@a9i5k4/dsh-literature 0.2.2 → 0.2.3

package/lib/index.js

@@ -1856,12 +1856,237 @@ var init_local_api = __esm({
   }
 });
 
+// src/node/importer.js
+var importer_exports = {};
+__export(importer_exports, {
+  importDir: () => importDir,
+  importFromZotero: () => importFromZotero,
+  startWatcher: () => startWatcher,
+  stopWatcher: () => stopWatcher,
+  titleFromFilename: () => titleFromFilename,
+  zoteroItemToRecord: () => zoteroItemToRecord
+});
+import { readdir as readdir2, readFile as readFile3, stat as stat2 } from "node:fs/promises";
+import { join as join4, resolve as resolve4, extname, basename } from "node:path";
+import { writeFile as writeFile3, mkdir as mkdir3 } from "node:fs/promises";
+import { dirname as dirname2 } from "node:path";
+function titleFromFilename(name2) {
+  let t = String(name2).replace(/\.pdf$/i, "");
+  const doi = /(10\.\d{4,9}[\/_][^\s]+)/i.exec(t);
+  if (doi) return { kind: "doi", value: doi[1].replace(/_/g, "/") };
+  const arxiv = /\b(\d{4}\.\d{4,5})(v\d+)?\b/.exec(t);
+  if (arxiv) return { kind: "arxiv", value: arxiv[1] + (arxiv[2] ?? "") };
+  t = t.replace(/^\[\d+\]\s*/, "").replace(/\s*\(?\d{4}[a-z]?\)?\s*$/i, "").replace(/[_\-]+/g, " ").replace(/^\d+[\s.]+/, "").trim();
+  if (!t) return null;
+  return { kind: "title", value: t };
+}
+function pathForPdfKey(key) {
+  return join4(PDF_DIR, `${key.replace(/[^\w.-]+/g, "_")}.pdf`);
+}
+async function savePdfBuffer(key, buffer) {
+  const path = pathForPdfKey(key);
+  await mkdir3(dirname2(path), { recursive: true });
+  await writeFile3(path, buffer);
+  return { path, size: buffer.length, source: "library-import" };
+}
+async function importDir(dir, { autoResolve = true } = {}) {
+  const target = resolve4(dir || "");
+  if (!target) throw Object.assign(new Error("\u672A\u914D\u7F6E\u5BFC\u5165\u6587\u4EF6\u5939"), { code: "no_dir" });
+  let files;
+  try {
+    files = await readdir2(target);
+  } catch (e) {
+    throw Object.assign(new Error(`\u65E0\u6CD5\u8BFB\u53D6\u6587\u4EF6\u5939 ${target}`), { code: "no_dir", cause: e });
+  }
+  const imported = [];
+  const skipped = [];
+  for (const name2 of files.filter((f) => extname(f).toLowerCase() === ".pdf")) {
+    const abs = join4(target, name2);
+    const seen = await getImportedFile(abs);
+    let st;
+    try {
+      st = await stat2(abs);
+    } catch {
+      continue;
+    }
+    if (seen && seen.mtimeMs === st.mtimeMs) {
+      skipped.push({ file: name2, reason: "already-imported" });
+      continue;
+    }
+    let item;
+    try {
+      const hit = titleFromFilename(name2);
+      if (!hit) {
+        skipped.push({ file: name2, reason: "unparseable-name" });
+        continue;
+      }
+      const provisional = buildItem({
+        doi: hit.kind === "doi" ? hit.value : "",
+        arxiv: hit.kind === "arxiv" ? hit.value : "",
+        title: hit.kind === "title" ? hit.value : ""
+      });
+      if (!provisional.key) {
+        skipped.push({ file: name2, reason: "no-identity" });
+        continue;
+      }
+      const clash = await getItem(provisional.key);
+      if (clash) {
+        await addImportedFile(abs, st.mtimeMs);
+        skipped.push({ file: name2, reason: "duplicate" });
+        continue;
+      }
+      item = await putItem({
+        ...provisional,
+        kind: hit.kind,
+        rawValue: hit.value,
+        display: hit.value,
+        title: provisional.title || hit.value,
+        state: "discovered",
+        sourceFile: abs,
+        createdAt: Date.now()
+      });
+      emitItem(item);
+    } catch (e) {
+      warn(`import ${name2} failed:`, e.message);
+      skipped.push({ file: name2, reason: e.message });
+      continue;
+    }
+    if (autoResolve) {
+      try {
+        const { resolveItem: resolveItem2 } = await Promise.resolve().then(() => (init_pipeline(), pipeline_exports));
+        item = await resolveItem2(item.key);
+      } catch (e) {
+        warn(`auto-resolve ${name2} failed:`, e.message);
+      }
+    }
+    await addImportedFile(abs, st.mtimeMs);
+    imported.push(item?.key ?? provisionalKeyOf(item));
+  }
+  return { imported, skipped, dir: target };
+}
+function provisionalKeyOf(item) {
+  return item?.key ?? "";
+}
+function zoteroItemToRecord(it) {
+  const d = it?.data ?? it ?? {};
+  const year = /^(\d{4})/.exec(String(d.date ?? ""))?.[1];
+  return {
+    source: "zotero",
+    itemType: d.itemType ?? "journalArticle",
+    title: d.title ?? "",
+    authors: (d.creators ?? []).map((c) => ({
+      creatorType: c.creatorType ?? "author",
+      firstName: c.firstName ?? "",
+      lastName: c.lastName ?? ""
+    })),
+    year: year ? Number(year) : null,
+    container: d.publicationTitle ?? d.bookTitle ?? d.proceedingsTitle ?? "",
+    publisher: d.publisher ?? "",
+    volume: d.volume ?? "",
+    issue: d.issue ?? "",
+    pages: d.pages ?? "",
+    doi: d.DOI ?? "",
+    isbn: d.ISBN ?? "",
+    url: d.url ?? "",
+    abstract: d.abstractNote ?? ""
+  };
+}
+async function importFromZotero({ limit = 50 } = {}) {
+  const items = await searchItems("", { limit });
+  const imported = [];
+  const skipped = [];
+  for (const it of items) {
+    const rec = zoteroItemToRecord(it);
+    if (!rec.title) {
+      skipped.push({ reason: "no-title" });
+      continue;
+    }
+    const provisional = buildItem(rec);
+    if (!provisional.key) {
+      skipped.push({ reason: "no-identity" });
+      continue;
+    }
+    const existing = await getItem(provisional.key);
+    if (existing) {
+      skipped.push({ reason: "duplicate" });
+      continue;
+    }
+    let pdf = null;
+    try {
+      const children = await getItemChildren(it.key);
+      const attach = (children ?? []).find(
+        (c) => c.itemType === "attachment" && /pdf/i.test(c.contentType ?? c.contentType ?? "")
+      );
+      if (attach?.key) {
+        const { buffer } = await getFileBuffer(attach.key).catch(() => ({ buffer: null }));
+        if (buffer?.length && buffer.subarray(0, 5).equals(PDF_MAGIC2)) {
+          pdf = await savePdfBuffer(provisional.key, buffer);
+        }
+      }
+    } catch (e) {
+      warn(`attachment import failed for ${it.key}:`, e.message);
+    }
+    const item = await putItem({
+      ...provisional,
+      record: rec,
+      state: pdf ? "fetched" : "resolved",
+      pdf,
+      createdAt: Date.now()
+    });
+    emitItem(item);
+    imported.push(item.key);
+  }
+  return { imported, skipped, count: imported.length };
+}
+function startWatcher() {
+  if (watcherTimer) return () => stopWatcher();
+  let stopping = false;
+  const sweep = async () => {
+    if (stopping) return;
+    const config = await loadConfig();
+    if (!config.watchImport || !config.importDir) return;
+    try {
+      const r = await importDir(config.importDir, { autoResolve: config.autoResolve !== false });
+      if (r.imported.length) log(`folder watch imported ${r.imported.length} new file(s)`);
+    } catch (e) {
+      warn("folder watch sweep failed:", e.message);
+    }
+  };
+  const initial = setTimeout(sweep, 5e3);
+  initial.unref?.();
+  const timer = setInterval(sweep, 3e4);
+  timer.unref?.();
+  watcherTimer = { timer, initial };
+  return () => stopWatcher();
+}
+function stopWatcher() {
+  if (!watcherTimer) return;
+  clearInterval(watcherTimer.timer);
+  clearTimeout(watcherTimer.initial);
+  watcherTimer = null;
+}
+var PDF_MAGIC2, watcherTimer;
+var init_importer = __esm({
+  "src/node/importer.js"() {
+    init_config();
+    init_db();
+    init_sse();
+    init_dedupe();
+    init_metadata();
+    init_local_api();
+    init_log();
+    PDF_MAGIC2 = Buffer.from("%PDF-", "latin1");
+    watcherTimer = null;
+  }
+});
+
 // src/node/pipeline.js
 var pipeline_exports = {};
 __export(pipeline_exports, {
   arxivBase: () => arxivBase,
   discardItem: () => discardItem,
   fetchItemPdf: () => fetchItemPdf,
+  importDroppedPdf: () => importDroppedPdf,
   importPdf: () => importPdf,
   log: () => log,
   normalizeDoi: () => normalizeDoi,
@@ -1873,8 +2098,8 @@ __export(pipeline_exports, {
   startTask: () => startTask
 });
 import { randomUUID as randomUUID2 } from "node:crypto";
-import { writeFile as writeFile3, unlink, mkdir as mkdir3 } from "node:fs/promises";
-import { join as join4, dirname as dirname2 } from "node:path";
+import { writeFile as writeFile4, unlink, mkdir as mkdir4 } from "node:fs/promises";
+import { join as join5, dirname as dirname3 } from "node:path";
 function failure(code, message, extra = {}) {
   return {
     code,
@@ -1885,7 +2110,7 @@ function failure(code, message, extra = {}) {
   };
 }
 function pdfPathFor(key) {
-  return join4(PDF_DIR, `${key.replace(/[^\w.-]+/g, "_")}.pdf`);
+  return join5(PDF_DIR, `${key.replace(/[^\w.-]+/g, "_")}.pdf`);
 }
 async function update(key, patch) {
   const next = await patchItem(key, patch);
@@ -2001,8 +2226,8 @@ async function fetchItemPdf(key) {
       { ...config.retry, label: `fetch ${key}` }
     );
     const path = pdfPathFor(key);
-    await mkdir3(dirname2(path), { recursive: true });
-    await writeFile3(path, result.buffer);
+    await mkdir4(dirname3(path), { recursive: true });
+    await writeFile4(path, result.buffer);
     const updated = await patchItem(key, {
       state: "fetched",
       pdf: { path, size: result.buffer.length, source: result.source, url: result.url },
@@ -2154,8 +2379,8 @@ async function importPdf(key, buffer, { filename = "imported.pdf", autoSave = tr
     throw failure("network", "\u6240\u9009\u6587\u4EF6\u4E0D\u662F\u6709\u6548\u7684 PDF");
   }
   const path = pdfPathFor(key);
-  await mkdir3(dirname2(path), { recursive: true });
-  await writeFile3(path, buffer);
+  await mkdir4(dirname3(path), { recursive: true });
+  await writeFile4(path, buffer);
   const updated = await update(key, {
     state: "fetched",
     pdf: { path, size: buffer.length, source: "local-import", url: "", filename },
@@ -2163,10 +2388,66 @@ async function importPdf(key, buffer, { filename = "imported.pdf", autoSave = tr
   });
   let saved = updated;
   if (autoSave) {
-    saved = await saveItem(key);
+    saved = await saveItem(key, { mode: "builtin" });
   }
   return saved;
 }
+async function importDroppedPdf(buffer, { filename = "dropped.pdf" } = {}) {
+  if (!buffer || buffer.length < 8 || !buffer.subarray(0, 5).equals(Buffer.from("%PDF-", "latin1"))) {
+    throw failure("network", "\u62D6\u5165\u7684\u6587\u4EF6\u4E0D\u662F\u6709\u6548\u7684 PDF");
+  }
+  const { titleFromFilename: titleFromFilename2 } = await Promise.resolve().then(() => (init_importer(), importer_exports));
+  const hit = titleFromFilename2(filename);
+  const provisional = buildItem(
+    hit ? {
+      doi: hit.kind === "doi" ? hit.value : "",
+      arxiv: hit.kind === "arxiv" ? hit.value : "",
+      title: hit.kind === "title" ? hit.value : ""
+    } : { title: String(filename).replace(/\.pdf$/i, "") }
+  );
+  if (!provisional.key) throw failure("network", "\u65E0\u6CD5\u4ECE\u6587\u4EF6\u540D\u8BC6\u522B\u8BE5\u6587\u732E");
+  const clash = await getItem(provisional.key);
+  if (clash) {
+    const path2 = pdfPathFor(provisional.key);
+    await mkdir4(dirname3(path2), { recursive: true });
+    await writeFile4(path2, buffer);
+    const updated = await patchItem(provisional.key, {
+      state: "fetched",
+      pdf: { path: path2, size: buffer.length, source: "local-import", url: "", filename },
+      error: null,
+      updatedAt: Date.now()
+    });
+    emitItem(updated);
+    return updated;
+  }
+  const path = pdfPathFor(provisional.key);
+  await mkdir4(dirname3(path), { recursive: true });
+  await writeFile4(path, buffer);
+  let item = await putItem({
+    ...provisional,
+    kind: hit?.kind ?? "title",
+    rawValue: hit?.value ?? "",
+    display: hit?.value ?? provisional.title,
+    sourceFile: filename,
+    state: "fetched",
+    pdf: { path, size: buffer.length, source: "drop-import", url: "", filename },
+    createdAt: Date.now()
+  });
+  emitItem(item);
+  try {
+    item = await resolveItem(item.key);
+  } catch {
+  }
+  if (!item.record) {
+    item = await patchItem(item.key, {
+      record: { itemType: "journalArticle", title: item.title || provisional.title, authors: [], year: null },
+      state: "resolved"
+    });
+  }
+  const saved = await saveItem(item.key, { mode: "builtin" });
+  emitItem(saved);
+  return saved;
+}
 var FAILURE_MESSAGES;
 var init_pipeline = __esm({
   "src/node/pipeline.js"() {
@@ -2367,230 +2648,6 @@ var init_cite = __esm({
   }
 });
 
-// src/node/importer.js
-var importer_exports = {};
-__export(importer_exports, {
-  importDir: () => importDir,
-  importFromZotero: () => importFromZotero,
-  startWatcher: () => startWatcher,
-  stopWatcher: () => stopWatcher,
-  titleFromFilename: () => titleFromFilename,
-  zoteroItemToRecord: () => zoteroItemToRecord
-});
-import { readdir as readdir2, readFile as readFile3, stat as stat2 } from "node:fs/promises";
-import { join as join5, resolve as resolve4, extname, basename } from "node:path";
-import { writeFile as writeFile4, mkdir as mkdir4 } from "node:fs/promises";
-import { dirname as dirname3 } from "node:path";
-function titleFromFilename(name2) {
-  let t = String(name2).replace(/\.pdf$/i, "");
-  const doi = /(10\.\d{4,9}[\/_][^\s]+)/i.exec(t);
-  if (doi) return { kind: "doi", value: doi[1].replace(/_/g, "/") };
-  const arxiv = /\b(\d{4}\.\d{4,5})(v\d+)?\b/.exec(t);
-  if (arxiv) return { kind: "arxiv", value: arxiv[1] + (arxiv[2] ?? "") };
-  t = t.replace(/^\[\d+\]\s*/, "").replace(/\s*\(?\d{4}[a-z]?\)?\s*$/i, "").replace(/[_\-]+/g, " ").replace(/^\d+[\s.]+/, "").trim();
-  if (!t) return null;
-  return { kind: "title", value: t };
-}
-function pathForPdfKey(key) {
-  return join5(PDF_DIR, `${key.replace(/[^\w.-]+/g, "_")}.pdf`);
-}
-async function savePdfBuffer(key, buffer) {
-  const path = pathForPdfKey(key);
-  await mkdir4(dirname3(path), { recursive: true });
-  await writeFile4(path, buffer);
-  return { path, size: buffer.length, source: "library-import" };
-}
-async function importDir(dir, { autoResolve = true } = {}) {
-  const target = resolve4(dir || "");
-  if (!target) throw Object.assign(new Error("\u672A\u914D\u7F6E\u5BFC\u5165\u6587\u4EF6\u5939"), { code: "no_dir" });
-  let files;
-  try {
-    files = await readdir2(target);
-  } catch (e) {
-    throw Object.assign(new Error(`\u65E0\u6CD5\u8BFB\u53D6\u6587\u4EF6\u5939 ${target}`), { code: "no_dir", cause: e });
-  }
-  const imported = [];
-  const skipped = [];
-  for (const name2 of files.filter((f) => extname(f).toLowerCase() === ".pdf")) {
-    const abs = join5(target, name2);
-    const seen = await getImportedFile(abs);
-    let st;
-    try {
-      st = await stat2(abs);
-    } catch {
-      continue;
-    }
-    if (seen && seen.mtimeMs === st.mtimeMs) {
-      skipped.push({ file: name2, reason: "already-imported" });
-      continue;
-    }
-    let item;
-    try {
-      const hit = titleFromFilename(name2);
-      if (!hit) {
-        skipped.push({ file: name2, reason: "unparseable-name" });
-        continue;
-      }
-      const provisional = buildItem({
-        doi: hit.kind === "doi" ? hit.value : "",
-        arxiv: hit.kind === "arxiv" ? hit.value : "",
-        title: hit.kind === "title" ? hit.value : ""
-      });
-      if (!provisional.key) {
-        skipped.push({ file: name2, reason: "no-identity" });
-        continue;
-      }
-      const clash = await getItem(provisional.key);
-      if (clash) {
-        await addImportedFile(abs, st.mtimeMs);
-        skipped.push({ file: name2, reason: "duplicate" });
-        continue;
-      }
-      item = await putItem({
-        ...provisional,
-        kind: hit.kind,
-        rawValue: hit.value,
-        display: hit.value,
-        title: provisional.title || hit.value,
-        state: "discovered",
-        sourceFile: abs,
-        createdAt: Date.now()
-      });
-      emitItem(item);
-    } catch (e) {
-      warn(`import ${name2} failed:`, e.message);
-      skipped.push({ file: name2, reason: e.message });
-      continue;
-    }
-    if (autoResolve) {
-      try {
-        const { resolveItem: resolveItem2 } = await Promise.resolve().then(() => (init_pipeline(), pipeline_exports));
-        item = await resolveItem2(item.key);
-      } catch (e) {
-        warn(`auto-resolve ${name2} failed:`, e.message);
-      }
-    }
-    await addImportedFile(abs, st.mtimeMs);
-    imported.push(item?.key ?? provisionalKeyOf(item));
-  }
-  return { imported, skipped, dir: target };
-}
-function provisionalKeyOf(item) {
-  return item?.key ?? "";
-}
-function zoteroItemToRecord(it) {
-  const d = it?.data ?? it ?? {};
-  const year = /^(\d{4})/.exec(String(d.date ?? ""))?.[1];
-  return {
-    source: "zotero",
-    itemType: d.itemType ?? "journalArticle",
-    title: d.title ?? "",
-    authors: (d.creators ?? []).map((c) => ({
-      creatorType: c.creatorType ?? "author",
-      firstName: c.firstName ?? "",
-      lastName: c.lastName ?? ""
-    })),
-    year: year ? Number(year) : null,
-    container: d.publicationTitle ?? d.bookTitle ?? d.proceedingsTitle ?? "",
-    publisher: d.publisher ?? "",
-    volume: d.volume ?? "",
-    issue: d.issue ?? "",
-    pages: d.pages ?? "",
-    doi: d.DOI ?? "",
-    isbn: d.ISBN ?? "",
-    url: d.url ?? "",
-    abstract: d.abstractNote ?? ""
-  };
-}
-async function importFromZotero({ limit = 50 } = {}) {
-  const items = await searchItems("", { limit });
-  const imported = [];
-  const skipped = [];
-  for (const it of items) {
-    const rec = zoteroItemToRecord(it);
-    if (!rec.title) {
-      skipped.push({ reason: "no-title" });
-      continue;
-    }
-    const provisional = buildItem(rec);
-    if (!provisional.key) {
-      skipped.push({ reason: "no-identity" });
-      continue;
-    }
-    const existing = await getItem(provisional.key);
-    if (existing) {
-      skipped.push({ reason: "duplicate" });
-      continue;
-    }
-    let pdf = null;
-    try {
-      const children = await getItemChildren(it.key);
-      const attach = (children ?? []).find(
-        (c) => c.itemType === "attachment" && /pdf/i.test(c.contentType ?? c.contentType ?? "")
-      );
-      if (attach?.key) {
-        const { buffer } = await getFileBuffer(attach.key).catch(() => ({ buffer: null }));
-        if (buffer?.length && buffer.subarray(0, 5).equals(PDF_MAGIC2)) {
-          pdf = await savePdfBuffer(provisional.key, buffer);
-        }
-      }
-    } catch (e) {
-      warn(`attachment import failed for ${it.key}:`, e.message);
-    }
-    const item = await putItem({
-      ...provisional,
-      record: rec,
-      state: pdf ? "fetched" : "resolved",
-      pdf,
-      createdAt: Date.now()
-    });
-    emitItem(item);
-    imported.push(item.key);
-  }
-  return { imported, skipped, count: imported.length };
-}
-function startWatcher() {
-  if (watcherTimer) return () => stopWatcher();
-  let stopping = false;
-  const sweep = async () => {
-    if (stopping) return;
-    const config = await loadConfig();
-    if (!config.watchImport || !config.importDir) return;
-    try {
-      const r = await importDir(config.importDir, { autoResolve: config.autoResolve !== false });
-      if (r.imported.length) log(`folder watch imported ${r.imported.length} new file(s)`);
-    } catch (e) {
-      warn("folder watch sweep failed:", e.message);
-    }
-  };
-  const initial = setTimeout(sweep, 5e3);
-  initial.unref?.();
-  const timer = setInterval(sweep, 3e4);
-  timer.unref?.();
-  watcherTimer = { timer, initial };
-  return () => stopWatcher();
-}
-function stopWatcher() {
-  if (!watcherTimer) return;
-  clearInterval(watcherTimer.timer);
-  clearTimeout(watcherTimer.initial);
-  watcherTimer = null;
-}
-var PDF_MAGIC2, watcherTimer;
-var init_importer = __esm({
-  "src/node/importer.js"() {
-    init_config();
-    init_db();
-    init_sse();
-    init_dedupe();
-    init_metadata();
-    init_local_api();
-    init_log();
-    PDF_MAGIC2 = Buffer.from("%PDF-", "latin1");
-    watcherTimer = null;
-  }
-});
-
 // src/node/index.js
 init_config();
 init_db();
@@ -2821,6 +2878,14 @@ async function handler(req, res) {
       writeJson(res, 200, result);
       return;
     }
+    if (head === "drop" && methodOk(req, "POST")) {
+      const url = new URL(req.url ?? "/", "http://127.0.0.1");
+      const filename = url.searchParams.get("filename") || "dropped.pdf";
+      const buffer = await readRawBody(req, 128 * 1024 * 1024);
+      const item = await importDroppedPdf(buffer, { filename });
+      writeJson(res, 200, { item });
+      return;
+    }
     if (head === "import-zotero" && methodOk(req, "POST")) {
       const body = await readJsonBody2(req);
       const { importFromZotero: importFromZotero2 } = await Promise.resolve().then(() => (init_importer(), importer_exports));

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@a9i5k4/dsh-literature",
-  "version": "0.2.2",
+  "version": "0.2.3",
   "description": "DSH Literature 文献侧窗：在 DeepSeek Harness 侧边栏识别 DOI/arXiv/标题、抓取元数据与全文、写入本地文献库或导出目录，并提供内置 PDF 阅读器（缩放/翻页/目录/搜索/高亮/笔记）。",
   "type": "module",
   "main": "lib/index.js",