Initial Commit

playwright/debugger.spec.js

@@ -1,11 +1,12 @@
 import { test } from '@playwright/test';
+import path from 'path';
 
-test.use({ storageState: '/home/pwuser/ig-session.json' });
+test.use({ storageState: path.resolve(__dirname, '.auth', 'ig-session.json') });
 
-test('inspect non-text message shapes', async ({ page }) => {
+test('inspect IMAGES message shape', async ({ page }) => {
     test.setTimeout(120_000);
 
-    const interesting = [];
+    const found = [];
 
     page.on('response', async (response) => {
         const url = response.url();
@@ -17,10 +18,8 @@ test('inspect non-text message shapes', async ({ page }) => {
             const t = json?.data?.get_slide_thread_nullable?.as_ig_direct_thread;
             if (!t) return;
             for (const edge of t.slide_messages?.edges || []) {
-                const node = edge.node;
-                const hasReactions = (node.reactions?.length || 0) > 0 || (node.msg_reactions?.length || 0) > 0;
-                if (node.content_type !== 'TEXT' || hasReactions || node.replied_to_message) {
-                    interesting.push({ thread: t.thread_title, node });
+                if (edge.node?.content_type === 'IMAGES') {
+                    found.push(edge.node);
                 }
             }
         } catch (e) { /* skip */ }
@@ -29,13 +28,14 @@ test('inspect non-text message shapes', async ({ page }) => {
     await page.goto('https://www.instagram.com/direct/inbox/');
     await page.waitForTimeout(6000);
 
-    const threadHrefs = await page.$$eval('a[href*="/direct/t/"]', els =>
-        [...new Set(els.map(el => el.getAttribute('href')))]);
+    const threadHrefs = await page.$$eval('a[href*="/direct/t/"]', (els) => [
+        ...new Set(els.map((el) => el.getAttribute('href'))),
+    ]);
 
     for (const href of threadHrefs) {
         await page.goto(`https://www.instagram.com${href}`);
         await page.waitForTimeout(3500);
     }
 
-    console.log('[INTERESTING MESSAGES]', JSON.stringify(interesting, null, 2));
-});
+    console.log('[IMAGES MESSAGES]', JSON.stringify(found, null, 2));
+});

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,