[PRISM] Handle RubyVM.keep_script_lines

iseq.c

@@ -1050,7 +1050,7 @@ pm_iseq_new_with_opt(pm_scope_node_t *node, VALUE name, VALUE path, VALUE realpa
     };
 
     prepare_iseq_build(iseq, name, path, realpath, first_lineno, &code_location, -1,
-                       parent, isolated_depth, type, Qnil, option);
+                       parent, isolated_depth, type, node->script_lines == NULL ? Qnil : *node->script_lines, option);
 
     pm_iseq_compile_node(iseq, node);
     finish_iseq_build(iseq);
@@ -1296,15 +1296,17 @@ pm_iseq_compile_with_option(VALUE src, VALUE file, VALUE realpath, VALUE line, V
         break;
     }
 
+    VALUE script_lines;
     VALUE error;
+
     if (RB_TYPE_P(src, T_FILE)) {
         VALUE filepath = rb_io_path(src);
-        error = pm_load_parse_file(&result, filepath);
+        error = pm_load_parse_file(&result, filepath, ruby_vm_keep_script_lines ? &script_lines : NULL);
         RB_GC_GUARD(filepath);
     }
     else {
         src = StringValue(src);
-        error = pm_parse_string(&result, src, file);
+        error = pm_parse_string(&result, src, file, ruby_vm_keep_script_lines ? &script_lines : NULL);
     }
 
     if (error == Qnil) {
@@ -1718,7 +1720,8 @@ iseqw_s_compile_file_prism(int argc, VALUE *argv, VALUE self)
     result.options.line = 1;
     result.node.coverage_enabled = 1;
 
-    VALUE error = pm_load_parse_file(&result, file);
+    VALUE script_lines;
+    VALUE error = pm_load_parse_file(&result, file, ruby_vm_keep_script_lines ? &script_lines : NULL);
 
     if (error == Qnil) {
         make_compile_option(&option, opt);

load.c

@@ -748,7 +748,7 @@ load_iseq_eval(rb_execution_context_t *ec, VALUE fname)
             result.options.line = 1;
             result.node.coverage_enabled = 1;
 
-            VALUE error = pm_load_parse_file(&result, fname);
+            VALUE error = pm_load_parse_file(&result, fname, NULL);
 
             if (error == Qnil) {
                 iseq = pm_iseq_new_top(&result.node, rb_fstring_lit("<top (required)>"), fname, realpath_internal_cached(realpath_map, fname), NULL);

prism_compile.c

@@ -3107,6 +3107,7 @@ pm_scope_node_init(const pm_node_t *node, pm_scope_node_t *scope, pm_scope_node_
         scope->filepath_encoding = previous->filepath_encoding;
         scope->constants = previous->constants;
         scope->coverage_enabled = previous->coverage_enabled;
+        scope->script_lines = previous->script_lines;
     }
 
     switch (PM_NODE_TYPE(node)) {
@@ -10379,7 +10380,7 @@ pm_parse_process_error(const pm_parse_result_t *result)
  * result object is zeroed out.
  */
 static VALUE
-pm_parse_process(pm_parse_result_t *result, pm_node_t *node)
+pm_parse_process(pm_parse_result_t *result, pm_node_t *node, VALUE *script_lines)
 {
     pm_parser_t *parser = &result->parser;
 
@@ -10397,6 +10398,20 @@ pm_parse_process(pm_parse_result_t *result, pm_node_t *node)
 
     scope_node->coverage_enabled = coverage_enabled;
 
+    // If RubyVM.keep_script_lines is set to true, then we need to create that
+    // array of script lines here.
+    if (script_lines != NULL) {
+        *script_lines = rb_ary_new_capa(parser->newline_list.size);
+
+        for (size_t index = 0; index < parser->newline_list.size; index++) {
+            size_t offset = parser->newline_list.offsets[index];
+            size_t length = index == parser->newline_list.size - 1 ? (parser->end - (parser->start + offset)) : (parser->newline_list.offsets[index + 1] - offset);
+            rb_ary_push(*script_lines, rb_enc_str_new((const char *) parser->start + offset, length, scope_node->encoding));
+        }
+
+        scope_node->script_lines = script_lines;
+    }
+
     // Emit all of the various warnings from the parse.
     const pm_diagnostic_t *warning;
     const char *warning_filepath = (const char *) pm_string_source(&parser->filepath);
@@ -10663,7 +10678,7 @@ pm_load_file(pm_parse_result_t *result, VALUE filepath, bool load_error)
  * is zeroed out.
  */
 VALUE
-pm_parse_file(pm_parse_result_t *result, VALUE filepath)
+pm_parse_file(pm_parse_result_t *result, VALUE filepath, VALUE *script_lines)
 {
     result->node.filepath_encoding = rb_enc_get(filepath);
     pm_options_filepath_set(&result->options, RSTRING_PTR(filepath));
@@ -10672,7 +10687,7 @@ pm_parse_file(pm_parse_result_t *result, VALUE filepath)
     pm_parser_init(&result->parser, pm_string_source(&result->input), pm_string_length(&result->input), &result->options);
     pm_node_t *node = pm_parse(&result->parser);
 
-    VALUE error = pm_parse_process(result, node);
+    VALUE error = pm_parse_process(result, node, script_lines);
 
     // If we're parsing a filepath, then we need to potentially support the
     // SCRIPT_LINES__ constant, which can be a hash that has an array of lines
@@ -10680,10 +10695,10 @@ pm_parse_file(pm_parse_result_t *result, VALUE filepath)
     ID id_script_lines = rb_intern("SCRIPT_LINES__");
 
     if (rb_const_defined_at(rb_cObject, id_script_lines)) {
-        VALUE script_lines = rb_const_get_at(rb_cObject, id_script_lines);
+        VALUE constant_script_lines = rb_const_get_at(rb_cObject, id_script_lines);
 
-        if (RB_TYPE_P(script_lines, T_HASH)) {
-            rb_hash_aset(script_lines, filepath, pm_parse_file_script_lines(&result->node, &result->parser));
+        if (RB_TYPE_P(constant_script_lines, T_HASH)) {
+            rb_hash_aset(constant_script_lines, filepath, pm_parse_file_script_lines(&result->node, &result->parser));
         }
     }
 
@@ -10695,11 +10710,11 @@ pm_parse_file(pm_parse_result_t *result, VALUE filepath)
  * cannot be read or if it cannot be parsed properly.
  */
 VALUE
-pm_load_parse_file(pm_parse_result_t *result, VALUE filepath)
+pm_load_parse_file(pm_parse_result_t *result, VALUE filepath, VALUE *script_lines)
 {
     VALUE error = pm_load_file(result, filepath, false);
     if (NIL_P(error)) {
-        error = pm_parse_file(result, filepath);
+        error = pm_parse_file(result, filepath, script_lines);
     }
 
     return error;
@@ -10712,7 +10727,7 @@ pm_load_parse_file(pm_parse_result_t *result, VALUE filepath)
  * error is returned.
  */
 VALUE
-pm_parse_string(pm_parse_result_t *result, VALUE source, VALUE filepath)
+pm_parse_string(pm_parse_result_t *result, VALUE source, VALUE filepath, VALUE *script_lines)
 {
     rb_encoding *encoding = rb_enc_get(source);
     if (!rb_enc_asciicompat(encoding)) {
@@ -10730,7 +10745,7 @@ pm_parse_string(pm_parse_result_t *result, VALUE source, VALUE filepath)
     pm_parser_init(&result->parser, pm_string_source(&result->input), pm_string_length(&result->input), &result->options);
     pm_node_t *node = pm_parse(&result->parser);
 
-    return pm_parse_process(result, node);
+    return pm_parse_process(result, node, script_lines);
 }
 
 /**
@@ -10780,7 +10795,7 @@ pm_parse_stdin(pm_parse_result_t *result)
     // we went through an IO object to be visible to the user.
     rb_reset_argf_lineno(0);
 
-    return pm_parse_process(result, node);
+    return pm_parse_process(result, node, NULL);
 }
 
 #undef NEW_ISEQ

prism_compile.h

@@ -29,6 +29,16 @@ typedef struct pm_scope_node {
     const pm_parser_t *parser;
     rb_encoding *encoding;
 
+    /**
+     * This is a pointer to the list of script lines for the ISEQs that will be
+     * associated with this scope node. It is only set if
+     * RubyVM.keep_script_lines is true. If it is set, it will be set to a
+     * pointer to an array that is always stack allocated (so no GC marking is
+     * needed by this struct). If it is not set, it will be NULL. It is
+     * inherited by all child scopes.
+     */
+    VALUE *script_lines;
+
     /**
      * This is the encoding of the actual filepath object that will be used when
      * a __FILE__ node is compiled or when the path has to be set on a syntax
@@ -75,9 +85,9 @@ typedef struct {
 } pm_parse_result_t;
 
 VALUE pm_load_file(pm_parse_result_t *result, VALUE filepath, bool load_error);
-VALUE pm_parse_file(pm_parse_result_t *result, VALUE filepath);
-VALUE pm_load_parse_file(pm_parse_result_t *result, VALUE filepath);
-VALUE pm_parse_string(pm_parse_result_t *result, VALUE source, VALUE filepath);
+VALUE pm_parse_file(pm_parse_result_t *result, VALUE filepath, VALUE *script_lines);
+VALUE pm_load_parse_file(pm_parse_result_t *result, VALUE filepath, VALUE *script_lines);
+VALUE pm_parse_string(pm_parse_result_t *result, VALUE source, VALUE filepath, VALUE *script_lines);
 VALUE pm_parse_stdin(pm_parse_result_t *result);
 void pm_parse_result_free(pm_parse_result_t *result);
 

ruby.c

@@ -2188,7 +2188,7 @@ prism_script(ruby_cmdline_options_t *opt, pm_parse_result_t *result)
 
         ruby_opt_init(opt);
         result->node.coverage_enabled = 0;
-        error = pm_parse_string(result, opt->e_script, rb_str_new2("-e"));
+        error = pm_parse_string(result, opt->e_script, rb_str_new2("-e"), NULL);
     }
     else {
         pm_options_command_line_set(options, command_line);
@@ -2201,7 +2201,7 @@ prism_script(ruby_cmdline_options_t *opt, pm_parse_result_t *result)
         // to load, it doesn't require files required by -r.
         if (NIL_P(error)) {
             ruby_opt_init(opt);
-            error = pm_parse_file(result, opt->script_name);
+            error = pm_parse_file(result, opt->script_name, NULL);
         }
 
         // Check if (after requiring all of the files through -r flags) we have

template/prelude.c.tmpl

@@ -158,7 +158,7 @@ static void
 pm_prelude_load(pm_parse_result_t *result, VALUE name, VALUE code, int line)
 {
     pm_options_line_set(&result->options, line);
-    VALUE error = pm_parse_string(result, code, name);
+    VALUE error = pm_parse_string(result, code, name, NULL);
 
     if (!NIL_P(error)) {
         pm_parse_result_free(result);

vm_eval.c

@@ -1719,7 +1719,7 @@ pm_eval_make_iseq(VALUE src, VALUE fname, int line,
     // Add our empty local scope at the very end of the array for our eval
     // scope's locals.
     pm_options_scope_init(&result.options.scopes[scopes_count], 0);
-    VALUE error = pm_parse_string(&result, src, fname);
+    VALUE error = pm_parse_string(&result, src, fname, NULL);
 
     // If the parse failed, clean up and raise.
     if (error != Qnil) {