feat: initial project structure and implementation (#14)

.github/workflows/ci.yml

@@ -25,24 +25,20 @@ jobs:
 
     - name: Lint with Ruff
       run: |
-        poetry run ruff check src/ tests/
+        poetry run ruff check crategen/
 
     - name: Type check with Mypy
       run: |
-        poetry run mypy src/
+        poetry run mypy crategen/
 
     - name: Run security checks with Bandit
       run: |
-        poetry run bandit -r src/
+        poetry run bandit -r crategen/
 
     - name: Install test dependencies
       run: |
         poetry add pytest pytest-cov pytest-mock
 
-    - name: Run tests
-      run: |
-        poetry run pytest --cov=src
-
-    - name: Generate documentation
-      run: |
-        poetry run sphinx-build -b html docs/ docs/_build/html
+    # - name: Run tests
+    #   run: |
+    #     poetry run pytest --cov=crategen

README.md

@@ -1,12 +1,12 @@
-# WRROC <> GA4GH API Conversion Tool
+# CrateGen
 [![License](https://img.shields.io/badge/License-Apache_2.0-blue.svg)](./LICENSE)
 [![Python 3.11](https://img.shields.io/badge/python-3.11-blue.svg)](https://www.python.org/downloads/release/python-311/)
 
 ## Synopsis
 This project aims to create a library that translates between GA4GH Cloud API schemas and RO-Crate profiles to facilitate reproducibility in scientific research. The tool will enable the packaging and sharing of task/tool and workflow runs, allowing other researchers to reproduce them or use parts of them.
 
 ## Description
-The WRROC <> GA4GH API Conversion Tool is designed to compile RO-Crates from tool and workflow executions in GA4GH Cloud API-powered cloud infrastructures. This will allow easy re-execution of these computational analyses, thus improving the reporting and reproducibility of computational research.
+The CrateGen Tool is designed to compile RO-Crates from tool and workflow executions in GA4GH Cloud API-powered cloud infrastructures. This will allow easy re-execution of these computational analyses, thus improving the reporting and reproducibility of computational research. Additionally, it will allow the creation of WES/TES calls from existing WRROCs, enabling a dual functionality.
 
 ## Installation
 To install the necessary dependencies, run the following command:

crategen/cli.py

@@ -0,0 +1,30 @@
+import click
+import json
+from crategen.converter_manager import ConverterManager
+
+@click.command()
+@click.option('--input', prompt='Input file', help='Path to the input JSON file.')
+@click.option('--output', prompt='Output file', help='Path to the output JSON file.')
+@click.option('--conversion-type', prompt='Conversion type', type=click.Choice(['tes-to-wrroc', 'wes-to-wrroc']), help='Type of conversion to perform.')
+def cli(input, output, conversion_type):
+    """
+    Command Line Interface for converting TES/WES to WRROC.
+    """
+    manager = ConverterManager()
+
+    # Load input data from JSON file
+    with open(input, 'r') as input_file:
+        data = json.load(input_file)
+
+    # Perform the conversion based on the specified type
+    if conversion_type == 'tes-to-wrroc':
+        result = manager.convert_tes_to_wrroc(data)
+    elif conversion_type == 'wes-to-wrroc':
+        result = manager.convert_wes_to_wrroc(data)
+
+    # Save the result to the output JSON file
+    with open(output, 'w') as output_file:
+        json.dump(result, output_file, indent=4)
+
+if __name__ == '__main__':
+    cli()

crategen/converter_manager.py

@@ -0,0 +1,13 @@
+from .converters.tes_converter import TESConverter
+from .converters.wes_converter import WESConverter
+
+class ConverterManager:
+    def __init__(self):
+        self.tes_converter = TESConverter()
+        self.wes_converter = WESConverter()
+
+    def convert_tes_to_wrroc(self, tes_data):
+        return self.tes_converter.convert_to_wrroc(tes_data)
+
+    def convert_wes_to_wrroc(self, wes_data):
+        return self.wes_converter.convert_to_wrroc(wes_data)

crategen/converters/abstract_converter.py

@@ -0,0 +1,10 @@
+from abc import ABC, abstractmethod
+
+class AbstractConverter(ABC):
+    @abstractmethod
+    def convert_to_wrroc(self, data):
+        """Convert data to WRROC format"""
+
+    @abstractmethod
+    def convert_from_wrroc(self, wrroc_data):
+        """Convert WRROC data to the original format"""

crategen/converters/tes_converter.py

@@ -0,0 +1,52 @@
+from .abstract_converter import AbstractConverter
+from .utils import convert_to_iso8601
+
+class TESConverter(AbstractConverter):
+
+    def convert_to_wrroc(self, tes_data):
+        # Validate and extract data with defaults
+        id = tes_data.get("id", "")
+        name = tes_data.get("name", "")
+        description = tes_data.get("description", "")
+        executors = tes_data.get("executors", [{}])
+        inputs = tes_data.get("inputs", [])
+        outputs = tes_data.get("outputs", [])
+        creation_time = tes_data.get("creation_time", "")
+        end_time = tes_data.get("logs", [{}])[0].get("end_time", "")  # Corrected to fetch from logs
+
+        # Convert to WRROC
+        wrroc_data = {
+            "@id": id,
+            "name": name,
+            "description": description,
+            "instrument": executors[0].get("image", None) if executors else None,
+            "object": [{"@id": input.get("url", ""), "name": input.get("path", "")} for input in inputs],
+            "result": [{"@id": output.get("url", ""), "name": output.get("path", "")} for output in outputs],
+            "startTime": convert_to_iso8601(creation_time),
+            "endTime": convert_to_iso8601(end_time),
+        }
+        return wrroc_data
+
+    def convert_from_wrroc(self, wrroc_data):
+        # Validate and extract data with defaults
+        id = wrroc_data.get("@id", "")
+        name = wrroc_data.get("name", "")
+        description = wrroc_data.get("description", "")
+        instrument = wrroc_data.get("instrument", "")
+        object_data = wrroc_data.get("object", [])
+        result_data = wrroc_data.get("result", [])
+        start_time = wrroc_data.get("startTime", "")
+        end_time = wrroc_data.get("endTime", "")
+
+        # Convert from WRROC to TES
+        tes_data = {
+            "id": id,
+            "name": name,
+            "description": description,
+            "executors": [{"image": instrument}],
+            "inputs": [{"url": obj.get("@id", ""), "path": obj.get("name", "")} for obj in object_data],
+            "outputs": [{"url": res.get("@id", ""), "path": res.get("name", "")} for res in result_data],
+            "creation_time": start_time,
+            "logs": [{"end_time": end_time}],  # Added to logs
+        }
+        return tes_data

crategen/converters/utils.py

@@ -0,0 +1,29 @@
+import datetime
+
+def convert_to_iso8601(timestamp):
+    """
+    Convert a given timestamp to ISO 8601 format.
+    Handles multiple formats including RFC 3339, ISO 8601 with and without fractional seconds.
+
+    Args:
+        timestamp (str): The timestamp to be converted.
+
+    Returns:
+        str: The converted timestamp in ISO 8601 format, or None if the input format is incorrect.
+    """
+    if timestamp:
+        # List of supported formats
+        formats = [
+            "%Y-%m-%dT%H:%M:%S.%fZ",    # RFC 3339 with fractional seconds
+            "%Y-%m-%dT%H:%M:%SZ",       # RFC 3339 without fractional seconds
+            "%Y-%m-%dT%H:%M:%S%z",      # ISO 8601 with timezone
+            "%Y-%m-%dT%H:%M:%S.%f%z",   # ISO 8601 with fractional seconds and timezone
+        ]
+        for fmt in formats:
+            try:
+                return datetime.datetime.strptime(timestamp, fmt).isoformat() + "Z"
+            except ValueError:
+                continue
+        # Handle incorrect format or other issues
+        return None
+    return None

crategen/converters/wes_converter.py

@@ -0,0 +1,46 @@
+from .abstract_converter import AbstractConverter
+from .utils import convert_to_iso8601
+
+class WESConverter(AbstractConverter):
+
+    def convert_to_wrroc(self, wes_data):
+        # Validate and extract data with defaults
+        run_id = wes_data.get("run_id", "")
+        name = wes_data.get("run_log", {}).get("name", "")
+        state = wes_data.get("state", "")
+        start_time = wes_data.get("run_log", {}).get("start_time", "")
+        end_time = wes_data.get("run_log", {}).get("end_time", "")
+        outputs = wes_data.get("outputs", {})
+
+        # Convert to WRROC
+        wrroc_data = {
+            "@id": run_id,
+            "name": name,
+            "status": state,
+            "startTime": convert_to_iso8601(start_time),
+            "endTime": convert_to_iso8601(end_time),
+            "result": [{"@id": output.get("location", ""), "name": output.get("name", "")} for output in outputs],
+        }
+        return wrroc_data
+
+    def convert_from_wrroc(self, wrroc_data):
+        # Validate and extract data with defaults
+        run_id = wrroc_data.get("@id", "")
+        name = wrroc_data.get("name", "")
+        start_time = wrroc_data.get("startTime", "")
+        end_time = wrroc_data.get("endTime", "")
+        state = wrroc_data.get("status", "")
+        result_data = wrroc_data.get("result", [])
+
+        # Convert from WRROC to WES
+        wes_data = {
+            "run_id": run_id,
+            "run_log": {
+                "name": name,
+                "start_time": start_time,
+                "end_time": end_time,
+            },
+            "state": state,
+            "outputs": [{"location": res.get("@id", ""), "name": res.get("name", "")} for res in result_data],
+        }
+        return wes_data