fi

.github/workflows/ci.yml

@@ -1,7 +1,7 @@
 name: CI
 
 env:
-  R2V: 6.0.8
+  R2V: 6.1.4
 
 permissions:
   contents: write
@@ -72,7 +72,7 @@ jobs:
           ls -l libhbc*
 
       - name: Create GitHub Release
-        uses: softprops/action-gh-release@v2.6.1
+        uses: softprops/action-gh-release@v3.0.0
         with:
           tag_name: ${{ needs.build.outputs.tag }}
           files: ./libhbc-${{ needs.build.outputs.tag }}.tgz

.github/workflows/meson.yml

@@ -1,7 +1,7 @@
 name: Meson Build
 
 env:
-  R2V: 6.0.8
+  R2V: 6.1.4
 
 on:
   push:

.github/workflows/r2plugin.yml

@@ -1,7 +1,7 @@
 name: CI
 
 env:
-  R2V: 6.0.8
+  R2V: 6.1.4
 
 permissions:
   contents: write

.github/workflows/windows.yml

@@ -1,7 +1,7 @@
 name: Windows Build
 
 env:
-  R2V: 6.0.8
+  R2V: 6.1.4
 
 on:
   push:

.gitignore

@@ -1,5 +1,5 @@
 build/
-bin/libhbctool
+bin/
 third_party/
 *.d
 *.o

Makefile

@@ -18,8 +18,9 @@ DISASM_SRC = $(wildcard $(SRC_DIR)/disassembly/*.c)
 DECOMPILE_SRC = $(wildcard $(SRC_DIR)/decompilation/*.c)
 OPCODES_SRC = $(filter-out $(SRC_DIR)/opcodes/%.inc.c,$(wildcard $(SRC_DIR)/opcodes/*.c))
 LIB_SRC = $(UTILS_SRC) $(PARSERS_SRC) $(DISASM_SRC) $(DECOMPILE_SRC) $(OPCODES_SRC) \
-          $(SRC_DIR)/hbc.c $(SRC_DIR)/opcodes/encoder.c $(SRC_DIR)/opcodes/decoder.c $(SRC_DIR)/r2.c
+          $(SRC_DIR)/hbc.c $(SRC_DIR)/literals_api.c $(SRC_DIR)/opcodes/encoder.c $(SRC_DIR)/opcodes/decoder.c $(SRC_DIR)/r2.c
 MAIN_SRC = src/tool/libhbctool.c
+HBC_HEADERS = $(shell find include -name '*.h')
 
 ## Object files
 LIB_OBJ = $(patsubst $(SRC_DIR)/%.c,$(BUILD_DIR)/%.o,$(LIB_SRC))
@@ -55,11 +56,11 @@ $(STATIC_LIB): $(LIB_OBJ)
 $(BIN_FILE): $(STATIC_LIB) $(MAIN_OBJ) | $(shell mkdir -p $(BIN_DIR))
 	$(CC) $(CFLAGS) -o $@ $(MAIN_OBJ) -L$(BUILD_DIR) -lhbc
 
-$(BUILD_DIR)/%.o: $(SRC_DIR)/%.c | $(VH)
+$(BUILD_DIR)/%.o: $(SRC_DIR)/%.c $(HBC_HEADERS) | $(VH)
 	@mkdir -p $(dir $@)
 	$(CC) $(CFLAGS) $(INCLUDES) -c $< -o $@
 
-$(MAIN_OBJ): $(MAIN_SRC) | $(VH)
+$(MAIN_OBJ): $(MAIN_SRC) $(HBC_HEADERS) | $(VH)
 	$(CC) $(CFLAGS) $(INCLUDES) -c $< -o $@
 
 clean:
@@ -72,6 +73,14 @@ TEST_DIR = tests
 TEST_SRC = $(wildcard $(TEST_DIR)/*.c)
 TEST_BIN = $(BIN_DIR)/run_tests
 
+$(TEST_BIN): $(STATIC_LIB) $(TEST_SRC) | $(shell mkdir -p $(BIN_DIR))
+	$(CC) $(CFLAGS) $(INCLUDES) -o $@ $(TEST_SRC) -L$(BUILD_DIR) -lhbc
+
+test: $(BIN_FILE) $(TEST_BIN)
+	$(TEST_BIN)
+	$(BIN_FILE) sl test/bins/hbc/bespoke_eval.hbc | grep -q '0x000000b4 f=0 +0x4 yes.js:1:5 stmt=0'
+	$(BIN_FILE) h test/bins/hbc/bespoke_eval.hbc | grep -q 'Source Locations: 13 bytes'
+	$(BIN_FILE) h test/bins/hbc/index.android.bundle | grep -q 'Source Locations: 0 bytes'
 
 r2:
 	$(MAKE) clean

include/hbc/common.h

@@ -9,7 +9,6 @@
 #include <string.h>
 #include <errno.h>
 #include <stdarg.h>
-#include "version.h"
 
 #ifndef R2_VERSION
 #define R_RETURN_VAL_IF_FAIL(x, y) \

include/hbc/decompilation/decompiler.h

@@ -144,6 +144,8 @@ typedef struct HermesDecompiler {
 	bool *function_in_progress; /* Track functions currently on the emit stack */
 	int indent_level;
 	bool inlining_function; /* True when outputting a nested function inline */
+	bool output_truncated; /* Sticky flag: budget exceeded; stop emitting new content */
+	bool truncation_marker_emitted; /* Sticky flag: the truncation comment was already written */
 	HBCDecompOptions options;
 	StringBuffer output;
 } HermesDecompiler;

include/hbc/hbc.h

@@ -63,6 +63,34 @@ typedef struct {
 	u32 param_count; // Number of parameters
 } HBCFunc;
 
+typedef struct {
+	u32 function_id;
+	u32 address;
+	u32 function_address;
+	u32 line;
+	u32 column;
+	u32 statement;
+	const char *filename;
+} HBCSourceLine;
+
+typedef struct {
+	HBCSourceLine *lines;
+	u32 count;
+} HBCSourceLineArray;
+
+typedef struct {
+	bool has_debug_info;
+	u32 filename_count;
+	u32 filename_storage_size;
+	u32 file_region_count;
+	u32 source_locations_size;
+	u32 scope_desc_data_size;
+	u32 textified_data_size;
+	u32 string_table_size;
+	u32 debug_data_size;
+	u32 functions_with_debug_info;
+} HBCDebugInfo;
+
 /* String table data */
 typedef struct {
 	u32 string_count;
@@ -104,6 +132,10 @@ typedef struct {
 	bool force_dispatch; /* Force switch/case dispatch loop for linear functions */
 	bool inline_closures; /* Inline closure definitions (default: true) */
 	int inline_threshold; /* Max instruction count to inline (0 = no limit, -1 = no inline) */
+	/* Early-stop budgets (0 = unlimited). max_ast_statements caps pass2 AST
+	 * growth per function; max_output_bytes caps total emitted output. */
+	int max_ast_statements;
+	int max_output_bytes;
 } HBCDecompOptions;
 
 /* Per-instruction details */
@@ -216,6 +248,14 @@ Result hbc_get_string_tables(HBC *hbc, HBCStrs *out);
  */
 Result hbc_get_function_source(HBC *hbc, u32 function_id, const char **out_src);
 
+/**
+ * Get decoded source-line records from debug information.
+ */
+Result hbc_get_source_lines(HBC *hbc, HBCSourceLineArray *out);
+void hbc_free_source_lines(HBCSourceLineArray *arr);
+bool hbc_has_source_lines(HBC *hbc);
+Result hbc_get_debug_info(HBC *hbc, HBCDebugInfo *out);
+
 /**
  * Get the raw bytecode bytes for a function.
  */
@@ -284,11 +324,39 @@ typedef struct {
 	u32 count;
 } HBCInsns;
 
+typedef enum {
+	HBC_BINDING_IMPORT = 0,
+	HBC_BINDING_EXPORT = 1
+} HBCBindingType;
+
+typedef struct {
+	HBCBindingType type;
+	const char *kind; /* js, native, cjs, global, module */
+	char *name;
+	char *module;
+	u32 function_id;
+	u32 offset;
+	u32 string_id;
+} HBCBinding;
+
+typedef struct {
+	HBCBinding *bindings;
+	u32 count;
+} HBCBindings;
+
 /**
  * Free instruction array.
  */
 void hbc_free_insns(HBCInsn *insns, u32 count);
 
+/**
+ * Scan bytecode/string references for probable JS/native import/export
+ * boundaries. Hermes has no native import table, so these entries are derived
+ * from bytecode operands and CJS metadata.
+ */
+Result hbc_scan_bindings(HBC *hbc, HBCBindings *out);
+void hbc_free_bindings(HBCBindings *bindings);
+
 /**
  * Encoding buffer and functions
  */

include/hbc/literals.h

@@ -0,0 +1,114 @@
+#ifndef LIBHBC_LITERALS_PUBLIC_H
+#define LIBHBC_LITERALS_PUBLIC_H
+
+#include <hbc/common.h>
+#include <hbc/hbc.h>
+
+/* ============================================================================
+ * Literals API (public)
+ *
+ * Covers the buffer-backed literals produced by:
+ *   OP_NewArrayWithBuffer / OP_NewArrayWithBufferLong
+ *   OP_NewObjectWithBuffer / OP_NewObjectWithBufferLong
+ *
+ * Responsibilities:
+ *   - Format a literal from opcode immediates into a JS-like string.
+ *   - Maintain a lazy cache of distinct literals so the disassembler and
+ *     decompiler can look up previously-formatted literals in O(1) and avoid
+ *     re-walking the SLP buffer on every call.
+ *   - Track the set of instruction addresses that reference each literal.
+ *   - Enumerate literals from the pool side (scan SLP stream) and from the
+ *     code side (walk function bodies).
+ *   - Allow resetting the cache to recover from a bad scan.
+ *
+ * Caching policy: entries are keyed by (kind, num_items, primary_id,
+ * secondary_id); formatting happens at registration time and is memoized.
+ * The cache can be cleared at any time.
+ * ============================================================================ */
+
+typedef enum {
+	HBC_LIT_ARRAY = 0,
+	HBC_LIT_OBJECT = 1
+} HBCLiteralKind;
+
+/* One entry in the literal cache. The strings / xref array are owned by the
+ * HBC and freed on hbc_close() or hbc_literals_reset(). */
+typedef struct {
+	HBCLiteralKind kind;
+	u32 num_items;
+	u32 primary_id; /* array_id (array) or keys/shape id (object) */
+	u32 secondary_id; /* values_id for objects, 0 for arrays */
+	u32 paddr; /* absolute file offset of the literal data in the SLP pool (0 if unknown) */
+	char *formatted; /* cached JS literal text; NULL if not yet formatted */
+	u32 *xref_addrs; /* instruction addresses that construct this literal */
+	u32 xref_count;
+} HBCLiteralEntry;
+
+/* Format a literal directly from raw parameters. Does not touch the cache.
+ * Returned string is heap-allocated; caller frees with free(). */
+Result hbc_literals_format_raw(HBC *hbc, HBCLiteralKind kind, u32 num_items,
+	u32 primary_id, u32 secondary_id, char **out);
+
+/* Convenience wrapper taking opcode + raw arg3/arg4/arg5 as decoded by
+ * hbc_dec(). Knows the per-opcode mapping; useful as the "give me the string
+ * for these parameters" helper. */
+Result hbc_literals_format_for_opcode(HBC *hbc, u8 opcode, u32 arg3, u32 arg4,
+	u32 arg5, char **out);
+
+/* Look up (or create) a cache entry for a literal and return the formatted
+ * text. The returned pointer is owned by the cache — do not free it. */
+Result hbc_literals_get(HBC *hbc, HBCLiteralKind kind, u32 num_items,
+	u32 primary_id, u32 secondary_id, const char **out_formatted);
+
+/* Register that `from_addr` references the literal described by the key.
+ * Creates the cache entry (and formats) on first reference. */
+Result hbc_literals_register(HBC *hbc, HBCLiteralKind kind, u32 num_items,
+	u32 primary_id, u32 secondary_id, u32 from_addr);
+
+/* Full-binary scan: walk every function body; record every
+ * New{Array,Object}WithBuffer[Long] call site into the cache (with xrefs).
+ * Returns the number of distinct literals found in *out_count. */
+Result hbc_literals_scan_code(HBC *hbc, u32 *out_count);
+
+/* Pool-side scan: enumerate group boundaries in the SLP pool for the given
+ * kind. For arrays and object-values it walks the byte stream linearly.
+ * For object-keys on v97+ it enumerates the shape table instead.
+ * Does not populate the cache — this is an inspection helper. Caller frees
+ * *out via free(). */
+typedef struct {
+	u32 paddr; /* absolute file offset */
+	u32 pool_offset; /* byte offset within the pool */
+	u32 num_items; /* number of values in this group (or shape prop_count) */
+	u8 tag; /* SLP tag 0..7 (for arrays/values); 0xff if not applicable */
+} HBCPoolGroup;
+
+Result hbc_literals_scan_pool(HBC *hbc, HBCLiteralKind kind,
+	HBCPoolGroup **out, u32 *out_count);
+
+/* Enumerate the current cache contents. The returned array is owned by the
+ * HBC and valid until the next cache mutation. */
+Result hbc_literals_list(HBC *hbc, const HBCLiteralEntry **out,
+	u32 *out_count);
+
+/* Drop every cached entry and every xref. Fast. Does not affect the SLP
+ * pools themselves. */
+void hbc_literals_reset(HBC *hbc);
+
+/* Number of cached entries. 0 before any register / scan call. */
+u32 hbc_literals_count(HBC *hbc);
+
+/* Absolute file offset of the named pool, or 0 if the pool is empty/absent. */
+u32 hbc_get_pool_paddr(HBC *hbc, HBCLiteralKind kind);
+u32 hbc_get_pool_size(HBC *hbc, HBCLiteralKind kind);
+
+/* Value pool paddr/size for objects (arrays & object-keys reuse the kind-based
+ * accessors; object values is a distinct pool). */
+u32 hbc_get_object_values_paddr(HBC *hbc);
+u32 hbc_get_object_values_size(HBC *hbc);
+
+/* Global toggle: whether hbc_dec() should inline the formatted literal as a
+ * trailing " ; <text>" comment on the disasm line. Default: off. */
+void hbc_set_inline_literals(bool on);
+bool hbc_get_inline_literals(void);
+
+#endif /* LIBHBC_LITERALS_PUBLIC_H */

include/hbc/parser.h

@@ -239,6 +239,11 @@ struct HBCReader {
 	ShapeTableEntry *object_shapes;
 	size_t object_shape_count;
 
+	/* Absolute file offsets of the SLP pools (0 if the pool is absent). */
+	u32 arrays_paddr;
+	u32 object_keys_paddr;
+	u32 object_values_paddr;
+
 	/* BigInt data */
 	i64 *bigint_values;
 	size_t bigint_count;
@@ -258,11 +263,21 @@ struct HBCReader {
 	FunctionSourceEntry *function_sources;
 	size_t function_source_count;
 
-	/* Debug info */
+	/* Debug info — split into two phases:
+	 *   debug_info_header_loaded: cheap parse of the DebugInfoHeader fields
+	 *     only (a handful of u32s), populated eagerly at load time so callers
+	 *     can report accurate "has source lines" / "stripped" without paying
+	 *     for the full parse.
+	 *   debug_info_loaded: full parse of filenames, file regions, and the
+	 *     sources / scope / textified / string-table storage blobs. Populated
+	 *     lazily on first hbc_get_source_lines / hbc_get_debug_info call. */
+	bool debug_info_header_loaded;
+	bool debug_info_loaded;
 	DebugInfoHeader debug_info_header;
 	OffsetLengthPair *debug_string_table;
 	u8 *debug_string_storage;
 	size_t debug_string_storage_size;
+	char **debug_filenames;
 	DebugFileRegion *debug_file_regions;
 	size_t debug_file_region_count;
 	u8 *sources_data_storage;
@@ -293,6 +308,7 @@ Result _hbc_reader_read_bigints(HBCReader *reader);
 Result _hbc_reader_read_regexp(HBCReader *reader);
 Result _hbc_reader_read_cjs_modules(HBCReader *reader);
 Result _hbc_reader_read_function_sources(HBCReader *reader);
+Result _hbc_reader_read_debug_info_header(HBCReader *reader);
 Result _hbc_reader_read_debug_info(HBCReader *reader);
 Result _hbc_reader_read_whole_file(HBCReader *reader, const char *filename);
 

meson.build

@@ -1,7 +1,7 @@
 project(
   'r2hermes',
   'c',
-  version: '1.2.4',
+  version: '1.2.6',
   license: 'MIT',
   default_options: [
     'c_std=c11',
@@ -67,6 +67,7 @@ lib_opcodes_src = files(
 
 lib_core_src = files(
   'src/lib/hbc.c',
+  'src/lib/literals_api.c',
   'src/lib/r2.c'
 )
 

r2plugin/meson.build

@@ -1,66 +1,26 @@
-# Version
-# XXX reuse from parent meson.build or wtf
-version = '1.0.0'
+# r2hermes XPS integration for radare2.
+# Included from libr/xps/meson.build when r2hermes sources live under
+# libr/xps/p/r2hermes/. Appends sources and include paths to the r_core
+# aggregate so the static plugin is linked into libr_core.
 
-# Version extraction
-version_parts = version.split('.')
-version_major = version_parts[0]
-version_minor = version_parts.length() > 1 ? version_parts[1] : '0'
-version_patch = version_parts.length() > 2 ? version_parts[2] : '0'
-
-# Generate version.h
-configure_file(
-  input: '../include/hbc/version.h.in',
-  output: 'version.h',
-  configuration: {
-    'VERSION': version,
-    'VERSION_MAJOR': version_major,
-    'VERSION_MINOR': version_minor,
-    'VERSION_PATCH': version_patch,
-  },
-  install_dir: 'libr/hbc',
-  install: true,
-)
-
-# Register plugins in the static plugin lists
 core_plugins += ['r2hermes']
 
-# Include directories (relative to this meson.build file location)
-include_dirs = include_directories(join_paths('..', 'include'))
-r_core_additional_inc += include_dirs
+r_core_additional_inc += include_directories(join_paths('..', 'include'))
 
-# Library source files
-lib_utils_src = files(
+r_core_additional_sources += files(
   '../src/lib/utils/string_buffer.c',
   '../src/lib/utils/buffer_reader.c',
-)
-
-lib_parsers_src = files(
   '../src/lib/parsers/hbc_file_parser.c',
   '../src/lib/parsers/hbc_bytecode_parser.c',
-)
-
-lib_disasm_src = []  # Disassembly sources (if any)
-
-lib_decompile_src = files(
   '../src/lib/decompilation/translator.c',
   '../src/lib/decompilation/token.c',
   '../src/lib/decompilation/literals.c',
   '../src/lib/decompilation/decompiler.c',
-)
-
-lib_opcodes_src = files(
   '../src/lib/opcodes/isa.c',
   '../src/lib/opcodes/encoder.c',
   '../src/lib/opcodes/decoder.c',
-)
-
-lib_core_src = files(
   '../src/lib/hbc.c',
+  '../src/lib/literals_api.c',
   '../src/lib/r2.c',
   '../src/r2/core_hbc_one.c',
 )
-
-lib_src = lib_utils_src + lib_parsers_src + lib_disasm_src + lib_decompile_src + lib_opcodes_src + lib_core_src
-
-r_core_additional_sources += lib_src

src/lib/decompilation/decompiler.c

@@ -149,7 +149,7 @@ static bool u32set_contains(const U32Set *s, u32 v) {
 	if (!s || !s->bitmap || v / 8 >= s->bitmap_size) {
 		return false;
 	}
-	return (s->bitmap[v / 8] &(1 << (v % 8))) != 0;
+	return (s->bitmap[v / 8] & (1 << (v % 8))) != 0;
 }
 static Result u32set_add(U32Set *s, u32 v) {
 	if (!s) {
@@ -710,6 +710,8 @@ Result _hbc_decompiler_init(HermesDecompiler *decompiler) {
 	decompiler->function_in_progress = NULL;
 	decompiler->indent_level = 0;
 	decompiler->inlining_function = false;
+	decompiler->output_truncated = false;
+	decompiler->truncation_marker_emitted = false;
 	decompiler->hbc = NULL; /* Will be set if using state-based API */
 	decompiler->options.pretty_literals = true;
 	decompiler->options.suppress_comments = false;
@@ -883,6 +885,9 @@ Result _hbc_decompile_all_to_buffer(HBCReader *reader, HBCDecompOptions options,
 		if (dec.decompiled_functions[i]) {
 			continue;
 		}
+		if (dec.output_truncated) {
+			break;
+		}
 		Result r = _hbc_decompile_function (&dec, i, NULL, -1, false, false, false);
 		if (r.code != RESULT_SUCCESS) {
 			_hbc_decompiler_cleanup (&dec);
@@ -1090,6 +1095,9 @@ Result _hbc_decompile_all_with_state(HBC *hbc, HBCDecompOptions options, StringB
 		if (dec.decompiled_functions[i]) {
 			continue;
 		}
+		if (dec.output_truncated) {
+			break;
+		}
 		Result r = _hbc_decompile_function (&dec, i, NULL, -1, false, false, false);
 		if (r.code != RESULT_SUCCESS) {
 			_hbc_decompiler_cleanup (&dec);
@@ -1428,8 +1436,13 @@ Result _hbc_pass2_transform_code(HermesDecompiler *state, DecompiledFunctionBody
 	if (!state || !function_body) {
 		return ERROR_RESULT (RESULT_ERROR_INVALID_ARGUMENT, "_hbc_pass2_transform_code args");
 	}
-	/* Translate each parsed instruction into a TokenString line */
+	int ast_budget = state->options.max_ast_statements;
 	for (u32 i = 0; i < function_body->instructions.count; i++) {
+		/* AST-size cap: bail mid-build so pass3/4 and output stay cheap. */
+		if (ast_budget > 0 && function_body->statements_count >= (u32)ast_budget) {
+			state->output_truncated = true;
+			break;
+		}
 		ParsedInstruction *ins = &function_body->instructions.instructions[i];
 		if (!ins->inst) {
 			continue;
@@ -1497,14 +1510,16 @@ Result _hbc_pass3_parse_forin_loops(HermesDecompiler *state, DecompiledFunctionB
 		if (!filni || function_body->statements[other].head->type != TOKEN_TYPE_FOR_IN_LOOP_NEXT_ITER) {
 			continue;
 		}
+		int next_value_register = filni->next_value_register;
+		int obj_register = fili->obj_register;
 
 		/* Replace GetPNameList line with a `for (<next> in <obj>)` header */
 		RETURN_IF_ERROR (token_string_clear_tokens (line));
 		RETURN_IF_ERROR (_hbc_token_string_add_token (line, create_raw_token ("for")));
 		RETURN_IF_ERROR (_hbc_token_string_add_token (line, create_left_parenthesis_token ()));
-		RETURN_IF_ERROR (_hbc_token_string_add_token (line, create_left_hand_reg_token (filni->next_value_register)));
+		RETURN_IF_ERROR (_hbc_token_string_add_token (line, create_left_hand_reg_token (next_value_register)));
 		RETURN_IF_ERROR (_hbc_token_string_add_token (line, create_raw_token ("in")));
-		RETURN_IF_ERROR (_hbc_token_string_add_token (line, create_right_hand_reg_token (fili->obj_register)));
+		RETURN_IF_ERROR (_hbc_token_string_add_token (line, create_right_hand_reg_token (obj_register)));
 		RETURN_IF_ERROR (_hbc_token_string_add_token (line, create_right_parenthesis_token ()));
 
 		/* Silence the loop plumbing instructions */
@@ -1579,6 +1594,7 @@ Result _hbc_pass4_name_closure_vars(HermesDecompiler *state, DecompiledFunctionB
 				}
 			} else if (tok->type == TOKEN_TYPE_STORE_TO_ENVIRONMENT) {
 				StoreToEnvironmentToken *t = (StoreToEnvironmentToken *)tok;
+				int value_register = t->value_register;
 				Environment *env = envmap_get (function_body, t->env_register);
 				if (!env) {
 					continue;
@@ -1600,7 +1616,7 @@ Result _hbc_pass4_name_closure_vars(HermesDecompiler *state, DecompiledFunctionB
 				}
 				RETURN_IF_ERROR (_hbc_token_string_add_token (line, create_raw_token (existing)));
 				RETURN_IF_ERROR (_hbc_token_string_add_token (line, create_assignment_token ()));
-				RETURN_IF_ERROR (_hbc_token_string_add_token (line, create_right_hand_reg_token (t->value_register)));
+				RETURN_IF_ERROR (_hbc_token_string_add_token (line, create_right_hand_reg_token (value_register)));
 				break;
 			} else if (tok->type == TOKEN_TYPE_LOAD_FROM_ENVIRONMENT) {
 				LoadFromEnvironmentToken *t = (LoadFromEnvironmentToken *)tok;
@@ -1876,6 +1892,17 @@ Result _hbc_output_code(HermesDecompiler *state, DecompiledFunctionBody *functio
 		(unsigned long long)state->options.function_base,
 		function_body->statements_count);
 	for (u32 si = 0; si < function_body->statements_count; si++) {
+		/* Stop when budget exhausted or earlier phase (pass2) flagged it. */
+		if (state->output_truncated || (state->options.max_output_bytes > 0 && out->length >= (size_t)state->options.max_output_bytes)) {
+			state->output_truncated = true;
+			if (!state->truncation_marker_emitted) {
+				state->truncation_marker_emitted = true;
+				append_indent (out, state->indent_level);
+				_hbc_string_buffer_append (out,
+					"// [output truncated: raise 'hbc.max_ast'/'hbc.max_bytes' (0=unlimited) for full output]\n");
+			}
+			break;
+		}
 		/* Skip dead code */
 		if (dce && dce[si]) {
 			continue;
@@ -2193,6 +2220,9 @@ Result _hbc_decompile_function(HermesDecompiler *state, u32 function_id, Environ
 	if (!state || !state->hbc_reader) {
 		return ERROR_RESULT (RESULT_ERROR_INVALID_ARGUMENT, "_hbc_decompile_function args");
 	}
+	if (state->output_truncated) {
+		return SUCCESS_RESULT ();
+	}
 	HBCReader *reader = state->hbc_reader;
 	if (function_id >= reader->header.functionCount) {
 		return ERROR_RESULT (RESULT_ERROR_INVALID_ARGUMENT, "_hbc_decompile_function bad id");

src/lib/decompilation/literals.c

@@ -91,7 +91,7 @@ static bool slp_read_u16(const u8 *base, size_t size, size_t *pos, u16 *out) {
 	if (*pos + 2 > size) {
 		return false;
 	}
-	*out = (u16) (base[*pos] | (base[*pos + 1] << 8));
+	*out = (u16) ((u16)base[*pos] | ((u16)base[*pos + 1] << 8));
 	(*pos) += 2;
 	return true;
 }
@@ -99,7 +99,7 @@ static bool slp_read_u32(const u8 *base, size_t size, size_t *pos, u32 *out) {
 	if (*pos + 4 > size) {
 		return false;
 	}
-	*out = (u32) (base[*pos] | (base[*pos + 1] << 8) | (base[*pos + 2] << 16) | (base[*pos + 3] << 24));
+	*out = (u32)base[*pos] | ((u32)base[*pos + 1] << 8) | ((u32)base[*pos + 2] << 16) | ((u32)base[*pos + 3] << 24);
 	(*pos) += 4;
 	return true;
 }
@@ -216,8 +216,8 @@ static Result slp_append_value_js(HBCReader *r, const SLPValue *v, bool for_key,
 	case SLP_ShortStringTag:
 	case SLP_ByteStringTag:
 		{
-		const char *s = (r && r->strings && v->v.str_id < r->header.stringCount)? r->strings[v->v.str_id]: NULL;
-		if (for_key && s && _hbc_is_js_identifier (s)) {
+			const char *s = (r && r->strings && v->v.str_id < r->header.stringCount)? r->strings[v->v.str_id]: NULL;
+			if (for_key && s && _hbc_is_js_identifier (s)) {
 				return _hbc_string_buffer_append (out, s);
 			}
 			if (s) {

src/lib/decompilation/token.c

@@ -326,10 +326,10 @@ Result _hbc_token_to_string(Token *token, StringBuffer *buffer) {
 		{
 			FunctionTableIndexToken *t = (FunctionTableIndexToken *)token;
 			/* Try to resolve name if possible */
-		if (t->state && t->state->hbc_reader && t->function_id < t->state->hbc_reader->header.functionCount) {
+			if (t->state && t->state->hbc_reader && t->function_id < t->state->hbc_reader->header.functionCount) {
 				u32 name_id = t->state->hbc_reader->function_headers[t->function_id].functionName;
-			const char *name = (name_id < t->state->hbc_reader->header.stringCount && t->state->hbc_reader->strings)? t->state->hbc_reader->strings[name_id]: NULL;
-			if (name && *name) {
+				const char *name = (name_id < t->state->hbc_reader->header.stringCount && t->state->hbc_reader->strings)? t->state->hbc_reader->strings[name_id]: NULL;
+				if (name && *name) {
 					RETURN_IF_ERROR (_hbc_string_buffer_append (buffer, name));
 					return SUCCESS_RESULT ();
 				}

src/lib/decompilation/translator.c

@@ -220,8 +220,10 @@ static Result emit_call_fixed(TokenString *out, const ParsedInstruction *insn, i
 	RETURN_IF_ERROR (add (out, reg_l_safe (insn, 0))); /* dest */
 	RETURN_IF_ERROR (add (out, create_assignment_token ()));
 	RETURN_IF_ERROR (add (out, reg_r_safe (insn, 1))); /* callee */
+	/* Call1..Call4 encode an explicit `this` register before the user arguments. */
+	RETURN_IF_ERROR (add (out, create_raw_token (".call")));
 	RETURN_IF_ERROR (add (out, create_left_parenthesis_token ()));
-	for (int i = 2; i < operand_count; i++) { /* args start at operand 2 */
+	for (int i = 2; i < operand_count; i++) {
 		if (i > 2) {
 			RETURN_IF_ERROR (add (out, create_raw_token (",")));
 		}
@@ -574,7 +576,7 @@ Result _hbc_translate_instruction_to_tokens(const ParsedInstruction *insn_c, Tok
 		}
 	case OP_CreateInnerEnvironment:
 		{
-		if (is_operand_register (insn->inst, 0) && is_operand_register (insn->inst, 1)) {
+			if (is_operand_register (insn->inst, 0) && is_operand_register (insn->inst, 1)) {
 				RETURN_IF_ERROR (add (out, create_new_inner_environment_token ((int)get_operand_value (insn_c, 0), (int)get_operand_value (insn_c, 1), (int)get_operand_value (insn_c, 2))));
 			} else {
 				RETURN_IF_ERROR (add (out, create_raw_token ("/*new_inner_env_invalid*/")));
@@ -643,18 +645,18 @@ Result _hbc_translate_instruction_to_tokens(const ParsedInstruction *insn_c, Tok
 			RETURN_IF_ERROR (add (out, reg_r_safe (insn_c, 1)));
 			u32 sid = insn->arg4;
 			const char *s = NULL;
-		if (insn->hbc_reader && insn->hbc_reader->strings && sid < insn->hbc_reader->header.stringCount) {
+			if (insn->hbc_reader && insn->hbc_reader->strings && sid < insn->hbc_reader->header.stringCount) {
 				s = insn->hbc_reader->strings[sid];
 			}
 			bool ident = true;
-		if (!s || !*s) {
+			if (!s || !*s) {
 				ident = false;
 			} else {
-			if (! (isalpha ((unsigned char)*s) || *s == '_' || *s == '$')) {
+				if (! (isalpha ((unsigned char)*s) || *s == '_' || *s == '$')) {
 					ident = false;
 				}
-			for (const char *p = s + 1; ident && *p; p++) {
-				ident = (isalnum ((unsigned char)*p) || *p == '_' || *p == '$');
+				for (const char *p = s + 1; ident && *p; p++) {
+					ident = (isalnum ((unsigned char)*p) || *p == '_' || *p == '$');
 				}
 			}
 			if (ident) {
@@ -707,8 +709,8 @@ Result _hbc_translate_instruction_to_tokens(const ParsedInstruction *insn_c, Tok
 			RETURN_IF_ERROR (add (out, reg_r_safe (insn_c, 0)));
 			u32 sid = (op == OP_PutNewOwnByIdShort)? insn->arg3: (op == OP_PutNewOwnById)? insn->arg3
 												: insn->arg4;
-		const char *s = (insn->hbc_reader && insn->hbc_reader->strings && sid < insn->hbc_reader->header.stringCount)? insn->hbc_reader->strings[sid]: NULL;
-		if (s && _hbc_is_js_identifier (s)) {
+			const char *s = (insn->hbc_reader && insn->hbc_reader->strings && sid < insn->hbc_reader->header.stringCount)? insn->hbc_reader->strings[sid]: NULL;
+			if (s && _hbc_is_js_identifier (s)) {
 				RETURN_IF_ERROR (add (out, create_dot_accessor_token ()));
 				RETURN_IF_ERROR (add (out, create_raw_token (s)));
 			} else {
@@ -809,7 +811,7 @@ Result _hbc_translate_instruction_to_tokens(const ParsedInstruction *insn_c, Tok
 	case OP_ResumeGenerator:
 		{
 			/* args: result_out, return_bool_out */
-		if (is_operand_register (insn->inst, 0) && is_operand_register (insn->inst, 1)) {
+			if (is_operand_register (insn->inst, 0) && is_operand_register (insn->inst, 1)) {
 				RETURN_IF_ERROR (add (out, create_resume_generator_token ((int)insn->arg1, (int)insn->arg2)));
 			} else {
 				RETURN_IF_ERROR (add (out, create_raw_token ("/*resume_gen_invalid*/")));
@@ -835,7 +837,7 @@ Result _hbc_translate_instruction_to_tokens(const ParsedInstruction *insn_c, Tok
 	case OP_StoreToEnvironment:
 	case OP_StoreToEnvironmentL:
 		{
-		if (is_operand_register (insn->inst, 0) && is_operand_register (insn->inst, 2)) {
+			if (is_operand_register (insn->inst, 0) && is_operand_register (insn->inst, 2)) {
 				RETURN_IF_ERROR (add (out, create_store_to_environment_token ((int)get_operand_value (insn_c, 0), (int)get_operand_value (insn_c, 1), (int)get_operand_value (insn_c, 2))));
 			} else {
 				RETURN_IF_ERROR (add (out, create_raw_token ("/*env_store_invalid*/")));
@@ -878,7 +880,7 @@ Result _hbc_translate_instruction_to_tokens(const ParsedInstruction *insn_c, Tok
 				return r;
 			}
 			r = _hbc_format_object_literal (insn->hbc_reader, insn->arg2, insn->arg3, insn->arg4, insn->arg5, &sb, LITERALS_PRETTY_AUTO, false);
-		Token *t = create_raw_token ((r.code == RESULT_SUCCESS && sb.data)? sb.data: "{ /*object*/ }");
+			Token *t = create_raw_token ((r.code == RESULT_SUCCESS && sb.data)? sb.data: "{ /*object*/ }");
 			_hbc_string_buffer_free (&sb);
 			if (!t) {
 				return ERROR_RESULT (RESULT_ERROR_MEMORY_ALLOCATION, "oom");
@@ -897,7 +899,7 @@ Result _hbc_translate_instruction_to_tokens(const ParsedInstruction *insn_c, Tok
 				return r;
 			}
 			r = _hbc_format_array_literal (insn->hbc_reader, insn->arg3, insn->arg4, &sb, LITERALS_PRETTY_AUTO, false);
-		Token *t = create_raw_token ((r.code == RESULT_SUCCESS && sb.data)? sb.data: "[ /*array*/ ]");
+			Token *t = create_raw_token ((r.code == RESULT_SUCCESS && sb.data)? sb.data: "[ /*array*/ ]");
 			_hbc_string_buffer_free (&sb);
 			if (!t) {
 				return ERROR_RESULT (RESULT_ERROR_MEMORY_ALLOCATION, "oom");
@@ -926,10 +928,10 @@ Result _hbc_translate_instruction_to_tokens(const ParsedInstruction *insn_c, Tok
 			RETURN_IF_ERROR (add (out, reg_r_safe (insn_c, 0)));
 			u32 sid = insn->arg2;
 			const char *s = NULL;
-		if (insn->hbc_reader && insn->hbc_reader->strings && sid < insn->hbc_reader->header.stringCount) {
+			if (insn->hbc_reader && insn->hbc_reader->strings && sid < insn->hbc_reader->header.stringCount) {
 				s = insn->hbc_reader->strings[sid];
 			}
-		if (s && _hbc_is_js_identifier (s)) {
+			if (s && _hbc_is_js_identifier (s)) {
 				RETURN_IF_ERROR (add (out, create_dot_accessor_token ()));
 				RETURN_IF_ERROR (add (out, create_raw_token (s)));
 			} else {
@@ -1096,11 +1098,14 @@ Result _hbc_translate_instruction_to_tokens(const ParsedInstruction *insn_c, Tok
 		{
 			RETURN_IF_ERROR (add (out, reg_l_safe (insn_c, 0)));
 			RETURN_IF_ERROR (add (out, create_assignment_token ()));
-			RETURN_IF_ERROR (add (out, create_raw_token ("fn_")));
-			RETURN_IF_ERROR (add (out, num_token_u32 (insn->arg3)));
+			Token *t = create_function_table_index_token (get_operand_value (insn_c, 2), NULL);
+			if (!t) {
+				return ERROR_RESULT (RESULT_ERROR_MEMORY_ALLOCATION, "oom fti");
+			}
+			RETURN_IF_ERROR (add (out, t));
 			RETURN_IF_ERROR (add (out, create_left_parenthesis_token ()));
 			RETURN_IF_ERROR (add (out, create_raw_token ("/*argc:")));
-			RETURN_IF_ERROR (add (out, num_token_u32 (insn->arg2)));
+			RETURN_IF_ERROR (add (out, num_token_u32 (get_operand_value (insn_c, 1))));
 			RETURN_IF_ERROR (add (out, create_raw_token ("*/")));
 			RETURN_IF_ERROR (add (out, create_right_parenthesis_token ()));
 			break;
@@ -1161,10 +1166,10 @@ Result _hbc_translate_instruction_to_tokens(const ParsedInstruction *insn_c, Tok
 			RETURN_IF_ERROR (add (out, reg_r_safe (insn_c, 0)));
 			u32 sid = insn->arg3;
 			const char *s = NULL;
-		if (insn->hbc_reader && insn->hbc_reader->strings && sid < insn->hbc_reader->header.stringCount) {
+			if (insn->hbc_reader && insn->hbc_reader->strings && sid < insn->hbc_reader->header.stringCount) {
 				s = insn->hbc_reader->strings[sid];
 			}
-		if (s && _hbc_is_js_identifier (s)) {
+			if (s && _hbc_is_js_identifier (s)) {
 				RETURN_IF_ERROR (add (out, create_dot_accessor_token ()));
 				RETURN_IF_ERROR (add (out, create_raw_token (s)));
 			} else {
@@ -1182,10 +1187,10 @@ Result _hbc_translate_instruction_to_tokens(const ParsedInstruction *insn_c, Tok
 			RETURN_IF_ERROR (add (out, reg_r_safe (insn_c, 0)));
 			u32 sid = (op == OP_PutNewOwnNEById)? insn->arg3: insn->arg4;
 			const char *s = NULL;
-		if (insn->hbc_reader && insn->hbc_reader->strings && sid < insn->hbc_reader->header.stringCount) {
+			if (insn->hbc_reader && insn->hbc_reader->strings && sid < insn->hbc_reader->header.stringCount) {
 				s = insn->hbc_reader->strings[sid];
 			}
-		if (s && _hbc_is_js_identifier (s)) {
+			if (s && _hbc_is_js_identifier (s)) {
 				RETURN_IF_ERROR (add (out, create_dot_accessor_token ()));
 				RETURN_IF_ERROR (add (out, create_raw_token (s)));
 			} else {
@@ -1223,8 +1228,8 @@ Result _hbc_translate_instruction_to_tokens(const ParsedInstruction *insn_c, Tok
 	/* Property list operations */
 	case OP_GetPNameList:
 		{
-		if (is_operand_register (insn->inst, 0) && is_operand_register (insn->inst, 1) &&
-			is_operand_register (insn->inst, 2) && is_operand_register (insn->inst, 3)) {
+			if (is_operand_register (insn->inst, 0) && is_operand_register (insn->inst, 1) &&
+				is_operand_register (insn->inst, 2) && is_operand_register (insn->inst, 3)) {
 				RETURN_IF_ERROR (add (out, create_for_in_loop_init_token ((int)get_operand_value (insn_c, 0), (int)get_operand_value (insn_c, 1), (int)get_operand_value (insn_c, 2), (int)get_operand_value (insn_c, 3))));
 			} else {
 				RETURN_IF_ERROR (add (out, create_raw_token ("/*forin_init_invalid*/")));
@@ -1233,8 +1238,8 @@ Result _hbc_translate_instruction_to_tokens(const ParsedInstruction *insn_c, Tok
 		}
 	case OP_GetNextPName:
 		{
-		if (is_operand_register (insn->inst, 0) && is_operand_register (insn->inst, 1) &&
-			is_operand_register (insn->inst, 2) && is_operand_register (insn->inst, 3) &&
+			if (is_operand_register (insn->inst, 0) && is_operand_register (insn->inst, 1) &&
+				is_operand_register (insn->inst, 2) && is_operand_register (insn->inst, 3) &&
 				is_operand_register (insn->inst, 4)) {
 				RETURN_IF_ERROR (add (out, create_for_in_loop_next_iter_token ((int)get_operand_value (insn_c, 0), (int)get_operand_value (insn_c, 1), (int)get_operand_value (insn_c, 2), (int)get_operand_value (insn_c, 3), (int)get_operand_value (insn_c, 4))));
 			} else {
@@ -1300,7 +1305,7 @@ Result _hbc_translate_instruction_to_tokens(const ParsedInstruction *insn_c, Tok
 		}
 	default:
 		{
-		if (!insn->inst->name || strcmp (insn->inst->name, "Unknown") == 0) {
+			if (!insn->inst->name || strcmp (insn->inst->name, "Unknown") == 0) {
 				char buf[96];
 				u32 skipped = insn->arg1;
 				if (skipped) {
@@ -1318,7 +1323,7 @@ Result _hbc_translate_instruction_to_tokens(const ParsedInstruction *insn_c, Tok
 
 			/* Generic compare-jump fallback (covers variants not explicitly handled above). */
 			const char *cmp = jump_cmp_operator (op);
-		if (cmp && is_operand_addr (insn->inst, 0) && is_operand_register (insn->inst, 1) && is_operand_register (insn->inst, 2)) {
+			if (cmp && is_operand_addr (insn->inst, 0) && is_operand_register (insn->inst, 1) && is_operand_register (insn->inst, 2)) {
 				i32 rel = (i32)get_operand_value (insn_c, 0);
 				u32 target = compute_target_address (insn, 0);
 				if (rel > 0) {

src/lib/hbc.c

@@ -3,6 +3,7 @@
 #include <hbc/disasm.h>
 #include <hbc/decompilation/decompiler.h>
 #include <hbc/decompilation/literals.h>
+#include <hbc/literals.h>
 #include <hbc/hermes_encoder.h>
 #include <hbc/parser.h>
 #include <hbc/bytecode.h>
@@ -117,7 +118,16 @@ Result hbc_open_from_memory(const u8 *data, size_t size, HBC **out) {
 		hbc_close (hbc);
 		return res;
 	}
-	(void)_hbc_reader_read_debug_info;
+	/* Read only the DebugInfoHeader eagerly — a handful of u32s — so
+	 * hbc_has_source_lines() can answer accurately without committing to the
+	 * full body parse. The filename, file region and sources/scope/textified
+	 * storage blobs are read lazily on first hbc_get_source_lines /
+	 * hbc_get_debug_info call. */
+	res = _hbc_reader_read_debug_info_header (&hbc->reader);
+	if (res.code != RESULT_SUCCESS) {
+		hbc_close (hbc);
+		return res;
+	}
 
 	*out = hbc;
 	return SUCCESS_RESULT ();
@@ -127,6 +137,7 @@ void hbc_close(HBC *hbc) {
 	if (!hbc) {
 		return;
 	}
+	hbc_literals_reset (hbc);
 	_hbc_reader_cleanup (&hbc->reader);
 	free (hbc);
 }
@@ -237,10 +248,238 @@ Result hbc_get_function_source(HBC *hbc, u32 function_id, const char **out_src)
 		return ERROR_RESULT (RESULT_ERROR_INVALID_ARGUMENT, "Function ID out of range");
 	}
 	*out_src = NULL;
-	if (hbc->reader.function_sources && function_id < hbc->reader.header.functionSourceCount) {
-		u32 string_id = hbc->reader.function_sources[function_id].string_id;
-		if (string_id < hbc->reader.header.stringCount) {
-			*out_src = hbc->reader.strings[string_id];
+	if (hbc->reader.function_sources) {
+		for (u32 i = 0; i < hbc->reader.header.functionSourceCount; i++) {
+			if (hbc->reader.function_sources[i].function_id != function_id) {
+				continue;
+			}
+			u32 string_id = hbc->reader.function_sources[i].string_id;
+			if (string_id < hbc->reader.header.stringCount) {
+				*out_src = hbc->reader.strings[string_id];
+			}
+			break;
+		}
+	}
+	return SUCCESS_RESULT ();
+}
+
+static bool read_sleb128(const u8 *data, size_t size, size_t *pos, int64_t *out) {
+	uint64_t result = 0;
+	unsigned shift = 0;
+	u8 byte = 0;
+
+	do {
+		if (*pos >= size || shift > 63) {
+			return false;
+		}
+		byte = data[(*pos)++];
+		if (shift == 63 && (byte & 0x7e)) {
+			return false;
+		}
+		result |= ((uint64_t)(byte & 0x7f)) << shift;
+		shift += 7;
+	} while (byte & 0x80);
+
+	if ((shift < 64) && (byte & 0x40)) {
+		result |= (~0ULL) << shift;
+	}
+	*out = (int64_t)result;
+	return true;
+}
+
+static const char *debug_filename_at(HBCReader *reader, u32 index) {
+	const u32 count = reader->debug_info_header.filename_count;
+	if (index >= count || !reader->debug_string_table || !reader->debug_string_storage) {
+		return NULL;
+	}
+	if (!reader->debug_filenames) {
+		reader->debug_filenames = (char **)calloc (count, sizeof (char *));
+		if (!reader->debug_filenames) {
+			return NULL;
+		}
+	}
+	if (!reader->debug_filenames[index]) {
+		const OffsetLengthPair *entry = &reader->debug_string_table[index];
+		const size_t storage_size = reader->debug_string_storage_size;
+		if ((size_t)entry->offset > storage_size || (size_t)entry->length > storage_size - entry->offset) {
+			return NULL;
+		}
+		char *filename = (char *)malloc ((size_t)entry->length + 1);
+		if (!filename) {
+			return NULL;
+		}
+		memcpy (filename, reader->debug_string_storage + entry->offset, entry->length);
+		filename[entry->length] = '\0';
+		reader->debug_filenames[index] = filename;
+	}
+	return reader->debug_filenames[index];
+}
+
+static const char *debug_filename_for_source_offset(HBCReader *reader, u32 source_offset) {
+	const DebugFileRegion *best = NULL;
+	for (u32 i = 0; i < reader->debug_file_region_count; i++) {
+		const DebugFileRegion *region = &reader->debug_file_regions[i];
+		if (region->from_address <= source_offset && (!best || region->from_address >= best->from_address)) {
+			best = region;
+		}
+	}
+	return best? debug_filename_at (reader, best->filename_id): NULL;
+}
+
+static Result append_source_line(HBCSourceLineArray *arr, u32 *cap, const HBCSourceLine *line) {
+	if (arr->count >= *cap) {
+		u32 next = *cap? (*cap * 2): 64;
+		HBCSourceLine *lines = (HBCSourceLine *)realloc (arr->lines, (size_t)next * sizeof (HBCSourceLine));
+		if (!lines) {
+			return ERROR_RESULT (RESULT_ERROR_MEMORY_ALLOCATION, "Failed to allocate source lines");
+		}
+		arr->lines = lines;
+		*cap = next;
+	}
+	arr->lines[arr->count++] = *line;
+	return SUCCESS_RESULT ();
+}
+
+#define PARSE_FAIL(msg) do { res = ERROR_RESULT (RESULT_ERROR_PARSING_FAILED, msg); goto fail; } while (0)
+
+Result hbc_get_source_lines(HBC *hbc, HBCSourceLineArray *out) {
+	if (!hbc || !out) {
+		return ERROR_RESULT (RESULT_ERROR_INVALID_ARGUMENT, "Invalid arguments");
+	}
+	memset (out, 0, sizeof (*out));
+	HBCReader *reader = &hbc->reader;
+	RETURN_IF_ERROR (_hbc_reader_read_debug_info (reader));
+	const u8 *data = reader->sources_data_storage;
+	const size_t size = reader->sources_data_storage_size;
+	if (!data || !size) {
+		return SUCCESS_RESULT ();
+	}
+
+	Result res = SUCCESS_RESULT ();
+	u32 cap = 0;
+	size_t pos = 0;
+	while (pos < size) {
+		const size_t source_offset = pos;
+		int64_t function_id = 0, line = 0, column = 0;
+		if (source_offset > UINT32_MAX
+				|| !read_sleb128 (data, size, &pos, &function_id)
+				|| !read_sleb128 (data, size, &pos, &line)
+				|| !read_sleb128 (data, size, &pos, &column)) {
+			PARSE_FAIL ("Invalid source location header");
+		}
+		if (function_id < 0 || function_id >= (int64_t)reader->header.functionCount
+				|| line < 0 || line > UINT32_MAX || column < 0 || column > UINT32_MAX) {
+			PARSE_FAIL ("Source location header out of range");
+		}
+
+		HBCFunc fi = { 0 };
+		(void)hbc_get_function_info (hbc, (u32)function_id, &fi);
+		HBCSourceLine sl = {
+			.function_id = (u32)function_id,
+			.address = fi.offset,
+			.line = (u32)line,
+			.column = (u32)column,
+			.filename = debug_filename_for_source_offset (reader, (u32)source_offset)
+		};
+		res = append_source_line (out, &cap, &sl);
+		if (res.code != RESULT_SUCCESS) {
+			goto fail;
+		}
+
+		int64_t address = 0, statement = 0;
+		for (;;) {
+			int64_t adelta = 0;
+			if (!read_sleb128 (data, size, &pos, &adelta)) {
+				PARSE_FAIL ("Invalid source location address delta");
+			}
+			if (adelta == -1) {
+				break;
+			}
+			int64_t ldelta = 0, cdelta = 0, scope = 0, env = 0, sdelta = 0;
+			if (adelta < 0
+					|| !read_sleb128 (data, size, &pos, &ldelta)
+					|| !read_sleb128 (data, size, &pos, &cdelta)
+					|| !read_sleb128 (data, size, &pos, &scope)
+					|| !read_sleb128 (data, size, &pos, &env)) {
+				PARSE_FAIL ("Invalid source location entry");
+			}
+			if ((ldelta & 1) && !read_sleb128 (data, size, &pos, &sdelta)) {
+				PARSE_FAIL ("Invalid source location statement delta");
+			}
+			ldelta = (ldelta - (ldelta & 1)) / 2;
+			address += adelta;
+			line += ldelta;
+			column += cdelta;
+			statement += sdelta;
+			if (address < 0 || address > UINT32_MAX || (uint64_t)fi.offset + (uint64_t)address > UINT32_MAX
+					|| line < 0 || line > UINT32_MAX || column < 0 || column > UINT32_MAX
+					|| statement < 0 || statement > UINT32_MAX) {
+				PARSE_FAIL ("Source location entry out of range");
+			}
+			sl.address = fi.offset + (u32)address;
+			sl.function_address = (u32)address;
+			sl.line = (u32)line;
+			sl.column = (u32)column;
+			sl.statement = (u32)statement;
+			res = append_source_line (out, &cap, &sl);
+			if (res.code != RESULT_SUCCESS) {
+				goto fail;
+			}
+		}
+	}
+	return SUCCESS_RESULT ();
+fail:
+	hbc_free_source_lines (out);
+	return res;
+}
+
+#undef PARSE_FAIL
+
+void hbc_free_source_lines(HBCSourceLineArray *arr) {
+	if (!arr) {
+		return;
+	}
+	free (arr->lines);
+	memset (arr, 0, sizeof (*arr));
+}
+
+/* Describes the file, not the runtime preference: true iff the file carries
+ * a non-empty source-locations blob. Relies on the DebugInfoHeader parsed at
+ * load time, so callers pay no body-parse cost for this query. */
+bool hbc_has_source_lines(HBC *hbc) {
+	if (!hbc || hbc->reader.header.debugInfoOffset == 0) {
+		return false;
+	}
+	if (_hbc_reader_read_debug_info_header (&hbc->reader).code != RESULT_SUCCESS) {
+		return false;
+	}
+	return hbc->reader.debug_info_header.scope_desc_data_offset > 0;
+}
+
+Result hbc_get_debug_info(HBC *hbc, HBCDebugInfo *out) {
+	if (!hbc || !out) {
+		return ERROR_RESULT (RESULT_ERROR_INVALID_ARGUMENT, "Invalid arguments");
+	}
+	memset (out, 0, sizeof (*out));
+	HBCReader *reader = &hbc->reader;
+	out->has_debug_info = reader->header.debugInfoOffset != 0;
+	if (!out->has_debug_info) {
+		return SUCCESS_RESULT ();
+	}
+	RETURN_IF_ERROR (_hbc_reader_read_debug_info (reader));
+	out->filename_count = reader->debug_info_header.filename_count;
+	out->filename_storage_size = reader->debug_info_header.filename_storage_size;
+	out->file_region_count = reader->debug_info_header.file_region_count;
+	out->source_locations_size = (u32)reader->sources_data_storage_size;
+	out->scope_desc_data_size = (u32)reader->scope_desc_data_storage_size;
+	out->textified_data_size = (u32)reader->textified_data_storage_size;
+	out->string_table_size = (u32)reader->string_table_storage_size;
+	out->debug_data_size = reader->debug_info_header.debug_data_size;
+	if (reader->function_headers) {
+		for (u32 i = 0; i < reader->header.functionCount; i++) {
+			if (reader->function_headers[i].hasDebugInfo) {
+				out->functions_with_debug_info++;
+			}
 		}
 	}
 	return SUCCESS_RESULT ();
@@ -412,7 +651,7 @@ Result hbc_dec(const HBCDecodeCtx *ctx, HBCInsnInfo *out) {
 	u8 opcode = ctx->bytes[0];
 
 	/* Get instruction definition */
-	const Instruction *inst = (opcode < isa.count) ? &isa.instructions[opcode] : NULL;
+	const Instruction *inst = (opcode < isa.count)? &isa.instructions[opcode]: NULL;
 	if (!inst || !inst->name) {
 		out->text = strdup ("unk");
 		out->size = 1;
@@ -459,19 +698,19 @@ Result hbc_dec(const HBCDecodeCtx *ctx, HBCInsnInfo *out) {
 			pos += 1;
 			break;
 		case OPERAND_TYPE_ADDR8:
-			operand_values[i] = (u32)(i32)(i8)ctx->bytes[pos];
+			operand_values[i] = (u32) (i32) (i8)ctx->bytes[pos];
 			pos += 1;
 			break;
 		case OPERAND_TYPE_UINT16:
-			operand_values[i] = ctx->bytes[pos] | (ctx->bytes[pos + 1] << 8);
+			operand_values[i] = (u32)ctx->bytes[pos] | ((u32)ctx->bytes[pos + 1] << 8);
 			pos += 2;
 			break;
 		case OPERAND_TYPE_REG32:
 		case OPERAND_TYPE_UINT32:
 		case OPERAND_TYPE_IMM32:
 		case OPERAND_TYPE_ADDR32:
-			operand_values[i] = ctx->bytes[pos] | (ctx->bytes[pos + 1] << 8) |
-				(ctx->bytes[pos + 2] << 16) | (ctx->bytes[pos + 3] << 24);
+			operand_values[i] = (u32)ctx->bytes[pos] | ((u32)ctx->bytes[pos + 1] << 8) |
+				((u32)ctx->bytes[pos + 2] << 16) | ((u32)ctx->bytes[pos + 3] << 24);
 			pos += 4;
 			break;
 		case OPERAND_TYPE_DOUBLE:
@@ -564,7 +803,7 @@ Result hbc_dec(const HBCDecodeCtx *ctx, HBCInsnInfo *out) {
 			!strncmp (mnemonic, "new_array_with_buffer", strlen ("new_array_with_buffer"));
 		if (is_literal_op) {
 			char *with_comment = build_literal_comment (&ctx->hbc->reader, mnemonic, operand_values, buf);
-			out->text = with_comment ? with_comment : strdup (buf);
+			out->text = with_comment? with_comment: strdup (buf);
 		} else {
 			out->text = strdup (buf);
 		}
@@ -592,6 +831,237 @@ Result hbc_dec(const HBCDecodeCtx *ctx, HBCInsnInfo *out) {
 	return SUCCESS_RESULT ();
 }
 
+typedef struct {
+	const char *name;    /* exact match, else match prefix */
+	const char *prefix;
+	const char *kind;
+	const char *module;  /* NULL for exports */
+	HBCBindingType type;
+} BindingHint;
+
+static const BindingHint g_hints[] = {
+	{ "NativeModules",          NULL,     "native", "react-native", HBC_BINDING_IMPORT },
+	{ "TurboModuleRegistry",    NULL,     "native", "react-native", HBC_BINDING_IMPORT },
+	{ "__turboModuleProxy",     NULL,     "native", "react-native", HBC_BINDING_IMPORT },
+	{ "requireNativeComponent", NULL,     "native", "react-native", HBC_BINDING_IMPORT },
+	{ "codegenNativeComponent", NULL,     "native", "react-native", HBC_BINDING_IMPORT },
+	{ "NativeEventEmitter",     NULL,     "native", "react-native", HBC_BINDING_IMPORT },
+	{ "RCTDeviceEventEmitter",  NULL,     "native", "react-native", HBC_BINDING_IMPORT },
+	{ "HermesInternal",         NULL,     "native", "react-native", HBC_BINDING_IMPORT },
+	{ "require",                NULL,     "js",     "react-native", HBC_BINDING_IMPORT },
+	{ "metroRequire",           NULL,     "js",     "react-native", HBC_BINDING_IMPORT },
+	{ "__r",                    NULL,     "js",     "react-native", HBC_BINDING_IMPORT },
+	{ "importScripts",          NULL,     "js",     "react-native", HBC_BINDING_IMPORT },
+	{ "global",                 NULL,     "global", "react-native", HBC_BINDING_IMPORT },
+	{ "globalThis",             NULL,     "global", "react-native", HBC_BINDING_IMPORT },
+	{ "exports",                NULL,     "module", NULL,           HBC_BINDING_EXPORT },
+	{ "module",                 NULL,     "module", NULL,           HBC_BINDING_EXPORT },
+	{ "__esModule",             NULL,     "module", NULL,           HBC_BINDING_EXPORT },
+	{ "default",                NULL,     "module", NULL,           HBC_BINDING_EXPORT },
+	{ NULL,                     "RCT",    "native", "react-native", HBC_BINDING_IMPORT },
+	{ NULL,                     "Native", "native", "react-native", HBC_BINDING_IMPORT },
+};
+#define HINTS_N (sizeof (g_hints) / sizeof (g_hints[0]))
+
+static bool str_eq(const char *a, const char *b) {
+	if (a == b) {
+		return true;
+	}
+	return a && b && !strcmp (a, b);
+}
+
+static bool is_export_marker(const char *s) {
+	return !strcmp (s, "exports") || !strcmp (s, "module");
+}
+
+static bool is_export_store(const char *name) {
+	return strstr (name, "put") && strstr (name, "by_id");
+}
+
+static Result binding_add(HBCBindings *out, HBCBindingType type, const char *kind,
+	const char *name, const char *module, u32 function_id, u32 offset, u32 string_id) {
+	for (u32 i = 0; i < out->count; i++) {
+		HBCBinding *b = &out->bindings[i];
+		if (b->type == type && !strcmp (b->kind, kind)
+				&& !strcmp (b->name, name) && str_eq (b->module, module)) {
+			return SUCCESS_RESULT ();
+		}
+	}
+	HBCBinding *nb = realloc (out->bindings, (out->count + 1) * sizeof (HBCBinding));
+	if (!nb) {
+		return ERROR_RESULT (RESULT_ERROR_MEMORY_ALLOCATION, "OOM");
+	}
+	out->bindings = nb;
+	HBCBinding *b = &nb[out->count++];
+	*b = (HBCBinding){
+		.type = type, .kind = kind, .name = strdup (name),
+		.module = module? strdup (module): NULL,
+		.function_id = function_id, .offset = offset, .string_id = string_id,
+	};
+	if (!b->name || (module && !b->module)) {
+		return ERROR_RESULT (RESULT_ERROR_MEMORY_ALLOCATION, "OOM");
+	}
+	return SUCCESS_RESULT ();
+}
+
+static size_t decode_operands(const Instruction *inst, const u8 *bytes, size_t len,
+	u32 values[6]) {
+	size_t pos = 1;
+	memset (values, 0, 6 * sizeof (u32));
+	for (int i = 0; i < 6 && inst->operands[i].operand_type != OPERAND_TYPE_NONE; i++) {
+		OperandType t = inst->operands[i].operand_type;
+		size_t need;
+		switch (t) {
+		case OPERAND_TYPE_REG8:
+		case OPERAND_TYPE_UINT8:
+			need = 1;
+			if (pos + need > len) { return 0; }
+			values[i] = bytes[pos];
+			break;
+		case OPERAND_TYPE_ADDR8:
+			need = 1;
+			if (pos + need > len) { return 0; }
+			values[i] = (u32)(i32)(i8)bytes[pos];
+			break;
+		case OPERAND_TYPE_UINT16:
+			need = 2;
+			if (pos + need > len) { return 0; }
+			values[i] = (u32)bytes[pos] | ((u32)bytes[pos + 1] << 8);
+			break;
+		case OPERAND_TYPE_REG32:
+		case OPERAND_TYPE_UINT32:
+		case OPERAND_TYPE_IMM32:
+		case OPERAND_TYPE_ADDR32:
+			need = 4;
+			if (pos + need > len) { return 0; }
+			values[i] = (u32)bytes[pos] | ((u32)bytes[pos + 1] << 8)
+				| ((u32)bytes[pos + 2] << 16) | ((u32)bytes[pos + 3] << 24);
+			break;
+		case OPERAND_TYPE_DOUBLE:
+			need = 8;
+			if (pos + need > len) { return 0; }
+			break;
+		default:
+			return 0;
+		}
+		pos += need;
+	}
+	return pos;
+}
+
+static Result scan_hint_matches(HBCBindings *out, const char *s, u32 fid, u32 off, u32 sid) {
+	for (size_t h = 0; h < HINTS_N; h++) {
+		const BindingHint *hint = &g_hints[h];
+		bool match = hint->name? !strcmp (s, hint->name):
+			!strncmp (s, hint->prefix, strlen (hint->prefix));
+		if (!match) {
+			continue;
+		}
+		RETURN_IF_ERROR (binding_add (out, hint->type, hint->kind, s,
+			hint->module, fid, off, sid));
+	}
+	return SUCCESS_RESULT ();
+}
+
+static bool function_has_export_marker(HBCReader *r, const u8 *code, u32 size, HBCISA isa) {
+	for (u32 pc = 0; pc < size; ) {
+		u8 op = code[pc];
+		const Instruction *inst = (op < isa.count)? &isa.instructions[op]: NULL;
+		if (!inst || !inst->name || !inst->binary_size) {
+			break;
+		}
+		u32 values[6];
+		if (!decode_operands (inst, code + pc, size - pc, values)) {
+			break;
+		}
+		for (int i = 0; i < 6 && inst->operands[i].operand_type != OPERAND_TYPE_NONE; i++) {
+			if (inst->operands[i].operand_meaning == OPERAND_MEANING_STRING_ID) {
+				u32 sid = values[i];
+				const char *s = (sid < r->header.stringCount && r->strings)? r->strings[sid]: NULL;
+				if (s && is_export_marker (s)) {
+					return true;
+				}
+			}
+		}
+		pc += inst->binary_size;
+	}
+	return false;
+}
+
+Result hbc_scan_bindings(HBC *hbc, HBCBindings *out) {
+	if (!hbc || !out) {
+		return ERROR_RESULT (RESULT_ERROR_INVALID_ARGUMENT, "invalid args");
+	}
+	memset (out, 0, sizeof (*out));
+	HBCReader *r = &hbc->reader;
+	HBCISA isa = hbc_isa_getv (r->header.version);
+	if (!isa.instructions) {
+		return ERROR_RESULT (RESULT_ERROR_INVALID_ARGUMENT, "no isa for version");
+	}
+	if (r->header.cjsModuleCount > 0 && r->header.version >= 77 && r->cjs_modules) {
+		for (u32 i = 0; i < r->header.cjsModuleCount; i++) {
+			u32 sid = r->cjs_modules[i].symbol_id;
+			const char *name = (sid < r->header.stringCount && r->strings)? r->strings[sid]: NULL;
+			if (name && *name) {
+				RETURN_IF_ERROR (binding_add (out, HBC_BINDING_IMPORT,
+					"cjs", name, "commonjs", 0, r->cjs_modules[i].offset, sid));
+			}
+		}
+	}
+	for (u32 fid = 0; fid < r->header.functionCount; fid++) {
+		const u8 *code = NULL;
+		u32 size = 0;
+		if (hbc_get_function_bytecode (hbc, fid, &code, &size).code != RESULT_SUCCESS
+				|| !code || !size) {
+			continue;
+		}
+		HBCFunc fi = { 0 };
+		(void)hbc_get_function_info (hbc, fid, &fi);
+		bool has_export_marker = function_has_export_marker (r, code, size, isa);
+		for (u32 pc = 0; pc < size; ) {
+			u8 op = code[pc];
+			const Instruction *inst = (op < isa.count)? &isa.instructions[op]: NULL;
+			if (!inst || !inst->name || !inst->binary_size) {
+				break;
+			}
+			u32 values[6];
+			if (!decode_operands (inst, code + pc, size - pc, values)) {
+				break;
+			}
+			for (int i = 0; i < 6 && inst->operands[i].operand_type != OPERAND_TYPE_NONE; i++) {
+				if (inst->operands[i].operand_meaning != OPERAND_MEANING_STRING_ID) {
+					continue;
+				}
+				u32 sid = values[i];
+				const char *s = (sid < r->header.stringCount && r->strings)? r->strings[sid]: NULL;
+				if (!s || !*s) {
+					continue;
+				}
+				RETURN_IF_ERROR (scan_hint_matches (out, s, fid, fi.offset + pc, sid));
+				if (has_export_marker && is_export_store (inst->name) && !is_export_marker (s)) {
+					RETURN_IF_ERROR (binding_add (out, HBC_BINDING_EXPORT,
+						"module", s, NULL, fid, fi.offset + pc, sid));
+				}
+			}
+			pc += inst->binary_size;
+		}
+	}
+	return SUCCESS_RESULT ();
+}
+
+void hbc_free_bindings(HBCBindings *bindings) {
+	if (!bindings || !bindings->bindings) {
+		return;
+	}
+	for (u32 i = 0; i < bindings->count; i++) {
+		free (bindings->bindings[i].name);
+		free (bindings->bindings[i].module);
+	}
+	free (bindings->bindings);
+	bindings->bindings = NULL;
+	bindings->count = 0;
+}
+
 /* Encoding functions */
 
 Result hbc_enc(

src/lib/hbc_internal.h

@@ -3,6 +3,13 @@
 
 #include <hbc/hbc.h>
 #include <hbc/parser.h>
+#include <hbc/literals.h>
+
+typedef struct {
+	HBCLiteralEntry *entries;
+	u32 count;
+	u32 cap;
+} HBCLiteralCache;
 
 struct HBC {
 	HBCReader reader;
@@ -11,6 +18,7 @@ struct HBC {
 	u32 string_count;
 	const char **strings;
 	u32 function_count;
+	HBCLiteralCache lit_cache;
 };
 
 #endif /* HBC_INTERNAL_H */

src/lib/literals_api.c

@@ -0,0 +1,519 @@
+/* Public literals API: lazy cache, formatting, scanning. See
+ * include/hbc/literals.h for the contract. */
+
+#include <hbc/literals.h>
+#include <hbc/opcodes.h>
+#include <hbc/bytecode.h>
+#include <hbc/decompilation/literals.h>
+#include "hbc_internal.h"
+#include <stdbool.h>
+#include <stdlib.h>
+#include <string.h>
+
+/* ---------- Global toggles ---------- */
+
+static bool g_inline_literals = false;
+
+void hbc_set_inline_literals(bool on) {
+	g_inline_literals = on;
+}
+
+bool hbc_get_inline_literals(void) {
+	return g_inline_literals;
+}
+
+/* ---------- Pool accessors ---------- */
+
+u32 hbc_get_pool_paddr(HBC *hbc, HBCLiteralKind kind) {
+	if (!hbc) {
+		return 0;
+	}
+	const HBCReader *r = &hbc->reader;
+	switch (kind) {
+	case HBC_LIT_ARRAY:
+		return r->arrays? r->arrays_paddr: 0;
+	case HBC_LIT_OBJECT:
+		return r->object_keys? r->object_keys_paddr: 0;
+	}
+	return 0;
+}
+
+u32 hbc_get_pool_size(HBC *hbc, HBCLiteralKind kind) {
+	if (!hbc) {
+		return 0;
+	}
+	const HBCReader *r = &hbc->reader;
+	switch (kind) {
+	case HBC_LIT_ARRAY: return r->header.arrayBufferSize;
+	case HBC_LIT_OBJECT: return r->header.objKeyBufferSize;
+	}
+	return 0;
+}
+
+u32 hbc_get_object_values_paddr(HBC *hbc) {
+	return (hbc && hbc->reader.object_values)? hbc->reader.object_values_paddr: 0;
+}
+
+u32 hbc_get_object_values_size(HBC *hbc) {
+	return hbc? hbc->reader.header.objValueBufferSize: 0;
+}
+
+/* ---------- Formatting ---------- */
+
+static Result format_raw_impl(HBCReader *r, HBCLiteralKind kind, u32 num_items,
+	u32 primary_id, u32 secondary_id, char **out) {
+	*out = NULL;
+	StringBuffer sb;
+	Result ir = _hbc_string_buffer_init (&sb, 128);
+	if (ir.code != RESULT_SUCCESS) {
+		return ir;
+	}
+	Result fr;
+	if (kind == HBC_LIT_ARRAY) {
+		fr = _hbc_format_array_literal (r, num_items, primary_id, &sb,
+			LITERALS_PRETTY_NEVER, true);
+	} else {
+		/* For v97+ primary_id is the shape index; the internal formatter
+		 * expects that convention already (first arg is "key_count" which is
+		 * reinterpreted as shape id when version >= 97). */
+		fr = _hbc_format_object_literal (r, num_items, num_items, primary_id,
+			secondary_id, &sb, LITERALS_PRETTY_NEVER, true);
+	}
+	if (fr.code == RESULT_SUCCESS && sb.data && sb.length > 0) {
+		*out = strdup (sb.data);
+		if (!*out) {
+			fr = ERROR_RESULT (RESULT_ERROR_MEMORY_ALLOCATION, "oom");
+		}
+	}
+	_hbc_string_buffer_free (&sb);
+	return fr;
+}
+
+Result hbc_literals_format_raw(HBC *hbc, HBCLiteralKind kind, u32 num_items,
+	u32 primary_id, u32 secondary_id, char **out) {
+	if (!hbc || !out) {
+		return ERROR_RESULT (RESULT_ERROR_INVALID_ARGUMENT, "invalid args");
+	}
+	return format_raw_impl (&hbc->reader, kind, num_items, primary_id,
+		secondary_id, out);
+}
+
+Result hbc_literals_format_for_opcode(HBC *hbc, u8 opcode, u32 arg3, u32 arg4,
+	u32 arg5, char **out) {
+	if (!hbc || !out) {
+		return ERROR_RESULT (RESULT_ERROR_INVALID_ARGUMENT, "invalid args");
+	}
+	switch (opcode) {
+	case OP_NewArrayWithBuffer:
+	case OP_NewArrayWithBufferLong:
+		return hbc_literals_format_raw (hbc, HBC_LIT_ARRAY, arg3, arg4, 0, out);
+	case OP_NewObjectWithBuffer:
+	case OP_NewObjectWithBufferLong:
+		return hbc_literals_format_raw (hbc, HBC_LIT_OBJECT, arg3, arg4, arg5, out);
+	}
+	return ERROR_RESULT (RESULT_ERROR_INVALID_ARGUMENT,
+		"opcode is not a buffer-literal constructor");
+}
+
+/* ---------- Cache ---------- */
+
+static HBCLiteralEntry *cache_find(HBCLiteralCache *c, HBCLiteralKind kind,
+	u32 num_items, u32 primary_id, u32 secondary_id) {
+	for (u32 i = 0; i < c->count; i++) {
+		HBCLiteralEntry *e = &c->entries[i];
+		if (e->kind == kind && e->num_items == num_items
+			&& e->primary_id == primary_id
+			&& e->secondary_id == secondary_id) {
+			return e;
+		}
+	}
+	return NULL;
+}
+
+static HBCLiteralEntry *cache_alloc(HBCLiteralCache *c) {
+	if (c->count == c->cap) {
+		u32 nc = c->cap? c->cap * 2: 64;
+		HBCLiteralEntry *ne = realloc (c->entries, nc * sizeof (HBCLiteralEntry));
+		if (!ne) {
+			return NULL;
+		}
+		c->entries = ne;
+		c->cap = nc;
+	}
+	HBCLiteralEntry *e = &c->entries[c->count++];
+	memset (e, 0, sizeof (*e));
+	return e;
+}
+
+/* Compute paddr for an entry based on its kind/primary_id. For v97+ objects,
+ * primary_id is a shape index — resolve via the shape table. */
+static u32 entry_paddr(HBC *hbc, HBCLiteralKind kind, u32 primary_id) {
+	const HBCReader *r = &hbc->reader;
+	if (kind == HBC_LIT_ARRAY) {
+		if (!r->arrays) {
+			return 0;
+		}
+		if (primary_id >= r->header.arrayBufferSize) {
+			return 0;
+		}
+		return r->arrays_paddr + primary_id;
+	}
+	/* object */
+	if (r->header.version >= 97 && r->object_shapes
+		&& primary_id < r->object_shape_count) {
+		u32 key_off = r->object_shapes[primary_id].key_buffer_offset;
+		if (r->object_keys && key_off < r->header.objKeyBufferSize) {
+			return r->object_keys_paddr + key_off;
+		}
+		return 0;
+	}
+	if (!r->object_keys || primary_id >= r->header.objKeyBufferSize) {
+		return 0;
+	}
+	return r->object_keys_paddr + primary_id;
+}
+
+static Result cache_get_or_create(HBC *hbc, HBCLiteralKind kind, u32 num_items,
+	u32 primary_id, u32 secondary_id, HBCLiteralEntry **out_entry) {
+	HBCLiteralEntry *e = cache_find (&hbc->lit_cache, kind, num_items,
+		primary_id, secondary_id);
+	if (e) {
+		*out_entry = e;
+		return SUCCESS_RESULT ();
+	}
+	e = cache_alloc (&hbc->lit_cache);
+	if (!e) {
+		return ERROR_RESULT (RESULT_ERROR_MEMORY_ALLOCATION, "oom");
+	}
+	e->kind = kind;
+	e->num_items = num_items;
+	e->primary_id = primary_id;
+	e->secondary_id = secondary_id;
+	e->paddr = entry_paddr (hbc, kind, primary_id);
+	/* Format now (lazy across calls, but eager per key once we've decided to
+	 * cache it). The formatted string is small and reused on every access. */
+	char *txt = NULL;
+	Result fr = format_raw_impl (&hbc->reader, kind, num_items, primary_id,
+		secondary_id, &txt);
+	if (fr.code == RESULT_SUCCESS) {
+		e->formatted = txt;
+	}
+	*out_entry = e;
+	return SUCCESS_RESULT ();
+}
+
+Result hbc_literals_get(HBC *hbc, HBCLiteralKind kind, u32 num_items,
+	u32 primary_id, u32 secondary_id, const char **out_formatted) {
+	if (!hbc || !out_formatted) {
+		return ERROR_RESULT (RESULT_ERROR_INVALID_ARGUMENT, "invalid args");
+	}
+	HBCLiteralEntry *e;
+	Result r = cache_get_or_create (hbc, kind, num_items, primary_id,
+		secondary_id, &e);
+	if (r.code != RESULT_SUCCESS) {
+		return r;
+	}
+	*out_formatted = e->formatted;
+	return SUCCESS_RESULT ();
+}
+
+static void xref_add(HBCLiteralEntry *e, u32 from_addr) {
+	/* Keep xrefs as a small sorted-by-insertion array. Linear dedup. */
+	for (u32 i = 0; i < e->xref_count; i++) {
+		if (e->xref_addrs[i] == from_addr) {
+			return;
+		}
+	}
+	u32 *na = realloc (e->xref_addrs, (e->xref_count + 1) * sizeof (u32));
+	if (!na) {
+		return;
+	}
+	e->xref_addrs = na;
+	e->xref_addrs[e->xref_count++] = from_addr;
+}
+
+Result hbc_literals_register(HBC *hbc, HBCLiteralKind kind, u32 num_items,
+	u32 primary_id, u32 secondary_id, u32 from_addr) {
+	if (!hbc) {
+		return ERROR_RESULT (RESULT_ERROR_INVALID_ARGUMENT, "invalid args");
+	}
+	HBCLiteralEntry *e;
+	Result r = cache_get_or_create (hbc, kind, num_items, primary_id,
+		secondary_id, &e);
+	if (r.code != RESULT_SUCCESS) {
+		return r;
+	}
+	if (from_addr) {
+		xref_add (e, from_addr);
+	}
+	return SUCCESS_RESULT ();
+}
+
+Result hbc_literals_list(HBC *hbc, const HBCLiteralEntry **out, u32 *out_count) {
+	if (!hbc || !out || !out_count) {
+		return ERROR_RESULT (RESULT_ERROR_INVALID_ARGUMENT, "invalid args");
+	}
+	*out = hbc->lit_cache.entries;
+	*out_count = hbc->lit_cache.count;
+	return SUCCESS_RESULT ();
+}
+
+u32 hbc_literals_count(HBC *hbc) {
+	return hbc? hbc->lit_cache.count: 0;
+}
+
+void hbc_literals_reset(HBC *hbc) {
+	if (!hbc) {
+		return;
+	}
+	HBCLiteralCache *c = &hbc->lit_cache;
+	for (u32 i = 0; i < c->count; i++) {
+		free (c->entries[i].formatted);
+		free (c->entries[i].xref_addrs);
+	}
+	free (c->entries);
+	c->entries = NULL;
+	c->count = 0;
+	c->cap = 0;
+}
+
+/* ---------- Code-side scan ---------- */
+
+static bool is_buffer_op(u8 opcode, HBCLiteralKind *out_kind) {
+	switch (opcode) {
+	case OP_NewArrayWithBuffer:
+	case OP_NewArrayWithBufferLong:
+		*out_kind = HBC_LIT_ARRAY;
+		return true;
+	case OP_NewObjectWithBuffer:
+	case OP_NewObjectWithBufferLong:
+		*out_kind = HBC_LIT_OBJECT;
+		return true;
+	}
+	return false;
+}
+
+/* Decode opcode immediates into (num_items, primary, secondary), version-aware.
+ * Returns false if the opcode is not a buffer-literal constructor or if the
+ * buffer would be read out of bounds. */
+static bool decode_buffer_op(u8 op, u32 version, const u8 *code, u32 size,
+	u32 pc, u32 inst_size, HBCLiteralKind *kind, u32 *num_items, u32 *primary,
+	u32 *secondary) {
+	if (pc + inst_size > size) {
+		return false;
+	}
+	*secondary = 0;
+	if (op == OP_NewArrayWithBuffer) {
+		/* reg8 + u16 prealloc + u16 num + u16 id (v<97 AND v97+) */
+		*kind = HBC_LIT_ARRAY;
+		*num_items = (u32)code[pc + 4] | ((u32)code[pc + 5] << 8);
+		*primary = (u32)code[pc + 6] | ((u32)code[pc + 7] << 8);
+		return true;
+	}
+	if (op == OP_NewArrayWithBufferLong) {
+		/* reg8 + u16 prealloc + u16 num + u32 id */
+		*kind = HBC_LIT_ARRAY;
+		*num_items = (u32)code[pc + 4] | ((u32)code[pc + 5] << 8);
+		*primary = (u32)code[pc + 6] | ((u32)code[pc + 7] << 8)
+			| ((u32)code[pc + 8] << 16) | ((u32)code[pc + 9] << 24);
+		return true;
+	}
+	if (op == OP_NewObjectWithBuffer) {
+		*kind = HBC_LIT_OBJECT;
+		if (version >= 97) {
+			/* reg8 + u16 shape_id + u16 values_id */
+			*primary = (u32)code[pc + 2] | ((u32)code[pc + 3] << 8);
+			*secondary = (u32)code[pc + 4] | ((u32)code[pc + 5] << 8);
+			*num_items = 0; /* resolved via shape table by the formatter */
+		} else {
+			/* reg8 + u16 prealloc + u16 num + u16 keys + u16 vals */
+			*num_items = (u32)code[pc + 4] | ((u32)code[pc + 5] << 8);
+			*primary = (u32)code[pc + 6] | ((u32)code[pc + 7] << 8);
+			*secondary = (u32)code[pc + 8] | ((u32)code[pc + 9] << 8);
+		}
+		return true;
+	}
+	if (op == OP_NewObjectWithBufferLong) {
+		*kind = HBC_LIT_OBJECT;
+		if (version >= 97) {
+			/* reg8 + u32 shape_id + u32 values_id */
+			*primary = (u32)code[pc + 2] | ((u32)code[pc + 3] << 8)
+				| ((u32)code[pc + 4] << 16) | ((u32)code[pc + 5] << 24);
+			*secondary = (u32)code[pc + 6] | ((u32)code[pc + 7] << 8)
+				| ((u32)code[pc + 8] << 16) | ((u32)code[pc + 9] << 24);
+			*num_items = 0;
+		} else {
+			/* reg8 + u16 prealloc + u16 num + u32 keys + u32 vals */
+			*num_items = (u32)code[pc + 4] | ((u32)code[pc + 5] << 8);
+			*primary = (u32)code[pc + 6] | ((u32)code[pc + 7] << 8)
+				| ((u32)code[pc + 8] << 16) | ((u32)code[pc + 9] << 24);
+			*secondary = (u32)code[pc + 10] | ((u32)code[pc + 11] << 8)
+				| ((u32)code[pc + 12] << 16) | ((u32)code[pc + 13] << 24);
+		}
+		return true;
+	}
+	return false;
+}
+
+Result hbc_literals_scan_code(HBC *hbc, u32 *out_count) {
+	if (!hbc) {
+		return ERROR_RESULT (RESULT_ERROR_INVALID_ARGUMENT, "invalid args");
+	}
+	u32 nfuncs = hbc->reader.header.functionCount;
+	u32 version = hbc->reader.header.version;
+	HBCISA isa = hbc_isa_getv (version);
+	if (!isa.instructions) {
+		return ERROR_RESULT (RESULT_ERROR_INVALID_ARGUMENT, "no isa for version");
+	}
+	for (u32 fid = 0; fid < nfuncs; fid++) {
+		const u8 *code = NULL;
+		u32 size = 0;
+		if (hbc_get_function_bytecode (hbc, fid, &code, &size).code
+			!= RESULT_SUCCESS || !code || size == 0) {
+			continue;
+		}
+		HBCFunc fi;
+		u32 fn_paddr = 0;
+		if (hbc_get_function_info (hbc, fid, &fi).code == RESULT_SUCCESS) {
+			fn_paddr = fi.offset;
+		}
+		u32 pc = 0;
+		while (pc < size) {
+			u8 op = code[pc];
+			const Instruction *ins = (op < isa.count)? &isa.instructions[op]: NULL;
+			if (!ins || !ins->name || ins->binary_size == 0) {
+				break;
+			}
+			HBCLiteralKind kind;
+			u32 num_items = 0, primary = 0, secondary = 0;
+			if (decode_buffer_op (op, version, code, size, pc, ins->binary_size,
+					&kind, &num_items, &primary, &secondary)) {
+				u32 call_addr = fn_paddr + pc;
+				hbc_literals_register (hbc, kind, num_items, primary, secondary,
+					call_addr);
+			}
+			(void)is_buffer_op; /* retained for tools that need the predicate */
+			pc += ins->binary_size;
+		}
+	}
+	if (out_count) {
+		*out_count = hbc->lit_cache.count;
+	}
+	return SUCCESS_RESULT ();
+}
+
+/* ---------- Pool-side scan ---------- */
+
+static bool slp_skip_group(const u8 *base, size_t size, size_t *pos,
+	u8 *out_tag, u32 *out_length) {
+	if (*pos >= size) {
+		return false;
+	}
+	u8 taglen = base[(*pos)++];
+	u8 tag = (taglen >> 4) & 0x7;
+	u32 length = taglen & 0x0F;
+	if (taglen & 0x80) {
+		if (*pos >= size) {
+			return false;
+		}
+		length = ((u32)(taglen & 0x0F) << 8) | base[(*pos)++];
+	}
+	size_t per_value = 0;
+	switch (tag) {
+	case 0: case 1: case 2: per_value = 0; break; /* null/true/false */
+	case 3: per_value = 8; break; /* number */
+	case 4: case 7: per_value = 4; break; /* longstr/int */
+	case 5: per_value = 2; break; /* shortstr */
+	case 6: per_value = 1; break; /* bytestr */
+	default: return false;
+	}
+	if (*pos + per_value * length > size) {
+		return false;
+	}
+	*pos += per_value * length;
+	if (out_tag) {
+		*out_tag = tag;
+	}
+	if (out_length) {
+		*out_length = length;
+	}
+	return true;
+}
+
+Result hbc_literals_scan_pool(HBC *hbc, HBCLiteralKind kind,
+	HBCPoolGroup **out, u32 *out_count) {
+	if (!hbc || !out || !out_count) {
+		return ERROR_RESULT (RESULT_ERROR_INVALID_ARGUMENT, "invalid args");
+	}
+	*out = NULL;
+	*out_count = 0;
+
+	/* Object-keys on v97+: enumerate shape table instead of the raw pool. */
+	if (kind == HBC_LIT_OBJECT && hbc->reader.header.version >= 97
+		&& hbc->reader.object_shapes
+		&& hbc->reader.object_shape_count > 0) {
+		size_t n = hbc->reader.object_shape_count;
+		HBCPoolGroup *arr = calloc (n, sizeof (HBCPoolGroup));
+		if (!arr) {
+			return ERROR_RESULT (RESULT_ERROR_MEMORY_ALLOCATION, "oom");
+		}
+		for (size_t i = 0; i < n; i++) {
+			arr[i].pool_offset = hbc->reader.object_shapes[i].key_buffer_offset;
+			arr[i].num_items = hbc->reader.object_shapes[i].prop_count;
+			arr[i].paddr = hbc->reader.object_keys_paddr + arr[i].pool_offset;
+			arr[i].tag = 0xff; /* N/A for a shape row */
+		}
+		*out = arr;
+		*out_count = (u32)n;
+		return SUCCESS_RESULT ();
+	}
+
+	const u8 *base;
+	u32 size;
+	u32 paddr;
+	if (kind == HBC_LIT_ARRAY) {
+		base = hbc->reader.arrays;
+		size = hbc->reader.header.arrayBufferSize;
+		paddr = hbc->reader.arrays_paddr;
+	} else {
+		base = hbc->reader.object_keys;
+		size = hbc->reader.header.objKeyBufferSize;
+		paddr = hbc->reader.object_keys_paddr;
+	}
+	if (!base || !size) {
+		return SUCCESS_RESULT ();
+	}
+
+	u32 cap = 64;
+	HBCPoolGroup *arr = calloc (cap, sizeof (HBCPoolGroup));
+	if (!arr) {
+		return ERROR_RESULT (RESULT_ERROR_MEMORY_ALLOCATION, "oom");
+	}
+	u32 n = 0;
+	size_t pos = 0;
+	while (pos < size) {
+		size_t group_start = pos;
+		u8 tag = 0;
+		u32 length = 0;
+		if (!slp_skip_group (base, size, &pos, &tag, &length)) {
+			break;
+		}
+		if (n == cap) {
+			u32 nc = cap * 2;
+			HBCPoolGroup *na = realloc (arr, nc * sizeof (HBCPoolGroup));
+			if (!na) {
+				break;
+			}
+			arr = na;
+			cap = nc;
+		}
+		arr[n].pool_offset = (u32)group_start;
+		arr[n].paddr = paddr + (u32)group_start;
+		arr[n].num_items = length;
+		arr[n].tag = tag;
+		n++;
+	}
+	*out = arr;
+	*out_count = n;
+	return SUCCESS_RESULT ();
+}

src/lib/opcodes/encoder.c

@@ -187,89 +187,89 @@ Result hbc_encoder_parse_instruction(HBCEncoder *encoder, const char *asm_line,
 		bool parse_success = false;
 
 		switch (expected_type) {
-			case OPERAND_TYPE_REG8:
-			case OPERAND_TYPE_REG32:
-				/* Register: rN or RN */
-				if (trim_start[0] == 'r' || trim_start[0] == 'R') {
-					operand_values[operand_count] = (u32)strtoul (trim_start + 1, &endptr, 10);
-					if (*endptr == '\0') {
-						parse_success = true;
-						/* Validate register range */
-						if (expected_type == OPERAND_TYPE_REG8 && operand_values[operand_count] > 255) {
-							return ERROR_RESULT (RESULT_ERROR_PARSING_FAILED, "Register number out of range for REG8");
-						}
-					}
-				} else {
-					/* Allow bare numbers for registers too */
-					operand_values[operand_count] = (u32)strtoul (trim_start, &endptr, 10);
-					if (*endptr == '\0') {
-						parse_success = true;
-						/* Validate register range */
-						if (expected_type == OPERAND_TYPE_REG8 && operand_values[operand_count] > 255) {
-							return ERROR_RESULT (RESULT_ERROR_PARSING_FAILED, "Register number out of range for REG8");
-						}
+		case OPERAND_TYPE_REG8:
+		case OPERAND_TYPE_REG32:
+			/* Register: rN or RN */
+			if (trim_start[0] == 'r' || trim_start[0] == 'R') {
+				operand_values[operand_count] = (u32)strtoul (trim_start + 1, &endptr, 10);
+				if (*endptr == '\0') {
+					parse_success = true;
+					/* Validate register range */
+					if (expected_type == OPERAND_TYPE_REG8 && operand_values[operand_count] > 255) {
+						return ERROR_RESULT (RESULT_ERROR_PARSING_FAILED, "Register number out of range for REG8");
 					}
 				}
-				break;
-
-			case OPERAND_TYPE_UINT8:
-			case OPERAND_TYPE_UINT16:
-			case OPERAND_TYPE_UINT32:
-			case OPERAND_TYPE_IMM32:
-			case OPERAND_TYPE_ADDR8:
-			case OPERAND_TYPE_ADDR32:
-				/* Immediate or address */
-				if (trim_start[0] == '0' && (trim_start[1] == 'x' || trim_start[1] == 'X')) {
-					/* Hex */
-					operand_values[operand_count] = (u32)strtoul (trim_start, &endptr, 16);
-					if (*endptr == '\0') {
-						parse_success = true;
-					}
-				} else {
-					/* Decimal */
-					operand_values[operand_count] = (u32)strtoul (trim_start, &endptr, 10);
-					if (*endptr == '\0') {
-						parse_success = true;
+			} else {
+				/* Allow bare numbers for registers too */
+				operand_values[operand_count] = (u32)strtoul (trim_start, &endptr, 10);
+				if (*endptr == '\0') {
+					parse_success = true;
+					/* Validate register range */
+					if (expected_type == OPERAND_TYPE_REG8 && operand_values[operand_count] > 255) {
+						return ERROR_RESULT (RESULT_ERROR_PARSING_FAILED, "Register number out of range for REG8");
 					}
 				}
-				break;
-
-			case OPERAND_TYPE_DOUBLE:
-				/* Parse as double literal */
-				{
-					char *endptr_double;
-					double dval = strtod (trim_start, &endptr_double);
-					if (*endptr_double == '\0') {
-						/* Convert double to IEEE 754 bits */
-						u64 bits;
-						memcpy (&bits, &dval, sizeof (double));
-						operand_values[operand_count] = bits;
-						parse_success = true;
-					} else if (trim_start[0] == '0' && (trim_start[1] == 'x' || trim_start[1] == 'X')) {
-						/* Allow hex for bit patterns */
-						operand_values[operand_count] = (u64)strtoull (trim_start, &endptr_double, 16);
-						if (*endptr_double == '\0') {
-							parse_success = true;
-						}
-					}
-				}
-				break;
-
-			default:
-				/* Unknown type, try to parse as number */
-				if (trim_start[0] == '0' && (trim_start[1] == 'x' || trim_start[1] == 'X')) {
-					operand_values[operand_count] = (u32)strtoul (trim_start, &endptr, 16);
-					if (*endptr == '\0') {
-						parse_success = true;
-					}
-				} else {
-					operand_values[operand_count] = (u32)strtoul (trim_start, &endptr, 10);
-					if (*endptr == '\0') {
-						parse_success = true;
-					}
-				}
-				break;
 			}
+			break;
+
+		case OPERAND_TYPE_UINT8:
+		case OPERAND_TYPE_UINT16:
+		case OPERAND_TYPE_UINT32:
+		case OPERAND_TYPE_IMM32:
+		case OPERAND_TYPE_ADDR8:
+		case OPERAND_TYPE_ADDR32:
+			/* Immediate or address */
+			if (trim_start[0] == '0' && (trim_start[1] == 'x' || trim_start[1] == 'X')) {
+				/* Hex */
+				operand_values[operand_count] = (u32)strtoul (trim_start, &endptr, 16);
+				if (*endptr == '\0') {
+					parse_success = true;
+				}
+			} else {
+				/* Decimal */
+				operand_values[operand_count] = (u32)strtoul (trim_start, &endptr, 10);
+				if (*endptr == '\0') {
+					parse_success = true;
+				}
+			}
+			break;
+
+		case OPERAND_TYPE_DOUBLE:
+			/* Parse as double literal */
+			{
+				char *endptr_double;
+				double dval = strtod (trim_start, &endptr_double);
+				if (*endptr_double == '\0') {
+					/* Convert double to IEEE 754 bits */
+					u64 bits;
+					memcpy (&bits, &dval, sizeof (double));
+					operand_values[operand_count] = bits;
+					parse_success = true;
+				} else if (trim_start[0] == '0' && (trim_start[1] == 'x' || trim_start[1] == 'X')) {
+					/* Allow hex for bit patterns */
+					operand_values[operand_count] = (u64)strtoull (trim_start, &endptr_double, 16);
+					if (*endptr_double == '\0') {
+						parse_success = true;
+					}
+				}
+			}
+			break;
+
+		default:
+			/* Unknown type, try to parse as number */
+			if (trim_start[0] == '0' && (trim_start[1] == 'x' || trim_start[1] == 'X')) {
+				operand_values[operand_count] = (u32)strtoul (trim_start, &endptr, 16);
+				if (*endptr == '\0') {
+					parse_success = true;
+				}
+			} else {
+				operand_values[operand_count] = (u32)strtoul (trim_start, &endptr, 10);
+				if (*endptr == '\0') {
+					parse_success = true;
+				}
+			}
+			break;
+		}
 
 		if (!parse_success) {
 			return ERROR_RESULT (RESULT_ERROR_PARSING_FAILED, "Invalid operand format");

src/lib/parsers/hbc_file_parser.c

@@ -81,6 +81,12 @@ Result _hbc_reader_cleanup(HBCReader *reader) {
 	/* Clean up debug info */
 	free (reader->debug_string_table);
 	free (reader->debug_string_storage);
+	if (reader->debug_filenames) {
+		for (u32 i = 0; i < reader->debug_info_header.filename_count; i++) {
+			free (reader->debug_filenames[i]);
+		}
+		free (reader->debug_filenames);
+	}
 	free (reader->debug_file_regions);
 	free (reader->sources_data_storage);
 	free (reader->scope_desc_data_storage);
@@ -1077,6 +1083,7 @@ Result _hbc_reader_read_arrays(HBCReader *reader) {
 
 	/* Read array buffer */
 	if (reader->header.arrayBufferSize > 0) {
+		reader->arrays_paddr = (u32)reader->file_buffer.position;
 		reader->arrays = (u8 *)malloc (reader->header.arrayBufferSize);
 		if (!reader->arrays) {
 			hbc_debug_printf ("Failed to allocate %u bytes for array buffer\n", reader->header.arrayBufferSize);
@@ -1108,6 +1115,7 @@ Result _hbc_reader_read_arrays(HBCReader *reader) {
 			reader->header.objKeyBufferSize = 0;
 		} else {
 			/* Safe to read object keys */
+			reader->object_keys_paddr = (u32)reader->file_buffer.position;
 			reader->object_keys = (u8 *)malloc (reader->header.objKeyBufferSize);
 			if (!reader->object_keys) {
 				fprintf (stderr, "Failed to allocate %u bytes for object key buffer\n", reader->header.objKeyBufferSize);
@@ -1149,6 +1157,7 @@ Result _hbc_reader_read_arrays(HBCReader *reader) {
 		}
 		reader->literal_values = reader->arrays;
 		reader->object_values = reader->arrays;
+		reader->object_values_paddr = reader->arrays_paddr;
 		return SUCCESS_RESULT ();
 	}
 
@@ -1164,6 +1173,7 @@ Result _hbc_reader_read_arrays(HBCReader *reader) {
 			reader->header.objValueBufferSize = 0;
 		} else {
 			/* Safe to read object values */
+			reader->object_values_paddr = (u32)reader->file_buffer.position;
 			reader->object_values = (u8 *)malloc (reader->header.objValueBufferSize);
 			if (!reader->object_values) {
 				fprintf (stderr, "Failed to allocate %u bytes for object value buffer\n", reader->header.objValueBufferSize);
@@ -1574,22 +1584,31 @@ Result _hbc_reader_read_function_sources(HBCReader *reader) {
 	return SUCCESS_RESULT ();
 }
 
-/* Read debug information */
-Result _hbc_reader_read_debug_info(HBCReader *reader) {
+/* Size of the on-disk DebugInfoHeader block, in bytes. */
+static u32 debug_info_header_bytes(u32 version) {
+	/* v<91: 5 u32s (filename_count, filename_storage_size, file_region_count,
+	 * scope_desc_data_offset, debug_data_size).
+	 * v>=91: 7 u32s (adds textified_data_offset, string_table_offset). */
+	return (version >= 91? 7u: 5u) * sizeof (u32);
+}
+
+/* Read just the DebugInfoHeader fields (no storage blobs, no allocations).
+ * Cheap enough to run eagerly at load time so "has source lines" is knowable
+ * without committing to a full debug-info parse. Idempotent. */
+Result _hbc_reader_read_debug_info_header(HBCReader *reader) {
 	if (!reader) {
 		return ERROR_RESULT (RESULT_ERROR_INVALID_ARGUMENT, "Reader is NULL");
 	}
-	if (reader->header.debugInfoOffset == 0) {
+	if (reader->debug_info_header_loaded) {
 		return SUCCESS_RESULT ();
 	}
-	if (reader->header.version >= 97) {
+	if (reader->header.debugInfoOffset == 0) {
+		reader->debug_info_header_loaded = true;
 		return SUCCESS_RESULT ();
 	}
 
-	/* Seek to debug info offset */
 	RETURN_IF_ERROR (_hbc_buffer_reader_seek (&reader->file_buffer, reader->header.debugInfoOffset));
 
-	/* Read debug info header */
 	RETURN_IF_ERROR (_hbc_buffer_reader_read_u32 (&reader->file_buffer, &reader->debug_info_header.filename_count));
 	RETURN_IF_ERROR (_hbc_buffer_reader_read_u32 (&reader->file_buffer, &reader->debug_info_header.filename_storage_size));
 	RETURN_IF_ERROR (_hbc_buffer_reader_read_u32 (&reader->file_buffer, &reader->debug_info_header.file_region_count));
@@ -1602,6 +1621,30 @@ Result _hbc_reader_read_debug_info(HBCReader *reader) {
 
 	RETURN_IF_ERROR (_hbc_buffer_reader_read_u32 (&reader->file_buffer, &reader->debug_info_header.debug_data_size));
 
+	reader->debug_info_header_loaded = true;
+	return SUCCESS_RESULT ();
+}
+
+/* Read the full debug-info body (filenames, file regions, sources/scope/
+ * textified/string-table storage blobs). Builds on the cheap header parse.
+ * Idempotent: subsequent calls after a successful load are no-ops. */
+Result _hbc_reader_read_debug_info(HBCReader *reader) {
+	if (!reader) {
+		return ERROR_RESULT (RESULT_ERROR_INVALID_ARGUMENT, "Reader is NULL");
+	}
+	if (reader->debug_info_loaded) {
+		return SUCCESS_RESULT ();
+	}
+	RETURN_IF_ERROR (_hbc_reader_read_debug_info_header (reader));
+	if (reader->header.debugInfoOffset == 0) {
+		reader->debug_info_loaded = true;
+		return SUCCESS_RESULT ();
+	}
+
+	/* Seek past the already-parsed header to the start of the body. */
+	RETURN_IF_ERROR (_hbc_buffer_reader_seek (&reader->file_buffer,
+		reader->header.debugInfoOffset + debug_info_header_bytes (reader->header.version)));
+
 	/* Allocate debug string table */
 	if (reader->debug_info_header.filename_count > 0) {
 		reader->debug_string_table = (OffsetLengthPair *)malloc (
@@ -1703,6 +1746,7 @@ Result _hbc_reader_read_debug_info(HBCReader *reader) {
 		reader->string_table_storage_size = string_table_size;
 	}
 
+	reader->debug_info_loaded = true;
 	return SUCCESS_RESULT ();
 }
 
@@ -2025,10 +2069,11 @@ Result _hbc_reader_read_whole_file(HBCReader *reader, const char *filename) {
 		}
 	}
 
-	/* Read debug info */
-	result = _hbc_reader_read_debug_info (reader);
+	/* Read only the DebugInfoHeader eagerly (cheap). The body is parsed lazily
+	 * via _hbc_reader_read_debug_info() on first API use. */
+	result = _hbc_reader_read_debug_info_header (reader);
 	if (result.code != RESULT_SUCCESS) {
-		fprintf (stderr, "Error reading debug info: %s\n", result.error_message);
+		fprintf (stderr, "Error reading debug info header: %s\n", result.error_message);
 		return result;
 	}
 

src/r2/Makefile

@@ -33,6 +33,7 @@ BIN_SRC = bin_hbc.c
 ASM_SRC = asm_hbc.c
 CORE_SRC = core_hbc.c
 CORE_ONE_SRC = core_hbc_one.c
+HBC_HEADERS = $(shell find ../../include -name '*.h') utils.inc.c
 
 # Object files
 ARCH_OBJ = arch_hbc.o
@@ -107,19 +108,19 @@ $(CORE_ONE_PLUGIN): $(CORE_ONE_OBJ) $(HERMES_LIB)
 	$(CC) $(LDFLAGS) -o $@ $(CORE_ONE_OBJ) $(HERMES_LIB)
 
 # Compile object files
-$(ARCH_OBJ): $(ARCH_SRC)
+$(ARCH_OBJ): $(ARCH_SRC) $(HBC_HEADERS)
 	$(CC) $(CFLAGS) -c $< -o $@
 
-$(BIN_OBJ): $(BIN_SRC)
+$(BIN_OBJ): $(BIN_SRC) $(HBC_HEADERS)
 	$(CC) $(CFLAGS) -c $< -o $@
 
-$(ASM_OBJ): $(ASM_SRC)
+$(ASM_OBJ): $(ASM_SRC) $(HBC_HEADERS)
 	$(CC) $(CFLAGS) -c $< -o $@
 
-$(CORE_OBJ): $(CORE_SRC)
+$(CORE_OBJ): $(CORE_SRC) $(HBC_HEADERS)
 	$(CC) $(CFLAGS) -c $< -o $@
 
-$(CORE_ONE_OBJ): $(CORE_ONE_SRC)
+$(CORE_ONE_OBJ): $(CORE_ONE_SRC) $(HBC_HEADERS)
 	$(CC) $(CFLAGS) -c $< -o $@
 
 

src/r2/arch_hbc.c

@@ -2,11 +2,13 @@
 
 #include <r_anal.h>
 #include <hbc/hbc.h>
+#include <hbc/literals.h>
 #include <hbc/opcodes.h>
 #include <hbc/bytecode.h>
 #include <hbc/parser.h>
 
 #define MAX_OP_SIZE 16
+#define HBC_VADDR_BASE 0x10000000
 #ifndef R_ARCH_SYNTAX_CAMEL
 #define R_ARCH_SYNTAX_CAMEL 6
 #endif
@@ -169,7 +171,7 @@ static void parse_operands_and_set_ptr(RAnalOp *op, const ut8 *bytes, ut32 size,
 				HBCStringMeta meta;
 				Result meta_result = hbc_get_string_meta (hs->hbc, string_id, &meta);
 				if (meta_result.code == RESULT_SUCCESS) {
-					op->ptr = (st64)meta.offset;
+					op->ptr = (st64)(HBC_VADDR_BASE + meta.offset);
 				}
 			}
 		}
@@ -185,7 +187,7 @@ static void parse_operands_and_set_ptr(RAnalOp *op, const ut8 *bytes, ut32 size,
 					offset = fi.offset;
 					(void)fi.size;
 					(void)fi.param_count;
-					op->ptr = (st64)offset;
+					op->ptr = (st64)(HBC_VADDR_BASE + offset);
 				}
 			}
 		}
@@ -231,8 +233,12 @@ static bool decode(RArchSession *s, RAnalOp *op, RArchDecodeMask mask) {
 		.resolve_string_ids = true,
 		.string_tables = &string_tables,
 		.hbc = hs->hbc,
-	};
-	if (mask & R_ARCH_OP_MASK_ESIL && mask & R_ARCH_OP_MASK_DISASM) {
+};
+	/* Inlining of buffer-literal contents into the disasm line is opt-in —
+	 * it makes arch.decode O(n) in literal size and breaks r2's O(1) per-op
+	 * disasm assumption. Toggle via `e hbc.inline_buffer_literals=true`
+	 * (core plugin updates a process-wide flag). */
+	if ((mask & R_ARCH_OP_MASK_DISASM) && hbc_get_inline_literals ()) {
 		ctx.build_objects = true;
 	}
 

src/r2/asm_hbc.c

@@ -5,206 +5,206 @@
 /* Pseudo transformation rules for Hermes bytecode instructions */
 static const char *pseudo_rules[] = {
 	/* Mov/Load operations */
-	"mov/2/$1 = $2",
-	"mov_long/2/$1 = $2",
-	"loadi8/2/$1 = (int8)$2",
-	"loadu8/2/$1 = (uint8)$2",
-	"loadi16/2/$1 = (int16)$2",
-	"loadu16/2/$1 = (uint16)$2",
-	"loadi32/2/$1 = (int32)$2",
-	"loadu32/2/$1 = (uint32)$2",
-	"load_const_uint8/2/$1 = $2",
-	"load_const_u_int8/2/$1 = $2",
-	"load_const_int/2/$1 = $2",
-	"load_const_double/2/$1 = $2",
-	"load_const_bigint/2/$1 = $2",
-	"load_const_string/2/$1 = $2",
-	"load_const_string_long_index/2/$1 = $2",
-	"load_const_empty/1/$1 = empty",
-	"load_const_undefined/1/$1 = undefined",
-	"load_const_null/1/$1 = null",
-	"load_const_true/1/$1 = true",
-	"load_const_false/1/$1 = false",
-	"load_const_zero/1/$1 = 0",
+		"mov/2/$1 = $2",
+		"mov_long/2/$1 = $2",
+		"loadi8/2/$1 = (int8)$2",
+		"loadu8/2/$1 = (uint8)$2",
+		"loadi16/2/$1 = (int16)$2",
+		"loadu16/2/$1 = (uint16)$2",
+		"loadi32/2/$1 = (int32)$2",
+		"loadu32/2/$1 = (uint32)$2",
+		"load_const_uint8/2/$1 = $2",
+		"load_const_u_int8/2/$1 = $2",
+		"load_const_int/2/$1 = $2",
+		"load_const_double/2/$1 = $2",
+		"load_const_bigint/2/$1 = $2",
+		"load_const_string/2/$1 = $2",
+		"load_const_string_long_index/2/$1 = $2",
+		"load_const_empty/1/$1 = empty",
+		"load_const_undefined/1/$1 = undefined",
+		"load_const_null/1/$1 = null",
+		"load_const_true/1/$1 = true",
+		"load_const_false/1/$1 = false",
+		"load_const_zero/1/$1 = 0",
 
 	/* Property access - get_by_id has 4 args: dest, obj, cache_idx, prop_name */
-	"get_by_id/4/$1 = $2.$4",
-	"get_by_id_long/4/$1 = $2.$4",
-	"get_by_id_short/4/$1 = $2.$4",
-	"get_by_val/3/$1 = $2[$3]",
-	"put_by_id/4/$1.$4 = $3",
-	"put_by_id_long/4/$1.$4 = $3",
-	"put_by_val/3/$1[$2] = $3",
-	"put_new_own_by_id/3/$1.$3 = $2",
-	"put_new_own_by_id_short/3/$1.$3 = $2",
-	"put_new_own_by_id_long/3/$1.$3 = $2",
-	"put_new_own_ne_by_id/3/$1.$3 = $2",
-	"put_new_own_ne_by_id_long/3/$1.$3 = $2",
-	"try_get_by_id/4/$1 = try_get($2, $4)",
-	"try_get_by_id_long/4/$1 = try_get($2, $4)",
-	"try_put_by_id/4/try_put($1, $4, $3)",
-	"try_put_by_id_long/4/try_put($1, $4, $3)",
-	"del_by_id/3/delete $1.$3",
-	"del_by_id_long/3/delete $1.$3",
-	"del_by_val/3/delete $1[$2]",
+		"get_by_id/4/$1 = $2.$4",
+		"get_by_id_long/4/$1 = $2.$4",
+		"get_by_id_short/4/$1 = $2.$4",
+		"get_by_val/3/$1 = $2[$3]",
+		"put_by_id/4/$1.$4 = $3",
+		"put_by_id_long/4/$1.$4 = $3",
+		"put_by_val/3/$1[$2] = $3",
+		"put_new_own_by_id/3/$1.$3 = $2",
+		"put_new_own_by_id_short/3/$1.$3 = $2",
+		"put_new_own_by_id_long/3/$1.$3 = $2",
+		"put_new_own_ne_by_id/3/$1.$3 = $2",
+		"put_new_own_ne_by_id_long/3/$1.$3 = $2",
+		"try_get_by_id/4/$1 = try_get($2, $4)",
+		"try_get_by_id_long/4/$1 = try_get($2, $4)",
+		"try_put_by_id/4/try_put($1, $4, $3)",
+		"try_put_by_id_long/4/try_put($1, $4, $3)",
+		"del_by_id/3/delete $1.$3",
+		"del_by_id_long/3/delete $1.$3",
+		"del_by_val/3/delete $1[$2]",
 
 	/* Arithmetic operations */
-	"add/3/$1 = $2 + $3",
-	"add_n/3/$1 = $2 + $3",
-	"add_32/3/$1 = (int32)($2 + $3)",
-	"add_empty_string/2/$1 = \"\" + $2",
-	"sub/3/$1 = $2 - $3",
-	"sub_n/3/$1 = $2 - $3",
-	"sub_32/3/$1 = (int32)($2 - $3)",
-	"mul/3/$1 = $2 * $3",
-	"mul_n/3/$1 = $2 * $3",
-	"mul_32/3/$1 = (int32)($2 * $3)",
-	"div/3/$1 = $2 / $3",
-	"div_n/3/$1 = $2 / $3",
-	"divi32/3/$1 = (int32)($2 / $3)",
-	"divu32/3/$1 = (uint32)($2 / $3)",
-	"mod/3/$1 = $2 % $3",
-	"inc/2/$1 = $2 + 1",
-	"dec/2/$1 = $2 - 1",
-	"negate/2/$1 = -$2",
+		"add/3/$1 = $2 + $3",
+		"add_n/3/$1 = $2 + $3",
+		"add_32/3/$1 = (int32)($2 + $3)",
+		"add_empty_string/2/$1 = \"\" + $2",
+		"sub/3/$1 = $2 - $3",
+		"sub_n/3/$1 = $2 - $3",
+		"sub_32/3/$1 = (int32)($2 - $3)",
+		"mul/3/$1 = $2 * $3",
+		"mul_n/3/$1 = $2 * $3",
+		"mul_32/3/$1 = (int32)($2 * $3)",
+		"div/3/$1 = $2 / $3",
+		"div_n/3/$1 = $2 / $3",
+		"divi32/3/$1 = (int32)($2 / $3)",
+		"divu32/3/$1 = (uint32)($2 / $3)",
+		"mod/3/$1 = $2 % $3",
+		"inc/2/$1 = $2 + 1",
+		"dec/2/$1 = $2 - 1",
+		"negate/2/$1 = -$2",
 
 	/* Bitwise operations */
-	"bit_and/3/$1 = $2 & $3",
-	"bit_or/3/$1 = $2 | $3",
-	"bit_xor/3/$1 = $2 ^ $3",
-	"bit_not/2/$1 = ~$2",
-	"lshift/3/$1 = $2 << $3",
-	"rshift/3/$1 = $2 >> $3",
-	"urshift/3/$1 = (unsigned)$2 >> $3",
+		"bit_and/3/$1 = $2 & $3",
+		"bit_or/3/$1 = $2 | $3",
+		"bit_xor/3/$1 = $2 ^ $3",
+		"bit_not/2/$1 = ~$2",
+		"lshift/3/$1 = $2 << $3",
+		"rshift/3/$1 = $2 >> $3",
+		"urshift/3/$1 = (unsigned)$2 >> $3",
 
 	/* Logical operations */
-	"not/2/$1 = !$2",
-	"eq/3/$1 = $2 == $3",
-	"strict_eq/3/$1 = $2 === $3",
-	"neq/3/$1 = $2 != $3",
-	"strict_neq/3/$1 = $2 !== $3",
-	"less/3/$1 = $2 < $3",
-	"greater/3/$1 = $2 > $3",
-	"less_eq/3/$1 = $2 <= $3",
-	"greater_eq/3/$1 = $2 >= $3",
+		"not/2/$1 = !$2",
+		"eq/3/$1 = $2 == $3",
+		"strict_eq/3/$1 = $2 === $3",
+		"neq/3/$1 = $2 != $3",
+		"strict_neq/3/$1 = $2 !== $3",
+		"less/3/$1 = $2 < $3",
+		"greater/3/$1 = $2 > $3",
+		"less_eq/3/$1 = $2 <= $3",
+		"greater_eq/3/$1 = $2 >= $3",
 
 	/* Type operations */
-	"typeof/2/$1 = typeof($2)",
-	"is_in/3/$1 = ($2 in $3)",
-	"instanceof/3/$1 = ($2 instanceof $3)",
-	"to_number/2/$1 = Number($2)",
-	"to_numeric/2/$1 = Numeric($2)",
-	"to_int32/2/$1 = (int32)$2",
+		"typeof/2/$1 = typeof($2)",
+		"is_in/3/$1 = ($2 in $3)",
+		"instanceof/3/$1 = ($2 instanceof $3)",
+		"to_number/2/$1 = Number($2)",
+		"to_numeric/2/$1 = Numeric($2)",
+		"to_int32/2/$1 = (int32)$2",
 
 	/* Object/Array creation */
-	"new_object/1/$1 = {}",
-	"new_object_with_parent/2/$1 = Object.create($2)",
-	"new_object_with_buffer/5/$1 = {/*buffer*/}",
-	"new_object_with_buffer_long/5/$1 = {/*buffer*/}",
-	"new_array/2/$1 = []",
-	"new_array_with_buffer/4/$1 = [/*buffer*/]",
-	"new_array_with_buffer_long/4/$1 = [/*buffer*/]",
-	"create_closure/3/$1 = closure($3)",
-	"create_closure_long_index/3/$1 = closure($3)",
-	"create_generator_closure/3/$1 = gen_closure($3)",
-	"create_generator/3/$1 = generator($3)",
-	"create_regexp/3/$1 = regexp($3)",
+		"new_object/1/$1 = {}",
+		"new_object_with_parent/2/$1 = Object.create($2)",
+		"new_object_with_buffer/5/$1 = {/*buffer*/}",
+		"new_object_with_buffer_long/5/$1 = {/*buffer*/}",
+		"new_array/2/$1 = []",
+		"new_array_with_buffer/4/$1 = [/*buffer*/]",
+		"new_array_with_buffer_long/4/$1 = [/*buffer*/]",
+		"create_closure/3/$1 = closure($3)",
+		"create_closure_long_index/3/$1 = closure($3)",
+		"create_generator_closure/3/$1 = gen_closure($3)",
+		"create_generator/3/$1 = generator($3)",
+		"create_regexp/3/$1 = regexp($3)",
 
 	/* Control flow */
-	"unreachable/0/unreachable",
-	"jmp/1/goto $1",
-	"jmp_long/1/goto $1",
-	"name/3/if ($2) goto $1",
-	"name##_long/2/if ($2) goto $1",
-	"jmp_true/2/if ($2) goto $1",
-	"jmp_true_long/2/if ($2) goto $1",
-	"jmp_false/2/if (!$2) goto $1",
-	"jmp_false_long/2/if (!$2) goto $1",
-	"jmp_undefined/2/if ($2 === undefined) goto $1",
-	"jmp_undefined_long/2/if ($2 === undefined) goto $1",
-	"jless/3/if ($1 < $2) goto $3",
-	"jless_long/3/if ($1 < $2) goto $3",
-	"jless_eq/3/if ($1 <= $2) goto $3",
-	"jless_eq_long/3/if ($1 <= $2) goto $3",
-	"jgreater/3/if ($1 > $2) goto $3",
-	"jgreater_long/3/if ($1 > $2) goto $3",
-	"jgreater_eq/3/if ($1 >= $2) goto $3",
-	"jgreater_eq_long/3/if ($1 >= $2) goto $3",
-	"jequal/3/if ($1 == $2) goto $3",
-	"jequal_long/3/if ($1 == $2) goto $3",
-	"jstrict_equal/3/if ($1 === $2) goto $3",
-	"jstrict_equal_long/3/if ($1 === $2) goto $3",
-	"jnot_equal/3/if ($1 != $2) goto $3",
-	"jnot_equal_long/3/if ($1 != $2) goto $3",
-	"jstrict_not_equal/3/if ($1 !== $2) goto $3",
-	"jstrict_not_equal_long/3/if ($1 !== $2) goto $3",
+		"unreachable/0/unreachable",
+		"jmp/1/goto $1",
+		"jmp_long/1/goto $1",
+		"name/3/if ($2) goto $1",
+		"name##_long/2/if ($2) goto $1",
+		"jmp_true/2/if ($2) goto $1",
+		"jmp_true_long/2/if ($2) goto $1",
+		"jmp_false/2/if (!$2) goto $1",
+		"jmp_false_long/2/if (!$2) goto $1",
+		"jmp_undefined/2/if ($2 === undefined) goto $1",
+		"jmp_undefined_long/2/if ($2 === undefined) goto $1",
+		"jless/3/if ($1 < $2) goto $3",
+		"jless_long/3/if ($1 < $2) goto $3",
+		"jless_eq/3/if ($1 <= $2) goto $3",
+		"jless_eq_long/3/if ($1 <= $2) goto $3",
+		"jgreater/3/if ($1 > $2) goto $3",
+		"jgreater_long/3/if ($1 > $2) goto $3",
+		"jgreater_eq/3/if ($1 >= $2) goto $3",
+		"jgreater_eq_long/3/if ($1 >= $2) goto $3",
+		"jequal/3/if ($1 == $2) goto $3",
+		"jequal_long/3/if ($1 == $2) goto $3",
+		"jstrict_equal/3/if ($1 === $2) goto $3",
+		"jstrict_equal_long/3/if ($1 === $2) goto $3",
+		"jnot_equal/3/if ($1 != $2) goto $3",
+		"jnot_equal_long/3/if ($1 != $2) goto $3",
+		"jstrict_not_equal/3/if ($1 !== $2) goto $3",
+		"jstrict_not_equal_long/3/if ($1 !== $2) goto $3",
 
 	/* Function calls and returns */
 	/* call: result = callee (this, args...) */
-	"call/2/$1($2)",
-	"call1/3/$1 = $2($3)",
-	"call2/4/$1 = $2($4)",
-	"call3/5/$1 = $2($4, $5)",
-	"call4/6/$1 = $2($4, $5, $6)",
-	"call_long/2/$1($2)",
-	"call_builtin/3/$1 = builtin.$2($3)",
-	"call_builtin_long/3/$1 = builtin.$2($3)",
-	"call_direct/3/$1 = $2($3)",
-	"call_direct_long_index/3/$1 = $2($3)",
-	"construct/3/$1 = new $2($3)",
-	"construct_long/3/$1 = new $2($3)",
-	"ret/1/return $1",
-	"ret_undefined/0/return",
+		"call/2/$1($2)",
+		"call1/3/$1 = $2($3)",
+		"call2/4/$1 = $2($4)",
+		"call3/5/$1 = $2($4, $5)",
+		"call4/6/$1 = $2($4, $5, $6)",
+		"call_long/2/$1($2)",
+		"call_builtin/3/$1 = builtin.$2($3)",
+		"call_builtin_long/3/$1 = builtin.$2($3)",
+		"call_direct/3/$1 = $2($3)",
+		"call_direct_long_index/3/$1 = $2($3)",
+		"construct/3/$1 = new $2($3)",
+		"construct_long/3/$1 = new $2($3)",
+		"ret/1/return $1",
+		"ret_undefined/0/return",
 
 	/* Exception handling */
-	"throw/1/throw $1",
-	"throw_if_empty/2/if (empty($1)) throw $2",
-	"catch/2/$1 = catch $2",
+		"throw/1/throw $1",
+		"throw_if_empty/2/if (empty($1)) throw $2",
+		"catch/2/$1 = catch $2",
 
 	/* Parameter handling */
-	"load_param/2/$1 = param[$2]",
-	"load_param_long/2/$1 = param[$2]",
-	"reify_arguments/1/arguments = reify($1)",
-	"get_arguments_length/2/$1 = arguments.length",
-	"get_arguments_prop_by_val/3/$1 = arguments[$2]",
+		"load_param/2/$1 = param[$2]",
+		"load_param_long/2/$1 = param[$2]",
+		"reify_arguments/1/arguments = reify($1)",
+		"get_arguments_length/2/$1 = arguments.length",
+		"get_arguments_prop_by_val/3/$1 = arguments[$2]",
 
 	/* Environment operations */
-	"get_environment/2/$1 = env[$2]",
-	"load_from_environment/3/$1 = $2[$3]",
-	"load_from_environment_l/3/$1 = $2[$3]",
-	"store_to_environment/3/$2[$3] = $1",
-	"store_to_environment_l/3/$2[$3] = $1",
-	"store_np_to_environment/3/$2[$3] = $1",
-	"store_np_to_environment_l/3/$2[$3] = $1",
-	"create_environment/1/$1 = new_env()",
-	"create_inner_environment/1/$1 = new_inner_env()",
+		"get_environment/2/$1 = env[$2]",
+		"load_from_environment/3/$1 = $2[$3]",
+		"load_from_environment_l/3/$1 = $2[$3]",
+		"store_to_environment/3/$2[$3] = $1",
+		"store_to_environment_l/3/$2[$3] = $1",
+		"store_np_to_environment/3/$2[$3] = $1",
+		"store_np_to_environment_l/3/$2[$3] = $1",
+		"create_environment/1/$1 = new_env()",
+		"create_inner_environment/1/$1 = new_inner_env()",
 
 	/* Generator operations */
-	"start_generator/0/start_generator()",
-	"resume_generator/2/$1 = resume_generator($2)",
-	"save_generator/1/save_generator($1)",
-	"complete_generator/0/complete_generator()",
+		"start_generator/0/start_generator()",
+		"resume_generator/2/$1 = resume_generator($2)",
+		"save_generator/1/save_generator($1)",
+		"complete_generator/0/complete_generator()",
 
 	/* Miscellaneous */
-	"debugger/0/debugger",
-	"profile_point/1/profile($1)",
-	"select_object/3/$1 = select($2, $3)",
-	"catch/2/$1 = catch_value($2)",
-	"get_builtin_closure/2/$1 = builtin_closure($2)",
-	"get_global_object/1/$1 = global",
-	"get_new_target/1/$1 = new.target",
-	"get_this_ns/1/$1 = this",
-	"coerce_this_ns/1/$1 = coerce_this()",
+		"debugger/0/debugger",
+		"profile_point/1/profile($1)",
+		"select_object/3/$1 = select($2, $3)",
+		"catch/2/$1 = catch_value($2)",
+		"get_builtin_closure/2/$1 = builtin_closure($2)",
+		"get_global_object/1/$1 = global",
+		"get_new_target/1/$1 = new.target",
+		"get_this_ns/1/$1 = this",
+		"coerce_this_ns/1/$1 = coerce_this()",
 
 	/* Iterator operations */
-	"iterator_begin/2/$1 = iterator_begin($2)",
-	"iterator_next/3/$1 = iterator_next($2, $3)",
-	"iterator_close/2/iterator_close($1, $2)",
-	"get_pname_list/2/$1 = property_names($2)",
-	"get_next_pname/4/$1 = next_property_name($2, $3, $4)",
+		"iterator_begin/2/$1 = iterator_begin($2)",
+		"iterator_next/3/$1 = iterator_next($2, $3)",
+		"iterator_close/2/iterator_close($1, $2)",
+		"get_pname_list/2/$1 = property_names($2)",
+		"get_next_pname/4/$1 = next_property_name($2, $3, $4)",
 
 	/* Switch */
-	"switch_imm/3/switch($1) { cases in $2 }",
+		"switch_imm/3/switch($1) { cases in $2 }",
 
 	NULL
 };

src/r2/bin_hbc.c

@@ -1,7 +1,9 @@
 /* radare2 - LGPL - Copyright 2025-2026 - pancake */
 
 #include <r_bin.h>
+#include <r_bin_dwarf.h>
 #include <hbc/hbc.h>
+#include <hbc/literals.h>
 #include "utils.inc.c"
 
 #define HEADER_MAGIC 0x1f1903c103bc1fc6ULL
@@ -131,7 +133,7 @@ static ut64 resolve_entrypoint(RBinFile *bf, HBC *hbc) {
 	return 0;
 }
 
-static void fill_info(RBinInfo *ret, const char *file_path, bool has_version, ut32 version) {
+static void fill_info(RBinInfo *ret, const char *file_path, bool has_version, ut32 version, bool has_source_lines) {
 	ret->file = file_path? strdup (file_path): NULL;
 	ret->bclass = strdup ("Hermes bytecode");
 	ret->rclass = strdup ("hermes");
@@ -141,12 +143,15 @@ static void fill_info(RBinInfo *ret, const char *file_path, bool has_version, ut
 	ret->type = strdup ("Hermes bytecode");
 	ret->machine = strdup ("Hermes VM");
 	ret->cpu = has_version? r_str_newf ("%u", version): strdup ("unknown");
+	ret->has_va = true;
+	ret->dbg_info = has_source_lines? R_BIN_DBG_LINENUMS: R_BIN_DBG_STRIPPED;
 }
 
 static RBinInfo *bininfo(RBinFile *bf) {
 	RBinInfo *ret = R_NEW0 (RBinInfo);
 	bool has_version = false;
 	ut32 version = 0;
+	bool has_source_lines = false;
 
 	HBC *hbc = get_hbc (bf);
 	if (hbc) {
@@ -155,6 +160,10 @@ static RBinInfo *bininfo(RBinFile *bf) {
 			has_version = true;
 			version = hh.version;
 		}
+		/* Cheap: answered from the DebugInfoHeader parsed at load time.
+		 * Reflects file truth (does this file carry source lines?) regardless
+		 * of bin.dbginfo — that setting only gates whether we decode them. */
+		has_source_lines = hbc_has_source_lines (hbc);
 	}
 
 	/* Fallback: parse version directly from buffer */
@@ -167,10 +176,30 @@ static RBinInfo *bininfo(RBinFile *bf) {
 		}
 	}
 
-	fill_info (ret, bf->file, has_version, version);
+	fill_info (ret, bf->file, has_version, version, has_source_lines);
 	return ret;
 }
 
+/* Helper: append a section if the pool is present. Owns no strings — `name`
+ * is duplicated into a heap-allocated section name via strdup. */
+static void add_pool_section(RList *list, const char *name, ut64 paddr, ut64 size) {
+	if (!size || !paddr) {
+		return;
+	}
+	RBinSection *s = R_NEW0 (RBinSection);
+	if (!s) {
+		return;
+	}
+	s->name = strdup (name);
+	s->size = size;
+	s->vsize = size;
+	s->paddr = paddr;
+	s->vaddr = HBC_VADDR_BASE + paddr;
+	s->perm = R_PERM_R;
+	s->add = true;
+	r_list_append (list, s);
+}
+
 static RList *sections(RBinFile *bf) {
 	RList *sections = r_list_newf ((RListFree)free);
 	if (!sections) {
@@ -183,9 +212,25 @@ static RList *sections(RBinFile *bf) {
 	section->vsize = section->size;
 	section->paddr = 0;
 	section->vaddr = HBC_VADDR_BASE;
-	section->perm = R_PERM_R;
+	section->perm = R_PERM_RX;
+	section->add = true;
 	r_list_append (sections, section);
 
+	/* SLP pool subsections — gives `iS` a quick map of where the buffer
+	 * literals live, and lets users `s` / `pd @` straight into them. */
+	HBC *hbc = get_hbc (bf);
+	if (hbc) {
+		add_pool_section (sections, "slp.arrays",
+			hbc_get_pool_paddr (hbc, HBC_LIT_ARRAY),
+			hbc_get_pool_size (hbc, HBC_LIT_ARRAY));
+		add_pool_section (sections, "slp.object_keys",
+			hbc_get_pool_paddr (hbc, HBC_LIT_OBJECT),
+			hbc_get_pool_size (hbc, HBC_LIT_OBJECT));
+		add_pool_section (sections, "slp.object_values",
+			hbc_get_object_values_paddr (hbc),
+			hbc_get_object_values_size (hbc));
+	}
+
 	return sections;
 }
 
@@ -210,6 +255,33 @@ static ut64 baddr(RBinFile *bf R_UNUSED) {
 	return HBC_VADDR_BASE;
 }
 
+static void append_binding_symbols(RList *symbols, HBC *hbc) {
+	HBCBindings bindings = { 0 };
+	if (hbc_scan_bindings (hbc, &bindings).code != RESULT_SUCCESS) {
+		return;
+	}
+	for (u32 i = 0; i < bindings.count; i++) {
+		HBCBinding *b = &bindings.bindings[i];
+		if (b->type != HBC_BINDING_EXPORT) {
+			continue;
+		}
+		RBinSymbol *sym = R_NEW0 (RBinSymbol);
+		char *nm = r_str_newf ("hbc.export.%s.%s", b->kind? b->kind: "js", b->name);
+		sym->name = r_bin_name_new (nm);
+		r_bin_name_filtered (sym->name, nm);
+		sym->paddr = b->offset;
+		sym->vaddr = HBC_VADDR_BASE + b->offset;
+		sym->size = 1;
+		sym->ordinal = b->string_id;
+		sym->type = R_BIN_TYPE_OBJECT_STR;
+		sym->bind = R_BIN_BIND_GLOBAL_STR;
+		sym->bits = 32;
+		r_list_append (symbols, sym);
+		free (nm);
+	}
+	hbc_free_bindings (&bindings);
+}
+
 static RList *symbols(RBinFile *bf) {
 	RList *symbols = r_list_newf ((RListFree)r_bin_symbol_free);
 	HBC *hbc = get_hbc (bf);
@@ -263,9 +335,118 @@ static RList *symbols(RBinFile *bf) {
 		free (san);
 	}
 
+	/* SLP pool groups as symbols — one per group discovered by the
+	 * pool-side linear scan. Lets users see `lit.arr.* / lit.obj.*` in `is`
+	 * without needing to run pd:hLs. The scan is linear in pool size and
+	 * runs at RBin load time. */
+	for (int k = 0; k < 2; k++) {
+		HBCLiteralKind kind = k? HBC_LIT_OBJECT: HBC_LIT_ARRAY;
+		const char *prefix = k? "lit.obj.": "lit.arr.";
+		HBCPoolGroup *groups = NULL;
+		u32 n = 0;
+		if (hbc_literals_scan_pool (hbc, kind, &groups, &n).code
+				!= RESULT_SUCCESS || !groups) {
+			continue;
+		}
+		for (u32 i = 0; i < n; i++) {
+			RBinSymbol *sym = R_NEW0 (RBinSymbol);
+			char *nm = r_str_newf ("%s0x%x", prefix, groups[i].paddr);
+			sym->name = r_bin_name_new (nm);
+			r_bin_name_filtered (sym->name, nm);
+			sym->paddr = groups[i].paddr;
+			sym->vaddr = HBC_VADDR_BASE + groups[i].paddr;
+			sym->size = 1;
+			sym->type = R_BIN_TYPE_OBJECT_STR;
+			sym->bits = 32;
+			r_list_append (symbols, sym);
+			free (nm);
+		}
+		free (groups);
+	}
+
+	append_binding_symbols (symbols, hbc);
 	return symbols;
 }
 
+static RList *imports(RBinFile *bf) {
+	RList *imports = r_list_newf ((RListFree)r_bin_import_free);
+	HBC *hbc = get_hbc (bf);
+	if (!imports || !hbc) {
+		return imports;
+	}
+	HBCBindings bindings = { 0 };
+	if (hbc_scan_bindings (hbc, &bindings).code != RESULT_SUCCESS) {
+		return imports;
+	}
+	for (u32 i = 0; i < bindings.count; i++) {
+		HBCBinding *b = &bindings.bindings[i];
+		if (b->type != HBC_BINDING_IMPORT) {
+			continue;
+		}
+		RBinImport *imp = R_NEW0 (RBinImport);
+		char *nm = r_str_newf ("%s.%s", b->kind? b->kind: "js", b->name);
+		imp->name = r_bin_name_new (nm);
+		r_bin_name_filtered (imp->name, nm);
+		imp->libname = strdup (b->module? b->module: "hermes");
+		imp->bind = "NONE";
+		imp->type = "FUNC";
+		imp->ordinal = b->string_id;
+		imp->is_imported = true;
+		r_list_append (imports, imp);
+		free (nm);
+	}
+	hbc_free_bindings (&bindings);
+	return imports;
+}
+
+static R_UNOWNED RList *lines(RBinFile *bf) {
+	HBC *hbc = get_hbc (bf);
+	if (!hbc) {
+		return NULL;
+	}
+	/* Respect bin.dbginfo=false — return early so the lazy debug-info parse
+	 * (triggered by hbc_get_source_lines) never runs for users who opted out. */
+	if (bf->rbin && !bf->rbin->want_dbginfo) {
+		return NULL;
+	}
+	HBCSourceLineArray lines = { 0 };
+	RList *ret = NULL;
+	if (hbc_get_source_lines (hbc, &lines).code != RESULT_SUCCESS || !lines.count) {
+		goto done;
+	}
+	ret = r_list_newf (free);
+	if (!ret) {
+		goto done;
+	}
+	const bool has_al = bf->addrline.al_add && bf->addrline.al_get;
+	for (u32 i = 0; i < lines.count; i++) {
+		HBCSourceLine *sl = &lines.lines[i];
+		ut64 addr = HBC_VADDR_BASE + sl->address;
+		const char *file = sl->filename? sl->filename: "";
+		if (has_al) {
+			bf->addrline.al_add (&bf->addrline, addr, file, NULL, sl->line, sl->column);
+		}
+		RBinAddrline *row = R_NEW0 (RBinAddrline);
+		if (!row) {
+			continue;
+		}
+		const RBinAddrline *stored = has_al? bf->addrline.al_get (&bf->addrline, addr): NULL;
+		if (stored) {
+			*row = *stored;
+		} else {
+			row->addr = addr;
+			row->file = UT32_MAX;
+			row->path = UT32_MAX;
+			row->line = sl->line;
+			row->column = sl->column;
+		}
+		r_list_append (ret, row);
+	}
+done:
+	hbc_free_source_lines (&lines);
+	return ret;
+}
+
 static RList *strings(RBinFile *bf) {
 	RList *ret = r_list_newf ((RListFree)r_bin_string_free);
 	HBC *hbc = get_hbc (bf);
@@ -304,7 +485,7 @@ static RList *strings(RBinFile *bf) {
 		}
 
 		ptr->paddr = meta.offset;
-		ptr->vaddr = meta.offset;
+		ptr->vaddr = HBC_VADDR_BASE + meta.offset;
 		ptr->size = meta.isUTF16? (meta.length * 2): meta.length;
 		ptr->length = meta.length;
 		ptr->ordinal = i;
@@ -329,6 +510,8 @@ const RBinPlugin r_bin_plugin_r2hermes = {
 	.sections = &sections,
 	.baddr = &baddr,
 	.symbols = &symbols,
+	.imports = &imports,
+	.lines = &lines,
 	.strings = &strings,
 };
 

src/r2/core_hbc.c

@@ -4,8 +4,11 @@
 
 #include <r_core.h>
 #include <hbc/hbc.h>
+#include <hbc/literals.h>
 #include "utils.inc.c"
 
+#define HBC_VADDR_BASE 0x10000000ULL
+
 /* Plugin registration - need these when HBC_CORE_REGISTER_PLUGINS is enabled */
 #ifdef HBC_CORE_REGISTER_PLUGINS
 extern const RArchPlugin r_arch_plugin_r2hermes;
@@ -139,7 +142,9 @@ static HBCDecompOptions make_decompile_options(RCore *core, bool show_offsets, u
 		.skip_pass4_closure = r_config_get_b (core->config, "hbc.skip_pass4"),
 		.force_dispatch = r_config_get_b (core->config, "hbc.force_dispatch"),
 		.inline_closures = r_config_get_b (core->config, "hbc.inline_closures"),
-		.inline_threshold = r_config_get_i (core->config, "hbc.inline_threshold")
+		.inline_threshold = r_config_get_i (core->config, "hbc.inline_threshold"),
+		.max_ast_statements = r_config_get_i (core->config, "hbc.max_ast"),
+		.max_output_bytes = r_config_get_i (core->config, "hbc.max_bytes")
 	};
 	return opts;
 }
@@ -277,19 +282,76 @@ static void cmd_file_info(HbcContext *ctx, RCore *core) {
 		"  File Length: %u bytes\n"
 		"  Functions: %u\n"
 		"  Strings: %u\n"
+		"  String Kinds: %u\n"
+		"  Overflow Strings: %u\n"
+		"  String Storage Size: %u\n"
 		"  Identifiers: %u\n"
 		"  Global Code Index: %u\n"
+		"  Function Sources: %u\n"
+		"  Debug Info Offset: %u\n"
+		"  BigInts: %u (%u bytes)\n"
+		"  RegExps: %u (%u bytes)\n"
+		"  Literal Value Buffer: %u bytes\n"
+		"  Array Buffer: %u bytes\n"
+		"  Object Key Buffer: %u bytes\n"
+		"  Object Value Buffer: %u bytes\n"
+		"  Object Shapes: %u\n"
+		"  CJS Modules: %u\n"
 		"  Static Builtins: %s\n"
+		"  CJS Modules Statically Resolved: %s\n"
 		"  Has Async: %s\n",
 		header.version,
 		header.fileLength,
 		header.functionCount,
 		header.stringCount,
+		header.stringKindCount,
+		header.overflowStringCount,
+		header.stringStorageSize,
 		header.identifierCount,
 		header.globalCodeIndex,
+		header.functionSourceCount,
+		header.debugInfoOffset,
+		header.bigIntCount,
+		header.bigIntStorageSize,
+		header.regExpCount,
+		header.regExpStorageSize,
+		header.literalValueBufferSize,
+		header.arrayBufferSize,
+		header.objKeyBufferSize,
+		header.objValueBufferSize,
+		header.objShapeTableCount,
+		header.cjsModuleCount,
 		header.staticBuiltins? "yes": "no",
+		header.cjsModulesStaticallyResolved? "yes": "no",
 		header.hasAsync? "yes": "no");
 
+	HBCDebugInfo di = { 0 };
+	if (hbc_get_debug_info (ctx->hbc, &di).code == RESULT_SUCCESS) {
+		r_cons_printf (cons,
+			"  Debug Info Present: %s\n",
+			di.has_debug_info? "yes": "no");
+		if (di.has_debug_info) {
+			r_cons_printf (cons,
+				"  Debug Files: %u (%u bytes)\n"
+				"  Debug File Regions: %u\n"
+				"  Functions With Debug Info: %u\n"
+				"  Source Locations: %u bytes\n"
+				"  Scope Descriptors: %u bytes\n"
+				"  Textified Callees: %u bytes\n"
+				"  Debug String Table: %u bytes\n"
+				"  Debug Data Size: %u bytes\n",
+				di.filename_count,
+				di.filename_storage_size,
+				di.file_region_count,
+				di.functions_with_debug_info,
+				di.source_locations_size,
+				di.scope_desc_data_size,
+				di.textified_data_size,
+				di.string_table_size,
+				di.debug_data_size);
+		}
+	}
+
 	char hex[41] = { 0 };
 	r_hex_bin2str (header.sourceHash, 20, hex);
 	r_cons_printf (cons,
@@ -355,62 +417,428 @@ static void cmd_file_info(HbcContext *ctx, RCore *core) {
 	}
 }
 
-/* JSON output for current function */
-static void cmd_json(HbcContext *ctx, RCore *core, const char *addr_str) {
+static void cmd_source_lines(HbcContext *ctx, RCore *core, const char *arg) {
 	Result res = hbc_load_current_binary (ctx, core);
 	if (res.code != RESULT_SUCCESS) {
-		PJ *pj = r_core_pj_new (core);
-		pj_o (pj);
-		pj_ks (pj, "error", safe_errmsg (res.error_message));
-		pj_end (pj);
-		char *s = pj_drain (pj);
-		r_cons_println (core->cons, s);
-		free (s);
+		R_LOG_ERROR ("%s", safe_errmsg (res.error_message));
 		return;
 	}
-
-	u32 function_id = parse_function_id (addr_str);
-	HBCDecompOptions opts = { .pretty_literals = true, .suppress_comments = false };
-	u32 count = hbc_function_count (ctx->hbc);
+	HBCSourceLineArray lines = { 0 };
+	res = hbc_get_source_lines (ctx->hbc, &lines);
 	if (res.code != RESULT_SUCCESS) {
-		PJ *pj = r_core_pj_new (core);
-		pj_o (pj);
-		pj_ks (pj, "error", "cannot get function count");
-		pj_end (pj);
-		char *s = pj_drain (pj);
-		r_cons_println (core->cons, s);
-		free (s);
+		R_LOG_ERROR ("source lines: %s", safe_errmsg (res.error_message));
+		hbc_free_source_lines (&lines);
 		return;
 	}
+	const char *as = r_str_trim_head_ro (arg);
+	const bool filter = *as != '\0';
+	const ut32 filter_id = filter? parse_function_id (as): 0;
+	for (u32 i = 0; i < lines.count; i++) {
+		const HBCSourceLine *sl = &lines.lines[i];
+		if (filter && sl->function_id != filter_id) {
+			continue;
+		}
+		r_cons_printf (core->cons, "0x%08x f=%u +0x%x %s:%u:%u stmt=%u\n",
+			sl->address, sl->function_id, sl->function_address,
+			sl->filename? sl->filename: "",
+			sl->line, sl->column, sl->statement);
+	}
+	hbc_free_source_lines (&lines);
+}
+
+/* Emit a JSON error envelope compatible with the {code, annotations, errors} format */
+static void emit_json_error(RCore *core, const char *msg) {
 	PJ *pj = r_core_pj_new (core);
 	pj_o (pj);
-	pj_kn (pj, "function_id", function_id);
-	pj_k (pj, "decompilation");
-	if (function_id >= count) {
-		pj_null (pj);
-		pj_ks (pj, "error", "function id out of range");
-		pj_kn (pj, "count", count);
-		pj_end (pj);
-		char *s = pj_drain (pj);
-		r_cons_println (core->cons, s);
-		free (s);
-		return;
-	}
-	char *output = NULL;
-	res = hbc_decomp_fn (ctx->hbc, function_id, opts, &output);
-
-	if (res.code == RESULT_SUCCESS && output) {
-		pj_s (pj, output);
-	} else {
-		pj_null (pj);
-	}
+	pj_ks (pj, "code", "");
+	pj_ka (pj, "annotations");
+	pj_end (pj);
+	pj_ka (pj, "errors");
+	pj_s (pj, msg);
+	pj_end (pj);
 	pj_end (pj);
 	char *s = pj_drain (pj);
 	r_cons_println (core->cons, s);
 	free (s);
+}
+
+typedef struct {
+	size_t start;
+	size_t end;
+	u64 offset;
+} LineAnn;
+
+/* JSON output for a function in the codemeta-compatible {code, annotations, errors} shape.
+ * Uses show_offsets=true internally and parses the "0xHHHHHHHH: " line prefixes to
+ * produce per-line offset annotations, then strips the prefix from the emitted code. */
+static void cmd_json(HbcContext *ctx, RCore *core, const char *addr_str) {
+	Result res = hbc_load_current_binary (ctx, core);
+	if (res.code != RESULT_SUCCESS) {
+		emit_json_error (core, safe_errmsg (res.error_message));
+		return;
+	}
+
+	u32 count = hbc_function_count (ctx->hbc);
+	u32 function_id = 0;
+	const char *as = r_str_trim_head_ro (addr_str);
+	if (*as) {
+		function_id = parse_function_id (addr_str);
+	} else if (find_function_at_offset (ctx, (u32)core->addr, &function_id) != 0) {
+		emit_json_error (core, "No function at current offset");
+		return;
+	}
+	if (function_id >= count) {
+		emit_json_error (core, "function id out of range");
+		return;
+	}
+
+	HBCFunc fi;
+	Result fres = hbc_get_function_info (ctx->hbc, function_id, &fi);
+	u64 func_base = (fres.code == RESULT_SUCCESS)? fi.offset: 0;
+
+	HBCDecompOptions opts = make_decompile_options (core, true, func_base);
+	opts.pretty_literals = true;
+
+	char *output = NULL;
+	res = hbc_decomp_fn (ctx->hbc, function_id, opts, &output);
+	if (res.code != RESULT_SUCCESS || !output) {
+		emit_json_error (core, safe_errmsg (res.error_message));
+		free (output);
+		return;
+	}
+
+	RStrBuf *code_buf = r_strbuf_new ("");
+	LineAnn *anns = NULL;
+	size_t anns_n = 0, anns_cap = 0;
+
+	const char *p = output;
+	while (*p) {
+		const char *eol = strchr (p, '\n');
+		size_t line_len = eol? (size_t) (eol - p): strlen (p);
+
+		bool has_offset = false;
+		u64 line_offset = 0;
+		if (line_len >= 12 && p[0] == '0' && p[1] == 'x' && p[10] == ':' && p[11] == ' ') {
+			bool valid = true;
+			for (int i = 2; i < 10; i++) {
+				if (!isxdigit ((unsigned char)p[i])) {
+					valid = false;
+					break;
+				}
+			}
+			if (valid) {
+				char hexbuf[9];
+				memcpy (hexbuf, p + 2, 8);
+				hexbuf[8] = 0;
+				line_offset = strtoull (hexbuf, NULL, 16);
+				has_offset = true;
+			}
+		}
+
+		size_t start = r_strbuf_length (code_buf);
+		if (has_offset) {
+			r_strbuf_append_n (code_buf, p + 12, line_len - 12);
+		} else {
+			r_strbuf_append_n (code_buf, p, line_len);
+		}
+		r_strbuf_append (code_buf, "\n");
+		size_t end = r_strbuf_length (code_buf);
+
+		if (has_offset) {
+			if (anns_n == anns_cap) {
+				anns_cap = anns_cap? anns_cap * 2: 64;
+				LineAnn *grown = realloc (anns, anns_cap * sizeof (LineAnn));
+				if (!grown) {
+					break;
+				}
+				anns = grown;
+			}
+			anns[anns_n].start = start;
+			anns[anns_n].end = end;
+			anns[anns_n].offset = line_offset;
+			anns_n++;
+		}
+
+		if (!eol) {
+			break;
+		}
+		p = eol + 1;
+	}
+
+	char *cleaned = r_strbuf_drain (code_buf);
+
+	PJ *pj = r_core_pj_new (core);
+	pj_o (pj);
+	pj_ks (pj, "code", cleaned? cleaned: "");
+	pj_ka (pj, "annotations");
+	for (size_t i = 0; i < anns_n; i++) {
+		pj_o (pj);
+		pj_ks (pj, "type", "offset");
+		pj_kn (pj, "start", anns[i].start);
+		pj_kn (pj, "end", anns[i].end);
+		pj_kn (pj, "offset", anns[i].offset);
+		pj_end (pj);
+	}
+	pj_end (pj);
+	pj_ka (pj, "errors");
+	pj_end (pj);
+	pj_end (pj);
+	char *s = pj_drain (pj);
+	r_cons_println (core->cons, s);
+	free (s);
+	free (cleaned);
+	free (anns);
 	free (output);
 }
 
+/* ==========================================================================
+ * Literal-buffer commands (pd:hL*)
+ *
+ * Work against the lazy HBC literal cache. Registration in r2 is additive:
+ * flags (lit.arr.* / lit.obj.*) and CC comments are created at the literal
+ * paddr; code→data xrefs are added for every call site. None of this is
+ * attempted while plain disasm runs — the cache only populates on demand.
+ * ========================================================================== */
+
+static const char *lit_kind_prefix(HBCLiteralKind k) {
+	return k == HBC_LIT_ARRAY? "lit.arr.": "lit.obj.";
+}
+
+/* Register r2 artifacts for one cache entry. Safe to call repeatedly. */
+static void register_r2_artifacts(RCore *core, const HBCLiteralEntry *e) {
+	if (!e->paddr) {
+		return;
+	}
+	ut64 vaddr = HBC_VADDR_BASE + (ut64)e->paddr;
+	char flag_name[64];
+	snprintf (flag_name, sizeof (flag_name), "%s0x%x", lit_kind_prefix (e->kind), e->paddr);
+	r_flag_set (core->flags, flag_name, vaddr, 1);
+	if (e->formatted && *e->formatted) {
+		r_meta_set_string (core->anal, R_META_TYPE_COMMENT, vaddr, e->formatted);
+	}
+	for (u32 i = 0; i < e->xref_count; i++) {
+		ut64 from = HBC_VADDR_BASE + (ut64)e->xref_addrs[i];
+		r_anal_xrefs_set (core->anal, from, vaddr, R_ANAL_REF_TYPE_DATA);
+	}
+}
+
+static void register_all_artifacts(RCore *core, HBC *hbc) {
+	const HBCLiteralEntry *arr = NULL;
+	u32 n = 0;
+	if (hbc_literals_list (hbc, &arr, &n).code != RESULT_SUCCESS) {
+		return;
+	}
+	for (u32 i = 0; i < n; i++) {
+		register_r2_artifacts (core, &arr[i]);
+	}
+}
+
+static const char *kind_label(HBCLiteralKind k) {
+	return k == HBC_LIT_ARRAY? "array": "object";
+}
+
+/* Make sure the current binary is loaded into the core plugin's HbcContext. */
+static Result ensure_hbc_loaded(HbcContext *ctx, RCore *core, HBC **out) {
+	Result res = hbc_load_current_binary (ctx, core);
+	if (res.code == RESULT_SUCCESS) {
+		*out = ctx->hbc;
+	}
+	return res;
+}
+
+static void cmd_lit_list(HbcContext *ctx, RCore *core, bool as_json) {
+	HBC *hbc = NULL;
+	Result r = ensure_hbc_loaded (ctx, core, &hbc);
+	if (r.code != RESULT_SUCCESS) {
+		R_LOG_ERROR ("%s", safe_errmsg (r.error_message));
+		return;
+	}
+	const HBCLiteralEntry *arr = NULL;
+	u32 n = 0;
+	if (hbc_literals_list (hbc, &arr, &n).code != RESULT_SUCCESS) {
+		R_LOG_ERROR ("list failed");
+		return;
+	}
+	if (n == 0) {
+		r_cons_println (core->cons, "(cache empty — run pd:hLs to scan)");
+		return;
+	}
+	if (as_json) {
+		PJ *pj = r_core_pj_new (core);
+		pj_a (pj);
+		for (u32 i = 0; i < n; i++) {
+			const HBCLiteralEntry *e = &arr[i];
+			pj_o (pj);
+			pj_ks (pj, "kind", kind_label (e->kind));
+			pj_kn (pj, "num_items", e->num_items);
+			pj_kn (pj, "primary_id", e->primary_id);
+			pj_kn (pj, "secondary_id", e->secondary_id);
+			pj_kn (pj, "paddr", e->paddr);
+			pj_kn (pj, "vaddr", HBC_VADDR_BASE + e->paddr);
+			pj_ks (pj, "formatted", e->formatted? e->formatted: "");
+			pj_ka (pj, "xrefs");
+			for (u32 j = 0; j < e->xref_count; j++) {
+				pj_N (pj, HBC_VADDR_BASE + (ut64)e->xref_addrs[j]);
+			}
+			pj_end (pj);
+			pj_end (pj);
+		}
+		pj_end (pj);
+		char *s = pj_drain (pj);
+		r_cons_println (core->cons, s);
+		free (s);
+		return;
+	}
+	r_cons_printf (core->cons, "literals: %u\n", n);
+	for (u32 i = 0; i < n; i++) {
+		const HBCLiteralEntry *e = &arr[i];
+		r_cons_printf (core->cons,
+			"%-6s n=%-4u id=(%u,%u) paddr=0x%08x xrefs=%u  %s\n",
+			kind_label (e->kind), e->num_items, e->primary_id,
+			e->secondary_id, e->paddr, e->xref_count,
+			e->formatted? e->formatted: "");
+	}
+}
+
+static void cmd_lit_scan_code(HbcContext *ctx, RCore *core) {
+	HBC *hbc = NULL;
+	Result r = ensure_hbc_loaded (ctx, core, &hbc);
+	if (r.code != RESULT_SUCCESS) {
+		R_LOG_ERROR ("%s", safe_errmsg (r.error_message));
+		return;
+	}
+	u32 n = 0;
+	r = hbc_literals_scan_code (hbc, &n);
+	if (r.code != RESULT_SUCCESS) {
+		R_LOG_ERROR ("scan failed: %s", safe_errmsg (r.error_message));
+		return;
+	}
+	register_all_artifacts (core, hbc);
+	r_cons_printf (core->cons, "scanned code, %u distinct literals (flags + xrefs registered)\n", n);
+}
+
+static void cmd_lit_scan_pool(HbcContext *ctx, RCore *core, HBCLiteralKind kind) {
+	HBC *hbc = NULL;
+	Result r = ensure_hbc_loaded (ctx, core, &hbc);
+	if (r.code != RESULT_SUCCESS) {
+		R_LOG_ERROR ("%s", safe_errmsg (r.error_message));
+		return;
+	}
+	HBCPoolGroup *groups = NULL;
+	u32 n = 0;
+	r = hbc_literals_scan_pool (hbc, kind, &groups, &n);
+	if (r.code != RESULT_SUCCESS) {
+		R_LOG_ERROR ("pool scan failed: %s", safe_errmsg (r.error_message));
+		return;
+	}
+	r_cons_printf (core->cons, "%s pool: %u groups\n", kind_label (kind), n);
+	for (u32 i = 0; i < n; i++) {
+		r_cons_printf (core->cons,
+			"  paddr=0x%08x pool_off=0x%08x n=%-4u tag=%u\n",
+			groups[i].paddr, groups[i].pool_offset, groups[i].num_items,
+			groups[i].tag);
+	}
+	free (groups);
+}
+
+static void cmd_lit_reset(HbcContext *ctx, RCore *core) {
+	HBC *hbc = NULL;
+	Result r = ensure_hbc_loaded (ctx, core, &hbc);
+	if (r.code != RESULT_SUCCESS) {
+		R_LOG_ERROR ("%s", safe_errmsg (r.error_message));
+		return;
+	}
+	hbc_literals_reset (hbc);
+	r_cons_println (core->cons, "literal cache reset (r2 flags/comments/xrefs kept — use 'f- lit.*' to drop)");
+}
+
+static void cmd_lit_get(HbcContext *ctx, RCore *core, const char *args) {
+	HBC *hbc = NULL;
+	Result r = ensure_hbc_loaded (ctx, core, &hbc);
+	if (r.code != RESULT_SUCCESS) {
+		R_LOG_ERROR ("%s", safe_errmsg (r.error_message));
+		return;
+	}
+	/* Syntax: pd:hLg a <num> <array_id>        (array)
+	 *         pd:hLg o <num> <primary> [<sec>] (object)  */
+	char kindc = 0;
+	u64 num = 0, primary = 0, secondary = 0;
+	if (!args || sscanf (args, " %c %" SCNu64 " %" SCNu64 " %" SCNu64,
+			&kindc, &num, &primary, &secondary) < 3) {
+		R_LOG_ERROR ("usage: pd:hLg {a|o} <num_items> <primary_id> [<secondary_id>]");
+		return;
+	}
+	HBCLiteralKind kind = (kindc == 'a')? HBC_LIT_ARRAY: HBC_LIT_OBJECT;
+	char *text = NULL;
+	r = hbc_literals_format_raw (hbc, kind, (u32)num, (u32)primary,
+		(u32)secondary, &text);
+	if (r.code != RESULT_SUCCESS || !text) {
+		R_LOG_ERROR ("format failed: %s", safe_errmsg (r.error_message));
+		free (text);
+		return;
+	}
+	r_cons_println (core->cons, text);
+	free (text);
+}
+
+static void cmd_lit_toggle_inline(RCore *core) {
+	bool cur = hbc_get_inline_literals ();
+	bool next = !cur;
+	hbc_set_inline_literals (next);
+	r_config_set_b (core->config, "hbc.inline_buffer_literals", next);
+	r_cons_printf (core->cons, "inline buffer literals: %s\n", next? "on": "off");
+}
+
+static const char LIT_HELP[] =
+	"Usage: pd:hL[subcmd]\n"
+	" pd:hL            List cached literals\n"
+	" pd:hLj           List as JSON\n"
+	" pd:hLs           Scan all code; register literals as flags/xrefs/comments\n"
+	" pd:hLp[ao]       Scan SLP pool (default: arrays; a=arrays, o=objects)\n"
+	" pd:hLr           Reset literal cache (does not remove r2 flags/comments)\n"
+	" pd:hLg <k> <n> <primary> [<sec>]\n"
+	"                  Format a literal from raw params (k=a|o)\n"
+	" pd:hLi           Toggle inline literal comments in disasm\n"
+	" pd:hL?           This help\n";
+
+static void cmd_literals(HbcContext *ctx, RCore *core, const char *arg) {
+	while (*arg && isspace ((unsigned char)*arg)) {
+		arg++;
+	}
+	switch (*arg) {
+	case 0:
+		cmd_lit_list (ctx, core, false);
+		break;
+	case 'j':
+		cmd_lit_list (ctx, core, true);
+		break;
+	case 's':
+		cmd_lit_scan_code (ctx, core);
+		break;
+	case 'p': {
+		char k = arg[1];
+		HBCLiteralKind kind = (k == 'o')? HBC_LIT_OBJECT: HBC_LIT_ARRAY;
+		cmd_lit_scan_pool (ctx, core, kind);
+		break;
+	}
+	case 'r':
+		cmd_lit_reset (ctx, core);
+		break;
+	case 'g':
+		cmd_lit_get (ctx, core, arg + 1);
+		break;
+	case 'i':
+		cmd_lit_toggle_inline (core);
+		break;
+	case '?':
+		r_cons_print (core->cons, LIT_HELP);
+		break;
+	default:
+		R_LOG_ERROR ("Unknown subcommand. Use pd:hL? for help");
+		break;
+	}
+}
+
 /* Show help */
 static void cmd_help(RCore *core) {
 	r_cons_print (core->cons,
@@ -425,6 +853,8 @@ static void cmd_help(RCore *core) {
 		"  pd:hj [id]     - JSON output for function\n"
 		"  pd:ho [id]     - Decompile with offsets (addresses) per statement\n"
 		"  pd:hoa         - Decompile all with offsets\n"
+		"  pd:hs [id]     - List source-line information\n"
+		"  pd:hL[?]       - SLP literal cache: list/scan/reset (see pd:hL?)\n"
 		"  pd:h?          - Show this help\n"
 		"\nNote: r2 comments (CC command) are automatically inlined in decompiler output.\n");
 }
@@ -491,6 +921,12 @@ static bool cmd_handler(RCorePluginSession *s, const char *input) {
 		}
 		break;
 	}
+	case 'L': /* pd:hL* — literal cache / buffer-literal commands */
+		cmd_literals (ctx, core, arg + 1);
+		break;
+	case 's': /* pd:hs [id] */
+		cmd_source_lines (ctx, core, arg + 1);
+		break;
 	case '?': /* pd:h? */
 		cmd_help (core);
 		break;
@@ -502,6 +938,13 @@ static bool cmd_handler(RCorePluginSession *s, const char *input) {
 	return true;
 }
 
+static bool cb_inline_buffer_literals(void *user, void *data) {
+	(void)user;
+	RConfigNode *node = (RConfigNode *)data;
+	hbc_set_inline_literals (node && node->i_value);
+	return true;
+}
+
 static bool plugin_init(RCorePluginSession *s) {
 	RCore *core = s->core;
 	RConfig *cfg = core->config;
@@ -582,6 +1025,25 @@ static bool plugin_init(RCorePluginSession *s) {
 		r_config_node_desc (node, "Max bytecode size (bytes) for closure inlining (0=no limit, -1=never inline)");
 	}
 
+	node = r_config_set_i (cfg, "hbc.max_ast", 5000);
+	if (node) {
+		r_config_node_desc (node, "Abort AST build once N statements per function (0=unlimited)");
+	}
+
+	node = r_config_set_i (cfg, "hbc.max_bytes", 262144);
+	if (node) {
+		r_config_node_desc (node, "Stop decompilation once total output reaches N bytes (0=unlimited)");
+	}
+
+	/* Default off: inlining the SLP literal as a disasm-line comment makes
+	 * arch.decode O(n) in literal size. Use pd:hL* to inspect literals
+	 * without impacting disasm speed. */
+	hbc_set_inline_literals (false);
+	node = r_config_set_b_cb (cfg, "hbc.inline_buffer_literals", false, cb_inline_buffer_literals);
+	if (node) {
+		r_config_node_desc (node, "Inline SLP buffer literals as comments in disasm (slow; default off)");
+	}
+
 	r_config_lock (cfg, true);
 	return true;
 }

test/db/extras/basic

@@ -1,4 +1,4 @@
-NAME=r2hermes test
+NAME=r2hermes test (va)
 FILE=bins/hbc/bespoke_eval.hbc
 ARGS=-e asm.sub.varmin=16
 CMDS=<<EOF
@@ -9,7 +9,23 @@ EXPECT=<<EOF
 hbc
             ;-- entry0:
             ;-- global_0x000000b0:
-            0x000000b0      73000100       load_const_string r0, str.print_123__ ; 0xa2 ; "print(123);"
+            0x100000b0      73000100       load_const_string r0, 0xa2  ; yes.js:1:1 ; str.print_123__
+                                                                       ; 0x100000a2 ; "print(123);"
+            0x100000b4      5e000000       direct_eval r0, r0, 0       ; yes.js:1:5
+EOF
+RUN
+
+NAME=r2hermes test (no va)
+FILE=bins/hbc/bespoke_eval.hbc
+ARGS=-e asm.sub.varmin=16 -e io.va=false
+CMDS=<<EOF
+-a
+pd 2 @ 0xb0
+EOF
+EXPECT=<<EOF
+hbc
+            0x000000b0      73000100       load_const_string r0, 0xa2  ; str.print_123__
+                                                                       ; 0x100000a2
             0x000000b4      5e000000       direct_eval r0, r0, 0
 EOF
 RUN

test/db/extras/hbc-decompiler

@@ -45,6 +45,7 @@ pd:hi~Functions
 EOF
 EXPECT=<<EOF
   Functions: 1
+  Functions With Debug Info: 1
 EOF
 RUN
 
@@ -55,6 +56,7 @@ pd:hi~Strings
 EOF
 EXPECT=<<EOF
   Strings: 2
+  Overflow Strings: 0
 EOF
 RUN
 
@@ -65,6 +67,7 @@ pd:hi~Functions
 EOF
 EXPECT=<<EOF
   Functions: 5388
+  Functions With Debug Info: 0
 EOF
 RUN
 
@@ -75,6 +78,7 @@ pd:hi~Strings
 EOF
 EXPECT=<<EOF
   Strings: 6170
+  Overflow Strings: 7
 EOF
 RUN
 
@@ -178,13 +182,25 @@ EXPECT=<<EOF
 EOF
 RUN
 
-NAME=pd:hj json output
+NAME=pd:hj json code field
 FILE=bins/hbc/bespoke_eval.hbc
 CMDS=<<EOF
-pd:hj 0~function_id
+pd:hj 0~{code}
 EOF
 EXPECT=<<EOF
-{"function_id":0,"decompilation":"function global() {\n  r0 = \"print(123);\";\n  r0 = eval (r0);\n  return r0;\n}\n\n\n"}
+function global() {\n  r0 = \"print(123);\";\n  r0 = eval (r0);\n  return r0;\n}\n\n\n
+EOF
+RUN
+
+NAME=pd:hj json annotation offsets
+FILE=bins/hbc/bespoke_eval.hbc
+CMDS=<<EOF
+pd:hj 0~{annotations[0].type}
+pd:hj 0~{annotations[0].offset}
+EOF
+EXPECT=<<EOF
+offset
+176
 EOF
 RUN
 

test/db/extras/hbc-disasm

@@ -1,4 +1,4 @@
-NAME=hbc disasm basic instructions
+NAME=hbc disasm basic instructions (va)
 FILE=bins/hbc/bespoke_eval.hbc
 ARGS=-e asm.sub.varmin=16
 CMDS=<<EOF
@@ -7,7 +7,22 @@ EOF
 EXPECT=<<EOF
             ;-- entry0:
             ;-- global_0x000000b0:
-            0x000000b0      73000100       load_const_string r0, str.print_123__ ; 0xa2 ; "print(123);"
+            0x100000b0      73000100       load_const_string r0, 0xa2  ; yes.js:1:1 ; str.print_123__
+                                                                       ; 0x100000a2 ; "print(123);"
+            0x100000b4      5e000000       direct_eval r0, r0, 0       ; yes.js:1:5
+            0x100000b8      5c00           ret r0
+EOF
+RUN
+
+NAME=hbc disasm basic instructions (no va)
+FILE=bins/hbc/bespoke_eval.hbc
+ARGS=-e asm.sub.varmin=16 -e io.va=false
+CMDS=<<EOF
+pd 3 @ 0xb0
+EOF
+EXPECT=<<EOF
+            0x000000b0      73000100       load_const_string r0, 0xa2  ; str.print_123__
+                                                                       ; 0x100000a2
             0x000000b4      5e000000       direct_eval r0, r0, 0
             0x000000b8      5c00           ret r0
 EOF
@@ -47,32 +62,66 @@ cpu      96
 EOF
 RUN
 
-NAME=hbc symbols list
+NAME=hbc symbols list (va)
 FILE=bins/hbc/bespoke_eval.hbc
 ARGS=-e asm.sub.varmin=16
 CMDS=<<EOF
 is~FUNC
 EOF
 EXPECT=<<EOF
+0   0x000000b0 0x100000b0 NONE FUNC 10       global_0x000000b0
+EOF
+RUN
+
+NAME=hbc symbols list (no va)
+FILE=bins/hbc/bespoke_eval.hbc
+ARGS=-e asm.sub.varmin=16 -e io.va=false
+CMDS=<<EOF
+is~FUNC
+EOF
+EXPECT=<<EOF
 0   0x000000b0 0x000000b0 NONE FUNC 10       global_0x000000b0
 EOF
 RUN
 
-NAME=hbc entry point
+NAME=hbc entry point (va)
 FILE=bins/hbc/bespoke_eval.hbc
 ARGS=-e asm.sub.varmin=16
 CMDS=<<EOF
 ie~program
 EOF
 EXPECT=<<EOF
+0x000000b0 0x100000b0 ---------- ---------- program
+EOF
+RUN
+
+NAME=hbc entry point (no va)
+FILE=bins/hbc/bespoke_eval.hbc
+ARGS=-e asm.sub.varmin=16 -e io.va=false
+CMDS=<<EOF
+ie~program
+EOF
+EXPECT=<<EOF
 0x000000b0 0x000000b0 ---------- ---------- program
 EOF
 RUN
 
-NAME=hbc disasm ret instruction
+NAME=hbc disasm ret instruction (va)
 FILE=bins/hbc/bespoke_eval.hbc
 ARGS=-e asm.sub.varmin=16
 CMDS=<<EOF
+s 0x100000b8
+pd 1
+EOF
+EXPECT=<<EOF
+            0x100000b8      5c00           ret r0
+EOF
+RUN
+
+NAME=hbc disasm ret instruction (no va)
+FILE=bins/hbc/bespoke_eval.hbc
+ARGS=-e asm.sub.varmin=16 -e io.va=false
+CMDS=<<EOF
 s 0xb8
 pd 1
 EOF
@@ -81,24 +130,50 @@ EXPECT=<<EOF
 EOF
 RUN
 
-NAME=hbc disasm load_const_string
+NAME=hbc disasm load_const_string (va)
 ARGS=-e asm.sub.varmin=16
 FILE=bins/hbc/bespoke_eval.hbc
 CMDS=<<EOF
+s 0x100000b0
+pd 1
+EOF
+EXPECT=<<EOF
+            ;-- entry0:
+            ;-- global_0x000000b0:
+            0x100000b0      73000100       load_const_string r0, 0xa2  ; yes.js:1:1 ; str.print_123__
+                                                                       ; 0x100000a2 ; "print(123);"
+EOF
+RUN
+
+NAME=hbc disasm load_const_string (no va)
+ARGS=-e asm.sub.varmin=16 -e io.va=false
+FILE=bins/hbc/bespoke_eval.hbc
+CMDS=<<EOF
 s 0xb0
 pd 1
 EOF
 EXPECT=<<EOF
-            ;-- entry0:
-            ;-- global_0x000000b0:
-            0x000000b0      73000100       load_const_string r0, str.print_123__ ; 0xa2 ; "print(123);"
+            0x000000b0      73000100       load_const_string r0, 0xa2  ; str.print_123__
+                                                                       ; 0x100000a2
 EOF
 RUN
 
-NAME=hbc disasm direct_eval
+NAME=hbc disasm direct_eval (va)
 FILE=bins/hbc/bespoke_eval.hbc
 ARGS=-e asm.sub.varmin=16
 CMDS=<<EOF
+s 0x100000b4
+pd 1
+EOF
+EXPECT=<<EOF
+            0x100000b4      5e000000       direct_eval r0, r0, 0       ; yes.js:1:5
+EOF
+RUN
+
+NAME=hbc disasm direct_eval (no va)
+FILE=bins/hbc/bespoke_eval.hbc
+ARGS=-e asm.sub.varmin=16 -e io.va=false
+CMDS=<<EOF
 s 0xb4
 pd 1
 EOF
@@ -107,18 +182,29 @@ EXPECT=<<EOF
 EOF
 RUN
 
-NAME=hbc sections
+NAME=hbc sections (va)
 FILE=bins/hbc/bespoke_eval.hbc
 ARGS=-e asm.sub.varmin=16
 CMDS=<<EOF
 iS~hermes
 EOF
 EXPECT=<<EOF
-0   0x00000000  0x124 0x00000000  0x124 -r-- 0x0   ---- hermes_bytecode
+0   0x00000000  0x124 0x10000000  0x124 -r-x 0x0   ---- hermes_bytecode
 EOF
 RUN
 
-NAME=hbc seek and disasm
+NAME=hbc sections (no va)
+FILE=bins/hbc/bespoke_eval.hbc
+ARGS=-e asm.sub.varmin=16 -e io.va=false
+CMDS=<<EOF
+iS~hermes
+EOF
+EXPECT=<<EOF
+0   0x00000000  0x124 0x00000000  0x124 -r-x 0x0   ---- hermes_bytecode
+EOF
+RUN
+
+NAME=hbc seek and disasm (va)
 FILE=bins/hbc/bespoke_eval.hbc
 ARGS=-e asm.sub.varmin=16
 CMDS=<<EOF
@@ -128,16 +214,32 @@ EOF
 EXPECT=<<EOF
             ;-- entry0:
             ;-- global_0x000000b0:
-            0x000000b0      73000100       load_const_string r0, str.print_123__ ; 0xa2 ; "print(123);"
-            0x000000b4      5e000000       direct_eval r0, r0, 0
+            0x100000b0      73000100       load_const_string r0, 0xa2  ; yes.js:1:1 ; str.print_123__
+                                                                       ; 0x100000a2 ; "print(123);"
+            0x100000b4      5e000000       direct_eval r0, r0, 0       ; yes.js:1:5
 EOF
 RUN
 
-NAME=hbc print hex bytes
+NAME=hbc print hex bytes (va)
 FILE=bins/hbc/bespoke_eval.hbc
 ARGS=-e asm.sub.varmin=16
 CMDS=<<EOF
+s 0x100000b0
+-e hex.header=true
+px 10
+EOF
+EXPECT=<<EOF
+- offset -  B0B1 B2B3 B4B5 B6B7 B8B9 BABB BCBD BEBF  0123456789ABCDEF
+0x100000b0  7300 0100 5e00 0000 5c00                 s...^...\.
+EOF
+RUN
+
+NAME=hbc print hex bytes (no va)
+FILE=bins/hbc/bespoke_eval.hbc
+ARGS=-e asm.sub.varmin=16 -e io.va=false
+CMDS=<<EOF
 s 0xb0
+-e hex.header=true
 px 10
 EOF
 EXPECT=<<EOF
@@ -146,10 +248,23 @@ EXPECT=<<EOF
 EOF
 RUN
 
-NAME=hbc unreachable padding
+NAME=hbc unreachable padding (va)
 FILE=bins/hbc/bespoke_eval.hbc
 ARGS=-e asm.sub.varmin=16
 CMDS=<<EOF
+s 0x100000ba
+pd 2
+EOF
+EXPECT=<<EOF
+            0x100000ba      00             unreachable
+            0x100000bb      00             unreachable
+EOF
+RUN
+
+NAME=hbc unreachable padding (no va)
+FILE=bins/hbc/bespoke_eval.hbc
+ARGS=-e asm.sub.varmin=16 -e io.va=false
+CMDS=<<EOF
 s 0xba
 pd 2
 EOF
@@ -169,21 +284,82 @@ EXPECT=<<EOF
 EOF
 RUN
 
-NAME=hbc get_global_object instruction
+NAME=hbc imports js require
 FILE=bins/hbc/index.android.bundle
+ARGS=-e asm.sub.varmin=16
+CMDS=<<EOF
+ii~js.__r
+ii~?
+EOF
+EXPECT=<<EOF
+3048 0x00000000 NONE FUNC react-native js.__r
+58
+EOF
+RUN
+
+NAME=hbc imports native module
+FILE=bins/hbc/index.android.bundle
+ARGS=-e asm.sub.varmin=16
+CMDS=<<EOF
+ii~native.NativeModules
+EOF
+EXPECT=<<EOF
+3894 0x00000000 NONE FUNC react-native native.NativeModules
+EOF
+RUN
+
+NAME=hbc exports module symbol
+FILE=bins/hbc/index.android.bundle
+ARGS=-e asm.sub.varmin=16
+CMDS=<<EOF
+iE~hbc.export.module.RN$AppRegistry
+iE~?
+EOF
+EXPECT=<<EOF
+3630 0x00071961 0x10071961 GLOBAL OBJ  1        hbc.export.module.RN$AppRegistry
+451
+EOF
+RUN
+
+NAME=hbc get_global_object instruction (va)
+FILE=bins/hbc/index.android.bundle
+CMDS=<<EOF
+s 0x1003e510
+pd 1
+EOF
+EXPECT=<<EOF
+            ;-- clear_0x0003e510:
+            0x1003e510      3000           get_global_object r0
+EOF
+RUN
+
+NAME=hbc get_global_object instruction (no va)
+FILE=bins/hbc/index.android.bundle
+ARGS=-e io.va=false
 CMDS=<<EOF
 s 0x3e510
 pd 1
 EOF
 EXPECT=<<EOF
-            ;-- clear_0x0003e510:
             0x0003e510      3000           get_global_object r0
 EOF
 RUN
 
-NAME=hbc try_get_by_id instruction
+NAME=hbc try_get_by_id instruction (va)
 FILE=bins/hbc/index.android.bundle
 CMDS=<<EOF
+s 0x1003e512
+ao~mnemonic
+EOF
+EXPECT=<<EOF
+mnemonic: try_get_by_id
+EOF
+RUN
+
+NAME=hbc try_get_by_id instruction (no va)
+FILE=bins/hbc/index.android.bundle
+ARGS=-e io.va=false
+CMDS=<<EOF
 s 0x3e512
 ao~mnemonic
 EOF
@@ -192,9 +368,21 @@ mnemonic: try_get_by_id
 EOF
 RUN
 
-NAME=hbc get_by_id_short instruction
+NAME=hbc get_by_id_short instruction (va)
 FILE=bins/hbc/index.android.bundle
 CMDS=<<EOF
+s 0x1003e518
+ao~mnemonic
+EOF
+EXPECT=<<EOF
+mnemonic: get_by_id_short
+EOF
+RUN
+
+NAME=hbc get_by_id_short instruction (no va)
+FILE=bins/hbc/index.android.bundle
+ARGS=-e io.va=false
+CMDS=<<EOF
 s 0x3e518
 ao~mnemonic
 EOF
@@ -203,9 +391,21 @@ mnemonic: get_by_id_short
 EOF
 RUN
 
-NAME=hbc create_this instruction
+NAME=hbc create_this instruction (va)
 FILE=bins/hbc/index.android.bundle
 CMDS=<<EOF
+s 0x1003e51d
+pd 1
+EOF
+EXPECT=<<EOF
+            0x1003e51d      6a010100       create_this r1, r1, r0
+EOF
+RUN
+
+NAME=hbc create_this instruction (no va)
+FILE=bins/hbc/index.android.bundle
+ARGS=-e io.va=false
+CMDS=<<EOF
 s 0x3e51d
 pd 1
 EOF
@@ -214,9 +414,21 @@ EXPECT=<<EOF
 EOF
 RUN
 
-NAME=hbc mov instruction
+NAME=hbc mov instruction (va)
 FILE=bins/hbc/index.android.bundle
 CMDS=<<EOF
+s 0x1003e521
+pd 1
+EOF
+EXPECT=<<EOF
+            0x1003e521      080201         mov r2, r1
+EOF
+RUN
+
+NAME=hbc mov instruction (no va)
+FILE=bins/hbc/index.android.bundle
+ARGS=-e io.va=false
+CMDS=<<EOF
 s 0x3e521
 pd 1
 EOF
@@ -225,9 +437,21 @@ EXPECT=<<EOF
 EOF
 RUN
 
-NAME=hbc construct instruction
+NAME=hbc construct instruction (va)
 FILE=bins/hbc/index.android.bundle
 CMDS=<<EOF
+s 0x1003e524
+pd 1
+EOF
+EXPECT=<<EOF
+            0x1003e524      50000001       construct r0, r0, 1
+EOF
+RUN
+
+NAME=hbc construct instruction (no va)
+FILE=bins/hbc/index.android.bundle
+ARGS=-e io.va=false
+CMDS=<<EOF
 s 0x3e524
 pd 1
 EOF
@@ -236,9 +460,21 @@ EXPECT=<<EOF
 EOF
 RUN
 
-NAME=hbc select_object instruction
+NAME=hbc select_object instruction (va)
 FILE=bins/hbc/index.android.bundle
 CMDS=<<EOF
+s 0x1003e528
+pd 1
+EOF
+EXPECT=<<EOF
+            0x1003e528      6b000100       select_object r0, r1, r0
+EOF
+RUN
+
+NAME=hbc select_object instruction (no va)
+FILE=bins/hbc/index.android.bundle
+ARGS=-e io.va=false
+CMDS=<<EOF
 s 0x3e528
 pd 1
 EOF
@@ -247,9 +483,21 @@ EXPECT=<<EOF
 EOF
 RUN
 
-NAME=hbc get_environment instruction
+NAME=hbc get_environment instruction (va)
 FILE=bins/hbc/index.android.bundle
 CMDS=<<EOF
+s 0x1003e52c
+pd 1
+EOF
+EXPECT=<<EOF
+            0x1003e52c      290100         get_environment r1, 0
+EOF
+RUN
+
+NAME=hbc get_environment instruction (no va)
+FILE=bins/hbc/index.android.bundle
+ARGS=-e io.va=false
+CMDS=<<EOF
 s 0x3e52c
 pd 1
 EOF
@@ -258,9 +506,21 @@ EXPECT=<<EOF
 EOF
 RUN
 
-NAME=hbc store_to_environment instruction
+NAME=hbc store_to_environment instruction (va)
 FILE=bins/hbc/index.android.bundle
 CMDS=<<EOF
+s 0x1003e52f
+pd 1
+EOF
+EXPECT=<<EOF
+            0x1003e52f      2a010100       store_to_environment r1, 1, r0
+EOF
+RUN
+
+NAME=hbc store_to_environment instruction (no va)
+FILE=bins/hbc/index.android.bundle
+ARGS=-e io.va=false
+CMDS=<<EOF
 s 0x3e52f
 pd 1
 EOF
@@ -269,9 +529,25 @@ EXPECT=<<EOF
 EOF
 RUN
 
-NAME=hbc complete function disasm
+NAME=hbc complete function disasm (va)
 FILE=bins/hbc/index.android.bundle
 CMDS=<<EOF
+s 0x1003e510
+pd 3~get_global
+pd 1 @ 0x1003e51d~create_this
+pd 1 @ 0x1003e533~ret
+EOF
+EXPECT=<<EOF
+            0x1003e510      3000           get_global_object r0
+            0x1003e51d      6a010100       create_this r1, r1, r0
+            0x1003e533      5c00           ret r0
+EOF
+RUN
+
+NAME=hbc complete function disasm (no va)
+FILE=bins/hbc/index.android.bundle
+ARGS=-e io.va=false
+CMDS=<<EOF
 s 0x3e510
 pd 3~get_global
 pd 1 @ 0x3e51d~create_this
@@ -580,4 +856,3 @@ EXPECT=<<EOF
 store_to_environment r1, 1, r0
 EOF
 RUN
-

tests/meson.build

@@ -1,13 +1,14 @@
 test_src = files(
-  # Add test files here as needed
+  'test_debug_info.c',
 )
 
 if test_src.length() > 0
-  executable(
+  run_tests = executable(
     'run_tests',
     test_src,
     link_with: libhbc,
     include_directories: include_dirs,
     install: false,
   )
+  test('debug-info', run_tests, args: [meson.project_source_root()])
 endif

tests/test_debug_info.c

@@ -0,0 +1,81 @@
+#include <hbc/hbc.h>
+
+#include <stdio.h>
+#include <string.h>
+
+#define CHECK(x) do { \
+	if (!(x)) { \
+		fprintf (stderr, "check failed: %s:%d: %s\n", __FILE__, __LINE__, #x); \
+		return 1; \
+	} \
+} while (0)
+
+static int test_small_debug_info(const char *root) {
+	char path[512];
+	snprintf (path, sizeof (path), "%s/test/bins/hbc/bespoke_eval.hbc", root);
+
+	HBC *hbc = NULL;
+	CHECK (hbc_open (path, &hbc).code == RESULT_SUCCESS);
+
+	HBCDebugInfo di = { 0 };
+	CHECK (hbc_get_debug_info (hbc, &di).code == RESULT_SUCCESS);
+	CHECK (di.has_debug_info);
+	CHECK (di.filename_count == 1);
+	CHECK (di.filename_storage_size == 6);
+	CHECK (di.file_region_count == 1);
+	CHECK (di.functions_with_debug_info == 1);
+	CHECK (di.source_locations_size == 13);
+	CHECK (di.scope_desc_data_size == 3);
+	CHECK (di.textified_data_size == 1);
+
+	HBCSourceLineArray lines = { 0 };
+	CHECK (hbc_get_source_lines (hbc, &lines).code == RESULT_SUCCESS);
+	CHECK (hbc_has_source_lines (hbc));
+	CHECK (lines.count == 2);
+	CHECK (lines.lines[0].address == 0xb0);
+	CHECK (lines.lines[0].function_address == 0);
+	CHECK (lines.lines[0].line == 1);
+	CHECK (lines.lines[0].column == 1);
+	CHECK (!strcmp (lines.lines[0].filename, "yes.js"));
+	CHECK (lines.lines[1].address == 0xb4);
+	CHECK (lines.lines[1].function_address == 4);
+	CHECK (lines.lines[1].line == 1);
+	CHECK (lines.lines[1].column == 5);
+
+	hbc_free_source_lines (&lines);
+	hbc_close (hbc);
+	return 0;
+}
+
+static int test_empty_debug_info(const char *root) {
+	char path[512];
+	snprintf (path, sizeof (path), "%s/test/bins/hbc/index.android.bundle", root);
+
+	HBC *hbc = NULL;
+	CHECK (hbc_open (path, &hbc).code == RESULT_SUCCESS);
+
+	HBCDebugInfo di = { 0 };
+	CHECK (hbc_get_debug_info (hbc, &di).code == RESULT_SUCCESS);
+	CHECK (di.has_debug_info);
+	CHECK (di.filename_count == 0);
+	CHECK (di.file_region_count == 0);
+	CHECK (di.functions_with_debug_info == 0);
+	CHECK (di.source_locations_size == 0);
+	CHECK (!hbc_has_source_lines (hbc));
+
+	HBCSourceLineArray lines = { 0 };
+	CHECK (hbc_get_source_lines (hbc, &lines).code == RESULT_SUCCESS);
+	CHECK (lines.count == 0);
+
+	hbc_free_source_lines (&lines);
+	hbc_close (hbc);
+	return 0;
+}
+
+int main(int argc, char **argv) {
+	const char *root = argc > 1? argv[1]: ".";
+	if (test_small_debug_info (root) || test_empty_debug_info (root)) {
+		return 1;
+	}
+	return 0;
+}