Herb: Generate Parser Options from config.yml

.gitignore

@@ -91,19 +91,25 @@ pkg/*
 
 # Templated Files
 ext/herb/error_helpers.c
+ext/herb/parser_options_helpers.c
+ext/herb/parser_options_helpers.h
 ext/herb/error_helpers.h
 ext/herb/nodes.c
 ext/herb/nodes.h
 java/error_helpers.c
+java/parser_options_helpers.c
+java/parser_options_helpers.h
 java/error_helpers.h
 java/nodes.c
 java/nodes.h
 java/org/herb/ast/Errors.java
+java/org/herb/ParserOptions.java
 java/org/herb/ast/Nodes.java
 java/org/herb/ast/NodeVisitor.java
 java/org/herb/ast/HelperRegistry.java
 java/org/herb/ast/Visitor.java
 javascript/packages/core/src/action-view-helpers.ts
+javascript/packages/core/src/parser-options.ts
 javascript/packages/core/src/errors.ts
 javascript/packages/core/src/html-entities.json
 javascript/packages/core/src/node-type-guards.ts
@@ -117,11 +123,13 @@ javascript/packages/node/extension/nodes.h
 lib/herb/action_view/helper_registry.rb
 lib/herb/ast/nodes.rb
 lib/herb/errors.rb
+lib/herb/parser_options.rb
 lib/herb/visitor.rb
 rust/src/action_view_helpers.rs
 rust/src/ast/nodes.rs
 rust/src/config.rs
 rust/src/errors.rs
+rust/src/parser_options.rs
 rust/src/nodes.rs
 rust/src/union_types.rs
 rust/src/visitor.rs
@@ -135,17 +143,21 @@ src/analyze/transform.c
 src/ast/ast_nodes.c
 src/ast/ast_pretty_print.c
 src/errors.c
+src/parser_options_defaults.c
 src/include/analyze/action_view/helper_registry.h
 src/include/ast/ast_nodes.h
 src/include/ast/ast_pretty_print.h
 src/include/errors.h
+src/include/parser/parser_options.h
 src/include/lib/hb_foreach.h
 src/parser/match_tags.c
 src/diff/herb_diff_helpers.c
 src/diff/herb_diff_nodes.c
 src/diff/herb_hash_tree.c
 src/visitor.c
 wasm/error_helpers.cpp
+wasm/parser_options_helpers.cpp
+wasm/parser_options_helpers.h
 wasm/error_helpers.h
 wasm/nodes.cpp
 wasm/nodes.h

config.yml

@@ -500,6 +500,98 @@ warnings:
   fields: []
   types: []
 
+parser_options:
+  fields:
+    - name: track_whitespace
+      type: boolean
+      default: false
+      description: "Whether to track whitespace tokens during parsing."
+
+    - name: analyze
+      type: boolean
+      default: true
+      description: "Whether analysis is performed during parsing."
+
+    - name: strict
+      type: boolean
+      default: true
+      description: "Whether strict mode is enabled during parsing."
+
+    - name: action_view_helpers
+      type: boolean
+      default: false
+      description: "Whether ActionView tag helper transformation is enabled during parsing."
+
+    - name: transform_conditionals
+      type: boolean
+      default: false
+      description: "Whether postfix conditional transformation is enabled during parsing."
+
+    - name: render_nodes
+      type: boolean
+      default: false
+      description: "Whether ActionView render call detection is enabled during parsing."
+
+    - name: strict_locals
+      type: boolean
+      default: false
+      description: "Whether strict locals analysis is enabled during parsing."
+
+    - name: prism_program
+      type: boolean
+      default: false
+      description: "Whether the full Prism ProgramNode is serialized on the DocumentNode."
+
+    - name: prism_nodes
+      type: boolean
+      default: false
+      description: "Whether Prism node serialization is enabled during parsing."
+
+    - name: prism_nodes_deep
+      type: boolean
+      default: false
+      description: "Whether deep Prism node serialization is enabled during parsing."
+
+    - name: dot_notation_tags
+      type: boolean
+      default: false
+      description: "Whether dot-notation component tags (e.g. Dialog.Button) are parsed as HTML elements."
+
+    - name: html
+      type: boolean
+      default: true
+      description: "Whether HTML tag parsing is enabled. When false, HTML-like content is treated as literal text."
+
+    - name: timeout
+      type: uint32
+      c_name: timeout_ms
+      default_ms: 1000
+      ruby_unit: seconds
+      description: "Parse timeout."
+
+    - name: max_errors
+      type: uint32
+      default: 25
+      nullable: true
+      null_sentinel: 0
+      description: "Maximum number of errors to report. nil/null means unlimited."
+
+  internal_fields:
+    - name: start_line
+      type: uint32
+      default: 0
+
+    - name: start_column
+      type: uint32
+      default: 0
+
+    - name: error_count
+      type: pointer_uint32
+
+    - name: deadline_ms
+      type: uint64
+      default: 0
+
 nodes:
   # fields:
   #   - name: type

ext/herb/extconf.rb

@@ -101,7 +101,8 @@
   "extension.c",
   "nodes.c",
   "error_helpers.c",
-  "extension_helpers.c"
+  "extension_helpers.c",
+  "parser_options_helpers.c"
 ]
 
 $srcs = core_src_files + herb_src_files + prism_main_files + prism_util_files

ext/herb/extension.c

@@ -7,6 +7,7 @@
 #include "extension.h"
 #include "extension_helpers.h"
 #include "nodes.h"
+#include "parser_options_helpers.h"
 
 VALUE mHerb;
 VALUE cPosition;
@@ -116,79 +117,9 @@ static VALUE Herb_parse(int argc, VALUE* argv, VALUE self) {
   bool print_arena_stats = false;
 
   parser_options_T parser_options = HERB_DEFAULT_PARSER_OPTIONS;
+  herb_extract_parser_options(options, &parser_options);
 
   if (!NIL_P(options)) {
-    VALUE track_whitespace = rb_hash_lookup(options, rb_utf8_str_new_cstr("track_whitespace"));
-    if (NIL_P(track_whitespace)) { track_whitespace = rb_hash_lookup(options, ID2SYM(rb_intern("track_whitespace"))); }
-    if (!NIL_P(track_whitespace) && RTEST(track_whitespace)) { parser_options.track_whitespace = true; }
-
-    VALUE analyze = rb_hash_lookup(options, rb_utf8_str_new_cstr("analyze"));
-    if (NIL_P(analyze)) { analyze = rb_hash_lookup(options, ID2SYM(rb_intern("analyze"))); }
-    if (!NIL_P(analyze) && !RTEST(analyze)) { parser_options.analyze = false; }
-
-    VALUE strict = rb_hash_lookup(options, rb_utf8_str_new_cstr("strict"));
-    if (NIL_P(strict)) { strict = rb_hash_lookup(options, ID2SYM(rb_intern("strict"))); }
-    if (!NIL_P(strict)) { parser_options.strict = RTEST(strict); }
-
-    VALUE action_view_helpers = rb_hash_lookup(options, rb_utf8_str_new_cstr("action_view_helpers"));
-    if (NIL_P(action_view_helpers)) {
-      action_view_helpers = rb_hash_lookup(options, ID2SYM(rb_intern("action_view_helpers")));
-    }
-    if (!NIL_P(action_view_helpers) && RTEST(action_view_helpers)) { parser_options.action_view_helpers = true; }
-
-    VALUE transform_conditionals = rb_hash_lookup(options, rb_utf8_str_new_cstr("transform_conditionals"));
-    if (NIL_P(transform_conditionals)) {
-      transform_conditionals = rb_hash_lookup(options, ID2SYM(rb_intern("transform_conditionals")));
-    }
-    if (!NIL_P(transform_conditionals) && RTEST(transform_conditionals)) {
-      parser_options.transform_conditionals = true;
-    }
-
-    VALUE dot_notation_tags = rb_hash_lookup(options, rb_utf8_str_new_cstr("dot_notation_tags"));
-    if (NIL_P(dot_notation_tags)) {
-      dot_notation_tags = rb_hash_lookup(options, ID2SYM(rb_intern("dot_notation_tags")));
-    }
-    if (!NIL_P(dot_notation_tags) && RTEST(dot_notation_tags)) { parser_options.dot_notation_tags = true; }
-
-    VALUE render_nodes = rb_hash_lookup(options, rb_utf8_str_new_cstr("render_nodes"));
-    if (NIL_P(render_nodes)) { render_nodes = rb_hash_lookup(options, ID2SYM(rb_intern("render_nodes"))); }
-    if (!NIL_P(render_nodes) && RTEST(render_nodes)) { parser_options.render_nodes = true; }
-
-    VALUE strict_locals = rb_hash_lookup(options, rb_utf8_str_new_cstr("strict_locals"));
-    if (NIL_P(strict_locals)) { strict_locals = rb_hash_lookup(options, ID2SYM(rb_intern("strict_locals"))); }
-    if (!NIL_P(strict_locals) && RTEST(strict_locals)) { parser_options.strict_locals = true; }
-
-    VALUE prism_nodes = rb_hash_lookup(options, rb_utf8_str_new_cstr("prism_nodes"));
-    if (NIL_P(prism_nodes)) { prism_nodes = rb_hash_lookup(options, ID2SYM(rb_intern("prism_nodes"))); }
-    if (!NIL_P(prism_nodes) && RTEST(prism_nodes)) { parser_options.prism_nodes = true; }
-
-    VALUE prism_nodes_deep = rb_hash_lookup(options, rb_utf8_str_new_cstr("prism_nodes_deep"));
-    if (NIL_P(prism_nodes_deep)) { prism_nodes_deep = rb_hash_lookup(options, ID2SYM(rb_intern("prism_nodes_deep"))); }
-    if (!NIL_P(prism_nodes_deep) && RTEST(prism_nodes_deep)) { parser_options.prism_nodes_deep = true; }
-
-    VALUE prism_program = rb_hash_lookup(options, rb_utf8_str_new_cstr("prism_program"));
-    if (NIL_P(prism_program)) { prism_program = rb_hash_lookup(options, ID2SYM(rb_intern("prism_program"))); }
-    if (!NIL_P(prism_program) && RTEST(prism_program)) { parser_options.prism_program = true; }
-
-    VALUE html = rb_hash_lookup(options, rb_utf8_str_new_cstr("html"));
-    if (NIL_P(html)) { html = rb_hash_lookup(options, ID2SYM(rb_intern("html"))); }
-    if (!NIL_P(html) && !RTEST(html)) { parser_options.html = false; }
-
-    VALUE timeout = rb_hash_lookup(options, rb_utf8_str_new_cstr("timeout"));
-    if (NIL_P(timeout)) { timeout = rb_hash_lookup(options, ID2SYM(rb_intern("timeout"))); }
-    if (!NIL_P(timeout)) { parser_options.timeout_ms = (uint32_t) (NUM2DBL(timeout) * 1000); }
-
-    VALUE max_errors_sentinel = ID2SYM(rb_intern("__not_set__"));
-    VALUE max_errors = rb_hash_lookup2(options, rb_utf8_str_new_cstr("max_errors"), max_errors_sentinel);
-
-    if (max_errors == max_errors_sentinel) {
-      max_errors = rb_hash_lookup2(options, ID2SYM(rb_intern("max_errors")), max_errors_sentinel);
-    }
-
-    if (max_errors != max_errors_sentinel) {
-      parser_options.max_errors = NIL_P(max_errors) ? 0 : (uint32_t) NUM2UINT(max_errors);
-    }
-
     VALUE arena_stats = rb_hash_lookup(options, rb_utf8_str_new_cstr("arena_stats"));
     if (NIL_P(arena_stats)) { arena_stats = rb_hash_lookup(options, ID2SYM(rb_intern("arena_stats"))); }
     if (!NIL_P(arena_stats) && RTEST(arena_stats)) { print_arena_stats = true; }

ext/herb/extension_helpers.c

@@ -3,6 +3,7 @@
 #include "extension.h"
 #include "extension_helpers.h"
 #include "nodes.h"
+#include "parser_options_helpers.h"
 
 #include "../../src/include/herb.h"
 #include "../../src/include/lexer/token.h"
@@ -83,25 +84,7 @@ VALUE create_parse_result(AST_DOCUMENT_NODE_T* root, VALUE source, const parser_
   VALUE value = rb_node_from_c_struct((AST_NODE_T*) root);
   VALUE warnings = rb_ary_new();
   VALUE errors = rb_ary_new();
-
-  VALUE kwargs = rb_hash_new();
-  rb_hash_aset(kwargs, ID2SYM(rb_intern("strict")), options->strict ? Qtrue : Qfalse);
-  rb_hash_aset(kwargs, ID2SYM(rb_intern("track_whitespace")), options->track_whitespace ? Qtrue : Qfalse);
-  rb_hash_aset(kwargs, ID2SYM(rb_intern("analyze")), options->analyze ? Qtrue : Qfalse);
-  rb_hash_aset(kwargs, ID2SYM(rb_intern("action_view_helpers")), options->action_view_helpers ? Qtrue : Qfalse);
-  rb_hash_aset(kwargs, ID2SYM(rb_intern("transform_conditionals")), options->transform_conditionals ? Qtrue : Qfalse);
-  rb_hash_aset(kwargs, ID2SYM(rb_intern("render_nodes")), options->render_nodes ? Qtrue : Qfalse);
-  rb_hash_aset(kwargs, ID2SYM(rb_intern("strict_locals")), options->strict_locals ? Qtrue : Qfalse);
-  rb_hash_aset(kwargs, ID2SYM(rb_intern("prism_nodes")), options->prism_nodes ? Qtrue : Qfalse);
-  rb_hash_aset(kwargs, ID2SYM(rb_intern("prism_nodes_deep")), options->prism_nodes_deep ? Qtrue : Qfalse);
-  rb_hash_aset(kwargs, ID2SYM(rb_intern("prism_program")), options->prism_program ? Qtrue : Qfalse);
-  rb_hash_aset(kwargs, ID2SYM(rb_intern("timeout")), DBL2NUM((double) options->timeout_ms / 1000.0));
-
-  rb_hash_aset(
-    kwargs,
-    ID2SYM(rb_intern("max_errors")),
-    options->max_errors == 0 ? Qnil : UINT2NUM(options->max_errors)
-  );
+  VALUE kwargs = herb_build_parser_options_hash(options);
 
   VALUE parser_options_args[1] = { kwargs };
   VALUE parser_options = rb_class_new_instance_kw(1, parser_options_args, cParserOptions, RB_PASS_KEYWORDS);

java/Makefile

@@ -39,7 +39,7 @@ HERB_SOURCES = $(shell find $(SRC_DIR) -name '*.c')
 HERB_OBJECTS = $(filter-out $(SRC_DIR)/main.o, $(HERB_SOURCES:.c=.o))
 
 JNI_SOURCES = herb_jni.c extension_helpers.c
-JNI_GENERATED_SOURCES = nodes.c error_helpers.c
+JNI_GENERATED_SOURCES = nodes.c error_helpers.c parser_options_helpers.c
 
 JNI_OBJECTS = $(JNI_SOURCES:.c=.o) $(JNI_GENERATED_SOURCES:.c=.o)