Issue 3871 configure bin hashes

librz/bin/bfile.c

@@ -396,7 +396,7 @@ static ut64 buf_compute_hashes(const ut8 *buf, ut64 size, void *user) {
 }
 
 /**
- * Return a pvector of RzBinFileHash structures with the hashes md5, sha1, sha256, crc32 and entropy
+ * Return a pvector of RzBinFileHash structures with the hashes configured in bin.hashes.default
  * computed over the whole \p bf .
  */
 RZ_API RZ_OWN RzPVector /*<RzBinFileHash *>*/ *rz_bin_file_compute_hashes(RzBin *bin, RzBinFile *bf, ut64 limit) {
@@ -426,43 +426,47 @@ RZ_API RZ_OWN RzPVector /*<RzBinFileHash *>*/ *rz_bin_file_compute_hashes(RzBin
 	if (!md) {
 		goto rz_bin_file_compute_hashes_bad;
 	}
+	RzListIter *hash_algos_iter = bin->default_hashes ? rz_list_iterator(bin->default_hashes) : NULL;
 
-	if (!rz_hash_cfg_configure(md, "md5") ||
-		!rz_hash_cfg_configure(md, "sha1") ||
-		!rz_hash_cfg_configure(md, "sha256") ||
-		!rz_hash_cfg_configure(md, "crc32") ||
-		!rz_hash_cfg_configure(md, "entropy")) {
-		goto rz_bin_file_compute_hashes_bad;
-	}
-	if (!rz_hash_cfg_init(md)) {
-		goto rz_bin_file_compute_hashes_bad;
-	}
+	if (hash_algos_iter) {
+		while (hash_algos_iter) {
+			char *algo = rz_list_iter_get(hash_algos_iter);
+			if (!rz_hash_cfg_configure(md, algo)) {
+				goto rz_bin_file_compute_hashes_bad;
+			}
+		}
 
-	if (rz_buf_fwd_scan(buf, 0, buf_size, buf_compute_hashes, md) != buf_size) {
-		goto rz_bin_file_compute_hashes_bad;
-	}
+		if (!rz_hash_cfg_init(md)) {
+			goto rz_bin_file_compute_hashes_bad;
+		}
 
-	if (!rz_hash_cfg_final(md)) {
-		goto rz_bin_file_compute_hashes_bad;
-	}
+		if (rz_buf_fwd_scan(buf, 0, buf_size, buf_compute_hashes, md) != buf_size) {
+			goto rz_bin_file_compute_hashes_bad;
+		}
 
-	if (!add_file_hash(md, "md5", file_hashes) ||
-		!add_file_hash(md, "sha1", file_hashes) ||
-		!add_file_hash(md, "sha256", file_hashes) ||
-		!add_file_hash(md, "crc32", file_hashes) ||
-		!add_file_hash(md, "entropy", file_hashes)) {
-		goto rz_bin_file_compute_hashes_bad;
-	}
+		if (!rz_hash_cfg_final(md)) {
+			goto rz_bin_file_compute_hashes_bad;
+		}
 
-	if (o->plugin && o->plugin->hashes) {
-		RzPVector *plugin_hashes = o->plugin->hashes(bf);
-		void **it;
-		rz_pvector_foreach (plugin_hashes, it) {
-			RzBinFileHash *h = *it;
-			rz_pvector_push(file_hashes, h);
+		hash_algos_iter = rz_list_iterator(bin->default_hashes);
+
+		while (hash_algos_iter) {
+			char *algo = rz_list_iter_get(hash_algos_iter);
+			if (!add_file_hash(md, algo, file_hashes)) {
+				goto rz_bin_file_compute_hashes_bad;
+			}
+		}
+
+		if (o->plugin && o->plugin->hashes) {
+			RzPVector *plugin_hashes = o->plugin->hashes(bf);
+			void **it;
+			rz_pvector_foreach (plugin_hashes, it) {
+				RzBinFileHash *h = *it;
+				rz_pvector_push(file_hashes, h);
+			}
+			plugin_hashes->v.free = NULL;
+			rz_pvector_free(plugin_hashes);
 		}
-		plugin_hashes->v.free = NULL;
-		rz_pvector_free(plugin_hashes);
 	}
 
 	// TODO: add here more rows

librz/core/cconfig.c

@@ -2450,6 +2450,68 @@ static bool cb_binverbose(void *user, void *data) {
 	return true;
 }
 
+static bool cb_binhashesdefault(void *user, void *data) {
+	RzCore *core = (RzCore *)user;
+	RzConfigNode *node = (RzConfigNode *)data;
+	if (*node->value == '?') {
+		print_node_options(node);
+		rz_cons_printf("Multiple algorithms can be specified in a comma-separated list (no spaces).\n");
+		return false;
+	}
+
+	char *algos[] = {
+		"md5",
+		"sha1",
+		"sha256",
+		"crc32",
+		"crc64",
+		"entropy",
+		"blake3",
+		"sm3",
+	};
+
+	free(core->bin->default_hashes);
+	if (*node->value) {
+		char *delim = ",";
+		char *value = strdup(node->value);
+		if (!value) {
+			rz_sys_perror("strdup");
+			return false;
+		}
+		RzList *algo_list = rz_list_newf((RzListFree)free);
+		char *token = strtok(value, delim);
+		while (token) {
+			for (int i = 0; i < RZ_ARRAY_SIZE(algos); i++) {
+				if (algos[i] != NULL && !strcmp(token, algos[i])) {
+					char *hash = strdup(algos[i]);
+					if (hash) {
+						rz_list_append(algo_list, hash);
+					} else {
+						rz_sys_perror("strdup");
+						rz_list_free(algo_list);
+						free(value);
+						return false;
+					}
+					algos[i] = NULL;
+					break;
+				} else if (i == RZ_ARRAY_SIZE(algos) - 1) {
+					RZ_LOG_ERROR("core: bin.hashes.default: invalid or duplicate value: %s\n", node->value);
+					rz_list_free(algo_list);
+					free(value);
+					return false;
+				}
+			}
+			token = strtok(NULL, delim);
+		}
+		free(value);
+		core->bin->default_hashes = algo_list;
+		return true;
+	} else {
+		core->bin->default_hashes = NULL;
+		return true;
+	}
+}
+
 static bool cb_debase64(void *user, void *data) {
 	RzCore *core = (RzCore *)user;
 	RzConfigNode *node = (RzConfigNode *)data;
@@ -3245,6 +3307,10 @@ RZ_API int rz_core_config_init(RzCore *core) {
 	SETCB("bin.debase64", "false", &cb_debase64, "Try to debase64 all strings");
 	SETBPREF("bin.classes", "true", "Load classes from rbin on startup");
 	SETCB("bin.verbose", "false", &cb_binverbose, "Show RzBin warnings when loading binaries");
+	SETCB("bin.hashes.default", "md5,sha1,sha256,crc32,entropy", &cb_binhashesdefault, "Select hash algorithms");
+	n = NODECB("bin.hashes.default", "md5,sha1,sha256,crc32,entropy", &cb_binhashesdefault);
+	SETDESC(n, "Select hashing algorithms");
+	SETOPTIONS(n, "md5", "sha1", "sha256", "crc32", "entropy", "blake3", "sm3", NULL);
 
 	/* prj */
 	SETPREF("prj.file", "", "Path of the currently opened project");

librz/core/cmd/cmd_info.c

@@ -622,10 +622,16 @@ RZ_IPI RzCmdStatus rz_cmd_info_hashes_handler(RzCore *core, int argc, const char
 	GET_CHECK_CUR_BINFILE(core);
 
 	RzPVector *new_hashes = rz_bin_file_compute_hashes(core->bin, bf, limit);
+	if (!new_hashes) {
+		RZ_LOG_ERROR("core: Computing file hashes failed\n")
+		return RZ_CMD_STATUS_ERROR;
+	}
 	RzPVector *old_hashes = rz_bin_file_set_hashes(core->bin, new_hashes);
-	bool equal = true;
+	bool equal = false;
 	if (new_hashes && old_hashes && !rz_pvector_empty(new_hashes) && !rz_pvector_empty(old_hashes)) {
-		if (!is_equal_file_hashes(new_hashes, old_hashes, &equal)) {
+		if (is_equal_file_hashes(new_hashes, old_hashes, &equal)) {
+			equal = true;
+		} else {
 			RZ_LOG_ERROR("core: Cannot compare file hashes\n");
 			rz_pvector_free(old_hashes);
 			return RZ_CMD_STATUS_ERROR;
@@ -653,20 +659,23 @@ RZ_IPI RzCmdStatus rz_cmd_info_hashes_handler(RzCore *core, int argc, const char
 		break;
 	case RZ_OUTPUT_MODE_STANDARD:
 		if (!equal) {
-			size_t min_len = RZ_MIN(rz_pvector_len(old_hashes), rz_pvector_len(new_hashes));
-			for (int i = 0; i < min_len; i++) {
-				fh_new = (RzBinFileHash *)rz_pvector_at(new_hashes, i);
-				fh_old = (RzBinFileHash *)rz_pvector_at(old_hashes, i);
-				if (strcmp(fh_new->type, fh_old->type)) {
+			size_t max_len = RZ_MAX(rz_pvector_len(old_hashes), rz_pvector_len(new_hashes));
+			for (int i = 0; i < max_len; i++) {
+				fh_new = i < rz_pvector_len(new_hashes) ? (RzBinFileHash *)rz_pvector_at(new_hashes, i) : NULL;
+				fh_old = i < rz_pvector_len(old_hashes) ? (RzBinFileHash *)rz_pvector_at(old_hashes, i) : NULL;
+				if (fh_new && fh_old && strcmp(fh_new->type, fh_old->type)) {
 					RZ_LOG_ERROR("core: Wrong file hashes structure");
 				}
-				if (!strcmp(fh_new->hex, fh_old->hex)) {
+				if (fh_new && fh_old && !strcmp(fh_new->hex, fh_old->hex)) {
 					fprintf(stderr, "= %s %s\n", fh_new->type, fh_new->hex); // output one line because hash remains same `= hashtype hashval`
 				} else {
 					// output diff-like two lines, one with old hash val `- hashtype hashval` and one with new `+ hashtype hashval`
-					fprintf(stderr, "- %s %s\n+ %s %s\n",
-						fh_old->type, fh_old->hex,
-						fh_new->type, fh_new->hex);
+					if (fh_old) {
+						fprintf(stderr, "- %s %s\n", fh_old->type, fh_old->hex);
+					}
+					if (fh_new) {
+						fprintf(stderr, "+ %s %s\n", fh_new->type, fh_new->hex);
+					}
 				}
 			}
 		} else { // hashes are equal

librz/hash/hash.c

@@ -261,8 +261,8 @@ RZ_API void rz_hash_cfg_free(RZ_NONNULL RzHashCfg *md) {
 /**
  * \brief Allocates and configures the plugin message digest context
  *
- * message digest allocates internally a HashCfgConfig which
- * contains all the needed informations to the plugin to work.
+ * message digest internally allocates a HashCfgConfig which
+ * contains all the information needed for the plugin to work.
  * */
 RZ_API bool rz_hash_cfg_configure(RZ_NONNULL RzHashCfg *md, RZ_NONNULL const char *name) {
 	rz_return_val_if_fail(md && name, false);

librz/include/rz_bin.h

@@ -369,6 +369,7 @@ struct rz_bin_t {
 	bool is_reloc_patched; // used to indicate whether relocations were patched or not
 	RzDemangler *demangler;
 	RzHash *hash;
+	RzList /*<char *>*/ *default_hashes; // bin.hashes.default
 };
 
 typedef struct rz_bin_xtr_metadata_t {

test/db/cmd/cmd_i

@@ -4637,7 +4637,7 @@ EXPECT_ERR=<<EOF
 EOF
 RUN
 
-NAME=it2
+NAME=iT2
 FILE=--
 CMDS=<<EOF
 mkdir .tmp
@@ -4671,6 +4671,20 @@ entropy 2.959094
 EOF
 RUN
 
+NAME=iT3
+FILE==
+CMDS=iT; e bin.hashes.default=sm3,sha1; iT
+EXPECT=<<EOF
+md5 bf619eac0cdf3f68d496ea9344137e8b
+sha1 5c3eb80066420002bc3dcc7ca4ab6efad7ed4ae5
+sha256 076a27c79e5ace2a3d47f9dd2e83e4ff6ea8872b3c2218f66c92b89b55f36560
+crc32 b2aa7578
+entropy 0.000000
+sm3 8b63e00c479bc948f58d360e3831bc32e3924fa25d126a221ec9ad7fc6223462
+sha1 5c3eb80066420002bc3dcc7ca4ab6efad7ed4ae5
+EOF
+RUN
+
 NAME=isq. (malloc)
 FILE=malloc://512
 CMDS=isq.

test/integration/test_bin.c

@@ -8,6 +8,8 @@
 
 bool test_rz_bin(void) {
 	RzBin *bin = rz_bin_new();
+	const char* default_algos[] = {"md5", "sha1", "sha256", "crc32", "entropy"};
+	bin->default_hashes = rz_list_new_from_array(default_algos, sizeof(default_algos)/sizeof(default_algos[0]));
 	RzIO *io = rz_io_new();
 	rz_io_bind(io, &bin->iob);