TM Extractor

The TM Extractor script is designed to trigger extraction requests for Tasking Manager projects. It utilizes the HOTOSM Tasking Manager API and the Raw Data API for data extraction. For more complex queries navigate to osm-rawdata module

Table of Contents

• Getting Started
  • Prerequisites
  • Installation
• Usage
  • Command Line
  • AWS Lambda
  • Streamlit APP
• Configuration
  • Environment Variables
  • Config JSON
• Releasing
• Script Overview
• Result Path

Getting Started

Prerequisites

• Python 3.10+
• uv installed (curl -LsSf https://astral.sh/uv/install.sh | sh)
• Access token for Raw Data API

Installation

The package is published on PyPI. The fastest path requires no cloning and no manual install, uvx fetches the package into an ephemeral environment and runs it:

uvx tm-extractor --fetch-active-projects 24

For local development, clone and sync:

git clone https://github.com/kshitijrajsharma/TM-Extractor
cd TM-Extractor
uv sync

Usage

Command Line

Head over to Env to verify you have setup correct env variables & run the command with the following options:

• For Specific TM Projects :

uvx tm-extractor --projects 123 456 789

• For fetching active projects within last 24hr

uvx tm-extractor --fetch-active-projects 24

• For tracking request and Dumping result

uvx tm-extractor --projects 123 --track

During local development, swap uvx tm-extractor for uv run tm-extractor to use the checked-out source.

You can set it up as systemd service or cronjob in your PC if required or run manually.

AWS Lambda

1. Create an AWS Lambda Function:

   • In the AWS Management Console, navigate to the Lambda service, Choose role and create one with necessary permissions

2. Set Environment Variables:

   • Add the following environment variables to your Lambda function configuration:

     • CONFIG_JSON: [Optional] Path to a custom config JSON file. If unset, the packaged default bundled inside tm-extractor is used.
     • Refer to Configurations for more env variables as required

3. Deploy the Script as a Lambda Function:

   • Build a deployment bundle with uv pip install --target ./package 'tm-extractor' (add [sentry] for Sentry support).
   • Zip the contents of ./package together with any custom config.json override and upload to your Lambda function.
   • Set the handler to tm_extractor.lambda_handler.

4. Configure Lambda Trigger:

   • Configure an appropriate event source for your Lambda function. This could be an API Gateway, CloudWatch Event, or another trigger depending on your requirements.

5. Invoke the Lambda Function:

   • Trigger the Lambda function manually or wait for the configured event source to invoke it. Event shape: {"fetch_active_projects": 24} or {"projects": [123, 456]}.

   Your Lambda function should be able to execute the script with the provided configurations.

Using Terraform

1. Download & install Terraform:

   • Install terraform from the HashiCorp install guide.

2. Create AWS Credentials

   • Create an IAM programmatic user :

     • Configure .aws/credentials or use AWS Environment varibles for terraform authentication. Check Here

3. Environment Variables You can set terraform input variables using TF_VAR_<varible-name-from-vars.tf> if you don't want to provide them each time.

4. Run Terraform Plan/Apply

   • Run terraform init to download required providers
   • Run terraform plan plan to check for infrastructure changes.
   • Run terraform apply to apply terraform configurations.

NOTE: Check main.tf for resources terraform creates.

Streamlit

The Streamlit frontend lives in this repo but is not shipped in the PyPI package.

1. Run locally :

   uv run --with streamlit streamlit run streamlit_app.py

2. To use hosted service : Go to tm-extractor.streamlit.app

Configuration

Environment Variables

Set the following environment variables for proper configuration. .env files in the working directory are loaded automatically.

Example :

export RAWDATA_API_AUTH_TOKEN='my_token'

• RAWDATA_API_AUTH_TOKEN: API token for Raw Data API authentication, Request admins for yours to RAW DATA API

• RAW_DATA_API_BASE_URL: Base URL for the Raw Data API. Default is https://api-prod.raw-data.hotosm.org/v1.

• TM_API_BASE_URL: Base URL for the Tasking Manager API. Default is https://tasking-manager-production-api.hotosm.org/api/v2.

• CONFIG_JSON: Path to a custom config JSON file. If unset, the default config.json bundled inside the package is used.

• TASKING_MANAGER_API_KEY: [Optional] Tasking manager API key . Example : Token your_token_key_from_tasking_manager. Only required to fetch projects that requires authentication.

• SENTRY_DSN: [Optional] Enables Sentry error reporting. Install with the [sentry] extra: uvx --from 'tm-extractor[sentry]' tm-extractor ....

Config JSON

The config.json file contains configuration settings for the extraction process. It includes details about the dataset, categories, and geometry of the extraction area. A default config.json ships inside the package (src/tm_extractor/config.json) and is used automatically when no override is provided; pass --config path/to/config.json or set CONFIG_JSON=... to override.

{
    "geometry": {...},
    "dataset": {...},
    "categories": [...]
}

Explanation

geometry

Defines the geographical area for extraction. Typically auto-populated with Tasking Manager (TM) geometry.

queue

Specifies the Raw Data API queue, often set as "raw_default" for default processing, This can be changed if there is disaster activation and special services are deployed so that those requests can be prioritized.

dataset

Contains information about the dataset:

• dataset_prefix: Prefix for the dataset extraction eg : hotosm_project_123.
• dataset_folder: Default Mother folder to place during extraction eg : TM , Mindful to change this.
• dataset_title: Title of the Tasking Manager project eg : Tasking Manger Project 123.

categories

Array of extraction categories, each represented by a dictionary with:

• Category Name: Name of the extraction category (e.g., "Buildings", "Roads").
  • types: Types of geometries to extract (e.g., "polygons", "lines", "points").
  • select: Attributes to select during extraction (e.g., "name", "highway", "surface").
  • where: Conditions for filtering the data during extraction (e.g., filtering by tags).
  • formats: File formats for export (e.g., "geojson", "shp", "kml").

Adjust these settings based on your project requirements and the types of features you want to extract.

Refer to the sample config.json for default config.

Releasing

Versioning is managed by commitizen using conventional commits.

uv run cz bump          # bumps version in pyproject.toml, updates CHANGELOG.md, creates v$version tag
git push --follow-tags  # push the tag, then create a GitHub Release to trigger PyPI publish

The publish.yml workflow builds with uv build and publishes to PyPI on release.

Script Overview

Purpose

The script is designed to trigger extraction requests for Tasking Manager projects using the Raw Data API. It automates the extraction process based on predefined configurations.

Features

• Supports both command line and AWS Lambda execution.
• Dynamically fetches project details, including mapping types and geometry, from the Tasking Manager API.
• Configurable extraction settings using a config.json file.
• Helps debugging the service and track the api requests

Result Path

• Your export download link will be generated based on the config , with raw-data-api logic it will be Base_download_url/dataset_folder/dataset_prefix/Category_name/feature_type/dataset_prefix_category_name_export_format.zip
• Example for Waterways configuration : Here Category Name is Waterways, dataset_prefix is hotosm_project_9, dataset_folder is TM , feature_type is lines and export format is geojson

"Waterways": {
    "resources": [
    {
        "name": "hotosm_project_9_waterways_lines_geojson.zip",
        "format": "geojson",
        "description": "GeoJSON",
        "url": "https://s3.sample.your_domain.org/TM/hotosm_project_9/waterways/lines/hotosm_project_9_waterways_lines_geojson.zip",
        "last_modifed": "2023-12-28T17:48:21.378667"
    },
    ]
}

See sample result to go through how result will look like