feat: add windows support

.github/workflows/ci.yml

@@ -53,6 +53,76 @@ jobs:
           just ext docker-stop
           just ext docker-rm
 
+  build-and-test-extension-windows-pg14:
+    runs-on: windows-latest
+
+    steps:
+      - uses: actions/checkout@v4
+      - uses: taiki-e/install-action@just
+
+      - name: Install PostgreSQL
+        uses: ikalnytskyi/action-setup-postgres@v7
+        with:
+          postgres-version: 16
+
+      - name: Configure Passwordless Authentication
+        shell: bash
+        run: |
+          # This is the PGDATA dir used on the previous step.
+          PGDATA="$RUNNER_TEMP/pgdata"
+          # Add trust authentication to pg_hba.conf the extension tests
+          # required this, otherwise they will fail.
+          echo "host all all 127.0.0.1/32 trust" > "$PGDATA/pg_hba.conf"
+          echo "host all all ::1/128 trust" >> "$PGDATA/pg_hba.conf"
+
+          # Restart PostgreSQL to apply changes
+          pg_ctl -D "$PGDATA" restart
+
+      - name: Install pgvector
+        shell: cmd
+        env:
+          PGROOT: "C:/Program Files/PostgreSQL/16"
+        run: |
+          git clone --branch v0.8.0 https://github.com/pgvector/pgvector.git
+          cd pgvector
+          call "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvars64.bat" && ^
+          nmake /NOLOGO /F Makefile.win && ^
+          nmake /NOLOGO /F Makefile.win install
+
+      - name: Create pip cache directory
+        run: |
+          New-Item -Path "C:\Users\runneradmin\AppData\Local\pip\cache" -ItemType Directory -Force
+
+      - name: Set up Python
+        uses: actions/setup-python@v5
+        with:
+          python-version: "3.10"
+          cache: "pip" # caching pip dependencies
+
+      - name: Install test dependencies
+        shell: cmd
+        run: |
+          pip3 install -r projects/extension/requirements-test.txt
+
+      - name: Build and install extension
+        shell: cmd
+        env:
+          PGAI_EXT_SKIP_BUILD_CHECKS: "true"
+        run: |
+          just ext build-install
+
+      - name: Run test server
+        shell: cmd
+        env:
+          PYTHONUTF8: 1
+        run: |
+          start /b python3 -m fastapi dev projects/extension/tests/vectorizer/server.py
+
+      - name: Run test
+        shell: cmd
+        run: |
+          just ext test
+
   build-and-test-pgai:
     runs-on: ubuntu-latest
 

projects/extension/build.py

@@ -1,14 +1,14 @@
 #!/usr/bin/env python3
 import hashlib
-import re
 import os
+import platform
+import re
 import shutil
 import subprocess
 import sys
 import tempfile
 from pathlib import Path
 
-
 HELP = """Available targets:
 - help             displays this message and exits
 - build-install    runs build followed by install
@@ -325,9 +325,10 @@ def build_feature_flags() -> str:
 
 
 def build_sql() -> None:
-    check_versions()
-    check_incremental_sql_files(incremental_sql_files())
-    check_idempotent_sql_files(idempotent_sql_files())
+    if not os.environ.get("PGAI_EXT_SKIP_BUILD_CHECKS", "").lower() == "true":
+        check_versions()
+        check_incremental_sql_files(incremental_sql_files())
+        check_idempotent_sql_files(idempotent_sql_files())
     build_control_file()
     hr = "".rjust(80, "-")  # "horizontal rule"
     osf = output_sql_file()
@@ -375,34 +376,51 @@ def clean_sql() -> None:
 
 
 def postgres_bin_dir() -> Path:
-    bin_dir = os.getenv("PG_BIN")
-    if bin_dir:
-        return Path(bin_dir).resolve()
-    else:
-        bin_dir = Path(f"/usr/lib/postgresql/{pg_major()}/bin")
+    bin_dir_env = os.getenv("PG_BIN")
+    if bin_dir_env:
+        bin_dir = Path(bin_dir_env)
         if bin_dir.is_dir():
-            return bin_dir.resolve()
-        else:
-            p = shutil.which("pg_config")
-            if not p:
-                fatal("pg_config not found")
-            return Path(p).parent.resolve()
+            dir = bin_dir.resolve()
+            print(f"pg bin dir: using PG_BIN environment variable {dir}")
+            return dir
+
+    bin_dir = Path(f"/usr/lib/postgresql/{pg_major()}/bin")
+    if platform.system() == "Windows":
+        program_files = os.environ.get("ProgramFiles", r"C:\Program Files")
+        bin_dir = Path(f"{program_files}\\PostgreSQL\\{pg_major()}\\bin")
+    if bin_dir.is_dir():
+        dir = bin_dir.resolve()
+        print(f"pg bin dir: using default postgres bin directory {dir}")
+        return dir
+    else:
+        p = shutil.which("pg_config")
+        if p is None:
+            fatal("pg_config not found")
+        dir = Path(p).parent.resolve()
+        print(f"pg bin dir: parent directory of pg_config {dir}")
+        return dir
 
 
 def pg_config() -> Path:
-    return postgres_bin_dir() / "pg_config"
+    cmd = "pg_config"
+    if platform.system() == "Windows":
+        cmd = "pg_config.exe"
+    return postgres_bin_dir() / cmd
 
 
 def extension_install_dir() -> Path:
-    proc = subprocess.run(
-        f"{pg_config()} --sharedir",
-        check=True,
-        shell=True,
-        env=os.environ,
-        text=True,
-        capture_output=True,
-    )
-    return Path(str(proc.stdout).strip()).resolve() / "extension"
+    try:
+        proc = subprocess.run(
+            [pg_config(), "--sharedir"],
+            check=True,
+            env=os.environ,
+            text=True,
+            capture_output=True,
+        )
+        return Path(str(proc.stdout).strip()).resolve() / "extension"
+    except subprocess.CalledProcessError as e:
+        print(f"Error running pg_config: {e.stderr}", file=sys.stderr)
+        raise e
 
 
 def install_sql() -> None:
@@ -437,9 +455,12 @@ def python_install_dir() -> Path:
     # don't do it. i'm warning you
     # seriously.
     # you'll wreck old versions. look at build_idempotent_sql_file()
-    return Path(
-        "/usr/local/lib/pgai"
-    ).resolve()  # CONTROLS WHERE THE PYTHON LIB AND DEPS ARE INSTALLED
+    if platform.system() == "Windows":
+        program_files = Path(os.environ.get("ProgramFiles", r"C:\Program Files"))
+        return (program_files / "pgai" / "lib").resolve()
+    else:
+        # Use the existing Unix-like directory structure
+        return Path("/usr/local/lib/pgai").resolve()
 
 
 def install_old_py_deps() -> None:
@@ -468,25 +489,25 @@ def install_prior_py() -> None:
             fatal(f"'{os.sep}' in version {version}. this is not supported")
         version_target_dir = python_install_dir().joinpath(version)
         if version_target_dir.exists():
+            print(f"Version {version} already installed in {version_target_dir}")
             continue
-        tmp_dir = Path(tempfile.gettempdir()).joinpath("pgai", version)
-        tmp_dir.mkdir(parents=True, exist_ok=True)
-        branch = git_tag(version)
-        subprocess.run(
-            f"git clone https://github.com/timescale/pgai.git --branch {branch} {tmp_dir}",
-            shell=True,
-            check=True,
-            env=os.environ,
-        )
-        tmp_src_dir = tmp_dir.joinpath("projects", "extension").resolve()
-        subprocess.run(
-            f'pip3 install -v --compile -t "{version_target_dir}" "{tmp_src_dir}"',
-            check=True,
-            shell=True,
-            env=os.environ,
-            cwd=str(tmp_src_dir),
-        )
-        shutil.rmtree(tmp_dir)
+
+        with tempfile.TemporaryDirectory() as tmp_dir:
+            branch = git_tag(version)
+            subprocess.run(
+                f"git clone https://github.com/timescale/pgai.git --branch {branch} {tmp_dir}",
+                shell=True,
+                check=True,
+                env=os.environ,
+            )
+            tmp_src_dir = Path(tmp_dir).joinpath("projects", "extension").resolve()
+            subprocess.run(
+                f'pip3 install -v --compile -t "{version_target_dir}" "{tmp_src_dir}"',
+                check=True,
+                shell=True,
+                env=os.environ,
+                cwd=str(tmp_src_dir),
+            )
 
 
 def build_init_py() -> None:

projects/extension/justfile

@@ -3,7 +3,7 @@ PG_BIN := "/usr/lib/postgresql/" + PG_MAJOR + "/bin"
 
 # Show list of recipes
 default:
-    @just --list
+  @just --list
 
 ci: docker-build docker-run
   #!/usr/bin/env bash

projects/extension/tests/vectorizer/db.py

@@ -0,0 +1,2 @@
+def db_url(user: str) -> str:
+    return f"postgres://{user}@127.0.0.1:5432/test"

projects/extension/tests/vectorizer/test_chunking.py

@@ -3,16 +3,14 @@
 import psycopg
 import pytest
 
+from .db import db_url
+
 # skip tests in this module if disabled
 enable_vectorizer_tests = os.getenv("ENABLE_VECTORIZER_TESTS")
 if enable_vectorizer_tests == "0":
     pytest.skip(allow_module_level=True)
 
 
-def db_url(user: str) -> str:
-    return f"postgres://{user}@127.0.0.1:5432/test"
-
-
 def test_chunking_character_text_splitter():
     tests = [
         (

projects/extension/tests/vectorizer/test_embedding.py

@@ -2,17 +2,14 @@
 
 import psycopg
 import pytest
+from .db import db_url
 
 # skip tests in this module if disabled
 enable_vectorizer_tests = os.getenv("ENABLE_VECTORIZER_TESTS")
 if enable_vectorizer_tests == "0":
     pytest.skip(allow_module_level=True)
 
 
-def db_url(user: str) -> str:
-    return f"postgres://{user}@127.0.0.1:5432/test"
-
-
 def test_embedding_openai():
     tests = [
         (

projects/extension/tests/vectorizer/test_formatting.py

@@ -3,16 +3,14 @@
 import psycopg
 import pytest
 
+from .db import db_url
+
 # skip tests in this module if disabled
 enable_vectorizer_tests = os.getenv("ENABLE_VECTORIZER_TESTS")
 if enable_vectorizer_tests == "0":
     pytest.skip(allow_module_level=True)
 
 
-def db_url(user: str) -> str:
-    return f"postgres://{user}@127.0.0.1:5432/test"
-
-
 def test_formatting_python_template():
     tests = [
         (

projects/extension/tests/vectorizer/test_grants.py

@@ -3,16 +3,14 @@
 import psycopg
 import pytest
 
+from .db import db_url
+
 # skip tests in this module if disabled
 enable_vectorizer_tests = os.getenv("ENABLE_VECTORIZER_TESTS")
 if enable_vectorizer_tests == "0":
     pytest.skip(allow_module_level=True)
 
 
-def db_url(user: str) -> str:
-    return f"postgres://{user}@127.0.0.1:5432/test"
-
-
 def test_grant_to():
     tests = [
         (

projects/extension/tests/vectorizer/test_indexing.py

@@ -3,16 +3,15 @@
 import psycopg
 import pytest
 
+
+from .db import db_url
+
 # skip tests in this module if disabled
 enable_vectorizer_tests = os.getenv("ENABLE_VECTORIZER_TESTS")
 if enable_vectorizer_tests == "0":
     pytest.skip(allow_module_level=True)
 
 
-def db_url(user: str) -> str:
-    return f"postgres://{user}@127.0.0.1:5432/test"
-
-
 def test_indexing_none():
     tests = [
         (

projects/extension/tests/vectorizer/test_processing.py

@@ -3,16 +3,14 @@
 import psycopg
 import pytest
 
+from .db import db_url
+
 # skip tests in this module if disabled
 enable_vectorizer_tests = os.getenv("ENABLE_VECTORIZER_TESTS")
 if enable_vectorizer_tests == "0":
     pytest.skip(allow_module_level=True)
 
 
-def db_url(user: str) -> str:
-    return f"postgres://{user}@127.0.0.1:5432/test"
-
-
 def test_processing_default():
     tests = [
         (

projects/extension/tests/vectorizer/test_scheduling.py

@@ -3,16 +3,14 @@
 import psycopg
 import pytest
 
+from .db import db_url
+
 # skip tests in this module if disabled
 enable_vectorizer_tests = os.getenv("ENABLE_VECTORIZER_TESTS")
 if enable_vectorizer_tests == "0":
     pytest.skip(allow_module_level=True)
 
 
-def db_url(user: str) -> str:
-    return f"postgres://{user}@127.0.0.1:5432/test"
-
-
 def test_scheduling_none():
     tests = [
         (