[refactor] expanded ConfigFile with additional modular structures and default implementations for improved configuration handling and extensibility

backend-rust/src/config.rs

@@ -5,18 +5,58 @@ use std::path::PathBuf;
 #[derive(Deserialize, Serialize, Debug, Clone)]
 #[serde(default)]
 pub struct ConfigFile {
+    #[serde(default)]
     pub server: Server,
+    #[serde(default)]
     pub display: ConfDisplay,
+    #[serde(default)]
     pub analytics: Analytics,
+    #[serde(default)]
     pub filtering: Filtering,
+    #[serde(default)]
     pub sharing: Sharing,
+    #[serde(default)]
     pub ai: Ai,
+    #[serde(default)]
     pub scraping: Scraping,
+    #[serde(default)]
     pub processing: Processing,
+    #[serde(default)]
     pub migration: ConfMigration,
+    #[serde(default)]
     pub cli: Cli,
 }
 
+#[derive(Deserialize, Serialize, Debug, Clone)]
+#[serde(default)]
+pub struct ContentQuality {
+    pub min_content_length: u32,
+    pub max_content_length: u32,
+    pub min_text_html_ratio: f32,
+    pub readability_threshold: f32,
+}
+
+#[derive(Deserialize, Serialize, Debug, Clone)]
+#[serde(default)]
+pub struct DuplicateDetection {
+    pub enabled: bool,
+    pub title_similarity_threshold: f32,
+    pub content_similarity_threshold: f32,
+    pub check_historical_days: u32,
+    pub store_fingerprints: bool,
+}
+
+#[derive(Deserialize, Serialize, Debug, Clone)]
+#[serde(default)]
+pub struct AdPrevention {
+    pub enabled: bool,
+    pub block_iframes: bool,
+    pub clean_content: bool,
+    pub ad_patterns: Vec<String>,
+    pub preserved_elements: Vec<String>,
+    pub removed_elements: Vec<String>,
+}
+
 #[derive(Deserialize, Serialize, Debug, Clone, PartialEq)]
 pub enum DefaultView {
     Compact,
@@ -74,122 +114,457 @@ pub enum DefaultOutput {
 }
 
 #[derive(Deserialize, Serialize, Debug, Clone)]
+#[serde(default)]
 pub struct Server {
+    #[serde(default = "Server::default_host")]
     pub host: String,
+    #[serde(default = "Server::default_port")]
     pub port: u16,
 }
+impl Server {
+    fn default_host() -> String {
+        Server::default().host
+    }
+    fn default_port() -> u16 {
+        Server::default().port
+    }
+}
 
 #[derive(Deserialize, Serialize, Debug, Clone)]
+#[serde(default)]
 pub struct ConfDisplay {
+    #[serde(default)]
     pub default_view: DefaultView,
+    #[serde(default = "ConfDisplay::default_articles_per_page")]
     pub articles_per_page: u32,
+    #[serde(default = "ConfDisplay::default_show_reading_time")]
     pub show_reading_time: bool,
+    #[serde(default = "ConfDisplay::default_show_word_count")]
     pub show_word_count: bool,
+    #[serde(default = "ConfDisplay::default_highlight_unread")]
     pub highlight_unread: bool,
+    #[serde(default)]
     pub theme: Theme,
 }
+impl ConfDisplay {
+    fn default_articles_per_page() -> u32 {
+        ConfDisplay::default().articles_per_page
+    }
+    fn default_show_reading_time() -> bool {
+        ConfDisplay::default().show_reading_time
+    }
+    fn default_show_word_count() -> bool {
+        ConfDisplay::default().show_word_count
+    }
+    fn default_highlight_unread() -> bool {
+        ConfDisplay::default().highlight_unread
+    }
+}
 
 #[derive(Deserialize, Serialize, Debug, Clone)]
+#[serde(default)]
 pub struct Analytics {
+    #[serde(default = "Analytics::default_enabled")]
     pub enabled: bool,
+    #[serde(default = "Analytics::default_track_reading_time")]
     pub track_reading_time: bool,
+    #[serde(default = "Analytics::default_track_scroll_position")]
     pub track_scroll_position: bool,
+    #[serde(default = "Analytics::default_retention_days")]
     pub retention_days: u32,
+    #[serde(default = "Analytics::default_aggregate_older_data")]
     pub aggregate_older_data: bool,
 }
+impl Analytics {
+    fn default_enabled() -> bool {
+        Analytics::default().enabled
+    }
+    fn default_track_reading_time() -> bool {
+        Analytics::default().track_reading_time
+    }
+    fn default_track_scroll_position() -> bool {
+        Analytics::default().track_scroll_position
+    }
+    fn default_retention_days() -> u32 {
+        Analytics::default().retention_days
+    }
+    fn default_aggregate_older_data() -> bool {
+        Analytics::default().aggregate_older_data
+    }
+}
 
 #[derive(Deserialize, Serialize, Debug, Clone)]
+#[serde(default)]
 pub struct Filtering {
+    #[serde(default = "Filtering::default_enable_smart_suggestions")]
     pub enable_smart_suggestions: bool,
+    #[serde(default = "Filtering::default_max_recent_filters")]
     pub max_recent_filters: u32,
+    #[serde(default = "Filtering::default_auto_save_filters")]
     pub auto_save_filters: bool,
+    #[serde(default)]
     pub default_sort: DefaultSort,
+    #[serde(default = "Filtering::default_enable_geographic_hierarchy")]
     pub enable_geographic_hierarchy: bool,
+    #[serde(default = "Filtering::default_auto_migrate_country_filters")]
     pub auto_migrate_country_filters: bool,
 }
-
-#[derive(Deserialize, Serialize, Debug, Clone)]
-pub struct Sharing {
-    pub default_format: DefaultFormat,
-    pub include_summary: bool,
-    pub include_tags: bool,
-    pub include_source: bool,
-    pub copy_to_clipboard: bool,
-    pub templates: SharingTemplates,
+impl Filtering {
+    fn default_enable_smart_suggestions() -> bool {
+        Filtering::default().enable_smart_suggestions
+    }
+    fn default_max_recent_filters() -> u32 {
+        Filtering::default().max_recent_filters
+    }
+    fn default_auto_save_filters() -> bool {
+        Filtering::default().auto_save_filters
+    }
+    fn default_enable_geographic_hierarchy() -> bool {
+        Filtering::default().enable_geographic_hierarchy
+    }
+    fn default_auto_migrate_country_filters() -> bool {
+        Filtering::default().auto_migrate_country_filters
+    }
 }
 
 #[derive(Deserialize, Serialize, Debug, Clone)]
+#[serde(default)]
+pub struct Sharing {
+    #[serde(default)]
+    pub default_format: DefaultFormat,
+    #[serde(default = "Sharing::default_include_summary")]
+    pub include_summary: bool,
+    #[serde(default = "Sharing::default_include_tags")]
+    pub include_tags: bool,
+    #[serde(default = "Sharing::default_include_source")]
+    pub include_source: bool,
+    #[serde(default = "Sharing::default_copy_to_clipboard")]
+    pub copy_to_clipboard: bool,
+    #[serde(default)]
+    pub templates: SharingTemplates,
+}
+impl Sharing {
+    fn default_include_summary() -> bool {
+        Sharing::default().include_summary
+    }
+    fn default_include_tags() -> bool {
+        Sharing::default().include_tags
+    }
+    fn default_include_source() -> bool {
+        Sharing::default().include_source
+    }
+    fn default_copy_to_clipboard() -> bool {
+        Sharing::default().copy_to_clipboard
+    }
+}
+
+#[derive(Deserialize, Serialize, Debug, Clone)]
+#[serde(default)]
 pub struct SharingTemplates {
+    #[serde(default = "SharingTemplate::default_text")]
     pub text: SharingTemplate,
+    #[serde(default = "SharingTemplate::default_markdown")]
     pub markdown: SharingTemplate,
 }
 
 #[derive(Deserialize, Serialize, Debug, Clone)]
+#[serde(default)]
 pub struct SharingTemplate {
     pub format: String,
 }
+impl Default for SharingTemplate {
+    fn default() -> Self {
+        // Fallback empty only if used generically; specific fields use the richer defaults below.
+        Self {
+            format: String::new(),
+        }
+    }
+}
+impl SharingTemplate {
+    pub fn default_text() -> Self {
+        SharingTemplates::default().text
+    }
+    pub fn default_markdown() -> Self {
+        SharingTemplates::default().markdown
+    }
+}
 
 #[derive(Deserialize, Serialize, Debug, Clone)]
+#[serde(default)]
 pub struct Ai {
+    #[serde(default = "Ai::default_enabled")]
     pub enabled: bool,
+    #[serde(default)]
     pub provider: AiProvider,
+    #[serde(default = "Ai::default_timeout_seconds")]
     pub timeout_seconds: u32,
+    #[serde(default)]
     pub summary: AiSummary,
+    #[serde(default)]
     pub tagging: AiTagging,
 }
-
-#[derive(Deserialize, Serialize, Debug, Clone)]
-pub struct AiSummary {
-    pub enabled: bool,
-    pub temperature: f32,
-    pub max_tokens: u32,
+impl Ai {
+    fn default_enabled() -> bool {
+        Ai::default().enabled
+    }
+    fn default_timeout_seconds() -> u32 {
+        Ai::default().timeout_seconds
+    }
 }
 
 #[derive(Deserialize, Serialize, Debug, Clone)]
-pub struct AiTagging {
+#[serde(default)]
+pub struct AiSummary {
+    #[serde(default = "AiSummary::default_enabled")]
     pub enabled: bool,
+    #[serde(default = "AiSummary::default_temperature")]
     pub temperature: f32,
+    #[serde(default = "AiSummary::default_max_tokens")]
     pub max_tokens: u32,
+}
+impl AiSummary {
+    fn default_enabled() -> bool {
+        AiSummary::default().enabled
+    }
+    fn default_temperature() -> f32 {
+        AiSummary::default().temperature
+    }
+    fn default_max_tokens() -> u32 {
+        AiSummary::default().max_tokens
+    }
+}
+
+#[derive(Deserialize, Serialize, Debug, Clone)]
+#[serde(default)]
+pub struct AiTagging {
+    #[serde(default = "AiTagging::default_enabled")]
+    pub enabled: bool,
+    #[serde(default = "AiTagging::default_temperature")]
+    pub temperature: f32,
+    #[serde(default = "AiTagging::default_max_tokens")]
+    pub max_tokens: u32,
+    #[serde(default = "AiTagging::default_max_tags_per_article")]
     pub max_tags_per_article: u32,
+    #[serde(default = "AiTagging::default_min_confidence_threshold")]
     pub min_confidence_threshold: f32,
+    #[serde(default = "AiTagging::default_enable_geographic_tagging")]
     pub enable_geographic_tagging: bool,
+    #[serde(default = "AiTagging::default_enable_category_tagging")]
     pub enable_category_tagging: bool,
+    #[serde(default = "AiTagging::default_geographic_hierarchy_levels")]
     pub geographic_hierarchy_levels: u32,
 }
-
-#[derive(Deserialize, Serialize, Debug, Clone)]
-pub struct Scraping {
-    pub timeout_seconds: u32,
-    pub max_retries: u32,
-    pub max_content_length: u32,
-    pub respect_robots_txt: bool,
-    pub rate_limit_delay_ms: u32,
+impl AiTagging {
+    fn default_enabled() -> bool {
+        AiTagging::default().enabled
+    }
+    fn default_temperature() -> f32 {
+        AiTagging::default().temperature
+    }
+    fn default_max_tokens() -> u32 {
+        AiTagging::default().max_tokens
+    }
+    fn default_max_tags_per_article() -> u32 {
+        AiTagging::default().max_tags_per_article
+    }
+    fn default_min_confidence_threshold() -> f32 {
+        AiTagging::default().min_confidence_threshold
+    }
+    fn default_enable_geographic_tagging() -> bool {
+        AiTagging::default().enable_geographic_tagging
+    }
+    fn default_enable_category_tagging() -> bool {
+        AiTagging::default().enable_category_tagging
+    }
+    fn default_geographic_hierarchy_levels() -> u32 {
+        AiTagging::default().geographic_hierarchy_levels
+    }
 }
 
 #[derive(Deserialize, Serialize, Debug, Clone)]
+#[serde(default)]
+pub struct Scraping {
+    #[serde(default = "Scraping::default_timeout_seconds")]
+    pub timeout_seconds: u32,
+    #[serde(default = "Scraping::default_max_retries")]
+    pub max_retries: u32,
+    #[serde(default = "Scraping::default_max_content_length")]
+    pub max_content_length: u32,
+    #[serde(default = "Scraping::default_respect_robots_txt")]
+    pub respect_robots_txt: bool,
+    #[serde(default = "Scraping::default_rate_limit_delay_ms")]
+    pub rate_limit_delay_ms: u32,
+    #[serde(default)]
+    pub content_quality: ContentQuality,
+    #[serde(default)]
+    pub duplicate_detection: DuplicateDetection,
+    #[serde(default)]
+    pub ad_prevention: AdPrevention,
+}
+impl Scraping {
+    fn default_timeout_seconds() -> u32 {
+        Scraping::default().timeout_seconds
+    }
+    fn default_max_retries() -> u32 {
+        Scraping::default().max_retries
+    }
+    fn default_max_content_length() -> u32 {
+        Scraping::default().max_content_length
+    }
+    fn default_respect_robots_txt() -> bool {
+        Scraping::default().respect_robots_txt
+    }
+    fn default_rate_limit_delay_ms() -> u32 {
+        Scraping::default().rate_limit_delay_ms
+    }
+}
+
+#[derive(Deserialize, Serialize, Debug, Clone)]
+#[serde(default)]
 pub struct Processing {
+    #[serde(default = "Processing::default_batch_size")]
     pub batch_size: u32,
+    #[serde(default = "Processing::default_max_concurrent")]
     pub max_concurrent: u32,
+    #[serde(default = "Processing::default_retry_attempts")]
     pub retry_attempts: u32,
+    #[serde(default = "Processing::default_priority_manual")]
     pub priority_manual: bool,
+    #[serde(default = "Processing::default_auto_mark_read_on_view")]
     pub auto_mark_read_on_view: bool,
 }
-
-#[derive(Deserialize, Serialize, Debug, Clone)]
-pub struct ConfMigration {
-    pub auto_convert_country_filters: bool,
-    pub preserve_legacy_data: bool,
-    pub migration_batch_size: u32,
+impl Processing {
+    fn default_batch_size() -> u32 {
+        Processing::default().batch_size
+    }
+    fn default_max_concurrent() -> u32 {
+        Processing::default().max_concurrent
+    }
+    fn default_retry_attempts() -> u32 {
+        Processing::default().retry_attempts
+    }
+    fn default_priority_manual() -> bool {
+        Processing::default().priority_manual
+    }
+    fn default_auto_mark_read_on_view() -> bool {
+        Processing::default().auto_mark_read_on_view
+    }
 }
 
 #[derive(Deserialize, Serialize, Debug, Clone)]
+#[serde(default)]
+pub struct ConfMigration {
+    #[serde(default = "ConfMigration::default_auto_convert_country_filters")]
+    pub auto_convert_country_filters: bool,
+    #[serde(default = "ConfMigration::default_preserve_legacy_data")]
+    pub preserve_legacy_data: bool,
+    #[serde(default = "ConfMigration::default_migration_batch_size")]
+    pub migration_batch_size: u32,
+}
+impl ConfMigration {
+    fn default_auto_convert_country_filters() -> bool {
+        ConfMigration::default().auto_convert_country_filters
+    }
+    fn default_preserve_legacy_data() -> bool {
+        ConfMigration::default().preserve_legacy_data
+    }
+    fn default_migration_batch_size() -> u32 {
+        ConfMigration::default().migration_batch_size
+    }
+}
+
+#[derive(Deserialize, Serialize, Debug, Clone)]
+#[serde(default)]
 pub struct Cli {
+    #[serde(default)]
     pub default_output: DefaultOutput,
+    #[serde(default = "Cli::default_pager_command")]
     pub pager_command: String,
+    #[serde(default = "Cli::default_show_progress")]
     pub show_progress: bool,
+    #[serde(default = "Cli::default_auto_confirm_bulk")]
     pub auto_confirm_bulk: bool,
+    #[serde(default = "Cli::default_show_geographic_hierarchy")]
     pub show_geographic_hierarchy: bool,
 }
+impl Cli {
+    fn default_pager_command() -> String {
+        Cli::default().pager_command
+    }
+    fn default_show_progress() -> bool {
+        Cli::default().show_progress
+    }
+    fn default_auto_confirm_bulk() -> bool {
+        Cli::default().auto_confirm_bulk
+    }
+    fn default_show_geographic_hierarchy() -> bool {
+        Cli::default().show_geographic_hierarchy
+    }
+}
+
+impl Default for ContentQuality {
+    fn default() -> Self {
+        Self {
+            min_content_length: 100,
+            max_content_length: 500_000,
+            min_text_html_ratio: 0.3,
+            readability_threshold: 0.6,
+        }
+    }
+}
+
+impl Default for DuplicateDetection {
+    fn default() -> Self {
+        Self {
+            enabled: true,
+            title_similarity_threshold: 0.90,
+            content_similarity_threshold: 0.85,
+            check_historical_days: 30,
+            store_fingerprints: true,
+        }
+    }
+}
+
+impl Default for AdPrevention {
+    fn default() -> Self {
+        Self {
+            enabled: true,
+            block_iframes: true,
+            clean_content: true,
+            ad_patterns: vec![
+                "sponsored-content".to_string(),
+                "advertisement".to_string(),
+                "promoted-post".to_string(),
+                "partner-content".to_string(),
+                "ad-wrapper".to_string(),
+                "sponsored".to_string(),
+            ],
+            preserved_elements: vec![
+                "article".to_string(),
+                "p".to_string(),
+                "h1".to_string(),
+                "h2".to_string(),
+                "h3".to_string(),
+                "img".to_string(),
+                "figure".to_string(),
+                "figcaption".to_string(),
+            ],
+            removed_elements: vec![
+                "aside".to_string(),
+                "iframe".to_string(),
+                "script".to_string(),
+                "style".to_string(),
+                "ins".to_string(),
+                "form".to_string(),
+                "button".to_string(),
+                ".ad".to_string(),
+                "#ad".to_string(),
+                "[class*='ad-']".to_string(),
+                "[id*='ad-']".to_string(),
+            ],
+        }
+    }
+}
 
 impl DefaultView {
     pub fn display_name(&self) -> &str {
@@ -364,6 +739,9 @@ impl Default for Scraping {
             max_content_length: 100_000,
             respect_robots_txt: true,
             rate_limit_delay_ms: 1000,
+            content_quality: ContentQuality::default(),
+            duplicate_detection: DuplicateDetection::default(),
+            ad_prevention: AdPrevention::default(),
         }
     }
 }