|
| 1 | +import fs from 'fs' |
| 2 | +import path from 'path' |
| 3 | + |
| 4 | +import { addError } from 'markdownlint-rule-helpers' |
| 5 | + |
| 6 | +import { getFrontmatter } from '../helpers/utils' |
| 7 | +import type { RuleParams, RuleErrorCallback, Rule } from '../../types' |
| 8 | + |
| 9 | +// --------------------------------------------------------------------------- |
| 10 | +// Template parser — derives all validation data from templates.md |
| 11 | +// --------------------------------------------------------------------------- |
| 12 | + |
| 13 | +const TEMPLATES_PATH = path.resolve('content/contributing/writing-for-github-docs/templates.md') |
| 14 | +const SENTINEL = '<!-- rai-card-template-source' |
| 15 | +const OPTIONAL_MARKER = '<!-- optional-section -->' |
| 16 | +const PLACEHOLDER = 'APPLICATION-OR-PLATFORM-SERVICE' |
| 17 | + |
| 18 | +interface TemplateHeading { |
| 19 | + level: number |
| 20 | + text: string |
| 21 | + /** Regex pattern for matching this heading in actual articles. */ |
| 22 | + pattern: RegExp |
| 23 | + /** Human-readable label for error messages. */ |
| 24 | + label: string |
| 25 | + /** If true, the section may be removed from a real article. */ |
| 26 | + optional: boolean |
| 27 | + /** For H3 headings, the pattern of the parent H2. */ |
| 28 | + parentPattern?: RegExp |
| 29 | +} |
| 30 | + |
| 31 | +export interface ParsedTemplate { |
| 32 | + h2s: TemplateHeading[] |
| 33 | + h3s: TemplateHeading[] |
| 34 | + reusables: string[] |
| 35 | +} |
| 36 | + |
| 37 | +/** |
| 38 | + * Extract the RAI card template code block from templates.md. |
| 39 | + * Finds the sentinel HTML comment, then captures the next fenced code block. |
| 40 | + * Strips {% raw %} / {% endraw %} and {% comment %}...{% endcomment %} blocks. |
| 41 | + */ |
| 42 | +function extractTemplateBlock(): string { |
| 43 | + const content = fs.readFileSync(TEMPLATES_PATH, 'utf-8') |
| 44 | + const sentinelIndex = content.indexOf(SENTINEL) |
| 45 | + if (sentinelIndex === -1) { |
| 46 | + throw new Error( |
| 47 | + `Could not find "${SENTINEL}" marker in ${TEMPLATES_PATH}. ` + |
| 48 | + 'This marker is required for the GHD064 linter rule.', |
| 49 | + ) |
| 50 | + } |
| 51 | + |
| 52 | + const afterSentinel = content.slice(sentinelIndex) |
| 53 | + const codeBlockMatch = afterSentinel.match(/```yaml\s*\n([\s\S]*?)```/) |
| 54 | + if (!codeBlockMatch) { |
| 55 | + throw new Error( |
| 56 | + `Could not find a fenced yaml code block after the "${SENTINEL}" marker in ${TEMPLATES_PATH}.`, |
| 57 | + ) |
| 58 | + } |
| 59 | + |
| 60 | + return codeBlockMatch[1] |
| 61 | + .replace(/\{%\s*raw\s*%\}\s*/, '') |
| 62 | + .replace(/\{%\s*endraw\s*%\}\s*/, '') |
| 63 | + .replace(/\{%\s*comment\s*%\}[\s\S]*?\{%\s*endcomment\s*%\}/g, '') |
| 64 | +} |
| 65 | + |
| 66 | +/** |
| 67 | + * Build a regex pattern from a template heading text. |
| 68 | + * Headings containing the placeholder get a pattern that matches any text |
| 69 | + * in place of the placeholder. Fixed headings get an exact match. |
| 70 | + */ |
| 71 | +function headingToPattern(text: string): RegExp { |
| 72 | + if (text.includes(PLACEHOLDER)) { |
| 73 | + const escaped = text |
| 74 | + .replace(/[.*+?^${}()|[\]\\]/g, '\\$&') |
| 75 | + .replace(new RegExp(PLACEHOLDER.replace(/[.*+?^${}()|[\]\\]/g, '\\$&'), 'g'), '.+') |
| 76 | + return new RegExp(`^${escaped}$`, 'i') |
| 77 | + } |
| 78 | + return new RegExp(`^${text.replace(/[.*+?^${}()|[\]\\]/g, '\\$&')}$`, 'i') |
| 79 | +} |
| 80 | + |
| 81 | +/** |
| 82 | + * Build a human-readable label for error messages. |
| 83 | + * Replaces the placeholder with "..." to keep messages concise. |
| 84 | + */ |
| 85 | +function headingLabel(level: number, text: string): string { |
| 86 | + const prefix = '#'.repeat(level) |
| 87 | + const label = text.includes(PLACEHOLDER) ? text.replace(PLACEHOLDER, '...') : text |
| 88 | + return `${prefix} ${label}` |
| 89 | +} |
| 90 | + |
| 91 | +/** |
| 92 | + * Parse the RAI template into structured heading and reusable data. |
| 93 | + * This is the single source of truth for validation — no hardcoded constants. |
| 94 | + */ |
| 95 | +function parseTemplate(): ParsedTemplate { |
| 96 | + const block = extractTemplateBlock() |
| 97 | + const lines = block.split('\n') |
| 98 | + |
| 99 | + const h2s: TemplateHeading[] = [] |
| 100 | + const h3s: TemplateHeading[] = [] |
| 101 | + const reusables: string[] = [] |
| 102 | + |
| 103 | + let currentH2Pattern: RegExp | undefined |
| 104 | + let nextIsOptional = false |
| 105 | + |
| 106 | + for (const line of lines) { |
| 107 | + if (line.trim() === OPTIONAL_MARKER) { |
| 108 | + nextIsOptional = true |
| 109 | + continue |
| 110 | + } |
| 111 | + |
| 112 | + const h2Match = line.match(/^## (.+)$/) |
| 113 | + if (h2Match) { |
| 114 | + const text = h2Match[1].trim() |
| 115 | + const pattern = headingToPattern(text) |
| 116 | + currentH2Pattern = pattern |
| 117 | + h2s.push({ |
| 118 | + level: 2, |
| 119 | + text, |
| 120 | + pattern, |
| 121 | + label: headingLabel(2, text), |
| 122 | + optional: nextIsOptional, |
| 123 | + }) |
| 124 | + nextIsOptional = false |
| 125 | + continue |
| 126 | + } |
| 127 | + |
| 128 | + const h3Match = line.match(/^### (.+)$/) |
| 129 | + if (h3Match) { |
| 130 | + const text = h3Match[1].trim() |
| 131 | + h3s.push({ |
| 132 | + level: 3, |
| 133 | + text, |
| 134 | + pattern: headingToPattern(text), |
| 135 | + label: headingLabel(3, text), |
| 136 | + optional: nextIsOptional, |
| 137 | + parentPattern: currentH2Pattern, |
| 138 | + }) |
| 139 | + nextIsOptional = false |
| 140 | + continue |
| 141 | + } |
| 142 | + |
| 143 | + // Parse reusable references |
| 144 | + const reusableMatches = line.matchAll(/\{%\s*data\s+([\w.-]+)\s*%\}/g) |
| 145 | + for (const m of reusableMatches) { |
| 146 | + if (!reusables.includes(m[1])) { |
| 147 | + reusables.push(m[1]) |
| 148 | + } |
| 149 | + } |
| 150 | + |
| 151 | + // Reset optional flag if line has non-whitespace content (not a heading or marker) |
| 152 | + if (line.trim() !== '') { |
| 153 | + nextIsOptional = false |
| 154 | + } |
| 155 | + } |
| 156 | + |
| 157 | + return { h2s, h3s, reusables } |
| 158 | +} |
| 159 | + |
| 160 | +// Lazy singleton — parsed once on first use |
| 161 | +let _parsed: ParsedTemplate | null = null |
| 162 | + |
| 163 | +export function getTemplate(): ParsedTemplate { |
| 164 | + if (!_parsed) _parsed = parseTemplate() |
| 165 | + return _parsed |
| 166 | +} |
| 167 | + |
| 168 | +// --------------------------------------------------------------------------- |
| 169 | +// File-level heading extraction |
| 170 | +// --------------------------------------------------------------------------- |
| 171 | + |
| 172 | +interface Heading { |
| 173 | + level: number |
| 174 | + text: string |
| 175 | + lineNumber: number |
| 176 | +} |
| 177 | + |
| 178 | +function extractHeadings(lines: string[]): Heading[] { |
| 179 | + const headings: Heading[] = [] |
| 180 | + for (let i = 0; i < lines.length; i++) { |
| 181 | + const match = lines[i].match(/^(#{1,6})\s+(.+)$/) |
| 182 | + if (match) { |
| 183 | + headings.push({ |
| 184 | + level: match[1].length, |
| 185 | + text: match[2].trim(), |
| 186 | + lineNumber: i + 1, |
| 187 | + }) |
| 188 | + } |
| 189 | + } |
| 190 | + return headings |
| 191 | +} |
| 192 | + |
| 193 | +// --------------------------------------------------------------------------- |
| 194 | +// Validators |
| 195 | +// --------------------------------------------------------------------------- |
| 196 | + |
| 197 | +/** |
| 198 | + * Validate required H2 sections exist and appear in the correct order. |
| 199 | + */ |
| 200 | +function validateH2Sections( |
| 201 | + headings: Heading[], |
| 202 | + template: ParsedTemplate, |
| 203 | + onError: RuleErrorCallback, |
| 204 | +): void { |
| 205 | + const h2Headings = headings.filter((h) => h.level === 2) |
| 206 | + let lastMatchedIndex = -1 |
| 207 | + |
| 208 | + for (const required of template.h2s) { |
| 209 | + const matchIndex = h2Headings.findIndex((h) => required.pattern.test(h.text)) |
| 210 | + |
| 211 | + if (matchIndex === -1) { |
| 212 | + addError( |
| 213 | + onError, |
| 214 | + 1, |
| 215 | + `Missing required section: ${required.label}. RAI application/platform cards must include all sections defined in the template.`, |
| 216 | + undefined, |
| 217 | + undefined, |
| 218 | + null, |
| 219 | + ) |
| 220 | + } else if (matchIndex <= lastMatchedIndex) { |
| 221 | + addError( |
| 222 | + onError, |
| 223 | + h2Headings[matchIndex].lineNumber, |
| 224 | + `Section out of order: ${required.label}. RAI application/platform card sections must appear in the order defined by the template.`, |
| 225 | + h2Headings[matchIndex].text, |
| 226 | + undefined, |
| 227 | + null, |
| 228 | + ) |
| 229 | + } else { |
| 230 | + lastMatchedIndex = matchIndex |
| 231 | + } |
| 232 | + } |
| 233 | +} |
| 234 | + |
| 235 | +/** |
| 236 | + * Validate H3 subsections in a single pass: required ones must exist, |
| 237 | + * and all H3s under structured parents must match a known template heading. |
| 238 | + */ |
| 239 | +function validateH3Subsections( |
| 240 | + headings: Heading[], |
| 241 | + template: ParsedTemplate, |
| 242 | + onError: RuleErrorCallback, |
| 243 | +): void { |
| 244 | + // Group template H3s by parent pattern (keyed by pattern source string to |
| 245 | + // avoid relying on RegExp reference equality). |
| 246 | + const h3sByParent = new Map<string, { parentPattern: RegExp; h3s: TemplateHeading[] }>() |
| 247 | + for (const h3 of template.h3s) { |
| 248 | + if (!h3.parentPattern) continue |
| 249 | + const key = h3.parentPattern.source |
| 250 | + const entry = h3sByParent.get(key) || { parentPattern: h3.parentPattern, h3s: [] } |
| 251 | + entry.h3s.push(h3) |
| 252 | + h3sByParent.set(key, entry) |
| 253 | + } |
| 254 | + |
| 255 | + for (const [, { parentPattern, h3s: templateH3s }] of h3sByParent) { |
| 256 | + const parentIndex = headings.findIndex((h) => h.level === 2 && parentPattern.test(h.text)) |
| 257 | + if (parentIndex === -1) continue // Missing parent caught by validateH2Sections |
| 258 | + |
| 259 | + // Collect actual H3s under this parent |
| 260 | + const childH3s: Heading[] = [] |
| 261 | + for (let i = parentIndex + 1; i < headings.length; i++) { |
| 262 | + if (headings[i].level <= 2) break |
| 263 | + if (headings[i].level === 3) childH3s.push(headings[i]) |
| 264 | + } |
| 265 | + |
| 266 | + // Check required H3s exist |
| 267 | + for (const required of templateH3s) { |
| 268 | + if (required.optional) continue |
| 269 | + const found = childH3s.some((h) => required.pattern.test(h.text)) |
| 270 | + if (!found) { |
| 271 | + addError( |
| 272 | + onError, |
| 273 | + headings[parentIndex].lineNumber, |
| 274 | + `Missing required subsection: ${required.label}. This subsection is required under the parent section per the RAI application/platform card template.`, |
| 275 | + undefined, |
| 276 | + undefined, |
| 277 | + null, |
| 278 | + ) |
| 279 | + } |
| 280 | + } |
| 281 | + |
| 282 | + // Check all actual H3s match a known template heading |
| 283 | + for (const child of childH3s) { |
| 284 | + const matchesKnown = templateH3s.some((t) => t.pattern.test(child.text)) |
| 285 | + if (!matchesKnown) { |
| 286 | + addError( |
| 287 | + onError, |
| 288 | + child.lineNumber, |
| 289 | + `Unexpected subsection heading: "${child.text}". H3 headings under this section must match the RAI application/platform card template exactly.`, |
| 290 | + child.text, |
| 291 | + undefined, |
| 292 | + null, |
| 293 | + ) |
| 294 | + } |
| 295 | + } |
| 296 | + } |
| 297 | +} |
| 298 | + |
| 299 | +/** |
| 300 | + * Validate that all required boilerplate reusable references are present. |
| 301 | + */ |
| 302 | +function validateReusables( |
| 303 | + lines: string[], |
| 304 | + template: ParsedTemplate, |
| 305 | + onError: RuleErrorCallback, |
| 306 | +): void { |
| 307 | + const content = lines.join('\n') |
| 308 | + |
| 309 | + for (const reusable of template.reusables) { |
| 310 | + if (!content.includes(reusable)) { |
| 311 | + addError( |
| 312 | + onError, |
| 313 | + 1, |
| 314 | + `Missing required boilerplate reusable: {% data ${reusable} %}. This reusable contains legally-mandated text that must be included in all RAI application/platform cards.`, |
| 315 | + undefined, |
| 316 | + undefined, |
| 317 | + null, |
| 318 | + ) |
| 319 | + } |
| 320 | + } |
| 321 | +} |
| 322 | + |
| 323 | +// --------------------------------------------------------------------------- |
| 324 | +// Rule export |
| 325 | +// --------------------------------------------------------------------------- |
| 326 | + |
| 327 | +interface Frontmatter { |
| 328 | + contentType?: string |
| 329 | + type?: string |
| 330 | + [key: string]: unknown |
| 331 | +} |
| 332 | + |
| 333 | +function isFileRaiCard(params: RuleParams): boolean { |
| 334 | + const fm: Frontmatter = (getFrontmatter(params.frontMatterLines) as Frontmatter) || {} |
| 335 | + return fm.contentType === 'rai' || fm.type === 'rai' |
| 336 | +} |
| 337 | + |
| 338 | +export const raiAppCardStructure: Rule = { |
| 339 | + names: ['GHD064', 'rai-app-card-structure'], |
| 340 | + description: 'RAI application/platform card articles must follow the required template structure', |
| 341 | + tags: ['feature', 'rai'], |
| 342 | + function: (params: RuleParams, onError: RuleErrorCallback) => { |
| 343 | + if (!isFileRaiCard(params)) return |
| 344 | + |
| 345 | + const template = getTemplate() |
| 346 | + const headings = extractHeadings(params.lines) |
| 347 | + validateH2Sections(headings, template, onError) |
| 348 | + validateH3Subsections(headings, template, onError) |
| 349 | + validateReusables(params.lines, template, onError) |
| 350 | + }, |
| 351 | +} |
0 commit comments