[Bug Report] Bugs of JinaAPI and HTTP error

[pprp]

Self Checks

• This is only for bug report, if you would like to ask a question, please head to Discussions.
• I have searched for existing issues search for existing issues, including closed ones.
• I confirm that I am using English to submit this report (我已阅读并同意 Language Policy).
• [FOR CHINESE USERS] 请务必使用英文提交 Issue，否则会被关闭。谢谢！:）
• Please do not modify this template :) and fill in all the required fields.

Dify version

0.10.0

Cloud or Self Hosted

Cloud

Steps to reproduce

Bug1: Jina error. I found that Dify can not update the Jina api properly. When my Jina API is out of tokens, I bought some tokens and update the API key. However, I find Dify still remembers the old API key so that Jina can not run and just give bugs:

Bug2: HTTP request. I am using imgrender api which is implemented using HTTP Request node in Dify. This API is properly tested on local device using python to get http results. Today, I just find that the returned img urls can not be open. (Also, it worked fine yesterday on dify.) I am sure that this bug has no relation with imgrender.

Bug3: I have both image and text in the output file. However, there is only text being shown. The images are now properly shown. For example, the figure below is have image link but it can not properly shown in the output.

✔️ Expected Behavior

Bug 1: After the Jina API is set, it can operate properly.

Bug 2: Please resolve the HTTPS error. As it can run locally, it should also function well on dify.

Bug 3: I wish to ensure that the image is displayed properly.

❌ Actual Behavior

No response

[dosubot]

@pprp 你好! 我是Dosu，一个可以帮助你解决 bug、回答问题并成为贡献者的机器人。我会帮你等待人类维护者的过程中。

For the issues you're experiencing with Dify, here are some relevant insights and potential solutions:

1. Jina API Token Management:

   • There is a known issue with Dify not updating the Jina API key properly, which can lead to errors like "InsufficientBalanceError" if the old API key is still being used. Ensure that the new API key is correctly updated in Dify's configuration. This might involve checking if the key is cached or stored in a way that needs manual updating ^[1].

2. HTTP Request Issues:

   • When HTTP requests work locally but not in the cloud, it is important to verify the API endpoint configuration, including the URL and API key. Additionally, using tools like Ngrok can help expose local services to the cloud for testing. Ensure that the API response format is correct and that there are no network restrictions or misconfigurations in the cloud environment ^[2][3].

3. Image Display in Outputs:

   • If images are not displaying properly in the output, ensure that the workflow is configured to handle multimedia content correctly. This might involve checking the configuration settings related to file uploads and ensuring that the image URLs are accessible and correctly formatted. There have been similar issues where images were not displayed due to configuration problems ^[4][5][6].

These steps should help address the issues you're facing with Dify. If the problems persist, further investigation into the specific configurations and logs might be necessary.

To continue talking to Dosu, mention @dosu.

[crazywoola]

Bug 1: After the Jina API is set, it can operate properly.

If the key is changed, you need to remove the credentials in jina reader tool and add it again.

Bug 2: Please resolve the HTTPS error. As it can run locally, it should also function well on dify.

Bug 3: I wish to ensure that the image is displayed properly.

Please provide the DSL.

[pprp]

@crazywoola how to remove the credential?

I have updated the key in these two places.

[crazywoola]

Here

[mapleeit]

Hi, I'm Jina AI's engineer. After some checking with @pprp

We found the place that made the confusion: the credentials of DataSource, Model(Provider), Tools are managed separately. And Jina have all of them, unfortunately. Wondering if you have some advice or plan to solve this kind of problem.🤔

[pprp]

@crazywoola for the 3rd bug, here is the dsl file:

app:
  description: 对某个话题进行消化
  icon: grinning
  icon_background: '#D5D9EB'
  mode: workflow
  name: 图文混排测试
  use_icon_as_answer_icon: false
kind: app
version: 0.1.2
workflow:
  conversation_variables: []
  environment_variables: []
  features:
    file_upload:
      allowed_file_extensions:
      - .JPG
      - .JPEG
      - .PNG
      - .GIF
      - .WEBP
      - .SVG
      allowed_file_types:
      - image
      allowed_file_upload_methods:
      - local_file
      - remote_url
      enabled: false
      image:
        enabled: false
        number_limits: 3
        transfer_methods:
        - local_file
        - remote_url
      number_limits: 3
    opening_statement: ''
    retriever_resource:
      enabled: false
    sensitive_word_avoidance:
      enabled: false
    speech_to_text:
      enabled: false
    suggested_questions: []
    suggested_questions_after_answer:
      enabled: false
    text_to_speech:
      enabled: false
      language: ''
      voice: ''
  graph:
    edges:
    - data:
        isInIteration: false
        sourceType: llm
        targetType: code
      id: 1725590604530-source-1725599428552-target
      source: '1725590604530'
      sourceHandle: source
      target: '1725599428552'
      targetHandle: target
      type: custom
      zIndex: 0
    - data:
        isInIteration: false
        sourceType: code
        targetType: iteration
      id: 1725599428552-source-1725600094234-target
      source: '1725599428552'
      sourceHandle: source
      target: '1725600094234'
      targetHandle: target
      type: custom
      zIndex: 0
    - data:
        isInIteration: true
        iteration_id: '1725600094234'
        sourceType: iteration-start
        targetType: tool
      id: 1725600094234start0-source-1725600107123-target
      source: 1725600094234start0
      sourceHandle: source
      target: '1725600107123'
      targetHandle: target
      type: custom
      zIndex: 1002
    - data:
        isInIteration: false
        sourceType: start
        targetType: tool
      id: start-source-1729433574725-target
      source: start
      sourceHandle: source
      target: '1729433574725'
      targetHandle: target
      type: custom
      zIndex: 0
    - data:
        isInIteration: false
        sourceType: llm
        targetType: end
      id: 1725602633045-source-end-target
      source: '1725602633045'
      sourceHandle: source
      target: end
      targetHandle: target
      type: custom
      zIndex: 0
    - data:
        isInIteration: false
        sourceType: tool
        targetType: llm
      id: 1729433574725-source-1725590604530-target
      source: '1729433574725'
      sourceHandle: source
      target: '1725590604530'
      targetHandle: target
      type: custom
      zIndex: 0
    - data:
        isInIteration: false
        sourceType: iteration
        targetType: llm
      id: 1725600094234-source-1725602633045-target
      source: '1725600094234'
      sourceHandle: source
      target: '1725602633045'
      targetHandle: target
      type: custom
      zIndex: 0
    nodes:
    - data:
        selected: false
        title: START
        type: start
        variables:
        - label: 网页链接
          max_length: 256
          options: []
          required: true
          type: text-input
          variable: URL
      height: 90
      id: start
      position:
        x: 30
        y: 349
      positionAbsolute:
        x: 30
        y: 349
      selected: false
      type: custom
      width: 244
    - data:
        outputs:
        - value_selector:
          - '1725602633045'
          - text
          variable: output
        selected: false
        title: END
        type: end
      height: 90
      id: end
      position:
        x: 2595
        y: 349
      positionAbsolute:
        x: 2595
        y: 349
      selected: false
      type: custom
      width: 244
    - data:
        context:
          enabled: false
          variable_selector: []
        desc: ''
        model:
          completion_params:
            frequency_penalty: 0.1
            presence_penalty: 0.1
            temperature: 0.8
            top_p: 0.9
          mode: chat
          name: deepseek-ai/DeepSeek-V2.5
          provider: siliconflow
        prompt_template:
        - id: af8ee5d9-ae1e-4610-9675-8db26e3f48ec
          role: system
          text: "## 工作描述：根据提供的文章撰写一系列prompt用于文生图 \n\n## 工作流程 \n\n第一步，根据提供的文章，均匀地划分4个模块；\n\
            第二步，对不同模块内容进行抽取核心概念，并且正确地编写2个prompt用于文生图；\n\n返回文生图所用的一系列Prompt\n\n###\
            \ 输出 ###\n1. 注意，接收你输出的程序，只能接受JSON格式\n2. 格式模版为 `{\"outline\":\" [{{第一个prompt}}\"\
            ,\"{{第二个prompt}}\", \"{{第三个prompt}}\", {{第四个prompt}}]`\n3. 不要添加markdown标记。\n\
            4. Prompt本身应该使用英文来尽可能详细的描述。\n5. Prompt内容尽量不要涉及文字，比如图片中写着xxx字。\n\n### 输出例子\
            \ ### \n\n{\n  \"outline\": [\n    \"Describe a future cityscape, with\
            \ neon lights and flying cars filling the night scene\",\n    \"Describe\
            \ a scene of elderly people and children living harmoniously together,\
            \ in a realistic style\",\n  ]\n}\n\n\n### 输入内容为：\n\n{{#1729433574725.text#}}"
        selected: false
        title: PromptGenerate
        type: llm
        variables: []
        vision:
          enabled: false
      height: 98
      id: '1725590604530'
      position:
        x: 724.8571428571429
        y: 336.1428571428571
      positionAbsolute:
        x: 724.8571428571429
        y: 336.1428571428571
      selected: false
      sourcePosition: right
      targetPosition: left
      type: custom
      width: 244
    - data:
        code: "function main({ text }) {\n    const json = JSON.parse(text);\n   \
          \ return {\n        prompts: json.outline\n    };\n}"
        code_language: javascript
        desc: ''
        outputs:
          prompts:
            children: null
            type: array[string]
        selected: false
        title: 代码执行
        type: code
        variables:
        - value_selector:
          - '1725590604530'
          - text
          variable: text
      height: 54
      id: '1725599428552'
      position:
        x: 1007.4285714285716
        y: 393.28571428571433
      positionAbsolute:
        x: 1007.4285714285716
        y: 393.28571428571433
      selected: false
      sourcePosition: right
      targetPosition: left
      type: custom
      width: 244
    - data:
        desc: ''
        height: 273
        iterator_selector:
        - '1725599428552'
        - prompts
        output_selector:
        - '1725600107123'
        - text
        output_type: array[string]
        selected: false
        startNodeType: tool
        start_node_id: 1725600094234start0
        title: 迭代
        type: iteration
        width: 377
      height: 273
      id: '1725600094234'
      position:
        x: 1344.2857142857144
        y: 363.2857142857143
      positionAbsolute:
        x: 1344.2857142857144
        y: 363.2857142857143
      selected: false
      sourcePosition: right
      targetPosition: left
      type: custom
      width: 377
      zIndex: 1
    - data:
        desc: ''
        isInIteration: true
        isIterationStart: true
        iteration_id: '1725600094234'
        provider_id: siliconflow
        provider_name: siliconflow
        provider_type: builtin
        selected: false
        title: Flux
        tool_configurations:
          image_size: 768x512
          model: schnell
          num_inference_steps: 12
          seed: null
        tool_label: Flux
        tool_name: flux
        tool_parameters:
          prompt:
            type: mixed
            value: '{{#1725600094234.item#}}'
        type: tool
      extent: parent
      height: 168
      id: '1725600107123'
      parentId: '1725600094234'
      position:
        x: 105.88230746836871
        y: 85
      positionAbsolute:
        x: 1450.1680217540832
        y: 448.2857142857143
      selected: false
      sourcePosition: right
      targetPosition: left
      type: custom
      width: 244
      zIndex: 1001
    - data:
        context:
          enabled: false
          variable_selector: []
        desc: 生成MD格式
        model:
          completion_params:
            frequency_penalty: 0.3
            presence_penalty: 0.2
            temperature: 0.5
            top_p: 0.85
          mode: chat
          name: gemini-1.5-flash-002
          provider: google
        prompt_template:
        - edition_type: basic
          id: b1131b36-0da7-4864-a019-e9fbdb937a9e
          role: system
          text: "### 要求：根据给定内容将图片通过markdown形式以合适的方法插入文本\n\n### 内容 \n\n将 {{#1729433574725.text#}}\
            \ 作为核心文本，将 {{#1725600094234.output#}} 中提供的一些markdown url 在合适的位置，以markdown语法形式插入文本中。\n\
            \n所谓合适的位置，即均匀地插入文本中，不是太靠前或者太靠后。 \n\n请添加合适的标题，需要符合公众号命名习惯。\n\n### 输出 \n\
            \n输出为markdown格式文件。"
        selected: true
        title: MDGEN
        type: llm
        variables: []
        vision:
          configs:
            detail: high
          enabled: false
      height: 126
      id: '1725602633045'
      position:
        x: 2053.8571428571427
        y: 331.85714285714283
      positionAbsolute:
        x: 2053.8571428571427
        y: 331.85714285714283
      selected: true
      sourcePosition: right
      targetPosition: left
      type: custom
      width: 244
    - data:
        desc: ''
        isInIteration: true
        selected: false
        title: ''
        type: iteration-start
      draggable: false
      height: 48
      id: 1725600094234start0
      parentId: '1725600094234'
      position:
        x: 24
        y: 68
      positionAbsolute:
        x: 1368.2857142857144
        y: 431.2857142857143
      selectable: false
      sourcePosition: right
      targetPosition: left
      type: custom-iteration-start
      width: 44
      zIndex: 1002
    - data:
        desc: ''
        provider_id: jina
        provider_name: jina
        provider_type: builtin
        selected: false
        title: 获取单页面
        tool_configurations:
          gather_all_images_at_the_end: 0
          gather_all_links_at_the_end: 0
          image_caption: 0
          max_retries: 3
          no_cache: 0
          proxy_server: null
          summary: 0
          target_selector: null
          wait_for_selector: null
        tool_label: 获取单页面
        tool_name: jina_reader
        tool_parameters:
          url:
            type: mixed
            value: '{{#start.URL#}}'
        type: tool
      height: 298
      id: '1729433574725'
      position:
        x: 334
        y: 349
      positionAbsolute:
        x: 334
        y: 349
      selected: false
      sourcePosition: right
      targetPosition: left
      type: custom
      width: 244
    viewport:
      x: -1048.8999999999996
      y: 390.20000000000005
      zoom: 0.7

Taking this input as example: https://mp.weixin.qq.com/s/XO7Qb6LlYPr7frxuJw2BuA

here is the final output:

Here are urls in the output of LLM. Unfortunatelly, there is no image shown as below:

[pprp]

@crazywoola For the 2nd bug, here is the code for testing locally:

import requests
import json

url = "https://api.imgrender.cn/open/v1/pics"


payload = json.dumps(
    {
        "width": 640,
        "height": 850,
        "backgroundColor": "#fff",
        "texts": [
            {
                "x": 32,
                "y": 185,
                "text": "这是一个测试，为什么本地可以，dify上有问题？",
                "font": "SourceHanSerifCN-Heavy",
                "fontSize": 80,
                "color": "#000",
                "width": 620,
                "textAlign": "left",
                "zIndex": 1,
            },
        ],
        "lines": [],
        "images": [],
    }
)

headers = {
    "X-API-Key": "YOUR API KEY",
    "Content-Type": "application/json",
}

response = requests.request("POST", url, headers=headers, data=payload)
response_data = json.loads(response.text)
print(response_data)
image_url = response_data["data"]["url"]

# Download the image
image_response = requests.get(image_url)
with open("rendered_image.jpg", "wb") as f:
    f.write(image_response.content)

print("Image saved as 'rendered_image.jpg'")

This is the test result:

here is how to use it in dify:

We got output:

{
  "status_code": 200,
  "body": "{\"code\":0,\"message\":\"OK\",\"data\":{\"url\":\"https://res.imgrender.net/e839a782d64ddfd997b80ac072f8cd9a.jpg?auth_key=1729583835-ixc6ciq4LYoCg0ca-0-510d609b8448d6918b05687ce5e32ce4\"}}",
  "headers": {
    "date": "Tue, 22 Oct 2024 07:57:15 GMT",
    "content-type": "application/json; charset=utf-8",
    "content-length": "177",
    "connection": "keep-alive",
    "keep-alive": "timeout=25",
    "server": "Kaede/3.5.3.899 (bj0026odl)",
    "x-request-id": "c1d989d43c1f3f9af6eb",
    "x-ca-request-id": "70A2B2D2-223D-4777-B5D5-955C97814FEA",
    "access-control-expose-headers": "Date,x-fc-request-id",
    "x-fc-request-id": "70A2B2D2-223D-4777-B5D5-955C97814FEA",
    "vary": "Origin"
  },
  "files": []
}

It is so odd. Why can't I open this url?

---

Additionally, there is other person who has the same issue as Bug3:

[pprp]

@crazywoola is there any progress? Now, the second bug is fixed using python. But the 3rd bug still exists.

[crazywoola]

For bug 3, I think you could use some nodes to extract the url and display it. But if you want to get the url from the uploaded file, please take a look at this. #9567