How could I use aws lambda to write file to s3 (python)?

Question

I have tried to use lambda function to write a file to S3, then test shows "succeeded" ,but nothing appeared in my S3 bucket. What happened? Does anyone can give me some advice or solutions? Thanks a lot. Here's my code.

import json
import boto3

def lambda_handler(event, context):

string = "dfghj"

file_name = "hello.txt"
lambda_path = "/tmp/" + file_name
s3_path = "/100001/20180223/" + file_name

with open(lambda_path, 'w+') as file:
    file.write(string)
    file.close()

s3 = boto3.resource('s3')
s3.meta.client.upload_file(lambda_path, 's3bucket', s3_path)

Solution 1

I've had success streaming data to S3, it has to be encoded to do this:

import boto3

def lambda_handler(event, context):
    string = "dfghj"
    encoded_string = string.encode("utf-8")

    bucket_name = "s3bucket"
    file_name = "hello.txt"
    s3_path = "100001/20180223/" + file_name

    s3 = boto3.resource("s3")
    s3.Bucket(bucket_name).put_object(Key=s3_path, Body=encoded_string)

If the data is in a file, you can read this file and send it up:

with open(filename) as f:
    string = f.read()

encoded_string = string.encode("utf-8")

Solution 2

My response is very similar to Tim B but the most import part is

1.Go to S3 bucket and create a bucket you want to write to

2.Follow the below steps otherwise you lambda will fail due to permission/access. I've copied and pasted it the link content here for you too just in case if they change the url /move it to some other page.

a. Open the roles page in the IAM console.

b. Choose Create role.

c. Create a role with the following properties.

-Trusted entity – AWS Lambda.

-Permissions – AWSLambdaExecute.

-Role name – lambda-s3-role.

The AWSLambdaExecute policy has the permissions that the function needs to manage objects in Amazon S3 and write logs to CloudWatch Logs.

3. Copy and past this into your Lambda python function

   import json, boto3,os, sys, uuid
   from urllib.parse import unquote_plus
   
   s3_client = boto3.client('s3')
   
   def lambda_handler(event, context):
       some_text = "test"
       #put the bucket name you create in step 1
       bucket_name = "my_buck_name"
       file_name = "my_test_file.csv"
       lambda_path = "/tmp/" + file_name
       s3_path = "output/" + file_name
       os.system('echo testing... >'+lambda_path)
       s3 = boto3.resource("s3")
       s3.meta.client.upload_file(lambda_path, bucket_name, file_name)
   
       return {
           'statusCode': 200,
           'body': json.dumps('file is created in:'+s3_path)
       }
   

Comments

[Rick.Wang]

Thanks a lot. My code is also valid,I forgot to reload S3 bucket. I also try your method,it also works.Thank you so much.

[Robert Swift]

I found that the S3 path shouldn't have a leading / otherwise an empty folder is created, essentially becoming //100001 so I think the line should read: s3_path = "100001/20180223/" + file_name

[Matt Klein]

Note that this isn't streaming, but buffering to disk and then sending

[acaruci]

This solution worked perfectly. I needed to write a csv file , so I actually wrote to an io.StringIO and then encoded the buffer content to utf-8 and saved it to an S3 file

[user3821178]

why do you need a lambda path variable?

[vesperknight]

that variable isn't used so he doesn't need it, he just forgot to remove it