Add initial `define' support for eqn(7).

This works by iterating over a simple list.  It's a slow, auditable
early implementation.  Data is read (the reading function will be
reused) then parsed, then the line re-run if remaining stuff exists.
Note this function isn't the same as mandoc_getarg(), as eqn(7) uses a
different system for reading quoted strings.
This doesn't actually use the defines.

3 files changed, 129 insertions, 14 deletions

diff --git a/eqn.c b/eqn.c
index 323b0442..3d58d8e6 100644
--- a/eqn.c
+++ b/eqn.c
@@ -1,4 +1,4 @@
-/*	$Id: eqn.c,v 1.5 2011/07/12 21:32:43 kristaps Exp $ */
+/*	$Id: eqn.c,v 1.6 2011/07/17 12:13:37 kristaps Exp $ */
 /*
  * Copyright (c) 2011 Kristaps Dzonsons <kristaps@bsd.lv>
  *
@@ -28,12 +28,18 @@
 #include "libmandoc.h"
 #include "libroff.h"
 
+static const char	*eqn_nexttok(struct mparse *, int, int,
+				const char **, size_t *);
+
 /* ARGSUSED */
 enum rofferr
-eqn_read(struct eqn_node **epp, int ln, const char *p, int offs)
+eqn_read(struct eqn_node **epp, int ln, 
+		const char *p, int pos, int *offs)
 {
-	size_t		 sz;
-	struct eqn_node	*ep;
+	size_t		  sz;
+	struct eqn_node	 *ep;
+	const char	 *start, *end;
+	int		  i;
 
 	if (0 == strcmp(p, ".EN")) {
 		*epp = NULL;
@@ -41,14 +47,69 @@ eqn_read(struct eqn_node **epp, int ln, const char *p, int offs)
 	}
 
 	ep = *epp;
+	end = p + pos;
+	start = eqn_nexttok(ep->parse, ln, pos, &end, &sz);
+
+	if (NULL == start)
+		return(ROFF_IGN);
+
+	if (6 == sz && 0 == strncmp("define", start, 6)) {
+		/* 
+		 * TODO: warn if key is quoted: groff doesn't seem to
+		 * like this (I don't know why).
+		 */
+		start = eqn_nexttok(ep->parse, ln, pos, &end, &sz);
+		for (i = 0; i < (int)ep->defsz; i++)  {
+			if (ep->defs[i].keysz != sz)
+				continue;
+			if (0 == strncmp(ep->defs[i].key, start, sz))
+				break;
+		}
+
+		/*
+		 * TODO: merge this code with roff_getstr().
+		 */
+
+		if (i == (int)ep->defsz) {
+			ep->defsz++;
+			ep->defs = mandoc_realloc
+				(ep->defs, ep->defsz * 
+				 sizeof(struct eqn_def));
+			ep->defs[i].keysz = sz;
+			ep->defs[i].key = mandoc_malloc(sz + 1);
+			memcpy(ep->defs[i].key, start, sz);
+			ep->defs[i].key[(int)sz] = '\0';
+			ep->defs[i].val = NULL;
+			ep->defs[i].valsz = 0;
+		}
+
+		start = eqn_nexttok(ep->parse, ln, pos, &end, &sz);
+
+		ep->defs[i].valsz = sz;
+		ep->defs[i].val = mandoc_realloc
+			(ep->defs[i].val, sz + 1);
+		memcpy(ep->defs[i].val, start, sz);
+		ep->defs[i].val[(int)sz] = '\0';
+
+		if ('\0' == *end)
+			return(ROFF_IGN);
+
+		*offs = end - (p + pos);
+		assert(*offs > 0);
+
+		return(ROFF_RERUN);
+	}  else
+		end = p + pos;
+
+	if (0 == (sz = strlen(end)))
+		return(ROFF_IGN);
 
-	sz = strlen(&p[offs]);
 	ep->eqn.data = mandoc_realloc(ep->eqn.data, ep->eqn.sz + sz + 1);
 	if (0 == ep->eqn.sz)
 		*ep->eqn.data = '\0';
 
 	ep->eqn.sz += sz;
-	strlcat(ep->eqn.data, &p[offs], ep->eqn.sz + 1);
+	strlcat(ep->eqn.data, end, ep->eqn.sz + 1);
 	return(ROFF_IGN);
 }
 
@@ -59,7 +120,6 @@ eqn_alloc(int pos, int line, struct mparse *parse)
 
 	p = mandoc_calloc(1, sizeof(struct eqn_node));
 	p->parse = parse;
-
 	p->eqn.line = line;
 	p->eqn.pos = pos;
 
@@ -77,7 +137,52 @@ eqn_end(struct eqn_node *e)
 void
 eqn_free(struct eqn_node *p)
 {
+	int		 i;
 
 	free(p->eqn.data);
+
+	for (i = 0; i < (int)p->defsz; i++) {
+		free(p->defs[i].key);
+		free(p->defs[i].val);
+	}
+
+	free(p->defs);
 	free(p);
 }
+
+static const char *
+eqn_nexttok(struct mparse *mp, int ln, int pos,
+		const char **next, size_t *sz)
+{
+	const char	*start;
+	int		 q;
+
+	start = *next;
+	q = 0;
+
+	if ('\0' == *start)
+		return(NULL);
+
+	if ('"' == *start) {
+		start++;
+		q = 1;
+	}
+
+	*next = q ? strchr(start, '"') : strchr(start, ' ');
+
+	if (NULL != *next) {
+		*sz = (size_t)(*next - start);
+		if (q)
+			(*next)++;
+		while (' ' == **next)
+			(*next)++;
+	} else {
+		if (q)
+			mandoc_msg(MANDOCERR_BADQUOTE, 
+					mp, ln, pos, NULL);
+		*next = strchr(start, '\0');
+		*sz = (size_t)(*next - start);
+	}
+
+	return(start);
+}
diff --git a/libroff.h b/libroff.h
index 9831b19e..58dc86c9 100644
--- a/libroff.h
+++ b/libroff.h
@@ -1,4 +1,4 @@
-/*	$Id: libroff.h,v 1.21 2011/07/12 21:32:43 kristaps Exp $ */
+/*	$Id: libroff.h,v 1.22 2011/07/17 12:13:37 kristaps Exp $ */
 /*
  * Copyright (c) 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
  *
@@ -42,10 +42,19 @@ struct	tbl_node {
 	struct tbl_node	 *next;
 };
 
+struct	eqn_def {
+	char		 *key;
+	size_t		  keysz;
+	char		 *val;
+	size_t		  valsz;
+};
+
 struct	eqn_node {
+	struct eqn_def	 *defs;
+	size_t		  defsz;
 	struct eqn	  eqn;
-	struct eqn_node	 *next;
 	struct mparse	 *parse;
+	struct eqn_node  *next;
 };
 
 struct tbl_node	*tbl_alloc(int, int, struct mparse *);
@@ -62,7 +71,8 @@ void		 tbl_end(struct tbl_node *);
 struct eqn_node	*eqn_alloc(int, int, struct mparse *);
 void		 eqn_end(struct eqn_node *);
 void		 eqn_free(struct eqn_node *);
-enum rofferr 	 eqn_read(struct eqn_node **, int, const char *, int);
+enum rofferr 	 eqn_read(struct eqn_node **, int, 
+			const char *, int, int *);
 
 __END_DECLS
 
diff --git a/roff.c b/roff.c
index b6fa4682..f2b2a743 100644
--- a/roff.c
+++ b/roff.c
@@ -1,4 +1,4 @@
-/*	$Id: roff.c,v 1.145 2011/07/12 21:32:43 kristaps Exp $ */
+/*	$Id: roff.c,v 1.146 2011/07/17 12:13:37 kristaps Exp $ */
 /*
  * Copyright (c) 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
  * Copyright (c) 2010, 2011 Ingo Schwarze <schwarze@openbsd.org>
@@ -519,18 +519,18 @@ roff_parseln(struct roff *r, int ln, char **bufp,
 		if (ROFF_CONT != e)
 			return(e);
 		if (r->eqn)
-			return(eqn_read(&r->eqn, ln, *bufp, pos));
+			return(eqn_read(&r->eqn, ln, *bufp, pos, offs));
 		if (r->tbl)
 			return(tbl_read(r->tbl, ln, *bufp, pos));
 		return(ROFF_CONT);
 	} else if ( ! ctl) {
 		if (r->eqn)
-			return(eqn_read(&r->eqn, ln, *bufp, pos));
+			return(eqn_read(&r->eqn, ln, *bufp, pos, offs));
 		if (r->tbl)
 			return(tbl_read(r->tbl, ln, *bufp, pos));
 		return(ROFF_CONT);
 	} else if (r->eqn)
-		return(eqn_read(&r->eqn, ln, *bufp, ppos));
+		return(eqn_read(&r->eqn, ln, *bufp, ppos, offs));
 
 	/*
 	 * If a scope is open, go to the child handler for that macro,