fix(instagram-api): env reprompt + runtime wedge hardening + post-login challenge#72
Open
tnunamak wants to merge 3 commits intofix/instagram-api-compatfrom
Open
fix(instagram-api): env reprompt + runtime wedge hardening + post-login challenge#72tnunamak wants to merge 3 commits intofix/instagram-api-compatfrom
tnunamak wants to merge 3 commits intofix/instagram-api-compatfrom
Conversation
…ials When env-provided credentials (PLATFORM_LOGIN / PLATFORM_PASSWORD) fail with invalid_credentials, fall through to the existing promptWithRetry loop instead of throwing immediately. Matches the no-env path behavior and supports the bad-password canary mode in context-gateway. Other error kinds (login_api_error, network, etc.) still throw as before. Real users who fat-finger their password now re-prompt instead of seeing a generic failure. Ported from an ad-hoc snapshot edit in context-gateway PR #185.
…nge handling
Bundles the improvements from context-gateway PR #186 into the upstream
connector. Aimed at making instagram × full mode viable in canary and
improving real-user UX when Instagram flags the login.
Changes, roughly in order of impact:
1. Mid-run setData pressure reduction. setAuthState and
setCollectorTraceSection now accumulate in-memory; terminal flush
in main IIFE only. ~93 page.setData round-trips per run → ~26.
User-visible 'status' breadcrumbs unchanged.
2. handlePostLoginChallenge(): after OTP, if session lands on
/challenge/ or /checkpoint/, programmatically walk the page:
- Click 'This was me' / 'Confirm it's you' / 'It was me'.
- Click 'Dismiss' / 'Continue' / 'Next' (for 'automated behavior'
and 'help us confirm it's you' pages respectively).
- Email-code prompt with 'email' keyword so canary email worker
polls and supplies.
- Full event dispatch (pointer/mouse/click) for React onClick
handlers that ignore bare element.click().
3. Accounts Center HTML shrinking. fetchAccountsCenterHtml previously
returned document.documentElement.outerHTML (multi-MB) across the
remote-browser boundary. Now projects inside the page: extracts
data-sjs script bodies containing fxcal_settings and a small meta
slice; reconstructs minimal HTML for existing downstream parsers.
4. readCapturedGraphql(matchSubstring) — pre-filters inside the page
and caps body/response length. Previously returned the full
__igAllGraphql__ array every 250ms during adCategories discovery.
5. Runtime-wedge circuit breaker. Module-scope runtimeWedgeDetected
flag set when any caught error matches /Command timeout:/. Used
to skip remaining ads sub-phases and optional terminal breadcrumbs;
cuts wedged-run duration from ~5min to ~2min.
Still limited by the Accounts Center pages consistently wedging the
remote browser runtime after 1-2 navigations — a remote-browser-
service capacity issue that needs infra-layer work.
tnunamak
changed the title
Ports additional improvements from context-gateway side that went in after the initial port on this branch. Matches the code merged in: - CG #187 (profile capture wedge check): short-circuit profile-capture polling loop on runtime wedge. - CG #188 (XHR Accounts Center): fetch data-sjs HTML blocks via same- origin XHR from an instagram.com anchor page instead of navigating the remote browser to each heavy Accounts Center page. Verified on staging parallel full-mode runs: acct 3 got first advertiser via XHR-fetched ads page; acct 1 consistently collects profile + posts + followers + following + advertisers with this code.
Member
Author
|
Updated with additional improvements from CG #187 + #188:
Verified across multiple staging parallel full-mode runs. Not 6/6 green (accounts flagged with /challenge/ 'Help us confirm it's you' can't be automated through; categories phase still wedges because it needs UI tab click for GraphQL discovery), but consistently 3-4/6 partial outcomes with real data collection — up from 0/6 before. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Bundles two rounds of hardening for the API connector, both validated against staging canary in `vana-com/context-gateway#186`.
Round 1 (original scope): env-credential reprompt
When env-provided credentials fail with invalid_credentials, fall through to promptWithRetry instead of throwing. Supports bad-password canary mode + improves real-user UX. Already verified green:
Round 2: runtime wedge + post-login challenge
Additional changes to make `instagram × full` viable in canary and handle Instagram's post-OTP challenge pages without a headed fallback:
Verification
`instagram × full` (rotation) on context-gateway#186 got past login + challenge + profile + posts + followers + following into the Accounts Center phase, where the remote browser runtime wedges regardless of payload shrinks. That's a separate infra-layer concern.
Merge order
Targeting `fix/instagram-api-compat` (PR #70). Once #70 lands and this is included, CG can repin to canonical main ref instead of the branch index.