avoid reading uninitialized memory in __map_file

[musl] / src / locale / iconv.c

diff --git a/src/locale/iconv.c b/src/locale/iconv.c
index 30ea8da..1eeea94 100644 (file)
--- a/src/locale/iconv.c
+++ b/src/locale/iconv.c
@@ -5,6 +5,7 @@
 #include <stdlib.h>
 #include <limits.h>
 #include <stdint.h>
+#include "locale_impl.h"
 
 #define UTF_32BE    0300
 #define UTF_16LE    0301
@@ -23,19 +24,13 @@
 #define BIG5        0340
 #define EUC_KR      0350
 
-/* FIXME: these are not implemented yet
- * EUC:   A1-FE A1-FE
- * GBK:   81-FE 40-7E,80-FE
- * Big5:  A1-FE 40-7E,A1-FE
- */
-
 /* Definitions of charmaps. Each charmap consists of:
  * 1. Empty-string-terminated list of null-terminated aliases.
  * 2. Special type code or number of elided entries.
  * 3. Character table (size determined by field 2). */
 
 static const unsigned char charmaps[] =
-"utf8\0\0\310"
+"utf8\0char\0\0\310"
 "wchart\0\0\306"
 "ucs2\0ucs2be\0\0\304"
 "ucs2le\0\0\305"
@@ -49,7 +44,7 @@ static const unsigned char charmaps[] =
 "gb18030\0\0\330"
 "gbk\0\0\331"
 "gb2312\0\0\332"
-"big5\0bigfive\0cp950\0\0\340"
+"big5\0bigfive\0cp950\0big5hkscs\0\0\340"
 "euckr\0ksc5601\0ksx1001\0cp949\0\0\350"
 #include "codepages.h"
 ;
@@ -70,6 +65,10 @@ static const unsigned short big5[89][157] = {
 #include "big5.h"
 };
 
+static const unsigned short hkscs[] = {
+#include "hkscs.h"
+};
+
 static const unsigned short ksc[93][94] = {
 #include "ksc.h"
 };
@@ -86,6 +85,7 @@ static int fuzzycmp(const unsigned char *a, const unsigned char *b)
 static size_t find_charmap(const void *name)
 {
        const unsigned char *s;
+       if (!*(char *)name) name=charmaps; /* "utf8" */
        for (s=charmaps; *s; ) {
                if (!fuzzycmp(name, s)) {
                        for (; *s; s+=strlen((void *)s)+1);
@@ -166,9 +166,12 @@ size_t iconv(iconv_t cd0, char **restrict in, size_t *restrict inb, char **restr
        int err;
        unsigned char type = map[-1];
        unsigned char totype = tomap[-1];
+       locale_t *ploc = &CURRENT_LOCALE, loc = *ploc;
 
        if (!in || !*in || !*inb) return 0;
 
+       *ploc = UTF8_LOCALE;
+
        for (; *inb; *in+=l, *inb-=l) {
                c = *(unsigned char *)*in;
                l = 1;
@@ -294,12 +297,37 @@ size_t iconv(iconv_t cd0, char **restrict in, size_t *restrict inb, char **restr
                        l = 2;
                        if (*inb < 2) goto starved;
                        d = *((unsigned char *)*in + 1);
-                       if (c-0xa1>=0xfa-0xa1) goto ilseq;
-                       c -= 0xa1;
                        if (d-0x40>=0xff-0x40 || d-0x7f<0xa1-0x7f) goto ilseq;
                        d -= 0x40;
                        if (d > 0x3e) d -= 0x22;
-                       c = big5[c][d];
+                       if (c-0xa1>=0xfa-0xa1) {
+                               if (c-0x87>=0xff-0x87) goto ilseq;
+                               if (c < 0xa1) c -= 0x87;
+                               else c -= 0x87 + (0xfa-0xa1);
+                               c = (hkscs[4867+(c*157+d)/16]>>(c*157+d)%16)%2<<17
+                                       | hkscs[c*157+d];
+                               /* A few HKSCS characters map to pairs of UCS
+                                * characters. These are mapped to surrogate
+                                * range in the hkscs table then hard-coded
+                                * here. Ugly, yes. */
+                               if (c/256 == 0xdc) {
+                                       if (totype-0300U > 8) k = 2;
+                                       else k = "\10\4\4\10\4\4\10\2\4"[totype-0300];
+                                       if (k > *outb) goto toobig;
+                                       x += iconv((iconv_t)(uintptr_t)to,
+                                               &(char *){"\303\212\314\204"
+                                               "\303\212\314\214"
+                                               "\303\252\314\204"
+                                               "\303\252\314\214"
+                                               +c%256}, &(size_t){4},
+                                               out, outb);
+                                       continue;
+                               }
+                               if (!c) goto ilseq;
+                               break;
+                       }
+                       c -= 0xa1;
+                       c = big5[c][d]|(c==0x27&&(d==0x3a||d==0x3c||d==0x42))<<17;
                        if (!c) goto ilseq;
                        break;
                case EUC_KR:
@@ -407,6 +435,7 @@ size_t iconv(iconv_t cd0, char **restrict in, size_t *restrict inb, char **restr
                        break;
                }
        }
+       *ploc = loc;
        return x;
 ilseq:
        err = EILSEQ;
@@ -421,5 +450,6 @@ starved:
        x = -1;
 end:
        errno = err;
+       *ploc = loc;
        return x;
 }