Fixes for rules

- Whitespace is now collapsed before evaluating rules - Feed tests are fixed to retrieve a dumy set of rules - Rule evaluation during feed parsing also filled out
mensbeam · Jan 7, 2021 · 6dba8aa · 6dba8aa
1 parent 461e256
commit 6dba8aa
Show file tree

Hide file tree

Showing 4 changed files with 28 additions and 19 deletions.
diff --git a/lib/Feed.php b/lib/Feed.php
@@ -7,6 +7,7 @@
 namespace JKingWeb\Arsse;
 
 use JKingWeb\Arsse\Misc\Date;
+use JKingWeb\Arsse\Rule\Rule;
 use PicoFeed\PicoFeedException;
 use PicoFeed\Config\Config;
 use PicoFeed\Client\Client;
@@ -25,6 +26,7 @@ class Feed {
     public $nextFetch;
     public $newItems = [];
     public $changedItems = [];
+    public $filteredItems = [];
 
     public static function discover(string $url, string $username = '', string $password = ''): string {
         // fetch the candidate feed
@@ -452,14 +454,16 @@ protected function scrape(): void {
     }
 
     protected function computeFilterRules(int $feedID): void {
-        return;
         $rules = Arsse::$db->feedRulesGet($feedID);
         foreach ($rules as $r) {
-            $keep = "";
-            $block = "";
-            if (strlen($r['keep'])) {
-
+            $stats = ['new' => [], 'changed' => []];
+            foreach ($this->newItems as $index => $item) {
+                $stats['new'][$index] = Rule::apply($r['keep'], $r['block'], $item->title, $item->categories);
             }
+            foreach ($this->changedItems as $index => $item) {
+                $stats['changed'][$index] = Rule::apply($r['keep'], $r['block'], $item->title, $item->categories);
+            }
+            $this->filteredItems[$r['owner']] = $stats;
         }
     }
 }
diff --git a/lib/Rule/Rule.php b/lib/Rule/Rule.php
@@ -45,8 +45,10 @@ public static function validate(string $pattern): bool {
     public static function apply(string $keepRule, string $blockRule, string $title, array $categories = []): bool {
         // if neither rule is processed we should keep
         $keep = true;
-        // add the title to the front of the category array
-        array_unshift($categories, $title);
+        // merge and clean the data to match
+        $data = array_map(function($str) {
+            return preg_replace('/\s+/', " ", $str);
+        }, array_merge([$title], $categories));
         // process the keep rule if it exists
         if (strlen($keepRule)) {
             try {
@@ -56,7 +58,7 @@ public static function apply(string $keepRule, string $blockRule, string $title,
             }
             // if a keep rule is specified the default state is now not to keep
             $keep = false;
-            foreach ($categories as $str) {
+            foreach ($data as $str) {
                 if (is_string($str)) {
                     if (preg_match($rule, $str)) {
                         // keep if the keep-rule matches one of the strings
@@ -73,7 +75,7 @@ public static function apply(string $keepRule, string $blockRule, string $title,
             } catch (Exception $e) {
                 return true;
             }
-            foreach ($categories as $str) {
+            foreach ($data as $str) {
                 if (is_string($str)) {
                     if (preg_match($rule, $str)) {
                         // do not keep if the block-rule matches one of the strings

diff --git a/tests/cases/Feed/TestFeed.php b/tests/cases/Feed/TestFeed.php
@@ -95,6 +95,7 @@ public function setUp(): void {
         self::clearData();
         self::setConf();
         Arsse::$db = \Phake::mock(Database::class);
+        \Phake::when(Arsse::$db)->feedRulesGet->thenReturn(new Result([]));
     }
 
     public function testParseAFeed(): void {

diff --git a/tests/cases/Misc/TestRule.php b/tests/cases/Misc/TestRule.php
@@ -35,16 +35,18 @@ public function testApplyRules(string $keepRule, string $blockRule, string $titl
 
     public function provideApplications(): iterable {
         return [
-            ["",           "",           "Title", ["Dummy", "Category"], true],
-            ["^Title$",    "",           "Title", ["Dummy", "Category"], true],
-            ["^Category$", "",           "Title", ["Dummy", "Category"], true],
-            ["^Naught$",   "",           "Title", ["Dummy", "Category"], false],
-            ["",           "^Title$",    "Title", ["Dummy", "Category"], false],
-            ["",           "^Category$", "Title", ["Dummy", "Category"], false],
-            ["",           "^Naught$",   "Title", ["Dummy", "Category"], true],
-            ["^Category$", "^Category$", "Title", ["Dummy", "Category"], false],
-            ["[",          "",           "Title", ["Dummy", "Category"], true],
-            ["",           "[",          "Title", ["Dummy", "Category"], true],
+            ["",           "",           "Title",   ["Dummy", "Category"], true],
+            ["^Title$",    "",           "Title",   ["Dummy", "Category"], true],
+            ["^Category$", "",           "Title",   ["Dummy", "Category"], true],
+            ["^Naught$",   "",           "Title",   ["Dummy", "Category"], false],
+            ["",           "^Title$",    "Title",   ["Dummy", "Category"], false],
+            ["",           "^Category$", "Title",   ["Dummy", "Category"], false],
+            ["",           "^Naught$",   "Title",   ["Dummy", "Category"], true],
+            ["^Category$", "^Category$", "Title",   ["Dummy", "Category"], false],
+            ["[",          "",           "Title",   ["Dummy", "Category"], true],
+            ["",           "[",          "Title",   ["Dummy", "Category"], true],
+            ["",           "^A B C$",    "A  B\nC", ["X\n   Y  \t  \r Z"], false],
+            ["",           "^X Y Z$",    "A  B\nC", ["X\n   Y  \t  \r Z"], false],
         ];
     }
 }