Reliable AI categories + image fallback for source-less articles

src/Admin/Admin.php

@@ -136,6 +136,7 @@ public function assets( string $hook ): void {
 		}
 
 		wp_enqueue_style( 'aggregate-it-admin', AGGREGATE_IT_URL . 'assets/css/admin.css', [], $this->asset_version( 'assets/css/admin.css' ) );
+		wp_enqueue_media();
 		wp_enqueue_script( 'jquery-ui-sortable' );
 		wp_enqueue_script( 'aggregate-it-charts', AGGREGATE_IT_URL . 'assets/js/charts.js', [], $this->asset_version( 'assets/js/charts.js' ), true );
 		wp_enqueue_script( 'aggregate-it-admin', AGGREGATE_IT_URL . 'assets/js/admin.js', [ 'aggregate-it-charts' ], $this->asset_version( 'assets/js/admin.js' ), true );
@@ -908,6 +909,7 @@ public function handle_save_settings(): void {
 				'daily_spend_cap_usd'    => max( 0, (float) ( $_POST['daily_spend_cap_usd'] ?? 5 ) ),
 				'image_mode'             => sanitize_key( wp_unslash( $_POST['image_mode'] ?? 'import' ) ),
 				'image_source'           => sanitize_key( wp_unslash( $_POST['image_source'] ?? 'share' ) ),
+				'default_image_id'       => max( 0, (int) ( $_POST['default_image_id'] ?? 0 ) ),
 				'indexnow_enabled'       => isset( $_POST['indexnow_enabled'] ),
 				'wikipedia_research'     => isset( $_POST['wikipedia_research'] ),
 				'related_articles'       => isset( $_POST['related_articles'] ),

src/Admin/views/settings.php

@@ -149,6 +149,22 @@
 					</select>
 				</td>
 			</tr>
+			<tr class="ai-default-image-row">
+				<th><label><?php esc_html_e( 'Fallback image', 'aggregate-it' ); ?></label></th>
+				<td>
+					<?php
+					$default_image_id  = $settings->default_image_id();
+					$default_image_src = $default_image_id ? wp_get_attachment_image_url( $default_image_id, 'medium' ) : '';
+					?>
+					<input type="hidden" name="default_image_id" id="default_image_id" value="<?php echo esc_attr( (string) $default_image_id ); ?>">
+					<img id="default_image_preview" src="<?php echo esc_url( (string) $default_image_src ); ?>" alt="" style="max-width:160px;height:auto;border-radius:6px;margin-bottom:6px;display:<?php echo $default_image_src ? 'block' : 'none'; ?>;">
+					<p>
+						<button type="button" class="button" id="default_image_choose"><?php esc_html_e( 'Choose image', 'aggregate-it' ); ?></button>
+						<button type="button" class="button-link" id="default_image_clear" style="display:<?php echo $default_image_id ? 'inline' : 'none'; ?>;"><?php esc_html_e( 'Remove', 'aggregate-it' ); ?></button>
+					</p>
+					<p class="description"><?php esc_html_e( 'Featured image used only when an article has none of its own. Leave empty to allow image-less posts.', 'aggregate-it' ); ?></p>
+				</td>
+			</tr>
 			<tr>
 				<th><?php esc_html_e( 'Tell search engines instantly', 'aggregate-it' ); ?></th>
 				<td><label><input name="indexnow_enabled" type="checkbox" <?php checked( $settings->indexnow_enabled() ); ?>> <?php esc_html_e( 'Let search engines know right away when an article is published or updated', 'aggregate-it' ); ?></label></td>
@@ -244,6 +260,35 @@ function sync() {
 		if ( imageMode ) { imageMode.addEventListener( 'change', sync ); }
 		sync();
 
+		var chooseBtn = document.getElementById( 'default_image_choose' );
+		var clearBtn = document.getElementById( 'default_image_clear' );
+		var imageIdField = document.getElementById( 'default_image_id' );
+		var imagePreview = document.getElementById( 'default_image_preview' );
+		var mediaFrame;
+		if ( chooseBtn && imageIdField && window.wp && wp.media ) {
+			chooseBtn.addEventListener( 'click', function () {
+				if ( mediaFrame ) { mediaFrame.open(); return; }
+				mediaFrame = wp.media( { title: chooseBtn.textContent, library: { type: 'image' }, multiple: false } );
+				mediaFrame.on( 'select', function () {
+					var att = mediaFrame.state().get( 'selection' ).first().toJSON();
+					imageIdField.value = att.id;
+					if ( imagePreview ) {
+						imagePreview.src = ( att.sizes && att.sizes.medium ) ? att.sizes.medium.url : att.url;
+						imagePreview.style.display = 'block';
+					}
+					if ( clearBtn ) { clearBtn.style.display = 'inline'; }
+				} );
+				mediaFrame.open();
+			} );
+		}
+		if ( clearBtn && imageIdField ) {
+			clearBtn.addEventListener( 'click', function () {
+				imageIdField.value = '0';
+				if ( imagePreview ) { imagePreview.src = ''; imagePreview.style.display = 'none'; }
+				clearBtn.style.display = 'none';
+			} );
+		}
+
 		var test = document.getElementById( 'ai-test-key' );
 		var result = document.getElementById( 'ai-test-result' );
 		if ( test && result ) {

src/Ai/OpenAiProvider.php

@@ -28,8 +28,38 @@ public function key(): string {
 	}
 
 	public function structured( string $prompt, array $schema, array $opts = [] ): array {
-		$model = $this->settings->ai_model() ?: self::DEFAULT_MODEL;
+		$model   = $this->settings->ai_model() ?: self::DEFAULT_MODEL;
+		$message = $prompt . $this->schema_hint( $schema );
+
+		// Strict Structured Outputs guarantees every required field (e.g. category) is
+		// present; plain JSON mode does not, and a dropped category silently sends the post
+		// to the default WordPress category. Fall back to JSON mode if the model rejects it.
+		try {
+			$data = $this->chat(
+				$model,
+				$message,
+				[
+					'type'        => 'json_schema',
+					'json_schema' => [ 'name' => 'article', 'strict' => true, 'schema' => $this->strict_schema( $schema ) ],
+				]
+			);
+		} catch ( \RuntimeException $e ) {
+			$data = $this->chat( $model, $message, [ 'type' => 'json_object' ] );
+		}
+
+		$text = (string) ( $data['choices'][0]['message']['content'] ?? '' );
+		$in   = (int) ( $data['usage']['prompt_tokens'] ?? 0 );
+		$out  = (int) ( $data['usage']['completion_tokens'] ?? 0 );
+
+		return [
+			'result'   => $this->first_json( $text ),
+			'tokens'   => $in + $out,
+			'cost_usd' => $this->cost( self::PRICING, $model, self::DEFAULT_MODEL, $in, $out ),
+		];
+	}
 
+	/** @param array<string,mixed> $response_format */
+	private function chat( string $model, string $message, array $response_format ): array {
 		$response = wp_remote_post(
 			self::CHAT_ENDPOINT,
 			[
@@ -41,24 +71,40 @@ public function structured( string $prompt, array $schema, array $opts = [] ): a
 				'body'    => wp_json_encode(
 					[
 						'model'           => $model,
-						'messages'        => [ [ 'role' => 'user', 'content' => $prompt . $this->schema_hint( $schema ) ] ],
-						'response_format' => [ 'type' => 'json_object' ],
+						'messages'        => [ [ 'role' => 'user', 'content' => $message ] ],
+						'response_format' => $response_format,
 					]
 				),
 			]
 		);
 
-		$data = $this->decode( $response );
+		return $this->decode( $response );
+	}
 
-		$text = (string) ( $data['choices'][0]['message']['content'] ?? '' );
-		$in   = (int) ( $data['usage']['prompt_tokens'] ?? 0 );
-		$out  = (int) ( $data['usage']['completion_tokens'] ?? 0 );
+	/**
+	 * Rewrite our schema into the subset OpenAI's strict mode accepts: every object must list
+	 * all properties as required and forbid extras, and string/number constraints it rejects
+	 * (maxLength and friends) are dropped — length is still guided by the prompt.
+	 *
+	 * @param array<string,mixed> $schema
+	 * @return array<string,mixed>
+	 */
+	private function strict_schema( array $schema ): array {
+		unset( $schema['minLength'], $schema['maxLength'], $schema['pattern'], $schema['format'], $schema['minimum'], $schema['maximum'] );
+
+		if ( ( $schema['type'] ?? '' ) === 'object' && isset( $schema['properties'] ) && is_array( $schema['properties'] ) ) {
+			foreach ( $schema['properties'] as $key => $prop ) {
+				$schema['properties'][ $key ] = $this->strict_schema( (array) $prop );
+			}
+			$schema['required']             = array_keys( $schema['properties'] );
+			$schema['additionalProperties'] = false;
+		}
 
-		return [
-			'result'   => $this->first_json( $text ),
-			'tokens'   => $in + $out,
-			'cost_usd' => $this->cost( self::PRICING, $model, self::DEFAULT_MODEL, $in, $out ),
-		];
+		if ( ( $schema['type'] ?? '' ) === 'array' && isset( $schema['items'] ) && is_array( $schema['items'] ) ) {
+			$schema['items'] = $this->strict_schema( (array) $schema['items'] );
+		}
+
+		return $schema;
 	}
 
 	public function embed( string $text ): array {

src/Publish/ImageImporter.php

@@ -30,23 +30,31 @@ public function maybe_import( int $post_id, string $image_url, string $alt, bool
 		if ( ! $replace && $this->settings->image_mode() === 'off' ) {
 			return;
 		}
-		if ( $image_url === '' ) {
-			EventLog::warning( sprintf( 'Post #%d: no image found for this article.', $post_id ) );
-			return;
-		}
 		if ( has_post_thumbnail( $post_id ) && ! $replace ) {
 			return;
 		}
 
-		$this->load_media_deps();
+		if ( $image_url !== '' ) {
+			$this->load_media_deps();
+			$attachment_id = $this->sideload( esc_url_raw( $image_url ), $post_id, $alt );
+			if ( $attachment_id !== 0 ) {
+				set_post_thumbnail( $post_id, $attachment_id );
+				update_post_meta( $attachment_id, '_wp_attachment_image_alt', sanitize_text_field( $alt ) );
+				return;
+			}
+		}
 
-		$attachment_id = $this->sideload( esc_url_raw( $image_url ), $post_id, $alt );
-		if ( $attachment_id === 0 ) {
+		// No usable source image (missing, or download/import failed): fall back to the
+		// configured default so no post is left image-less.
+		$default = $this->settings->default_image_id();
+		if ( $default > 0 && wp_attachment_is_image( $default ) ) {
+			set_post_thumbnail( $post_id, $default );
 			return;
 		}
 
-		set_post_thumbnail( $post_id, $attachment_id );
-		update_post_meta( $attachment_id, '_wp_attachment_image_alt', sanitize_text_field( $alt ) );
+		if ( $image_url === '' ) {
+			EventLog::warning( sprintf( 'Post #%d: no image found for this article.', $post_id ) );
+		}
 	}
 
 	/**

src/Publish/PostFactory.php

@@ -6,6 +6,7 @@
 use AggregateIt\Seo\SlugGenerator;
 use AggregateIt\Settings;
 use AggregateIt\Source\SourceRepository;
+use AggregateIt\Support\EventLog;
 
 defined( 'ABSPATH' ) || exit;
 
@@ -80,6 +81,14 @@ private function assign_terms( int $post_id, string $post_type, int $source_id,
 			}
 			if ( $ids ) {
 				wp_set_post_terms( $post_id, $ids, 'category', false );
+			} elseif ( $this->settings->ai_categorize() ) {
+				EventLog::warning(
+					sprintf(
+						'Post #%d kept the site default category: the AI returned %s and the source has no category set.',
+						$post_id,
+						$ai_category === '' ? 'no category' : sprintf( '"%s", which could not be created', $ai_category )
+					)
+				);
 			}
 		}
 

src/Settings.php

@@ -159,6 +159,10 @@ public function image_source(): string {
 		return in_array( $src, [ 'share', 'feed' ], true ) ? $src : 'share';
 	}
 
+	public function default_image_id(): int {
+		return max( 0, (int) $this->get( 'default_image_id', 0 ) );
+	}
+
 	public function indexnow_enabled(): bool {
 		return (bool) $this->get( 'indexnow_enabled', true );
 	}

src/Source/ContentExtractor.php

@@ -85,18 +85,37 @@ private function link_image( string $html ): string {
 	}
 
 	private function first_content_image( string $html ): string {
-		if ( ! preg_match_all( '/<img[^>]+src=["\']([^"\']+)["\']/i', $html, $m ) ) {
+		if ( ! preg_match_all( '/<img\b[^>]*>/i', $html, $m ) ) {
 			return '';
 		}
-		foreach ( $m[1] as $src ) {
-			$src = trim( html_entity_decode( $src ) );
-			if ( preg_match( '#^https?://#i', $src ) && ! $this->is_junk_image( $src ) ) {
+		foreach ( $m[0] as $tag ) {
+			$src = $this->img_src( $tag );
+			if ( $src !== '' && preg_match( '#^https?://#i', $src ) && ! $this->is_junk_image( $src ) ) {
 				return $src;
 			}
 		}
 		return '';
 	}
 
+	/** Publishers lazy-load heroes: the real URL sits in data-src/srcset while src holds a placeholder. */
+	private function img_src( string $tag ): string {
+		foreach ( [ 'data-src', 'data-lazy-src', 'data-original', 'data-lazy', 'src' ] as $attr ) {
+			if ( preg_match( '/\b' . preg_quote( $attr, '/' ) . '=(["\'])(.*?)\1/i', $tag, $m ) ) {
+				$url = trim( html_entity_decode( $m[2] ) );
+				if ( $url !== '' && strpos( $url, 'data:' ) !== 0 ) {
+					return $url;
+				}
+			}
+		}
+		if ( preg_match( '/\bsrcset=(["\'])(.*?)\1/i', $tag, $m ) ) {
+			$first = trim( (string) strtok( trim( html_entity_decode( $m[2] ) ), ' ' ) );
+			if ( $first !== '' && strpos( $first, 'data:' ) !== 0 ) {
+				return $first;
+			}
+		}
+		return '';
+	}
+
 	private function meta_content( string $html, string $key ): string {
 		$e = preg_quote( $key, '/' );
 		// Capture the opening quote and match the same one, so a URL containing an apostrophe

tests/ContentExtractorTest.php

@@ -73,6 +73,21 @@ public function test_share_image_rethrows_transient_only_when_requested(): void
 		$ext->share_image( 'https://example.com/x', true );
 	}
 
+	public function test_reads_lazy_loaded_hero_from_data_src(): void {
+		$html = '<article><img src="data:image/svg+xml;base64,PLACEHOLDER" data-src="https://cdn/hero.jpg"></article>';
+		$this->assertSame( 'https://cdn/hero.jpg', $this->best->invoke( $this->extractor, $html ) );
+	}
+
+	public function test_reads_lazy_loaded_hero_from_srcset(): void {
+		$html = '<article><img src="data:image/gif;base64,R0lGOD" srcset="https://cdn/small.jpg 480w, https://cdn/large.jpg 1200w"></article>';
+		$this->assertSame( 'https://cdn/small.jpg', $this->best->invoke( $this->extractor, $html ) );
+	}
+
+	public function test_skips_placeholder_only_image(): void {
+		$html = '<article><img src="data:image/svg+xml;base64,PLACEHOLDER"></article>';
+		$this->assertSame( '', $this->best->invoke( $this->extractor, $html ) );
+	}
+
 	public function test_readability_keeps_article_drops_chrome(): void {
 		$html = '<html><body><nav>Home About</nav><header>Site</header>'
 			. '<article><h2>Big News</h2><p>First paragraph with real content.</p>'

tests/OpenAiProviderTest.php

@@ -0,0 +1,69 @@
+<?php
+
+namespace AggregateIt\Tests;
+
+use AggregateIt\Ai\OpenAiProvider;
+use AggregateIt\Settings;
+use PHPUnit\Framework\TestCase;
+
+final class OpenAiProviderTest extends TestCase {
+
+	private function strict( array $schema ): array {
+		$ref    = new \ReflectionMethod( OpenAiProvider::class, 'strict_schema' );
+		$ref->setAccessible( true );
+		return $ref->invoke( new OpenAiProvider( new Settings() ), $schema );
+	}
+
+	public function test_makes_every_property_required_and_forbids_extras(): void {
+		$out = $this->strict(
+			[
+				'type'       => 'object',
+				'required'   => [ 'category' ],
+				'properties' => [
+					'category' => [ 'type' => 'string' ],
+					'entities' => [ 'type' => 'array', 'items' => [ 'type' => 'string' ] ],
+				],
+			]
+		);
+
+		$this->assertSame( [ 'category', 'entities' ], $out['required'] );
+		$this->assertFalse( $out['additionalProperties'] );
+	}
+
+	public function test_strips_unsupported_string_constraints(): void {
+		$out = $this->strict(
+			[
+				'type'       => 'object',
+				'properties' => [
+					'seo_title' => [ 'type' => 'string', 'maxLength' => 70 ],
+				],
+			]
+		);
+
+		$this->assertArrayNotHasKey( 'maxLength', $out['properties']['seo_title'] );
+	}
+
+	public function test_recurses_into_nested_object_items(): void {
+		$out = $this->strict(
+			[
+				'type'       => 'object',
+				'properties' => [
+					'entities' => [
+						'type'  => 'array',
+						'items' => [
+							'type'       => 'object',
+							'properties' => [
+								'name' => [ 'type' => 'string' ],
+								'type' => [ 'type' => 'string' ],
+							],
+						],
+					],
+				],
+			]
+		);
+
+		$item = $out['properties']['entities']['items'];
+		$this->assertSame( [ 'name', 'type' ], $item['required'] );
+		$this->assertFalse( $item['additionalProperties'] );
+	}
+}