=================================================================== RCS file: /cvs/mandoc/Attic/validate.c,v retrieving revision 1.38 retrieving revision 1.47 diff -u -p -r1.38 -r1.47 --- mandoc/Attic/validate.c 2009/01/17 20:10:36 1.38 +++ mandoc/Attic/validate.c 2009/01/21 17:56:32 1.47 @@ -1,4 +1,4 @@ -/* $Id: validate.c,v 1.38 2009/01/17 20:10:36 kristaps Exp $ */ +/* $Id: validate.c,v 1.47 2009/01/21 17:56:32 kristaps Exp $ */ /* * Copyright (c) 2008 Kristaps Dzonsons * @@ -21,14 +21,22 @@ #include "private.h" +/* + * Pre- and post-validate macros as they're parsed. Pre-validation + * occurs when the macro has been detected and its arguments parsed. + * Post-validation occurs when all child macros have also been parsed. + * In the ELEMENT case, this is simply the parameters of the macro; in + * the BLOCK case, this is the HEAD, BODY, TAIL and so on. + */ + typedef int (*v_pre)(struct mdoc *, struct mdoc_node *); typedef int (*v_post)(struct mdoc *); /* FIXME: some sections should only occur in specific msecs. */ /* FIXME: ignoring Pp. */ /* FIXME: math symbols. */ -/* FIXME: make sure prologue is complete. */ /* FIXME: valid character-escape checks. */ +/* FIXME: make sure required sections are included (NAME, ...). */ struct valids { v_pre *pre; @@ -43,9 +51,11 @@ static int pre_check_msecs(struct mdoc *, struct mdoc_ int, enum mdoc_msec *); static int pre_check_stdarg(struct mdoc *, struct mdoc_node *); static int post_check_children_count(struct mdoc *); -static int post_check_children_lt(struct mdoc *, int); -static int post_check_children_gt(struct mdoc *, int); -static int post_check_children_eq(struct mdoc *, int); +static int post_check_children_lt(struct mdoc *, const char *, int); +static int post_check_children_gt(struct mdoc *, const char *, int); +static int post_check_children_wgt(struct mdoc *, const char *, int); +static int post_check_children_eq(struct mdoc *, const char *, int); +static int post_check_children_weq(struct mdoc *, const char *, int); /* Specific pre-child-parse routines. */ @@ -68,7 +78,7 @@ static int pre_prologue(struct mdoc *, struct mdoc_nod /* Specific post-child-parse routines. */ static int herr_ge1(struct mdoc *); -static int hwarn_ge1(struct mdoc *); +static int herr_le1(struct mdoc *); static int herr_eq0(struct mdoc *); static int eerr_eq0(struct mdoc *); static int eerr_le1(struct mdoc *); @@ -76,11 +86,13 @@ static int eerr_le2(struct mdoc *); static int eerr_eq1(struct mdoc *); static int eerr_ge1(struct mdoc *); static int ewarn_eq0(struct mdoc *); +static int ewarn_eq1(struct mdoc *); static int bwarn_ge1(struct mdoc *); -static int berr_eq0(struct mdoc *); static int ewarn_ge1(struct mdoc *); static int ebool(struct mdoc *); static int post_sh(struct mdoc *); +static int post_sh_body(struct mdoc *); +static int post_sh_head(struct mdoc *); static int post_bl(struct mdoc *); static int post_it(struct mdoc *); static int post_ex(struct mdoc *); @@ -88,6 +100,7 @@ static int post_an(struct mdoc *); static int post_at(struct mdoc *); static int post_xr(struct mdoc *); static int post_nm(struct mdoc *); +static int post_bf(struct mdoc *); static int post_root(struct mdoc *); /* Collections of pre-child-parse routines. */ @@ -113,18 +126,22 @@ static v_post posts_bd[] = { herr_eq0, bwarn_ge1, NULL static v_post posts_text[] = { eerr_ge1, NULL }; static v_post posts_wtext[] = { ewarn_ge1, NULL }; static v_post posts_notext[] = { eerr_eq0, NULL }; -static v_post posts_wline[] = { hwarn_ge1, berr_eq0, NULL }; +static v_post posts_wline[] = { bwarn_ge1, herr_eq0, NULL }; static v_post posts_sh[] = { herr_ge1, bwarn_ge1, post_sh, NULL }; static v_post posts_bl[] = { herr_eq0, bwarn_ge1, post_bl, NULL }; static v_post posts_it[] = { post_it, NULL }; +static v_post posts_in[] = { ewarn_eq1, NULL }; static v_post posts_ss[] = { herr_ge1, NULL }; static v_post posts_pp[] = { ewarn_eq0, NULL }; -static v_post posts_d1[] = { herr_ge1, NULL }; static v_post posts_ex[] = { eerr_le1, post_ex, NULL }; static v_post posts_an[] = { post_an, NULL }; static v_post posts_at[] = { post_at, NULL }; static v_post posts_xr[] = { eerr_ge1, eerr_le2, post_xr, NULL }; static v_post posts_nm[] = { post_nm, NULL }; +static v_post posts_bf[] = { herr_le1, post_bf, NULL }; +static v_post posts_rs[] = { herr_eq0, bwarn_ge1, NULL }; +static v_post posts_fo[] = { bwarn_ge1, NULL }; +static v_post posts_bk[] = { herr_eq0, bwarn_ge1, NULL }; /* Per-macro pre- and post-child-check routine collections. */ @@ -133,12 +150,11 @@ const struct valids mdoc_valids[MDOC_MAX] = { { pres_prologue, posts_text }, /* Dd */ { pres_prologue, NULL }, /* Dt */ { pres_prologue, NULL }, /* Os */ - /* FIXME: NAME section internal ordering. */ { pres_sh, posts_sh }, /* Sh */ { pres_ss, posts_ss }, /* Ss */ { NULL, posts_pp }, /* Pp */ - { pres_d1, posts_d1 }, /* D1 */ - { pres_d1, posts_d1 }, /* Dl */ + { pres_d1, posts_wline }, /* D1 */ + { pres_d1, posts_wline }, /* Dl */ { pres_bd, posts_bd }, /* Bd */ { NULL, NULL }, /* Ed */ { pres_bl, posts_bl }, /* Bl */ @@ -155,12 +171,12 @@ const struct valids mdoc_valids[MDOC_MAX] = { { pres_ex, posts_ex }, /* Ex */ { NULL, posts_text }, /* Fa */ /* FIXME: only in SYNOPSIS section. */ - { NULL, NULL }, /* Fd */ + { NULL, posts_wtext }, /* Fd */ { NULL, NULL }, /* Fl */ { NULL, posts_text }, /* Fn */ - { NULL, NULL }, /* Ft */ + { NULL, posts_wtext }, /* Ft */ { NULL, posts_text }, /* Ic */ - { NULL, posts_wtext }, /* In */ + { NULL, posts_in }, /* In */ { NULL, posts_text }, /* Li */ { NULL, posts_wtext }, /* Nd */ { NULL, posts_nm }, /* Nm */ @@ -188,7 +204,7 @@ const struct valids mdoc_valids[MDOC_MAX] = { { NULL, posts_wline }, /* Aq */ { NULL, posts_at }, /* At */ { NULL, NULL }, /* Bc */ - { NULL, NULL }, /* Bf */ + { NULL, posts_bf }, /* Bf */ { NULL, NULL }, /* Bo */ { NULL, posts_wline }, /* Bq */ { NULL, NULL }, /* Bsx */ @@ -216,7 +232,7 @@ const struct valids mdoc_valids[MDOC_MAX] = { { NULL, NULL }, /* Qo */ { NULL, posts_wline }, /* Qq */ { NULL, NULL }, /* Re */ - { NULL, NULL }, /* Rs */ + { NULL, posts_rs }, /* Rs */ { NULL, NULL }, /* Sc */ { NULL, NULL }, /* So */ { NULL, posts_wline }, /* Sq */ @@ -227,11 +243,11 @@ const struct valids mdoc_valids[MDOC_MAX] = { { NULL, NULL }, /* Ux */ { NULL, NULL }, /* Xc */ { NULL, NULL }, /* Xo */ - { NULL, NULL }, /* Fo */ + { NULL, posts_fo }, /* Fo */ { NULL, NULL }, /* Fc */ { NULL, NULL }, /* Oo */ { NULL, NULL }, /* Oc */ - { NULL, NULL }, /* Bk */ + { NULL, posts_bk }, /* Bk */ { NULL, NULL }, /* Ek */ { NULL, posts_notext }, /* Bt */ { NULL, NULL }, /* Hf */ @@ -253,38 +269,62 @@ post_check_children_count(struct mdoc *mdoc) static int -post_check_children_gt(struct mdoc *mdoc, int sz) +post_check_children_wgt(struct mdoc *mdoc, const char *p, int sz) { int i; if ((i = post_check_children_count(mdoc)) > sz) return(1); + return(mdoc_warn(mdoc, WARN_SYNTAX, "macro suggests more " + "than %d %s (has %d)", sz, p, i)); +} + + +static int +post_check_children_gt(struct mdoc *mdoc, const char *p, int sz) +{ + int i; + + if ((i = post_check_children_count(mdoc)) > sz) + return(1); return(mdoc_err(mdoc, "macro requires more than %d " - "parameters (have %d)", sz, i)); + "%s (has %d)", sz, p, i)); } static int -post_check_children_eq(struct mdoc *mdoc, int sz) +post_check_children_weq(struct mdoc *mdoc, const char *p, int sz) { int i; if ((i = post_check_children_count(mdoc)) == sz) return(1); - return(mdoc_err(mdoc, "macro requires %d parameters " - "(have %d)", sz, i)); + return(mdoc_warn(mdoc, WARN_SYNTAX, "macro suggests %d " + "%s (has %d)", sz, p, i)); } static int -post_check_children_lt(struct mdoc *mdoc, int sz) +post_check_children_eq(struct mdoc *mdoc, const char *p, int sz) { int i; + if ((i = post_check_children_count(mdoc)) == sz) + return(1); + return(mdoc_err(mdoc, "macro requires %d %s " + "(have %d)", sz, p, i)); +} + + +static int +post_check_children_lt(struct mdoc *mdoc, const char *p, int sz) +{ + int i; + if ((i = post_check_children_count(mdoc)) < sz) return(1); return(mdoc_err(mdoc, "macro requires less than %d " - "parameters (have %d)", sz, i)); + "%s (have %d)", sz, p, i)); } @@ -333,26 +373,21 @@ pre_check_parent(struct mdoc *mdoc, struct mdoc_node * static int -berr_eq0(struct mdoc *mdoc) +bwarn_ge1(struct mdoc *mdoc) { if (MDOC_BODY != mdoc->last->type) return(1); - if (NULL == mdoc->last->child) - return(1); - return(mdoc_warn(mdoc, WARN_SYNTAX, "macro suggests no body children")); + return(post_check_children_wgt(mdoc, "body children", 0)); } static int -bwarn_ge1(struct mdoc *mdoc) +ewarn_eq1(struct mdoc *mdoc) { - if (MDOC_BODY != mdoc->last->type) - return(1); - if (mdoc->last->child) - return(1); - return(mdoc_warn(mdoc, WARN_SYNTAX, "macro suggests one or more body children")); + assert(MDOC_ELEM == mdoc->last->type); + return(post_check_children_weq(mdoc, "parameters", 1)); } @@ -361,10 +396,7 @@ ewarn_eq0(struct mdoc *mdoc) { assert(MDOC_ELEM == mdoc->last->type); - if (NULL == mdoc->last->child) - return(1); - return(mdoc_pwarn(mdoc, mdoc->last->child->line, - mdoc->last->child->pos, WARN_SYNTAX, "macro suggests no parameters")); + return(post_check_children_weq(mdoc, "parameters", 0)); } @@ -373,9 +405,7 @@ ewarn_ge1(struct mdoc *mdoc) { assert(MDOC_ELEM == mdoc->last->type); - if (mdoc->last->child) - return(1); - return(mdoc_warn(mdoc, WARN_SYNTAX, "macro suggests one or more parameters")); + return(post_check_children_wgt(mdoc, "parameters", 0)); } @@ -384,7 +414,7 @@ eerr_eq1(struct mdoc *mdoc) { assert(MDOC_ELEM == mdoc->last->type); - return(post_check_children_eq(mdoc, 1)); + return(post_check_children_eq(mdoc, "parameters", 1)); } @@ -393,7 +423,7 @@ eerr_le2(struct mdoc *mdoc) { assert(MDOC_ELEM == mdoc->last->type); - return(post_check_children_lt(mdoc, 3)); + return(post_check_children_lt(mdoc, "parameters", 3)); } @@ -402,7 +432,7 @@ eerr_le1(struct mdoc *mdoc) { assert(MDOC_ELEM == mdoc->last->type); - return(post_check_children_lt(mdoc, 2)); + return(post_check_children_lt(mdoc, "parameters", 2)); } @@ -411,7 +441,7 @@ eerr_eq0(struct mdoc *mdoc) { assert(MDOC_ELEM == mdoc->last->type); - return(post_check_children_eq(mdoc, 0)); + return(post_check_children_eq(mdoc, "parameters", 0)); } @@ -420,7 +450,7 @@ eerr_ge1(struct mdoc *mdoc) { assert(MDOC_ELEM == mdoc->last->type); - return(post_check_children_gt(mdoc, 0)); + return(post_check_children_gt(mdoc, "parameters", 0)); } @@ -430,19 +460,16 @@ herr_eq0(struct mdoc *mdoc) if (MDOC_HEAD != mdoc->last->type) return(1); - return(post_check_children_eq(mdoc, 0)); + return(post_check_children_eq(mdoc, "parameters", 0)); } static int -hwarn_ge1(struct mdoc *mdoc) +herr_le1(struct mdoc *mdoc) { - if (MDOC_HEAD != mdoc->last->type) return(1); - if (mdoc->last->child) - return(1); - return(mdoc_warn(mdoc, WARN_SYNTAX, "macro suggests one or more parameters")); + return(post_check_children_lt(mdoc, "parameters", 2)); } @@ -452,7 +479,7 @@ herr_ge1(struct mdoc *mdoc) if (MDOC_HEAD != mdoc->last->type) return(1); - return(post_check_children_gt(mdoc, 0)); + return(post_check_children_gt(mdoc, "parameters", 0)); } @@ -691,7 +718,7 @@ static int pre_prologue(struct mdoc *mdoc, struct mdoc_node *node) { - if (SEC_PROLOGUE != mdoc->sec_lastn) + if (SEC_PROLOGUE != mdoc->lastnamed) return(mdoc_nerr(mdoc, node, "macro may only be invoked in the prologue")); assert(MDOC_ELEM == node->type); @@ -740,6 +767,40 @@ pre_prologue(struct mdoc *mdoc, struct mdoc_node *node static int +post_bf(struct mdoc *mdoc) +{ + char *p; + struct mdoc_node *head; + + if (MDOC_BLOCK != mdoc->last->type) + return(1); + assert(MDOC_Bf == mdoc->last->tok); + head = mdoc->last->data.block.head; + assert(head); + + if (0 == mdoc->last->data.block.argc) { + if (head->child) { + assert(MDOC_TEXT == head->child->type); + p = head->child->data.text.string; + if (xstrcmp(p, "Em")) + return(1); + else if (xstrcmp(p, "Li")) + return(1); + else if (xstrcmp(p, "Sm")) + return(1); + return(mdoc_nerr(mdoc, head->child, "invalid font mode")); + } + return(mdoc_err(mdoc, "macro expects an argument or parameter")); + } + if (head->child) + return(mdoc_err(mdoc, "macro expects an argument or parameter")); + if (1 == mdoc->last->data.block.argc) + return(1); + return(mdoc_err(mdoc, "macro expects an argument or parameter")); +} + + +static int post_nm(struct mdoc *mdoc) { @@ -843,6 +904,7 @@ post_it(struct mdoc *mdoc) #define TYPE_NONE (0) #define TYPE_BODY (1) #define TYPE_HEAD (2) +#define TYPE_OHEAD (3) size_t i, argc; struct mdoc_node *n; @@ -889,20 +951,23 @@ post_it(struct mdoc *mdoc) case (MDOC_Hyphen): /* FALLTHROUGH */ case (MDOC_Item): - /* FALLTHROUGH */ - case (MDOC_Column): type = TYPE_BODY; sv = n->data.block.argv[(int)i].arg; break; + case (MDOC_Column): + type = TYPE_OHEAD; + sv = n->data.block.argv[(int)i].arg; + break; default: break; } assert(TYPE_NONE != type); + n = mdoc->last->data.block.head; + assert(n); + if (TYPE_HEAD == type) { - n = mdoc->last->data.block.head; - assert(n); if (NULL == n->child) if ( ! mdoc_warn(mdoc, WARN_SYNTAX, "macro suggests line parameters")) return(0); @@ -913,26 +978,29 @@ post_it(struct mdoc *mdoc) if ( ! mdoc_warn(mdoc, WARN_SYNTAX, "macro suggests body children")) return(0); - return(1); + } else if (TYPE_BODY == type) { + if (n->child) + if ( ! mdoc_warn(mdoc, WARN_SYNTAX, "macro suggests no line parameters")) + return(0); + + n = mdoc->last->data.block.body; + assert(n); + if (NULL == n->child) + if ( ! mdoc_warn(mdoc, WARN_SYNTAX, "macro suggests body children")) + return(0); + } else { + if (NULL == n->child) + if ( ! mdoc_warn(mdoc, WARN_SYNTAX, "macro suggests line parameters")) + return(0); + + n = mdoc->last->data.block.body; + assert(n); + if (n->child) + if ( ! mdoc_warn(mdoc, WARN_SYNTAX, "macro suggests no body children")) + return(0); } - assert(TYPE_BODY == type); - assert(mdoc->last->data.block.head); - - n = mdoc->last->data.block.head; - assert(n); - if (n->child) - if ( ! mdoc_warn(mdoc, WARN_SYNTAX, "macro suggests no line parameters")) - return(0); - - n = mdoc->last->data.block.body; - assert(n); - if (NULL == n->child) - if ( ! mdoc_warn(mdoc, WARN_SYNTAX, "macro suggests body children")) - return(0); - - assert(-1 != sv); - if (MDOC_Column != sv) + if (MDOC_Column != sv) return(1); /* Make sure the number of columns is sane. */ @@ -949,6 +1017,7 @@ post_it(struct mdoc *mdoc) #undef TYPE_NONE #undef TYPE_BODY #undef TYPE_HEAD +#undef TYPE_OHEAD } @@ -1000,39 +1069,78 @@ static int post_root(struct mdoc *mdoc) { - if (NULL == mdoc->last->child) + if (NULL == mdoc->first->child) return(mdoc_err(mdoc, "document has no data")); - if (NULL == mdoc->meta.title) - return(mdoc_err(mdoc, "document has no incomplete prologue")); - if (NULL == mdoc->meta.os) - return(mdoc_err(mdoc, "document has no incomplete prologue")); - if (0 == mdoc->meta.date) - return(mdoc_err(mdoc, "document has no incomplete prologue")); + if (SEC_PROLOGUE == mdoc->lastnamed) + return(mdoc_err(mdoc, "document has incomplete prologue")); + if (MDOC_BLOCK != mdoc->first->child->type) + return(mdoc_err(mdoc, "document expects `%s' macro after prologue", mdoc_macronames[MDOC_Sh])); + if (MDOC_Sh != mdoc->first->child->tok) + return(mdoc_err(mdoc, "document expects `%s' macro after prologue", mdoc_macronames[MDOC_Sh])); return(1); } -/* Warn if conventional sections are out of order. */ static int post_sh(struct mdoc *mdoc) { + + if (MDOC_HEAD == mdoc->last->type) + return(post_sh_head(mdoc)); + if (MDOC_BODY == mdoc->last->type) + return(post_sh_body(mdoc)); + return(1); +} + + +static int +post_sh_body(struct mdoc *mdoc) +{ + struct mdoc_node *n; + + assert(MDOC_Sh == mdoc->last->tok); + if (SEC_NAME != mdoc->lastnamed) + return(1); + + if (NULL == (n = mdoc->last->child)) + return(mdoc_err(mdoc, "section NAME must contain %s as the first body child", mdoc_macronames[MDOC_Nm])); + if (MDOC_ELEM != n->type || MDOC_Nm != n->tok) + return(mdoc_err(mdoc, "section NAME must contain %s as the first body child", mdoc_macronames[MDOC_Nm])); + if (NULL == (n = n->next)) + return(mdoc_err(mdoc, "section NAME must contain %s as the second body child", mdoc_macronames[MDOC_Nd])); + if (MDOC_ELEM != n->type || MDOC_Nd != n->tok) + return(mdoc_err(mdoc, "section NAME must contain %s as the second body child", mdoc_macronames[MDOC_Nd])); + if (NULL == (n = n->next)) + return(1); + + return(mdoc_warn(mdoc, WARN_SYNTAX, "section NAME usually limited to %s and %s body children", + mdoc_macronames[MDOC_Nm], mdoc_macronames[MDOC_Nd])); +} + + +static int +post_sh_head(struct mdoc *mdoc) +{ char buf[64]; enum mdoc_sec sec; - if (MDOC_HEAD != mdoc->last->type) - return(1); assert(MDOC_Sh == mdoc->last->tok); if ( ! xstrlcats(buf, mdoc->last->child, 64)) return(mdoc_err(mdoc, "macro parameters too long")); - if (SEC_CUSTOM == (sec = mdoc_atosec(buf))) + sec = mdoc_atosec(buf); + + if (SEC_BODY == mdoc->lastnamed && SEC_NAME != sec) + return(mdoc_err(mdoc, "section NAME must be first")); + if (SEC_CUSTOM == sec) return(1); - if (sec > mdoc->sec_lastn) - return(1); - if (sec == mdoc->sec_lastn) + if (sec == mdoc->lastnamed) return(mdoc_warn(mdoc, WARN_SYNTAX, "section repeated")); - return(mdoc_warn(mdoc, WARN_SYNTAX, "section out of conventional order")); + if (sec < mdoc->lastnamed) + return(mdoc_warn(mdoc, WARN_SYNTAX, "section out of conventional order")); + + return(1); } @@ -1058,6 +1166,10 @@ int mdoc_valid_post(struct mdoc *mdoc) { v_post *p; + + if (MDOC_VALID & mdoc->last->flags) + return(1); + mdoc->last->flags |= MDOC_VALID; if (MDOC_TEXT == mdoc->last->type) return(1);