har.test

API integration test runner and reporter using .har files

How it works

./main.sh http://your.host
finds queries in all .har files in the folder
which have http://your.host in their target
builds queries and runs them                 -> saves responses -> compares -> reports -> logs failures

*.har         -> ./queries.sh                   ./responses        ./expected             ./log
...other.host...
...your.host...  curl -XPOST ...1 -d '{..1}'    {"a":1}            {"a":1}     Pass       Diff at line: 2. query: curl '...'
...your.host...  curl -XPOST ...2 -d '{..2}'    {"b":1}            {"b":2}     Fail       Response: { "userId": 1, ... }
...your.host...                                                                           Expected: { "userId": 2, ... }
...other.host...

compare does sanitization both on ./responses and ./expected using ./sanitize

Intro

1. Grab your favorite web browser save all the network traffic you have recorded in you current tab; and copy your .har file to this folder.
2. Run ./main.sh https://yourdomain.here.
3. Har.test will filter all the requests made to your domain and replays them in the original order.
4. On the first run har.test will generate the ./expected file based on your .har files; ./expected will be your snapshot but you can edit it; is just simply text.
5. Discover the jq tutorial and the local ./sanitize file here in the repo, which serves not only as an example but har.test actively uses it.
6. Discover the source code main.sh.
7. Tip: Don't forget to check out json.test a similar tool for when you don't have .har files.
8. Give it a star and fork it, create your private clone, add your .gitignore to your needs and link it into your CI flow.

Output

• The output will go to stdout so you could pipe it into a file for example, but there is a more detailed ./log file. Also, the exit code can be 0 if all test passed, or 1 if any of the tests have failed, so is compatible with the standard CI tools.

Goal

• The original goal was to create an easy setup integration test runner and reporter using .har file.

Advantages

• Small set of dependencies: some common GNU commands plus jq,
• ~100 lines of shell script - lightweight codebase,
• and so its easy to fork and hack it to your own needs.

Dependencies

• jq - stedolan.github.io
• curl - you probably have one already
• diff - same here
• awk - again

Limitations that might change in the future

• har.test is running each .har file in alphabetical order syncronously
• also, each test cases are processed syncronously
• there is no HTTP response code checking
• there is no option for any kind of timing

References and similar or other .har tools

• https://github.com/mrichman/hargo
• http://andydude.github.io/harcurl/
• https://har.tech/

Similar tool without .har

• https://github.com/slve/json.test