Add a test case for nesting of .RS/.RE

[mandoc.git] / roff.c

diff --git a/roff.c b/roff.c
index d35c9a7efea14cf72db40339616b2cb32e5c0ce5..a6544c9fdadb2b69cf4a0dd9b13ef4d4c0e46404 100644 (file)
--- a/roff.c
+++ b/roff.c
@@ -1,4 +1,4 @@
-/*     $Id: roff.c,v 1.71 2010/05/15 21:53:11 kristaps Exp $ */
+/*     $Id: roff.c,v 1.78 2010/05/16 22:28:33 kristaps Exp $ */

 /*
  * Copyright (c) 2010 Kristaps Dzonsons <kristaps@bsd.lv>
  *
 /*
  * Copyright (c) 2010 Kristaps Dzonsons <kristaps@bsd.lv>
  *


@@ -21,17 +21,34 @@
 #include <assert.h>
 #include <stdlib.h>
 #include <string.h>
 #include <assert.h>
 #include <stdlib.h>
 #include <string.h>

+#include <stdio.h>

 
 #include "mandoc.h"
 #include "roff.h"
 
 
 #include "mandoc.h"
 #include "roff.h"
 

+#define        ROFF_CTL(c) \
+       ('.' == (c) || '\'' == (c))
+#if    0
+#define        ROFF_MDEBUG(p, str) \
+       fprintf(stderr, "%s: %s (%d:%d)\n", (str), \
+               roffs[(p)->last->tok].name, \
+               (p)->last->line, (p)->last->col)
+#else
+#define        ROFF_MDEBUG(p, str) while (/* CONSTCOND */ 0)
+#endif
+

 enum   rofft {
 enum   rofft {

-       ROFF_de,
-       ROFF_dei,
+       ROFF_if,
+       ROFF_ig,
+       ROFF_cblock,
+       ROFF_ccond,
+#if 0

        ROFF_am,
        ROFF_ami,
        ROFF_am,
        ROFF_ami,

-       ROFF_ig,
+       ROFF_de,
+       ROFF_dei,

        ROFF_close,
        ROFF_close,

+#endif

        ROFF_MAX
 };
 
        ROFF_MAX
 };
 


@@ -44,41 +61,46 @@ struct      roff {
 struct roffnode {
        enum rofft       tok; /* type of node */
        struct roffnode *parent; /* up one in stack */
 struct roffnode {
        enum rofft       tok; /* type of node */
        struct roffnode *parent; /* up one in stack */

-       char            *end; /* custom end-token */
+       char            *end; /* end-token: custom */

        int              line; /* parse line */
        int              col; /* parse col */
        int              line; /* parse line */
        int              col; /* parse col */

+       int              endspan;

 };
 
 #define        ROFF_ARGS        struct roff *r, /* parse ctx */ \
 };
 
 #define        ROFF_ARGS        struct roff *r, /* parse ctx */ \

+                        enum rofft tok, /* tok of macro */ \

                         char **bufp, /* input buffer */ \
                         size_t *szp, /* size of input buffer */ \
                         int ln, /* parse line */ \
                         char **bufp, /* input buffer */ \
                         size_t *szp, /* size of input buffer */ \
                         int ln, /* parse line */ \

-                        int ppos /* current pos in buffer */
+                        int ppos, /* original pos in buffer */ \
+                        int pos, /* current pos in buffer */ \
+                        int *offs /* reset offset of buffer data */

 
 typedef        enum rofferr (*roffproc)(ROFF_ARGS);
 
 struct roffmac {
        const char      *name; /* macro name */
 
 typedef        enum rofferr (*roffproc)(ROFF_ARGS);
 
 struct roffmac {
        const char      *name; /* macro name */

-       roffproc         sub; /* child of control black */
-       roffproc         new; /* root of stack (type = ROFF_MAX) */
+       roffproc         proc;
+       roffproc         text;

 };
 
 };
 

-static enum rofferr     roff_ignore(ROFF_ARGS);
-static enum rofferr     roff_new_close(ROFF_ARGS);
-static enum rofferr     roff_new_ig(ROFF_ARGS);
-static enum rofferr     roff_sub_ig(ROFF_ARGS);
+static enum rofferr     roff_if(ROFF_ARGS);
+static enum rofferr     roff_if_text(ROFF_ARGS);
+static enum rofferr     roff_ig(ROFF_ARGS);
+static enum rofferr     roff_ig_text(ROFF_ARGS);
+static enum rofferr     roff_cblock(ROFF_ARGS);
+static enum rofferr     roff_ccond(ROFF_ARGS);

 
 const  struct roffmac   roffs[ROFF_MAX] = {
 
 const  struct roffmac   roffs[ROFF_MAX] = {

-       { "de", NULL, roff_ignore },
-       { "dei", NULL, roff_ignore },
-       { "am", NULL, roff_ignore },
-       { "ami", NULL, roff_ignore },
-       { "ig", roff_sub_ig, roff_new_ig },
-       { ".", NULL, roff_new_close },
+       { "if", roff_if, roff_if_text },
+       { "ig", roff_ig, roff_ig_text },
+       { ".", roff_cblock, NULL },
+       { "\\}", roff_ccond, NULL },

 };
 
 static void             roff_free1(struct roff *);
 static enum rofft       roff_hash_find(const char *);
 };
 
 static void             roff_free1(struct roff *);
 static enum rofft       roff_hash_find(const char *);

+static void             roffnode_cleanscope(struct roff *);

 static int              roffnode_push(struct roff *, 
                                enum rofft, int, int);
 static void             roffnode_pop(struct roff *);
 static int              roffnode_push(struct roff *, 
                                enum rofft, int, int);
 static void             roffnode_pop(struct roff *);


@@ -113,9 +135,11 @@ roffnode_pop(struct roff *r)
 {
        struct roffnode *p;
 
 {
        struct roffnode *p;
 

-       if (NULL == (p = r->last))
-               return;
-       r->last = p->parent;
+       assert(r->last);
+       p = r->last; 
+       r->last = r->last->parent;
+       if (p->end)
+               free(p->end);

        free(p);
 }
 
        free(p);
 }
 


@@ -187,30 +211,87 @@ roff_alloc(const mandocmsg msg, void *data)
 
 
 enum rofferr
 
 
 enum rofferr

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

 {
        enum rofft       t;
 {
        enum rofft       t;

-       int              ppos;
+       int              ppos, i, j, wtf;

 
 

-       if (NULL != r->last) {
+       if (r->last && ! ROFF_CTL((*bufp)[pos])) {

                /*
                /*

-                * If there's a node on the stack, then jump directly
-                * into its processing function.
+                * If a scope is open and we're not a macro, pass it
+                * through our text detector and continue as quickly as
+                * possible.

                 */
                t = r->last->tok;
                 */
                t = r->last->tok;

-               assert(roffs[t].sub);
-               return((*roffs[t].sub)(r, bufp, szp, ln, 0));
-       } else if ('.' != (*bufp)[0] && NULL == r->last)
-               /* Return when in free text without a context. */
+               assert(roffs[t].text);
+               return((*roffs[t].text)
+                               (r, t, bufp, szp, ln, pos, pos, offs));
+       } else if ( ! ROFF_CTL((*bufp)[pos]))
+               /*
+                * Don't do anything if we're free-form text.
+                */

                return(ROFF_CONT);
 
                return(ROFF_CONT);
 

-       /* There's nothing on the stack: make us anew. */
+       /* A macro-ish line with a possibly-open macro context. */

 
 

-       if (ROFF_MAX == (t = roff_parse(*bufp, &ppos)))
-               return(ROFF_CONT);
+       wtf = 0;
+
+       if (r->last && r->last->end) {
+               /*
+                * We have a scope open that has a custom end-macro
+                * handler.  Try to match it against the input.
+                */
+               i = pos + 1;
+               while (' ' == (*bufp)[i] || '\t' == (*bufp)[i])
+                       i++;

 
 

-       assert(roffs[t].new);
-       return((*roffs[t].new)(r, bufp, szp, ln, ppos));
+               for (j = 0; r->last->end[j]; j++, i++)
+                       if ((*bufp)[i] != r->last->end[j])
+                               break;
+
+               if ('\0' == r->last->end[j] && 
+                               ('\0' == (*bufp)[i] ||
+                                ' ' == (*bufp)[i] ||
+                                '\t' == (*bufp)[i])) {
+                       roffnode_pop(r);
+                       roffnode_cleanscope(r);
+                       wtf = 1;
+               }
+       }
+
+       ppos = pos;
+       if (ROFF_MAX == (t = roff_parse(*bufp, &pos))) {
+               /* 
+                * This is some of groff's stranger behaviours.  If we
+                * encountered a custom end-scope tag and that tag also
+                * happens to be a "real" macro, then we need to try
+                * interpreting it again as a real macro.  If it's not,
+                * then return ignore.  Else continue.
+                */
+               if (wtf)
+                       return(ROFF_IGN);
+               else if (NULL == r->last)
+                       return(ROFF_CONT);
+
+               /* FIXME: this assumes that we ignore!? */
+               return(ROFF_IGN);
+       }
+
+       assert(roffs[t].proc);
+       return((*roffs[t].proc)
+                       (r, t, bufp, szp, ln, ppos, pos, offs));
+}
+
+
+int
+roff_endparse(struct roff *r)
+{
+
+       if (NULL == r->last)
+               return(1);
+       return((*r->msg)(MANDOCERR_SCOPEEXIT, r->data, r->last->line, 
+                               r->last->col, NULL));

 }
 
 
 }
 
 


@@ -225,8 +306,8 @@ roff_parse(const char *buf, int *pos)
        char             mac[5];
        enum rofft       t;
 
        char             mac[5];
        enum rofft       t;
 

-       assert('.' == buf[0]);
-       *pos = 1;
+       assert(ROFF_CTL(buf[*pos]));
+       (*pos)++;

 
        while (buf[*pos] && (' ' == buf[*pos] || '\t' == buf[*pos]))
                (*pos)++;
 
        while (buf[*pos] && (' ' == buf[*pos] || '\t' == buf[*pos]))
                (*pos)++;


@@ -257,62 +338,124 @@ roff_parse(const char *buf, int *pos)
 
 /* ARGSUSED */
 static enum rofferr
 
 /* ARGSUSED */
 static enum rofferr

-roff_ignore(ROFF_ARGS)
+roff_cblock(ROFF_ARGS)

 {
 
 {
 

+       if (NULL == r->last) {
+               if ( ! (*r->msg)(MANDOCERR_NOSCOPE, r->data, ln, ppos, NULL))
+                       return(ROFF_ERR);
+               return(ROFF_IGN);
+       }
+
+       if (ROFF_ig != r->last->tok) {
+               if ( ! (*r->msg)(MANDOCERR_NOSCOPE, r->data, ln, ppos, NULL))
+                       return(ROFF_ERR);
+               return(ROFF_IGN);
+       }
+
+       if ((*bufp)[pos])
+               if ( ! (*r->msg)(MANDOCERR_ARGSLOST, r->data, ln, pos, NULL))
+                       return(ROFF_ERR);
+
+       ROFF_MDEBUG(r, "closing ignore block");
+       roffnode_pop(r);
+       roffnode_cleanscope(r);

        return(ROFF_IGN);
        return(ROFF_IGN);

+
+}
+
+
+static void
+roffnode_cleanscope(struct roff *r)
+{
+
+       while (r->last) {
+               if (--r->last->endspan < 0)
+                       break;
+               ROFF_MDEBUG(r, "closing implicit scope");
+               roffnode_pop(r);
+       }

 }
 
 
 /* ARGSUSED */
 static enum rofferr
 }
 
 
 /* ARGSUSED */
 static enum rofferr

-roff_sub_ig(ROFF_ARGS)
+roff_ccond(ROFF_ARGS)

 {
 {

-       int              i, j;

 
 

-       /* Ignore free-text lines. */
+       if (NULL == r->last) {
+               if ( ! (*r->msg)(MANDOCERR_NOSCOPE, r->data, ln, ppos, NULL))
+                       return(ROFF_ERR);
+               return(ROFF_IGN);
+       }

 
 

-       if ('.' != (*bufp)[ppos])
+       if (ROFF_if != r->last->tok) {
+               if ( ! (*r->msg)(MANDOCERR_NOSCOPE, r->data, ln, ppos, NULL))
+                       return(ROFF_ERR);

                return(ROFF_IGN);
                return(ROFF_IGN);

+       }

 
 

-       if (r->last->end) {
-               /*
-                * Allow a macro to break us, if we've defined a special
-                * one for the case.  Old groff didn't allow spaces to
-                * buffer the macro, but new groff does.  Whee!
-                */
-               i = ppos + 1;
-               while ((*bufp)[i] && ' ' == (*bufp)[i])
-                       i++;
+       if (r->last->endspan > -1) {
+               if ( ! (*r->msg)(MANDOCERR_NOSCOPE, r->data, ln, ppos, NULL))
+                       return(ROFF_ERR);
+               return(ROFF_IGN);
+       }

 
 

-               if ('\0' == (*bufp)[i])
-                       return(ROFF_IGN);
+       if ((*bufp)[pos])
+               if ( ! (*r->msg)(MANDOCERR_ARGSLOST, r->data, ln, pos, NULL))
+                       return(ROFF_ERR);

 
 

-               for (j = 0; r->last->end[j]; i++, j++)
-                       if ((*bufp)[i] != r->last->end[j])
-                               return(ROFF_IGN);
+       ROFF_MDEBUG(r, "closing explicit scope");
+       roffnode_pop(r);
+       roffnode_cleanscope(r);
+       return(ROFF_IGN);
+}

 
 

-               if (r->last->end[j])
-                       return(ROFF_IGN);
-               if ((*bufp)[i] && ' ' != (*bufp)[i])
-                       return(ROFF_IGN);

 
 

-               while (' ' == (*bufp)[i])
-                       i++;
-       } else if (ROFF_close != roff_parse(*bufp, &i))
+/* ARGSUSED */
+static enum rofferr
+roff_ig(ROFF_ARGS)
+{
+       int             sv;
+       size_t          sz;
+
+       if ( ! roffnode_push(r, tok, ln, ppos))
+               return(ROFF_ERR);
+
+       if ('\0' == (*bufp)[pos]) {
+               ROFF_MDEBUG(r, "opening ignore block");

                return(ROFF_IGN);
                return(ROFF_IGN);

+       }
+
+       sv = pos;
+       while ((*bufp)[pos] && ' ' != (*bufp)[pos] && 
+                       '\t' != (*bufp)[pos])
+               pos++;

 
        /*
 
        /*

-        * Pop off the ignoring context and warn if we're going to lose
-        * any of our remaining arguments.
+        * Note: groff does NOT like escape characters in the input.
+        * Instead of detecting this, we're just going to let it fly and
+        * to hell with it.

         */
 
         */
 

-       roffnode_pop(r);
+       assert(pos > sv);
+       sz = (size_t)(pos - sv);

 
 

-       if ('\0' == (*bufp)[i])
-               return(ROFF_IGN);
-       if ( ! (*r->msg)(MANDOCERR_ARGSLOST, r->data, ln, i, NULL))
+       r->last->end = malloc(sz + 1);
+
+       if (NULL == r->last->end) {
+               (*r->msg)(MANDOCERR_MEM, r->data, ln, pos, NULL);

                return(ROFF_ERR);
                return(ROFF_ERR);

+       }
+
+       memcpy(r->last->end, *bufp + sv, sz);
+       r->last->end[(int)sz] = '\0';
+
+       ROFF_MDEBUG(r, "opening explicit ignore block");
+
+       if ((*bufp)[pos])
+               if ( ! (*r->msg)(MANDOCERR_ARGSLOST, r->data, ln, pos, NULL))
+                       return(ROFF_ERR);

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


@@ -320,66 +463,92 @@ roff_sub_ig(ROFF_ARGS)
 
 /* ARGSUSED */
 static enum rofferr
 
 /* ARGSUSED */
 static enum rofferr

-roff_new_close(ROFF_ARGS)
+roff_ig_text(ROFF_ARGS)

 {
 
 {
 

-       /*
-       if ( ! (*r->msg)(MANDOCERR_NOSCOPE, r->data, ln, ppos, NULL))
-               return(ROFF_ERR);
-       */

        return(ROFF_IGN);
 }
 
 
 /* ARGSUSED */
 static enum rofferr
        return(ROFF_IGN);
 }
 
 
 /* ARGSUSED */
 static enum rofferr

-roff_new_ig(ROFF_ARGS)
+roff_if_text(ROFF_ARGS)

 {
 {

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

 
 

-       if ( ! roffnode_push(r, ROFF_ig, ln, ppos))
-               return(ROFF_ERR);
+       st = &(*bufp)[pos];
+       if (NULL == (ep = strstr(st, "\\}"))) {
+               roffnode_cleanscope(r);
+               return(ROFF_IGN);
+       }
+
+       if (ep > st && '\\' != *(ep - 1)) {
+               ROFF_MDEBUG(r, "closing explicit scope (in-line)");
+               roffnode_pop(r);
+       }

 
 

-       i = (int)ppos;
-       while ((*bufp)[i] && ' ' != (*bufp)[i])
-               i++;
+       roffnode_cleanscope(r);
+       return(ROFF_IGN);
+}

 
 

-       if (i == (int)ppos)
-               return(ROFF_IGN);
-       if ((*bufp)[i])
-               if ( ! (*r->msg)(MANDOCERR_ARGSLOST, r->data, ln, i, NULL))
-                       return(ROFF_ERR);
+
+/* ARGSUSED */
+static enum rofferr
+roff_if(ROFF_ARGS)
+{
+       int              sv;

 
        /*
 
        /*

-        * If `.ig' has arguments, the first argument (up to the next
-        * whitespace) is interpreted as an argument marking the macro
-        * close.  Thus, `.ig foo' will close at `.foo'.
-        *
-        * NOTE: the closing macro `.foo' in the above case is not
-        * allowed to have leading spaces with old groff!  Thus `.foo'
-        * != `. foo'.  Oh yeah, everything after the `.foo' is lost.
-        * Merry fucking Christmas.
+        * Read ahead past the conditional.
+        * FIXME: this does not work, as conditionals don't end on
+        * whitespace, but are parsed according to a formal grammar.
+        * It's good enough for now, however.

         */
 
         */
 

-       r->last->end = malloc((size_t)i - ppos + 1);
-       if (NULL == r->last->end) {
-               (*r->msg)(MANDOCERR_MEM, r->data, ln, ppos, NULL);
-               return(ROFF_ERR);
+       while ((*bufp)[pos] && ' ' != (*bufp)[pos])
+               pos++;
+
+       sv = pos;
+       while (' ' == (*bufp)[pos])
+               pos++;
+
+       /*
+        * Roff is weird.  If we have just white-space after the
+        * conditional, it's considered the BODY and we exit without
+        * really doing anything.  Warn about this.  It's probably
+        * wrong.
+        */
+
+       if ('\0' == (*bufp)[pos] && sv != pos) {
+               if ( ! (*r->msg)(MANDOCERR_NOARGS, r->data, ln, ppos, NULL))
+                       return(ROFF_ERR);
+               return(ROFF_IGN);

        }
 
        }
 

-       memcpy(r->last->end, &(*bufp)[ppos], (size_t)i - ppos);
-       r->last->end[(size_t)i - ppos] = '\0';
+       if ( ! roffnode_push(r, tok, ln, ppos))
+               return(ROFF_ERR);

 
 

-       return(ROFF_IGN);
-}
+       /* Don't evaluate: just assume NO. */

 
 

+       r->last->endspan = 1;

 
 

-int
-roff_endparse(struct roff *r)
-{
+       if ('\\' == (*bufp)[pos] && '{' == (*bufp)[pos + 1]) {
+               ROFF_MDEBUG(r, "opening explicit scope");
+               r->last->endspan = -1;
+               pos += 2;
+       } else
+               ROFF_MDEBUG(r, "opening implicit scope");

 
 

-       if (NULL == r->last)
-               return(1);
-       return((*r->msg)(MANDOCERR_SCOPEEXIT, r->data, 
-                               r->last->line, r->last->col, NULL));
+       /*
+        * If there are no arguments on the line, the next-line scope is
+        * assumed.
+        */
+
+       if ('\0' == (*bufp)[pos])
+               return(ROFF_IGN);
+
+       /* Otherwise re-run the roff parser after recalculating. */
+
+       *offs = pos;
+       return(ROFF_RERUN);

 }
 }