=================================================================== RCS file: /cvs/mandoc/man.c,v retrieving revision 1.167 retrieving revision 1.182 diff -u -p -r1.167 -r1.182 --- mandoc/man.c 2017/01/10 13:47:00 1.167 +++ mandoc/man.c 2018/12/30 00:49:55 1.182 @@ -1,7 +1,7 @@ -/* $Id: man.c,v 1.167 2017/01/10 13:47:00 schwarze Exp $ */ +/* $Id: man.c,v 1.182 2018/12/30 00:49:55 schwarze Exp $ */ /* * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons - * Copyright (c) 2013, 2014, 2015 Ingo Schwarze + * Copyright (c) 2013,2014,2015,2017,2018 Ingo Schwarze * Copyright (c) 2011 Joerg Sonnenberger * * Permission to use, copy, modify, and distribute this software for any @@ -35,22 +35,7 @@ #include "roff_int.h" #include "libman.h" -const char *const __man_macronames[MAN_MAX] = { - "br", "TH", "SH", "SS", - "TP", "LP", "PP", "P", - "IP", "HP", "SM", "SB", - "BI", "IB", "BR", "RB", - "R", "B", "I", "IR", - "RI", "sp", "nf", - "fi", "RE", "RS", "DT", - "UC", "PD", "AT", "in", - "ft", "OP", "EX", "EE", - "UR", "UE", "ll" - }; - -const char * const *man_macronames = __man_macronames; - -static void man_descope(struct roff_man *, int, int); +static char *man_hasc(char *); static int man_ptext(struct roff_man *, int, char *, int); static int man_pmacro(struct roff_man *, int, char *, int); @@ -67,18 +52,43 @@ man_parseln(struct roff_man *man, int ln, char *buf, i man_ptext(man, ln, buf, offs); } -static void -man_descope(struct roff_man *man, int line, int offs) +/* + * If the string ends with \c, return a pointer to the backslash. + * Otherwise, return NULL. + */ +static char * +man_hasc(char *start) { + char *cp, *ep; + + ep = strchr(start, '\0') - 2; + if (ep < start || ep[0] != '\\' || ep[1] != 'c') + return NULL; + for (cp = ep; cp > start; cp--) + if (cp[-1] != '\\') + break; + return (ep - cp) % 2 ? NULL : ep; +} + +void +man_descope(struct roff_man *man, int line, int offs, char *start) +{ + /* Trailing \c keeps next-line scope open. */ + + if (start != NULL && man_hasc(start) != NULL) + return; + /* * Co-ordinate what happens with having a next-line scope open: - * first close out the element scope (if applicable), then close - * out the block scope (also if applicable). + * first close out the element scopes (if applicable), + * then close out the block scope (also if applicable). */ if (man->flags & MAN_ELINE) { + while (man->last->parent->type != ROFFT_ROOT && + man_macro(man->last->parent->tok)->flags & MAN_ESCOPED) + man_unscope(man, man->last->parent); man->flags &= ~MAN_ELINE; - man_unscope(man, man->last->parent); } if ( ! (man->flags & MAN_BLINE)) return; @@ -91,12 +101,13 @@ static int man_ptext(struct roff_man *man, int line, char *buf, int offs) { int i; + char *ep; /* Literal free-form text whitespace is preserved. */ if (man->flags & MAN_LITERAL) { roff_word_alloc(man, line, offs, buf + offs); - man_descope(man, line, offs); + man_descope(man, line, offs, buf + offs); return 1; } @@ -104,17 +115,25 @@ man_ptext(struct roff_man *man, int line, char *buf, i /* Skip leading whitespace. */ ; /* - * Blank lines are ignored right after headings + * Blank lines are ignored in next line scope + * and right after headings and cancel preceding \c, * but add a single vertical space elsewhere. */ if (buf[i] == '\0') { - /* Allocate a blank entry. */ - if (man->last->tok != MAN_SH && - man->last->tok != MAN_SS) { - roff_elem_alloc(man, line, offs, MAN_sp); - man->next = ROFF_NEXT_SIBLING; + if (man->flags & (MAN_ELINE | MAN_BLINE)) { + mandoc_msg(MANDOCERR_BLK_BLANK, line, 0, NULL); + return 1; } + if (man->last->tok == MAN_SH || man->last->tok == MAN_SS) + return 1; + if (man->last->type == ROFFT_TEXT && + ((ep = man_hasc(man->last->string)) != NULL)) { + *ep = '\0'; + return 1; + } + roff_elem_alloc(man, line, offs, ROFF_sp); + man->next = ROFF_NEXT_SIBLING; return 1; } @@ -128,8 +147,7 @@ man_ptext(struct roff_man *man, int line, char *buf, i if (' ' == buf[i - 1] || '\t' == buf[i - 1]) { if (i > 1 && '\\' != buf[i - 2]) - mandoc_msg(MANDOCERR_SPACE_EOL, man->parse, - line, i - 1, NULL); + mandoc_msg(MANDOCERR_SPACE_EOL, line, i - 1, NULL); for (--i; i && ' ' == buf[i]; i--) /* Spin back to non-space. */ ; @@ -151,7 +169,7 @@ man_ptext(struct roff_man *man, int line, char *buf, i if (mandoc_eos(buf, (size_t)i)) man->last->flags |= NODE_EOS; - man_descope(man, line, offs); + man_descope(man, line, offs, buf + offs); return 1; } @@ -160,29 +178,21 @@ man_pmacro(struct roff_man *man, int ln, char *buf, in { struct roff_node *n; const char *cp; - int tok; - int i, ppos; + size_t sz; + enum roff_tok tok; + int ppos; int bline; - char mac[5]; - ppos = offs; + /* Determine the line macro. */ - /* - * Copy the first word into a nil-terminated buffer. - * Stop when a space, tab, escape, or eoln is encountered. - */ - - i = 0; - while (i < 4 && strchr(" \t\\", buf[offs]) == NULL) - mac[i++] = buf[offs++]; - - mac[i] = '\0'; - - tok = (i > 0 && i < 4) ? man_hash_find(mac) : TOKEN_NONE; - + ppos = offs; + tok = TOKEN_NONE; + for (sz = 0; sz < 4 && strchr(" \t\\", buf[offs]) == NULL; sz++) + offs++; + if (sz > 0 && sz < 4) + tok = roffhash_find(man->manmac, buf + ppos, sz); if (tok == TOKEN_NONE) { - mandoc_msg(MANDOCERR_MACRO, man->parse, - ln, ppos, buf + ppos - 1); + mandoc_msg(MANDOCERR_MACRO, ln, ppos, "%s", buf + ppos - 1); return 1; } @@ -203,7 +213,7 @@ man_pmacro(struct roff_man *man, int ln, char *buf, in /* Jump to the next non-whitespace word. */ - while (buf[offs] && buf[offs] == ' ') + while (buf[offs] == ' ') offs++; /* @@ -212,8 +222,7 @@ man_pmacro(struct roff_man *man, int ln, char *buf, in */ if (buf[offs] == '\0' && buf[offs - 1] == ' ') - mandoc_msg(MANDOCERR_SPACE_EOL, man->parse, - ln, offs - 1, NULL); + mandoc_msg(MANDOCERR_SPACE_EOL, ln, offs - 1, NULL); /* * Some macros break next-line scopes; otherwise, remember @@ -223,10 +232,20 @@ man_pmacro(struct roff_man *man, int ln, char *buf, in man_breakscope(man, tok); bline = man->flags & MAN_BLINE; + /* + * If the line in next-line scope ends with \c, keep the + * next-line scope open for the subsequent input line. + * That is not at all portable, only groff >= 1.22.4 + * does it, but *if* this weird idiom occurs in a manual + * page, that's very likely what the author intended. + */ + + if (bline && man_hasc(buf + offs)) + bline = 0; + /* Call to handler... */ - assert(man_macros[tok].fp); - (*man_macros[tok].fp)(man, tok, ln, ppos, &offs, buf); + (*man_macro(tok)->fp)(man, tok, ln, ppos, &offs, buf); /* In quick mode (for mandocdb), abort after the NAME section. */ @@ -243,13 +262,13 @@ man_pmacro(struct roff_man *man, int ln, char *buf, in * unless the next-line scope is allowed to continue. */ - if ( ! bline || man->flags & MAN_ELINE || - man_macros[tok].flags & MAN_NSCOPED) + if (bline == 0 || + (man->flags & MAN_BLINE) == 0 || + man->flags & MAN_ELINE || + man_macro(tok)->flags & MAN_NSCOPED) return 1; - assert(man->flags & MAN_BLINE); man->flags &= ~MAN_BLINE; - man_unscope(man, man->last->parent); roff_body_alloc(man, ln, ppos, man->last->tok); return 1; @@ -266,17 +285,18 @@ man_breakscope(struct roff_man *man, int tok) * Delete the element that is being broken. */ - if (man->flags & MAN_ELINE && (tok == TOKEN_NONE || - ! (man_macros[tok].flags & MAN_NSCOPED))) { + if (man->flags & MAN_ELINE && (tok < MAN_TH || + (man_macro(tok)->flags & MAN_NSCOPED) == 0)) { n = man->last; - assert(n->type != ROFFT_TEXT); - if (man_macros[n->tok].flags & MAN_NSCOPED) + if (n->type == ROFFT_TEXT) n = n->parent; + if (n->tok < MAN_TH || + (man_macro(n->tok)->flags & (MAN_NSCOPED | MAN_ESCOPED)) + == MAN_NSCOPED) + n = n->parent; - mandoc_vmsg(MANDOCERR_BLK_LINE, man->parse, - n->line, n->pos, "%s breaks %s", - tok == TOKEN_NONE ? "TS" : man_macronames[tok], - man_macronames[n->tok]); + mandoc_msg(MANDOCERR_BLK_LINE, n->line, n->pos, + "%s breaks %s", roff_name[tok], roff_name[n->tok]); roff_node_delete(man, n); man->flags &= ~MAN_ELINE; @@ -302,37 +322,28 @@ man_breakscope(struct roff_man *man, int tok) * Delete the block that is being broken. */ - if (man->flags & MAN_BLINE && (tok == TOKEN_NONE || - man_macros[tok].flags & MAN_BSCOPE)) { + if (man->flags & MAN_BLINE && (tok < MAN_TH || + man_macro(tok)->flags & MAN_XSCOPE)) { n = man->last; if (n->type == ROFFT_TEXT) n = n->parent; - if ( ! (man_macros[n->tok].flags & MAN_BSCOPE)) + if (n->tok < MAN_TH || + (man_macro(n->tok)->flags & MAN_XSCOPE) == 0) n = n->parent; assert(n->type == ROFFT_HEAD); n = n->parent; assert(n->type == ROFFT_BLOCK); - assert(man_macros[n->tok].flags & MAN_SCOPED); + assert(man_macro(n->tok)->flags & MAN_BSCOPED); - mandoc_vmsg(MANDOCERR_BLK_LINE, man->parse, - n->line, n->pos, "%s breaks %s", - tok == TOKEN_NONE ? "TS" : man_macronames[tok], - man_macronames[n->tok]); + mandoc_msg(MANDOCERR_BLK_LINE, n->line, n->pos, + "%s breaks %s", roff_name[tok], roff_name[n->tok]); roff_node_delete(man, n); man->flags &= ~MAN_BLINE; } } -const struct mparse * -man_mparse(const struct roff_man *man) -{ - - assert(man && man->parse); - return man->parse; -} - void man_state(struct roff_man *man, struct roff_node *n) { @@ -341,16 +352,14 @@ man_state(struct roff_man *man, struct roff_node *n) case MAN_nf: case MAN_EX: if (man->flags & MAN_LITERAL && ! (n->flags & NODE_VALID)) - mandoc_msg(MANDOCERR_NF_SKIP, man->parse, - n->line, n->pos, "nf"); + mandoc_msg(MANDOCERR_NF_SKIP, n->line, n->pos, "nf"); man->flags |= MAN_LITERAL; break; case MAN_fi: case MAN_EE: if ( ! (man->flags & MAN_LITERAL) && ! (n->flags & NODE_VALID)) - mandoc_msg(MANDOCERR_FI_SKIP, man->parse, - n->line, n->pos, "fi"); + mandoc_msg(MANDOCERR_FI_SKIP, n->line, n->pos, "fi"); man->flags &= ~MAN_LITERAL; break; default: @@ -363,7 +372,7 @@ void man_validate(struct roff_man *man) { - man->last = man->first; + man->last = man->meta.first; man_node_validate(man); man->flags &= ~MAN_LITERAL; }