=================================================================== RCS file: /cvs/mandoc/roff.c,v retrieving revision 1.300 retrieving revision 1.304 diff -u -p -r1.300 -r1.304 --- mandoc/roff.c 2017/05/07 17:31:45 1.300 +++ mandoc/roff.c 2017/06/04 22:44:15 1.304 @@ -1,4 +1,4 @@ -/* $Id: roff.c,v 1.300 2017/05/07 17:31:45 schwarze Exp $ */ +/* $Id: roff.c,v 1.304 2017/06/04 22:44:15 schwarze Exp $ */ /* * Copyright (c) 2008-2012, 2014 Kristaps Dzonsons * Copyright (c) 2010-2015, 2017 Ingo Schwarze @@ -99,6 +99,7 @@ struct roff { int format; /* current file in mdoc or man format */ int argc; /* number of args of the last macro */ char control; /* control character */ + char escape; /* escape character */ }; struct roffnode { @@ -155,6 +156,8 @@ static enum rofferr roff_cond(ROFF_ARGS); static enum rofferr roff_cond_text(ROFF_ARGS); static enum rofferr roff_cond_sub(ROFF_ARGS); static enum rofferr roff_ds(ROFF_ARGS); +static enum rofferr roff_ec(ROFF_ARGS); +static enum rofferr roff_eo(ROFF_ARGS); static enum rofferr roff_eqndelim(struct roff *, struct buf *, int); static int roff_evalcond(struct roff *r, int, char *, int *); static int roff_evalnum(struct roff *, int, @@ -212,8 +215,8 @@ static enum rofferr roff_userdef(ROFF_ARGS); #define ROFFNUM_WHITE (1 << 1) /* Skip whitespace in roff_evalnum(). */ const char *__roff_name[MAN_MAX + 1] = { - "br", "ft", "ll", "sp", - "ta", NULL, + "br", "ft", "ll", "mc", + "sp", "ta", "ti", NULL, "ab", "ad", "af", "aln", "als", "am", "am1", "ami", "ami1", "as", "as1", "asciify", @@ -246,7 +249,7 @@ const char *__roff_name[MAN_MAX + 1] = { "lc", "lc_ctype", "lds", "length", "letadj", "lf", "lg", "lhang", "linetabs", "lnr", "lnrf", "lpfx", - "ls", "lsm", "lt", "mc", + "ls", "lsm", "lt", "mediasize", "minss", "mk", "mso", "na", "ne", "nh", "nhychar", "nm", "nn", "nop", "nr", @@ -264,7 +267,7 @@ const char *__roff_name[MAN_MAX + 1] = { "spacewidth", "special", "spreadwarn", "ss", "sty", "substring", "sv", "sy", "T&", "tc", "TE", - "TH", "ti", "tkf", "tl", + "TH", "tkf", "tl", "tm", "tm1", "tmc", "tr", "track", "transchar", "trf", "trimat", "trin", "trnt", "troff", "TS", @@ -322,8 +325,10 @@ static struct roffmac roffs[TOKEN_NONE] = { { roff_br, NULL, NULL, 0 }, /* br */ { roff_onearg, NULL, NULL, 0 }, /* ft */ { roff_onearg, NULL, NULL, 0 }, /* ll */ + { roff_onearg, NULL, NULL, 0 }, /* mc */ { roff_onearg, NULL, NULL, 0 }, /* sp */ { roff_manyarg, NULL, NULL, 0 }, /* ta */ + { roff_onearg, NULL, NULL, 0 }, /* ti */ { NULL, NULL, NULL, 0 }, /* ROFF_MAX */ { roff_unsupp, NULL, NULL, 0 }, /* ab */ { roff_line_ignore, NULL, NULL, 0 }, /* ad */ @@ -384,13 +389,13 @@ static struct roffmac roffs[TOKEN_NONE] = { { roff_ds, NULL, NULL, 0 }, /* ds1 */ { roff_unsupp, NULL, NULL, 0 }, /* dwh */ { roff_unsupp, NULL, NULL, 0 }, /* dt */ - { roff_unsupp, NULL, NULL, 0 }, /* ec */ + { roff_ec, NULL, NULL, 0 }, /* ec */ { roff_unsupp, NULL, NULL, 0 }, /* ecr */ { roff_unsupp, NULL, NULL, 0 }, /* ecs */ { roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT }, /* el */ { roff_unsupp, NULL, NULL, 0 }, /* em */ { roff_EN, NULL, NULL, 0 }, /* EN */ - { roff_unsupp, NULL, NULL, 0 }, /* eo */ + { roff_eo, NULL, NULL, 0 }, /* eo */ { roff_unsupp, NULL, NULL, 0 }, /* EP */ { roff_EQ, NULL, NULL, 0 }, /* EQ */ { roff_line_ignore, NULL, NULL, 0 }, /* errprint */ @@ -456,7 +461,6 @@ static struct roffmac roffs[TOKEN_NONE] = { { roff_line_ignore, NULL, NULL, 0 }, /* ls */ { roff_unsupp, NULL, NULL, 0 }, /* lsm */ { roff_line_ignore, NULL, NULL, 0 }, /* lt */ - { roff_line_ignore, NULL, NULL, 0 }, /* mc */ { roff_line_ignore, NULL, NULL, 0 }, /* mediasize */ { roff_line_ignore, NULL, NULL, 0 }, /* minss */ { roff_line_ignore, NULL, NULL, 0 }, /* mk */ @@ -525,7 +529,6 @@ static struct roffmac roffs[TOKEN_NONE] = { { roff_unsupp, NULL, NULL, 0 }, /* tc */ { roff_TE, NULL, NULL, 0 }, /* TE */ { roff_TH, NULL, NULL, 0 }, /* TH */ - { roff_unsupp, NULL, NULL, 0 }, /* ti */ { roff_line_ignore, NULL, NULL, 0 }, /* tkf */ { roff_unsupp, NULL, NULL, 0 }, /* tl */ { roff_line_ignore, NULL, NULL, 0 }, /* tm */ @@ -751,7 +754,8 @@ roff_reset(struct roff *r) { roff_free1(r); r->format = r->options & (MPARSE_MDOC | MPARSE_MAN); - r->control = 0; + r->control = '\0'; + r->escape = '\\'; } void @@ -773,6 +777,7 @@ roff_alloc(struct mparse *parse, int options) r->options = options; r->format = options & (MPARSE_MDOC | MPARSE_MAN); r->rstackpos = -1; + r->escape = '\\'; return r; } @@ -1009,7 +1014,7 @@ roff_addtbl(struct roff_man *man, const struct tbl_spa struct roff_node *n; if (man->macroset == MACROSET_MAN) - man_breakscope(man, TOKEN_NONE); + man_breakscope(man, ROFF_TS); n = roff_node_alloc(man, tbl->line, 0, ROFFT_TBL, TOKEN_NONE); n->span = tbl; roff_node_append(man, n); @@ -1149,27 +1154,80 @@ roff_res(struct roff *r, struct buf *buf, int ln, int int expand_count; /* to avoid infinite loops */ int npos; /* position in numeric expression */ int arg_complete; /* argument not interrupted by eol */ + int done; /* no more input available */ char term; /* character terminating the escape */ - expand_count = 0; + /* Search forward for comments. */ + + done = 0; start = buf->buf + pos; - stesc = strchr(start, '\0') - 1; - while (stesc-- > start) { + for (stesc = buf->buf + pos; *stesc != '\0'; stesc++) { + if (stesc[0] != r->escape || stesc[1] == '\0') + continue; + stesc++; + if (*stesc != '"' && *stesc != '#') + continue; + cp = strchr(stesc--, '\0') - 1; + if (*cp == '\n') { + done = 1; + cp--; + } + if (*cp == ' ' || *cp == '\t') + mandoc_msg(MANDOCERR_SPACE_EOL, r->parse, + ln, cp - buf->buf, NULL); + while (stesc > start && stesc[-1] == ' ') + stesc--; + *stesc = '\0'; + break; + } + if (stesc == start) + return ROFF_CONT; + stesc--; + /* Notice the end of the input. */ + + if (*stesc == '\n') { + *stesc-- = '\0'; + done = 1; + } + + expand_count = 0; + while (stesc >= start) { + /* Search backwards for the next backslash. */ - if (*stesc != '\\') + if (*stesc != r->escape) { + if (*stesc == '\\') { + *stesc = '\0'; + buf->sz = mandoc_asprintf(&nbuf, "%s\\e%s", + buf->buf, stesc + 1) + 1; + start = nbuf + pos; + stesc = nbuf + (stesc - buf->buf); + free(buf->buf); + buf->buf = nbuf; + } + stesc--; continue; + } /* If it is escaped, skip it. */ for (cp = stesc - 1; cp >= start; cp--) - if (*cp != '\\') + if (*cp != r->escape) break; if ((stesc - cp) % 2 == 0) { - stesc = (char *)cp; + while (stesc > cp) + *stesc-- = '\\'; continue; + } else if (stesc[1] != '\0') { + *stesc = '\\'; + } else { + *stesc-- = '\0'; + if (done) + continue; + else + return ROFF_APPEND; } /* Decide whether to expand or to check only. */ @@ -1195,6 +1253,7 @@ roff_res(struct roff *r, struct buf *buf, int ln, int mandoc_vmsg(MANDOCERR_ESC_BAD, r->parse, ln, (int)(stesc - buf->buf), "%.*s", (int)(cp - stesc), stesc); + stesc--; continue; } @@ -1409,7 +1468,7 @@ roff_parseln(struct roff *r, int ln, struct buf *buf, /* Expand some escape sequences. */ e = roff_res(r, buf, ln, pos); - if (e == ROFF_IGN) + if (e == ROFF_IGN || e == ROFF_APPEND) return e; assert(e == ROFF_CONT); @@ -2778,6 +2837,10 @@ roff_onearg(ROFF_ARGS) struct roff_node *n; char *cp; + if (r->man->flags & (MAN_BLINE | MAN_ELINE) && + (tok == ROFF_sp || tok == ROFF_ti)) + man_breakscope(r->man, tok); + roff_elem_alloc(r->man, ln, ppos, tok); n = r->man->last; @@ -2826,6 +2889,8 @@ roff_manyarg(ROFF_ARGS) static enum rofferr roff_br(ROFF_ARGS) { + if (r->man->flags & (MAN_BLINE | MAN_ELINE)) + man_breakscope(r->man, ROFF_br); roff_elem_alloc(r->man, ln, ppos, ROFF_br); if (buf->buf[pos] != '\0') mandoc_vmsg(MANDOCERR_ARG_SKIP, r->parse, ln, pos, @@ -2843,7 +2908,7 @@ roff_cc(ROFF_ARGS) p = buf->buf + pos; if (*p == '\0' || (r->control = *p++) == '.') - r->control = 0; + r->control = '\0'; if (*p != '\0') mandoc_vmsg(MANDOCERR_ARG_EXCESS, r->parse, @@ -2853,6 +2918,33 @@ roff_cc(ROFF_ARGS) } static enum rofferr +roff_ec(ROFF_ARGS) +{ + const char *p; + + p = buf->buf + pos; + if (*p == '\0') + r->escape = '\\'; + else { + r->escape = *p; + if (*++p != '\0') + mandoc_vmsg(MANDOCERR_ARG_EXCESS, r->parse, + ln, p - buf->buf, "ec ... %s", p); + } + return ROFF_IGN; +} + +static enum rofferr +roff_eo(ROFF_ARGS) +{ + r->escape = '\0'; + if (buf->buf[pos] != '\0') + mandoc_vmsg(MANDOCERR_ARG_SKIP, r->parse, + ln, pos, "eo %s", buf->buf + pos); + return ROFF_IGN; +} + +static enum rofferr roff_tr(ROFF_ARGS) { const char *p, *first, *second; @@ -3379,9 +3471,9 @@ roff_getcontrol(const struct roff *r, const char *cp, pos = *ppos; - if (0 != r->control && cp[pos] == r->control) + if (r->control != '\0' && cp[pos] == r->control) pos++; - else if (0 != r->control) + else if (r->control != '\0') return 0; else if ('\\' == cp[pos] && '.' == cp[pos + 1]) pos += 2;