bwf-analyser/build/test-pdf.js

/**
 * The PDF, checked geometrically rather than by eye.
 *
 * With 26 fields selected the old writer gave every column 1/26th of A4
 * landscape — 31pt — while a header like "Originator Reference" needs 69pt at
 * that size, so labels printed straight over their neighbours and values
 * truncated to junk. That's a measurable defect: parse the text-drawing
 * operators out of the generated PDF, measure each string with the same font
 * metrics jsPDF used, and assert nothing overlaps the column to its right.
 *
 * Run: npm i jsdom jspdf && node build/test-pdf.js
 */

const fs = require("fs");
const os = require("os");
const path = require("path");
const zlib = require("zlib");
const assert = require("assert");
const { JSDOM, VirtualConsole } = require("jsdom");
const { jsPDF } = require("jspdf");
const { build } = require("./make-sample.js");

const INDEX = path.join(__dirname, "..", "mac-app", "dist", "index.html");

const results = [];
function check(name, fn) {
	try {
		fn();
		results.push(["PASS", name]);
	} catch (e) {
		results.push(["FAIL", name + " — " + e.message]);
	}
}

function waitFor(fn, label, timeout = 20000) {
	return new Promise((resolve, reject) => {
		const started = Date.now();
		(function poll() {
			let value;
			try {
				value = fn();
			} catch (e) {
				return reject(e);
			}
			if (value) return resolve(value);
			if (Date.now() - started > timeout) return reject(new Error("timed out waiting for " + label));
			setTimeout(poll, 40);
		})();
	});
}

/* ------------------------------------------------------------------ */
/* Reading a PDF back                                                  */
/* ------------------------------------------------------------------ */

function pageWidthOf(bytes) {
	const match = /\/MediaBox\s*\[([^\]]*)\]/.exec(bytes.toString("latin1"));
	assert(match, "no MediaBox in the PDF");
	return parseFloat(match[1].trim().split(/\s+/)[2]);
}

/** Every text-drawing op on the first page, with its font size and position. */
function textOps(bytes) {
	const raw = bytes.toString("latin1");
	const streams = [];
	const re = /stream\r?\n([\s\S]*?)endstream/g;
	let m;
	while ((m = re.exec(raw)) !== null) {
		const body = Buffer.from(m[1], "latin1");
		try {
			streams.push(zlib.inflateSync(body).toString("latin1"));
		} catch (e) {
			streams.push(m[1]);
		}
	}
	const ops = [];
	streams.forEach((content) => {
		const opRe = /BT\s*\/F\d+\s+([\d.]+)\s+Tf[\s\S]*?([\d.]+)\s+([\d.]+)\s+Td\s*\((.*?)\)\s*Tj/g;
		let op;
		while ((op = opRe.exec(content)) !== null) {
			ops.push({
				size: parseFloat(op[1]),
				x: parseFloat(op[2]),
				y: parseFloat(op[3]),
				text: op[4].replace(/\\([()\\])/g, "$1"),
			});
		}
	});
	return ops;
}

/** The header row: the widest band of ops sharing a y, below the title. */
function headerRow(ops) {
	const byY = new Map();
	ops.forEach((op) => {
		if (op.size > 10) return; // title and its metadata line
		byY.set(op.y, (byY.get(op.y) || []).concat(op));
	});
	let best = [];
	byY.forEach((row) => {
		if (row.length > best.length) best = row;
	});
	return best.slice().sort((a, b) => a.x - b.x);
}

function measure(text, size, bold) {
	const probe = new jsPDF({ orientation: "landscape", unit: "pt", format: "a4" });
	probe.setFont("helvetica", bold ? "bold" : "normal");
	probe.setFontSize(size);
	return probe.getTextWidth(text);
}

/* ------------------------------------------------------------------ */
/* Driving a real export                                              */
/* ------------------------------------------------------------------ */

async function exportWith(fieldKeys, details) {
	const pdfBytes = { value: null };
	const consoleErrors = [];
	const virtualConsole = new VirtualConsole();
	virtualConsole.on("jsdomError", (e) => consoleErrors.push(e.message));

	const dom = new JSDOM(fs.readFileSync(INDEX, "utf8"), {
		runScripts: "dangerously",
		url: "http://tauri.localhost/",
		pretendToBeVisual: true,
		virtualConsole,
		beforeParse(win) {
			const ctxStub = new Proxy({}, {
				get: (target, prop) => (prop === "canvas" ? null : () => {}),
				set: () => true,
			});
			win.HTMLCanvasElement.prototype.getContext = () => ctxStub;
			win.URL.createObjectURL = () => "blob:stub";
			win.URL.revokeObjectURL = () => {};
		},
	});

	const { window } = dom;
	await new Promise((r) => window.addEventListener("load", r));
	const doc = window.document;

	// jsPDF writes via an anchor; intercept at the instance instead.
	const RealJsPDF = window.jspdf.jsPDF;
	window.jspdf.jsPDF = function (options) {
		const instance = new RealJsPDF(options);
		instance.save = function () {
			pdfBytes.value = Buffer.from(new Uint8Array(this.output("arraybuffer")));
			return this;
		};
		return instance;
	};
	Object.keys(RealJsPDF).forEach((key) => { window.jspdf.jsPDF[key] = RealJsPDF[key]; });

	const files = [
		["A001_12A_T1.wav", {
			scene: "12A", take: 1,
			// A note with a line break in it, and the multi-line coding history
			// EBU 3285 actually describes: both used to be drawn as extra lines
			// straight through the rows below.
			note: "boom a little hot on the wide,\r\nwatch it on the close",
			codingHistory: "A=PCM,F=48000,W=24,M=stereo,T=833\r\n" +
				"A=ANALOGUE,M=stereo,T=Schoeps CMIT 5U\r\n",
		}],
		["A002_12A_T2.wav", { scene: "12A", take: 2, note: "plane overhead from 00:12" }],
		["A003_14B_T3.wav", { scene: "14B", take: 3, circled: true, description: "aSPEED=025.000-ND" }],
	].map(([name, opts]) => {
		const bytes = build(opts);
		const view = new window.Uint8Array(bytes.length);
		view.set(bytes);
		return new window.File([view], name, { type: "audio/wav" });
	});

	const input = doc.querySelector("[data-bwfa-files-input]");
	files.item = (i) => files[i];
	Object.defineProperty(input, "files", { value: files, configurable: true });
	input.dispatchEvent(new window.Event("change", { bubbles: true }));
	await waitFor(() => /Done/i.test(doc.querySelector("[data-bwfa-status]").textContent), "parse");

	// Set the export selection through the picker, as a user would.
	doc.querySelector("[data-bwfa-export-fields-toggle]")
		.dispatchEvent(new window.MouseEvent("click", { bubbles: true }));
	Array.from(doc.querySelectorAll("[data-bwfa-export-field]")).forEach((box) => {
		const wanted = fieldKeys === "all" || fieldKeys.includes(box.getAttribute("data-bwfa-export-field"));
		if (box.checked !== wanted) {
			box.checked = wanted;
			box.dispatchEvent(new window.Event("change", { bubbles: true }));
		}
	});
	doc.querySelector("[data-bwfa-export-close]")
		.dispatchEvent(new window.MouseEvent("click", { bubbles: true }));

	if (details) {
		// Through the report modal, the way a user gets here: fill in the
		// production details, leave the format on PDF, press Create Report.
		doc.querySelector("[data-bwfa-report-open]")
			.dispatchEvent(new window.MouseEvent("click", { bubbles: true }));
		Object.keys(details).forEach((key) => {
			const input = doc.querySelector('[data-bwfa-report-field="' + key + '"]');
			assert(input, "no report field called " + key);
			input.value = details[key];
		});
		doc.querySelector("[data-bwfa-report-format]").value = "pdf";
		doc.querySelector("[data-bwfa-report-create]")
			.dispatchEvent(new window.MouseEvent("click", { bubbles: true }));
	} else {
		doc.querySelector("[data-bwfa-export-pdf]")
			.dispatchEvent(new window.MouseEvent("click", { bubbles: true }));
	}
	await waitFor(() => pdfBytes.value, "pdf");
	assert.strictEqual(consoleErrors.length, 0, consoleErrors.join(" | "));
	window.close();
	return pdfBytes.value;
}

/* ------------------------------------------------------------------ */

(async () => {
	const everything = await exportWith("all");
	const withDetails = await exportWith(
		["fileName", "scene", "take", "startTimecode"],
		{
			company: "Acme Films",
			project: "The Long Weekend",
			director: "R. Okonjo",
			mixer: "Vincent Rozenberg",
			phone: "+31 6 1234 5678",
			email: "vincent@example.com",
			note: "Day 4, ext. night",
		}
	);
	const compact = await exportWith([
		"fileName", "scene", "take", "startTimecode", "frameRate", "durationSeconds",
	]);

	const wideHeader = headerRow(textOps(everything));
	const compactHeader = headerRow(textOps(compact));

	check("all 26 fields reach the page", () => {
		assert(wideHeader.length >= 24,
			"only " + wideHeader.length + " header cells: " + wideHeader.map((o) => o.text).join(" | "));
	});

	check("no header prints over the column to its right", () => {
		for (let i = 0; i < wideHeader.length - 1; i++) {
			const cell = wideHeader[i];
			const width = measure(cell.text, cell.size, true);
			assert(cell.x + width <= wideHeader[i + 1].x + 0.5,
				'"' + cell.text + '" ends at ' + (cell.x + width).toFixed(1) +
					' but "' + wideHeader[i + 1].text + '" starts at ' + wideHeader[i + 1].x.toFixed(1));
		}
	});

	check("the page grew sideways to hold them", () => {
		const width = pageWidthOf(everything);
		assert(width > 841.9, "still A4 landscape at " + width.toFixed(0) + "pt");
		assert(width <= 2400, "page ran away to " + width.toFixed(0) + "pt");
	});

	check("every column is wide enough for its own header", () => {
		for (let i = 0; i < wideHeader.length - 1; i++) {
			const available = wideHeader[i + 1].x - wideHeader[i].x;
			assert(available > 8, '"' + wideHeader[i].text + '" got only ' + available.toFixed(1) + "pt");
		}
	});

	check("nothing anywhere is cut off", () => {
		// The ellipsis was the writer's own doing — nothing in a BWF file
		// contains one — so its presence is the defect, wherever it turns up.
		[["all 26 fields", everything], ["a short selection", compact],
			["a report with details", withDetails]].forEach(([what, bytes]) => {
			const cut = textOps(bytes).filter((op) => /\u2026|\u0085/.test(op.text));
			assert.deepStrictEqual(cut.map((op) => op.text), [],
				what + " came out clipped: " + cut.map((op) => op.text).join(" | "));
		});
	});

	check("no value prints over the column to its right", () => {
		const ops = textOps(everything);
		// The first data row: every cell measured as drawn, against where the
		// next column starts.
		const row = ops.filter((op) => op.y === ops.filter((o) => /^A001_/.test(o.text))[0].y)
			.slice().sort((a, b) => a.x - b.x);
		assert(row.length >= 20, "expected a cell per column, got " + row.length);
		for (let i = 0; i < row.length - 1; i++) {
			const width = measure(row[i].text, row[i].size, false);
			assert(row[i].x + width <= row[i + 1].x + 0.5,
				'"' + row[i].text + '" runs into "' + row[i + 1].text + '"');
		}
	});

	check("a multi-line field is flattened, not spread over the rows below", () => {
		const ops = textOps(everything);
		ops.forEach((op) => {
			assert(!/[\r\n]/.test(op.text), "a drawn string still has a line break: " + op.text);
		});

		const history = ops.filter((op) => /^A=PCM,F=48000,W=24,M=stereo,T=833/.test(op.text));
		assert(history.length, "the coding history didn't make it into the report");
		assert(/A=ANALOGUE/.test(history[0].text),
			"the second history line was dropped rather than joined: " + history[0].text);
		assert(/ \u00b7 /.test(history[0].text),
			"the lines were run together with no separator: " + history[0].text);

		const note = ops.filter((op) => /^boom a little hot/.test(op.text));
		assert(note.length, "the note didn't make it into the report");
		assert(/watch it on the close$/.test(note[0].text),
			"the note lost its second line: " + note[0].text);

		// Every row of the table on one baseline pitch: an extra line drawn by
		// jsPDF for a \n would show up as a y that isn't on the grid.
		const rowYs = Array.from(new Set(ops.filter((op) => op.size < 10).map((op) => op.y)))
			.sort((a, b) => b - a);
		const gaps = rowYs.slice(1).map((y, i) => rowYs[i] - y).filter((gap) => gap < 30);
		gaps.forEach((gap) => assert(Math.abs(gap - 18) < 0.5 || Math.abs(gap - 12) < 0.5,
			"an unexpected baseline gap of " + gap.toFixed(1) + "pt"));
	});

	check("values are not truncated to nonsense", () => {
		const ops = textOps(everything);
		const filenames = ops.filter((op) => /^A00\d_/.test(op.text));
		assert(filenames.length >= 3, "expected a filename per row, got " + filenames.length);
		assert(filenames.every((op) => /\.wav$/.test(op.text)),
			"filenames lost their extension: " + filenames.map((o) => o.text).join(", "));
	});

	check("a short selection still fits A4 landscape", () => {
		const width = pageWidthOf(compact);
		assert(Math.abs(width - 841.89) < 1, "six columns should not resize the page: " + width.toFixed(1));
		assert(compactHeader.length >= 6, "expected six header cells, got " + compactHeader.length);
	});

	check("a short selection spans the full page width", () => {
		// Columns share out the spare room rather than huddling on the left.
		const last = compactHeader[compactHeader.length - 1];
		assert(last.x > 600, "the table stops at " + last.x.toFixed(0) + "pt of 842");
	});

	console.log("");
	/** Where the table's own header sits, found by one of its column labels —
	 *  headerRow() picks the widest band, and with four columns selected the
	 *  detail block ties with it. */
	const tableHeaderY = (ops) => {
		const scene = ops.filter((op) => op.text === "Scene");
		assert(scene.length, "no Scene column header in the PDF");
		return scene[0].y;
	};

	check("the report carries the production details, above the table", () => {
		const ops = textOps(withDetails);
		const text = ops.map((op) => op.text).join(" | ");
		["Acme Films", "The Long Weekend", "R. Okonjo", "Vincent Rozenberg",
			"+31 6 1234 5678", "vincent@example.com", "Day 4, ext. night"].forEach((value) => {
			assert(text.indexOf(value) !== -1, "missing from the report: " + value);
		});

		// In PDF space y counts up from the bottom, so "above" means larger.
		const headerY = tableHeaderY(ops);
		const detail = ops.filter((op) => /Acme Films|R\. Okonjo/.test(op.text));
		detail.forEach((op) => {
			assert(op.y > headerY,
				"a detail line at y=" + op.y + " is below the table header at y=" + headerY);
		});

		// Two to a line, so seven details cost four lines, not seven.
		const lines = new Set(detail.concat(ops.filter((op) =>
			/Production company|Sound mixer|Mixer phone/.test(op.text))).map((op) => op.y));
		assert(lines.size <= 4, "the details take " + lines.size + " lines");
	});

	check("the details push the table down rather than printing over it", () => {
		const plain = tableHeaderY(textOps(compact));
		const withHeader = tableHeaderY(textOps(withDetails));
		assert(withHeader < plain,
			"the table header didn't move: " + withHeader + " vs " + plain);
	});

	results.forEach(([s, n]) => console.log((s === "PASS" ? "  ok  " : "  FAIL") + "  " + n));
	const failed = results.filter(([s]) => s === "FAIL").length;
	console.log("\n" + (results.length - failed) + "/" + results.length + " checks passed");
	process.exit(failed ? 1 : 0);
})().catch((e) => {
	console.error("harness error:", e);
	process.exit(1);
});