Add support for Perl quoted and Python named capturing group syntax v…

…ariants
asmblah · May 10, 2024 · 683b72f · 683b72f
1 parent 086fbd4
commit 683b72f
Show file tree

Hide file tree

Showing 3 changed files with 94 additions and 5 deletions.
diff --git a/src/spec/parserGrammar.ts b/src/spec/parserGrammar.ts
@@ -368,9 +368,14 @@ export default {
         },
         'N_NAMED_CAPTURING_GROUP': {
             components: [
-                /\(\?</,
-                { name: 'groupName', what: /[^>]+/ },
-                />/,
+                {
+                    oneOf: [
+                        // Chevron-bracketed Perl or Python named capturing group.
+                        [/\(\?P?</, { name: 'groupName', what: /[^>]+/ }, />/],
+                        // Quoted Perl named capturing group.
+                        [/\(\?'/, { name: 'groupName', what: /[^']+/ }, /'/],
+                    ],
+                },
                 { name: 'components', zeroOrMoreOf: 'N_COMPONENT' },
                 /\)/,
             ],

diff --git a/test/integration/match/group/namedCapturingGroupTest.ts b/test/integration/match/group/namedCapturingGroupTest.ts
@@ -12,7 +12,7 @@ import { expect } from 'chai';
 
 describe('Named capturing group match integration', () => {
     describe('in optimised mode', () => {
-        it('should be able to capture a named capturing group', () => {
+        it('should be able to capture a single chevron-bracketed Perl named capturing group', () => {
             const matcher = emulator.compile('my (?<grabbed>captured) text');
 
             const match = matcher.matchOne('my captured text');
@@ -25,6 +25,32 @@ describe('Named capturing group match integration', () => {
             expect(match?.getNamedCapture('grabbed')).to.equal('captured');
         });
 
+        it('should be able to capture a single-quoted Perl named capturing group', () => {
+            const matcher = emulator.compile("my (?'grabbed'captured) text");
+
+            const match = matcher.matchOne('my captured text');
+
+            expect(match).not.to.be.null;
+            expect(match?.getCaptureCount()).to.equal(2);
+            // Named captures are also available by their index.
+            expect(match?.getNumberedCapture(0)).to.equal('my captured text');
+            expect(match?.getNumberedCapture(1)).to.equal('captured');
+            expect(match?.getNamedCapture('grabbed')).to.equal('captured');
+        });
+
+        it('should be able to capture a single Python named capturing group', () => {
+            const matcher = emulator.compile('my (?P<grabbed>captured) text');
+
+            const match = matcher.matchOne('my captured text');
+
+            expect(match).not.to.be.null;
+            expect(match?.getCaptureCount()).to.equal(2);
+            // Named captures are also available by their index.
+            expect(match?.getNumberedCapture(0)).to.equal('my captured text');
+            expect(match?.getNumberedCapture(1)).to.equal('captured');
+            expect(match?.getNamedCapture('grabbed')).to.equal('captured');
+        });
+
         it('should be able to backtrack into a capturing group', () => {
             const matcher = emulator.compile('my (?<grabbed>a+)aa text');
 

diff --git a/test/integration/parser/group/namedCapturingGroupTest.ts b/test/integration/parser/group/namedCapturingGroupTest.ts
@@ -18,7 +18,7 @@ describe('Parser named capturing group integration', () => {
         parser = emulator.createParser();
     });
 
-    it('should be able to parse a regex pattern containing a single capturing group', () => {
+    it('should be able to parse a regex pattern containing a single chevron-bracketed Perl named capturing group', () => {
         const ast = parser.parse('my (?<found>stuff inside) here');
 
         expect(ast.getPattern()).to.equal('my (?<found>stuff inside) here');
@@ -47,6 +47,64 @@ describe('Parser named capturing group integration', () => {
         });
     });
 
+    it('should be able to parse a regex pattern containing a single-quoted Perl named capturing group', () => {
+        const ast = parser.parse("my (?'found'stuff inside) here");
+
+        expect(ast.getPattern()).to.equal("my (?'found'stuff inside) here");
+        expect(ast.getParsingAst()).to.deep.equal({
+            'name': 'N_PATTERN',
+            'components': [
+                {
+                    'name': 'N_LITERAL',
+                    'text': 'my ',
+                },
+                {
+                    'name': 'N_NAMED_CAPTURING_GROUP',
+                    'groupName': 'found',
+                    'components': [
+                        {
+                            'name': 'N_LITERAL',
+                            'text': 'stuff inside',
+                        },
+                    ],
+                },
+                {
+                    'name': 'N_LITERAL',
+                    'text': ' here',
+                },
+            ],
+        });
+    });
+
+    it('should be able to parse a regex pattern containing a single Python named capturing group', () => {
+        const ast = parser.parse('my (?P<found>stuff inside) here');
+
+        expect(ast.getPattern()).to.equal('my (?P<found>stuff inside) here');
+        expect(ast.getParsingAst()).to.deep.equal({
+            'name': 'N_PATTERN',
+            'components': [
+                {
+                    'name': 'N_LITERAL',
+                    'text': 'my ',
+                },
+                {
+                    'name': 'N_NAMED_CAPTURING_GROUP',
+                    'groupName': 'found',
+                    'components': [
+                        {
+                            'name': 'N_LITERAL',
+                            'text': 'stuff inside',
+                        },
+                    ],
+                },
+                {
+                    'name': 'N_LITERAL',
+                    'text': ' here',
+                },
+            ],
+        });
+    });
+
     it('should be able to parse a regex pattern containing a named capturing group nested inside a numbered one', () => {
         const ast = parser.parse('my (stuff (?<inner>goes) inside) here');