@anthusai/papyrus 5.51.0 → 5.53.0

package/amplify/functions/content-actions/requirements.txt

@@ -4,3 +4,5 @@ anthus-markus==0.7.0
 Pillow>=11.3,<12
 awscrt>=0.23,<1
 humanus>=0.33,<0.34
+pypdf>=6.12.1,<7
+youtube-transcript-api>=1.2.4,<2

package/amplify/functions/content-actions/resource.js

@@ -7,7 +7,7 @@ import { fileURLToPath } from "node:url";
 import { pythonLambdaCode } from "../shared/python-bundle.js";
 const dirname = path.dirname(fileURLToPath(import.meta.url));
 const projectRoot = path.resolve(dirname, "../../..");
-export const contentActionsRequirements = ["PyYAML>=6.0.2,<7", "citeproc-py>=0.8.2,<1", "anthus-markus==0.7.0", "Pillow>=11.3,<12", "awscrt>=0.23,<1", "humanus>=0.33,<0.34"];
+export const contentActionsRequirements = ["PyYAML>=6.0.2,<7", "citeproc-py>=0.8.2,<1", "anthus-markus==0.7.0", "Pillow>=11.3,<12", "awscrt>=0.23,<1", "humanus>=0.33,<0.34", "pypdf>=6.12.1,<7", "youtube-transcript-api>=1.2.4,<2"];
 function installCheckoutRequirements(outputDir) {
     const requirementsPath = path.join(projectRoot, "amplify/functions/content-actions/requirements.txt");
     execSync(`python3 -m pip install -r "${requirementsPath}" -t "${outputDir}" --no-cache-dir --platform manylinux2014_aarch64 --implementation cp --python-version 3.12 --only-binary=:all: --ignore-requires-python`, { cwd: projectRoot, stdio: "inherit" });

package/amplify/functions/content-actions/resource.ts

@@ -10,7 +10,7 @@ import { pythonLambdaCode, type SiteBundleFile } from "../shared/python-bundle";
 const dirname = path.dirname(fileURLToPath(import.meta.url));
 const projectRoot = path.resolve(dirname, "../../..");
 
-export const contentActionsRequirements = ["PyYAML>=6.0.2,<7", "citeproc-py>=0.8.2,<1", "anthus-markus==0.7.0", "Pillow>=11.3,<12", "awscrt>=0.23,<1", "humanus>=0.33,<0.34"];
+export const contentActionsRequirements = ["PyYAML>=6.0.2,<7", "citeproc-py>=0.8.2,<1", "anthus-markus==0.7.0", "Pillow>=11.3,<12", "awscrt>=0.23,<1", "humanus>=0.33,<0.34", "pypdf>=6.12.1,<7", "youtube-transcript-api>=1.2.4,<2"];
 
 function installCheckoutRequirements(outputDir: string): void {
   const requirementsPath = path.join(projectRoot, "amplify/functions/content-actions/requirements.txt");

package/components/reference-capture-client.ts

@@ -7,7 +7,7 @@ import {
   captureStatus, countWords, curationCardOf, grobidRows, primaryTextAttachment, structuredRows,
   type CaptureAssessment as Assessment, type CaptureStatus, type CurationCard,
 } from "../lib/reference-capture";
-import { loadModelPayloadsForOwner, loadStoragePathText } from "./news-desk-taxonomy-client";
+import { loadModelPayloadsForOwner, loadStoragePathText, uploadModelPayloadForOwner } from "./news-desk-taxonomy-client";
 
 export type CaptureAssessment = Assessment;
 
@@ -23,8 +23,16 @@ export type CaptureView = {
   rows: Array<{ label: string; value: string }>;
   /** Saves an editor's pasted or corrected text and their verdict on the capture (the `correctReferenceCapture` action). */
   correct?: (input: { text?: string; assessment: CaptureAssessment; note?: string }) => Promise<void>;
+  /** True when the current text was read from a PDF the editor uploaded. */
+  fromPdf: boolean;
+  /** Stores a PDF the editor downloaded as an attachment of the Reference, then has the server read its text (`correctReferenceCapture` with `pdfAttachmentId`). */
+  uploadPdf?: (input: { file: File; assessment: CaptureAssessment; note?: string }, onStep?: (step: "uploading" | "reading") => void) => Promise<void>;
 };
 
+/** The server reads PDFs up to this size (ideas.source_text.MAXIMUM_PDF_BYTES); say so before a slow upload that would be refused. */
+export const MAXIMUM_PDF_UPLOAD_BYTES = 8_000_000;
+export const PDF_UPLOAD_ROLE = "curation-upload-pdf";
+
 function previewText(attachment: ReferenceAttachmentRecord): string | null {
   // Local preview only: a fixture can carry the captured text inline instead of in storage.
   if (!isLocalPreviewEnabled()) return null;
@@ -42,7 +50,7 @@ type Payloads = { key: string; text: string | null; grobid: unknown; summary: st
  * Reads what the newsroom stored for one Reference. The curation card comes first (status, word count, signals, summary, GROBID fields,
  * and the `curation-text` body its `current` points at); the older `extracted_text` attachment is the fallback.
  */
-export type CorrectCaptureAction = (input: { referenceId: string; text?: string; assessment: CaptureAssessment; note?: string }) => Promise<unknown>;
+export type CorrectCaptureAction = (input: { referenceId: string; text?: string; pdfAttachmentId?: string; assessment: CaptureAssessment; note?: string }) => Promise<unknown>;
 
 export function useReferenceCapture(reference: ReferenceRecord, attachments: readonly ReferenceAttachmentRecord[], teaserLine: number, correctAction?: CorrectCaptureAction): CaptureView {
   const card: CurationCard | null = useMemo(() => curationCardOf(attachments), [attachments]);
@@ -51,7 +59,7 @@ export function useReferenceCapture(reference: ReferenceRecord, attachments: rea
   const grobidSortKey = card?.current?.grobidSortKey ?? card?.grobid?.fieldsSortKey ?? null;
   const [payloads, setPayloads] = useState<Payloads>({ key: "", text: null, grobid: null, summary: null, loaded: false });
   const [fallback, setFallback] = useState<{ key: string; text: string | null; error: string | null; loading: boolean }>({ key: "", text: null, error: null, loading: false });
-  const [corrected, setCorrected] = useState<{ key: string; text: string; assessment: CaptureAssessment } | null>(null);
+  const [corrected, setCorrected] = useState<{ key: string; text: string; assessment: CaptureAssessment; fromPdf?: boolean } | null>(null);
   const referenceKey = `${reference.id}:${textSortKey ?? ""}:${grobidSortKey ?? ""}`;
 
   // The card's bodies and the Reference's metadata.json, in one read of the Reference's ModelAttachments.
@@ -92,6 +100,7 @@ export function useReferenceCapture(reference: ReferenceRecord, attachments: rea
 
   const current = payloads.key === referenceKey ? payloads : null;
   const mine = corrected && corrected.key === reference.id ? corrected : null;
+  const fromPdf = mine ? mine.fromPdf === true : card?.current?.textVia === "editor_pdf";
   const text = mine?.text ?? (textSortKey ? current?.text ?? null : fallback.key === fallbackKey ? fallback.text : null);
   const loading = mine ? false : textSortKey ? !(current?.loaded) : useFallback && fallback.key === fallbackKey ? fallback.loading : useFallback;
   const error = textSortKey || mine ? null : fallback.key === fallbackKey ? fallback.error : null;
@@ -103,8 +112,27 @@ export function useReferenceCapture(reference: ReferenceRecord, attachments: rea
 
   const correct = useCallback(async (input: { text?: string; assessment: CaptureAssessment; note?: string }) => {
     if (correctAction) await correctAction({ referenceId: reference.id, ...(input.text !== undefined ? { text: input.text } : {}), assessment: input.assessment, ...(input.note ? { note: input.note } : {}) });
-    setCorrected({ key: reference.id, text: input.text ?? text ?? "", assessment: input.assessment });
-  }, [reference.id, text, correctAction]);
+    setCorrected({ key: reference.id, text: input.text ?? text ?? "", assessment: input.assessment, fromPdf: input.text === undefined && fromPdf });
+  }, [reference.id, text, fromPdf, correctAction]);
+
+  const uploadPdf = useCallback(async (input: { file: File; assessment: CaptureAssessment; note?: string }, onStep?: (step: "uploading" | "reading") => void) => {
+    if (!correctAction) throw new Error("Uploading is not available here.");
+    if (input.file.size > MAXIMUM_PDF_UPLOAD_BYTES) throw new Error(`That PDF is larger than we read (${MAXIMUM_PDF_UPLOAD_BYTES / 1_000_000} MB). Paste the text instead.`);
+    onStep?.("uploading");
+    const attachment = await uploadModelPayloadForOwner({
+      ownerKind: "reference", ownerId: reference.id, ownerLineageId: reference.lineageId ?? reference.id, ownerVersionNumber: reference.versionNumber ?? null,
+      ownerVersionKey: `reference#${reference.id}`, role: PDF_UPLOAD_ROLE, sortKey: `${PDF_UPLOAD_ROLE}-${Date.now().toString(36)}`,
+      filename: input.file.name || "upload.pdf", mediaType: "application/pdf", content: input.file,
+    });
+    onStep?.("reading");
+    const result = await correctAction({ referenceId: reference.id, pdfAttachmentId: attachment.id, assessment: input.assessment, ...(input.note ? { note: input.note } : {}) });
+    // The server read the text; fetch the new body it stored so the panel shows it.
+    const sortKey = (result as { edits?: Array<{ sortKey?: string; kind?: string }> } | null | undefined)?.edits?.find((edit) => edit.kind === "text")?.sortKey;
+    const rows = sortKey ? await loadModelPayloadsForOwner("reference", reference.id, ["metadata"]) : [];
+    const stored = rows.find((row) => row.attachment.sortKey === sortKey)?.text;
+    if (typeof stored !== "string") throw new Error("The PDF was read, but its text could not be loaded. Reload to see it.");
+    setCorrected({ key: reference.id, text: stored, assessment: input.assessment, fromPdf: true });
+  }, [reference.id, reference.lineageId, reference.versionNumber, correctAction]);
 
-  return { status, text, loading, error, summary, summaryStale: Boolean(mine) || status.summaryStale === true, rows, correct: correctAction || isLocalPreviewEnabled() ? correct : undefined };
+  return { status, text, loading, error, summary, summaryStale: Boolean(mine) || status.summaryStale === true, rows, fromPdf, correct: correctAction || isLocalPreviewEnabled() ? correct : undefined, uploadPdf: correctAction ? uploadPdf : undefined };
 }

package/components/reference-capture-panel.tsx

@@ -1,7 +1,7 @@
 "use client";
 
-import { FileCheckIcon, FileClockIcon, FileWarningIcon, FileXIcon, Loader2Icon } from "lucide-react";
-import { useEffect, useState } from "react";
+import { FileCheckIcon, FileClockIcon, FileUpIcon, FileWarningIcon, FileXIcon, Loader2Icon } from "lucide-react";
+import { useEffect, useRef, useState } from "react";
 import { captureLine, type CaptureAssessment, type CaptureState, type CaptureStatus } from "../lib/reference-capture";
 import { cn } from "../lib/utils";
 import { AutoGrowTextarea } from "./ui/auto-grow-textarea";
@@ -42,6 +42,87 @@ const ASSESSMENTS: Array<{ key: CaptureAssessment; label: string; hint: string }
   { key: "incorrect", label: "Incorrect", hint: "This is not the right text." },
 ];
 
+/** The editor's verdict on the machine capture; asked the same way whether they paste text or upload a PDF. */
+function AssessmentChoice({ value, onChange }: { value: CaptureAssessment | null; onChange: (value: CaptureAssessment) => void }) {
+  return (
+    <div aria-label="Was the capture complete?" className="flex flex-wrap gap-2" role="group">
+      {ASSESSMENTS.map((option) => (
+        <button
+          aria-pressed={value === option.key}
+          className={cn("inline-flex min-h-9 items-center rounded-full px-3 text-sm font-medium outline-none focus-visible:ring-2 focus-visible:ring-ring", value === option.key ? "bg-[var(--primary)] text-[var(--primary-foreground)]" : "bg-[var(--feed-card)]")}
+          data-capture-assessment={option.key}
+          key={option.key}
+          onClick={() => onChange(option.key)}
+          title={option.hint}
+          type="button"
+        >
+          {option.label}
+        </button>
+      ))}
+    </div>
+  );
+}
+
+/** "Upload a PDF": the editor's verdict, then the PDF they downloaded themselves; the server reads its text. */
+function PdfUploadForm({ capture, onDone, onCancel }: { capture: CaptureView; onDone: () => void; onCancel: () => void }) {
+  const [assessment, setAssessment] = useState<CaptureAssessment | null>(null);
+  const [note, setNote] = useState("");
+  const [phase, setPhase] = useState<"idle" | "uploading" | "reading" | "error">("idle");
+  const [message, setMessage] = useState<string | null>(null);
+  const input = useRef<HTMLInputElement>(null);
+  const reason = !assessment ? "Say whether the capture was complete, incomplete or incorrect, then choose the PDF." : null;
+  const busy = phase === "uploading" || phase === "reading";
+  async function upload(file: File) {
+    if (!assessment || !capture.uploadPdf) return;
+    setMessage(null);
+    try {
+      await capture.uploadPdf({ file, assessment, ...(note.trim() ? { note: note.trim() } : {}) }, (step) => setPhase(step));
+      onDone();
+    } catch (failure) {
+      setPhase("error");
+      setMessage(failure instanceof Error ? failure.message : "The PDF could not be uploaded.");
+    }
+  }
+  return (
+    <div className="space-y-3 rounded-2xl bg-[var(--feed-control)] p-3" data-capture-upload-form>
+      <p className="text-sm font-medium">Upload the PDF you downloaded. Its text replaces the capture, and the capture is kept.</p>
+      <AssessmentChoice onChange={setAssessment} value={assessment} />
+      <label className="block text-sm font-medium">
+        Note (optional)
+        <AutoGrowTextarea aria-label="Note about the capture (optional)" className="mt-1 block min-h-16 w-full rounded-2xl bg-[var(--feed-card)] p-3 text-sm font-normal" onChange={(event) => setNote(event.target.value)} value={note} />
+      </label>
+      <input
+        accept="application/pdf"
+        aria-label="PDF file"
+        className="hidden"
+        data-capture-upload-input
+        onChange={(event) => {
+          const file = event.target.files?.[0];
+          event.target.value = "";
+          if (file) void upload(file);
+        }}
+        ref={input}
+        type="file"
+      />
+      {reason && !busy ? <p className="text-xs text-muted-foreground" data-capture-upload-reason>{reason}</p> : null}
+      {busy ? <p className="inline-flex items-center gap-1.5 text-xs text-muted-foreground" data-capture-upload-status role="status"><Loader2Icon aria-hidden="true" className="size-3.5 animate-spin" />{phase === "uploading" ? "Uploading the PDF…" : "Reading the text from the PDF…"}</p> : null}
+      {phase === "error" ? <p className="text-xs text-destructive" data-capture-upload-error role="alert">{message}</p> : null}
+      <div className="flex gap-2">
+        <button
+          aria-disabled={reason || busy ? true : undefined}
+          className={cn("inline-flex min-h-9 items-center gap-1.5 rounded-full bg-[var(--primary)] px-4 text-sm font-semibold text-[var(--primary-foreground)] outline-none focus-visible:ring-2 focus-visible:ring-ring", (reason || busy) && "opacity-60")}
+          data-capture-upload-choose
+          onClick={() => { if (!reason && !busy) input.current?.click(); }}
+          type="button"
+        >
+          <FileUpIcon aria-hidden="true" className="size-4" />Choose the PDF
+        </button>
+        <button className="inline-flex min-h-9 items-center rounded-full bg-[var(--feed-card)] px-4 text-sm font-medium outline-none focus-visible:ring-2 focus-visible:ring-ring" disabled={busy} onClick={onCancel} type="button">Cancel</button>
+      </div>
+    </div>
+  );
+}
+
 /** "Paste or correct the text": an editor's own text and verdict on the capture, saved with `correctReferenceCapture`. */
 function CorrectionForm({ capture, onDone }: { capture: CaptureView; onDone: () => void }) {
   const [text, setText] = useState(capture.text ?? "");
@@ -77,21 +158,7 @@ function CorrectionForm({ capture, onDone }: { capture: CaptureView; onDone: ()
           value={text}
         />
       </label>
-      <div aria-label="Was the capture complete?" className="flex flex-wrap gap-2" role="group">
-        {ASSESSMENTS.map((option) => (
-          <button
-            aria-pressed={assessment === option.key}
-            className={cn("inline-flex min-h-9 items-center rounded-full px-3 text-sm font-medium outline-none focus-visible:ring-2 focus-visible:ring-ring", assessment === option.key ? "bg-[var(--primary)] text-[var(--primary-foreground)]" : "bg-[var(--feed-card)]")}
-            data-capture-assessment={option.key}
-            key={option.key}
-            onClick={() => setAssessment(option.key)}
-            title={option.hint}
-            type="button"
-          >
-            {option.label}
-          </button>
-        ))}
-      </div>
+      <AssessmentChoice onChange={setAssessment} value={assessment} />
       <label className="block text-sm font-medium">
         Note (optional)
         <AutoGrowTextarea aria-label="Note about the capture (optional)" className="mt-1 block min-h-16 w-full rounded-2xl bg-[var(--feed-card)] p-3 text-sm font-normal" onChange={(event) => setNote(event.target.value)} value={note} />
@@ -123,6 +190,7 @@ export function ReferenceCapturePanel({ capture, showSummary = true, startEditin
 }) {
   const [expanded, setExpanded] = useState(false);
   const [editing, setEditing] = useState(startEditing && Boolean(capture.correct));
+  const [uploading, setUploading] = useState(false);
   const [savedNote, setSavedNote] = useState(false);
   const line = captureLine(capture.status);
   const Icon = ICON[line.state];
@@ -145,6 +213,7 @@ export function ReferenceCapturePanel({ capture, showSummary = true, startEditin
       {capture.loading ? <p className="inline-flex items-center gap-1.5 text-xs text-muted-foreground" role="status"><Loader2Icon aria-hidden="true" className="size-3.5 animate-spin" />Reading the stored text…</p> : null}
       {capture.error ? <p className="text-xs text-destructive" role="alert">The stored text could not be read. {capture.error}</p> : null}
       {/* While the editor is open it holds the text, so the read-only copy steps aside: the text is never shown twice. */}
+      {text && capture.fromPdf ? <p className="text-xs font-medium" data-capture-from-pdf>Text from a PDF you uploaded</p> : null}
       {text && !editing ? (
         <div className="space-y-2">
           {/* No scroll box of its own: on a phone a swipe here scrolls the page. "Show all" grows it in place. */}
@@ -175,15 +244,24 @@ export function ReferenceCapturePanel({ capture, showSummary = true, startEditin
           ))}
         </dl>
       ) : null}
-      {capture.correct ? (
-        editing ? <CorrectionForm capture={capture} onDone={() => { setEditing(false); setSavedNote(true); }} /> : (
-          <div className="flex flex-wrap items-center gap-2">
-            <button className="inline-flex min-h-9 items-center rounded-full bg-[var(--feed-control)] px-4 text-sm font-medium outline-none focus-visible:ring-2 focus-visible:ring-ring" data-capture-correct onClick={() => { setSavedNote(false); setEditing(true); }} type="button">
-              Paste or correct the text
-            </button>
-            {savedNote ? <span className="text-xs text-muted-foreground" data-capture-correction-saved role="status">Saved. Your verdict trains the completeness check.</span> : null}
-          </div>
-        )
+      {capture.correct || capture.uploadPdf ? (
+        editing && capture.correct ? <CorrectionForm capture={capture} onDone={() => { setEditing(false); setSavedNote(true); }} />
+          : uploading && capture.uploadPdf ? <PdfUploadForm capture={capture} onCancel={() => setUploading(false)} onDone={() => { setUploading(false); setSavedNote(true); }} />
+          : (
+            <div className="flex flex-wrap items-center gap-2">
+              {capture.correct ? (
+                <button className="inline-flex min-h-9 items-center rounded-full bg-[var(--feed-control)] px-4 text-sm font-medium outline-none focus-visible:ring-2 focus-visible:ring-ring" data-capture-correct onClick={() => { setSavedNote(false); setEditing(true); }} type="button">
+                  Paste or correct the text
+                </button>
+              ) : null}
+              {capture.uploadPdf ? (
+                <button className="inline-flex min-h-9 items-center gap-1.5 rounded-full bg-[var(--feed-control)] px-4 text-sm font-medium outline-none focus-visible:ring-2 focus-visible:ring-ring" data-capture-upload onClick={() => { setSavedNote(false); setUploading(true); }} type="button">
+                  <FileUpIcon aria-hidden="true" className="size-4" />Upload a PDF
+                </button>
+              ) : null}
+              {savedNote ? <span className="text-xs text-muted-foreground" data-capture-correction-saved role="status">Saved. Your verdict trains the completeness check.</span> : null}
+            </div>
+          )
       ) : null}
     </section>
   );

package/lib/content-actions-client.ts

@@ -35,12 +35,14 @@ export type CorrectReferenceCaptureInput = {
   referenceId: string;
   /** The full corrected or pasted text. */
   text?: string;
+  /** The id of a PDF the editor uploaded to this Reference (a ModelAttachment); the server reads its text. Sent instead of `text`. */
+  pdfAttachmentId?: string;
   grobid?: Record<string, unknown>;
   /** Required with text; alone it is just a label on the capture. */
   assessment?: CaptureAssessment;
   note?: string;
 };
-export type CorrectReferenceCaptureResult = { ok: true; referenceId: string; status: string; edits: unknown[]; label: unknown; summaryStale: boolean };
+export type CorrectReferenceCaptureResult = { ok: true; referenceId: string; status: string; edits: Array<{ sortKey?: string; via?: string }>; label: unknown; summaryStale: boolean };
 
 export class ContentActionFailure extends Error {
   readonly errors: ContentActionError[];

package/lib/reference-capture.ts

@@ -23,7 +23,7 @@ export type CurationCard = {
   editedSignals: Record<string, unknown> | null;
   summary: { status?: string; text?: string; stale?: boolean; reason?: string } | null;
   grobid: { status?: string; reason?: string; title?: string; authorCount?: number; hasAbstract?: boolean; sectionCount?: number; referenceCount?: number; fieldsSortKey?: string } | null;
-  current: { textSortKey?: string; grobidSortKey?: string } | null;
+  current: { textSortKey?: string; grobidSortKey?: string; textVia?: string } | null;
   labels: Array<{ kind?: string; assessment?: string; labeledAt?: string }>;
 };
 

package/package.json

@@ -500,7 +500,7 @@
       "optional": true
     }
   },
-  "version": "5.51.0",
+  "version": "5.53.0",
   "dependencies": {
     "@aws-amplify/adapter-nextjs": "^1.7.3",
     "@aws-crypto/sha256-js": "5.2.0",

package/routes.manifest.json

@@ -1,5 +1,5 @@
 {
-  "version": "5.51.0",
+  "version": "5.53.0",
   "routes": [
     {
       "file": "app/%5F_preview/[[...path]]/route.ts",