Giters

nan-wang / jina_rest_demo

a demo for jina rest api

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

jina_rest_demo

a demo for jina rest api

Goals for Debug Tooling

Start Jina Flow

cd rest_api_example
pip install -r requirements.txt
python app.py
           pod0@38127[I]:starting jina.peapods.runtimes.zmq.zed.ZEDRuntime...
           pod0@38127[I]:input tcp://0.0.0.0:61219 (PULL_BIND) output tcp://0.0.0.0:61223 (PUSH_CONNECT) control over tcp://0.0.0.0:61218 (PAIR_BIND)
           pod1@38128[I]:starting jina.peapods.runtimes.zmq.zed.ZEDRuntime...
           pod1@38128[I]:input tcp://0.0.0.0:61223 (PULL_BIND) output tcp://0.0.0.0:61224 (PUSH_BIND) control over tcp://0.0.0.0:61222 (PAIR_BIND)
        gateway@38129[I]:starting jina.peapods.runtimes.asyncio.grpc.GRPCRuntime...
           JINA@38127[S]:successfully built ChunksSegmenter from a yaml config
        gateway@38129[I]:input tcp://0.0.0.0:61224 (PULL_CONNECT) output tcp://0.0.0.0:61219 (PUSH_CONNECT) control over ipc:///var/folders/gw/d7zfntgd7z56bytfb2w662ww0000gn/T/tmpz012mrbm (PAIR_BIND)
           pod0@38115[S]:ready and listening
        gateway@38129[S]:GRPCRuntime is listening at: 0.0.0.0:61229
           JINA@38128[S]:successfully built MatchesAppender from a yaml config
           pod1@38115[S]:ready and listening
        gateway@38115[S]:ready and listening
           Flow@38115[I]:3 Pods (i.e. 3 Peas) are running in this Flow
           Flow@38115[S]:🎉 Flow is ready to use, accepting gRPC request
           Flow@38115[I]:
            ...
           JINA@38130[S]:ready and listening
        gateway@38130[S]:RESTRuntime is listening at: 0.0.0.0:45678
        gateway@38115[S]:ready and listening

Note that the jina Flow is starting successfully and listening to 0.0.0.0:45678.

Send Requests

You can use curl to send requests to the REST APIs

Send text data

curl --request POST -d '{"data": [{"text": "hello world"}], "parameters": {"mode": "text"}}' -H 'Content-Type: application/json' http://localhost:45678/search

Change the parameters.mode can change the uri in the response.

parameter.mode response.data.docs[0].uri response.data.docs[0].mime_type
"parameters": {"mode": "text"} N/A "text/plain
"parameters": {"mode": "image"} data:image/jpeg;charset=utf-8,%FF... data:image/jpeg
"parameters": {"mode": "audio"} data:audio/mpeg;charset=utf-8,ID3%... audio/mpeg
"parameters": {"mode": "video"} data:video/mp4;charset=utf-8,ABC%... video/mp4

Send multimedia data

Use Case: As a jina developer, I would like to upload multimedia data, including image(.jpeg, .png)/audio(.wav, .mp3)/video(.mp4), and send it via the debug tool as a data uri in the format of base64.

To send multimedia data, you need to change data parameter to {"data": [{"uri": "data:image/jpeg;charset=utf-8,%FF..."}]} where uri is the data uri in the format of base64

You will see the following response in JSON format. Please refer to https://api.jina.ai/rest/ for more details about the RESTful API,

{
  "request_id": "93c528f2-117d-44ee-ad1e-c45553bbf444",
  "data": {
    "docs": [
      {
        "id": "3b2faa54-c228-11eb-8589-8c8590467aa7",
        "chunks": [
          {
            "id": "3e2bb180-c228-11eb-9cda-8c8590467aa7",
            "mime_type": "text/plain",
            "text": "chunk_0",
            "granularity": 1,
            "parent_id": "3b2faa54-c228-11eb-8589-8c8590467aa7",
            "content_hash": "ac17ad4d1eea81e9"
          },
          {
            "id": "3e2bba0e-c228-11eb-9cda-8c8590467aa7",
            "mime_type": "text/plain",
            "text": "chunk_1",
            "granularity": 1,
            "parent_id": "3b2faa54-c228-11eb-8589-8c8590467aa7",
            "content_hash": "adedb4c0da5a123e"
          }
        ],
        "matches": [
          {
            "id": "3ec5b1f4-c228-11eb-bcfb-8c8590467aa7",
            "mime_type": "text/plain",
            "text": "match_0",
            "score": {
              "ref_id": "3b2faa54-c228-11eb-8589-8c8590467aa7"
            },
            "adjacency": 1
          },
          {
            "id": "3ec5ba32-c228-11eb-bcfb-8c8590467aa7",
            "mime_type": "text/plain",
            "text": "match_1",
            "score": {
              "ref_id": "3b2faa54-c228-11eb-8589-8c8590467aa7"
            },
            "adjacency": 1
          },
          {
            "id": "3ec5bc30-c228-11eb-bcfb-8c8590467aa7",
            "mime_type": "text/plain",
            "text": "match_2",
            "score": {
              "ref_id": "3b2faa54-c228-11eb-8589-8c8590467aa7"
            },
            "adjacency": 1
          }
        ],
        "tags": {},
        "text": "hello world"
      }
    ]
  },
  "header": {
    "exec_endpoint": "/search"
  },
  "routes": [
    {
      "pod": "gateway",
      "pod_id": "6772cb69-27bd-4656-9d44-df37af6e4b7b",
      "start_time": "2021-05-31T15:52:41.239503Z",
      "end_time": "2021-05-31T15:52:47.258237Z"
    },
    {
      "pod": "pod0/ZEDRuntime",
      "pod_id": "ea48efe9-720f-4440-a5a2-fe7880508709",
      "start_time": "2021-05-31T15:52:41.241344Z",
      "end_time": "2021-05-31T15:52:46.248215Z"
    },
    {
      "pod": "pod1/ZEDRuntime",
      "pod_id": "2a061ca1-b22e-4765-830f-b4b45a972948",
      "start_time": "2021-05-31T15:52:46.251329Z",
      "end_time": "2021-05-31T15:52:47.256879Z"
    },
    {
      "pod": "gateway",
      "pod_id": "4b014887-9a8f-4fee-aa4f-61e9918cf93e",
      "start_time": "2021-05-31T15:52:47.258202Z"
    }
  ],
  "status": {}
}

Send multifields data

The full specification of the flow RESTful APIs is defined at https://api.jina.ai/rest/#operation/post_post__endpoint__post.

Use Case: As a jina developer, I would like to specify the fields of data and parameters. Specifically, I would like to send the following fields via thee REST API,

  • data.id: string
  • data.uri: string
  • data.text: string
  • data.mimeType: string
  • data.tags: (via RAW inputs)
  • data.modality: string
  • parameters: (via RAW inputs)
curl --request POST -d '{"data": [{"text": "a multi-field doc", "modality": "customized_modality"}], "parameters": {"mode": "text"}}' -H 'Content-Type: application/json' http://localhost:45678/post/multifields

Visualize Response

In the response, we want to visualize the following fields

routes

In the routes, you will see the route information of all the Pods that has processed the request.

{
  ...
  "routes": [
    {
      "pod": "gateway",
      "pod_id": "6772cb69-27bd-4656-9d44-df37af6e4b7b",
      "start_time": "2021-05-31T15:52:41.239503Z",
      "end_time": "2021-05-31T15:52:47.258237Z"
    },
    {
      "pod": "pod0/ZEDRuntime",
      "pod_id": "ea48efe9-720f-4440-a5a2-fe7880508709",
      "start_time": "2021-05-31T15:52:41.241344Z",
      "end_time": "2021-05-31T15:52:46.248215Z"
    },
    {
      "pod": "pod1/ZEDRuntime",
      "pod_id": "2a061ca1-b22e-4765-830f-b4b45a972948",
      "start_time": "2021-05-31T15:52:46.251329Z",
      "end_time": "2021-05-31T15:52:47.256879Z"
    },
    {
      "pod": "gateway",
      "pod_id": "4b014887-9a8f-4fee-aa4f-61e9918cf93e",
      "start_time": "2021-05-31T15:52:47.258202Z"
    }
  ],
  ...
}

USE CASE: As a jina developer, I want to see how much time spent on each Pod, which is equal to end_time - start_time.

docs[i]

Each Document in the docs has chunks, which is a list of Document again.

USE CASE: As a jina developer, I want to see the details of each chunk. including

  • mime_type
  • text
  • uri (display images in the case of files images, otherwise show in plain text)
  • tags (TBD)
  • blob (TBD)
  • embedding (TBD)
{
  "request_id": "93c528f2-117d-44ee-ad1e-c45553bbf444",
  "data": {
    "docs": [
      {
        "id": "3b2faa54-c228-11eb-8589-8c8590467aa7",
        "chunks": [...],
        "matches": [...],
        "tags": {},
        "text": "hello world"
      }
    ]
  },
  ...
}

docs[i].chunks

Each Document in the docs has chunks, which is a list of Document again.

USE CASE: As a jina developer, I want to see the details of each chunk. including

  • mime_type
  • text
  • granularity
  • uri (display images in the case of files images, otherwise show in plain text)
  • blob (TBD)
  • embedding (TBD)
{
  "request_id": "93c528f2-117d-44ee-ad1e-c45553bbf444",
  "data": {
    "docs": [
      {
        "id": "3b2faa54-c228-11eb-8589-8c8590467aa7",
        "chunks": [
          {
            "id": "3e2bb180-c228-11eb-9cda-8c8590467aa7",
            "mime_type": "text/plain",
            "text": "chunk_0",
            "granularity": 1,
            "parent_id": "3b2faa54-c228-11eb-8589-8c8590467aa7",
            "content_hash": "ac17ad4d1eea81e9"
          },
          {
            "id": "3e2bba0e-c228-11eb-9cda-8c8590467aa7",
            "mime_type": "text/plain",
            "text": "chunk_1",
            "granularity": 1,
            "parent_id": "3b2faa54-c228-11eb-8589-8c8590467aa7",
            "content_hash": "adedb4c0da5a123e"
          }
        ],
        "text": "hello world"
      }
    ]
  },
  ...
}

docs[i].matches

Each Document in the docs has matches, which is a list of Document again.

USE CASE: As a jina developer, I want to see the details of each match. including

  • mime_type
  • text
  • granularity
  • uri (display images in the case of files images, otherwise show in plain text)
  • blob (TBD)
  • embedding (TBD)
  • parent_id (TBD)
{
  "request_id": "93c528f2-117d-44ee-ad1e-c45553bbf444",
  "data": {
    "docs": [
      {
        ...
        "matches": [
          {
            "id": "3ec5b1f4-c228-11eb-bcfb-8c8590467aa7",
            "mime_type": "text/plain",
            "text": "match_0",
            "score": {
              "ref_id": "3b2faa54-c228-11eb-8589-8c8590467aa7"
            },
            "adjacency": 1
          },
          {
            "id": "3ec5ba32-c228-11eb-bcfb-8c8590467aa7",
            "mime_type": "text/plain",
            "text": "match_1",
            "score": {
              "ref_id": "3b2faa54-c228-11eb-8589-8c8590467aa7"
            },
            "adjacency": 1
          },
          {
            "id": "3ec5bc30-c228-11eb-bcfb-8c8590467aa7",
            "mime_type": "text/plain",
            "text": "match_2",
            "score": {
              "ref_id": "3b2faa54-c228-11eb-8589-8c8590467aa7"
            },
            "adjacency": 1
          }
        ],
        ...
      }
    ]
  },
  ...
}

docs[i].uri

USE CASE: As a jina developer, I would like to visualize/play the uri if it is a data uri in the format of base64, in particular when the uri is an image(.jpeg, .png)/audio(.wav, .mp3)/video(.mp4).

About

a demo for jina rest api

License:Apache License 2.0

Languages

Language:Python 100.0%