=================================================================== RCS file: /cvs/mandoc/mdoc.c,v retrieving revision 1.123 retrieving revision 1.133 diff -u -p -r1.123 -r1.133 --- mandoc/mdoc.c 2010/04/08 07:53:01 1.123 +++ mandoc/mdoc.c 2010/05/15 16:24:37 1.133 @@ -1,4 +1,4 @@ -/* $Id: mdoc.c,v 1.123 2010/04/08 07:53:01 kristaps Exp $ */ +/* $Id: mdoc.c,v 1.133 2010/05/15 16:24:37 kristaps Exp $ */ /* * Copyright (c) 2008, 2009 Kristaps Dzonsons * @@ -69,7 +69,6 @@ const char *const __mdoc_merrnames[MERRMAX] = { "prologue macro out of conventional order", /* EPROLOOO */ "prologue macro repeated", /* EPROLREP */ "invalid manual section", /* EBADMSEC */ - "invalid section", /* EBADSEC */ "invalid font mode", /* EFONT */ "invalid date syntax", /* EBADDATE */ "invalid number format", /* ENUMFMT */ @@ -154,9 +153,8 @@ static int node_append(struct mdoc *, static int mdoc_ptext(struct mdoc *, int, char *); static int mdoc_pmacro(struct mdoc *, int, char *); static int macrowarn(struct mdoc *, int, const char *); -static int pstring(struct mdoc *, int, int, - const char *, size_t); + const struct mdoc_node * mdoc_node(const struct mdoc *m) { @@ -192,6 +190,8 @@ mdoc_free1(struct mdoc *mdoc) free(mdoc->meta.arch); if (mdoc->meta.vol) free(mdoc->meta.vol); + if (mdoc->meta.msec) + free(mdoc->meta.msec); } @@ -290,7 +290,9 @@ mdoc_parseln(struct mdoc *m, int ln, char *buf) if (MDOC_HALT & m->flags) return(0); - return('.' == *buf ? mdoc_pmacro(m, ln, buf) : + m->flags |= MDOC_NEWLINE; + return('.' == *buf ? + mdoc_pmacro(m, ln, buf) : mdoc_ptext(m, ln, buf)); } @@ -454,7 +456,9 @@ node_alloc(struct mdoc *m, int line, int pos, p->pos = pos; p->tok = tok; p->type = type; - + if (MDOC_NEWLINE & m->flags) + p->flags |= MDOC_LINE; + m->flags &= ~MDOC_NEWLINE; return(p); } @@ -535,13 +539,15 @@ mdoc_elem_alloc(struct mdoc *m, int line, int pos, } -static int -pstring(struct mdoc *m, int line, int pos, const char *p, size_t len) +int +mdoc_word_alloc(struct mdoc *m, int line, int pos, const char *p) { struct mdoc_node *n; - size_t sv; + size_t sv, len; - n = node_alloc(m, line, pos, -1, MDOC_TEXT); + len = strlen(p); + + n = node_alloc(m, line, pos, MDOC_MAX, MDOC_TEXT); n->string = mandoc_malloc(len + 1); sv = strlcpy(n->string, p, len + 1); @@ -550,19 +556,12 @@ pstring(struct mdoc *m, int line, int pos, const char if ( ! node_append(m, n)) return(0); + m->next = MDOC_NEXT_SIBLING; return(1); } -int -mdoc_word_alloc(struct mdoc *m, int line, int pos, const char *p) -{ - - return(pstring(m, line, pos, p, strlen(p))); -} - - void mdoc_node_free(struct mdoc_node *p) { @@ -633,98 +632,93 @@ mdoc_node_delete(struct mdoc *m, struct mdoc_node *p) static int mdoc_ptext(struct mdoc *m, int line, char *buf) { - int i, j; - char sv; + int i; /* Ignore bogus comments. */ if ('\\' == buf[0] && '.' == buf[1] && '\"' == buf[2]) return(mdoc_pwarn(m, line, 0, EBADCOMMENT)); + /* No text before an initial macro. */ + if (SEC_NONE == m->lastnamed) return(mdoc_perr(m, line, 0, ETEXTPROL)); - - /* - * If in literal mode, then pass the buffer directly to the - * back-end, as it should be preserved as a single term. - */ + /* Literal just gets pulled in as-is. */ + if (MDOC_LITERAL & m->flags) return(mdoc_word_alloc(m, line, 0, buf)); - /* Disallow blank/white-space lines in non-literal mode. */ + /* Check for a blank line, which may also consist of spaces. */ for (i = 0; ' ' == buf[i]; i++) - /* Skip leading whitespace. */ ; + /* Skip to first non-space. */ ; if ('\0' == buf[i]) { if ( ! mdoc_pwarn(m, line, 0, ENOBLANK)) return(0); + /* - * Assume that a `Pp' should be inserted in the case of - * a blank line. Technically, blank lines aren't - * allowed, but enough manuals assume this behaviour - * that we want to work around it. + * Insert a `Pp' in the case of a blank line. Technically, + * blank lines aren't allowed, but enough manuals assume this + * behaviour that we want to work around it. */ if ( ! mdoc_elem_alloc(m, line, 0, MDOC_Pp, NULL)) return(0); + + m->next = MDOC_NEXT_SIBLING; + return(1); } - /* - * Break apart a free-form line into tokens. Spaces are - * stripped out of the input. + /* + * Warn if the last un-escaped character is whitespace. Then + * strip away the remaining spaces (tabs stay!). */ - for (j = i; buf[i]; i++) { - if (' ' != buf[i]) - continue; + i = (int)strlen(buf); + assert(i); - /* Escaped whitespace. */ - if (i && ' ' == buf[i] && '\\' == buf[i - 1]) - continue; + if (' ' == buf[i - 1] || '\t' == buf[i - 1]) { + if (i > 1 && '\\' != buf[i - 2]) + if ( ! mdoc_pwarn(m, line, i - 1, ETAILWS)) + return(0); - sv = buf[i]; - buf[i++] = '\0'; + for (--i; i && ' ' == buf[i]; i--) + /* Spin back to non-space. */ ; - if ( ! pstring(m, line, j, &buf[j], (size_t)(i - j))) - return(0); + /* Jump ahead of escaped whitespace. */ + i += '\\' == buf[i] ? 2 : 1; - /* Trailing whitespace? Check at overwritten byte. */ + buf[i] = '\0'; + } - if (' ' == sv && '\0' == buf[i]) - if ( ! mdoc_pwarn(m, line, i - 1, ETAILWS)) - return(0); + /* Allocate the whole word. */ - for ( ; ' ' == buf[i]; i++) - /* Skip trailing whitespace. */ ; + if ( ! mdoc_word_alloc(m, line, 0, buf)) + return(0); - j = i; + /* + * End-of-sentence check. If the last character is an unescaped + * EOS character, then flag the node as being the end of a + * sentence. The front-end will know how to interpret this. + */ - /* Trailing whitespace? */ + /* FIXME: chain of close delims. */ - if (' ' == buf[i - 1] && '\0' == buf[i]) - if ( ! mdoc_pwarn(m, line, i - 1, ETAILWS)) - return(0); + assert(i); - if ('\0' == buf[i]) - break; - } + if (mandoc_eos(buf, (size_t)i)) + m->last->flags |= MDOC_EOS; - if (j != i && ! pstring(m, line, j, &buf[j], (size_t)(i - j))) - return(0); - - m->next = MDOC_NEXT_SIBLING; return(1); } - static int macrowarn(struct mdoc *m, int ln, const char *buf) { if ( ! (MDOC_IGN_MACRO & m->pflags)) - return(mdoc_verr(m, ln, 0, - "unknown macro: %s%s", + return(mdoc_verr(m, ln, 0, "unknown macro: %s%s", buf, strlen(buf) > 3 ? "..." : "")); return(mdoc_vwarn(m, ln, 0, "unknown macro: %s%s", buf, strlen(buf) > 3 ? "..." : "")); @@ -738,8 +732,9 @@ macrowarn(struct mdoc *m, int ln, const char *buf) int mdoc_pmacro(struct mdoc *m, int ln, char *buf) { - int i, j, c; - char mac[5]; + enum mdoct tok; + int i, j, sv; + char mac[5]; /* Empty lines are ignored. */ @@ -758,6 +753,8 @@ mdoc_pmacro(struct mdoc *m, int ln, char *buf) return(1); } + sv = i; + /* Copy the first word into a nil-terminated buffer. */ for (j = 0; j < 4; j++, i++) { @@ -781,7 +778,7 @@ mdoc_pmacro(struct mdoc *m, int ln, char *buf) return(1); } - if (MDOC_MAX == (c = mdoc_hash_find(mac))) { + if (MDOC_MAX == (tok = mdoc_hash_find(mac))) { if ( ! macrowarn(m, ln, mac)) goto err; return(1); @@ -792,7 +789,10 @@ mdoc_pmacro(struct mdoc *m, int ln, char *buf) while (buf[i] && ' ' == buf[i]) i++; - /* Trailing whitespace? */ + /* + * Trailing whitespace. Note that tabs are allowed to be passed + * into the parser as "text", so we only warn about spaces here. + */ if ('\0' == buf[i] && ' ' == buf[i - 1]) if ( ! mdoc_pwarn(m, ln, i - 1, ETAILWS)) @@ -802,7 +802,7 @@ mdoc_pmacro(struct mdoc *m, int ln, char *buf) * Begin recursive parse sequence. Since we're at the start of * the line, we don't need to do callable/parseable checks. */ - if ( ! mdoc_macro(m, c, ln, 1, &i, buf)) + if ( ! mdoc_macro(m, tok, ln, sv, &i, buf)) goto err; return(1);