content.config.ts

import { defineCollection, z } from 'astro:content';
import { parseNewsFrontmatterDate } from './lib/dates';
import { glob } from 'astro/loaders';

/** `src/content/<name>/` — absolute file URL avoids resolving `./src/...` from the wrong cwd. */
function contentGlobBase(segment: string) {
	return new URL(`./content/${segment}/`, import.meta.url);
}

const mastheadTopicItemSchema = z.union([
	z.string(),
	z.object({
		label: z.string(),
		tone: z.enum(['core', 'code', 'apps', 'seminar', 'neutral']).optional(),
	}),
]);

/**
 * Dispatch labels — named slots (flattened to pills in order:
 * researchField → applicationDomain → task).
 */
const structuredTopicsSchema = z
	.object({
		researchField: mastheadTopicItemSchema.optional(),
		applicationDomain: mastheadTopicItemSchema,
		task: mastheadTopicItemSchema,
	})
	.optional();

const people = defineCollection({
	loader: glob({ pattern: '**/*.md', base: contentGlobBase('people') }),
	schema: z.object({
		name: z.string(),
		role: z.string(),
		/** Section on the People page (current vs alumni subsections). */
		peopleTier: z
			.enum(['faculty', 'postdoc', 'phd', 'ms', 'undergrad', 'research_assistant', 'visitor'])
			.default('visitor'),
		current: z.boolean().default(true),
		/** Cohort / program label shown when term range isn’t used */
		cohort: z.string().optional(),
		/** If set, shown under the name on `/people/` and the profile hero instead of term range/cohort */
		directoryMeta: z.string().optional(),
		/** Academic start semester, e.g. `Fall 2021` — drives People sort order */
		startTerm: z.string().optional(),
		/** End semester or `Present` if omitted while `current` */
		endTerm: z.string().optional(),
		website: z.string().url().optional(),
		github: z.string().url().optional(),
		linkedin: z.string().url().optional(),
		/** Google Scholar profile URL — prefer if set; Semantic Scholar remains optional separately. */
		googleScholar: z.string().url().optional(),
		semanticScholar: z.string().url().optional(),
		photo: z.string().optional(),
		/**
		 * Optional wide photo shown after Education / theses / publications / achievements on the profile page.
		 * Path under `public/` (e.g. `/news/…`).
		 */
		footerPhoto: z
			.object({
				src: z.string(),
				alt: z.string().optional(),
				caption: z.string().optional(),
			})
			.optional(),
		/** Optional extra photos after Education / thesis / publications / achievements (after `footerPhoto` if set). */
		footerPhotos: z
			.array(
				z.object({
					src: z.string(),
					alt: z.string().optional(),
					caption: z.string().optional(),
				}),
			)
			.optional(),
		aliases: z.array(z.string()).optional(),
		/**
		 * Alternate author strings used on papers/theses only (normalized + linked on `/papers/`).
		 * Omit from roster display / profile parentheticals (`aliases` is for preferred/nickname display).
		 */
		publicationAuthorAliases: z.array(z.string()).optional(),
		/** When `false`, omit the auto-linked Publications section on this profile. */
		profilePublications: z.boolean().optional(),
		/**
		 * Honors, grants, competition results, etc.
		 * Prefer `title` (what happened) + `when` (timing) + optional `issuer` (who gave it) instead of one long `title` string.
		 * Legacy entries may still embed dates in `title`; the site splits trailing `(...)` when it looks date-like.
		 * Editorial style: short noun-first lines; em dash for results (`Contest — 1st place (Team)`); awards as `Award, Emory Computer Science`.
		 * Do not list undergraduate thesis honor level here (`Highest Honor in …`, etc.) — use `honorsLevel` on the linked thesis entry for the thesis detail page; the people profile thesis list omits it (see Achievements).
		 */
		achievements: z
			.array(
				z.object({
					title: z.string(),
					when: z.string().optional(),
					issuer: z.string().optional(),
					url: z.string().optional(),
				}),
			)
			.optional(),
		/** Degrees/postdoc training etc.; Education lists show `endYear` only. Composed bios use `endTerm` when set (`Spring 2026`), else `endYear`. Optional `startTerm` / `endTerm` also affect bio sort order. */
		education: z
			.array(
				z.object({
					degree: z.string(),
					institution: z.string(),
					/** e.g. `Fall 2021` — optional; composed-bio chronological sort. */
					startTerm: z.string().optional(),
					/** e.g. `Spring 2026` — optional; shown in composed bio; Education list still shows `endYear`. */
					endTerm: z.string().optional(),
					endYear: z.number().int().optional(),
					ongoing: z.boolean().optional(),
					notes: z.string().optional(),
				}),
			)
			.optional(),
	}),
});

const papers = defineCollection({
	loader: glob({ pattern: '**/*.md', base: contentGlobBase('papers') }),
	schema: z.object({
		title: z.string(),
		/** Trim each line — avoids stray spaces before/after commas in listings. */
		authors: z.array(z.string()).transform((authors) => authors.map((s) => s.trim())),
		venue: z.string(),
		/** Short listing badge (rail); inferred from venue when omitted. */
		venueAbbrev: z.string().optional(),
		year: z.number().int(),
		abstract: z.string().optional(),
		/** Repository, Hugging Face model/dataset, or other artifact (shown as “Resources” on the paper page). */
		resourceUrl: z.string().optional(),
		/** @deprecated Use `resourceUrl` — still accepted for older posts. */
		codeUrl: z.string().optional(),
		/** Canonical paper/preprint landing (ACL page, DOI, arXiv abs, …). Detail page Links row omits PDF/Bib only. */
		paperUrl: z.string().url().optional(),
		/** Poster PDF or landing URL — shown in the Links row after Paper / Resources. */
		posterUrl: z.string().url().optional(),
		/** Slides PDF or deck URL — shown in the Links row. */
		slidesUrl: z.string().url().optional(),
		/** Venue row hyperlink — used verbatim when set (e.g. journal portal). */
		venueUrl: z.string().url().optional(),
		publicationType: z.enum(['conference', 'journal', 'preprint', 'workshop', 'other']).optional(),
		/** Optional masthead line (reserved; not shown on the `/papers/` cards). */
		masthead: z.string().optional(),
		/** Named Dispatch labels: `applicationDomain` + `task`; optional `researchField`. */
		topics: structuredTopicsSchema,
		/** Listing teaser copy; otherwise first sentence of `abstract`, or truncated abstract. See `publicationDek`. */
		dek: z.string().optional(),
		/** Detail page: optional footnote after authors, shown as `*: …` (e.g. proceedings vs arXiv). */
		authorFootnote: z.string().optional(),
		/** Must match an `authors` entry exactly; that name is shown with a trailing `*`. */
		authorFootnoteFor: z.string().optional(),
		/** Detail page row below Links (“Present”). Multiple people: comma-separated (split on commas); each segment is matched to `/people/` like `authors`. */
		presenter: z.string().optional(),
		/**
		 * Calendar date when the paper was published or presented (`YYYY-MM-DD` recommended).
		 * Drives `/papers/` listing order (newer first); if omitted, sort uses Jan 1 of `year`.
		 * When `forthcoming` is true, omit this (no placeholder date); listing sort uses end of `year` instead.
		 */
		published: z
			.union([z.string(), z.date()])
			.optional()
			.transform((val): Date | undefined => {
				if (val === undefined) return undefined;
				return parseNewsFrontmatterDate(val);
			}),
		/** Accepted or in press — not yet published; do not set a fake `published` date. Sorting stays stable (see `publicationSortInstant`). */
		forthcoming: z.boolean().optional(),
	}),
});

const theses = defineCollection({
	loader: glob({ pattern: '**/*.md', base: contentGlobBase('theses') }),
	schema: z.object({
		title: z.string(),
		author: z.string(),
		degree: z.enum([
			'PhD',
			'MS',
			'BS',
			'BA',
			'Undergraduate Honors',
			'Undergraduate Thesis',
			'Masters Thesis',
			'Dissertation',
			'Other',
		]),
		abstract: z.string().optional(),
		term: z.string().optional(),
		department: z.string().optional(),
		paperUrl: z.string().url().optional(),
		slidesUrl: z.string().url().optional(),
		committee: z.array(z.string()).optional(),
		honorsLevel: z.enum(['Highest Honor', 'High Honor', 'Honor']).optional(),
		photo: z.string().optional(),
		sourceUrl: z.string().url().optional(),
		/** Optional topic labels for Dispatch; when omitted, inferred from title and abstract. */
		topics: structuredTopicsSchema,
	}),
});

const dispatchActivityKind = z.enum(['award', 'media', 'service', 'travel', 'welcome', 'social', 'other']);

const dispatchDistinctionKind = z.enum(['award', 'honor', 'other']);

const dispatchFeaturedFlag = z.literal(true).optional();

const dispatchNewsLineItem = z.object({
	date: z.union([z.string(), z.date()]).transform((val) => parseNewsFrontmatterDate(val)),
	text: z.string(),
	highlight: z.string().optional(),
	featured: dispatchFeaturedFlag,
	blurb: z.string().optional(),
});

const dispatchPublicationRef = z.union([
	z.string(),
	z.object({
		publication: z.string(),
		featured: dispatchFeaturedFlag,
		blurb: z.string().optional(),
	}),
]);

const dispatchThesisRef = z.union([
	z.string(),
	z.object({
		thesis: z.string(),
		featured: dispatchFeaturedFlag,
		blurb: z.string().optional(),
	}),
]);

const dispatchColumnFormat = z.enum(['full', 'external']);

const dispatchColumns = defineCollection({
	loader: glob({ pattern: '**/*.md', base: contentGlobBase('dispatch-columns') }),
	schema: z
		.object({
			/** Dispatch issue slug this column belongs to (e.g. `2026-spring`). */
			issue: z.string(),
			/** `full` — markdown body on the issue page; `external` — teaser linking to source. */
			format: dispatchColumnFormat.default('full'),
			headline: z.string(),
			/** Full-format pull quote shown above the body. */
			pullQuote: z.string().optional(),
			/** External-format subtitle (e.g. Substack dek). */
			subtitle: z.string().optional(),
			/** External-format hook line under the subtitle. */
			tagline: z.string().optional(),
			/** External-format episode or article summary. */
			summary: z.string().optional(),
			/** External-format SEO / social meta description from the source page. */
			seoDescription: z.string().optional(),
			/** Cover image path under `public/` (e.g. `/dispatch/2026-spring-lights-out-web-1.webp`). */
			coverImage: z.string().optional(),
			/** CSS object-position for the cover image. */
			coverImagePosition: z.string().optional(),
			byline: z.string().optional(),
			/** Publication date (e.g. on Substack). */
			published: z.coerce.date().optional(),
			/** Photo path under `public/` (e.g. `/people/jinho-choi.webp`). */
			avatar: z.string().optional(),
			/** Canonical URL when republished from elsewhere (e.g. Substack). */
			sourceUrl: z.string().url().optional(),
			sourceLabel: z.string().optional(),
			/** iframe `src` for an embedded player (e.g. Substack post embed). */
			embedUrl: z.string().url().optional(),
			/** Footnote shown below external teasers (e.g. podcast disclaimer). */
			note: z.string().optional(),
		})
		.superRefine((data, ctx) => {
			if (data.format === 'full' && !data.pullQuote?.trim()) {
				ctx.addIssue({
					code: 'custom',
					message: 'pullQuote is required when format is full',
					path: ['pullQuote'],
				});
			}
			if (data.format === 'external') {
				if (!data.sourceUrl) {
					ctx.addIssue({
						code: 'custom',
						message: 'sourceUrl is required when format is external',
						path: ['sourceUrl'],
					});
				}
				for (const field of ['subtitle', 'summary'] as const) {
					if (!data[field]?.trim()) {
						ctx.addIssue({
							code: 'custom',
							message: `${field} is required when format is external`,
							path: [field],
						});
					}
				}
			}
		}),
});

const dispatchHomepageHighlight = z
	.object({
		kind: z.string(),
		label: z.string().optional(),
		publication: z.string().optional(),
		column: z.string().optional(),
		highlight: z.string().optional(),
		tagTone: z.enum(['core', 'code', 'apps', 'seminar', 'neutral']).optional(),
	})
	.superRefine((item, ctx) => {
		const refs = [item.publication, item.column, item.highlight].filter(Boolean);
		if (refs.length !== 1) {
			ctx.addIssue({
				code: 'custom',
				message: 'Each homepage highlight must set exactly one of publication, column, or highlight',
				path: ['publication'],
			});
		}
	});

const dispatch = defineCollection({
	loader: glob({ pattern: '**/*.md', base: contentGlobBase('dispatch') }),
	schema: z.object({
		issueTitle: z.string(),
		volume: z.number().int(),
		issue: z.number().int(),
		periodLabel: z.string(),
		periodStart: z.coerce.date(),
		periodEnd: z.coerce.date(),
		published: z.coerce.date(),
		tagline: z.string().optional(),
		papers: z.array(dispatchPublicationRef),
		/** Omit or leave empty to hide the theses section on the issue page. */
		theses: z.array(dispatchThesisRef).optional(),
		/** Highlights shown on the home page from the current dispatch issue. */
		homepageHighlights: z.array(dispatchHomepageHighlight).optional(),
		/** Ordered column article slugs (`src/content/dispatch-columns/`). */
		columns: z.array(z.string()),
		/** Omit to hide the student spotlight section on the issue page. */
		studentSpotlight: z
			.object({
				name: z.string(),
				degree: z.string(),
				blurb: z.string(),
				project: z.string(),
				personSlug: z.string().optional(),
				photo: z.string().optional(),
			})
			.optional(),
		/** Omit or leave empty to hide the Distinctions section on the issue page. */
		distinctions: z
			.array(
				dispatchNewsLineItem.extend({
					kind: dispatchDistinctionKind,
				}),
			)
			.optional(),
		activityNews: z.array(
			dispatchNewsLineItem.extend({
				kind: dispatchActivityKind.optional(),
			}),
		),
		/** Omit to hide the reading list section on the issue page. */
		readingList: z
			.array(
				z.object({
					why: z.string(),
					publication: z.string().optional(),
					title: z.string().optional(),
					authors: z.string().optional(),
					venue: z.string().optional(),
					url: z.string().url().optional(),
				}),
			)
			.optional(),
		heroVisual: z.enum(['radar', 'graph', 'tree']).optional(),
		/** Short paragraph summarizing the issue’s key points (cover + archive). */
		issueSummary: z.string().optional(),
	}),
});

const seminars = defineCollection({
	loader: glob({ pattern: '**/*.md', base: contentGlobBase('seminars') }),
	schema: z.object({
		title: z.string(),
		date: z.coerce.date(),
		term: z.string(),
		/** Wall-clock time line from the legacy seminar page (e.g. `3:00 - 4:00 PM`). */
		time: z.string().optional(),
		speakers: z.array(z.string()).optional(),
		abstract: z.string().optional(),
		/** Guest or speaker biography; rendered like the abstract on the detail page. */
		bio: z.string().optional(),
		location: z.string().optional(),
		videoUrl: z.string().url().optional(),
		slidesUrl: z.string().url().optional(),
		/** Speaker homepage (guest talks without a people profile). */
		speakerUrl: z.string().url().optional(),
		/** Cover image path under `public/` (e.g. `/seminars/*.webp`). */
		coverImage: z.string().optional(),
	}),
});

const news = defineCollection({
	loader: glob({ pattern: '**/*.md', base: contentGlobBase('news') }),
	schema: z.object({
		title: z.string(),
		/** `YYYY-MM-DD` (UTC), `YYYY-MM-DD-HH:MM` (Eastern wall time), or any string JS `Date` accepts */
		date: z.union([z.string(), z.date()]).transform((val): Date => parseNewsFrontmatterDate(val)),
		featured: z.boolean().optional(),
		/** Cover image path under `public/` (e.g. `/news/*.jpg`) — listing thumbnails and home carousel */
		coverImage: z.string().optional(),
		/** CSS object-position for listing and home cover thumbnails only (cards, carousel). Not applied to images inside the article body. */
		coverImagePosition: z.string().optional(),
		/** Free-form labels rendered as pills, e.g. `social`, `paper`, `visit`, `conferences` */
		labels: z.array(z.string()).optional(),
		/** People slugs (`src/content/people/{slug}.md`). Missing files still render (name derived from slug); add profiles to enable links + People page listings. */
		participants: z.array(z.string()).optional(),
	}),
});

export const collections = {
	people,
	papers,
	theses,
	dispatch,
	dispatchColumns,
	seminars,
	news,
};