Browse Source

Start on refactoring of feed processing

microsub
J. King 7 years ago
parent
commit
4e57e56ca7
  1. 274
      lib/Database.php
  2. 2
      locale/en.php
  3. 22
      sql/SQLite3/0.sql

274
lib/Database.php

@ -387,57 +387,146 @@ class Database {
}
public function subscriptionAdd(string $user, string $url, string $fetchUser = "", string $fetchPassword = ""): int {
// If the user isn't authorized to perform this action then throw an exception.
if(!Data::$user->authorize($user, __FUNCTION__)) {
throw new User\ExceptionAuthz("notAuthorized", ["action" => __FUNCTION__, "user" => $user]);
if(!Data::$user->authorize($user, __FUNCTION__)) throw new User\ExceptionAuthz("notAuthorized", ["action" => __FUNCTION__, "user" => $user]);
if(!$this->userExists($user)) throw new User\Exception("doesNotExist", ["user" => $user, "action" => __FUNCTION__]);
// If the feed doesn't already exist in the database then add it to the database
// after determining its validity with PicoFeed.
$feedID = $this->db->prepare("SELECT id from arsse_feeds where url is ? and username is ? and password is ?", "str", "str", "str")->run($url, $fetchUser, $fetchPassword)->getValue();
if($feedID === null) {
$feedID = $this->feedAdd($url, $fetchUser, $fetchPassword);
}
// If the user doesn't exist throw an exception.
if(!$this->userExists($user)) {
throw new User\Exception("doesNotExist", ["user" => $user, "action" => __FUNCTION__]);
// Add the feed to the user's subscriptions.
return $this->db->prepare('INSERT INTO arsse_subscriptions(owner,feed) values(?,?)', 'str', 'int')->run($user, $feedID)->lastId();
}
public function subscriptionRemove(string $user, int $id): bool {
if(!Data::$user->authorize($user, __FUNCTION__)) throw new User\ExceptionAuthz("notAuthorized", ["action" => __FUNCTION__, "user" => $user]);
return (bool) $this->db->prepare("DELETE from arsse_subscriptions where owner is ? and id is ?", "str", "int")->run($user, $id)->changes();
}
public function feedAdd(string $url, string $fetchUser = "", string $fetchPassword = ""): int {
$feed = new Feed($url, "", "", $fetchUser, $fetchPassword);
$feed->parse();
$feedID = $this->db->prepare('INSERT INTO arsse_feeds(url,username,password) values(?,?,?)', 'str', 'str', 'str')->run($url, $fetchUser, $fetchPassword)->lastId();
// Add the feed to the database and return its Id which will be used when adding
// its articles to the database.
try {
$this->feedUpdate($feedID, $feed);
} catch(\Throwable $e) {
$this->db->prepare('DELETE from arsse_feeds where id is ?', 'int')->run($feedID);
throw $e;
}
return $feedID;
}
public function feedUpdate(int $feedID, Feed $feed = null): bool {
$this->db->begin();
try {
// If the feed doesn't already exist in the database then add it to the database
// after determining its validity with PicoFeed.
$feedID = $this->db->prepare("SELECT id from arsse_feeds where url is ? and username is ? and password is ?", "str", "str", "str")->run($url, $fetchUser, $fetchPassword)->getValue();
if($feedID === null) {
$feed = new Feed($url);
$feed->parse();
// Add the feed to the database and return its Id which will be used when adding
// its articles to the database.
$feedID = $this->db->prepare(
'INSERT INTO arsse_feeds(url,title,favicon,source,updated,modified,etag,username,password) values(?,?,?,?,?,?,?,?,?)',
'str', 'str', 'str', 'str', 'datetime', 'datetime', 'str', 'str', 'str'
)->run(
$url,
$feed->data->title,
// Grab the favicon for the feed; returns an empty string if it cannot find one.
$feed->favicon,
$feed->data->siteUrl,
$feed->data->date,
\DateTime::createFromFormat("!D, d M Y H:i:s e", $feed->resource->getLastModified()),
$feed->resource->getEtag(),
$fetchUser,
$fetchPassword
)->lastId();
// Add each of the articles to the database.
foreach($feed->data->items as $i) {
// upon the very first update of a feed the $feed object is already supplied and already parsed; for all other updates we must parse it ourselves here
if(!$feed) {
$f = $this->db->prepare('SELECT url, username, password, DATEFORMAT("http", modified) AS lastmodified, etag FROM arsse_feeds where id is ?', "int")->run($feedID)->getRow();
if(!$f) throw new Db\ExceptionInput("idMissing", ["action" => __FUNCTION__, "field" => "feed", 'id' => $feedID]);
// Feed object throws an exception when there are problems, but that isn't ideal
// here. When an exception is occurred it should update the database with the
// error instead of failing.
try {
$feed = new Feed($f['url'], $f['lastmodified'], $f['etag'], $f['username'], $f['password']);
if($feed->resource->isModified()) {
$feed->parse();
} else {
$this->db->rollback();
return false;
}
} catch (Feed\Exception $e) {
$this->db->prepare('UPDATE arsse_feeds SET err_count = err_count + 1, err_msg = ? WHERE id is ?', 'str', 'int')->run($e->getMessage(),$feedID);
$this->db->commit();
return false;
}
}
$articles = $this->db->prepare('SELECT id, url, title, author, DATEFORMAT("http", edited) AS edited_date, guid, content, url_title_hash, url_content_hash, title_content_hash FROM arsse_articles WHERE feed is ? ORDER BY id', 'int')->run($feedID)->getAll();
foreach($feed->data->items as $i) {
// Iterate through the articles in the database to determine a match for the one
// in the just-parsed feed.
$match = null;
foreach($articles as $a) {
// If the id exists and is equal to one in the database then this is the post.
if($i->id) {
if($i->id === $a['guid']) {
$match = $a;
}
}
// Otherwise if the id doesn't exist and any of the hashes match then this is
// the post.
elseif($i->urlTitleHash === $a['url_title_hash'] || $i->urlContentHash === $a['url_content_hash'] || $i->titleContentHash === $a['title_content_hash']) {
$match = $a;
}
}
// If there is no match then this is a new post and must be added to the
// database.
if(!$match) {
$this->articleAdd($feedID, $i);
continue;
}
// With that out of the way determine if the post has been updated.
// If there is an updated date, and it doesn't match the database's then update
// the post.
$update = false;
if($i->updatedDate) {
if($i->updatedDate !== $match['edited_date']) {
$update = true;
}
}
// Otherwise if there isn't an updated date and any of the hashes don't match
// then update the post.
elseif($i->urlTitleHash !== $match['url_title_hash'] || $i->urlContentHash !== $match['url_content_hash'] || $i->titleContentHash !== $match['title_content_hash']) {
$update = true;
}
if($update) {
$this->db->prepare(
'UPDATE arsse_articles SET url = ?, title = ?, author = ?, published = ?, edited = ?, modified = CURRENT_TIMESTAMP, guid = ?, content = ?, url_title_hash = ?, url_content_hash = ?, title_content_hash = ? WHERE id is ?',
'str', 'str', 'str', 'datetime', 'datetime', 'str', 'str', 'str', 'str', 'str', 'int'
)->run(
$i->url,
$i->title,
$i->author,
$i->publishedDate,
$i->updatedDate,
$i->id,
$i->content,
$i->urlTitleHash,
$i->urlContentHash,
$i->titleContentHash,
$match['id']
);
// If the article has categories update them.
$this->db->prepare('DELETE FROM arsse_categories WHERE article is ?', 'int')->run($match['id']);
$this->categoriesAdd($i, $match['id']);
}
}
// Add the feed to the user's subscriptions.
$sub = $this->db->prepare('INSERT INTO arsse_subscriptions(owner,feed) values(?,?)', 'str', 'int')->run($user, $feedID)->lastId();
// Lastly update the feed database itself with updated information.
$this->db->prepare('UPDATE arsse_feeds SET url = ?, title = ?, favicon = ?, source = ?, updated = ?, modified = ?, etag = ?, err_count = 0, err_msg = "" WHERE id is ?', 'str', 'str', 'str', 'str', 'datetime', 'datetime', 'str', 'int')->run(
$feed->data->feedUrl,
$feed->data->title,
$feed->favicon,
$feed->data->siteUrl,
$feed->data->date,
\DateTime::createFromFormat("!D, d M Y H:i:s e", $feed->resource->getLastModified()),
$feed->resource->getEtag(),
$feedID
);
} catch(\Throwable $e) {
$this->db->rollback();
throw $e;
}
$this->db->commit();
return $sub;
}
public function subscriptionRemove(string $user, int $id): bool {
if(!Data::$user->authorize($user, __FUNCTION__)) throw new User\ExceptionAuthz("notAuthorized", ["action" => __FUNCTION__, "user" => $user]);
return (bool) $this->db->prepare("DELETE from arsse_subscriptions where owner is ? and id is ?", "str", "int")->run($user, $id)->changes();
return true;
}
public function articleAdd(int $feedID, \PicoFeed\Parser\Item $article): int {
@ -485,111 +574,4 @@ class Database {
$this->db->commit();
return count($categories);
}
public function updateFeeds(): int {
$feeds = $this->db->query('SELECT id, url, username, password, DATEFORMAT("http", modified) AS lastmodified, etag FROM arsse_feeds')->getAll();
foreach($feeds as $f) {
// Feed object throws an exception when there are problems, but that isn't ideal
// here. When an exception is occurred it should update the database with the
// error instead of failing.
try {
$feed = new Feed($f['url'], $f['lastmodified'], $f['etag'], $f['username'], $f['password']);
} catch (Feed\Exception $e) {
$this->db->prepare('UPDATE arsse_feeds SET err_count = err_count + 1, err_msg = ? WHERE id is ?', 'str', 'int')->run(
$e->getMessage(),
$f['id']
);
continue;
}
// If the feed has been updated then update the database.
if($feed->resource->isModified()) {
$feed->parse();
$this->db->begin();
$articles = $this->db->prepare('SELECT id, url, title, author, DATEFORMAT("http", edited) AS edited_date, guid, content, url_title_hash, url_content_hash, title_content_hash FROM arsse_articles WHERE feed is ? ORDER BY id', 'int')->run($f['id'])->getAll();
foreach($feed->data->items as $i) {
// Iterate through the articles in the database to determine a match for the one
// in the just-parsed feed.
$match = null;
foreach($articles as $a) {
// If the id exists and is equal to one in the database then this is the post.
if($i->id) {
if($i->id === $a['guid']) {
$match = $a;
}
}
// Otherwise if the id doesn't exist and any of the hashes match then this is
// the post.
elseif($i->urlTitleHash === $a['url_title_hash'] || $i->urlContentHash === $a['url_content_hash'] || $i->titleContentHash === $a['title_content_hash']) {
$match = $a;
}
}
// If there is no match then this is a new post and must be added to the
// database.
if(!$match) {
$this->articleAdd($i);
continue;
}
// With that out of the way determine if the post has been updated.
// If there is an updated date, and it doesn't match the database's then update
// the post.
$update = false;
if($i->updatedDate) {
if($i->updatedDate !== $match['edited_date']) {
$update = true;
}
}
// Otherwise if there isn't an updated date and any of the hashes don't match
// then update the post.
elseif($i->urlTitleHash !== $match['url_title_hash'] || $i->urlContentHash !== $match['url_content_hash'] || $i->titleContentHash !== $match['title_content_hash']) {
$update = true;
}
if($update) {
$this->db->prepare(
'UPDATE arsse_articles SET url = ?, title = ?, author = ?, published = ?, edited = ?, modified = CURRENT_TIMESTAMP, guid = ?, content = ?, url_title_hash = ?, url_content_hash = ?, title_content_hash = ? WHERE id is ?',
'str', 'str', 'str', 'datetime', 'datetime', 'str', 'str', 'str', 'str', 'str', 'int'
)->run(
$i->url,
$i->title,
$i->author,
$i->publishedDate,
$i->updatedDate,
$i->id,
$i->content,
$i->urlTitleHash,
$i->urlContentHash,
$i->titleContentHash,
$match['id']
);
// If the article has categories update them.
$this->db->prepare('DELETE FROM arsse_categories WHERE article is ?', 'int')->run($match['id']);
$this->categoriesAdd($i, $match['id']);
}
}
// Lastly update the feed database itself with updated information.
$this->db->prepare('UPDATE arsse_feeds SET url = ?, title = ?, favicon = ?, source = ?, updated = ?, modified = ?, etag = ?, err_count = 0, err_msg = "" WHERE id is ?', 'str', 'str', 'str', 'str', 'datetime', 'datetime', 'str', 'int')->run(
$feed->feedUrl,
$feed->title,
$feed->favicon,
$feed->siteUrl,
$feed->date,
$feed->resource->getLastModified(),
$feed->resource->getEtag(),
$f['id']
);
}
}
$this->db->commit();
return 1;
}
}

2
locale/en.php

@ -79,7 +79,7 @@ return [
other {Authenticated user is not authorized to perform the action "{action}" on behalf of {user}}
}',
'Exception.JKingWeb/Arsse/Feed/Exception.invalidCertificate' => 'Could not download feed "{url}" because its server is serving an invalid SSL certificate',
'Exception.JKingWeb/Arsse/Feed/Exception.invalidURL' => 'Feed URL "{url}" is invalid',
'Exception.JKingWeb/Arsse/Feed/Exception.invalidUrl' => 'Feed URL "{url}" is invalid',
'Exception.JKingWeb/Arsse/Feed/Exception.maxRedirect' => 'Could not download feed "{url}" because its server reached its maximum number of HTTP redirections',
'Exception.JKingWeb/Arsse/Feed/Exception.maxSize' => 'Could not download feed "{url}" because its size exceeds the maximum allowed on its server',
'Exception.JKingWeb/Arsse/Feed/Exception.timeout' => 'Could not download feed "{url}" because its server timed out',

22
sql/SQLite3/0.sql

@ -20,14 +20,14 @@ create table arsse_users(
-- newsfeeds, deduplicated
create table arsse_feeds(
id integer primary key not null, -- sequence number
id integer primary key, -- sequence number
url TEXT not null, -- URL of feed
title TEXT, -- default title of feed
favicon TEXT, -- URL of favicon
source TEXT, -- URL of site to which the feed belongs
updated datetime, -- time at which the feed was last fetched
modified datetime, -- time at which the feed last actually changed
etag TEXT, -- HTTP ETag hash used for cache validation, changes each time the content changes
etag TEXT not null default '', -- HTTP ETag hash used for cache validation, changes each time the content changes
err_count integer not null default 0, -- count of successive times update resulted in error since last successful update
err_msg TEXT, -- last error message
username TEXT not null default '', -- HTTP authentication username
@ -37,7 +37,7 @@ create table arsse_feeds(
-- users' subscriptions to newsfeeds, with settings
create table arsse_subscriptions(
id integer primary key not null, -- sequence number
id integer primary key, -- sequence number
owner TEXT not null references arsse_users(id) on delete cascade on update cascade, -- owner of subscription
feed integer not null references arsse_feeds(id) on delete cascade, -- feed for the subscription
added datetime not null default CURRENT_TIMESTAMP, -- time at which feed was added
@ -51,7 +51,7 @@ create table arsse_subscriptions(
-- TT-RSS categories and NextCloud folders
create table arsse_folders(
id integer primary key not null, -- sequence number
id integer primary key, -- sequence number
owner TEXT not null references arsse_users(id) on delete cascade on update cascade, -- owner of folder
parent integer references arsse_folders(id) on delete cascade, -- parent folder id
name TEXT not null, -- folder name
@ -61,7 +61,7 @@ create table arsse_folders(
-- entries in newsfeeds
create table arsse_articles(
id integer primary key not null, -- sequence number
id integer primary key, -- sequence number
feed integer not null references arsse_feeds(id) on delete cascade, -- feed for the subscription
url TEXT not null, -- URL of article
title TEXT, -- article title
@ -85,17 +85,23 @@ create table arsse_enclosures(
-- users' actions on newsfeed entries
create table arsse_subscription_articles(
id integer primary key not null,
id integer primary key,
article integer not null references arsse_articles(id) on delete cascade,
owner TEXT not null references arsse_users(id) on delete cascade on update cascade,
read boolean not null default 0,
starred boolean not null default 0,
modified datetime not null default CURRENT_TIMESTAMP
);
-- IDs for specific editions of articles (required for at least NextCloud News)
create table arsse_editions(
id integer primary key,
article integer not null references arsse_articles(id) on delete cascade
);
-- user labels associated with newsfeed entries
create table arsse_labels(
sub_article integer not null references arsse_subscription_articles(id) on delete cascade, --
owner TEXT not null references arsse_users(id) on delete cascade on update cascade,
sub_article integer not null references arsse_subscription_articles(id) on delete cascade,
name TEXT
);
create index arsse_label_names on arsse_labels(name);

Loading…
Cancel
Save