AWSM-CLI

Combines the power of Unix pipes and the official AWS awscli cli tool on the command line, plus more. Simply written in Bash.

awsm-cli is written and maintained by Stephen Chu ([email protected]).

Mailing list: https://groups.google.com/forum/#!forum/awsm-cli-users

Basic Usage

# Find ec2 instances within all us-west-2 vpcs named 'production' alike:

$ awsm ec2 describe-vpcs -r us-west-2 | grep -w production | awsm ec2 describe-instances
Region     tag:Name  InstanceId           AvailabilityZone  InstanceType  State    PublicIpAddress  PrivateIpAddress  PrivateDnsName                              VpcId         ImageId       LaunchTime                
us-west-2  Foo       i-015ef79784b536dbe  us-west-2a        r3.xlarge     running  n/a              10.42.17.213      ip-10-42-17-213.us-west-2.compute.internal  vpc-d85397bc  ami-1411c474  2016-08-24T20:39:01.000Z

# Count all the instance types of all us-west-2 EC2 instances:

$ awsm ec2 describe-instances -r us-west-2 | tail -n +2 | awk -F $'\t' '{ print $5 }' | sort | uniq -c
    14 c4.large
     3 c4.xlarge
     2 i2.2xlarge
     1 m3.2xlarge

Installation

Dependencies

First you will have to install these yourself, as different versions are available/supported on different Linux distros.

• awscli >= 1.10.63
• jq >= 1.6
• GNU awk >= 4.1.1
• GNU parallel >= 20160922

Tip: Run awsm _ dependencies-check to ensure you have a working environment:

$ awsm _ dependencies-check
[INFO] Checking for dependencies used by awsm-cli...
[INFO] Your awscli is working properly.
[INFO] Your jq is working properly.
[INFO] Your GNU awk is working properly.
[INFO] Your GNU parallel is working properly.
[INFO] Summary: Awesome! All required dependencies are installed correctly. Enjoy awsm-cli!

Bash

$ bash -c 'export DIR=$(mktemp --directory); cd $DIR && git clone --recursive https://github.com/stephenchu/awsm-cli.git .; mv files/* /usr/local/bin/'

For shell auto completion, see AUTOCOMPLETE.md here.

Problems awsm-cli Aims to Solve Over the AWS Official awscli

1. Defaults to human-readable, line-based table format instead of JSON

JSON is great for machine-to-machine parseability and is reasonably humanly readable, but it becomes unreadable quickly when the amount of JSON text exceeds a few pages worth. Line-based, tabular format for each JSON entities allow compact information to be presented to human, at the expense of completeness, is more readable.

awsm-cli parses it for you and presents to you a readable, line-based table format.

2. Changing its default output format is still humanly unreadable

The awscli option --output table yields an ugly table border. Using --output text yields a difficult-to-parse multi-line textual output as well.

With awsm-cli, you append --jq '.' (a "dot" means everything in jq) to swiftly change from tabular format to JSON format that the human-readable results are derived from, plus any JSON filtering/transforming you wish to alter your results using the powerful features of Jq:

$ awsm ec2 describe-instances -r us-west-2 --jq '[.] | map(select(.InstanceType == "t2.small"))'

Each line of output in awsm-cli is one entity. Plain simple and expected. But if you need to, displaying its representing JSON is only one command line option away.

3. Traversing AWS resource object graph involves repetitive conditional logic

Common Laborious Pattern:

vpc_ids=$(aws ec2 --region us-west-2 describe-vpcs --output text --query 'Vpcs[*].VpcId')
if [ ! -z "$vpc_ids" ]; then
  aws ec2 --region us-west-2 describe-instances ... --filters Name=vpc-id,Values=$(tr $'\t' ',' <<< "$vpc_ids")
else
  aws ec2 --region us-west-2 describe-instances ...
fi

Pattern Explained:

1. Find some AWS resource identifiers. Make sure you know your --query JSON hierachy. Store them in a variable.
2. Do a if condition check for the above variable, becuase you then need to conditionally put the --filters clause.
3. Conditionally execute your subsequent awscli command with the filtering clause containing (or not) your pre-fetched values (in CSV).
4. Don't forget you also have to customize your --output, or pick out relevant information you need. Again remember your JSON hierarchy.

Now this example is a simple relationship of VPC -> instances. Try: VPC -> subnets -> autoscaling groups -> instances.

Instead, awsm-cli allows you to just do:

$ awsm ec2 describe-vpcs -r us-west-2 | awsm ec2 describe-instances

awsm-cli uses Unix pipes to intuitively traverse AWS resource relationships from one to another. You can even combine other command line tools that works with stdin/stdout (e.g. grep, sed, awk, etc.) to avoid writing complicated shell scripts to get what you want.

Unix pipes is simply the most intuitive and error-free way in command line to avoid writing shell scripts imperatively to get what you want.

Features

• Uses Unix pipes to walk the AWS resources relationships
• All subcommands support option --jq '.' to change from textual outputs to their native JSON outputs via the more powerful JQ filters.
• All subcommands support printing the underlying native awscli executed commands via --log-aws-cli
• All subcommands support printing the underlying jq filter used via --log-jq
• All region-specific subcommands support displaying multiple regions' worth of data via --region "us-west-1 us-west-2", something that takes multiple page loads on the AWS web console
• Nicely align any awsm-cli output using column by piping any stdout into | awsm _ column.
• Supports multi-column sort piping any stdout into | awsm _ sort -k 2,2 -k 3,3.

Supported AWS Subcommands

Look in files/awsm-cli/*.sh, or, better yet, run:

$ awsm _ subcommands
_
autoscaling
cloudformation
ec2
.
.
.

$ awsm _ subcommand-actions --subcommand ec2
describe-availability-zones
describe-images
describe-instances
describe-instance-status
describe-regions
.
.
.

Or use shell autocomplete to discover. See AUTOSCOMPLETE.md.

How It Works

See DESIGN.md.

Contribute

Yes, please, thank you! See CONTRIBUTE.md.

About the Author

Stephen Chu ([email protected])

A software developer, who had done a little bit of Ruby, Python, Java, C#, JRuby, Bash, and he played a little with Rails, Postgres, MySql, Oracle, Terraform, Packer, Cassandra, Kafka, ZooKeeper, Docker, and AWS. He specializes in writing low maintenance-cost software.