packages/cli/src/lib/markdown.ts 10.3 K raw
1
import * as fs from "node:fs/promises";
2
import * as path from "node:path";
3
import { glob } from "glob";
4
import yaml from "js-yaml";
5
import { minimatch } from "minimatch";
6
import * as toml from "smol-toml";
7
import type { BlogPost, FrontmatterMapping, PostFrontmatter } from "./types";
8
9
export function parseFrontmatter(
10
	content: string,
11
	mapping?: FrontmatterMapping,
12
): {
13
	frontmatter: PostFrontmatter;
14
	body: string;
15
	rawFrontmatter: Record<string, unknown>;
16
} {
17
	// Support multiple frontmatter delimiters:
18
	// --- (YAML) - Jekyll, Astro, most SSGs
19
	// +++ (TOML) - Hugo
20
	// *** - Alternative format
21
	const frontmatterRegex = /^(---|\+\+\+|\*\*\*)\n([\s\S]*?)\n\1\n([\s\S]*)$/;
22
	const match = content.match(frontmatterRegex);
23
24
	if (!match) {
25
		throw new Error("Could not parse frontmatter");
26
	}
27
28
	const delimiter = match[1];
29
	const frontmatterStr = match[2] ?? "";
30
	const body = match[3] ?? "";
31
32
	// Determine format based on delimiter:
33
	// +++ uses TOML (key = value)
34
	// --- and *** use YAML (key: value)
35
	const isToml = delimiter === "+++";
36
37
	// Parse frontmatter using the appropriate library
38
	let raw: Record<string, unknown>;
39
	if (isToml) {
40
		raw = toml.parse(frontmatterStr) as Record<string, unknown>;
41
	} else {
42
		// Use CORE_SCHEMA to keep dates as strings rather than Date objects
43
		raw =
44
			(yaml.load(frontmatterStr, { schema: yaml.CORE_SCHEMA }) as Record<
45
				string,
46
				unknown
47
			>) ?? {};
48
	}
49
50
	// Apply field mappings to normalize to standard PostFrontmatter fields
51
	const frontmatter: Record<string, unknown> = {};
52
53
	// Title mapping
54
	const titleField = mapping?.title || "title";
55
	frontmatter.title = raw[titleField] || raw.title;
56
57
	// Description mapping
58
	const descField = mapping?.description || "description";
59
	frontmatter.description = raw[descField] || raw.description;
60
61
	// Publish date mapping - check custom field first, then fallbacks
62
	const dateField = mapping?.publishDate;
63
	if (dateField && raw[dateField]) {
64
		frontmatter.publishDate = raw[dateField];
65
	} else if (raw.publishDate) {
66
		frontmatter.publishDate = raw.publishDate;
67
	} else {
68
		// Fallback to common date field names
69
		const dateFields = ["pubDate", "date", "createdAt", "created_at"];
70
		for (const field of dateFields) {
71
			if (raw[field]) {
72
				frontmatter.publishDate = raw[field];
73
				break;
74
			}
75
		}
76
	}
77
78
	// Cover image mapping
79
	const coverField = mapping?.coverImage || "ogImage";
80
	frontmatter.ogImage = raw[coverField] || raw.ogImage;
81
82
	// Tags mapping
83
	const tagsField = mapping?.tags || "tags";
84
	frontmatter.tags = raw[tagsField] || raw.tags;
85
86
	// Draft mapping
87
	const draftField = mapping?.draft || "draft";
88
	const draftValue = raw[draftField] ?? raw.draft;
89
	if (draftValue !== undefined) {
90
		frontmatter.draft = draftValue === true || draftValue === "true";
91
	}
92
93
	// Always preserve atUri (internal field)
94
	frontmatter.atUri = raw.atUri;
95
96
	return {
97
		frontmatter: frontmatter as unknown as PostFrontmatter,
98
		body,
99
		rawFrontmatter: raw,
100
	};
101
}
102
103
export function getSlugFromFilename(filename: string): string {
104
	return filename
105
		.replace(/\.mdx?$/, "")
106
		.toLowerCase()
107
		.replace(/\s+/g, "-");
108
}
109
110
export interface SlugOptions {
111
	slugField?: string;
112
	removeIndexFromSlug?: boolean;
113
	stripDatePrefix?: boolean;
114
}
115
116
export function getSlugFromOptions(
117
	relativePath: string,
118
	rawFrontmatter: Record<string, unknown>,
119
	options: SlugOptions = {},
120
): string {
121
	const {
122
		slugField,
123
		removeIndexFromSlug = false,
124
		stripDatePrefix = false,
125
	} = options;
126
127
	let slug: string;
128
129
	// If slugField is set, try to get the value from frontmatter
130
	if (slugField) {
131
		const frontmatterValue = rawFrontmatter[slugField];
132
		if (frontmatterValue && typeof frontmatterValue === "string") {
133
			// Remove leading slash if present
134
			slug = frontmatterValue
135
				.replace(/^\//, "")
136
				.toLowerCase()
137
				.replace(/\s+/g, "-");
138
		} else {
139
			// Fallback to filepath if frontmatter field not found
140
			slug = relativePath
141
				.replace(/\.mdx?$/, "")
142
				.toLowerCase()
143
				.replace(/\s+/g, "-");
144
		}
145
	} else {
146
		// Default: use filepath
147
		slug = relativePath
148
			.replace(/\.mdx?$/, "")
149
			.toLowerCase()
150
			.replace(/\s+/g, "-");
151
	}
152
153
	// Remove /index or /_index suffix if configured
154
	if (removeIndexFromSlug) {
155
		slug = slug.replace(/\/_?index$/, "");
156
	}
157
158
	// Strip Jekyll-style date prefix (YYYY-MM-DD-) from filename
159
	if (stripDatePrefix) {
160
		slug = slug.replace(/(^|\/)(\d{4}-\d{2}-\d{2})-/g, "$1");
161
	}
162
163
	return slug;
164
}
165
166
export function resolvePathTemplate(template: string, post: BlogPost): string {
167
	const publishDate = new Date(post.frontmatter.publishDate);
168
	const year = String(publishDate.getFullYear());
169
	const yearUTC = String(publishDate.getUTCFullYear());
170
	const month = String(publishDate.getMonth() + 1).padStart(2, "0");
171
	const monthUTC = String(publishDate.getUTCMonth() + 1).padStart(2, "0");
172
	const day = String(publishDate.getDate()).padStart(2, "0");
173
	const dayUTC = String(publishDate.getUTCDate()).padStart(2, "0");
174
175
	const slugifiedTitle = (post.frontmatter.title || "")
176
		.toLowerCase()
177
		.replace(/\s+/g, "-")
178
		.replace(/[^\w-]/g, "");
179
180
	// Replace known tokens
181
	let result = template
182
		.replace(/\{slug\}/g, post.slug)
183
		.replace(/\{year\}/g, year)
184
		.replace(/\{yearUTC\}/g, yearUTC)
185
		.replace(/\{month\}/g, month)
186
		.replace(/\{monthUTC\}/g, monthUTC)
187
		.replace(/\{day\}/g, day)
188
		.replace(/\{dayUTC\}/g, dayUTC)
189
		.replace(/\{title\}/g, slugifiedTitle);
190
191
	// Replace any remaining {field} tokens with raw frontmatter values
192
	result = result.replace(/\{(\w+)\}/g, (_match, field: string) => {
193
		const value = post.rawFrontmatter[field];
194
		if (value != null && typeof value === "string") {
195
			return value;
196
		}
197
		return "";
198
	});
199
200
	// Ensure leading slash
201
	if (!result.startsWith("/")) {
202
		result = `/${result}`;
203
	}
204
205
	return result;
206
}
207
208
export function resolvePostPath(
209
	post: BlogPost,
210
	pathPrefix?: string,
211
	pathTemplate?: string,
212
): string {
213
	if (pathTemplate) {
214
		return resolvePathTemplate(pathTemplate, post);
215
	}
216
	const prefix = pathPrefix || "/posts";
217
	return `${prefix}/${post.slug}`;
218
}
219
220
export async function getContentHash(content: string): Promise<string> {
221
	const encoder = new TextEncoder();
222
	const data = encoder.encode(content);
223
	const hashBuffer = await crypto.subtle.digest("SHA-256", data);
224
	const hashArray = Array.from(new Uint8Array(hashBuffer));
225
	return hashArray.map((b) => b.toString(16).padStart(2, "0")).join("");
226
}
227
228
function shouldIgnore(relativePath: string, ignorePatterns: string[]): boolean {
229
	for (const pattern of ignorePatterns) {
230
		if (minimatch(relativePath, pattern)) {
231
			return true;
232
		}
233
	}
234
	return false;
235
}
236
237
export interface ScanOptions {
238
	frontmatterMapping?: FrontmatterMapping;
239
	ignorePatterns?: string[];
240
	slugField?: string;
241
	removeIndexFromSlug?: boolean;
242
	stripDatePrefix?: boolean;
243
}
244
245
export async function scanContentDirectory(
246
	contentDir: string,
247
	frontmatterMappingOrOptions?: FrontmatterMapping | ScanOptions,
248
	ignorePatterns: string[] = [],
249
): Promise<BlogPost[]> {
250
	// Handle both old signature (frontmatterMapping, ignorePatterns) and new signature (options)
251
	let options: ScanOptions;
252
	if (
253
		frontmatterMappingOrOptions &&
254
		("frontmatterMapping" in frontmatterMappingOrOptions ||
255
			"ignorePatterns" in frontmatterMappingOrOptions ||
256
			"slugField" in frontmatterMappingOrOptions)
257
	) {
258
		options = frontmatterMappingOrOptions as ScanOptions;
259
	} else {
260
		// Old signature: (contentDir, frontmatterMapping?, ignorePatterns?)
261
		options = {
262
			frontmatterMapping: frontmatterMappingOrOptions as
263
				| FrontmatterMapping
264
				| undefined,
265
			ignorePatterns,
266
		};
267
	}
268
269
	const {
270
		frontmatterMapping,
271
		ignorePatterns: ignore = [],
272
		slugField,
273
		removeIndexFromSlug,
274
		stripDatePrefix,
275
	} = options;
276
277
	const patterns = ["**/*.md", "**/*.mdx"];
278
	const posts: BlogPost[] = [];
279
280
	for (const pattern of patterns) {
281
		const files = await glob(pattern, {
282
			cwd: contentDir,
283
			absolute: false,
284
		});
285
286
		for (const relativePath of files) {
287
			// Skip files matching ignore patterns
288
			if (shouldIgnore(relativePath, ignore)) {
289
				continue;
290
			}
291
292
			const filePath = path.join(contentDir, relativePath);
293
			const rawContent = await fs.readFile(filePath, "utf-8");
294
295
			try {
296
				const { frontmatter, body, rawFrontmatter } = parseFrontmatter(
297
					rawContent,
298
					frontmatterMapping,
299
				);
300
				const slug = getSlugFromOptions(relativePath, rawFrontmatter, {
301
					slugField,
302
					removeIndexFromSlug,
303
					stripDatePrefix,
304
				});
305
306
				posts.push({
307
					filePath,
308
					slug,
309
					frontmatter,
310
					content: body,
311
					rawContent,
312
					rawFrontmatter,
313
				});
314
			} catch (error) {
315
				console.error(`Error parsing ${relativePath}:`, error);
316
			}
317
		}
318
	}
319
320
	// Sort by publish date (newest first)
321
	posts.sort((a, b) => {
322
		const dateA = new Date(a.frontmatter.publishDate);
323
		const dateB = new Date(b.frontmatter.publishDate);
324
		return dateB.getTime() - dateA.getTime();
325
	});
326
327
	return posts;
328
}
329
330
export function updateFrontmatterWithAtUri(
331
	rawContent: string,
332
	atUri: string,
333
): string {
334
	// Detect which delimiter is used (---, +++, or ***)
335
	const delimiterMatch = rawContent.match(/^(---|\+\+\+|\*\*\*)/);
336
	const delimiter = delimiterMatch?.[1] ?? "---";
337
	const isToml = delimiter === "+++";
338
339
	// Format the atUri entry based on frontmatter type
340
	const atUriEntry = isToml ? `atUri = "${atUri}"` : `atUri: "${atUri}"`;
341
342
	// Check if atUri already exists in frontmatter (handle both formats)
343
	if (rawContent.includes("atUri:") || rawContent.includes("atUri =")) {
344
		// Replace existing atUri (match both YAML and TOML formats)
345
		return rawContent.replace(
346
			/atUri\s*[=:]\s*["']?[^"'\n]+["']?\n?/,
347
			`${atUriEntry}\n`,
348
		);
349
	}
350
351
	// Insert atUri before the closing delimiter
352
	const frontmatterEndIndex = rawContent.indexOf(delimiter, 4);
353
	if (frontmatterEndIndex === -1) {
354
		throw new Error("Could not find frontmatter end");
355
	}
356
357
	const beforeEnd = rawContent.slice(0, frontmatterEndIndex);
358
	const afterEnd = rawContent.slice(frontmatterEndIndex);
359
360
	return `${beforeEnd}${atUriEntry}\n${afterEnd}`;
361
}
362
363
export function stripMarkdownForText(markdown: string): string {
364
	return markdown
365
		.replace(/#{1,6}\s/g, "") // Remove headers
366
		.replace(/\*\*([^*]+)\*\*/g, "$1") // Remove bold
367
		.replace(/\*([^*]+)\*/g, "$1") // Remove italic
368
		.replace(/\[([^\]]+)\]\([^)]+\)/g, "$1") // Remove links, keep text
369
		.replace(/`{3}[\s\S]*?`{3}/g, "") // Remove code blocks
370
		.replace(/`([^`]+)`/g, "$1") // Remove inline code formatting
371
		.replace(/!\[.*?\]\(.*?\)/g, "") // Remove images
372
		.replace(/\n{3,}/g, "\n\n") // Normalize multiple newlines
373
		.trim();
374
}