first commit

src/modules/pdf/context.ts

@@ -0,0 +1,77 @@
+import type { ChatMessage } from "../llm/types";
+import { getSummaryPrompt, getSystemPrompt } from "../llm/prompts";
+import {
+  buildMetadataBlock,
+  extractPdfContext,
+  type ExtractResult,
+} from "./extractor";
+
+export interface BuiltContext {
+  messages: ChatMessage[];
+  extract: ExtractResult;
+  warning?: string;
+}
+
+export async function buildChatMessages(options: {
+  item: Zotero.Item;
+  history: ChatMessage[];
+  userText: string;
+  selection?: string;
+  mode?: "chat" | "summary";
+}): Promise<BuiltContext> {
+  const extract = await extractPdfContext(options.item);
+  const warnings: string[] = [];
+
+  if (!extract.text) {
+    warnings.push(
+      "未能提取 PDF 文本（可能是扫描版或尚无文本层）。请检查附件后重试。",
+    );
+  } else if (extract.truncated) {
+    warnings.push("论文较长，已按设置截断上下文。");
+  }
+
+  const meta = extract.parentItem
+    ? buildMetadataBlock(extract.parentItem)
+    : "";
+
+  const systemParts = [getSystemPrompt()];
+  if (meta) {
+    systemParts.push(`论文元数据：\n${meta}`);
+  }
+  if (extract.text) {
+    systemParts.push(
+      `论文正文（页之间可能以换页符分隔）：\n${extract.text}`,
+    );
+  }
+
+  const messages: ChatMessage[] = [
+    { role: "system", content: systemParts.join("\n\n") },
+  ];
+
+  // Keep recent history (exclude old system)
+  const history = options.history
+    .filter((m) => m.role !== "system")
+    .slice(-20);
+  messages.push(...history);
+
+  if (options.mode === "summary") {
+    const summaryAsk = getSummaryPrompt();
+    messages.push({
+      role: "user",
+      content: summaryAsk,
+    });
+  } else {
+    let content = options.userText.trim();
+    if (options.selection?.trim()) {
+      content =
+        `【选中原文】\n${options.selection.trim()}\n\n【问题】\n${content}`;
+    }
+    messages.push({ role: "user", content });
+  }
+
+  return {
+    messages,
+    extract,
+    warning: warnings.length ? warnings.join(" ") : undefined,
+  };
+}