[ruby/prism] Wire up options through the Java parser

https://github.com/ruby/prism/commit/13fa262669

6 files changed, 23 insertions, 90 deletions

diff --git a/lib/prism/ffi.rb b/lib/prism/ffi.rb
index cc84d00f14..931e277ef5 100644
--- a/lib/prism/ffi.rb
+++ b/lib/prism/ffi.rb
@@ -231,6 +231,7 @@ module Prism
 
         loader.load_header
         loader.load_force_encoding
+        loader.load_start_line
         loader.load_comments
       end
     end
@@ -283,7 +284,7 @@ module Prism
       end
 
       template << "L"
-      values << options.fetch(:line, 0)
+      values << options.fetch(:line, 1)
 
       template << "L"
       if (encoding = options[:encoding])
diff --git a/lib/prism/parse_result.rb b/lib/prism/parse_result.rb
index 92651cf766..c7f6cdcd16 100644
--- a/lib/prism/parse_result.rb
+++ b/lib/prism/parse_result.rb
@@ -9,7 +9,7 @@ module Prism
     attr_reader :source
 
     # The line number where this source starts.
-    attr_reader :start_line
+    attr_accessor :start_line
 
     # The list of newline byte offsets in the source code.
     attr_reader :offsets
diff --git a/prism/options.h b/prism/options.h
index 6faadc2c47..f1b0254fff 100644
--- a/prism/options.h
+++ b/prism/options.h
@@ -160,28 +160,28 @@ PRISM_EXPORTED_FUNCTION void pm_options_free(pm_options_t *options);
  *
  * | # bytes | field                      |
  * | ------- | -------------------------- |
- * | 4       | the length of the filepath |
+ * | `4`     | the length of the filepath |
  * | ...     | the filepath bytes         |
- * | 4       | the line number            |
- * | 4       | the length the encoding    |
+ * | `4`     | the line number            |
+ * | `4`     | the length the encoding    |
  * | ...     | the encoding bytes         |
- * | 1       | frozen string literal      |
- * | 1       | suppress warnings          |
- * | 4       | the number of scopes       |
+ * | `1`     | frozen string literal      |
+ * | `1`     | suppress warnings          |
+ * | `4`     | the number of scopes       |
  * | ...     | the scopes                 |
  *
  * Each scope is layed out as follows:
  * 
  * | # bytes | field                      |
  * | ------- | -------------------------- |
- * | 4       | the number of locals       |
+ * | `4`     | the number of locals       |
  * | ...     | the locals                 |
  *
  * Each local is layed out as follows:
  *
  * | # bytes | field                      |
  * | ------- | -------------------------- |
- * | 4       | the length of the local    |
+ * | `4`     | the length of the local    |
  * | ...     | the local bytes            |
  *
  * Some additional things to note about this layout:
diff --git a/prism/prism.c b/prism/prism.c
index be8dd38ccf..a362961a1f 100644
--- a/prism/prism.c
+++ b/prism/prism.c
@@ -16239,82 +16239,6 @@ parse_program(pm_parser_t *parser) {
     return (pm_node_t *) pm_program_node_create(parser, &locals, statements);
 }
 
-/**
- * Read a 32-bit unsigned integer from a pointer. This function is used to read
- * the metadata that is passed into the parser from the Ruby implementation. It
- * handles aligned and unaligned reads.
- */
-static uint32_t
-pm_metadata_read_u32(const char *ptr) {
-    if (((uintptr_t) ptr) % sizeof(uint32_t) == 0) {
-        return *((uint32_t *) ptr);
-    } else {
-        uint32_t value;
-        memcpy(&value, ptr, sizeof(uint32_t));
-        return value;
-    }
-}
-
-/**
- * Process any additional metadata being passed into a call to the parser via
- * the pm_parse_serialize function. Since the source of these calls will be from
- * Ruby implementation internals we assume it is from a trusted source.
- *
- * Currently, this is only passing in variable scoping surrounding an eval, but
- * eventually it will be extended to hold any additional metadata.  This data
- * is serialized to reduce the calling complexity for a foreign function call
- * vs a foreign runtime making a bindable in-memory version of a C structure.
- *
- * metadata is assumed to be a valid pointer pointing to well-formed data. The
- * format is described below:
- *
- * ```text
- * [
- *   filepath_size: uint32_t,
- *   filepath: char*,
- *   scopes_count: uint32_t,
- *   [
- *     locals_count: uint32_t,
- *     [local_size: uint32_t, local: char*]*
- *   ]*
- * ]
- * ```
- */
-void
-pm_parser_metadata(pm_parser_t *parser, const char *metadata) {
-    uint32_t filepath_size = pm_metadata_read_u32(metadata);
-    metadata += 4;
-
-    if (filepath_size) {
-        pm_string_t filepath_string;
-        pm_string_constant_init(&filepath_string, metadata, filepath_size);
-
-        parser->filepath_string = filepath_string;
-        metadata += filepath_size;
-    }
-
-    uint32_t scopes_count = pm_metadata_read_u32(metadata);
-    metadata += 4;
-
-    for (size_t scope_index = 0; scope_index < scopes_count; scope_index++) {
-        uint32_t locals_count = pm_metadata_read_u32(metadata);
-        metadata += 4;
-
-        pm_parser_scope_push(parser, scope_index == 0);
-
-        for (size_t local_index = 0; local_index < locals_count; local_index++) {
-            uint32_t local_size = pm_metadata_read_u32(metadata);
-            metadata += 4;
-
-            uint8_t *constant = malloc(local_size);
-            memcpy(constant, metadata, local_size);
-
-            pm_parser_local_add_owned(parser, constant, (size_t) local_size);
-            metadata += local_size;
-        }
-    }
-}
-
 /******************************************************************************/
 /* External functions                                                         */
 /******************************************************************************/
@@ -16561,7 +16485,7 @@ PRISM_EXPORTED_FUNCTION void
 pm_serialize(pm_parser_t *parser, pm_node_t *node, pm_buffer_t *buffer) {
     pm_serialize_header(buffer);
     pm_serialize_content(parser, node, buffer);
-    pm_buffer_append_string(buffer, "\0", 1);
+    pm_buffer_append_byte(buffer, '\0');
 }
 
 /**
@@ -16601,6 +16525,7 @@ pm_serialize_parse_comments(pm_buffer_t *buffer, const uint8_t *source, size_t s
     pm_node_t *node = pm_parse(&parser);
     pm_serialize_header(buffer);
     pm_serialize_encoding(&parser.encoding, buffer);
+    pm_buffer_append_varint(buffer, parser.start_line);
     pm_serialize_comment_list(&parser, &parser.comment_list, buffer);
 
     pm_node_destroy(&parser, node);
diff --git a/prism/prism.h b/prism/prism.h
index 2f7208929b..211d2957f2 100644
--- a/prism/prism.h
+++ b/prism/prism.h
@@ -233,7 +233,7 @@ PRISM_EXPORTED_FUNCTION const char * pm_token_type_to_str(pm_token_type_t token_
  * * `pm_buffer_t` - a small buffer object that will hold the serialized AST
  * * `pm_buffer_free` - free the memory associated with the buffer
  * * `pm_serialize` - serialize the AST into a buffer
- * * `pm_parse_serialize` - parse and serialize the AST into a buffer
+ * * `pm_serialize_parse` - parse and serialize the AST into a buffer
  *
  * Putting all of this together would look something like:
  *
@@ -241,7 +241,7 @@ PRISM_EXPORTED_FUNCTION const char * pm_token_type_to_str(pm_token_type_t token_
  * void serialize(const uint8_t *source, size_t length) {
  *     pm_buffer_t buffer = { 0 };
  *
- *     pm_parse_serialize(source, length, &buffer, NULL);
+ *     pm_serialize_parse(&buffer, source, length, NULL);
  *     printf("SERIALIZED!\n");
  *
  *     pm_buffer_free(&buffer);
diff --git a/prism/templates/lib/prism/serialize.rb.erb b/prism/templates/lib/prism/serialize.rb.erb
index a8a6a2dd47..ef8ebbbae8 100644
--- a/prism/templates/lib/prism/serialize.rb.erb
+++ b/prism/templates/lib/prism/serialize.rb.erb
@@ -46,6 +46,7 @@ module Prism
     class Loader # :nodoc:
       attr_reader :encoding, :input, :serialized, :io
       attr_reader :constant_pool_offset, :constant_pool, :source
+      attr_reader :start_line
 
       def initialize(source, serialized)
         @encoding = Encoding::UTF_8
@@ -59,7 +60,7 @@ module Prism
         @constant_pool = nil
 
         @source = source
-        define_load_node_lambdas unless RUBY_ENGINE == 'ruby'
+        define_load_node_lambdas unless RUBY_ENGINE == "ruby"
       end
 
       def load_header
@@ -80,6 +81,10 @@ module Prism
         @input = input.force_encoding(@encoding).freeze
       end
 
+      def load_start_line
+        source.start_line = load_varint
+      end
+
       def load_comments
         load_varint.times.map { Comment.new(Comment::TYPES.fetch(load_varint), load_location) }
       end
@@ -108,6 +113,7 @@ module Prism
       def load_tokens_result
         tokens = load_tokens
         encoding = load_encoding
+        load_start_line
         comments, magic_comments, errors, warnings = load_metadata
 
         if encoding != @encoding
@@ -121,6 +127,7 @@ module Prism
       def load_nodes
         load_header
         load_force_encoding
+        load_start_line
 
         comments, magic_comments, errors, warnings = load_metadata