=================================================================== RCS file: /cvs/mandoc/mdoc.c,v retrieving revision 1.35 retrieving revision 1.50 diff -u -p -r1.35 -r1.50 --- mandoc/mdoc.c 2009/01/17 20:10:36 1.35 +++ mandoc/mdoc.c 2009/03/02 17:14:46 1.50 @@ -1,4 +1,4 @@ -/* $Id: mdoc.c,v 1.35 2009/01/17 20:10:36 kristaps Exp $ */ +/* $Id: mdoc.c,v 1.50 2009/03/02 17:14:46 kristaps Exp $ */ /* * Copyright (c) 2008 Kristaps Dzonsons * @@ -26,6 +26,24 @@ #include "private.h" +/* + * Main caller in the libmdoc library. This begins the parsing routine, + * handles allocation of data, and so forth. Most of the "work" is done + * in macro.c and validate.c. + */ + +static struct mdoc_arg *argdup(size_t, const struct mdoc_arg *); +static void argfree(size_t, struct mdoc_arg *); +static void argcpy(struct mdoc_arg *, + const struct mdoc_arg *); + +static struct mdoc_node *mdoc_node_alloc(const struct mdoc *); +static int mdoc_node_append(struct mdoc *, + struct mdoc_node *); +static void mdoc_elem_free(struct mdoc_elem *); +static void mdoc_text_free(struct mdoc_text *); + + const char *const __mdoc_macronames[MDOC_MAX] = { "\\\"", "Dd", "Dt", "Os", "Sh", "Ss", "Pp", "D1", @@ -83,153 +101,23 @@ const char *const __mdoc_argnames[MDOC_ARG_MAX] = { "emphasis", "symbolic", }; -const struct mdoc_macro __mdoc_macros[MDOC_MAX] = { - { NULL, 0 }, /* \" */ - { macro_constant, MDOC_PROLOGUE }, /* Dd */ - { macro_constant, MDOC_PROLOGUE }, /* Dt */ - { macro_constant, MDOC_PROLOGUE }, /* Os */ - { macro_scoped, 0 }, /* Sh */ - { macro_scoped, 0 }, /* Ss */ - { macro_text, 0 }, /* Pp */ - { macro_scoped_line, MDOC_PARSED }, /* D1 */ - { macro_scoped_line, MDOC_PARSED }, /* Dl */ - { macro_scoped, MDOC_EXPLICIT }, /* Bd */ - { macro_scoped_close, MDOC_EXPLICIT }, /* Ed */ - { macro_scoped, MDOC_EXPLICIT }, /* Bl */ - { macro_scoped_close, MDOC_EXPLICIT }, /* El */ - { macro_scoped, MDOC_PARSED | MDOC_TABSEP}, /* It */ - { macro_text, MDOC_CALLABLE | MDOC_PARSED }, /* Ad */ - { macro_text, MDOC_PARSED }, /* An */ - { macro_text, MDOC_CALLABLE | MDOC_PARSED }, /* Ar */ - { macro_constant, MDOC_QUOTABLE }, /* Cd */ - { macro_text, MDOC_CALLABLE | MDOC_PARSED }, /* Cm */ - { macro_text, MDOC_CALLABLE | MDOC_PARSED }, /* Dv */ - { macro_text, MDOC_CALLABLE | MDOC_PARSED }, /* Er */ - { macro_text, MDOC_CALLABLE | MDOC_PARSED }, /* Ev */ - { macro_constant, 0 }, /* Ex */ - { macro_text, MDOC_CALLABLE | MDOC_QUOTABLE | MDOC_PARSED }, /* Fa */ - { macro_constant, 0 }, /* Fd */ - { macro_text, MDOC_CALLABLE | MDOC_PARSED }, /* Fl */ - { macro_text, MDOC_CALLABLE | MDOC_QUOTABLE | MDOC_PARSED }, /* Fn */ - { macro_text, MDOC_PARSED | MDOC_QUOTABLE }, /* Ft */ - { macro_text, MDOC_CALLABLE | MDOC_PARSED }, /* Ic */ - { macro_constant, 0 }, /* In */ - { macro_text, MDOC_CALLABLE | MDOC_PARSED }, /* Li */ - { macro_constant, 0 }, /* Nd */ - { macro_text, MDOC_CALLABLE | MDOC_PARSED }, /* Nm */ - { macro_scoped_line, MDOC_CALLABLE | MDOC_PARSED }, /* Op */ - { macro_obsolete, 0 }, /* Ot */ - { macro_text, MDOC_CALLABLE | MDOC_PARSED }, /* Pa */ - { macro_constant, 0 }, /* Rv */ - /* XXX - .St supposed to be (but isn't) callable. */ - { macro_constant_delimited, MDOC_PARSED }, /* St */ - { macro_text, MDOC_CALLABLE | MDOC_PARSED }, /* Va */ - { macro_text, MDOC_CALLABLE | MDOC_PARSED }, /* Vt */ - { macro_text, MDOC_CALLABLE | MDOC_PARSED }, /* Xr */ - { macro_constant, MDOC_QUOTABLE }, /* %A */ - { macro_constant, MDOC_QUOTABLE }, /* %B */ - { macro_constant, MDOC_QUOTABLE }, /* %D */ - { macro_constant, MDOC_QUOTABLE }, /* %I */ - { macro_constant, MDOC_QUOTABLE }, /* %J */ - { macro_constant, MDOC_QUOTABLE }, /* %N */ - { macro_constant, MDOC_QUOTABLE }, /* %O */ - { macro_constant, MDOC_QUOTABLE }, /* %P */ - { macro_constant, MDOC_QUOTABLE }, /* %R */ - { macro_constant, MDOC_QUOTABLE }, /* %T */ - { macro_constant, MDOC_QUOTABLE }, /* %V */ - { macro_scoped_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Ac */ - { macro_constant_scoped, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Ao */ - { macro_scoped_line, MDOC_CALLABLE | MDOC_PARSED }, /* Aq */ - { macro_constant_delimited, 0 }, /* At */ - { macro_scoped_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Bc */ - { macro_scoped, MDOC_EXPLICIT }, /* Bf */ - { macro_constant_scoped, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Bo */ - { macro_scoped_line, MDOC_CALLABLE | MDOC_PARSED }, /* Bq */ - { macro_constant_delimited, MDOC_PARSED }, /* Bsx */ - { macro_constant_delimited, MDOC_PARSED }, /* Bx */ - { macro_constant, 0 }, /* Db */ - { macro_scoped_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Dc */ - { macro_constant_scoped, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Do */ - { macro_scoped_line, MDOC_CALLABLE | MDOC_PARSED }, /* Dq */ - { macro_scoped_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Ec */ - { macro_scoped_close, MDOC_EXPLICIT }, /* Ef */ - { macro_text, MDOC_CALLABLE | MDOC_PARSED }, /* Em */ - { macro_constant_scoped, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Eo */ - { macro_constant_delimited, MDOC_PARSED }, /* Fx */ - { macro_text, MDOC_PARSED }, /* Ms */ - { macro_constant_delimited, MDOC_CALLABLE | MDOC_PARSED }, /* No */ - { macro_constant_delimited, MDOC_CALLABLE | MDOC_PARSED }, /* Ns */ - { macro_constant_delimited, MDOC_PARSED }, /* Nx */ - { macro_constant_delimited, MDOC_PARSED }, /* Ox */ - { macro_scoped_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Pc */ - { macro_constant_delimited, MDOC_PARSED }, /* Pf */ - { macro_constant_scoped, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Po */ - { macro_scoped_line, MDOC_CALLABLE | MDOC_PARSED }, /* Pq */ - { macro_scoped_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Qc */ - { macro_scoped_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ql */ - { macro_constant_scoped, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Qo */ - { macro_scoped_line, MDOC_CALLABLE | MDOC_PARSED }, /* Qq */ - { macro_scoped_close, MDOC_EXPLICIT }, /* Re */ - { macro_scoped, MDOC_EXPLICIT }, /* Rs */ - { macro_scoped_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Sc */ - { macro_constant_scoped, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* So */ - { macro_scoped_line, MDOC_CALLABLE | MDOC_PARSED }, /* Sq */ - { macro_constant, 0 }, /* Sm */ - { macro_text, MDOC_CALLABLE | MDOC_PARSED }, /* Sx */ - { macro_text, MDOC_CALLABLE | MDOC_PARSED }, /* Sy */ - { macro_text, MDOC_CALLABLE | MDOC_PARSED }, /* Tn */ - { macro_constant_delimited, MDOC_PARSED }, /* Ux */ - { macro_scoped_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Xc */ - { macro_constant_scoped, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Xo */ - /* XXX - .Fo supposed to be (but isn't) callable. */ - { macro_scoped, MDOC_EXPLICIT | MDOC_PARSED }, /* Fo */ - /* XXX - .Fc supposed to be (but isn't) callable. */ - { macro_scoped_close, MDOC_EXPLICIT | MDOC_PARSED }, /* Fc */ - { macro_constant_scoped, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Oo */ - { macro_scoped_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Oc */ - { macro_scoped, MDOC_EXPLICIT }, /* Bk */ - { macro_scoped_close, MDOC_EXPLICIT }, /* Ek */ - { macro_constant, 0 }, /* Bt */ - { macro_constant, 0 }, /* Hf */ - { macro_obsolete, 0 }, /* Fr */ - { macro_constant, 0 }, /* Ud */ -}; - const char * const *mdoc_macronames = __mdoc_macronames; const char * const *mdoc_argnames = __mdoc_argnames; -const struct mdoc_macro * const mdoc_macros = __mdoc_macros; -static struct mdoc_arg *argdup(size_t, const struct mdoc_arg *); -static void argfree(size_t, struct mdoc_arg *); -static void argcpy(struct mdoc_arg *, - const struct mdoc_arg *); - -static void mdoc_node_freelist(struct mdoc_node *); -static int mdoc_node_append(struct mdoc *, - struct mdoc_node *); -static void mdoc_elem_free(struct mdoc_elem *); -static void mdoc_text_free(struct mdoc_text *); - - const struct mdoc_node * -mdoc_result(struct mdoc *mdoc) +mdoc_node(const struct mdoc *mdoc) { return(mdoc->first); } -void -mdoc_meta_free(struct mdoc *mdoc) +const struct mdoc_meta * +mdoc_meta(const struct mdoc *mdoc) { - if (mdoc->meta.title) - free(mdoc->meta.title); - if (mdoc->meta.os) - free(mdoc->meta.os); - if (mdoc->meta.name) - free(mdoc->meta.name); + return(&mdoc->meta); } @@ -241,7 +129,13 @@ mdoc_free(struct mdoc *mdoc) mdoc_node_freelist(mdoc->first); if (mdoc->htab) mdoc_tokhash_free(mdoc->htab); - + if (mdoc->meta.title) + free(mdoc->meta.title); + if (mdoc->meta.os) + free(mdoc->meta.os); + if (mdoc->meta.name) + free(mdoc->meta.name); + free(mdoc); } @@ -286,6 +180,12 @@ mdoc_endparse(struct mdoc *mdoc) } +/* + * Main line-parsing routine. If the line is a macro-line (started with + * a '.' control character), then pass along to the parser, which parses + * subsequent macros until the end of line. If normal text, simply + * append the entire line to the chain. + */ int mdoc_parseln(struct mdoc *mdoc, int line, char *buf) { @@ -295,22 +195,32 @@ mdoc_parseln(struct mdoc *mdoc, int line, char *buf) if (MDOC_HALT & mdoc->flags) return(0); + mdoc->linetok = 0; + if ('.' != *buf) { - if (SEC_PROLOGUE != mdoc->sec_lastn) { - if ( ! mdoc_word_alloc(mdoc, line, 0, buf)) - return(0); - mdoc->next = MDOC_NEXT_SIBLING; - return(1); - } - return(mdoc_perr(mdoc, line, 0, "text disallowed")); + /* + * Free-form text. Not allowed in the prologue. + */ + if (SEC_PROLOGUE == mdoc->lastnamed) + return(mdoc_perr(mdoc, line, 0, + "no text in prologue")); + + if ( ! mdoc_word_alloc(mdoc, line, 0, buf)) + return(0); + mdoc->next = MDOC_NEXT_SIBLING; + return(1); } + /* + * Control-character detected. Begin the parsing sequence. + */ + if (buf[1] && '\\' == buf[1]) if (buf[2] && '\"' == buf[2]) return(1); i = 1; - while (buf[i] && ! isspace(buf[i]) && i < (int)sizeof(tmp)) + while (buf[i] && ! isspace((int)buf[i]) && i < (int)sizeof(tmp)) i++; if (i == (int)sizeof(tmp)) { @@ -331,13 +241,14 @@ mdoc_parseln(struct mdoc *mdoc, int line, char *buf) return(mdoc_perr(mdoc, line, 1, "unknown macro")); } - while (buf[i] && isspace(buf[i])) + while (buf[i] && isspace((int)buf[i])) i++; if ( ! mdoc_macro(mdoc, c, line, 1, &i, buf)) { mdoc->flags |= MDOC_HALT; return(0); } + return(1); } @@ -399,8 +310,11 @@ mdoc_macro(struct mdoc *mdoc, int tok, assert(mdoc_macros[tok].fp); - if ( ! (MDOC_PROLOGUE & mdoc_macros[tok].flags) && - SEC_PROLOGUE == mdoc->sec_lastn) + if (MDOC_PROLOGUE & mdoc_macros[tok].flags && + SEC_PROLOGUE != mdoc->lastnamed) + return(mdoc_perr(mdoc, ln, ppos, "macro disallowed in document body")); + if ( ! (MDOC_PROLOGUE & mdoc_macros[tok].flags) && + SEC_PROLOGUE == mdoc->lastnamed) return(mdoc_perr(mdoc, ln, ppos, "macro disallowed in document prologue")); if (1 != ppos && ! (MDOC_CALLABLE & mdoc_macros[tok].flags)) return(mdoc_perr(mdoc, ln, ppos, "macro not callable")); @@ -417,6 +331,27 @@ mdoc_node_append(struct mdoc *mdoc, struct mdoc_node * assert(mdoc->first); assert(MDOC_ROOT != p->type); + /* See if we exceed the suggest line-max. */ + + switch (p->type) { + case (MDOC_TEXT): + /* FALLTHROUGH */ + case (MDOC_ELEM): + /* FALLTHROUGH */ + case (MDOC_BLOCK): + mdoc->linetok++; + break; + default: + break; + } + + /* This sort-of works (re-opening of text macros...). */ + if (mdoc->linetok > MDOC_LINEARG_SOFTMAX) + if ( ! mdoc_nwarn(mdoc, p, WARN_COMPAT, + "suggested %d tokens per line exceeded (has %d)", + MDOC_LINEARG_SOFTMAX, mdoc->linetok)) + return(0); + if (MDOC_TEXT == mdoc->last->type) on = ""; else if (MDOC_ROOT == mdoc->last->type) @@ -478,6 +413,18 @@ mdoc_node_append(struct mdoc *mdoc, struct mdoc_node * } +static struct mdoc_node * +mdoc_node_alloc(const struct mdoc *mdoc) +{ + struct mdoc_node *p; + + p = xcalloc(1, sizeof(struct mdoc_node)); + p->sec = mdoc->lastsec; + + return(p); +} + + int mdoc_tail_alloc(struct mdoc *mdoc, int line, int pos, int tok) { @@ -486,7 +433,7 @@ mdoc_tail_alloc(struct mdoc *mdoc, int line, int pos, assert(mdoc->first); assert(mdoc->last); - p = xcalloc(1, sizeof(struct mdoc_node)); + p = mdoc_node_alloc(mdoc); p->line = line; p->pos = pos; @@ -505,7 +452,7 @@ mdoc_head_alloc(struct mdoc *mdoc, int line, int pos, assert(mdoc->first); assert(mdoc->last); - p = xcalloc(1, sizeof(struct mdoc_node)); + p = mdoc_node_alloc(mdoc); p->line = line; p->pos = pos; @@ -524,7 +471,7 @@ mdoc_body_alloc(struct mdoc *mdoc, int line, int pos, assert(mdoc->first); assert(mdoc->last); - p = xcalloc(1, sizeof(struct mdoc_node)); + p = mdoc_node_alloc(mdoc); p->line = line; p->pos = pos; @@ -540,7 +487,7 @@ mdoc_root_alloc(struct mdoc *mdoc) { struct mdoc_node *p; - p = xcalloc(1, sizeof(struct mdoc_node)); + p = mdoc_node_alloc(mdoc); p->type = MDOC_ROOT; @@ -554,7 +501,7 @@ mdoc_block_alloc(struct mdoc *mdoc, int line, int pos, { struct mdoc_node *p; - p = xcalloc(1, sizeof(struct mdoc_node)); + p = mdoc_node_alloc(mdoc); p->pos = pos; p->line = line; @@ -573,7 +520,7 @@ mdoc_elem_alloc(struct mdoc *mdoc, int line, int pos, { struct mdoc_node *p; - p = xcalloc(1, sizeof(struct mdoc_node)); + p = mdoc_node_alloc(mdoc); p->line = line; p->pos = pos; @@ -592,7 +539,8 @@ mdoc_word_alloc(struct mdoc *mdoc, { struct mdoc_node *p; - p = xcalloc(1, sizeof(struct mdoc_node)); + p = mdoc_node_alloc(mdoc); + p->line = line; p->pos = pos; p->type = MDOC_TEXT; @@ -671,7 +619,7 @@ mdoc_node_free(struct mdoc_node *p) } -static void +void mdoc_node_freelist(struct mdoc_node *p) { @@ -725,6 +673,7 @@ argdup(size_t argsz, const struct mdoc_arg *args) } +/* FIXME: deprecate. */ char * mdoc_node2a(struct mdoc_node *node) { @@ -741,6 +690,7 @@ mdoc_node2a(struct mdoc_node *node) (void)xstrlcat(buf, node->data.text.string, 64); else (void)xstrlcat(buf, mdoc_macronames[node->tok], 64); + (void)xstrlcat(buf, "'", 64); return(buf); }