parses xjustiz and orders attachments

.idea/workspace.xml

@@ -2,8 +2,8 @@
 <project version="4">
   <component name="ChangeListManager">
     <list default="true" id="89f9fc3a-63ad-41cb-8d25-f07df352eb98" name="Changes" comment="">
-      <change beforePath="$PROJECT_DIR$/package.json" beforeDir="false" afterPath="$PROJECT_DIR$/package.json" afterDir="false" />
+      <change beforePath="$PROJECT_DIR$/.idea/workspace.xml" beforeDir="false" afterPath="$PROJECT_DIR$/.idea/workspace.xml" afterDir="false" />
-      <change beforePath="$PROJECT_DIR$/pnpm-lock.yaml" beforeDir="false" afterPath="$PROJECT_DIR$/pnpm-lock.yaml" afterDir="false" />
+      <change beforePath="$PROJECT_DIR$/src/lib/zip-processing.ts" beforeDir="false" afterPath="$PROJECT_DIR$/src/lib/zip-processing.ts" afterDir="false" />
       <change beforePath="$PROJECT_DIR$/src/routes/+page.svelte" beforeDir="false" afterPath="$PROJECT_DIR$/src/routes/+page.svelte" afterDir="false" />
     </list>
     <option name="SHOW_DIALOG" value="false" />
@@ -34,7 +34,7 @@
     "RunOnceActivity.typescript.service.memoryLimit.init": "true",
     "cidr.known.project.marker": "true",
     "codeWithMe.voiceChat.enabledByDefault": "false",
-    "git-widget-placeholder": "main",
+    "git-widget-placeholder": "features/zip-processing",
     "last_opened_file_path": "/home/rehlert/repos/juri-merger",
     "nodejs_package_manager_path": "pnpm",
     "settings.editor.selected.configurable": "preferences.pluginManager",
@@ -50,7 +50,7 @@
       <option name="presentableId" value="Default" />
       <updated>1781169139290</updated>
       <workItem from="1781169141138" duration="309000" />
-      <workItem from="1781251694220" duration="5995000" />
+      <workItem from="1781251694220" duration="7259000" />
     </task>
     <task id="LOCAL-00001" summary="adapts styling of rvg.legal">
       <option name="closed" value="true" />
@@ -68,7 +68,15 @@
       <option name="project" value="LOCAL" />
       <updated>1781257473070</updated>
     </task>
-    <option name="localTasksCounter" value="3" />
+    <task id="LOCAL-00003" summary="zips are getting extracted">
+      <option name="closed" value="true" />
+      <created>1781258303003</created>
+      <option name="number" value="00003" />
+      <option name="presentableId" value="LOCAL-00003" />
+      <option name="project" value="LOCAL" />
+      <updated>1781258303003</updated>
+    </task>
+    <option name="localTasksCounter" value="4" />
     <servers />
   </component>
   <component name="TypeScriptGeneratedFilesManager">
@@ -80,7 +88,8 @@
     <option name="CLEAR_INITIAL_COMMIT_MESSAGE" value="true" />
     <MESSAGE value="adapts styling of rvg.legal" />
     <MESSAGE value="adds dropzone to add zips" />
-    <option name="LAST_COMMIT_MESSAGE" value="adds dropzone to add zips" />
+    <MESSAGE value="zips are getting extracted" />
+    <option name="LAST_COMMIT_MESSAGE" value="zips are getting extracted" />
   </component>
   <component name="XDebuggerManager">
     <breakpoint-manager>

src/lib/zip-processing.ts

@@ -10,9 +10,8 @@ export type ZipAttachment = {
 };
 
 export type ProcessedZipArchive = {
-	name: string;
+	name: string
-	attachments: ZipAttachment[];
+	attachments: ZipAttachment[]
-	xjustizNachrichtXml: Uint8Array | null;
 };
 
 const IMAGE_EXTENSIONS = new Set([
@@ -30,6 +29,7 @@ const IMAGE_EXTENSIONS = new Set([
 ]);
 
 const ZIP_META_FILE_NAME = "xjustiz_nachricht.xml";
+const XML_TEXT_DECODER = new TextDecoder();
 
 const unzipArchive = (data: Uint8Array) =>
 	new Promise<Record<string, Uint8Array>>((resolve, reject) => {
@@ -56,6 +56,126 @@ const isPdf = (path: string) => getExtension(path) === ".pdf";
 
 const isImage = (path: string) => IMAGE_EXTENSIONS.has(getExtension(path));
 
+const normalizeKey = (value: string) => value.replaceAll("\\", "/").toLowerCase().trim();
+
+const matchesElementName = (element: Element, requestedName: string) => {
+	const requestedLocalName = requestedName.split(".").pop() ?? requestedName;
+
+	return (
+		element.localName === requestedName ||
+		element.localName === requestedLocalName ||
+		element.tagName === requestedName ||
+		element.tagName === requestedLocalName ||
+		element.tagName.endsWith(`:${requestedLocalName}`)
+	);
+};
+
+const getDirectChildByLocalName = (element: Element, localName: string) =>
+	Array.from(element.children).find((child) => matchesElementName(child, localName)) ?? null;
+
+const getFirstDescendantByLocalName = (root: Document | Element, localName: string) =>
+	Array.from(root.querySelectorAll("*")).find((element) => matchesElementName(element, localName)) ?? null;
+
+const getTextContentByPath = (root: Element, path: string[]) => {
+	let current: Element | null = root;
+
+	for (const localName of path) {
+		current = current ? getDirectChildByLocalName(current, localName) : null;
+
+		if (!current) {
+			return null;
+		}
+	}
+
+	const textContent = current.textContent?.trim();
+
+	return textContent ? textContent : null;
+};
+
+const parseXmlDocument = (xmlBytes: Uint8Array) => {
+	const parser = new DOMParser();
+	const document = parser.parseFromString(XML_TEXT_DECODER.decode(xmlBytes), "application/xml");
+
+	if (document.querySelector("parsererror")) {
+		return null;
+	}
+
+	return document;
+};
+
+const getXJustizDocumentNames = (xmlBytes: Uint8Array) => {
+	const document = parseXmlDocument(xmlBytes);
+
+	if (!document) {
+		return [];
+	}
+
+	const schriftgutobjekte = getFirstDescendantByLocalName(document, "schriftgutobjekte");
+
+	if (!schriftgutobjekte) {
+		return [];
+	}
+
+	const dokumentNodes = Array.from(schriftgutobjekte.children).filter((child) =>
+		matchesElementName(child, "dokument"),
+	);
+
+	return dokumentNodes
+		.map((dokument) => getTextContentByPath(dokument, ["xjustiz.fachspezifischeDaten", "datei", "dateiname"]))
+		.filter((name): name is string => Boolean(name));
+};
+
+const orderAttachmentsByDocumentNames = (attachments: ZipAttachment[], documentNames: string[]) => {
+	const buckets = new Map<string, ZipAttachment[]>();
+
+	for (const attachment of attachments) {
+		const keys = new Set([normalizeKey(attachment.path), normalizeKey(attachment.name)]);
+
+		for (const key of keys) {
+			const bucket = buckets.get(key) ?? [];
+			bucket.push(attachment);
+			buckets.set(key, bucket);
+		}
+	}
+
+	const usedAttachments = new Set<ZipAttachment>();
+	const orderedAttachments: ZipAttachment[] = [];
+
+	const takeAttachment = (key: string) => {
+		const bucket = buckets.get(key);
+
+		if (!bucket) {
+			return null;
+		}
+
+		const nextAttachment = bucket.find((attachment) => !usedAttachments.has(attachment)) ?? null;
+
+		if (nextAttachment) {
+			usedAttachments.add(nextAttachment);
+		}
+
+		return nextAttachment;
+	};
+
+	for (const documentName of documentNames) {
+		const normalizedName = normalizeKey(documentName);
+		const normalizedBaseName = normalizeKey(getBaseName(documentName));
+		const matchedAttachment = takeAttachment(normalizedName) ?? takeAttachment(normalizedBaseName);
+
+		if (matchedAttachment) {
+			orderedAttachments.push(matchedAttachment);
+		}
+	}
+
+	for (const attachment of attachments) {
+		if (!usedAttachments.has(attachment)) {
+			orderedAttachments.push(attachment);
+		}
+	}
+
+	return orderedAttachments;
+};
+
 export const extractZipArchive = async (file: File): Promise<ProcessedZipArchive> => {
 	const archiveBytes = new Uint8Array(await file.arrayBuffer());
 	const files = await unzipArchive(archiveBytes);
@@ -91,9 +211,12 @@ export const extractZipArchive = async (file: File): Promise<ProcessedZipArchive
 		}
 	}
 
+	const documentNames = xjustizNachrichtXml ? getXJustizDocumentNames(xjustizNachrichtXml) : [];
+	const orderedAttachments = orderAttachmentsByDocumentNames(attachments, documentNames);
+	const archiveName = documentNames[0] ? getBaseName(documentNames[0]) : file.name;
+
 	return {
-		name: file.name,
+		name: archiveName,
-		attachments,
+		attachments: orderedAttachments
-		xjustizNachrichtXml,
 	};
 };

src/routes/+page.svelte

@@ -20,7 +20,6 @@
 
         try {
             const processedZipArchive = await extractZipArchive(file);
-            processedZipArchive.xjustizNachrichtXml
             processedZipFiles = [...processedZipFiles, processedZipArchive];
         } catch (error) {
             console.error(`Failed to process ${file.name}`, error);
@@ -53,9 +52,6 @@
                         {#each processedZipFiles as file}
                             <li class="rounded-container bg-primary-50 px-4 py-3 text-[14px] font-medium text-primary-900">
                                 {file.name} - {file.attachments.length} relevante Datei{file.attachments.length === 1 ? "" : "en"}
-                                {#if file.xjustizNachrichtXml}
-                                    - xjustiz_nachricht.xml vorhanden
-                                {/if}
                             </li>
                         {/each}
                     </ul>