From bf4a2232f49e46582ea5f81f8a96cfb83394d869 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Thibault=20Cl=C3=A9rice?= <leponteineptique@gmail.com>
Date: Fri, 23 Aug 2024 17:28:16 +0200
Subject: [PATCH] [WIP] Adding a catalog solution

---
 dapitains/local/collections.py | 55 ++++++++++++++++++++++++++++
 dapitains/local/ingester.py    | 67 ++++++++++++++++++++++++++++++++++
 2 files changed, 122 insertions(+)
 create mode 100644 dapitains/local/collections.py
 create mode 100644 dapitains/local/ingester.py

diff --git a/dapitains/local/collections.py b/dapitains/local/collections.py
new file mode 100644
index 0000000..dfc1a51
--- /dev/null
+++ b/dapitains/local/collections.py
@@ -0,0 +1,55 @@
+from dataclasses import dataclass, field
+from typing import List, Optional
+
+
+@dataclass
+class DublinCore:
+    term: str
+    value: str
+    language: Optional[str] = None
+
+    def json(self):
+        return {"property": f"http://purl.org/dc/terms/{self.term}", "value": self.value, "language": self.language}
+
+
+class Extension(DublinCore):
+    term: str
+    value: str
+    language: Optional[str] = None
+
+    def json(self):
+        return {"property": self.term, "value": self.value, "language": self.language}
+
+
+@dataclass
+class Collection:
+    identifier: str
+    title: str
+    description: Optional[str] = None
+    parents: List[str] = field(default_factory=list)
+    children: List[str] = field(default_factory=list)
+    dublin_core: List[DublinCore] = field(default_factory=list)
+    extension: List[Extension] = field(default_factory=list)
+    resource: bool = False
+    filepath: Optional[str] = None
+
+    def json(self):
+        return {
+            "identifier": self.identifier,
+            "title": self.title,
+            "description": self.description,
+            "parents": self.parents,
+            "children": self.children,
+            "dublin_core": self.dublin_core,
+            "extension": self.extension,
+            "resource": self.resource,
+            "filepath": self.filepath
+        }
+
+@dataclass
+class CitableUnit:
+    resource: str
+    reference: str
+    children: List[str] = field(default_factory=list)
+    dublin_core: List[DublinCore] = field(default_factory=list)
+    extension: List[Extension] = field(default_factory=list)
diff --git a/dapitains/local/ingester.py b/dapitains/local/ingester.py
new file mode 100644
index 0000000..e485fc0
--- /dev/null
+++ b/dapitains/local/ingester.py
@@ -0,0 +1,67 @@
+import os.path
+import re
+from typing import Dict, Optional
+import lxml.etree as ET
+from dapitains.local.collections import DublinCore, Extension, Collection, CitableUnit
+
+
+_re_tag = re.compile(r"[{}]")
+
+
+def parse_metadata(xml: ET.Element):
+    obj = {
+        "identifier": xml.attrib["identifier"],
+        "title": xml.xpath("./title[1]/text()")[0],
+        "description": (xml.xpath("./description[1]/text()") or [None])[0]
+    }
+    # Treat Dublin Core
+    dublin_core = []
+    for node in xml.xpath("./dublinCore/*"):
+        tag = node.tag.split("}")[-1]
+        language = node.attrib.get("{http://www.w3.org/XML/1998/namespace}lang")
+        text = node.text
+        dublin_core.append(DublinCore(tag, text, language))
+    if dublin_core:
+        obj["dublin_core"] = dublin_core
+
+    # Treat Extension
+    extensions = []
+    for node in xml.xpath("./extension/*"):
+        tag = _re_tag.sub("", node.tag)
+        language = node.attrib.get("{http://www.w3.org/XML/1998/namespace}lang")
+        text = node.text
+        extensions.append(Extension(tag, text, language))
+    if extensions:
+        obj["extensions"] = extensions
+
+    return obj
+
+
+def parse_collection(xml: ET.Element, basedir: str, tree: Dict[str, Collection]) -> Collection:
+    obj = parse_metadata(xml)
+    obj = Collection(**obj, resource=xml.tag == "resource")
+    tree[obj.identifier] = obj
+    for member in xml.xpath("./members/*"):
+        if member.xpath("./title"):
+            parse_collection(member, basedir, tree)
+            # ToDo: deal with children ?
+        else:
+            ingest_catalog(os.path.join(basedir, member.attrib["filepath"]), tree)
+    return obj
+
+
+def ingest_catalog(path: str, tree: Optional[Dict[str, Collection]] = None) -> Dict[str, Collection]:
+    """
+
+    :param path:
+    :return:
+
+    >>> ingest_catalog("../../tests/catalog/example-collection.xml")
+    """
+    xml = ET.parse(path)
+    current_dir = os.path.abspath(os.path.dirname(path))
+
+    root: ET.Element = xml.getroot()
+    tree = tree or {}
+    parse_collection(root, basedir=current_dir, tree=tree)
+    return tree