bom processing for native automation (#2906)

* bom processing for native automation

* add server test

src/processing/resources/page.ts

@@ -5,7 +5,7 @@ import DomProcessor from '../dom';
 import DomAdapter from '../dom/parse5-dom-adapter';
 import ResourceProcessorBase from './resource-processor-base';
 import * as parse5Utils from '../../utils/parse5';
-import getBOM from '../../utils/get-bom';
+import { getBOM, getBOMDecoded } from '../../utils/get-bom';
 import getStorageKey from '../../utils/get-storage-key';
 import SELF_REMOVING_SCRIPTS from '../../utils/self-removing-scripts';
 import RequestPipelineContext from '../../request-pipeline/context';
@@ -275,6 +275,10 @@ class PageProcessor extends ResourceProcessorBase {
 
     // NOTE: API for new implementation without request pipeline
     injectResources (html: string, resources: PageInjectableResources, options?: PageRestoreStoragesOptions): string {
+        const bom = getBOMDecoded(html);
+
+        html = bom ? html.replace(bom, '') : html;
+
         const root     = parse5.parse(html);
         const elements = parse5Utils.findElementsByTagNames(root, ['head', 'body']);
         const head     = elements.head[0];
@@ -283,7 +287,7 @@ class PageProcessor extends ResourceProcessorBase {
         PageProcessor._addPageResources(head, resources, options);
         PageProcessor._addBodyCreatedEventScript(body);
 
-        return parse5.serialize(root);
+        return (bom || '') + parse5.serialize(root);
     }
 }
 

src/processing/script/index.ts

@@ -9,7 +9,7 @@ import { add as addHeader, remove as removeHeader } from './header';
 import { parse } from 'acorn-hammerhead';
 import { generate, Syntax } from 'esotope-hammerhead';
 import reEscape from '../../utils/regexp-escape';
-import getBOM from '../../utils/get-bom';
+import { getBOM } from '../../utils/get-bom';
 
 const HTML_COMMENT_RE       = /(^|\n)\s*<!--[^\n]*(\n|$)/g;
 const OBJECT_RE             = /^\s*\{.*\}\s*$/;

src/utils/get-bom.ts

@@ -4,10 +4,17 @@
 // -------------------------------------------------------------
 /* eslint hammerhead/proto-methods: 2 */
 
-const BOM_RE = /^(\xEF\xBB\xBF|\xFE\xFF|\xFF\xFE|\x00\x00\xFE\xFF|\xFF\xFE\x00\x00|\x2B\x2F\x76\x38|\x2B\x2F\x76\x39|\x2B\x2F\x76\x2B|\x2B\x2F\x76\x2F|\xF7\x64\x4C|\xDD\x73\x66\x73|\x0E\xFE\xFF|\xFB\xEE\x28|\x84\x31\x95\x33)/; // eslint-disable-line no-control-regex
+const BOM_RE         = /^(\xEF\xBB\xBF|\xFE\xFF|\xFF\xFE|\x00\x00\xFE\xFF|\xFF\xFE\x00\x00|\x2B\x2F\x76\x38|\x2B\x2F\x76\x39|\x2B\x2F\x76\x2B|\x2B\x2F\x76\x2F|\xF7\x64\x4C|\xDD\x73\x66\x73|\x0E\xFE\xFF|\xFB\xEE\x28|\x84\x31\x95\x33)/; // eslint-disable-line no-control-regex
+const BOM_DECODED_RE = /^\uFEFF/;
 
-export default function (str: string): string | null {
+export function getBOM (str: string): string | null {
     const match = str.match(BOM_RE);
 
     return match ? match[0] : null;
 }
+
+export function getBOMDecoded (str: string): string | null {
+    const match = str.match(BOM_DECODED_RE);
+
+    return match ? match[0] : null;
+}

test/server/proxy/without-request-pipeline-test.js

@@ -3,18 +3,27 @@ const { injectResources } = require('../../../lib');
 const { compareCode }     = require('../common/utils');
 
 describe('New API', () => {
+    const resources = {
+        stylesheets:     ['./styles.css'],
+        scripts:         ['common/script1.js', './common/script2.js'],
+        embeddedScripts: ['var script1 = 1;', 'var script2 = 2;'],
+        userScripts:     ['/custom-script1.js', '/custom-script2.js'],
+    };
+
+    const pageContent         = fs.readFileSync('test/server/data/without-request-pipeline/src.html').toString();
+    const expectedPageContent = fs.readFileSync('test/server/data/without-request-pipeline/expected.html').toString();
+
     it('PageProcessor.injectResources', () => {
-        const pageContent = fs.readFileSync('test/server/data/without-request-pipeline/src.html').toString();
+        const updatedPageContent = injectResources(pageContent, resources);
+
+        compareCode(updatedPageContent, expectedPageContent);
+    });
 
-        const resources = {
-            stylesheets:     ['./styles.css'],
-            scripts:         ['common/script1.js', './common/script2.js'],
-            embeddedScripts: ['var script1 = 1;', 'var script2 = 2;'],
-            userScripts:     ['/custom-script1.js', '/custom-script2.js'],
-        };
+    it('PageProcessor.injectResources - trim BOM', () => {
+        const bomSymbol          = String.fromCharCode(65279);
+        const updatedPageContent = injectResources(bomSymbol + pageContent, resources);
 
-        const updatedPageContent  = injectResources(pageContent, resources);
-        const expectedPageContent = fs.readFileSync('test/server/data/without-request-pipeline/expected.html').toString();
+        updatedPageContent.charCodeAt(0) === bomSymbol;
 
         compareCode(updatedPageContent, expectedPageContent);
     });