justin
0e625553e5
CORPUS — 4,299 GH plot reports added (3,797 written + 502 from the
earlier slow run + 319 sitemap-listed URLs that 404'd as
discontinued). Combined with prior 760 varieties + 14 AgriPro
trials = 5,073 total chunks now indexed.
scrape/sources/gh_plot_reports.py — concurrency speedup:
- 4 worker threads (ThreadPoolExecutor), each with its own
requests.Session for connection-pool efficiency.
- Shared class-level rate limiter (0.25 sec between ANY two
requests across all threads). Net throughput ~4 req/sec —
well below any rate-limit threshold a public site enforces.
- Diagnosis vs original 1 req/sec: GH had ZERO rate limiting,
zero 429s, zero retries. The 1 sec self-throttle was just too
conservative. Bench:
1 worker / 1.0 sec throttle: ~0.4 plots/sec (190 min ETA)
4 workers / 0.25 sec throttle: ~3 plots/sec (~25 min actual)
rag/chunk.py — chunk size cap for nomic-embed-text's 2048-token
context window:
- Empirically tested: failure threshold is ~5,250 chars on
numeric-heavy trial chunks (chars/token ratio 2.4 vs 3.5 for
prose). Cap at 4,500 chars to be safely under at worst-case
2.2 chars/token.
- Applied to BOTH variety and trial chunks. Marked truncated
chunks with metadata.embed_truncated = True; FULL text stays
in the on-disk .md for get_page to return verbatim.
.gitea/workflows/{refresh,image-only}.yml — OLLAMA_URL pool
restructured for the 4 GPU-pinned endpoints. Bench (50-chunk
batches on nomic-embed-text):
.0.125:11434 (RTX 40-series) 242 embeds/sec ← weight ×4
.0.2:11436 (GPU-pinned) 108 embeds/sec ← weight ×2
.0.2:11435 (GPU-pinned) 72 embeds/sec ← weight ×1
localhost (TITAN X) 37 embeds/sec ← weight ×1
Weighting is done by listing the URL multiple times in
OLLAMA_URL since the embedder uses round-robin. .0.2:11434 is
explicitly EXCLUDED — it isn't pinned to a specific GPU.
Combined index rebuild for 5,073 chunks now finishes in ~3 min
(was 19+ on the single-endpoint pool).
Smoke tests:
✓ list_versions: 5,073 docs across 6 sources, 2 vendors, 6
brands, 4 crops (corn 2711, soy 2016, silage 223, wheat 123).
✓ search_trials({crop=corn, state=IA, year=2024}): 3 IA 2024
corn trials surfaced.
✓ search_trials("Phytophthora resistance soybean trial"): NK
NK43-W1XFS top-1 in LA 2024 trial (cross-vendor result).
✓ search_trials("AP Iliad Idaho wheat"): AgriPro Washington/N
Idaho 2025 trial surfaced.
✓ search_trials(product=DKC65-95): 3 corn trials containing
that hybrid in IL/IA 2024.
✓ search_trials(product=NK1701): 3 corn trials in AR/MS 2024.
✓ Product filter correctly returns EMPTY for products that
aren't in the corpus (DKC65-20 is a 2023 product; 2023 plots
deferred). Anti-hallucination contract preserved.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
scrape/
Per-vendor seed catalog scrapers + the runner that dispatches to
them. Each source lives in scrape/sources/<name>.py with a main()
entrypoint. The runner is a thin shim:
python -m scrape.runner --source bayer_seeds --force
python -m scrape.runner --source golden_harvest --limit 20
python -m scrape.runner --all # only GREEN sources
Output layout
Each scraper writes:
corpus/<source>/<source_key>.md— LLM-visible body (chunk_0 preamble + the variety's marketing + agronomic narrative)corpus/<source>/<source_key>.json— sidecar metadata (per CLAUDE.md's canonical schema)
source_key is a stable per-vendor slug — typically <brand>-<sku>
lowercased, e.g. dekalb-dkc62-08rib. Stability matters: it's the
join key the MCP uses for get_page(source, source_key).
Sources
| Source | Module | Verdict | Notes |
|---|---|---|---|
bayer_seeds |
bayer_seeds.py |
🟢 | DEKALB + Asgrow + WestBred, ~475 varieties |
golden_harvest |
golden_harvest.py |
🟢 | ~175 varieties, 9-to-1 disease scale (reverse) |
nk |
nk.py |
🟢 | 29 varieties, ratings in CDN PDFs |
agripro |
agripro.py |
🟢 | 24 wheat varieties |
becks_pfr |
becks_pfr.py |
🟡 | 2,089 research docs via public Sanity GROQ |
becks_products |
becks_products.py |
🟡 | 860 products, identity-only (SeedIQ-gated) |
Pioneer is intentionally absent — see CLAUDE.md and the curated
Pioneer fallback in docs_mcp/lessons.md.
Tips
- Sniff before you scrape. Most catalogs are SPAs that call a
backend API. The recon docs in
~/.claude/projects/-home-justin/ memory/reference_seed_vendor_recon.mdalready capture the endpoints; if you find new ones, update that file. - Idempotent re-scrapes. Without
--force, skip pages already on disk. With--force, re-fetch everything — that's the monthly cron mode. - Respect the portals. Backoff on 429s. Set a recognizable
user-agent (
seed-mcp-scraper/<version>). - Normalize at chunk time, not at scrape time. The chunker
(Phase 2) handles the 9-to-1 → 1-9 disease-scale flip for Golden
Harvest, NOT this scraper. Sidecar JSON should preserve the
vendor's raw values + a
_scale_directionfield; the chunker reads that and normalizes the markdown body.
changelog.py
Reusable as-is from the template. Walks git diff --name-status
output for the commit summary, and git log for the digest history
(Phase 13).