Fix expression visitor to make escape work in nested entity definition (

#1106) * fix expression lexer * fix expression parser
microsoft · Feb 5, 2021 · fe56306 · fe56306
1 parent c622c6f
commit fe56306
Show file tree

Hide file tree

Showing 2 changed files with 102 additions and 46 deletions.
diff --git a/packages/lu/src/parser/lufile/visitor.js b/packages/lu/src/parser/lufile/visitor.js
@@ -95,58 +95,56 @@ class Visitor {
         let entityNameCapture = false;
         let entityValueCapture = false;
         let entityRoleCapture = false;
-        exp.split('').forEach(char => {
-            switch(char) 
-            {
-                case '{':
-                    let newEntity = {entityName : '', role : '', entityValue : undefined, parent : curEntity};
-                    curList.push(newEntity);
-                    curEntity = newEntity;
-                    entityNameCapture = true;
-                    entityRoleCapture = false;
-                    entityValueCapture = false;
-                    break;
-                case '}':
-                    curEntity = curEntity.parent || undefined;
-                    curList = curEntity != undefined ? curEntity.entityValue : splitString;
-                    entityValueCapture = false;
-                    entityRoleCapture = false;
-                    entityNameCapture = false;
-                    break;
-                case '=':
-                    curEntity.entityValue = [];
-                    curList = curEntity.entityValue;
+        let expChars = exp.split('');
+        let escapeChar = false;
+        expChars.forEach(function (char, index) {
+            if (char === '\\' && expChars.length > index + 1 && EscapeCharsInUtterance.includes(expChars[index + 1])) {
+                escapeChar = true;
+            } else if (char === '{' && !escapeChar) {
+                let newEntity = {entityName : '', role : '', entityValue : undefined, parent : curEntity};
+                curList.push(newEntity);
+                curEntity = newEntity;
+                entityNameCapture = true;
+                entityRoleCapture = false;
+                entityValueCapture = false;
+            } else if (char === '}' && !escapeChar) {
+                curEntity = curEntity.parent || undefined;
+                curList = curEntity != undefined ? curEntity.entityValue : splitString;
+                entityValueCapture = false;
+                entityRoleCapture = false;
+                entityNameCapture = false;
+            } else if (char === '=' && !entityValueCapture) {
+                curEntity.entityValue = [];
+                curList = curEntity.entityValue;
+                entityNameCapture = false;
+                entityValueCapture = true;
+                entityRoleCapture = false;
+            } else if (char === ':' && !entityRoleCapture) {
+                if (curEntity !== undefined && curEntity.entityName !== '' && entityNameCapture === true) {
+                    entityRoleCapture = true;
                     entityNameCapture = false;
-                    entityValueCapture = true;
-                    entityRoleCapture = false;
-                    break;
-                case ':':
-                    if (curEntity !== undefined && curEntity.entityName !== '' && entityNameCapture === true) {
-                        entityRoleCapture = true;
-                        entityNameCapture = false;
-                        entityValueCapture = false;
-                    } else {
-                        curList.push(char);
-                    }
-                    break;
-                default :
-                    if (entityNameCapture) {
-                        curEntity.entityName += char;
-                    } else if (entityValueCapture) {
-                        if (char === ' ') {
-                            // we do not want leading spaces
-                            if (curList.length !== 0) {
-                                curList.push(char);
-                            }
-                        } else {
+                    entityValueCapture = false;
+                } else {
+                    curList.push(char);
+                }
+            } else {
+                escapeChar = false;
+                if (entityNameCapture) {
+                    curEntity.entityName += char;
+                } else if (entityValueCapture) {
+                    if (char === ' ') {
+                        // we do not want leading spaces
+                        if (curList.length !== 0) {
                             curList.push(char);
                         }
-                    } else if (entityRoleCapture) {
-                        curEntity.role += char;
                     } else {
                         curList.push(char);
                     }
-                    break;
+                } else if (entityRoleCapture) {
+                    curEntity.role += char;
+                } else {
+                    curList.push(char);
+                }
             }
         });
         return splitString;

diff --git a/packages/lu/test/parser/lufile/parseFileContents.parseFile.test.js b/packages/lu/test/parser/lufile/parseFileContents.parseFile.test.js
@@ -1147,4 +1147,62 @@ describe('parseFile correctly parses utterances', function () {
             })
             .catch((err) => done(err));
         });
+
+        it("Correctly parses utterance that escape char \\ in nested entity definition", function (done) {
+          let testLU = `
+                # test
+                - {@Command={@Action={@BoldAction=bold}} \\{{@ActionTargetPhrase=directions from seattle to portland}\\}}`;
+          parseFile
+            .parseFile(testLU)
+            .then((res) => {
+              assert.equal(
+                res.LUISJsonStructure.utterances[0].text,
+                "bold {directions from seattle to portland}"
+              );
+              assert.equal(res.LUISJsonStructure.entities.length, 4);
+              assert.equal(res.LUISJsonStructure.entities[0].name, "BoldAction");
+              assert.equal(res.LUISJsonStructure.entities[1].name, "Action");
+              assert.equal(res.LUISJsonStructure.entities[2].name, "ActionTargetPhrase");
+              assert.equal(res.LUISJsonStructure.entities[3].name, "Command");
+              done();
+            })
+            .catch((err) => done(err));
+        });
+
+        it("Correctly parses utterance that with equal sign as entity value", function (done) {
+          let testLU = `
+                # test
+                - {@Command={@Action={@BoldAction=emphasise from}} {@ActionTargetRange={@ActionTargetStart=@username} {@ActionTargetSeparator=through} {@ActionTargetEnd==}}}`;
+          parseFile
+            .parseFile(testLU)
+            .then((res) => {
+                assert.equal(res.LUISJsonStructure.utterances[0].text, "emphasise from @username through =");
+                assert.equal(res.LUISJsonStructure.entities.length, 7);
+                assert.equal(res.LUISJsonStructure.entities[0].name, "BoldAction");
+                assert.equal(res.LUISJsonStructure.entities[1].name, "Action");
+                assert.equal(res.LUISJsonStructure.entities[2].name, "ActionTargetStart");
+                assert.equal(res.LUISJsonStructure.entities[3].name, "ActionTargetSeparator");
+                assert.equal(res.LUISJsonStructure.entities[4].name, "ActionTargetEnd");
+                assert.equal(res.LUISJsonStructure.entities[5].name, "ActionTargetRange");
+                assert.equal(res.LUISJsonStructure.entities[6].name, "Command");
+                done();
+            })
+            .catch((err) => done(err));
+        });
+
+        it("Correctly parses utterance that with colon sign in entity role", function (done) {
+          let testLU = `
+                      # test
+                      - {@city::startCity=Seattle}`;
+          parseFile
+            .parseFile(testLU)
+            .then((res) => {
+                assert.equal(res.LUISJsonStructure.utterances[0].text, "Seattle");
+                assert.equal(res.LUISJsonStructure.entities.length, 1);
+                assert.equal(res.LUISJsonStructure.entities[0].name, "city");
+                assert.equal(res.LUISJsonStructure.entities[0].roles[0], ":startCity");
+                done();
+            })
+            .catch((err) => done(err));
+        });
 })