@a9i5k4/dsh-literature 0.3.5 → 0.3.6

package/lib/index.js

@@ -757,7 +757,7 @@ var UA, FetchFailure;
 var init_net = __esm({
   "src/node/net.js"() {
     init_log();
-    UA = "dsh-literature/0.3.5 (+https://github.com/deepseek-ai/deepseek-harness)";
+    UA = "dsh-literature/0.3.6 (+https://github.com/deepseek-ai/deepseek-harness)";
     FetchFailure = class extends Error {
       constructor(message, { code = "network", status = 0, retryable = true, cause } = {}) {
         super(message, { cause });
@@ -2421,6 +2421,8 @@ function titleFromFilename(name2) {
   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] ?? "") };
+  const springer = /^([a-z]\d{4,6}-\d{2,4}-\d{2,6}(?:-[a-z0-9]{1,10})?)(?:\s*\(\d+\))?\s*$/i.exec(t.trim());
+  if (springer) return { kind: "doi", value: `10.1007/${springer[1].toLowerCase()}` };
   t = t.replace(/^\[\d+\]\s*/, "").replace(/\s*\(?\d{4}[a-z]?\)?\s*$/i, "").replace(/[_\-]+/g, " ").replace(/^\d+[\s.]+/, "").trim();
   if (!t) return null;
   return { kind: "title", value: t };
@@ -2825,6 +2827,7 @@ __export(pipeline_exports, {
   log: () => log,
   normalizeDoi: () => normalizeDoi,
   previewConflict: () => previewConflict,
+  reidentifyItem: () => reidentifyItem,
   resolveItem: () => resolveItem,
   retryItem: () => retryItem,
   saveItem: () => saveItem,
@@ -3275,6 +3278,33 @@ async function importDroppedPdf(buffer, { filename = "dropped.pdf" } = {}) {
   emitItem(saved);
   return saved;
 }
+async function reidentifyItem(key) {
+  const item = await getItem(key);
+  if (!item) throw failure("not_found", "\u6761\u76EE\u4E0D\u5B58\u5728");
+  const name2 = item.pdf?.filename ?? item.sourceFile ?? (item.rawValue ? `${item.rawValue}.pdf` : "");
+  const { titleFromFilename: titleFromFilename2 } = await Promise.resolve().then(() => (init_importer(), importer_exports));
+  const hit = titleFromFilename2(name2);
+  if (!hit || hit.kind === "title") {
+    throw failure("network", "\u65E0\u6CD5\u4ECE\u6587\u4EF6\u540D\u8BC6\u522B\u51FA DOI / arXiv \u6807\u8BC6\u7B26\uFF0C\u8BF7\u624B\u52A8\u7C98\u8D34\u6807\u8BC6\u7B26\u540E\u91CD\u8BD5");
+  }
+  let next = await patchItem(key, {
+    kind: hit.kind,
+    rawValue: hit.value,
+    display: hit.value,
+    doi: hit.kind === "doi" ? hit.value : "",
+    arxiv: hit.kind === "arxiv" ? hit.value : "",
+    isbn: "",
+    record: null,
+    error: null
+  });
+  emitItem(next);
+  try {
+    next = await resolveItem(key);
+  } catch {
+  }
+  emitItem(next);
+  return next;
+}
 async function importDroppedCourseware(buffer, { filename = "deck.pptx" } = {}) {
   if (!buffer || buffer.length < 4 || buffer.subarray(0, 2).toString("latin1") !== "PK") {
     throw failure("network", "\u62D6\u5165\u7684\u6587\u4EF6\u4E0D\u662F\u6709\u6548\u7684 .pptx\uFF08\u65E7\u7248 .ppt \u8BF7\u5148\u53E6\u5B58\u4E3A .pptx \u6216\u5BFC\u51FA PDF\uFF09");
@@ -36262,7 +36292,7 @@ init_health();
 init_local_api();
 init_connector();
 init_log();
-import { readFile as readFile5 } from "node:fs/promises";
+import { readFile as readFile5, open as openFile, stat as statFile } from "node:fs/promises";
 import { resolve as resolve5 } from "node:path";
 var PREFIX2 = "/api/dsh-literature";
 function pathOf(req) {
@@ -36311,30 +36341,44 @@ async function servePdf(req, res, key) {
     writeJson(res, 400, { error: "invalid key", code: "invalidKey" });
     return;
   }
-  let buffer;
+  let size = 0;
   try {
-    buffer = await readFile5(path);
+    size = (await statFile(path)).size;
   } catch {
     writeJson(res, 404, { error: "pdf not downloaded yet", code: "pdfNotDownloaded" });
     return;
   }
-  const range = parseRange(req.headers.range, buffer.length);
+  const range = parseRange(req.headers.range, size);
   if (range?.invalid) {
-    res.writeHead(416, { "content-range": `bytes */${buffer.length}`, "content-length": 0 });
+    res.writeHead(416, { "content-range": `bytes */${size}`, "content-length": 0 });
     res.end();
     return;
   }
   if (range) {
-    const slice = buffer.subarray(range.start, range.end + 1);
+    const len = range.end - range.start + 1;
+    const slice = Buffer.alloc(len);
+    const fh = await openFile(path, "r");
+    try {
+      await fh.read(slice, 0, len, range.start);
+    } finally {
+      await fh.close();
+    }
     res.writeHead(206, {
       "content-type": "application/pdf",
       "content-length": slice.length,
-      "content-range": `bytes ${range.start}-${range.end}/${buffer.length}`,
+      "content-range": `bytes ${range.start}-${range.end}/${size}`,
       "accept-ranges": "bytes"
     });
     res.end(req.method === "HEAD" ? void 0 : slice);
     return;
   }
+  let buffer;
+  try {
+    buffer = await readFile5(path);
+  } catch {
+    writeJson(res, 404, { error: "pdf not downloaded yet", code: "pdfNotDownloaded" });
+    return;
+  }
   res.writeHead(200, {
     "content-type": "application/pdf",
     "content-length": buffer.length,
@@ -36485,6 +36529,12 @@ async function handler(req, res, ctx2, { activation } = {}) {
       writeJson(res, 200, { item });
       return;
     }
+    if (head === "reidentify" && methodOk(req, "POST")) {
+      const body = await readJsonBody2(req);
+      const item = await reidentifyItem(String(body?.key ?? ""));
+      writeJson(res, 200, { item });
+      return;
+    }
     if (head === "save" && methodOk(req, "POST")) {
       const body = await readJsonBody2(req);
       const item = await saveItem(String(body?.key ?? ""), { mode: body?.mode, tags: body?.tags });

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@a9i5k4/dsh-literature",
-  "version": "0.3.5",
+  "version": "0.3.6",
   "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",