tree-sitter-delphi13-pure 1.1.0 → 1.2.1

package/grammar.js

@@ -104,8 +104,15 @@ function enable_if(cond, ...args) {
 // Generate rules for trailing & non-trailing statements
 function statements(trailing) {
 	let rn            = x => trailing ? x + 'Tr' : x
-	let lastStatement = $ => trailing ? optional(tr($,'_statement')) : $._statement;
-	let lastStatement1= $ => trailing ? tr($,'_statement') : $._statement;
+	// Ported from root 2026-07-16: a goto label may prefix the
+	// single-statement BODY of then/else/do/on (`if I = 0 then Found: ...`,
+	// `else FoundMismatch: Exit`, `while true do redo: case ...`).
+	// prec.dynamic(-1) keeps the case-ARM interpretation winning where a
+	// label could also open an arm; requires [caseCase]/[caseCaseTr]
+	// conflicts.
+	let bodyLabel     = $ => optional(prec.dynamic(-1, $.label));
+	let lastStatement = $ => trailing ? optional(seq(bodyLabel($), tr($,'_statement'))) : seq(bodyLabel($), $._statement);
+	let lastStatement1= $ => trailing ? seq(bodyLabel($), tr($,'_statement')) : seq(bodyLabel($), $._statement);
 	let semicolon     = trailing ? [] : [';'];
 
 	return Object.fromEntries([
@@ -272,6 +279,37 @@ function statements(trailing) {
 	]);
 }
 
+// Ported from root 2026-07-16: shared front of a routine declaration
+// (everything before the first `;`) — used by _declProc (strict tail,
+// backs defProc) and _declProcLenient (separator-form tail, interface
+// declaration lists only).
+function declProcFront($) {
+	return [
+		...enable_if(fpc, optional($.kGeneric)),
+		optional($.kClass),
+		choice($.kProcedure, $.kFunction, $.kConstructor, $.kDestructor),
+		field('name', $._genericName),
+		field('args', optional($.declArgs)),
+		optional(seq(
+			':',
+			field('type', $.typeref),
+		)),
+		field('assign', optional($.defaultValue)),
+		// Inline calling-convention or procAttribute BEFORE the terminating
+		// ';' — Embarcadero RTL / AzureAPI use lenient forms:
+		//   function DbiInitFn(...): DBIResult stdcall;   (iter 56)
+		//   procedure PreflightBlobRequest(...) overload; (iter 60)
+		// `repeat` allows multiple chained attributes like `... cdecl overload`.
+		repeat(choice(
+			$.kStdcall, $.kCdecl, $.kSafecall, $.kPascal,
+			$.kRegister, $.kWinapi, $.kInline,
+			$.kOverload, $.kVirtual, $.kAbstract, $.kOverride,
+			$.kReintroduce, $.kStatic, $.kDynamic, $.kFinal,
+			$.kUnsafe   // ARC/AUTOREFCOUNT method directive (ported v1.1.0)
+		)),
+	];
+}
+
 module.exports = grammar({
 	name: "delphi13_pure",
 
@@ -308,6 +346,26 @@ module.exports = grammar({
 		// optional _definitions before AND after declContains. GLR forks
 		// on which slot absorbs a given definition.
 		[$.package],
+		// Iter 2026-07-16 (implicit-init, ported from root): the unit tail's
+		// `tr($,'block')` arm forks against `implementation` absorbing the
+		// same `begin` into its `_definitions` (prec(-1) blockTr recovery).
+		// GLR keeps both; precedence picks the unit-tail block.
+		[$.implementation],
+		// Iter 2026-07-16 (label-as-body, ported from root): body label vs
+		// the arm's own jumpLabel slot (`1: Foo: x`); prec.dynamic(-1) keeps
+		// the arm interpretation winning.
+		[$.caseCase],
+		[$.caseCaseTr],
+		// Iter 2026-07-16 (lenient directive tail, ported from root): group
+		// separator `;` vs optional final `;`, and lenient-tail groups vs
+		// declProcFwd's strict groups while both are live.
+		[$.procAttribute],
+		[$._declProcLenient, $._procAttributeNoExt],
+		// Iter 2026-07-16 (Register as a field name, ported from root):
+		// previous field's trailing callconv vs next field's name; `;` vs `:`
+		// one token later decides.
+		[$.declFieldNoSemi, $.declField],
+		[$.declField],
 		// Anonymous record/class types in type position conflict with the
 		// type-declaration form (`type Foo = record ... end;`). Both reach
 		// declClass but at different parser states.
@@ -381,13 +439,29 @@ module.exports = grammar({
 
 	rules: {
 		root:               $ => optional(choice(
-			$.program,
-			$.library,
-			$.unit,
-			$.package,
+			seq(
+				choice(
+					$.program,
+					$.library,
+					$.unit,
+					$.package,
+				),
+				// Iter 2026-07-16 (ported from root): dcc ignores everything
+				// after the final `end.` (warning W1011 only). One
+				// low-precedence catch-all token; linters can flag the node.
+				optional($.trailingText),
+			),
 			$._definitions // For include files
 		)),
 
+		// Everything from the first non-blank char after the module's final
+		// `end.` to EOF. token(prec(-2)) so real tokens (incl. comments)
+		// always win over it.
+		// NOTE: `(.|\s)*` rather than `[\s\S]*` — tree-sitter's regex engine
+		// compiles the latter to a non-greedy/empty match (token stopped after
+		// one char; observed 2026-07-16).
+		trailingText:       $ => token(prec(-2, /[^\s](.|\s)*/)),
+
 		// HIGH LEVEL ----------------------------------------------------------
 
 		program:            $ => seq(
@@ -420,13 +494,32 @@ module.exports = grammar({
 				$.kLibrary,   // ported from root v1.1.0
 			)),
 			';',
+			// Unit tail, restructured 2026-07-16 (ported from root, iter:
+			// implicit-init): `initialization`/`finalization` moved out of the
+			// flat repeat into the kEnd tail arm, plus a mutually-exclusive
+			// `tr($,'block')` arm for the Turbo-Pascal implicit initialization
+			// block (`begin <stmts> end.` — dcc 13 still accepts it). Because
+			// the block arm is unreachable once `initialization` appears, the
+			// statement-list fork the old shape suffered cannot arise; the only
+			// fork is blockTr-inside-_definitions (prec -1 recovery) vs
+			// blockTr-as-unit-tail (prec 0), same as `library`. Requires the
+			// [$.implementation] conflict. CST: the implicit block renders as
+			// the same `block` node `program`/`library` produce.
 			repeat(choice(
 				$.interface,
 				$.implementation,
-				$.initialization,
-				$.finalization,
 			)),
-			$.kEnd, $.kEndDot
+			choice(
+				seq(
+					repeat(choice(
+						$.initialization,
+						$.finalization,
+					)),
+					$.kEnd
+				),
+				tr($,'block'),
+			),
+			$.kEndDot
 		),
 
 		// Delphi package file (.dpk). Structure:
@@ -515,6 +608,13 @@ module.exports = grammar({
 		// en/EN not followed by d/D, end immediately followed by ident-char
 		// (so it's NOT the keyword, e.g. 'endif' / 'endloop').
 		asmBody: $ => token(prec(-1, repeat1(choice(
+			// Ported from root 2026-07-16: consume a whole COMMENT as one chunk so a
+			// bare `end` *inside* a comment doesn't falsely terminate the asm block.
+			//   add edi,ecx {point EDI to end of destination}   <-- AsyncPro AwFView
+			//   mov eax,1   // end of loop
+			// Must precede the single-char alternatives: longest match wins.
+			/\{[^}]*\}/,
+			/\/\/[^\n]*/,
 			// Identifier not starting with e/E — consume as a unit so an
 			// embedded `end` substring (like in `addend` or `xchg ext`)
 			// doesn't falsely terminate the body.
@@ -753,7 +853,15 @@ module.exports = grammar({
 		typeref:         $ => seq(
 			...enable_if(fpc, field('_dummy', optional($.kSpecialize))),
 			$._typeref,
-			...enable_if(delphi, optional(seq($.kDeprecated, $._expr))),
+			// Ported from root 2026-07-16: trailing kPlatform hint on the
+			// typeref — makes `Default8087CW: Word platform = $033F;`
+			// (System.pas x6) parse with declVar's normal defaultValue.
+			// Mirrors the existing kDeprecated role-set; the hint-then-value
+			// ARM inside declVar is a bisect-confirmed table bomb (TODO.md).
+			...enable_if(delphi, optional(choice(
+				seq($.kDeprecated, $._expr),
+				$.kPlatform,
+			))),
 		),
 
 		_typeref:        $ => choice(
@@ -805,8 +913,18 @@ module.exports = grammar({
 			$.identifier, $.genericDot, ...enable_if(templates, $.genericTpl)
 		),
 		genericArgs:     $ => delimited1($.genericArg, ';'),
+		// Ported from root 2026-07-16: nested generic instantiation as a
+		// genericArg name element (method resolution clause `function
+		// TFunc<T1, IEnumerable<TResult>>.Invoke = Bind;`). Self-contained
+		// identifier-only recursion — referencing genericTpl/_genericName
+		// here explodes table construction.
+		genericArgTpl:   $ => seq(
+			$.identifier, $.kLt,
+			delimited1(choice($.identifier, $.genericArgTpl)),
+			$.kGt
+		),
 		genericArg:      $ => seq(
-			field('name', delimited1($.identifier)),
+			field('name', delimited1(choice($.identifier, $.genericArgTpl))),
 			// Type constraint: `T: BaseType` or one/more of the special
 			// constraint keywords `class` / `record` / `constructor`, comma-
 			// separated. Examples:
@@ -943,7 +1061,11 @@ module.exports = grammar({
 		),
 
 		_declarations:   $ => repeat1(choice(
-			$.declTypes, $.declVars, $.declConsts, $.declProc, $.declProp,
+			$.declTypes, $.declVars, $.declConsts,
+			// Lenient directive tail in interface declaration lists (ported
+			// from root); rendered as a plain declProc node.
+			alias($.declProcLenient, $.declProc),
+			$.declProp,
 			alias($.declProcFwd, $.declProc),
 			$.declUses, $.declLabels, $.declExports,
 			// Phase 3 (pure): CIL/.NET [assembly: ...] attributes inside
@@ -1110,6 +1232,12 @@ module.exports = grammar({
 			...enable_if(rtti, optional($.rttiAttributes)),
 			choice($._declProc, $._declOperator),
 		),
+		// Interface/forward flavor with the lenient directive tail; aliased
+		// back to `declProc` at its use site (CST-identical).
+		declProcLenient: $ => seq(
+			...enable_if(rtti, optional($.rttiAttributes)),
+			choice($._declProcLenient, $._declOperator),
+		),
 
 		declVar:         $ => seq(
 			...enable_if(rtti, optional($.rttiAttributes)),
@@ -1140,6 +1268,13 @@ module.exports = grammar({
 				alias($.kDeprecated,   $.identifier),
 				alias($.kExperimental, $.identifier),
 				alias($.kRegister,     $.identifier),
+				// Ported from root 2026-07-16: same family — `local` is a
+				// procAttribute, so `var X: Integer; Local: Integer;` had the
+				// parser eat `Local` as a trailing directive on X.
+				alias($.kLocal,        $.identifier),
+				// Ported from root 2026-07-16: `dispid` is a property/method
+				// directive — same trap (System.Win.ObjComAuto, Vcl.OleCtrls).
+				alias($.kDispId,       $.identifier),
 			))),
 			':',
 			// Phase 3b iter 36: declVar type may be a subrange:
@@ -1405,7 +1540,13 @@ module.exports = grammar({
 
 		declField:       $ =>  seq(
 			...enable_if(rtti, optional($.rttiAttributes)),
-			field('name', delimited1($.identifier)),
+			// Ported from root 2026-07-16: `Register` as a FIELD NAME after a
+			// prior field (Winapi.D3D10 shader reflection records). Narrow —
+			// only kRegister; see the root note for why not all callconvs.
+			field('name', delimited1(choice(
+				$.identifier,
+				alias($.kRegister, $.identifier),
+			))),
 			':',
 			// Ported from root v1.1.x: field type may be a subrange
 			// (`FtrListCount: 0 .. FTRRECMAXCOUNT;`, ORM3 ImportCadFile.pas).
@@ -1500,32 +1641,27 @@ module.exports = grammar({
 		// Stuff for procedure / function / operator declarations
 
 		_declProc:       $ => seq(
-			...enable_if(fpc, optional($.kGeneric)),
-			optional($.kClass),
-			choice($.kProcedure, $.kFunction, $.kConstructor, $.kDestructor),
-			field('name', $._genericName),
-			field('args', optional($.declArgs)),
-			optional(seq(
-				':',
-				field('type', $.typeref),
-			)),
-			field('assign', optional($.defaultValue)),
-			// Inline calling-convention or procAttribute BEFORE the terminating
-			// ';' — Embarcadero RTL / AzureAPI use lenient forms:
-			//   function DbiInitFn(...): DBIResult stdcall;   (iter 56)
-			//   procedure PreflightBlobRequest(...) overload; (iter 60)
-			// `repeat` allows multiple chained attributes like `... cdecl overload`.
-			repeat(choice(
-				$.kStdcall, $.kCdecl, $.kSafecall, $.kPascal,
-				$.kRegister, $.kWinapi, $.kInline,
-				$.kOverload, $.kVirtual, $.kAbstract, $.kOverride,
-				$.kReintroduce, $.kStatic, $.kDynamic, $.kFinal,
-				$.kUnsafe   // ARC/AUTOREFCOUNT method directive (ported v1.1.0)
-			)),
+			...declProcFront($),
 			';',
+			// STRICT tail — backs defProc, where the `;` before the body is
+			// mandatory. Lenient separator-form twin below (ported from root).
 			repeat($._procAttributeNoExt)
 		),
 
+		// Lenient twin for interface declaration lists: directive groups are
+		// `;`-SEPARATED with the final `;` optional, so
+		// `function F: Boolean; overload` before another declaration parses
+		// (dxCryptoAPI, dxServerModeUtils, dxGDIPlusAPI — dcc32-verified).
+		_declProcLenient: $ => seq(
+			...declProcFront($),
+			';',
+			optional(seq(
+				repeat1(field('attribute', $.procAttribute)),
+				repeat(seq(';', repeat1(field('attribute', $.procAttribute)))),
+				optional(';')
+			))
+		),
+
 		_declOperator:   $ => seq(
 			optional($.kClass),
 			$.kOperator,
@@ -1921,7 +2057,10 @@ module.exports = grammar({
 		// so structural tokens stay intact.
 		identifier:        $ => /&{0,2}[a-zA-Z_€-￿][0-9_a-zA-Z€-￿]*/,
 
-	  	_space:            $ => /[\s\r\n\t]+/,
+		// Classic Pascal treats every control char <= #31 as a token
+		// separator (dcc compiles sources with stray bytes like the 0x12 in
+		// DevExpress dxPDFForm.pas), so the whitespace extra does too.
+	  	_space:            $ => /[\s\x00-\x1F]+/,
 		// Single preprocessor directive. The external scanner (src/scanner.c)
 		// handles multi-line `{$IF*}...{$END*}` blocks via the pp_block token,
 		// refusing blocks that wrap structural decls so they parse normally.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "tree-sitter-delphi13-pure",
-  "version": "1.1.0",
+  "version": "1.2.1",
   "description": "Pure Delphi 13 tree-sitter sub-grammar — drops {$IFDEF}/pp_* tokens entirely and expects preprocessor-resolved source as input. Pairs with delphi13-preprocessor to reach 99%+ pass on real Delphi 13 corpora.",
   "main": "bindings/node",
   "types": "bindings/node",