Apply ICU patches for ICU tickets 6175 (ISO-2022 and ...

third_party/icu38/source/common/ucnv_ext.c

 /*
 ******************************************************************************
 *
 *   Copyright (C) 2003-2007, International Business Machines
 *   Corporation and others.  All Rights Reserved.
 *
 ******************************************************************************
 *   file name:  ucnv_ext.c
 *   encoding:   US-ASCII
 *   tab size:   8 (not used)
@@ skipping 928 matching lines @@
 
         /* set the error code for unassigned */
         *pErrorCode=U_INVALID_CHAR_FOUND;
     }
 }
 
 static void
 ucnv_extGetUnicodeSetString(const UConverterSharedData *sharedData,
                             const int32_t *cx,
                             const USetAdder *sa,
-                            UConverterUnicodeSet which,
+                            UBool useFallback,
                             int32_t minLength,
                             UChar32 c,
                             UChar s[UCNV_EXT_MAX_UCHARS], int32_t length,
                             int32_t sectionIndex,
                             UErrorCode *pErrorCode) {
     const UChar *fromUSectionUChars;
     const uint32_t *fromUSectionValues;
 
     uint32_t value;
     int32_t i, count;
 
     fromUSectionUChars=UCNV_EXT_ARRAY(cx, UCNV_EXT_FROM_U_UCHARS_INDEX, UChar)+sectionIndex;
     fromUSectionValues=UCNV_EXT_ARRAY(cx, UCNV_EXT_FROM_U_VALUES_INDEX, uint32_t)+sectionIndex;
 
     /* read first pair of the section */
     count=*fromUSectionUChars++;
     value=*fromUSectionValues++;
 
     if( value!=0 &&
-        UCNV_EXT_FROM_U_IS_ROUNDTRIP(value) &&
+        (UCNV_EXT_FROM_U_IS_ROUNDTRIP(value) || useFallback) &&
         UCNV_EXT_FROM_U_GET_LENGTH(value)>=minLength
     ) {
         if(c>=0) {
             /* add the initial code point */
             sa->add(sa->set, c);
         } else {
             /* add the string so far */
             sa->addString(sa->set, s, length);
         }
     }
 
     for(i=0; i<count; ++i) {
         /* append this code unit and recurse or add the string */
         s[length]=fromUSectionUChars[i];
         value=fromUSectionValues[i];
 
         if(value==0) {
             /* no mapping, do nothing */
         } else if(UCNV_EXT_FROM_U_IS_PARTIAL(value)) {
             ucnv_extGetUnicodeSetString(
-                sharedData, cx, sa, which, minLength,
+                sharedData, cx, sa, useFallback, minLength,
                 U_SENTINEL, s, length+1,
                 (int32_t)UCNV_EXT_FROM_U_GET_PARTIAL_INDEX(value),
                 pErrorCode);
-        } else if(((value&(UCNV_EXT_FROM_U_ROUNDTRIP_FLAG|UCNV_EXT_FROM_U_RESERVED_MASK))==
-                           UCNV_EXT_FROM_U_ROUNDTRIP_FLAG) &&
+        } else if((useFallback ?
+                      (value&UCNV_EXT_FROM_U_RESERVED_MASK)==0 :
+                      ((value&(UCNV_EXT_FROM_U_ROUNDTRIP_FLAG|UCNV_EXT_FROM_U_RESERVED_MASK))==
+                          UCNV_EXT_FROM_U_ROUNDTRIP_FLAG)) &&
                   UCNV_EXT_FROM_U_GET_LENGTH(value)>=minLength
         ) {
             sa->addString(sa->set, s, length+1);
         }
     }
 }
 
 U_CFUNC void
 ucnv_extGetUnicodeSet(const UConverterSharedData *sharedData,
                       const USetAdder *sa,
                       UConverterUnicodeSet which,
+                      UConverterSetFilter filter,
                       UErrorCode *pErrorCode) {
     const int32_t *cx;
     const uint16_t *stage12, *stage3, *ps2, *ps3;
     const uint32_t *stage3b;
 
     uint32_t value;
     int32_t st1, stage1Length, st2, st3, minLength;
+    UBool useFallback;
 
     UChar s[UCNV_EXT_MAX_UCHARS];
     UChar32 c;
     int32_t length;
 
     cx=sharedData->mbcs.extIndexes;
     if(cx==NULL) {
         return;
     }
 
     stage12=UCNV_EXT_ARRAY(cx, UCNV_EXT_FROM_U_STAGE_12_INDEX, uint16_t);
     stage3=UCNV_EXT_ARRAY(cx, UCNV_EXT_FROM_U_STAGE_3_INDEX, uint16_t);
     stage3b=UCNV_EXT_ARRAY(cx, UCNV_EXT_FROM_U_STAGE_3B_INDEX, uint32_t);
 
     stage1Length=cx[UCNV_EXT_FROM_U_STAGE_1_LENGTH];
 
+    useFallback=(UBool)(which==UCNV_ROUNDTRIP_AND_FALLBACK_SET);
+
     /* enumerate the from-Unicode trie table */
     c=0; /* keep track of the current code point while enumerating */
 
-    if(sharedData->mbcs.outputType==MBCS_OUTPUT_DBCS_ONLY) {
+    if(filter==UCNV_SET_FILTER_2022_CN) {
+        minLength=3;
+    } else if( sharedData->mbcs.outputType==MBCS_OUTPUT_DBCS_ONLY ||
+               filter!=UCNV_SET_FILTER_NONE
+    ) {
         /* DBCS-only, ignore single-byte results */
         minLength=2;
     } else {
         minLength=1;
     }
 
     /*
      * the trie enumeration is almost the same as
      * in MBCSGetUnicodeSet() for MBCS_OUTPUT_1
      */
@@ skipping 13 matching lines @@
                      * Recurse for partial results.
                      */
                     do {
                         value=stage3b[*ps3++];
                         if(value==0) {
                             /* no mapping, do nothing */
                         } else if(UCNV_EXT_FROM_U_IS_PARTIAL(value)) {
                             length=0;
                             U16_APPEND_UNSAFE(s, length, c);
                             ucnv_extGetUnicodeSetString(
-                                sharedData, cx, sa, which, minLength,
+                                sharedData, cx, sa, useFallback, minLength,
                                 c, s, length,
                                 (int32_t)UCNV_EXT_FROM_U_GET_PARTIAL_INDEX(value),
                                 pErrorCode);
-                        } else if(((value&(UCNV_EXT_FROM_U_ROUNDTRIP_FLAG|UCNV_EXT_FROM_U_RESERVED_MASK))==
-                                           UCNV_EXT_FROM_U_ROUNDTRIP_FLAG) &&
+                        } else if((useFallback ?
+                                      (value&UCNV_EXT_FROM_U_RESERVED_MASK)==0 :
+                                      ((value&(UCNV_EXT_FROM_U_ROUNDTRIP_FLAG|UCNV_EXT_FROM_U_RESERVED_MASK))==
+                                          UCNV_EXT_FROM_U_ROUNDTRIP_FLAG)) &&
                                   UCNV_EXT_FROM_U_GET_LENGTH(value)>=minLength
                         ) {
+                            switch(filter) {
+                            case UCNV_SET_FILTER_2022_CN:
+                                if(!(UCNV_EXT_FROM_U_GET_LENGTH(value)==3 && UCNV_EXT_FROM_U_GET_DATA(value)<=0x82ffff)) {
+                                    continue;
+                                }
+                                break;
+                            case UCNV_SET_FILTER_SJIS:
+                                if(!(UCNV_EXT_FROM_U_GET_LENGTH(value)==2 && (value=UCNV_EXT_FROM_U_GET_DATA(value))>=0x8140 && value<=0xeffc)) {
+                                    continue;
+                                }
+                                break;
+                            case UCNV_SET_FILTER_GR94DBCS:
+                                if(!(UCNV_EXT_FROM_U_GET_LENGTH(value)==2 &&
+                                     (uint16_t)((value=UCNV_EXT_FROM_U_GET_DATA(value))-0xa1a1)<=(0xfefe - 0xa1a1) &&
+                                     (uint8_t)(value-0xa1)<=(0xfe - 0xa1))) {
+                                    continue;
+                                }
+                                break;
+                            case UCNV_SET_FILTER_HZ:
+                                if(!(UCNV_EXT_FROM_U_GET_LENGTH(value)==2 &&
+                                     (uint16_t)((value=UCNV_EXT_FROM_U_GET_DATA(value))-0xa1a1)<=(0xfdfe - 0xa1a1) &&
+                                     (uint8_t)(value-0xa1)<=(0xfe - 0xa1))) {
+                                    continue;
+                                }
+                                break;
+                            default:
+                                /*
+                                 * UCNV_SET_FILTER_NONE,
+                                 * or UCNV_SET_FILTER_DBCS_ONLY which is handled via minLength
+                                 */
+                                break;
+                            }
                             sa->add(sa->set, c);
                         }
                     } while((++c&0xf)!=0);
                 } else {
                     c+=16; /* empty stage 3 block */
                 }
             }
         } else {
             c+=1024; /* empty stage 2 block */
         }
     }
 }
 
 #endif /* #if !UCONFIG_NO_LEGACY_CONVERSION */