Introduction

This article mainly explains how to build an API inversion service based on Cloudflare Worker to realize API original data recording + custom model rewriting function.

Occasionally, some users want to customize the model name to be compatible with some clients that behave strangely, and some users want to obtain the raw data of API interactions to troubleshoot errors. However, for the sake of compatibility and protecting customer privacy, we are not willing to directly add support for these functions on the server side. Therefore, we designed a small script based on CF Worker to simply implement these functions while taking into account security and privacy.

Effect display:

API model rewrite claude-3-haiku=>claude-3-5-haiku,claude-3.5-haiku-8080=>claude-3-5-sonnet:

API interaction raw data records:

#Configuration

Create Worker

Go into your Cloudflare Dashboard

Click Worker & Pages => Overveiw => Create to create a new Worker

You can name it as you like, just click Create

Edit Worker code

After successful creation, click Edit Code:

Clear the original code, then copy the entire code below, then click Deploy to save and update the deployment

Create and initialize D1 database

Worker & Pages => D1 => Create button to create a new database

The name is arbitrary, just click Create after entering it.

Initialize the database and execute the table creation statement:

Copy the above SQL into the blue box and execute it

This is considered successful:

Configure Worker

Bind D1 database

Find the Worker you just created

Bind D1 database:

Fill in DB here for the variable name. For the database, select the one just initialized. Click Deploy to deploy.

Configure Worker environment variables

By default, this Worker can forward the API at this time, but will not perform any actions.

You can set the Worker environment variables here

If logging needs to be enabled:

New environment variable: API_LOGGING
Set it to enabled

If you need to use the log download interface, you need to configure the API key:

Environment variable: API_SEC_KEY
Set to a private key known only to you

If you need model override functionality:

Set environment variable: API_MODEL_REWRITE
Format: model1=>model2,model3=>model4

Start request

Assume your Worker address is:
wandering-poetry-3106.hash070.workers.dev

Then if you want to access OpenAI's Chat.Completions API interface, you can make a request to this interface https://wandering-poetry-3106.hash070.workers.dev/v1/chat/completions. If you want to access the Messages API interface, you can make a request to https://wandering-poetry-3106.hash070.workers.dev/v1/messages.

If you need to download logs:

You can open the following URL in the browser (note that the address and Key are replaced with your own in this format)

https://[your Worker address]/logs?key=[your API_SEC_KEY]&format=[export data format]

For example:
https://wandering-poetry-3106.hash070.workers.dev/logs?key=your-secret-key&format=csv

The key parameter in the URL should be filled in with the API_SEC_KEY variable set for you. format guides the data format, which can be a json or csv table.

If you need to clear data:

You can access this URL directly in your browser

https://[your Worker address]/clear-logs?key=[your API_SEC_KEY]

{"success":true,"message":{"en":"Logs cleared successfully","zh":"Logs cleared successfully"}}

Attached:

API interface description

Log export

Address:/logs
Method: GET
parameter:

key: API_SEC_KEY (required)
format: export format (optional, default csv)

csv: CSV format
json: JSON format

Example:

/logs?key=your-api-key
/logs?key=your-api-key&format=json

Log cleaning

Address:/clear-logs
Method: GET
parameter:

key: API_SEC_KEY (required)

Example:

/clear-logs?key=your-api-key

Self-built Cloudflare Worker anti-generation tutorial

Introduction#

Create Worker#

Edit Worker code#

Create and initialize D1 database#

Configure Worker#

Bind D1 database#

Configure Worker environment variables#

Start request#