fix: fix http node content decode

[fatelei]

Important

1. Make sure you have read our contribution guidelines
2. Ensure there is an associated issue and you have been assigned to it
3. Use the correct syntax to link this PR: Fixes #<issue number>.

Summary

fix #31695

because cp949 and gbk structural overlap

• The trailing byte range of GBK is quite broad: 0x40 - 0xFE (excluding 0x7F).
• CP949 also has a very wide trailing byte range: 0x41 - 0xFE (excluding 0x7F).

so use the charset_normalizer may detect chinese as korean, the output will be wrong

when httpx response has charset, use the origin httpx response content, if it missing, use charset_normalizer to guess the encoding.

Candidate List and Orthogonal Evaluation
define a clear list of candidates with corresponding scoring functions, then perform "orthogonal evaluation" on them (scoring each independently before comparing), thereby avoiding reliance on any specific trial order.
Bonus Points for CP949

if encoding == "cp949" and score > 0:
    score += 3

This is an interesting heuristic adjustment. The logic is: "If the original detector believes it's CP949 and it passes the round-trip test (score > 0), we give it a slight preference, respecting the original detector's judgment." This helps maintain stability when both encodings are genuinely valid (an extremely rare occurrence). While the number 3 is an empirical value, it causes no harm since character-count differences are typically large.
Dynamic Threshold

threshold = max(2, min(8, text_length // 10))

This dynamic threshold is ingeniously designed. For very short text, it requires at least 2 matching characters; for longer text, it demands more matches, but caps the requirement at 8. This prevents incorrect judgments in long texts based solely on one or two accidental matches.

Screenshots

Before	After
...	...

Checklist

• This change requires a documentation update, included: Dify Document
• I understand that this PR may be closed in case there was no previous discussion or issues. (This doesn't apply to typos!)
• I've added a test for each change that was introduced, and I tried as much as possible to make a single atomic change.
• I've updated the documentation accordingly.
• I ran make lint and make type-check (backend) and cd web && npx lint-staged (frontend) to appease the lint gods

[gemini-code-assist]

Summary of Changes

Hello @fatelei, I'm Gemini Code Assist¹! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!

This pull request resolves an issue where HTTP response content, particularly with Chinese characters, was being incorrectly decoded due to charset_normalizer misidentifying GBK as CP949. It introduces a sophisticated safe_decode function that intelligently handles encoding detection by prioritizing explicit charset headers and employing an orthogonal evaluation strategy for ambiguous encodings, thereby ensuring accurate text representation across diverse character sets.

Highlights

• Improved HTTP Response Content Decoding: The pull request introduces a more robust mechanism for decoding HTTP response content, specifically addressing issues where charset_normalizer might misidentify Chinese encodings (like GBK) as Korean (CP949) due to structural overlaps in their byte ranges.
• New safe_decode Utility Function: A new utility function, safe_decode, has been added to libs/encode.py. This function intelligently detects and decodes byte data, prioritizing httpx's built-in charset detection when available, and falling back to an enhanced charset_normalizer approach otherwise.
• Orthogonal Encoding Evaluation: The safe_decode function now performs an orthogonal evaluation for ambiguous encodings (specifically Chinese and Korean). It considers multiple candidate encodings (GBK, GB18030, GB2312, CP949) and scores them based on character counts and round-trip encoding validation, preventing misidentification.
• Dynamic Thresholding and CP949 Preference: The decoding logic incorporates a dynamic threshold for score validation, adapting to text length, and applies a slight preference (bonus points) to CP949 if it's initially detected and passes validation, ensuring stability for genuinely valid CP949 content.
• Comprehensive Test Coverage: Extensive unit tests have been added for the safe_decode function, covering a wide range of languages and encodings, including various Chinese, Korean, Japanese, English, and other international character sets, to ensure the reliability of the new decoding logic.

🧠 New Feature in Public Preview: You can now enable Memory to help Gemini Code Assist learn from your team's feedback. This makes future code reviews more consistent and personalized to your project's style. Click here to enable Memory in your admin console.

Using Gemini Code Assist

The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.

Invoking Gemini

You can request assistance from Gemini at any point by creating a comment using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands on the current page.

Feature	Command	Description
Code Review	/gemini review	Performs a code review for the current pull request in its current state.
Pull Request Summary	/gemini summary	Provides a summary of the current pull request in its current state.
Comment	@gemini-code-assist	Responds in comments when explicitly tagged, both in pull request comments and review comments.
Help	/gemini help	Displays a list of available commands.

Customization

To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for Github and other Google products, sign up here.

You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension.

Footnotes

1. Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩

[gemini-code-assist]

Code Review

This pull request introduces a robust fix for decoding HTTP response content, specifically addressing the ambiguity between Chinese (GBK) and Korean (CP949) encodings. The new safe_decode function is a clever approach that evaluates multiple encoding candidates and scores them to find the best fit. The logic is sound and is backed by a comprehensive set of unit tests. My review identifies a couple of minor improvements in the new api/libs/encode.py file to enhance readability and simplify the code. Overall, this is a high-quality contribution that effectively solves a tricky encoding problem.