Throw exception when parsing entity name contains invalid char

packages/lu/src/parser/lufile/entitySection.js

@@ -2,6 +2,7 @@ const EntitySectionContext = require('./generated/LUFileParser').LUFileParser.En
 const DiagnosticSeverity = require('./diagnostic').DiagnosticSeverity;
 const BuildDiagnostic = require('./diagnostic').BuildDiagnostic;
 const LUSectionTypes = require('./../utils/enums/lusectiontypes');
+const InvalidCharsInIntentOrEntityName = require('./../utils/enums/invalidchars').InvalidCharsInIntentOrEntityName;
 
 class EntitySection {
     /**
@@ -19,13 +20,23 @@ class EntitySection {
     }
 
     ExtractName(parseTree) {
+        let entityName;
         if (parseTree.entityDefinition().entityLine().entityName()) {
-            return parseTree.entityDefinition().entityLine().entityName().getText().trim();
+            entityName = parseTree.entityDefinition().entityLine().entityName().getText().trim();
         } else {
             this.Errors.push(BuildDiagnostic({
                 message: "Invalid entity line, did you miss entity name after $",
                 context: parseTree.entityDefinition().entityLine()
-            }))     
+            }));
+        }
+
+        if (entityName && InvalidCharsInIntentOrEntityName.some(x => entityName.includes(x))) {
+            this.Errors.push(BuildDiagnostic({
+                message: `Invalid entity line, entity name ${entityName} cannot contain any of the following characters: [<, >, *, %, &, :, \\, $]`,
+                context: parseTree.newEntityDefinition().newEntityLine()
+            }));
+        } else {
+            return entityName;
         }
     }
 

packages/lu/src/parser/lufile/newEntitySection.js

@@ -2,6 +2,7 @@ const NewEntitySectionContext = require('./generated/LUFileParser').LUFileParser
 const DiagnosticSeverity = require('./diagnostic').DiagnosticSeverity;
 const BuildDiagnostic = require('./diagnostic').BuildDiagnostic;
 const LUSectionTypes = require('./../utils/enums/lusectiontypes');
+const InvalidCharsInIntentOrEntityName = require('./../utils/enums/invalidchars').InvalidCharsInIntentOrEntityName;
 
 class NewEntitySection {
     /**
@@ -23,16 +24,26 @@ class NewEntitySection {
     }
 
     ExtractName(parseTree) {
+        let entityName
         if (parseTree.newEntityDefinition().newEntityLine().newEntityName()) {
-            return parseTree.newEntityDefinition().newEntityLine().newEntityName().getText().trim();
+            entityName = parseTree.newEntityDefinition().newEntityLine().newEntityName().getText().trim();
         } else if (parseTree.newEntityDefinition().newEntityLine().newEntityNameWithWS()) {
-            return parseTree.newEntityDefinition().newEntityLine().newEntityNameWithWS().getText().trim();
+            entityName = parseTree.newEntityDefinition().newEntityLine().newEntityNameWithWS().getText().trim();
         } else {
             this.Errors.push(BuildDiagnostic({
                 message: "Invalid entity line, did you miss entity name after @",
                 context: parseTree.newEntityDefinition().newEntityLine()
             }))
         }
+
+        if (entityName && InvalidCharsInIntentOrEntityName.some(x => entityName.includes(x))) {
+            this.Errors.push(BuildDiagnostic({
+                message: `Invalid entity line, entity name ${entityName} cannot contain any of the following characters: [<, >, *, %, &, :, \\, $]`,
+                context: parseTree.newEntityDefinition().newEntityLine()
+            }));
+        } else {
+            return entityName;
+        }
     }
 
     ExtractType(parseTree) {

packages/lu/src/parser/lufile/visitor.js

@@ -1,5 +1,6 @@
 const lp = require('./generated/LUFileParser').LUFileParser;
 const LUISObjNameEnum = require('./../utils/enums/luisobjenum');
+const InvalidCharsInIntentOrEntityName = require('./../utils/enums/invalidchars').InvalidCharsInIntentOrEntityName;
 
 class Visitor {
     /**
@@ -38,6 +39,12 @@ class Visitor {
     static recurselyResolveTokenizedUtterance(tokUtt, entities, errorMsgs, srcUtterance) {
         for (const item of tokUtt) {
             if (item === Object(item)) {
+                let entityName = item.entityName.trim()
+                if (entityName && InvalidCharsInIntentOrEntityName.some(x => entityName.includes(x))) {
+                    errorMsgs.push(`Invalid utterance line, entity name ${entityName} cannot contain any of the following characters: [<, >, *, %, &, :, \\, $]`);
+                    continue;
+                }
+
                 if (item.entityValue === undefined) {
                     // we have a pattern.any entity
                     const patternStr = item.role ? `{${item.entityName}:${item.role}}` : `{${item.entityName}}`

packages/lu/src/parser/utils/enums/invalidchars.js

@@ -0,0 +1,8 @@
+/**
+ * Copyright (c) Microsoft Corporation. All rights reserved.
+ * Licensed under the MIT License.
+ */
+ // Invalid chars in intent or entity name
+module.exports = {
+    InvalidCharsInIntentOrEntityName: ['<', '>', '*', '%', '&', ':', '\\', '$']
+};

packages/lu/test/commands/luis/convert.test.js

@@ -233,6 +233,19 @@ describe('luis:convert negative tests', () => {
             })
 
     })
+
+    it('luis:convert should show ERR message when entity name contains invalid char', (done) => {
+        loadLuFile('./../../fixtures/testcases/bad5.lu')
+            .then(res => {
+                LuisBuilder.fromLUAsync(res)
+                    .then(res => done(res))
+                    .catch(err => {
+                        assert.isTrue(err.text.includes('[ERROR] line 2:0 - line 2:26: Invalid utterance line, entity name @addto*Property cannot contain any of the following characters: [<, >, *, %, &, :, \\, $]'))
+                        assert.isTrue(err.text.includes('[ERROR] line 4:0 - line 4:20: Invalid entity line, entity name delete$Property cannot contain any of the following characters: [<, >, *, %, &, :, \\, $]'))
+                        done()
+                    })
+            })
+    })
   })
 
 describe('luis:convert new entity format', () => {

packages/lu/test/fixtures/testcases/bad5.lu

@@ -0,0 +1,4 @@
+# greeting
+- hi {@addto*Property=foo}
+
+@ ml delete$Property