Configuration

Environment variables consumed by the brgybnd toolchain: OPENAI_API_KEY, OPENAI_BASE_URL, OPENAI_MODEL, PSGC_DATA_DIR.

Author

Manuel Balmeo

The brgybnd toolchain is configured via environment variables (read by config.py with python-dotenv). The LLM-related variables are only required for the optional Pipeline A agent batch-extraction step.

Variables

Variable	Default	Purpose
`OPENAI_API_KEY`	(none — required for LLM agent pipeline)	API key for the OpenAI-compatible LLM client
`OPENAI_BASE_URL`	`https://api.z.ai/api/paas/v4/` (GLM API)	Base URL of the OpenAI-compatible endpoint
`OPENAI_MODEL`	`glm-4.7`	Model name for structured-output extraction
`PSGC_DATA_DIR`	`~/commons/barangay`	Directory holding the PSGC snapshot source files

.env example

A .env in the repository root (loaded by config.py via python-dotenv) only needs the data directory in practice; the three OPENAI_* vars are for the optional Pipeline A LLM step and can be omitted for offline runs:

# required to point at the PSGC snapshot source tree
PSGC_DATA_DIR=~/commons/barangay

# optional — only for the LLM batch-extraction step of Pipeline A
OPENAI_API_KEY=your-key
OPENAI_BASE_URL=https://api.z.ai/api/paas/v4/
OPENAI_MODEL=glm-4.7

PSGC_DATA_DIR is the root of the snapshot tree. ingest/scanner.py expects each snapshot as a per-quarter subdirectory named YYYY-MM-DD/ (e.g. 2023-10-24/) containing the PSA Excel datafile, the changes sheet(s), and the PDF press release — it classifies the files inside by filename regex. Pointing PSGC_DATA_DIR anywhere else just relocates that tree.

The GLM defaults (OPENAI_BASE_URL / OPENAI_MODEL) target the Z.ai GLM endpoint; set them to any OpenAI-compatible base URL and structured-output model of your choice.