publishport-opencli 1.0.20 → 1.0.21

package/clis/zhihu/write-shared.js

@@ -1,9 +1,9 @@
-import{readFile as h,stat as A}from"node:fs/promises";import{CliError as Q}from"@jackwener/opencli/errors";const T=new Set(["status","outcome","message","target_type","target"]),$='header, nav, [role="banner"], [role="navigation"]',J='a[href^="/people/"]',W='img, [class*="Avatar"], [data-testid*="avatar" i], [aria-label*="头像"]',Y=["我","我的","个人主页"],z=[["self"],["current","user"],["account","profile"],["my","profile"],["my","account"]],C=JSON.stringify(z);function F(){return{readFile:h,stat:(j)=>A(j),decodeUtf8:(j)=>new TextDecoder("utf-8",{fatal:!0}).decode(j)}}function f(j){const q=j.toLowerCase();return Y.some((G)=>j.includes(G))||q.includes("my profile")||q.includes("my account")}function I(j){return j.toLowerCase().split(/[^a-z0-9]+/).filter(Boolean)}function _(j){const q=new Set(I(j));return z.some((G)=>G.every((B)=>q.has(B)))}function O(j){return typeof j==="object"&&j!==null&&"querySelectorAll"in j&&typeof j.querySelectorAll==="function"}function c(j){return typeof j==="object"&&j!==null&&"getAttribute"in j&&"querySelector"in j&&typeof j.getAttribute==="function"&&typeof j.querySelector==="function"}function L(j){const q=j?.topstory?.me?.slug||j?.me?.slug||j?.initialState?.me?.slug;return typeof q==="string"&&q?q:null}function x(j,q){const B=(j.getAttribute("href")||"").match(/^\/people\/([A-Za-z0-9_-]+)/);if(!B)return null;const H=j.getAttribute("aria-label")||"",X=j.getAttribute("title")||"",Z=j.getAttribute("data-testid")||"",M=j.getAttribute("class")||"",K=j.getAttribute("rel")||"",U=`${H} ${X} ${j.textContent||""}`,V=`${Z} ${M} ${K}`,b=Boolean(j.querySelector(W)),P=f(U),N=_(V);if(P||N)return B[1];if(q&&b)return B[1];return null}function D(j,q){for(const G of j)for(const B of Array.from(G.querySelectorAll(J)).filter(c)){const H=x(B,q);if(H)return H}return null}export function resolveCurrentUserSlugFromDom(j,q){const G=L(j);if(G)return G;const B=Array.from(q.querySelectorAll($)).filter(O);return D(B,!0)||D([q],!1)}export function requireExecute(j){if(!j.execute)throw new Q("INVALID_INPUT","此知乎写命令需要 --execute 才会真正执行。注意：本报错仅表示缺少 --execute，并未校验其余参数，不能当 dry-run 预览用")}export async function resolvePayload(j,q=F()){const G=typeof j.text==="string"?j.text:void 0,B=typeof j.file==="string"?j.file:void 0;if(G&&B)throw new Q("INVALID_INPUT","Use either <text> or --file, not both");let H=G??"";if(B){let X;try{X=await q.stat(B)}catch{throw new Q("INVALID_INPUT",`File not found: ${B}`)}if(!X.isFile())throw new Q("INVALID_INPUT",`File must be a readable text file: ${B}`);let Z;try{Z=await q.readFile(B)}catch{throw new Q("INVALID_INPUT",`File could not be read: ${B}`)}try{H=q.decodeUtf8(Z)}catch{throw new Q("INVALID_INPUT",`File could not be decoded as UTF-8 text: ${B}`)}}if(!H.trim())throw new Q("INVALID_INPUT","Payload cannot be empty or whitespace only");return H}function R(){return`(() => {
-    const selfLabelTokens = ${JSON.stringify(Y)};
-    const explicitIdentityMetaTokenGroups = ${C};
-    const navScopeSelector = ${JSON.stringify($)};
-    const profileLinkSelector = ${JSON.stringify(J)};
-    const avatarSelector = ${JSON.stringify(W)};
+import{readFile as b,stat as _}from"node:fs/promises";import{CliError as i}from"@jackwener/opencli/errors";const L=new Set(["status","outcome","message","target_type","target"]),s='header, nav, [role="banner"], [role="navigation"]',c='a[href^="/people/"]',u='img, [class*="Avatar"], [data-testid*="avatar" i], [aria-label*="头像"]',f=["我","我的","个人主页"],d=[["self"],["current","user"],["account","profile"],["my","profile"],["my","account"]],w=JSON.stringify(d);function T(){return{readFile:b,stat:(t)=>_(t),decodeUtf8:(t)=>new TextDecoder("utf-8",{fatal:!0}).decode(t)}}function E(t){const e=t.toLowerCase();return f.some((r)=>t.includes(r))||e.includes("my profile")||e.includes("my account")}function N(t){return t.toLowerCase().split(/[^a-z0-9]+/).filter(Boolean)}function x(t){const e=new Set(N(t));return d.some((r)=>r.every((o)=>e.has(o)))}function v(t){return typeof t==="object"&&t!==null&&"querySelectorAll"in t&&typeof t.querySelectorAll==="function"}function F(t){return typeof t==="object"&&t!==null&&"getAttribute"in t&&"querySelector"in t&&typeof t.getAttribute==="function"&&typeof t.querySelector==="function"}function O(t){const e=t?.topstory?.me?.slug||t?.me?.slug||t?.initialState?.me?.slug;return typeof e==="string"&&e?e:null}function k(t,e){const o=(t.getAttribute("href")||"").match(/^\/people\/([A-Za-z0-9_-]+)/);if(!o)return null;const n=t.getAttribute("aria-label")||"",l=t.getAttribute("title")||"",a=t.getAttribute("data-testid")||"",g=t.getAttribute("class")||"",m=t.getAttribute("rel")||"",S=`${n} ${l} ${t.textContent||""}`,I=`${a} ${g} ${m}`,p=Boolean(t.querySelector(u)),A=E(S),h=x(I);if(A||h)return o[1];if(e&&p)return o[1];return null}function y(t,e){for(const r of t)for(const o of Array.from(r.querySelectorAll(c)).filter(F)){const n=k(o,e);if(n)return n}return null}export function resolveCurrentUserSlugFromDom(t,e){const r=O(t);if(r)return r;const o=Array.from(e.querySelectorAll(s)).filter(v);return y(o,!0)||y([e],!1)}export function requireExecute(t){if(!t.execute)throw new i("INVALID_INPUT","此知乎写命令需要 --execute 才会真正执行。注意：本报错仅表示缺少 --execute，并未校验其余参数，不能当 dry-run 预览用")}export async function resolvePayload(t,e=T()){const r=typeof t.text==="string"?t.text:void 0,o=typeof t.file==="string"?t.file:void 0;if(r&&o)throw new i("INVALID_INPUT","Use either <text> or --file, not both");let n=r??"";if(o){let l;try{l=await e.stat(o)}catch{throw new i("INVALID_INPUT",`File not found: ${o}`)}if(!l.isFile())throw new i("INVALID_INPUT",`File must be a readable text file: ${o}`);let a;try{a=await e.readFile(o)}catch{throw new i("INVALID_INPUT",`File could not be read: ${o}`)}try{n=e.decodeUtf8(a)}catch{throw new i("INVALID_INPUT",`File could not be decoded as UTF-8 text: ${o}`)}}if(!n.trim())throw new i("INVALID_INPUT","Payload cannot be empty or whitespace only");return n}function C(){return`(() => {
+    const selfLabelTokens = ${JSON.stringify(f)};
+    const explicitIdentityMetaTokenGroups = ${w};
+    const navScopeSelector = ${JSON.stringify(s)};
+    const profileLinkSelector = ${JSON.stringify(c)};
+    const avatarSelector = ${JSON.stringify(u)};
 
     const hasExplicitIdentityLabel = (text) => {
       const normalized = String(text || '').toLowerCase();
@@ -68,4 +68,4 @@ import{readFile as h,stat as A}from"node:fs/promises";import{CliError as Q}from"
       if (altMatch) return { slug: altMatch[1] };
     }
     return null;
-  })()`}export async function resolveCurrentUserIdentity(j){let q=null;for(let G=0;G<30;G++){q=await j.evaluate(R());if(q?.slug)break;await j.wait({time:0.5})}if(!q?.slug)throw new Q("ACTION_NOT_AVAILABLE","Could not resolve the logged-in Zhihu user identity before write");return q.slug}export function buildResultRow(j,q,G,B,H={}){for(const X of Object.keys(H))if(T.has(X))throw new Q("INVALID_INPUT",`Result extra field cannot overwrite reserved key: ${X}`);return[{status:"success",outcome:B,message:j,target_type:q,target:G,...H}]}export const __test__={requireExecute,resolvePayload,resolveCurrentUserIdentity,resolveCurrentUserSlugFromDom,buildResultRow};
+  })()`}export async function resolveCurrentUserIdentity(t){let e=null;for(let r=0;r<30;r++){e=await t.evaluate(C());if(e?.slug)break;await t.wait({time:0.5})}if(!e?.slug)throw new i("ACTION_NOT_AVAILABLE","Could not resolve the logged-in Zhihu user identity before write");return e.slug}export function buildResultRow(t,e,r,o,n={}){for(const l of Object.keys(n))if(L.has(l))throw new i("INVALID_INPUT",`Result extra field cannot overwrite reserved key: ${l}`);return[{status:"success",outcome:o,message:t,target_type:e,target:r,...n}]}export const __test__={requireExecute,resolvePayload,resolveCurrentUserIdentity,resolveCurrentUserSlugFromDom,buildResultRow};

package/clis/zlibrary/utils.js

@@ -1,4 +1,4 @@
-import{ArgumentError as K}from"@jackwener/opencli/errors";const ZLIBRARY_DOMAIN="z-library.im",ZLIBRARY_ORIGIN=`https://${ZLIBRARY_DOMAIN}`,P=new Set([ZLIBRARY_DOMAIN,`www.${ZLIBRARY_DOMAIN}`]);export function normalizeZlibraryBookUrl(C){const F=String(C||"").trim();let J;try{J=new URL(F)}catch{throw new K("Z-Library book URL must be a valid http(s) URL",`Example: ${ZLIBRARY_ORIGIN}/book/...`)}if(!["http:","https:"].includes(J.protocol)||!P.has(J.hostname))throw new K(`Unsupported Z-Library URL host: ${J.hostname}`,`Pass a book URL under ${ZLIBRARY_DOMAIN}, for example ${ZLIBRARY_ORIGIN}/book/...`);return J.toString()}export function buildSearchUrl(C){const F=String(C||"").trim();if(!F)throw new K("zlibrary search query cannot be empty");return`${ZLIBRARY_ORIGIN}/s/${encodeURIComponent(F)}`}export async function extractBookTitle(C){const F=await C.evaluate(`
+import{ArgumentError as o}from"@jackwener/opencli/errors";const ZLIBRARY_DOMAIN="z-library.im",ZLIBRARY_ORIGIN=`https://${ZLIBRARY_DOMAIN}`,n=new Set([ZLIBRARY_DOMAIN,`www.${ZLIBRARY_DOMAIN}`]);export function normalizeZlibraryBookUrl(t){const e=String(t||"").trim();let r;try{r=new URL(e)}catch{throw new o("Z-Library book URL must be a valid http(s) URL",`Example: ${ZLIBRARY_ORIGIN}/book/...`)}if(!["http:","https:"].includes(r.protocol)||!n.has(r.hostname))throw new o(`Unsupported Z-Library URL host: ${r.hostname}`,`Pass a book URL under ${ZLIBRARY_DOMAIN}, for example ${ZLIBRARY_ORIGIN}/book/...`);return r.toString()}export function buildSearchUrl(t){const e=String(t||"").trim();if(!e)throw new o("zlibrary search query cannot be empty");return`${ZLIBRARY_ORIGIN}/s/${encodeURIComponent(e)}`}export async function extractBookTitle(t){const e=await t.evaluate(`
       (() => {
         const card = document.querySelector('z-bookcard');
         if (card && card.shadowRoot) {
@@ -7,14 +7,14 @@ import{ArgumentError as K}from"@jackwener/opencli/errors";const ZLIBRARY_DOMAIN=
         }
         return document.title.replace(/\\s*[-|].*$/, '').trim();
       })()
-    `);return String(F||"").trim()}export async function extractFormats(C){await C.evaluate(`
+    `);return String(e||"").trim()}export async function extractFormats(t){await t.evaluate(`
       (() => {
         const btn = document.querySelector(
           'button[aria-label*="more" i], [class*="dots" i], [class*="more" i]'
         );
         if (btn) btn.click();
       })()
-    `);await C.wait({time:3});const F=await C.evaluate(`
+    `);await t.wait({time:3});const e=await t.evaluate(`
       JSON.stringify((() => {
         const res = { pdf: '', epub: '' };
         document.querySelectorAll('a[href]').forEach(a => {
@@ -25,10 +25,10 @@ import{ArgumentError as K}from"@jackwener/opencli/errors";const ZLIBRARY_DOMAIN=
         });
         return res;
       })())
-    `);return JSON.parse(F)}export async function extractSearchResults(C,F){const J=await C.evaluate(`
+    `);return JSON.parse(e)}export async function extractSearchResults(t,e){const r=await t.evaluate(`
     JSON.stringify(
       Array.from(document.querySelectorAll('z-bookcard'))
-        .slice(0, ${F})
+        .slice(0, ${e})
         .map((card, index) => {
           const text = card.textContent.trim();
           const lines = text.split('\\n').map(l => l.trim()).filter(Boolean);
@@ -45,4 +45,4 @@ import{ArgumentError as K}from"@jackwener/opencli/errors";const ZLIBRARY_DOMAIN=
         })
         .filter(item => item.url && item.title)
     )
-  `);try{return JSON.parse(J)}catch{return[]}}export{ZLIBRARY_DOMAIN,ZLIBRARY_ORIGIN};
+  `);try{return JSON.parse(r)}catch{return[]}}export{ZLIBRARY_DOMAIN,ZLIBRARY_ORIGIN};

package/clis/zsxq/auth.js

@@ -1,4 +1,4 @@
-import{AuthRequiredError as D,CommandExecutionError as F}from"@jackwener/opencli/errors";import{registerSiteAuthCommands as H}from"../_shared/site-auth.js";async function G(B){await B.goto("https://wx.zsxq.com/");await B.wait(2);const z=await B.evaluate(`
+import{AuthRequiredError as s,CommandExecutionError as r}from"@jackwener/opencli/errors";import{registerSiteAuthCommands as n}from"../_shared/site-auth.js";async function i(t){await t.goto("https://wx.zsxq.com/");await t.wait(2);const e=await t.evaluate(`
     (async () => {
       const url = location.href;
       if (/\\/login(\\b|$)/.test(url)) {
@@ -23,11 +23,11 @@ import{AuthRequiredError as D,CommandExecutionError as F}from"@jackwener/opencli
         return { kind: 'exception', detail: String(e && e.message || e) };
       }
     })()
-  `);if(z?.kind==="auth")throw new D("zsxq.com",z.detail);if(z?.kind==="http")throw new F(`HTTP ${z.httpStatus} from zsxq /v2/users/self`);if(z?.kind==="exception")throw new F(`zsxq whoami failed: ${z.detail}`);if(!z?.ok)throw new F(`Unexpected zsxq probe: ${JSON.stringify(z)}`);if(!z.user_id)throw new D("zsxq.com","zsxq /v2/users/self 200 but user_id missing — incomplete session");return{user_id:z.user_id,name:z.name}}H({site:"zsxq",domain:"zsxq.com",loginUrl:"https://wx.zsxq.com/login",columns:["user_id","name"],verify:G,poll:async(B)=>{if(!await B.evaluate(`(async () => {
+  `);if(e?.kind==="auth")throw new s("zsxq.com",e.detail);if(e?.kind==="http")throw new r(`HTTP ${e.httpStatus} from zsxq /v2/users/self`);if(e?.kind==="exception")throw new r(`zsxq whoami failed: ${e.detail}`);if(!e?.ok)throw new r(`Unexpected zsxq probe: ${JSON.stringify(e)}`);if(!e.user_id)throw new s("zsxq.com","zsxq /v2/users/self 200 but user_id missing — incomplete session");return{user_id:e.user_id,name:e.name}}n({site:"zsxq",domain:"zsxq.com",loginUrl:"https://wx.zsxq.com/login",columns:["user_id","name"],verify:i,poll:async(t)=>{if(!await t.evaluate(`(async () => {
       try {
         const r = await fetch('https://api.zsxq.com/v2/users/self', { credentials: 'include', headers: { Accept: 'application/json' } });
         if (!r.ok) return false;
         const d = await r.json();
         return !!(d?.resp_data?.user);
       } catch { return false; }
-    })()`))throw new D("zsxq.com","Waiting for zsxq login");return G(B)}});
+    })()`))throw new s("zsxq.com","Waiting for zsxq login");return i(t)}});

package/clis/zsxq/search.js

@@ -1 +1 @@
-import{cli as V,Strategy as W}from"@jackwener/opencli/registry";import{getActiveGroupId as X,ensureZsxqAuth as Y,ensureZsxqPage as _,fetchFirstJson as L,getGroupsFromResponse as $,getTopicsFromResponse as D,toTopicRow as S}from"./utils.js";V({site:"zsxq",name:"search",access:"read",description:"搜索星球内容",domain:"wx.zsxq.com",strategy:W.COOKIE,browser:!0,args:[{name:"keyword",required:!0,positional:!0,help:"Search keyword"},{name:"limit",type:"int",default:20,help:"Number of results to return"},{name:"group_id",help:"Optional group id; defaults to the active group in Chrome"}],columns:["topic_id","group","author","title","comments","likes","time","url"],func:async(b,z)=>{await _(b);await Y(b);const M=String(z.keyword||"").trim(),E=Math.max(1,Number(z.limit)||20),B=String(z.group_id||await X(b)),O=encodeURIComponent(M);let H=B;try{const{data:C}=await L(b,["https://api.zsxq.com/v2/groups"]),K=$(C).find((U)=>String(U.group_id)===B);if(K?.name)H=K.name}catch{}const{data:Q}=await L(b,[`https://api.zsxq.com/v2/search/groups/${B}/topics?keyword=${O}&count=${E}`]);return D(Q).slice(0,E).map((C)=>({...S(C),group:H}))}});
+import{cli as d,Strategy as g}from"@jackwener/opencli/registry";import{getActiveGroupId as l,ensureZsxqAuth as h,ensureZsxqPage as w,fetchFirstJson as n,getGroupsFromResponse as y,getTopicsFromResponse as f,toTopicRow as q}from"./utils.js";d({site:"zsxq",name:"search",access:"read",description:"搜索星球内容",domain:"wx.zsxq.com",strategy:g.COOKIE,browser:!0,args:[{name:"keyword",required:!0,positional:!0,help:"Search keyword"},{name:"limit",type:"int",default:20,help:"Number of results to return"},{name:"group_id",help:"Optional group id; defaults to the active group in Chrome"}],columns:["topic_id","group","author","title","comments","likes","time","url"],func:async(t,o)=>{await w(t);await h(t);const u=String(o.keyword||"").trim(),i=Math.max(1,Number(o.limit)||20),e=String(o.group_id||await l(t)),c=encodeURIComponent(u);let s=e;try{const{data:r}=await n(t,["https://api.zsxq.com/v2/groups"]),a=y(r).find((m)=>String(m.group_id)===e);if(a?.name)s=a.name}catch{}const{data:p}=await n(t,[`https://api.zsxq.com/v2/search/groups/${e}/topics?keyword=${c}&count=${i}`]);return f(p).slice(0,i).map((r)=>({...q(r),group:s}))}});

package/clis/zsxq/topic.js

@@ -1 +1 @@
-import{cli as u,Strategy as d}from"@jackwener/opencli/registry";import{CliError as s}from"@jackwener/opencli/errors";import{getActiveGroupId as l,browserJsonRequest as h,ensureZsxqAuth as w,ensureZsxqPage as f,fetchFirstJson as _,getCommentsFromResponse as g,getTopicFromResponse as R,getTopicUrl as T,summarizeComments as $,toTopicRow as q}from"./utils.js";u({site:"zsxq",name:"topic",access:"read",description:"获取单个话题详情和评论",domain:"wx.zsxq.com",strategy:d.COOKIE,browser:!0,args:[{name:"id",required:!0,positional:!0,help:"Topic ID"},{name:"group_id",help:"Group ID (optional; defaults to active group in Chrome)"},{name:"comment_limit",type:"int",default:20,help:"Number of comments to fetch"}],columns:["topic_id","type","author","title","comments","likes","comment_preview","url"],func:async(o,n)=>{await f(o);await w(o);const t=String(n.id),c=String(n.group_id||await l(o)),p=Math.max(1,Number(n.comment_limit)||20),m=`https://api.zsxq.com/v2/groups/${c}/topics/${t}`,e=await h(o,m);if(e.status===404)throw new s("NOT_FOUND",`Topic ${t} not found`);if(!e.ok)throw new s("FETCH_ERROR",e.error||`Failed to fetch topic ${t}`,`Checked endpoint: ${m}`);const a=await _(o,[`https://api.zsxq.com/v2/groups/${c}/topics/${t}/comments?sort=asc&count=${p}`]),i=R(e.data);if(!i)throw new s("NOT_FOUND",`Topic ${t} not found`);const r=g(a.data);return[{...q({...i,comments:r,comments_count:i.comments_count??r.length}),comment_preview:$(r,5),url:T(i.topic_id??t)}]}});
+import{cli as u,Strategy as d}from"@jackwener/opencli/registry";import{CliError as s}from"@jackwener/opencli/errors";import{getActiveGroupId as l,browserJsonRequest as h,ensureZsxqAuth as w,ensureZsxqPage as f,fetchFirstJson as _,getCommentsFromResponse as g,getTopicFromResponse as R,getTopicUrl as T,summarizeComments as q,toTopicRow as x}from"./utils.js";u({site:"zsxq",name:"topic",access:"read",description:"获取单个话题详情和评论",domain:"wx.zsxq.com",strategy:d.COOKIE,browser:!0,args:[{name:"id",required:!0,positional:!0,help:"Topic ID"},{name:"group_id",help:"Group ID (optional; defaults to active group in Chrome)"},{name:"comment_limit",type:"int",default:20,help:"Number of comments to fetch"}],columns:["topic_id","type","author","title","comments","likes","comment_preview","url"],func:async(o,n)=>{await f(o);await w(o);const t=String(n.id),c=String(n.group_id||await l(o)),p=Math.max(1,Number(n.comment_limit)||20),m=`https://api.zsxq.com/v2/groups/${c}/topics/${t}`,e=await h(o,m);if(e.status===404)throw new s("NOT_FOUND",`Topic ${t} not found`);if(!e.ok)throw new s("FETCH_ERROR",e.error||`Failed to fetch topic ${t}`,`Checked endpoint: ${m}`);const a=await _(o,[`https://api.zsxq.com/v2/groups/${c}/topics/${t}/comments?sort=asc&count=${p}`]),i=R(e.data);if(!i)throw new s("NOT_FOUND",`Topic ${t} not found`);const r=g(a.data);return[{...x({...i,comments:r,comments_count:i.comments_count??r.length}),comment_preview:q(r,5),url:T(i.topic_id??t)}]}});

package/clis/zsxq/utils.js

@@ -1,4 +1,4 @@
-import{ArgumentError as z,AuthRequiredError as Y,CliError as W}from"@jackwener/opencli/errors";const J="wx.zsxq.com",Z="https://wx.zsxq.com";function $(B){return B&&typeof B==="object"&&!Array.isArray(B)?B:null}function X(...B){for(const G of B)if(Array.isArray(G))return G;return[]}export async function ensureZsxqPage(B){await B.goto(Z)}export async function ensureZsxqAuth(B){try{if(!await B.evaluate(`
+import{ArgumentError as l,AuthRequiredError as i,CliError as s}from"@jackwener/opencli/errors";const f="wx.zsxq.com",u="https://wx.zsxq.com";function c(r){return r&&typeof r==="object"&&!Array.isArray(r)?r:null}function a(...r){for(const e of r)if(Array.isArray(e))return e;return[]}export async function ensureZsxqPage(r){await r.goto(u)}export async function ensureZsxqAuth(r){try{if(!await r.evaluate(`
       (async () => {
         try {
           const r = await new Promise((resolve, reject) => {
@@ -18,7 +18,7 @@ import{ArgumentError as z,AuthRequiredError as Y,CliError as W}from"@jackwener/o
           return r !== null;
         } catch { return false; }
       })()
-    `))throw new Y("zsxq.com")}catch(G){if(G instanceof Y)throw G;throw new Y("zsxq.com")}}export async function getCookieValue(B,G){return(await B.getCookies({domain:J})).find((K)=>K.name===G)?.value}export async function getActiveGroupId(B){const G=await B.evaluate(`
+    `))throw new i("zsxq.com")}catch(e){if(e instanceof i)throw e;throw new i("zsxq.com")}}export async function getCookieValue(r,e){return(await r.getCookies({domain:f})).find((n)=>n.name===e)?.value}export async function getActiveGroupId(r){const e=await r.evaluate(`
     (() => {
       const target = localStorage.getItem('target_group');
       if (target) {
@@ -29,9 +29,9 @@ import{ArgumentError as z,AuthRequiredError as Y,CliError as W}from"@jackwener/o
       }
       return null;
     })()
-  `);if(G)return G;throw new z("Cannot determine active group_id","Pass --group_id <id> or open the target 知识星球 page in Chrome first")}export async function browserJsonRequest(B,G){return await B.evaluate(`
+  `);if(e)return e;throw new l("Cannot determine active group_id","Pass --group_id <id> or open the target 知识星球 page in Chrome first")}export async function browserJsonRequest(r,e){return await r.evaluate(`
     (async () => {
-      const path = ${JSON.stringify(G)};
+      const path = ${JSON.stringify(e)};
 
       try {
         return await new Promise((resolve) => {
@@ -69,4 +69,4 @@ import{ArgumentError as z,AuthRequiredError as Y,CliError as W}from"@jackwener/o
         };
       }
     })()
-  `)}export async function fetchFirstJson(B,G){let H=null;for(const K of G){const Q=await browserJsonRequest(B,K);if(Q.ok)return Q;H=Q}if(!H)throw new W("FETCH_ERROR","No candidate endpoint returned JSON",`Checked endpoints: ${G.join(", ")}`);throw new W("FETCH_ERROR",H.error||"Failed to fetch ZSXQ API",`Checked endpoints: ${G.join(", ")}`)}export function unwrapRespData(B){const G=$(B);if(!G)throw new W("PARSE_ERROR","Invalid ZSXQ API response");if(G.succeeded===!1){const H=typeof G.code==="number"?String(G.code):"API_ERROR",K=typeof G.info==="string"?G.info:typeof G.error==="string"?G.error:"ZSXQ API returned an error";throw new W(H,K)}return G.resp_data??G.data??B}export function getTopicsFromResponse(B){const G=unwrapRespData(B);if(Array.isArray(G))return G;return X(G.topics,G.list,G.records,G.items,G.search_result)}export function getCommentsFromResponse(B){const G=unwrapRespData(B);if(Array.isArray(G))return G;return X(G.comments,G.list,G.items)}export function getGroupsFromResponse(B){const G=unwrapRespData(B);if(Array.isArray(G))return G;return X(G.groups,G.list,G.items)}export function getTopicFromResponse(B){const G=unwrapRespData(B);if(Array.isArray(G))return G[0]??null;if(typeof G.topic_id==="number"||typeof G.topic_id==="string")return G;const H=$(G);if(!H)return null;const K=H.topic;return K&&typeof K==="object"?K:null}export function getTopicAuthor(B){return B.owner?.name||B.talk?.owner?.name||B.question?.owner?.name||B.answer?.owner?.name||B.task?.owner?.name||B.solution?.owner?.name||""}export function getTopicText(B){return(B.title||"").replace(/\s+/g," ").trim()||getTopicContent(B)}export function getTopicContent(B){return([B.talk?.text,B.question?.text,B.answer?.text,B.task?.text,B.solution?.text].find((H)=>typeof H==="string"&&H.trim())||"").replace(/\s+/g," ").trim()}export function getTopicUrl(B){return B?`${Z}/topic/${B}`:Z}export function summarizeComments(B,G=3){return B.slice(0,G).map((H)=>{const K=H.owner?.name||"匿名",Q=H.repliee?.name?` -> ${H.repliee.name}`:"",j=(H.text||"").replace(/\s+/g," ").trim();return`${K}${Q}: ${j}`}).join(" | ")}export function toTopicRow(B){const G=B.topic_id??"",H=X(B.show_comments,B.comments);return{topic_id:G,type:B.type||"",group:B.group?.name||"",author:getTopicAuthor(B),title:getTopicText(B),content:getTopicContent(B),comments:B.comments_count??H.length??0,likes:B.likes_count??0,readers:B.readers_count??B.reading_count??0,time:B.create_time||"",comment_preview:summarizeComments(H),url:getTopicUrl(G)}}
+  `)}export async function fetchFirstJson(r,e){let t=null;for(const n of e){const o=await browserJsonRequest(r,n);if(o.ok)return o;t=o}if(!t)throw new s("FETCH_ERROR","No candidate endpoint returned JSON",`Checked endpoints: ${e.join(", ")}`);throw new s("FETCH_ERROR",t.error||"Failed to fetch ZSXQ API",`Checked endpoints: ${e.join(", ")}`)}export function unwrapRespData(r){const e=c(r);if(!e)throw new s("PARSE_ERROR","Invalid ZSXQ API response");if(e.succeeded===!1){const t=typeof e.code==="number"?String(e.code):"API_ERROR",n=typeof e.info==="string"?e.info:typeof e.error==="string"?e.error:"ZSXQ API returned an error";throw new s(t,n)}return e.resp_data??e.data??r}export function getTopicsFromResponse(r){const e=unwrapRespData(r);if(Array.isArray(e))return e;return a(e.topics,e.list,e.records,e.items,e.search_result)}export function getCommentsFromResponse(r){const e=unwrapRespData(r);if(Array.isArray(e))return e;return a(e.comments,e.list,e.items)}export function getGroupsFromResponse(r){const e=unwrapRespData(r);if(Array.isArray(e))return e;return a(e.groups,e.list,e.items)}export function getTopicFromResponse(r){const e=unwrapRespData(r);if(Array.isArray(e))return e[0]??null;if(typeof e.topic_id==="number"||typeof e.topic_id==="string")return e;const t=c(e);if(!t)return null;const n=t.topic;return n&&typeof n==="object"?n:null}export function getTopicAuthor(r){return r.owner?.name||r.talk?.owner?.name||r.question?.owner?.name||r.answer?.owner?.name||r.task?.owner?.name||r.solution?.owner?.name||""}export function getTopicText(r){return(r.title||"").replace(/\s+/g," ").trim()||getTopicContent(r)}export function getTopicContent(r){return([r.talk?.text,r.question?.text,r.answer?.text,r.task?.text,r.solution?.text].find((t)=>typeof t==="string"&&t.trim())||"").replace(/\s+/g," ").trim()}export function getTopicUrl(r){return r?`${u}/topic/${r}`:u}export function summarizeComments(r,e=3){return r.slice(0,e).map((t)=>{const n=t.owner?.name||"匿名",o=t.repliee?.name?` -> ${t.repliee.name}`:"",p=(t.text||"").replace(/\s+/g," ").trim();return`${n}${o}: ${p}`}).join(" | ")}export function toTopicRow(r){const e=r.topic_id??"",t=a(r.show_comments,r.comments);return{topic_id:e,type:r.type||"",group:r.group?.name||"",author:getTopicAuthor(r),title:getTopicText(r),content:getTopicContent(r),comments:r.comments_count??t.length??0,likes:r.likes_count??0,readers:r.readers_count??r.reading_count??0,time:r.create_time||"",comment_preview:summarizeComments(t),url:getTopicUrl(e)}}

package/dist/src/adapter-source.js

@@ -1 +1 @@
-import*as w from"node:fs";export function resolveAdapterSourcePath(g){const j=[];if(g.source&&!g.source.startsWith("manifest:"))j.push(g.source);if(g._modulePath)j.push(g._modulePath);for(const q of j)if(w.existsSync(q))return q;return j[0]}
+import*as o from"node:fs";export function resolveAdapterSourcePath(t){const e=[];if(t.source&&!t.source.startsWith("manifest:"))e.push(t.source);if(t._modulePath)e.push(t._modulePath);for(const s of e)if(o.existsSync(s))return s;return e[0]}

package/dist/src/browser/analyze.js

@@ -1 +1 @@
-const H=[{vendor:"aliyun_waf",cookiePatterns:[/^acw_sc__v2$/,/^acw_tc$/,/^ssxmod_itna/],bodyPatterns:[/arg1\s*=\s*['"][0-9A-F]{30,}/,/\/ntc_captcha\//i],implication:"Direct Node-side fetch/curl will return the slider HTML. Validate the endpoint in browser context first; HTML COOKIE adapters still finish with Node-side fetch + page.getCookies."},{vendor:"cloudflare",cookiePatterns:[/^__cf_bm$/,/^cf_clearance$/,/^__cfduid$/],bodyPatterns:[/Cloudflare Ray ID/i,/Checking your browser before accessing/i,/cf-chl-/i],implication:"Cloudflare bot check. Start from a real browser session; probe in browser context first. HTML COOKIE adapters still finish with Node-side fetch + page.getCookies."},{vendor:"akamai",cookiePatterns:[/^_abck$/,/^bm_sz$/,/^bm_sv$/],bodyPatterns:[/akamai/i],implication:"Akamai Bot Manager. Probe in browser context first; keep final HTML COOKIE adapters on Node-side fetch + page.getCookies."},{vendor:"geetest",cookiePatterns:[],bodyPatterns:[/geetest/i,/gt_captcha/i],implication:"Geetest slider/puzzle captcha. Agent cannot bypass programmatically — requires UI strategy or human-in-loop."}];export function detectAntiBot(q){const M=[];let Q=null;for(const V of H){const Z=[];for(const $ of V.cookiePatterns){const X=q.cookieNames.find((O)=>$.test(O));if(X)Z.push(`cookie:${X}`)}for(const $ of V.bodyPatterns)for(const X of q.networkEntries)if(X.bodyPreview&&$.test(X.bodyPreview)){Z.push(`body:${X.url}`);break}if(Z.length>0&&!Q){Q=V;M.push(...Z)}}if(!Q)return{detected:!1,vendor:null,evidence:[],implication:"No known anti-bot signatures. Try Node-side COOKIE fetch first; if endpoint validation is blocked, retry from browser context."};return{detected:!0,vendor:Q.vendor,evidence:M,implication:Q.implication}}const I=/(?:analytics|beacon|collect|telemetry|tracking|sentry|doubleclick|google-analytics|googletagmanager|adservice|\/ads?(?:[/?#]|$)|metrics?|pixel|personalization|experiment|\/events?(?:[/?#]|$))/i,U=/^(?:data|items?|results?|records?|list|rows?|edges?|nodes?|timeline|users?|title|name|text|content|body|price|amount|id|url|avatar|nickname|desc|comments?|likes?|shares?|total|page|cursor|next|rank|score|date|time|author)$/i,B=/^(?:event|events|trace|traceid|sessionid|clientid|visitorid|experiment|abtest|beacon|analytics|metrics?|pixel|log|logs)$/i;function P(q){return Math.max(0,Math.min(1,Math.round(q*100)/100))}function R(q){if(!q)return null;const M=q.trim();if(!M)return null;try{return JSON.parse(M)}catch{return M}}function Y(q,M="$",Q=[],V=0){if(V>4||Q.length>=24)return Q;if(Array.isArray(q)){Q.push(`${M}:array(${q.length})`);if(q.length>0)Y(q[0],`${M}[0]`,Q,V+1);return Q}if(!q||typeof q!=="object"){Q.push(`${M}:${typeof q}`);return Q}const Z=Object.entries(q).slice(0,12);for(const[$,X]of Z){Q.push(`${M}.${$}`);if(Array.isArray(X))Q.push(`${M}.${$}:array(${X.length})`);else if(X&&typeof X==="object")Y(X,`${M}.${$}`,Q,V+1);else Q.push(`${M}.${$}:${typeof X}`)}return Q}function G(q,M,Q=0){if(Q>4||!q||typeof q!=="object")return 0;if(Array.isArray(q))return q.slice(0,3).reduce((V,Z)=>V+G(Z,M,Q+1),0);return Object.entries(q).reduce((V,[Z,$])=>V+(M(Z)?1:0)+G($,M,Q+1),0)}function T(q,M=0){if(M>4||!q||typeof q!=="object")return!1;if(Array.isArray(q))return q.length>0;return Object.values(q).some((Q)=>T(Q,M+1))}export function scoreEndpointEvidence(q){const M=[],Q=R(q.bodyPreview);let V=0;if(q.status>=200&&q.status<300){V+=0.15;M.push("2xx status")}else if(q.status===401||q.status===403)return{url:q.url,status:q.status,contentType:q.contentType,real_data_score:0.05,verdict:"blocked",reasons:["auth-blocked status"],sample_paths:[]};else M.push(`non-2xx status ${q.status}`);if(/json/i.test(q.contentType)){V+=0.2;M.push("json content-type")}else if(/html/i.test(q.contentType)){V-=0.25;M.push("html content-type")}else if(/javascript|text/i.test(q.contentType)){V+=0.05;M.push("text/script content-type")}if(I.test(q.url)){V-=0.3;M.push("telemetry-like url")}const Z=Y(Q).slice(0,8);if(typeof Q==="string"){if(/^\s*</.test(Q)||/<!doctype html|<html/i.test(Q)){V-=0.25;M.push("html body")}else if(Q.trim().length>20){V+=0.05;M.push("non-empty text body")}}else if(Array.isArray(Q))if(Q.length>0){V+=0.3;M.push("non-empty top-level array")}else{V-=0.15;M.push("empty array")}else if(Q&&typeof Q==="object"){if(Object.keys(Q).length===0){V-=0.15;M.push("empty object")}else{V+=0.12;M.push("json object body")}const L=G(Q,(W)=>U.test(W));if(L>0){V+=Math.min(0.3,L*0.05);M.push(`${L} business-like key${L===1?"":"s"}`)}if(T(Q)){V+=0.2;M.push("nested non-empty array")}const z=G(Q,(W)=>B.test(W));if(z>0&&L===0){V-=Math.min(0.25,z*0.08);M.push(`${z} tracking-like key${z===1?"":"s"} without business keys`)}}const $=P(V),X=$>=0.65?"likely_data":$>=0.35?"maybe_data":"noise";return{url:q.url,status:q.status,contentType:q.contentType,real_data_score:$,verdict:X,reasons:M,sample_paths:Z}}export function scoreNetworkEvidence(q){return q.networkEntries.map(scoreEndpointEvidence).filter((M)=>M.verdict!=="noise"||M.real_data_score>0).sort((M,Q)=>Q.real_data_score-M.real_data_score).slice(0,8)}export function classifyPattern(q){const M=q.networkEntries.filter((X)=>/json/i.test(X.contentType)),V=scoreNetworkEvidence(q).filter((X)=>X.verdict==="likely_data"||X.verdict==="maybe_data").length,Z=q.networkEntries.filter((X)=>X.status===401||X.status===403).length,$=q.initialState.__INITIAL_STATE__||q.initialState.__NUXT__||q.initialState.__NEXT_DATA__||q.initialState.__APOLLO_STATE__;if(Z>=2&&M.length>=1)return{pattern:"D",reason:`${Z} auth-failing API responses seen — endpoint is token-gated`,json_responses:M.length,real_data_candidates:V,auth_failures:Z};if($)return{pattern:"B",reason:`SSR state global present: ${Object.entries(q.initialState).filter(([,O])=>O).map(([O])=>O).join(", ")}`,json_responses:M.length,real_data_candidates:V,auth_failures:Z};if(V>=1)return{pattern:"A",reason:`${V} captured response${V===1?"":"s"} look like real data — inspect api_candidates before choosing a strategy`,json_responses:M.length,real_data_candidates:V,auth_failures:Z};return{pattern:"C",reason:M.length>0?`${M.length} JSON response${M.length===1?"":"s"} observed, but none look like target data — likely telemetry/side-channel; treat as HTML/DOM until an endpoint validates`:"No JSON XHR and no SSR state — HTML scrape (Pattern C); escalate to E manually if WebSocket traffic appears",json_responses:M.length,real_data_candidates:V,auth_failures:Z}}export function findNearestAdapter(q,M){let Q;try{Q=new URL(q).hostname}catch{return null}const V=Q.replace(/^www\./,""),Z=V.split("."),$=Z.slice(-2).join("."),X=Z.length>1?Z[Z.length-2]:V,O=new Map;for(const z of M.values()){const W=z.domain?.toLowerCase();if(W&&(V.endsWith(W)||W.endsWith($))||z.site.toLowerCase()===X?.toLowerCase()||V.includes(z.site.toLowerCase())){const J=O.get(z.site)??[];J.push(z);O.set(z.site,J)}}if(O.size===0)return null;let L=null;for(const z of O)if(!L||z[1].length>L[1].length)L=z;if(!L)return null;return{site:L[0],example_commands:L[1].slice(0,5).map((z)=>`${z.site} ${z.name}`),reason:`${L[1].length} existing adapter${L[1].length===1?"":"s"} target this site — reuse strategy/cookie config`}}export function analyzeSite(q,M){const Q=classifyPattern(q),V=detectAntiBot(q),Z=scoreNetworkEvidence(q),$=findNearestAdapter(q.finalUrl,M);let X;if(V.detected)X=V.implication;else if(Q.pattern==="A")X="Inspect `api_candidates`, then replay the best endpoint and record the status/content-type/sample shape in your strategy note; do not choose API strategy from XHR count alone.";else if(Q.pattern==="B")X='Read the SSR global via `ppcli browser eval "JSON.stringify(window.__INITIAL_STATE__ ?? window.__NUXT__ ?? window.__NEXT_DATA__ ?? window.__APOLLO_STATE__)"` — no API needed.';else if(Q.pattern==="C")X="No API visible — use SSR HTML scrape (e.g. `ppcli browser extract`) against the rendered page.";else if(Q.pattern==="D")X="Endpoints need auth. Re-open the page from a signed-in session, then retry analyze; see `field-decode-playbook` §4 for token tracing.";else if(Q.pattern==="E")X="WebSocket stream detected — find the underlying HTTP poll/long-poll endpoint; raw WS is not supported.";else X="No strong signal. Manually inspect `ppcli browser network --all` and pick a pattern.";return{requested_url:q.requestedUrl,final_url:q.finalUrl,title:q.title,pattern:Q,anti_bot:V,initial_state:q.initialState,api_candidates:Z,nearest_adapter:$,recommended_next_step:X}}
+const _=[{vendor:"aliyun_waf",cookiePatterns:[/^acw_sc__v2$/,/^acw_tc$/,/^ssxmod_itna/],bodyPatterns:[/arg1\s*=\s*['"][0-9A-F]{30,}/,/\/ntc_captcha\//i],implication:"Direct Node-side fetch/curl will return the slider HTML. Validate the endpoint in browser context first; HTML COOKIE adapters still finish with Node-side fetch + page.getCookies."},{vendor:"cloudflare",cookiePatterns:[/^__cf_bm$/,/^cf_clearance$/,/^__cfduid$/],bodyPatterns:[/Cloudflare Ray ID/i,/Checking your browser before accessing/i,/cf-chl-/i],implication:"Cloudflare bot check. Start from a real browser session; probe in browser context first. HTML COOKIE adapters still finish with Node-side fetch + page.getCookies."},{vendor:"akamai",cookiePatterns:[/^_abck$/,/^bm_sz$/,/^bm_sv$/],bodyPatterns:[/akamai/i],implication:"Akamai Bot Manager. Probe in browser context first; keep final HTML COOKIE adapters on Node-side fetch + page.getCookies."},{vendor:"geetest",cookiePatterns:[],bodyPatterns:[/geetest/i,/gt_captcha/i],implication:"Geetest slider/puzzle captcha. Agent cannot bypass programmatically — requires UI strategy or human-in-loop."}];export function detectAntiBot(e){const t=[];let n=null;for(const s of _){const i=[];for(const r of s.cookiePatterns){const a=e.cookieNames.find((l)=>r.test(l));if(a)i.push(`cookie:${a}`)}for(const r of s.bodyPatterns)for(const a of e.networkEntries)if(a.bodyPreview&&r.test(a.bodyPreview)){i.push(`body:${a.url}`);break}if(i.length>0&&!n){n=s;t.push(...i)}}if(!n)return{detected:!1,vendor:null,evidence:[],implication:"No known anti-bot signatures. Try Node-side COOKIE fetch first; if endpoint validation is blocked, retry from browser context."};return{detected:!0,vendor:n.vendor,evidence:t,implication:n.implication}}const y=/(?:analytics|beacon|collect|telemetry|tracking|sentry|doubleclick|google-analytics|googletagmanager|adservice|\/ads?(?:[/?#]|$)|metrics?|pixel|personalization|experiment|\/events?(?:[/?#]|$))/i,m=/^(?:data|items?|results?|records?|list|rows?|edges?|nodes?|timeline|users?|title|name|text|content|body|price|amount|id|url|avatar|nickname|desc|comments?|likes?|shares?|total|page|cursor|next|rank|score|date|time|author)$/i,b=/^(?:event|events|trace|traceid|sessionid|clientid|visitorid|experiment|abtest|beacon|analytics|metrics?|pixel|log|logs)$/i;function g(e){return Math.max(0,Math.min(1,Math.round(e*100)/100))}function k(e){if(!e)return null;const t=e.trim();if(!t)return null;try{return JSON.parse(t)}catch{return t}}function u(e,t="$",n=[],s=0){if(s>4||n.length>=24)return n;if(Array.isArray(e)){n.push(`${t}:array(${e.length})`);if(e.length>0)u(e[0],`${t}[0]`,n,s+1);return n}if(!e||typeof e!=="object"){n.push(`${t}:${typeof e}`);return n}const i=Object.entries(e).slice(0,12);for(const[r,a]of i){n.push(`${t}.${r}`);if(Array.isArray(a))n.push(`${t}.${r}:array(${a.length})`);else if(a&&typeof a==="object")u(a,`${t}.${r}`,n,s+1);else n.push(`${t}.${r}:${typeof a}`)}return n}function p(e,t,n=0){if(n>4||!e||typeof e!=="object")return 0;if(Array.isArray(e))return e.slice(0,3).reduce((s,i)=>s+p(i,t,n+1),0);return Object.entries(e).reduce((s,[i,r])=>s+(t(i)?1:0)+p(r,t,n+1),0)}function h(e,t=0){if(t>4||!e||typeof e!=="object")return!1;if(Array.isArray(e))return e.length>0;return Object.values(e).some((n)=>h(n,t+1))}export function scoreEndpointEvidence(e){const t=[],n=k(e.bodyPreview);let s=0;if(e.status>=200&&e.status<300){s+=0.15;t.push("2xx status")}else if(e.status===401||e.status===403)return{url:e.url,status:e.status,contentType:e.contentType,real_data_score:0.05,verdict:"blocked",reasons:["auth-blocked status"],sample_paths:[]};else t.push(`non-2xx status ${e.status}`);if(/json/i.test(e.contentType)){s+=0.2;t.push("json content-type")}else if(/html/i.test(e.contentType)){s-=0.25;t.push("html content-type")}else if(/javascript|text/i.test(e.contentType)){s+=0.05;t.push("text/script content-type")}if(y.test(e.url)){s-=0.3;t.push("telemetry-like url")}const i=u(n).slice(0,8);if(typeof n==="string"){if(/^\s*</.test(n)||/<!doctype html|<html/i.test(n)){s-=0.25;t.push("html body")}else if(n.trim().length>20){s+=0.05;t.push("non-empty text body")}}else if(Array.isArray(n))if(n.length>0){s+=0.3;t.push("non-empty top-level array")}else{s-=0.15;t.push("empty array")}else if(n&&typeof n==="object"){if(Object.keys(n).length===0){s-=0.15;t.push("empty object")}else{s+=0.12;t.push("json object body")}const c=p(n,(d)=>m.test(d));if(c>0){s+=Math.min(0.3,c*0.05);t.push(`${c} business-like key${c===1?"":"s"}`)}if(h(n)){s+=0.2;t.push("nested non-empty array")}const o=p(n,(d)=>b.test(d));if(o>0&&c===0){s-=Math.min(0.25,o*0.08);t.push(`${o} tracking-like key${o===1?"":"s"} without business keys`)}}const r=g(s),a=r>=0.65?"likely_data":r>=0.35?"maybe_data":"noise";return{url:e.url,status:e.status,contentType:e.contentType,real_data_score:r,verdict:a,reasons:t,sample_paths:i}}export function scoreNetworkEvidence(e){return e.networkEntries.map(scoreEndpointEvidence).filter((t)=>t.verdict!=="noise"||t.real_data_score>0).sort((t,n)=>n.real_data_score-t.real_data_score).slice(0,8)}export function classifyPattern(e){const t=e.networkEntries.filter((a)=>/json/i.test(a.contentType)),s=scoreNetworkEvidence(e).filter((a)=>a.verdict==="likely_data"||a.verdict==="maybe_data").length,i=e.networkEntries.filter((a)=>a.status===401||a.status===403).length,r=e.initialState.__INITIAL_STATE__||e.initialState.__NUXT__||e.initialState.__NEXT_DATA__||e.initialState.__APOLLO_STATE__;if(i>=2&&t.length>=1)return{pattern:"D",reason:`${i} auth-failing API responses seen — endpoint is token-gated`,json_responses:t.length,real_data_candidates:s,auth_failures:i};if(r)return{pattern:"B",reason:`SSR state global present: ${Object.entries(e.initialState).filter(([,l])=>l).map(([l])=>l).join(", ")}`,json_responses:t.length,real_data_candidates:s,auth_failures:i};if(s>=1)return{pattern:"A",reason:`${s} captured response${s===1?"":"s"} look like real data — inspect api_candidates before choosing a strategy`,json_responses:t.length,real_data_candidates:s,auth_failures:i};return{pattern:"C",reason:t.length>0?`${t.length} JSON response${t.length===1?"":"s"} observed, but none look like target data — likely telemetry/side-channel; treat as HTML/DOM until an endpoint validates`:"No JSON XHR and no SSR state — HTML scrape (Pattern C); escalate to E manually if WebSocket traffic appears",json_responses:t.length,real_data_candidates:s,auth_failures:i}}export function findNearestAdapter(e,t){let n;try{n=new URL(e).hostname}catch{return null}const s=n.replace(/^www\./,""),i=s.split("."),r=i.slice(-2).join("."),a=i.length>1?i[i.length-2]:s,l=new Map;for(const o of t.values()){const d=o.domain?.toLowerCase();if(d&&(s.endsWith(d)||d.endsWith(r))||o.site.toLowerCase()===a?.toLowerCase()||s.includes(o.site.toLowerCase())){const f=l.get(o.site)??[];f.push(o);l.set(o.site,f)}}if(l.size===0)return null;let c=null;for(const o of l)if(!c||o[1].length>c[1].length)c=o;if(!c)return null;return{site:c[0],example_commands:c[1].slice(0,5).map((o)=>`${o.site} ${o.name}`),reason:`${c[1].length} existing adapter${c[1].length===1?"":"s"} target this site — reuse strategy/cookie config`}}export function analyzeSite(e,t){const n=classifyPattern(e),s=detectAntiBot(e),i=scoreNetworkEvidence(e),r=findNearestAdapter(e.finalUrl,t);let a;if(s.detected)a=s.implication;else if(n.pattern==="A")a="Inspect `api_candidates`, then replay the best endpoint and record the status/content-type/sample shape in your strategy note; do not choose API strategy from XHR count alone.";else if(n.pattern==="B")a='Read the SSR global via `ppcli browser eval "JSON.stringify(window.__INITIAL_STATE__ ?? window.__NUXT__ ?? window.__NEXT_DATA__ ?? window.__APOLLO_STATE__)"` — no API needed.';else if(n.pattern==="C")a="No API visible — use SSR HTML scrape (e.g. `ppcli browser extract`) against the rendered page.";else if(n.pattern==="D")a="Endpoints need auth. Re-open the page from a signed-in session, then retry analyze; see `field-decode-playbook` §4 for token tracing.";else if(n.pattern==="E")a="WebSocket stream detected — find the underlying HTTP poll/long-poll endpoint; raw WS is not supported.";else a="No strong signal. Manually inspect `ppcli browser network --all` and pick a pattern.";return{requested_url:e.requestedUrl,final_url:e.finalUrl,title:e.title,pattern:n,anti_bot:s,initial_state:e.initialState,api_candidates:i,nearest_adapter:r,recommended_next_step:a}}

package/dist/src/browser/article-extract.js

@@ -1,2 +1,2 @@
-import*as j from"node:fs";import{createRequire as O}from"node:module";const v=O(import.meta.url);let Y=null;function U(){if(Y)return Y;const Q=v.resolve("@mozilla/readability/Readability.js"),V=v.resolve("@mozilla/readability/Readability-readerable.js");Y={readability:j.readFileSync(Q,"utf8"),readerable:j.readFileSync(V,"utf8")};return Y}export const DEFAULT_FALLBACK_SELECTORS=["main",'[role="main"]',"#main-content","#main","#content",".content","article","body"];const X=80;export function buildExtractArticleJs(Q={}){const{readability:V,readerable:Z}=U(),W=Q.cleanSelectors??[],z=Q.fallbackSelectors??DEFAULT_FALLBACK_SELECTORS,$=!!Q.force,D=JSON.stringify(V),G=JSON.stringify(Z),I=JSON.stringify(W),J=JSON.stringify(z),M=JSON.stringify($);return["(() => {","  const cleanSelectors = "+I+";","  const fallbackSelectors = "+J+";","  const force = "+M+";","  const minFallbackText = "+X+";","  const readabilitySrc = "+D+";","  const readerableSrc = "+G+";","","  function escapeHtml(s) {",'    return String(s).replace(/[&<>]/g, c => ({ "&": "&amp;", "<": "&lt;", ">": "&gt;" }[c]));',"  }","","  // Short-circuit 1: non-HTML document",'  const ct = document.contentType || "";','  if (ct && ct !== "text/html" && ct !== "application/xhtml+xml") {','    const body = document.body ? (document.body.textContent || "") : "";','    return { source: "raw-text", html: "<pre>" + escapeHtml(body) + "</pre>", title: document.title || "" };',"  }","","  // Short-circuit 2: body is a single <pre>","  if (document.body) {","    const kids = document.body.children;",'    if (kids.length === 1 && kids[0] && kids[0].tagName === "PRE") {','      return { source: "pre", html: document.body.outerHTML, title: document.title || "" };',"    }","  }","","  // Deep-clone + adapter-supplied dirty-node removal","  const cloneDoc = document.cloneNode(true);","  for (const sel of cleanSelectors) {","    try { for (const n of cloneDoc.querySelectorAll(sel)) n.remove(); }","    catch (e) { /* ignore invalid selector */ }","  }","","  // Inject Readability into an isolated Function scope and extract the","  // constructors we need. Library sources use their own module.exports",'  // guard (if typeof module === "object"), which is falsy here.',"  const libs = (new Function(",'    readabilitySrc + "\\n" + readerableSrc + "\\nreturn {" +','    " Readability: typeof Readability !== \\"undefined\\" ? Readability : null," +','    " isProbablyReaderable: typeof isProbablyReaderable !== \\"undefined\\" ? isProbablyReaderable : null" +','    " };"',"  ))();","  const Readability = libs.Readability;","  const isProbablyReaderable = libs.isProbablyReaderable;","",'  const readerableOk = force || (typeof isProbablyReaderable === "function" ? isProbablyReaderable(cloneDoc) : true);',"  let article = null;",'  if (readerableOk && typeof Readability === "function") {',"    try { article = new Readability(cloneDoc).parse(); } catch (e) { article = null; }","  }","  if (article && article.content) {","    return {",'      source: "readability",',"      html: article.content,",'      title: article.title || document.title || "",',"      byline: article.byline || undefined,","      publishedTime: article.publishedTime || undefined,","      siteName: article.siteName || undefined,","    };","  }","","  // Fallback chain","  for (const sel of fallbackSelectors) {","    let el = null;","    try { el = cloneDoc.querySelector(sel); } catch (e) { continue; }","    if (!el) continue;",'    const text = (el.textContent || "").trim();',"    if (text.length < minFallbackText) continue;",'    return { source: "fallback", html: el.outerHTML, title: document.title || "" };',"  }","","  return null;","})()"].join(`
-`)}export async function extractArticle(Q,V={}){const Z=buildExtractArticleJs(V),W=await Q.evaluate(Z);if(W==null||typeof W!=="object")return null;const z=W;if(typeof z.html!=="string"||typeof z.source!=="string")return null;const $=z.source;return{html:z.html,title:typeof z.title==="string"?z.title:"",...z.byline&&{byline:z.byline},...z.publishedTime&&{publishedTime:z.publishedTime},...z.siteName&&{siteName:z.siteName},source:$}}
+import*as o from"node:fs";import{createRequire as f}from"node:module";const a=f(import.meta.url);let r=null;function m(){if(r)return r;const t=a.resolve("@mozilla/readability/Readability.js"),l=a.resolve("@mozilla/readability/Readability-readerable.js");r={readability:o.readFileSync(t,"utf8"),readerable:o.readFileSync(l,"utf8")};return r}export const DEFAULT_FALLBACK_SELECTORS=["main",'[role="main"]',"#main-content","#main","#content",".content","article","body"];const p=80;export function buildExtractArticleJs(t={}){const{readability:l,readerable:n}=m(),i=t.cleanSelectors??[],e=t.fallbackSelectors??DEFAULT_FALLBACK_SELECTORS,c=!!t.force,s=JSON.stringify(l),d=JSON.stringify(n),u=JSON.stringify(i),b=JSON.stringify(e),y=JSON.stringify(c);return["(() => {","  const cleanSelectors = "+u+";","  const fallbackSelectors = "+b+";","  const force = "+y+";","  const minFallbackText = "+p+";","  const readabilitySrc = "+s+";","  const readerableSrc = "+d+";","","  function escapeHtml(s) {",'    return String(s).replace(/[&<>]/g, c => ({ "&": "&amp;", "<": "&lt;", ">": "&gt;" }[c]));',"  }","","  // Short-circuit 1: non-HTML document",'  const ct = document.contentType || "";','  if (ct && ct !== "text/html" && ct !== "application/xhtml+xml") {','    const body = document.body ? (document.body.textContent || "") : "";','    return { source: "raw-text", html: "<pre>" + escapeHtml(body) + "</pre>", title: document.title || "" };',"  }","","  // Short-circuit 2: body is a single <pre>","  if (document.body) {","    const kids = document.body.children;",'    if (kids.length === 1 && kids[0] && kids[0].tagName === "PRE") {','      return { source: "pre", html: document.body.outerHTML, title: document.title || "" };',"    }","  }","","  // Deep-clone + adapter-supplied dirty-node removal","  const cloneDoc = document.cloneNode(true);","  for (const sel of cleanSelectors) {","    try { for (const n of cloneDoc.querySelectorAll(sel)) n.remove(); }","    catch (e) { /* ignore invalid selector */ }","  }","","  // Inject Readability into an isolated Function scope and extract the","  // constructors we need. Library sources use their own module.exports",'  // guard (if typeof module === "object"), which is falsy here.',"  const libs = (new Function(",'    readabilitySrc + "\\n" + readerableSrc + "\\nreturn {" +','    " Readability: typeof Readability !== \\"undefined\\" ? Readability : null," +','    " isProbablyReaderable: typeof isProbablyReaderable !== \\"undefined\\" ? isProbablyReaderable : null" +','    " };"',"  ))();","  const Readability = libs.Readability;","  const isProbablyReaderable = libs.isProbablyReaderable;","",'  const readerableOk = force || (typeof isProbablyReaderable === "function" ? isProbablyReaderable(cloneDoc) : true);',"  let article = null;",'  if (readerableOk && typeof Readability === "function") {',"    try { article = new Readability(cloneDoc).parse(); } catch (e) { article = null; }","  }","  if (article && article.content) {","    return {",'      source: "readability",',"      html: article.content,",'      title: article.title || document.title || "",',"      byline: article.byline || undefined,","      publishedTime: article.publishedTime || undefined,","      siteName: article.siteName || undefined,","    };","  }","","  // Fallback chain","  for (const sel of fallbackSelectors) {","    let el = null;","    try { el = cloneDoc.querySelector(sel); } catch (e) { continue; }","    if (!el) continue;",'    const text = (el.textContent || "").trim();',"    if (text.length < minFallbackText) continue;",'    return { source: "fallback", html: el.outerHTML, title: document.title || "" };',"  }","","  return null;","})()"].join(`
+`)}export async function extractArticle(t,l={}){const n=buildExtractArticleJs(l),i=await t.evaluate(n);if(i==null||typeof i!=="object")return null;const e=i;if(typeof e.html!=="string"||typeof e.source!=="string")return null;const c=e.source;return{html:e.html,title:typeof e.title==="string"?e.title:"",...e.byline&&{byline:e.byline},...e.publishedTime&&{publishedTime:e.publishedTime},...e.siteName&&{siteName:e.siteName},source:c}}

package/dist/src/browser/ax-snapshot.js

@@ -1,2 +1,2 @@
-const _=new Set(["button","link","textbox","searchbox","checkbox","radio","combobox","listbox","menuitem","menuitemcheckbox","menuitemradio","option","slider","spinbutton","switch","tab","treeitem"]),T=new Set(["article","cell","columnheader","gridcell","heading","listitem","main","navigation","region","rowheader"]),j=new Set(["generic","group","list","none","presentation","RootWebArea","WebArea"]);export function buildAxSnapshot(n,e={}){return buildAxSnapshotFromTrees([{tree:n}],e)}export function buildAxSnapshotFromTrees(n,e={}){const r=["source: ax","---"],i=new Map;let o=1;for(const[s,a]of n.entries()){if(s>0){const d=a.frame?.url?JSON.stringify(a.frame.url):JSON.stringify(a.frame?.frameId??`frame:${s}`);r.push(`frame ${d}:`)}o=U(a.tree,r,i,o,{...e,frame:a.frame,baseDepth:s>0?1:0})}r.push("---");r.push(`interactive: ${i.size}`);return{text:r.join(`
-`),refs:i}}function U(n,e,r,i,o){const a=(Array.isArray(n?.nodes)?n.nodes:[]).filter((t)=>t&&!t.ignored),d=new Map,S=new Set;for(const t of a){if(typeof t.nodeId==="string")d.set(t.nodeId,t);for(const f of t.childIds??[])S.add(f)}const J=a.filter((t)=>{if(!t.nodeId)return!1;const f=c(t.role);return!S.has(t.nodeId)||f==="RootWebArea"||f==="WebArea"})[0]??a[0],L=Math.max(1,Math.min(Number(o.maxDepth)||50,200)),W=X(a),N=new Map;function I(t,f){if(!t||f>L)return!1;const h=c(t.role)||"generic",u=b(c(t.name)),m=b(c(t.value)||p(t,"value")),V=p(t,"disabled"),w=p(t,"checked"),k=p(t,"expanded"),A=p(t,"selected"),O=E(h,u,t.backendDOMNodeId),$=O||!!u||!!m||T.has(h)||!o.interactiveOnly&&!j.has(h),M=e.length;let R=!1;for(const g of t.childIds??[])if(I(d.get(g),f+1))R=!0;if(!$&&!R){e.length=M;return!1}if($){const g="  ".repeat(f),l=[];let v="";if(O){const x=String(i++);v=`[${x}]`;const y=C(h,u),D=N.get(y)??0;N.set(y,D+1);r.set(x,{ref:x,backendNodeId:t.backendDOMNodeId,role:h,name:u,...W.get(y)>1?{nth:D}:{},...o.frame?{frame:o.frame}:{}})}if(u)l.push(JSON.stringify(u));if(m&&m!==u)l.push(`value=${JSON.stringify(m)}`);if(w)l.push(`checked=${w}`);if(k)l.push(`expanded=${k}`);if(A)l.push(`selected=${A}`);if(V==="true")l.push("disabled");e.splice(M,0,`${g}${v}${h}${l.length?` ${l.join(" ")}`:""}`)}return!0}I(J,o.baseDepth);return i}export function findAxRefReplacement(n,e){const r=Array.isArray(n?.nodes)?n.nodes:[],i=e.nth??0;let o=0;for(const s of r){if(!s||s.ignored)continue;const a=c(s.role),d=b(c(s.name));if(a!==e.role||d!==e.name)continue;if(o===i){if(typeof s.backendDOMNodeId!=="number")return null;return{...e,backendNodeId:s.backendDOMNodeId}}o++}return null}function E(n,e,r){if(typeof r!=="number")return!1;if(_.has(n))return!0;return T.has(n)&&!!e}function X(n){const e=new Map;for(const r of n){if(!r||r.ignored)continue;const i=c(r.role),o=b(c(r.name));if(!E(i,o,r.backendDOMNodeId))continue;const s=C(i,o);e.set(s,(e.get(s)??0)+1)}return e}function C(n,e){return`${n}\x00${e}`}function c(n){const e=n?.value;if(typeof e==="string")return e;if(typeof e==="number"||typeof e==="boolean")return String(e);return""}function p(n,e){const r=n.properties?.find((i)=>i.name===e);return c(r?.value)}function b(n){return n.replace(/\s+/g," ").trim().slice(0,160)}
+const j=new Set(["button","link","textbox","searchbox","checkbox","radio","combobox","listbox","menuitem","menuitemcheckbox","menuitemradio","option","slider","spinbutton","switch","tab","treeitem"]),T=new Set(["article","cell","columnheader","gridcell","heading","listitem","main","navigation","region","rowheader"]),U=new Set(["generic","group","list","none","presentation","RootWebArea","WebArea"]);export function buildAxSnapshot(t,e={}){return buildAxSnapshotFromTrees([{tree:t}],e)}export function buildAxSnapshotFromTrees(t,e={}){const r=["source: ax","---"],i=new Map;let o=1;for(const[s,c]of t.entries()){if(s>0){const d=c.frame?.url?JSON.stringify(c.frame.url):JSON.stringify(c.frame?.frameId??`frame:${s}`);r.push(`frame ${d}:`)}o=z(c.tree,r,i,o,{...e,frame:c.frame,baseDepth:s>0?1:0})}r.push("---");r.push(`interactive: ${i.size}`);return{text:r.join(`
+`),refs:i}}function z(t,e,r,i,o){const c=(Array.isArray(t?.nodes)?t.nodes:[]).filter((n)=>n&&!n.ignored),d=new Map,y=new Set;for(const n of c){if(typeof n.nodeId==="string")d.set(n.nodeId,n);for(const f of n.childIds??[])y.add(f)}const L=c.filter((n)=>{if(!n.nodeId)return!1;const f=a(n.role);return!y.has(n.nodeId)||f==="RootWebArea"||f==="WebArea"})[0]??c[0],W=Math.max(1,Math.min(Number(o.maxDepth)||50,200)),V=F(c),S=new Map;function I(n,f){if(!n||f>W)return!1;const h=a(n.role)||"generic",l=b(a(n.name)),p=b(a(n.value)||m(n,"value")),_=m(n,"disabled"),k=m(n,"checked"),w=m(n,"expanded"),A=m(n,"selected"),O=C(h,l,n.backendDOMNodeId),R=O||!!l||!!p||T.has(h)||!o.interactiveOnly&&!U.has(h),M=e.length;let v=!1;for(const g of n.childIds??[])if(I(d.get(g),f+1))v=!0;if(!R&&!v){e.length=M;return!1}if(R){const g="  ".repeat(f),u=[];let D="";if(O){const x=String(i++);D=`[${x}]`;const N=J(h,l),E=S.get(N)??0;S.set(N,E+1);r.set(x,{ref:x,backendNodeId:n.backendDOMNodeId,role:h,name:l,...V.get(N)>1?{nth:E}:{},...o.frame?{frame:o.frame}:{}})}if(l)u.push(JSON.stringify(l));if(p&&p!==l)u.push(`value=${JSON.stringify(p)}`);if(k)u.push(`checked=${k}`);if(w)u.push(`expanded=${w}`);if(A)u.push(`selected=${A}`);if(_==="true")u.push("disabled");e.splice(M,0,`${g}${D}${h}${u.length?` ${u.join(" ")}`:""}`)}return!0}I(L,o.baseDepth);return i}export function findAxRefReplacement(t,e){const r=Array.isArray(t?.nodes)?t.nodes:[],i=e.nth??0;let o=0;for(const s of r){if(!s||s.ignored)continue;const c=a(s.role),d=b(a(s.name));if(c!==e.role||d!==e.name)continue;if(o===i){if(typeof s.backendDOMNodeId!=="number")return null;return{...e,backendNodeId:s.backendDOMNodeId}}o++}return null}function C(t,e,r){if(typeof r!=="number")return!1;if(j.has(t))return!0;return T.has(t)&&!!e}function F(t){const e=new Map;for(const r of t){if(!r||r.ignored)continue;const i=a(r.role),o=b(a(r.name));if(!C(i,o,r.backendDOMNodeId))continue;const s=J(i,o);e.set(s,(e.get(s)??0)+1)}return e}function J(t,e){return`${t}\x00${e}`}function a(t){const e=t?.value;if(typeof e==="string")return e;if(typeof e==="number"||typeof e==="boolean")return String(e);return""}function m(t,e){const r=t.properties?.find((i)=>i.name===e);return a(r?.value)}function b(t){return t.replace(/\s+/g," ").trim().slice(0,160)}