[project @ 2006-01-22 17:25:30 by jmb]

Convert local files to UTF-8 when dragged in (assumes they're encoded in the system local encoding).
Fix potential buffer overflow in utf8_from_local_encoding

svn path=/import/netsurf/; revision=2025

riscos/ucstables.c

@@ -18,6 +18,7 @@
 #include "oslib/territory.h"
 
 #include "netsurf/riscos/ucstables.h"
+#include "netsurf/utils/log.h"
 #include "netsurf/utils/utf8.h"
 #include "netsurf/utils/utils.h"
 
@@ -582,7 +583,11 @@ utf8_convert_ret utf8_from_local_encoding(const char *string, size_t len,
 
 	/* UTF-8 -> simply copy string */
 	if (alphabet == 111 /* UTF-8 */) {
-		*result = strndup(string, len);
+		temp = strndup(string, len);
+		if (!temp)
+			return UTF8_CONVERT_NOMEM;
+
+		*result = temp;
 		return UTF8_CONVERT_OK;
 	}
 
@@ -616,8 +621,8 @@ utf8_convert_ret utf8_from_local_encoding(const char *string, size_t len,
 		return utf8_from_enc(string, enc, len, result);
 	}
 
-	/* create output buffer (oversized, but not by much) */
+	/* create output buffer (oversized) */
-	*(result) = malloc(len + (3 * offset_count) + 1);
+	*(result) = malloc((len * 4) + (3 * offset_count) + 1);
 	if (!(*result))
 		return UTF8_CONVERT_NOMEM;
 	*(*result) = '\0';
@@ -625,13 +630,13 @@ utf8_convert_ret utf8_from_local_encoding(const char *string, size_t len,
 	/* convert the chunks between offsets, then copy stripped
 	 * UTF-8 character into output string */
 	for (i = 0; i != offset_count; i++) {
-		off = (i > 0 ? offsets[i-1].offset + offsets[i-1].local->len
+		off = (i > 0 ? offsets[i-1].offset + 1 : 0);
-			     : 0);
 
 		err = utf8_from_enc(string + off, enc,
 				offsets[i].offset - off, &temp);
 		if (err != UTF8_CONVERT_OK) {
 			assert(err != UTF8_CONVERT_BADENC);
+			LOG(("utf8_from_enc failed"));
 			free(*result);
 			return UTF8_CONVERT_NOMEM;
 		}
@@ -644,12 +649,12 @@ utf8_convert_ret utf8_from_local_encoding(const char *string, size_t len,
 
 	/* handle last chunk */
 	if (offsets[offset_count - 1].offset < len) {
-		off = offsets[offset_count - 1].offset +
+		off = offsets[offset_count - 1].offset + 1;
-				offsets[offset_count - 1].local->len;
 
 		err = utf8_from_enc(string + off, enc, len - off, &temp);
 		if (err != UTF8_CONVERT_OK) {
 			assert(err != UTF8_CONVERT_BADENC);
+			LOG(("utf8_from_enc failed"));
 			free(*result);
 			return UTF8_CONVERT_NOMEM;
 		}
@@ -659,5 +664,18 @@ utf8_convert_ret utf8_from_local_encoding(const char *string, size_t len,
 		free(temp);
 	}
 
+	/* and copy into more reasonably-sized buffer */
+	temp = malloc(strlen((*result)) + 1);
+	if (!temp) {
+		LOG(("malloc failed"));
+		free(*result);
+		return UTF8_CONVERT_NOMEM;
+	}
+	*temp = '\0';
+
+	strcpy(temp, (*result));
+	free(*result);
+	*result = temp;
+
 	return UTF8_CONVERT_OK;
 }

riscos/window.c

@@ -2209,11 +2209,23 @@ bool ro_gui_window_dataload(struct gui_window *g, wimp_message *message)
 		return false;
 
 	if (file_box) {
+		utf8_convert_ret ret;
+		char *utf8_fn;
+
+		ret = utf8_from_local_encoding(
+				message->data.data_xfer.file_name, 0,
+				&utf8_fn);
+		if (ret != UTF8_CONVERT_OK) {
+			/* A bad encoding should never happen */
+			assert(ret != UTF8_CONVERT_BADENC);
+			LOG(("utf8_from_local_encoding failed"));
+			/* Load was for us - just no memory */
+			return true;
+		}
 
 		/* Found: update form input. */
 		free(file_box->gadget->value);
-		file_box->gadget->value =
+		file_box->gadget->value = utf8_fn;
-				strdup(message->data.data_xfer.file_name);
 
 		/* Redraw box. */
 		box_coords(file_box, &x, &y);
@@ -2754,8 +2766,9 @@ bool ro_gui_window_import_text(struct gui_window *g, const char *filename,
 {
 	fileswitch_object_type obj_type;
 	os_error *error;
-	char *buf;
+	char *buf, *utf8_buf;
 	int size;
+	utf8_convert_ret ret;
 
 	error = xosfile_read_stamped(filename, &obj_type, NULL, NULL,
 			&size, NULL, NULL);
@@ -2782,10 +2795,20 @@ bool ro_gui_window_import_text(struct gui_window *g, const char *filename,
 		return true;
 	}
 
+	ret = utf8_from_local_encoding(buf, size + 1, &utf8_buf);
+	if (ret != UTF8_CONVERT_OK) {
+		/* bad encoding shouldn't happen */
+		assert(ret != UTF8_CONVERT_BADENC);
+		LOG(("utf8_from_local_encoding failed"));
+		free(buf);
+		warn_user("NoMemory", NULL);
+		return true;
+	}
+
 	if (toolbar) {
 		const char *ep = buf + size;
 		const char *sp;
-		char *p = buf;
+		char *p = utf8_buf;
 
 		/* skip leading whitespace */
 		while (p < ep && isspace(*p)) p++;
@@ -2801,8 +2824,10 @@ bool ro_gui_window_import_text(struct gui_window *g, const char *filename,
 			ro_gui_window_launch_url(g, sp);
 	}
 	else
-		browser_window_paste_text(g->bw, buf, size, true);
+		browser_window_paste_text(g->bw, utf8_buf,
+				strlen(utf8_buf), true);
 
 	free(buf);
+	free(utf8_buf);
 	return true;
 }