| 1 | import { isLocalFileHref } from "./localFileUrl"; |
| 2 | |
| 3 | const PDF_PAGE_WIDTH = 595.28; |
| 4 | const PDF_PAGE_HEIGHT = 841.89; |
| 5 | const PDF_MARGIN = 36; |
| 6 | |
| 7 | export const PDF_CONTENT_ASPECT = |
| 8 | (PDF_PAGE_HEIGHT - PDF_MARGIN * 2) / (PDF_PAGE_WIDTH - PDF_MARGIN * 2); |
| 9 | |
| 10 | export interface RasterSlice { |
| 11 | offset: number; |
| 12 | height: number; |
| 13 | } |
| 14 | |
| 15 | export interface RasterPdfImage { |
| 16 | bytes: Uint8Array; |
| 17 | width: number; |
| 18 | height: number; |
| 19 | } |
| 20 | |
| 21 | export function planRasterSlices( |
| 22 | totalHeight: number, |
| 23 | maxSliceHeight: number, |
| 24 | naturalBreakpoints: number[] = [], |
| 25 | contentEnd?: number, |
| 26 | ): RasterSlice[] { |
| 27 | const total = Math.max(1, Math.ceil(Number.isFinite(totalHeight) ? totalHeight : 1)); |
| 28 | const limit = Math.max(1, Math.floor(Number.isFinite(maxSliceHeight) ? maxSliceHeight : 1)); |
| 29 | // contentEnd lets callers distinguish meaningful content from trailing |
| 30 | // container whitespace. Plan pages through the content first, then retain |
| 31 | // only the trailing whitespace that still fits on the final content page. |
| 32 | const plannedTotal = Math.max( |
| 33 | 1, |
| 34 | Math.min(total, Math.ceil(contentEnd !== undefined && Number.isFinite(contentEnd) ? contentEnd : total)), |
| 35 | ); |
| 36 | const breakpoints = naturalBreakpoints |
| 37 | .filter((value) => Number.isFinite(value) && value > 0 && value < plannedTotal) |
| 38 | .map((value) => Math.floor(value)) |
| 39 | .sort((a, b) => a - b); |
| 40 | const slices: RasterSlice[] = []; |
| 41 | let offset = 0; |
| 42 | while (offset < plannedTotal) { |
| 43 | const target = Math.min(plannedTotal, offset + limit); |
| 44 | let end = target; |
| 45 | if (target < plannedTotal) { |
| 46 | const earliestNaturalBreak = offset + Math.floor(limit * 0.55); |
| 47 | for (const breakpoint of breakpoints) { |
| 48 | if (breakpoint > target) break; |
| 49 | if (breakpoint >= earliestNaturalBreak) end = breakpoint; |
| 50 | } |
| 51 | } |
| 52 | if (end <= offset) end = target; |
| 53 | slices.push({ offset, height: end - offset }); |
| 54 | offset = end; |
| 55 | } |
| 56 | const last = slices[slices.length - 1]; |
| 57 | if (last && plannedTotal < total && last.height < limit) { |
| 58 | last.height += Math.min(total - plannedTotal, limit - last.height); |
| 59 | } |
| 60 | return slices; |
| 61 | } |
| 62 | |
| 63 | // SVG foreignObject rendering becomes origin-tainted when its CSS references a |
| 64 | // font or image URL. Export surfaces use system fonts, so external resources are |
| 65 | // intentionally neutralised before the SVG is drawn onto a canvas. |
| 66 | export function neutralizeExternalCssResources(css: string): string { |
| 67 | return css.replace(/url\(\s*(?:"[^"]*"|'[^']*'|[^)]*)\s*\)/gi, "none"); |
| 68 | } |
| 69 | |
| 70 | export function isSafeInlineExportImage(src: string | undefined): boolean { |
| 71 | return /^data:image\/(?:png|jpe?g|webp|gif);base64,[a-z0-9+/]+={0,2}$/i.test(src?.trim() ?? ""); |
| 72 | } |
| 73 | |
| 74 | export function transformExportMarkdownUrl( |
| 75 | value: string, |
| 76 | key: string, |
| 77 | fallback: (value: string) => string, |
| 78 | ): string { |
| 79 | const trimmed = value.trim(); |
| 80 | if (key === "src" && isSafeInlineExportImage(trimmed)) return trimmed; |
| 81 | // Local-path anchors from remarkLocalPathLinks are kept so an exported |
| 82 | // document stays clickable; everything else goes through the default |
| 83 | // transform which blanks javascript: and other unsafe schemes. |
| 84 | if (key === "href" && isLocalFileHref(trimmed)) return trimmed; |
| 85 | return fallback(value); |
| 86 | } |
| 87 | |
| 88 | function bytesFromString(value: string): Uint8Array { |
| 89 | const bytes = new Uint8Array(value.length); |
| 90 | for (let i = 0; i < value.length; i++) { |
| 91 | bytes[i] = value.charCodeAt(i) & 0xff; |
| 92 | } |
| 93 | return bytes; |
| 94 | } |
| 95 | |
| 96 | function concatBytes(chunks: Uint8Array[]): Uint8Array { |
| 97 | const total = chunks.reduce((sum, chunk) => sum + chunk.length, 0); |
| 98 | const out = new Uint8Array(total); |
| 99 | let offset = 0; |
| 100 | for (const chunk of chunks) { |
| 101 | out.set(chunk, offset); |
| 102 | offset += chunk.length; |
| 103 | } |
| 104 | return out; |
| 105 | } |
| 106 | |
| 107 | function pdfNumber(value: number): string { |
| 108 | return value.toFixed(3).replace(/\.?0+$/, ""); |
| 109 | } |
| 110 | |
| 111 | function pdfString(value: string): string { |
| 112 | return value |
| 113 | .replace(/[^\x20-\x7e]/g, "") |
| 114 | .replace(/\\/g, "\\\\") |
| 115 | .replace(/\(/g, "\\(") |
| 116 | .replace(/\)/g, "\\)"); |
| 117 | } |
| 118 | |
| 119 | export function createRasterPdf(images: RasterPdfImage[], title: string): Uint8Array { |
| 120 | if (images.length === 0) throw new Error("Cannot create a PDF without pages"); |
| 121 | |
| 122 | const contentWidth = PDF_PAGE_WIDTH - PDF_MARGIN * 2; |
| 123 | const contentHeight = PDF_PAGE_HEIGHT - PDF_MARGIN * 2; |
| 124 | const infoObjectId = 3 + images.length * 3; |
| 125 | const objectCount = infoObjectId; |
| 126 | const chunks: Uint8Array[] = []; |
| 127 | const offsets: number[] = new Array(objectCount + 1).fill(0); |
| 128 | let position = 0; |
| 129 | |
| 130 | const push = (value: string | Uint8Array) => { |
| 131 | const bytes = typeof value === "string" ? bytesFromString(value) : value; |
| 132 | chunks.push(bytes); |
| 133 | position += bytes.length; |
| 134 | }; |
| 135 | const addObject = (id: number, body: string) => { |
| 136 | offsets[id] = position; |
| 137 | push(`${id} 0 obj\n${body}\nendobj\n`); |
| 138 | }; |
| 139 | const addStreamObject = (id: number, header: string, body: Uint8Array) => { |
| 140 | offsets[id] = position; |
| 141 | push(`${id} 0 obj\n${header}\nstream\n`); |
| 142 | push(body); |
| 143 | push("\nendstream\nendobj\n"); |
| 144 | }; |
| 145 | |
| 146 | push("%PDF-1.4\n%\xff\xff\xff\xff\n"); |
| 147 | addObject(1, "<< /Type /Catalog /Pages 2 0 R >>"); |
| 148 | const kids = images.map((_, index) => `${3 + index * 3} 0 R`).join(" "); |
| 149 | addObject(2, `<< /Type /Pages /Kids [ ${kids} ] /Count ${images.length} >>`); |
| 150 | |
| 151 | images.forEach((image, index) => { |
| 152 | const pageId = 3 + index * 3; |
| 153 | const contentId = pageId + 1; |
| 154 | const imageId = pageId + 2; |
| 155 | const renderedHeight = Math.min(contentHeight, image.height * (contentWidth / image.width)); |
| 156 | const y = PDF_PAGE_HEIGHT - PDF_MARGIN - renderedHeight; |
| 157 | const stream = bytesFromString( |
| 158 | `q\n${pdfNumber(contentWidth)} 0 0 ${pdfNumber(renderedHeight)} ${pdfNumber(PDF_MARGIN)} ${pdfNumber(y)} cm\n/Im0 Do\nQ\n`, |
| 159 | ); |
| 160 | addObject( |
| 161 | pageId, |
| 162 | `<< /Type /Page /Parent 2 0 R /MediaBox [0 0 ${pdfNumber(PDF_PAGE_WIDTH)} ${pdfNumber(PDF_PAGE_HEIGHT)}] /Resources << /XObject << /Im0 ${imageId} 0 R >> /ProcSet [/PDF /ImageC] >> /Contents ${contentId} 0 R >>`, |
| 163 | ); |
| 164 | addStreamObject(contentId, `<< /Length ${stream.length} >>`, stream); |
| 165 | addStreamObject( |
| 166 | imageId, |
| 167 | `<< /Type /XObject /Subtype /Image /Width ${image.width} /Height ${image.height} /ColorSpace /DeviceRGB /BitsPerComponent 8 /Filter /DCTDecode /Length ${image.bytes.length} >>`, |
| 168 | image.bytes, |
| 169 | ); |
| 170 | }); |
| 171 | |
| 172 | addObject(infoObjectId, `<< /Title (${pdfString(title)}) /Producer (Reasonix) >>`); |
| 173 | const xrefStart = position; |
| 174 | push(`xref\n0 ${objectCount + 1}\n0000000000 65535 f \n`); |
| 175 | for (let id = 1; id <= objectCount; id++) { |
| 176 | push(`${String(offsets[id]).padStart(10, "0")} 00000 n \n`); |
| 177 | } |
| 178 | push(`trailer\n<< /Size ${objectCount + 1} /Root 1 0 R /Info ${infoObjectId} 0 R >>\nstartxref\n${xrefStart}\n%%EOF\n`); |
| 179 | return concatBytes(chunks); |
| 180 | } |
| 181 |