cc-webView.ts

import { tool } from "@opencode-ai/plugin"
import { spawn } from "node:child_process"
import { readFile } from "node:fs/promises"
import { existsSync } from "node:fs"

export default tool({
  description:
    "Fetch a URL or local HTML file, extract structured content into blocks, " +
    "and download images. Writes <out-dir>/article.json and <out-dir>/images/*. " +
    "Returns a summary with block count, image stats, and login-wall detection. " +
    "Use this BEFORE the ccParseWebArticle skill to translate the page into bilingual Markdown.",
  args: {
    input: tool.schema
      .string()
      .describe("URL (https://...) or path to local HTML file"),
    out_dir: tool.schema
      .string()
      .describe("Output directory (will be created if missing)"),
    images_dir: tool.schema
      .string()
      .describe("Subdirectory for images. Default: images"),
    json_out: tool.schema
      .string()
      .describe("Path for article.json. Default: <out-dir>/article.json"),
    timeout: tool.schema
      .number()
      .describe("Fetch timeout in seconds. Default: 30"),
    max_image_size: tool.schema
      .number()
      .describe("Skip images larger than this (bytes). Default: 10485760 (10MB)"),
    fail_on_image_error: tool.schema
      .boolean()
      .describe("Exit with code 3 on image download failure"),
    cookies: tool.schema
      .string()
      .describe(
        'HTTP cookies in "key=value; key2=value2" format. ' +
          "Copy from browser DevTools → Application → Cookies.",
      ),
  },
  async execute(args, _ctx) {
    // Build cc-webView CLI argv from typed args
    const argv: string[] = [
      "extract",
      args.input,
      "--out-dir",
      args.out_dir,
    ]
    if (args.images_dir !== undefined)
      argv.push("--images-dir", args.images_dir)
    if (args.json_out !== undefined) argv.push("--json-out", args.json_out)
    if (args.timeout !== undefined) argv.push("--timeout", String(args.timeout))
    if (args.max_image_size !== undefined)
      argv.push("--max-image-size", String(args.max_image_size))
    if (args.fail_on_image_error === true) argv.push("--fail-on-image-error")
    if (args.cookies !== undefined) argv.push("--cookies", args.cookies)

    // Run cc-webView, capture stderr (cc-webView logs progress to stderr)
    const { code, stderr } = await new Promise<{
      code: number | null
      stderr: string
    }>((resolve, reject) => {
      const proc = spawn("cc-webView", argv, {
        stdio: ["ignore", "pipe", "pipe"],
      })
      let err = ""
      proc.stderr.on("data", (d: Buffer) => {
        err += d.toString()
      })
      proc.on("error", reject)
      proc.on("close", (code: number | null) => resolve({ code, stderr: err }))
    })

    // Exit codes: 0 ok, 2 login wall (still success — user must authenticate),
    // 1 fetch error, 3 image error.
    if (code !== 0 && code !== 2) {
      throw new Error(
        `cc-webView failed (exit ${code}).\n` +
          (stderr.trim() || "no stderr output"),
      )
    }

    // Read the resulting article.json and return a structured summary
    const jsonPath = args.json_out ?? `${args.out_dir}/article.json`
    if (!existsSync(jsonPath)) {
      return (
        `cc-webView exited with code ${code} but ${jsonPath} was not produced.\n` +
        `stderr:\n${stderr.trim()}`
      )
    }

    const article = JSON.parse(await readFile(jsonPath, "utf-8"))
    const summary = {
      status: code === 2 ? "login_wall" : "ok",
      title: article.title ?? null,
      source: article.source ?? args.input,
      blocks: article.blocks?.length ?? 0,
      images: {
        total: article.images_summary?.total ?? 0,
        downloaded: article.images_summary?.downloaded ?? 0,
        failed: article.images_summary?.failed ?? 0,
        dir: `${args.out_dir}/${args.images_dir ?? "images"}`,
      },
      login_wall: article.login_wall?.detected
        ? {
            patterns: article.login_wall.patterns_matched,
            hint: article.login_wall.hint ?? null,
          }
        : null,
      article_json: jsonPath,
    }
    return JSON.stringify(summary, null, 2)
  },
})