This is a multi-stage tool for generating climate action priorities for cities.
This project is licensed under the Affero General Public License v3.0. See the LICENSE file for details.
data/: Contains data files used by the different stages.schema/: Contains JSON schemas for the data files.extractor/: Extracts action data and generates a JSON file.prioritizer/: Prioritizes actions based on the extracted data.scripts/: Contains helper scripts like adding ccras, ghgis or uploading files to AWS S3. This folder also contains arun_pipelinescript which automatically prioritizes all actions inside the long list of actions against a chosen city, transforms the data to match the required frontend schema and uploads the created file to the AWS S3 bucket for updating the frontend.
Run the run_pipeline script inside /scripts folder with bash run_pipeline.sh LOCODE where LOCODE needs to be replaced by a matching locode from the cities data file inside data/cities.
This script will automatically create the ranked actions for this city, format and upload those actions to the AWS S3 bucket for displaying in the frontend app.
To run this pipeline, AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY must be set in an .env file with permissions to write to the S3 bucket /data folder.
To run the script, the packages inside /extractor/requirements.txt needs to be installed. The script will be looking for a virtual environment .cap inside the /.cap top level folder.
If you create the virtual environment with a different name then .cap the path inside run_pipeline.sh must be adjusted accordingly.
Under Windows OS, this script is best executed under WSL oder Git bash.
If the LLM logs should be logged by langsmith, the .env variables must be set
LANGCHAIN_TRACING_V2=true
as well as the corresponding project name and api key.