From 6018bd89a5ca3f26cbba912285c17fd1a0d9d2e3 Mon Sep 17 00:00:00 2001
From: JK <jovi.koikkara@matfyz.cuni.cz>
Date: Mon, 18 May 2026 14:21:03 +0200
Subject: [PATCH] feat: add Zotero cleanup feature to remove duplicate
 annotations from active note

---
 src/features/settings/settings.ts    |   6 ++
 src/features/settings/tab.ts         |  15 ++++
 src/features/zotero-cleanup/index.ts | 118 +++++++++++++++++++++++++++
 src/main.ts                          |   9 ++
 4 files changed, 148 insertions(+)
 create mode 100644 src/features/zotero-cleanup/index.ts

diff --git a/src/features/settings/settings.ts b/src/features/settings/settings.ts
index 707e007..9ec2623 100644
--- a/src/features/settings/settings.ts
+++ b/src/features/settings/settings.ts
@@ -60,6 +60,9 @@ export interface PluginSettings {
     debug: boolean;
     enabledCss: boolean;
     concurrency: string;
+
+    // Zotero Cleanup
+    zoteroCleanDirectories: string;
 }
 
 export const DEFAULT_SETTINGS: Required<PluginSettings> = {
@@ -116,4 +119,7 @@ export const DEFAULT_SETTINGS: Required<PluginSettings> = {
     debug: false,
     enabledCss: false,
     concurrency: "5",
+    
+    // Zotero Cleanup
+    zoteroCleanDirectories: "",
 };
diff --git a/src/features/settings/tab.ts b/src/features/settings/tab.ts
index 787c1e5..6cae806 100644
--- a/src/features/settings/tab.ts
+++ b/src/features/settings/tab.ts
@@ -273,6 +273,21 @@ export class MathSettingTab extends PluginSettingTab {
                 });
             });
 
+        containerEl.createEl("h2", { text: "Zotero Cleanup" });
+
+        new Setting(containerEl)
+            .setName("Directories to search")
+            .setDesc("Comma-separated list of directories to search recursively for Zotero annotations (e.g. 'Zotero,Notes/Readings').")
+            .addTextArea((textArea) => {
+                textArea
+                    .setValue(this.plugin.settings.zoteroCleanDirectories)
+                    .onChange(async (value) => {
+                        this.plugin.settings.zoteroCleanDirectories = value;
+                        await this.plugin.saveSettings();
+                    });
+                textArea.inputEl.setAttr("rows", 3);
+            });
+
 
         new Setting(containerEl).setName("Debug").setHeading();
         new Setting(containerEl)
diff --git a/src/features/zotero-cleanup/index.ts b/src/features/zotero-cleanup/index.ts
new file mode 100644
index 0000000..ce633b1
--- /dev/null
+++ b/src/features/zotero-cleanup/index.ts
@@ -0,0 +1,118 @@
+import { MarkdownView, Notice } from 'obsidian';
+import LatexReferencer from '../../main';
+
+export const processZoteroCleanup = async (plugin: LatexReferencer, activeView: MarkdownView) => {
+    const activeFile = activeView.file;
+    if (!activeFile) {
+        new Notice("No active file to process.");
+        return;
+    }
+
+    const dirsSetting = plugin.settings.zoteroCleanDirectories;
+    if (!dirsSetting || dirsSetting.trim().length === 0) {
+        new Notice("No Zotero cleanup directories configured in settings.");
+        return;
+    }
+
+    const directories = dirsSetting
+        .split(',')
+        .map(d => d.trim())
+        .filter(d => d.length > 0)
+        .map(d => d.replace(/^\/+|\/+$/g, '')); // remove leading/trailing slashes
+
+    if (directories.length === 0) {
+        new Notice("No valid Zotero cleanup directories configured.");
+        return;
+    }
+
+    const app = plugin.app;
+    const activeContent = activeView.editor.getValue();
+    const urlRegex = /zotero:\/\/[^\s\)<>"]+/g;
+    
+    const activeUrlsMatches = activeContent.match(urlRegex);
+    if (!activeUrlsMatches || activeUrlsMatches.length === 0) {
+        new Notice("No Zotero URLs found in the active note.");
+        return;
+    }
+
+    const activeUrls = Array.from(new Set(activeUrlsMatches));
+    
+    // Find all markdown files that match the directories, excluding the active file
+    const otherFiles = app.vault.getMarkdownFiles().filter(file => {
+        if (file.path === activeFile.path) return false;
+        return directories.some(dir => file.path === dir || file.path.startsWith(dir + '/'));
+    });
+
+    if (otherFiles.length === 0) {
+        new Notice("No other markdown files found in the specified directories.");
+        return;
+    }
+
+    const existingUrls = new Set<string>();
+    const notice = new Notice("Scanning directories for duplicate Zotero annotations...", 0);
+
+    try {
+        for (const file of otherFiles) {
+            const text = await app.vault.cachedRead(file);
+            for (const url of activeUrls) {
+                if (!existingUrls.has(url) && text.includes(url)) {
+                    existingUrls.add(url);
+                }
+            }
+            if (existingUrls.size === activeUrls.length) break;
+        }
+    } finally {
+        notice.hide();
+    }
+
+    if (existingUrls.size === 0) {
+        new Notice("No duplicate Zotero annotations found.");
+        return;
+    }
+
+    const lines = activeContent.split('\n');
+    const blocks: string[][] = [];
+    let currentBlock: string[] = [];
+
+    // Group lines into blocks. A new block starts when a line contains '<mark class='
+    for (let i = 0; i < lines.length; i++) {
+        const line = lines[i];
+        if (line.includes('<mark class=') && currentBlock.length > 0) {
+            blocks.push(currentBlock);
+            currentBlock = [];
+        }
+        currentBlock.push(line);
+    }
+    if (currentBlock.length > 0) {
+        blocks.push(currentBlock);
+    }
+
+    const newLines: string[] = [];
+    for (const block of blocks) {
+        const blockText = block.join('\n');
+        let shouldDelete = false;
+        
+        const isAnnotationBlock = block.some(l => l.includes('<mark class='));
+
+        if (isAnnotationBlock) {
+            for (const url of existingUrls) {
+                if (blockText.includes(url)) {
+                    shouldDelete = true;
+                    break;
+                }
+            }
+        }
+
+        if (!shouldDelete) {
+            newLines.push(...block);
+        }
+    }
+
+    const newContent = newLines.join('\n');
+    if (newContent !== activeContent) {
+        activeView.editor.setValue(newContent);
+        new Notice(`Removed ${existingUrls.size} duplicate Zotero annotation(s).`);
+    } else {
+        new Notice("No changes made.");
+    }
+};
diff --git a/src/main.ts b/src/main.ts
index ab6db99..f7e1c58 100644
--- a/src/main.ts
+++ b/src/main.ts
@@ -27,6 +27,7 @@ import { ExportConfigModal } from "./features/export-pdf/modal";
 import { checkAndFixCalloutMath } from 'utils/fixer';
 import { traverseFolder } from "./features/export-pdf/utils";
 import { SnippetManager } from 'features/snippets/manager';
+import { processZoteroCleanup } from 'features/zotero-cleanup';
 
 
 const isDev = process.env.NODE_ENV === "development";
@@ -49,6 +50,14 @@ export default class LatexReferencer extends Plugin {
 		this.addSettingTab(new MathSettingTab(this.app, this));
 
 		// Commands
+		this.addCommand({
+			id: 'remove-duplicate-zotero-annotations',
+			name: 'Remove duplicate Zotero annotations',
+			editorCallback: (editor: Editor, view: MarkdownView) => {
+				processZoteroCleanup(this, view);
+			}
+		});
+
 		this.addCommand({
 			id: 'fix-callout-equations',
 			name: 'Fix callout equations in active note',