web feed intermediate

This commit is contained in:
Fabian Freund
2025-02-28 14:31:07 +01:00
parent 3c4d479bd9
commit 2a2311253b
51 changed files with 5503 additions and 0 deletions
@@ -0,0 +1,21 @@
import 'dart:convert';
import 'package:drift/drift.dart';
import 'package:lensai/features/web_feed/data/models/feed_author.dart';
class FeedAuthorsConverter extends TypeConverter<List<FeedAuthor>, String> {
const FeedAuthorsConverter();
@override
List<FeedAuthor> fromSql(String fromDb) {
final authors = jsonDecode(fromDb) as List<dynamic>;
return authors
.map((author) => FeedAuthor.fromJson(author as Map<String, dynamic>))
.toList();
}
@override
String toSql(List<FeedAuthor> value) {
return jsonEncode(value.map((author) => author.toJson()).toList());
}
}
@@ -0,0 +1,24 @@
import 'dart:convert';
import 'package:drift/drift.dart';
import 'package:lensai/features/web_feed/data/models/feed_category.dart';
class FeedCategoriesConverter
extends TypeConverter<List<FeedCategory>, String> {
const FeedCategoriesConverter();
@override
List<FeedCategory> fromSql(String fromDb) {
final categories = jsonDecode(fromDb) as List<dynamic>;
return categories
.map(
(category) => FeedCategory.fromJson(category as Map<String, dynamic>),
)
.toList();
}
@override
String toSql(List<FeedCategory> value) {
return jsonEncode(value.map((category) => category.toJson()).toList());
}
}
@@ -0,0 +1,16 @@
import 'package:json_annotation/json_annotation.dart';
import 'package:lensai/features/web_feed/data/database/database.dart';
class FeedDataConverter extends JsonConverter<FeedData, Map<String, dynamic>> {
const FeedDataConverter();
@override
FeedData fromJson(Map<String, dynamic> json) {
return FeedData.fromJson(json);
}
@override
Map<String, dynamic> toJson(FeedData object) {
return object.toJson();
}
}
@@ -0,0 +1,21 @@
import 'dart:convert';
import 'package:drift/drift.dart';
import 'package:lensai/features/web_feed/data/models/feed_link.dart';
class FeedLinksConverter extends TypeConverter<List<FeedLink>, String> {
const FeedLinksConverter();
@override
List<FeedLink> fromSql(String fromDb) {
final links = jsonDecode(fromDb) as List<dynamic>;
return links
.map((link) => FeedLink.fromJson(link as Map<String, dynamic>))
.toList();
}
@override
String toSql(List<FeedLink> value) {
return jsonEncode(value.map((link) => link.toJson()).toList());
}
}
@@ -0,0 +1,119 @@
import 'package:drift/drift.dart';
import 'package:lensai/features/web_feed/data/database/database.dart';
import 'package:lensai/features/web_feed/data/models/feed_article.dart';
import 'package:lensai/features/web_feed/data/models/feed_article_query_result.dart';
part 'article.g.dart';
@DriftAccessor()
class ArticleDao extends DatabaseAccessor<FeedDatabase> with _$ArticleDaoMixin {
ArticleDao(super.attachedDatabase);
Selectable<FeedArticle> getFeedArticles(Uri? url) {
final select = db.article.select();
if (url != null) {
select.where((article) => article.feedId.equalsValue(url));
}
return select..orderBy([
(row) => OrderingTerm(
expression: coalesce([row.updated, row.created]),
mode: OrderingMode.desc,
),
]);
}
SingleOrNullSelectable<FeedArticle> getArticleById(String articleId) {
return db.article.select()..where((row) => row.id.equals(articleId));
}
Future<void> upsertArticles(List<FeedArticle> articles) {
return db.transaction(() async {
await Future.wait(
articles
.map(
(article) => db.article.insertOne(
article,
onConflict: DoUpdate(
(old) {
return ArticleCompanion(
authors: Value(article.authors),
contentMarkdown: Value(article.contentMarkdown),
contentPlain: Value(article.contentPlain),
links: Value(article.links),
summaryMarkdown: Value(article.summaryMarkdown),
summaryPlain: Value(article.summaryPlain),
tags: Value(article.tags),
title: Value(article.title),
updated: Value(article.updated),
);
},
where:
(old) =>
old.updated.isNotNull() &
old.updated.isSmallerThanValue(
article.updated ?? DateTime(0),
),
),
),
)
.toList(),
);
});
}
Future<int> updateArticleRead(String articleId, DateTime? read) {
final statement =
db.article.update()..where((article) => article.id.equals(articleId));
return statement.write(ArticleCompanion(lastRead: Value(read)));
}
Selectable<(String, int)> getUnreadArticleCount() {
final count = countAll();
final countByFeed =
db.article.selectOnly()
..addColumns([db.article.feedId, count])
..where(
db.article.lastRead.isNull() |
(db.article.updated.isNotNull() &
db.article.lastRead.isSmallerThan(db.article.lastRead)),
)
..groupBy([db.article.feedId]);
return countByFeed.map(
(result) => (result.read(db.article.feedId)!, result.read(count)!),
);
}
Selectable<FeedArticleQueryResult> queryArticles({
required String matchPrefix,
required String matchSuffix,
required String ellipsis,
required int snippetLength,
required String searchString,
required Uri? feedId,
}) {
final ftsQuery = db.buildFtsQuery(searchString);
if (ftsQuery.isNotEmpty) {
return db.queryArticlesFullContent(
feedId: feedId?.toString(),
query: ftsQuery,
snippetLength: snippetLength,
beforeMatch: matchPrefix,
afterMatch: matchSuffix,
ellipsis: ellipsis,
);
} else {
return db.queryArticlesBasic(
feedId: feedId?.toString(),
query: db.buildLikeQuery(searchString),
beforeMatch: matchPrefix,
afterMatch: matchSuffix,
);
}
}
}
@@ -0,0 +1,6 @@
// GENERATED CODE - DO NOT MODIFY BY HAND
part of 'article.dart';
// ignore_for_file: type=lint
mixin _$ArticleDaoMixin on DatabaseAccessor<FeedDatabase> {}
@@ -0,0 +1,38 @@
import 'package:drift/drift.dart';
import 'package:lensai/features/web_feed/data/database/database.dart';
part 'feed.g.dart';
@DriftAccessor()
class FeedDao extends DatabaseAccessor<FeedDatabase> with _$FeedDaoMixin {
FeedDao(super.attachedDatabase);
Selectable<FeedData> getFeeds() {
return db.feed.select();
}
Future<int> updateFeedFetched(Uri url, DateTime fetched) {
final statement =
db.feed.update()..where((feed) => feed.url.equalsValue(url));
return statement.write(FeedCompanion(lastFetched: Value(fetched)));
}
Future<int> deleteFeed(Uri url) {
return db.feed.deleteWhere((feed) => feed.url.equals(url.toString()));
}
Future<int> upsertFeed(FeedData feedData) {
return db.feed.insertOne(
feedData,
onConflict: DoUpdate((old) {
return FeedCompanion(
authors: Value(feedData.authors),
title: Value(feedData.title),
description: Value(feedData.description),
tags: Value(feedData.tags),
);
}),
);
}
}
@@ -0,0 +1,6 @@
// GENERATED CODE - DO NOT MODIFY BY HAND
part of 'feed.dart';
// ignore_for_file: type=lint
mixin _$FeedDaoMixin on DatabaseAccessor<FeedDatabase> {}
@@ -0,0 +1,36 @@
import 'package:drift/drift.dart';
import 'package:lensai/data/database/converters/uri.dart';
import 'package:lensai/features/search/domain/fts_tokenizer.dart';
import 'package:lensai/features/web_feed/data/database/converters/feed_authors.dart';
import 'package:lensai/features/web_feed/data/database/converters/feed_categories.dart';
import 'package:lensai/features/web_feed/data/database/converters/feed_links.dart';
import 'package:lensai/features/web_feed/data/database/daos/article.dart';
import 'package:lensai/features/web_feed/data/database/daos/feed.dart';
import 'package:lensai/features/web_feed/data/models/feed_article.dart';
import 'package:lensai/features/web_feed/data/models/feed_article_query_result.dart';
import 'package:lensai/features/web_feed/data/models/feed_author.dart';
import 'package:lensai/features/web_feed/data/models/feed_category.dart';
import 'package:lensai/features/web_feed/data/models/feed_link.dart';
part 'database.g.dart';
@DriftDatabase(include: {'database.drift'}, daos: [ArticleDao, FeedDao])
class FeedDatabase extends _$FeedDatabase with TrigramQueryBuilderMixin {
@override
final int schemaVersion = 1;
@override
final int ftsTokenLimit = 10;
@override
final int ftsMinTokenLength = 3;
@override
MigrationStrategy get migration => MigrationStrategy(
beforeOpen: (details) async {
await customStatement('PRAGMA foreign_keys = ON;');
await optimizeFtsIndex();
},
);
FeedDatabase(super.e);
}
@@ -0,0 +1,116 @@
import 'package:lensai/data/database/converters/uri.dart';
import 'package:lensai/features/web_feed/data/models/feed_article_query_result.dart';
import 'package:lensai/features/web_feed/data/database/converters/feed_authors.dart';
import 'package:lensai/features/web_feed/data/database/converters/feed_categories.dart';
import 'package:lensai/features/web_feed/data/database/converters/feed_links.dart';
import 'package:lensai/features/web_feed/data/models/feed_article.dart';
CREATE TABLE feed (
url TEXT PRIMARY KEY NOT NULL MAPPED BY `const UriConverter()`,
title TEXT,
description TEXT,
authors TEXT MAPPED BY `const FeedAuthorsConverter()`,
tags TEXT MAPPED BY `const FeedCategoriesConverter()`,
last_fetched DATETIME
);
CREATE TABLE article (
id TEXT PRIMARY KEY NOT NULL,
feed_id TEXT NOT NULL MAPPED BY `const UriConverter()` REFERENCES feed ("url") ON DELETE CASCADE,
fetched DATETIME NOT NULL,
created DATETIME,
updated DATETIME,
last_read DATETIME,
title TEXT,
authors TEXT MAPPED BY `const FeedAuthorsConverter()`,
tags TEXT MAPPED BY `const FeedCategoriesConverter()`,
links TEXT MAPPED BY `const FeedLinksConverter()`,
summaryMarkdown TEXT,
summaryPlain TEXT,
contentMarkdown TEXT,
contentPlain TEXT
) WITH FeedArticle;
CREATE INDEX article_feed_id ON article (feed_id);
CREATE VIRTUAL TABLE article_fts
USING fts5(
title,
summaryPlain,
contentPlain,
content=article,
tokenize="trigram"
);
-- Triggers to keep the FTS index up to date.
CREATE TRIGGER article_after_insert AFTER INSERT ON article BEGIN
INSERT INTO
article_fts(rowid, title, summaryPlain, contentPlain)
VALUES (new.rowid, new.title, new.summaryPlain, new.contentPlain);
END;
CREATE TRIGGER article_after_delete AFTER DELETE ON article BEGIN
INSERT INTO
article_fts(article_fts, rowid, title, summaryPlain, contentPlain)
VALUES('delete', old.rowid, old.title, old.summaryPlain, old.contentPlain);
END;
CREATE TRIGGER article_after_update AFTER UPDATE ON article BEGIN
INSERT INTO
article_fts(article_fts, rowid, title, summaryPlain, contentPlain)
VALUES('delete', old.rowid, old.title, old.summaryPlain, old.contentPlain);
INSERT INTO
article_fts(rowid, title, summaryPlain, contentPlain)
VALUES (new.rowid, new.title, new.summaryPlain, new.contentPlain);
END;
optimizeFtsIndex:
INSERT INTO article_fts(article_fts) VALUES ('optimize');
queryArticlesBasic(:feed_id AS TEXT OR NULL) WITH FeedArticleQueryResult:
WITH weights AS (
SELECT
-- Customize these weights (higher = more important)
1.0 as title_weight -- Title matches are most important
)
SELECT
a.*,
highlight(article_fts, 0, :beforeMatch, :afterMatch) AS title,
(
bm25(article_fts, weights.title_weight)
) AS weighted_rank
FROM article_fts(:query) fts
INNER JOIN
article a ON a.rowid = fts.rowid
CROSS JOIN weights
WHERE
fts.title LIKE :query AND
:feed_id IS NULL OR a.feed_id = :feed_id
ORDER BY
weighted_rank ASC,
a.created DESC NULLS LAST;
queryArticlesFullContent(:feed_id AS TEXT OR NULL) WITH FeedArticleQueryResult:
WITH weights AS (
SELECT
-- Customize these weights (higher = more important)
5.0 as title_weight, -- Title matches are most important
2.0 as summary_weight, -- Summary matches are quite important
1.0 as content_weight -- Content matches are basic
)
SELECT
a.*,
highlight(article_fts, 0, :beforeMatch, :afterMatch) AS title,
snippet(article_fts, 1, :beforeMatch, :afterMatch, :ellipsis, :snippetLength) AS summary,
snippet(article_fts, 2, :beforeMatch, :afterMatch, :ellipsis, :snippetLength) AS content,
(
bm25(article_fts, weights.title_weight, weights.summary_weight,
weights.content_weight)
) AS weighted_rank
FROM article_fts(:query) fts
INNER JOIN
article a ON a.rowid = fts.rowid
CROSS JOIN weights
WHERE
:feed_id IS NULL OR a.feed_id = :feed_id
ORDER BY
weighted_rank ASC,
a.created DESC NULLS LAST;
File diff suppressed because it is too large Load Diff