tscrape

twitter scraper
git clone git://git.codemadness.org/tscrape
Log | Files | Refs | README | LICENSE

commit 573905aec2e99fbe31a1cabe5864853ef9015a41
parent 426522824e719e081c9c5e47ba8771779b0fdc85
Author: Hiltjo Posthuma <hiltjo@codemadness.org>
Date:   Fri, 20 Mar 2020 12:00:16 +0100

sync printutf8pad from sfeed

changes:
- util: printutf8pad: proper counting of multiwidth characters
  for example the string "\xef\xbc\xb5".
- optimization

Diffstat:
Mutil.c | 30++++++++++++++++++------------
1 file changed, 18 insertions(+), 12 deletions(-)

diff --git a/util.c b/util.c @@ -72,32 +72,38 @@ xmlencode(const char *s, FILE *fp) } } -/* print `len' columns of characters. If string is shorter pad the rest - * with characters `pad`. */ +/* print `len' columns of characters. If string is shorter pad the rest with + * characters `pad`. */ void printutf8pad(FILE *fp, const char *s, size_t len, int pad) { - wchar_t w; + wchar_t wc; size_t col = 0, i, slen; - int rl, wc; + int rl, w; if (!len) return; slen = strlen(s); - for (i = 0; i < slen && col < len + 1; i += rl) { - if ((rl = mbtowc(&w, &s[i], slen - i < 4 ? slen - i : 4)) <= 0) - break; - if ((wc = wcwidth(w)) == -1) - wc = 1; - col += (size_t)wc; - if (col >= len && s[i + rl]) { + for (i = 0; i < slen; i += rl) { + rl = w = 1; + if ((unsigned char)s[i] < 32) + continue; + if ((unsigned char)s[i] >= 127) { + if ((rl = mbtowc(&wc, s + i, slen - i < 4 ? slen - i : 4)) <= 0) + break; + if ((w = wcwidth(wc)) == -1) + continue; + } + if (col + w > len || (col + w == len && s[i + rl])) { fputs("\xe2\x80\xa6", fp); + col++; break; } fwrite(&s[i], 1, rl, fp); + col += w; } - for (; col < len; col++) + for (; col < len; ++col) putc(pad, fp); }