Add backend support for PostgreSQL

Author: amotl

.github/workflows/main.yml

@@ -37,6 +37,13 @@ jobs:
       max-parallel: 4
       matrix:
         python-version: [3.7, 3.8, 3.9, '3.10', '3.11']
+    services:
+      postgresql:
+        image: postgres:16
+        ports:
+          - 5432:5432
+        env:
+          POSTGRES_HOST_AUTH_METHOD: trust
     steps:
       - uses: actions/checkout@v3
       - name: Set up Python ${{ matrix.python-version }}
@@ -66,6 +73,13 @@ jobs:
           ls -l dist
   documentation:
     runs-on: ubuntu-latest
+    services:
+      postgresql:
+        image: postgres:16
+        ports:
+          - 5432:5432
+        env:
+          POSTGRES_HOST_AUTH_METHOD: trust
     steps:
       - uses: actions/checkout@v3
       - name: Setup python

README.rst

@@ -79,6 +79,28 @@ Usage
   Please access http://localhost:8000/search.html
 
 
+Development
+===========
+
+Install package in development mode::
+
+    pip install --editable='.[cli,docs,test]' --prefer-binary
+
+Start PostgreSQL server::
+
+    docker run --rm -it --publish=5432:5432 --env "POSTGRES_HOST_AUTH_METHOD=trust" postgres:16 postgres -c log_statement=all
+
+Invoke software tests::
+
+    export POSTGRES_LOG_STATEMENT=all
+    pytest -vvv
+
+Invoke linters::
+
+    pip install pre-commit
+    pre-commit run --all-files
+
+
 .. _atsphinx-sqlite3fts: https://pypi.org/project/atsphinx-sqlite3fts/
 .. _Kazuya Takei: https://github.com/attakei
 .. _readthedocs-sphinx-search: https://github.com/readthedocs/readthedocs-sphinx-search

docs/conf.py

@@ -32,6 +32,7 @@
 }
 # atsphinx-sqlite3fts
 sqlite3fts_use_search_html = True
+sqlite3fts_database_url = "postgresql://postgres@localhost:5432"
 
 
 def setup(app):  # noqa: D103

docs/getting-started.rst

@@ -47,8 +47,11 @@ You can build database by ``sqlite`` builder.
 
 .. code-block:: console
 
-   make sqlite
-   sqlite3 _build/sqlite/db.sqlite
+   make fts-index
+
+.. code-block:: console
+
+   psql postgresql://postgres@localhost:5432/ --command 'SELECT * FROM document;'
 
 .. code-block:: sqlite3
 

pyproject.toml

@@ -35,6 +35,7 @@ dynamic = ["version", "description"]
 dependencies = [
   "docutils",
   "peewee",
+  "psycopg2[binary]",
   "sphinx<7",
 ]
 

src/atsphinx/sqlite3fts/__init__.py

@@ -1,4 +1,4 @@
-"""Sphinx document searcher using SQLite3."""
+"""Sphinx document searcher using SQL database."""
 from sphinx.application import Sphinx
 
 from . import builders, events
@@ -10,9 +10,10 @@ def setup(app: Sphinx):
     """Entrypoint as Sphinx extension."""
     app.add_config_value("sqlite3fts_exclude_pages", [], "env")
     app.add_config_value("sqlite3fts_use_search_html", False, "env")
-    app.add_builder(builders.SqliteBuilder)
+    app.add_config_value("sqlite3fts_database_url", None, "env")
+    app.add_builder(builders.FtsIndexer)
     app.connect("config-inited", events.setup_search_html)
-    app.connect("builder-inited", events.configure_database)
+    app.connect("config-inited", events.configure_database)
     app.connect("html-page-context", events.register_document)
     app.connect("build-finished", events.save_database)
     return {

src/atsphinx/sqlite3fts/builders.py

@@ -7,13 +7,14 @@
 from . import models, services
 
 
-class SqliteBuilder(Builder):
-    """Single database generation builder.
+class FtsIndexer(Builder):
+    """
+    Fulltext index builder for databases.
 
-    This is custom builder to generate only SQLite database file
+    A custom builder to generate fulltext indexes, stored in SQL databases.
     """
 
-    name = "sqlite"
+    name = "fts-index"
     allow_parallel = True
 
     def get_target_uri(self, docname: str, typ: str = None) -> str:  # noqa: D102
@@ -23,7 +24,11 @@ def get_outdated_docs(self) -> str:  # noqa: D102
         return "db.sqlite"
 
     def prepare_writing(self, docnames: Set[str]) -> None:  # noqa: D102
-        pass
+        from atsphinx.sqlite3fts.models import Content, Document, Section
+
+        Document.truncate_table(cascade=True)
+        Section.truncate_table(cascade=True)
+        Content.truncate_table(cascade=True)
 
     def write_doc(self, docname: str, doctree: nodes.document) -> None:
         """Register content of document into database.

src/atsphinx/sqlite3fts/events.py

@@ -33,12 +33,23 @@ def _generate_search_html(app: Sphinx):
     app.connect("html-collect-pages", _generate_search_html)
 
 
-def configure_database(app: Sphinx):
-    """Connect database for project output."""
+def configure_database(app: Sphinx, config: Config):
+    """
+    Connect database for project output.
+
+    TODO: Add support for multiple database backends?
+    """
+    # SQLite
+    """
     db_path = Path(app.outdir) / "db.sqlite"
     if db_path.exists():
         db_path.unlink()
-    models.initialize(db_path)
+    models.initialize("sqlite", db_path)
+    """
+    # PostgreSQL
+    if not app.config.sqlite3fts_database_url:
+        raise ValueError("Configuring database failed")
+    models.initialize("postgresql", app.config.sqlite3fts_database_url)
 
 
 def register_document(

src/atsphinx/sqlite3fts/models.py

@@ -6,47 +6,52 @@
 
 TODO: Add support for multiple database backends?
 """
+import os
 from pathlib import Path
 from typing import Iterable
 
-from playhouse import sqlite_ext
+from peewee import fn, SQL
+from playhouse import postgres_ext as ext
 
-db_proxy = sqlite_ext.DatabaseProxy()
+from atsphinx.sqlite3fts.playhouse import TSVectorFieldPlus
 
+db_proxy = ext.DatabaseProxy()
 
-class Document(sqlite_ext.Model):
+
+class Document(ext.Model):
     """Document main model."""
 
-    page = sqlite_ext.TextField(null=False, unique=True)
-    title = sqlite_ext.TextField(null=False)
+    page = ext.TextField(null=False, unique=True)
+    title = ext.TextField(null=False)
 
     class Meta:  # noqa: D106
         database = db_proxy
 
 
-class Section(sqlite_ext.Model):
+class Section(ext.Model):
     """Section unit of document."""
 
-    document = sqlite_ext.ForeignKeyField(Document)
-    root = sqlite_ext.BooleanField(default=False, null=False)
-    ref = sqlite_ext.TextField(null=False)
-    title = sqlite_ext.TextField(null=False)
-    body = sqlite_ext.TextField(null=False)
+    document = ext.ForeignKeyField(Document)
+    root = ext.BooleanField(default=False, null=False)
+    ref = ext.TextField(null=False)
+    title = ext.TextField(null=False)
+    body = ext.TextField(null=False)
 
     class Meta:  # noqa: D106
         database = db_proxy
 
 
-class Content(sqlite_ext.FTS5Model):
+class Content(ext.Model):
     """Searching model."""
 
-    rowid = sqlite_ext.RowIDField()
-    title = sqlite_ext.SearchField()
-    body = sqlite_ext.SearchField()
+    rowid = ext.IntegerField()
+    title = TSVectorFieldPlus()
+    body = TSVectorFieldPlus()
 
     class Meta:  # noqa: D106
         database = db_proxy
-        options = {"tokenize": "trigram"}
+        # TODO: This is an option from SQLite, it does not work on other DBMS.
+        # options = {"tokenize": "trigram"}
 
 
 def store_document(document: Document, sections: Iterable[Section]):
@@ -58,32 +63,69 @@ def store_document(document: Document, sections: Iterable[Section]):
         Content.insert(
             {
                 Content.rowid: section.id,
-                Content.title: section.title or document.title,
-                Content.body: section.body,
+                Content.title: fn.to_tsvector(section.title or document.title),
+                Content.body: fn.to_tsvector(section.body),
             }
         ).execute()
 
 
 def search_documents(keyword: str) -> Iterable[Section]:
     """Search documents from keyword by full-text-search."""
+
+    # SQLite.
+    """
     return (
         Section.select()
         .join(Content, on=(Section.id == Content.rowid))
         .where(Content.match(keyword))
         .order_by(Content.bm25())
     )
+    """
+
+    # PostgreSQL.
+    # https://www.postgresql.org/docs/current/textsearch-controls.html
+    # https://stackoverflow.com/questions/25033184/postgresql-full-text-search-performance-not-acceptable-when-ordering-by-ts-rank/25245291#25245291
+    return (
+        Section.select(
+            Section,
+            fn.ts_rank_cd(Content.title, fn.websearch_to_tsquery(keyword), 32).alias(
+                "rank_title"
+            ),
+            fn.ts_rank_cd(Content.body, fn.websearch_to_tsquery(keyword), 32).alias(
+                "rank_body"
+            ),
+        )
+        .join(Content, on=(Section.id == Content.rowid))
+        .where(
+            Content.title.match(keyword, web=True)
+            | Content.body.match(keyword, web=True)
+        )
+        .order_by(
+            SQL("rank_title").desc(),
+            SQL("rank_body").desc(),
+        )
+    )
 
 
-def bind(db_path: Path):
+def bind(db_type: str, db_path: Path):
     """Bind connection.
 
     This works only set db into proxy, not included creating tables.
     """
-    db = sqlite_ext.SqliteExtDatabase(db_path)
+    if db_type == "sqlite":
+        db = ext.SqliteExtDatabase(db_path)
+    elif db_type == "postgresql":
+        db = ext.PostgresqlExtDatabase(db_path)
+        if "POSTGRES_LOG_STATEMENT" in os.environ:
+            db.execute_sql(
+                f"SET log_statement='{os.environ['POSTGRES_LOG_STATEMENT']}';"
+            )
+    else:
+        raise ValueError(f"Unknown database type: {db_type}")
     db_proxy.initialize(db)
 
 
-def initialize(db_path: Path):
+def initialize(db_type: str, db_path: Path):
     """Bind connection and create tables."""
-    bind(db_path)
+    bind(db_type, db_path)
     db_proxy.create_tables([Document, Section, Content])

src/atsphinx/sqlite3fts/playhouse.py

@@ -0,0 +1,18 @@
+from peewee import TextField, Field, fn, Expression
+from playhouse.postgres_ext import IndexedFieldMixin, TS_MATCH
+
+
+class TSVectorFieldPlus(IndexedFieldMixin, TextField):
+    """
+    An advanced `TSVectorField`, capable to use `websearch_to_tsquery`.
+    """
+
+    field_type = "TSVECTOR"
+    __hash__ = Field.__hash__
+
+    def match(self, query, language=None, plain=False, web=False):
+        params = (language, query) if language is not None else (query,)
+        func = fn.plainto_tsquery if plain else fn.to_tsquery
+        if web:
+            func = fn.websearch_to_tsquery
+        return Expression(self, TS_MATCH, func(*params))