diff options
| -rw-r--r-- | argv.c | 188 | ||||
| -rw-r--r-- | macro.c | 476 | ||||
| -rw-r--r-- | mdoc.c | 16 | ||||
| -rw-r--r-- | mdoc.h | 15 | ||||
| -rw-r--r-- | mdocml.c | 24 | ||||
| -rw-r--r-- | private.h | 15 | ||||
| -rw-r--r-- | strings.c | 21 |
7 files changed, 488 insertions, 267 deletions
@@ -26,17 +26,49 @@ #include "private.h" -static int parse_next(struct mdoc *, int, - int *, char *, char **); +static int lookup(int, const char *); +static int parse(struct mdoc *, int, + struct mdoc_arg *, int *, char *); +static int postparse(struct mdoc *, int, + const struct mdoc_arg *, int); -static int -parse_next(struct mdoc *mdoc, int tok, - int *pos, char *buf, char **v) +int +mdoc_args(struct mdoc *mdoc, int tok, int *pos, char *buf, int fl, char **v) { + int i; if (0 == buf[*pos]) - return(0); + return(ARGS_EOLN); + + if ('\"' == buf[*pos] && ! (fl & ARGS_QUOTED)) + if ( ! mdoc_warn(mdoc, tok, *pos, WARN_SYNTAX_QUOTED)) + return(ARGS_ERROR); + + if ('-' == buf[*pos]) + if ( ! mdoc_warn(mdoc, tok, *pos, WARN_SYNTAX_ARGLIKE)) + return(ARGS_ERROR); + + if ((fl & ARGS_DELIM) && mdoc_iscdelim(buf[*pos])) { + for (i = *pos; buf[i]; ) { + if ( ! mdoc_iscdelim(buf[i])) + break; + i++; + if (0 == buf[i] || ! isspace(buf[i])) + break; + i++; + while (buf[i] && isspace(buf[i])) + i++; + } + if (0 == buf[i]) { + *v = &buf[*pos]; + return(ARGS_PUNCT); + } + } + + /* + * Parse routine for non-quoted string. + */ if ('\"' != buf[*pos]) { *v = &buf[*pos]; @@ -45,26 +77,31 @@ parse_next(struct mdoc *mdoc, int tok, (*pos)++; if (0 == buf[*pos]) - return(1); + return(ARGS_WORD); buf[(*pos)++] = 0; if (0 == buf[*pos]) - return(1); + return(ARGS_WORD); while (buf[*pos] && isspace(buf[*pos])) (*pos)++; if (buf[*pos]) - return(1); + return(ARGS_WORD); if ( ! mdoc_warn(mdoc, tok, *pos, WARN_SYNTAX_WS_EOLN)) - return(-1); - return(1); - } + return(ARGS_ERROR); - if ('-' == buf[*pos]) - if ( ! mdoc_warn(mdoc, tok, *pos, WARN_SYNTAX_ARGLIKE)) - return(-1); + return(ARGS_WORD); + } + + /* + * If we're a quoted string (and quoted strings are allowed), + * then parse ahead to the next quote. If none's found, it's an + * error. After, parse to the next word. We're not allowed to + * also be DELIM requests (for now). + */ + assert( ! (fl & ARGS_DELIM)); *v = &buf[++(*pos)]; @@ -73,31 +110,32 @@ parse_next(struct mdoc *mdoc, int tok, if (0 == buf[*pos]) { (void)mdoc_err(mdoc, tok, *pos, ERR_SYNTAX_UNQUOTE); - return(-1); + return(ARGS_ERROR); } buf[(*pos)++] = 0; if (0 == buf[*pos]) - return(1); + return(ARGS_WORD); while (buf[*pos] && isspace(buf[*pos])) (*pos)++; if (buf[*pos]) - return(1); + return(ARGS_WORD); if ( ! mdoc_warn(mdoc, tok, *pos, WARN_SYNTAX_WS_EOLN)) - return(-1); - return(1); + return(ARGS_ERROR); + + return(ARGS_WORD); } -int -mdoc_argv_lookup(int tok, const char *argv) +static int +lookup(int tok, const char *argv) { switch (tok) { - case (MDOC_Bl): + case (MDOC_Bd): if (xstrcmp(argv, "ragged")) return(MDOC_Ragged); else if (xstrcmp(argv, "unfilled")) @@ -108,7 +146,10 @@ mdoc_argv_lookup(int tok, const char *argv) return(MDOC_File); else if (xstrcmp(argv, "offset")) return(MDOC_Offset); - else if (xstrcmp(argv, "bullet")) + break; + + case (MDOC_Bl): + if (xstrcmp(argv, "bullet")) return(MDOC_Bullet); else if (xstrcmp(argv, "dash")) return(MDOC_Dash); @@ -132,10 +173,12 @@ mdoc_argv_lookup(int tok, const char *argv) return(MDOC_Column); else if (xstrcmp(argv, "width")) return(MDOC_Width); + else if (xstrcmp(argv, "offset")) + return(MDOC_Offset); else if (xstrcmp(argv, "compact")) return(MDOC_Compact); - break; + default: abort(); /* NOTREACHED */ @@ -145,17 +188,43 @@ mdoc_argv_lookup(int tok, const char *argv) } -int -mdoc_argv_parse(struct mdoc *mdoc, int tok, int arg, +static int +postparse(struct mdoc *mdoc, int tok, const struct mdoc_arg *v, int pos) +{ + + switch (v->arg) { + case (MDOC_Offset): + assert(v->value); + assert(v->value[0]); + if (xstrcmp(v->value[0], "left")) + break; + if (xstrcmp(v->value[0], "right")) + break; + if (xstrcmp(v->value[0], "center")) + break; + if (xstrcmp(v->value[0], "indent")) + break; + if (xstrcmp(v->value[0], "indent-two")) + break; + return(mdoc_err(mdoc, tok, pos, ERR_SYNTAX_ARGBAD)); + default: + break; + } + + return(1); +} + + +static int +parse(struct mdoc *mdoc, int tok, struct mdoc_arg *v, int *pos, char *buf) { char *p; int c, ppos, i; - v->arg = arg; ppos = *pos; - switch (arg) { + switch (v->arg) { case(MDOC_Compact): /* FALLTHROUGH */ case(MDOC_Ragged): @@ -195,10 +264,10 @@ mdoc_argv_parse(struct mdoc *mdoc, int tok, int arg, /* * This has a single value for an argument. */ - c = parse_next(mdoc, tok, pos, buf, &p); - if (-1 == c) + c = mdoc_args(mdoc, tok, pos, buf, ARGS_QUOTED, &p); + if (ARGS_ERROR == c) return(0); - else if (0 == c) + else if (ARGS_EOLN == c) return(mdoc_err(mdoc, tok, ppos, ERR_SYNTAX_ARGVAL)); v->sz = 1; @@ -215,11 +284,11 @@ mdoc_argv_parse(struct mdoc *mdoc, int tok, int arg, v->sz = 0; v->value = xcalloc(MDOC_LINEARG_MAX, sizeof(char *)); for (i = 0; i < MDOC_LINEARG_MAX; i++) { - c = parse_next(mdoc, tok, pos, buf, &p); - if (-1 == c) { + c = mdoc_args(mdoc, tok, pos, buf, ARGS_QUOTED, &p); + if (ARGS_ERROR == c) { free(v->value); return(0); - } else if (0 == c) + } else if (ARGS_EOLN == c) break; v->value[i] = p; } @@ -240,6 +309,54 @@ mdoc_argv_parse(struct mdoc *mdoc, int tok, int arg, } +int +mdoc_argv(struct mdoc *mdoc, int tok, + struct mdoc_arg *v, int *pos, char *buf) +{ + int i, ppos; + char *argv; + + (void)memset(v, 0, sizeof(struct mdoc_arg)); + + if (0 == buf[*pos]) + return(0); + + assert( ! isspace(buf[*pos])); + + if ('-' != buf[*pos]) { + (void)mdoc_err(mdoc, tok, *pos, ERR_SYNTAX_ARGFORM); + return(-1); + } + + i = *pos; + argv = &buf[++(*pos)]; + + while (buf[*pos] && ! isspace(buf[*pos])) + (*pos)++; + + if (buf[*pos]) + buf[(*pos)++] = 0; + + if (MDOC_ARG_MAX == (v->arg = lookup(tok, argv))) { + (void)mdoc_err(mdoc, tok, i, ERR_SYNTAX_ARG); + return(-1); + } + + while (buf[*pos] && isspace(buf[*pos])) + (*pos)++; + + /* FIXME: whitespace if no value. */ + + ppos = *pos; + if ( ! parse(mdoc, tok, v, pos, buf)) + return(-1); + if ( ! postparse(mdoc, tok, v, ppos)) + return(-1); + + return(1); +} + + void mdoc_argv_free(int sz, struct mdoc_arg *arg) { @@ -254,3 +371,4 @@ mdoc_argv_free(int sz, struct mdoc_arg *arg) free(arg[i].value); } } + @@ -27,168 +27,93 @@ #define _CC(p) ((const char **)p) static int scope_rewind_exp(struct mdoc *, int, int, int); +static int scope_rewind_imp(struct mdoc *, int, int); static int append_text(struct mdoc *, int, int, int, char *[]); static int append_scoped(struct mdoc *, int, int, int, const char *[], int, const struct mdoc_arg *); -static int args_next(struct mdoc *, int, - int *, char *, char **); -static int argv_next(struct mdoc *, int, - int *, char *, struct mdoc_arg *); -static int args_next_quoted(struct mdoc *, int, - int *, char *, char **); +static int append_delims(struct mdoc *, int, int *, char *); static int -args_next_quoted(struct mdoc *mdoc, int tok, - int *pos, char *buf, char **v) +append_delims(struct mdoc *mdoc, int tok, int *pos, char *buf) { + int c, lastarg; + char *p; if (0 == buf[*pos]) - return(0); - - assert( ! isspace(buf[*pos])); - - if ('\"' != buf[*pos]) - return(args_next(mdoc, tok, pos, buf, v)); - - *v = &buf[++(*pos)]; - - while (buf[*pos] && '\"' != buf[*pos]) - (*pos)++; - - if (0 == buf[*pos]) { - (void)mdoc_err(mdoc, tok, *pos, ERR_SYNTAX_UNQUOTE); - return(-1); - } - - buf[(*pos)++] = 0; - if (0 == buf[*pos]) return(1); - while (buf[*pos] && isspace(buf[*pos])) - (*pos)++; + mdoc_msg(mdoc, *pos, "appending delimiters"); - if (0 == buf[*pos]) - if ( ! mdoc_warn(mdoc, tok, *pos, WARN_SYNTAX_WS_EOLN)) - return(-1); + for (;;) { + lastarg = *pos; + c = mdoc_args(mdoc, tok, pos, buf, 0, &p); + if (ARGS_ERROR == c) + return(0); + else if (ARGS_EOLN == c) + break; + assert(mdoc_isdelim(p)); + mdoc_word_alloc(mdoc, lastarg, p); + } return(1); } static int -scope_rewind_exp(struct mdoc *mdoc, int ppos, int tok, int dst) +scope_rewind_imp(struct mdoc *mdoc, int ppos, int tok) { struct mdoc_node *n; + int t; + + n = mdoc->last ? mdoc->last->parent : NULL; /* LINTED */ - for (n = mdoc->last; n; n = n->parent) { + for ( ; n; n = n->parent) { if (MDOC_BLOCK != n->type) continue; - if (dst == n->data.block.tok) + if (tok == (t = n->data.block.tok)) break; + if ( ! (MDOC_EXPLICIT & mdoc_macros[t].flags)) + continue; return(mdoc_err(mdoc, tok, ppos, ERR_SCOPE_BREAK)); } - if (NULL == (mdoc->last = n)) - return(mdoc_err(mdoc, tok, ppos, ERR_SCOPE_NOCTX)); - - mdoc_msg(mdoc, ppos, "scope: rewound `%s' to `%s'", - mdoc_macronames[tok], mdoc_macronames[dst]); - - return(1); -} - - -static int -argv_next(struct mdoc *mdoc, int tok, - int *pos, char *buf, struct mdoc_arg *v) -{ - char *argv; - int i, val; - - if (0 == buf[*pos]) - return(0); - - assert( ! isspace(buf[*pos])); - - if ('-' != buf[*pos]) { - (void)mdoc_err(mdoc, tok, *pos, ERR_SYNTAX_ARGS); - return(-1); - } - - i = *pos; - argv = &buf[++(*pos)]; - - while (buf[*pos] && ! isspace(buf[*pos])) - (*pos)++; - - if (buf[*pos]) - buf[(*pos)++] = 0; - - /* - * XXX This is a little bit ugly. The mdoc_argv structure - * points to a pointer array, which we allocate on-the-fly in - * this function. If there's any failure, we need to release - * this memory, which is done by the caller of this function - * with mdoc_argv_free. Ew. This should be simpler. - */ - - if (MDOC_ARG_MAX == (val = mdoc_argv_lookup(tok, argv))) { - (void)mdoc_err(mdoc, tok, i, ERR_SYNTAX_BADARG); - return(-1); - } - - while (buf[*pos] && isspace(buf[*pos])) - (*pos)++; - - if ( ! mdoc_argv_parse(mdoc, tok, val, v, pos, buf)) - return(-1); + if (n) { + mdoc->last = n; + mdoc_msg(mdoc, ppos, "scope: rewound implicit `%s'", + mdoc_macronames[tok]); + return(1); + } + mdoc_msg(mdoc, ppos, "scope: new implicit `%s'", + mdoc_macronames[tok]); return(1); } static int -args_next(struct mdoc *mdoc, int tok, - int *pos, char *buf, char **v) +scope_rewind_exp(struct mdoc *mdoc, int ppos, int tok, int dst) { + struct mdoc_node *n; - if (0 == buf[*pos]) - return(0); - - assert( ! isspace(buf[*pos])); - - if ('\"' == buf[*pos]) { - (void)mdoc_err(mdoc, tok, *pos, ERR_SYNTAX_QUOTE); - return(-1); - } - - *v = &buf[*pos]; - - /* Scan ahead to end of token. */ - - while (buf[*pos] && ! isspace(buf[*pos])) - (*pos)++; + assert(mdoc->last); - if (buf[*pos] && buf[*pos + 1] && '\\' == buf[*pos]) { - (void)mdoc_err(mdoc, tok, *pos, ERR_SYNTAX_WS); - return(-1); + /* LINTED */ + for (n = mdoc->last->parent; n; n = n->parent) { + if (MDOC_BLOCK != n->type) + continue; + if (dst == n->data.block.tok) + break; + return(mdoc_err(mdoc, tok, ppos, ERR_SCOPE_BREAK)); } - if (0 == buf[*pos]) - return(1); - - /* Scan ahead over trailing whitespace. */ - - buf[(*pos)++] = 0; - while (buf[*pos] && isspace(buf[*pos])) - (*pos)++; + if (NULL == (mdoc->last = n)) + return(mdoc_err(mdoc, tok, ppos, ERR_SCOPE_NOCTX)); - if (0 == buf[*pos]) - if ( ! mdoc_warn(mdoc, tok, *pos, WARN_SYNTAX_WS_EOLN)) - return(-1); + mdoc_msg(mdoc, ppos, "scope: rewound explicit `%s' to `%s'", + mdoc_macronames[tok], mdoc_macronames[dst]); return(1); } @@ -199,7 +124,8 @@ append_scoped(struct mdoc *mdoc, int tok, int pos, int sz, const char *args[], int argc, const struct mdoc_arg *argv) { - enum mdoc_sec sec; + enum mdoc_sec sec; + struct mdoc_node *node; switch (tok) { /* ======= ADD MORE MACRO CHECKS BELOW. ======= */ @@ -224,6 +150,17 @@ append_scoped(struct mdoc *mdoc, int tok, int pos, if (0 == sz) return(mdoc_err(mdoc, tok, pos, ERR_ARGS_GE1)); break; + + case (MDOC_Bd): + assert(mdoc->last); + for (node = mdoc->last->parent; node; node = node->parent) { + if (node->type != MDOC_BLOCK) + continue; + if (node->data.block.tok != MDOC_Bd) + continue; + return(mdoc_err(mdoc, tok, pos, ERR_SCOPE_NONEST)); + } + break; case (MDOC_Bl): break; @@ -251,6 +188,13 @@ append_text(struct mdoc *mdoc, int tok, switch (tok) { /* ======= ADD MORE MACRO CHECKS BELOW. ======= */ + case (MDOC_Pp): + if (0 == sz) + break; + if ( ! mdoc_warn(mdoc, tok, pos, WARN_ARGS_EQ0)) + return(0); + break; + case (MDOC_Ft): /* FALLTHROUGH */ case (MDOC_Li): @@ -265,12 +209,15 @@ append_text(struct mdoc *mdoc, int tok, if ( ! mdoc_warn(mdoc, tok, pos, WARN_ARGS_GE1)) return(0); break; + case (MDOC_Ar): /* FALLTHROUGH */ case (MDOC_Cm): /* FALLTHROUGH */ case (MDOC_Fl): + /* These can have no arguments. */ break; + case (MDOC_Ad): /* FALLTHROUGH */ case (MDOC_Em): @@ -297,8 +244,7 @@ append_text(struct mdoc *mdoc, int tok, /* NOTREACHED */ } - mdoc_elem_alloc(mdoc, pos, tok, 0, - NULL, (size_t)sz, _CC(args)); + mdoc_elem_alloc(mdoc, pos, tok, 0, NULL, (size_t)sz, _CC(args)); return(1); } @@ -306,59 +252,109 @@ append_text(struct mdoc *mdoc, int tok, int macro_text(MACRO_PROT_ARGS) { - int lastarg, c, lasttok, lastpunct, j; + int lastarg, lastpunct, c, j; char *args[MDOC_LINEARG_MAX], *p; - lasttok = ppos; - lastpunct = 0; - j = 0; - if (SEC_PROLOGUE == mdoc->sec_lastn) return(mdoc_err(mdoc, tok, ppos, ERR_SEC_PROLOGUE)); -again: - lastarg = *pos; + /* Token pre-processing. */ + + switch (tok) { + case (MDOC_Pp): + /* `.Pp' ignored when following `.Sh' or `.Ss'. */ + assert(mdoc->last); + if (MDOC_BODY != mdoc->last->type) + break; + switch (mdoc->last->data.body.tok) { + case (MDOC_Ss): + /* FALLTHROUGH */ + case (MDOC_Sh): + if ( ! mdoc_warn(mdoc, tok, ppos, WARN_IGN_AFTER_BLK)) + return(0); + return(1); + default: + break; + } + break; + default: + break; + } + + /* Process line parameters. */ + + j = 0; + lastarg = ppos; + lastpunct = 0; +again: if (j == MDOC_LINEARG_MAX) return(mdoc_err(mdoc, tok, lastarg, ERR_ARGS_MANY)); - c = args_next(mdoc, tok, pos, buf, &args[j]); - - if (-1 == c) + + /* + * Parse out the next argument, unquoted and unescaped. If + * we're a word (which may be punctuation followed eventually by + * a real word), then fall into checking for callables. If + * only punctuation remains and we're the first, then flush + * arguments, punctuation and exit; else, return to the caller. + */ + + lastarg = *pos; + + switch (mdoc_args(mdoc, tok, pos, buf, ARGS_DELIM, &args[j])) { + case (ARGS_ERROR): return(0); - if (0 == c && ! lastpunct) - return(append_text(mdoc, tok, lasttok, j, args)); - else if (0 == c) - return(1); + case (ARGS_WORD): + break; + case (ARGS_PUNCT): + if ( ! lastpunct && ! append_text(mdoc, tok, ppos, j, args)) + return(0); + return(append_delims(mdoc, tok, pos, buf)); + case (ARGS_EOLN): + return(append_text(mdoc, tok, ppos, j, args)); + default: + abort(); + /* NOTREACHED */ + } - /* Command found. */ + /* + * Command found. First flush out arguments, then call the + * command. If we're the line macro when it exits, flush + * terminal punctuation. + */ if (MDOC_MAX != (c = mdoc_find(mdoc, args[j]))) { - if ( ! lastpunct) - if ( ! append_text(mdoc, tok, lasttok, j, args)) - return(0); - return(mdoc_macro(mdoc, c, lastarg, pos, buf)); + if ( ! lastpunct && ! append_text(mdoc, tok, ppos, j, args)) + return(0); + if ( ! mdoc_macro(mdoc, c, lastarg, pos, buf)) + return(0); + if (ppos > 1) + return(1); + return(append_delims(mdoc, tok, pos, buf)); } - /* Word found. */ + /* Word/non-term-punctuation found. */ if ( ! mdoc_isdelim(args[j])) { + /* Words are appended to the array of arguments. */ j++; + lastpunct = 1; goto again; } - /* Punctuation found. */ - - p = args[j]; /* Save argument (NULL-ified in append). */ - - if ( ! lastpunct) - if ( ! append_text(mdoc, tok, lasttok, j, args)) - return(0); + /* + * For punctuation, flush all collected words, then flush + * punctuation, then start collecting again. Of course, this + * is non-terminal punctuation. + */ - args[j] = p; + p = args[j]; + if ( ! lastpunct && ! append_text(mdoc, tok, ppos, j, args)) + return(0); - mdoc_word_alloc(mdoc, lastarg, args[j]); - lastpunct = 1; + mdoc_word_alloc(mdoc, lastarg, p); j = 0; + lastpunct = 1; goto again; @@ -369,7 +365,7 @@ again: int macro_prologue_dtitle(MACRO_PROT_ARGS) { - int c, lastarg, j; + int lastarg, j; char *args[MDOC_LINEARG_MAX]; if (SEC_PROLOGUE != mdoc->sec_lastn) @@ -380,15 +376,16 @@ macro_prologue_dtitle(MACRO_PROT_ARGS) return(mdoc_err(mdoc, tok, ppos, ERR_SEC_PROLOGUE_REP)); j = -1; + lastarg = ppos; again: - lastarg = *pos; - if (j == MDOC_LINEARG_MAX) return(mdoc_err(mdoc, tok, lastarg, ERR_ARGS_MANY)); - c = args_next(mdoc, tok, pos, buf, &args[++j]); - if (0 == c) { + lastarg = *pos; + + switch (mdoc_args(mdoc, tok, pos, buf, 0, &args[++j])) { + case (ARGS_EOLN): if (mdoc->meta.title) return(1); if ( ! mdoc_warn(mdoc, tok, ppos, WARN_ARGS_GE1)) @@ -396,9 +393,12 @@ again: (void)xstrlcpy(mdoc->meta.title, "UNTITLED", META_TITLE_SZ); return(1); - } else if (-1 == c) + case (ARGS_ERROR): return(0); - + default: + break; + } + if (MDOC_MAX != mdoc_find(mdoc, args[j]) && ! mdoc_warn (mdoc, tok, lastarg, WARN_SYNTAX_MACLIKE)) return(0); @@ -406,13 +406,13 @@ again: if (0 == j) { if (xstrlcpy(mdoc->meta.title, args[0], META_TITLE_SZ)) goto again; - return(mdoc_err(mdoc, tok, lastarg, ERR_SYNTAX_ARGS)); + return(mdoc_err(mdoc, tok, lastarg, ERR_SYNTAX_ARGFORM)); } else if (1 == j) { mdoc->meta.msec = mdoc_atomsec(args[1]); if (MSEC_DEFAULT != mdoc->meta.msec) goto again; - return(mdoc_err(mdoc, tok, -1, ERR_SYNTAX_ARGS)); + return(mdoc_err(mdoc, tok, -1, ERR_SYNTAX_ARGFORM)); } else if (2 == j) { mdoc->meta.vol = mdoc_atovol(args[2]); @@ -421,7 +421,7 @@ again: mdoc->meta.arch = mdoc_atoarch(args[2]); if (ARCH_DEFAULT != mdoc->meta.arch) goto again; - return(mdoc_err(mdoc, tok, lastarg, ERR_SYNTAX_ARGS)); + return(mdoc_err(mdoc, tok, lastarg, ERR_SYNTAX_ARGFORM)); } return(mdoc_err(mdoc, tok, lastarg, ERR_ARGS_MANY)); @@ -431,7 +431,7 @@ again: int macro_prologue_os(MACRO_PROT_ARGS) { - int c, lastarg, j; + int lastarg, j; char *args[MDOC_LINEARG_MAX]; if (SEC_PROLOGUE != mdoc->sec_lastn) @@ -442,24 +442,29 @@ macro_prologue_os(MACRO_PROT_ARGS) return(mdoc_err(mdoc, tok, ppos, ERR_SEC_PROLOGUE_REP)); j = -1; + lastarg = ppos; again: - lastarg = *pos; - if (j == MDOC_LINEARG_MAX) return(mdoc_err(mdoc, tok, lastarg, ERR_ARGS_MANY)); - c = args_next_quoted(mdoc, tok, pos, buf, &args[++j]); - if (0 == c) { + lastarg = *pos; + + switch (mdoc_args(mdoc, tok, pos, buf, + ARGS_QUOTED, &args[++j])) { + case (ARGS_EOLN): mdoc->sec_lastn = mdoc->sec_last = SEC_BODY; return(1); - } else if (-1 == c) + case (ARGS_ERROR): return(0); + default: + break; + } if ( ! xstrlcat(mdoc->meta.os, args[j], sizeof(mdoc->meta.os))) - return(mdoc_err(mdoc, tok, lastarg, ERR_SYNTAX_ARGS)); + return(mdoc_err(mdoc, tok, lastarg, ERR_SYNTAX_ARGFORM)); if ( ! xstrlcat(mdoc->meta.os, " ", sizeof(mdoc->meta.os))) - return(mdoc_err(mdoc, tok, lastarg, ERR_SYNTAX_ARGS)); + return(mdoc_err(mdoc, tok, lastarg, ERR_SYNTAX_ARGFORM)); goto again; /* NOTREACHED */ @@ -469,7 +474,7 @@ again: int macro_prologue_ddate(MACRO_PROT_ARGS) { - int c, lastarg, j; + int lastarg, j; char *args[MDOC_LINEARG_MAX], date[64]; if (SEC_PROLOGUE != mdoc->sec_lastn) @@ -481,23 +486,26 @@ macro_prologue_ddate(MACRO_PROT_ARGS) j = -1; date[0] = 0; + lastarg = ppos; again: - lastarg = *pos; - if (j == MDOC_LINEARG_MAX) return(mdoc_err(mdoc, tok, lastarg, ERR_ARGS_MANY)); - c = args_next(mdoc, tok, pos, buf, &args[++j]); - if (0 == c) { + lastarg = *pos; + switch (mdoc_args(mdoc, tok, pos, buf, 0, &args[++j])) { + case (ARGS_EOLN): if (mdoc->meta.date) return(1); mdoc->meta.date = mdoc_atotime(date); if (mdoc->meta.date) return(1); - return(mdoc_err(mdoc, tok, ppos, ERR_SYNTAX_ARGS)); - } else if (-1 == c) + return(mdoc_err(mdoc, tok, ppos, ERR_SYNTAX_ARGFORM)); + case (ARGS_ERROR): return(0); + default: + break; + } if (MDOC_MAX != mdoc_find(mdoc, args[j]) && ! mdoc_warn (mdoc, tok, lastarg, WARN_SYNTAX_MACLIKE)) @@ -514,9 +522,9 @@ again: goto again; if ( ! xstrlcat(date, args[j], sizeof(date))) - return(mdoc_err(mdoc, tok, lastarg, ERR_SYNTAX_ARGS)); + return(mdoc_err(mdoc, tok, lastarg, ERR_SYNTAX_ARGFORM)); if ( ! xstrlcat(date, " ", sizeof(date))) - return(mdoc_err(mdoc, tok, lastarg, ERR_SYNTAX_ARGS)); + return(mdoc_err(mdoc, tok, lastarg, ERR_SYNTAX_ARGFORM)); goto again; /* NOTREACHED */ @@ -528,6 +536,7 @@ macro_scoped_explicit(MACRO_PROT_ARGS) { int c, lastarg, j; struct mdoc_arg argv[MDOC_LINEARG_MAX]; + struct mdoc_node *n; if (SEC_PROLOGUE == mdoc->sec_lastn) return(mdoc_err(mdoc, tok, ppos, ERR_SEC_PROLOGUE)); @@ -541,17 +550,43 @@ macro_scoped_explicit(MACRO_PROT_ARGS) switch (tok) { case (MDOC_El): return(scope_rewind_exp(mdoc, ppos, tok, MDOC_Bl)); + case (MDOC_Ed): + return(scope_rewind_exp(mdoc, ppos, tok, MDOC_Bd)); default: break; } assert(MDOC_EXPLICIT & mdoc_macros[tok].flags); + /* Token pre-processing. */ + + switch (tok) { + case (MDOC_Bl): + /* FALLTHROUGH */ + case (MDOC_Bd): + /* `.Pp' ignored when preceding `.Bl' or `.Bd'. */ + assert(mdoc->last); + if (MDOC_ELEM != mdoc->last->type) + break; + if (MDOC_Pp != mdoc->last->data.elem.tok) + break; + if ( ! mdoc_warn(mdoc, tok, ppos, WARN_IGN_BEFORE_BLK)) + return(0); + assert(mdoc->last->prev); + n = mdoc->last; + mdoc->last = mdoc->last->prev; + mdoc->last->next = NULL; + mdoc_node_free(n); + break; + default: + break; + } + lastarg = *pos; for (j = 0; j < MDOC_LINEARG_MAX; j++) { lastarg = *pos; - c = argv_next(mdoc, tok, pos, buf, &argv[j]); + c = mdoc_argv(mdoc, tok, &argv[j], pos, buf); if (0 == c) break; else if (1 == c) @@ -575,7 +610,7 @@ macro_scoped_explicit(MACRO_PROT_ARGS) int macro_scoped_implicit(MACRO_PROT_ARGS) { - int t, c, lastarg, j; + int lastarg, j; char *args[MDOC_LINEARG_MAX]; struct mdoc_node *n; @@ -584,45 +619,63 @@ macro_scoped_implicit(MACRO_PROT_ARGS) if (SEC_PROLOGUE == mdoc->sec_lastn) return(mdoc_err(mdoc, tok, ppos, ERR_SEC_PROLOGUE)); - /* FIXME: put into scope_rewind_imp(). */ + /* Token pre-processing. */ - /* LINTED */ - for (n = mdoc->last; n; n = n->parent) { - if (MDOC_BLOCK != n->type) - continue; - if (tok == (t = n->data.block.tok)) + switch (tok) { + case (MDOC_Ss): + /* FALLTHROUGH */ + case (MDOC_Sh): + /* `.Pp' ignored when preceding `.Ss' or `.Sh'. */ + if (NULL == mdoc->last) break; - if ( ! (MDOC_EXPLICIT & mdoc_macros[t].flags)) - continue; - return(mdoc_err(mdoc, tok, ppos, ERR_SCOPE_BREAK)); + if (MDOC_ELEM != mdoc->last->type) + break; + if (MDOC_Pp != mdoc->last->data.elem.tok) + break; + if ( ! mdoc_warn(mdoc, tok, ppos, WARN_IGN_BEFORE_BLK)) + return(0); + assert(mdoc->last->prev); + n = mdoc->last; + mdoc->last = mdoc->last->prev; + mdoc->last->next = NULL; + mdoc_node_free(n); + break; + default: + break; } - if (n) { - mdoc->last = n; - mdoc_msg(mdoc, ppos, "scope: rewound `%s'", - mdoc_macronames[tok]); - } else - mdoc_msg(mdoc, ppos, "scope: new `%s'", - mdoc_macronames[tok]); + switch (tok) { + case (MDOC_Sh): + /* FALLTHROUGH */ + case (MDOC_Ss): + if ( ! scope_rewind_imp(mdoc, ppos, tok)) + return(0); + break; + default: + break; + } j = 0; + lastarg = ppos; again: - lastarg = *pos; - if (j == MDOC_LINEARG_MAX) return(mdoc_err(mdoc, tok, lastarg, ERR_ARGS_MANY)); - c = args_next(mdoc, tok, pos, buf, &args[j]); - - if (-1 == c) + + lastarg = *pos; + + switch (mdoc_args(mdoc, tok, pos, buf, 0, &args[j])) { + case (ARGS_ERROR): return(0); - if (0 == c) - return(append_scoped(mdoc, tok, ppos, - j, _CC(args), 0, NULL)); + case (ARGS_EOLN): + return(append_scoped(mdoc, tok, ppos, j, _CC(args), 0, NULL)); + default: + break; + } /* Command found. */ - if (MDOC_MAX != (c = mdoc_find(mdoc, args[j]))) + if (MDOC_MAX != mdoc_find(mdoc, args[j])) if ( ! mdoc_warn(mdoc, tok, lastarg, WARN_SYNTAX_MACLIKE)) return(0); @@ -634,3 +687,10 @@ again: /* NOTREACHED */ } + +int +macro_scoped_line(MACRO_PROT_ARGS) +{ + + return(1); +} @@ -89,11 +89,11 @@ const struct mdoc_macro __mdoc_macros[MDOC_MAX] = { { macro_prologue_os, 0 }, /* Os */ { macro_scoped_implicit, 0 }, /* Sh */ { macro_scoped_implicit, 0 }, /* Ss */ - { NULL, 0 }, /* Pp */ + { macro_text, 0 }, /* Pp */ { NULL, 0 }, /* D1 */ { NULL, 0 }, /* Dl */ - { NULL, 0 }, /* Bd */ - { NULL, 0 }, /* Ed */ + { macro_scoped_explicit, MDOC_EXPLICIT }, /* Bd */ + { macro_scoped_explicit, 0 }, /* Ed */ { macro_scoped_explicit, MDOC_EXPLICIT }, /* Bl */ { macro_scoped_explicit, 0 }, /* El */ { NULL, 0 }, /* It */ @@ -252,9 +252,9 @@ mdoc_parseln(struct mdoc *mdoc, char *buf) int c, i; char tmp[5]; - if ('.' != *buf) { - /* TODO. */ - return(1); + if ('.' != *buf) { + mdoc_word_alloc(mdoc, 0, buf); + return(1); } if (buf[1] && '\\' == buf[1]) @@ -347,7 +347,7 @@ mdoc_node_append(struct mdoc *mdoc, int pos, struct mdoc_node *p) switch (p->type) { case (MDOC_TEXT): - nn = "<text>"; + nn = p->data.text.string; nt = "text"; break; case (MDOC_BODY): @@ -416,6 +416,7 @@ mdoc_node_append(struct mdoc *mdoc, int pos, struct mdoc_node *p) case (MDOC_HEAD): p->parent = mdoc->last->parent; mdoc->last->next = p; + p->prev = mdoc->last; act = "sibling"; break; default: @@ -440,6 +441,7 @@ mdoc_node_append(struct mdoc *mdoc, int pos, struct mdoc_node *p) break; default: p->parent = mdoc->last->parent; + p->prev = mdoc->last; mdoc->last->next = p; act = "sibling"; break; @@ -192,17 +192,19 @@ #define MDOC_ARG_MAX 60 enum mdoc_err { - ERR_SYNTAX_QUOTE, + ERR_SYNTAX_QUOTE, /* NOTUSED */ ERR_SYNTAX_UNQUOTE, ERR_SYNTAX_WS, - ERR_SYNTAX_ARGS, - ERR_SYNTAX_BADARG, + ERR_SYNTAX_ARG, + ERR_SYNTAX_ARGFORM, ERR_SYNTAX_ARGVAL, + ERR_SYNTAX_ARGBAD, ERR_SYNTAX_ARGMANY, ERR_MACRO_NOTSUP, ERR_MACRO_NOTCALL, ERR_SCOPE_BREAK, ERR_SCOPE_NOCTX, + ERR_SCOPE_NONEST, ERR_SEC_PROLOGUE, ERR_SEC_NPROLOGUE, ERR_SEC_PROLOGUE_OO, @@ -215,9 +217,13 @@ enum mdoc_err { enum mdoc_warn { WARN_SYNTAX_WS_EOLN, WARN_SYNTAX_MACLIKE, + WARN_SYNTAX_ARGLIKE, + WARN_SYNTAX_QUOTED, + WARN_IGN_AFTER_BLK, + WARN_IGN_BEFORE_BLK, WARN_SEC_OO, WARN_ARGS_GE1, - WARN_SYNTAX_ARGLIKE + WARN_ARGS_EQ0, }; struct mdoc_arg { @@ -370,6 +376,7 @@ struct mdoc_node { struct mdoc_node *parent; struct mdoc_node *child; struct mdoc_node *next; + struct mdoc_node *prev; enum mdoc_type type; union mdoc_data data; }; @@ -192,7 +192,7 @@ print_node(const struct mdoc_node *n, int indent) switch (n->type) { case (MDOC_TEXT): assert(NULL == n->child); - p = "<text>"; + p = n->data.text.string; t = "text"; break; case (MDOC_BODY): @@ -326,10 +326,10 @@ msg_err(void *arg, int tok, int col, enum mdoc_err type) case (ERR_SYNTAX_WS): lit = "syntax: whitespace in argument"; break; - case (ERR_SYNTAX_ARGS): + case (ERR_SYNTAX_ARGFORM): fmt = "syntax: macro `%s' arguments malformed"; break; - case (ERR_SYNTAX_BADARG): + case (ERR_SYNTAX_ARG): fmt = "syntax: unknown argument for macro `%s'"; break; case (ERR_SCOPE_BREAK): @@ -339,6 +339,9 @@ msg_err(void *arg, int tok, int col, enum mdoc_err type) case (ERR_SCOPE_NOCTX): fmt = "scope: closure macro `%s' has no context"; break; + case (ERR_SCOPE_NONEST): + fmt = "scope: macro `%s' may not be nested"; + break; case (ERR_MACRO_NOTSUP): fmt = "macro `%s' not supported"; break; @@ -369,6 +372,9 @@ msg_err(void *arg, int tok, int col, enum mdoc_err type) case (ERR_SYNTAX_ARGVAL): lit = "syntax: expected value for macro argument"; break; + case (ERR_SYNTAX_ARGBAD): + lit = "syntax: invalid value for macro argument"; + break; case (ERR_SYNTAX_ARGMANY): lit = "syntax: too many values for macro argument"; break; @@ -451,6 +457,9 @@ msg_warn(void *arg, int tok, int col, enum mdoc_warn type) case (WARN_SYNTAX_WS_EOLN): lit = "syntax: whitespace at end-of-line"; break; + case (WARN_SYNTAX_QUOTED): + lit = "syntax: quotation mark starting string"; + break; case (WARN_SYNTAX_MACLIKE): lit = "syntax: macro-like argument"; break; @@ -463,6 +472,15 @@ msg_warn(void *arg, int tok, int col, enum mdoc_warn type) case (WARN_ARGS_GE1): fmt = "macro `%s' suggests one or more arguments"; break; + case (WARN_ARGS_EQ0): + fmt = "macro `%s' suggests zero arguments"; + break; + case (WARN_IGN_AFTER_BLK): + fmt = "ignore: macro `%s' ignored after block macro"; + break; + case (WARN_IGN_BEFORE_BLK): + fmt = "ignore: macro before block macro `%s' ignored"; + break; default: abort(); /* NOTREACHED */ @@ -67,16 +67,25 @@ void *mdoc_tokhash_alloc(void); int mdoc_tokhash_find(const void *, const char *); void mdoc_tokhash_free(void *); int mdoc_isdelim(const char *); +int mdoc_iscdelim(char); enum mdoc_sec mdoc_atosec(size_t, const char **); enum mdoc_msec mdoc_atomsec(const char *); enum mdoc_vol mdoc_atovol(const char *); enum mdoc_arch mdoc_atoarch(const char *); time_t mdoc_atotime(const char *); -int mdoc_argv_lookup(int, const char *); -int mdoc_argv_parse(struct mdoc *, int, - int, struct mdoc_arg *, int *, char *); +int mdoc_argv(struct mdoc *, int, + struct mdoc_arg *, int *, char *); void mdoc_argv_free(int, struct mdoc_arg *); +int mdoc_args(struct mdoc *, int, + int *, char *, int, char **); +#define ARGS_ERROR (-1) +#define ARGS_EOLN (0) +#define ARGS_WORD (1) +#define ARGS_PUNCT (2) + +#define ARGS_QUOTED (1 << 0) +#define ARGS_DELIM (1 << 1) int xstrlcat(char *, const char *, size_t); int xstrlcpy(char *, const char *, size_t); @@ -25,15 +25,10 @@ #include "private.h" int -mdoc_isdelim(const char *p) +mdoc_iscdelim(char p) { - if (0 == *p) - return(0); - if (0 != *(p + 1)) - return(0); - - switch (*p) { + switch (p) { case('{'): /* FALLTHROUGH */ case('.'): @@ -66,6 +61,18 @@ mdoc_isdelim(const char *p) } +int +mdoc_isdelim(const char *p) +{ + + if (0 == *p) + return(0); + if (0 != *(p + 1)) + return(0); + return(mdoc_iscdelim(*p)); +} + + enum mdoc_sec mdoc_atosec(size_t sz, const char **p) { |