Prevent duplicate files for existing Readwise books

src/main.ts

@@ -13,7 +13,7 @@ import {
 import * as zip from "@zip.js/zip.js";
 import { Md5 } from "ts-md5";
 import { getErrorDetailsFromResponse, ReadwiseSyncError } from "./errors";
-import { readwiseSyncFilePath } from "./paths";
+import { readwiseSyncFilePath, stablePathForBookExport } from "./paths";
 import { StatusBar } from "./status";
 
 // keep pluginVersion in sync with manifest.json
@@ -398,24 +398,25 @@ export default class ReadwisePlugin extends Plugin {
             }
           }
 
-          // write the actual files
-          let contentToSave = data.full_content ?? data.append_only_content;
-          if (contentToSave) {
-            // track the book
-            this.settings.booksIDsMap[processedFileName] = bookID;
-            // ensure the directory exists
-            await this.createDirForFile(processedFileName);
-            if (await this.fs.exists(processedFileName)) {
-              // if the file already exists we need to append content to existing one
-              const existingContent = await this.fs.read(processedFileName);
-              const existingContentHash = Md5.hashStr(existingContent).toString();
-              if (existingContentHash !== data.last_content_hash) {
-                // content has been modified (it differs from the previously exported full document)
-                contentToSave = existingContent.trimEnd() + "\n" + data.append_only_content;
-              }
-            }
-            await this.fs.write(processedFileName, contentToSave);
-          }
+          // write the actual files
+          let contentToSave = data.full_content ?? data.append_only_content;
+          if (contentToSave) {
+            const pathToSave = stablePathForBookExport(this.settings.booksIDsMap, bookID, processedFileName);
+            // track the book
+            this.settings.booksIDsMap[pathToSave] = bookID;
+            // ensure the directory exists
+            await this.createDirForFile(pathToSave);
+            if (await this.fs.exists(pathToSave)) {
+              // if the file already exists we need to append content to existing one
+              const existingContent = await this.fs.read(pathToSave);
+              const existingContentHash = Md5.hashStr(existingContent).toString();
+              if (existingContentHash !== data.last_content_hash) {
+                // content has been modified (it differs from the previously exported full document)
+                contentToSave = existingContent.trimEnd() + "\n" + data.append_only_content;
+              }
+            }
+            await this.fs.write(pathToSave, contentToSave);
+          }
 
           // save the entry in settings to ensure that it can be
           // retried later when deleted files are re-synced if

src/paths.ts

@@ -20,3 +20,58 @@ export function readwiseSyncFilePath(
 ): string {
   return normalizePath(`${readwiseDir}/${READWISE_SYNC_FILENAME}.md`);
 }
+
+export type BooksIDsMap = { [filePath: string]: string };
+
+function isFullDocumentContentPath(path: string): boolean {
+  return path.split("/").includes("Full Document Contents");
+}
+
+function hasNumericSuffixBeforeExtension(path: string): boolean {
+  return /\s\(\d+\)\.md$/i.test(path);
+}
+
+function compareCandidatePaths(left: string, right: string): number {
+  const leftHasNumericSuffix = hasNumericSuffixBeforeExtension(left);
+  const rightHasNumericSuffix = hasNumericSuffixBeforeExtension(right);
+
+  if (leftHasNumericSuffix !== rightHasNumericSuffix) {
+    return leftHasNumericSuffix ? 1 : -1;
+  }
+
+  if (left.length !== right.length) {
+    return left.length - right.length;
+  }
+
+  return left.localeCompare(right);
+}
+
+/**
+ * Resolve the vault path to update for an exported Readwise book.
+ *
+ * Readwise export artifacts can occasionally contain a new filename for a book
+ * that already exists in the vault, such as `Book Title (467).md`. When the
+ * settings map already knows a path for the same book ID, keep updating that
+ * stable path instead of creating a duplicate suffixed file.
+ */
+export function stablePathForBookExport(
+  booksIDsMap: BooksIDsMap,
+  bookID: string,
+  exportedPath: string,
+): string {
+  if (!bookID || booksIDsMap[exportedPath] === bookID) {
+    return exportedPath;
+  }
+
+  const exportedIsFullDocumentContent = isFullDocumentContentPath(exportedPath);
+  const candidates = Object.keys(booksIDsMap).filter((path) => {
+    return booksIDsMap[path] === bookID
+      && isFullDocumentContentPath(path) === exportedIsFullDocumentContent;
+  });
+
+  if (!candidates.length) {
+    return exportedPath;
+  }
+
+  return candidates.sort(compareCandidatePaths)[0];
+}

tests/paths.test.js

@@ -1,6 +1,6 @@
 const assert = require("node:assert/strict");
 const { test } = require("node:test");
-const { readwiseSyncFilePath } = require("../.test-build/src/paths");
+const { readwiseSyncFilePath, stablePathForBookExport } = require("../.test-build/src/paths");
 
 // Faithful copy of Obsidian's normalizePath (the real one lives in the
 // "obsidian" runtime, which is not available under `node --test`).
@@ -35,3 +35,66 @@ test("sync file path matches the processed entry for default and nested base fol
     assert.equal(readwiseSyncFilePath(dir, normalizePath), processedSyncName(dir));
   }
 });
+
+test("book export reuses an existing path for the same book ID", () => {
+  const booksIDsMap = {
+    "Readwise/Books/Seven Brief Lessons on Physics.md": "book:35774970",
+  };
+
+  assert.equal(
+    stablePathForBookExport(
+      booksIDsMap,
+      "book:35774970",
+      "Readwise/Books/Seven Brief Lessons on Physics (467).md",
+    ),
+    "Readwise/Books/Seven Brief Lessons on Physics.md",
+  );
+});
+
+test("book export preserves a user-renamed path for the same book ID", () => {
+  const booksIDsMap = {
+    "Readwise/Books/Favorites/Physics notes.md": "book:35774970",
+  };
+
+  assert.equal(
+    stablePathForBookExport(
+      booksIDsMap,
+      "book:35774970",
+      "Readwise/Books/Seven Brief Lessons on Physics (467).md",
+    ),
+    "Readwise/Books/Favorites/Physics notes.md",
+  );
+});
+
+test("book export prefers a non-suffixed candidate when duplicates already exist", () => {
+  const booksIDsMap = {
+    "Readwise/Books/Seven Brief Lessons on Physics (467).md": "book:35774970",
+    "Readwise/Books/Seven Brief Lessons on Physics.md": "book:35774970",
+    "Readwise/Books/Seven Brief Lessons on Physics (485).md": "book:35774970",
+  };
+
+  assert.equal(
+    stablePathForBookExport(
+      booksIDsMap,
+      "book:35774970",
+      "Readwise/Books/Seven Brief Lessons on Physics (486).md",
+    ),
+    "Readwise/Books/Seven Brief Lessons on Physics.md",
+  );
+});
+
+test("book export keeps full document content paths separate from highlight paths", () => {
+  const booksIDsMap = {
+    "Readwise/Books/Seven Brief Lessons on Physics.md": "book:35774970",
+    "Readwise/Full Document Contents/Books/Seven Brief Lessons on Physics.md": "book:35774970",
+  };
+
+  assert.equal(
+    stablePathForBookExport(
+      booksIDsMap,
+      "book:35774970",
+      "Readwise/Full Document Contents/Books/Seven Brief Lessons on Physics (467).md",
+    ),
+    "Readwise/Full Document Contents/Books/Seven Brief Lessons on Physics.md",
+  );
+});