WeChat Article Search & Text Extractor — Sogou

Search public WeChat Official Account articles indexed by Sogou and optionally extract readable article text. No WeChat login is required.

通过搜狗微信按关键词搜索公开的微信公众号文章，解析微信原文地址，并可提取可读正文。支持 JSON、CSV 导出，无需微信登录。

Quick start

Use this input for your first run:

{
  "query": "新能源汽车",
  "limit": 3,
  "fetchContent": true
}

1. Enter a keyword in query.
2. Choose a result limit from 1 to 1000. The Actor automatically walks Sogou result pages, up to 10 results per page.
3. Set fetchContent to true to retrieve article text.
4. Start the Actor and open its Output tab.
5. Check each row's status before using the results; export the dataset as JSON or CSV.

Input

Field	Type	Default	Description
query	string	Required	Non-empty search keyword. Chinese keywords generally match Chinese-language content more directly.
limit	integer	3	Maximum articles to collect across Sogou result pages; allowed range 1–1000. Actual count can be lower if Sogou has fewer results or verification interrupts the run.
fetchContent	boolean	false	When true, retrieve readable text after resolving the article URL. When false, still resolve article links.
proxyConfiguration	object	No proxy	Optional Apify Proxy or your own HTTP(S) proxies, selected in the Proxy configuration editor.

Suggested starting settings: 512 MB memory and a 300-second timeout. Large limits, especially with fetchContent: true, can require substantially more runtime and are more likely to encounter verification challenges.

Optional proxy configuration

If Sogou returns SOGOU_ANTI_BOT, you can configure a proxy you are authorized to use. No proxy is enabled by default. Omitted configuration, {}, and an empty proxy URL list remain direct mode. Only useApifyProxy: true or a non-empty list of valid custom proxy URLs enables a proxy; inactive group/country selections are ignored. Invalid custom URLs or conflicting modes are rejected. The input editor supports Apify datacenter/residential groups, available countries, and your own HTTP(S) proxy URLs. Access and proxy charges depend on your Apify account or proxy provider; no developer-owned proxy credentials are supplied. Core features are not gated by the user's subscription tier.

For Apify Proxy, add:

"proxyConfiguration": { "useApifyProxy": true }

For an available residential group, use apifyProxyGroups: ["RESIDENTIAL"]; optionally select a supported apifyProxyCountry. For your own proxy, use:

"proxyConfiguration": {
  "useApifyProxy": false,
  "proxyUrls": ["http://USERNAME:PASSWORD@PROXY_HOST:PORT"]
}

These are fields to add to the input object, not complete inputs. Percent-encode special characters in usernames/passwords. Do not post real proxy credentials in Issues or share your INPUT record. The Actor excludes proxy URLs and credentials from its own result summaries and diagnostics.

Search, link resolution and article fetches use the same proxy configuration and one session per run. One custom endpoint is selected per run; provider-side IP rotation is outside the Actor's control. Cookies are kept in a run-local cookie jar with domain/path scoping. Selecting a proxy never silently falls back to direct traffic. Requests are serial, spaced by at least 500 ms, with at most one retry for temporary network failures or 429/5xx responses within the request time budget. Long Retry-After responses are returned without an early retry. Human-verification pages are not automatically retried or solved.

A proxy can improve reachability but does not guarantee access. This release extracts readable text, not a complete offline webpage or image archive. This proxy option is for the Apify Actor; the Cloudflare Worker keeps its original native-fetch behavior and MCP inputs.

Output

One article is saved per dataset row. The Output tab provides an overview and an article-text view. Missing source metadata is omitted rather than invented.

Field	Meaning
query	Original search keyword
title	Article title
source	Official Account name, when available
publishedAt	Publication time or date as extracted from the search result, when available
snippet	Search-result excerpt, when available
url	Original Sogou result URL
realUrl	Resolved WeChat article URL, when successful
resolvedUrl	Alias of the successfully resolved article URL
content	Readable article text when requested and successfully fetched
resolutionStatus	resolved, failed, or deferred
status	resolved, fetched, or failed
error	Processing error for a failed row
errorCode	Machine-readable error category, when a row failed
stage	resolve or content, when a row failed
resolutionError	Link-resolution diagnostic, when applicable

• resolved: the WeChat URL was resolved; article text was not requested.
• fetched: readable article text was retrieved.
• failed: inspect error. A title or search excerpt is not proof that article text was fetched.

The default key-value store record ERROR records the failure stage (input, proxy, or search), a message and a troubleshooting hint when the run fails. OUTPUT.proxyMode reports none, apify, or custom without credentials.

The default key-value store record OUTPUT contains the query, requested limit, count, succeeded and failed counts, and a summary status: complete, partial, failed, or empty.

Article processing failures do not fail the Actor run, even if every returned article fails. In that case the run succeeds with exit code 0, while OUTPUT.status is failed and all failed rows retain status, errorCode, and stage. The summary status describes data quality, not the platform run status. A search with no results is recorded as empty. Invalid input, proxy initialization failures, and search or storage errors still fail the run. An explicitly selected but unavailable proxy never silently falls back to direct traffic. Previously saved rows remain available if a later operation fails.

What you can use it for

• Discover articles about a topic or brand in Chinese-language public content.
• Collect source text for research or downstream summarization.
• Export structured article records into your own analysis workflow.
• Run keyword searches through the Apify API or your own scheduled tasks.

This Actor produces source data; it does not summarize articles or track changes between runs.

Current limitations

• One keyword per run; limit can request up to 1000 articles and the Actor paginates automatically, but the actual count depends on Sogou's available result pages.
• Search covers Sogou's available index, not every WeChat article.
• No full Official Account history, guaranteed real-time coverage, or strict account filtering.
• No private, paid-access, or deleted article access.
• No engagement metrics such as reads, likes, or comments.
• Text extraction does not perform OCR on images or download article media.
• Verification challenges, expired links, and source availability can prevent resolution or text extraction.
• Links are handled serially; already resolved URLs are reused. A Sogou verification challenge stops further Sogou link resolution. Optional proxies are supported; there is no automatic CAPTCHA-solving or browser fallback.

Pricing

Check this Actor's Pricing tab for the currently effective charges and any platform usage charges.

The dataset intentionally includes failed rows for diagnosis. If default-dataset-item billing is enabled, those rows may also incur result charges. This version does not implement custom success-only billing; do not interpret a failed run as a guarantee of zero charges.

Support

Use this Actor's Issues tab and include the input JSON, run ID, and exact error message. Never include API tokens or other credentials.

本工具支持按结果页自动翻页，单次最多请求 1000 条；大批量抓取更容易触发搜狗验证，也会显著增加运行时间。遇到失败请查看 error，并提供关键词、运行 ID 和报错信息；不要提供 Token。

This is an independent tool and is not affiliated with Tencent, WeChat, or Sogou.