Add formatter

Author: SSlinky

.vscode/launch.json

@@ -57,6 +57,16 @@
 			"printParseTree": true,
 			"visualParseTree": true
 		},
+		{
+			"name": "Debug ANTLR4 Fmt grammar",
+			"type": "antlr-debug",
+			"request": "launch",
+			"input": "${file}",
+			"grammar": "server/src/antlr/vbafmt.g4",
+			"startRule": "startRule",
+			"printParseTree": true,
+			"visualParseTree": true
+		},
 		{
 			"name": "Extension Tests",
 			"type": "extensionHost",

package.json

@@ -127,6 +127,7 @@
 		"devMate": "npx js-yaml client/src/syntaxes/dev.tmLanguage.yaml > client/out/vba.tmLanguage.json",
 		"antlr4ng": "antlr4ng -Dlanguage=TypeScript -visitor ./server/src/antlr/vba.g4 -o ./server/src/antlr/out/",
 		"antlr4ngPre": "antlr4ng -Dlanguage=TypeScript -visitor ./server/src/antlr/vbapre.g4 -o ./server/src/antlr/out/",
+		"antlr4ngFmt": "antlr4ng -Dlanguage=TypeScript -visitor ./server/src/antlr/vbafmt.g4 -o ./server/src/antlr/out/",
 		"test": "npm run tmSnapTest && npm run tmUnitTest && npm run vsctest",
 		"testsh": "sh ./scripts/e2e.sh",
 		"testps": "powershell ./scripts/e2e.ps1",

server/src/antlr/vbafmt.g4

@@ -0,0 +1,200 @@
+grammar vbafmt;
+
+options {
+    caseInsensitive = true;
+}
+
+// PARSER RULES
+startRule
+    : document EOF
+    ;
+
+endOfFile
+    : ws? EOF
+    ;
+
+document
+	: documentElement* endOfFile
+	;
+
+basicStatement
+    : ws (ambiguousComponent ws?)+ endOfStatement
+    ;
+
+// endOfStatement
+//     : bols? NEWLINE
+//     ;
+
+documentElement
+	: comment
+	| methodDeclaration
+	| attributeStatement
+    | basicStatement
+	;
+
+attributeStatement
+	: ws? ATTRIBUTE (ANYCHARS | ws | continuation | STRINGLITERAL)* endOfStatement
+	;
+
+asType
+    : ws AS ws (ANYCHARS | LPAREN | RPAREN | continuation)+
+    ;
+
+visibility
+	: PUBLIC
+	| PRIVATE
+	| GLOBAL
+	;
+
+methodOpen
+    : ws? (visibility ws)? (SUB | FUNCTION | (PROPERTY ws ANYCHARS)) ws ANYCHARS
+    ;
+
+methodParameters
+    : LPAREN (ANYCHARS | AS | STRINGLITERAL | ws)* RPAREN asType? endOfStatement
+    ;
+
+methodClose
+    : ws? END ws (SUB | FUNCTION | PROPERTY) endOfStatement
+    ;
+
+methodBody
+    : documentElement*
+    ;
+
+methodDeclaration
+	: methodOpen
+		methodParameters
+        attributeStatement?
+		methodBody
+		methodClose
+	;
+
+comment
+	: ws? (COMMENT | REMCOMMENT) (ws | ANYCHARS)*
+	;
+
+endOfStatement
+    : comment? (NEWLINE+ | endOfFile)
+    ;
+
+continuation
+    : LINE_CONTINUATION
+    ;
+
+ambiguousComponent
+    : ANYCHARS
+    | STRINGLITERAL
+    ;
+
+ws
+	: (TAB
+	| SPACE
+	| continuation)+
+	;
+
+// COMPUND LEXER RULES
+
+REMCOMMENT
+    : COLON? REM (TAB | SPACE) (LINE_CONTINUATION | ~[\r\n\u2028\u2029])*
+    ;
+
+COMMENT
+    : SINGLEQUOTE (LINE_CONTINUATION | ~[\r\n\u2028\u2029])*
+    ;
+
+
+// LEXER RULES
+
+STRINGLITERAL
+    : '"' (~["\r\n] | '""')* '"'
+    ;
+
+LPAREN
+    : '('
+    ;
+
+RPAREN
+	: ')'
+	;
+
+COLON
+    : ':'
+    ;
+
+AS
+    : 'AS'
+    ;
+
+ATTRIBUTE
+	: 'ATTRIBUTE'
+	;
+
+END
+	: 'END'
+	;
+
+ENUM
+    : 'ENUM'
+    ;
+
+FUNCTION
+    : 'FUNCTION'
+    ;
+
+PROPERTY
+    : 'PROPERTY'
+    ;
+
+GLOBAL
+    : 'GLOBAL'
+    ;
+
+LINE_CONTINUATION
+    : (TAB | SPACE) UNDERSCORE (TAB | SPACE)? '\r'? '\n'
+    ;
+
+PRIVATE
+    : 'PRIVATE'
+    ;
+
+PUBLIC
+    : 'PUBLIC'
+    ;
+
+REM
+    : 'REM'
+    ;
+
+SINGLEQUOTE
+    : '\''
+    ;
+
+SUB
+    : 'SUB'
+    ;
+
+UNDERSCORE
+    : '_'
+    ;
+
+TAB
+	: '\t'+
+	;
+
+SPACE
+    : [ \u3000]+
+    ;
+
+NEWLINE
+	: [\r\n\u2028\u2029\u0019]+
+	;
+
+// Any non-whitespace or new line characters.
+ANYCHARS
+    : ANYCHAR+
+    ;
+
+fragment ANYCHAR
+    : ~[\r\n\u2028\u2029 \t\u0019\u3000()]
+    ;

server/src/extensions/parserExtensions.ts

@@ -30,6 +30,7 @@ declare module 'antlr4ng' {
 		toRange(doc: TextDocument): Range;
       startIndex(): number;
       stopIndex(): number;
+      hasPositionOf(ctx: ParserRuleContext): boolean;
 	}
 
    interface TerminalNode {
@@ -120,6 +121,10 @@ ParserRuleContext.prototype.stopIndex = function (): number {
    return this.stop?.stop ?? this.startIndex();
 }
 
+ParserRuleContext.prototype.hasPositionOf = function (ctx: ParserRuleContext): boolean {
+   return this.startIndex() === ctx.startIndex() && this.stopIndex() === ctx.stopIndex();
+}
+
 
 TerminalNode.prototype.toRange = function (doc: TextDocument): Range {
    return Range.create(

server/src/project/document.ts

@@ -25,6 +25,7 @@ import { PropertyDeclarationElement,
 	PropertyLetDeclarationElement,
 	PropertySetDeclarationElement
 } from './elements/procedure';
+import { VbaFmtListener } from './parser/vbaListener';
 
 
 export interface DocumentSettings {
@@ -177,7 +178,7 @@ export abstract class BaseProjectDocument {
 		}
 
 		// Parse the document.
-		await (new SyntaxParser()).parseAsync(this)
+		await (new SyntaxParser()).parseAsync(this);
 
 		// Evaluate the diagnostics.
 		this.diagnostics = this.hasDiagnosticElements
@@ -187,6 +188,28 @@ export abstract class BaseProjectDocument {
 		this._isBusy = false;
 	};
 
+	async formatParseAsync(token: CancellationToken): Promise<VbaFmtListener | undefined> {
+		// Handle already cancelled.
+		if (token.isCancellationRequested) {
+			throw new ParseCancellationException(Error('Parse operation cancelled before it started.'));
+		}
+
+		// Listen for cancellation event.
+		token.onCancellationRequested(() => {
+			throw new ParseCancellationException(new Error('Parse operation cancelled during parse.'));
+		})
+
+		// Don't parse oversize documents.
+		if (await this.isOversize) {
+			console.log(`Document oversize: ${this.textDocument.lineCount} lines.`);
+            console.warn(`Syntax parsing has been disabled to prevent crashing.`);
+			return;
+		}
+
+		// Parse the document.
+		return await (new SyntaxParser()).formatParseAsync(this);
+	}
+
 	/**
 	 * Auto registers the element based on capabilities.
 	 * @returns This for chaining.

server/src/project/formatter.ts

@@ -0,0 +1,76 @@
+// Core
+import { Range, TextEdit } from 'vscode-languageserver';
+import { TextDocument } from 'vscode-languageserver-textdocument';
+import { VbaFmtListener } from './parser/vbaListener';
+
+
+export function getFormattingEdits(document: TextDocument, listener: VbaFmtListener, range?: Range): TextEdit[] {
+	// Return nothing if we have nothing.
+	if (document.getText() == '')
+		return [];
+
+	const result: TextEdit[] = [];
+
+	const startLine = range?.start.line ?? 0;
+	const endLine = (range?.end.line ?? document.lineCount) + 1;
+
+	let trackedIndentLevel = 0;
+	const baseIndentLevel = getIndentLevel(document.getText(range));
+
+	for (let i = startLine; i < endLine; i++) {
+		const text = getLine(document, i);
+
+		// Ignore comment lines.
+		if (/^\s*'/.test(text)) continue;
+
+		// Actual indent level
+		const currentIndentLevel = getIndentLevel(text);
+		const newIndentLevel = listener.getIndent(i);
+		if (currentIndentLevel != newIndentLevel) {
+			result.push({
+				range: getIndentRange(text, i)!,
+				newText: ' '.repeat(newIndentLevel * 2)
+			});
+		}
+	}
+
+	return result;
+}
+
+function getExpectedIndent(listener: VbaFmtListener, range: Range, n: number) {
+	// The listener will be offset by the range, e.g., if the range.start.line
+	// is 5 then getting line 2 from the listener will be document line 7.
+	return listener.getIndent(n - range.start.line + 1)
+}
+
+function getIndentRange(text: string, n: number): Range | undefined {
+	const match = /^(?!\s*')(\s*)/m.exec(text);
+	if (match) {
+		return {
+			start: { line: n, character: 0 },
+			end: { line: n, character: match[0].length }
+		}
+	}
+}
+
+function getIndentLevel(text: string): number {
+	// Get spaces at start of non-comment lines (tab is four spaces)
+	const normalised = text.replace(/\t/g, '    ')
+	const match = /^(?!\s*')(\s*)/m.exec(normalised);
+	
+	// Default is no indent.
+	if (!match) {
+		return 0;
+	}
+
+	// Four spaces per indent.
+	return (match[0].length / 4) | 0;
+}
+
+function getLine(d: TextDocument, n: number): string {
+	return d.getText({
+		start: { line: n, character: 0 },
+		end: { line: n + 1, character: 0
+		}
+	})
+}