fix：agent eval and doc file #6158

YYH211 · 2025-12-26T11:18:11Z

doc 文件太大导致 html 不转 md （增加了限制的大小）
用户创建的测评任务管理员删不掉
用户创建的测评任务管理员删除但是界面还是显示的问题

Note

HTML → Markdown limits

Adds MAX_HTML_TRANSFORM_CHARS env (default 1,000,000) to control HTML→MD conversion size; wired through docker-compose, .env.template, env.d.ts, and worker getSafeEnv.
htmlStr2Md: uses the new limit and, when exceeding it, returns original HTML while preserving extracted base64 imageList; performance/defensive tests updated to 1MB thresholds.

Evaluation auth

authEval: query now scoped by teamId; enforces app-level read/manage checks; on AppErrEnum.unExist (app deleted) allows operation to enable eval cleanup.

^{Written by Cursor Bugbot for commit 757cc47. This will update automatically on new commits. Configure here.}

gru-agent · 2025-12-26T11:18:22Z

TestGru Assignment

Summary

Link	CommitId	Status	Reason
Detail	`af1855a`	🚫 Skipped	No files need to be tested {"packages/service/support/permission/evaluation/auth.ts":"File path does not match include patterns.","packages/service/worker/htmlStr2Md/utils.ts":"File path does not match include patterns.","test/cases/service/worker/htmlStr2Md.test.ts":"File path does not match include patterns."}

History Assignment

Tip

You can @gru-agent and leave your feedback. TestGru will make adjustments based on your input

cla-assistant · 2025-12-26T11:18:27Z

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you all sign our Contributor License Agreement before we can accept your contribution.
1 out of 2 committers have signed the CLA.

✅ c121914yu
❌ xxyyh

xxyyh seems not to be a GitHub user. You need a GitHub account to be able to sign the CLA. If you have already a GitHub account, please add the email address used for this commit to your account.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}

cla-assistant · 2025-12-26T11:18:27Z

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.

xxyyh seems not to be a GitHub user. You need a GitHub account to be able to sign the CLA. If you have already a GitHub account, please add the email address used for this commit to your account.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}

github-actions · 2025-12-26T11:18:57Z

Preview sandbox Image:

registry.cn-hangzhou.aliyuncs.com/fastgpt/fastgpt-pr:fatsgpt_sandbox_757cc47c4a1c102877e4193b6c9c79ee5b133d66

github-actions · 2025-12-26T11:20:08Z

Preview mcp_server Image:

registry.cn-hangzhou.aliyuncs.com/fastgpt/fastgpt-pr:fatsgpt_mcp_server_757cc47c4a1c102877e4193b6c9c79ee5b133d66

github-actions · 2025-12-26T11:27:09Z

Preview fastgpt Image:

registry.cn-hangzhou.aliyuncs.com/fastgpt/fastgpt-pr:fatsgpt_757cc47c4a1c102877e4193b6c9c79ee5b133d66

Copilot

Pull request overview

This PR addresses three issues: preventing HTML-to-Markdown conversion failures for large documents, fixing admin deletion permissions for user-created evaluation tasks, and resolving UI display issues after deletion.

Key Changes:

Introduces MAX_HTML_TRANSFORM_CHARS environment variable (default 1MB) to control HTML-to-MD conversion size limits
Modifies evaluation authorization to scope queries by teamId and enables cleanup of evaluations when their parent app is deleted
Updates tests to align with the new 1MB threshold

Reviewed changes

Copilot reviewed 6 out of 7 changed files in this pull request and generated 5 comments.

Show a summary per file

File	Description
projects/app/.env.template	Adds environment variable template for HTML transform character limit
deploy/templates/docker-compose.prod.yml	Configures MAX_HTML_TRANSFORM_CHARS=1000000 for production deployment
packages/service/type/env.d.ts	Adds TypeScript type definition for MAX_HTML_TRANSFORM_CHARS
packages/service/worker/utils.ts	Exposes MAX_HTML_TRANSFORM_CHARS through getSafeEnv utility
packages/service/worker/htmlStr2Md/utils.ts	Implements configurable size limit and preserves extracted images on overflow
packages/service/support/permission/evaluation/auth.ts	Adds teamId scoping to evaluation queries and special handling for deleted apps
test/cases/service/worker/htmlStr2Md.test.ts	Updates test thresholds from 100KB to 1MB and adds boundary test cases

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2025-12-30T03:17:34Z

packages/service/support/permission/evaluation/auth.ts

-    };
+  } catch (error) {
+    // If app does not exist, allow operation (app was deleted, allow eval cleanup)
+    if (error !== AppErrEnum.unExist) {


The error comparison using !== is incorrect. Error enums are typically objects with properties like code, statusText, and message, not primitive strings. This comparison will always evaluate to true because the error object will never strictly equal the enum value.

The correct approach is to compare the error's statusText property or use a type-safe error checking function. This should be:
if (error?.statusText !== AppErrEnum.unExist) or access the error's identifier property that matches the enum value.

Suggested change

if (error !== AppErrEnum.unExist) {

if ((error as any)?.statusText !== AppErrEnum.unExist) {

Copilot · 2025-12-30T03:17:35Z

packages/service/worker/htmlStr2Md/utils.ts

    if (processedHtml.length > MAX_HTML_SIZE) {
-      return { rawText: processedHtml, imageList: [] };
+      return { rawText: processedHtml, imageList: images };


When the HTML exceeds the size limit, the function now returns the original HTML with extracted base64 images. However, this creates an inconsistent behavior: the caller receives HTML in rawText (not Markdown) but also gets the extracted images. This could cause issues for consumers expecting either:

HTML with intact base64 images (original behavior), or

Markdown text with separate image references

Consider whether it's more appropriate to return the original HTML string without extracting images when the size limit is exceeded, maintaining consistency with the input format.

Copilot · 2025-12-30T03:17:35Z

projects/app/.env.template

 SHOW_DISCOUNT_COUPON=false
 # 自定义 config.json 路径
 CONFIG_JSON_PATH=
+# HTML 转 Markdown 最大字符数（超过字符数不执行转化）


The comment states "HTML 转 Markdown 最大字符数（超过字符数不执行转化）" but doesn't specify a default value or expected format. This makes it unclear for users what happens if this variable is not set. Consider adding the default value (1000000) to the comment for better clarity.

Suggested change

# HTML 转 Markdown 最大字符数（超过字符数不执行转化）

# HTML 转 Markdown 最大字符数（超过字符数不执行转化，默认 1000000）

Copilot · 2025-12-30T03:17:35Z

packages/service/support/permission/evaluation/auth.ts

+  try {
+    // App read per
+    if (per === ReadPermissionVal) {
+      await authAppByTmbId({
+        tmbId,
+        appId: evaluation.appId,
+        per: ReadPermissionVal,
+        isRoot
+      });
+
+      return {
+        teamId,
+        tmbId,
+        evaluation
+      };
+    }
+
+    // Write per
    await authAppByTmbId({
      tmbId,
      appId: evaluation.appId,
-      per: ReadPermissionVal,
+      per: ManagePermissionVal,
      isRoot
    });
-    return {
-      teamId,
-      tmbId,
-      evaluation
-    };
+  } catch (error) {
+    // If app does not exist, allow operation (app was deleted, allow eval cleanup)
+    if (error !== AppErrEnum.unExist) {
+      throw error;
+    }
  }


When a ReadPermissionVal is requested and the check passes (lines 46-58), the function returns early. However, if the check fails with an error other than AppErrEnum.unExist, that error is re-thrown (line 71). But if it's AppErrEnum.unExist, the function continues to line 75 and returns successfully - this means users could get read access to an evaluation even if the app doesn't exist and they're not the owner. This bypasses the intended access control for read operations.

The try-catch block should only apply to write permission checks (ManagePermissionVal), not read permission checks. Consider restructuring so that read permission failures are always propagated.

Copilot · 2025-12-30T03:17:35Z

test/cases/service/worker/htmlStr2Md.test.ts

      const result = html2md(hugeHtml);

      expect(result.rawText).toBe(hugeHtml);
      expect(result.imageList).toHaveLength(0);


The test expects imageList to have length 0 when HTML exceeds the size limit, but the implementation was changed to return the extracted images even when the size limit is exceeded (line 75 in utils.ts returns imageList: images). This test assertion is now incorrect and will fail.

Suggested change

expect(result.imageList).toHaveLength(0);

// 超大 HTML 被拒绝时，保留原始文本，但实现仍可能返回已提取的图片列表

expect(Array.isArray(result.imageList)).toBe(true);

* fix: santinize the upload filename (labring#6159) * feat: add user ip filter for chat log table (labring#6162) * feat: add user ip filter for chat log table * chore: fix menu placement * feat: export all chunks in collection (labring#6163) * feat: export all chunks in collection * perf: export collection api * doc --------- Co-authored-by: archer <545436317@qq.com> * fix：agent eval and doc file (labring#6158) * agent eval * eval auth * html transofrm size * fix: test --------- Co-authored-by: xxyyh <2289112474@qq> Co-authored-by: archer <545436317@qq.com> * upgrade: MongoDB 5.0.18 to 5.0.32 (labring#6148) * upgrade: MongoDB 5.0.18 to 5.0.32 Update MongoDB version from 5.0.18 to 5.0.32 in all deployment configurations. 🤖 Generated with [Claude Code](https://claude.com/claude-code) via [Happy](https://happy.engineering) Co-Authored-By: Claude <noreply@anthropic.com> Co-Authored-By: Happy <yesreply@happy.engineering> * doc --------- Co-authored-by: Claude <noreply@anthropic.com> Co-authored-by: Happy <yesreply@happy.engineering> Co-authored-by: archer <545436317@qq.com> * perf: vector db log;perf: s3 mock (labring#6165) * stop design doc * remove invalid doc * remove log * perf: s3 mock * perf: vector db log * update lock * feat: workflow presentation ui (labring#6156) * icon & gradient * line & header ui * basic presentation mode * ui & icon * node * fix * source color schema * delete code * fix build * fix * fix * fix icon * function position * fix: workflow fixview padding & context menu position (labring#6169) * fix: vector ts (labring#6166) * stop design doc * remove invalid doc * rename * fix: ts * perf: auto fit * perf: comment menu * remove invalid checker * save button colorschema * fix: icon * perf: color schema * fix: icon * perf: icon * perf: icon * perf: icon * perf: icon --------- Co-authored-by: Roy <whoeverimf5@gmail.com> Co-authored-by: archer <545436317@qq.com> Co-authored-by: YeYuheng <57035043+YYH211@users.noreply.github.com> Co-authored-by: xxyyh <2289112474@qq> Co-authored-by: Finley Ge <32237950+FinleyGe@users.noreply.github.com> Co-authored-by: Claude <noreply@anthropic.com> Co-authored-by: Happy <yesreply@happy.engineering> Co-authored-by: heheer <heheer@sealos.io>

agent eval

af1855a

pull-request-size bot added the size/M label Dec 26, 2025

eval auth

e06c4fe

c121914yu previously approved these changes Dec 30, 2025

View reviewed changes

c121914yu dismissed their stale review via e06c4fe December 30, 2025 03:01

c121914yu added 2 commits December 30, 2025 11:10

html transofrm size

9f3a054

fix: test

757cc47

c121914yu requested review from c121914yu and Copilot December 30, 2025 03:13

Copilot started reviewing on behalf of c121914yu December 30, 2025 03:13 View session

c121914yu approved these changes Dec 30, 2025

View reviewed changes

Copilot AI reviewed Dec 30, 2025

View reviewed changes

c121914yu merged commit 84570bd into labring:main Dec 30, 2025
11 of 12 checks passed

	if (error !== AppErrEnum.unExist) {
	if ((error as any)?.statusText !== AppErrEnum.unExist) {

	# HTML 转 Markdown 最大字符数（超过字符数不执行转化）
	# HTML 转 Markdown 最大字符数（超过字符数不执行转化，默认 1000000）

	expect(result.imageList).toHaveLength(0);
	// 超大 HTML 被拒绝时，保留原始文本，但实现仍可能返回已提取的图片列表
	expect(Array.isArray(result.imageList)).toBe(true);

fix：agent eval and doc file #6158

fix：agent eval and doc file #6158

Uh oh!

Conversation

YYH211 commented Dec 26, 2025 • edited by cursor bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

gru-agent bot commented Dec 26, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

TestGru Assignment

Summary

Uh oh!

cla-assistant bot commented Dec 26, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

cla-assistant bot commented Dec 26, 2025

Uh oh!

github-actions bot commented Dec 26, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Preview sandbox Image:

Uh oh!

github-actions bot commented Dec 26, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Preview mcp_server Image:

Uh oh!

github-actions bot commented Dec 26, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Preview fastgpt Image:

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Copilot AI Dec 30, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Dec 30, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Dec 30, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Dec 30, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Dec 30, 2025

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

YYH211 commented Dec 26, 2025 •

edited by cursor bot

Loading

gru-agent bot commented Dec 26, 2025 •

edited

Loading

cla-assistant bot commented Dec 26, 2025 •

edited

Loading

github-actions bot commented Dec 26, 2025 •

edited

Loading

github-actions bot commented Dec 26, 2025 •

edited

Loading

github-actions bot commented Dec 26, 2025 •

edited

Loading