version 1.63, 2011/03/16 17:55:39 |
version 1.74, 2011/04/07 01:08:42 |
|
|
#include <stdio.h> |
#include <stdio.h> |
#include <string.h> |
#include <string.h> |
|
|
|
#include "mdoc.h" |
#include "mandoc.h" |
#include "mandoc.h" |
#include "libmdoc.h" |
#include "libmdoc.h" |
#include "libmandoc.h" |
#include "libmandoc.h" |
|
|
/* |
#define MULTI_STEP 5 /* pre-allocate argument values */ |
* Routines to parse arguments of macros. Arguments follow the syntax |
|
* of `-arg [val [valN...]]'. Arguments come in all types: quoted |
|
* arguments, multiple arguments per value, no-value arguments, etc. |
|
* |
|
* There's no limit to the number or arguments that may be allocated. |
|
*/ |
|
|
|
#define ARGV_NONE (1 << 0) |
|
#define ARGV_SINGLE (1 << 1) |
|
#define ARGV_MULTI (1 << 2) |
|
#define ARGV_OPT_SINGLE (1 << 3) |
|
|
|
#define MULTI_STEP 5 |
|
|
|
static enum mdocargt argv_a2arg(enum mdoct, const char *); |
static enum mdocargt argv_a2arg(enum mdoct, const char *); |
static enum margserr args(struct mdoc *, int, int *, |
static enum margserr args(struct mdoc *, int, int *, |
char *, int, char **); |
char *, int, char **); |
|
static int args_checkpunct(struct mdoc *, |
|
const char *, int, int, int); |
static int argv(struct mdoc *, int, |
static int argv(struct mdoc *, int, |
struct mdoc_argv *, int *, char *); |
struct mdoc_argv *, int *, char *); |
static int argv_single(struct mdoc *, int, |
static int argv_single(struct mdoc *, int, |
Line 56 static int argv_opt_single(struct mdoc *, int, |
|
Line 46 static int argv_opt_single(struct mdoc *, int, |
|
struct mdoc_argv *, int *, char *); |
struct mdoc_argv *, int *, char *); |
static int argv_multi(struct mdoc *, int, |
static int argv_multi(struct mdoc *, int, |
struct mdoc_argv *, int *, char *); |
struct mdoc_argv *, int *, char *); |
|
static void argn_free(struct mdoc_arg *, int); |
|
|
/* Per-argument flags. */ |
enum argvflag { |
|
ARGV_NONE, /* no args to flag (e.g., -split) */ |
|
ARGV_SINGLE, /* one arg to flag (e.g., -file xxx) */ |
|
ARGV_MULTI, /* multiple args (e.g., -column xxx yyy) */ |
|
ARGV_OPT_SINGLE /* optional arg (e.g., -offset [xxx]) */ |
|
}; |
|
|
static int mdoc_argvflags[MDOC_ARG_MAX] = { |
static const enum argvflag argvflags[MDOC_ARG_MAX] = { |
ARGV_NONE, /* MDOC_Split */ |
ARGV_NONE, /* MDOC_Split */ |
ARGV_NONE, /* MDOC_Nosplit */ |
ARGV_NONE, /* MDOC_Nosplit */ |
ARGV_NONE, /* MDOC_Ragged */ |
ARGV_NONE, /* MDOC_Ragged */ |
Line 88 static int mdoc_argvflags[MDOC_ARG_MAX] = { |
|
Line 84 static int mdoc_argvflags[MDOC_ARG_MAX] = { |
|
ARGV_NONE /* MDOC_Symbolic */ |
ARGV_NONE /* MDOC_Symbolic */ |
}; |
}; |
|
|
static int mdoc_argflags[MDOC_MAX] = { |
static const int argflags[MDOC_MAX] = { |
0, /* Ap */ |
0, /* Ap */ |
0, /* Dd */ |
0, /* Dd */ |
0, /* Dt */ |
0, /* Dt */ |
Line 213 static int mdoc_argflags[MDOC_MAX] = { |
|
Line 209 static int mdoc_argflags[MDOC_MAX] = { |
|
0, /* Ta */ |
0, /* Ta */ |
}; |
}; |
|
|
|
static const enum mdocargt args_Ex[] = { |
|
MDOC_Std, |
|
MDOC_ARG_MAX |
|
}; |
|
|
|
static const enum mdocargt args_An[] = { |
|
MDOC_Split, |
|
MDOC_Nosplit, |
|
MDOC_ARG_MAX |
|
}; |
|
|
|
static const enum mdocargt args_Bd[] = { |
|
MDOC_Ragged, |
|
MDOC_Unfilled, |
|
MDOC_Filled, |
|
MDOC_Literal, |
|
MDOC_File, |
|
MDOC_Offset, |
|
MDOC_Compact, |
|
MDOC_Centred, |
|
MDOC_ARG_MAX |
|
}; |
|
|
|
static const enum mdocargt args_Bf[] = { |
|
MDOC_Emphasis, |
|
MDOC_Literal, |
|
MDOC_Symbolic, |
|
MDOC_ARG_MAX |
|
}; |
|
|
|
static const enum mdocargt args_Bk[] = { |
|
MDOC_Words, |
|
MDOC_ARG_MAX |
|
}; |
|
|
|
static const enum mdocargt args_Bl[] = { |
|
MDOC_Bullet, |
|
MDOC_Dash, |
|
MDOC_Hyphen, |
|
MDOC_Item, |
|
MDOC_Enum, |
|
MDOC_Tag, |
|
MDOC_Diag, |
|
MDOC_Hang, |
|
MDOC_Ohang, |
|
MDOC_Inset, |
|
MDOC_Column, |
|
MDOC_Width, |
|
MDOC_Offset, |
|
MDOC_Compact, |
|
MDOC_Nested, |
|
MDOC_ARG_MAX |
|
}; |
|
|
/* |
/* |
* Parse an argument from line text. This comes in the form of -key |
* Parse an argument from line text. This comes in the form of -key |
* [value0...], which may either have a single mandatory value, at least |
* [value0...], which may either have a single mandatory value, at least |
Line 254 mdoc_argv(struct mdoc *m, int line, enum mdoct tok, |
|
Line 303 mdoc_argv(struct mdoc *m, int line, enum mdoct tok, |
|
buf[(*pos)++] = '\0'; |
buf[(*pos)++] = '\0'; |
} |
} |
|
|
(void)memset(&tmp, 0, sizeof(struct mdoc_argv)); |
memset(&tmp, 0, sizeof(struct mdoc_argv)); |
tmp.line = line; |
tmp.line = line; |
tmp.pos = *pos; |
tmp.pos = *pos; |
|
|
Line 280 mdoc_argv(struct mdoc *m, int line, enum mdoct tok, |
|
Line 329 mdoc_argv(struct mdoc *m, int line, enum mdoct tok, |
|
arg->argv = mandoc_realloc |
arg->argv = mandoc_realloc |
(arg->argv, arg->argc * sizeof(struct mdoc_argv)); |
(arg->argv, arg->argc * sizeof(struct mdoc_argv)); |
|
|
(void)memcpy(&arg->argv[(int)arg->argc - 1], |
memcpy(&arg->argv[(int)arg->argc - 1], |
&tmp, sizeof(struct mdoc_argv)); |
&tmp, sizeof(struct mdoc_argv)); |
|
|
return(ARGV_ARG); |
return(ARGV_ARG); |
} |
} |
|
|
|
|
void |
void |
mdoc_argv_free(struct mdoc_arg *p) |
mdoc_argv_free(struct mdoc_arg *p) |
{ |
{ |
Line 303 mdoc_argv_free(struct mdoc_arg *p) |
|
Line 351 mdoc_argv_free(struct mdoc_arg *p) |
|
assert(p->argc); |
assert(p->argc); |
|
|
for (i = (int)p->argc - 1; i >= 0; i--) |
for (i = (int)p->argc - 1; i >= 0; i--) |
mdoc_argn_free(p, i); |
argn_free(p, i); |
|
|
free(p->argv); |
free(p->argv); |
free(p); |
free(p); |
} |
} |
|
|
|
static void |
void |
argn_free(struct mdoc_arg *p, int iarg) |
mdoc_argn_free(struct mdoc_arg *p, int iarg) |
|
{ |
{ |
struct mdoc_argv *arg; |
struct mdoc_argv *arg; |
int j; |
int j; |
Line 328 mdoc_argn_free(struct mdoc_arg *p, int iarg) |
|
Line 375 mdoc_argn_free(struct mdoc_arg *p, int iarg) |
|
p->argv[iarg] = p->argv[iarg+1]; |
p->argv[iarg] = p->argv[iarg+1]; |
} |
} |
|
|
|
|
enum margserr |
enum margserr |
mdoc_zargs(struct mdoc *m, int line, int *pos, |
mdoc_zargs(struct mdoc *m, int line, int *pos, |
char *buf, int flags, char **v) |
char *buf, int flags, char **v) |
Line 337 mdoc_zargs(struct mdoc *m, int line, int *pos, |
|
Line 383 mdoc_zargs(struct mdoc *m, int line, int *pos, |
|
return(args(m, line, pos, buf, flags, v)); |
return(args(m, line, pos, buf, flags, v)); |
} |
} |
|
|
|
|
enum margserr |
enum margserr |
mdoc_args(struct mdoc *m, int line, int *pos, |
mdoc_args(struct mdoc *m, int line, int *pos, |
char *buf, enum mdoct tok, char **v) |
char *buf, enum mdoct tok, char **v) |
Line 345 mdoc_args(struct mdoc *m, int line, int *pos, |
|
Line 390 mdoc_args(struct mdoc *m, int line, int *pos, |
|
int fl; |
int fl; |
struct mdoc_node *n; |
struct mdoc_node *n; |
|
|
fl = mdoc_argflags[tok]; |
fl = argflags[tok]; |
|
|
if (MDOC_It != tok) |
if (MDOC_It != tok) |
return(args(m, line, pos, buf, fl, v)); |
return(args(m, line, pos, buf, fl, v)); |
Line 369 mdoc_args(struct mdoc *m, int line, int *pos, |
|
Line 414 mdoc_args(struct mdoc *m, int line, int *pos, |
|
return(args(m, line, pos, buf, fl, v)); |
return(args(m, line, pos, buf, fl, v)); |
} |
} |
|
|
|
|
static enum margserr |
static enum margserr |
args(struct mdoc *m, int line, int *pos, |
args(struct mdoc *m, int line, int *pos, |
char *buf, int fl, char **v) |
char *buf, int fl, char **v) |
{ |
{ |
int i; |
|
char *p, *pp; |
char *p, *pp; |
enum margserr rc; |
enum margserr rc; |
enum mdelim d; |
|
|
|
/* |
/* |
* Parse out the terms (like `val' in `.Xx -arg val' or simply |
* Parse out the terms (like `val' in `.Xx -arg val' or simply |
Line 405 args(struct mdoc *m, int line, int *pos, |
|
Line 447 args(struct mdoc *m, int line, int *pos, |
|
* is unterminated. |
* is unterminated. |
*/ |
*/ |
if (MDOC_PHRASELIT & m->flags) |
if (MDOC_PHRASELIT & m->flags) |
if ( ! mdoc_pmsg(m, line, *pos, MANDOCERR_BADQUOTE)) |
mdoc_pmsg(m, line, *pos, MANDOCERR_BADQUOTE); |
return(ARGS_ERROR); |
|
|
|
m->flags &= ~MDOC_PHRASELIT; |
m->flags &= ~MDOC_PHRASELIT; |
return(ARGS_EOLN); |
return(ARGS_EOLN); |
} |
} |
|
|
/* |
|
* If the first character is a closing delimiter and we're to |
|
* look for delimited strings, then pass down the buffer seeing |
|
* if it follows the pattern of [[::delim::][ ]+]+. Note that |
|
* we ONLY care about closing delimiters. |
|
*/ |
|
|
|
if ((fl & ARGS_DELIM) && DELIM_CLOSE == mdoc_iscdelim(buf[*pos])) { |
|
for (i = *pos; buf[i]; ) { |
|
d = mdoc_iscdelim(buf[i]); |
|
if (DELIM_NONE == d || DELIM_OPEN == d) |
|
break; |
|
i++; |
|
if ('\0' == buf[i] || ' ' != buf[i]) |
|
break; |
|
i++; |
|
while (buf[i] && ' ' == buf[i]) |
|
i++; |
|
} |
|
|
|
if ('\0' == buf[i]) { |
|
*v = &buf[*pos]; |
|
if (i && ' ' != buf[i - 1]) |
|
return(ARGS_PUNCT); |
|
if (ARGS_NOWARN & fl) |
|
return(ARGS_PUNCT); |
|
if ( ! mdoc_pmsg(m, line, *pos, MANDOCERR_EOLNSPACE)) |
|
return(ARGS_ERROR); |
|
return(ARGS_PUNCT); |
|
} |
|
} |
|
|
|
*v = &buf[*pos]; |
*v = &buf[*pos]; |
|
|
|
if (ARGS_DELIM & fl && args_checkpunct(m, buf, *pos, line, fl)) |
|
return(ARGS_PUNCT); |
|
|
/* |
/* |
* First handle TABSEP items, restricted to `Bl -column'. This |
* First handle TABSEP items, restricted to `Bl -column'. This |
* ignores conventional token parsing and instead uses tabs or |
* ignores conventional token parsing and instead uses tabs or |
Line 493 args(struct mdoc *m, int line, int *pos, |
|
Line 505 args(struct mdoc *m, int line, int *pos, |
|
|
|
/* Whitespace check for eoln case... */ |
/* Whitespace check for eoln case... */ |
if ('\0' == *p && ' ' == *(p - 1) && ! (ARGS_NOWARN & fl)) |
if ('\0' == *p && ' ' == *(p - 1) && ! (ARGS_NOWARN & fl)) |
if ( ! mdoc_pmsg(m, line, *pos, MANDOCERR_EOLNSPACE)) |
mdoc_pmsg(m, line, *pos, MANDOCERR_EOLNSPACE); |
return(ARGS_ERROR); |
|
|
|
*pos += (int)(p - *v); |
*pos += (int)(p - *v); |
|
|
Line 538 args(struct mdoc *m, int line, int *pos, |
|
Line 549 args(struct mdoc *m, int line, int *pos, |
|
if ('\0' == buf[*pos]) { |
if ('\0' == buf[*pos]) { |
if (ARGS_NOWARN & fl || MDOC_PPHRASE & m->flags) |
if (ARGS_NOWARN & fl || MDOC_PPHRASE & m->flags) |
return(ARGS_QWORD); |
return(ARGS_QWORD); |
if ( ! mdoc_pmsg(m, line, *pos, MANDOCERR_BADQUOTE)) |
mdoc_pmsg(m, line, *pos, MANDOCERR_BADQUOTE); |
return(ARGS_ERROR); |
|
return(ARGS_QWORD); |
return(ARGS_QWORD); |
} |
} |
|
|
Line 553 args(struct mdoc *m, int line, int *pos, |
|
Line 563 args(struct mdoc *m, int line, int *pos, |
|
(*pos)++; |
(*pos)++; |
|
|
if (0 == buf[*pos] && ! (ARGS_NOWARN & fl)) |
if (0 == buf[*pos] && ! (ARGS_NOWARN & fl)) |
if ( ! mdoc_pmsg(m, line, *pos, MANDOCERR_EOLNSPACE)) |
mdoc_pmsg(m, line, *pos, MANDOCERR_EOLNSPACE); |
return(ARGS_ERROR); |
|
|
|
return(ARGS_QWORD); |
return(ARGS_QWORD); |
} |
} |
Line 577 args(struct mdoc *m, int line, int *pos, |
|
Line 586 args(struct mdoc *m, int line, int *pos, |
|
(*pos)++; |
(*pos)++; |
|
|
if ('\0' == buf[*pos] && ! (ARGS_NOWARN & fl)) |
if ('\0' == buf[*pos] && ! (ARGS_NOWARN & fl)) |
if ( ! mdoc_pmsg(m, line, *pos, MANDOCERR_EOLNSPACE)) |
mdoc_pmsg(m, line, *pos, MANDOCERR_EOLNSPACE); |
return(ARGS_ERROR); |
|
|
|
return(ARGS_WORD); |
return(ARGS_WORD); |
} |
} |
|
|
|
/* |
|
* Check if the string consists only of space-separated closing |
|
* delimiters. This is a bit of a dance: the first must be a close |
|
* delimiter, but it may be followed by middle delimiters. Arbitrary |
|
* whitespace may separate these tokens. |
|
*/ |
|
static int |
|
args_checkpunct(struct mdoc *m, const char *buf, int i, int ln, int fl) |
|
{ |
|
int j; |
|
char dbuf[DELIMSZ]; |
|
enum mdelim d; |
|
|
|
/* First token must be a close-delimiter. */ |
|
|
|
for (j = 0; buf[i] && ' ' != buf[i] && j < DELIMSZ; j++, i++) |
|
dbuf[j] = buf[i]; |
|
|
|
if (DELIMSZ == j) |
|
return(0); |
|
|
|
dbuf[j] = '\0'; |
|
if (DELIM_CLOSE != mdoc_isdelim(dbuf)) |
|
return(0); |
|
|
|
while (' ' == buf[i]) |
|
i++; |
|
|
|
/* Remaining must NOT be open/none. */ |
|
|
|
while (buf[i]) { |
|
j = 0; |
|
while (buf[i] && ' ' != buf[i] && j < DELIMSZ) |
|
dbuf[j++] = buf[i++]; |
|
|
|
if (DELIMSZ == j) |
|
return(0); |
|
|
|
dbuf[j] = '\0'; |
|
d = mdoc_isdelim(dbuf); |
|
if (DELIM_NONE == d || DELIM_OPEN == d) |
|
return(0); |
|
|
|
while (' ' == buf[i]) |
|
i++; |
|
} |
|
|
|
if ( ! (ARGS_NOWARN & fl) && i && ' ' == buf[i - 1]) |
|
mdoc_pmsg(m, ln, i - 1, MANDOCERR_EOLNSPACE); |
|
|
|
return('\0' == buf[i]); |
|
} |
|
|
/* |
/* |
* Match up an argument string (e.g., `-foo bar' having "foo") with the |
* Match up an argument string (e.g., `-foo bar' having "foo") with the |
* correrct identifier. It must apply to the given macro. If none was |
* correrct identifier. It must apply to the given macro. If none was |
Line 591 args(struct mdoc *m, int line, int *pos, |
|
Line 652 args(struct mdoc *m, int line, int *pos, |
|
static enum mdocargt |
static enum mdocargt |
argv_a2arg(enum mdoct tok, const char *p) |
argv_a2arg(enum mdoct tok, const char *p) |
{ |
{ |
enum mdocargt args[MDOC_ARG_MAX]; |
const enum mdocargt *argsp; |
int i, len; |
|
|
|
len = 0; |
argsp = NULL; |
|
|
switch (tok) { |
switch (tok) { |
case (MDOC_An): |
case (MDOC_An): |
args[len++] = MDOC_Split; |
argsp = args_An; |
args[len++] = MDOC_Nosplit; |
|
break; |
break; |
case (MDOC_Bd): |
case (MDOC_Bd): |
args[len++] = MDOC_Ragged; |
argsp = args_Bd; |
args[len++] = MDOC_Unfilled; |
|
args[len++] = MDOC_Filled; |
|
args[len++] = MDOC_Literal; |
|
args[len++] = MDOC_File; |
|
args[len++] = MDOC_Offset; |
|
args[len++] = MDOC_Compact; |
|
args[len++] = MDOC_Centred; |
|
break; |
break; |
case (MDOC_Bf): |
case (MDOC_Bf): |
args[len++] = MDOC_Emphasis; |
argsp = args_Bf; |
args[len++] = MDOC_Literal; |
|
args[len++] = MDOC_Symbolic; |
|
break; |
break; |
case (MDOC_Bk): |
case (MDOC_Bk): |
args[len++] = MDOC_Words; |
argsp = args_Bk; |
break; |
break; |
case (MDOC_Bl): |
case (MDOC_Bl): |
args[len++] = MDOC_Bullet; |
argsp = args_Bl; |
args[len++] = MDOC_Dash; |
|
args[len++] = MDOC_Hyphen; |
|
args[len++] = MDOC_Item; |
|
args[len++] = MDOC_Enum; |
|
args[len++] = MDOC_Tag; |
|
args[len++] = MDOC_Diag; |
|
args[len++] = MDOC_Hang; |
|
args[len++] = MDOC_Ohang; |
|
args[len++] = MDOC_Inset; |
|
args[len++] = MDOC_Column; |
|
args[len++] = MDOC_Width; |
|
args[len++] = MDOC_Offset; |
|
args[len++] = MDOC_Compact; |
|
args[len++] = MDOC_Nested; |
|
break; |
break; |
case (MDOC_Rv): |
case (MDOC_Rv): |
/* FALLTHROUGH */ |
/* FALLTHROUGH */ |
case (MDOC_Ex): |
case (MDOC_Ex): |
args[len++] = MDOC_Std; |
argsp = args_Ex; |
break; |
break; |
default: |
default: |
break; |
return(MDOC_ARG_MAX); |
} |
} |
|
|
for (i = 0; i < len; i++) |
assert(argsp); |
if (0 == strcmp(p, mdoc_argnames[args[i]])) |
|
return(args[i]); |
|
|
|
|
for ( ; MDOC_ARG_MAX != *argsp ; argsp++) |
|
if (0 == strcmp(p, mdoc_argnames[*argsp])) |
|
return(*argsp); |
|
|
return(MDOC_ARG_MAX); |
return(MDOC_ARG_MAX); |
} |
} |
|
|
|
|
static int |
static int |
argv_multi(struct mdoc *m, int line, |
argv_multi(struct mdoc *m, int line, |
struct mdoc_argv *v, int *pos, char *buf) |
struct mdoc_argv *v, int *pos, char *buf) |
Line 679 argv_multi(struct mdoc *m, int line, |
|
Line 716 argv_multi(struct mdoc *m, int line, |
|
return(1); |
return(1); |
} |
} |
|
|
|
|
static int |
static int |
argv_opt_single(struct mdoc *m, int line, |
argv_opt_single(struct mdoc *m, int line, |
struct mdoc_argv *v, int *pos, char *buf) |
struct mdoc_argv *v, int *pos, char *buf) |
Line 703 argv_opt_single(struct mdoc *m, int line, |
|
Line 739 argv_opt_single(struct mdoc *m, int line, |
|
return(1); |
return(1); |
} |
} |
|
|
|
|
/* |
/* |
* Parse a single, mandatory value from the stream. |
* Parse a single, mandatory value from the stream. |
*/ |
*/ |
Line 731 argv_single(struct mdoc *m, int line, |
|
Line 766 argv_single(struct mdoc *m, int line, |
|
return(1); |
return(1); |
} |
} |
|
|
|
|
/* |
/* |
* Determine rules for parsing arguments. Arguments can either accept |
* Determine rules for parsing arguments. Arguments can either accept |
* no parameters, an optional single parameter, one parameter, or |
* no parameters, an optional single parameter, one parameter, or |
Line 745 argv(struct mdoc *mdoc, int line, |
|
Line 779 argv(struct mdoc *mdoc, int line, |
|
v->sz = 0; |
v->sz = 0; |
v->value = NULL; |
v->value = NULL; |
|
|
switch (mdoc_argvflags[v->arg]) { |
switch (argvflags[v->arg]) { |
case (ARGV_SINGLE): |
case (ARGV_SINGLE): |
return(argv_single(mdoc, line, v, pos, buf)); |
return(argv_single(mdoc, line, v, pos, buf)); |
case (ARGV_MULTI): |
case (ARGV_MULTI): |
return(argv_multi(mdoc, line, v, pos, buf)); |
return(argv_multi(mdoc, line, v, pos, buf)); |
case (ARGV_OPT_SINGLE): |
case (ARGV_OPT_SINGLE): |
return(argv_opt_single(mdoc, line, v, pos, buf)); |
return(argv_opt_single(mdoc, line, v, pos, buf)); |
default: |
case (ARGV_NONE): |
/* ARGV_NONE */ |
|
break; |
break; |
|
default: |
|
abort(); |
|
/* NOTREACHED */ |
} |
} |
|
|
return(1); |
return(1); |