Repository Map

All repositories live in the sanskrit-lexicon GitHub organization.

Per-dictionary repositories

Each dictionary has its own repo, named by its code. Examples (see the full catalog):

Code	Dictionary	Repo
MWS	Monier-Williams	MWS
AP90	Apte (1890)	AP90
PWG	Böhtlingk-Roth (large)	PWG
PWK	Böhtlingk (shorter)	PWK
GRA	Grassmann (Rig-Veda)	GRA

Infrastructure repositories

Repo	Role
csl-orig	Canonical source: v02/{dict}/{dict}.txt for every dictionary
csl-pywork	Build tooling: generate_dict.sh, make_xml.py, validators
csl-corrections	Audit trail: change files grouped in batch_YYYYMMDD/
csl-websanlexicon	Web/display assets consumed by generation
csl-apidev	Web backend — the RESTful + Salt API (PHP)
csl-app	Cross-platform app (Android/iOS/macOS/Linux/Windows; Dart/Flutter)
csl-standards	Shared standards (incl. the normative Salt API contract)
csl-inflect	Inflected-forms generation (e.g. MW inflected forms)
csl-sqlite	The per-dictionary SQLite search databases (published via GitHub Releases)
csl-json	JSON form of the dictionary data ({words, text})
cologne-stardict	StarDict / offline packaging (via Babylon export)
csl-doc	Sphinx per-dictionary front-matter / prefaces documentation

Other infrastructure repos (csl-atlas, csl-observatory, …) support tooling and observability. See Data Formats for how the SQLite, JSON, and StarDict artifacts are produced.

Dictionary code vs. repository name

A dictionary's code (used in URLs, csl-orig, and the API) is not always its repo name. The clearest case: the shorter Petersburg dictionary has code PW (csl-orig/v02/pw/) but lives in the repo PWK. The catalog lists each dictionary's actual repo.

Conventions shared across repos

• Session state: each repo keeps a tracked .ai_state.md journal.
• Correction pattern: most repos apply corrections via updateByLine.py change files (see Change Files).
• Input files for the large German dictionaries live in sibling *xml repos (e.g. ../pwgxml/pwg.xml, ../mwsxml/mws.xml).