=================================================================== RCS file: /cvs/mandoc/man.c,v retrieving revision 1.157 retrieving revision 1.186 diff -u -p -r1.157 -r1.186 --- mandoc/man.c 2015/04/18 17:53:21 1.157 +++ mandoc/man.c 2018/12/31 10:04:39 1.186 @@ -1,7 +1,7 @@ -/* $Id: man.c,v 1.157 2015/04/18 17:53:21 schwarze Exp $ */ +/* $Id: man.c,v 1.186 2018/12/31 10:04:39 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 @@ -32,43 +32,14 @@ #include "roff.h" #include "man.h" #include "libmandoc.h" +#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_breakscope(struct roff_man *, int); -static void man_descope(struct roff_man *, int, int); -static struct roff_node *man_node_alloc(struct roff_man *, int, int, - enum roff_type, int); -static void man_node_append(struct roff_man *, - struct roff_node *); -static void man_node_free(struct roff_node *); -static void man_node_unlink(struct roff_man *, - struct roff_node *); +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); -void -man_endparse(struct roff_man *man) -{ - - man_macroend(man); -} - int man_parseln(struct roff_man *man, int ln, char *buf, int offs) { @@ -76,250 +47,94 @@ man_parseln(struct roff_man *man, int ln, char *buf, i if (man->last->type != ROFFT_EQN || ln > man->last->line) man->flags |= MAN_NEWLINE; - return (roff_getcontrol(man->roff, buf, &offs) ? + return roff_getcontrol(man->roff, buf, &offs) ? man_pmacro(man, ln, buf, offs) : - man_ptext(man, ln, buf, offs)); + man_ptext(man, ln, buf, offs); } -static void -man_node_append(struct roff_man *man, struct roff_node *p) -{ - - assert(man->last); - assert(man->first); - assert(p->type != ROFFT_ROOT); - - switch (man->next) { - case ROFF_NEXT_SIBLING: - man->last->next = p; - p->prev = man->last; - p->parent = man->last->parent; - break; - case ROFF_NEXT_CHILD: - man->last->child = p; - p->parent = man->last; - break; - default: - abort(); - /* NOTREACHED */ - } - - assert(p->parent); - p->parent->nchild++; - - switch (p->type) { - case ROFFT_BLOCK: - if (p->tok == MAN_SH || p->tok == MAN_SS) - man->flags &= ~MAN_LITERAL; - break; - case ROFFT_HEAD: - assert(p->parent->type == ROFFT_BLOCK); - p->parent->head = p; - break; - case ROFFT_BODY: - assert(p->parent->type == ROFFT_BLOCK); - p->parent->body = p; - break; - default: - break; - } - - man->last = p; - - switch (p->type) { - case ROFFT_TBL: - /* FALLTHROUGH */ - case ROFFT_TEXT: - man_valid_post(man); - break; - default: - break; - } -} - -static struct roff_node * -man_node_alloc(struct roff_man *man, int line, int pos, - enum roff_type type, int tok) -{ - struct roff_node *p; - - p = mandoc_calloc(1, sizeof(*p)); - p->line = line; - p->pos = pos; - p->type = type; - p->tok = tok; - - if (man->flags & MAN_NEWLINE) - p->flags |= MAN_LINE; - man->flags &= ~MAN_NEWLINE; - return(p); -} - -void -man_elem_alloc(struct roff_man *man, int line, int pos, int tok) -{ - struct roff_node *p; - - p = man_node_alloc(man, line, pos, ROFFT_ELEM, tok); - man_node_append(man, p); - man->next = ROFF_NEXT_CHILD; -} - -void -man_head_alloc(struct roff_man *man, int line, int pos, int tok) -{ - struct roff_node *p; - - p = man_node_alloc(man, line, pos, ROFFT_HEAD, tok); - man_node_append(man, p); - man->next = ROFF_NEXT_CHILD; -} - -void -man_body_alloc(struct roff_man *man, int line, int pos, int tok) -{ - struct roff_node *p; - - p = man_node_alloc(man, line, pos, ROFFT_BODY, tok); - man_node_append(man, p); - man->next = ROFF_NEXT_CHILD; -} - -void -man_block_alloc(struct roff_man *man, int line, int pos, int tok) -{ - struct roff_node *p; - - p = man_node_alloc(man, line, pos, ROFFT_BLOCK, tok); - man_node_append(man, p); - man->next = ROFF_NEXT_CHILD; -} - -void -man_word_alloc(struct roff_man *man, int line, int pos, const char *word) -{ - struct roff_node *n; - - n = man_node_alloc(man, line, pos, ROFFT_TEXT, MAN_MAX); - n->string = roff_strdup(man->roff, word); - man_node_append(man, n); - man->next = ROFF_NEXT_SIBLING; -} - -void -man_word_append(struct roff_man *man, const char *word) -{ - struct roff_node *n; - char *addstr, *newstr; - - n = man->last; - addstr = roff_strdup(man->roff, word); - mandoc_asprintf(&newstr, "%s %s", n->string, addstr); - free(addstr); - free(n->string); - n->string = newstr; - man->next = ROFF_NEXT_SIBLING; -} - /* - * Free all of the resources held by a node. This does NOT unlink a - * node from its context; for that, see man_node_unlink(). + * If the string ends with \c, return a pointer to the backslash. + * Otherwise, return NULL. */ -static void -man_node_free(struct roff_node *p) +static char * +man_hasc(char *start) { + char *cp, *ep; - free(p->string); - free(p); + 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_node_delete(struct roff_man *man, struct roff_node *p) +man_descope(struct roff_man *man, int line, int offs, char *start) { + /* Trailing \c keeps next-line scope open. */ - while (p->child) - man_node_delete(man, p->child); + if (start != NULL && man_hasc(start) != NULL) + return; - man_node_unlink(man, p); - man_node_free(p); -} - -void -man_addeqn(struct roff_man *man, const struct eqn *ep) -{ - struct roff_node *n; - - n = man_node_alloc(man, ep->ln, ep->pos, ROFFT_EQN, MAN_MAX); - n->eqn = ep; - if (ep->ln > man->last->line) - n->flags |= MAN_LINE; - man_node_append(man, n); - man->next = ROFF_NEXT_SIBLING; - man_descope(man, ep->ln, ep->pos); -} - -void -man_addspan(struct roff_man *man, const struct tbl_span *sp) -{ - struct roff_node *n; - - man_breakscope(man, MAN_MAX); - n = man_node_alloc(man, sp->line, 0, ROFFT_TBL, MAN_MAX); - n->span = sp; - man_node_append(man, n); - man->next = ROFF_NEXT_SIBLING; - man_descope(man, sp->line, 0); -} - -static void -man_descope(struct roff_man *man, int line, int offs) -{ /* * 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; man->flags &= ~MAN_BLINE; man_unscope(man, man->last->parent); - man_body_alloc(man, line, offs, man->last->tok); + roff_body_alloc(man, line, offs, man->last->tok); } 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. */ + /* In no-fill mode, whitespace is preserved on text lines. */ - if (man->flags & MAN_LITERAL) { - man_word_alloc(man, line, offs, buf + offs); - man_descope(man, line, offs); - return(1); + if (man->flags & ROFF_NOFILL) { + roff_word_alloc(man, line, offs, buf + offs); + man_descope(man, line, offs, buf + offs); + return 1; } for (i = offs; buf[i] == ' '; 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) { - man_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; } - 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; } /* @@ -332,8 +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. */ ; @@ -343,7 +157,7 @@ man_ptext(struct roff_man *man, int line, char *buf, i buf[i] = '\0'; } - man_word_alloc(man, line, offs, buf + offs); + roff_word_alloc(man, line, offs, buf + offs); /* * End-of-sentence check. If the last character is an unescaped @@ -353,10 +167,10 @@ man_ptext(struct roff_man *man, int line, char *buf, i assert(i); if (mandoc_eos(buf, (size_t)i)) - man->last->flags |= MAN_EOS; + man->last->flags |= NODE_EOS; - man_descope(man, line, offs); - return(1); + man_descope(man, line, offs, buf + offs); + return 1; } static int @@ -364,30 +178,22 @@ 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) : MAN_MAX; - - if (tok == MAN_MAX) { - mandoc_msg(MANDOCERR_MACRO, man->parse, - ln, ppos, buf + ppos - 1); - return(1); + 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, ln, ppos, "%s", buf + ppos - 1); + return 1; } /* Skip a leading escape sequence or tab. */ @@ -407,7 +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++; /* @@ -416,8 +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 @@ -427,10 +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. */ @@ -438,7 +253,7 @@ man_pmacro(struct roff_man *man, int ln, char *buf, in n = man->last; if (n->type == ROFFT_BODY && strcmp(n->prev->child->string, "NAME")) - return(2); + return 2; } /* @@ -447,16 +262,16 @@ 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) - return(1); + 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); - man_body_alloc(man, ln, ppos, man->last->tok); - return(1); + roff_body_alloc(man, ln, ppos, man->last->tok); + return 1; } void @@ -470,142 +285,61 @@ man_breakscope(struct roff_man *man, int tok) * Delete the element that is being broken. */ - if (man->flags & MAN_ELINE && (tok == MAN_MAX || - ! (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 == MAN_MAX ? "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]); - man_node_delete(man, n); + roff_node_delete(man, n); man->flags &= ~MAN_ELINE; } /* + * Weird special case: + * Switching fill mode closes section headers. + */ + + if (man->flags & MAN_BLINE && + (tok == ROFF_nf || tok == ROFF_fi) && + (man->last->tok == MAN_SH || man->last->tok == MAN_SS)) { + n = man->last; + man_unscope(man, n); + roff_body_alloc(man, n->line, n->pos, n->tok); + man->flags &= ~MAN_BLINE; + } + + /* * A block header next line scope is open, * and the new macro is not allowed inside block headers. * Delete the block that is being broken. */ - if (man->flags & MAN_BLINE && (tok == MAN_MAX || - man_macros[tok].flags & MAN_BSCOPE)) { + if (man->flags & MAN_BLINE && tok != ROFF_nf && tok != ROFF_fi && + (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 == MAN_MAX ? "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]); - man_node_delete(man, n); + roff_node_delete(man, n); man->flags &= ~MAN_BLINE; } -} - -/* - * Unlink a node from its context. If "man" is provided, the last parse - * point will also be adjusted accordingly. - */ -static void -man_node_unlink(struct roff_man *man, struct roff_node *n) -{ - - /* Adjust siblings. */ - - if (n->prev) - n->prev->next = n->next; - if (n->next) - n->next->prev = n->prev; - - /* Adjust parent. */ - - if (n->parent) { - n->parent->nchild--; - if (n->parent->child == n) - n->parent->child = n->prev ? n->prev : n->next; - } - - /* Adjust parse point, if applicable. */ - - if (man && man->last == n) { - /*XXX: this can occur when bailing from validation. */ - /*assert(NULL == n->next);*/ - if (n->prev) { - man->last = n->prev; - man->next = ROFF_NEXT_SIBLING; - } else { - man->last = n->parent; - man->next = ROFF_NEXT_CHILD; - } - } - - if (man && man->first == n) - man->first = NULL; -} - -const struct mparse * -man_mparse(const struct roff_man *man) -{ - - assert(man && man->parse); - return(man->parse); -} - -void -man_deroff(char **dest, const struct roff_node *n) -{ - char *cp; - size_t sz; - - if (n->type != ROFFT_TEXT) { - for (n = n->child; n; n = n->next) - man_deroff(dest, n); - return; - } - - /* Skip leading whitespace and escape sequences. */ - - cp = n->string; - while ('\0' != *cp) { - if ('\\' == *cp) { - cp++; - mandoc_escape((const char **)&cp, NULL, NULL); - } else if (isspace((unsigned char)*cp)) - cp++; - else - break; - } - - /* Skip trailing whitespace. */ - - for (sz = strlen(cp); sz; sz--) - if (0 == isspace((unsigned char)cp[sz-1])) - break; - - /* Skip empty strings. */ - - if (0 == sz) - return; - - if (NULL == *dest) { - *dest = mandoc_strndup(cp, sz); - return; - } - - mandoc_asprintf(&cp, "%s %*s", *dest, (int)sz, cp); - free(*dest); - *dest = cp; }