paper-manager 0.10.0 → 0.10.2

package/dist/commands/dep.js

@@ -6,15 +6,21 @@ const KNOWN_DEPS = new Set(["opendataloader"]);
 export function createDepCommand() {
     const dep = new Command("dep").description("Manage external dependencies");
     dep
-        .command("check <dep>")
+        .command("check [dep]")
         .description("Check if an external dependency is available")
         .action(async (depName) => {
-        if (!KNOWN_DEPS.has(depName)) {
-            log.error(`Unknown dependency: ${depName}`);
-            log.step(`Available: ${[...KNOWN_DEPS].join(", ")}`);
-            process.exit(1);
+        if (depName != null) {
+            if (!KNOWN_DEPS.has(depName)) {
+                log.error(`Unknown dependency: ${depName}`);
+                log.step(`Available: ${[...KNOWN_DEPS].join(", ")}`);
+                process.exit(1);
+            }
+            if (depName === "opendataloader") {
+                await checkOpendataLoader();
+            }
         }
-        if (depName === "opendataloader") {
+        else {
+            // Check all known dependencies
             await checkOpendataLoader();
         }
     });

package/dist/commands/knowledge-base.js

@@ -6,6 +6,7 @@ import * as projectKb from "../db/project/knowledge-bases.js";
 import * as projectLit from "../db/project/literatures.js";
 import * as userKb from "../db/user/knowledge-bases.js";
 import * as userLit from "../db/user/literatures.js";
+import { removeImageDir } from "../extractor/markdown.js";
 import { log } from "../logger.js";
 import { queryVectorStore } from "../vector-store/index.js";
 function resolveKnowledgeBase(id) {
@@ -136,7 +137,7 @@ export function createKnowledgeBaseCommand() {
         const kbOps = scope === "project" ? projectKb : userKb;
         // 1. Get all literatures in this KB
         const literatures = litOps.getLiteraturesByKnowledgeBaseId(id);
-        // 2. Delete stored files
+        // 2. Delete stored files and image directories
         const filesDir = getFilesDir(baseDir);
         if (fs.existsSync(filesDir)) {
             for (const lit of literatures) {
@@ -145,6 +146,7 @@ export function createKnowledgeBaseCommand() {
                         fs.unlinkSync(path.join(filesDir, entry.name));
                     }
                 }
+                removeImageDir(filesDir, lit.id);
             }
         }
         // 3. Delete literatures from DB

package/dist/commands/literature.js

@@ -9,7 +9,7 @@ import * as projectLit from "../db/project/literatures.js";
 import * as userKb from "../db/user/knowledge-bases.js";
 import * as userLit from "../db/user/literatures.js";
 import { extractContent, extractPdfMetadata } from "../extractor/index.js";
-import { convertPdfToMarkdown, isOpendataLoaderAvailable } from "../extractor/markdown.js";
+import { convertPdfToMarkdown, isOpendataLoaderAvailable, removeImageDir, saveConvertResult, } from "../extractor/markdown.js";
 import { log } from "../logger.js";
 import { splitDocuments } from "../text-splitter.js";
 import { addDocuments, createVectorStore } from "../vector-store/index.js";
@@ -97,9 +97,9 @@ export function createLiteratureCommand() {
         fs.copyFileSync(absolutePath, path.join(filesDir, `${literature.id}${ext}`));
         // Convert PDF to Markdown if opendataloader is available
         if (isPdf && (await isOpendataLoaderAvailable())) {
-            const markdown = await convertPdfToMarkdown(absolutePath);
-            if (markdown) {
-                fs.writeFileSync(path.join(filesDir, `${literature.id}.md`), markdown, "utf-8");
+            const result = await convertPdfToMarkdown(absolutePath);
+            if (result) {
+                saveConvertResult(filesDir, literature.id, result);
                 log.step("Converted to Markdown via opendataloader-pdf.");
             }
         }
@@ -164,12 +164,12 @@ export function createLiteratureCommand() {
             process.exit(1);
         }
         log.info("Converting PDF to Markdown...");
-        const markdown = await convertPdfToMarkdown(path.join(filesDir, pdfFile));
-        if (!markdown) {
+        const result = await convertPdfToMarkdown(path.join(filesDir, pdfFile));
+        if (!result) {
             log.error("Conversion failed.");
             process.exit(1);
         }
-        fs.writeFileSync(mdPath, markdown, "utf-8");
+        saveConvertResult(filesDir, id, result);
         log.success(`Markdown saved: ${id}.md`);
     });
     // ─── lit remove ────────────────────────────────────────────
@@ -190,7 +190,7 @@ export function createLiteratureCommand() {
             log.error(`Literature not found: ${id}`);
             process.exit(1);
         }
-        // Delete stored file (find by pattern <id>.*)
+        // Delete stored files and image directory
         const filesDir = getFilesDir(baseDir);
         if (fs.existsSync(filesDir)) {
             for (const entry of fs.readdirSync(filesDir, { withFileTypes: true })) {
@@ -198,6 +198,7 @@ export function createLiteratureCommand() {
                     fs.unlinkSync(path.join(filesDir, entry.name));
                 }
             }
+            removeImageDir(filesDir, id);
         }
         // Delete literature record
         litOps.deleteLiterature(id);

package/dist/extractor/markdown.js

@@ -1,5 +1,5 @@
 import { execFile } from "node:child_process";
-import { mkdirSync, readdirSync, readFileSync, rmSync } from "node:fs";
+import { existsSync, mkdirSync, readdirSync, readFileSync, rmSync, writeFileSync } from "node:fs";
 import { tmpdir } from "node:os";
 import * as path from "node:path";
 /**
@@ -14,7 +14,7 @@ export async function isOpendataLoaderAvailable() {
 }
 /**
  * Convert a PDF file to Markdown using opendataloader-pdf.
- * Returns the markdown content on success, or null on failure.
+ * Returns the markdown content and extracted images on success, or null on failure.
  */
 export async function convertPdfToMarkdown(pdfPath) {
     const outDir = path.join(tmpdir(), `odl-${Date.now()}`);
@@ -26,10 +26,19 @@ export async function convertPdfToMarkdown(pdfPath) {
             format: "markdown",
             quiet: true,
         });
-        const mdFile = readdirSync(outDir).find((f) => f.endsWith(".md"));
+        const files = readdirSync(outDir);
+        const mdFile = files.find((f) => f.endsWith(".md"));
         if (!mdFile)
             return null;
-        return readFileSync(path.join(outDir, mdFile), "utf-8");
+        const markdown = readFileSync(path.join(outDir, mdFile), "utf-8");
+        const imageExtensions = new Set([".png", ".jpg", ".jpeg", ".gif", ".svg", ".webp", ".bmp"]);
+        const images = new Map();
+        for (const file of files) {
+            if (imageExtensions.has(path.extname(file).toLowerCase())) {
+                images.set(file, readFileSync(path.join(outDir, file)));
+            }
+        }
+        return { markdown, images };
     }
     catch {
         return null;
@@ -38,6 +47,33 @@ export async function convertPdfToMarkdown(pdfPath) {
         rmSync(outDir, { recursive: true, force: true });
     }
 }
+/**
+ * Save a ConvertResult to disk: writes the markdown file and any extracted images.
+ * Images are stored in `filesDir/<id>/` and image references in the markdown are
+ * rewritten to use the `<id>/` prefix.
+ */
+export function saveConvertResult(filesDir, id, result) {
+    let { markdown } = result;
+    if (result.images.size > 0) {
+        const imageSubDir = path.join(filesDir, id);
+        mkdirSync(imageSubDir, { recursive: true });
+        for (const [filename, data] of result.images) {
+            writeFileSync(path.join(imageSubDir, filename), data);
+            // Rewrite image/link references: ](filename) → ](<id>/filename)
+            markdown = markdown.replaceAll(`](${filename})`, `](${id}/${filename})`);
+        }
+    }
+    writeFileSync(path.join(filesDir, `${id}.md`), markdown, "utf-8");
+}
+/**
+ * Remove the extracted images directory for a literature, if it exists.
+ */
+export function removeImageDir(filesDir, id) {
+    const imageDir = path.join(filesDir, id);
+    if (existsSync(imageDir)) {
+        rmSync(imageDir, { recursive: true, force: true });
+    }
+}
 // ─── Internal ────────────────────────────────────────────
 let cachedAvailability;
 async function detectAvailability() {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "paper-manager",
-  "version": "0.10.0",
+  "version": "0.10.2",
   "description": "A paper management system.",
   "keywords": [],
   "homepage": "https://github.com/EurFelux/paper-manager",