=================================================================== RCS file: /cvs/mandoc/man.c,v retrieving revision 1.22 retrieving revision 1.31 diff -u -p -r1.22 -r1.31 --- mandoc/man.c 2009/06/18 10:32:00 1.22 +++ mandoc/man.c 2009/08/19 11:58:32 1.31 @@ -1,4 +1,4 @@ -/* $Id: man.c,v 1.22 2009/06/18 10:32:00 kristaps Exp $ */ +/* $Id: man.c,v 1.31 2009/08/19 11:58:32 kristaps Exp $ */ /* * Copyright (c) 2008, 2009 Kristaps Dzonsons * @@ -23,13 +23,36 @@ #include "libman.h" +const char *const __man_merrnames[WERRMAX] = { + "invalid character", /* WNPRINT */ + "system: malloc error", /* WNMEM */ + "invalid manual section", /* WMSEC */ + "invalid date format", /* WDATE */ + "scope of prior line violated", /* WLNSCOPE */ + "trailing whitespace", /* WTSPACE */ + "unterminated quoted parameter", /* WTQUOTE */ + "document has no body", /* WNODATA */ + "document has no title/section", /* WNOTITLE */ + "invalid escape sequence", /* WESCAPE */ + "invalid number format", /* WNUMFMT */ + "expected block head arguments", /* WHEADARGS */ + "expected block body arguments", /* WBODYARGS */ + "expected empty block head", /* WNHEADARGS */ + "unknown macro", /* WMACRO */ + "ill-formed macro", /* WMACROFORM */ + "scope open on exit", /* WEXITSCOPE */ + "no scope context" /* WNOSCOPE */ +}; + 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", "na", "i" + "RI", "na", "i", "sp", + "nf", "fi", "r", "RE", + "RS" }; const char * const *man_macronames = __man_macronames; @@ -42,6 +65,8 @@ static int man_ptext(struct man *, int, char *); static int man_pmacro(struct man *, int, char *); static void man_free1(struct man *); static int man_alloc1(struct man *); +static int pstring(struct man *, int, int, + const char *, size_t); const struct man_node * @@ -185,6 +210,22 @@ man_node_append(struct man *man, struct man_node *p) p->parent->nchild++; + if ( ! man_valid_pre(man, p)) + return(0); + + switch (p->type) { + case (MAN_HEAD): + assert(MAN_BLOCK == p->parent->type); + p->parent->head = p; + break; + case (MAN_BODY): + assert(MAN_BLOCK == p->parent->type); + p->parent->body = p; + break; + default: + break; + } + man->last = p; switch (p->type) { @@ -220,38 +261,110 @@ man_node_alloc(int line, int pos, enum man_type type, int -man_elem_alloc(struct man *man, int line, int pos, int tok) +man_elem_alloc(struct man *m, int line, int pos, int tok) { struct man_node *p; p = man_node_alloc(line, pos, MAN_ELEM, tok); if (NULL == p) return(0); - return(man_node_append(man, p)); + if ( ! man_node_append(m, p)) + return(0); + m->next = MAN_NEXT_CHILD; + return(1); } int -man_word_alloc(struct man *man, - int line, int pos, const char *word) +man_head_alloc(struct man *m, int line, int pos, int tok) { - struct man_node *p; + struct man_node *p; - p = man_node_alloc(line, pos, MAN_TEXT, -1); + p = man_node_alloc(line, pos, MAN_HEAD, tok); if (NULL == p) return(0); - if (NULL == (p->string = strdup(word))) + if ( ! man_node_append(m, p)) return(0); - return(man_node_append(man, p)); + m->next = MAN_NEXT_CHILD; + return(1); } +int +man_body_alloc(struct man *m, int line, int pos, int tok) +{ + struct man_node *p; + + p = man_node_alloc(line, pos, MAN_BODY, tok); + if (NULL == p) + return(0); + if ( ! man_node_append(m, p)) + return(0); + m->next = MAN_NEXT_CHILD; + return(1); +} + + +int +man_block_alloc(struct man *m, int line, int pos, int tok) +{ + struct man_node *p; + + p = man_node_alloc(line, pos, MAN_BLOCK, tok); + if (NULL == p) + return(0); + if ( ! man_node_append(m, p)) + return(0); + m->next = MAN_NEXT_CHILD; + return(1); +} + + +static int +pstring(struct man *m, int line, int pos, + const char *p, size_t len) +{ + struct man_node *n; + size_t sv; + + n = man_node_alloc(line, pos, MAN_TEXT, -1); + if (NULL == n) + return(0); + + n->string = malloc(len + 1); + if (NULL == n->string) { + free(n); + return(0); + } + + sv = strlcpy(n->string, p, len + 1); + + /* Prohibit truncation. */ + assert(sv < len + 1); + + if ( ! man_node_append(m, n)) + return(0); + m->next = MAN_NEXT_SIBLING; + return(1); +} + + +int +man_word_alloc(struct man *m, int line, int pos, const char *word) +{ + + return(pstring(m, line, pos, word, strlen(word))); +} + + void man_node_free(struct man_node *p) { if (p->string) free(p->string); + if (p->parent) + p->parent->nchild--; free(p); } @@ -265,6 +378,7 @@ man_node_freelist(struct man_node *p) if (p->next) man_node_freelist(p->next); + assert(0 == p->nchild); man_node_free(p); } @@ -272,30 +386,64 @@ man_node_freelist(struct man_node *p) static int man_ptext(struct man *m, int line, char *buf) { + int i, j; - if ( ! man_word_alloc(m, line, 0, buf)) + /* First de-chunk and allocate words. */ + + for (i = 0; ' ' == buf[i]; i++) + /* Skip leading whitespace. */ ; + if (0 == buf[i]) { + if ( ! pstring(m, line, 0, &buf[i], 0)) + return(0); + goto descope; + } + + for (j = i; buf[i]; i++) { + if (' ' != buf[i]) + continue; + + /* Escaped whitespace. */ + if (i && ' ' == buf[i] && '\\' == buf[i - 1]) + continue; + + buf[i++] = 0; + if ( ! pstring(m, line, j, &buf[j], (size_t)(i - j))) + return(0); + + for ( ; ' ' == buf[i]; i++) + /* Skip trailing whitespace. */ ; + + j = i; + if (0 == buf[i]) + break; + } + + if (j != i && ! pstring(m, line, j, &buf[j], (size_t)(i - j))) return(0); - m->next = MAN_NEXT_SIBLING; +descope: + /* - * If this is one of the zany NLINE macros that consumes the - * next line of input as being influenced, then close out the - * existing macro "scope" and continue processing. + * 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). */ - if ( ! (MAN_NLINE & m->flags)) - return(1); + /* XXX - this should be in man_action.c. */ - m->flags &= ~MAN_NLINE; - m->last = m->last->parent; + if (MAN_ELINE & m->flags) { + m->flags &= ~MAN_ELINE; + if ( ! man_unscope(m, m->last->parent)) + return(0); + } - assert(MAN_ROOT != m->last->type); - if ( ! man_valid_post(m)) - return(0); - if ( ! man_action_post(m)) - return(0); + if ( ! (MAN_BLINE & m->flags)) + return(1); + m->flags &= ~MAN_BLINE; - return(1); + if ( ! man_unscope(m, m->last->parent)) + return(0); + return(man_body_alloc(m, line, 0, m->last->tok)); } @@ -304,12 +452,10 @@ man_pmacro(struct man *m, int ln, char *buf) { int i, j, c, ppos, fl; char mac[5]; - struct man_node *n; /* Comments and empties are quickly ignored. */ - n = m->last; - fl = MAN_NLINE & m->flags; + fl = m->flags; if (0 == buf[1]) goto out; @@ -339,24 +485,20 @@ man_pmacro(struct man *m, int ln, char *buf) if (j == 4 || j < 1) { if ( ! (MAN_IGN_MACRO & m->pflags)) { - (void)man_verr(m, ln, ppos, - "ill-formed macro: %s", mac); + (void)man_perr(m, ln, ppos, WMACROFORM); goto err; } - if ( ! man_vwarn(m, ln, ppos, - "ill-formed macro: %s", mac)) + if ( ! man_pwarn(m, ln, ppos, WMACROFORM)) goto err; return(1); } if (MAN_MAX == (c = man_hash_find(m->htab, mac))) { if ( ! (MAN_IGN_MACRO & m->pflags)) { - (void)man_verr(m, ln, ppos, - "unknown macro: %s", mac); + (void)man_perr(m, ln, ppos, WMACRO); goto err; } - if ( ! man_vwarn(m, ln, ppos, - "unknown macro: %s", mac)) + if ( ! man_pwarn(m, ln, ppos, WMACRO)) goto err; return(1); } @@ -368,33 +510,34 @@ man_pmacro(struct man *m, int ln, char *buf) /* Begin recursive parse sequence. */ - if ( ! man_macro(m, c, ln, ppos, &i, buf)) + assert(man_macros[c].fp); + + if ( ! (*man_macros[c].fp)(m, c, ln, ppos, &i, buf)) goto err; out: - if (fl) { - /* - * A NLINE macro has been immediately followed with - * another. Close out the preceding macro's scope, and - * continue. - */ - assert(MAN_ROOT != m->last->type); - assert(m->last->parent); - assert(MAN_ROOT != m->last->parent->type); + if ( ! (MAN_BLINE & fl)) + return(1); - if (n != m->last) - m->last = m->last->parent; + /* + * If we've opened a new next-line element scope, then return + * now, as the next line will close out the block scope. + */ - if ( ! man_valid_post(m)) - return(0); - if ( ! man_action_post(m)) - return(0); - m->next = MAN_NEXT_SIBLING; - m->flags &= ~MAN_NLINE; - } + if (MAN_ELINE & m->flags) + return(1); - return(1); + /* Close out the block scope opened in the prior line. */ + /* XXX - this should be in man_action.c. */ + + assert(MAN_BLINE & m->flags); + m->flags &= ~MAN_BLINE; + + if ( ! man_unscope(m, m->last->parent)) + return(0); + return(man_body_alloc(m, ln, 0, m->last->tok)); + err: /* Error out. */ m->flags |= MAN_HALT; @@ -434,3 +577,16 @@ man_vwarn(struct man *man, int ln, int pos, const char } +int +man_err(struct man *m, int line, int pos, int iserr, enum merr type) +{ + const char *p; + + p = __man_merrnames[(int)type]; + assert(p); + + if (iserr) + return(man_verr(m, line, pos, p)); + + return(man_vwarn(m, line, pos, p)); +}