new setup.cfg parser. Fixes #988

docs/setupcfg.md

@@ -0,0 +1,52 @@
+The following metadata fields can be extracted from a setup.cfg file.   
+These fields are defined in the [setuptools declarative configuration specification](https://setuptools.pypa.io/en/latest/userguide/declarative_config.html), and are mapped according to the [CodeMeta crosswalk for Python Distutils](https://github.com/codemeta/codemeta/blob/master/crosswalks/Python%20Distutils%20(PyPI).csv).
+
+| Software metadata category  |    SOMEF metadata JSON path    | SETUP.CFG metadata file field     |
+|--------------------------------|-----------------------------|----------------------------------------|
+| author - value                   |  author[i].result.value       |     metadata.author                  |
+| author - email                   |  author[i].result.email       |     metadata.author_email            |
+| author - name                   |  author[i].result.name        |     metadata.author                |
+| code_repository           |  code_repository[i].result.value  |  project_urls (source, repository, code) |
+| description               |  description[i].result.value |  metadata.description                |
+| documentation             |  documentation[i].result.value  |  project_urls (Documentation, docs) |       
+| license - value                   |  license[i].result.value  |   metadata.license or metadata.license_files                   |
+| license - name                   |  license[i].result.name    |  metadata.license           *(1)*   |
+| license - spdx id                   |  license[i].result.spdx_id |    metadata.license if "spdx.org/licenses/"   *(1)* |  
+| has_package_file          |  has_package_file[i].result.value    |   URL of the setup.cfg file      |
+| homepage                  |  homepage[i].result.value          |   metadata.url or project_urls (Homepage)     |
+| keywords                  |  keywords[i].result.value          |     metadata.keywords                 |
+| package_id                |  package_id[i].result.value |   metadata.name  |
+| requirements - value              | requirements[i].result.value  |   options.install_requires or options.setup_requires *(2)*   |
+| requirements - name              |  requirements[i].result.name   |   options.install_requires or options.setup_requires -> name  *(2)*     |
+| requirements - version              |  requirements[i].result.version   | options.install_requires or options.setup_requires -> version *(2)*   |
+| runtime_platform - value         |  runtime_platform[i].result.value |  options.python_requires -> "Python" + version  *(3)*  |
+| runtime_platform - name          |  runtime_platform[i].result.name |   options.python_requires -> "Python"  *(3)* |
+| runtime_platform - version          |  runtime_platform[i].result.version |   options.python_requires  *(3)* |
+| version - value                   |  version[i].result.value |  metadata.version |
+| version - tag                  |  version[i].result.tag                            |     metadata.version                            |
+
+---
+
+*(1)*  
+- Look for the name and spdx_id in a local dictionary with all licenses.
+
+*(2)*  
+- Examples of requirements
+```
+[options]
+install_requires =
+    astropy
+    ctapipe >= 0.12
+    h5py ~= 3.1.0
+
+setup_requires =
+    setuptools >= 40.6.0
+    wheel
+
+```
+
+*(3)*  
+- Example:
+```
+python_requires = >= 3.10.0
+```

docs/supported_languages.md

@@ -12,7 +12,7 @@ To know more about the extraction details for each type of file, click on it.
 | JavaScript | [`package.json`](./packagejson.md), [`bower.json`](./bower.md) | 
 | Julia | [`Project.toml`](./julia.md) | 
 | PHP | [`composer.json`](./composer.md) |  
-| Python | [`setup.py`](./setuppy.md), [`pyproject.toml`](./pyprojecttoml.md), [`requirements.txt`](./requirementstxt.md) | 
+| Python | [`setup.py`](./setuppy.md), [`setup.cfg`](./setupcfg.md), [`pyproject.toml`](./pyprojecttoml.md), [`requirements.txt`](./requirementstxt.md) | 
 | R | [`DESCRIPTION`](./description.md) | 
 | Ruby | [`*.gemspec`](./gemspec.md) |  
 | Rust | [`Cargo.toml`](./cargo.md) |  

docs/supported_metadata_files.md

@@ -21,6 +21,7 @@ SOMEF can extract metadata from a wide range of files commonly found in software
 | `pyproject.toml`   | Python                     | Modern Python project configuration file used by tools like Poetry and Flit |   [🔍](./pyprojecttoml.md)| [📄](https://packaging.python.org/en/latest/guides/writing-pyproject-toml/)| [PEP 621](https://peps.python.org/pep-0621/)| [Example](https://github.com/KnowledgeCaptureAndDiscovery/somef/blob/master/pyproject.toml) | 
 | `requirements.txt` | Python                     | Lists Python package dependencies | [🔍](./requirementstxt.md)| [📄](https://pip.pypa.io/en/stable/reference/requirements-file-format/)| [Latest](https://pip.pypa.io/en/stable/reference/requirements-file-format/)| [Example](https://github.com/oeg-upm/FAIR-Research-Object/blob/main/requirements.txt) |
 | `setup.py`         | Python                     | Package file format used in python projects | [🔍](./setuppy.md)| [📄](https://setuptools.pypa.io/en/latest/references/keywords.html)| [v75.0.0](https://github.com/pypa/setuptools)| [Example](https://github.com/oeg-upm/soca/blob/main/setup.py) | 
+| `setup.cfg` | Python | Configuration file for setuptools used to define package metadata and options in a declarative way |  [🔍](./setupcfg.md)| [📄](https://setuptools.pypa.io/en/latest/userguide/declarative_config.html) | [v75.0.0](https://github.com/pypa/setuptools)|[Example](https://github.com/oeg-upm/soca/blob/main/setup.cfg)|
 | `DESCRIPTION`      | R                          | Metadata file for R packages including title, author, and version |   [🔍](./description.md) | [📄](https://cran.r-project.org/doc/manuals/R-exts.html#The-DESCRIPTION-file)| [v4.4.1](https://cran.r-project.org/doc/manuals/r-release/R-exts.html) |  [Example](https://github.com/cran/ggplot2/blob/master/DESCRIPTION) |
 | `*.gemspec`        | Ruby                       | Manifest file serves as the package descriptor used in Ruby gem projects. |  [🔍](./gemspec.md)| [📄](https://guides.rubygems.org/specification-reference/)| [v3.5.22](https://github.com/rubygems/rubygems)|[Example](https://github.com/rubygems/rubygems/blob/master/bundler/bundler.gemspec) |
 | `cargo.toml`       | Rust                       | Manifest file serves as the package descriptor used in Rust projects |  [🔍](./cargo.md) | [📄](https://doc.rust-lang.org/cargo/reference/manifest.html)| [v0.85.0](https://github.com/rust-lang/cargo) | [Example](https://github.com/rust-lang/cargo/blob/master/Cargo.toml) |

src/somef/parser/setupcfg_parser.py

@@ -0,0 +1,192 @@
+import re
+import os
+import logging
+import configparser
+from pathlib import Path
+from ..process_results import Result
+from ..utils import constants
+from ..regular_expressions import detect_license_spdx, detect_spdx_from_declared
+
+def parse_setup_cfg(file_path, metadata_result: Result, source):
+    """
+    Parser for setup.cfg files. Very similar to the one for pyproject.toml, but using configparser instead of toml library.
+    """
+
+    try:
+        metadata_result.add_result(
+            constants.CAT_HAS_PACKAGE_FILE,
+            {"value": source, "type": constants.URL},
+            1,
+            constants.TECHNIQUE_CODE_CONFIG_PARSER,
+            source
+        )
+
+        config = configparser.ConfigParser()
+        config.read(file_path, encoding="utf-8")
+
+        metadata = dict(config["metadata"]) if "metadata" in config else {}
+        options = dict(config["options"]) if "options" in config else {}
+
+        if "name" in metadata:
+            metadata_result.add_result(
+                constants.CAT_PACKAGE_ID,
+                {"value": metadata["name"], "type": constants.STRING},
+                1, constants.TECHNIQUE_CODE_CONFIG_PARSER, source
+            )
+
+        if "version" in metadata:
+            version_value = metadata["version"]
+            if not version_value.startswith("attr:"):
+                metadata_result.add_result(
+                    constants.CAT_VERSION,
+                    {"value": version_value, "type": constants.RELEASE, "tag": version_value},
+                    1, constants.TECHNIQUE_CODE_CONFIG_PARSER, source
+                )
+
+        if "description" in metadata:
+            metadata_result.add_result(
+                constants.CAT_DESCRIPTION,
+                {"value": metadata["description"], "type": constants.STRING},
+                1, constants.TECHNIQUE_CODE_CONFIG_PARSER, source
+            )
+
+        if "author" in metadata or "author_email" in metadata:
+            author_data = {
+                "name": metadata.get("author"),
+                "email": metadata.get("author_email"),
+                "type": constants.AGENT,
+                "value": metadata.get("author")
+            }
+            metadata_result.add_result(
+                constants.CAT_AUTHORS, author_data,
+                1, constants.TECHNIQUE_CODE_CONFIG_PARSER, source
+            )
+
+        if "license" in metadata:
+            license_value = metadata["license"]
+            license_info_spdx = detect_spdx_from_declared(license_value)
+            if not license_info_spdx:
+                license_info_spdx = detect_license_spdx(license_value, 'JSON')
+            if license_info_spdx:
+                license_data = {
+                    "value": license_value,
+                    "spdx_id": license_info_spdx.get('spdx_id'),
+                    "name": license_info_spdx.get('name'),
+                    "type": constants.LICENSE
+                }
+            else:
+                license_data = {"value": license_value, "type": constants.LICENSE}
+
+            metadata_result.add_result(
+                constants.CAT_LICENSE, license_data,
+                1, constants.TECHNIQUE_CODE_CONFIG_PARSER, source
+            )
+
+        if "keywords" in metadata:
+            for kw in re.split(r'[,\n]', metadata["keywords"]):
+                kw = kw.strip()
+                if kw:
+                    metadata_result.add_result(
+                        constants.CAT_KEYWORDS,
+                        {"value": kw, "type": constants.STRING},
+                        1, constants.TECHNIQUE_CODE_CONFIG_PARSER, source
+                    )
+
+        if "url" in metadata:
+            metadata_result.add_result(
+                constants.CAT_HOMEPAGE,
+                {"value": metadata["url"], "type": constants.URL},
+                1, constants.TECHNIQUE_CODE_CONFIG_PARSER, source
+            )
+
+        if "install_requires" in options:
+            for req in options["install_requires"].strip().splitlines():
+                req = req.strip()
+                if req:
+                    name, version = parse_dependency(req)
+                    if name:
+                        metadata_result.add_result(
+                            constants.CAT_REQUIREMENTS,
+                            {
+                                "value": req,
+                                "name": name,
+                                "version": version,
+                                "type": constants.SOFTWARE_DEPENDENCY,
+                                "dependency_type": constants.DEPENDENCY_TYPE_RUNTIME,
+                                "dependency_resolver": "python"
+                            },
+                            1, constants.TECHNIQUE_CODE_CONFIG_PARSER, source
+                        )
+
+        if "setup_requires" in options:
+                for req in options["setup_requires"].strip().splitlines():
+                    req = req.strip()
+                    if req:
+                        name, version = parse_dependency(req)
+                        if name:
+                            metadata_result.add_result(
+                                constants.CAT_REQUIREMENTS,
+                                {
+                                    "value": req,
+                                    "name": name,
+                                    "version": version,
+                                    "type": constants.SOFTWARE_DEPENDENCY,
+                                    "dependency_type": constants.DEPENDENCY_TYPE_DEVELOPMENT,
+                                    "dependency_resolver": "python"
+                                },
+                                1, constants.TECHNIQUE_CODE_CONFIG_PARSER, source
+                            )
+
+        if "python_requires" in options:
+            metadata_result.add_result(
+                constants.CAT_RUNTIME_PLATFORM,
+                {
+                    "value": f"Python{options['python_requires']}",
+                    "name": "Python",
+                    "version": options["python_requires"],
+                    "type": constants.STRING
+                },
+                1, constants.TECHNIQUE_CODE_CONFIG_PARSER, source
+            )
+
+        if "project_urls" in metadata:
+            lines = metadata["project_urls"].split('\n')
+            for line in lines:
+                if '=' in line:
+                    label, url_val = [part.strip() for part in line.split('=', 1)]
+                    label_lower = label.lower()
+
+                    if label_lower in ["documentation", "docs", "doc"]:
+                        metadata_result.add_result(
+                            constants.CAT_DOCUMENTATION,
+                            {"value": url_val, "type": constants.URL},
+                            1, constants.TECHNIQUE_CODE_CONFIG_PARSER, source
+                        )
+
+                    elif label_lower in ["repository", "source", "code"]:
+                         metadata_result.add_result(
+                            constants.CAT_CODE_REPOSITORY,
+                            {"value": url_val, "type": constants.URL},
+                            1, constants.TECHNIQUE_CODE_CONFIG_PARSER, source
+                        )
+
+    except Exception as e:
+        logging.error(f"Error parsing setup.cfg file {file_path}: {str(e)}")
+
+    return metadata_result
+
+def parse_dependency(dependency_str):
+    """Parse a dependency string to extract name and version."""
+    if not dependency_str:
+        return None, None
+
+    parts = re.split(r'(>=|<=|==|!=|>|<|~=)', dependency_str, 1)
+    name = parts[0].strip()
+    if len(parts) > 1:
+        version = ''.join(parts[1:])
+    else:
+        version = ""
+
+    version = re.sub(r'[\[\]]', '', version)
+
+    return name, version

src/somef/parser/toml_parser.py

@@ -118,7 +118,8 @@ def extract_common_version_field(data, metadata_result, source, file_type):
     For Project.toml: data["version"]
     """
     version_value = None
-
+    version_type = None
+
     if file_type == "cargo" and "package" in data and "version" in data["package"]:
         version_value = data["package"]["version"]
         version_type = constants.RELEASE

src/somef/process_files.py

@@ -25,6 +25,7 @@
 from .parser.publiccode_parser import parse_publiccode_file
 from .parser.codeowners_parser import parse_codeowners_file
 from .parser.conda_environment_parser import parse_conda_environment_file
+from .parser.setupcfg_parser import parse_setup_cfg
 from chardet import detect
 
 
@@ -277,7 +278,8 @@ def process_repository_files(repo_dir, metadata_result: Result, repo_type, owner
                         (filename.lower() == "environment.yml" or filename.lower() == "environment.yaml") or \
                         (filename.lower() == ".zenodo.json") or \
                         (filename.lower() == "cargo.toml" and repo_relative_path == ".") or (filename.lower() == "composer.json" and repo_relative_path == ".") or \
-                        (filename == "Project.toml" or (filename.lower()== "publiccode.yml" or filename.lower()== "publiccode.yaml") and repo_relative_path == "."):
+                        (filename == "Project.toml" or (filename.lower()== "publiccode.yml" or filename.lower()== "publiccode.yaml") and repo_relative_path == ".") or \
+                        filename.lower() == "setup.cfg":
                         if filename.lower() in parsed_build_files and repo_relative_path != ".":
                             logging.info(f"Ignoring secondary {filename} in {dir_path}")
                             continue
@@ -318,6 +320,8 @@ def process_repository_files(repo_dir, metadata_result: Result, repo_type, owner
                             metadata_result = parse_publiccode_file(os.path.join(dir_path, filename), metadata_result, build_file_url)
                         if filename.lower() == "environment.yml" or filename.lower() == "environment.yaml":
                             metadata_result = parse_conda_environment_file(os.path.join(dir_path, filename), metadata_result, build_file_url)
+                        if filename.lower() == "setup.cfg":
+                            metadata_result = parse_setup_cfg(os.path.join(dir_path, filename), metadata_result, build_file_url)
                         # if filename.lower() == ".zenodo":
                         #     metadata_result = parse_zenodo_file(os.path.join(dir_path, filename), metadata_result, build_file_url)
                         parsed_build_files.add(filename.lower())