paper-manager 0.10.2 → 0.10.4

package/dist/commands/dep.js

@@ -31,6 +31,7 @@ async function checkOpendataLoader() {
     const status = await checkOpendataLoaderStatus();
     const pkgIcon = status.packageInstalled ? chalk.green("✔") : chalk.red("✖");
     const javaIcon = status.javaAvailable ? chalk.green("✔") : chalk.red("✖");
+    const hybridIcon = status.hybridBackendAvailable ? chalk.green("✔") : chalk.dim("○");
     log.plain(`  ${pkgIcon} @opendataloader/pdf package`);
     if (status.javaAvailable) {
         log.plain(`  ${javaIcon} Java runtime (${status.javaVersion})`);
@@ -38,9 +39,18 @@ async function checkOpendataLoader() {
     else {
         log.plain(`  ${javaIcon} Java runtime (not found)`);
     }
+    if (status.hybridBackendAvailable) {
+        log.plain(`  ${hybridIcon} Hybrid backend (localhost:5002)`);
+    }
+    else {
+        log.plain(`  ${hybridIcon} Hybrid backend (not running, optional)`);
+    }
     log.newline();
     if (status.packageInstalled && status.javaAvailable) {
         log.success("opendataloader-pdf is ready.");
+        if (status.hybridBackendAvailable) {
+            log.step("Hybrid mode enabled — using docling backend for improved extraction.");
+        }
     }
     else {
         log.error("opendataloader-pdf is not available.");

package/dist/extractor/markdown.js

@@ -1,5 +1,6 @@
 import { execFile } from "node:child_process";
 import { existsSync, mkdirSync, readdirSync, readFileSync, rmSync, writeFileSync } from "node:fs";
+import { request } from "node:http";
 import { tmpdir } from "node:os";
 import * as path from "node:path";
 /**
@@ -21,23 +22,20 @@ export async function convertPdfToMarkdown(pdfPath) {
     mkdirSync(outDir, { recursive: true });
     try {
         const { convert } = await import("@opendataloader/pdf");
+        const hybridAvailable = await isHybridBackendAvailable();
         await convert([pdfPath], {
             outputDir: outDir,
             format: "markdown",
             quiet: true,
+            ...(hybridAvailable ? { hybrid: "docling-fast", hybridFallback: true } : {}),
         });
-        const files = readdirSync(outDir);
-        const mdFile = files.find((f) => f.endsWith(".md"));
+        const mdFile = readdirSync(outDir).find((f) => f.endsWith(".md"));
         if (!mdFile)
             return null;
         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)));
-            }
-        }
+        collectImages(outDir, outDir, imageExtensions, images);
         return { markdown, images };
     }
     catch {
@@ -57,10 +55,11 @@ export function saveConvertResult(filesDir, id, 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}/${filename})`);
+        for (const [relPath, data] of result.images) {
+            const basename = path.basename(relPath);
+            writeFileSync(path.join(imageSubDir, basename), data);
+            // Rewrite image/link references: ](old/path.png) → ](<id>/basename.png)
+            markdown = markdown.replaceAll(`](${relPath})`, `](${id}/${basename})`);
         }
     }
     writeFileSync(path.join(filesDir, `${id}.md`), markdown, "utf-8");
@@ -75,6 +74,18 @@ export function removeImageDir(filesDir, id) {
     }
 }
 // ─── Internal ────────────────────────────────────────────
+/** Recursively collect image files under `dir`, keyed by path relative to `root`. */
+function collectImages(dir, root, extensions, out) {
+    for (const entry of readdirSync(dir, { withFileTypes: true })) {
+        const full = path.join(dir, entry.name);
+        if (entry.isDirectory()) {
+            collectImages(full, root, extensions, out);
+        }
+        else if (extensions.has(path.extname(entry.name).toLowerCase())) {
+            out.set(path.relative(root, full), readFileSync(full));
+        }
+    }
+}
 let cachedAvailability;
 async function detectAvailability() {
     const [hasPackage, hasJava] = await Promise.all([checkPackage(), checkJava()]);
@@ -101,13 +112,36 @@ async function checkJava() {
  * Detailed availability check for the `dep check` command.
  */
 export async function checkOpendataLoaderStatus() {
-    const [packageInstalled, javaResult] = await Promise.all([checkPackage(), getJavaVersion()]);
+    const [packageInstalled, javaResult, hybridBackendAvailable] = await Promise.all([
+        checkPackage(),
+        getJavaVersion(),
+        isHybridBackendAvailable(),
+    ]);
     return {
         packageInstalled,
         javaAvailable: javaResult !== null,
         javaVersion: javaResult,
+        hybridBackendAvailable,
     };
 }
+const HYBRID_BACKEND_URL = "http://localhost:5002";
+const HYBRID_PROBE_TIMEOUT_MS = 1500;
+/** Check if the opendataloader hybrid backend is reachable at localhost:5002. */
+function isHybridBackendAvailable() {
+    return new Promise((resolve) => {
+        const req = request(HYBRID_BACKEND_URL, { method: "GET", timeout: HYBRID_PROBE_TIMEOUT_MS }, (res) => {
+            // Any response means the server is running
+            res.resume();
+            resolve(true);
+        });
+        req.on("error", () => resolve(false));
+        req.on("timeout", () => {
+            req.destroy();
+            resolve(false);
+        });
+        req.end();
+    });
+}
 // execFile is safe — arguments are passed as an array, no shell interpolation.
 function getJavaVersion() {
     return new Promise((resolve) => {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "paper-manager",
-  "version": "0.10.2",
+  "version": "0.10.4",
   "description": "A paper management system.",
   "keywords": [],
   "homepage": "https://github.com/EurFelux/paper-manager",