utils.js
1 import { AuthRequiredError, CommandExecutionError } from '@jackwener/opencli/errors'; 2 export function stripHtml(html) { 3 if (!html) 4 return ''; 5 const decoded = html 6 .replace(/\\u003c/g, '<') 7 .replace(/\\u003e/g, '>') 8 .replace(/\\n/g, '\n') 9 .replace(/\\r/g, ''); 10 return decoded.replace(/<[^>]+>/g, '').trim(); 11 } 12 export function decodeHtmlEntities(html) { 13 if (!html) 14 return ''; 15 return html.replace(/ /g, ' ') 16 .replace(/</g, '<') 17 .replace(/>/g, '>') 18 .replace(/&/g, '&') 19 .replace(/"/g, '"') 20 .replace(/'/g, "'"); 21 } 22 export function getHupuThreadUrl(tid) { 23 return `https://bbs.hupu.com/${encodeURIComponent(String(tid))}-1.html`; 24 } 25 export function getHupuSearchUrl(query, page, forum, sort) { 26 const searchParams = new URLSearchParams(); 27 searchParams.append('q', String(query)); 28 searchParams.append('page', String(page)); 29 if (forum) { 30 searchParams.append('topicId', String(forum)); 31 } 32 if (sort) { 33 searchParams.append('sortby', String(sort)); 34 } 35 return `https://bbs.hupu.com/search?${searchParams.toString()}`; 36 } 37 export async function readHupuNextData(page, url, actionLabel, options = {}) { 38 await page.goto(url); 39 const result = await page.evaluate(` 40 (async () => { 41 const wait = (ms) => new Promise((resolve) => setTimeout(resolve, ms)); 42 const expectedTid = ${JSON.stringify(options.expectedTid || '')}; 43 const timeoutMs = ${JSON.stringify(options.timeoutMs ?? 5000)}; 44 let lastSeenTid = ''; 45 let lastSeenHref = ''; 46 47 const waitFor = async (predicate, limitMs = timeoutMs) => { 48 const start = Date.now(); 49 while (Date.now() - start < limitMs) { 50 if (predicate()) return true; 51 await wait(100); 52 } 53 return false; 54 }; 55 56 const ready = await waitFor(() => { 57 const script = document.getElementById('__NEXT_DATA__'); 58 if (!script?.textContent) return false; 59 60 lastSeenHref = location.href; 61 62 try { 63 const parsed = JSON.parse(script.textContent); 64 const threadTid = parsed?.props?.pageProps?.detail?.thread?.tid; 65 lastSeenTid = typeof threadTid === 'string' ? threadTid : ''; 66 67 if (!expectedTid) return true; 68 return threadTid === expectedTid; 69 } catch { 70 return false; 71 } 72 }); 73 if (!ready) { 74 return { 75 ok: false, 76 error: expectedTid 77 ? \`帖子数据未就绪或tid不匹配（expected=\${expectedTid}, actual=\${lastSeenTid || 'unknown'}, href=\${lastSeenHref || location.href}）\` 78 : '无法找到帖子数据' 79 }; 80 } 81 82 try { 83 const text = document.getElementById('__NEXT_DATA__')?.textContent || ''; 84 return { 85 ok: true, 86 data: JSON.parse(text) 87 }; 88 } catch (error) { 89 return { 90 ok: false, 91 error: error instanceof Error ? error.message : String(error) 92 }; 93 } 94 })() 95 `); 96 if (!result || typeof result !== 'object' || !result.ok) { 97 throw new CommandExecutionError(`${actionLabel} failed: ${result?.error || 'invalid browser response'}`); 98 } 99 return result.data; 100 } 101 export async function readHupuSearchData(page, url, actionLabel) { 102 await page.goto(url); 103 const result = await page.evaluate(` 104 (async () => { 105 const wait = (ms) => new Promise((resolve) => setTimeout(resolve, ms)); 106 const waitFor = async (predicate, timeoutMs = 5000) => { 107 const start = Date.now(); 108 while (Date.now() - start < timeoutMs) { 109 if (predicate()) return true; 110 await wait(100); 111 } 112 return false; 113 }; 114 115 const extractFromScript = () => { 116 const marker = 'window.$$data='; 117 for (const script of Array.from(document.scripts)) { 118 const text = script.textContent || ''; 119 const dataIndex = text.indexOf(marker); 120 if (dataIndex === -1) continue; 121 122 const jsonStart = dataIndex + marker.length; 123 let braceCount = 0; 124 let jsonEnd = jsonStart; 125 let inString = false; 126 let escapeNext = false; 127 128 for (let i = jsonStart; i < text.length; i++) { 129 const char = text[i]; 130 131 if (escapeNext) { 132 escapeNext = false; 133 continue; 134 } 135 136 if (char === '\\\\') { 137 escapeNext = true; 138 continue; 139 } 140 141 if (char === '"') { 142 inString = !inString; 143 continue; 144 } 145 146 if (!inString) { 147 if (char === '{') { 148 braceCount++; 149 } else if (char === '}') { 150 braceCount--; 151 if (braceCount === 0) { 152 jsonEnd = i; 153 break; 154 } 155 } 156 } 157 } 158 159 if (jsonEnd > jsonStart) { 160 return text.substring(jsonStart, jsonEnd + 1); 161 } 162 } 163 return ''; 164 }; 165 166 const ready = await waitFor(() => { 167 return typeof window.$$data !== 'undefined' || Boolean(extractFromScript()); 168 }); 169 if (!ready) { 170 return { ok: false, error: '无法找到搜索数据' }; 171 } 172 173 try { 174 if (typeof window.$$data !== 'undefined') { 175 return { 176 ok: true, 177 data: JSON.parse(JSON.stringify(window.$$data)) 178 }; 179 } 180 181 const jsonString = extractFromScript(); 182 return { 183 ok: true, 184 data: JSON.parse(jsonString) 185 }; 186 } catch (error) { 187 return { 188 ok: false, 189 error: error instanceof Error ? error.message : String(error) 190 }; 191 } 192 })() 193 `); 194 if (!result || typeof result !== 'object' || !result.ok) { 195 throw new CommandExecutionError(`${actionLabel} failed: ${result?.error || 'invalid browser response'}`); 196 } 197 return result.data; 198 } 199 function buildBrowserJsonPostScript(apiUrl, body, mode) { 200 return ` 201 (async () => { 202 const url = ${JSON.stringify(apiUrl)}; 203 const payload = ${JSON.stringify(body)}; 204 const mode = ${JSON.stringify(mode)}; 205 const getCookie = (name) => document.cookie 206 .split('; ') 207 .find((item) => item.startsWith(name + '=')) 208 ?.slice(name.length + 1) || ''; 209 210 const findThumbcacheValue = () => { 211 const rawEntry = document.cookie 212 .split('; ') 213 .find((item) => item.startsWith('.thumbcache_')); 214 if (rawEntry && rawEntry.includes('=')) { 215 const rawValue = rawEntry.slice(rawEntry.indexOf('=') + 1); 216 try { 217 return decodeURIComponent(rawValue); 218 } catch { 219 return rawValue; 220 } 221 } 222 223 const storageKey = Object.keys(localStorage).find((key) => key.startsWith('.thumbcache_')); 224 if (!storageKey) return ''; 225 return localStorage.getItem(storageKey) || ''; 226 }; 227 228 const resolveDefaultPayload = (input) => { 229 const next = { ...input }; 230 const sensorsRaw = decodeURIComponent(getCookie('sensorsdata2015jssdkcross') || ''); 231 let deviceid = ''; 232 try { 233 const sensors = JSON.parse(sensorsRaw); 234 deviceid = sensors?.props?.['$device_id'] || sensors?.distinct_id || ''; 235 } catch {} 236 237 if ((next.puid === '' || next.puid == null) && getCookie('ua')) { 238 next.puid = getCookie('ua'); 239 } 240 if ((next.shumei_id === '' || next.shumei_id == null) && getCookie('smidV2')) { 241 next.shumei_id = getCookie('smidV2'); 242 } 243 if ((next.deviceid === '' || next.deviceid == null) && deviceid) { 244 next.deviceid = deviceid; 245 } 246 return next; 247 }; 248 249 const resolveReplyPayload = (input) => { 250 const next = { ...input }; 251 const thumbcache = findThumbcacheValue(); 252 if ((next.shumeiId === '' || next.shumeiId == null) && thumbcache) { 253 next.shumeiId = thumbcache; 254 } 255 if ((next.deviceid === '' || next.deviceid == null) && thumbcache) { 256 next.deviceid = thumbcache; 257 } 258 return next; 259 }; 260 261 const resolvedPayload = mode === 'reply' 262 ? resolveReplyPayload(payload) 263 : resolveDefaultPayload(payload); 264 265 try { 266 const response = await fetch(url, { 267 method: 'POST', 268 credentials: 'include', 269 headers: { 270 'Content-Type': 'application/json' 271 }, 272 body: JSON.stringify(resolvedPayload) 273 }); 274 275 const text = await response.text(); 276 let data = null; 277 try { 278 data = text ? JSON.parse(text) : null; 279 } catch { 280 data = text ? { message: text } : null; 281 } 282 283 return { 284 ok: response.ok, 285 status: response.status, 286 data 287 }; 288 } catch (error) { 289 return { 290 ok: false, 291 error: error instanceof Error ? error.message : String(error) 292 }; 293 } 294 })() 295 `; 296 } 297 /** 298 * Execute authenticated Hupu JSON requests inside the browser page so 299 * cookies and the thread referer come from the live logged-in session. 300 */ 301 export async function postHupuJson(page, tid, apiUrl, body, actionLabel, mode = 'default') { 302 const referer = getHupuThreadUrl(tid); 303 await page.goto(referer); 304 const result = await page.evaluate(buildBrowserJsonPostScript(apiUrl, body, mode)); 305 if (!result || typeof result !== 'object') { 306 throw new CommandExecutionError(`${actionLabel} failed: invalid browser response`); 307 } 308 if (result.status === 401 || result.status === 403) { 309 throw new AuthRequiredError('bbs.hupu.com', `${actionLabel} failed: please log in to Hupu first`); 310 } 311 if (result.error) { 312 throw new CommandExecutionError(`${actionLabel} failed: ${result.error}`); 313 } 314 if (!result.ok) { 315 const detail = result.data?.msg || result.data?.message || `HTTP ${result.status ?? 'unknown'}`; 316 throw new CommandExecutionError(`${actionLabel} failed: ${detail}`); 317 } 318 return result.data ?? {}; 319 }