Note version 1.11.4 bits. Here we go!

[mandoc.git] / roff.c

diff --git a/roff.c b/roff.c
index 548e2f94cbbd95d9c18afa831c68ec7dd8fa0f2b..79d0897ef3a79b22efa07b5c2867d0875d8ee43c 100644 (file)
--- a/roff.c
+++ b/roff.c
@@ -1,4 +1,4 @@
-/*     $Id: roff.c,v 1.135 2011/04/17 09:08:19 kristaps Exp $ */
+/*     $Id: roff.c,v 1.144 2011/07/08 09:28:33 kristaps Exp $ */

 /*
  * Copyright (c) 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
  * Copyright (c) 2010, 2011 Ingo Schwarze <schwarze@openbsd.org>
 /*
  * Copyright (c) 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
  * Copyright (c) 2010, 2011 Ingo Schwarze <schwarze@openbsd.org>


@@ -20,17 +20,15 @@
 #endif
 
 #include <assert.h>
 #endif
 
 #include <assert.h>

-#include <errno.h>

 #include <ctype.h>
 #include <ctype.h>

-#include <limits.h>

 #include <stdlib.h>
 #include <string.h>
 #include <stdlib.h>
 #include <string.h>

-#include <stdio.h>

 
 #include "mandoc.h"
 #include "libroff.h"
 #include "libmandoc.h"
 
 
 #include "mandoc.h"
 #include "libroff.h"
 #include "libmandoc.h"
 

+/* Maximum number of nested if-else conditionals. */

 #define        RSTACK_MAX      128
 
 enum   rofft {
 #define        RSTACK_MAX      128
 
 enum   rofft {


@@ -63,7 +61,7 @@ enum  rofft {
        ROFF_EQ,
        ROFF_EN,
        ROFF_cblock,
        ROFF_EQ,
        ROFF_EN,
        ROFF_cblock,

-       ROFF_ccond, /* FIXME: remove this. */
+       ROFF_ccond,

        ROFF_USERDEF,
        ROFF_MAX
 };
        ROFF_USERDEF,
        ROFF_MAX
 };


@@ -127,6 +125,14 @@ struct     roffmac {
        struct roffmac  *next;
 };
 
        struct roffmac  *next;
 };
 

+struct predef {
+       const char      *name; /* predefined input name */
+       const char      *str; /* replacement symbol */
+};
+
+#define        PREDEF(__name, __str) \
+       { (__name), (__str) },
+

 static enum rofferr     roff_block(ROFF_ARGS);
 static enum rofferr     roff_block_text(ROFF_ARGS);
 static enum rofferr     roff_block_sub(ROFF_ARGS);
 static enum rofferr     roff_block(ROFF_ARGS);
 static enum rofferr     roff_block_text(ROFF_ARGS);
 static enum rofferr     roff_block_sub(ROFF_ARGS);


@@ -144,7 +150,7 @@ static      const char      *roff_getstrn(const struct roff *,
 static enum rofferr     roff_line_ignore(ROFF_ARGS);
 static enum rofferr     roff_nr(ROFF_ARGS);
 static int              roff_res(struct roff *, 
 static enum rofferr     roff_line_ignore(ROFF_ARGS);
 static enum rofferr     roff_nr(ROFF_ARGS);
 static int              roff_res(struct roff *, 

-                               char **, size_t *, int);
+                               char **, size_t *, int, int);

 static enum rofferr     roff_rm(ROFF_ARGS);
 static void             roff_setstr(struct roff *,
                                const char *, const char *, int);
 static enum rofferr     roff_rm(ROFF_ARGS);
 static void             roff_setstr(struct roff *,
                                const char *, const char *, int);


@@ -198,6 +204,12 @@ static     struct roffmac   roffs[ROFF_MAX] = {
        { NULL, roff_userdef, NULL, NULL, 0, NULL },
 };
 
        { NULL, roff_userdef, NULL, NULL, 0, NULL },
 };
 

+/* Array of injected predefined strings. */
+#define        PREDEFS_MAX      38
+static const struct predef predefs[PREDEFS_MAX] = {
+#include "predefs.in"
+};
+

 static void             roff_free1(struct roff *);
 static enum rofft       roff_hash_find(const char *, size_t);
 static void             roff_hash_init(void);
 static void             roff_free1(struct roff *);
 static enum rofft       roff_hash_find(const char *, size_t);
 static void             roff_hash_init(void);


@@ -206,7 +218,6 @@ static      void             roffnode_push(struct roff *, enum rofft,
                                const char *, int, int);
 static void             roffnode_pop(struct roff *);
 static enum rofft       roff_parse(struct roff *, const char *, int *);
                                const char *, int, int);
 static void             roffnode_pop(struct roff *);
 static enum rofft       roff_parse(struct roff *, const char *, int *);

-static int              roff_parse_nat(const char *, unsigned int *);

 
 /* See roff_hash_find() */
 #define        ROFF_HASH(p)    (p[0] - ASCII_LO)
 
 /* See roff_hash_find() */
 #define        ROFF_HASH(p)    (p[0] - ASCII_LO)


@@ -232,7 +243,6 @@ roff_hash_init(void)
        }
 }
 
        }
 }
 

-

 /*
  * Look up a roff token by its name.  Returns ROFF_MAX if no macro by
  * the nil-terminated string name could be found.
 /*
  * Look up a roff token by its name.  Returns ROFF_MAX if no macro by
  * the nil-terminated string name could be found.


@@ -337,8 +347,12 @@ roff_free1(struct roff *r)
 void
 roff_reset(struct roff *r)
 {
 void
 roff_reset(struct roff *r)
 {

+       int              i;

 
        roff_free1(r);
 
        roff_free1(r);

+
+       for (i = 0; i < PREDEFS_MAX; i++) 
+               roff_setstr(r, predefs[i].name, predefs[i].str, 0);

 }
 
 
 }
 
 


@@ -355,6 +369,7 @@ struct roff *
 roff_alloc(struct regset *regs, struct mparse *parse)
 {
        struct roff     *r;
 roff_alloc(struct regset *regs, struct mparse *parse)
 {
        struct roff     *r;

+       int              i;

 
        r = mandoc_calloc(1, sizeof(struct roff));
        r->regs = regs;
 
        r = mandoc_calloc(1, sizeof(struct roff));
        r->regs = regs;


@@ -362,6 +377,10 @@ roff_alloc(struct regset *regs, struct mparse *parse)
        r->rstackpos = -1;
        
        roff_hash_init();
        r->rstackpos = -1;
        
        roff_hash_init();

+
+       for (i = 0; i < PREDEFS_MAX; i++) 
+               roff_setstr(r, predefs[i].name, predefs[i].str, 0);
+

        return(r);
 }
 
        return(r);
 }
 


@@ -372,7 +391,7 @@ roff_alloc(struct regset *regs, struct mparse *parse)
  * is processed. 
  */
 static int
  * is processed. 
  */
 static int

-roff_res(struct roff *r, char **bufp, size_t *szp, int pos)
+roff_res(struct roff *r, char **bufp, size_t *szp, int ln, int pos)

 {
        const char      *stesc; /* start of an escape sequence ('\\') */
        const char      *stnam; /* start of the name, after "[(*" */
 {
        const char      *stesc; /* start of an escape sequence ('\\') */
        const char      *stnam; /* start of the name, after "[(*" */


@@ -439,8 +458,9 @@ roff_res(struct roff *r, char **bufp, size_t *szp, int pos)
                res = roff_getstrn(r, stnam, (size_t)i);
 
                if (NULL == res) {
                res = roff_getstrn(r, stnam, (size_t)i);
 
                if (NULL == res) {

-                       cp -= maxl ? 1 : 0;
-                       continue;
+                       /* TODO: keep track of the correct position. */
+                       mandoc_msg(MANDOCERR_BADESCAPE, r->parse, ln, pos, NULL);
+                       res = "";

                }
 
                /* Replace the escape sequence by the string. */
                }
 
                /* Replace the escape sequence by the string. */


@@ -476,7 +496,7 @@ roff_parseln(struct roff *r, int ln, char **bufp,
         * words to fill in.
         */
 
         * words to fill in.
         */
 

-       if (r->first_string && ! roff_res(r, bufp, szp, pos))
+       if (r->first_string && ! roff_res(r, bufp, szp, ln, pos))

                return(ROFF_REPARSE);
 
        ppos = pos;
                return(ROFF_REPARSE);
 
        ppos = pos;


@@ -576,11 +596,18 @@ roff_parse(struct roff *r, const char *buf, int *pos)
        size_t           maclen;
        enum rofft       t;
 
        size_t           maclen;
        enum rofft       t;
 

-       if ('\0' == buf[*pos] || '"' == buf[*pos])
+       if ('\0' == buf[*pos] || '"' == buf[*pos] || 
+                       '\t' == buf[*pos] || ' ' == buf[*pos])

                return(ROFF_MAX);
 
                return(ROFF_MAX);
 

+       /*
+        * We stop the macro parse at an escape, tab, space, or nil.
+        * However, `\}' is also a valid macro, so make sure we don't
+        * clobber it by seeing the `\' as the end of token.
+        */
+

        mac = buf + *pos;
        mac = buf + *pos;

-       maclen = strcspn(mac, " \\\t\0");
+       maclen = strcspn(mac + 1, " \\\t\0") + 1;

 
        t = (r->current_string = roff_getstrn(r, mac, maclen))
            ? ROFF_USERDEF : roff_hash_find(mac, maclen);
 
        t = (r->current_string = roff_getstrn(r, mac, maclen))
            ? ROFF_USERDEF : roff_hash_find(mac, maclen);


@@ -593,27 +620,6 @@ roff_parse(struct roff *r, const char *buf, int *pos)
        return(t);
 }
 
        return(t);
 }
 

-
-static int
-roff_parse_nat(const char *buf, unsigned int *res)
-{
-       char            *ep;
-       long             lval;
-
-       errno = 0;
-       lval = strtol(buf, &ep, 10);
-       if (buf[0] == '\0' || *ep != '\0')
-               return(0);
-       if ((errno == ERANGE && 
-                       (lval == LONG_MAX || lval == LONG_MIN)) ||
-                       (lval > INT_MAX || lval < 0))
-               return(0);
-
-       *res = (unsigned int)lval;
-       return(1);
-}
-
-

 /* ARGSUSED */
 static enum rofferr
 roff_cblock(ROFF_ARGS)
 /* ARGSUSED */
 static enum rofferr
 roff_cblock(ROFF_ARGS)


@@ -829,8 +835,7 @@ roff_block_sub(ROFF_ARGS)
         * pulling it out of the hashtable.
         */
 
         * pulling it out of the hashtable.
         */
 

-       if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos)))
-               return(ROFF_IGN);
+       t = roff_parse(r, *bufp, &pos);

 
        /*
         * Macros other than block-end are only significant
 
        /*
         * Macros other than block-end are only significant


@@ -866,21 +871,44 @@ roff_cond_sub(ROFF_ARGS)
 {
        enum rofft       t;
        enum roffrule    rr;
 {
        enum rofft       t;
        enum roffrule    rr;

+       char            *ep;

 
        rr = r->last->rule;
 
        rr = r->last->rule;

+       roffnode_cleanscope(r);

 
 

-       /* 
-        * Clean out scope.  If we've closed ourselves, then don't
-        * continue. 
+       /*
+        * If the macro is unknown, first check if it contains a closing
+        * delimiter `\}'.  If it does, close out our scope and return
+        * the currently-scoped rule (ignore or continue).  Else, drop
+        * into the currently-scoped rule.

         */
 
         */
 

-       roffnode_cleanscope(r);
-

        if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos))) {
        if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos))) {

-               if ('\\' == (*bufp)[pos] && '}' == (*bufp)[pos + 1])
-                       return(roff_ccond
-                               (r, ROFF_ccond, bufp, szp,
-                                ln, pos, pos + 2, offs));
+               ep = &(*bufp)[pos];
+               for ( ; NULL != (ep = strchr(ep, '\\')); ep++) {
+                       ep++;
+                       if ('}' != *ep)
+                               continue;
+
+                       /*
+                        * Make the \} go away.
+                        * This is a little haphazard, as it's not quite
+                        * clear how nroff does this.
+                        * If we're at the end of line, then just chop
+                        * off the \} and resize the buffer.
+                        * If we aren't, then conver it to spaces.
+                        */
+
+                       if ('\0' == *(ep + 1)) {
+                               *--ep = '\0';
+                               *szp -= 2;
+                       } else
+                               *(ep - 1) = *ep = ' ';
+
+                       roff_ccond(r, ROFF_ccond, bufp, szp, 
+                                       ln, pos, pos + 2, offs);
+                       break;
+               }

                return(ROFFRULE_DENY == rr ? ROFF_IGN : ROFF_CONT);
        }
 
                return(ROFFRULE_DENY == rr ? ROFF_IGN : ROFF_CONT);
        }
 


@@ -889,6 +917,7 @@ roff_cond_sub(ROFF_ARGS)
         * if they're either structurally required (such as loops and
         * conditionals) or a closing macro.
         */
         * if they're either structurally required (such as loops and
         * conditionals) or a closing macro.
         */

+

        if (ROFFRULE_DENY == rr)
                if ( ! (ROFFMAC_STRUCT & roffs[t].flags))
                        if (ROFF_ccond != t)
        if (ROFFRULE_DENY == rr)
                if ( ! (ROFFMAC_STRUCT & roffs[t].flags))
                        if (ROFF_ccond != t)


@@ -899,37 +928,28 @@ roff_cond_sub(ROFF_ARGS)
                                ln, ppos, pos, offs));
 }
 
                                ln, ppos, pos, offs));
 }
 

-

 /* ARGSUSED */
 static enum rofferr
 roff_cond_text(ROFF_ARGS)
 {
 /* ARGSUSED */
 static enum rofferr
 roff_cond_text(ROFF_ARGS)
 {

-       char            *ep, *st;
+       char            *ep;

        enum roffrule    rr;
 
        rr = r->last->rule;
        enum roffrule    rr;
 
        rr = r->last->rule;

+       roffnode_cleanscope(r);

 
 

-       /*
-        * We display the value of the text if out current evaluation
-        * scope permits us to do so.
-        */
-
-       /* FIXME: use roff_ccond? */
-
-       st = &(*bufp)[pos];
-       if (NULL == (ep = strstr(st, "\\}"))) {
-               roffnode_cleanscope(r);
-               return(ROFFRULE_DENY == rr ? ROFF_IGN : ROFF_CONT);
+       ep = &(*bufp)[pos];
+       for ( ; NULL != (ep = strchr(ep, '\\')); ep++) {
+               ep++;
+               if ('}' != *ep)
+                       continue;
+               *ep = '&';
+               roff_ccond(r, ROFF_ccond, bufp, szp, 
+                               ln, pos, pos + 2, offs);

        }
        }

-
-       if (ep == st || (ep > st && '\\' != *(ep - 1)))
-               roffnode_pop(r);
-
-       roffnode_cleanscope(r);

        return(ROFFRULE_DENY == rr ? ROFF_IGN : ROFF_CONT);
 }
 
        return(ROFFRULE_DENY == rr ? ROFF_IGN : ROFF_CONT);
 }
 

-

 static enum roffrule
 roff_evalcond(const char *v, int *pos)
 {
 static enum roffrule
 roff_evalcond(const char *v, int *pos)
 {


@@ -1091,6 +1111,7 @@ roff_nr(ROFF_ARGS)
 {
        const char      *key;
        char            *val;
 {
        const char      *key;
        char            *val;

+       int              iv;

        struct reg      *rg;
 
        val = *bufp + pos;
        struct reg      *rg;
 
        val = *bufp + pos;


@@ -1099,8 +1120,10 @@ roff_nr(ROFF_ARGS)
 
        if (0 == strcmp(key, "nS")) {
                rg[(int)REG_nS].set = 1;
 
        if (0 == strcmp(key, "nS")) {
                rg[(int)REG_nS].set = 1;

-               if ( ! roff_parse_nat(val, &rg[(int)REG_nS].v.u))
-                       rg[(int)REG_nS].v.u = 0;
+               if ((iv = mandoc_strntou(val, strlen(val), 10)) >= 0)
+                       rg[REG_nS].v.u = (unsigned)iv;
+               else
+                       rg[(int)REG_nS].v.u = 0u;

        }
 
        return(ROFF_IGN);
        }
 
        return(ROFF_IGN);


@@ -1238,7 +1261,7 @@ roff_userdef(ROFF_ARGS)
        cp = *bufp + pos;
        for (i = 0; i < 9; i++)
                arg[i] = '\0' == *cp ? "" :
        cp = *bufp + pos;
        for (i = 0; i < 9; i++)
                arg[i] = '\0' == *cp ? "" :

-                   mandoc_getarg(r->parse, &cp, ln, 1, &pos);
+                   mandoc_getarg(r->parse, &cp, ln, &pos);

 
        /*
         * Expand macro arguments.
 
        /*
         * Expand macro arguments.