=================================================================== RCS file: /cvs/mandoc/mdoc.c,v retrieving revision 1.80 retrieving revision 1.107 diff -u -p -r1.80 -r1.107 --- mandoc/mdoc.c 2009/06/15 10:36:01 1.80 +++ mandoc/mdoc.c 2009/09/24 15:01:06 1.107 @@ -1,4 +1,4 @@ -/* $Id: mdoc.c,v 1.80 2009/06/15 10:36:01 kristaps Exp $ */ +/* $Id: mdoc.c,v 1.107 2009/09/24 15:01:06 kristaps Exp $ */ /* * Copyright (c) 2008, 2009 Kristaps Dzonsons * @@ -14,6 +14,8 @@ * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. */ +#include + #include #include #include @@ -23,18 +25,63 @@ #include "libmdoc.h" -enum merr { - ENOCALL, - EBODYPROL, - EPROLBODY, - ESPACE, - ETEXTPROL, - ENOBLANK, - EMALLOC +const char *const __mdoc_merrnames[MERRMAX] = { + "trailing whitespace", /* ETAILWS */ + "unexpected quoted parameter", /* EQUOTPARM */ + "unterminated quoted parameter", /* EQUOTTERM */ + "system: malloc error", /* EMALLOC */ + "argument parameter suggested", /* EARGVAL */ + "macro disallowed in prologue", /* EBODYPROL */ + "macro disallowed in body", /* EPROLBODY */ + "text disallowed in prologue", /* ETEXTPROL */ + "blank line disallowed", /* ENOBLANK */ + "text parameter too long", /* ETOOLONG */ + "invalid escape sequence", /* EESCAPE */ + "invalid character", /* EPRINT */ + "document has no body", /* ENODAT */ + "document has no prologue", /* ENOPROLOGUE */ + "expected line arguments", /* ELINE */ + "invalid AT&T argument", /* EATT */ + "default name not yet set", /* ENAME */ + "missing list type", /* ELISTTYPE */ + "missing display type", /* EDISPTYPE */ + "too many display types", /* EMULTIDISP */ + "too many list types", /* EMULTILIST */ + "NAME section must be first", /* ESECNAME */ + "badly-formed NAME section", /* ENAMESECINC */ + "argument repeated", /* EARGREP */ + "expected boolean parameter", /* EBOOL */ + "inconsistent column syntax", /* ECOLMIS */ + "nested display invalid", /* ENESTDISP */ + "width argument missing", /* EMISSWIDTH */ + "invalid section for this manual section", /* EWRONGMSEC */ + "section out of conventional order", /* ESECOOO */ + "section repeated", /* ESECREP */ + "invalid standard argument", /* EBADSTAND */ + "multi-line arguments discouraged", /* ENOMULTILINE */ + "multi-line arguments suggested", /* EMULTILINE */ + "line arguments discouraged", /* ENOLINE */ + "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 */ + "superfluous width argument", /* ENOWIDTH */ + "system: utsname error", /* EUTSNAME */ + "obsolete macro", /* EOBS */ + "end-of-line scope violation", /* EIMPBRK */ + "empty macro ignored", /* EIGNE */ + "unclosed explicit scope", /* EOPEN */ + "unterminated quoted phrase", /* EQUOTPHR */ + "closure macro without prior context", /* ENOCTX */ + "no description found for library", /* ELIB */ + "bad child for parent context", /* EBADCHILD */ }; const char *const __mdoc_macronames[MDOC_MAX] = { - "\\\"", "Dd", "Dt", "Os", + "Ap", "Dd", "Dt", "Os", "Sh", "Ss", "Pp", "D1", "Dl", "Bd", "Ed", "Bl", "El", "It", "Ad", "An", @@ -63,12 +110,12 @@ const char *const __mdoc_macronames[MDOC_MAX] = { "Tn", "Ux", "Xc", "Xo", "Fo", "Fc", "Oo", "Oc", "Bk", "Ek", "Bt", "Hf", - "Fr", "Ud", "Lb", "Ap", - "Lp", "Lk", "Mt", "Brq", + "Fr", "Ud", "Lb", "Lp", + "Lk", "Mt", "Brq", "Bro", /* LINTED */ - "Bro", "Brc", "\%C", "Es", + "Brc", "\%C", "Es", "En", /* LINTED */ - "En", "Dx", "\%Q" + "Dx", "\%Q", "br", "sp" }; const char *const __mdoc_argnames[MDOC_ARG_MAX] = { @@ -95,13 +142,14 @@ static int node_append(struct mdoc *, static int parsetext(struct mdoc *, int, char *); static int parsemacro(struct mdoc *, int, char *); static int macrowarn(struct mdoc *, int, const char *); -static int perr(struct mdoc *, int, int, enum merr); +static int pstring(struct mdoc *, int, int, + const char *, size_t); -#define verr(m, t) perr((m), (m)->last->line, (m)->last->pos, (t)) +#ifdef __linux__ +extern size_t strlcpy(char *, const char *, size_t); +#endif -/* - * Get the first (root) node of the parse tree. - */ + const struct mdoc_node * mdoc_node(const struct mdoc *m) { @@ -118,6 +166,9 @@ mdoc_meta(const struct mdoc *m) } +/* + * Frees volatile resources (parse tree, meta-data, fields). + */ static void mdoc_free1(struct mdoc *mdoc) { @@ -137,13 +188,16 @@ mdoc_free1(struct mdoc *mdoc) } +/* + * Allocate all volatile resources (parse tree, meta-data, fields). + */ static int mdoc_alloc1(struct mdoc *mdoc) { bzero(&mdoc->meta, sizeof(struct mdoc_meta)); mdoc->flags = 0; - mdoc->lastnamed = mdoc->lastsec = 0; + mdoc->lastnamed = mdoc->lastsec = SEC_NONE; mdoc->last = calloc(1, sizeof(struct mdoc_node)); if (NULL == mdoc->last) return(0); @@ -156,9 +210,10 @@ mdoc_alloc1(struct mdoc *mdoc) /* - * Free up all resources contributed by a parse: the node tree, - * meta-data and so on. Then reallocate the root node for another - * parse. + * Free up volatile resources (see mdoc_free1()) then re-initialises the + * data with mdoc_alloc1(). After invocation, parse data has been reset + * and the parser is ready for re-invocation on a new tree; however, + * cross-parse non-volatile data is kept intact. */ int mdoc_reset(struct mdoc *mdoc) @@ -170,19 +225,21 @@ mdoc_reset(struct mdoc *mdoc) /* - * Completely free up all resources. + * Completely free up all volatile and non-volatile parse resources. + * After invocation, the pointer is no longer usable. */ void mdoc_free(struct mdoc *mdoc) { mdoc_free1(mdoc); - if (mdoc->htab) - mdoc_hash_free(mdoc->htab); free(mdoc); } +/* + * Allocate volatile and non-volatile parse resources. + */ struct mdoc * mdoc_alloc(void *data, int pflags, const struct mdoc_cb *cb) { @@ -193,13 +250,12 @@ mdoc_alloc(void *data, int pflags, const struct mdoc_c if (cb) (void)memcpy(&p->cb, cb, sizeof(struct mdoc_cb)); + mdoc_hash_init(); + p->data = data; p->pflags = pflags; - if (NULL == (p->htab = mdoc_hash_alloc())) { - free(p); - return(NULL); - } else if (mdoc_alloc1(p)) + if (mdoc_alloc1(p)) return(p); free(p); @@ -209,7 +265,7 @@ mdoc_alloc(void *data, int pflags, const struct mdoc_c /* * Climb back up the parse tree, validating open scopes. Mostly calls - * through to macro_end in macro.c. + * through to macro_end() in macro.c. */ int mdoc_endparse(struct mdoc *m) @@ -226,14 +282,12 @@ mdoc_endparse(struct mdoc *m) /* * Main parse routine. Parses a single line -- really just hands off to - * the macro or text parser. + * the macro (parsemacro()) or text parser (parsetext()). */ int mdoc_parseln(struct mdoc *m, int ln, char *buf) { - /* If in error-mode, then we parse no more. */ - if (MDOC_HALT & m->flags) return(0); @@ -255,13 +309,13 @@ mdoc_verr(struct mdoc *mdoc, int ln, int pos, va_start(ap, fmt); (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap); va_end(ap); + return((*mdoc->cb.mdoc_err)(mdoc->data, ln, pos, buf)); } int -mdoc_vwarn(struct mdoc *mdoc, int ln, int pos, - enum mdoc_warn type, const char *fmt, ...) +mdoc_vwarn(struct mdoc *mdoc, int ln, int pos, const char *fmt, ...) { char buf[256]; va_list ap; @@ -272,149 +326,46 @@ mdoc_vwarn(struct mdoc *mdoc, int ln, int pos, va_start(ap, fmt); (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap); va_end(ap); - return((*mdoc->cb.mdoc_warn)(mdoc->data, ln, pos, type, buf)); -} - -int -mdoc_nerr(struct mdoc *mdoc, const struct mdoc_node *node, const char *fmt, ...) -{ - char buf[256]; - va_list ap; - - if (NULL == mdoc->cb.mdoc_err) - return(0); - - va_start(ap, fmt); - (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap); - va_end(ap); - return((*mdoc->cb.mdoc_err)(mdoc->data, node->line, node->pos, buf)); + return((*mdoc->cb.mdoc_warn)(mdoc->data, ln, pos, buf)); } int -mdoc_warn(struct mdoc *mdoc, enum mdoc_warn type, const char *fmt, ...) +mdoc_err(struct mdoc *m, int line, int pos, int iserr, enum merr type) { - char buf[256]; - va_list ap; + const char *p; - if (NULL == mdoc->cb.mdoc_warn) - return(0); + p = __mdoc_merrnames[(int)type]; + assert(p); - va_start(ap, fmt); - (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap); - va_end(ap); - return((*mdoc->cb.mdoc_warn)(mdoc->data, mdoc->last->line, - mdoc->last->pos, type, buf)); -} + if (iserr) + return(mdoc_verr(m, line, pos, p)); - -int -mdoc_err(struct mdoc *mdoc, const char *fmt, ...) -{ - char buf[256]; - va_list ap; - - if (NULL == mdoc->cb.mdoc_err) - return(0); - - va_start(ap, fmt); - (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap); - va_end(ap); - return((*mdoc->cb.mdoc_err)(mdoc->data, mdoc->last->line, - mdoc->last->pos, buf)); + return(mdoc_vwarn(m, line, pos, p)); } int -mdoc_pwarn(struct mdoc *mdoc, int line, int pos, enum mdoc_warn type, - const char *fmt, ...) -{ - char buf[256]; - va_list ap; - - if (NULL == mdoc->cb.mdoc_warn) - return(0); - - va_start(ap, fmt); - (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap); - va_end(ap); - return((*mdoc->cb.mdoc_warn)(mdoc->data, line, pos, type, buf)); -} - -int -mdoc_perr(struct mdoc *mdoc, int line, int pos, const char *fmt, ...) -{ - char buf[256]; - va_list ap; - - if (NULL == mdoc->cb.mdoc_err) - return(0); - - va_start(ap, fmt); - (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap); - va_end(ap); - return((*mdoc->cb.mdoc_err)(mdoc->data, line, pos, buf)); -} - - -int mdoc_macro(struct mdoc *m, int tok, int ln, int pp, int *pos, char *buf) { - - /* FIXME - these should happen during validation. */ - + /* + * If we're in the prologue, deny "body" macros. Similarly, if + * we're in the body, deny prologue calls. + */ if (MDOC_PROLOGUE & mdoc_macros[tok].flags && - SEC_PROLOGUE != m->lastnamed) - return(perr(m, ln, pp, EPROLBODY)); - + MDOC_PBODY & m->flags) + return(mdoc_perr(m, ln, pp, EPROLBODY)); if ( ! (MDOC_PROLOGUE & mdoc_macros[tok].flags) && - SEC_PROLOGUE == m->lastnamed) - return(perr(m, ln, pp, EBODYPROL)); + ! (MDOC_PBODY & m->flags)) + return(mdoc_perr(m, ln, pp, EBODYPROL)); - if (1 != pp && ! (MDOC_CALLABLE & mdoc_macros[tok].flags)) - return(perr(m, ln, pp, ENOCALL)); - return((*mdoc_macros[tok].fp)(m, tok, ln, pp, pos, buf)); } static int -perr(struct mdoc *m, int line, int pos, enum merr type) -{ - char *p; - - p = NULL; - switch (type) { - case (ENOCALL): - p = "not callable"; - break; - case (EPROLBODY): - p = "macro disallowed in document body"; - break; - case (EBODYPROL): - p = "macro disallowed in document prologue"; - break; - case (EMALLOC): - p = "memory exhausted"; - break; - case (ETEXTPROL): - p = "text disallowed in document prologue"; - break; - case (ENOBLANK): - p = "blank lines disallowed in non-literal contexts"; - break; - case (ESPACE): - p = "whitespace disallowed after delimiter"; - break; - } - assert(p); - return(mdoc_perr(m, line, pos, p)); -} - - -static int node_append(struct mdoc *mdoc, struct mdoc_node *p) { @@ -437,6 +388,8 @@ node_append(struct mdoc *mdoc, struct mdoc_node *p) /* NOTREACHED */ } + p->parent->nchild++; + if ( ! mdoc_valid_pre(mdoc, p)) return(0); if ( ! mdoc_action_pre(mdoc, p)) @@ -477,17 +430,17 @@ node_append(struct mdoc *mdoc, struct mdoc_node *p) static struct mdoc_node * -node_alloc(struct mdoc *mdoc, int line, +node_alloc(struct mdoc *m, int line, int pos, int tok, enum mdoc_type type) { struct mdoc_node *p; if (NULL == (p = calloc(1, sizeof(struct mdoc_node)))) { - (void)verr(mdoc, EMALLOC); + (void)mdoc_nerr(m, m->last, EMALLOC); return(NULL); } - p->sec = mdoc->lastsec; + p->sec = m->lastsec; p->line = line; p->pos = pos; p->tok = tok; @@ -499,97 +452,133 @@ node_alloc(struct mdoc *mdoc, int line, int -mdoc_tail_alloc(struct mdoc *mdoc, int line, int pos, int tok) +mdoc_tail_alloc(struct mdoc *m, int line, int pos, int tok) { struct mdoc_node *p; - p = node_alloc(mdoc, line, pos, tok, MDOC_TAIL); + p = node_alloc(m, line, pos, tok, MDOC_TAIL); if (NULL == p) return(0); - return(node_append(mdoc, p)); + if ( ! node_append(m, p)) + return(0); + m->next = MDOC_NEXT_CHILD; + return(1); } int -mdoc_head_alloc(struct mdoc *mdoc, int line, int pos, int tok) +mdoc_head_alloc(struct mdoc *m, int line, int pos, int tok) { struct mdoc_node *p; - assert(mdoc->first); - assert(mdoc->last); + assert(m->first); + assert(m->last); - p = node_alloc(mdoc, line, pos, tok, MDOC_HEAD); + p = node_alloc(m, line, pos, tok, MDOC_HEAD); if (NULL == p) return(0); - return(node_append(mdoc, p)); + if ( ! node_append(m, p)) + return(0); + m->next = MDOC_NEXT_CHILD; + return(1); } int -mdoc_body_alloc(struct mdoc *mdoc, int line, int pos, int tok) +mdoc_body_alloc(struct mdoc *m, int line, int pos, int tok) { struct mdoc_node *p; - p = node_alloc(mdoc, line, pos, tok, MDOC_BODY); + p = node_alloc(m, line, pos, tok, MDOC_BODY); if (NULL == p) return(0); - return(node_append(mdoc, p)); + if ( ! node_append(m, p)) + return(0); + m->next = MDOC_NEXT_CHILD; + return(1); } int -mdoc_block_alloc(struct mdoc *mdoc, int line, int pos, +mdoc_block_alloc(struct mdoc *m, int line, int pos, int tok, struct mdoc_arg *args) { struct mdoc_node *p; - p = node_alloc(mdoc, line, pos, tok, MDOC_BLOCK); + p = node_alloc(m, line, pos, tok, MDOC_BLOCK); if (NULL == p) return(0); p->args = args; if (p->args) (args->refcnt)++; - return(node_append(mdoc, p)); + if ( ! node_append(m, p)) + return(0); + m->next = MDOC_NEXT_CHILD; + return(1); } int -mdoc_elem_alloc(struct mdoc *mdoc, int line, int pos, +mdoc_elem_alloc(struct mdoc *m, int line, int pos, int tok, struct mdoc_arg *args) { struct mdoc_node *p; - p = node_alloc(mdoc, line, pos, tok, MDOC_ELEM); + p = node_alloc(m, line, pos, tok, MDOC_ELEM); if (NULL == p) return(0); p->args = args; if (p->args) (args->refcnt)++; - return(node_append(mdoc, p)); + if ( ! node_append(m, p)) + return(0); + m->next = MDOC_NEXT_CHILD; + return(1); } -int -mdoc_word_alloc(struct mdoc *mdoc, - int line, int pos, const char *word) +static int +pstring(struct mdoc *m, int line, int pos, const char *p, size_t len) { - struct mdoc_node *p; + struct mdoc_node *n; + size_t sv; - p = node_alloc(mdoc, line, pos, -1, MDOC_TEXT); - if (NULL == p) - return(0); - if (NULL == (p->string = strdup(word))) { - (void)verr(mdoc, EMALLOC); - return(0); + n = node_alloc(m, line, pos, -1, MDOC_TEXT); + if (NULL == n) + return(mdoc_nerr(m, m->last, EMALLOC)); + + n->string = malloc(len + 1); + if (NULL == n->string) { + free(n); + return(mdoc_nerr(m, m->last, EMALLOC)); } - return(node_append(mdoc, p)); + + sv = strlcpy(n->string, p, len + 1); + + /* Prohibit truncation. */ + assert(sv < len + 1); + + 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) { + if (p->parent) + p->parent->nchild--; if (p->string) free(p->string); if (p->args) @@ -607,6 +596,7 @@ mdoc_node_freelist(struct mdoc_node *p) if (p->next) mdoc_node_freelist(p->next); + assert(0 == p->nchild); mdoc_node_free(p); } @@ -618,14 +608,52 @@ mdoc_node_freelist(struct mdoc_node *p) static int parsetext(struct mdoc *m, int line, char *buf) { + int i, j; - if (SEC_PROLOGUE == m->lastnamed) - return(perr(m, line, 0, ETEXTPROL)); + 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. + */ - if (0 == buf[0] && ! (MDOC_LITERAL & m->flags)) - return(perr(m, line, 0, ENOBLANK)); + if (MDOC_LITERAL & m->flags) + return(mdoc_word_alloc(m, line, 0, buf)); - if ( ! mdoc_word_alloc(m, line, 0, buf)) + /* Disallow blank/white-space lines in non-literal mode. */ + + for (i = 0; ' ' == buf[i]; i++) + /* Skip leading whitespace. */ ; + if (0 == buf[i]) + return(mdoc_perr(m, line, 0, ENOBLANK)); + + /* + * Break apart a free-form line into tokens. Spaces are + * stripped out of the input. + */ + + 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 = MDOC_NEXT_SIBLING; @@ -633,20 +661,20 @@ parsetext(struct mdoc *m, int line, char *buf) } + + static int macrowarn(struct mdoc *m, int ln, const char *buf) { if ( ! (MDOC_IGN_MACRO & m->pflags)) - return(mdoc_perr(m, ln, 1, + return(mdoc_verr(m, ln, 0, "unknown macro: %s%s", buf, strlen(buf) > 3 ? "..." : "")); - return(mdoc_pwarn(m, ln, 1, WARN_SYNTAX, - "unknown macro: %s%s", + return(mdoc_vwarn(m, ln, 0, "unknown macro: %s%s", buf, strlen(buf) > 3 ? "..." : "")); } - /* * Parse a macro line, that is, a line beginning with the control * character. @@ -654,45 +682,50 @@ macrowarn(struct mdoc *m, int ln, const char *buf) int parsemacro(struct mdoc *m, int ln, char *buf) { - int i, c; + int i, j, c; char mac[5]; - /* Comments and empties are quickly ignored. */ + /* Empty lines are ignored. */ if (0 == buf[1]) return(1); - if (' ' == buf[1]) { - i = 2; + i = 1; + + /* Accept whitespace after the initial control char. */ + + if (' ' == buf[i]) { + i++; while (buf[i] && ' ' == buf[i]) i++; if (0 == buf[i]) return(1); - return(perr(m, ln, 1, ESPACE)); } - if (buf[1] && '\\' == buf[1]) - if (buf[2] && '\"' == buf[2]) - return(1); - /* Copy the first word into a nil-terminated buffer. */ - for (i = 1; i < 5; i++) { - if (0 == (mac[i - 1] = buf[i])) + for (j = 0; j < 4; j++, i++) { + if (0 == (mac[j] = buf[i])) break; else if (' ' == buf[i]) break; + + /* Check for invalid characters. */ + + if (isgraph((u_char)buf[i])) + continue; + return(mdoc_perr(m, ln, i, EPRINT)); } - mac[i - 1] = 0; + mac[j] = 0; - if (i == 5 || i <= 2) { + if (j == 4 || j < 2) { if ( ! macrowarn(m, ln, mac)) goto err; return(1); } - if (MDOC_MAX == (c = mdoc_hash_find(m->htab, mac))) { + if (MDOC_MAX == (c = mdoc_hash_find(mac))) { if ( ! macrowarn(m, ln, mac)) goto err; return(1); @@ -703,8 +736,10 @@ parsemacro(struct mdoc *m, int ln, char *buf) while (buf[i] && ' ' == buf[i]) i++; - /* Begin recursive parse sequence. */ - + /* + * 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)) goto err; @@ -715,3 +750,5 @@ err: /* Error out. */ m->flags |= MDOC_HALT; return(0); } + +