ConusAI vendor v0.1.0-dev.20261027+9b1131fb

modules/documents/core/export.ts

@@ -0,0 +1,73 @@
+// The export: one workbook of the lines a person approved. It reads `documents_line`, which only an
+// approval writes (target.ts), so there is no filter here that a later change could forget: a line
+// that is pending or rejected is not in the table this reads.
+//
+// The columns are fixed and in English on purpose. The file is an interchange format -- what a
+// product's own import or a spreadsheet formula refers to by name -- and a header that changed with
+// the reviewer's language would break both. A product that needs another layout (a customer's ERP
+// import, say) adds its own exporter beside this one; that belongs to the customer's module.
+
+import { conflict } from "@conusai-kit/core"
+import { getFile } from "@conusai-kit/files"
+import { documents, lines } from "../contract/index.js"
+import type { DocumentScope, DocumentsDeps } from "./lines.js"
+import { buildWorkbook, type Cell } from "./xlsx.js"
+
+export const WORKBOOK_CONTENT_TYPE = "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet"
+const PAGE = 200
+
+export const EXPORT_HEADER = ["Label", "Value", "Page", "Source quote", "Edited by a reviewer"] as const
+
+/** `report.final.pdf` becomes `report.final-approved.xlsx`; nothing a path or a header could misread. */
+export function exportFilename(source: string): string {
+  const stem = source.replace(/\.[A-Za-z0-9]{1,5}$/u, "")
+  // biome-ignore lint/suspicious/noControlCharactersInRegex: control characters are what is removed.
+  const safe = stem.replace(/[\u0000-\u001f\u007f"\\/:*?<>|]/gu, "_").trim()
+  return `${safe.length > 0 ? safe : "document"}-approved.xlsx`
+}
+
+export async function exportApproved(
+  deps: DocumentsDeps,
+  scope: DocumentScope,
+  fileId: string,
+): Promise<{ filename: string; contentType: string; bytes: Uint8Array; lineCount: number }> {
+  const file = await getFile(deps.files, scope, fileId)
+  const document = (
+    await deps.data
+      .forScope(scope)
+      .of(documents)
+      .list({ where: [{ field: "fileId", op: "eq", value: file.fileId }], limit: 1 })
+  ).items[0]
+  if (!document) throw conflict("this document has not been read yet")
+
+  const rows: Cell[][] = [[...EXPORT_HEADER]]
+  let cursor: string | undefined
+  do {
+    const page = await deps.data
+      .forScope(scope)
+      .of(lines)
+      .list({
+        where: [{ field: "documentId", op: "eq", value: document.id }],
+        order: [
+          { field: "page", direction: "asc" },
+          { field: "position", direction: "asc" },
+        ],
+        limit: PAGE,
+        ...(cursor === undefined ? {} : { cursor }),
+      })
+    for (const line of page.items) {
+      rows.push([line.label, line.value, line.page, line.quote, line.edited ? "yes" : "no"])
+    }
+    cursor = page.hasMore ? page.cursor : undefined
+  } while (cursor !== undefined)
+
+  // An empty workbook would read as "the document had nothing in it", which is not what was
+  // decided: nobody has approved anything yet.
+  if (rows.length === 1) throw conflict("no line of this document has been approved")
+  return {
+    filename: exportFilename(file.filename),
+    contentType: WORKBOOK_CONTENT_TYPE,
+    bytes: buildWorkbook("Approved lines", rows),
+    lineCount: rows.length - 1,
+  }
+}