Initial Commit

playwright/scrape-messages.spec.js

@@ -61,6 +61,18 @@ function extractShare(node) {
     };
 }
 
+// A raw photo/video message has content_type 'IMAGES' with one or more
+// attachments (Instagram allows sending several photos in a single message).
+function extractAttachments(node) {
+    const attachments = node.content?.attachments;
+    if (!attachments?.length) return null;
+    return attachments.map((a) => ({
+        url: a.attachment_cdn_url || a.preview_cdn_url || null,
+        width: a.preview_width || null,
+        height: a.preview_height || null,
+    })).filter((a) => a.url);
+}
+
 function extractMessage(node, thread, fbidLookup) {
     const senderDict = node.sender?.user_dict;
     const resolved = senderDict
@@ -85,6 +97,7 @@ function extractMessage(node, thread, fbidLookup) {
         text: node.text_body || node.content?.text_body || null,
         timestamp_ms: node.timestamp_ms ? Number(node.timestamp_ms) : null,
         share: extractShare(node),
+        attachments: extractAttachments(node),
         reactions: reactions.length ? reactions : null,
         replied_to_id: node.replied_to_message_id || null,
         replied_to_text: node.replied_to_message?.text_body || null,