Fixed null-pointer dereference.

[mandoc.git] / argv.c

diff --git a/argv.c b/argv.c
index bd2e2c52b969cca45e298e5458a231d0c7ffa09d..9499bd85dc4618a1805d9e6a9a63b53feca2428e 100644 (file)
--- a/argv.c
+++ b/argv.c
@@ -1,4 +1,4 @@
-/* $Id: argv.c,v 1.5 2009/01/01 20:40:16 kristaps Exp $ */
+/* $Id: argv.c,v 1.41 2009/03/08 12:46:38 kristaps Exp $ */

 /*
  * Copyright (c) 2008 Kristaps Dzonsons <kristaps@kth.se>
  *
 /*
  * Copyright (c) 2008 Kristaps Dzonsons <kristaps@kth.se>
  *


@@ -25,39 +25,448 @@
 
 #include "private.h"
 
 
 #include "private.h"
 

+/*
+ * 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.
+ */

 
 

-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);
+#define        ARGS_QUOTED     (1 << 0)
+#define        ARGS_DELIM      (1 << 1)
+#define        ARGS_TABSEP     (1 << 2)
+#define        ARGS_ARGVLIKE   (1 << 3)
+
+#define        ARGV_NONE       (1 << 0)
+#define        ARGV_SINGLE     (1 << 1)
+#define        ARGV_MULTI      (1 << 2)
+#define        ARGV_OPT_SINGLE (1 << 3)
+
+enum   mwarn {
+       WQUOTPARM,
+       WARGVPARM,
+       WCOLEMPTY,
+       WTAILWS 
+};
+
+enum   merr {
+       EQUOTTERM,
+       EARGVAL 
+};
+
+static int              argv_a2arg(int, const char *);
+static int              args(struct mdoc *, int, int *, 
+                               char *, int, char **);
+static int              argv(struct mdoc *, int, int,
+                               struct mdoc_argv *, int *, char *);
+static int              argv_single(struct mdoc *, int, 
+                               struct mdoc_argv *, int *, char *);
+static int              argv_opt_single(struct mdoc *, int, 
+                               struct mdoc_argv *, int *, char *);
+static int              argv_multi(struct mdoc *, int, 
+                               struct mdoc_argv *, int *, char *);
+static int              pwarn(struct mdoc *, int, int, enum mwarn);
+static int              perr(struct mdoc *, int, int, enum merr);
+
+/* Per-argument flags. */
+
+static int mdoc_argvflags[MDOC_ARG_MAX] = {
+       ARGV_NONE,      /* MDOC_Split */
+       ARGV_NONE,      /* MDOC_Nosplit */
+       ARGV_NONE,      /* MDOC_Ragged */
+       ARGV_NONE,      /* MDOC_Unfilled */
+       ARGV_NONE,      /* MDOC_Literal */
+       ARGV_NONE,      /* MDOC_File */
+       ARGV_SINGLE,    /* MDOC_Offset */
+       ARGV_NONE,      /* MDOC_Bullet */
+       ARGV_NONE,      /* MDOC_Dash */
+       ARGV_NONE,      /* MDOC_Hyphen */
+       ARGV_NONE,      /* MDOC_Item */
+       ARGV_NONE,      /* MDOC_Enum */
+       ARGV_NONE,      /* MDOC_Tag */
+       ARGV_NONE,      /* MDOC_Diag */
+       ARGV_NONE,      /* MDOC_Hang */
+       ARGV_NONE,      /* MDOC_Ohang */
+       ARGV_NONE,      /* MDOC_Inset */
+       ARGV_MULTI,     /* MDOC_Column */
+       ARGV_SINGLE,    /* MDOC_Width */
+       ARGV_NONE,      /* MDOC_Compact */
+       ARGV_SINGLE,    /* MDOC_Std */
+       ARGV_NONE,      /* MDOC_Filled */
+       ARGV_NONE,      /* MDOC_Words */
+       ARGV_NONE,      /* MDOC_Emphasis */
+       ARGV_NONE       /* MDOC_Symbolic */
+};
+
+static int mdoc_argflags[MDOC_MAX] = {
+       0, /* \" */
+       0, /* Dd */
+       0, /* Dt */
+       0, /* Os */
+       0, /* Sh */
+       0, /* Ss */ 
+       ARGS_DELIM, /* Pp */ 
+       ARGS_DELIM, /* D1 */
+       ARGS_DELIM, /* Dl */
+       0, /* Bd */
+       0, /* Ed */
+       0, /* Bl */
+       0, /* El */
+       0, /* It */
+       ARGS_DELIM, /* Ad */ 
+       ARGS_DELIM, /* An */
+       ARGS_DELIM, /* Ar */
+       ARGS_QUOTED, /* Cd */
+       ARGS_DELIM, /* Cm */
+       ARGS_DELIM, /* Dv */ 
+       ARGS_DELIM, /* Er */ 
+       ARGS_DELIM, /* Ev */ 
+       0, /* Ex */
+       ARGS_DELIM | ARGS_QUOTED, /* Fa */ 
+       0, /* Fd */ 
+       ARGS_DELIM, /* Fl */
+       ARGS_DELIM | ARGS_QUOTED, /* Fn */ 
+       ARGS_DELIM | ARGS_QUOTED, /* Ft */ 
+       ARGS_DELIM, /* Ic */ 
+       0, /* In */ 
+       ARGS_DELIM, /* Li */
+       0, /* Nd */ 
+       ARGS_DELIM, /* Nm */ 
+       ARGS_DELIM, /* Op */
+       0, /* Ot */
+       ARGS_DELIM, /* Pa */
+       0, /* Rv */
+       ARGS_DELIM | ARGS_ARGVLIKE, /* St */ 
+       ARGS_DELIM, /* Va */
+       ARGS_DELIM, /* Vt */ 
+       ARGS_DELIM, /* Xr */
+       ARGS_QUOTED, /* %A */
+       ARGS_QUOTED, /* %B */
+       ARGS_QUOTED, /* %D */
+       ARGS_QUOTED, /* %I */
+       ARGS_QUOTED, /* %J */
+       ARGS_QUOTED, /* %N */
+       ARGS_QUOTED, /* %O */
+       ARGS_QUOTED, /* %P */
+       ARGS_QUOTED, /* %R */
+       ARGS_QUOTED, /* %T */
+       ARGS_QUOTED, /* %V */
+       ARGS_DELIM, /* Ac */
+       0, /* Ao */
+       ARGS_DELIM, /* Aq */
+       ARGS_DELIM, /* At */
+       ARGS_DELIM, /* Bc */
+       0, /* Bf */ 
+       0, /* Bo */
+       ARGS_DELIM, /* Bq */
+       ARGS_DELIM, /* Bsx */
+       ARGS_DELIM, /* Bx */
+       0, /* Db */
+       ARGS_DELIM, /* Dc */
+       0, /* Do */
+       ARGS_DELIM, /* Dq */
+       ARGS_DELIM, /* Ec */
+       0, /* Ef */
+       ARGS_DELIM, /* Em */ 
+       0, /* Eo */
+       ARGS_DELIM, /* Fx */
+       ARGS_DELIM, /* Ms */
+       ARGS_DELIM, /* No */
+       ARGS_DELIM, /* Ns */
+       ARGS_DELIM, /* Nx */
+       ARGS_DELIM, /* Ox */
+       ARGS_DELIM, /* Pc */
+       ARGS_DELIM, /* Pf */
+       0, /* Po */
+       ARGS_DELIM, /* Pq */
+       ARGS_DELIM, /* Qc */
+       ARGS_DELIM, /* Ql */
+       0, /* Qo */
+       ARGS_DELIM, /* Qq */
+       0, /* Re */
+       0, /* Rs */
+       ARGS_DELIM, /* Sc */
+       0, /* So */
+       ARGS_DELIM, /* Sq */
+       0, /* Sm */
+       ARGS_DELIM, /* Sx */
+       ARGS_DELIM, /* Sy */
+       ARGS_DELIM, /* Tn */
+       ARGS_DELIM, /* Ux */
+       ARGS_DELIM, /* Xc */
+       0, /* Xo */
+       0, /* Fo */ 
+       0, /* Fc */ 
+       0, /* Oo */
+       ARGS_DELIM, /* Oc */
+       0, /* Bk */
+       0, /* Ek */
+       0, /* Bt */
+       0, /* Hf */
+       0, /* Fr */
+       0, /* Ud */
+       0, /* Lb */
+};

 
 
 
 

+/*
+ * Parse an argument from line text.  This comes in the form of -key
+ * [value0...], which may either have a single mandatory value, at least
+ * one mandatory value, an optional single value, or no value.
+ */

 int
 int

-mdoc_args(struct mdoc *mdoc, int tok, int *pos, char *buf, int fl, char **v)
+mdoc_argv(struct mdoc *mdoc, int line, int tok,
+               struct mdoc_arg **v, int *pos, char *buf)

 {
 {

-       int              i;
+       int               i;
+       char             *p;
+       struct mdoc_argv tmp;
+       struct mdoc_arg  *arg;
+
+       if (0 == buf[*pos])
+               return(ARGV_EOLN);
+
+       assert( ! isspace((u_char)buf[*pos]));
+
+       if ('-' != buf[*pos])
+               return(ARGV_WORD);
+
+       i = *pos;
+       p = &buf[++(*pos)];
+
+       assert(*pos > 0);
+
+       /* LINTED */
+       while (buf[*pos]) {
+               if (isspace((u_char)buf[*pos])) 
+                       if ('\\' != buf[*pos - 1])
+                               break;
+               (*pos)++;
+       }
+
+       if (buf[*pos])
+               buf[(*pos)++] = 0;
+
+       (void)memset(&tmp, 0, sizeof(struct mdoc_argv));
+       tmp.line = line;
+       tmp.pos = *pos;
+
+       /*
+        * We now parse out the per-macro arguments.  XXX - this can be
+        * made much cleaner using per-argument tables.  See argv_a2arg
+        * for details.
+        */
+
+       if (MDOC_ARG_MAX == (tmp.arg = argv_a2arg(tok, p))) {
+               if ( ! pwarn(mdoc, line, i, WARGVPARM))
+                       return(ARGV_ERROR);
+               return(ARGV_WORD);
+       }
+
+       while (buf[*pos] && isspace((u_char)buf[*pos]))
+               (*pos)++;
+
+       /* FIXME: whitespace if no value. */
+
+       if ( ! argv(mdoc, tok, line, &tmp, pos, buf))
+               return(ARGV_ERROR);
+
+       if (NULL == (arg = *v)) {
+               *v = xcalloc(1, sizeof(struct mdoc_arg));
+               arg = *v;
+       } 
+
+       arg->argc++;
+       arg->argv = xrealloc(arg->argv, arg->argc * 
+                       sizeof(struct mdoc_argv));
+
+       (void)memcpy(&arg->argv[(int)arg->argc - 1], 
+                       &tmp, sizeof(struct mdoc_argv));
+
+       return(ARGV_ARG);
+}
+
+
+void
+mdoc_argv_free(struct mdoc_arg *p)
+{
+       int              i, j;
+
+       if (NULL == p)
+               return;
+
+       if (p->refcnt) {
+               --(p->refcnt);
+               if (p->refcnt)
+                       return;
+       }
+
+       assert(p->argc);
+
+       /* LINTED */
+       for (i = 0; i < (int)p->argc; i++) {
+               if (0 == p->argv[i].sz)
+                       continue;
+               /* LINTED */
+               for (j = 0; j < (int)p->argv[i].sz; j++)
+                       free(p->argv[i].value[j]);
+
+               free(p->argv[i].value);
+       }
+
+       free(p->argv);
+       free(p);
+}
+
+
+
+static int
+perr(struct mdoc *mdoc, int line, int pos, enum merr code)
+{
+       char            *p;
+
+       p = NULL;
+
+       switch (code) {
+       case (EQUOTTERM):
+               p = "unterminated quoted parameter";
+               break;
+       case (EARGVAL):
+               p = "argument requires a value";
+               break;
+       }
+
+       assert(p);
+       return(mdoc_perr(mdoc, line, pos, p));
+}
+
+
+static int
+pwarn(struct mdoc *mdoc, int line, int pos, enum mwarn code)
+{
+       char            *p;
+       int              c;
+
+       p = NULL;
+       c = WARN_SYNTAX;
+
+       switch (code) {
+       case (WQUOTPARM):
+               p = "unexpected quoted parameter";
+               break;
+       case (WARGVPARM):
+               p = "argument-like parameter";
+               break;
+       case (WCOLEMPTY):
+               p = "last list column is empty";
+               c = WARN_COMPAT;
+               break;
+       case (WTAILWS):
+               p = "trailing whitespace";
+               c = WARN_COMPAT;
+               break;
+       }
+
+       assert(p);
+       return(mdoc_pwarn(mdoc, line, pos, c, p));
+}
+
+
+int
+mdoc_args(struct mdoc *mdoc, int line, 
+               int *pos, char *buf, int tok, char **v)
+{
+       int               fl, c, i;
+       struct mdoc_node *n;
+
+       fl = (0 == tok) ? 0 : mdoc_argflags[tok];
+
+       /* 
+        * Override per-macro argument flags with context-specific ones.
+        * As of now, this is only valid for `It' depending on its list
+        * context.
+        */
+
+       switch (tok) {
+       case (MDOC_It):
+               for (n = mdoc->last; n; n = n->parent)
+                       if (MDOC_BLOCK == n->type && MDOC_Bl == n->tok)
+                               break;
+
+               assert(n);
+               c = (int)(n->args ? n->args->argc : 0);
+               assert(c > 0);
+
+               /*
+                * Using `Bl -column' adds ARGS_TABSEP to the arguments
+                * and invalidates ARGS_DELIM.  Using `Bl -diag' allows
+                * for quoted arguments.
+                */
+
+               /* LINTED */
+               for (i = 0; i < c; i++) {
+                       switch (n->args->argv[i].arg) {
+                       case (MDOC_Column):
+                               fl |= ARGS_TABSEP;
+                               fl &= ~ARGS_DELIM;
+                               i = c;
+                               break;
+                       case (MDOC_Diag):
+                               fl |= ARGS_QUOTED;
+                               i = c;
+                               break;
+                       default:
+                               break;
+                       }
+               }
+               break;
+       default:
+               break;
+       }
+
+       /* Continue parsing the arguments themselves...  */
+
+       return(args(mdoc, line, pos, buf, fl, v));
+}
+
+
+static int
+args(struct mdoc *mdoc, int line, 
+               int *pos, char *buf, int fl, char **v)
+{
+       int               i;
+       char             *p, *pp;
+
+       assert(*pos > 0);

 
        if (0 == buf[*pos])
                return(ARGS_EOLN);
 
        if ('\"' == buf[*pos] && ! (fl & ARGS_QUOTED))
 
        if (0 == buf[*pos])
                return(ARGS_EOLN);
 
        if ('\"' == buf[*pos] && ! (fl & ARGS_QUOTED))

-               if ( ! mdoc_warn(mdoc, tok, *pos, WARN_SYNTAX_QUOTED))
+               if ( ! pwarn(mdoc, line, *pos, WQUOTPARM))

                        return(ARGS_ERROR);
 
                        return(ARGS_ERROR);
 

-       if ('-' == buf[*pos]) 
-               if ( ! mdoc_warn(mdoc, tok, *pos, WARN_SYNTAX_ARGLIKE))
+       if ( ! (fl & ARGS_ARGVLIKE) && '-' == buf[*pos]) 
+               if ( ! pwarn(mdoc, line, *pos, WARGVPARM))

                        return(ARGS_ERROR);
 
                        return(ARGS_ERROR);
 

+       /* 
+        * If the first character is a delimiter and we're to look for
+        * delimited strings, then pass down the buffer seeing if it
+        * follows the pattern of [[::delim::][ ]+]+.
+        */
+

        if ((fl & ARGS_DELIM) && mdoc_iscdelim(buf[*pos])) {
                for (i = *pos; buf[i]; ) {
                        if ( ! mdoc_iscdelim(buf[i]))
                                break;
                        i++;
        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]))
+                       /* There must be at least one space... */
+                       if (0 == buf[i] || ! isspace((u_char)buf[i]))

                                break;
                        i++;
                                break;
                        i++;

-                       while (buf[i] && isspace(buf[i]))
+                       while (buf[i] && isspace((u_char)buf[i]))

                                i++;
                }
                if (0 == buf[i]) {
                                i++;
                }
                if (0 == buf[i]) {


@@ -66,30 +475,127 @@ mdoc_args(struct mdoc *mdoc, int tok, int *pos, char *buf, int fl, char **v)
                }
        }
 
                }
        }
 

-       /*
-        * Parse routine for non-quoted string.  
-        */
+       /* First parse non-quoted strings. */

 
 

-       if ('\"' != buf[*pos]) {
+       if ('\"' != buf[*pos] || ! (ARGS_QUOTED & fl)) {

                *v = &buf[*pos];
 
                *v = &buf[*pos];
 

-               while (buf[*pos] && ! isspace(buf[*pos]))
-                       (*pos)++;
+               /* 
+                * Thar be dragons here!  If we're tab-separated, search
+                * ahead for either a tab or the `Ta' macro.  If a tab
+                * is detected, it mustn't be escaped; if a `Ta' is
+                * detected, it must be space-buffered before and after.
+                * If either of these hold true, then prune out the
+                * extra spaces and call it an argument.
+                */
+
+               if (ARGS_TABSEP & fl) {
+                       /* Scan ahead to unescaped tab. */
+
+                       for (p = *v; ; p++) {
+                               if (NULL == (p = strchr(p, '\t')))
+                                       break;
+                               if (p == *v)
+                                       break;
+                               if ('\\' != *(p - 1))
+                                       break;
+                       }
+
+                       /* Scan ahead to unescaped `Ta'. */
+
+                       for (pp = *v; ; pp++) {
+                               if (NULL == (pp = strstr(pp, "Ta")))
+                                       break;
+                               if (pp > *v && ' ' != *(pp - 1))
+                                       continue;
+                               if (' ' == *(pp + 2) || 0 == *(pp + 2))
+                                       break;
+                       }
+
+                       /* Choose delimiter tab/Ta. */
+
+                       if (p && pp)
+                               p = (p < pp ? p : pp);
+                       else if ( ! p && pp)
+                               p = pp;
+
+                       /* Strip delimiter's preceding whitespace. */
+
+                       if (p && p > *v) {
+                               pp = p - 1;
+                               while (pp > *v && ' ' == *pp)
+                                       pp--;
+                               if (pp == *v && ' ' == *pp) 
+                                       *pp = 0;
+                               else if (' ' == *pp)
+                                       *(pp + 1) = 0;
+                       }
+
+                       /* ...in- and proceding whitespace. */
+
+                       if (p && ('\t' != *p)) {
+                               *p++ = 0;
+                               *p++ = 0;
+                       } else if (p)
+                               *p++ = 0;
+
+                       if (p) {
+                               while (' ' == *p)
+                                       p++;
+                               if (0 != *p)
+                                       *(p - 1) = 0;
+                               *pos += (int)(p - *v);
+                       } 
+
+                       if (p && 0 == *p)
+                               if ( ! pwarn(mdoc, line, *pos, WCOLEMPTY))
+                                       return(0);
+                       if (p && 0 == *p && p > *v && ' ' == *(p - 1))
+                               if ( ! pwarn(mdoc, line, *pos, WTAILWS))
+                                       return(0);
+
+                       if (p)
+                               return(ARGS_WORD);
+
+                       /* Configure the eoln case, too. */
+
+                       p = strchr(*v, 0);
+                       assert(p);
+
+                       if (p > *v && ' ' == *(p - 1))
+                               if ( ! pwarn(mdoc, line, *pos, WTAILWS))
+                                       return(0);
+                       *pos += (int)(p - *v);
+
+                       return(ARGS_WORD);
+               } 
+
+               /* Do non-tabsep look-ahead here. */
+               
+               if ( ! (ARGS_TABSEP & fl))
+                       while (buf[*pos]) {
+                               if (isspace((u_char)buf[*pos]))
+                                       if ('\\' != buf[*pos - 1])
+                                               break;
+                               (*pos)++;
+                       }

 
                if (0 == buf[*pos])
                        return(ARGS_WORD);
 
                buf[(*pos)++] = 0;
 
                if (0 == buf[*pos])
                        return(ARGS_WORD);
 
                buf[(*pos)++] = 0;

+

                if (0 == buf[*pos])
                        return(ARGS_WORD);
 
                if (0 == buf[*pos])
                        return(ARGS_WORD);
 

-               while (buf[*pos] && isspace(buf[*pos]))
-                       (*pos)++;
+               if ( ! (ARGS_TABSEP & fl))
+                       while (buf[*pos] && isspace((u_char)buf[*pos]))
+                               (*pos)++;

 
                if (buf[*pos])
                        return(ARGS_WORD);
 
 
                if (buf[*pos])
                        return(ARGS_WORD);
 

-               if ( ! mdoc_warn(mdoc, tok, *pos, WARN_SYNTAX_WS_EOLN))
+               if ( ! pwarn(mdoc, line, *pos, WTAILWS))

                        return(ARGS_ERROR);
 
                return(ARGS_WORD);
                        return(ARGS_ERROR);
 
                return(ARGS_WORD);


@@ -107,37 +613,54 @@ mdoc_args(struct mdoc *mdoc, int tok, int *pos, char *buf, int fl, char **v)
                (*pos)++;
 
        if (0 == buf[*pos]) {
                (*pos)++;
 
        if (0 == buf[*pos]) {

-               (void)mdoc_err(mdoc, tok, *pos, ERR_SYNTAX_UNQUOTE);
+               (void)perr(mdoc, line, *pos, EQUOTTERM);

                return(ARGS_ERROR);
        }
 
        buf[(*pos)++] = 0;
        if (0 == buf[*pos])
                return(ARGS_ERROR);
        }
 
        buf[(*pos)++] = 0;
        if (0 == buf[*pos])

-               return(ARGS_WORD);
+               return(ARGS_QWORD);

 
 

-       while (buf[*pos] && isspace(buf[*pos]))
+       while (buf[*pos] && isspace((u_char)buf[*pos]))

                (*pos)++;
 
        if (buf[*pos])
                (*pos)++;
 
        if (buf[*pos])

-               return(ARGS_WORD);
+               return(ARGS_QWORD);

 
 

-       if ( ! mdoc_warn(mdoc, tok, *pos, WARN_SYNTAX_WS_EOLN))
+       if ( ! pwarn(mdoc, line, *pos, WTAILWS))

                return(ARGS_ERROR);
 
                return(ARGS_ERROR);
 

-       return(ARGS_WORD);
+       return(ARGS_QWORD);

 }
 
 
 static int
 }
 
 
 static int

-lookup(int tok, const char *argv)
+argv_a2arg(int tok, const char *argv)

 {
 
 {
 

+       /*
+        * Parse an argument identifier from its text.  XXX - this
+        * should really be table-driven to clarify the code.
+        *
+        * If you add an argument to the list, make sure that you
+        * register it here with its one or more macros!
+        */
+

        switch (tok) {
        switch (tok) {

+       case (MDOC_An):
+               if (xstrcmp(argv, "split"))
+                       return(MDOC_Split);
+               else if (xstrcmp(argv, "nosplit"))
+                       return(MDOC_Nosplit);
+               break;
+

        case (MDOC_Bd):
                if (xstrcmp(argv, "ragged"))
                        return(MDOC_Ragged);
                else if (xstrcmp(argv, "unfilled"))
                        return(MDOC_Unfilled);
        case (MDOC_Bd):
                if (xstrcmp(argv, "ragged"))
                        return(MDOC_Ragged);
                else if (xstrcmp(argv, "unfilled"))
                        return(MDOC_Unfilled);

+               else if (xstrcmp(argv, "filled"))
+                       return(MDOC_Filled);

                else if (xstrcmp(argv, "literal"))
                        return(MDOC_Literal);
                else if (xstrcmp(argv, "file"))
                else if (xstrcmp(argv, "literal"))
                        return(MDOC_Literal);
                else if (xstrcmp(argv, "file"))


@@ -146,6 +669,20 @@ lookup(int tok, const char *argv)
                        return(MDOC_Offset);
                break;
 
                        return(MDOC_Offset);
                break;
 

+       case (MDOC_Bf):
+               if (xstrcmp(argv, "emphasis"))
+                       return(MDOC_Emphasis);
+               else if (xstrcmp(argv, "literal"))
+                       return(MDOC_Literal);
+               else if (xstrcmp(argv, "symbolic"))
+                       return(MDOC_Symbolic);
+               break;
+
+       case (MDOC_Bk):
+               if (xstrcmp(argv, "words"))
+                       return(MDOC_Words);
+               break;
+

        case (MDOC_Bl):
                if (xstrcmp(argv, "bullet"))
                        return(MDOC_Bullet);
        case (MDOC_Bl):
                if (xstrcmp(argv, "bullet"))
                        return(MDOC_Bullet);


@@ -183,274 +720,132 @@ lookup(int tok, const char *argv)
                if (xstrcmp(argv, "std"))
                        return(MDOC_Std);
                break;
                if (xstrcmp(argv, "std"))
                        return(MDOC_Std);
                break;

-       
-       case (MDOC_St):
-               if (xstrcmp(argv, "p1003.1-88"))
-                       return(MDOC_p1003_1_88);
-               else if (xstrcmp(argv, "p1003.1-90"))
-                       return(MDOC_p1003_1_90);
-               else if (xstrcmp(argv, "p1003.1-96"))
-                       return(MDOC_p1003_1_96);
-               else if (xstrcmp(argv, "p1003.1-2001"))
-                       return(MDOC_p1003_1_2001);
-               else if (xstrcmp(argv, "p1003.1-2004"))
-                       return(MDOC_p1003_1_2004);
-               else if (xstrcmp(argv, "p1003.1"))
-                       return(MDOC_p1003_1);
-               else if (xstrcmp(argv, "p1003.1b"))
-                       return(MDOC_p1003_1b);
-               else if (xstrcmp(argv, "p1003.1b-93"))
-                       return(MDOC_p1003_1b_93);
-               else if (xstrcmp(argv, "p1003.1c-95"))
-                       return(MDOC_p1003_1c_95);
-               else if (xstrcmp(argv, "p1003.1g-2000"))
-                       return(MDOC_p1003_1g_2000);
-               else if (xstrcmp(argv, "p1003.2-92"))
-                       return(MDOC_p1003_2_92);
-               else if (xstrcmp(argv, "p1003.2-95"))
-                       return(MDOC_p1387_2_95);
-               else if (xstrcmp(argv, "p1003.2"))
-                       return(MDOC_p1003_2);
-               else if (xstrcmp(argv, "p1387.2-95"))
-                       return(MDOC_p1387_2);
-               else if (xstrcmp(argv, "isoC-90"))
-                       return(MDOC_isoC_90);
-               else if (xstrcmp(argv, "isoC-amd1"))
-                       return(MDOC_isoC_amd1);
-               else if (xstrcmp(argv, "isoC-tcor1"))
-                       return(MDOC_isoC_tcor1);
-               else if (xstrcmp(argv, "isoC-tcor2"))
-                       return(MDOC_isoC_tcor2);
-               else if (xstrcmp(argv, "isoC-99"))
-                       return(MDOC_isoC_99);
-               else if (xstrcmp(argv, "ansiC"))
-                       return(MDOC_ansiC);
-               else if (xstrcmp(argv, "ansiC-89"))
-                       return(MDOC_ansiC_89);
-               else if (xstrcmp(argv, "ansiC-99"))
-                       return(MDOC_ansiC_99);
-               else if (xstrcmp(argv, "ieee754"))
-                       return(MDOC_ieee754);
-               else if (xstrcmp(argv, "iso8802-3"))
-                       return(MDOC_iso8802_3);
-               else if (xstrcmp(argv, "xpg3"))
-                       return(MDOC_xpg3);
-               else if (xstrcmp(argv, "xpg4"))
-                       return(MDOC_xpg4);
-               else if (xstrcmp(argv, "xpg4.2"))
-                       return(MDOC_xpg4_2);
-               else if (xstrcmp(argv, "xpg4.3"))
-                       return(MDOC_xpg4_3);
-               else if (xstrcmp(argv, "xbd5"))
-                       return(MDOC_xbd5);
-               else if (xstrcmp(argv, "xcu5"))
-                       return(MDOC_xcu5);
-               else if (xstrcmp(argv, "xsh5"))
-                       return(MDOC_xsh5);
-               else if (xstrcmp(argv, "xns5"))
-                       return(MDOC_xns5);
-               else if (xstrcmp(argv, "xns5.2d2.0"))
-                       return(MDOC_xns5_2d2_0);
-               else if (xstrcmp(argv, "xcurses4.2"))
-                       return(MDOC_xcurses4_2);
-               else if (xstrcmp(argv, "susv2"))
-                       return(MDOC_susv2);
-               else if (xstrcmp(argv, "susv3"))
-                       return(MDOC_susv3);
-               else if (xstrcmp(argv, "svid4"))
-                       return(MDOC_svid4);
-               break;
-
-       default:
-               abort();
-               /* NOTREACHED */
-       }
-
-       return(MDOC_ARG_MAX);
-}
-
-
-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;
        }
 
        default:
                break;
        }
 

-       return(1);
+       return(MDOC_ARG_MAX);

 }
 
 
 static int
 }
 
 
 static int

-parse(struct mdoc *mdoc, int tok, 
-               struct mdoc_arg *v, int *pos, char *buf)
+argv_multi(struct mdoc *mdoc, int line, 
+               struct mdoc_argv *v, int *pos, char *buf)

 {
 {

+       int              c, ppos;

        char            *p;
        char            *p;

-       int              c, ppos, i;

 
        ppos = *pos;
 
 
        ppos = *pos;
 

-       switch (v->arg) {
-       case(MDOC_Compact):
-               /* FALLTHROUGH */
-       case(MDOC_Ragged):
-               /* FALLTHROUGH */
-       case(MDOC_Unfilled):
-               /* FALLTHROUGH */
-       case(MDOC_Literal):
-               /* FALLTHROUGH */
-       case(MDOC_File):
-               /* FALLTHROUGH */
-       case(MDOC_Bullet):
-               /* FALLTHROUGH */
-       case(MDOC_Dash):
-               /* FALLTHROUGH */
-       case(MDOC_Hyphen):
-               /* FALLTHROUGH */
-       case(MDOC_Item):
-               /* FALLTHROUGH */
-       case(MDOC_Enum):
-               /* FALLTHROUGH */
-       case(MDOC_Tag):
-               /* FALLTHROUGH */
-       case(MDOC_Diag):
-               /* FALLTHROUGH */
-       case(MDOC_Hang):
-               /* FALLTHROUGH */
-       case(MDOC_Ohang):
-               /* FALLTHROUGH */
-       case(MDOC_Inset):
-               v->sz = 0;
-               v->value = NULL;
-               break;
-
-       case(MDOC_Std):
-               /* FALLTHROUGH */
-       case(MDOC_Width):
-               /* FALLTHROUGH */
-       case(MDOC_Offset):
-               /*
-                * This has a single value for an argument.
-                */
-               c = mdoc_args(mdoc, tok, pos, buf, ARGS_QUOTED, &p);
+       for (v->sz = 0; ; v->sz++) {
+               if ('-' == buf[*pos])
+                       break;
+               c = args(mdoc, line, pos, buf, ARGS_QUOTED, &p);

                if (ARGS_ERROR == c)
                        return(0);
                else if (ARGS_EOLN == c)
                if (ARGS_ERROR == c)
                        return(0);
                else if (ARGS_EOLN == c)

-                       return(mdoc_err(mdoc, tok, ppos, ERR_SYNTAX_ARGVAL));
-                       
-               v->sz = 1;
-               v->value = xcalloc(1, sizeof(char *));
-               v->value[0] = p;
-               break;
+                       break;

 
 

-       case(MDOC_Column):
-               /*
-                * This has several value for a single argument.  We
-                * pre-allocate a pointer array and don't let it exceed
-                * this size.
-                */
-               v->sz = 0;
-               v->value = xcalloc(MDOC_LINEARG_MAX, sizeof(char *));
-               for (i = 0; i < MDOC_LINEARG_MAX; i++) {
-                       c = mdoc_args(mdoc, tok, pos, buf, ARGS_QUOTED, &p);
-                       if (ARGS_ERROR == c) {
-                               free(v->value);
-                               return(0);
-                       } else if (ARGS_EOLN == c)
-                               break;
-                       v->value[i] = p;
-               }
-               if (0 == i) {
-                       free(v->value);
-                       return(mdoc_err(mdoc, tok, ppos, ERR_SYNTAX_ARGVAL));
-               } else if (MDOC_LINEARG_MAX == i)
-                       return(mdoc_err(mdoc, tok, ppos, ERR_SYNTAX_ARGMANY));
+               if (0 == v->sz % 5)
+                       v->value = xrealloc(v->value, 
+                               (v->sz + 5) * sizeof(char *));

 
 

-               v->sz = i;
-               break;
-       default:
-               abort();
-               /* NOTREACHED */
+               v->value[(int)v->sz] = xstrdup(p);

        }
 
        }
 

-       return(1);
+       if (v->sz)
+               return(1);
+
+       return(perr(mdoc, line, ppos, EARGVAL));

 }
 
 
 }
 
 

-int
-mdoc_argv(struct mdoc *mdoc, int tok, 
-               struct mdoc_arg *v, int *pos, char *buf)
+static int
+argv_opt_single(struct mdoc *mdoc, int line, 
+               struct mdoc_argv *v, int *pos, char *buf)

 {
 {

-       int              i, ppos;
-       char            *argv;
+       int              c;
+       char            *p;

 
 

-       (void)memset(v, 0, sizeof(struct mdoc_arg));
+       if ('-' == buf[*pos])
+               return(1);

 
 

-       if (0 == buf[*pos])
-               return(ARGV_EOLN);
+       c = args(mdoc, line, pos, buf, ARGS_QUOTED, &p);
+       if (ARGS_ERROR == c)
+               return(0);
+       if (ARGS_EOLN == c)
+               return(1);

 
 

-       assert( ! isspace(buf[*pos]));
-
-       if ('-' != buf[*pos])
-               return(ARGV_WORD);
+       v->sz = 1;
+       v->value = xcalloc(1, sizeof(char *));
+       v->value[0] = xstrdup(p);
+       return(1);
+}

 
 

-       i = *pos;
-       argv = &buf[++(*pos)];

 
 

-       while (buf[*pos] && ! isspace(buf[*pos]))
-               (*pos)++;
+/*
+ * Parse a single, mandatory value from the stream.
+ */
+static int
+argv_single(struct mdoc *mdoc, int line, 
+               struct mdoc_argv *v, int *pos, char *buf)
+{
+       int              c, ppos;
+       char            *p;

 
 

-       if (buf[*pos])
-               buf[(*pos)++] = 0;
+       ppos = *pos;

 
 

-       if (MDOC_ARG_MAX == (v->arg = lookup(tok, argv))) {
-               (void)mdoc_err(mdoc, tok, i, ERR_SYNTAX_ARG);
-               return(ARGV_ERROR);
-       }
+       c = args(mdoc, line, pos, buf, ARGS_QUOTED, &p);
+       if (ARGS_ERROR == c)
+               return(0);
+       if (ARGS_EOLN == c)
+               return(perr(mdoc, line, ppos, EARGVAL));

 
 

-       while (buf[*pos] && isspace(buf[*pos]))
-               (*pos)++;
+       v->sz = 1;
+       v->value = xcalloc(1, sizeof(char *));
+       v->value[0] = xstrdup(p);
+       return(1);
+}

 
 

-       /* FIXME: whitespace if no value. */

 
 

-       ppos = *pos;
-       if ( ! parse(mdoc, tok, v, pos, buf))
-               return(ARGV_ERROR);
-       if ( ! postparse(mdoc, tok, v, ppos))
-               return(ARGV_ERROR);
+/*
+ * Determine rules for parsing arguments.  Arguments can either accept
+ * no parameters, an optional single parameter, one parameter, or
+ * multiple parameters.
+ */
+static int
+argv(struct mdoc *mdoc, int tok, int line, 
+               struct mdoc_argv *v, int *pos, char *buf)
+{
+       int              fl;

 
 

-       return(ARGV_ARG);
-}
+       v->sz = 0;
+       v->value = NULL;
+       fl = mdoc_argvflags[v->arg];

 
 

+       /*
+        * Override the default per-argument value.
+        */

 
 

-void
-mdoc_argv_free(int sz, struct mdoc_arg *arg)
-{
-       int              i;
+       switch (tok) {
+       case (MDOC_Ex):
+               fl = ARGV_OPT_SINGLE;
+               break;
+       default:
+               break;
+       }

 
 

-       for (i = 0; i < sz; i++) {
-               if (0 == arg[i].sz) {
-                       assert(NULL == arg[i].value);
-                       continue;
-               }
-               assert(arg[i].value);
-               free(arg[i].value);
+       switch (fl) {
+       case (ARGV_SINGLE):
+               return(argv_single(mdoc, line, v, pos, buf));
+       case (ARGV_MULTI):
+               return(argv_multi(mdoc, line, v, pos, buf));
+       case (ARGV_OPT_SINGLE):
+               return(argv_opt_single(mdoc, line, v, pos, buf));
+       default:
+               /* ARGV_NONE */
+               break;

        }
        }

-}

 
 

+       return(1);
+}