ose-plugin-hierarchical-spreadsheets 0.2.5__tar.gz

ose_plugin_hierarchical_spreadsheets-0.2.5/PKG-INFO

@@ -0,0 +1,34 @@
+Metadata-Version: 2.4
+Name: ose-plugin-hierarchical-spreadsheets
+Version: 0.2.5
+Summary: OntoSpreadEd plugin for hierarchical spreadsheet generation
+Requires-Python: >=3.12
+Description-Content-Type: text/markdown
+Requires-Dist: ose-core==0.2.5
+
+# OSE Plugin: Hierarchical Spreadsheets
+
+OntoSpreadEd plugin for generating hierarchical spreadsheet exports during ontology releases.
+
+## Description
+
+This plugin adds a release step that generates hierarchical spreadsheet representations of ontologies. It exports ontology structures in a format that preserves parent-child relationships in an easy-to-read tabular format.
+
+## Installation
+
+```bash
+pip install ose-plugin-hierarchical-spreadsheets
+```
+
+## Requirements
+
+- Python 3.12+
+- ose-core
+
+## Usage
+
+Add the `GenerateHierarchicalSpreadsheetReleaseStep` to your release script configuration to include hierarchical spreadsheet generation in your release pipeline.
+
+## License
+
+LGPL-3.0-or-later

ose_plugin_hierarchical_spreadsheets-0.2.5/README.md

@@ -0,0 +1,26 @@
+# OSE Plugin: Hierarchical Spreadsheets
+
+OntoSpreadEd plugin for generating hierarchical spreadsheet exports during ontology releases.
+
+## Description
+
+This plugin adds a release step that generates hierarchical spreadsheet representations of ontologies. It exports ontology structures in a format that preserves parent-child relationships in an easy-to-read tabular format.
+
+## Installation
+
+```bash
+pip install ose-plugin-hierarchical-spreadsheets
+```
+
+## Requirements
+
+- Python 3.12+
+- ose-core
+
+## Usage
+
+Add the `GenerateHierarchicalSpreadsheetReleaseStep` to your release script configuration to include hierarchical spreadsheet generation in your release pipeline.
+
+## License
+
+LGPL-3.0-or-later

ose_plugin_hierarchical_spreadsheets-0.2.5/pyproject.toml

@@ -0,0 +1,23 @@
+[build-system]
+requires = ["setuptools >= 61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "ose-plugin-hierarchical-spreadsheets"
+version = "0.2.5"
+description = "OntoSpreadEd plugin for hierarchical spreadsheet generation"
+readme = "README.md"
+requires-python = ">=3.12"
+dependencies = [
+    "ose-core==0.2.5",
+]
+
+[project.entry-points.'ose.plugins']
+hierarchical_spreadsheets = "ose_plugin_hierarchical_spreadsheets:HierarchicalSpreadsheetsPlugin"
+
+[tool.uv.sources]
+ose-core = { workspace = true }
+
+[tool.setuptools.packages.find]
+where = ["src"]
+

ose_plugin_hierarchical_spreadsheets-0.2.5/setup.cfg

@@ -0,0 +1,4 @@
+[egg_info]
+tag_build = 
+tag_date = 0
+

ose_plugin_hierarchical_spreadsheets-0.2.5/src/ose_plugin_hierarchical_spreadsheets/GenerateHierarchicalSpreadsheetReleaseStep.py

@@ -0,0 +1,185 @@
+import dataclasses
+import re
+from typing import List, Optional, Dict, Callable, Any, Tuple
+
+import openpyxl
+import pyhornedowl
+from flask_github import GitHub
+from flask_sqlalchemy import SQLAlchemy
+from typing_extensions import Self
+
+from ose.release.ReleaseStep import ReleaseStep
+from ose.model.ReleaseScript import ReleaseScript, ReleaseScriptFile
+from ose.model.Result import Result
+from ose.services.ConfigurationService import ConfigurationService
+from ose.utils import letters
+from ose.utils.github import parse_spreadsheet
+
+
+@dataclasses.dataclass
+class Node:
+    item: str
+    label: str
+    definition: str
+    children: List[Self] = dataclasses.field(default_factory=list)
+    parent: Optional[Self] = None
+    annotations: Dict[str, str] = dataclasses.field(default_factory=dict)
+
+    def to_plain(self):
+        plain_children = []
+        for c in self.children:
+            plain_children.append(c.to_plain())
+
+        return dict(item=self.item, label=self.label, definition=self.definition, children=plain_children)
+
+    def height(self) -> int:
+        return max((c.height() for c in self.children), default=0) + 1
+
+    def recurse(self, fn: Callable[[Self], Any]):
+        fn(self)
+        for c in self.children:
+            c.recurse(fn)
+
+
+def form_tree(edges: List[Tuple[Tuple[str, str | None, str | None], Optional[str]]]) -> List[Node]:
+    all_nodes = set(n for n, _ in edges)
+    item_to_node = dict((c, Node(item=c, label=lbl if lbl is not None else c, definition=d or "<no definition>")) for (c, lbl, d) in all_nodes)
+
+    for (child, _, _), parent in edges:
+        if parent is None:
+            continue
+
+        if child == parent:
+            continue
+
+        child_node = item_to_node[child]
+        parent_node = item_to_node.setdefault(parent, Node(item=parent, label=parent, definition=""))
+
+        child_node.parent = parent_node
+        parent_node.children.append(child_node)
+
+    return [n for n in item_to_node.values() if n.parent is None]
+
+
+class GenerateHierarchicalSpreadsheetReleaseStep(ReleaseStep):
+
+    def __init__(self, db: SQLAlchemy, gh: GitHub, release_script: ReleaseScript, release_id: int, tmp: str,
+                 config: ConfigurationService, *, included_files: Dict[str, str]) -> None:
+        super().__init__(db, gh, release_script, release_id, tmp, config)
+
+        self._included_files = included_files
+
+    def run(self) -> bool:
+        result = Result(())
+
+        files = [f for k, f in self._release_script.files.items() if k in self._included_files]
+        self._total_items = len(files)
+
+        for file in files:
+            self._next_item(item=file.target.file, message="Generating hierarchical spreadsheet for")
+
+            hierarchies, ontology = self.build_hierarchy(file)
+
+            wb = openpyxl.Workbook()
+            assert wb.active is not None
+            sheet = wb.active
+
+            height = max(h.height() for h in hierarchies)
+            annotations = list({k for h in hierarchies for k in h.annotations.keys()})
+
+            sheet.append(["ID", "Label"] + [""] * (height - 1) + ["Definition"] + annotations)
+
+            def write_line(n: Node, depth: int) -> None:
+                sheet.append([ontology.get_id_for_iri(n.item)] +
+                             [""] * depth +
+                             [n.label] + [""] * (height - depth - 1) +
+                             [n.definition] +
+                             [n.annotations.get(a, None) for a in annotations])
+
+                for child in n.children:
+                    write_line(child, depth + 1)
+
+            for hierarchy in hierarchies:
+                write_line(hierarchy, 0)
+
+            [path, name] = file.target.file.rsplit("/", 1)
+            sub_name = name.rsplit(".", 1)[0]
+            sub_name = re.sub(f"^{self._release_script.short_repository_name}[_]?", "", sub_name, flags=re.IGNORECASE)
+
+            file_name = f"{self._release_script.short_repository_name}-{sub_name}-hierarchy.xlsx"
+
+            wb.save(self._local_name(file_name))
+
+            self._store_artifact(self._local_name(file_name), f"{path}/{file_name}")
+
+        result.warnings = []
+        self._set_release_result(result)
+        return result.ok()
+
+    @classmethod
+    def name(cls) -> str:
+        return "HIERARCHICAL_SPREADSHEETS"
+
+    def build_hierarchy(self, file: ReleaseScriptFile) -> Tuple[List[Node], pyhornedowl.PyIndexedOntology]:
+        # Excel files to extract annotations
+        excel_files: List[str]
+        release_file: str
+
+        excel_files = [self._local_name(s.file) for s in file.sources]
+        release_file = next((a.local_path for a in self._artifacts()
+                             if a.target_path == file.target.file and a.kind == 'final'),
+                            self._local_name(file.target.file))
+
+        ontology = pyhornedowl.open_ontology_from_file(release_file)
+
+        for p, d in self._repo_config.prefixes.items():
+            ontology.prefix_mapping.add_prefix(p, d)
+
+        classes = [(c, ontology.get_annotation(c, "http://www.w3.org/2000/01/rdf-schema#label"),
+                    ontology.get_annotation(c, "http://purl.obolibrary.org/obo/IAO_0000115")) for c in
+                   ontology.get_classes()]
+        child_parent: List[Tuple[Tuple[str, Optional[str], Optional[str]], Optional[str]]] = []
+        for c in classes:
+            for p in ontology.get_superclasses(c[0]):
+                child_parent.append((c, p))
+            else:
+                child_parent.append((c, None))
+
+        # child_parent = [(c, p) for c in classes for p in ontology.get_superclasses(c[0])]
+        hierarchies = form_tree(child_parent)
+
+        # If we do not collapse (or do not import) imported ontologies a root will always contain no label
+        # We remove these roots as they only indicate where the subontology should be mounted in the overall ontology
+        hierarchies = [c for h in hierarchies for c in h.children]
+
+        for excel_file in excel_files:
+            with open(excel_file, "rb") as f:
+                file_data = f.read()
+
+            rows, header = parse_spreadsheet(file_data)
+
+            data = dict((r["ID"], r) for r in rows if "ID" in r)
+
+            def annotate(n: Node):
+                id = ontology.get_id_for_iri(n.item)
+                if id is None:
+                    return
+                
+                fields = {
+                    "comment": "Comment",
+                    "subontology": "Sub-ontology",
+                    "examples": "Examples",
+                    "synonyms": "Synonyms",
+                    "crossreference": "Cross reference",
+                    "informaldefinition": "Informal definition",
+                }
+                for field_key, field in fields.items():
+                    node_data = data.get(id, dict())
+                    key = next((k for k in node_data.keys() if letters(k) == field_key), None)
+                    if key is not None and n.annotations.get(field, None) is None:
+                        n.annotations[field] = node_data[key]
+
+            for h in hierarchies:
+                h.recurse(annotate)
+
+        return hierarchies, ontology

ose_plugin_hierarchical_spreadsheets-0.2.5/src/ose_plugin_hierarchical_spreadsheets/__init__.py

@@ -0,0 +1,14 @@
+from ose.model.Plugin import Plugin
+from .GenerateHierarchicalSpreadsheetReleaseStep import GenerateHierarchicalSpreadsheetReleaseStep
+
+
+
+HierarchicalSpreadsheetsPlugin = Plugin(
+    id="org.bssofoundry.hierarchicalspreadsheets",
+    name="HierarchicalSpreadsheet Plugin",
+    version="0.1.0",
+    description="Plugin to generate hierarchical spreadsheets during release.",
+    contents=[
+      GenerateHierarchicalSpreadsheetReleaseStep,
+    ],
+)

ose_plugin_hierarchical_spreadsheets-0.2.5/src/ose_plugin_hierarchical_spreadsheets.egg-info/PKG-INFO

@@ -0,0 +1,34 @@
+Metadata-Version: 2.4
+Name: ose-plugin-hierarchical-spreadsheets
+Version: 0.2.5
+Summary: OntoSpreadEd plugin for hierarchical spreadsheet generation
+Requires-Python: >=3.12
+Description-Content-Type: text/markdown
+Requires-Dist: ose-core==0.2.5
+
+# OSE Plugin: Hierarchical Spreadsheets
+
+OntoSpreadEd plugin for generating hierarchical spreadsheet exports during ontology releases.
+
+## Description
+
+This plugin adds a release step that generates hierarchical spreadsheet representations of ontologies. It exports ontology structures in a format that preserves parent-child relationships in an easy-to-read tabular format.
+
+## Installation
+
+```bash
+pip install ose-plugin-hierarchical-spreadsheets
+```
+
+## Requirements
+
+- Python 3.12+
+- ose-core
+
+## Usage
+
+Add the `GenerateHierarchicalSpreadsheetReleaseStep` to your release script configuration to include hierarchical spreadsheet generation in your release pipeline.
+
+## License
+
+LGPL-3.0-or-later

ose_plugin_hierarchical_spreadsheets-0.2.5/src/ose_plugin_hierarchical_spreadsheets.egg-info/SOURCES.txt

@@ -0,0 +1,10 @@
+README.md
+pyproject.toml
+src/ose_plugin_hierarchical_spreadsheets/GenerateHierarchicalSpreadsheetReleaseStep.py
+src/ose_plugin_hierarchical_spreadsheets/__init__.py
+src/ose_plugin_hierarchical_spreadsheets.egg-info/PKG-INFO
+src/ose_plugin_hierarchical_spreadsheets.egg-info/SOURCES.txt
+src/ose_plugin_hierarchical_spreadsheets.egg-info/dependency_links.txt
+src/ose_plugin_hierarchical_spreadsheets.egg-info/entry_points.txt
+src/ose_plugin_hierarchical_spreadsheets.egg-info/requires.txt
+src/ose_plugin_hierarchical_spreadsheets.egg-info/top_level.txt

ose_plugin_hierarchical_spreadsheets-0.2.5/src/ose_plugin_hierarchical_spreadsheets.egg-info/dependency_links.txt

@@ -0,0 +1 @@
+

ose_plugin_hierarchical_spreadsheets-0.2.5/src/ose_plugin_hierarchical_spreadsheets.egg-info/entry_points.txt

@@ -0,0 +1,2 @@
+[ose.plugins]
+hierarchical_spreadsheets = ose_plugin_hierarchical_spreadsheets:HierarchicalSpreadsheetsPlugin

ose_plugin_hierarchical_spreadsheets-0.2.5/src/ose_plugin_hierarchical_spreadsheets.egg-info/requires.txt

@@ -0,0 +1 @@
+ose-core==0.2.5

ose_plugin_hierarchical_spreadsheets-0.2.5/src/ose_plugin_hierarchical_spreadsheets.egg-info/top_level.txt

@@ -0,0 +1 @@
+ose_plugin_hierarchical_spreadsheets