fix: remove kreuzberg-tesseract version constraint and add PHP test assertions

Author: Goldziher

crates/kreuzberg/Cargo.toml

@@ -173,7 +173,7 @@ rst_parser = { version = "0.4", optional = true }
 fb2 = { version = "0.4", optional = true }
 typst-syntax = { version = "0.14", optional = true }
 
-kreuzberg-tesseract = { path = "../kreuzberg-tesseract", version = "4.0", optional = true }
+kreuzberg-tesseract = { path = "../kreuzberg-tesseract", optional = true }
 image = { workspace = true, default-features = false, features = [
     "png",
     "jpeg",

packages/php/tests/Integration/ImageExtractionTest.php

@@ -543,6 +543,11 @@ public function it_returns_empty_array_for_documents_without_images(): void
         $result = $kreuzberg->extractFile($filePath);
 
         // Images should be null or empty array for documents without images
+        $this->assertTrue(
+            $result->images === null || is_array($result->images),
+            'Images should be null or array',
+        );
+
         if ($result->images !== null) {
             $this->assertIsArray(
                 $result->images,
@@ -578,6 +583,11 @@ public function it_skips_image_extraction_when_disabled(): void
         $result = $kreuzberg->extractFile($pdfFiles[0]);
 
         // When disabled, images should be null or empty
+        $this->assertTrue(
+            $result->images === null || (is_array($result->images) && empty($result->images)),
+            'Images should be null or empty array when extraction is disabled',
+        );
+
         if ($result->images !== null) {
             $this->assertEmpty(
                 $result->images,

packages/php/tests/Integration/KeywordExtractionTest.php

@@ -244,12 +244,27 @@ public function it_returns_readonly_keyword_objects(): void
         $kreuzberg = new Kreuzberg($config);
         $result = $kreuzberg->extractFile($filePath);
 
+        // Verify keywords array exists and is populated
+        $this->assertNotNull($result->keywords, 'Keywords should not be null');
+        $this->assertIsArray($result->keywords, 'Keywords should be an array');
+        $this->assertNotEmpty($result->keywords, 'Should return at least one keyword');
+
         if (!empty($result->keywords)) {
             $keyword = $result->keywords[0];
 
-            // Verify readonly by attempting to access and verify immutability
-            $this->assertIsString($keyword->text ?? '');
-            $this->assertIsFloat($keyword->score ?? 0.0);
+            // Verify readonly by checking that objects have expected properties
+            $this->assertTrue(
+                property_exists($keyword, 'text'),
+                'Keyword object should have text property',
+            );
+            $this->assertTrue(
+                property_exists($keyword, 'score'),
+                'Keyword object should have score property',
+            );
+
+            // Verify property types
+            $this->assertIsString($keyword->text, 'Keyword text should be a string');
+            $this->assertIsFloat($keyword->score, 'Keyword score should be a float');
         }
     }