Total rewrite of rz_search

rizinorg · Nov 25, 2024 · fd68bba · fd68bba
1 parent abd0c6c
commit fd68bba
Show file tree

Hide file tree

Showing 27 changed files with 1,135 additions and 1,672 deletions.
diff --git a/librz/arch/analysis.c b/librz/arch/analysis.c
@@ -795,28 +795,23 @@ RZ_API void rz_analysis_bind(RzAnalysis *analysis, RzAnalysisBind *b) {
 	}
 }
 
-RZ_API RzList /*<RzSearchKeyword *>*/ *rz_analysis_preludes(RzAnalysis *analysis) {
+RZ_API RzSearchCollection *rz_analysis_preludes(RzAnalysis *analysis) {
 	if (analysis->cur && analysis->cur->preludes) {
 		return analysis->cur->preludes(analysis);
 	}
 	return NULL;
 }
 
 RZ_API bool rz_analysis_is_prelude(RzAnalysis *analysis, const ut8 *data, int len) {
-	RzList *l = rz_analysis_preludes(analysis);
-	if (l) {
-		RzSearchKeyword *kw;
-		RzListIter *iter;
-		rz_list_foreach (l, iter, kw) {
-			int ks = kw->keyword_length;
-			if (len >= ks && !memcmp(data, kw->bin_keyword, ks)) {
-				rz_list_free(l);
-				return true;
-			}
-		}
-		rz_list_free(l);
+	RzSearchCollection *col = rz_analysis_preludes(analysis);
+	if (!col || len < 1) {
+		rz_search_collection_free(col);
+		return false;
 	}
-	return false;
+
+	bool any = rz_search_collection_match_any(col, data, len);
+	rz_search_collection_free(col);
+	return any;
 }
 
 RZ_API void rz_analysis_add_import(RzAnalysis *analysis, const char *imp) {

diff --git a/librz/arch/p/analysis/analysis_arm_cs.c b/librz/arch/p/analysis/analysis_arm_cs.c
@@ -2701,29 +2701,32 @@ static ut8 *analysis_mask(RzAnalysis *analysis, int size, const ut8 *data, ut64
 	return ret;
 }
 
-static RzList /*<RzSearchKeyword *>*/ *analysis_preludes(RzAnalysis *analysis) {
-#define KW(d, ds, m, ms) rz_list_append(l, rz_search_keyword_new((const ut8 *)d, ds, (const ut8 *)m, ms, NULL))
-	RzList *l = rz_list_newf((RzListFree)rz_search_keyword_free);
+static RzSearchCollection *analysis_preludes(RzAnalysis *analysis) {
+	RzSearchCollection *sc = rz_search_collection_bytes();
+	if (!sc) {
+		return NULL;
+	}
+#define ADD_PRELUDE(d, m, l) rz_search_collection_bytes_add(sc, NULL, (const ut8 *)d, (const ut8 *)m, l)
 	switch (analysis->bits) {
 	case 16:
-		KW("\x00\xb5", 2, "\x0f\xff", 2);
-		KW("\x08\xb5", 2, "\x0f\xff", 2);
+		ADD_PRELUDE("\x00\xb5", "\x0f\xff", 2);
+		ADD_PRELUDE("\x08\xb5", "\x0f\xff", 2);
 		break;
 	case 32:
-		KW("\x00\x00\x2d\xe9", 4, "\x0f\x0f\xff\xff", 4);
+		ADD_PRELUDE("\x00\x00\x2d\xe9", "\x0f\x0f\xff\xff", 4);
 		break;
 	case 64:
-		KW("\x7f\x23\x03\xd5", 4, "\xff\xff\xff\xff", 4); // pacibsp - Pointer auth
-		KW("\xf0\x0f\x00\xf8", 4, "\xf0\x0f\x00\xff", 4);
-		KW("\xf0\x00\x00\xd1", 4, "\xf0\x00\x00\xff", 4);
-		KW("\xf0\x00\x00\xa9", 4, "\xf0\x00\x00\xff", 4);
-		KW("\x7f\x23\x03\xd5\xff", 5, NULL, 0);
+		ADD_PRELUDE("\x7f\x23\x03\xd5", "\xff\xff\xff\xff", 4); // pacibsp - Pointer auth
+		ADD_PRELUDE("\xf0\x0f\x00\xf8", "\xf0\x0f\x00\xff", 4);
+		ADD_PRELUDE("\xf0\x00\x00\xd1", "\xf0\x00\x00\xff", 4);
+		ADD_PRELUDE("\xf0\x00\x00\xa9", "\xf0\x00\x00\xff", 4);
+		ADD_PRELUDE("\x7f\x23\x03\xd5\xff", NULL, 5);
 		break;
 	default:
-		rz_list_free(l);
-		l = NULL;
+		rz_search_collection_free(sc);
+		sc = NULL;
 	}
-	return l;
+	return sc;
 }
 
 static int address_bits(RzAnalysis *analysis, int bits) {

diff --git a/librz/arch/p/analysis/analysis_mips_cs.c b/librz/arch/p/analysis/analysis_mips_cs.c
@@ -1234,11 +1234,14 @@ static int archinfo(RzAnalysis *a, RzAnalysisInfoType query) {
 	}
 }
 
-static RzList /*<RzSearchKeyword *>*/ *analysis_preludes(RzAnalysis *analysis) {
-#define KW(d, ds, m, ms) rz_list_append(l, rz_search_keyword_new((const ut8 *)d, ds, (const ut8 *)m, ms, NULL))
-	RzList *l = rz_list_newf((RzListFree)rz_search_keyword_free);
-	KW("\x27\xbd\x00", 3, NULL, 0);
-	return l;
+static RzSearchCollection *analysis_preludes(RzAnalysis *analysis) {
+	RzSearchCollection *sc = rz_search_collection_bytes();
+	if (!sc) {
+		return NULL;
+	}
+#define ADD_PRELUDE(d, m, l) rz_search_collection_bytes_add(sc, NULL, (const ut8 *)d, (const ut8 *)m, l)
+	ADD_PRELUDE("\x27\xbd\x00", NULL, 3);
+	return sc;
 }
 
 static bool mips_fini(void *user) {

diff --git a/librz/arch/p/analysis/analysis_ppc_cs.c b/librz/arch/p/analysis/analysis_ppc_cs.c
@@ -1763,11 +1763,14 @@ static int archinfo(RzAnalysis *a, RzAnalysisInfoType query) {
 	}
 }
 
-static RzList /*<RzSearchKeyword *>*/ *analysis_preludes(RzAnalysis *analysis) {
-#define KW(d, ds, m, ms) rz_list_append(l, rz_search_keyword_new((const ut8 *)d, ds, (const ut8 *)m, ms, NULL))
-	RzList *l = rz_list_newf((RzListFree)rz_search_keyword_free);
-	KW("\x7c\x08\x02\xa6", 4, NULL, 0);
-	return l;
+static RzSearchCollection *analysis_preludes(RzAnalysis *analysis) {
+	RzSearchCollection *sc = rz_search_collection_bytes();
+	if (!sc) {
+		return NULL;
+	}
+#define ADD_PRELUDE(d, m, l) rz_search_collection_bytes_add(sc, NULL, (const ut8 *)d, (const ut8 *)m, l)
+	ADD_PRELUDE("\x7c\x08\x02\xa6", NULL, 4);
+	return sc;
 }
 
 static RzAnalysisILConfig *il_config(RzAnalysis *analysis) {

diff --git a/librz/arch/p/analysis/analysis_v850.c b/librz/arch/p/analysis/analysis_v850.c
@@ -421,38 +421,42 @@ static char *get_reg_profile(RzAnalysis *analysis) {
 /**
  * All preludes are guessed by looking at the instruction at the beginning of the function
  */
-static RzList /*<RzSearchKeyword *>*/ *analysis_preludes(RzAnalysis *analysis) {
-#define KW(d, ds, m, ms) rz_list_append(l, rz_search_keyword_new((const ut8 *)d, ds, (const ut8 *)m, ms, NULL))
-	RzList *l = rz_list_newf((RzListFree)rz_search_keyword_free);
+
+static RzSearchCollection *analysis_preludes(RzAnalysis *analysis) {
+	RzSearchCollection *sc = rz_search_collection_bytes();
+	if (!sc) {
+		return NULL;
+	}
+#define ADD_PRELUDE(d, m, l) rz_search_collection_bytes_add(sc, NULL, (const ut8 *)d, (const ut8 *)m, l)
 
 	// movea 0xff, r0, r20
-	KW("\x20\xa6\xff\x00", 4, "\xff\xff\xff\xff", 4);
+	ADD_PRELUDE("\x20\xa6\xff\x00", "\xff\xff\xff\xff", 4);
 
 	// mov r6, r7
 	// ld.w ?[gp], r6
 	// prepare {lp}, 0
-	KW("\x06\x38\x24\x37\x01\x00\x80\x07\x21\x00", 10, "\xff\xff\xff\xff\x01\x00\xff\xff\xff\xff", 10);
+	ADD_PRELUDE("\x06\x38\x24\x37\x01\x00\x80\x07\x21\x00", "\xff\xff\xff\xff\x01\x00\xff\xff\xff\xff", 10);
 
 	// ld.w ?[gp], r6
 	// prepare {lp}, 0
-	KW("\x24\x37\x01\x00\x80\x07\x21\x00", 8, "\xff\xff\x01\x00\xff\xff\xff\xff", 8);
+	ADD_PRELUDE("\x24\x37\x01\x00\x80\x07\x21\x00", "\xff\xff\x01\x00\xff\xff\xff\xff", 8);
 
 	// prepare
-	KW("\x80\x07\x01\x00", 4, "\xc0\xff\x1f\x00", 4);
-	KW("\x80\x07\x03\x00", 4, "\xc0\xff\x1f\x00", 4);
-	KW("\x80\x07\x0b\x00\x00\x00", 6, "\xc0\xff\x1f\x00\x00\x00", 6);
-	KW("\x80\x07\x13\x00\x00\x00", 6, "\xc0\xff\x1f\x00\x00\x00", 6);
-	KW("\x80\x07\x1b\x00\x00\x00\x00\x00", 8, "\xc0\xff\x1f\x00\x00\x00\x00\x00", 8);
+	ADD_PRELUDE("\x80\x07\x01\x00", "\xc0\xff\x1f\x00", 4);
+	ADD_PRELUDE("\x80\x07\x03\x00", "\xc0\xff\x1f\x00", 4);
+	ADD_PRELUDE("\x80\x07\x0b\x00\x00\x00", "\xc0\xff\x1f\x00\x00\x00", 6);
+	ADD_PRELUDE("\x80\x07\x13\x00\x00\x00", "\xc0\xff\x1f\x00\x00\x00", 6);
+	ADD_PRELUDE("\x80\x07\x1b\x00\x00\x00\x00\x00", "\xc0\xff\x1f\x00\x00\x00\x00\x00", 8);
 
 	// trap
-	KW("\xe0\x07\x00\x01", 4, "\xe0\xff\xff\xff", 4);
+	ADD_PRELUDE("\xe0\x07\x00\x01", "\xe0\xff\xff\xff", 4);
 
 	// addi ?, sp, sp
-	KW("\x03\x1e\xd0\xff", 4, "\xff\xff\xff\xff", 4);
+	ADD_PRELUDE("\x03\x1e\xd0\xff", "\xff\xff\xff\xff", 4);
 
 	// add ?, sp
-	KW("\x50\x1a", 2, "\xf0\xff", 2);
-	return l;
+	ADD_PRELUDE("\x50\x1a", "\xf0\xff", 2);
+	return sc;
 }
 
 static int archinfo(RzAnalysis *a, RzAnalysisInfoType query) {

diff --git a/librz/arch/p/analysis/analysis_x86_cs.c b/librz/arch/p/analysis/analysis_x86_cs.c
@@ -3702,27 +3702,29 @@ static int archinfo(RzAnalysis *a, RzAnalysisInfoType query) {
 	}
 }
 
-static RzList /*<RzSearchKeyword *>*/ *analysis_preludes(RzAnalysis *analysis) {
-#define KW(d, ds, m, ms) rz_list_append(l, rz_search_keyword_new((const ut8 *)d, ds, (const ut8 *)m, ms, NULL))
-	RzList *l = rz_list_newf((RzListFree)rz_search_keyword_free);
+static RzSearchCollection *analysis_preludes(RzAnalysis *analysis) {
+	RzSearchCollection *sc = rz_search_collection_bytes();
+	if (!sc) {
+		return NULL;
+	}
+#define ADD_PRELUDE(d, m, l) rz_search_collection_bytes_add(sc, NULL, (const ut8 *)d, (const ut8 *)m, l)
 	switch (analysis->bits) {
 	case 32:
-		KW("\x8b\xff\x55\x8b\xec", 5, NULL, 0);
-		KW("\x55\x89\xe5", 3, NULL, 0);
-		KW("\x55\x8b\xec", 3, NULL, 0);
-		KW("\xf3\x0f\x1e\xfb", 4, NULL, 0); // endbr32
+		ADD_PRELUDE("\x8b\xff\x55\x8b\xec", NULL, 5);
+		ADD_PRELUDE("\x55\x89\xe5", NULL, 3);
+		ADD_PRELUDE("\x55\x8b\xec", NULL, 3);
+		ADD_PRELUDE("\xf3\x0f\x1e\xfb", NULL, 4); // endbr32
 		break;
 	case 64:
-		KW("\x55\x48\x89\xe5", 4, NULL, 0);
-		KW("\x55\x48\x8b\xec", 4, NULL, 0);
-		KW("\xf3\x0f\x1e\xfa", 4, NULL, 0); // endbr64
+		ADD_PRELUDE("\x55\x48\x89\xe5", NULL, 4);
+		ADD_PRELUDE("\x55\x48\x8b\xec", NULL, 4);
+		ADD_PRELUDE("\xf3\x0f\x1e\xfa", NULL, 4); // endbr64
 		break;
 	default:
-		rz_list_free(l);
-		l = NULL;
-		break;
+		rz_search_collection_free(sc);
+		sc = NULL;
 	}
-	return l;
+	return sc;
 }
 
 RzAnalysisPlugin rz_analysis_plugin_x86_cs = {

diff --git a/librz/include/rz_analysis.h b/librz/include/rz_analysis.h
@@ -1372,7 +1372,7 @@ typedef struct rz_analysis_plugin_t {
 	// int (*reset_counter) (RzAnalysis *analysis, ut64 start_addr);
 	int (*archinfo)(RzAnalysis *analysis, RzAnalysisInfoType query);
 	ut8 *(*analysis_mask)(RzAnalysis *analysis, int size, const ut8 *data, ut64 at);
-	RzList /*<RzSearchKeyword *>*/ *(*preludes)(RzAnalysis *analysis);
+	RzSearchCollection *(*preludes)(RzAnalysis *analysis);
 
 	/**
 	 * The actual bit-size of an address for given analysis.bits.
@@ -2275,7 +2275,7 @@ RZ_API void rz_analysis_rtti_print_at_vtable(RzAnalysis *analysis, ut64 addr, Rz
 RZ_API void rz_analysis_rtti_print_all(RzAnalysis *analysis, RzOutputMode mode);
 RZ_API void rz_analysis_rtti_recover_all(RzAnalysis *analysis);
 
-RZ_API RzList /*<RzSearchKeyword *>*/ *rz_analysis_preludes(RzAnalysis *analysis);
+RZ_API RzSearchCollection *rz_analysis_preludes(RzAnalysis *analysis);
 RZ_API bool rz_analysis_is_prelude(RzAnalysis *analysis, const ut8 *data, int len);
 
 /* classes */

diff --git a/librz/include/rz_core.h b/librz/include/rz_core.h
@@ -282,7 +282,7 @@ struct rz_core_t {
 	RzLang *lang;
 	RzDebug *dbg;
 	RzFlag *flags;
-	RzSearch *search;
+	RzSearchOpt *search;
 	RzEgg *egg;
 	RzCrypto *crypto;
 	RzAGraph *graph;