Does AWS apigateway change http body? How can I stop it from doing this?
My application:
(1) A front end "UI" that sends a "http request" using "POST method" that contains a "zip file" in "body" through "form-data".
(2) AWS "apigateway" receives this request and forward it to "Lambda Proxy"
(3) AWS "Lambda" implemented by python coding receives this request and decompresses this zip file to a temporary folder.
The problem I'm facing:
(1) and (2) works fine, but in (3) the pythong program at lambda failed to decompress the file.
My finding:
It seems that when sending from the "UI" the body contains the binary data of the zip file
like below:
"PK\x03\x04\n\x00\x00\x00\x00\x00\xd6;TO\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x06\x00\x00\x00x2.txtPK\x03\x04\n\x00\x00\x00\x00\x00\xd6;TO\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x06\x00\x00\x00x1.txtPK\x01\x02\x14\x00\n\x00\x00\x00\x00\x00\xd6;TO\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x06\x00\x00\x00\x00\x00\x00\x00\x00\x00
\x00\x00\x00\x00\x00\x00\x00x2.txtPK\x01\x02\x14\x00\n\x00\x00\x00\x00\x00\xd6;TO\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x06\x00\x00\x00\x00\x00\x00\x00\x00\x00
\x00\x00\x00$\x00\x00\x00x1.txtPK\x05\x06\x00\x00\x00\x00\x02\x00\x02\x00h\x00\x00\x00H\x00\x00\x00\x00\x00"
But at (3) the python code at lambda, if we just simply returns the response like below:
response = {
"statusCode": 200,
"headers": {
"lambda-response": str(event["body"])
},
"body": "",
"isBase64Encoded": False
}
return response
will find that the binary data in the body,
seems like apigateway has changed the content
from:
"PK\x03\x04\n\x00\x00\x00\x00\x00\xd6;TO\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x06\x00\x00\x00x2.txtPK\x03\x04\n\x00\x00\x00\x00\x00\xd6;TO\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x06\x00\x00\x00x1.txtPK\x01\x02\x14\x00\n\x00\x00\x00\x00\x00\xd6;TO\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x06\x00\x00\x00\x00\x00\x00\x00\x00\x00
\x00\x00\x00\x00\x00\x00\x00x2.txtPK\x01\x02\x14\x00\n\x00\x00\x00\x00\x00\xd6;TO\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x06\x00\x00\x00\x00\x00\x00\x00\x00\x00
\x00\x00\x00$\x00\x00\x00x1.txtPK\x05\x06\x00\x00\x00\x00\x02\x00\x02\x00h\x00\x00\x00H\x00\x00\x00\x00\x00"
into:
"PK\u0003\u0004\n\u0000\u0000\u0000\u0000\u0000\ufffd;TO\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0006\u0000\u0000\u0000x2.txtPK\u0003\u0004\n\u0000\u0000\u0000\u0000\u0000\ufffd;TO\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0006\u0000\u0000\u0000x1.txtPK\u0001\u0002\u0014\u0000\n\u0000\u0000\u0000\u0000\u0000\ufffd;TO\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0006\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000
\u0000\u0000\u0000\u0000\u0000\u0000\u0000x2.txtPK\u0001\u0002\u0014\u0000\n\u0000\u0000\u0000\u0000\u0000\ufffd;TO\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0006\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000
\u0000\u0000\u0000$\u0000\u0000\u0000x1.txtPK\u0005\u0006\u0000\u0000\u0000\u0000\u0002\u0000\u0002\u0000h\u0000\u0000\u0000H\u0000\u0000\u0000\u0000\u0000\r\n"
Which is weird, what can I do to stop this?
(2019/12/17 update) below the lambda code I'm using.
import json # to decode json
import os # file IO
import shutil # file IO (use this to recursively force remove a directory)
print('Loading function')
def decompress_zip_file(src_file_path, dest_dir_path):
'''
Decompress a zip file into a directory.
Args:
src_file_path (Srting): source zip file's path.
dest_dir_path (Srting): the destination of the output directory.
Returns:
isSuccess (bool): the operation is successful or not.
'''
error_msg = "Nothing."
try:
if(os.path.isdir(dest_dir_path)):
shutil.rmtree(dest_dir_path)
with zipfile.ZipFile(src_file_path, 'r') as zip_ref:
zip_ref.extractall(dest_dir_path)
except Exception as ep:
error_msg = "Error in decompress_zip_file(), ep={}:{}".format(type(ep).__name__, str(ep))
print(error_msg)
return (False, error_msg)
return (True, error_msg)
def decompress_zip_file_from_content_in_binary(src_file_in_binary, dest_dir_path):
'''
Decompress a zip file content into a directory.
Args:
src_file_in_binary (byte array): source zip file's content in binary format.
dest_dir_path (Srting): the destination of the output directory.
Returns:
isSuccess (bool): the operation is successful or not.
'''
# write the obtained binary data into a tmp zip file
tmp_file_path = "/tmp/tmp.zip"
if(os.path.isfile(tmp_file_path)):
os.remove(tmp_file_path)
output_file = open(tmp_file_path, 'wb')
output_file.write(src_file_in_binary)
output_file.close()
(isSuccess, error_msg) = decompress_zip_file(tmp_file_path, dest_dir_path)
return (isSuccess, error_msg)
def convert_from_http_body_encoding_to_local_binary(extracted_file_from_http_body_str):
'''
Extract the file (in binary string format) from event['body'] encoding to local binary encoding.
Args:
extracted_file_from_http_body_str (string): the event['body'] file (in binary string format),.
Returns:
zipfile_binary1 (binary array): the conversion result.
'''
zipfile_binary1 = bytes(extracted_file_from_http_body_str, encoding = "ascii") # convert into a zipfile in binary format
return zipfile_binary1
def extract_zipfile_binary_from_body(body_str):
'''
Extract the zipfile (in binary format) from event['body'] string.
Args:
body_str (string): the event['body'] string.
Returns:
(binary array): the conversion result.
'''
retValue = ""
tmpArray = body_str.split("application/zip") # split the content based on MIME part field data; cut the head
if(len(tmpArray) > 1):
retValue += "entered-Lv1."
tmpArray = tmpArray[1].split("PK") # split the content based on zip file header.
if(len(tmpArray) > 1):
retValue += "entered-Lv2."
zipfile_str = "PK" + 'PK'.join(tmpArray[1:]) # add back the zip file header
tmpArray = zipfile_str.split("------WebKitFormBoundary") # split the content based on MIME part field data; cut the tail
if(len(tmpArray) > 1):
zipfile_str = tmpArray[0]
zipfile_binary = convert_from_http_body_encoding_to_local_binary(zipfile_str)
retValue = zipfile_binary
return retValue
def handler(event, context):
'''Provide an event that contains the following keys:
- operation: one of the operations in the operations dict below
- payload: a parameter to pass to the operation being performed
'''
# set the mapping table for "operation" x "return value"
operations = {
'unzip': lambda x: decompress_zip_file_from_content_in_binary(**x), # unzip an uploaded file
'ping': lambda x: 'pong' # respond to ping req.
}
# because we use "Lambda Proxe", means we have api-gateway forward the whole packet without resolving it for lambda.
event_headers = event["headers"]
operation = event_headers['operation']
event_body = event["body"]
if(operation == 'unzip'):
src_file_in_binary = extract_zipfile_binary_from_body(event_body)
payload_json = {}
payload_json['src_file_in_binary'] = src_file_in_binary
payload_json['dest_dir_path'] = "/tmp/tmp_zipfile_output"
event_headers["payload"] = payload_json
if operation in operations:
responseBody = operations[operation](event_headers.get('payload'))
response = {
"statusCode": 200,
"headers": {
"lambda-response": str(responseBody) # the api-gateway will forward the header to the front end.
},
"body": "",
"isBase64Encoded": False
}
return response
else:
raise ValueError('Unrecognized operation "{}"'.format(operation))
Below is a response from AWS support. LGTM. Leave it here so that people can see the solution to this issue in the future.
=====================Below is the response from AWS support ==================
Hi,
Thank you for contacting AWS Premium Support. I am Jyoti, and I will assist you with this case today.
From the case correspondence, I understand that you are concerned that API Gateway modifies
the binary data payload before proxying to your Lambda function. Please correct me if my understanding is wrong.
Expected Behaviour:
API gateway does modify the binary data payload into UTF-8 encoded JSON strings if
the API is configured at its default settings. Hence this is an expected behaviour.
Kindly note, as per [1], we must configure the API to support binary payloads for
our API in API Gateway. API Gateway can not send binary as is, since it has to send
a JSON body to the lambda proxy. Hence, it encodes the data/payload in UTF-8 by default.
Solution:
In order to overcome the aforementioned challenge, we need to add the desired
binary media types (application/zip in this case) to the binaryMediaTypes list
on the RestApi resource's settings page. For further information on how to achieve
this, please refer here --> [2]. If this property is not defined, the payloads
are handled as UTF-8 encoded JSON strings as mentioned in [1].
This is why the file in your request looks UTF-8 encoded. After configuring the API,
the event received by the Lambda would be a Base64-encoded string.
If you want to conduct operations on this object (the encoded request body or 'event["body"]'),
then you may decode the base64-encoded string to its orginal binary form by following
the below lines (in case of python runtime) :
import base64
coded_string = str(event["body"])
base64.b64decode(coded_string)
Troubleshooting:
I tried to replicate your setup in my environment. Instead of the frontend 'UI' of the application,
I used Postman as a client, while the rest of the setup (API Gateway and Lambda) are similar.
I am making a POST request to my API from Postman, with the request headers 'Content-Type' and 'Accept',
both set to the value 'application/zip', which is the binary media type that is being sent and
also being expected in the response. My API has been configured to support binary media types being
passed in the request body. I have added 'application/zip' in the binaryMediaTypes list for the API.
Finally, in the Lambda function I am decoding the base64-encoded request body (i.e. event["body"])
to its original binary form by using the base64 library (in python).
If you still want to confirm the consistency of your request's form-data out by returning the binary
data in your response, you can refer to the following snippet:
response {
'isBase64Encoded': True, #Ensure the body is base encoded
'statusCode': 200,
'headers': { "Content-Type": "applicaiton/zip" }, #Define the Content-Type
'body': event["body"] #Response Body returns the Base64-encoded value
}
We set the isBase64Encoded parameter to True and API Gateway automatically decodes the
response body depending on the Content-Type (i.e. the binary data/media type) that the
client (in my case Postman), is set to receive (i.e. application/zip). Kindly note, the 'Accept'
header that I had sent in my header, is to validate that the response body contains the binary
data type, the request was made for.
The above response body was the same as the request body binary data that was first sent
through the API, in my setup.
Hope I have addressed your concerns. However, if you still need help with the implementation,
please contact us again and I will be happy to assist you.
References:
=-=-=-=--=-=-=-=-=-=
[1] Support Binary Payloads in API Gateway: https://docs.aws.amazon.com/apigateway/latest/developerguide/api-gateway-payload-encodings.html
[2] Enable Binary Support Using the API Gateway Console: https://docs.aws.amazon.com/apigateway/latest/developerguide/api-gateway-payload-encodings-configure-with-console.html
Best regards,
Jyoti Prakash P.
Amazon Web Services
2019/12/20 update
I realize that my content type is actually multipart instead of application/zip so I modified again the setting then it worked.
Below is the help from AWS support. Many thanks to their help.
Hi,
Thanks a lot for elaborating your application flow and the logs. I understand now that your HTTP Request header 'content-type' is set to 'multipart/form-data'. I agree that for a web form to upload a file it is quite common to set content type as form-data and AWS API Gateway does support it. You would like to know if you could prevent UTF-8 encoding without changing the front end code. Please correct me if my understanding is wrong.
For the ease of discussion, I would like to separate the approach of troubleshooting for the HTTP request and response.
For the request to the API:
Please add 'multipart/form-data' as one of the values in the binaryMediaType list in your "API settings page in the API Gateway console. You would not have to alter your code or HTTP request or any of it's headers. Kindly note to handle binary media/data in API Gateway, the HTTP Request Content-Type header must match the values in binaryMediaType list.
In your use case, if you want to send the binary media back in a response for your request, the HTTP Request 'Content-Type' and 'Accept' headers, the binaryMediaType value of the API and the HTTP Response 'Content-Type' must all be set to 'multipart/form-data'. I tried the above and it worked for me with Postman Client. The 'boundary' directive is set up by Postman automatically if the HTTP Request 'Content-Type' is set to 'multipart/form-data'. Hence, you would have to only add 'multipart/form-data' in the 'binaryMediaType' list. Please have a look at my HTTP request, below:
POST /stg-with-logs HTTP/1.1
Host: <some-api-id>.execute-api.us-east-1.amazonaws.com
Content-Type: multipart/form-data; boundary=----WebKitFormBoundary7MA4YWxkTrZu0gW
Accept: multipart/form-data
Cache-Control: no-cache
Postman-Token: 123b64f9-5669-f794-b9df-34a7561e9708
------WebKitFormBoundary7MA4YWxkTrZu0gW
Content-Disposition: form-data; name="File"; filename="archive.zip"
Content-Type: application/zip
------WebKitFormBoundary7MA4YWxkTrZu0gW--
For the response from the API:
I noticed while going through your API Gateway Logs, the header 'isBase64Encoded' was not set. Kindly set that to true. API Gateway automatically decodes any base64-encoded string in the body of your HTTP response, if 'isBase64Encoded' is set to true. Please have a look at the HTTP Response from my lambda below:
(a6729f56-b245-45a4-9ac4-7e00b23c8957) Endpoint response body before transformations:
{
"isBase64Encoded": true,
"statusCode": 200,
"headers": {
"Content-Type": "multipart/form-data",
"Accpet": "multipart/form-data"
},
"body": "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"
}
Along with this correspondence I am attaching my API Gateway Swagger file and Lambda function code for your reference. The setup worked fine for me and I was able to return the binary payload upon making an HTTP Request. If you want to test it out in your environment, please set the appropriate credentials and lambda uri in the Swagger file.
Hope this addresses your concern. However, if the issue still persists or you have any further questions, please contact us again and I will be happy to assist you.
To see the file named 'binaryPost-stg-with-logs-oas30-apigateway.yaml,python-binary-response.py' included with this correspondence, please use the case link given below the signature.
Best regards,
Jyoti Prakash P.
Amazon Web Services
Check out the AWS Support Knowledge Center, a knowledge base of articles and videos that answer customer questions about AWS services: https://aws.amazon.com/premiumsupport/knowledge-center/?icmpid=support_email_category
We are using JAVA implementation for one of the requests.In the request we are uploading a file. Request doesn't work when HTTPCLIENT4 implementation is selected.The request works fine with previous Jmeter version with HTTPCLIENT3.1 implementation. We need to capture bytes sent in results. How to capture bytes sent through JAVA implementation in HTTP Request sampler
Sent bytes is basically a combination of URL + headers + body so you can calculate it yourself using JSR223 PostProcessor and code like:
def url = sampler.getUrl().toString().length()
def headers = prev.getHeadersSize()
def body = 0;
sampler.getHTTPFiles().each {file ->
body += new File(file.getPath()).length()
}
prev.setSentBytes(url + headers + body)
Where:
sampler is an instance of HTTPSamplerProxy where you can get all files which you're sending with the request
prev is an instance of HTTPSampleResult where you can get URL and Headers and also override Sent Bytes field.
See The Groovy Templates Cheat Sheet for JMeter to see what else you can do with Groovy and how.
I'm trying to create a Data Structure
# Data Structures
## Incorrect Credentials (string)
- `Insufficient privileges.`
This is where it will be used:
+ Response 401 (application/vnd.api+json)
Not allowed.
+ Attributes (object)
+ errors (array[Incorrect Credentials])
This is what it outputs. I'm trying to replace the "Hello, world!" with "Insufficient privileges.".
Just changing the string to enum should solve it. When inheriting from primitive types you can't set a default value.
# API
# A [GET /]
+ Response 401 (application/vnd.api+json)
Not allowed.
+ Attributes (object)
+ errors (array[Incorrect Credentials])
# Data Structures
## Incorrect Credentials (enum)
- Insufficient privileges.
Can I resolve the following scenario by JMeter?:
Send request1
Insert response of request1 to request2
Send request2
Send request3
Compare response3 with responce1
If response3 = responce1 -> setResponseOK() for request2 ELSE Failed
Thanks in advance.
Suggestion to Resolve this situatuion
Send request 1
Capture the response of request 1 in var1(Using correlation-->depending up on your requirement)
Send request 3 first (as you don't need response of request 2 for input of req 3)
Capture response of request 3 in var2
Now hit Request 2 by passing the response of request 1 as an input which is store in var1
Apply beanshell post processor as a child of request 2
Now compare Respose of request 1 and 3 inside if condition:
if matches make the previous request (i.e. Request 2) pass, else fail
if(vars.get("var1").equals(vars.get("var2")))
{
prev.setSuccessful(true);
}
else
{
prev.setSuccessful(false);
}
Design your test as follows:
Request 1
Regular Expression Extractor to extract the whole response into a JMeter Variable configured as:
Reference Name: anything meaningful, i.e. response1
Regular Expression: (?s)(^.*)
Template: $1$
Request 2: use ${response1} where required
Request 3
Response Assertion configured like:
Pattern Matching Rules: Equals
Patterns to Test -> ${response1}
Reference material:
Regular Expressions
How to Use JMeter Assertions in Three Easy Steps
The API we are building allows the user to provide different request bodies. From what I have seen / found so far, it is only possible to provide a single request per action/resource. Is it somehow possible to specify different request bodies?
I tried it like:
### Create new line item for an existing order [POST]
Some description
+ Request (application/json)
### Create new line item and order [POST]
Some other description
+ Request (application/json)
And also like:
### Create new line item for an existing order [POST]
Some description
+ Request (application/json)
+ Request (application/json)
You can have multiple request and responses in a single action. Please see the example below.
### Create new line item for an existing order [POST]
Some description
+ Request (application/json)
+ Response 200
+ Request (application/json)
+ Response 404
+ Request (application/xml)
+ Request (application/json)
+ Response 200