Unit testing CI (#36)

**Summary**: Made GitHub Action which installs dependencies and runs all
unit tests in the repository.

**Demo**:
A successful run. You can view it
[here](https://github.com/cmu-db/dbgym/actions/runs/9831741840/job/27139551785).
![Screenshot 2024-07-07 at 17 56
58](https://github.com/cmu-db/dbgym/assets/20631215/c2613191-49eb-4d86-b0c2-a59a66e899c5)

**Details**:
* Also fixed some benign errors in the unit tests. Some paths were
previously relative and had to be changed to be absolute. I also started
counting index names from 0 instead of 1, which led to another test
failure.

Author: wangpatrick57

.github/workflows/sanity_test.yml

@@ -0,0 +1,35 @@
+name: Unit Tests
+
+on:
+  push: {}
+  pull_request:
+    branches: [main]
+
+jobs:
+  build:
+    runs-on: self-hosted
+
+    steps:
+    - uses: actions/checkout@v2
+
+    - name: Set up Python
+      uses: actions/setup-python@v4
+      with:
+        python-version: '3.10'
+
+    # We could choose to set up dependencies manually in the GHA runner instead of installing them during the GHA.
+    #
+    # However, I think it's better to do them in the GHA itself so that we're testing our dependency installation step
+    # in addition to our actual code. It also removes the need to manually reinstall dependencies on the GHA runners
+    # every time we add a new dependency.
+    #
+    # Note that the GHA runners are stateful. Dependencies installed from previous runs will still be on the runner.
+    # This means this step will usually be pretty fast as most dependencies will already be cached. However, it also
+    # means that past runs might interfere with the current run, so you sometimes may need to restart the GHA runners.
+    - name: Install dependencies
+      run: |
+        ./dependencies/install_dependencies.sh
+        . "$HOME/.cargo/env"
+
+    - name: Run unit tests
+      run: python scripts/run_unittests.py

.github/workflows/unittest_ci.yml

@@ -1,5 +1,5 @@
 #!/bin/bash
 # You may want to create a conda environment before doing this
-pip install -r dependency/requirements.txt
-cat dependency/apt_requirements.txt | xargs sudo apt-get install -y
-curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh
+pip install -r dependencies/requirements.txt
+cat dependencies/apt_requirements.txt | xargs sudo apt-get install -y
+curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- -y

dependencies/install_dependencies.sh

@@ -1,8 +1,11 @@
 import unittest
+import sys
 
 if __name__ == "__main__":
     loader = unittest.TestLoader()
     suite = loader.discover(".")
-    print(f"suite={suite}")
     runner = unittest.TextTestRunner()
-    runner.run(suite)
+    result = runner.run(suite)
+    if not result.wasSuccessful():
+        # This is needed so that the GHA fails if the unit tests fail.
+        sys.exit(1)

dependencies/rust.sh

@@ -223,7 +223,6 @@ def __init__(
         workload_timeout_penalty: float = 1.0,
         logger: Optional[Logger] = None,
     ) -> None:
-
         self.dbgym_cfg = dbgym_cfg
         self.workload_path = workload_path
         # Whether we should use benchbase or not.
@@ -257,7 +256,7 @@ def __init__(
             sqls = [
                 (
                     line.split(",")[0],
-                    Path(line.split(",")[1]),
+                    self.workload_path / Path(line.split(",")[1]),
                     1.0,
                 )
                 for line in lines
@@ -271,7 +270,7 @@ def __init__(
                 sqls = [
                     (
                         split[0],
-                        Path(split[1]),
+                        self.workload_path / Path(split[1]),
                         float(split[2]),
                     )
                     for split in splits

scripts/run_unittests.py

@@ -11,7 +11,7 @@
 class IndexSpaceTests(unittest.TestCase):
     @staticmethod
     def load(
-        config_path=Path("tune/protox/tests/unittest_benchmark_configs/unittest_tpch.yaml"),
+        config_path=Path("tune/protox/tests/unittest_benchmark_configs/unittest_tpch.yaml").resolve(),
         aux_type=True,
         aux_include=True,
     ):
@@ -27,7 +27,7 @@ def load(
             tables=benchmark_config["tables"],
             attributes=benchmark_config["attributes"],
             query_spec=benchmark_config["query_spec"],
-            workload_path=Path("tune/protox/tests/unittest_tpch_dir"),
+            workload_path=Path("tune/protox/tests/unittest_tpch_dir").resolve(),
             pid=None,
             workload_timeout=0,
             workload_timeout_penalty=1.0,

tune/protox/env/workload.py

@@ -98,7 +98,7 @@ def test_ia(self):
         IndexAction.index_counter = 0
         self.assertEqual(
             ia1.sql(add=True),
-            "CREATE INDEX  index1 ON tbl USING btree (a,b,c) INCLUDE (d,e)",
+            "CREATE INDEX  index0 ON tbl USING btree (a,b,c) INCLUDE (d,e)",
         )
 
         ia2 = IndexAction(

tune/protox/tests/test_index_space.py

@@ -56,7 +56,7 @@ def test_tpch(self):
                 for k, v in ref.items()
             }
 
-        w, i = WorkloadTests.load("tune/protox/tests/unittest_benchmark_configs/unittest_tpch.yaml", Path("tune/protox/tests/unittest_tpch_dir"))
+        w, i = WorkloadTests.load("tune/protox/tests/unittest_benchmark_configs/unittest_tpch.yaml", Path("tune/protox/tests/unittest_tpch_dir").resolve())
         self.assertEqual(i.class_mapping, ref)
 
     def test_job(self):
@@ -68,7 +68,7 @@ def test_job(self):
                 for k, v in ref.items()
             }
 
-        w, i = WorkloadTests.load("tune/protox/tests/unittest_benchmark_configs/unittest_job_full.yaml", Path("tune/protox/tests/unittest_job_full_dir"))
+        w, i = WorkloadTests.load("tune/protox/tests/unittest_benchmark_configs/unittest_job_full.yaml", Path("tune/protox/tests/unittest_job_full_dir").resolve())
         self.assertEqual(i.class_mapping, ref)
 
     def test_dsb(self):
@@ -80,7 +80,7 @@ def test_dsb(self):
                 for k, v in ref.items()
             }
 
-        w, i = WorkloadTests.load("tune/protox/tests/unittest_benchmark_configs/unittest_dsb.yaml", Path("tune/protox/tests/unittest_dsb_dir"))
+        w, i = WorkloadTests.load("tune/protox/tests/unittest_benchmark_configs/unittest_dsb.yaml", Path("tune/protox/tests/unittest_dsb_dir").resolve())
         self.diff_classmapping(ref, i.class_mapping)
 
     def test_tpcc(self):
@@ -92,5 +92,5 @@ def test_tpcc(self):
                 for k, v in ref.items()
             }
 
-        w, i = WorkloadTests.load("tune/protox/tests/unittest_benchmark_configs/unittest_tpcc.yaml", Path("tune/protox/tests/unittest_tpcc_dir"))
+        w, i = WorkloadTests.load("tune/protox/tests/unittest_benchmark_configs/unittest_tpcc.yaml", Path("tune/protox/tests/unittest_tpcc_dir").resolve())
         self.assertEqual(i.class_mapping, ref)