Example Lambda to illustrate concepts in lambda blog post

example-lambda.py

import hashlib
import flask
import boto3
import time

## CONFIG ##
BUCKET = 'evan-lambda-longform'
FILEPATH = 'hashes'
SEP = '\t'
metric_namespace = 'hasher.'
app = flask.Flask(__name__)

def log(metric, mtype='count', value=1, tags=[]):
    # format:
    # MONITORING|unix_epoch_timestamp|metric_value|metric_type|my.metric.name|#tag1:value,tag2
    print "MONITORING|{}|{}|{}|{}|#{}".format(int(time.time()), value, mtype, metric, ', '.join(tags))


def read_s3(bucket=BUCKET, path=FILEPATH):
    client = boto3.client('s3')
    resp = client.get_object(Bucket=bucket, Key=path)
    resp = resp['Body'].read()
    log(metric=metric_namespace + 'db_reads', tags=['hash-service'])
    return resp


def hash_exists(hash, address=False):
    resp = read_s3()
    resp = resp.split('\n')
    # Obviously not an ideal solution as hash db grows
    for each in resp:
        if each.split(SEP)[0] == hash:
            log(metric=metric_namespace + 'hashes_found', tags=['hash-service'])
            if address:  # True if we want to return the hash
                return each.split(SEP)[0]
            # Otherwise return the original (unhashed) string
            return each.split(SEP)[1]
    return False


@app.route('/messages/', methods=['POST', 'GET'])
@app.route('/messages/<hash>', methods=['GET'])
def main(hash=None, address=None):
    log(metric=metric_namespace + 'requests', tags=['hash-service'])
    if hash:
        # Check if we saw it before
        if hash_exists(hash):
            log(metric=metric_namespace + 'hashes_retrieved', tags=['hash-service'])
            log(metric=metric_namespace + 'responses', tags=['hash-service', 'status:200'])
            return hash_exists(hash)
        else:
            log(metric=metric_namespace + 'hashes_not_found', tags=['hash-service'])
            log(metric=metric_namespace + 'responses', tags=['hash-service', 'status:404'])
            flask.abort(404)

    elif flask.request.data:
        req = flask.request.get_json()
        message = req.get('message', None)
        if message:
            # log payload size
            payload_size = len(message)
            log(metric=metric_namespace + 'payload_size',
                   value=payload_size, tags=['hash-service'])
            # Store hash and return
            hash = hashlib.sha256(message).hexdigest()
            log(metric=metric_namespace + 'strings_received', tags=['hash-service'])

            if hash_exists(hash):
                # Search through the hashes to find it
                # address = true because we want to match on the hash
                log(metric=metric_namespace + 'responses', tags=['hash-service', 'status:200'])
                return hash_exists(hash, address=True)
            else:
                data = read_s3()
                data = "{}\n{}{}{}".format(data, hash, SEP, message)
                client = boto3.client('s3')
                client.put_object(Bucket=BUCKET, Key=FILEPATH, Body=data)
                log(metric=metric_namespace + 'response.200', tags=['hash-service'])
                return hash
        else:
            log(metric=metric_namespace + 'malformed_payloads', tags=['hash-service'])
            log(metric=metric_namespace + 'responses', tags=['hash-service', 'status:400'])
            flask.abort(400)

    else:
        log(metric='evan.hashes_not_found', tags=['hash-service'])
        log(metric=metric_namespace + 'responses', tags=['hash-service', 'status:200'])
        return "Nothing to see here"


if __name__ == "__main__":
    app.run()