diff --git a/index.html b/index.html index d377fc2..6f913e0 100644 --- a/index.html +++ b/index.html @@ -873,6 +873,7 @@ renderLlmProgress(job); statusEl.textContent = job.statusMessage || "Die Einschätzung wird erstellt. Das kann einen Moment dauern."; if (job.status === "done") return job; + if (job.status === "cancelled") throw new Error("Job wurde abgebrochen"); if (job.status === "error") throw new Error(job.errors?.join("\\n") || "Die Auswertung ist fehlgeschlagen."); await new Promise((resolve) => setTimeout(resolve, POLL_DELAY_MS)); } diff --git a/index.ts b/index.ts index b0e664a..f9ba041 100644 --- a/index.ts +++ b/index.ts @@ -2,9 +2,12 @@ import { routes as templatebuilder, workerPath } from "./packages/templatebuilde import { routes as extractor } from "./packages/extractor/index"; import { routes as summarizer } from "./packages/summarizer/index"; import { join } from "node:path"; +import { rm } from "node:fs/promises"; import { dataPath } from "./deployPaths"; import { addSecurityHeaders, + getCsrfCookieName, + isAuthenticated, isSafeStem, login, loginPage, @@ -17,6 +20,9 @@ import { const SUMMARIZER_OUTPUTS_DIR = process.env.BMP_SUMMARIZER_OUTPUTS_DIR ?? dataPath(join(import.meta.dir, "packages/summarizer/outputs"), "summarizer", "outputs"); +const EXTRACTOR_OUTPUTS_DIR = + process.env.BMP_EXTRACTOR_OUTPUTS_DIR ?? + dataPath(join(import.meta.dir, "packages/extractor/outputs"), "extractor", "outputs"); const ROOT_PAGE_PATH = join(import.meta.dir, "index.html"); const BMP_LOGO_PATH = join(import.meta.dir, "assets/bmp-logo.png"); const PORT = Number(process.env.PORT ?? 3000); @@ -74,6 +80,12 @@ Bun.serve({ "/logout": { POST: withAuth(async () => logout(), { csrf: true }), }, + "/api/auth/session": { + GET: async (req: Request) => addSecurityHeaders(Response.json({ + authenticated: await isAuthenticated(req), + csrfCookieName: getCsrfCookieName(), + })), + }, "/": { GET: withAuth(async () => new Response(Bun.file(ROOT_PAGE_PATH), { headers: { "Content-Type": "text/html; charset=utf-8" }, @@ -84,6 +96,24 @@ Bun.serve({ headers: { "Content-Type": "image/png", "Cache-Control": "private, max-age=86400" }, })), }, + "/api/results/:stem": { + DELETE: withAuth(async (req: Request) => { + const stem = (req as Request & { params: Record }).params.stem ?? ""; + if (!isSafeStem(stem)) { + return Response.json({ error: "Invalid result id" }, { status: 400 }); + } + + const targets = [ + join(EXTRACTOR_OUTPUTS_DIR, stem), + join(SUMMARIZER_OUTPUTS_DIR, stem), + ]; + await Promise.all(targets.map((target) => rm(target, { recursive: true, force: true }))); + return Response.json({ ok: true, stem, deleted: targets.length }); + }, { + csrf: true, + limit: { key: "result-delete", max: 20, windowMs: 60_000 }, + }), + }, "/api/pipeline/result/:stem": { GET: withAuth(async (req: Request) => { const stem = (req as Request & { params: Record }).params.stem ?? ""; diff --git a/packages/extractor/index.ts b/packages/extractor/index.ts index dfdab05..b4e580e 100644 --- a/packages/extractor/index.ts +++ b/packages/extractor/index.ts @@ -1,9 +1,9 @@ import { join } from "node:path"; import { dataPath } from "../../deployPaths"; -import { jsonError, normalizeStem, securityHeaders, withAuth } from "../../security"; +import { isSafeStem, jsonError, normalizeStem, securityHeaders, withAuth } from "../../security"; const UPLOAD_PAGE_PATH = join(import.meta.dir, "upload.html"); -import { mkdir } from "node:fs/promises"; +import { mkdir, readdir } from "node:fs/promises"; import { flattenTemplate, extractFields, buildOutput, getNested } from "./extractor"; const TEMPLATE_PATH = @@ -59,6 +59,60 @@ function extractScalarText(value: unknown): string | undefined { return undefined; } +function extractorDownloadUrl(stem: string, file: string): string { + return `/downloads/extractor/${encodeURIComponent(stem)}/${encodeURIComponent(file)}`; +} + +async function listExtractorResults() { + let entries: string[]; + try { + entries = await readdir(OUTPUTS_DIR); + } catch { + return []; + } + + const results = []; + for (const stem of entries.toSorted()) { + if (!isSafeStem(stem)) continue; + const jsonFile = `${stem}.json`; + const pdfFile = `${stem}.pdf`; + const jsonPath = join(OUTPUTS_DIR, stem, jsonFile); + const pdfPath = join(OUTPUTS_DIR, stem, pdfFile); + if (!(await Bun.file(jsonPath).exists())) continue; + results.push({ + stem, + files: { + json: (await Bun.file(jsonPath).exists()) ? jsonFile : undefined, + pdf: (await Bun.file(pdfPath).exists()) ? pdfFile : undefined, + }, + downloads: [ + { label: "JSON herunterladen", url: extractorDownloadUrl(stem, jsonFile) }, + ...((await Bun.file(pdfPath).exists()) ? [{ label: "PDF herunterladen", url: extractorDownloadUrl(stem, pdfFile) }] : []), + ], + }); + } + return results; +} + +async function serveExtractorDownload(stem: string, file: string): Promise { + if (!isSafeStem(stem) || (file !== `${stem}.json` && file !== `${stem}.pdf`)) { + return new Response("Not found", { status: 404, headers: securityHeaders }); + } + const path = join(OUTPUTS_DIR, stem, file); + const output = Bun.file(path); + if (!(await output.exists())) { + return new Response("Not found", { status: 404, headers: securityHeaders }); + } + return new Response(output, { + headers: { + ...securityHeaders, + "Content-Type": file.endsWith(".json") ? "application/json; charset=utf-8" : "application/pdf", + "Cache-Control": "private, no-store", + "Content-Disposition": `attachment; filename="${file.replaceAll('"', "")}"`, + }, + }); +} + export const routes = { "/extractor": { GET: withAuth(async () => new Response(Bun.file(UPLOAD_PAGE_PATH), { @@ -67,6 +121,41 @@ export const routes = { ), }, + "/api/extractor/results": { + GET: withAuth(async () => Response.json({ results: await listExtractorResults() })), + }, + + "/api/extractor/results/:stem": { + GET: withAuth(async (req: Request) => { + const stem = (req as Request & { params: Record }).params.stem ?? ""; + if (!isSafeStem(stem)) { + return jsonError("Invalid result id", 400); + } + const jsonPath = join(OUTPUTS_DIR, stem, `${stem}.json`); + const pdfPath = join(OUTPUTS_DIR, stem, `${stem}.pdf`); + const jsonFile = Bun.file(jsonPath); + if (!(await jsonFile.exists())) { + return Response.json({ error: "Extractor result not found" }, { status: 404 }); + } + return Response.json({ + ok: true, + stem, + result: await jsonFile.json(), + downloads: [ + { label: "JSON herunterladen", url: extractorDownloadUrl(stem, `${stem}.json`) }, + ...((await Bun.file(pdfPath).exists()) ? [{ label: "PDF herunterladen", url: extractorDownloadUrl(stem, `${stem}.pdf`) }] : []), + ], + }); + }), + }, + + "/downloads/extractor/:stem/:file": { + GET: withAuth(async (req: Request) => { + const params = (req as Request & { params: Record }).params; + return serveExtractorDownload(params.stem ?? "", params.file ?? ""); + }), + }, + "/extract": { POST: withAuth(async (req: Request) => { const contentLength = Number(req.headers.get("content-length") ?? 0); diff --git a/packages/summarizer/index.html b/packages/summarizer/index.html index bdfd0af..aad47f2 100644 --- a/packages/summarizer/index.html +++ b/packages/summarizer/index.html @@ -185,6 +185,7 @@ renderProgress(job); if (job.status === "done") return job; + if (job.status === "cancelled") throw new Error("Job wurde abgebrochen"); if (job.status === "error") throw new Error(job.errors?.join("\n") || "Job fehlgeschlagen"); await new Promise((resolve) => setTimeout(resolve, 700)); diff --git a/packages/summarizer/index.ts b/packages/summarizer/index.ts index 3c83c34..40530ff 100644 --- a/packages/summarizer/index.ts +++ b/packages/summarizer/index.ts @@ -11,7 +11,7 @@ const MAX_PARALLEL_SUMMARIES = 2; const MAX_ACTIVE_JOBS = Number(process.env.BMP_MAX_ACTIVE_SUMMARY_JOBS ?? 10); const EXPECTED_LLM_CALLS_PER_COMPANY = 13; -type JobStatus = "queued" | "running" | "done" | "error"; +type JobStatus = "queued" | "running" | "done" | "error" | "cancelled"; interface SummaryJob { id: string; @@ -203,6 +203,10 @@ function activeJobCount(): number { return [...jobs.values()].filter((job) => job.status === "queued" || job.status === "running").length; } +function isCancelled(job: SummaryJob): boolean { + return job.status === "cancelled"; +} + const jobs = new Map(); function createJob(stem: string, model: string, total: number): SummaryJob { @@ -262,12 +266,14 @@ async function runWithConcurrency( } async function executeJob(job: SummaryJob): Promise { + if (isCancelled(job)) return; job.status = "running"; job.statusMessage = "Auswertung wird vorbereitet."; const companies = job.stem === "__all__" ? await listCompanies() : [job.stem]; try { await runWithConcurrency(companies, MAX_PARALLEL_SUMMARIES, async (company) => { + if (isCancelled(job)) return; job.current = company; job.statusMessage = `LLM-Auswertung für ${company} läuft.`; const summary = await summarizeCompany(company, job.model, { @@ -286,17 +292,31 @@ async function executeJob(job: SummaryJob): Promise { }, }); job.statusMessage = `Downloads für ${company} werden vorbereitet.`; + if (isCancelled(job)) return; const outputPaths = await writeSummary(company, summary, OUTPUTS_DIR); job.files.push(...outputPaths.map((path) => basename(path))); job.downloads.push(...outputPaths.map((path) => buildDownload(company, path))); job.completed += 1; }); + if (isCancelled(job)) { + job.current = undefined; + job.statusMessage = "Auswertung abgebrochen."; + job.finishedAt = new Date().toISOString(); + return; + } + job.status = "done"; job.current = undefined; job.statusMessage = "Auswertung fertig."; job.finishedAt = new Date().toISOString(); } catch (error) { + if (isCancelled(job)) { + job.current = undefined; + job.statusMessage = "Auswertung abgebrochen."; + job.finishedAt = new Date().toISOString(); + return; + } job.status = "error"; job.errors.push(String(error)); job.statusMessage = "Auswertung fehlgeschlagen."; @@ -368,6 +388,15 @@ export const routes = { }), }, + "/api/summarizer/jobs": { + GET: withAuth(async () => { + const summaries = [...jobs.values()] + .toSorted((a, b) => Date.parse(b.createdAt) - Date.parse(a.createdAt)) + .map(publicJob); + return Response.json({ jobs: summaries }); + }), + }, + "/api/summarizer/jobs/:id/llm-calls": { GET: withAuth(async (req: Request) => { const id = (req as Request & { params: Record }).params.id; @@ -402,5 +431,24 @@ export const routes = { } return Response.json(publicJob(job)); }), + DELETE: withAuth(async (req: Request) => { + const id = (req as Request & { params: Record }).params.id; + const job = jobs.get(id); + if (!job) { + return Response.json({ error: "Job not found" }, { status: 404 }); + } + if (job.status === "queued" || job.status === "running") { + job.status = "cancelled"; + job.current = undefined; + job.statusMessage = "Auswertung abgebrochen."; + job.finishedAt = new Date().toISOString(); + return Response.json({ ok: true, job: publicJob(job) }); + } + jobs.delete(id); + return Response.json({ ok: true, deleted: true }); + }, { + csrf: true, + limit: { key: "summarizer-job-delete", max: 30, windowMs: 60_000 }, + }), }, } as const; diff --git a/packages/templatebuilder/index.ts b/packages/templatebuilder/index.ts index 214d5ac..9f350be 100644 --- a/packages/templatebuilder/index.ts +++ b/packages/templatebuilder/index.ts @@ -21,7 +21,9 @@ async function loadTemplate(): Promise> { } export const routes = { - "/builder": index, + "/builder": { + GET: withAuth(async () => index), + }, "/api/save": { POST: withAuth(async (req: Request) => { diff --git a/security.ts b/security.ts index 1f2cdfd..f7340b7 100644 --- a/security.ts +++ b/security.ts @@ -245,7 +245,7 @@ function hasValidCsrf(req: Request): boolean { return Boolean(header && cookie && timingSafeEqual(header, cookie)); } -async function isAuthenticated(req: Request): Promise { +export async function isAuthenticated(req: Request): Promise { const auth = parseCookies(req.headers.get("cookie")).get(AUTH_COOKIE); if (auth && await verifyAuthCookie(auth)) return true;