rene/transceiver-db
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
34 Commits 5 Branches 9 Tags
Commit Graph

16 Commits

Author SHA1 Message Date
Rene Fichtmueller
4a53f3c45d feat: blog engine v5 — Auto-Kill Layer, 16-step pipeline, longer content 
Upgrades FO Blog Pipeline from 14 to 16 steps:
- NEW Step 8d: Auto-Kill Layer v1.0 (10 systematic categories A-J)
- NEW Step 15: Auto-Kill Scoring (cleanliness, narrative, non-AI, relevance)
- Updated banned phrases from Gold-standard editorial feedback
- Soft Delete List for conditional phrases
- Auto-Kill categories: spec blocks, formulas, section leakage,
  generic transitions, repeated concepts, SKU mentions, false authority,
  over-explained basics, whitepaper tone, fake precision

Content length changes per user feedback:
- Blog target: 1,200-2,000 words (was 700-1,000) — thorough and detailed
- LinkedIn target: 2,000-2,800 chars (was 350-600) — use maximum length
- Reduction pass: 25-30% cut (was 15-25%) — remove weak, keep depth
 2026-04-04 11:02:45 +02:00
Rene Fichtmueller
4a501d4461 feat: blog engine v3 — 8-stage pipeline with Auto-Kill Layer 
Complete rewrite of blog prompts and pipeline based on editorial
Gold-standard feedback. Replaces 3-pass system with 8-stage pipeline:

1. Master generation (narrative voice, no spec dumps)
2. Narrative Control (kill visible structure, enforce flow)
3. Auto-Kill Layer (remove AI phrases, spec residue, repetition)
4. Reduction Engine (cut 40% — keep strongest ideas only)
5. Depth pass (add specifics where vague, no spec dumps)
6. Quality Control (hard delete list validation)
7. Procurement layer (optional, sales audience)
8. LinkedIn post generation (new)

Key changes:
- System prompt rewritten with Hard Delete List (29 banned phrases)
- Soft Delete List for conditional phrases
- Auto-Kill categories A-J (spec blocks, formulas, whitepaper tone, etc.)
- Master prompts enforce continuous narrative, no section headings
- Word count targets reduced (800-1200 instead of 1500+)
- Scoring pass added (cleanliness, narrative, non-AI feel, relevance)
- LinkedIn companion post auto-generated
- Context data injection reduced (fewer items, no dump instructions)
 2026-04-04 10:52:31 +02:00
Rene Fichtmueller
5a0cbed5a2 feat: dashboard v2, blog expansion, market/cable MCP tools, switch asset scrapers, scraper utilities 2026-03-30 08:07:12 +02:00
Rene Fichtmueller
f940bf2cd4 fix: remove non-existent vendor URL columns, fix text=uuid cast in transceiver lookup 2026-03-30 07:49:54 +02:00
Rene Fichtmueller
891bd018a8 fix: add trust proxy for Cloudflare — fixes ERR_ERL_UNEXPECTED_X_FORWARDED_FOR in rate limiter 2026-03-30 06:41:36 +02:00
Rene Fichtmueller
4b452ab49e feat(scrapers+mcp): ATGBICS + ProLabs scrapers, MCP HTTP/SSE server 
Scrapers:
- atgbics.ts: PlaywrightCrawler for UK vendor ATGBICS (Shopify store),
  scrapes SFP/SFP+/SFP28/QSFP+/QSFP28/QSFP-DD in GBP, max 50 pages/run
- prolabs.ts: HttpCrawler for ProLabs (Legrand subsidiary), USD pricing,
  category-driven crawl with reach/fiber/speed detection
- Both registered in scheduler (every 8h, staggered) and index.ts CLI

MCP HTTP Server:
- packages/mcp-server/src/http-server.ts: Express + SSEServerTransport
- Exposes all 12 TIP tools via GET /sse + POST /message
- Bearer token auth (MCP_SECRET env), CORS-configurable
- GET /health → { status: "ok", tools: 12 }
- Port: MCP_HTTP_PORT (default 3201)

SQL + tools:
- sql/006-009: seed scripts for whitebox switches, vendors, assets
- switch-docs.ts: MCP tool for switch documentation queries
 2026-03-29 02:26:45 +08:00
Rene Fichtmueller
66b722a5e4 feat: calibrate regional adoption model with research-backed parameters 
Update REGIONAL_LAGS with data from LightCounting, vendor earnings,
OFC market sessions, and Chinese IPO prospectuses. Add price index
per region and segment mix (hyperscaler/telco/enterprise) for
more accurate regional revenue modeling.
 2026-03-28 02:34:29 +13:00
Rene Fichtmueller
c6308e93c0 feat: massive scraper expansion + hype cycle engine + lifecycle prediction 
New scrapers:
- GBICS.com (BigCommerce, GBP prices, 10 categories, 78 products)
- Juniper HCT (Next.js SSR parser, 475 transceivers with specs/EOL)
- SFPcables.com (Magento store, 16 categories, 78 products)
- Fluxlight (BigCommerce, 6 pages, 118 products)
- Champion ONE (compatible vendor scraper)

Scraper fixes:
- 10Gtek: rewritten to parse HTML spec tables (152 products)
- Flexoptix: fix price extraction from Magento Hyva HTML
- Register all scrapers in CLI (--gbics, --juniper, --sfpcables, etc.)

Hype Cycle Engine enhancements:
- Data-driven enrichment from scraped vendor/price data
- Revenue lifecycle prediction (peak year, decline, revenue index)
- Regional adoption model (NA, China, APAC, Europe, RoW with lag coefficients)
- New API endpoints: /enriched, /lifecycle, /regional/:tech

DB growth: 89 → 1,168 transceivers, 0 → 416 prices, 6 vendors
Qdrant: 1,162 products embedded with nomic-embed-text

Research: Norton-Bass model, standards-to-market timelines, hype signals
 2026-03-28 02:30:19 +13:00
Rene Fichtmueller
46af736db3 fix: hype cycle findTechnology matched wrong tech (1G instead of 1.6T) 
findTechnology used loose includes() matching — '1.6T OSFP-XD' matched
'1G SFP' first because query contained '1'. Now matches exact name first,
then by speed prefix with proper unit parsing (G/T).
 2026-03-28 01:00:52 +13:00
Rene Fichtmueller
1b0b602aa4 feat: Phase 8 — Dashboard frontend + static serving 
Single-file dashboard with 6 tabs: Overview, Semantic Search,
Hype Cycle, Transceivers, News, Blog Engine. Dark theme, no
build step, served as static HTML from Express.

- Overview: health stats, vector collection counts, recent news
- Semantic Search: query across all 6 Qdrant collections
- Hype Cycle: Norton-Bass table with phase colors + position bars
- Transceivers: searchable table with form factor/speed/reach
- News: semantic news search with source links
- Blog: generate drafts from templates, view draft history

Live at: https://transceiver-db.context-x.org/dashboard/
 2026-03-28 00:37:10 +13:00
Rene Fichtmueller
f48a809e40 feat: Phase 7 — Blog generator + scraper scheduler activation 
Blog draft engine generates structured markdown from all Qdrant
collections (products, news, FAQ, troubleshooting). Supports 4
topic types: hype_cycle, comparison, new_product, tutorial.

- routes/blog.ts: POST /api/blog/generate, GET/PUT endpoints
- ecosystem.config.js: Added tip-scraper PM2 process
- Scraper scheduler (pg-boss) now running on Erik with 8 job queues
- News scraper running every 6 hours on Erik
 2026-03-28 00:32:08 +13:00
Rene Fichtmueller
0a63307505 feat: Phase 6 — FAQ + troubleshooting knowledge base embeddings 
19 curated FAQ entries covering form factors, fiber types, reach,
compatibility, WDM, power, and emerging tech (CPO, LPO, 400ZR).
10 troubleshooting guides with symptom/cause/solution format.

All 6 Qdrant collections now populated:
- product_embeddings: 89 transceivers
- datasheet_chunks: 40 chunks (OCR pipeline)
- faq_embeddings: 19 FAQ entries
- troubleshooting_embeddings: 10 guides
- news_embeddings: 33 articles
- manual_chunks: 0 (pending manual ingestion)
 2026-03-28 00:24:50 +13:00
Rene Fichtmueller
122ca8444d feat: Phase 5 — OCR pipeline + document/news search 
Docling-powered OCR pipeline: PDF → markdown → chunks → Ollama embed → Qdrant.
News embedding seeder for news_embeddings collection.
Document and news semantic search API endpoints.

- embeddings/ocr-pipeline.ts: Docling convert → chunk → embed pipeline
- embeddings/seed-news.ts: Batch embed news_articles into Qdrant
- routes/documents.ts: POST /api/documents/process, GET /api/documents
- routes/search.ts: GET /search/documents, GET /search/news endpoints
- sql/005-documents.sql: Add chunks_count, processed_at to documents table
- Ollama + nomic-embed-text installed on Erik (CPU mode)
- 89 products + 40 datasheet chunks + 33 news articles in Qdrant
 2026-03-28 00:22:01 +13:00
Rene Fichtmueller
0260d0b365 feat: Phase 4 — Vector embeddings + semantic search 
Ollama nomic-embed-text (768 dim) → Qdrant vector search pipeline.
Embeds all 89 transceivers with rich text representation and payload
filters (form_factor, speed_gbps, fiber_type, wdm_type).

- embeddings/client.ts: Ollama embed + Qdrant upsert/search
- embeddings/seed-products.ts: Batch seeder for product_embeddings
- routes/search.ts: GET /api/search, /search/products, /search/stats
- 6 Qdrant collections: products, datasheets, FAQs, manuals, troubleshooting, news
 2026-03-28 00:05:29 +13:00
Rene Fichtmueller
a6f7968393 feat: Phase 3 — Norton-Bass Hype Cycle Engine 
Implements the full Norton-Bass Multigenerational Diffusion Model for
transceiver technology lifecycle forecasting.

Math: Bass diffusion F(t) + logistic adoption S(t) = L / (1 + e^(-k(t-t0)))
Parameters: p (innovation ~0.03), q (imitation ~0.3-0.5), m (market potential)

Phase Classification Engine (composite score):
  30% Port shipment share + 20% ASP decline rate + 15% Standards maturity
  + 15% Interop validation + 10% Vendor trajectory + 10% Media sentiment

11 technologies tracked: 1G → 10G → 25G → 40G → 100G → 400G → 800G → 1.6T
  + CPO, LPO, 400ZR Coherent
5-year adoption forecast per technology

API: GET /api/hype-cycle (all) + GET /api/hype-cycle/:tech (detail)
Live: https://transceiver-db.context-x.org/api/hype-cycle
 2026-03-27 23:35:57 +13:00
Rene Fichtmueller
b43bdd3060 feat: TIP Phase 0+1 — monorepo, DB schema, API, scraper engine 
Phase 0 - Foundation:
- Restructure into npm workspace monorepo (packages/core, api, scraper)
- PostgreSQL 17 + TimescaleDB schema (15 tables incl. hypertables)
- Docker Compose for local dev (PostgreSQL on 5433 + Qdrant)
- Express 5 API on port 3200 with 6 routes
- Seed script to migrate 159 transceivers + 42 standards from npm package
- Erik server setup script + PM2 ecosystem config

Phase 1 - Scraper Engine:
- Crawlee + Playwright framework with pg-boss scheduler
- FS.com scraper (PlaywrightCrawler, anti-bot workaround)
- Optcore.net scraper (WP REST API enumeration + PlaywrightCrawler)
  - Uses /wp-json/wp/v2/product to get 2000+ product URLs
  - Playwright renders individual product pages for price extraction
- Cisco TMG Matrix scraper (compatibility data)
- News RSS aggregator (optics.org, SPIE, Network World, Nature Photonics)
  - Keyword relevance scoring for transceiver/fiber topics
  - xml2js with malformed XML sanitization
- SHA-256 content hashing for change detection (skip unchanged records)
- pg-boss v10 with explicit queue creation before scheduling
 2026-03-27 16:27:31 +13:00