muse/src/core/sources.ts

import { Glob } from "bun";
import { type LoadedFile, loadFileContent } from "./files";
import { type UnknownRecord, autoParseEntry } from "./records";
import { z } from "zod";
import { MuseError } from "#errors";

export const SourceSchema = z.union([
  z.object({
    file: z.string(),
  }),
  z.object({
    url: z.string().url(),
  }),
]);

export type SourceShorthand = z.infer<typeof SourceSchema>;

export function sourceShorthandToSource(
  shorthand: SourceShorthand,
): MuseSource {
  if ("file" in shorthand) {
    return {
      location: shorthand.file,
      type: "file",
    };
  }

  if ("url" in shorthand) {
    return {
      location: shorthand.url,
      type: "url",
    };
  }

  throw new MuseError("Invalid source shorthand");
}

export type SourceType = "file" | "url";

/** A descriptor of a location to load into a Muse binding */
export interface MuseSource {
  location: string;
  type: SourceType;
}

interface SourceOptions {
  cwd: string;
  contentKey: string;
  ignore: string[];
}

/** A single loaded entry from a Muse source */
export interface MuseEntry<MetaShape = UnknownRecord> {
  _raw: string;
  location: string;
  data: Record<string, unknown>;
  meta: MetaShape;
  source: MuseSource;
}

export async function parseMuseFile(
  rawFilePath: string,
  { contentKey = "content" }: SourceOptions,
): Promise<MuseEntry[]> {
  const file = await loadFileContent(rawFilePath);
  const entries = autoParseEntry(file, contentKey);

  const partial = {
    _raw: file.content,
    source: {
      location: file.filePath,
      type: "file" as SourceType,
    },
    location: file.filePath,
    meta: {},
  };

  return entries.map((data) => ({
    ...partial,
    data,
  }));
}

async function loadFromFileSource(
  source: MuseSource,
  options: SourceOptions,
): Promise<MuseEntry[]> {
  const paths = Array.from(
    new Glob(source.location).scanSync({
      cwd: options.cwd,
      absolute: true,
      followSymlinks: true,
      onlyFiles: true,
    }),
  );

  const filteredPaths = paths.filter((path) => !options.ignore.includes(path));

  const entries: MuseEntry[] = [];
  for (const filePath of filteredPaths) {
    const fileEntries = await parseMuseFile(filePath, options);
    entries.push(...fileEntries);
  }

  return entries;
}

function getFileExtensionFromURL(url: string): string | null {
  const parsedUrl = new URL(url);

  const pathname = parsedUrl.pathname;
  const filename = pathname.substring(pathname.lastIndexOf("/") + 1);
  const extension = filename.substring(filename.lastIndexOf(".") + 1);

  return extension === filename ? null : extension;
}

async function loadFromURLSource(
  source: MuseSource,
  options: SourceOptions,
): Promise<MuseEntry[]> {
  const { contentKey = "content" } = options;
  const response = await fetch(source.location);

  if (!response.ok) {
    throw new Error(`Failed to fetch URL: ${source.location}`);
  }
  const content = await response.text();
  const mimeType = response.headers.get("Content-Type") || "unknown";

  const parseType = getFileExtensionFromURL(source.location) ?? "unknown";

  const loadedFile: LoadedFile = {
    content,
    filePath: source.location,
    fileType: parseType,
    dirname: "",
    basename: "",
    mimeType,
  };

  const entries = autoParseEntry(loadedFile, contentKey);

  const partial = {
    _raw: content,
    source: {
      location: source.location,
      type: "url" as SourceType,
    },
    location: source.location,
    meta: {},
  };

  return entries.map((data) => ({
    ...partial,
    data,
  }));
}

export async function loadFromSource(
  source: MuseSource,
  options: SourceOptions,
): Promise<MuseEntry[]> {
  switch (source.type) {
    case "file":
      return loadFromFileSource(source, options);
    case "url":
      return loadFromURLSource(source, options);
    default:
      throw new Error(`Unsupported source type: ${source.type}`);
  }
}