diff options
Diffstat (limited to '')
| -rw-r--r-- | libwc/wtf.c | 67 |
1 files changed, 48 insertions, 19 deletions
diff --git a/libwc/wtf.c b/libwc/wtf.c index b8cfdc7..94d95c1 100644 --- a/libwc/wtf.c +++ b/libwc/wtf.c @@ -120,29 +120,36 @@ int wtf_strwidth(wc_uchar *p) { int w = 0; + wc_uchar *q = p + strlen(p); - while (*p) { + while (p < q) { w += wtf_width(p); p += WTF_LEN_MAP[*p]; } return w; } -/* size_t wtf_len1(wc_uchar *p) { - return (size_t)WTF_LEN_MAP[*p]; + size_t len, len_max = WTF_LEN_MAP[*p]; + + for (len = 0; *(p + len); len++) + if (len == len_max) + break; + if (len == 0) + len = 1; + return len; } -*/ size_t wtf_len(wc_uchar *p) { wc_uchar *q = p; + wc_uchar *strz = p + strlen(p); q += WTF_LEN_MAP[*q]; - while (*q && ! WTF_WIDTH_MAP[*q]) + while (q < strz && ! WTF_WIDTH_MAP[*q]) q += WTF_LEN_MAP[*q]; return q - p; } @@ -166,15 +173,17 @@ wtf_type(wc_uchar *p) ((p)[3] = (((c) >> 7) & 0x7f) | 0x80), \ ((p)[4] = ( (c) & 0x7f) | 0x80) #define wtf_to_wcs16(p) \ + ((p)[0] == 0 || (p)[1] == 0 || (p)[2] == 0 ? 0 : \ ((wc_uint32)((p)[0] & 0x03) << 14) \ | ((wc_uint32)((p)[1] & 0x7f) << 7) \ - | ((wc_uint32)((p)[2] & 0x7f) ) + | ((wc_uint32)((p)[2] & 0x7f) )) #define wtf_to_wcs32(p) \ + ((p)[0] == 0 || (p)[1] == 0 || (p)[2] == 0 || (p)[3] == 0 || (p)[4] == 0 ? 0 : \ ((wc_uint32)((p)[0] & 0x0f) << 28) \ | ((wc_uint32)((p)[1] & 0x7f) << 21) \ | ((wc_uint32)((p)[2] & 0x7f) << 14) \ | ((wc_uint32)((p)[3] & 0x7f) << 7) \ - | ((wc_uint32)((p)[4] & 0x7f) ) + | ((wc_uint32)((p)[4] & 0x7f) )) void wtf_push(Str os, wc_ccs ccs, wc_uint32 code) @@ -388,7 +397,7 @@ wtf_parse1(wc_uchar **p) cc.code = *(q++); } else if (*q > 0xa0) { cc.ccs = wtf_gr_ccs; - if (WC_CCS_IS_WIDE(cc.ccs)) { + if (WC_CCS_IS_WIDE(cc.ccs) && *(q+1)) { cc.code = ((wc_uint32)*q << 8) | *(q+1); q += 2; } else @@ -401,27 +410,47 @@ wtf_parse1(wc_uchar **p) case WC_CCS_A_CS942: case WC_CCS_A_PCS: case WC_CCS_A_UNKNOWN: - cc.ccs |= *(q++) & 0x7f; - cc.code = *(q++); + if (*q && *(q+1)) { + cc.ccs |= *(q++) & 0x7f; + cc.code = *(q++); + } else { + cc.ccs = WC_CCS_US_ASCII; + cc.code = (wc_uint32)' '; + } break; case WC_CCS_A_CS94W: case WC_CCS_A_CS96W: case WC_CCS_A_PCSW: - cc.ccs |= *(q++) & 0x7f; - cc.code = ((wc_uint32)*q << 8) | *(q+1); - q += 2; + if (*q && *(q+1) && *(q+2)) { + cc.ccs |= *(q++) & 0x7f; + cc.code = ((wc_uint32)*q << 8) | *(q+1); + q += 2; + } else { + cc.ccs = WC_CCS_US_ASCII; + cc.code = (wc_uint32)' '; + } break; case WC_CCS_A_WCS16: case WC_CCS_A_WCS16W: - cc.ccs |= (*q & 0x7c) >> 2; - cc.code = wtf_to_wcs16(q); - q += 3; + if (*q && *(q+1) && *(q+2)) { + cc.ccs |= (*q & 0x7c) >> 2; + cc.code = wtf_to_wcs16(q); + q += 3; + } else { + cc.ccs = WC_CCS_US_ASCII; + cc.code = (wc_uint32)' '; + } break; case WC_CCS_A_WCS32: case WC_CCS_A_WCS32W: - cc.ccs |= (*q & 0x70) >> 4; - cc.code = wtf_to_wcs32(q); - q += 5; + if (*q && *(q+1) && *(q+2) && *(q+3) && *(q+4)) { + cc.ccs |= (*q & 0x70) >> 4; + cc.code = wtf_to_wcs32(q); + q += 5; + } else { + cc.ccs = WC_CCS_US_ASCII; + cc.code = (wc_uint32)' '; + } break; default: /* case 0: */ |