Support big endian utf16 strings write+search ##search

radareorg · Nov 24, 2024 · 64da2da · 64da2da
1 parent 31acf6f
commit 64da2da
Show file tree

Hide file tree

Showing 7 changed files with 39 additions and 9 deletions.
diff --git a/libr/core/cmd_search.inc.c b/libr/core/cmd_search.inc.c
@@ -5003,10 +5003,11 @@ static int cmd_search(void *data, const char *input) {
 			shift++;
 		}
 		size_t strstart = shift + 1;
+		const bool be = r_config_get_b (core->config, "cfg.bigendian");
 		r_search_reset (core->search, R_SEARCH_KEYWORD);
 		r_search_set_distance (core->search, (int)
 				r_config_get_i (core->config, "search.distance"));
-		RSearchKeyword *skw = r_search_keyword_new_wide (input + strstart, NULL, NULL, ignorecase);
+		RSearchKeyword *skw = r_search_keyword_new_wide (input + strstart, NULL, NULL, ignorecase, be);
 		if (skw) {
 			r_search_kw_add (core->search, skw);
 			r_search_begin (core->search);

diff --git a/libr/core/cmd_write.inc.c b/libr/core/cmd_write.inc.c
@@ -1951,10 +1951,15 @@ static int cmd_ww(void *data, const char *input) {
 	str++;
 	len = (len - 1) << 1;
 	char *tmp = (len > 0) ? malloc (len + 1) : NULL;
+	bool be = r_config_get_b (core->config, "cfg.bigendian");
 	if (tmp) {
 		int i;
 		for (i = 0; i < len; i++) {
-			if (i % 2) {
+			bool match = i % 2;
+			if (be) {
+				match = !match;
+			}
+			if (match) {
 				tmp[i] = 0;
 			} else {
 				tmp[i] = str[i >> 1];

diff --git a/libr/include/r_search.h b/libr/include/r_search.h
@@ -118,7 +118,7 @@ R_API void r_search_keyword_free(RSearchKeyword *kw);
 R_API RSearchKeyword* r_search_keyword_new(const ut8 *kw, int kwlen, const ut8 *bm, int bmlen, const char *data);
 R_API RSearchKeyword* r_search_keyword_new_str(const char *kw, const char *bm, const char *data, bool icase);
 R_API RSearchKeyword* r_search_keyword_new_hexstr(const char *xs, const char *data);
-R_API RSearchKeyword* r_search_keyword_new_wide(const char *kw, const char *bm, const char *data, bool icase);
+R_API RSearchKeyword* r_search_keyword_new_wide(const char *kw, const char *bm, const char *data, bool icase, bool be);
 R_API RSearchKeyword* r_search_keyword_new_hex(const char *kwstr, const char *bmstr, const char *data);
 R_API RSearchKeyword* r_search_keyword_new_hexmask(const char *kwstr, const char *data);
 R_API RSearchKeyword *r_search_keyword_new_regexp(const char *str, const char *data);

diff --git a/libr/include/r_util/r_utf16.h b/libr/include/r_util/r_utf16.h
@@ -12,6 +12,7 @@ R_API int r_utf16_decode(const ut8 *ptr, int ptrlen, RRune *ch, bool bigendian);
 R_API int r_utf16le_decode(const ut8 *ptr, int ptrlen, RRune *ch);
 R_API int r_utf16be_decode(const ut8 *ptr, int ptrlen, RRune *ch);
 R_API int r_utf16le_encode(ut8 *ptr, RRune ch);
+R_API int r_utf16be_encode(ut8 *ptr, RRune ch);
 
 #ifdef __cplusplus
 }

diff --git a/libr/main/rafind2.c b/libr/main/rafind2.c
@@ -288,7 +288,7 @@ static int rafind_open_file(RafindOptions *ro, const char *file, const ut8 *data
 					k = r_search_keyword_new_hexmask (kw, NULL);
 				}
 			} else if (ro->widestr) {
-				k = r_search_keyword_new_wide (kw, ro->mask, NULL, 0);
+				k = r_search_keyword_new_wide (kw, ro->mask, NULL, 0, ro->bigendian);
 			} else {
 				k = r_search_keyword_new_str (kw, ro->mask, NULL, 0);
 			}

diff --git a/libr/search/keyword.c b/libr/search/keyword.c
@@ -91,7 +91,7 @@ R_API RSearchKeyword* r_search_keyword_new_str(const char *kwbuf, const char *bm
 	return kw;
 }
 
-R_API RSearchKeyword* r_search_keyword_new_wide(const char *kwbuf, const char *bmstr, const char *data, bool ignore_case) {
+R_API RSearchKeyword* r_search_keyword_new_wide(const char *kwbuf, const char *bmstr, const char *data, bool ignore_case, bool be) {
 	RSearchKeyword *kw;
 	int len;
 	const char *p2;
@@ -100,7 +100,7 @@ R_API RSearchKeyword* r_search_keyword_new_wide(const char *kwbuf, const char *b
 	int bmlen = 0;
 
 	if (bmstr) {
-		bmbuf = malloc (strlen (bmstr)+1);
+		bmbuf = malloc (strlen (bmstr) + 1);
 		if (!bmbuf) {
 			return NULL;
 		}
@@ -114,7 +114,7 @@ R_API RSearchKeyword* r_search_keyword_new_wide(const char *kwbuf, const char *b
 	str = malloc ((len + 1) * 2);
 	for (p2 = kwbuf, p = str; *p2; ) {
 		RRune ch;
-		int num_utf8_bytes = r_utf8_decode ((const ut8 *)p2, kwbuf + len - p2, &ch);
+		const int num_utf8_bytes = r_utf8_decode ((const ut8 *)p2, kwbuf + len - p2, &ch);
 		if (num_utf8_bytes < 1) {
 			R_LOG_WARN ("Malformed UTF8 at pos %d", (int)(p2 - kwbuf));
 			p[0] = *p2;
@@ -126,7 +126,9 @@ R_API RSearchKeyword* r_search_keyword_new_wide(const char *kwbuf, const char *b
 		if (ignore_case && ch <= 0xff) {
 			ch = tolower (ch);
 		}
-		int num_wide_bytes = r_utf16le_encode ((ut8 *)p, ch);
+		const int num_wide_bytes = be
+			? r_utf16be_encode ((ut8 *)p, ch)
+			: r_utf16le_encode ((ut8 *)p, ch);
 		r_warn_if_fail (num_wide_bytes != 0);
 		p2 += num_utf8_bytes;
 		p += num_wide_bytes;

diff --git a/libr/util/utf16.c b/libr/util/utf16.c
@@ -1,4 +1,4 @@
-/* radare2 - LGPL - Copyright 2017 - kazarmy */
+/* radare2 - LGPL - Copyright 2017-2024 - kazarmy */
 
 #include <r_types.h>
 #include <r_util.h>
@@ -62,3 +62,24 @@ R_API int r_utf16le_encode(ut8 *ptr, RRune ch) {
 	}
 	return 0;
 }
+
+/* Convert a unicode RRune into a UTF-16BE buf */
+R_API int r_utf16be_encode(ut8 *ptr, RRune ch) {
+	if (ch < 0x10000) {
+		ptr[1] = ch & 0xff;
+		ptr[0] = ch >> 8 & 0xff;
+		return 2;
+	}
+	if (ch < 0x110000) {
+		RRune high, low;
+		ch -= 0x10000;
+		high = 0xd800 + (ch >> 10 & 0x3ff);
+		low = 0xdc00 + (ch & 0x3ff);
+		ptr[3] = high & 0xff;
+		ptr[2] = high >> 8 & 0xff;
+		ptr[1] = low & 0xff;
+		ptr[0] = low >> 8 & 0xff;
+		return 4;
+	}
+	return 0;
+}