Adding python fuzz script that runs rust tests in a loop and logs to rollbar if it fails

fuzz/README.md

@@ -0,0 +1,10 @@
+This directory sets up running rust fuzz tests. We use python as the orchestration language
+to (1) gather fuzz testing for rust, (2) run individual tests, and (3) log any errors from tests to
+rollbar.
+
+From the `hyperdrive-rs` root directory:
+
+```
+pip install -r fuzz/requirements-fuzz.txt
+ROLLBAR_API_KEY=<rollbar_api_key> python fuzz/run-rust-fuzz.py
+```

fuzz/requirements-fuzz.txt

@@ -0,0 +1 @@
+rollbar

fuzz/run-rust-fuzz.py

@@ -0,0 +1,37 @@
+import os
+import subprocess
+
+import rollbar
+
+# Get rollbar api token from env variables
+token = os.environ['ROLLBAR_API_KEY']
+
+# Set up rollbar
+rollbar.init(
+  access_token=token,
+  environment='rust-fuzz',
+)
+
+# Gather all tests using cargo test
+tests = subprocess.run(["cargo", "test", "fuzz_", "--", "--list"], stdout=subprocess.PIPE, check=True)
+tests = tests.stdout.decode("utf-8").split("\n")
+
+# Remove the final : test from tests
+# The cargo test list also has some printouts that are not tests, so we remove those as well
+tests = [o.split(": test")[0] for o in tests if "::" in o]
+
+# Run forever
+while True:
+    # Loop through tests and run, while capturing failure output
+    for test in tests:
+        print(test)
+        # We don't throw exception if underlying test fails
+        output = subprocess.run(["cargo", "test", test], stdout=subprocess.PIPE, check=False)
+        # If the test failed
+        if output.returncode != 0:
+            print(f"{test} failed")
+            str_output = output.stdout.decode("utf-8")
+            # Log to rollbar
+            rollbar.report_message(f'Rust test {test} failed', 'error', extra_data={'output': str_output})
+
+