PyGeoModel 1.0.7

A Python package for integrating OpenGMS geographic model services.

PyGeoModel

PyGeoModel is a Python package for integrating OpenGMS geographic model services into Python-based urban data science workflows. It provides programmatic access to model-service discovery, metadata inspection, service invocation, task records, and result management. For exploratory notebook-based analysis, PyGeoModel also provides an optional Jupyter interface built on the same core API.

Installation

pip install PyGeoModel

Use the package in Python with:

from pygeomodel import GeoModeler

Core API

from pygeomodel import GeoModeler

modeler = GeoModeler()

models = modeler.search_models("photovoltaic")
model = modeler.get_model("Roof Photovoltaic Carbon Emission Reduction Potential Assessment Model")

print(model.description)
print(model.inputs)
print(model.outputs)

Model services can be invoked programmatically:

result = modeler.invoke(
    "Roof Photovoltaic Carbon Emission Reduction Potential Assessment Model",
    params={
        "system_efficiency": 0.8,
        "start_time": 201801,
        "end_time": 201812,
        "roof_vector_path": "data/rooftops.zip",
    },
)

result.to_json("execution_record.json")

OpenGMS execution uses the bundled public demo token by default. For stable or private use, configure your own OpenGMS token:

export OGMS_TOKEN="your-token"

Optional endpoint overrides:

export OGMS_BASE_PORTAL_URL="http://222.192.7.75"
export OGMS_BASE_MANAGER_URL="http://222.192.7.75/managerServer"
export OGMS_BASE_DATA_URL="http://222.192.7.75/dataTransferServer"

Notebook Interface

modeler.show_models()
modeler.invoke_model("SWAT_Model")

The notebook interface renders model search, metadata inspection, parameter entry, task execution, and output display. It uses the same search_models(), get_model(), and invoke() functions as the programmatic API so GUI operations can be converted into Python dictionaries and execution records.

Recommendation and Q&A Records

recommendation = modeler.suggest_model(return_result=True)
recommendation.to_json("recommendation_record.json")

answer = modeler.ask_model("SWAT_Model", "What input data are required?")
answer.to_json("qa_record.json")

The recommendation service automatically builds notebook/data context and calls the configured Dify workflow. Q&A uses OpenGMS model metadata and an OpenAI-compatible web-enabled model. PyGeoModel includes public demo credentials for these services so the main notebook workflow can run out of the box. For stable, private, or high-volume use, configure your own credentials with DIFY_API_KEY, OPENAI_API_KEY, OPENAI_BASE_URL, and OPENAI_MODEL.

Relation to OpenGMS

OpenGMS provides the model-service platform and online execution infrastructure. PyGeoModel is a Python client package that exposes OpenGMS model-service discovery, metadata inspection, task invocation, and result records to Python and notebook workflows.

Development

Run the lightweight test suite with:

python -m unittest discover -s tests

The source distribution should not include real API keys, local .env files, bytecode caches, generated C files, .pyd binaries, or previous build artifacts.