         }
 
-        *break_oppos = ctxt.bos;
     }
     else
         goto error;
 
+    if (*break_oppos == NULL) *break_oppos = ctxt.bos;
     if (ctxt.ods && ctxt.ods != local_ods) free(ctxt.ods);
     if (ctxt.bts && ctxt.bts != local_bts) free(ctxt.bts);
 
-    return cosumed;
+    return ctxt.n - 1;
 
 error:
-    if (ctxt.bos) free(ctxt.bos);
+    if (*break_oppos == NULL && ctxt.bos) free(ctxt.bos);
     if (ctxt.ods && ctxt.ods != local_ods) free(ctxt.ods);
     if (ctxt.bts && ctxt.bts != local_bts) free(ctxt.bts);
     return 0;