Upgrade to monetizacion.j-lab.tools: real AI chat, deeper analysis, J-Lab design#18
Open
luisassardo wants to merge 2 commits into
Open
Upgrade to monetizacion.j-lab.tools: real AI chat, deeper analysis, J-Lab design#18luisassardo wants to merge 2 commits into
luisassardo wants to merge 2 commits into
Conversation
…sign - Backend: replace the fake keyword "AI" with a real Claude tool-use chat (/api/ai/chat) grounded in the dataset via safe predefined query functions; graceful fallback to the local keyword engine when no ANTHROPIC_API_KEY or on any error. Refactor process_query() into reusable tool functions + registry. - Add 5 analysis endpoints for reporting: coordinated-bursts, program-migration, verified-gap, country-asymmetry, growth-outliers (leads, not verdicts). - Importer scripts/import_archive.py: normalize the archive's raw Export CSV to the 17-field schema, filter to Latin America by ISO-2 code, clean casing/encoding/languages/ghost-dates, dedupe. - Design: vendor the J-Lab tokens.css and restyle index/account/widget; recolor Chart.js from CSS vars; chat panel modeled on the J-Lab Chat component; EN/ES toggle + dark/light theme; server-side i18n. - Branding/domain: J-Lab tool with Vector Critico + WHAT TO FIX / CC BY-ND 4.0 attribution; update render.yaml (service + Anthropic env), .env.example, README. - Scope: Latin America (was Central America only). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
![github-advanced-security[bot]](https://avatars.githubusercontent.com/in/57789?s=60&v=4){kind=small}
| lang = resolve_lang() | ||
| resp = make_response(render_template(template, lang=lang, t=I18N[lang], **ctx)) | ||
| if request.args.get('lang') in ('es', 'en'): | ||
| resp.set_cookie('lang', lang, max_age=60 * 60 * 24 * 365, samesite='Lax') |
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 9b0228c57c
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| if any(w in q for w in ['top', 'mejores', 'mayores', 'más suscriptores', 'mas suscriptores', 'largest', 'biggest']): | ||
| nums = re.findall(r'\d+', q) | ||
| n = min(int(nums[0]), 50) if nums else 10 | ||
| return render_tool_result_md('top_accounts', tool_top_accounts(n), lang) |
There was a problem hiding this comment.
When the new /api/ai/chat path falls back to the local engine (no Anthropic key, rate limit, or API error), a country-scoped top query such as the page placeholder “¿Cuáles son las 10 cuentas mayores en Guatemala?” enters this branch but calls tool_top_accounts(n) without the country. With the current CSV, that returns the global top accounts (including El Salvador/Costa Rica rows) instead of Guatemala-only results, so the promised graceful fallback gives incorrect grounded answers whenever AI is unavailable.
Useful? React with 👍 / 👎.
…port
- Reverse-engineered the archive's CSV export endpoint (csv_data_table) and
refreshed data/accounts.csv to all 22 Latin American countries: 63,846
Facebook accounts, fresh through 2026-04-25 (was a ~3K Central America
snapshot through Jan 2026).
- scripts/import_archive.py gains a `--fetch` mode that pulls every LatAm
country live from that endpoint and normalizes in one command:
python scripts/import_archive.py --fetch --output data/accounts.csv
(manual --input export path still supported). Normalizes long ISO country
names to clean short names.
- Fix: load_data() now fills blank string cells (name/handle/language/location)
so empty names don't serialize as NaN / invalid JSON in the larger dataset.
- README documents --fetch as the primary refresh path.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
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
2 participants
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.
Rebuilds the Meta monetization explorer as a J-Lab tool ahead of the move to monetizacion.j-lab.tools.
What changed
Real AI chat (was fake)
The old "AI" was keyword matching. Now
POST /api/ai/chatuses Claude with tool-use over safe predefined query functions (top accounts, by country/year/program/language, search, summary). Claude picks tools and composes a grounded answer in the user's language — no arbitrary code/SQL, no hallucinated numbers.ANTHROPIC_API_KEY; default modelclaude-haiku-4-5(J-Lab standard).enginefield tells the UI which path answered)./api/querykept unchanged.Deeper analysis for reporting (
/api/analysis/*)coordinated-bursts(same-day/same-country creation),program-migration(cohort program shift),verified-gap(unverified accounts pulling revenue),country-asymmetry(normalized per-country),growth-outliers(young pages, big audiences). Surfaced as leads, not verdicts.Data: Latin America + importer
Scope widened from Central America to all Latin America. New
scripts/import_archive.pynormalizes the archive's raw "Export CSV" (no public API) → the 17-field schema: tolerant of column-casing/UTF-8/escaping quirks, maps messy languages, nulls ghost/epoch dates, filters to LatAm by ISO-2 code, dedupes.--dry-runprints a kept/dropped summary.Design → J-Lab
Vendored
static/css/tokens.css(Inter Tight / Instrument Serif / JetBrains Mono, dark default + light mode). Restyled all three templates; Chart.js recolored from CSS vars; chat panel modeled on the J-Lab Chat component; EN/ES toggle + theme toggle; server-side i18n (Spanish default). Keeps monetization.wtf / WHAT TO FIX / CC BY-ND 4.0 + Vector Crítico attribution.Config
render.yaml(servicemonetizacion-jlab, Anthropic env,ANTHROPIC_API_KEYas a dashboard secret),.env.example,requirements.txt(+anthropic), README rewrite.Verification
engine: local).Follow-ups for deploy
ANTHROPIC_API_KEYas a Render secret to enable the Claude path (works without it via fallback).monetizacion.j-lab.toolsDNS at the service.scripts/import_archive.pyto refreshdata/accounts.csv.🤖 Generated with Claude Code