@nitpicker/report-google-sheets 0.4.1 → 0.4.3

package/src/data/create-discrepancies.ts

@@ -1,81 +0,0 @@
-import type { CreateSheet } from '../sheets/types.js';
-import type { Cell } from '@d-zero/google-sheets';
-
-import { pLog } from '../debug.js';
-import { createCellData } from '../sheets/create-cell-data.js';
-import { defaultCellFormat } from '../sheets/default-cell-format.js';
-
-const log = pLog.extend('Discrepancies');
-
-/**
- * Creates the "Discrepancies" sheet configuration.
- *
- * Generates discrepancy data from two sources:
- *
- * 1. **Link Text vs Page Title** (via `eachPage`): For every anchor on
- *    each page, creates a row comparing the link's text content with the
- *    linked page's title. This helps identify misleading or inconsistent
- *    link labels.
- *
- * 2. **Plugin discrepancies** (via `addRows`): Includes any discrepancy data
- *    produced by analyze plugins (e.g. meta tag consistency checks).
- *
- * Both data sources are written to the same sheet, distinguished by
- * the "Type" column.
- * @param reports
- */
-export const createDiscrepancies: CreateSheet = (reports) => {
-	return {
-		name: 'Discrepancies',
-		createHeaders() {
-			return ['Type', 'Left URL', 'Left', 'Right', 'Right URL', 'Note'];
-		},
-		async eachPage(page) {
-			const anchors = await page.getAnchors();
-			const data: Cell[][] = [];
-			log('Create text link discrepancies');
-			log('Found %d anchors', anchors.length);
-			for (const anchor of anchors) {
-				data.push([
-					createCellData({ value: 'Link Text vs Page Title' }, defaultCellFormat),
-					createCellData({ value: page.url.href }, defaultCellFormat),
-					createCellData({ value: anchor.textContent }, defaultCellFormat),
-					createCellData({ value: anchor.title }, defaultCellFormat),
-					createCellData({ value: anchor.url }, defaultCellFormat),
-					createCellData({ value: '' }, defaultCellFormat),
-				]);
-			}
-			return data;
-		},
-		addRows() {
-			const data: Cell[][] = [];
-			for (const report of reports) {
-				if (!report.discrepancies) {
-					continue;
-				}
-				for (const discrepancy of report.discrepancies) {
-					data.push([
-						createCellData(
-							{ value: discrepancy.leftSourceUrl, note: discrepancy.leftSourceUrlNote },
-							defaultCellFormat,
-						),
-						createCellData(
-							{ value: discrepancy.left, note: discrepancy.leftNote },
-							defaultCellFormat,
-						),
-						createCellData(
-							{ value: discrepancy.right, note: discrepancy.rightNote },
-							defaultCellFormat,
-						),
-						createCellData(
-							{ value: discrepancy.rightSourceUrl, note: discrepancy.rightSourceUrlNote },
-							defaultCellFormat,
-						),
-						createCellData({ value: discrepancy.note }, defaultCellFormat),
-					]);
-				}
-			}
-			return data;
-		},
-	};
-};

package/src/data/create-image-list.ts

@@ -1,74 +0,0 @@
-import type { CreateSheet } from '../sheets/types.js';
-import type { Cell } from '@d-zero/google-sheets';
-
-import { JSDOM } from 'jsdom';
-
-import { createCellData } from '../sheets/create-cell-data.js';
-import { defaultCellFormat } from '../sheets/default-cell-format.js';
-
-/**
- * Creates the "Images" sheet configuration.
- *
- * Extracts all `<img>` elements from each internal page's HTML and
- * reports their attributes (src, currentSrc, alt, dimensions, lazy loading).
- * Uses JSDOM to parse the HTML so that resolved `src` URLs (after base URL
- * resolution) and `currentSrc` (for `<picture>` / `srcset`) are available.
- *
- * External pages are skipped because their HTML snapshots are not stored
- * in the archive.
- */
-export const createImageList: CreateSheet = () => {
-	return {
-		name: 'Images',
-		createHeaders() {
-			return [
-				'Page URL',
-				'Image path (src)',
-				'Image Path (currentSrc)',
-				'Alternative Text',
-				'Displayed Width',
-				'Displayed Height',
-				// 'Original Width',
-				// 'Original Height',
-				// 'Fit Width Rate',
-				// 'Fit Height Rate',
-				'Lazy Loading',
-				'Source Code',
-			];
-		},
-
-		async eachPage(page) {
-			if (!page.isInternalPage()) {
-				return;
-			}
-
-			const html = await page.getHtml();
-			if (!html) {
-				return;
-			}
-			const dom = new JSDOM(html, { url: page.url.href });
-			const imgs = dom.window.document.querySelectorAll('img');
-
-			const imgData: Cell[][] = [];
-
-			for (const img of imgs) {
-				const data = [
-					createCellData({ value: page.url.href }, defaultCellFormat),
-					createCellData({ value: img.src }, defaultCellFormat),
-					createCellData({ value: img.currentSrc }, defaultCellFormat),
-					createCellData({ value: img.alt }, defaultCellFormat),
-					createCellData({ value: img.width }, defaultCellFormat),
-					createCellData({ value: img.height }, defaultCellFormat),
-					// createCellData({ value: img.naturalWidth }, defaultCellFormat),
-					// createCellData({ value: img.naturalHeight }, defaultCellFormat),
-					createCellData({ value: img.loading === 'lazy' }, defaultCellFormat),
-					createCellData({ value: img.outerHTML }, defaultCellFormat),
-				];
-
-				imgData.push(data);
-			}
-
-			return imgData;
-		},
-	};
-};

package/src/data/create-links.ts

@@ -1,134 +0,0 @@
-import type { CreateSheet } from '../sheets/types.js';
-
-import { pLog } from '../debug.js';
-import { createCellData } from '../sheets/create-cell-data.js';
-import { defaultCellFormat } from '../sheets/default-cell-format.js';
-import { booleanFormatError } from '../sheets/format.js';
-
-const log = pLog.extend('Links');
-
-/**
- * Creates the "Links" sheet configuration.
- *
- * Produces one row per crawled page with URL, title, HTTP status,
- * content type, redirect chain, referrers, response headers, and remarks.
- * The remarks column shows the skip reason for pages that were skipped
- * during crawling (e.g., blocked by robots.txt, excluded by rules).
- * Applies conditional formatting to highlight:
- * - Status codes >= 400 (client/server errors)
- * - Status codes outside the 200-399 range (non-success)
- *
- * The header row and first column are frozen for easier scrolling.
- */
-export const createLinks: CreateSheet = () => {
-	return {
-		name: 'Links',
-		createHeaders() {
-			return [
-				'URL',
-				'Page Title',
-				'Status Code',
-				'Status Text',
-				'Content Type',
-				'Redirect From',
-				'Referrers',
-				'Headers',
-				'Remarks',
-			];
-		},
-		async eachPage(page, num, total) {
-			const p = Math.round((num / total) * 100);
-			log('Create links (%d%% %d/%d)', p, num, total);
-
-			// if (page.isInternalPage()) {
-			// 	return;
-			// }
-
-			const referrers = await page.getReferrers();
-			const data = [
-				createCellData(
-					{
-						value: page.url.href,
-						textFormat: { link: { uri: page.url.href } },
-					},
-					defaultCellFormat,
-				),
-				createCellData({ value: page.title || '-' }, defaultCellFormat),
-				createCellData({ value: page.status || -1 }, defaultCellFormat),
-				createCellData({ value: page.statusText || '' }, defaultCellFormat),
-				createCellData({ value: page.contentType || '' }, defaultCellFormat),
-				createCellData(
-					{
-						value: page.redirectFrom.length,
-						note: page.redirectFrom.map((r) => r.url).join('\n'),
-					},
-					defaultCellFormat,
-				),
-				createCellData(
-					{
-						value: `${referrers.length} Elements`,
-						note: referrers
-							.map((ref) => {
-								const text = ref.textContent || '__NO_TEXT_CONTENT__';
-								const url = ref.url + (ref.hash ? `#${ref.hash}` : '');
-								const pass =
-									page.url.href === ref.through
-										? ''
-										: ` => [REDIRECTED FROM] ${ref.through}`;
-								return `${text} (${url}${pass})`;
-							})
-							.join('\n'),
-					},
-					defaultCellFormat,
-				),
-				createCellData(
-					{
-						value: '{}',
-						note: JSON.stringify(page.responseHeaders, null, 2),
-					},
-					defaultCellFormat,
-				),
-				createCellData(
-					{ value: page.isSkipped ? page.skipReason || 'skipped' : '' },
-					defaultCellFormat,
-				),
-			];
-
-			return [data];
-		},
-		async updateSheet(sheet) {
-			await sheet.frozen(2, 1);
-
-			await sheet.conditionalFormat([sheet.getColNumByHeaderName('Status Code')], {
-				booleanRule: {
-					condition: {
-						type: 'NUMBER_GREATER_THAN_EQ',
-						values: [
-							{
-								userEnteredValue: '400',
-							},
-						],
-					},
-					format: booleanFormatError,
-				},
-			});
-
-			await sheet.conditionalFormat([sheet.getColNumByHeaderName('Status Code')], {
-				booleanRule: {
-					condition: {
-						type: 'NUMBER_NOT_BETWEEN',
-						values: [
-							{
-								userEnteredValue: '200',
-							},
-							{
-								userEnteredValue: '399',
-							},
-						],
-					},
-					format: booleanFormatError,
-				},
-			});
-		},
-	};
-};