fix(data): reject partial and malformed map data

trie.c

@@ -1,14 +1,33 @@
+#include <errno.h>
+#include <limits.h>
 #include "dat.h"
 #include "fn.h"
 
+static char*
+readline(Biobuf *b, char *path)
+{
+	char *line;
+
+	errno = 0;
+	line = Brdstr(b, '\n', 1);
+	if(errno != 0)
+		die("can't read %s: %s", path, strerror(errno));
+	return line;
+}
+
 static int
 newnode(Trie *t)
 {
+	int cap;
 	int i;
 
 	if(t->n >= t->cap){
-		t->cap *= 2;
-		t->nodes = erealloc(t->nodes, t->cap * sizeof(Tnode));
+		if(t->cap > INT_MAX/2 ||
+		   (ulong)t->cap > ULONG_MAX/(2*sizeof(Tnode)))
+			die("map is too large");
+		cap = t->cap * 2;
+		t->nodes = erealloc(t->nodes, (ulong)cap * sizeof(Tnode));
+		t->cap = cap;
 	}
 	i = t->n++;
 	memset(&t->nodes[i], 0, sizeof(Tnode));
@@ -54,7 +73,7 @@ insert(Trie *t, char *key, int klen, char *val, int vlen)
 		ni = ci;
 	}
 	free(t->nodes[ni].val);
-	t->nodes[ni].val = emalloc(vlen + 1);
+	t->nodes[ni].val = emalloc((ulong)vlen + 1);
 	memmove(t->nodes[ni].val, val, vlen);
 	t->nodes[ni].val[vlen] = '\0';
 	t->nodes[ni].vlen = vlen;
@@ -66,41 +85,45 @@ trieopen(char *path)
 	Trie *t;
 	Biobuf *b;
 	char *line, *tab, *key, *val;
+	Str s;
 	int klen, vlen;
 
 	b = Bopen(path, OREAD);
 	if(b == nil)
-		die("can't open: %s", path);
+		die("can't open %s: %s", path, strerror(errno));
 	t = emalloc(sizeof(*t));
 	t->cap = 1024;
-	t->nodes = emalloc(t->cap * sizeof(Tnode));
+	t->nodes = emalloc((ulong)t->cap * sizeof(Tnode));
 	t->n = 0;
 	t->root = newnode(t);
-	while((line = Brdstr(b, '\n', 1)) != nil){
-		vlen = strlen(line);
+	while((line = readline(b, path)) != nil){
+		vlen = Blinelen(b);
+		if(memchr(line, '\0', vlen) != nil)
+			die("NUL in map: %s", path);
 		if(vlen > 0 && line[vlen-1] == '\r')
 			line[--vlen] = '\0';
 		if(line[0] == '\0' || line[0] == ';'){
 			free(line);
 			continue;
 		}
-		tab = strchr(line, '\t');
-		if(tab == nil || tab == line || tab[1] == '\0' ||
-		   strchr(tab+1, '\t') != nil)
+		tab = memchr(line, '\t', vlen);
+		if(tab == nil || tab == line || tab == line+vlen-1 ||
+		   memchr(tab+1, '\t', line+vlen-(tab+1)) != nil)
 			die("malformed map: %s", path);
 		*tab = '\0';
 		key = line;
 		klen = tab - line;
-		if(utflen(key) > Maxrunes)
-			die("map key too long: %s", path);
+		if(!sinit(&s, key, klen))
+			die("invalid or oversized map key: %s", path);
 		val = tab + 1;
-		vlen = strlen(val);
-		if(utflen(val) > Maxrunes)
-			die("map value too long: %s", path);
+		vlen = line + vlen - val;
+		if(!sinit(&s, val, vlen))
+			die("invalid or oversized map value: %s", path);
 		insert(t, key, klen, val, vlen);
 		free(line);
 	}
-	Bterm(b);
+	if(Bterm(b) < 0)
+		die("can't close %s: %s", path, strerror(errno));
 	return t;
 }
 
@@ -117,46 +140,30 @@ trieclose(Trie *t)
 	free(t);
 }
 
-char*
-trieget(Trie *t, char *key, int klen, int *vlen)
-{
-	int ni;
-	int i;
-
-	if(t == nil || klen < 0 || (klen > 0 && key == nil))
-		return nil;
-	ni = t->root;
-	for(i = 0; i < klen; i++){
-		ni = find(t, ni, key[i]);
-		if(ni < 0)
-			return nil;
-	}
-	if(t->nodes[ni].val == nil)
-		return nil;
-	if(vlen != nil)
-		*vlen = t->nodes[ni].vlen;
-	return t->nodes[ni].val;
-}
-
 int
 trielookup(Trie *t, char *key, int klen, char **val, int *vlen)
 {
 	int ni;
 	int i;
 
+	if(val != nil)
+		*val = nil;
+	if(vlen != nil)
+		*vlen = 0;
 	if(t == nil || klen < 0 || (klen > 0 && key == nil))
-		return 0;
+		return TrieMiss;
 	ni = t->root;
 	for(i = 0; i < klen; i++){
 		ni = find(t, ni, key[i]);
 		if(ni < 0)
-			return 0;
+			return TrieMiss;
 	}
 	if(t->nodes[ni].val != nil){
 		if(val != nil)
 			*val = t->nodes[ni].val;
 		if(vlen != nil)
 			*vlen = t->nodes[ni].vlen;
+		return TrieExact;
 	}
-	return 1;
+	return TriePrefix;
 }