From 0299f2e2352773c53155ae815cd0a525d6408fe3 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Adeodato=20Sim=C3=B3?= Date: Fri, 24 Nov 2023 21:18:43 -0300 Subject: [PATCH 01/13] Add functional tests for search_vector triggers As metadata changes, search continues to work. --- bookwyrm/tests/test_book_search.py | 83 ++++++++++++++++++++++++++++++ 1 file changed, 83 insertions(+) diff --git a/bookwyrm/tests/test_book_search.py b/bookwyrm/tests/test_book_search.py index db6ba8353..2baff9bf1 100644 --- a/bookwyrm/tests/test_book_search.py +++ b/bookwyrm/tests/test_book_search.py @@ -134,3 +134,86 @@ class BookSearch(TestCase): # there's really not much to test here, it's just a dataclass self.assertEqual(result.confidence, 1) self.assertEqual(result.title, "Title") + + +class SearchVectorTriggers(TestCase): + """look for books as they change""" + + def setUp(self): + """we need basic test data and mocks""" + self.work = models.Work.objects.create(title="This Work") + self.author = models.Author.objects.create(name="Name") + self.edition = models.Edition.objects.create( + title="First Edition of Work", + subtitle="Some Extra Words Are Good", + series="A Fabulous Sequence of Items", + parent_work=self.work, + isbn_10="0000000000", + ) + self.edition.authors.add(self.author) + self.edition.save(broadcast=False) + + def test_search_after_changed_metadata(self): + """book found after updating metadata""" + self.assertEqual(self.edition, self._search_first("First")) # title + self.assertEqual(self.edition, self._search_first("Good")) # subtitle + self.assertEqual(self.edition, self._search_first("Sequence")) # series + + self.edition.title = "Second Title of Work" + self.edition.subtitle = "Fewer Words Is Better" + self.edition.series = "A Wondrous Bunch" + self.edition.save(broadcast=False) + + self.assertEqual(self.edition, self._search_first("Second")) # title new + self.assertEqual(self.edition, self._search_first("Fewer")) # subtitle new + self.assertEqual(self.edition, self._search_first("Wondrous")) # series new + + self.assertFalse(self._search_first("First")) # title old + self.assertFalse(self._search_first("Good")) # subtitle old + self.assertFalse(self._search_first("Sequence")) # series old + + def test_search_after_author_remove(self): + """book not found via removed author""" + self.assertEqual(self.edition, self._search_first("Name")) + + self.edition.authors.set([]) + self.edition.save(broadcast=False) + + self.assertFalse(self._search("Name")) + self.assertEqual(self.edition, self._search_first("Edition")) + + def test_search_after_author_add(self): + """book found by newly-added author""" + new_author = models.Author.objects.create(name="Mozilla") + + self.assertFalse(self._search("Mozilla")) + + self.edition.authors.add(new_author) + self.edition.save(broadcast=False) + + self.assertEqual(self.edition, self._search_first("Mozilla")) + self.assertEqual(self.edition, self._search_first("Name")) + + def test_search_after_updated_author_name(self): + """book found under new author name""" + self.assertEqual(self.edition, self._search_first("Name")) + self.assertFalse(self._search("Identifier")) + + self.author.name = "Identifier" + self.author.save(broadcast=False) + self.edition.refresh_from_db() + + self.assertFalse(self._search("Name")) + self.assertEqual(self.edition, self._search_first("Identifier")) + self.assertEqual(self.edition, self._search_first("Work")) + + def _search_first(self, query): + """wrapper around search_title_author""" + return self._search(query, return_first=True) + + # pylint: disable-next=no-self-use + def _search(self, query, *, return_first=False): + """wrapper around search_title_author""" + return book_search.search_title_author( + query, min_confidence=0, return_first=return_first + ) From e4d688665c7d54912b73752a270c4d4ead2a63b0 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Adeodato=20Sim=C3=B3?= Date: Fri, 24 Nov 2023 14:15:00 -0300 Subject: [PATCH 02/13] Remove index for `author.search_vector`, which is never used --- bookwyrm/migrations/0190_book_search_updates.py | 16 ++++++++++++++++ bookwyrm/models/author.py | 6 ------ 2 files changed, 16 insertions(+), 6 deletions(-) create mode 100644 bookwyrm/migrations/0190_book_search_updates.py diff --git a/bookwyrm/migrations/0190_book_search_updates.py b/bookwyrm/migrations/0190_book_search_updates.py new file mode 100644 index 000000000..52d80fcb9 --- /dev/null +++ b/bookwyrm/migrations/0190_book_search_updates.py @@ -0,0 +1,16 @@ +# Generated by Django 3.2.20 on 2023-11-24 17:11 + +from django.db import migrations + + +class Migration(migrations.Migration): + dependencies = [ + ("bookwyrm", "0188_theme_loads"), + ] + + operations = [ + migrations.RemoveIndex( + model_name="author", + name="bookwyrm_au_search__b050a8_gin", + ), + ] diff --git a/bookwyrm/models/author.py b/bookwyrm/models/author.py index 981e3c0cc..b4488d46a 100644 --- a/bookwyrm/models/author.py +++ b/bookwyrm/models/author.py @@ -2,7 +2,6 @@ import re from typing import Tuple, Any -from django.contrib.postgres.indexes import GinIndex from django.db import models from bookwyrm import activitypub @@ -68,8 +67,3 @@ class Author(BookDataModel): return f"https://{DOMAIN}/author/{self.id}" activity_serializer = activitypub.Author - - class Meta: - """sets up postgres GIN index field""" - - indexes = (GinIndex(fields=["search_vector"]),) From 44ef928c3ceb8ce9b4426113551baa53563630f7 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Adeodato=20Sim=C3=B3?= Date: Sat, 25 Nov 2023 16:08:36 -0300 Subject: [PATCH 03/13] Alter object row IDs to force test failure in original code --- bookwyrm/tests/test_book_search.py | 14 ++++++++++++++ 1 file changed, 14 insertions(+) diff --git a/bookwyrm/tests/test_book_search.py b/bookwyrm/tests/test_book_search.py index 2baff9bf1..e66ea97be 100644 --- a/bookwyrm/tests/test_book_search.py +++ b/bookwyrm/tests/test_book_search.py @@ -1,5 +1,7 @@ """ test searching for books """ import datetime +from unittest import expectedFailure +from django.db import connection from django.test import TestCase from django.utils import timezone @@ -153,6 +155,17 @@ class SearchVectorTriggers(TestCase): self.edition.authors.add(self.author) self.edition.save(broadcast=False) + @classmethod + def setUpTestData(cls): + """create conditions that trigger known old bugs""" + with connection.cursor() as cursor: + cursor.execute( + """ + ALTER SEQUENCE bookwyrm_author_id_seq RESTART WITH 20; + ALTER SEQUENCE bookwyrm_book_authors_id_seq RESTART WITH 300; + """ + ) + def test_search_after_changed_metadata(self): """book found after updating metadata""" self.assertEqual(self.edition, self._search_first("First")) # title @@ -194,6 +207,7 @@ class SearchVectorTriggers(TestCase): self.assertEqual(self.edition, self._search_first("Mozilla")) self.assertEqual(self.edition, self._search_first("Name")) + @expectedFailure def test_search_after_updated_author_name(self): """book found under new author name""" self.assertEqual(self.edition, self._search_first("Name")) From 416a6caf2d4124b884fd1a04841908e572d92c61 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Adeodato=20Sim=C3=B3?= Date: Fri, 24 Nov 2023 22:26:13 -0300 Subject: [PATCH 04/13] Define `author_search_vector_trigger` via Author.Meta.triggers Previously, triggers lived only in a particular migration file. With this change, code for the triggers resides in the model, and their lifecycle is managed through normal Django migrations. --- ...grate_search_vec_triggers_to_pgtriggers.py | 50 +++++++++++++++++++ bookwyrm/models/author.py | 29 +++++++++++ bookwyrm/settings.py | 1 + bookwyrm/utils/db.py | 22 ++++++++ requirements.txt | 1 + 5 files changed, 103 insertions(+) create mode 100644 bookwyrm/migrations/0191_migrate_search_vec_triggers_to_pgtriggers.py create mode 100644 bookwyrm/utils/db.py diff --git a/bookwyrm/migrations/0191_migrate_search_vec_triggers_to_pgtriggers.py b/bookwyrm/migrations/0191_migrate_search_vec_triggers_to_pgtriggers.py new file mode 100644 index 000000000..10354fa67 --- /dev/null +++ b/bookwyrm/migrations/0191_migrate_search_vec_triggers_to_pgtriggers.py @@ -0,0 +1,50 @@ +# Generated by Django 3.2.20 on 2023-11-25 00:47 + +from importlib import import_module +import re + +from django.db import migrations +import pgtrigger.compiler +import pgtrigger.migrations + +trigger_migration = import_module("bookwyrm.migrations.0077_auto_20210623_2155") + +# it's _very_ convenient for development that this migration be reversible +search_vector_trigger = trigger_migration.Migration.operations[4] +author_search_vector_trigger = trigger_migration.Migration.operations[5] + + +assert re.search(r"\bCREATE TRIGGER search_vector_trigger\b", search_vector_trigger.sql) +assert re.search( + r"\bCREATE TRIGGER author_search_vector_trigger\b", + author_search_vector_trigger.sql, +) + + +class Migration(migrations.Migration): + dependencies = [ + ("bookwyrm", "0190_book_search_updates"), + ] + + operations = [ + pgtrigger.migrations.AddTrigger( + model_name="author", + trigger=pgtrigger.compiler.Trigger( + name="reset_search_vector_on_author_edit", + sql=pgtrigger.compiler.UpsertTriggerSql( + func="WITH book AS (SELECT bookwyrm_book.id AS row_id FROM bookwyrm_author LEFT OUTER JOIN bookwyrm_book_authors ON bookwyrm_book_authors.id = new.id LEFT OUTER JOIN bookwyrm_book ON bookwyrm_book.id = bookwyrm_book_authors.book_id) UPDATE bookwyrm_book SET search_vector = '' FROM book WHERE id = book.row_id;RETURN NEW;", + hash="9c0a472e2bf60e63d593cce49f47972c7b227a00", + operation='UPDATE OF "name"', + pgid="pgtrigger_reset_search_vector_on_author_edit_a447c", + table="bookwyrm_author", + when="AFTER", + ), + ), + ), + migrations.RunSQL( + sql="""DROP TRIGGER IF EXISTS author_search_vector_trigger ON bookwyrm_author; + DROP FUNCTION IF EXISTS author_trigger; + """, + reverse_sql=author_search_vector_trigger.sql, + ), + ] diff --git a/bookwyrm/models/author.py b/bookwyrm/models/author.py index b4488d46a..9b4f3c1bd 100644 --- a/bookwyrm/models/author.py +++ b/bookwyrm/models/author.py @@ -3,9 +3,11 @@ import re from typing import Tuple, Any from django.db import models +import pgtrigger from bookwyrm import activitypub from bookwyrm.settings import DOMAIN +from bookwyrm.utils.db import format_trigger from .book import BookDataModel from . import fields @@ -66,4 +68,31 @@ class Author(BookDataModel): """editions and works both use "book" instead of model_name""" return f"https://{DOMAIN}/author/{self.id}" + class Meta: + """sets up indexes and triggers""" + + triggers = [ + pgtrigger.Trigger( + name="reset_search_vector_on_author_edit", + when=pgtrigger.After, + operation=pgtrigger.UpdateOf("name"), + func=format_trigger( + """WITH book AS ( + SELECT bookwyrm_book.id AS row_id + FROM bookwyrm_author + LEFT OUTER JOIN bookwyrm_book_authors + ON bookwyrm_book_authors.id = new.id + LEFT OUTER JOIN bookwyrm_book + ON bookwyrm_book.id = bookwyrm_book_authors.book_id + ) + UPDATE bookwyrm_book + SET search_vector = '' + FROM book + WHERE id = book.row_id; + RETURN new; + """ + ), + ) + ] + activity_serializer = activitypub.Author diff --git a/bookwyrm/settings.py b/bookwyrm/settings.py index 4cecc4df6..aaa50c56d 100644 --- a/bookwyrm/settings.py +++ b/bookwyrm/settings.py @@ -104,6 +104,7 @@ INSTALLED_APPS = [ "celery", "django_celery_beat", "imagekit", + "pgtrigger", "storages", ] diff --git a/bookwyrm/utils/db.py b/bookwyrm/utils/db.py new file mode 100644 index 000000000..8b74d9bf5 --- /dev/null +++ b/bookwyrm/utils/db.py @@ -0,0 +1,22 @@ +""" Database utilities """ + +from typing import cast +import sqlparse # type: ignore + + +def format_trigger(sql: str) -> str: + """format SQL trigger before storing + + we remove whitespace and use consistent casing so as to avoid migrations + due to formatting changes. + """ + return cast( + str, + sqlparse.format( + sql, + strip_comments=True, + strip_whitespace=True, + keyword_case="upper", + identifier_case="lower", + ), + ) diff --git a/requirements.txt b/requirements.txt index 36192f148..05fd9d2b4 100644 --- a/requirements.txt +++ b/requirements.txt @@ -7,6 +7,7 @@ django-celery-beat==2.4.0 django-compressor==4.3.1 django-imagekit==4.1.0 django-model-utils==4.3.1 +django-pgtrigger==4.10.0 django-sass-processor==1.2.2 django-csp==3.7 environs==9.5.0 From bcb3a343d4514fd9bc2674a918755aaa832c8886 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Adeodato=20Sim=C3=B3?= Date: Sat, 25 Nov 2023 01:49:14 -0300 Subject: [PATCH 05/13] Fix JOIN in `author_search_vector_trigger`, add missing WHERE clause --- .../0191_migrate_search_vec_triggers_to_pgtriggers.py | 4 ++-- bookwyrm/models/author.py | 3 ++- bookwyrm/tests/test_book_search.py | 2 -- 3 files changed, 4 insertions(+), 5 deletions(-) diff --git a/bookwyrm/migrations/0191_migrate_search_vec_triggers_to_pgtriggers.py b/bookwyrm/migrations/0191_migrate_search_vec_triggers_to_pgtriggers.py index 10354fa67..1e51e529e 100644 --- a/bookwyrm/migrations/0191_migrate_search_vec_triggers_to_pgtriggers.py +++ b/bookwyrm/migrations/0191_migrate_search_vec_triggers_to_pgtriggers.py @@ -32,8 +32,8 @@ class Migration(migrations.Migration): trigger=pgtrigger.compiler.Trigger( name="reset_search_vector_on_author_edit", sql=pgtrigger.compiler.UpsertTriggerSql( - func="WITH book AS (SELECT bookwyrm_book.id AS row_id FROM bookwyrm_author LEFT OUTER JOIN bookwyrm_book_authors ON bookwyrm_book_authors.id = new.id LEFT OUTER JOIN bookwyrm_book ON bookwyrm_book.id = bookwyrm_book_authors.book_id) UPDATE bookwyrm_book SET search_vector = '' FROM book WHERE id = book.row_id;RETURN NEW;", - hash="9c0a472e2bf60e63d593cce49f47972c7b227a00", + func="WITH book AS (SELECT bookwyrm_book.id AS row_id FROM bookwyrm_author LEFT OUTER JOIN bookwyrm_book_authors ON bookwyrm_book_authors.author_id = bookwyrm_author.id LEFT OUTER JOIN bookwyrm_book ON bookwyrm_book.id = bookwyrm_book_authors.book_id WHERE bookwyrm_author.id = new.id ) UPDATE bookwyrm_book SET search_vector = '' FROM book WHERE id = book.row_id;RETURN NEW;", + hash="abc8ea76fa1bf02a0f56aaae390c1b970bef1278", operation='UPDATE OF "name"', pgid="pgtrigger_reset_search_vector_on_author_edit_a447c", table="bookwyrm_author", diff --git a/bookwyrm/models/author.py b/bookwyrm/models/author.py index 9b4f3c1bd..6c5beba50 100644 --- a/bookwyrm/models/author.py +++ b/bookwyrm/models/author.py @@ -81,9 +81,10 @@ class Author(BookDataModel): SELECT bookwyrm_book.id AS row_id FROM bookwyrm_author LEFT OUTER JOIN bookwyrm_book_authors - ON bookwyrm_book_authors.id = new.id + ON bookwyrm_book_authors.author_id = bookwyrm_author.id LEFT OUTER JOIN bookwyrm_book ON bookwyrm_book.id = bookwyrm_book_authors.book_id + WHERE bookwyrm_author.id = new.id ) UPDATE bookwyrm_book SET search_vector = '' diff --git a/bookwyrm/tests/test_book_search.py b/bookwyrm/tests/test_book_search.py index e66ea97be..d2d5b692e 100644 --- a/bookwyrm/tests/test_book_search.py +++ b/bookwyrm/tests/test_book_search.py @@ -1,6 +1,5 @@ """ test searching for books """ import datetime -from unittest import expectedFailure from django.db import connection from django.test import TestCase from django.utils import timezone @@ -207,7 +206,6 @@ class SearchVectorTriggers(TestCase): self.assertEqual(self.edition, self._search_first("Mozilla")) self.assertEqual(self.edition, self._search_first("Name")) - @expectedFailure def test_search_after_updated_author_name(self): """book found under new author name""" self.assertEqual(self.edition, self._search_first("Name")) From 8df408e07eccefb934373f5c69b5068c1bee5046 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Adeodato=20Sim=C3=B3?= Date: Sat, 25 Nov 2023 16:46:54 -0300 Subject: [PATCH 06/13] Define `search_vector_trigger` via Book.Meta.triggers --- ...grate_search_vec_triggers_to_pgtriggers.py | 20 +++++++++++ bookwyrm/models/book.py | 33 ++++++++++++++++++- bookwyrm/tests/test_book_search.py | 1 - bookwyrm/utils/db.py | 1 + 4 files changed, 53 insertions(+), 2 deletions(-) diff --git a/bookwyrm/migrations/0191_migrate_search_vec_triggers_to_pgtriggers.py b/bookwyrm/migrations/0191_migrate_search_vec_triggers_to_pgtriggers.py index 1e51e529e..929063781 100644 --- a/bookwyrm/migrations/0191_migrate_search_vec_triggers_to_pgtriggers.py +++ b/bookwyrm/migrations/0191_migrate_search_vec_triggers_to_pgtriggers.py @@ -27,6 +27,20 @@ class Migration(migrations.Migration): ] operations = [ + pgtrigger.migrations.AddTrigger( + model_name="book", + trigger=pgtrigger.compiler.Trigger( + name="update_search_vector_on_book_edit", + sql=pgtrigger.compiler.UpsertTriggerSql( + func="new.search_vector := coalesce(nullif(setweight(to_tsvector('english', coalesce(new.title, '')), 'A'), ''), setweight(to_tsvector('simple', coalesce(new.title, '')), 'A')) || setweight(to_tsvector('english', coalesce(new.subtitle, '')), 'B') || (SELECT setweight(to_tsvector('simple', coalesce(array_to_string(array_agg(bookwyrm_author.name), ' '), '')), 'C') FROM bookwyrm_book LEFT OUTER JOIN bookwyrm_book_authors ON bookwyrm_book.id = bookwyrm_book_authors.book_id LEFT OUTER JOIN bookwyrm_author ON bookwyrm_book_authors.author_id = bookwyrm_author.id WHERE bookwyrm_book.id = new.id ) || setweight(to_tsvector('english', coalesce(new.series, '')), 'D');RETURN NEW;", + hash="9c898d46dfb7492ecd18f6c692bbecfa548f0e85", + operation='INSERT OR UPDATE OF "title", "subtitle", "series", "search_vector"', + pgid="pgtrigger_update_search_vector_on_book_edit_bec58", + table="bookwyrm_book", + when="BEFORE", + ), + ), + ), pgtrigger.migrations.AddTrigger( model_name="author", trigger=pgtrigger.compiler.Trigger( @@ -41,6 +55,12 @@ class Migration(migrations.Migration): ), ), ), + migrations.RunSQL( + sql="""DROP TRIGGER IF EXISTS search_vector_trigger ON bookwyrm_book; + DROP FUNCTION IF EXISTS book_trigger; + """, + reverse_sql=search_vector_trigger.sql, + ), migrations.RunSQL( sql="""DROP TRIGGER IF EXISTS author_search_vector_trigger ON bookwyrm_author; DROP FUNCTION IF EXISTS author_trigger; diff --git a/bookwyrm/models/book.py b/bookwyrm/models/book.py index 6893b9da1..ed26752e3 100644 --- a/bookwyrm/models/book.py +++ b/bookwyrm/models/book.py @@ -13,6 +13,7 @@ from django.utils.translation import gettext_lazy as _ from model_utils import FieldTracker from model_utils.managers import InheritanceManager from imagekit.models import ImageSpecField +import pgtrigger from bookwyrm import activitypub from bookwyrm.isbn.isbn import hyphenator_singleton as hyphenator @@ -24,6 +25,7 @@ from bookwyrm.settings import ( ENABLE_PREVIEW_IMAGES, ENABLE_THUMBNAIL_GENERATION, ) +from bookwyrm.utils.db import format_trigger from .activitypub_mixin import OrderedCollectionPageMixin, ObjectMixin from .base_model import BookWyrmModel @@ -232,9 +234,38 @@ class Book(BookDataModel): ) class Meta: - """sets up postgres GIN index field""" + """set up indexes and triggers""" + + # pylint: disable=line-too-long indexes = (GinIndex(fields=["search_vector"]),) + triggers = [ + pgtrigger.Trigger( + name="update_search_vector_on_book_edit", + when=pgtrigger.Before, + operation=pgtrigger.Insert + | pgtrigger.UpdateOf("title", "subtitle", "series", "search_vector"), + func=format_trigger( + """new.search_vector := + COALESCE( + NULLIF(setweight(to_tsvector('english', COALESCE(new.title, '')), 'A'), ''), + setweight(to_tsvector('simple', COALESCE(new.title, '')), 'A') + ) || + setweight(to_tsvector('english', COALESCE(new.subtitle, '')), 'B') || + (SELECT setweight(to_tsvector('simple', COALESCE(array_to_string(ARRAY_AGG(bookwyrm_author.name), ' '), '')), 'C') + FROM bookwyrm_book + LEFT OUTER JOIN bookwyrm_book_authors + ON bookwyrm_book.id = bookwyrm_book_authors.book_id + LEFT OUTER JOIN bookwyrm_author + ON bookwyrm_book_authors.author_id = bookwyrm_author.id + WHERE bookwyrm_book.id = new.id + ) || + setweight(to_tsvector('english', COALESCE(new.series, '')), 'D'); + RETURN new; + """ + ), + ) + ] class Work(OrderedCollectionPageMixin, Book): diff --git a/bookwyrm/tests/test_book_search.py b/bookwyrm/tests/test_book_search.py index d2d5b692e..f9c75d279 100644 --- a/bookwyrm/tests/test_book_search.py +++ b/bookwyrm/tests/test_book_search.py @@ -213,7 +213,6 @@ class SearchVectorTriggers(TestCase): self.author.name = "Identifier" self.author.save(broadcast=False) - self.edition.refresh_from_db() self.assertFalse(self._search("Name")) self.assertEqual(self.edition, self._search_first("Identifier")) diff --git a/bookwyrm/utils/db.py b/bookwyrm/utils/db.py index 8b74d9bf5..2bb3b9ff6 100644 --- a/bookwyrm/utils/db.py +++ b/bookwyrm/utils/db.py @@ -16,6 +16,7 @@ def format_trigger(sql: str) -> str: sql, strip_comments=True, strip_whitespace=True, + use_space_around_operators=True, keyword_case="upper", identifier_case="lower", ), From 9bcb5b80ea9dd4f11cae26ad1fc752a7cb71dcde Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Adeodato=20Sim=C3=B3?= Date: Sat, 25 Nov 2023 18:13:05 -0300 Subject: [PATCH 07/13] Further simplify bookwyrm_author trigger --- ..._migrate_search_vec_triggers_to_pgtriggers.py | 4 ++-- bookwyrm/models/author.py | 16 ++++++---------- 2 files changed, 8 insertions(+), 12 deletions(-) diff --git a/bookwyrm/migrations/0191_migrate_search_vec_triggers_to_pgtriggers.py b/bookwyrm/migrations/0191_migrate_search_vec_triggers_to_pgtriggers.py index 929063781..ddfe74a8b 100644 --- a/bookwyrm/migrations/0191_migrate_search_vec_triggers_to_pgtriggers.py +++ b/bookwyrm/migrations/0191_migrate_search_vec_triggers_to_pgtriggers.py @@ -46,8 +46,8 @@ class Migration(migrations.Migration): trigger=pgtrigger.compiler.Trigger( name="reset_search_vector_on_author_edit", sql=pgtrigger.compiler.UpsertTriggerSql( - func="WITH book AS (SELECT bookwyrm_book.id AS row_id FROM bookwyrm_author LEFT OUTER JOIN bookwyrm_book_authors ON bookwyrm_book_authors.author_id = bookwyrm_author.id LEFT OUTER JOIN bookwyrm_book ON bookwyrm_book.id = bookwyrm_book_authors.book_id WHERE bookwyrm_author.id = new.id ) UPDATE bookwyrm_book SET search_vector = '' FROM book WHERE id = book.row_id;RETURN NEW;", - hash="abc8ea76fa1bf02a0f56aaae390c1b970bef1278", + func="WITH updated_books AS (SELECT book_id FROM bookwyrm_book_authors WHERE author_id = new.id ) UPDATE bookwyrm_book SET search_vector = '' FROM updated_books WHERE id = updated_books.book_id;RETURN NEW;", + hash="e7bbf08711ff3724c58f4d92fb7a082ffb3d7826", operation='UPDATE OF "name"', pgid="pgtrigger_reset_search_vector_on_author_edit_a447c", table="bookwyrm_author", diff --git a/bookwyrm/models/author.py b/bookwyrm/models/author.py index 6c5beba50..154b00ccb 100644 --- a/bookwyrm/models/author.py +++ b/bookwyrm/models/author.py @@ -77,19 +77,15 @@ class Author(BookDataModel): when=pgtrigger.After, operation=pgtrigger.UpdateOf("name"), func=format_trigger( - """WITH book AS ( - SELECT bookwyrm_book.id AS row_id - FROM bookwyrm_author - LEFT OUTER JOIN bookwyrm_book_authors - ON bookwyrm_book_authors.author_id = bookwyrm_author.id - LEFT OUTER JOIN bookwyrm_book - ON bookwyrm_book.id = bookwyrm_book_authors.book_id - WHERE bookwyrm_author.id = new.id + """WITH updated_books AS ( + SELECT book_id + FROM bookwyrm_book_authors + WHERE author_id = new.id ) UPDATE bookwyrm_book SET search_vector = '' - FROM book - WHERE id = book.row_id; + FROM updated_books + WHERE id = updated_books.book_id; RETURN new; """ ), From bbfbd1e97ab201cd87c6414127c603f615923ce5 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Adeodato=20Sim=C3=B3?= Date: Sat, 25 Nov 2023 20:54:09 -0300 Subject: [PATCH 08/13] Add tests for trigger code (i.e. how search_vector is computed) --- bookwyrm/tests/test_book_search.py | 67 +++++++++++++++++++++++++++++- 1 file changed, 65 insertions(+), 2 deletions(-) diff --git a/bookwyrm/tests/test_book_search.py b/bookwyrm/tests/test_book_search.py index f9c75d279..99b62f0a0 100644 --- a/bookwyrm/tests/test_book_search.py +++ b/bookwyrm/tests/test_book_search.py @@ -137,6 +137,69 @@ class BookSearch(TestCase): self.assertEqual(result.title, "Title") +class SearchVectorTest(TestCase): + """check search_vector is computed correctly""" + + def test_search_vector_simple(self): + """simplest search vector""" + book = self._create_book("Book", "Mary") + self.assertEqual(book.search_vector, "'book':1A 'mary':2C") # A > C (priority) + + def test_search_vector_all_parts(self): + """search vector with subtitle and series""" + # for a book like this we call `to_tsvector("Book Long Mary Bunch")`, hence the + # indexes in the search vector. (priority "D" is the default, and never shown.) + book = self._create_book("Book", "Mary", subtitle="Long", series="Bunch") + self.assertEqual(book.search_vector, "'book':1A 'bunch':4 'long':2B 'mary':3C") + + def test_search_vector_parse_book(self): + """book parts are parsed in english""" + # FIXME: at some point this should stop being the default. + book = self._create_book( + "Edition", "Editor", series="Castle", subtitle="Writing" + ) + self.assertEqual( + book.search_vector, "'castl':4 'edit':1A 'editor':3C 'write':2B" + ) + + def test_search_vector_parse_author(self): + """author name is not stem'd or affected by stop words""" + book = self._create_book("Writing", "Writes") + self.assertEqual(book.search_vector, "'write':1A 'writes':2C") + + book = self._create_book("She Is Writing", "She Writes") + self.assertEqual(book.search_vector, "'she':4C 'write':3A 'writes':5C") + + def test_search_vector_parse_title_empty(self): + """empty parse in English retried as simple title""" + book = self._create_book("Here We", "John") + self.assertEqual(book.search_vector, "'here':1A 'john':3C 'we':2A") + + book = self._create_book("Hear We Come", "John") + self.assertEqual(book.search_vector, "'come':3A 'hear':1A 'john':4C") + + @staticmethod + def _create_book( + title, author_name, /, *, subtitle="", series="", author_alias=None + ): + """quickly create a book""" + work = models.Work.objects.create(title="work") + author = models.Author.objects.create( + name=author_name, aliases=author_alias or [] + ) + edition = models.Edition.objects.create( + title=title, + series=series or None, + subtitle=subtitle or None, + isbn_10="0000000000", + parent_work=work, + ) + edition.authors.add(author) + edition.save(broadcast=False) + edition.refresh_from_db() + return edition + + class SearchVectorTriggers(TestCase): """look for books as they change""" @@ -222,8 +285,8 @@ class SearchVectorTriggers(TestCase): """wrapper around search_title_author""" return self._search(query, return_first=True) - # pylint: disable-next=no-self-use - def _search(self, query, *, return_first=False): + @staticmethod + def _search(query, *, return_first=False): """wrapper around search_title_author""" return book_search.search_title_author( query, min_confidence=0, return_first=return_first From b5805accacb497be62cb4e7bb54a52258a1d12e3 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Adeodato=20Sim=C3=B3?= Date: Sat, 25 Nov 2023 21:47:13 -0300 Subject: [PATCH 09/13] Minor improvements to bookwyrm_book trigger code - do not COALESCE columns that cannot be NULL - do not bring bookwyrm_book to author names JOIN - add comments documenting the four steps --- ...grate_search_vec_triggers_to_pgtriggers.py | 4 ++-- bookwyrm/models/book.py | 21 ++++++++++--------- bookwyrm/tests/test_book_search.py | 16 +++++++++----- 3 files changed, 24 insertions(+), 17 deletions(-) diff --git a/bookwyrm/migrations/0191_migrate_search_vec_triggers_to_pgtriggers.py b/bookwyrm/migrations/0191_migrate_search_vec_triggers_to_pgtriggers.py index ddfe74a8b..5e798b654 100644 --- a/bookwyrm/migrations/0191_migrate_search_vec_triggers_to_pgtriggers.py +++ b/bookwyrm/migrations/0191_migrate_search_vec_triggers_to_pgtriggers.py @@ -32,8 +32,8 @@ class Migration(migrations.Migration): trigger=pgtrigger.compiler.Trigger( name="update_search_vector_on_book_edit", sql=pgtrigger.compiler.UpsertTriggerSql( - func="new.search_vector := coalesce(nullif(setweight(to_tsvector('english', coalesce(new.title, '')), 'A'), ''), setweight(to_tsvector('simple', coalesce(new.title, '')), 'A')) || setweight(to_tsvector('english', coalesce(new.subtitle, '')), 'B') || (SELECT setweight(to_tsvector('simple', coalesce(array_to_string(array_agg(bookwyrm_author.name), ' '), '')), 'C') FROM bookwyrm_book LEFT OUTER JOIN bookwyrm_book_authors ON bookwyrm_book.id = bookwyrm_book_authors.book_id LEFT OUTER JOIN bookwyrm_author ON bookwyrm_book_authors.author_id = bookwyrm_author.id WHERE bookwyrm_book.id = new.id ) || setweight(to_tsvector('english', coalesce(new.series, '')), 'D');RETURN NEW;", - hash="9c898d46dfb7492ecd18f6c692bbecfa548f0e85", + func="new.search_vector := setweight(coalesce(nullif(to_tsvector('english', new.title), ''), to_tsvector('simple', new.title)), 'A') || setweight(to_tsvector('english', coalesce(new.subtitle, '')), 'B') || (SELECT setweight(to_tsvector('simple', coalesce(array_to_string(array_agg(bookwyrm_author.name), ' '), '')), 'C') FROM bookwyrm_author LEFT JOIN bookwyrm_book_authors ON bookwyrm_author.id = bookwyrm_book_authors.author_id WHERE bookwyrm_book_authors.book_id = new.id ) || setweight(to_tsvector('english', coalesce(new.series, '')), 'D');RETURN NEW;", + hash="77d6399497c0a89b0bf09d296e33c396da63705c", operation='INSERT OR UPDATE OF "title", "subtitle", "series", "search_vector"', pgid="pgtrigger_update_search_vector_on_book_edit_bec58", table="bookwyrm_book", diff --git a/bookwyrm/models/book.py b/bookwyrm/models/book.py index ed26752e3..e167e2138 100644 --- a/bookwyrm/models/book.py +++ b/bookwyrm/models/book.py @@ -247,19 +247,20 @@ class Book(BookDataModel): | pgtrigger.UpdateOf("title", "subtitle", "series", "search_vector"), func=format_trigger( """new.search_vector := - COALESCE( - NULLIF(setweight(to_tsvector('english', COALESCE(new.title, '')), 'A'), ''), - setweight(to_tsvector('simple', COALESCE(new.title, '')), 'A') - ) || + -- title, with priority A (parse in English, default to simple if empty) + setweight(COALESCE(nullif( + to_tsvector('english', new.title), ''), + to_tsvector('simple', new.title)), 'A') || + -- subtitle, with priority B (always in English?) setweight(to_tsvector('english', COALESCE(new.subtitle, '')), 'B') || + -- list of authors, with priority C (TODO: add aliases?, bookwyrm-social#3063) (SELECT setweight(to_tsvector('simple', COALESCE(array_to_string(ARRAY_AGG(bookwyrm_author.name), ' '), '')), 'C') - FROM bookwyrm_book - LEFT OUTER JOIN bookwyrm_book_authors - ON bookwyrm_book.id = bookwyrm_book_authors.book_id - LEFT OUTER JOIN bookwyrm_author - ON bookwyrm_book_authors.author_id = bookwyrm_author.id - WHERE bookwyrm_book.id = new.id + FROM bookwyrm_author + LEFT JOIN bookwyrm_book_authors + ON bookwyrm_author.id = bookwyrm_book_authors.author_id + WHERE bookwyrm_book_authors.book_id = new.id ) || + --- last: series name, with lowest priority setweight(to_tsvector('english', COALESCE(new.series, '')), 'D'); RETURN new; """ diff --git a/bookwyrm/tests/test_book_search.py b/bookwyrm/tests/test_book_search.py index 99b62f0a0..e9f550a88 100644 --- a/bookwyrm/tests/test_book_search.py +++ b/bookwyrm/tests/test_book_search.py @@ -178,15 +178,17 @@ class SearchVectorTest(TestCase): book = self._create_book("Hear We Come", "John") self.assertEqual(book.search_vector, "'come':3A 'hear':1A 'john':4C") + def test_search_vector_no_author(self): + """book with no authors gets processed normally""" + book = self._create_book("Book", None, series="Bunch") + self.assertEqual(book.search_vector, "'book':1A 'bunch':2") + @staticmethod def _create_book( title, author_name, /, *, subtitle="", series="", author_alias=None ): """quickly create a book""" work = models.Work.objects.create(title="work") - author = models.Author.objects.create( - name=author_name, aliases=author_alias or [] - ) edition = models.Edition.objects.create( title=title, series=series or None, @@ -194,8 +196,12 @@ class SearchVectorTest(TestCase): isbn_10="0000000000", parent_work=work, ) - edition.authors.add(author) - edition.save(broadcast=False) + if author_name is not None: + author = models.Author.objects.create( + name=author_name, aliases=author_alias or [] + ) + edition.authors.add(author) + edition.save(broadcast=False) edition.refresh_from_db() return edition From d6eb390ceed8e6f63926aa8b54a4ef9314b93c75 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Adeodato=20Sim=C3=B3?= Date: Sun, 26 Nov 2023 15:57:51 -0300 Subject: [PATCH 10/13] Add test that forces `book_authors_search_vector_trigger` to execute --- bookwyrm/tests/test_book_search.py | 23 +++++++++++++++++++++++ 1 file changed, 23 insertions(+) diff --git a/bookwyrm/tests/test_book_search.py b/bookwyrm/tests/test_book_search.py index e9f550a88..42e57f773 100644 --- a/bookwyrm/tests/test_book_search.py +++ b/bookwyrm/tests/test_book_search.py @@ -275,6 +275,29 @@ class SearchVectorTriggers(TestCase): self.assertEqual(self.edition, self._search_first("Mozilla")) self.assertEqual(self.edition, self._search_first("Name")) + def test_search_after_author_add_remove_sql(self): + """add/remove author through SQL to ensure execution of book_authors trigger""" + # Tests calling edition.save(), above, pass even if the trigger in + # bookwyrm_book_authors is removed (probably because they trigger the one + # in bookwyrm_book directly). Here we make sure to exercise the former. + new_author = models.Author.objects.create(name="Mozilla") + + with connection.cursor() as cursor: + cursor.execute( + "DELETE FROM bookwyrm_book_authors WHERE book_id = %s", + [self.edition.id], + ) + self.assertFalse(self._search("Name")) + self.assertFalse(self._search("Mozilla")) + + with connection.cursor() as cursor: + cursor.execute( + "INSERT INTO bookwyrm_book_authors (book_id,author_id) VALUES (%s,%s)", + [self.edition.id, new_author.id], + ) + self.assertFalse(self._search("Name")) + self.assertEqual(self.edition, self._search_first("Mozilla")) + def test_search_after_updated_author_name(self): """book found under new author name""" self.assertEqual(self.edition, self._search_first("Name")) From beb49af51492e310c433eb4ed23bb947fc52787e Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Adeodato=20Sim=C3=B3?= Date: Sun, 17 Mar 2024 21:46:34 -0300 Subject: [PATCH 11/13] Upgade django-pgtrigger to 4.11 --- requirements.txt | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/requirements.txt b/requirements.txt index 05fd9d2b4..90ff18edb 100644 --- a/requirements.txt +++ b/requirements.txt @@ -7,7 +7,7 @@ django-celery-beat==2.4.0 django-compressor==4.3.1 django-imagekit==4.1.0 django-model-utils==4.3.1 -django-pgtrigger==4.10.0 +django-pgtrigger==4.11.0 django-sass-processor==1.2.2 django-csp==3.7 environs==9.5.0 From 2cf7ed477df2dddefa899a0feaca9e917f2b4738 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Adeodato=20Sim=C3=B3?= Date: Sun, 17 Mar 2024 22:37:28 -0300 Subject: [PATCH 12/13] Consolidate test_posgres.py into test_book_search.py These are tests I missed when first writing trigger tests in test_book_search.py. --- bookwyrm/tests/test_book_search.py | 60 ++++++++++++++++++++++- bookwyrm/tests/test_postgres.py | 77 ------------------------------ 2 files changed, 58 insertions(+), 79 deletions(-) delete mode 100644 bookwyrm/tests/test_postgres.py diff --git a/bookwyrm/tests/test_book_search.py b/bookwyrm/tests/test_book_search.py index 439d2edbf..0721cb142 100644 --- a/bookwyrm/tests/test_book_search.py +++ b/bookwyrm/tests/test_book_search.py @@ -184,11 +184,67 @@ class SearchVectorTest(TestCase): book = self._create_book("Hear We Come", "John") self.assertEqual(book.search_vector, "'come':3A 'hear':1A 'john':4C") + book = self._create_book("there there", "the") + self.assertEqual(book.search_vector, "'the':3C 'there':1A,2A") + def test_search_vector_no_author(self): """book with no authors gets processed normally""" book = self._create_book("Book", None, series="Bunch") self.assertEqual(book.search_vector, "'book':1A 'bunch':2") + book = self._create_book("there there", None) + self.assertEqual(book.search_vector, "'there':1A,2A") + + # n.b.: the following originally from test_posgres.py + + def test_search_vector_on_update(self): + """make sure that search_vector is being set correctly on edit""" + book = self._create_book("The Long Goodbye", None) + self.assertEqual(book.search_vector, "'goodby':3A 'long':2A") + + book.title = "The Even Longer Goodbye" + book.save(broadcast=False) + book.refresh_from_db() + self.assertEqual(book.search_vector, "'even':2A 'goodby':4A 'longer':3A") + + def test_search_vector_on_author_update(self): + """update search when an author name changes""" + book = self._create_book("The Long Goodbye", "The Rays") + self.assertEqual(book.search_vector, "'goodby':3A 'long':2A 'rays':5C 'the':4C") + + author = models.Author.objects.get(name="The Rays") + author.name = "Jeremy" + author.save(broadcast=False) + book.refresh_from_db() + self.assertEqual(book.search_vector, "'goodby':3A 'jeremy':4C 'long':2A") + + def test_search_vector_on_author_delete(self): + """update search when an author is deleted""" + book = self._create_book("The Long Goodbye", "The Rays") + self.assertEqual(book.search_vector, "'goodby':3A 'long':2A 'rays':5C 'the':4C") + + author = models.Author.objects.get(name="The Rays") + book.authors.remove(author) + book.refresh_from_db() + self.assertEqual(book.search_vector, "'goodby':3A 'long':2A") + + def test_search_vector_fields(self): + """language field irrelevant for search_vector""" + author = models.Author.objects.create(name="The Rays") + book = models.Edition.objects.create( + title="The Long Goodbye", + subtitle="wow cool", + series="series name", + languages=["irrelevant"], + ) + book.authors.add(author) + book.refresh_from_db() + self.assertEqual( + book.search_vector, + # pylint: disable-next=line-too-long + "'cool':5B 'goodby':3A 'long':2A 'name':9 'rays':7C 'seri':8 'the':6C 'wow':4B", + ) + @staticmethod def _create_book( title, author_name, /, *, subtitle="", series="", author_alias=None @@ -212,8 +268,8 @@ class SearchVectorTest(TestCase): return edition -class SearchVectorTriggers(TestCase): - """look for books as they change""" +class SearchVectorUpdates(TestCase): + """look for books as they change""" # functional tests of the above def setUp(self): """we need basic test data and mocks""" diff --git a/bookwyrm/tests/test_postgres.py b/bookwyrm/tests/test_postgres.py deleted file mode 100644 index 8fc3c9d59..000000000 --- a/bookwyrm/tests/test_postgres.py +++ /dev/null @@ -1,77 +0,0 @@ -""" django configuration of postgres """ -from unittest.mock import patch -from django.test import TestCase - -from bookwyrm import models - - -@patch("bookwyrm.models.activitypub_mixin.broadcast_task.apply_async") -class PostgresTriggers(TestCase): - """special migrations, fancy stuff ya know""" - - def test_search_vector_on_create(self, _): - """make sure that search_vector is being set correctly on create""" - book = models.Edition.objects.create(title="The Long Goodbye") - book.refresh_from_db() - self.assertEqual(book.search_vector, "'goodby':3A 'long':2A") - - def test_search_vector_on_update(self, _): - """make sure that search_vector is being set correctly on edit""" - book = models.Edition.objects.create(title="The Long Goodbye") - book.title = "The Even Longer Goodbye" - book.save(broadcast=False) - book.refresh_from_db() - self.assertEqual(book.search_vector, "'even':2A 'goodby':4A 'longer':3A") - - def test_search_vector_fields(self, _): - """use multiple fields to create search vector""" - author = models.Author.objects.create(name="The Rays") - book = models.Edition.objects.create( - title="The Long Goodbye", - subtitle="wow cool", - series="series name", - languages=["irrelevant"], - ) - book.authors.add(author) - book.refresh_from_db() - # pylint: disable=line-too-long - self.assertEqual( - book.search_vector, - "'cool':5B 'goodby':3A 'long':2A 'name':9 'rays':7C 'seri':8 'the':6C 'wow':4B", - ) - - def test_search_vector_on_author_update(self, _): - """update search when an author name changes""" - author = models.Author.objects.create(name="The Rays") - book = models.Edition.objects.create( - title="The Long Goodbye", - ) - book.authors.add(author) - author.name = "Jeremy" - author.save(broadcast=False) - book.refresh_from_db() - - self.assertEqual(book.search_vector, "'goodby':3A 'jeremy':4C 'long':2A") - - def test_search_vector_on_author_delete(self, _): - """update search when an author name changes""" - author = models.Author.objects.create(name="Jeremy") - book = models.Edition.objects.create( - title="The Long Goodbye", - ) - - book.authors.add(author) - book.refresh_from_db() - self.assertEqual(book.search_vector, "'goodby':3A 'jeremy':4C 'long':2A") - - book.authors.remove(author) - book.refresh_from_db() - self.assertEqual(book.search_vector, "'goodby':3A 'long':2A") - - def test_search_vector_stop_word_fallback(self, _): - """use a fallback when removing stop words leads to an empty vector""" - book = models.Edition.objects.create( - title="there there", - ) - book.refresh_from_db() - self.assertEqual(book.search_vector, "'there':1A,2A") From 6a87713f9f4f1c6dd9ad169f967b1335692e203a Mon Sep 17 00:00:00 2001 From: Bart Schuurmans Date: Wed, 20 Mar 2024 11:45:12 +0100 Subject: [PATCH 13/13] Recalculate all book search vectors after fixing the author trigger --- .../0191_migrate_search_vec_triggers_to_pgtriggers.py | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/bookwyrm/migrations/0191_migrate_search_vec_triggers_to_pgtriggers.py b/bookwyrm/migrations/0191_migrate_search_vec_triggers_to_pgtriggers.py index 5e798b654..03442298f 100644 --- a/bookwyrm/migrations/0191_migrate_search_vec_triggers_to_pgtriggers.py +++ b/bookwyrm/migrations/0191_migrate_search_vec_triggers_to_pgtriggers.py @@ -67,4 +67,10 @@ class Migration(migrations.Migration): """, reverse_sql=author_search_vector_trigger.sql, ), + migrations.RunSQL( + # Recalculate book search vector for any missed author name changes + # due to bug in JOIN in the old trigger. + sql="UPDATE bookwyrm_book SET search_vector = NULL;", + reverse_sql=migrations.RunSQL.noop, + ), ]