| 1 | /** @vitest-environment happy-dom */ |
| 2 | import { describe, expect, it } from 'vitest' |
| 3 | import fs from 'fs' |
| 4 | import path from 'path' |
| 5 | import { |
| 6 | buildHtmlToPptxExtractScript, |
| 7 | normalizeExtractedHtmlToPptxSlide |
| 8 | } from '@arcsin1/html2pptx' |
| 9 | |
| 10 | describe('buildHtmlToPptxExtractScript', () => { |
| 11 | const buildScript = () => |
| 12 | buildHtmlToPptxExtractScript({ |
| 13 | pageWidthPx: 1600, |
| 14 | pageHeightPx: 900 |
| 15 | }) |
| 16 | |
| 17 | const rect = (left: number, top: number, width: number, height: number) => ({ |
| 18 | x: left, |
| 19 | y: top, |
| 20 | left, |
| 21 | top, |
| 22 | width, |
| 23 | height, |
| 24 | right: left + width, |
| 25 | bottom: top + height |
| 26 | }) |
| 27 | |
| 28 | const assignRect = (selector: string, left: number, top: number, width: number, height: number) => { |
| 29 | const element = document.querySelector(selector) |
| 30 | if (!element) throw new Error(`Missing test node: ${selector}`) |
| 31 | Object.defineProperty(element, 'getBoundingClientRect', { |
| 32 | configurable: true, |
| 33 | value: () => rect(left, top, width, height) |
| 34 | }) |
| 35 | } |
| 36 | |
| 37 | const extractInDom = async ({ |
| 38 | maxTextBoxes = 360, |
| 39 | unsupportedTransformStrategy |
| 40 | }: { |
| 41 | maxTextBoxes?: number |
| 42 | unsupportedTransformStrategy?: 'native' | 'raster-fallback' |
| 43 | } = {}) => |
| 44 | new Function( |
| 45 | `return ${ |
| 46 | buildHtmlToPptxExtractScript({ |
| 47 | pageWidthPx: 1600, |
| 48 | pageHeightPx: 900, |
| 49 | maxTextBoxes, |
| 50 | unsupportedTransformStrategy |
| 51 | }).trim() |
| 52 | }` |
| 53 | )() as Promise<Record<string, unknown>> |
| 54 | |
| 55 | it('exports Tailwind rings with a visual hint map and computed spread fallback', () => { |
| 56 | const script = buildScript() |
| 57 | |
| 58 | expect(script).toContain("['ring-1', 1]") |
| 59 | expect(script).toContain("['ring-2', 2]") |
| 60 | expect(script).toContain("['ring-4', 4]") |
| 61 | expect(script).toContain('const parseTailwindRingWidth = (utility) => {') |
| 62 | expect(script).toContain('const arbitrary = utility.match(/^ring-') |
| 63 | expect(script).toContain('Number.parseFloat(arbitrary[1])') |
| 64 | expect(script).toContain('const [offsetX, offsetY, blur, spread] = lengths;') |
| 65 | expect(script).toContain('best = { w: spread, c: color, colorSource };') |
| 66 | }) |
| 67 | |
| 68 | it('supports common Tailwind color scales for visual hints', () => { |
| 69 | const script = buildScript() |
| 70 | |
| 71 | expect(script).toContain('const resolveTailwindColorToken = (name) => {') |
| 72 | expect(script).toContain("red: ['#FEF2F2'") |
| 73 | expect(script).toContain("'#EF4444'") |
| 74 | expect(script).toContain('const palette = tailwindColorPaletteMap.get(match[1]);') |
| 75 | }) |
| 76 | |
| 77 | it('generates parseable browser extraction JavaScript', () => { |
| 78 | expect(() => new Function(buildScript())).not.toThrow() |
| 79 | }) |
| 80 | |
| 81 | it('keeps an unexported text block in the raster fallback when the text limit is reached', async () => { |
| 82 | document.body.innerHTML = ` |
| 83 | <div class="ppt-page-root" style="display:block"> |
| 84 | <p id="first" style="display:block;font-size:20px">First text</p> |
| 85 | <p id="second" style="display:block;font-size:20px">Second text</p> |
| 86 | </div> |
| 87 | ` |
| 88 | assignRect('.ppt-page-root', 0, 0, 1600, 900) |
| 89 | assignRect('#first', 100, 100, 320, 36) |
| 90 | assignRect('#second', 100, 160, 320, 36) |
| 91 | |
| 92 | const extracted = await extractInDom({ maxTextBoxes: 1 }) |
| 93 | |
| 94 | expect(extracted.texts).toHaveLength(1) |
| 95 | expect(extracted.extractionReport).toMatchObject({ textLimitReached: true }) |
| 96 | expect(document.querySelector('#first')?.getAttribute('data-pptx-extracted-text')).toBe('1') |
| 97 | expect(document.querySelector('#second')?.hasAttribute('data-pptx-extracted-text')).toBe(false) |
| 98 | }) |
| 99 | |
| 100 | it('marks extracted inline KPI text so it is removed from the raster background', async () => { |
| 101 | document.body.innerHTML = ` |
| 102 | <div class="ppt-page-root" style="display:block"> |
| 103 | <div id="metrics" style="display:flex"> |
| 104 | <span id="market" style="font-size:48px">528</span> |
| 105 | <span id="share" style="font-size:48px">63<span id="percent" style="font-size:30px">%</span></span> |
| 106 | <span id="cost" style="font-size:48px">$2,950</span> |
| 107 | </div> |
| 108 | <p id="fallback" data-pptx-formula-block style="display:block;font-size:20px">Fallback text</p> |
| 109 | </div> |
| 110 | ` |
| 111 | assignRect('.ppt-page-root', 0, 0, 1600, 900) |
| 112 | assignRect('#metrics', 100, 180, 640, 80) |
| 113 | assignRect('#market', 100, 180, 120, 60) |
| 114 | assignRect('#share', 240, 180, 120, 60) |
| 115 | assignRect('#percent', 320, 205, 32, 36) |
| 116 | assignRect('#cost', 400, 180, 180, 60) |
| 117 | assignRect('#fallback', 100, 300, 220, 36) |
| 118 | |
| 119 | const originalCreateRange = document.createRange.bind(document) |
| 120 | let selectedNode: Node | null = null |
| 121 | const rangeRect = () => { |
| 122 | const id = selectedNode?.parentElement?.id |
| 123 | if (id === 'market') return rect(100, 180, 120, 60) |
| 124 | if (id === 'share') return rect(240, 180, 80, 60) |
| 125 | if (id === 'percent') return rect(320, 205, 32, 36) |
| 126 | if (id === 'cost') return rect(400, 180, 180, 60) |
| 127 | return rect(100, 300, 220, 36) |
| 128 | } |
| 129 | document.createRange = (() => ({ |
| 130 | selectNode: (node: Node) => { |
| 131 | selectedNode = node |
| 132 | }, |
| 133 | selectNodeContents: (node: Node) => { |
| 134 | selectedNode = node |
| 135 | }, |
| 136 | setStart: () => {}, |
| 137 | setEnd: () => {}, |
| 138 | getBoundingClientRect: () => rangeRect(), |
| 139 | getClientRects: () => [rangeRect()], |
| 140 | detach: () => {} |
| 141 | })) as typeof document.createRange |
| 142 | |
| 143 | try { |
| 144 | await extractInDom({ unsupportedTransformStrategy: 'raster-fallback' }) |
| 145 | |
| 146 | expect(document.querySelector('#market')?.getAttribute('data-pptx-extracted-text')).toBe('1') |
| 147 | expect(document.querySelector('#share')?.getAttribute('data-pptx-extracted-text')).toBe('1') |
| 148 | expect(document.querySelector('#percent')?.getAttribute('data-pptx-extracted-text')).toBe('1') |
| 149 | expect(document.querySelector('#cost')?.getAttribute('data-pptx-extracted-text')).toBe('1') |
| 150 | expect(document.querySelector('#fallback')?.hasAttribute('data-pptx-extracted-text')).toBe(false) |
| 151 | } finally { |
| 152 | document.createRange = originalCreateRange |
| 153 | } |
| 154 | }) |
| 155 | |
| 156 | it('keeps rotated shapes in the raster fallback instead of exporting a distorted rectangle', async () => { |
| 157 | document.body.innerHTML = ` |
| 158 | <div class="ppt-page-root" style="display:block"> |
| 159 | <div id="rotated" style="display:block;background-color:rgb(22, 96, 171);border:1px solid rgb(22, 96, 171);transform:rotate(25deg)"></div> |
| 160 | </div> |
| 161 | ` |
| 162 | assignRect('.ppt-page-root', 0, 0, 1600, 900) |
| 163 | assignRect('#rotated', 120, 90, 360, 180) |
| 164 | |
| 165 | const extracted = await extractInDom({ unsupportedTransformStrategy: 'raster-fallback' }) |
| 166 | |
| 167 | expect(extracted.shapes).toHaveLength(0) |
| 168 | expect(extracted.extractionReport).toMatchObject({ unsupportedTransformCount: 1 }) |
| 169 | expect(document.querySelector('#rotated')?.hasAttribute('data-pptx-extracted-shape')).toBe(false) |
| 170 | }) |
| 171 | |
| 172 | it('keeps transformed elements in the public API output unless raster fallback is requested', async () => { |
| 173 | document.body.innerHTML = ` |
| 174 | <div class="ppt-page-root" style="display:block"> |
| 175 | <div id="rotated" style="display:block;background-color:rgb(22, 96, 171);border:1px solid rgb(22, 96, 171);transform:rotate(25deg)"></div> |
| 176 | </div> |
| 177 | ` |
| 178 | assignRect('.ppt-page-root', 0, 0, 1600, 900) |
| 179 | assignRect('#rotated', 120, 90, 360, 180) |
| 180 | |
| 181 | const extracted = await extractInDom() |
| 182 | |
| 183 | expect(extracted.shapes).toHaveLength(1) |
| 184 | expect(extracted.extractionReport).toMatchObject({ unsupportedTransformCount: 0 }) |
| 185 | }) |
| 186 | |
| 187 | it('uses Tailwind font weight hints when no inline font weight overrides them', () => { |
| 188 | const script = buildHtmlToPptxExtractScript({ |
| 189 | pageWidthPx: 1600, |
| 190 | pageHeightPx: 900 |
| 191 | }) |
| 192 | |
| 193 | expect(script).toContain('const resolveTailwindFontWeight = (element) => {') |
| 194 | expect(script).toContain('const resolveInlineFontWeight = (element) => {') |
| 195 | expect(script).toContain('const inlineFontWeight = resolveInlineFontWeight(parentElement);') |
| 196 | expect(script).toContain('const tailwindFontWeight = resolveTailwindFontWeight(parentElement);') |
| 197 | expect(script).toContain( |
| 198 | 'const fontWeight = inlineFontWeight || tailwindFontWeight || computedFontWeight || 400;' |
| 199 | ) |
| 200 | expect(script).not.toContain('tailwindTextHints.fontWeight || computedFontWeight') |
| 201 | }) |
| 202 | |
| 203 | it('extracts inline text runs for styled spans inside block text', () => { |
| 204 | const script = buildScript() |
| 205 | |
| 206 | expect(script).toContain('const collectInlineTextRuns = (element, baseStyle) => {') |
| 207 | expect(script).toContain('const collectInlineTextLineRuns = (element, baseStyle) => {') |
| 208 | expect(script).toContain('const appendTextRun = (runs, run) => {') |
| 209 | expect(script).toContain('const textRunFor = (text, style, element) => {') |
| 210 | expect(script).toContain('const hasStyledRun = runs.some((run) => !sameTextRunStyle(run, baseRun));') |
| 211 | expect(script).toContain('const inlineRuns = collectInlineTextRuns(element, style);') |
| 212 | expect(script).toContain('const inlineLineRuns = inlineRuns?.length ? collectInlineTextLineRuns(element, style) : [];') |
| 213 | expect(script).toContain('inlineLineRuns.every((line) =>') |
| 214 | expect(script).toContain('const rollbackTextExtraction = (startIndex, seenBefore) =>') |
| 215 | expect(script).toContain('runs: inlineRuns') |
| 216 | expect(script).toContain('...(richTextRuns?.length ? { runs: richTextRuns } : {})') |
| 217 | }) |
| 218 | |
| 219 | it('preserves normalized rich text runs with color and bold weight', () => { |
| 220 | const slide = normalizeExtractedHtmlToPptxSlide({ |
| 221 | backgroundColor: 'FFFFFF', |
| 222 | texts: [ |
| 223 | { |
| 224 | text: '本页文字梳理：原始内容', |
| 225 | x: 1, |
| 226 | y: 1, |
| 227 | w: 3, |
| 228 | h: 0.4, |
| 229 | fontSize: 18, |
| 230 | color: '3E3A39', |
| 231 | runs: [ |
| 232 | { |
| 233 | text: '本页文字梳理：', |
| 234 | fontSize: 18, |
| 235 | fontFace: 'Noto Sans SC', |
| 236 | color: 'C00000', |
| 237 | bold: true |
| 238 | }, |
| 239 | { |
| 240 | text: '原始内容', |
| 241 | fontSize: 18, |
| 242 | fontFace: 'Noto Sans SC', |
| 243 | color: '3E3A39' |
| 244 | } |
| 245 | ] |
| 246 | } |
| 247 | ], |
| 248 | shapes: [], |
| 249 | images: [], |
| 250 | tables: [] |
| 251 | }) |
| 252 | |
| 253 | expect(slide.texts[0]?.runs?.[0]).toMatchObject({ |
| 254 | text: '本页文字梳理：', |
| 255 | color: 'C00000', |
| 256 | bold: true |
| 257 | }) |
| 258 | expect(slide.texts[0]?.runs?.[1]).toMatchObject({ |
| 259 | text: '原始内容', |
| 260 | color: '3E3A39', |
| 261 | bold: false |
| 262 | }) |
| 263 | }) |
| 264 | |
| 265 | it('uses CSS stacking order as paint-order fallback for z-indexed edits', () => { |
| 266 | const script = buildScript() |
| 267 | |
| 268 | expect(script).toContain('const parseCssZIndex = (style) => {') |
| 269 | expect(script).toContain( |
| 270 | "parseCssZIndex(style) !== undefined && (style.position !== 'static' || isFlexOrGridItem(element))" |
| 271 | ) |
| 272 | expect(script).toContain('const stackingKeyFor = (element) => {') |
| 273 | expect(script).toContain('const compareStackingOrder = (left, right) => {') |
| 274 | expect(script).toContain('const fallback = new Map(entries.map(([id, element]) => [id, stackingKeyFor(element)]));') |
| 275 | expect(script).toContain('if (!document.elementsFromPoint) return buildFallbackResult();') |
| 276 | expect(script).not.toContain('if (entries.length === 0 || !document.elementsFromPoint) return new Map();') |
| 277 | }) |
| 278 | |
| 279 | it('exports a single visible border side as a line instead of a full rectangle border', () => { |
| 280 | const script = buildScript() |
| 281 | |
| 282 | expect(script).toContain('const collectBorderSides = () => {') |
| 283 | expect(script).toContain('const parseTailwindBorderSideWidth = (utility) => {') |
| 284 | expect(script).toContain('const parseTailwindBorderSideColor = (utility) => {') |
| 285 | expect(script).toContain("sideKey === 'x' ? ['left', 'right']") |
| 286 | expect(script).toContain('borderSideWidthPx: {') |
| 287 | expect(script).toContain('borderSideColorSource: {') |
| 288 | expect(script).toContain('const hintedWidth = tailwindVisualHints.borderSideWidthPx?.[sideName];') |
| 289 | expect(script).toContain('const removeSide = (sideName) => {') |
| 290 | expect(script).toContain('if (hintedWidth !== undefined && hintedWidth <= 0) {') |
| 291 | expect(script).toContain('const hasUniformFourSideBorder =') |
| 292 | expect(script).toContain('const shouldSplitBorderSides =') |
| 293 | expect(script).toContain('const mainBorderInfo = shouldSplitBorderSides ? null : borderInfo;') |
| 294 | expect(script).toContain('const buildBorderLineShape = (borderSide) => {') |
| 295 | expect(script).toContain("shapeType: 'line'") |
| 296 | expect(script).toContain("if (borderSide.side === 'bottom')") |
| 297 | expect(script).toContain('border: hasMainBorder') |
| 298 | expect(script).toContain('splitBorderLineShapes.forEach((shape) => shapes.push(shape));') |
| 299 | expect(script).toContain("dash: borderSide.dash || 'solid'") |
| 300 | }) |
| 301 | |
| 302 | it('keeps small grid color cells as shapes for Tailwind visual panels', () => { |
| 303 | const script = buildScript() |
| 304 | |
| 305 | expect(script).toContain('const isGridPaintCell =') |
| 306 | expect(script).toContain('/grid/i.test(parentDisplay)') |
| 307 | expect(script).toContain('!normalize(element.innerText || element.textContent)') |
| 308 | expect(script).toContain('!hasBorder && !isSmallBadge && !isGridPaintCell') |
| 309 | expect(script).toContain('if (shapes.length >= maxShapes) {') |
| 310 | expect(script).toContain('extractionReport.shapeLimitReached = true;') |
| 311 | expect(script).not.toContain('if (shapes.length >= maxShapes) break;') |
| 312 | }) |
| 313 | |
| 314 | it('keeps thin filled strips such as Tailwind header bars', () => { |
| 315 | const script = buildScript() |
| 316 | |
| 317 | expect(script).toContain('const isThinPaintStrip =') |
| 318 | expect(script).toContain('(rect.width >= 24 && rect.height >= 2 && rect.height < 12)') |
| 319 | expect(script).toContain('(rect.height >= 24 && rect.width >= 2 && rect.width < 12)') |
| 320 | expect(script).toContain("if ((rect.width < 12 || rect.height < 12) && !isThinPaintStrip) continue;") |
| 321 | }) |
| 322 | |
| 323 | it('exports thin border connectors and CSS chevron arrows as PPT lines', () => { |
| 324 | const script = buildScript() |
| 325 | |
| 326 | expect(script).toContain('const hasVisibleBorder =') |
| 327 | expect(script).toContain("['Top', 'Right', 'Bottom', 'Left'].some") |
| 328 | expect(script).toContain('const isCssChevronBorder =') |
| 329 | expect(script).toContain("hasCornerBorderSides('top', 'right')") |
| 330 | expect(script).toContain('const buildCssChevronLineShapes = () => {') |
| 331 | expect(script).toContain('flipV: true') |
| 332 | expect(script).toContain('cssChevronLineShapes.forEach((shape) => shapes.push(shape));') |
| 333 | expect(script).toContain('const shouldExportOnlyBorderLines =') |
| 334 | }) |
| 335 | |
| 336 | it('uses enough text, shape and image budget for dense PPTX export', () => { |
| 337 | const rendererSource = fs.readFileSync( |
| 338 | path.join(process.cwd(), 'src/main/io/html-pptx/renderer.ts'), |
| 339 | 'utf-8' |
| 340 | ) |
| 341 | |
| 342 | expect(rendererSource).toContain('maxTextBoxes: 360') |
| 343 | expect(rendererSource).toContain('maxShapes: 400') |
| 344 | expect(rendererSource).toContain('maxImages: 80') |
| 345 | expect(rendererSource).not.toContain('maxShapes: 80') |
| 346 | }) |
| 347 | }) |
| 348 |