=================================================================== RCS file: /cvs/mandoc/roff.c,v retrieving revision 1.214 retrieving revision 1.229 diff -u -p -r1.214 -r1.229 --- mandoc/roff.c 2014/06/29 23:26:00 1.214 +++ mandoc/roff.c 2014/09/07 00:21:53 1.229 @@ -1,4 +1,4 @@ -/* $Id: roff.c,v 1.214 2014/06/29 23:26:00 schwarze Exp $ */ +/* $Id: roff.c,v 1.229 2014/09/07 00:21:53 schwarze Exp $ */ /* * Copyright (c) 2010, 2011, 2012 Kristaps Dzonsons * Copyright (c) 2010-2014 Ingo Schwarze @@ -15,10 +15,10 @@ * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. */ -#ifdef HAVE_CONFIG_H #include "config.h" -#endif +#include + #include #include #include @@ -27,8 +27,8 @@ #include "mandoc.h" #include "mandoc_aux.h" -#include "libroff.h" #include "libmandoc.h" +#include "libroff.h" /* Maximum number of nested if-else conditionals. */ #define RSTACK_MAX 128 @@ -60,6 +60,7 @@ enum rofft { ROFF_nh, ROFF_nr, ROFF_ns, + ROFF_pl, ROFF_ps, ROFF_rm, ROFF_rr, @@ -106,11 +107,8 @@ struct roffreg { struct roff { struct mparse *parse; /* parse point */ - int options; /* parse options */ struct roffnode *last; /* leaf of stack */ - int rstack[RSTACK_MAX]; /* stack of !`ie' rules */ - char control; /* control character */ - int rstackpos; /* position in rstack */ + int *rstack; /* stack of inverted `ie' values */ struct roffreg *regtab; /* number registers */ struct roffkv *strtab; /* user-defined strings & macros */ struct roffkv *xmbtab; /* multi-byte trans table (`tr') */ @@ -122,6 +120,11 @@ struct roff { struct eqn_node *last_eqn; /* last equation parsed */ struct eqn_node *first_eqn; /* first equation parsed */ struct eqn_node *eqn; /* current equation being parsed */ + int options; /* parse options */ + int rstacksz; /* current size limit of rstack */ + int rstackpos; /* position in rstack */ + int format; /* current file in mdoc or man format */ + char control; /* control character */ }; struct roffnode { @@ -254,6 +257,7 @@ static struct roffmac roffs[ROFF_MAX] = { { "nh", roff_line_ignore, NULL, NULL, 0, NULL }, { "nr", roff_nr, NULL, NULL, 0, NULL }, { "ns", roff_line_ignore, NULL, NULL, 0, NULL }, + { "pl", roff_line_ignore, NULL, NULL, 0, NULL }, { "ps", roff_line_ignore, NULL, NULL, 0, NULL }, { "rm", roff_rm, NULL, NULL, 0, NULL }, { "rr", roff_rr, NULL, NULL, 0, NULL }, @@ -420,32 +424,32 @@ roff_free1(struct roff *r) r->first_tbl = tbl->next; tbl_free(tbl); } - r->first_tbl = r->last_tbl = r->tbl = NULL; while (NULL != (e = r->first_eqn)) { r->first_eqn = e->next; eqn_free(e); } - r->first_eqn = r->last_eqn = r->eqn = NULL; while (r->last) roffnode_pop(r); - roff_freestr(r->strtab); - roff_freestr(r->xmbtab); + free (r->rstack); + r->rstack = NULL; + r->rstacksz = 0; + r->rstackpos = -1; - r->strtab = r->xmbtab = NULL; - roff_freereg(r->regtab); - r->regtab = NULL; + roff_freestr(r->strtab); + roff_freestr(r->xmbtab); + r->strtab = r->xmbtab = NULL; + if (r->xtab) for (i = 0; i < 128; i++) free(r->xtab[i].p); - free(r->xtab); r->xtab = NULL; } @@ -455,6 +459,7 @@ roff_reset(struct roff *r) { roff_free1(r); + r->format = r->options & (MPARSE_MDOC | MPARSE_MAN); r->control = 0; } @@ -474,6 +479,7 @@ roff_alloc(struct mparse *parse, int options) r = mandoc_calloc(1, sizeof(struct roff)); r->parse = parse; r->options = options; + r->format = options & (MPARSE_MDOC | MPARSE_MAN); r->rstackpos = -1; roffhash_init(); @@ -500,7 +506,7 @@ roff_res(struct roff *r, char **bufp, size_t *szp, int size_t naml; /* actual length of the escape name */ int expand_count; /* to avoid infinite loops */ int npos; /* position in numeric expression */ - int irc; /* return code from roff_evalnum() */ + int arg_complete; /* argument not interrupted by eol */ char term; /* character terminating the escape */ expand_count = 0; @@ -542,8 +548,9 @@ roff_res(struct roff *r, char **bufp, size_t *szp, int break; default: if (ESCAPE_ERROR == mandoc_escape(&cp, NULL, NULL)) - mandoc_msg(MANDOCERR_BADESCAPE, r->parse, - ln, (int)(stesc - *bufp), NULL); + mandoc_vmsg(MANDOCERR_ESC_BAD, + r->parse, ln, (int)(stesc - *bufp), + "%.*s", (int)(cp - stesc), stesc); continue; } @@ -585,10 +592,12 @@ roff_res(struct roff *r, char **bufp, size_t *szp, int /* Advance to the end of the name. */ + arg_complete = 1; for (naml = 0; 0 == maxl || naml < maxl; naml++, cp++) { if ('\0' == *cp) { - mandoc_msg(MANDOCERR_BADESCAPE, r->parse, - ln, (int)(stesc - *bufp), NULL); + mandoc_msg(MANDOCERR_ESC_BAD, r->parse, + ln, (int)(stesc - *bufp), stesc); + arg_complete = 0; break; } if (0 == maxl && *cp == term) { @@ -604,28 +613,34 @@ roff_res(struct roff *r, char **bufp, size_t *szp, int switch (stesc[1]) { case '*': - res = roff_getstrn(r, stnam, naml); + if (arg_complete) + res = roff_getstrn(r, stnam, naml); break; case 'B': npos = 0; - irc = roff_evalnum(stnam, &npos, NULL, 0); - ubuf[0] = irc && stnam + npos + 1 == cp - ? '1' : '0'; + ubuf[0] = arg_complete && + roff_evalnum(stnam, &npos, NULL, 0) && + stnam + npos + 1 == cp ? '1' : '0'; ubuf[1] = '\0'; break; case 'n': - (void)snprintf(ubuf, sizeof(ubuf), "%d", - roff_getregn(r, stnam, naml)); + if (arg_complete) + (void)snprintf(ubuf, sizeof(ubuf), "%d", + roff_getregn(r, stnam, naml)); + else + ubuf[0] = '\0'; break; case 'w': + /* use even incomplete args */ (void)snprintf(ubuf, sizeof(ubuf), "%d", 24 * (int)naml); break; } if (NULL == res) { - mandoc_msg(MANDOCERR_BADESCAPE, r->parse, - ln, (int)(stesc - *bufp), NULL); + mandoc_vmsg(MANDOCERR_STR_UNDEF, + r->parse, ln, (int)(stesc - *bufp), + "%.*s", (int)naml, stnam); res = ""; } @@ -746,6 +761,15 @@ roff_parseln(struct roff *r, int ln, char **bufp, return(roff_parsetext(bufp, szp, pos, offs)); } + /* Skip empty request lines. */ + + if ((*bufp)[pos] == '"') { + mandoc_msg(MANDOCERR_COMMENT_BAD, r->parse, + ln, pos, NULL); + return(ROFF_IGN); + } else if ((*bufp)[pos] == '\0') + return(ROFF_IGN); + /* * If a scope is open, go to the child handler for that macro, * as it may want to preprocess before doing anything with it. @@ -777,18 +801,19 @@ roff_endparse(struct roff *r) { if (r->last) - mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse, - r->last->line, r->last->col, NULL); + mandoc_msg(MANDOCERR_BLK_NOEND, r->parse, + r->last->line, r->last->col, + roffs[r->last->tok].name); if (r->eqn) { - mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse, - r->eqn->eqn.ln, r->eqn->eqn.pos, NULL); + mandoc_msg(MANDOCERR_BLK_NOEND, r->parse, + r->eqn->eqn.ln, r->eqn->eqn.pos, "EQ"); eqn_end(&r->eqn); } if (r->tbl) { - mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse, - r->tbl->line, r->tbl->pos, NULL); + mandoc_msg(MANDOCERR_BLK_NOEND, r->parse, + r->tbl->line, r->tbl->pos, "TS"); tbl_end(&r->tbl); } } @@ -832,17 +857,17 @@ roff_cblock(ROFF_ARGS) */ if (NULL == r->last) { - mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL); + mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse, + ln, ppos, ".."); return(ROFF_IGN); } switch (r->last->tok) { case ROFF_am: + /* ROFF_am1 is remapped to ROFF_am in roff_block(). */ /* FALLTHROUGH */ case ROFF_ami: /* FALLTHROUGH */ - case ROFF_am1: - /* FALLTHROUGH */ case ROFF_de: /* ROFF_de1 is remapped to ROFF_de in roff_block(). */ /* FALLTHROUGH */ @@ -851,12 +876,14 @@ roff_cblock(ROFF_ARGS) case ROFF_ig: break; default: - mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL); + mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse, + ln, ppos, ".."); return(ROFF_IGN); } if ((*bufp)[pos]) - mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL); + mandoc_vmsg(MANDOCERR_ARG_SKIP, r->parse, ln, pos, + ".. %s", *bufp + pos); roffnode_pop(r); roffnode_cleanscope(r); @@ -880,7 +907,8 @@ roff_ccond(struct roff *r, int ln, int ppos) { if (NULL == r->last) { - mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL); + mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse, + ln, ppos, "\\}"); return; } @@ -892,12 +920,14 @@ roff_ccond(struct roff *r, int ln, int ppos) case ROFF_if: break; default: - mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL); + mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse, + ln, ppos, "\\}"); return; } if (r->last->endspan > -1) { - mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL); + mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse, + ln, ppos, "\\}"); return; } @@ -909,34 +939,48 @@ roff_ccond(struct roff *r, int ln, int ppos) static enum rofferr roff_block(ROFF_ARGS) { - char *name, *cp; + const char *name; + char *iname, *cp; size_t namesz; - name = cp = *bufp + pos; - namesz = 0; + /* Ignore groff compatibility mode for now. */ - if (ROFF_ig != tok) { - if ('\0' == *cp) { - mandoc_msg(MANDOCERR_NOARGS, r->parse, ln, ppos, NULL); - return(ROFF_IGN); - } + if (ROFF_de1 == tok) + tok = ROFF_de; + else if (ROFF_am1 == tok) + tok = ROFF_am; - /* - * Re-write `de1', since we don't really care about - * groff's strange compatibility mode, into `de'. - */ + /* Parse the macro name argument. */ - if (ROFF_de1 == tok) - tok = ROFF_de; - else if (ROFF_de != tok) - mandoc_msg(MANDOCERR_REQUEST, r->parse, ln, ppos, - roffs[tok].name); - + cp = *bufp + pos; + if (ROFF_ig == tok) { + iname = NULL; + namesz = 0; + } else { + iname = cp; namesz = roff_getname(r, &cp, ln, ppos); - name[namesz] = '\0'; + iname[namesz] = '\0'; + } + + /* Resolve the macro name argument if it is indirect. */ + + if (namesz && (ROFF_dei == tok || ROFF_ami == tok)) { + if (NULL == (name = roff_getstrn(r, iname, namesz))) { + mandoc_vmsg(MANDOCERR_STR_UNDEF, + r->parse, ln, (int)(iname - *bufp), + "%.*s", (int)namesz, iname); + namesz = 0; + } else + namesz = strlen(name); } else - name = NULL; + name = iname; + if (0 == namesz && ROFF_ig != tok) { + mandoc_msg(MANDOCERR_REQ_EMPTY, r->parse, + ln, ppos, roffs[tok].name); + return(ROFF_IGN); + } + roffnode_push(r, tok, name, ln, ppos); /* @@ -945,21 +989,36 @@ roff_block(ROFF_ARGS) * appended from roff_block_text() in multiline mode. */ - if (namesz && ROFF_de == tok) + if (ROFF_de == tok || ROFF_dei == tok) roff_setstrn(&r->strtab, name, namesz, "", 0, 0); if ('\0' == *cp) return(ROFF_IGN); - /* If present, process the custom end-of-line marker. */ + /* Get the custom end marker. */ - name = cp; + iname = cp; namesz = roff_getname(r, &cp, ln, ppos); + + /* Resolve the end marker if it is indirect. */ + + if (namesz && (ROFF_dei == tok || ROFF_ami == tok)) { + if (NULL == (name = roff_getstrn(r, iname, namesz))) { + mandoc_vmsg(MANDOCERR_STR_UNDEF, + r->parse, ln, (int)(iname - *bufp), + "%.*s", (int)namesz, iname); + namesz = 0; + } else + namesz = strlen(name); + } else + name = iname; + if (namesz) r->last->end = mandoc_strndup(name, namesz); if ('\0' != *cp) - mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL); + mandoc_vmsg(MANDOCERR_ARG_EXCESS, r->parse, + ln, pos, ".%s ... %s", roffs[tok].name, cp); return(ROFF_IGN); } @@ -1008,12 +1067,8 @@ roff_block_sub(ROFF_ARGS) t = roff_parse(r, *bufp, &pos, ln, ppos); - /* - * Macros other than block-end are only significant - * in `de' blocks; elsewhere, simply throw them away. - */ if (ROFF_cblock != t) { - if (ROFF_de == tok) + if (ROFF_ig != tok) roff_setstr(r, r->last->name, *bufp + ppos, 2); return(ROFF_IGN); } @@ -1026,7 +1081,7 @@ static enum rofferr roff_block_text(ROFF_ARGS) { - if (ROFF_de == tok) + if (ROFF_ig != tok) roff_setstr(r, r->last->name, *bufp + pos, 2); return(ROFF_IGN); @@ -1243,10 +1298,10 @@ roff_cond(ROFF_ARGS) */ if (ROFF_ie == tok) { - if (r->rstackpos == RSTACK_MAX - 1) { - mandoc_msg(MANDOCERR_MEM, - r->parse, ln, ppos, NULL); - return(ROFF_ERR); + if (r->rstackpos + 1 == r->rstacksz) { + r->rstacksz += 16; + r->rstack = mandoc_reallocarray(r->rstack, + r->rstacksz, sizeof(int)); } r->rstack[++r->rstackpos] = !r->last->rule; } @@ -1285,7 +1340,8 @@ roff_cond(ROFF_ARGS) */ if ('\0' == (*bufp)[pos]) - mandoc_msg(MANDOCERR_NOARGS, r->parse, ln, ppos, NULL); + mandoc_msg(MANDOCERR_COND_EMPTY, r->parse, + ln, ppos, roffs[tok].name); r->last->endspan = 1; @@ -1717,7 +1773,7 @@ roff_it(ROFF_ARGS) len = strcspn(cp, " \t"); cp[len] = '\0'; if ((iv = mandoc_strntoi(cp, len, 10)) <= 0) { - mandoc_msg(MANDOCERR_NUMERIC, r->parse, + mandoc_msg(MANDOCERR_IT_NONUM, r->parse, ln, ppos, *bufp + 1); return(ROFF_IGN); } @@ -1734,10 +1790,13 @@ roff_Dd(ROFF_ARGS) { const char *const *cp; - if (0 == ((MPARSE_MDOC | MPARSE_QUICK) & r->options)) + if ((r->options & (MPARSE_MDOC | MPARSE_QUICK)) == 0) for (cp = __mdoc_reserved; *cp; cp++) roff_setstr(r, *cp, NULL, 0); + if (r->format == 0) + r->format = MPARSE_MDOC; + return(ROFF_CONT); } @@ -1746,10 +1805,13 @@ roff_TH(ROFF_ARGS) { const char *const *cp; - if (0 == (MPARSE_QUICK & r->options)) + if ((r->options & MPARSE_QUICK) == 0) for (cp = __man_reserved; *cp; cp++) roff_setstr(r, *cp, NULL, 0); + if (r->format == 0) + r->format = MPARSE_MAN; + return(ROFF_CONT); } @@ -1758,7 +1820,8 @@ roff_TE(ROFF_ARGS) { if (NULL == r->tbl) - mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL); + mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse, + ln, ppos, "TE"); else tbl_end(&r->tbl); @@ -1770,7 +1833,8 @@ roff_T_(ROFF_ARGS) { if (NULL == r->tbl) - mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL); + mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse, + ln, ppos, "T&"); else tbl_restart(ppos, ln, r->tbl); @@ -1821,7 +1885,7 @@ static enum rofferr roff_EN(ROFF_ARGS) { - mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL); + mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse, ln, ppos, "EN"); return(ROFF_IGN); } @@ -1831,7 +1895,8 @@ roff_TS(ROFF_ARGS) struct tbl_node *tbl; if (r->tbl) { - mandoc_msg(MANDOCERR_SCOPEBROKEN, r->parse, ln, ppos, NULL); + mandoc_msg(MANDOCERR_BLK_BROKEN, r->parse, + ln, ppos, "TS breaks TS"); tbl_end(&r->tbl); } @@ -1883,9 +1948,8 @@ roff_tr(ROFF_ARGS) if ('\\' == *first) { esc = mandoc_escape(&p, NULL, NULL); if (ESCAPE_ERROR == esc) { - mandoc_msg(MANDOCERR_BADESCAPE, - r->parse, ln, - (int)(p - *bufp), NULL); + mandoc_msg(MANDOCERR_ESC_BAD, r->parse, + ln, (int)(p - *bufp), first); return(ROFF_IGN); } fsz = (size_t)(p - first); @@ -1895,9 +1959,8 @@ roff_tr(ROFF_ARGS) if ('\\' == *second) { esc = mandoc_escape(&p, NULL, NULL); if (ESCAPE_ERROR == esc) { - mandoc_msg(MANDOCERR_BADESCAPE, - r->parse, ln, - (int)(p - *bufp), NULL); + mandoc_msg(MANDOCERR_ESC_BAD, r->parse, + ln, (int)(p - *bufp), second); return(ROFF_IGN); } ssz = (size_t)(p - second); @@ -1932,7 +1995,7 @@ roff_so(ROFF_ARGS) char *name; name = *bufp + pos; - mandoc_vmsg(MANDOCERR_SO, r->parse, ln, ppos, ".so %s", name); + mandoc_vmsg(MANDOCERR_SO, r->parse, ln, ppos, "so %s", name); /* * Handle `so'. Be EXTREMELY careful, as we shouldn't be @@ -2018,11 +2081,14 @@ roff_getname(struct roff *r, char **cpp, int ln, int p } if ('\\' != *cp) continue; + namesz = cp - name; + if ('{' == cp[1] || '}' == cp[1]) + break; cp++; if ('\\' == *cp) continue; - namesz = cp - name - 1; - mandoc_msg(MANDOCERR_NAMESC, r->parse, ln, pos, NULL); + mandoc_vmsg(MANDOCERR_NAMESC, r->parse, ln, pos, + "%.*s", (int)(cp - name + 1), name); mandoc_escape((const char **)&cp, NULL, NULL); break; } @@ -2259,6 +2325,13 @@ roff_strdup(const struct roff *r, const char *p) res[(int)ssz] = '\0'; return(res); +} + +int +roff_getformat(const struct roff *r) +{ + + return(r->format); } /*