Fixed delimiter end-of-line parsing.

[mandoc.git] / argv.c

diff --git a/argv.c b/argv.c
index 63dfde9eb8c0d72e92b18c155148f371c9d690de..8ad5d662ff223b4d3e4b395cbfdcd3be45bec86c 100644 (file)
--- a/argv.c
+++ b/argv.c
@@ -1,4 +1,4 @@
-/* $Id: argv.c,v 1.23 2009/01/20 22:55:46 kristaps Exp $ */
+/* $Id: argv.c,v 1.34 2009/02/28 12:16:02 kristaps Exp $ */

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


@@ -26,27 +26,39 @@
 #include "private.h"
 
 /*
 #include "private.h"
 
 /*

- * Parse arguments and parameters of macros.  Arguments follow the
- * syntax of `-arg [val [valN...]]', while parameters are free-form text
- * following arguments (if any).  This file must correctly handle the
- * strange punctuation rules dictated by groff.
+ * 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.

  */
 
  */
 

-/* FIXME: .It called with -column and quoted arguments. */
+#define        ARGS_QUOTED     (1 << 0)
+#define        ARGS_DELIM      (1 << 1)
+#define        ARGS_TABSEP     (1 << 2)

 
 

-static int              lookup(int, const char *);
-static int              parse(struct mdoc *, int,
+static int              argv_a2arg(int, const char *);
+static int              args(struct mdoc *, int, int *, 
+                               char *, int, char **);
+static int              argv(struct mdoc *, int,

                                struct mdoc_arg *, int *, char *);
                                struct mdoc_arg *, int *, char *);

-static int              parse_single(struct mdoc *, int, 
+static int              argv_single(struct mdoc *, int, 

                                struct mdoc_arg *, int *, char *);
                                struct mdoc_arg *, int *, char *);

-static int              parse_multi(struct mdoc *, int, 
+static int              argv_multi(struct mdoc *, int, 

                                struct mdoc_arg *, int *, char *);
                                struct mdoc_arg *, int *, char *);

-static int              postparse(struct mdoc *, int, 
-                               const struct mdoc_arg *, int);
+static int              pwarn(struct mdoc *, int, int, int);
+static int              perr(struct mdoc *, int, int, int);

 
 

-#define        ARGS_QUOTED     (1 << 0)
-#define        ARGS_DELIM      (1 << 1)
-#define        ARGS_TABSEP     (1 << 2)
+/* Warning messages. */
+
+#define        WQUOTPARM       (0)
+#define        WARGVPARM       (1)
+#define        WCOLEMPTY       (2)
+#define        WTAILWS         (3)
+
+/* Error messages. */
+
+#define        EQUOTTERM       (0)
+#define        EARGVAL         (1)
+#define        EARGMANY        (2)

 
 static int mdoc_argflags[MDOC_MAX] = {
        0, /* \" */
 
 static int mdoc_argflags[MDOC_MAX] = {
        0, /* \" */


@@ -62,7 +74,7 @@ static        int mdoc_argflags[MDOC_MAX] = {
        0, /* Ed */
        0, /* Bl */
        0, /* El */
        0, /* Ed */
        0, /* Bl */
        0, /* El */

-       ARGS_DELIM, /* It */
+       0, /* It */

        ARGS_DELIM, /* Ad */ 
        ARGS_DELIM, /* An */
        ARGS_DELIM, /* Ar */
        ARGS_DELIM, /* Ad */ 
        ARGS_DELIM, /* An */
        ARGS_DELIM, /* Ar */


@@ -158,39 +170,131 @@ static   int mdoc_argflags[MDOC_MAX] = {
 };
 
 
 };
 
 

+static int
+perr(struct mdoc *mdoc, int line, int pos, int code)
+{
+       int              c;
+
+       switch (code) {
+       case (EQUOTTERM):
+               c = mdoc_perr(mdoc, line, pos, 
+                               "unterminated quoted parameter");
+               break;
+       case (EARGVAL):
+               c = mdoc_perr(mdoc, line, pos, 
+                               "argument requires a value");
+               break;
+       case (EARGMANY):
+               c = mdoc_perr(mdoc, line, pos, 
+                               "too many values for argument");
+               break;
+       default:
+               abort();
+               /* NOTREACHED */
+       }
+       return(c);
+}
+
+
+static int
+pwarn(struct mdoc *mdoc, int line, int pos, int code)
+{
+       int              c;
+
+       switch (code) {
+       case (WQUOTPARM):
+               c = mdoc_pwarn(mdoc, line, pos, WARN_SYNTAX, 
+                               "unexpected quoted parameter");
+               break;
+       case (WARGVPARM):
+               c = mdoc_pwarn(mdoc, line, pos, WARN_SYNTAX, 
+                               "argument-like parameter");
+               break;
+       case (WCOLEMPTY):
+               c = mdoc_pwarn(mdoc, line, pos, WARN_SYNTAX, 
+                               "last list column is empty");
+               break;
+       case (WTAILWS):
+               c = mdoc_pwarn(mdoc, line, pos, WARN_COMPAT, 
+                               "trailing whitespace");
+               break;
+       default:
+               abort();
+               /* NOTREACHED */
+       }
+       return(c);
+}
+
+

 int
 mdoc_args(struct mdoc *mdoc, int line, 
                int *pos, char *buf, int tok, char **v)
 {
 int
 mdoc_args(struct mdoc *mdoc, int line, 
                int *pos, char *buf, int tok, char **v)
 {

-       int               i, fl;
+       int               fl, c, i;

        struct mdoc_node *n;
 
        struct mdoc_node *n;
 

-       fl = 0 == tok ? 0 : mdoc_argflags[tok];
+       fl = (0 == tok) ? 0 : mdoc_argflags[tok];
+
+       /* 
+        * First see if we should use TABSEP (Bl -column).  This
+        * invalidates the use of ARGS_DELIM.
+        */
+

        if (MDOC_It == tok) {
        if (MDOC_It == tok) {

-               n = mdoc->last->parent;
-               /* FIXME: scan for ARGS_TABSEP. */
+               for (n = mdoc->last; n; n = n->parent)
+                       if (MDOC_BLOCK == n->type)
+                               if (MDOC_Bl == n->tok)
+                                       break;
+               assert(n);
+               c = (int)n->data.block.argc;
+               assert(c > 0);
+
+               /* LINTED */
+               for (i = 0; i < c; i++) {
+                       if (MDOC_Column != n->data.block.argv[i].arg)
+                               continue;
+                       fl |= ARGS_TABSEP;
+                       fl &= ~ARGS_DELIM;
+                       break;
+               }

        }
 
        }
 

+       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_pwarn(mdoc, line, *pos, WARN_SYNTAX, "unexpected quoted parameter"))
+               if ( ! pwarn(mdoc, line, *pos, WQUOTPARM))

                        return(ARGS_ERROR);
 
        if ('-' == buf[*pos]) 
                        return(ARGS_ERROR);
 
        if ('-' == buf[*pos]) 

-               if ( ! mdoc_pwarn(mdoc, line, *pos, WARN_SYNTAX, "argument-like parameter"))
+               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])) {
        if ((fl & ARGS_DELIM) && mdoc_iscdelim(buf[*pos])) {

-               /* 
-                * If ARGS_DELIM, return ARGS_PUNCT if only space-separated
-                * punctuation remains.  
-                */

                for (i = *pos; buf[i]; ) {
                        if ( ! mdoc_iscdelim(buf[i]))
                                break;
                        i++;
                for (i = *pos; buf[i]; ) {
                        if ( ! mdoc_iscdelim(buf[i]))
                                break;
                        i++;

+                       /* There must be at least one space... */

                        if (0 == buf[i] || ! isspace((int)buf[i]))
                                break;
                        i++;
                        if (0 == buf[i] || ! isspace((int)buf[i]))
                                break;
                        i++;


@@ -203,37 +307,110 @@ mdoc_args(struct mdoc *mdoc, int line,
                }
        }
 
                }
        }
 

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

 
 

-       assert(*pos > 0);

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

-               /* FIXME: UGLY tab-sep processing. */
+               /* 
+                * 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. */

 
 

-               if (ARGS_TABSEP & fl)
-                       while (buf[*pos]) {
-                               if ('\t' == buf[*pos])
+                       for (p = *v; ; p++) {
+                               if (NULL == (p = strchr(p, '\t')))
+                                       break;
+                               if (p == *v)
+                                       break;
+                               if ('\\' != *(p - 1))

                                        break;
                                        break;

-                               if ('T' == buf[*pos]) {
-                                       (*pos)++;
-                                       if (0 == buf[*pos])
-                                               break;
-                                       if ('a' == buf[*pos]) {
-                                               buf[*pos - 1] = 0;
-                                               break;
-                                       }
-                               }
-                               (*pos)++;

                        }
                        }

-               else {
+
+                       /* 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((int)buf[*pos]))
                                        if ('\\' != buf[*pos - 1])
                                                break;
                                (*pos)++;
                        }
                        while (buf[*pos]) {
                                if (isspace((int)buf[*pos]))
                                        if ('\\' != buf[*pos - 1])
                                                break;
                                (*pos)++;
                        }

-               }

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


@@ -250,7 +427,7 @@ mdoc_args(struct mdoc *mdoc, int line,
                if (buf[*pos])
                        return(ARGS_WORD);
 
                if (buf[*pos])
                        return(ARGS_WORD);
 

-               if ( ! mdoc_pwarn(mdoc, line, *pos, WARN_COMPAT, "whitespace at end-of-line"))
+               if ( ! pwarn(mdoc, line, *pos, WTAILWS))

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


@@ -268,7 +445,7 @@ mdoc_args(struct mdoc *mdoc, int line,
                (*pos)++;
 
        if (0 == buf[*pos]) {
                (*pos)++;
 
        if (0 == buf[*pos]) {

-               (void)mdoc_perr(mdoc, line, *pos, "unterminated quoted parameter");
+               (void)perr(mdoc, line, *pos, EQUOTTERM);

                return(ARGS_ERROR);
        }
 
                return(ARGS_ERROR);
        }
 


@@ -282,7 +459,7 @@ mdoc_args(struct mdoc *mdoc, int line,
        if (buf[*pos])
                return(ARGS_QWORD);
 
        if (buf[*pos])
                return(ARGS_QWORD);
 

-       if ( ! mdoc_pwarn(mdoc, line, *pos, WARN_COMPAT, "whitespace at end-of-line"))
+       if ( ! pwarn(mdoc, line, *pos, WTAILWS))

                return(ARGS_ERROR);
 
        return(ARGS_QWORD);
                return(ARGS_ERROR);
 
        return(ARGS_QWORD);


@@ -290,7 +467,7 @@ mdoc_args(struct mdoc *mdoc, int line,
 
 
 static int
 
 
 static int

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

 {
 
        switch (tok) {
 {
 
        switch (tok) {


@@ -454,34 +631,7 @@ lookup(int tok, const char *argv)
 
 
 static int
 
 
 static int

-postparse(struct mdoc *mdoc, int line, 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_perr(mdoc, line, pos, "invalid offset value"));
-       default:
-               break;
-       }
-
-       return(1);
-}
-
-
-static int
-parse_multi(struct mdoc *mdoc, int line, 
+argv_multi(struct mdoc *mdoc, int line, 

                struct mdoc_arg *v, int *pos, char *buf)
 {
        int              c, ppos;
                struct mdoc_arg *v, int *pos, char *buf)
 {
        int              c, ppos;


@@ -495,27 +645,28 @@ parse_multi(struct mdoc *mdoc, int line,
        for (v->sz = 0; v->sz < MDOC_LINEARG_MAX; v->sz++) {
                if ('-' == buf[*pos])
                        break;
        for (v->sz = 0; v->sz < MDOC_LINEARG_MAX; v->sz++) {
                if ('-' == buf[*pos])
                        break;

-               c = mdoc_args(mdoc, line, pos, buf, ARGS_QUOTED, &p);
+               c = args(mdoc, line, pos, buf, ARGS_QUOTED, &p);

                if (ARGS_ERROR == c) {
                        free(v->value);
                        return(0);
                } else if (ARGS_EOLN == c)
                        break;
                if (ARGS_ERROR == c) {
                        free(v->value);
                        return(0);
                } else if (ARGS_EOLN == c)
                        break;

-               v->value[v->sz] = p;
+               v->value[(int)v->sz] = p;

        }
 
        if (0 < v->sz && v->sz < MDOC_LINEARG_MAX)
                return(1);
 
        free(v->value);
        }
 
        if (0 < v->sz && v->sz < MDOC_LINEARG_MAX)
                return(1);
 
        free(v->value);

-       return(mdoc_perr(mdoc, line, ppos, 0 == v->sz ?
-                               "argument requires a value" :
-                               "too many values to argument"));
+       if (0 == v->sz) 
+               return(perr(mdoc, line, ppos, EARGVAL));
+
+       return(perr(mdoc, line, ppos, EARGMANY));

 }
 
 
 static int
 }
 
 
 static int

-parse_single(struct mdoc *mdoc, int line, 
+argv_single(struct mdoc *mdoc, int line, 

                struct mdoc_arg *v, int *pos, char *buf)
 {
        int              c, ppos;
                struct mdoc_arg *v, int *pos, char *buf)
 {
        int              c, ppos;


@@ -523,11 +674,11 @@ parse_single(struct mdoc *mdoc, int line,
 
        ppos = *pos;
 
 
        ppos = *pos;
 

-       c = mdoc_args(mdoc, line, pos, buf, ARGS_QUOTED, &p);
+       c = args(mdoc, line, pos, buf, ARGS_QUOTED, &p);

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

-               return(mdoc_perr(mdoc, line, ppos,  "argument requires a value"));
+               return(perr(mdoc, line, ppos,  EARGVAL));

 
        v->sz = 1;
        v->value = xcalloc(1, sizeof(char *));
 
        v->sz = 1;
        v->value = xcalloc(1, sizeof(char *));


@@ -537,7 +688,7 @@ parse_single(struct mdoc *mdoc, int line,
 
 
 static int
 
 
 static int

-parse(struct mdoc *mdoc, int line, 
+argv(struct mdoc *mdoc, int line, 

                struct mdoc_arg *v, int *pos, char *buf)
 {
 
                struct mdoc_arg *v, int *pos, char *buf)
 {
 


@@ -550,9 +701,9 @@ parse(struct mdoc *mdoc, int line,
        case(MDOC_Width):
                /* FALLTHROUGH */
        case(MDOC_Offset):
        case(MDOC_Width):
                /* FALLTHROUGH */
        case(MDOC_Offset):

-               return(parse_single(mdoc, line, v, pos, buf));
+               return(argv_single(mdoc, line, v, pos, buf));

        case(MDOC_Column):
        case(MDOC_Column):

-               return(parse_multi(mdoc, line, v, pos, buf));
+               return(argv_multi(mdoc, line, v, pos, buf));

        default:
                break;
        }
        default:
                break;
        }


@@ -565,8 +716,8 @@ int
 mdoc_argv(struct mdoc *mdoc, int line, int tok,
                struct mdoc_arg *v, int *pos, char *buf)
 {
 mdoc_argv(struct mdoc *mdoc, int line, int tok,
                struct mdoc_arg *v, int *pos, char *buf)
 {

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

 
        (void)memset(v, 0, sizeof(struct mdoc_arg));
 
 
        (void)memset(v, 0, sizeof(struct mdoc_arg));
 


@@ -579,12 +730,14 @@ mdoc_argv(struct mdoc *mdoc, int line, int tok,
                return(ARGV_WORD);
 
        i = *pos;
                return(ARGV_WORD);
 
        i = *pos;

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

 
        v->line = line;
        v->pos = *pos;
 
        assert(*pos > 0);
 
        v->line = line;
        v->pos = *pos;
 
        assert(*pos > 0);

+
+       /* LINTED */

        while (buf[*pos]) {
                if (isspace((int)buf[*pos])) 
                        if ('\\' != buf[*pos - 1])
        while (buf[*pos]) {
                if (isspace((int)buf[*pos])) 
                        if ('\\' != buf[*pos - 1])


@@ -595,8 +748,8 @@ mdoc_argv(struct mdoc *mdoc, int line, int tok,
        if (buf[*pos])
                buf[(*pos)++] = 0;
 
        if (buf[*pos])
                buf[(*pos)++] = 0;
 

-       if (MDOC_ARG_MAX == (v->arg = lookup(tok, argv))) {
-               if ( ! mdoc_pwarn(mdoc, line, i, WARN_SYNTAX, "argument-like parameter"))
+       if (MDOC_ARG_MAX == (v->arg = argv_a2arg(tok, p))) {
+               if ( ! pwarn(mdoc, line, i, WARGVPARM))

                        return(ARGV_ERROR);
                return(ARGV_WORD);
        }
                        return(ARGV_ERROR);
                return(ARGV_WORD);
        }


@@ -606,10 +759,7 @@ mdoc_argv(struct mdoc *mdoc, int line, int tok,
 
        /* FIXME: whitespace if no value. */
 
 
        /* FIXME: whitespace if no value. */
 

-       ppos = *pos;
-       if ( ! parse(mdoc, line, v, pos, buf))
-               return(ARGV_ERROR);
-       if ( ! postparse(mdoc, line, v, ppos))
+       if ( ! argv(mdoc, line, v, pos, buf))

                return(ARGV_ERROR);
 
        return(ARGV_ARG);
                return(ARGV_ERROR);
 
        return(ARGV_ARG);