feat(import): parse a PDF resume without an AI provider (#3400)

* feat(import): parse a PDF resume without an AI provider

Importing a PDF required a connected AI provider, so anyone without a
paid API key could only import the three JSON formats. Almost nobody
arrives with one of those files; they arrive with a PDF. The first thing
a new user tries to do was blocked behind bringing their own key.

Adds a deterministic parser that reads the text out of the PDF in the
browser and prefills the builder. It pulls the contact block, segments
the body on conventional headings, and maps entries to real items,
reusing the ATS period parser for dates so a date range is not mistaken
for a phone number.

Nothing is thrown away: header parts that do not map to a field go into
the description, and unrecognized headings become custom sections. The
imported sections are placed on the page so the result renders straight
away. Output is validated against the resume schema before it is
returned.

Text extraction groups items by baseline rather than trusting hasEOL,
and turns wide column gaps into a double space, which is what lets a
row split into company, position and location.

The AI path still runs when a provider is connected. Word import is
unchanged and still requires one.

Closes #3334

* fix(import): keep every section and entry the PDF actually contains

Review found three ways the parser lost or mangled content, all of them
reproducible.

A document whose first heading was not one of the known aliases never
started a section, because unknown-heading detection was gated on a
section already being open. Everything after it was swallowed as contact
header text. The header block is now bounded by where the contact
details stop, so a heading is recognized wherever it appears.

An entry spreading company, position and dates over three lines was
imported as two malformed items. A line that introduces an entry now
merges into the open entry instead of starting a second one.

An uppercase company such as ACME CORPORATION was read as a section
heading and fragmented the entry. A heading candidate followed by a date
line is now treated as an entry header, which is what it is.

Also escape single quotes, and construct the PDF worker inside the try
so the nested worker is terminated even if construction throws.

Title-case headings are deliberately still not treated as headings:
company and school names are title case too, and splitting on them would
fragment real entries. Such a section stays in the preceding one with its
text intact rather than risking loss.

* fix(import): look past a multi-line preamble before calling a line a heading

The previous guard only inspected the next line, so an uppercase company
followed by a separate role line and then the dates was still read as a
section heading. The experience or education entry was moved into a
custom section and lost.

Heading detection now scans a two-line window for the date that marks an
entry, and stops early at a bullet so a genuine heading whose section
opens with bullet points is still recognized.

The window can suppress a real heading whose first entry puts a bare date
two lines below it. That is the deliberate direction to fail in: a missed
heading leaves the text in the preceding section, while a misread entry
fragments structured content.

* fix(import): collect an entry preamble until its dates appear

An entry that spread company, role, location and dates over four lines
was imported as two broken items: the company with no dates, and the
location carrying the period.

The cause was in entry grouping rather than heading detection. Lines
before a date were only folded into the entry header when the date sat
on the very next line; anything earlier fell through to the description.
Preamble lines are now collected into the entry header until the dates
turn up, bounded by the same lookahead and stopping at a bullet, so an
undated section cannot swallow itself.

The heading lookahead widens to four lines to match, which is the
realistic maximum for company, role, location and dates.

* fix(import): harden local PDF resume parsing

* chore(import): document audited HTML construction

---------

Co-authored-by: Amruth Pillai <im.amruth@gmail.com>

apps/web/src/dialogs/resume/import.dialog.test.tsx

@@ -88,25 +88,54 @@ const renderDialog = () => {
 	);
 };
 
-// A real "%PDF" header so the dialog auto-detects the type and shows the provider notice.
-const createPdfFile = () =>
-	new File([new Uint8Array([0x25, 0x50, 0x44, 0x46])], "resume.pdf", { type: "application/pdf" });
+// Word still needs a provider, so it is what surfaces the notice. PDF falls back to a local parse.
+const createWordFile = () =>
+	new File([new Uint8Array([0x50, 0x4b, 0x03, 0x04])], "resume.docx", {
+		type: "application/vnd.openxmlformats-officedocument.wordprocessingml.document",
+	});
 
 // The dialog renders through a portal, so query the document rather than the render container.
-async function selectPdfFile() {
+async function selectWordFile() {
 	const input = document.querySelector<HTMLInputElement>('input[type="file"]');
 	if (!input) throw new Error("File input not found");
 
-	fireEvent.change(input, { target: { files: [createPdfFile()] } });
+	fireEvent.change(input, { target: { files: [createWordFile()] } });
 
 	return await screen.findByText("Set up a provider");
 }
 
+const createPdfFile = () =>
+	new File([new Uint8Array([0x25, 0x50, 0x44, 0x46])], "resume.pdf", { type: "application/pdf" });
+
+describe("ImportResumeDialog — PDF without a provider", () => {
+	it("offers a local parse instead of demanding an AI provider", async () => {
+		renderDialog();
+		const input = document.querySelector<HTMLInputElement>('input[type="file"]');
+		if (!input) throw new Error("File input not found");
+
+		fireEvent.change(input, { target: { files: [createPdfFile()] } });
+
+		expect(await screen.findByText(/read the text out of the PDF here in your browser/)).toBeInTheDocument();
+		expect(screen.queryByText("Set up a provider")).not.toBeInTheDocument();
+	});
+
+	it("keeps the import button usable", async () => {
+		renderDialog();
+		const input = document.querySelector<HTMLInputElement>('input[type="file"]');
+		if (!input) throw new Error("File input not found");
+
+		fireEvent.change(input, { target: { files: [createPdfFile()] } });
+		await screen.findByText(/read the text out of the PDF here in your browser/);
+
+		expect(screen.getByRole("button", { name: "Import" })).not.toBeDisabled();
+	});
+});
+
 describe("ImportResumeDialog — Set up a provider", () => {
 	// https://github.com/amruthpillai/reactive-resume/issues/3307
 	it("confirms before leaving instead of navigating behind the dialog", async () => {
 		renderDialog();
-		const link = await selectPdfFile();
+		const link = await selectWordFile();
 
 		fireEvent.click(link);
 
@@ -118,7 +147,7 @@ describe("ImportResumeDialog — Set up a provider", () => {
 
 	it("stays put and keeps the selected file when the user cancels", async () => {
 		renderDialog();
-		const link = await selectPdfFile();
+		const link = await selectWordFile();
 
 		fireEvent.click(link);
 		fireEvent.click(await screen.findByText("Stay"));
@@ -130,12 +159,12 @@ describe("ImportResumeDialog — Set up a provider", () => {
 		expect(routerNavigate).not.toHaveBeenCalled();
 		expect(navigate).not.toHaveBeenCalled();
 		expect(useDialogStore.getState().open).toBe(true);
-		expect(screen.getByText("resume.pdf")).toBeInTheDocument();
+		expect(screen.getByText("resume.docx")).toBeInTheDocument();
 	});
 
 	it("closes the dialog and navigates once the user confirms", async () => {
 		renderDialog();
-		const link = await selectPdfFile();
+		const link = await selectWordFile();
 
 		fireEvent.click(link);
 		fireEvent.click(await screen.findByText("Leave"));
@@ -152,7 +181,7 @@ describe("ImportResumeDialog — Set up a provider", () => {
 
 	it("leaves modifier clicks to the browser so the link can open in a new tab", async () => {
 		renderDialog();
-		const link = await selectPdfFile();
+		const link = await selectWordFile();
 
 		const event = new MouseEvent("click", { bubbles: true, cancelable: true, metaKey: true });
 		fireEvent(link, event);
@@ -164,7 +193,7 @@ describe("ImportResumeDialog — Set up a provider", () => {
 
 	it("leaves middle clicks to the browser too", async () => {
 		renderDialog();
-		const link = await selectPdfFile();
+		const link = await selectWordFile();
 
 		const event = new MouseEvent("click", { bubbles: true, cancelable: true, button: 1 });
 		fireEvent(link, event);

apps/web/src/dialogs/resume/import.tsx

@@ -141,10 +141,15 @@ export function ImportResumeDialog(_: DialogProps<"resume.import">) {
 
 			setIsImporting(true);
 
+			// A PDF parsed in the browser never touches a provider, so promising one would be a lie.
+			const isLocalPdf = value.type === "pdf" && !hasUsableProvider;
+
 			const toastId = toast.add({
 				type: "loading",
 				title: t`Importing your resume...`,
-				description: t`This may take a few minutes, depending on the response of the AI provider. Please do not close the window or refresh the page.`,
+				description: isLocalPdf
+					? t`This may take a moment. Please do not close the window or refresh the page.`
+					: t`This may take a few minutes, depending on the response of the AI provider. Please do not close the window or refresh the page.`,
 			});
 
 			try {
@@ -160,14 +165,31 @@ export function ImportResumeDialog(_: DialogProps<"resume.import">) {
 
 				if (value.type === "pdf") {
 					if (isLoadingAiProviders) throw new Error(t`Loading AI providers. Please try again in a moment.`);
-					if (!hasUsableProvider)
-						throw new Error(t`This feature requires a connected AI provider. Please set one up in the settings.`);
 
-					const base64 = await fileToBase64(value.file);
+					if (hasUsableProvider) {
+						const base64 = await fileToBase64(value.file);
 
-					data = await client.ai.parsePdf({
-						file: { name: value.file.name, data: base64 },
-					});
+						data = await client.ai.parsePdf({
+							file: { name: value.file.name, data: base64 },
+						});
+					} else {
+						const [{ extractPdfLines }, { parseResumeText }] = await Promise.all([
+							import("@/features/resume/import/pdf-text"),
+							import("@reactive-resume/import/plain-text"),
+						]);
+
+						const lines = await extractPdfLines(value.file);
+						if (lines.length === 0) {
+							throw new Error(
+								t({
+									comment: "Error shown when a PDF has no extractable text layer during import",
+									message: "This PDF has no readable text. It is likely a scan, so there is nothing to import.",
+								}),
+							);
+						}
+
+						data = parseResumeText(lines.join("\n"));
+					}
 				}
 
 				if (value.type === "docx") {
@@ -236,7 +258,8 @@ export function ImportResumeDialog(_: DialogProps<"resume.import">) {
 
 	const type = useStore(form.store, (s) => s.values.type);
 	const file = useStore(form.store, (s) => s.values.file);
-	const aiRequired = type === "pdf" || type === "docx";
+	const aiRequired = type === "docx";
+	const pdfWithoutAi = type === "pdf" && !isLoadingAiProviders && !hasUsableProvider;
 
 	const onSelectFile = () => {
 		if (!inputRef.current) return;
@@ -372,12 +395,7 @@ export function ImportResumeDialog(_: DialogProps<"resume.import">) {
 												{
 													value: "pdf",
 													textValue: t({ comment: "File format label in import source selector", message: "PDF" }),
-													label: (
-														<div className="flex items-center gap-x-2">
-															{t({ comment: "File format label in import source selector", message: "PDF" })}{" "}
-															<Badge>{t`AI`}</Badge>
-														</div>
-													),
+													label: t({ comment: "File format label in import source selector", message: "PDF" }),
 												},
 												{
 													value: "docx",
@@ -413,7 +431,7 @@ export function ImportResumeDialog(_: DialogProps<"resume.import">) {
 				{aiRequired && !isLoadingAiProviders && !hasUsableProvider && (
 					<div className="flex flex-col gap-3 rounded-md border border-dashed p-3 text-sm lg:flex-row lg:items-center lg:justify-between">
 						<span className="text-muted-foreground">
-							<Trans>Importing from PDF or Word requires a connected AI provider.</Trans>
+							<Trans>Importing from Word requires a connected AI provider.</Trans>
 						</span>
 						<Button
 							size="sm"
@@ -428,6 +446,15 @@ export function ImportResumeDialog(_: DialogProps<"resume.import">) {
 					</div>
 				)}
 
+				{pdfWithoutAi && (
+					<div className="rounded-md border border-dashed p-3 text-muted-foreground text-sm">
+						<Trans>
+							No AI provider is connected, so we will read the text out of the PDF here in your browser and fill in what
+							we can recognize. Expect to tidy up the result.
+						</Trans>
+					</div>
+				)}
+
 				<DialogFooter>
 					<Button type="submit" disabled={!type || !file || isImporting || (aiRequired && !hasUsableProvider)}>
 						{isImporting ? <Spinner /> : null}

apps/web/src/features/ats-checker/extract-client.ts

@@ -20,6 +20,8 @@ export type ExtractProgress = HarvestProgress | { phase: "loading"; page: 0; pag
 export type ExtractOptions = {
 	onProgress?: (progress: ExtractProgress) => void;
 	signal?: AbortSignal;
+	/** Set to 0 to skip the operator pass; a caller that only wants the text layer has no use for it. */
+	operatorBudgetMs?: number;
 };
 
 export class PdfPasswordRequiredError extends Error {
@@ -81,6 +83,7 @@ export async function extractPdf(file: File, options: ExtractOptions = {}): Prom
 		return await harvestPdfDocument(document, {
 			file: { name: file.name, sizeBytes: file.size, magicBytesOk },
 			maxPages: HARVEST_DEFAULTS.maxPages,
+			...(options.operatorBudgetMs === undefined ? {} : { operatorBudgetMs: options.operatorBudgetMs }),
 			...(options.onProgress ? { onProgress: options.onProgress } : {}),
 			...(options.signal ? { signal: options.signal } : {}),
 		});

apps/web/src/features/resume/import/pdf-text.test.ts

@@ -0,0 +1,117 @@
+import type { RawExtraction } from "@reactive-resume/resume/ats-pdf";
+import { describe, expect, it } from "vitest";
+import { buildExtractedDocument } from "@reactive-resume/resume/ats-pdf";
+import { documentToLines } from "./pdf-text";
+
+const PAGE = { width: 595, height: 842 };
+const FONT_SIZE = 10;
+const CHAR_WIDTH = 5;
+
+type FixtureItem = { text: string; x: number; y: number; width?: number };
+
+/** `y` is measured from the top of the page; the builder flips it into the PDF's own space. */
+function rawExtraction(pages: FixtureItem[][]): RawExtraction {
+	return {
+		version: 1,
+		file: { name: "resume.pdf", sizeBytes: 1024, magicBytesOk: true },
+		pageCount: pages.length,
+		truncated: false,
+		pages: pages.map((items, index) => ({
+			pageNumber: index + 1,
+			width: PAGE.width,
+			height: PAGE.height,
+			rotation: 0,
+			items: items.map((item) => ({
+				str: item.text,
+				transform: [FONT_SIZE, 0, 0, FONT_SIZE, item.x, PAGE.height - item.y - FONT_SIZE] as const,
+				width: item.width ?? item.text.length * CHAR_WIDTH,
+				height: FONT_SIZE,
+				fontRef: "f1",
+				hasEol: false,
+			})),
+			operators: null,
+		})),
+		fonts: [],
+		links: [],
+		metadata: {
+			producer: null,
+			creator: null,
+			title: null,
+			author: null,
+			language: null,
+			pdfVersion: null,
+			isEncrypted: false,
+			isXfa: false,
+			isCollection: false,
+			isTagged: false,
+			hasAcroForm: false,
+		},
+		operatorsAvailable: false,
+	};
+}
+
+const linesOf = (...pages: FixtureItem[][]) => documentToLines(buildExtractedDocument(rawExtraction(pages)));
+
+describe("documentToLines", () => {
+	it("groups items on the same baseline into one line", () => {
+		expect(
+			linesOf([
+				{ text: "Ada", x: 40, y: 60 },
+				{ text: "Lovelace", x: 58, y: 60 },
+			]),
+		).toEqual(["Ada Lovelace"]);
+	});
+
+	it("orders lines from the top of the page down", () => {
+		expect(
+			linesOf([
+				{ text: "Second", x: 40, y: 80 },
+				{ text: "First", x: 40, y: 60 },
+			]),
+		).toEqual(["First", "Second"]);
+	});
+
+	it("turns a wide field gap into a double space the parser can split on", () => {
+		expect(
+			linesOf([
+				{ text: "Acme", x: 40, y: 60 },
+				{ text: "Engineer", x: 120, y: 60 },
+			]),
+		).toEqual(["Acme  Engineer"]);
+	});
+
+	it("reads every page in order", () => {
+		expect(linesOf([{ text: "Page one", x: 40, y: 60 }], [{ text: "Page two", x: 40, y: 60 }])).toEqual([
+			"Page one",
+			"Page two",
+		]);
+	});
+
+	it("drops blank items and blank lines", () => {
+		expect(
+			linesOf([
+				{ text: "   ", x: 40, y: 60 },
+				{ text: "Ada", x: 40, y: 80 },
+			]),
+		).toEqual(["Ada"]);
+	});
+
+	it("reads each column of a sidebar layout in full instead of interleaving them", () => {
+		const sidebar = ["SKILLS", "TypeScript", "Rust", "PostgreSQL", "Docker", "Figma"];
+		const main = [
+			"EXPERIENCE",
+			"Acme Corp",
+			"Senior Engineer",
+			"Jan 2020 - Present",
+			"Led the rewrite",
+			"Grew the team",
+		];
+
+		expect(
+			linesOf([
+				...sidebar.map((text, index) => ({ text, x: 40, y: 60 + index * 20 })),
+				...main.map((text, index) => ({ text, x: 220, y: 70 + index * 20 })),
+			]),
+		).toEqual([...sidebar, ...main]);
+	});
+});

apps/web/src/features/resume/import/pdf-text.ts

@@ -0,0 +1,82 @@
+import type { ExtractedDocument } from "@reactive-resume/resume/ats-pdf";
+import { buildExtractedDocument } from "@reactive-resume/resume/ats-pdf";
+import { extractPdf } from "@/features/ats-checker/extract-client";
+
+/**
+ * Turns a PDF into the plain lines `parseResumeText` reads.
+ *
+ * The extraction itself is the ATS checker's — same worker configuration, same size and password
+ * guards, same font-relative line clustering and column-gutter detection. Only the joining differs:
+ * the ATS text is prose, while the importer splits header rows on runs of two or more spaces, so a
+ * tabbed "Company    Role    Location" has to survive as three fields rather than one phrase.
+ */
+
+/** Horizontal gap, relative to font size, above which two spans are separate header fields. */
+const FIELD_GAP_RATIO = 1.5;
+/** Narrower gaps are still a word break; matches the ratio the ATS extractor joins prose with. */
+const SPACE_GAP_RATIO = 0.25;
+
+/** The evidence the ATS layout rules require before calling a page two-column. */
+const GUTTER_COVERAGE = 0.85;
+const GUTTER_SPLIT_RATIO = 0.25;
+
+type PageGeometry = ExtractedDocument["pages"][number];
+type TextSpan = PageGeometry["lines"][number]["spans"][number];
+
+function joinSpans(spans: readonly TextSpan[]): string {
+	let text = "";
+	let previous: TextSpan | null = null;
+
+	for (const span of spans) {
+		if (previous) {
+			const gap = span.x - (previous.x + previous.width);
+			const fontSize = Math.max(previous.fontSize, span.fontSize, 1);
+
+			if (gap > FIELD_GAP_RATIO * fontSize) text += "  ";
+			else if (gap > SPACE_GAP_RATIO * fontSize && !/\s$/.test(text) && !/^\s/.test(span.text)) text += " ";
+		}
+
+		text += span.text;
+		previous = span;
+	}
+
+	return text.trim();
+}
+
+/** Mid-point of a convincing gutter, or null on a page that reads as one column. */
+function columnSplitX(page: PageGeometry): number | null {
+	const gutter = page.gutter;
+	if (!gutter || gutter.coverage < GUTTER_COVERAGE || gutter.splitRatio < GUTTER_SPLIT_RATIO) return null;
+
+	return gutter.x + gutter.width / 2;
+}
+
+/**
+ * Lines grouped by baseline alone interleave a sidebar with the body, which drops a whole work
+ * history into the skills section. On a two-column page each column is read out in full instead.
+ */
+function pageToLines(page: PageGeometry): string[] {
+	const splitX = columnSplitX(page);
+	if (splitX === null) return page.lines.map((line) => joinSpans(line.spans));
+
+	const left: string[] = [];
+	const right: string[] = [];
+
+	for (const line of page.lines) {
+		const leftSpans = line.spans.filter((span) => span.x + span.width / 2 < splitX);
+		if (leftSpans.length > 0) left.push(joinSpans(leftSpans));
+
+		const rightSpans = line.spans.filter((span) => span.x + span.width / 2 >= splitX);
+		if (rightSpans.length > 0) right.push(joinSpans(rightSpans));
+	}
+
+	return [...left, ...right];
+}
+
+export function documentToLines(doc: ExtractedDocument): string[] {
+	return doc.pages.flatMap(pageToLines).filter((line) => line.length > 0);
+}
+
+export async function extractPdfLines(file: File): Promise<string[]> {
+	return documentToLines(buildExtractedDocument(await extractPdf(file, { operatorBudgetMs: 0 })));
+}

packages/import/package.json

@@ -5,6 +5,7 @@
 	"private": true,
 	"exports": {
 		"./json-resume": "./src/json-resume.tsx",
+		"./plain-text": "./src/plain-text.ts",
 		"./reactive-resume-json": "./src/reactive-resume-json.tsx",
 		"./reactive-resume-v4-json": "./src/reactive-resume-v4-json.tsx"
 	},

packages/import/src/plain-text.test.ts

@@ -0,0 +1,321 @@
+import { describe, expect, it } from "vitest";
+import { resumeDataSchema } from "@reactive-resume/schema/resume/data";
+import { parseResumeText } from "./plain-text";
+
+const SAMPLE = `Ada Lovelace
+Senior Software Engineer
+Berlin, Germany | ada@example.com | +44 20 7946 0100 | https://ada.dev
+
+SUMMARY
+Engineer with 10 years building analytical systems.
+
+WORK EXPERIENCE
+Analytical Engines  Senior Engineer  Berlin
+Jan 2020 - Present
+• Led the difference engine rewrite
+• Mentored four junior engineers
+Babbage Ltd  Engineer  London
+Mar 2016 - Dec 2019
+• Built the punch card pipeline
+
+EDUCATION
+University of London  BSc Mathematics
+2012 - 2016
+
+SKILLS
+TypeScript, Rust, PostgreSQL
+
+LANGUAGES
+English (Native)
+German (B2)
+
+CERTIFICATIONS
+AWS Solutions Architect  Amazon  2021
+`;
+
+describe("parseResumeText", () => {
+	const data = parseResumeText(SAMPLE);
+
+	it("always returns schema-valid resume data", () => {
+		expect(() => resumeDataSchema.parse(data)).not.toThrow();
+	});
+
+	it("reads the contact block", () => {
+		expect(data.basics).toMatchObject({
+			name: "Ada Lovelace",
+			headline: "Senior Software Engineer",
+			email: "ada@example.com",
+			phone: "+44 20 7946 0100",
+			location: "Berlin, Germany",
+		});
+		expect(data.basics.website.url).toBe("https://ada.dev");
+	});
+
+	it("reads the summary as rich text", () => {
+		expect(data.summary.content).toBe("<p>Engineer with 10 years building analytical systems.</p>");
+	});
+
+	it("splits experience into one entry per role", () => {
+		expect(data.sections.experience.items).toHaveLength(2);
+		expect(data.sections.experience.items[0]).toMatchObject({
+			company: "Analytical Engines",
+			position: "Senior Engineer",
+			location: "Berlin",
+			period: "Jan 2020 - Present",
+		});
+		expect(data.sections.experience.items[1]).toMatchObject({
+			company: "Babbage Ltd",
+			position: "Engineer",
+			period: "Mar 2016 - Dec 2019",
+		});
+	});
+
+	it("keeps bullets as a list in the description", () => {
+		expect(data.sections.experience.items[0]?.description).toBe(
+			"<ul><li>Led the difference engine rewrite</li><li>Mentored four junior engineers</li></ul>",
+		);
+	});
+
+	it("reads education", () => {
+		expect(data.sections.education.items[0]).toMatchObject({
+			school: "University of London",
+			degree: "BSc Mathematics",
+			period: "2012 - 2016",
+		});
+	});
+
+	it("splits a comma separated skills line", () => {
+		expect(data.sections.skills.items.map((item) => item.name)).toEqual(["TypeScript", "Rust", "PostgreSQL"]);
+	});
+
+	it("splits a language from its fluency", () => {
+		expect(data.sections.languages.items).toMatchObject([
+			{ language: "English", fluency: "Native" },
+			{ language: "German", fluency: "B2" },
+		]);
+	});
+
+	it("reads a trailing year as the certification date", () => {
+		expect(data.sections.certifications.items[0]).toMatchObject({
+			title: "AWS Solutions Architect",
+			issuer: "Amazon",
+			date: "2021",
+		});
+	});
+
+	it("places every populated section on the page in document order", () => {
+		expect(data.metadata.layout.pages[0]?.main).toEqual([
+			"summary",
+			"experience",
+			"education",
+			"skills",
+			"languages",
+			"certifications",
+		]);
+	});
+});
+
+describe("parseResumeText edge cases", () => {
+	it("returns usable data for empty input", () => {
+		const data = parseResumeText("");
+		expect(() => resumeDataSchema.parse(data)).not.toThrow();
+		expect(data.basics.name).toBe("");
+		expect(data.metadata.layout.pages[0]?.main).toEqual([]);
+	});
+
+	it("does not mistake a date range for a phone number", () => {
+		const data = parseResumeText("Ada Lovelace\nBerlin\n2016 - 2019\n");
+		expect(data.basics.phone).toBe("");
+	});
+
+	it("keeps an unrecognized heading as a custom section", () => {
+		const data = parseResumeText("Ada\n\nSKILLS\nRust\n\nSPEAKING\nGave a talk at a conference\n");
+		expect(data.customSections).toHaveLength(1);
+		expect(data.customSections[0]).toMatchObject({ type: "summary", title: "SPEAKING" });
+		expect(data.metadata.layout.pages[0]?.main).toContain(data.customSections[0]?.id);
+	});
+
+	it("keeps unclassified header parts in the description rather than dropping them", () => {
+		const data = parseResumeText("EXPERIENCE\nAcme  Engineer  Berlin  Remote  Contract\n2020 - 2022\n");
+		expect(data.sections.experience.items[0]?.description).toContain("Remote");
+		expect(data.sections.experience.items[0]?.description).toContain("Contract");
+	});
+
+	it("escapes markup found in the source text", () => {
+		const data = parseResumeText("SUMMARY\nI write <script>alert(1)</script> safely\n");
+		expect(data.summary.content).toContain("&lt;script&gt;");
+		expect(data.summary.content).not.toContain("<script>");
+	});
+
+	it("treats a heading with a trailing colon as a heading", () => {
+		const data = parseResumeText("Ada\n\nSkills:\nRust, Go\n");
+		expect(data.sections.skills.items.map((item) => item.name)).toEqual(["Rust", "Go"]);
+	});
+});
+
+describe("parseResumeText review findings", () => {
+	it("keeps a section whose heading is the first one in the document", () => {
+		const data = parseResumeText(
+			"Ada Lovelace\nada@example.com\n\nCAREER HIGHLIGHTS\nShipped the difference engine\nMentored the team\n",
+		);
+
+		expect(data.customSections).toHaveLength(1);
+		expect(data.customSections[0]).toMatchObject({ title: "CAREER HIGHLIGHTS" });
+		expect(JSON.stringify(data)).toContain("Shipped the difference engine");
+	});
+
+	it("keeps one entry when company, position and dates sit on separate lines", () => {
+		const data = parseResumeText(
+			"EXPERIENCE\nAnalytical Engines\nSenior Engineer\nJan 2020 - Present\n• Led the rewrite\n",
+		);
+
+		expect(data.sections.experience.items).toHaveLength(1);
+		expect(data.sections.experience.items[0]).toMatchObject({
+			company: "Analytical Engines",
+			position: "Senior Engineer",
+			period: "Jan 2020 - Present",
+		});
+	});
+
+	it("does not turn an uppercase company name into a section heading", () => {
+		const data = parseResumeText("EXPERIENCE\nACME CORPORATION\nJan 2020 - Present\n• Did the work\n");
+
+		expect(data.customSections).toHaveLength(0);
+		expect(data.sections.experience.items[0]).toMatchObject({ company: "ACME CORPORATION" });
+	});
+
+	it("escapes single quotes in extracted text", () => {
+		const data = parseResumeText("SUMMARY\nIt's a resume\n");
+		expect(data.summary.content).toContain("&#39;");
+	});
+});
+
+describe("parseResumeText multi-line entry preambles", () => {
+	it("keeps an uppercase company followed by a separate role line as one entry", () => {
+		const data = parseResumeText(
+			"EXPERIENCE\nACME CORPORATION\nSenior Engineer\nJan 2020 - Present\n• Led the rewrite\n",
+		);
+
+		expect(data.customSections).toHaveLength(0);
+		expect(data.sections.experience.items).toHaveLength(1);
+		expect(data.sections.experience.items[0]).toMatchObject({
+			company: "ACME CORPORATION",
+			position: "Senior Engineer",
+			period: "Jan 2020 - Present",
+		});
+	});
+
+	it("keeps an uppercase school followed by a separate degree line as one entry", () => {
+		const data = parseResumeText("EDUCATION\nUNIVERSITY OF LONDON\nBSc Mathematics\n2012 - 2016\n");
+
+		expect(data.customSections).toHaveLength(0);
+		expect(data.sections.education.items).toHaveLength(1);
+		expect(data.sections.education.items[0]).toMatchObject({
+			school: "UNIVERSITY OF LONDON",
+			degree: "BSc Mathematics",
+			period: "2012 - 2016",
+		});
+	});
+
+	it("still recognizes a real heading whose section starts with bullets", () => {
+		const data = parseResumeText("Ada\nada@example.com\n\nCAREER HIGHLIGHTS\n• Shipped in 2019\n• Grew the team\n");
+
+		expect(data.customSections).toHaveLength(1);
+		expect(data.customSections[0]).toMatchObject({ title: "CAREER HIGHLIGHTS" });
+	});
+});
+
+describe("parseResumeText four-line entry preambles", () => {
+	it("keeps company, role, location and dates as one entry", () => {
+		const data = parseResumeText(
+			"EXPERIENCE\nACME CORPORATION\nSenior Engineer\nBerlin, Germany\nJan 2020 - Present\n• Led the rewrite\n",
+		);
+
+		expect(data.customSections).toHaveLength(0);
+		expect(data.sections.experience.items).toHaveLength(1);
+		expect(data.sections.experience.items[0]).toMatchObject({
+			company: "ACME CORPORATION",
+			position: "Senior Engineer",
+			location: "Berlin, Germany",
+			period: "Jan 2020 - Present",
+		});
+	});
+
+	it("keeps school, degree, location and dates as one entry", () => {
+		const data = parseResumeText("EDUCATION\nUNIVERSITY OF LONDON\nBSc Mathematics\nLondon, UK\n2012 - 2016\n");
+
+		expect(data.customSections).toHaveLength(0);
+		expect(data.sections.education.items).toHaveLength(1);
+		expect(data.sections.education.items[0]).toMatchObject({
+			school: "UNIVERSITY OF LONDON",
+			degree: "BSc Mathematics",
+			location: "London, UK",
+			period: "2012 - 2016",
+		});
+	});
+});
+
+describe("parseResumeText keeps content the layout hides", () => {
+	it("recognizes isolated title-case custom section headings", () => {
+		const initial = parseResumeText(
+			"Ada Lovelace\nada@example.com\n\nConferences\nReactConf\nBerlin\n2021\nSpoke about parsers\n",
+		);
+		const subsequent = parseResumeText(
+			"Ada Lovelace\nada@example.com\n\nEXPERIENCE\nAcme  Engineer\n2020 - 2022\nBuilt products.\n\nConferences\nReactConf\nBerlin\n2021\nSpoke about parsers\n",
+		);
+
+		expect(initial.customSections[0]).toMatchObject({ title: "Conferences" });
+		expect(subsequent.customSections[0]).toMatchObject({ title: "Conferences" });
+		expect(subsequent.sections.experience.items).toHaveLength(1);
+	});
+
+	it("keeps a dated custom section that opens the body", () => {
+		const data = parseResumeText(
+			"Ada Lovelace\nada@example.com\n\nCONFERENCES\nReactConf\nBerlin\n2021\nSpoke about parsers\n",
+		);
+
+		expect(data.basics.headline).not.toBe("CONFERENCES");
+		expect(data.customSections).toHaveLength(1);
+		expect(data.customSections[0]).toMatchObject({ title: "CONFERENCES" });
+
+		const content = data.customSections[0]?.items[0]?.content ?? "";
+		for (const line of ["ReactConf", "Berlin", "2021", "Spoke about parsers"]) {
+			expect(content).toContain(line);
+		}
+	});
+
+	it("keeps unbulleted descriptions with their own role", () => {
+		const data = parseResumeText(
+			"EXPERIENCE\nAcme  Engineer  Berlin\nJan 2020 - Present\nBuilt the thing end to end.\nWorked with a team of five.\nBabbage Ltd  Engineer  London\nMar 2016 - Dec 2019\nDid other work.\n",
+		);
+
+		expect(data.sections.experience.items).toHaveLength(2);
+		expect(data.sections.experience.items[0]?.description).toContain("Built the thing end to end.");
+		expect(data.sections.experience.items[0]?.description).toContain("Worked with a team of five.");
+		expect(data.sections.experience.items[1]).toMatchObject({ company: "Babbage Ltd", location: "London" });
+	});
+
+	it("does not split an entry on a bare year in a section that ignores single dates", () => {
+		const data = parseResumeText(
+			"EXPERIENCE\nAcme  Engineer\nJan 2020 - Present\nGrew the team.\nMore work here.\n2022\n",
+		);
+
+		expect(data.sections.experience.items).toHaveLength(1);
+
+		const description = data.sections.experience.items[0]?.description ?? "";
+		for (const line of ["Grew the team.", "More work here.", "2022"]) {
+			expect(description).toContain(line);
+		}
+	});
+
+	it("stays schema-valid when a section starts with its dates", () => {
+		const data = parseResumeText(
+			"EXPERIENCE\nJan 2020 - Present\nAcme Corp\nSenior Engineer\n\nEDUCATION\n2012 - 2016\nUniversity of London\n\nCERTIFICATIONS\n2021\n",
+		);
+
+		expect(() => resumeDataSchema.parse(data)).not.toThrow();
+		expect(data.sections.experience.items[0]).toMatchObject({ company: "Acme Corp", period: "Jan 2020 - Present" });
+		expect(data.sections.education.items[0]).toMatchObject({ school: "University of London" });
+		expect(data.sections.certifications.items).toHaveLength(0);
+	});
+});

packages/import/src/plain-text.ts

@@ -0,0 +1,593 @@
+import type { ResumeData, SectionType } from "@reactive-resume/schema/resume/data";
+import { parsePeriod, parseSingleDate } from "@reactive-resume/resume/ats";
+import { parseResumeData } from "@reactive-resume/schema/resume/data";
+import { defaultResumeData } from "@reactive-resume/schema/resume/default";
+import { generateId } from "@reactive-resume/utils/string";
+
+type SectionKey = SectionType | "summary";
+
+type Segment = {
+	key: SectionKey | null;
+	title: string;
+	lines: string[];
+};
+
+type RawEntry = {
+	period: string;
+	headerParts: string[];
+	body: string[];
+};
+
+const MAX_HEADING_WORDS = 4;
+const MAX_HEADING_LENGTH = 48;
+const MAX_ENTRY_HEADER_WORDS = 8;
+const MAX_LIST_ITEMS = 60;
+const MIN_PHONE_DIGITS = 7;
+const MAX_PHONE_DIGITS = 15;
+
+const SECTION_ALIASES: Readonly<Record<string, SectionKey>> = {
+	summary: "summary",
+	"professional summary": "summary",
+	"career summary": "summary",
+	profile: "summary",
+	"personal profile": "summary",
+	about: "summary",
+	"about me": "summary",
+	objective: "summary",
+	"career objective": "summary",
+	experience: "experience",
+	"work experience": "experience",
+	"professional experience": "experience",
+	employment: "experience",
+	"employment history": "experience",
+	"work history": "experience",
+	"career history": "experience",
+	education: "education",
+	"academic background": "education",
+	"education and training": "education",
+	qualifications: "education",
+	skills: "skills",
+	"technical skills": "skills",
+	"key skills": "skills",
+	"core competencies": "skills",
+	competencies: "skills",
+	expertise: "skills",
+	projects: "projects",
+	"personal projects": "projects",
+	"selected projects": "projects",
+	"side projects": "projects",
+	languages: "languages",
+	interests: "interests",
+	hobbies: "interests",
+	"hobbies and interests": "interests",
+	awards: "awards",
+	honors: "awards",
+	honours: "awards",
+	"awards and honors": "awards",
+	achievements: "awards",
+	certifications: "certifications",
+	certificates: "certifications",
+	licenses: "certifications",
+	"licenses and certifications": "certifications",
+	publications: "publications",
+	papers: "publications",
+	research: "publications",
+	volunteer: "volunteer",
+	volunteering: "volunteer",
+	"volunteer experience": "volunteer",
+	"community involvement": "volunteer",
+	references: "references",
+	profiles: "profiles",
+	links: "profiles",
+	"social profiles": "profiles",
+};
+
+const BULLET_PATTERN = /^\s*[-–—•*◦‣·]\s+/;
+const EMAIL_PATTERN = /[\w.+-]+@[\w-]+\.[\w.-]*\w/;
+const URL_PATTERN = /\b(?:https?:\/\/|www\.)[^\s,;|•·]+/gi;
+const PHONE_CANDIDATE = /[+(]?\d[\d\s().+-]{5,}\d/g;
+const URL_TEST = /\b(?:https?:\/\/|www\.)\S+/i;
+const HEADER_SCAN_LINES = 6;
+const ENTRY_PREAMBLE_LOOKAHEAD = 4;
+const PERIOD_CANDIDATE =
+	/(?:\p{L}{3,}\.?\s+)?(?:\d{1,2}[/.])?\d{4}\s*(?:[-–—~]|to|until|through)\s*(?:(?:\p{L}{3,}\.?\s+)?(?:\d{1,2}[/.])?\d{4}|\p{L}+)/giu;
+const STRONG_SEPARATOR = /\s*[|•·]\s*|\s{2,}|\s+[–—]\s+/;
+const SENTENCE_END = /[.!?]$/;
+const TRAILING_DATES = [/(?:\p{L}{3,}\.?\s+)?(?:\d{1,2}[/.])?(?:19|20)\d{2}$/u, /(?:\d{1,2}[/.])?(?:19|20)\d{2}$/];
+
+const escapeHtml = (value: string) =>
+	value
+		.replace(/&/g, "&amp;")
+		.replace(/</g, "&lt;")
+		.replace(/>/g, "&gt;")
+		.replace(/"/g, "&quot;")
+		.replace(/'/g, "&#39;");
+
+const normalizeHeading = (line: string) =>
+	line
+		.replace(/[:：]\s*$/, "")
+		.replace(/&/g, " and ")
+		.replace(/[^\p{L}\p{N}\s]/gu, " ")
+		.replace(/\s+/g, " ")
+		.trim()
+		.toLowerCase();
+
+function knownHeading(line: string): SectionKey | null {
+	const normalized = normalizeHeading(line);
+	if (!normalized || normalized.split(" ").length > MAX_HEADING_WORDS) return null;
+
+	return SECTION_ALIASES[normalized] ?? null;
+}
+
+function looksLikeHeading(line: string): boolean {
+	const trimmed = line.trim().replace(/[:：]$/, "");
+	if (!trimmed || trimmed.length > MAX_HEADING_LENGTH || /\d/.test(trimmed)) return false;
+	if (trimmed.split(/\s+/).length > MAX_HEADING_WORDS) return false;
+
+	const letters = trimmed.replace(/[^\p{L}]/gu, "");
+	if (letters.length < 3) return false;
+
+	return letters === letters.toLocaleUpperCase() && letters !== letters.toLocaleLowerCase();
+}
+
+function looksLikeTitleCaseHeading(line: string): boolean {
+	const trimmed = line.trim().replace(/[:：]$/, "");
+	if (!trimmed || trimmed.length > MAX_HEADING_LENGTH || /\d/.test(trimmed)) return false;
+	const words = trimmed.split(/\s+/);
+	if (words.length > MAX_HEADING_WORDS) return false;
+
+	return words.every((word) => {
+		const letters = word.replace(/[^\p{L}]/gu, "");
+		if (letters.length === 0) return false;
+		return letters[0] === letters[0]?.toLocaleUpperCase() && letters.slice(1) === letters.slice(1).toLocaleLowerCase();
+	});
+}
+
+/**
+ * Whether a line could be the header of an entry rather than prose belonging to the previous one.
+ *
+ * Unbulleted descriptions are common, and without this test any narrative line that happens to sit
+ * within the lookahead of the next role's dates would be promoted to a header, stealing the current
+ * entry's description and seeding a garbage item from a sentence.
+ */
+function looksLikeEntryHeader(line: string): boolean {
+	const trimmed = line.trim();
+	if (STRONG_SEPARATOR.test(trimmed)) return true;
+	if (SENTENCE_END.test(trimmed)) return false;
+
+	return trimmed.split(/\s+/).length <= MAX_ENTRY_HEADER_WORDS;
+}
+
+function findPeriod(line: string): string {
+	PERIOD_CANDIDATE.lastIndex = 0;
+
+	for (const match of line.match(PERIOD_CANDIDATE) ?? []) {
+		if (parsePeriod(match)) return match.trim();
+	}
+
+	return "";
+}
+
+function findSingleDate(text: string): string {
+	const trimmed = text.trim();
+
+	for (const pattern of TRAILING_DATES) {
+		const value = pattern.exec(trimmed)?.[0]?.trim();
+		if (value && parseSingleDate(value)) return value;
+	}
+
+	return "";
+}
+
+function extractPhone(text: string): string {
+	for (const candidate of text.match(PHONE_CANDIDATE) ?? []) {
+		const digits = candidate.replace(/\D/g, "");
+		if (digits.length < MIN_PHONE_DIGITS || digits.length > MAX_PHONE_DIGITS) continue;
+		if (parsePeriod(candidate)) continue;
+
+		return candidate.trim();
+	}
+
+	return "";
+}
+
+function isDateLine(line: string, allowSingleDate = true): boolean {
+	if (findPeriod(line)) return true;
+	// A section grouped without single dates reads a bare "2022" as text, so the lookahead has to
+	// agree with it: otherwise that line closes an entry that groupEntries then never reopens.
+	if (!allowSingleDate) return false;
+
+	const bare = line.replace(BULLET_PATTERN, "").trim();
+	return bare !== "" && parseSingleDate(bare) !== null;
+}
+
+function introducesEntry(lines: readonly string[], index: number, allowSingleDate = true): boolean {
+	for (let offset = 1; offset <= ENTRY_PREAMBLE_LOOKAHEAD; offset++) {
+		const line = lines[index + offset];
+		if (line === undefined || BULLET_PATTERN.test(line)) return false;
+		if (isDateLine(line, allowSingleDate)) return true;
+	}
+
+	return false;
+}
+
+function headerBoundary(lines: string[]): number {
+	let boundary = 0;
+
+	for (const [index, line] of lines.slice(0, HEADER_SCAN_LINES).entries()) {
+		if (knownHeading(line)) break;
+		if (EMAIL_PATTERN.test(line) || URL_TEST.test(line) || extractPhone(line)) boundary = index;
+	}
+
+	return boundary;
+}
+
+function splitHeaderParts(text: string): string[] {
+	return text
+		.split(STRONG_SEPARATOR)
+		.map((part) => part.replace(/^[\s,;|•·–—-]+|[\s,;|•·–—-]+$/g, "").trim())
+		.filter(Boolean);
+}
+
+function toHtml(lines: string[]): string {
+	const cleaned = lines.map((line) => line.trim()).filter(Boolean);
+	if (cleaned.length === 0) return "";
+
+	const bulleted = cleaned.filter((line) => BULLET_PATTERN.test(line));
+	if (bulleted.length >= 2 && bulleted.length * 2 >= cleaned.length) {
+		const items = cleaned.map((line) => `<li>${escapeHtml(line.replace(BULLET_PATTERN, ""))}</li>`).join(""); // nosemgrep
+		return `<ul>${items}</ul>`; // nosemgrep
+	}
+
+	return cleaned.map((line) => `<p>${escapeHtml(line.replace(BULLET_PATTERN, ""))}</p>`).join(""); // nosemgrep
+}
+
+function splitList(lines: string[]): string[] {
+	const values: string[] = [];
+
+	for (const line of lines) {
+		for (const piece of line.replace(BULLET_PATTERN, "").split(/[,;|•·]|\s{3,}/)) {
+			const value = piece.trim();
+			if (value) values.push(value);
+		}
+	}
+
+	return [...new Set(values)].slice(0, MAX_LIST_ITEMS);
+}
+
+/**
+ * Guarantees an entry has header text, because every section shape maps `headerParts[0]` onto a
+ * field the resume schema requires to be non-empty. A section whose first line is a date opens an
+ * entry with an empty header, and without this the whole import fails validation on that one item.
+ */
+function withHeaderText(entry: RawEntry): RawEntry {
+	if (entry.headerParts.length > 0) return entry;
+
+	const [first, ...rest] = entry.body;
+	return { ...entry, headerParts: splitHeaderParts(first ?? ""), body: rest };
+}
+
+function groupEntries(lines: string[], allowSingleDate = false): RawEntry[] {
+	const cleaned = lines.map((line) => line.trim()).filter(Boolean);
+	const entries: RawEntry[] = [];
+	let current: RawEntry | null = null;
+
+	const dateOf = (line: string) => {
+		if (BULLET_PATTERN.test(line)) return "";
+
+		const period = findPeriod(line);
+		if (period) return period;
+
+		return allowSingleDate ? findSingleDate(line) : "";
+	};
+
+	for (const [index, line] of cleaned.entries()) {
+		const date = dateOf(line);
+
+		if (date) {
+			const remainder = splitHeaderParts(line.replace(date, " "));
+
+			if (current && !current.period) {
+				current.period = date;
+				current.headerParts.push(...remainder);
+				continue;
+			}
+
+			if (current) entries.push(current);
+			current = { period: date, headerParts: remainder, body: [] };
+			continue;
+		}
+
+		if (!current) {
+			current = { period: "", headerParts: splitHeaderParts(line), body: [] };
+			continue;
+		}
+
+		const isBullet = BULLET_PATTERN.test(line);
+		const leadsToDate = !isBullet && looksLikeEntryHeader(line) && introducesEntry(cleaned, index, allowSingleDate);
+
+		if (leadsToDate && !current.period && current.body.length === 0) {
+			current.headerParts.push(...splitHeaderParts(line));
+			continue;
+		}
+
+		if (leadsToDate) {
+			entries.push(current);
+			current = { period: "", headerParts: splitHeaderParts(line), body: [] };
+			continue;
+		}
+
+		current.body.push(line);
+	}
+
+	if (current) entries.push(current);
+
+	return entries.map(withHeaderText).filter((entry) => entry.headerParts.length > 0);
+}
+
+function entryDescription(entry: RawEntry, usedParts: number): string {
+	const leftover = entry.headerParts.slice(usedParts);
+	return toHtml([...leftover, ...entry.body]);
+}
+
+const baseItem = () => ({ id: generateId(), hidden: false });
+
+const emptyWebsite = { url: "", label: "", inlineLink: false };
+
+function buildSectionItems(key: SectionKey, lines: string[]): unknown[] {
+	if (key === "skills" || key === "interests") {
+		return splitList(lines).map((name) => ({
+			...baseItem(),
+			icon: "",
+			iconColor: "",
+			name,
+			...(key === "skills" ? { proficiency: "", level: 0, keywords: [] } : { keywords: [] }),
+		}));
+	}
+
+	if (key === "languages") {
+		return lines
+			.map((line) => line.replace(BULLET_PATTERN, "").trim())
+			.filter(Boolean)
+			.map((line) => {
+				const match = /^(.+?)\s*[([–—-]\s*(.+?)\s*[)\]]?$/.exec(line);
+				return {
+					...baseItem(),
+					language: (match?.[1] ?? line).trim(),
+					fluency: (match?.[2] ?? "").trim(),
+					level: 0,
+				};
+			});
+	}
+
+	if (key === "profiles") {
+		return lines
+			.map((line) => line.replace(BULLET_PATTERN, "").trim())
+			.filter(Boolean)
+			.map((line) => {
+				const url = line.match(URL_PATTERN)?.[0] ?? "";
+				const network = splitHeaderParts(line.replace(url, " "))[0] ?? line;
+				return {
+					...baseItem(),
+					icon: "",
+					iconColor: "",
+					network,
+					username: "",
+					website: { url, label: "", inlineLink: false },
+				};
+			});
+	}
+
+	const dated = key === "awards" || key === "certifications" || key === "publications";
+	const entries = groupEntries(lines, dated);
+
+	if (key === "experience") {
+		return entries.map((entry) => ({
+			...baseItem(),
+			company: entry.headerParts[0] ?? "",
+			position: entry.headerParts[1] ?? "",
+			location: entry.headerParts[2] ?? "",
+			period: entry.period,
+			website: emptyWebsite,
+			description: entryDescription(entry, 3),
+			roles: [],
+		}));
+	}
+
+	if (key === "education") {
+		return entries.map((entry) => ({
+			...baseItem(),
+			school: entry.headerParts[0] ?? "",
+			degree: entry.headerParts[1] ?? "",
+			area: "",
+			grade: "",
+			location: entry.headerParts[2] ?? "",
+			period: entry.period,
+			website: emptyWebsite,
+			description: entryDescription(entry, 3),
+		}));
+	}
+
+	if (key === "projects") {
+		return entries.map((entry) => ({
+			...baseItem(),
+			name: entry.headerParts[0] ?? "",
+			period: entry.period,
+			website: emptyWebsite,
+			description: entryDescription(entry, 1),
+		}));
+	}
+
+	if (key === "volunteer") {
+		return entries.map((entry) => ({
+			...baseItem(),
+			organization: entry.headerParts[0] ?? "",
+			location: entry.headerParts[1] ?? "",
+			period: entry.period,
+			website: emptyWebsite,
+			description: entryDescription(entry, 2),
+		}));
+	}
+
+	if (key === "awards") {
+		return entries.map((entry) => ({
+			...baseItem(),
+			title: entry.headerParts[0] ?? "",
+			awarder: entry.headerParts[1] ?? "",
+			date: entry.period,
+			website: emptyWebsite,
+			description: entryDescription(entry, 2),
+		}));
+	}
+
+	if (key === "certifications") {
+		return entries.map((entry) => ({
+			...baseItem(),
+			title: entry.headerParts[0] ?? "",
+			issuer: entry.headerParts[1] ?? "",
+			date: entry.period,
+			website: emptyWebsite,
+			description: entryDescription(entry, 2),
+		}));
+	}
+
+	if (key === "publications") {
+		return entries.map((entry) => ({
+			...baseItem(),
+			title: entry.headerParts[0] ?? "",
+			publisher: entry.headerParts[1] ?? "",
+			date: entry.period,
+			website: emptyWebsite,
+			description: entryDescription(entry, 2),
+		}));
+	}
+
+	return entries.map((entry) => ({
+		...baseItem(),
+		name: entry.headerParts[0] ?? "",
+		position: entry.headerParts[1] ?? "",
+		phone: "",
+		website: emptyWebsite,
+		description: entryDescription(entry, 2),
+	}));
+}
+
+function segment(lines: string[]): { header: string[]; segments: Segment[] } {
+	const records = lines
+		.map((line, index) => ({ line: line.trim(), precededByBlank: index > 0 && !lines[index - 1]?.trim() }))
+		.filter((record) => record.line);
+	const cleaned = records.map((record) => record.line);
+	const boundary = headerBoundary(cleaned);
+	const header: string[] = [];
+	const segments: Segment[] = [];
+	let current: Segment | null = null;
+
+	for (const [index, { line, precededByBlank }] of records.entries()) {
+		const key = knownHeading(line);
+		const isolatedTitleCase = precededByBlank && looksLikeTitleCaseHeading(line);
+		// With no section open the entry-preamble guard has nothing to protect: skipping it there keeps
+		// a dated custom section that opens the body from being swallowed into the contact header.
+		const unknown =
+			key === null &&
+			index > boundary &&
+			(looksLikeHeading(line) || isolatedTitleCase) &&
+			(current === null || isolatedTitleCase || !introducesEntry(cleaned, index));
+
+		if (key !== null || unknown) {
+			if (current) segments.push(current);
+			current = { key, title: line.replace(/[:：]\s*$/, "").trim(), lines: [] };
+			continue;
+		}
+
+		if (current) current.lines.push(line);
+		else header.push(line);
+	}
+
+	if (current) segments.push(current);
+
+	return { header, segments };
+}
+
+function parseHeader(lines: string[]) {
+	const joined = lines.join(" ");
+	const email = joined.match(EMAIL_PATTERN)?.[0] ?? "";
+	const phone = extractPhone(joined);
+	const urls = joined.match(URL_PATTERN) ?? [];
+
+	const strip = (value: string) => {
+		let result = value;
+		if (email) result = result.replace(email, " ");
+		if (phone) result = result.replace(phone, " ");
+		for (const url of urls) result = result.replace(url, " ");
+		return result.replace(/\s+/g, " ").trim();
+	};
+
+	const remaining = lines.map(strip).filter(Boolean);
+	const name = remaining[0] ?? "";
+	const rest = remaining.slice(1).flatMap(splitHeaderParts).filter(Boolean);
+	const locationIndex = rest.findIndex((part) => /,/.test(part) && !/\d{4}/.test(part));
+
+	return {
+		name,
+		headline: locationIndex === 0 ? (rest[1] ?? "") : (rest[0] ?? ""),
+		location: locationIndex === -1 ? "" : (rest[locationIndex] ?? ""),
+		email,
+		phone,
+		website: urls[0] ?? "",
+	};
+}
+
+export function parseResumeText(text: string): ResumeData {
+	const lines = text.replace(/\r\n?/g, "\n").split("\n");
+	const { header, segments } = segment(lines);
+	const contact = parseHeader(header);
+
+	const data: ResumeData = structuredClone(defaultResumeData);
+	const order: string[] = [];
+
+	data.basics.name = contact.name;
+	data.basics.headline = contact.headline;
+	data.basics.email = contact.email;
+	data.basics.phone = contact.phone;
+	data.basics.location = contact.location;
+	data.basics.website = { url: contact.website, label: "" };
+
+	for (const item of segments) {
+		if (item.lines.length === 0) continue;
+
+		if (item.key === "summary") {
+			const content = toHtml(item.lines);
+			data.summary.content = data.summary.content ? `${data.summary.content}${content}` : content;
+			if (!order.includes("summary")) order.push("summary");
+			continue;
+		}
+
+		if (item.key === null) {
+			const id = generateId();
+			data.customSections.push({
+				id,
+				type: "summary",
+				title: item.title,
+				icon: "",
+				columns: 1,
+				hidden: false,
+				keepTogether: false,
+				startOnNewPage: false,
+				items: [{ id: generateId(), hidden: false, content: toHtml(item.lines) }],
+			});
+			order.push(id);
+			continue;
+		}
+
+		const items = buildSectionItems(item.key, item.lines);
+		if (items.length === 0) continue;
+
+		const section = data.sections[item.key];
+		section.items = [...section.items, ...items] as typeof section.items;
+		if (!order.includes(item.key)) order.push(item.key);
+	}
+
+	data.metadata.layout.pages = [{ fullWidth: true, main: order, sidebar: [] }];
+
+	return parseResumeData(data);
+}