Files
stack/.gitignore
kert 31c9366b93 feat(dossier): the dossier page live at git.<domain>/dossier behind git-sso (S0, R76)
dev/scripts/dossier_page/build.py (from the session preview generator)
reads the replica and bib.sqlite read-only plus the public
regulations.gov mirror, and writes one complete HTML document with the
data inlined to infra/gitea/custom/public/assets/dossier/index.html
(gitignored; rebuilt, never committed). AMA safety kept: no RVU-file
short descriptors are read, descriptor_change notes are fixed text.
The page write is atomic and every "<" in the inlined JSON is escaped.

template.html is now a full document (doctype, lang, charset, viewport,
title); the artifact-only viewport-fit meta and safe-area hooks are gone;
the banner reads "Static snapshot of live data, built <timestamp>".

Traefik: routers dossier(-tls) rewrite ^/dossier/?$ to the static page,
and dossier-assets(-tls) keep /assets/dossier/ behind git-sso; both at
priority 1000 on the git service, with Cache-Control: no-cache to
override Gitea's 6 h asset cache.
2026-11-12 17:36:09 -05:00

65 lines
1.3 KiB
Plaintext

# Virtual environment
.venv/
# Bootstrap state (credentials, OAuth2 tokens)
.state/
# Data directories (user data, not tracked)
data/*
!data/.gitkeep
data/zotero/data/
!data/zotero/data/styles/
!data/zotero/data/translators/CMS 4innovation.js
data/zotero/profiles/
tuva/
logs/**
# Generated DAB SQL (regenerate via gen_config.py --dab-sql)
bundle/sql/
# Marimo cache
notebooks/__marimo__/
# Secrets and keys
.deploy_key
.deploy_key.pub
.env
.env.bak
infra/cloudflared/*.json
infra/cloudflared/cert.pem
# Python build artifacts
__pycache__/
*.pyc
.coverage
dist/
*.egg-info/
htmlcov/
# IDE / tool settings
.claude/settings.local.json
notebooks/aco.duckdb
# Package mirrors (local cache, not tracked)
mirrors/
# Docs (auto-generated at build time)
docs/docs/api/
docs/static/library.json
docs/node_modules/
docs/build/
docs/.docusaurus/
hw/node_modules/
infra/marimo/src/
# Claude Code worktrees (linked checkouts, never tracked)
.claude/worktrees/
# LEAD reference corpus (20 GB); its documents live in bib storage and
# data/public/lead_enrichment, never in git.
lead_corpus/
# Dossier page (dev/scripts/dossier_page/build.py): generated with the data
# inlined, served by Gitea at git.<domain>/dossier; rebuilt, never committed.
infra/gitea/custom/public/assets/dossier/index.html