return captured values from paths that reach the NFA end

Author: ColinEberhardt

assembly/__spec_tests__/generated.spec.ts

@@ -163,7 +163,7 @@ it("line: 34 - matches (a|b)c|a(b|c) against 'ac'", () => {
   expect(match.matches[0]).toBe("ac");
   expect(match.matches[1]).toBe("a");
 });
-xit("line: 35 - matches (a|b)c|a(b|c) against 'ab'", () => {
+it("line: 35 - matches (a|b)c|a(b|c) against 'ab'", () => {
   const match = exec("(a|b)c|a(b|c)", "ab", "");
   expect(match.matches[0]).toBe("ab");
   expect(match.matches[1]).toBe("");

assembly/__tests__/capture-group.spec.ts

@@ -27,7 +27,7 @@ it("supports capture groups", () => {
   expect(match.matches[2]).toBe("aaa");
 });
 
-xit("should not return captured values for non-matching alternations", () => {
+it("should not return captured values for non-matching alternations", () => {
   const match = exec("(a|b)c|a(b|c)", "ab");
   expect(match.matches[0]).toBe("ab");
   expect(match.matches[1]).toBe("");

assembly/nfa/nfa.ts

@@ -49,6 +49,8 @@ export class GroupStartMarkerState extends State {
 export class GroupEndMarkerState extends State {
   // a bit yucky - storing transient state in the state machine!
   capture: string = "";
+  // captures from the path through the NFA that reaches the end are flagged
+  flagged: bool = false;
 
   constructor(next: State, public startMarker: GroupStartMarkerState) {
     super();

assembly/regexp.ts

@@ -45,6 +45,10 @@ function recursiveBacktrackingSearch(
       position
     );
     if (match != null) {
+      // when unwinding the stack after a successful match, flag the captured values
+      if (state instanceof GroupEndMarkerState) {
+        (state as GroupEndMarkerState).flagged = true;
+      }
       return match;
     }
   }
@@ -132,7 +136,9 @@ export class RegExp {
       // we have found a match
       if (matchStr != null) {
         const match = new Match(
-          [matchStr!].concat(groupMarkers.map<string>((m) => m.capture)),
+          [matchStr!].concat(
+            groupMarkers.map<string>((m) => (m.flagged ? m.capture : ""))
+          ),
           matchIndex,
           str
         );

spec/test-generator.js

@@ -14,7 +14,6 @@ const knownIssues = {
   "BUG: doesn't support anchors within capture groups": [20],
   // https://github.com/ColinEberhardt/assemblyscript-regex/issues/2
   "BUG: Should not return captured values for non-matching alternations": [
-    35,
     133,
     150,
     187,

ts/index.ts

@@ -5,7 +5,7 @@ globalAny.log = console.log;
 
 import { RegExp } from "../assembly/regexp";
 
-const regexObj = new RegExp("^(([^!]+!)?([^!]+)|.+!([^!]+!)([^!]+))$");
-const match = regexObj.exec("foo!bar!bas");
+const regexObj = new RegExp("((a)(b)c)(d)");
+const match = regexObj.exec("abcd");
 
 console.log(match);