facebook: try to pull out post types from HTML content

lockdown-systems · Feb 21, 2025 · 0b56d2f · 0b56d2f
1 parent ab3cf81
commit 0b56d2f
Showing 1 changed file with 30 additions and 10 deletions.
diff --git a/src/account_facebook/facebook_account_controller.ts b/src/account_facebook/facebook_account_controller.ts
@@ -512,14 +512,11 @@ export class FacebookAccountController {
                     const postElements = dom.window.document.querySelectorAll('._a6-g');
 
                     for (const postElement of postElements) {
-                        // Status updates have exactly one text div inside _2pin
-                        const contentDiv = postElement.querySelector('._2pin');
-                        const directTextDiv = contentDiv?.querySelector(':scope > div');
-                        const hasComplexStructure = directTextDiv?.querySelector('div');
-
-                        if (hasComplexStructure) {
-                            log.info("FacebookAccountController.importFacebookArchive: skipping post with complex nested structure");
-                            continue; // Skip posts with complex nested structure (shared posts, groups, etc)
+                        const postType = getPostType(postElement);
+
+                        if (postType === 'shared_group') {
+                            log.info("FacebookAccountController.importFacebookArchive: skipping group posts");
+                            continue;
                         }
 
                         const titleElement = postElement.querySelector('._a6-h');
@@ -536,7 +533,7 @@ export class FacebookAccountController {
                                 title: titleElement.textContent || '',
                                 full_text: contentElement.textContent || '',
                                 created_at: dateElement.textContent || '',
-                                isReposted: false,
+                                isReposted: postType === 'shared_post',
                             });
                         }
                     };
@@ -608,4 +605,27 @@ export class FacebookAccountController {
             skipCount: skipCount,
         };
     }
-}
+}
+
+const getPostType = (element: Element): 'status' | 'shared_post' | 'shared_group' | 'other' => {
+    const pinDivs = element.querySelectorAll('._2pin');
+
+    if (pinDivs.length === 1) {
+        return 'status';
+    }
+
+    if (pinDivs.length === 2) {
+        // Check for group name structure
+        const firstPinContent = pinDivs[0].textContent?.trim();
+        if (firstPinContent && !firstPinContent.includes('div')) {
+            return 'shared_group';
+        }
+        // Shared posts have empty nested divs
+        const emptyDivs = pinDivs[0].querySelectorAll('div div div div');
+        if (emptyDivs.length > 0) {
+            return 'shared_post';
+        }
+    }
+
+    return 'other';
+};