Support non-ascii (unicode) characters in service name, operation name and parameter name.

This replaces regexp patterns that only worked with ascii characters with more proper matching that supports unicode identifiers in typescript/javascript.

The platform must support "unicode-aware mode" (the u flag) for this to work. https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/RegExp/unicode

Author: josstn

src/openApi/v2/parser/getOperationName.ts

@@ -1,18 +1,15 @@
 import camelCase from 'camelcase';
 
+import sanitizeOperationName from '../../../utils/sanitizeOperationName';
+
 /**
  * Convert the input value to a correct operation (method) classname.
  * This will use the operation ID - if available - and otherwise fallback
  * on a generated name from the URL
  */
 export const getOperationName = (url: string, method: string, operationId?: string): string => {
     if (operationId) {
-        return camelCase(
-            operationId
-                .replace(/^[^a-zA-Z]+/g, '')
-                .replace(/[^\w\-]+/g, '-')
-                .trim()
-        );
+        return camelCase(sanitizeOperationName(operationId).trim());
     }
 
     const urlWithoutPlaceholders = url

src/openApi/v2/parser/getOperationParameterName.ts

@@ -1,15 +1,13 @@
 import camelCase from 'camelcase';
 
 import { reservedWords } from '../../../utils/reservedWords';
+import sanitizeOperationParameterName from '../../../utils/sanitizeOperationParameterName';
 
 /**
  * Replaces any invalid characters from a parameter name.
  * For example: 'filter.someProperty' becomes 'filterSomeProperty'.
  */
 export const getOperationParameterName = (value: string): string => {
-    const clean = value
-        .replace(/^[^a-zA-Z]+/g, '')
-        .replace(/[^\w\-]+/g, '-')
-        .trim();
+    const clean = sanitizeOperationParameterName(value).trim();
     return camelCase(clean).replace(reservedWords, '_$1');
 };

src/openApi/v2/parser/getServiceName.ts

@@ -1,13 +1,12 @@
 import camelCase from 'camelcase';
 
+import sanitizeServiceName from '../../../utils/sanitizeServiceName';
+
 /**
  * Convert the input value to a correct service name. This converts
  * the input string to PascalCase.
  */
 export const getServiceName = (value: string): string => {
-    const clean = value
-        .replace(/^[^a-zA-Z]+/g, '')
-        .replace(/[^\w\-]+/g, '-')
-        .trim();
+    const clean = sanitizeServiceName(value).trim();
     return camelCase(clean, { pascalCase: true });
 };

src/openApi/v3/parser/getOperationName.ts

@@ -1,18 +1,15 @@
 import camelCase from 'camelcase';
 
+import sanitizeOperationName from '../../../utils/sanitizeOperationName';
+
 /**
  * Convert the input value to a correct operation (method) classname.
  * This will use the operation ID - if available - and otherwise fallback
  * on a generated name from the URL
  */
 export const getOperationName = (url: string, method: string, operationId?: string): string => {
     if (operationId) {
-        return camelCase(
-            operationId
-                .replace(/^[^a-zA-Z]+/g, '')
-                .replace(/[^\w\-]+/g, '-')
-                .trim()
-        );
+        return camelCase(sanitizeOperationName(operationId).trim());
     }
 
     const urlWithoutPlaceholders = url

src/openApi/v3/parser/getOperationParameterName.ts

@@ -1,16 +1,13 @@
 import camelCase from 'camelcase';
 
 import { reservedWords } from '../../../utils/reservedWords';
+import sanitizeOperationParameterName from '../../../utils/sanitizeOperationParameterName';
 
 /**
  * Replaces any invalid characters from a parameter name.
  * For example: 'filter.someProperty' becomes 'filterSomeProperty'.
  */
 export const getOperationParameterName = (value: string): string => {
-    const clean = value
-        .replace(/^[^a-zA-Z]+/g, '')
-        .replace('[]', 'Array')
-        .replace(/[^\w\-]+/g, '-')
-        .trim();
+    const clean = sanitizeOperationParameterName(value).trim();
     return camelCase(clean).replace(reservedWords, '_$1');
 };

src/openApi/v3/parser/getServiceName.spec.ts

@@ -9,5 +9,6 @@ describe('getServiceName', () => {
         expect(getServiceName('@fooBar')).toEqual('FooBar');
         expect(getServiceName('$fooBar')).toEqual('FooBar');
         expect(getServiceName('123fooBar')).toEqual('FooBar');
+        expect(getServiceName('non-ascii-æøåÆØÅöôêÊ字符串')).toEqual('NonAsciiÆøåÆøÅöôêÊ字符串');
     });
 });

src/openApi/v3/parser/getServiceName.ts

@@ -1,13 +1,12 @@
 import camelCase from 'camelcase';
 
+import sanitizeServiceName from '../../../utils/sanitizeServiceName';
+
 /**
  * Convert the input value to a correct service name. This converts
  * the input string to PascalCase.
  */
 export const getServiceName = (value: string): string => {
-    const clean = value
-        .replace(/^[^a-zA-Z]+/g, '')
-        .replace(/[^\w\-]+/g, '-')
-        .trim();
+    const clean = sanitizeServiceName(value).trim();
     return camelCase(clean, { pascalCase: true });
 };

src/utils/sanitizeOperationName.ts

@@ -0,0 +1,7 @@
+import sanitizeServiceName from './sanitizeServiceName';
+
+/**
+ * sanitizeOperationName does the same as sanitizeServiceName.
+ */
+const sanitizeOperationName = sanitizeServiceName;
+export default sanitizeOperationName;

src/utils/sanitizeOperationParameterName.ts

@@ -0,0 +1,7 @@
+import sanitizeOperationName from './sanitizeOperationName';
+
+const sanitizeOperationParameterName = (name: string): string => {
+    const withoutBrackets = name.replace('[]', 'Array');
+    return sanitizeOperationName(withoutBrackets);
+};
+export default sanitizeOperationParameterName;

src/utils/sanitizeServiceName.ts

@@ -0,0 +1,18 @@
+/**
+ * Sanitizes service names, so they are valid typescript identifiers of a certain form.
+ *
+ * 1: Remove any leading characters that are illegal as starting character of a typescript identifier.
+ * 2: Replace illegal characters in remaining part of type name with underscore (-).
+ *
+ * Step 1 should perhaps instead also replace illegal characters with underscore, or prefix with it, like sanitizeEnumName
+ * does. The way this is now one could perhaps end up removing all characters, if all are illegal start characters. It
+ * would be sort of a breaking change to do so, though, previously generated code might change then.
+ *
+ * Javascript identifier regexp pattern retrieved from https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Lexical_grammar#identifiers
+ *
+ * The output of this is expected to be converted to PascalCase
+ */
+const sanitizeServiceName = (name: string) =>
+    name.replace(/^[^\p{ID_Start}]+/u, '').replace(/[^$\u200c\u200d\p{ID_Continue}]/gu, '-');
+
+export default sanitizeServiceName;