justin
a0727da8da
Two new bundles:
* hvm_qualification_matrix (sd00006551en_us) — the "Qualification Matrix
for HVM Clusters Managed by HPE Morpheus Software". Single TOC bundle,
2 pages (parent + content). The content page is ~100 KB of HTML
containing five tables: Server Hardware Support, Storage Hardware
Support, Independent Software Vendor (ISV) Support, Hypervisor OS
Compatibility and Interoperability Matrix, and Guest OS. Scraped via
the same /hpesc/public/api/document/{docId}/render endpoint as every
other bundle on support.hpe.com — the API returns server-rendered
DITA HTML, so no JS/SPA shenanigans.
* hvm_quickspecs (a50004260enw) — HPE Morpheus VM Essentials Software
QuickSpecs, Version 4 (02-Feb-2026). SKUs: S5Q81AAE (1-yr per Socket
E-LTU), S5Q82AAE (3-yr), S5Q83AAE (5-yr); each includes Tech Care
Essentials. QuickSpecs lives at www.hpe.com (not support.hpe.com),
which drops connections at the edge for non-browser TLS fingerprints —
verified 2026-05-22 against curl, wget, urllib, and Anthropic's
WebFetch (all = 0 bytes / connection timeout in headers). Bypassed
here via curl_cffi impersonating Chrome 120's JA3/JA4 fingerprint.
HTTP 200, 255 KB on first try, all four sections + all three SKUs
cleanly parseable from the server-rendered HTML.
New module scrape/quickspecs.py drives the live fetch + parse for any
hvm_*_quickspecs bundle. CSS selectors taken from the captured DOM:
.lr-right-rail hpe-highlights-container .collateral-content
— one block per H3 section
h3.txto-title — section title
div.txto-description — section body
uc-table.uc-table-polaris — SKU and version-history tables
On any live failure the parser falls back to a committed HTML fixture
at scrape/quickspecs/<doc_id>.html so the build never breaks on a
transient edge hiccup.
scrape/runner.py learned a new mode "html-file" that dispatches to
scrape.quickspecs; bundles.py extended with an optional source_url on
BundleSpec for cases where the page lives outside support.hpe.com.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
28 lines
782 B
Plaintext
28 lines
782 B
Plaintext
# MCP server
|
|
mcp[fastmcp]>=1.0.0
|
|
pydantic>=2.0
|
|
httpx>=0.27
|
|
|
|
# Vector store + embeddings
|
|
chromadb>=0.5.0
|
|
ollama>=0.4.0 # if using Ollama-hosted embedder; swap if not
|
|
|
|
# Scraping (Phase 1; adjust per product)
|
|
beautifulsoup4>=4.12
|
|
requests>=2.31
|
|
curl_cffi>=0.7 # for HPE QuickSpecs scrape (Chrome TLS impersonation)
|
|
markdownify>=0.11
|
|
# playwright>=1.40 # uncomment if you need headless browser fallback
|
|
|
|
# Evaluation
|
|
numpy>=1.26
|
|
|
|
# Reranker is a sidecar (see deploy/docker-compose.yml). The MCP server
|
|
# only needs httpx (declared above) to call it. For the dev / CPU
|
|
# fallback reranker (scripts/rerank_server.py), install
|
|
# requirements-rerank.txt separately — it pulls in PyTorch which would
|
|
# triple the production image size.
|
|
|
|
# Dev / utility
|
|
python-dateutil>=2.8
|