Update sqlglot to V2 (#380)

In the current sqlglot normalized, conjunction string would be modified to its normalized form after the first iteration. And for the remaining iterations, normalization function was applied to the normalized form of the original conjunction string. Therefore, the performance of the first iteration would be different to the remaining iterations. This causes unsuitability in performance reports. To fix this issue, we created the V2 version, in which, a new conjunction string is created for every individual iteration.
python · Feb 4, 2025 · 727a0aa · 727a0aa
1 parent 15a82c9
commit 727a0aa
Show file tree

Hide file tree

Showing 7 changed files with 13 additions and 13 deletions.
diff --git a/pyperformance/data-files/benchmarks/MANIFEST b/pyperformance/data-files/benchmarks/MANIFEST
@@ -83,10 +83,10 @@ spectral_norm	<local>
 sphinx	<local>
 sqlalchemy_declarative	<local>
 sqlalchemy_imperative	<local>
-sqlglot	<local>
-sqlglot_parse	<local:sqlglot>
-sqlglot_transpile	<local:sqlglot>
-sqlglot_optimize	<local:sqlglot>
+sqlglot_v2	<local>
+sqlglot_v2_parse	<local:sqlglot_v2>
+sqlglot_v2_transpile	<local:sqlglot_v2>
+sqlglot_v2_optimize	<local:sqlglot_v2>
 sqlite_synth	<local>
 sympy	<local>
 telco	<local>

diff --git a/...marks/bm_sqlglot/bm_sqlglot_optimize.toml → ...bm_sqlglot_v2/bm_sqlglot_v2_optimize.toml b/...marks/bm_sqlglot/bm_sqlglot_optimize.toml → ...bm_sqlglot_v2/bm_sqlglot_v2_optimize.toml
@@ -1,3 +1,3 @@
 [tool.pyperformance]
-name = "sqlglot_optimize"
+name = "sqlglot_v2_optimize"
 extra_opts = ["optimize"]
diff --git a/...nchmarks/bm_sqlglot/bm_sqlglot_parse.toml → ...ks/bm_sqlglot_v2/bm_sqlglot_v2_parse.toml b/...nchmarks/bm_sqlglot/bm_sqlglot_parse.toml → ...ks/bm_sqlglot_v2/bm_sqlglot_v2_parse.toml
@@ -1,3 +1,3 @@
 [tool.pyperformance]
-name = "sqlglot_parse"
+name = "sqlglot_v2_parse"
 extra_opts = ["parse"]
diff --git a/...arks/bm_sqlglot/bm_sqlglot_transpile.toml → ...m_sqlglot_v2/bm_sqlglot_v2_transpile.toml b/...arks/bm_sqlglot/bm_sqlglot_transpile.toml → ...m_sqlglot_v2/bm_sqlglot_v2_transpile.toml
@@ -1,3 +1,3 @@
 [tool.pyperformance]
-name = "sqlglot_transpile"
+name = "sqlglot_v2_transpile"
 extra_opts = ["transpile"]
diff --git a/...iles/benchmarks/bm_sqlglot/pyproject.toml → ...s/benchmarks/bm_sqlglot_v2/pyproject.toml b/...iles/benchmarks/bm_sqlglot/pyproject.toml → ...s/benchmarks/bm_sqlglot_v2/pyproject.toml
@@ -1,13 +1,13 @@
 [project]
-name = "pyperformance_bm_sqlglot"
+name = "pyperformance_bm_sqlglot_v2"
 requires-python = ">=3.7"
 dependencies = [
     "pyperf",
-    "sqlglot",
+    "sqlglot_v2",
 ]
 urls = {repository = "https://github.com/python/pyperformance"}
 dynamic = ["version"]
 
 [tool.pyperformance]
-name = "sqlglot"
+name = "sqlglot_v2"
 extra_opts = ["normalize"]
diff --git a/...es/benchmarks/bm_sqlglot/requirements.txt → ...benchmarks/bm_sqlglot_v2/requirements.txt b/...es/benchmarks/bm_sqlglot/requirements.txt → ...benchmarks/bm_sqlglot_v2/requirements.txt
diff --git a/...es/benchmarks/bm_sqlglot/run_benchmark.py → ...benchmarks/bm_sqlglot_v2/run_benchmark.py b/...es/benchmarks/bm_sqlglot/run_benchmark.py → ...benchmarks/bm_sqlglot_v2/run_benchmark.py
@@ -156,8 +156,8 @@ def bench_optimize(loops):
 
 def bench_normalize(loops):
     elapsed = 0
-    conjunction = parse_one("(A AND B) OR (C AND D) OR (E AND F) OR (G AND H)")
     for _ in range(loops):
+        conjunction = parse_one("(A AND B) OR (C AND D) OR (E AND F) OR (G AND H)")
         t0 = pyperf.perf_counter()
         normalize.normalize(conjunction)
         elapsed += pyperf.perf_counter() - t0
@@ -186,9 +186,9 @@ def add_parser_args(parser):
 
 if __name__ == "__main__":
     runner = pyperf.Runner(add_cmdline_args=add_cmdline_args)
-    runner.metadata['description'] = "SQLGlot benchmark"
+    runner.metadata['description'] = "SQLGlot V2 benchmark"
     add_parser_args(runner.argparser)
     args = runner.parse_args()
     benchmark = args.benchmark
 
-    runner.bench_time_func(f"sqlglot_{benchmark}", BENCHMARKS[benchmark])
+    runner.bench_time_func(f"sqlglot_v2_{benchmark}", BENCHMARKS[benchmark])