fix: don't filter request-specific headers in the generator

This reverts commit 931516b.
apify · Jul 16, 2023 · 00ac285 · 00ac285
1 parent 90f592c
commit 00ac285
Show file tree

Hide file tree

Showing 2 changed files with 21 additions and 32 deletions.
diff --git a/packages/fingerprint-injector/src/fingerprint-injector.ts b/packages/fingerprint-injector/src/fingerprint-injector.ts
@@ -33,8 +33,24 @@ export class FingerprintInjector {
      * @returns Filtered headers.
      */
     private onlyInjectableHeaders(headers: Record<string, string>, browserName?: string): Record<string, string> {
+        const requestHeaders = [
+            'accept-encoding',
+            'accept',
+            'cache-control',
+            'pragma',
+            'sec-fetch-dest',
+            'sec-fetch-mode',
+            'sec-fetch-site',
+            'sec-fetch-user',
+            'upgrade-insecure-requests',
+        ];
+
         const filteredHeaders = { ...headers };
 
+        requestHeaders.forEach((header) => {
+            delete filteredHeaders[header];
+        });
+
         // Chromium-based controlled browsers do not support `te` header.
         // Probably needs more investigation, but for now, we can just remove it.
         if (!(browserName?.toLowerCase().includes('firefox') ?? false)) {

diff --git a/packages/header-generator/src/header-generator.ts b/packages/header-generator/src/header-generator.ts
@@ -284,44 +284,17 @@ export class HeaderGenerator {
             generatedSample[secFetchAttributeNames.dest] = 'document';
         }
 
+        for (const attribute of Object.keys(generatedSample)) {
+            if (attribute.startsWith('*') || generatedSample[attribute] === MISSING_VALUE_DATASET_TOKEN) delete generatedSample[attribute];
+        }
+
         // Order the headers in an order depending on the browser
         return this.orderHeaders({
-            ...this.filterClientHeaders(generatedSample),
+            ...generatedSample,
             ...requestDependentHeaders,
         }, (this.headersOrder as Record<string, any>)[generatedHttpAndBrowser.name]);
     }
 
-    /**
-     * Returns a new object that contains filtered headers from the input object. Filters out headers that are
-     * technically tied either with the individual browsers or individual requests and headers that have no value.
-     */
-    private filterClientHeaders(headers: Record<string, string>): Record<string, string> {
-        const clientHeaders = [
-            'accept-encoding',
-            'accept',
-            'cache-control',
-            'connection',
-            'pragma',
-            'sec-fetch-dest',
-            'sec-fetch-mode',
-            'sec-fetch-site',
-            'sec-fetch-user',
-            'upgrade-insecure-requests',
-        ];
-
-        const filteredHeaders = { ...headers };
-
-        return Object.fromEntries(
-            Object.entries(
-                filteredHeaders,
-            ).filter(([key, value]) => {
-                return !clientHeaders.includes(key.toLowerCase())
-                && value !== MISSING_VALUE_DATASET_TOKEN
-                && !key.startsWith('*');
-            }),
-        );
-    }
-
     /**
     * Returns a new object that contains ordered headers.
     * @param headers Specifies known values of headers dependent on the particular request.