Improve location info for syntax errors.

JerryScript-DCO-1.0-Signed-off-by: Zoltan Herczeg [email protected]

Author: zherczeg

docs/02.API-REFERENCE.md

@@ -1462,6 +1462,30 @@ typedef struct
 - [jerry_source_info_enabled_fields_t](#jerry_source_info_enabled_fields_t)
 - [jerry_get_source_info](#jerry_get_source_info)
 
+## jerry_syntax_error_location_t
+
+**Summary**
+
+Detailed location info for SyntaxErrors. It contains the
+start and end location of the token which caused the SyntaxError.
+
+**Prototype**
+
+```c
+typedef struct
+{
+  uint32_t line; /**< start line of the invalid token */
+  uint32_t column_start; /**< start column of the invalid token */
+  uint32_t column_end; /**< end column of the invalid token */
+} jerry_syntax_error_location_t;
+```
+
+*New in version [[NEXT_RELEASE]]*.
+
+**See also**
+
+- [jerry_get_syntax_error_location](#jerry_get_syntax_error_location)
+
 ## jerry_arraybuffer_type_t
 
 **Summary**
@@ -12236,8 +12260,8 @@ Returns a newly created source info structure corresponding to the passed script
 The function is lower level than `toString()` operation, but provides more contextual information.
 
 *Notes*:
-- Returned value must be freed with [jerry_free_source_info](#jerry_free_source_info) when it
-is no longer needed.
+- Returned value must be freed with [jerry_free_source_info](#jerry_free_source_info)
+  when it is no longer needed.
 - This API depends on a build option (`JERRY_FUNCTION_TO_STRING`) and can be checked
   in runtime with the `JERRY_FEATURE_FUNCTION_TO_STRING` feature enum value,
   see: [jerry_is_feature_enabled](#jerry_is_feature_enabled).
@@ -12320,6 +12344,69 @@ See [jerry_get_source_info](#jerry_get_source_info)
 - [jerry_get_source_info](#jerry_get_source_info)
 - [jerry_source_info_t](#jerry_source_info_t)
 
+## jerry_get_syntax_error_location
+
+**Summary**
+
+Gets the resource name and location info assigned to a SyntaxError object generated by the parser.
+
+*Notes*:
+- Returned value must be freed with [jerry_release_value](#jerry_release_value)
+  when it is no longer needed.
+- This API depends on a build option (`JERRY_ERROR_MESSAGES`) and can be checked
+  in runtime with the `JERRY_FEATURE_ERROR_MESSAGES` feature enum value,
+  see: [jerry_is_feature_enabled](#jerry_is_feature_enabled).
+
+**Prototype**
+
+```c
+jerry_value_t jerry_get_syntax_error_location (jerry_value_t value,
+                                               jerry_syntax_error_location_t *error_location_p);
+```
+- `value` - SyntaxError object
+- `error_location_p` - output location info
+- return
+  - resource name - if the `value` object has a location info data
+  - error - otherwise
+
+*New in version [[NEXT_RELEASE]]*.
+
+**Example**
+
+[doctest]: # ()
+
+```c
+#include "jerryscript.h"
+
+int
+main (void)
+{
+  jerry_init (JERRY_INIT_EMPTY);
+
+  const jerry_char_t script[] = "aa bb";
+
+  jerry_value_t result_value = jerry_parse (script, sizeof (script) - 1, NULL);
+
+  jerry_syntax_error_location_t error_location;
+  jerry_value_t resource_value = jerry_get_syntax_error_location (result_value, &error_location);
+
+  if (jerry_value_is_string (resource_value))
+  {
+    /* Prints the location of the error. */
+  }
+
+  jerry_release_value (resource_value);
+  jerry_release_value (result_value);
+
+  jerry_cleanup ();
+  return 0;
+}
+```
+
+**See also**
+
+- [jerry_syntax_error_location_t](#jerry_syntax_error_location_t)
+
 
 # Functions for realm objects
 

jerry-core/api/jerryscript.c

@@ -5645,6 +5645,64 @@ jerry_free_source_info (jerry_source_info_t *source_info_p) /**< source info blo
 #endif /* JERRY_FUNCTION_TO_STRING */
 } /* jerry_free_source_info */
 
+/**
+ * Gets the resource name and location info assigned to a SyntaxError object generated by the parser.
+ *
+ * @return resource name, if a location info is available
+ *         error, otherwise
+ */
+jerry_value_t
+jerry_get_syntax_error_location (jerry_value_t value, /**< SyntaxError object */
+                                 jerry_syntax_error_location_t *error_location_p) /**< [out] location info */
+{
+  jerry_assert_api_available ();
+
+#if JERRY_ERROR_MESSAGES
+  if (ecma_is_value_error_reference (value))
+  {
+    value = ecma_get_extended_primitive_from_value (value)->u.value;
+  }
+
+  if (ecma_is_value_object (value))
+  {
+    ecma_object_t *object_p = ecma_get_object_from_value (value);
+
+    ecma_string_t *name_p = ecma_get_internal_string (LIT_INTERNAL_MAGIC_STRING_SYNTAX_ERROR_LOCATION);
+    ecma_property_t *property_p = ecma_find_named_property (object_p, name_p);
+
+    if (property_p != NULL)
+    {
+      ecma_value_t error_property_value = ECMA_PROPERTY_VALUE_PTR (property_p)->value;
+      uint8_t *location_p = ECMA_GET_NON_NULL_POINTER_FROM_POINTER_TAG (uint8_t, error_property_value);
+      ecma_value_t result = *(ecma_value_t *) location_p;
+
+      if (error_location_p != NULL)
+      {
+        size_t size_data = error_property_value & ECMA_SYNTAX_ERROR_ALLOCATION_SIZE_MASK;
+        location_p += ((size_data + 1) << ECMA_SYNTAX_ERROR_ALLOCATION_UNIT_SHIFT);
+
+        error_location_p->line = ecma_extended_info_decode_vlq (&location_p);
+        error_location_p->column_start = ecma_extended_info_decode_vlq (&location_p);
+
+        uint32_t difference = ecma_extended_info_decode_vlq (&location_p);
+
+        error_location_p->column_end = error_location_p->column_start + difference;
+      }
+
+      ecma_ref_ecma_string (ecma_get_string_from_value (result));
+      return result;
+    }
+  }
+
+  return jerry_throw (ecma_raise_type_error (ECMA_ERR_MSG ("Location is not available")));
+#else /* !JERRY_ERROR_MESSAGES */
+  JERRY_UNUSED (value);
+  JERRY_UNUSED (error_location_p);
+
+  return jerry_throw (ecma_raise_type_error (ECMA_ERR_MSG ("Location is not available")));
+#endif /* JERRY_ERROR_MESSAGES */
+} /* jerry_get_syntax_error_location */
+
 /**
  * Replaces the currently active realm with another realm.
  *

jerry-core/ecma/base/ecma-extended-info.c

@@ -19,7 +19,7 @@
 
 #include "byte-code.h"
 
-#if JERRY_ESNEXT || JERRY_FUNCTION_TO_STRING
+#if JERRY_ESNEXT || JERRY_ERROR_MESSAGES || JERRY_FUNCTION_TO_STRING
 
 /** \addtogroup ecma ECMA
  * @{
@@ -105,6 +105,10 @@ ecma_extended_info_get_encoded_length (uint32_t value) /**< encoded value */
   return length;
 } /* ecma_extended_info_get_encoded_length */
 
+#endif /* JERRY_ESNEXT || JERRY_ERROR_MESSAGES || JERRY_FUNCTION_TO_STRING */
+
+#if JERRY_ESNEXT || JERRY_FUNCTION_TO_STRING
+
 /**
  * Get the extended info from a byte code
  *

jerry-core/ecma/base/ecma-extended-info.h

@@ -25,7 +25,7 @@
 
 #include "ecma-globals.h"
 
-#if JERRY_ESNEXT || JERRY_FUNCTION_TO_STRING
+#if JERRY_ESNEXT || JERRY_ERROR_MESSAGES || JERRY_FUNCTION_TO_STRING
 
 /**
  * Vlq encoding: flag which is set for all bytes except the last one.
@@ -46,6 +46,10 @@ uint32_t ecma_extended_info_decode_vlq (uint8_t **buffer_p);
 void ecma_extended_info_encode_vlq (uint8_t **buffer_p, uint32_t value);
 uint32_t ecma_extended_info_get_encoded_length (uint32_t value);
 
+#endif /* JERRY_ESNEXT || JERRY_ERROR_MESSAGES || JERRY_FUNCTION_TO_STRING */
+
+#if JERRY_ESNEXT || JERRY_FUNCTION_TO_STRING
+
 uint8_t *ecma_compiled_code_resolve_extended_info (const ecma_compiled_code_t *bytecode_header_p);
 
 #endif /* JERRY_ESNEXT || JERRY_FUNCTION_TO_STRING */

jerry-core/ecma/base/ecma-gc.c

@@ -1618,6 +1618,18 @@ ecma_gc_free_property (ecma_object_t *object_p, /**< object */
       break;
     }
 #endif /* JERRY_BUILTIN_CONTAINER */
+#if JERRY_ERROR_MESSAGES
+    case LIT_INTERNAL_MAGIC_STRING_SYNTAX_ERROR_LOCATION:
+    {
+      uint8_t *location_p = ECMA_GET_NON_NULL_POINTER_FROM_POINTER_TAG (uint8_t, value);
+
+      ecma_deref_ecma_string (ecma_get_string_from_value (*(ecma_value_t *) location_p));
+
+      size_t size_data = value & ECMA_SYNTAX_ERROR_ALLOCATION_SIZE_MASK;
+      jmem_heap_free_block (location_p, (size_data + 1) << ECMA_SYNTAX_ERROR_ALLOCATION_UNIT_SHIFT);
+      break;
+    }
+#endif /* JERRY_ERROR_MESSAGES */
     default:
     {
       JERRY_ASSERT (name_cp == LIT_INTERNAL_MAGIC_STRING_NATIVE_POINTER

jerry-core/ecma/base/ecma-globals.h

@@ -1920,6 +1920,20 @@ typedef struct
   ecma_stringbuilder_header_t *header_p; /**< pointer to header */
 } ecma_stringbuilder_t;
 
+#if JERRY_ERROR_MESSAGES
+
+/**
+ * Allocation block size shift for SyntaxError line info data.
+ */
+#define ECMA_SYNTAX_ERROR_ALLOCATION_UNIT_SHIFT 3
+
+/**
+ * Mask for extracting allocation size.
+ */
+#define ECMA_SYNTAX_ERROR_ALLOCATION_SIZE_MASK 0x3
+
+#endif /* JERRY_ERROR_MESSAGES */
+
 #ifndef JERRY_BUILTIN_BIGINT
 /**
  * BigInt type.

jerry-core/include/jerryscript-core.h

@@ -361,6 +361,7 @@ jerry_value_t jerry_get_user_value (const jerry_value_t value);
 bool jerry_is_eval_code (const jerry_value_t value);
 jerry_source_info_t *jerry_get_source_info (const jerry_value_t value);
 void jerry_free_source_info (jerry_source_info_t *source_info_p);
+jerry_value_t jerry_get_syntax_error_location (jerry_value_t value, jerry_syntax_error_location_t *error_location_p);
 
 /**
  * Array buffer components.

jerry-core/include/jerryscript-types.h

@@ -817,6 +817,16 @@ typedef struct
   uint32_t source_range_length; /**< source length of the function in the source code */
 } jerry_source_info_t;
 
+/**
+ * Detailed location info for SyntaxErrors.
+ */
+typedef struct
+{
+  uint32_t line; /**< start line of the invalid token */
+  uint32_t column_start; /**< start column of the invalid token */
+  uint32_t column_end; /**< end column of the invalid token */
+} jerry_syntax_error_location_t;
+
 /**
  * Array buffer types.
  */

jerry-core/jmem/jmem.h

@@ -262,12 +262,12 @@ void *JERRY_ATTR_PURE jmem_decompress_pointer (uintptr_t compressed_pointer);
  * Set value of pointer-tag value so that it will correspond
  * to specified non_compressed_pointer along with tag
  */
-#define JMEM_CP_SET_NON_NULL_POINTER_TAG(cp_value, pointer, tag)                   \
-  do                                                                               \
-  {                                                                                \
-    JERRY_ASSERT ((uintptr_t) tag < (uintptr_t) (JMEM_ALIGNMENT));                 \
-    jmem_cpointer_tag_t compressed_ptr = jmem_compress_pointer (pointer);          \
-    (cp_value) = (jmem_cpointer_tag_t) ((compressed_ptr << JMEM_TAG_SHIFT) | tag); \
+#define JMEM_CP_SET_NON_NULL_POINTER_TAG(cp_value, pointer, tag)                     \
+  do                                                                                 \
+  {                                                                                  \
+    JERRY_ASSERT ((uintptr_t) tag < (uintptr_t) (JMEM_ALIGNMENT));                   \
+    jmem_cpointer_tag_t compressed_ptr = jmem_compress_pointer (pointer);            \
+    (cp_value) = (jmem_cpointer_tag_t) ((compressed_ptr << JMEM_TAG_SHIFT) | (tag)); \
   } while (false);
 
 /**

jerry-core/lit/lit-magic-strings.h

@@ -64,6 +64,7 @@ typedef enum
                                                                                     *   data properties */
   LIT_INTERNAL_MAGIC_STRING_NATIVE_POINTER_WITH_REFERENCES, /**< native pointer info associated with an object
                                                              *   which contains references to other values */
+  LIT_INTERNAL_MAGIC_STRING_SYNTAX_ERROR_LOCATION, /**< location info for syntax error */
   LIT_INTERNAL_MAGIC_STRING_ENVIRONMENT_RECORD, /**< dynamic environment record needed by class constructors */
   LIT_INTERNAL_MAGIC_STRING_CLASS_FIELD_COMPUTED, /**< computed class field name list */
   LIT_INTERNAL_MAGIC_STRING_CONTAINER_WEAK_REFS, /**< Weak references to the current container object */