From 5cea49038068111d6a0cfde87487e06d7ed46b20 Mon Sep 17 00:00:00 2001 From: ache Date: Thu, 22 Sep 2016 16:49:53 +0000 Subject: MFC r305841 Implement multibyte encoding support for -v with fallback --- bin/cat/cat.c | 37 +++++++++++++++++++++++++++++++------ 1 file changed, 31 insertions(+), 6 deletions(-) diff --git a/bin/cat/cat.c b/bin/cat/cat.c index b5f99ee..570e431 100644 --- a/bin/cat/cat.c +++ b/bin/cat/cat.c @@ -64,6 +64,8 @@ __FBSDID("$FreeBSD$"); #include #include #include +#include +#include static int bflag, eflag, lflag, nflag, sflag, tflag, vflag; static int rval; @@ -207,6 +209,7 @@ static void cook_cat(FILE *fp) { int ch, gobble, line, prev; + wint_t wch; /* Reset EOF condition on stdin. */ if (fp == stdin && feof(stdin)) @@ -239,18 +242,40 @@ cook_cat(FILE *fp) continue; } } else if (vflag) { - if (!isascii(ch) && !isprint(ch)) { + (void)ungetc(ch, fp); + /* + * Our getwc(3) doesn't change file position + * on error. + */ + if ((wch = getwc(fp)) == WEOF) { + if (ferror(fp) && errno == EILSEQ) { + clearerr(fp); + /* Resync attempt. */ + memset(&fp->_mbstate, 0, sizeof(mbstate_t)); + if ((ch = getc(fp)) == EOF) + break; + wch = ch; + goto ilseq; + } else + break; + } + if (!iswascii(wch) && !iswprint(wch)) { +ilseq: if (putchar('M') == EOF || putchar('-') == EOF) break; - ch = toascii(ch); + wch = toascii(wch); } - if (iscntrl(ch)) { - if (putchar('^') == EOF || - putchar(ch == '\177' ? '?' : - ch | 0100) == EOF) + if (iswcntrl(wch)) { + ch = toascii(wch); + ch = (ch == '\177') ? '?' : (ch | 0100); + if (putchar('^') == EOF || putchar(ch) == EOF) break; continue; } + if (putwchar(wch) == WEOF) + break; + ch = -1; + continue; } if (putchar(ch) == EOF) break; -- cgit v1.1