Merge scriptx and bidi fields (#78)

Co-authored-by: Zoltan Herczeg <hzmester@freemail.hu>

maint/GenerateUcd.py

@@ -538,7 +538,11 @@ file.close()
 
 script_lists = [[]]
 last_script_extension = ""
-scriptx = read_table('Unicode.tables/ScriptExtensions.txt', get_script_extension, 0)
+scriptx_bidi_class = read_table('Unicode.tables/ScriptExtensions.txt', get_script_extension, 0)
+
+for idx in range(len(scriptx_bidi_class)):
+  scriptx_bidi_class[idx] = scriptx_bidi_class[idx] | (bidi_class[idx] << 11)
+bidi_class = None
 
 # Find the Boolean properties of each character. This next bit of magic creates
 # a list of empty lists. Using [[]] * MAX_UNICODE gives a list of references to
@@ -704,7 +708,7 @@ for s in caseless_sets:
 # Combine all the tables
 
 table, records = combine_tables(script, category, break_props,
-  caseless_offsets, other_case, scriptx, bidi_class, bool_props, padding_dummy)
+  caseless_offsets, other_case, scriptx_bidi_class, bool_props, padding_dummy)
 
 # Find the record size and create a string definition of the structure for
 # outputting as a comment.
@@ -794,8 +798,7 @@ const ucd_record PRIV(dummy_ucd_record)[] = {{
   ucp_gbOther,    /* grapheme break property */
   0,              /* case set */
   0,              /* other case */
-  ucp_Unknown,    /* script extension */
+  0 | (ucp_bidiL << UCD_BIDICLASS_SHIFT), /* script extension and bidi class */
-  ucp_bidiL,      /* bidi class */
   0,              /* bool properties offset */
   0               /* dummy filler */
   }};

src/pcre2_auto_possess.c

@@ -221,7 +221,7 @@ switch(ptype)
 
   case PT_SCX:
   ok = (pdata == prop->script
-        || MAPBIT(PRIV(ucd_script_sets) + prop->scriptx, pdata) != 0);
+        || MAPBIT(PRIV(ucd_script_sets) + UCD_SCRIPTX_PROP(prop), pdata) != 0);
   return ok == negated;
 
   /* These are specials */

src/pcre2_dfa_match.c

@@ -1195,7 +1195,7 @@ for (;;)
 
           case PT_SCX:
           OK = (prop->script == code[2] ||
-                MAPBIT(PRIV(ucd_script_sets) + prop->scriptx, code[2]) != 0);
+                MAPBIT(PRIV(ucd_script_sets) + UCD_SCRIPTX_PROP(prop), code[2]) != 0);
           break;
 
           /* These are specials for combination cases. */
@@ -1467,7 +1467,7 @@ for (;;)
 
           case PT_SCX:
           OK = (prop->script == code[3] ||
-                MAPBIT(PRIV(ucd_script_sets) + prop->scriptx, code[3]) != 0);
+                MAPBIT(PRIV(ucd_script_sets) + UCD_SCRIPTX_PROP(prop), code[3]) != 0);
           break;
 
           /* These are specials for combination cases. */
@@ -1722,7 +1722,7 @@ for (;;)
 
           case PT_SCX:
           OK = (prop->script == code[3] ||
-                MAPBIT(PRIV(ucd_script_sets) + prop->scriptx, code[3]) != 0);
+                MAPBIT(PRIV(ucd_script_sets) + UCD_SCRIPTX_PROP(prop), code[3]) != 0);
           break;
 
           /* These are specials for combination cases. */
@@ -2002,7 +2002,7 @@ for (;;)
 
           case PT_SCX:
           OK = (prop->script == code[1 + IMM2_SIZE + 2] ||
-                MAPBIT(PRIV(ucd_script_sets) + prop->scriptx,
+                MAPBIT(PRIV(ucd_script_sets) + UCD_SCRIPTX_PROP(prop),
                   code[1 + IMM2_SIZE + 2]) != 0);
           break;
 

src/pcre2_internal.h

@@ -1822,8 +1822,7 @@ typedef struct {
   uint8_t gbprop;     /* ucp_gbControl, etc. (grapheme break property) */
   uint8_t caseset;    /* offset to multichar other cases or zero */
   int32_t other_case; /* offset to other case, or zero if none */
-  uint8_t scriptx;    /* script extension value */
+  uint16_t scriptx_bidiclass; /* script extension (11 bit) and bidi class (5 bit) values */
-  uint8_t bidi;       /* bidi class */
   uint8_t bprops;     /* binary properties offset */
   uint8_t dummy;      /* spare - to round to multiple of 4 bytes */
 } ucd_record;
@@ -1842,15 +1841,21 @@ typedef struct {
 #define GET_UCD(ch) REAL_GET_UCD(ch)
 #endif
 
+#define UCD_SCRIPTX_MASK 0x3ff
+#define UCD_BIDICLASS_SHIFT 11
+
+#define UCD_SCRIPTX_PROP(prop) ((prop)->scriptx_bidiclass & UCD_SCRIPTX_MASK)
+#define UCD_BIDICLASS_PROP(prop) ((prop)->scriptx_bidiclass >> UCD_BIDICLASS_SHIFT)
+
 #define UCD_CHARTYPE(ch)    GET_UCD(ch)->chartype
 #define UCD_SCRIPT(ch)      GET_UCD(ch)->script
 #define UCD_CATEGORY(ch)    PRIV(ucp_gentype)[UCD_CHARTYPE(ch)]
 #define UCD_GRAPHBREAK(ch)  GET_UCD(ch)->gbprop
 #define UCD_CASESET(ch)     GET_UCD(ch)->caseset
 #define UCD_OTHERCASE(ch)   ((uint32_t)((int)ch + (int)(GET_UCD(ch)->other_case)))
-#define UCD_SCRIPTX(ch)     GET_UCD(ch)->scriptx
+#define UCD_SCRIPTX(ch)     UCD_SCRIPTX_PROP(GET_UCD(ch))
 #define UCD_BPROPS(ch)      GET_UCD(ch)->bprops
-#define UCD_BIDICLASS(ch)   GET_UCD(ch)->bidi
+#define UCD_BIDICLASS(ch)   UCD_BIDICLASS_PROP(GET_UCD(ch))
 
 /* The "scriptx" and bprops fields contain offsets into vectors of 32-bit words
 that form a bitmap representing a list of scripts or boolean properties. These

src/pcre2_jit_compile.c

@@ -7670,7 +7670,8 @@ if (unicode_status & XCLASS_NEEDS_UCD)
 
   if (unicode_status & XCLASS_HAS_BIDICL)
     {
-    OP1(SLJIT_MOV_U8, TMP1, 0, SLJIT_MEM1(TMP2), (sljit_sw)PRIV(ucd_records) + SLJIT_OFFSETOF(ucd_record, bidi));
+    OP1(SLJIT_MOV_U16, TMP1, 0, SLJIT_MEM1(TMP2), (sljit_sw)PRIV(ucd_records) + SLJIT_OFFSETOF(ucd_record, scriptx_bidiclass));
+    OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, UCD_BIDICLASS_SHIFT);
 
     while (*cc != XCL_END)
       {
@@ -7789,7 +7790,8 @@ if (unicode_status & XCLASS_NEEDS_UCD)
 
   if (unicode_status & XCLASS_HAS_SCRIPT_EXTENSION)
     {
-    OP1(SLJIT_MOV_U8, TMP1, 0, SLJIT_MEM1(TMP2), (sljit_sw)PRIV(ucd_records) + SLJIT_OFFSETOF(ucd_record, scriptx));
+    OP1(SLJIT_MOV_U16, TMP1, 0, SLJIT_MEM1(TMP2), (sljit_sw)PRIV(ucd_records) + SLJIT_OFFSETOF(ucd_record, scriptx_bidiclass));
+    OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, UCD_SCRIPTX_MASK);
     OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 2);
 
     if (unicode_status & XCLASS_SCRIPT_EXTENSION_NOTPROP)

src/pcre2_match.c

@@ -2455,7 +2455,7 @@ fprintf(stderr, "++ op=%d\n", *Fecode);
         case PT_SCX:
           {
           BOOL ok = (Fecode[2] == prop->script ||
-                     MAPBIT((PRIV(ucd_script_sets) + prop->scriptx), Fecode[2]) != 0);
+                     MAPBIT(PRIV(ucd_script_sets) + UCD_SCRIPTX_PROP(prop), Fecode[2]) != 0);
           if (ok == notmatch) RRETURN(MATCH_NOMATCH);
           }
         break;
@@ -2514,7 +2514,7 @@ fprintf(stderr, "++ op=%d\n", *Fecode);
         break;
 
         case PT_BIDICL:
-        if ((prop->bidi == Fecode[2]) == notmatch)
+        if ((UCD_BIDICLASS_PROP(prop) == Fecode[2]) == notmatch)
           RRETURN(MATCH_NOMATCH);
         break;
 
@@ -2737,7 +2737,7 @@ fprintf(stderr, "++ op=%d\n", *Fecode);
             GETCHARINCTEST(fc, Feptr);
             prop = GET_UCD(fc);
             ok = (prop->script == Lpropvalue ||
-                  MAPBIT(PRIV(ucd_script_sets) + prop->scriptx, Lpropvalue) != 0);
+                  MAPBIT(PRIV(ucd_script_sets) + UCD_SCRIPTX_PROP(prop), Lpropvalue) != 0);
             if (ok == notmatch)
               RRETURN(MATCH_NOMATCH);
             }
@@ -3535,7 +3535,7 @@ fprintf(stderr, "++ op=%d\n", *Fecode);
             GETCHARINCTEST(fc, Feptr);
             prop = GET_UCD(fc);
             ok = (prop->script == Lpropvalue
-                  || MAPBIT(PRIV(ucd_script_sets) + prop->scriptx, Lpropvalue) != 0);
+                  || MAPBIT(PRIV(ucd_script_sets) + UCD_SCRIPTX_PROP(prop), Lpropvalue) != 0);
             if (ok == (Lctype == OP_NOTPROP))
               RRETURN(MATCH_NOMATCH);
             }
@@ -4117,7 +4117,7 @@ fprintf(stderr, "++ op=%d\n", *Fecode);
             GETCHARLENTEST(fc, Feptr, len);
             prop = GET_UCD(fc);
             ok = (prop->script == Lpropvalue ||
-                  MAPBIT(PRIV(ucd_script_sets) + prop->scriptx, Lpropvalue) != 0);
+                  MAPBIT(PRIV(ucd_script_sets) + UCD_SCRIPTX_PROP(prop), Lpropvalue) != 0);
             if (ok == notmatch) break;
             Feptr+= len;
             }

src/pcre2_script_run.c

@@ -136,7 +136,7 @@ for (;;)
   Common is always accepted with any script. If there are extensions, the
   following processing happens for all scripts. */
 
-  if (ucd->scriptx != 0 || (script != ucp_Inherited && script != ucp_Common))
+  if (UCD_SCRIPTX_PROP(ucd) != 0 || (script != ucp_Inherited && script != ucp_Common))
     {
     BOOL OK;
 
@@ -146,7 +146,7 @@ for (;;)
     zero, and then, except for Common or Inherited, add this script's bit to
     the map. */
 
-    memcpy(map, PRIV(ucd_script_sets) + ucd->scriptx, UCD_MAPSIZE * sizeof(uint32_t));
+    memcpy(map, PRIV(ucd_script_sets) + UCD_SCRIPTX_PROP(ucd), UCD_MAPSIZE * sizeof(uint32_t));
     memset(map + UCD_MAPSIZE, 0, (FULL_MAPSIZE - UCD_MAPSIZE) * sizeof(uint32_t));
     if (script != ucp_Common && script != ucp_Inherited) MAPSET(map, script);
 

src/pcre2_xclass.c

@@ -163,7 +163,7 @@ while ((t = *data++) != XCL_END)
 
       case PT_SCX:
       ok = (data[1] == prop->script ||
-            MAPBIT(PRIV(ucd_script_sets) + prop->scriptx, data[1]) != 0);
+            MAPBIT(PRIV(ucd_script_sets) + UCD_SCRIPTX_PROP(prop), data[1]) != 0);
       if (ok == isprop) return !negated;
       break;
 
@@ -215,7 +215,7 @@ while ((t = *data++) != XCL_END)
       break;
       
       case PT_BIDICL:
-      if ((prop->bidi == data[1]) == isprop)
+      if ((UCD_BIDICLASS_PROP(prop) == data[1]) == isprop)
         return !negated;
       break;