=================================================================== RCS file: /cvs/mandoc/mdoc.c,v retrieving revision 1.121 retrieving revision 1.250 diff -u -p -r1.121 -r1.250 --- mandoc/mdoc.c 2010/04/06 11:33:00 1.121 +++ mandoc/mdoc.c 2015/04/19 14:57:38 1.250 @@ -1,22 +1,21 @@ -/* $Id: mdoc.c,v 1.121 2010/04/06 11:33:00 kristaps Exp $ */ +/* $Id: mdoc.c,v 1.250 2015/04/19 14:57:38 schwarze Exp $ */ /* - * Copyright (c) 2008, 2009 Kristaps Dzonsons + * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons + * Copyright (c) 2010, 2012-2015 Ingo Schwarze * * Permission to use, copy, modify, and distribute this software for any * purpose with or without fee is hereby granted, provided that the above * copyright notice and this permission notice appear in all copies. * - * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES + * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF - * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR + * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. */ -#ifdef HAVE_CONFIG_H #include "config.h" -#endif #include @@ -28,65 +27,16 @@ #include #include -#include "libmdoc.h" +#include "mandoc_aux.h" +#include "mandoc.h" +#include "roff.h" +#include "mdoc.h" #include "libmandoc.h" +#include "roff_int.h" +#include "libmdoc.h" -const char *const __mdoc_merrnames[MERRMAX] = { - "trailing whitespace", /* ETAILWS */ - "unexpected quoted parameter", /* EQUOTPARM */ - "unterminated quoted parameter", /* EQUOTTERM */ - "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 */ - "list arguments preceding type", /* ENOTYPE */ -}; - -const char *const __mdoc_macronames[MDOC_MAX] = { +const char *const __mdoc_macronames[MDOC_MAX + 1] = { + "text", "Ap", "Dd", "Dt", "Os", "Sh", "Ss", "Pp", "D1", "Dl", "Bd", "Ed", "Bl", @@ -97,11 +47,8 @@ const char *const __mdoc_macronames[MDOC_MAX] = { "Ic", "In", "Li", "Nd", "Nm", "Op", "Ot", "Pa", "Rv", "St", "Va", "Vt", - /* LINTED */ "Xr", "%A", "%B", "%D", - /* LINTED */ "%I", "%J", "%N", "%O", - /* LINTED */ "%P", "%R", "%T", "%V", "Ac", "Ao", "Aq", "At", "Bc", "Bf", "Bo", "Bq", @@ -118,695 +65,491 @@ const char *const __mdoc_macronames[MDOC_MAX] = { "Bk", "Ek", "Bt", "Hf", "Fr", "Ud", "Lb", "Lp", "Lk", "Mt", "Brq", "Bro", - /* LINTED */ "Brc", "%C", "Es", "En", - /* LINTED */ "Dx", "%Q", "br", "sp", - /* LINTED */ - "%U" - }; + "%U", "Ta", "ll", +}; -const char *const __mdoc_argnames[MDOC_ARG_MAX] = { +const char *const __mdoc_argnames[MDOC_ARG_MAX] = { "split", "nosplit", "ragged", - "unfilled", "literal", "file", - "offset", "bullet", "dash", - "hyphen", "item", "enum", - "tag", "diag", "hang", - "ohang", "inset", "column", - "width", "compact", "std", + "unfilled", "literal", "file", + "offset", "bullet", "dash", + "hyphen", "item", "enum", + "tag", "diag", "hang", + "ohang", "inset", "column", + "width", "compact", "std", "filled", "words", "emphasis", "symbolic", "nested", "centered" }; -const char * const *mdoc_macronames = __mdoc_macronames; +const char * const *mdoc_macronames = __mdoc_macronames + 1; const char * const *mdoc_argnames = __mdoc_argnames; -static void mdoc_node_free(struct mdoc_node *); -static void mdoc_node_unlink(struct mdoc *, - struct mdoc_node *); -static void mdoc_free1(struct mdoc *); -static void mdoc_alloc1(struct mdoc *); -static struct mdoc_node *node_alloc(struct mdoc *, int, int, - enum mdoct, enum mdoc_type); -static int node_append(struct mdoc *, - struct mdoc_node *); -static int parsetext(struct mdoc *, int, char *); -static int parsemacro(struct mdoc *, int, char *); -static int macrowarn(struct mdoc *, int, const char *); -static int pstring(struct mdoc *, int, int, - const char *, size_t); +static int mdoc_ptext(struct roff_man *, int, char *, int); +static int mdoc_pmacro(struct roff_man *, int, char *, int); -const struct mdoc_node * -mdoc_node(const struct mdoc *m) -{ - return(MDOC_HALT & m->flags ? NULL : m->first); -} - - -const struct mdoc_meta * -mdoc_meta(const struct mdoc *m) +void +mdoc_endparse(struct roff_man *mdoc) { - return(MDOC_HALT & m->flags ? NULL : &m->meta); + mdoc_macroend(mdoc); } - /* - * Frees volatile resources (parse tree, meta-data, fields). + * Main parse routine. Parses a single line -- really just hands off to + * the macro (mdoc_pmacro()) or text parser (mdoc_ptext()). */ -static void -mdoc_free1(struct mdoc *mdoc) +int +mdoc_parseln(struct roff_man *mdoc, int ln, char *buf, int offs) { - if (mdoc->first) - mdoc_node_delete(mdoc, mdoc->first); - if (mdoc->meta.title) - free(mdoc->meta.title); - if (mdoc->meta.os) - free(mdoc->meta.os); - if (mdoc->meta.name) - free(mdoc->meta.name); - if (mdoc->meta.arch) - free(mdoc->meta.arch); - if (mdoc->meta.vol) - free(mdoc->meta.vol); -} + if (mdoc->last->type != ROFFT_EQN || ln > mdoc->last->line) + mdoc->flags |= MDOC_NEWLINE; + /* + * Let the roff nS register switch SYNOPSIS mode early, + * such that the parser knows at all times + * whether this mode is on or off. + * Note that this mode is also switched by the Sh macro. + */ + if (roff_getreg(mdoc->roff, "nS")) + mdoc->flags |= MDOC_SYNOPSIS; + else + mdoc->flags &= ~MDOC_SYNOPSIS; -/* - * Allocate all volatile resources (parse tree, meta-data, fields). - */ -static void -mdoc_alloc1(struct mdoc *mdoc) -{ - - memset(&mdoc->meta, 0, sizeof(struct mdoc_meta)); - mdoc->flags = 0; - mdoc->lastnamed = mdoc->lastsec = SEC_NONE; - mdoc->last = mandoc_calloc(1, sizeof(struct mdoc_node)); - mdoc->first = mdoc->last; - mdoc->last->type = MDOC_ROOT; - mdoc->next = MDOC_NEXT_CHILD; + return(roff_getcontrol(mdoc->roff, buf, &offs) ? + mdoc_pmacro(mdoc, ln, buf, offs) : + mdoc_ptext(mdoc, ln, buf, offs)); } - -/* - * 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. - */ void -mdoc_reset(struct mdoc *mdoc) +mdoc_macro(MACRO_PROT_ARGS) { + assert(tok > TOKEN_NONE && tok < MDOC_MAX); - mdoc_free1(mdoc); - mdoc_alloc1(mdoc); + if (mdoc->flags & MDOC_PBODY) { + if (tok == MDOC_Dt) { + mandoc_vmsg(MANDOCERR_DT_LATE, + mdoc->parse, line, ppos, + "Dt %s", buf + *pos); + return; + } + } else if ( ! (mdoc_macros[tok].flags & MDOC_PROLOGUE)) { + if (mdoc->meta.title == NULL) { + mandoc_vmsg(MANDOCERR_DT_NOTITLE, + mdoc->parse, line, ppos, "%s %s", + mdoc_macronames[tok], buf + *pos); + mdoc->meta.title = mandoc_strdup("UNTITLED"); + } + if (NULL == mdoc->meta.vol) + mdoc->meta.vol = mandoc_strdup("LOCAL"); + mdoc->flags |= MDOC_PBODY; + } + (*mdoc_macros[tok].fp)(mdoc, tok, line, ppos, pos, buf); } - -/* - * 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_tail_alloc(struct roff_man *mdoc, int line, int pos, int tok) { + struct roff_node *p; - mdoc_free1(mdoc); - free(mdoc); + p = roff_node_alloc(mdoc, line, pos, ROFFT_TAIL, tok); + roff_node_append(mdoc, p); + mdoc->next = ROFF_NEXT_CHILD; } - -/* - * Allocate volatile and non-volatile parse resources. - */ -struct mdoc * -mdoc_alloc(void *data, int pflags, const struct mdoc_cb *cb) +struct roff_node * +mdoc_endbody_alloc(struct roff_man *mdoc, int line, int pos, int tok, + struct roff_node *body, enum mdoc_endbody end) { - struct mdoc *p; + struct roff_node *p; - p = mandoc_calloc(1, sizeof(struct mdoc)); - - if (cb) - memcpy(&p->cb, cb, sizeof(struct mdoc_cb)); - - p->data = data; - p->pflags = pflags; - - mdoc_hash_init(); - mdoc_alloc1(p); + body->flags |= MDOC_ENDED; + body->parent->flags |= MDOC_ENDED; + p = roff_node_alloc(mdoc, line, pos, ROFFT_BODY, tok); + p->body = body; + p->norm = body->norm; + p->end = end; + roff_node_append(mdoc, p); + mdoc->next = ROFF_NEXT_SIBLING; return(p); } - -/* - * Climb back up the parse tree, validating open scopes. Mostly calls - * through to macro_end() in macro.c. - */ -int -mdoc_endparse(struct mdoc *m) +struct roff_node * +mdoc_block_alloc(struct roff_man *mdoc, int line, int pos, + int tok, struct mdoc_arg *args) { + struct roff_node *p; - if (MDOC_HALT & m->flags) - return(0); - else if (mdoc_macroend(m)) - return(1); - m->flags |= MDOC_HALT; - return(0); -} + p = roff_node_alloc(mdoc, line, pos, ROFFT_BLOCK, tok); + p->args = args; + if (p->args) + (args->refcnt)++; - -/* - * Main parse routine. Parses a single line -- really just hands off to - * the macro (parsemacro()) or text parser (parsetext()). - */ -int -mdoc_parseln(struct mdoc *m, int ln, char *buf) -{ - - if (MDOC_HALT & m->flags) - return(0); - - return('.' == *buf ? parsemacro(m, ln, buf) : - parsetext(m, ln, buf)); + switch (tok) { + case MDOC_Bd: + /* FALLTHROUGH */ + case MDOC_Bf: + /* FALLTHROUGH */ + case MDOC_Bl: + /* FALLTHROUGH */ + case MDOC_En: + /* FALLTHROUGH */ + case MDOC_Rs: + p->norm = mandoc_calloc(1, sizeof(union mdoc_data)); + break; + default: + break; + } + roff_node_append(mdoc, p); + mdoc->next = ROFF_NEXT_CHILD; + return(p); } - -int -mdoc_verr(struct mdoc *mdoc, int ln, int pos, - const char *fmt, ...) +void +mdoc_elem_alloc(struct roff_man *mdoc, int line, int pos, + int tok, struct mdoc_arg *args) { - char buf[256]; - va_list ap; + struct roff_node *p; - if (NULL == mdoc->cb.mdoc_err) - return(0); + p = roff_node_alloc(mdoc, line, pos, ROFFT_ELEM, tok); + p->args = args; + if (p->args) + (args->refcnt)++; - 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)); + switch (tok) { + case MDOC_An: + p->norm = mandoc_calloc(1, sizeof(union mdoc_data)); + break; + default: + break; + } + roff_node_append(mdoc, p); + mdoc->next = ROFF_NEXT_CHILD; } - -int -mdoc_vwarn(struct mdoc *mdoc, int ln, int pos, const char *fmt, ...) +void +mdoc_node_relink(struct roff_man *mdoc, struct roff_node *p) { - 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, ln, pos, buf)); + roff_node_unlink(mdoc, p); + roff_node_append(mdoc, p); } - -int -mdoc_err(struct mdoc *m, int line, int pos, int iserr, enum merr type) +/* + * Parse free-form text, that is, a line that does not begin with the + * control character. + */ +static int +mdoc_ptext(struct roff_man *mdoc, int line, char *buf, int offs) { - const char *p; + struct roff_node *n; + char *c, *ws, *end; - p = __mdoc_merrnames[(int)type]; - assert(p); + assert(mdoc->last); + n = mdoc->last; - if (iserr) - return(mdoc_verr(m, line, pos, p)); - - return(mdoc_vwarn(m, line, pos, p)); -} - - -int -mdoc_macro(struct mdoc *m, enum mdoct tok, - int ln, int pp, int *pos, char *buf) -{ - - assert(tok < MDOC_MAX); /* - * If we're in the prologue, deny "body" macros. Similarly, if - * we're in the body, deny prologue calls. + * Divert directly to list processing if we're encountering a + * columnar ROFFT_BLOCK with or without a prior ROFFT_BLOCK entry + * (a ROFFT_BODY means it's already open, in which case we should + * process within its context in the normal way). */ - if (MDOC_PROLOGUE & mdoc_macros[tok].flags && - MDOC_PBODY & m->flags) { - if ( ! mdoc_pwarn(m, ln, pp, EBODYPROL)) - return(0); - /* - * FIXME: do this in mdoc_action.c. - */ - if (NULL == m->meta.title) - m->meta.title = mandoc_strdup("unknown"); - if (NULL == m->meta.vol) - m->meta.vol = mandoc_strdup("local"); - if (NULL == m->meta.os) - m->meta.os = mandoc_strdup("local"); - if (0 == m->meta.date) - m->meta.date = time(NULL); - m->flags |= MDOC_PBODY; - } - if ( ! (MDOC_PROLOGUE & mdoc_macros[tok].flags) && - ! (MDOC_PBODY & m->flags)) - return(mdoc_perr(m, ln, pp, EBODYPROL)); - return((*mdoc_macros[tok].fp)(m, tok, ln, pp, pos, buf)); -} - - -static int -node_append(struct mdoc *mdoc, struct mdoc_node *p) -{ - - assert(mdoc->last); - assert(mdoc->first); - assert(MDOC_ROOT != p->type); - - switch (mdoc->next) { - case (MDOC_NEXT_SIBLING): - mdoc->last->next = p; - p->prev = mdoc->last; - p->parent = mdoc->last->parent; - break; - case (MDOC_NEXT_CHILD): - mdoc->last->child = p; - p->parent = mdoc->last; - break; - default: - abort(); - /* NOTREACHED */ + if (n->tok == MDOC_Bl && n->type == ROFFT_BODY && + n->end == ENDBODY_NOT && n->norm->Bl.type == LIST_column) { + /* `Bl' is open without any children. */ + mdoc->flags |= MDOC_FREECOL; + mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf); + return(1); } - p->parent->nchild++; - - if ( ! mdoc_valid_pre(mdoc, p)) - return(0); - if ( ! mdoc_action_pre(mdoc, p)) - return(0); - - switch (p->type) { - case (MDOC_HEAD): - assert(MDOC_BLOCK == p->parent->type); - p->parent->head = p; - break; - case (MDOC_TAIL): - assert(MDOC_BLOCK == p->parent->type); - p->parent->tail = p; - break; - case (MDOC_BODY): - assert(MDOC_BLOCK == p->parent->type); - p->parent->body = p; - break; - default: - break; + if (n->tok == MDOC_It && n->type == ROFFT_BLOCK && + NULL != n->parent && + MDOC_Bl == n->parent->tok && + LIST_column == n->parent->norm->Bl.type) { + /* `Bl' has block-level `It' children. */ + mdoc->flags |= MDOC_FREECOL; + mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf); + return(1); } - mdoc->last = p; + /* + * Search for the beginning of unescaped trailing whitespace (ws) + * and for the first character not to be output (end). + */ - switch (p->type) { - case (MDOC_TEXT): - if ( ! mdoc_valid_post(mdoc)) - return(0); - if ( ! mdoc_action_post(mdoc)) - return(0); - break; - default: - break; + /* FIXME: replace with strcspn(). */ + ws = NULL; + for (c = end = buf + offs; *c; c++) { + switch (*c) { + case ' ': + if (NULL == ws) + ws = c; + continue; + case '\t': + /* + * Always warn about trailing tabs, + * even outside literal context, + * where they should be put on the next line. + */ + if (NULL == ws) + ws = c; + /* + * Strip trailing tabs in literal context only; + * outside, they affect the next line. + */ + if (MDOC_LITERAL & mdoc->flags) + continue; + break; + case '\\': + /* Skip the escaped character, too, if any. */ + if (c[1]) + c++; + /* FALLTHROUGH */ + default: + ws = NULL; + break; + } + end = c + 1; } + *end = '\0'; - return(1); -} + if (ws) + mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse, + line, (int)(ws-buf), NULL); + if (buf[offs] == '\0' && ! (mdoc->flags & MDOC_LITERAL)) { + mandoc_msg(MANDOCERR_FI_BLANK, mdoc->parse, + line, (int)(c - buf), NULL); -static struct mdoc_node * -node_alloc(struct mdoc *m, int line, int pos, - enum mdoct tok, enum mdoc_type type) -{ - struct mdoc_node *p; + /* + * Insert a `sp' 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. + */ + roff_elem_alloc(mdoc, line, offs, MDOC_sp); + mdoc->next = ROFF_NEXT_SIBLING; + mdoc_valid_post(mdoc); + return(1); + } - p = mandoc_calloc(1, sizeof(struct mdoc_node)); - p->sec = m->lastsec; - p->line = line; - p->pos = pos; - p->tok = tok; - p->type = type; + roff_word_alloc(mdoc, line, offs, buf+offs); - return(p); -} + if (mdoc->flags & MDOC_LITERAL) + return(1); + /* + * 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. + */ -int -mdoc_tail_alloc(struct mdoc *m, int line, int pos, enum mdoct tok) -{ - struct mdoc_node *p; + assert(buf < end); - p = node_alloc(m, line, pos, tok, MDOC_TAIL); - if ( ! node_append(m, p)) - return(0); - m->next = MDOC_NEXT_CHILD; + if (mandoc_eos(buf+offs, (size_t)(end-buf-offs))) + mdoc->last->flags |= MDOC_EOS; return(1); } - -int -mdoc_head_alloc(struct mdoc *m, int line, int pos, enum mdoct tok) -{ - struct mdoc_node *p; - - assert(m->first); - assert(m->last); - - p = node_alloc(m, line, pos, tok, MDOC_HEAD); - if ( ! node_append(m, p)) - return(0); - m->next = MDOC_NEXT_CHILD; - return(1); -} - - -int -mdoc_body_alloc(struct mdoc *m, int line, int pos, enum mdoct tok) -{ - struct mdoc_node *p; - - p = node_alloc(m, line, pos, tok, MDOC_BODY); - if ( ! node_append(m, p)) - return(0); - m->next = MDOC_NEXT_CHILD; - return(1); -} - - -int -mdoc_block_alloc(struct mdoc *m, int line, int pos, - enum mdoct tok, struct mdoc_arg *args) -{ - struct mdoc_node *p; - - p = node_alloc(m, line, pos, tok, MDOC_BLOCK); - p->args = args; - if (p->args) - (args->refcnt)++; - if ( ! node_append(m, p)) - return(0); - m->next = MDOC_NEXT_CHILD; - return(1); -} - - -int -mdoc_elem_alloc(struct mdoc *m, int line, int pos, - enum mdoct tok, struct mdoc_arg *args) -{ - struct mdoc_node *p; - - p = node_alloc(m, line, pos, tok, MDOC_ELEM); - p->args = args; - if (p->args) - (args->refcnt)++; - if ( ! node_append(m, p)) - return(0); - m->next = MDOC_NEXT_CHILD; - return(1); -} - - +/* + * Parse a macro line, that is, a line beginning with the control + * character. + */ static int -pstring(struct mdoc *m, int line, int pos, const char *p, size_t len) +mdoc_pmacro(struct roff_man *mdoc, int ln, char *buf, int offs) { - struct mdoc_node *n; - size_t sv; + struct roff_node *n; + const char *cp; + int tok; + int i, sv; + char mac[5]; - n = node_alloc(m, line, pos, -1, MDOC_TEXT); - n->string = mandoc_malloc(len + 1); - sv = strlcpy(n->string, p, len + 1); + sv = offs; - /* Prohibit truncation. */ - assert(sv < len + 1); + /* + * Copy the first word into a nil-terminated buffer. + * Stop when a space, tab, escape, or eoln is encountered. + */ - if ( ! node_append(m, n)) - return(0); - m->next = MDOC_NEXT_SIBLING; - return(1); -} + i = 0; + while (i < 4 && strchr(" \t\\", buf[offs]) == NULL) + mac[i++] = buf[offs++]; + mac[i] = '\0'; -int -mdoc_word_alloc(struct mdoc *m, int line, int pos, const char *p) -{ + tok = (i > 1 && i < 4) ? mdoc_hash_find(mac) : TOKEN_NONE; - return(pstring(m, line, pos, p, strlen(p))); -} + if (tok == TOKEN_NONE) { + mandoc_msg(MANDOCERR_MACRO, mdoc->parse, + ln, sv, buf + sv - 1); + return(1); + } + /* Skip a leading escape sequence or tab. */ -void -mdoc_node_free(struct mdoc_node *p) -{ - - if (p->string) - free(p->string); - if (p->args) - mdoc_argv_free(p->args); - free(p); -} - - -static void -mdoc_node_unlink(struct mdoc *m, struct mdoc_node *n) -{ - - /* Adjust siblings. */ - - if (n->prev) - n->prev->next = n->next; - if (n->next) - n->next->prev = n->prev; - - /* Adjust parent. */ - - if (n->parent) { - n->parent->nchild--; - if (n->parent->child == n) - n->parent->child = n->prev ? n->prev : n->next; + switch (buf[offs]) { + case '\\': + cp = buf + offs + 1; + mandoc_escape(&cp, NULL, NULL); + offs = cp - buf; + break; + case '\t': + offs++; + break; + default: + break; } - /* Adjust parse point, if applicable. */ + /* Jump to the next non-whitespace word. */ - if (m && m->last == n) { - if (n->prev) { - m->last = n->prev; - m->next = MDOC_NEXT_SIBLING; - } else { - m->last = n->parent; - m->next = MDOC_NEXT_CHILD; - } - } + while (buf[offs] && ' ' == buf[offs]) + offs++; - if (m && m->first == n) - m->first = NULL; -} + /* + * 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[offs] && ' ' == buf[offs - 1]) + mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse, + ln, offs - 1, NULL); -void -mdoc_node_delete(struct mdoc *m, struct mdoc_node *p) -{ + /* + * If an initial macro or a list invocation, divert directly + * into macro processing. + */ - while (p->child) { - assert(p->nchild); - mdoc_node_delete(m, p->child); + if (NULL == mdoc->last || MDOC_It == tok || MDOC_El == tok) { + mdoc_macro(mdoc, tok, ln, sv, &offs, buf); + return(1); } - assert(0 == p->nchild); - mdoc_node_unlink(m, p); - mdoc_node_free(p); -} + n = mdoc->last; + assert(mdoc->last); - -/* - * Parse free-form text, that is, a line that does not begin with the - * control character. - */ -static int -parsetext(struct mdoc *m, int line, char *buf) -{ - int i, j; - char sv; - - 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 the first macro of a `Bl -column', open an `It' block + * context around the parsed macro. */ - if (MDOC_LITERAL & m->flags) - return(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]) { - 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. - */ - if ( ! mdoc_elem_alloc(m, line, 0, MDOC_Pp, NULL)) - return(0); + if (n->tok == MDOC_Bl && n->type == ROFFT_BODY && + n->end == ENDBODY_NOT && n->norm->Bl.type == LIST_column) { + mdoc->flags |= MDOC_FREECOL; + mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf); + return(1); } /* - * Break apart a free-form line into tokens. Spaces are - * stripped out of the input. + * If we're following a block-level `It' within a `Bl -column' + * context (perhaps opened in the above block or in ptext()), + * then open an `It' block context around the parsed macro. */ - for (j = i; buf[i]; i++) { - if (' ' != buf[i]) - continue; + if (n->tok == MDOC_It && n->type == ROFFT_BLOCK && + NULL != n->parent && + MDOC_Bl == n->parent->tok && + LIST_column == n->parent->norm->Bl.type) { + mdoc->flags |= MDOC_FREECOL; + mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf); + return(1); + } - /* Escaped whitespace. */ - if (i && ' ' == buf[i] && '\\' == buf[i - 1]) - continue; + /* Normal processing of a macro. */ - sv = buf[i]; - buf[i++] = '\0'; + mdoc_macro(mdoc, tok, ln, sv, &offs, buf); - if ( ! pstring(m, line, j, &buf[j], (size_t)(i - j))) - return(0); + /* In quick mode (for mandocdb), abort after the NAME section. */ - /* Trailing whitespace? Check at overwritten byte. */ + if (mdoc->quick && MDOC_Sh == tok && + SEC_NAME != mdoc->last->sec) + return(2); - if (' ' == sv && '\0' == buf[i]) - if ( ! mdoc_pwarn(m, line, i - 1, ETAILWS)) - return(0); - - for ( ; ' ' == buf[i]; i++) - /* Skip trailing whitespace. */ ; - - j = i; - - /* Trailing whitespace? */ - - if (' ' == buf[i - 1] && '\0' == buf[i]) - if ( ! mdoc_pwarn(m, line, i - 1, ETAILWS)) - return(0); - - 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; return(1); } - - -static int -macrowarn(struct mdoc *m, int ln, const char *buf) +enum mdelim +mdoc_isdelim(const char *p) { - if ( ! (MDOC_IGN_MACRO & m->pflags)) - 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 ? "..." : "")); -} + if ('\0' == p[0]) + return(DELIM_NONE); -/* - * Parse a macro line, that is, a line beginning with the control - * character. - */ -int -parsemacro(struct mdoc *m, int ln, char *buf) -{ - int i, j, c; - char mac[5]; + if ('\0' == p[1]) + switch (p[0]) { + case '(': + /* FALLTHROUGH */ + case '[': + return(DELIM_OPEN); + case '|': + return(DELIM_MIDDLE); + case '.': + /* FALLTHROUGH */ + case ',': + /* FALLTHROUGH */ + case ';': + /* FALLTHROUGH */ + case ':': + /* FALLTHROUGH */ + case '?': + /* FALLTHROUGH */ + case '!': + /* FALLTHROUGH */ + case ')': + /* FALLTHROUGH */ + case ']': + return(DELIM_CLOSE); + default: + return(DELIM_NONE); + } - /* Empty lines are ignored. */ + if ('\\' != p[0]) + return(DELIM_NONE); - if ('\0' == buf[1]) - return(1); + if (0 == strcmp(p + 1, ".")) + return(DELIM_CLOSE); + if (0 == strcmp(p + 1, "fR|\\fP")) + return(DELIM_MIDDLE); - i = 1; + return(DELIM_NONE); +} - /* Accept whitespace after the initial control char. */ +void +mdoc_deroff(char **dest, const struct roff_node *n) +{ + char *cp; + size_t sz; - if (' ' == buf[i]) { - i++; - while (buf[i] && ' ' == buf[i]) - i++; - if ('\0' == buf[i]) - return(1); + if (n->type != ROFFT_TEXT) { + for (n = n->child; n; n = n->next) + mdoc_deroff(dest, n); + return; } - /* Copy the first word into a nil-terminated buffer. */ + /* Skip leading whitespace. */ - for (j = 0; j < 4; j++, i++) { - if ('\0' == (mac[j] = buf[i])) + for (cp = n->string; '\0' != *cp; cp++) + if (0 == isspace((unsigned char)*cp)) break; - else if (' ' == buf[i]) - break; - /* Check for invalid characters. */ + /* Skip trailing whitespace. */ - if (isgraph((u_char)buf[i])) - continue; - return(mdoc_perr(m, ln, i, EPRINT)); - } + for (sz = strlen(cp); sz; sz--) + if (0 == isspace((unsigned char)cp[sz-1])) + break; - mac[j] = 0; + /* Skip empty strings. */ - if (j == 4 || j < 2) { - if ( ! macrowarn(m, ln, mac)) - goto err; - return(1); - } - - if (MDOC_MAX == (c = mdoc_hash_find(mac))) { - if ( ! macrowarn(m, ln, mac)) - goto err; - return(1); + if (0 == sz) + return; + + if (NULL == *dest) { + *dest = mandoc_strndup(cp, sz); + return; } - /* The macro is sane. Jump to the next word. */ - - while (buf[i] && ' ' == buf[i]) - i++; - - /* Trailing whitespace? */ - - if ('\0' == buf[i] && ' ' == buf[i - 1]) - if ( ! mdoc_pwarn(m, ln, i - 1, ETAILWS)) - goto err; - - /* - * 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; - - return(1); - -err: /* Error out. */ - - m->flags |= MDOC_HALT; - return(0); + mandoc_asprintf(&cp, "%s %*s", *dest, (int)sz, cp); + free(*dest); + *dest = cp; } - -