crawlberg 1.7.1 → 1.7.2

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 4146ffc905ddf02d8ce12708033087570b86ea18e1fa0741f44cdcd0f99109d6
-  data.tar.gz: f675775b035824d0b0e545293dcbfa8453a564538a6f51d9a0cc6ab00cc31dfa
+  metadata.gz: 154f92c622f76efb16d88fb2075772251365a09de713990bbb1b63668ee22065
+  data.tar.gz: e60ce1d8f947cf1ea3afe0b3cf203467dc2fb1dc5d2a942bdf166a5ff2e198ea
 SHA512:
-  metadata.gz: 28328df6f4ca037a1026b95c4a9ff9f70c965150293d8c7d303af60d070d443329ba5635dae0c42179eed941407a0cb4375f9f03fa4ab7ec94e2f2e9e67b7973
-  data.tar.gz: fe72dea76f14a6e15038f27131e06108319c8726377f7c23de89eeb3e1abbe0905068a03fbd0577ac826b176f8b90a9a785326c88db17d92278c49dd43a072bd
+  metadata.gz: dea0dc85482926a5bd305ba5176ef280e6eca5d192ceb4ce4c24bcd49cb93383ec1358b3e3940ae712aadd22fbe32373efccc6032bbdeee89b1472281b598f33
+  data.tar.gz: a6ed384d2269b3dab4ca43ec5e5f5417b26e2e0391eb0329a64686d2d06e4f0b8bf89de0f5dccf3ca0a123114c0cc47aa4684c6dd6433ed0a7bc0b6a6f462d2c

data/ext/crawlberg_rb/native/Cargo.lock

@@ -555,9 +555,9 @@ dependencies = [
 
 [[package]]
 name = "crawlberg"
-version = "1.7.1"
+version = "1.7.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "2c9cce4104eda65a025494b1fff11843b2903f2a922b104d3865288ecfc9a312"
+checksum = "f65b7f99dea838e39d35c03e0acef6b16103a9626881e07b846d027fcd02d62f"
 dependencies = [
  "ahash",
  "aho-corasick",
@@ -599,7 +599,7 @@ dependencies = [
 
 [[package]]
 name = "crawlberg-rb"
-version = "1.7.1"
+version = "1.7.2"
 dependencies = [
  "crawlberg",
  "futures",
@@ -1553,9 +1553,9 @@ checksum = "112b39cec0b298b6c1999fee3e31427f74f676e4cb9879ed1a121b43661a4154"
 
 [[package]]
 name = "magnus"
-version = "0.8.2"
+version = "0.9.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "3b36a5b126bbe97eb0d02d07acfeb327036c6319fd816139a49824a83b7f9012"
+checksum = "379d50b6eea4ca84074ccd3e551eb9b07f7865194d5b3198325317e9ef78e5e9"
 dependencies = [
  "magnus-macros",
  "rb-sys",
@@ -1565,9 +1565,9 @@ dependencies = [
 
 [[package]]
 name = "magnus-macros"
-version = "0.8.0"
+version = "0.9.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "47607461fd8e1513cb4f2076c197d8092d921a1ea75bd08af97398f593751892"
+checksum = "4ca70566af98184bb92d4d9853daaf533c1a3418b9570d0547ea44970035f3fe"
 dependencies = [
  "proc-macro2",
  "quote",

data/ext/crawlberg_rb/native/Cargo.toml

@@ -1,11 +1,11 @@
 # This file is auto-generated by alef — DO NOT EDIT.
-# alef:hash:4d4e3f0cd2ee83b6be714f9fa4dc56a358a714d01aeeb6483de665ac6bf59c83
+# alef:hash:5335ae796018deac59899b2632bbcbf919394a15070c42cd96d6ad1719190d45
 # To regenerate: alef generate
 # To verify freshness: alef verify
 
 [package]
 name = "crawlberg-rb"
-version = "1.7.1"
+version = "1.7.2"
 edition = "2024"
 license = "MIT"
 description = "High-performance web crawling engine"
@@ -22,9 +22,9 @@ path = "../src/lib.rs"
 crate-type = ["cdylib"]
 
 [dependencies]
-crawlberg = { version = "1.7.1", features = ["interact", "browser-chromiumoxide"] }
+crawlberg = { version = "1.7.2", features = ["interact", "browser-chromiumoxide"] }
 futures = "0.3"
-magnus = "0.8"
+magnus = "0.9"
 rb-sys = ">=0.9.130, <0.10"
 serde = { version = "1", features = ["derive"] }
 serde_json = "1"

data/ext/crawlberg_rb/src/lib.rs

@@ -1,5 +1,5 @@
 // This file is auto-generated by alef. DO NOT EDIT.
-// alef:hash:202c4f469a62d77930ef13ac10e1718ac319e02c448ba94573bae7ab48c1d4ce
+// alef:hash:9563e80a11ae24fda8f480033a40a3dd9382715d77f7a2e60c254f7f3ab71ad5
 // Re-generate with: alef generate
 #![allow(dead_code, unused_imports, unused_variables)]
 #![allow(
@@ -6254,7 +6254,7 @@ impl SsrfPolicy {
     }
 }
 
-#[derive(Clone, Copy, PartialEq, Eq, Debug, serde::Serialize, serde::Deserialize)]
+#[derive(Clone, Copy, PartialEq, Eq, Hash, Debug, serde::Serialize, serde::Deserialize)]
 #[serde(rename_all = "snake_case")]
 pub enum AssetCategory {
     Document,
@@ -6295,7 +6295,14 @@ impl magnus::IntoValue for AssetCategory {
 
 impl magnus::TryConvert for AssetCategory {
     fn try_convert(val: magnus::Value) -> Result<Self, magnus::Error> {
-        let s: String = magnus::TryConvert::try_convert(val)?;
+        // IntoValue emits Symbols, including when this enum is a Hash key.
+        // Accept that exact representation as well as String inputs; do not
+        // coerce arbitrary objects through to_s.
+        let s: String = if let Some(symbol) = magnus::Symbol::from_value(val) {
+            symbol.name()?.into_owned()
+        } else {
+            magnus::TryConvert::try_convert(val)?
+        };
         // Accept the real serde wire value, the legacy always-snake_case symbol Magnus used to
         // emit unconditionally, and the verbatim PascalCase Rust variant name, so fixtures and
         // existing consumer code written in any of those styles keep working.
@@ -6352,7 +6359,8 @@ impl magnus::IntoValue for AuthConfig {
 
 impl magnus::TryConvert for AuthConfig {
     fn try_convert(val: magnus::Value) -> Result<Self, magnus::Error> {
-        // For data enums with fields (e.g., PageAction), try to deserialize from JSON first.
+        // Generated Data variants carry native DTOs, not JSON-compatible display strings.
+        let class_name: Option<String> = val.class().funcall("name", ())?; // For data enums with fields (e.g., PageAction), try to deserialize from JSON first.
         // For unit enums or when passed as a string, fall back to string-based conversion.
         let json_str: String = if let Ok(s) = <String as magnus::TryConvert>::try_convert(val) {
             s
@@ -6384,9 +6392,10 @@ impl magnus::TryConvert for AuthConfig {
 
 unsafe impl IntoValueFromNative for AuthConfig {}
 unsafe impl TryConvertOwned for AuthConfig {}
+
 impl AuthConfig {}
 
-#[derive(Clone, Copy, PartialEq, Eq, Debug, serde::Serialize, serde::Deserialize)]
+#[derive(Clone, Copy, PartialEq, Eq, Hash, Debug, serde::Serialize, serde::Deserialize)]
 #[serde(rename_all = "snake_case")]
 pub enum BrowserBackend {
     Chromiumoxide,
@@ -6411,7 +6420,14 @@ impl magnus::IntoValue for BrowserBackend {
 
 impl magnus::TryConvert for BrowserBackend {
     fn try_convert(val: magnus::Value) -> Result<Self, magnus::Error> {
-        let s: String = magnus::TryConvert::try_convert(val)?;
+        // IntoValue emits Symbols, including when this enum is a Hash key.
+        // Accept that exact representation as well as String inputs; do not
+        // coerce arbitrary objects through to_s.
+        let s: String = if let Some(symbol) = magnus::Symbol::from_value(val) {
+            symbol.name()?.into_owned()
+        } else {
+            magnus::TryConvert::try_convert(val)?
+        };
         // Accept the real serde wire value, the legacy always-snake_case symbol Magnus used to
         // emit unconditionally, and the verbatim PascalCase Rust variant name, so fixtures and
         // existing consumer code written in any of those styles keep working.
@@ -6429,7 +6445,7 @@ impl magnus::TryConvert for BrowserBackend {
 unsafe impl IntoValueFromNative for BrowserBackend {}
 unsafe impl TryConvertOwned for BrowserBackend {}
 
-#[derive(Clone, Copy, PartialEq, Eq, Debug, serde::Serialize, serde::Deserialize)]
+#[derive(Clone, Copy, PartialEq, Eq, Hash, Debug, serde::Serialize, serde::Deserialize)]
 #[serde(rename_all = "snake_case")]
 pub enum BrowserMode {
     Auto,
@@ -6458,7 +6474,14 @@ impl magnus::IntoValue for BrowserMode {
 
 impl magnus::TryConvert for BrowserMode {
     fn try_convert(val: magnus::Value) -> Result<Self, magnus::Error> {
-        let s: String = magnus::TryConvert::try_convert(val)?;
+        // IntoValue emits Symbols, including when this enum is a Hash key.
+        // Accept that exact representation as well as String inputs; do not
+        // coerce arbitrary objects through to_s.
+        let s: String = if let Some(symbol) = magnus::Symbol::from_value(val) {
+            symbol.name()?.into_owned()
+        } else {
+            magnus::TryConvert::try_convert(val)?
+        };
         // Accept the real serde wire value, the legacy always-snake_case symbol Magnus used to
         // emit unconditionally, and the verbatim PascalCase Rust variant name, so fixtures and
         // existing consumer code written in any of those styles keep working.
@@ -6478,7 +6501,7 @@ impl magnus::TryConvert for BrowserMode {
 unsafe impl IntoValueFromNative for BrowserMode {}
 unsafe impl TryConvertOwned for BrowserMode {}
 
-#[derive(Clone, Copy, PartialEq, Eq, Debug, serde::Serialize, serde::Deserialize)]
+#[derive(Clone, Copy, PartialEq, Eq, Hash, Debug, serde::Serialize, serde::Deserialize)]
 #[serde(rename_all = "snake_case")]
 pub enum BrowserWait {
     NetworkIdle,
@@ -6505,7 +6528,14 @@ impl magnus::IntoValue for BrowserWait {
 
 impl magnus::TryConvert for BrowserWait {
     fn try_convert(val: magnus::Value) -> Result<Self, magnus::Error> {
-        let s: String = magnus::TryConvert::try_convert(val)?;
+        // IntoValue emits Symbols, including when this enum is a Hash key.
+        // Accept that exact representation as well as String inputs; do not
+        // coerce arbitrary objects through to_s.
+        let s: String = if let Some(symbol) = magnus::Symbol::from_value(val) {
+            symbol.name()?.into_owned()
+        } else {
+            magnus::TryConvert::try_convert(val)?
+        };
         // Accept the real serde wire value, the legacy always-snake_case symbol Magnus used to
         // emit unconditionally, and the verbatim PascalCase Rust variant name, so fixtures and
         // existing consumer code written in any of those styles keep working.
@@ -6524,7 +6554,7 @@ impl magnus::TryConvert for BrowserWait {
 unsafe impl IntoValueFromNative for BrowserWait {}
 unsafe impl TryConvertOwned for BrowserWait {}
 
-#[derive(Clone, Copy, PartialEq, Eq, Debug, serde::Serialize, serde::Deserialize)]
+#[derive(Clone, Copy, PartialEq, Eq, Hash, Debug, serde::Serialize, serde::Deserialize)]
 #[serde(rename_all = "snake_case")]
 pub enum ContentFilterKind {
     Bm25,
@@ -6547,7 +6577,14 @@ impl magnus::IntoValue for ContentFilterKind {
 
 impl magnus::TryConvert for ContentFilterKind {
     fn try_convert(val: magnus::Value) -> Result<Self, magnus::Error> {
-        let s: String = magnus::TryConvert::try_convert(val)?;
+        // IntoValue emits Symbols, including when this enum is a Hash key.
+        // Accept that exact representation as well as String inputs; do not
+        // coerce arbitrary objects through to_s.
+        let s: String = if let Some(symbol) = magnus::Symbol::from_value(val) {
+            symbol.name()?.into_owned()
+        } else {
+            magnus::TryConvert::try_convert(val)?
+        };
         // Accept the real serde wire value, the legacy always-snake_case symbol Magnus used to
         // emit unconditionally, and the verbatim PascalCase Rust variant name, so fixtures and
         // existing consumer code written in any of those styles keep working.
@@ -6568,7 +6605,7 @@ unsafe impl TryConvertOwned for ContentFilterKind {}
 #[serde(tag = "type")]
 #[serde(rename_all = "snake_case")]
 pub enum CrawlEvent {
-    Page { result: CrawlPageResult },
+    Page { result: Box<CrawlPageResult> },
     Error { url: String, error: String },
     Complete { pages_crawled: usize },
 }
@@ -6592,7 +6629,8 @@ impl magnus::IntoValue for CrawlEvent {
 
 impl magnus::TryConvert for CrawlEvent {
     fn try_convert(val: magnus::Value) -> Result<Self, magnus::Error> {
-        // For data enums with fields (e.g., PageAction), try to deserialize from JSON first.
+        // Generated Data variants carry native DTOs, not JSON-compatible display strings.
+        let class_name: Option<String> = val.class().funcall("name", ())?; // For data enums with fields (e.g., PageAction), try to deserialize from JSON first.
         // For unit enums or when passed as a string, fall back to string-based conversion.
         let json_str: String = if let Ok(s) = <String as magnus::TryConvert>::try_convert(val) {
             s
@@ -6624,9 +6662,10 @@ impl magnus::TryConvert for CrawlEvent {
 
 unsafe impl IntoValueFromNative for CrawlEvent {}
 unsafe impl TryConvertOwned for CrawlEvent {}
+
 impl CrawlEvent {}
 
-#[derive(Clone, Copy, PartialEq, Eq, Debug, serde::Serialize, serde::Deserialize)]
+#[derive(Clone, Copy, PartialEq, Eq, Hash, Debug, serde::Serialize, serde::Deserialize)]
 #[serde(rename_all = "snake_case")]
 pub enum CrawlStrategyKind {
     Bfs,
@@ -6655,7 +6694,14 @@ impl magnus::IntoValue for CrawlStrategyKind {
 
 impl magnus::TryConvert for CrawlStrategyKind {
     fn try_convert(val: magnus::Value) -> Result<Self, magnus::Error> {
-        let s: String = magnus::TryConvert::try_convert(val)?;
+        // IntoValue emits Symbols, including when this enum is a Hash key.
+        // Accept that exact representation as well as String inputs; do not
+        // coerce arbitrary objects through to_s.
+        let s: String = if let Some(symbol) = magnus::Symbol::from_value(val) {
+            symbol.name()?.into_owned()
+        } else {
+            magnus::TryConvert::try_convert(val)?
+        };
         // Accept the real serde wire value, the legacy always-snake_case symbol Magnus used to
         // emit unconditionally, and the verbatim PascalCase Rust variant name, so fixtures and
         // existing consumer code written in any of those styles keep working.
@@ -6675,7 +6721,7 @@ impl magnus::TryConvert for CrawlStrategyKind {
 unsafe impl IntoValueFromNative for CrawlStrategyKind {}
 unsafe impl TryConvertOwned for CrawlStrategyKind {}
 
-#[derive(Clone, Copy, PartialEq, Eq, Debug, serde::Serialize, serde::Deserialize)]
+#[derive(Clone, Copy, PartialEq, Eq, Hash, Debug, serde::Serialize, serde::Deserialize)]
 #[serde(rename_all = "snake_case")]
 pub enum DocumentContentEncoding {
     Base64,
@@ -6698,7 +6744,14 @@ impl magnus::IntoValue for DocumentContentEncoding {
 
 impl magnus::TryConvert for DocumentContentEncoding {
     fn try_convert(val: magnus::Value) -> Result<Self, magnus::Error> {
-        let s: String = magnus::TryConvert::try_convert(val)?;
+        // IntoValue emits Symbols, including when this enum is a Hash key.
+        // Accept that exact representation as well as String inputs; do not
+        // coerce arbitrary objects through to_s.
+        let s: String = if let Some(symbol) = magnus::Symbol::from_value(val) {
+            symbol.name()?.into_owned()
+        } else {
+            magnus::TryConvert::try_convert(val)?
+        };
         // Accept the real serde wire value, the legacy always-snake_case symbol Magnus used to
         // emit unconditionally, and the verbatim PascalCase Rust variant name, so fixtures and
         // existing consumer code written in any of those styles keep working.
@@ -6715,7 +6768,7 @@ impl magnus::TryConvert for DocumentContentEncoding {
 unsafe impl IntoValueFromNative for DocumentContentEncoding {}
 unsafe impl TryConvertOwned for DocumentContentEncoding {}
 
-#[derive(Clone, Copy, PartialEq, Eq, Debug, serde::Serialize, serde::Deserialize)]
+#[derive(Clone, Copy, PartialEq, Eq, Hash, Debug, serde::Serialize, serde::Deserialize)]
 #[serde(rename_all = "snake_case")]
 pub enum FeedType {
     Rss,
@@ -6742,7 +6795,14 @@ impl magnus::IntoValue for FeedType {
 
 impl magnus::TryConvert for FeedType {
     fn try_convert(val: magnus::Value) -> Result<Self, magnus::Error> {
-        let s: String = magnus::TryConvert::try_convert(val)?;
+        // IntoValue emits Symbols, including when this enum is a Hash key.
+        // Accept that exact representation as well as String inputs; do not
+        // coerce arbitrary objects through to_s.
+        let s: String = if let Some(symbol) = magnus::Symbol::from_value(val) {
+            symbol.name()?.into_owned()
+        } else {
+            magnus::TryConvert::try_convert(val)?
+        };
         // Accept the real serde wire value, the legacy always-snake_case symbol Magnus used to
         // emit unconditionally, and the verbatim PascalCase Rust variant name, so fixtures and
         // existing consumer code written in any of those styles keep working.
@@ -6770,14 +6830,6 @@ pub enum HostMatcher {
     Cidr { value: String },
 }
 
-impl Default for HostMatcher {
-    fn default() -> Self {
-        Self::Exact {
-            value: Default::default(),
-        }
-    }
-}
-
 impl magnus::IntoValue for HostMatcher {
     fn into_value_with(self, handle: &Ruby) -> magnus::Value {
         match serde_json::to_value(&self) {
@@ -6789,7 +6841,8 @@ impl magnus::IntoValue for HostMatcher {
 
 impl magnus::TryConvert for HostMatcher {
     fn try_convert(val: magnus::Value) -> Result<Self, magnus::Error> {
-        // For data enums with fields (e.g., PageAction), try to deserialize from JSON first.
+        // Generated Data variants carry native DTOs, not JSON-compatible display strings.
+        let class_name: Option<String> = val.class().funcall("name", ())?; // For data enums with fields (e.g., PageAction), try to deserialize from JSON first.
         // For unit enums or when passed as a string, fall back to string-based conversion.
         let json_str: String = if let Ok(s) = <String as magnus::TryConvert>::try_convert(val) {
             s
@@ -6821,9 +6874,10 @@ impl magnus::TryConvert for HostMatcher {
 
 unsafe impl IntoValueFromNative for HostMatcher {}
 unsafe impl TryConvertOwned for HostMatcher {}
+
 impl HostMatcher {}
 
-#[derive(Clone, Copy, PartialEq, Eq, Debug, serde::Serialize, serde::Deserialize)]
+#[derive(Clone, Copy, PartialEq, Eq, Hash, Debug, serde::Serialize, serde::Deserialize)]
 #[serde(rename_all = "snake_case")]
 pub enum ImageSource {
     Img,
@@ -6854,7 +6908,14 @@ impl magnus::IntoValue for ImageSource {
 
 impl magnus::TryConvert for ImageSource {
     fn try_convert(val: magnus::Value) -> Result<Self, magnus::Error> {
-        let s: String = magnus::TryConvert::try_convert(val)?;
+        // IntoValue emits Symbols, including when this enum is a Hash key.
+        // Accept that exact representation as well as String inputs; do not
+        // coerce arbitrary objects through to_s.
+        let s: String = if let Some(symbol) = magnus::Symbol::from_value(val) {
+            symbol.name()?.into_owned()
+        } else {
+            magnus::TryConvert::try_convert(val)?
+        };
         // Accept the real serde wire value, the legacy always-snake_case symbol Magnus used to
         // emit unconditionally, and the verbatim PascalCase Rust variant name, so fixtures and
         // existing consumer code written in any of those styles keep working.
@@ -6874,7 +6935,7 @@ impl magnus::TryConvert for ImageSource {
 unsafe impl IntoValueFromNative for ImageSource {}
 unsafe impl TryConvertOwned for ImageSource {}
 
-#[derive(Clone, Copy, PartialEq, Eq, Debug, serde::Serialize, serde::Deserialize)]
+#[derive(Clone, Copy, PartialEq, Eq, Hash, Debug, serde::Serialize, serde::Deserialize)]
 #[serde(rename_all = "snake_case")]
 pub enum LinkType {
     Internal,
@@ -6903,7 +6964,14 @@ impl magnus::IntoValue for LinkType {
 
 impl magnus::TryConvert for LinkType {
     fn try_convert(val: magnus::Value) -> Result<Self, magnus::Error> {
-        let s: String = magnus::TryConvert::try_convert(val)?;
+        // IntoValue emits Symbols, including when this enum is a Hash key.
+        // Accept that exact representation as well as String inputs; do not
+        // coerce arbitrary objects through to_s.
+        let s: String = if let Some(symbol) = magnus::Symbol::from_value(val) {
+            symbol.name()?.into_owned()
+        } else {
+            magnus::TryConvert::try_convert(val)?
+        };
         // Accept the real serde wire value, the legacy always-snake_case symbol Magnus used to
         // emit unconditionally, and the verbatim PascalCase Rust variant name, so fixtures and
         // existing consumer code written in any of those styles keep working.
@@ -6974,7 +7042,8 @@ impl magnus::IntoValue for PageAction {
 
 impl magnus::TryConvert for PageAction {
     fn try_convert(val: magnus::Value) -> Result<Self, magnus::Error> {
-        // For data enums with fields (e.g., PageAction), try to deserialize from JSON first.
+        // Generated Data variants carry native DTOs, not JSON-compatible display strings.
+        let class_name: Option<String> = val.class().funcall("name", ())?; // For data enums with fields (e.g., PageAction), try to deserialize from JSON first.
         // For unit enums or when passed as a string, fall back to string-based conversion.
         let json_str: String = if let Ok(s) = <String as magnus::TryConvert>::try_convert(val) {
             s
@@ -7006,9 +7075,10 @@ impl magnus::TryConvert for PageAction {
 
 unsafe impl IntoValueFromNative for PageAction {}
 unsafe impl TryConvertOwned for PageAction {}
+
 impl PageAction {}
 
-#[derive(Clone, Copy, PartialEq, Eq, Debug, serde::Serialize, serde::Deserialize)]
+#[derive(Clone, Copy, PartialEq, Eq, Hash, Debug, serde::Serialize, serde::Deserialize)]
 #[serde(rename_all = "camelCase")]
 pub enum ScrollDirection {
     Up,
@@ -7033,7 +7103,14 @@ impl magnus::IntoValue for ScrollDirection {
 
 impl magnus::TryConvert for ScrollDirection {
     fn try_convert(val: magnus::Value) -> Result<Self, magnus::Error> {
-        let s: String = magnus::TryConvert::try_convert(val)?;
+        // IntoValue emits Symbols, including when this enum is a Hash key.
+        // Accept that exact representation as well as String inputs; do not
+        // coerce arbitrary objects through to_s.
+        let s: String = if let Some(symbol) = magnus::Symbol::from_value(val) {
+            symbol.name()?.into_owned()
+        } else {
+            magnus::TryConvert::try_convert(val)?
+        };
         // Accept the real serde wire value, the legacy always-snake_case symbol Magnus used to
         // emit unconditionally, and the verbatim PascalCase Rust variant name, so fixtures and
         // existing consumer code written in any of those styles keep working.
@@ -8619,7 +8696,7 @@ impl From<crawlberg::CrawlEvent> for CrawlEvent {
     fn from(val: crawlberg::CrawlEvent) -> Self {
         match val {
             crawlberg::CrawlEvent::Page { result } => Self::Page {
-                result: (*result).into(),
+                result: Box::new((*result).into()),
             },
             crawlberg::CrawlEvent::Error { url, error } => Self::Error {
                 url: url.to_string(),

data/lib/crawlberg/native.rb

@@ -1,5 +1,5 @@
 # This file is auto-generated by alef — DO NOT EDIT.
-# alef:hash:925532ba2f606ce36a0d3444d5c96407ca56a1c55bcd256372a1ea2a383d11e5
+# alef:hash:d323ed0aab9764791b48678b68c8fedcf8e7b3e1dbe50ec190a9023e68acc3bb
 # To regenerate: alef generate
 # To verify freshness: alef verify
 # frozen_string_literal: true
@@ -68,12 +68,10 @@ module Crawlberg
 
     # Username sent in the `Authorization: Basic` header.
     sig { returns(String) }
-    # rubocop:disable Lint/UselessMethodDefinition
-    def username = super
+    def username = Data.instance_method(:to_h).bind_call(self).fetch(:username)
     # Password sent in the `Authorization: Basic` header.
     sig { returns(String) }
-    # rubocop:disable Lint/UselessMethodDefinition
-    def password = super
+    def password = Data.instance_method(:to_h).bind_call(self).fetch(:password)
     sig { returns(T::Boolean) }
     def basic? = true
     sig { returns(T::Boolean) }
@@ -94,8 +92,7 @@ module Crawlberg
 
     # Token sent in the `Authorization: Bearer` header.
     sig { returns(String) }
-    # rubocop:disable Lint/UselessMethodDefinition
-    def token = super
+    def token = Data.instance_method(:to_h).bind_call(self).fetch(:token)
     sig { returns(T::Boolean) }
     def basic? = false
     sig { returns(T::Boolean) }
@@ -116,12 +113,10 @@ module Crawlberg
 
     # HTTP header name to set on each request.
     sig { returns(String) }
-    # rubocop:disable Lint/UselessMethodDefinition
-    def name = super
+    def name = Data.instance_method(:to_h).bind_call(self).fetch(:name)
     # HTTP header value to send.
     sig { returns(String) }
-    # rubocop:disable Lint/UselessMethodDefinition
-    def value = super
+    def value = Data.instance_method(:to_h).bind_call(self).fetch(:value)
     sig { returns(T::Boolean) }
     def basic? = false
     sig { returns(T::Boolean) }
@@ -175,8 +170,7 @@ module Crawlberg
 
     # The crawled page result.
     sig { returns(CrawlPageResult) }
-    # rubocop:disable Lint/UselessMethodDefinition
-    def result = super
+    def result = Data.instance_method(:to_h).bind_call(self).fetch(:result)
     sig { returns(T::Boolean) }
     def page? = true
     sig { returns(T::Boolean) }
@@ -197,12 +191,10 @@ module Crawlberg
 
     # The URL that failed.
     sig { returns(String) }
-    # rubocop:disable Lint/UselessMethodDefinition
-    def url = super
+    def url = Data.instance_method(:to_h).bind_call(self).fetch(:url)
     # The error message.
     sig { returns(String) }
-    # rubocop:disable Lint/UselessMethodDefinition
-    def error = super
+    def error = Data.instance_method(:to_h).bind_call(self).fetch(:error)
     sig { returns(T::Boolean) }
     def page? = false
     sig { returns(T::Boolean) }
@@ -223,8 +215,7 @@ module Crawlberg
 
     # Total number of pages crawled.
     sig { returns(Integer) }
-    # rubocop:disable Lint/UselessMethodDefinition
-    def pages_crawled = super
+    def pages_crawled = Data.instance_method(:to_h).bind_call(self).fetch(:pages_crawled)
     sig { returns(T::Boolean) }
     def page? = false
     sig { returns(T::Boolean) }
@@ -287,8 +278,7 @@ module Crawlberg
 
     # The hostname to match.
     sig { returns(String) }
-    # rubocop:disable Lint/UselessMethodDefinition
-    def value = super
+    def value = Data.instance_method(:to_h).bind_call(self).fetch(:value)
     sig { returns(T::Boolean) }
     def exact? = true
     sig { returns(T::Boolean) }
@@ -309,8 +299,7 @@ module Crawlberg
 
     # The dot-prefixed suffix to match. A leading dot is optional.
     sig { returns(String) }
-    # rubocop:disable Lint/UselessMethodDefinition
-    def value = super
+    def value = Data.instance_method(:to_h).bind_call(self).fetch(:value)
     sig { returns(T::Boolean) }
     def exact? = false
     sig { returns(T::Boolean) }
@@ -332,8 +321,7 @@ module Crawlberg
     # The CIDR block. Validated when built through [`HostMatcher::cidr`] or
     # deserialization.
     sig { returns(String) }
-    # rubocop:disable Lint/UselessMethodDefinition
-    def value = super
+    def value = Data.instance_method(:to_h).bind_call(self).fetch(:value)
     sig { returns(T::Boolean) }
     def exact? = false
     sig { returns(T::Boolean) }
@@ -395,8 +383,7 @@ module Crawlberg
 
     # CSS selector for the element to click.
     sig { returns(String) }
-    # rubocop:disable Lint/UselessMethodDefinition
-    def selector = super
+    def selector = Data.instance_method(:to_h).bind_call(self).fetch(:selector)
     sig { returns(T::Boolean) }
     def click? = true
     sig { returns(T::Boolean) }
@@ -427,12 +414,10 @@ module Crawlberg
 
     # CSS selector for the input element.
     sig { returns(String) }
-    # rubocop:disable Lint/UselessMethodDefinition
-    def selector = super
+    def selector = Data.instance_method(:to_h).bind_call(self).fetch(:selector)
     # Text to type into the element.
     sig { returns(String) }
-    # rubocop:disable Lint/UselessMethodDefinition
-    def text = super
+    def text = Data.instance_method(:to_h).bind_call(self).fetch(:text)
     sig { returns(T::Boolean) }
     def click? = false
     sig { returns(T::Boolean) }
@@ -463,8 +448,7 @@ module Crawlberg
 
     # Key name to press.
     sig { returns(String) }
-    # rubocop:disable Lint/UselessMethodDefinition
-    def key = super
+    def key = Data.instance_method(:to_h).bind_call(self).fetch(:key)
     sig { returns(T::Boolean) }
     def click? = false
     sig { returns(T::Boolean) }
@@ -495,16 +479,13 @@ module Crawlberg
 
     # Direction to scroll.
     sig { returns(ScrollDirection) }
-    # rubocop:disable Lint/UselessMethodDefinition
-    def direction = super
+    def direction = Data.instance_method(:to_h).bind_call(self).fetch(:direction)
     # Optional CSS selector for a scrollable element. Scrolls the page if absent.
     sig { returns(T.nilable(String)) }
-    # rubocop:disable Lint/UselessMethodDefinition
-    def selector = super
+    def selector = Data.instance_method(:to_h).bind_call(self).fetch(:selector)
     # Optional pixel amount to scroll. Uses a default if absent.
     sig { returns(T.nilable(Integer)) }
-    # rubocop:disable Lint/UselessMethodDefinition
-    def amount = super
+    def amount = Data.instance_method(:to_h).bind_call(self).fetch(:amount)
     sig { returns(T::Boolean) }
     def click? = false
     sig { returns(T::Boolean) }
@@ -539,12 +520,10 @@ module Crawlberg
 
     # Milliseconds to wait. Ignored if `selector` is provided.
     sig { returns(T.nilable(Integer)) }
-    # rubocop:disable Lint/UselessMethodDefinition
-    def milliseconds = super
+    def milliseconds = Data.instance_method(:to_h).bind_call(self).fetch(:milliseconds)
     # CSS selector to wait for.
     sig { returns(T.nilable(String)) }
-    # rubocop:disable Lint/UselessMethodDefinition
-    def selector = super
+    def selector = Data.instance_method(:to_h).bind_call(self).fetch(:selector)
     sig { returns(T::Boolean) }
     def click? = false
     sig { returns(T::Boolean) }
@@ -579,8 +558,7 @@ module Crawlberg
     # `full_page` (snake_case) alias so language bindings and fixtures can
     # use either convention without error.
     sig { returns(T.nilable(T::Boolean)) }
-    # rubocop:disable Lint/UselessMethodDefinition
-    def full_page = super
+    def full_page = Data.instance_method(:to_h).bind_call(self).fetch(:full_page)
     sig { returns(T::Boolean) }
     def click? = false
     sig { returns(T::Boolean) }
@@ -616,8 +594,7 @@ module Crawlberg
 
     # JavaScript source code to execute. Max 1 MB.
     sig { returns(String) }
-    # rubocop:disable Lint/UselessMethodDefinition
-    def script = super
+    def script = Data.instance_method(:to_h).bind_call(self).fetch(:script)
     sig { returns(T::Boolean) }
     def click? = false
     sig { returns(T::Boolean) }

data/lib/crawlberg/version.rb

@@ -1,10 +1,10 @@
 # This file is auto-generated by alef — DO NOT EDIT.
-# alef:hash:f45711d921997740dbd2259778d7ff9e5d0d75d2ce27ef835c9ce8916b1dc19d
+# alef:hash:fdaf5ebf5fb30639d0b25b5f62a7044ed59029769d49609135c3bfd37b6f18e8
 # To regenerate: alef generate
 # To verify freshness: alef verify
 # frozen_string_literal: true
 
 module Crawlberg
   ## The version string for this package.
-  VERSION = "1.7.1"
+  VERSION = "1.7.2"
 end

data/sig/types.rbs

@@ -1,5 +1,5 @@
 # This file is auto-generated by alef — DO NOT EDIT.
-# alef:hash:1a2c1bf0fb6b53091ffe21fcecbee4a7e5f8214d544addbfe5c39f34e5aa7590
+# alef:hash:ddb9c01b1ccd8b583329fc083873cbeaec5fed7790bc44b0db41df5ccc4b7d91
 # To regenerate: alef generate
 # To verify freshness: alef verify
 
@@ -70,11 +70,11 @@ def initialize: (?results: Array[BatchScrapeResult], ?total_count: Integer, ?com
     end
 
     class BrowserConfig
-        attr_reader mode: BrowserMode
-        attr_reader backend: BrowserBackend
+        attr_reader mode: enum_BrowserMode
+        attr_reader backend: enum_BrowserBackend
         attr_reader endpoint: String?
         attr_reader timeout: Integer
-        attr_reader wait: BrowserWait
+        attr_reader wait: enum_BrowserWait
         attr_reader wait_selector: String?
         attr_reader extra_wait: Integer?
         attr_reader proxy: ProxyConfig?
@@ -84,7 +84,7 @@ def initialize: (?results: Array[BatchScrapeResult], ?total_count: Integer, ?com
         attr_reader capture_network_events: bool
         attr_reader session_affinity: bool
 
-def initialize: (?mode: BrowserMode, ?backend: BrowserBackend, ?endpoint: String, ?timeout: Integer, ?wait: BrowserWait, ?wait_selector: String, ?extra_wait: Integer, ?proxy: ProxyConfig, ?block_url_patterns: Array[String], ?eval_script: String, ?robots_user_agent: String, ?capture_network_events: bool, ?session_affinity: bool) -> void
+def initialize: (?mode: enum_BrowserMode, ?backend: enum_BrowserBackend, ?endpoint: String, ?timeout: Integer, ?wait: enum_BrowserWait, ?wait_selector: String, ?extra_wait: Integer, ?proxy: ProxyConfig, ?block_url_patterns: Array[String], ?eval_script: String, ?robots_user_agent: String, ?capture_network_events: bool, ?session_affinity: bool) -> void
     end
 
     class BrowserExtras
@@ -141,8 +141,8 @@ def initialize: (?output_format: String, ?preprocessing_preset: String, ?remove_
         attr_reader max_pages: Integer?
         attr_reader max_links_per_page: Integer?
         attr_reader max_concurrent: Integer?
-        attr_reader crawl_strategy: CrawlStrategyKind
-        attr_reader content_filter: ContentFilterKind?
+        attr_reader crawl_strategy: enum_CrawlStrategyKind
+        attr_reader content_filter: enum_ContentFilterKind?
         attr_reader bm25_query: String?
         attr_reader bm25_threshold: Float?
         attr_reader respect_robots_txt: bool
@@ -166,7 +166,7 @@ def initialize: (?output_format: String, ?preprocessing_preset: String, ?remove_
         attr_reader map_limit: Integer?
         attr_reader map_search: String?
         attr_reader download_assets: bool
-    attr_reader asset_types: Array[AssetCategory]
+    attr_reader asset_types: Array[enum_AssetCategory]
         attr_reader max_asset_size: Integer?
         attr_reader browser: BrowserConfig
         attr_reader proxy: ProxyConfig?
@@ -178,14 +178,14 @@ def initialize: (?output_format: String, ?preprocessing_preset: String, ?remove_
         attr_reader document_max_size: Integer?
     attr_reader document_mime_types: Array[String]
         attr_reader document_output_dir: String?
-        attr_reader document_content_encoding: DocumentContentEncoding?
+        attr_reader document_content_encoding: enum_DocumentContentEncoding?
         attr_reader warc_output: String?
         attr_reader browser_profile: String?
         attr_reader save_browser_profile: bool
         attr_reader ssrf: SsrfPolicy
         attr_reader ssrf_deny_private_explicit: bool?
 
-def initialize: (?max_depth: Integer, ?max_pages: Integer, ?max_links_per_page: Integer, ?max_concurrent: Integer, ?crawl_strategy: CrawlStrategyKind, ?content_filter: ContentFilterKind, ?bm25_query: String, ?bm25_threshold: Float, ?respect_robots_txt: bool, ?soft_http_errors: bool, ?user_agent: String, ?stay_on_domain: bool, ?allow_subdomains: bool, ?include_paths: Array[String], ?exclude_paths: Array[String], ?custom_headers: Hash[String, String], ?request_timeout: Integer, ?rate_limit_ms: Integer, ?max_redirects: Integer, ?retry_count: Integer, ?retry_codes: Array[Integer], ?cookies_enabled: bool, ?auth: AuthConfig, ?max_body_size: Integer, ?remove_tags: Array[String], ?content: ContentConfig, ?map_limit: Integer, ?map_search: String, ?download_assets: bool, ?asset_types: Array[AssetCategory], ?max_asset_size: Integer, ?browser: BrowserConfig, ?proxy: ProxyConfig, ?user_agents: Array[String], ?capture_screenshot: bool, ?follow_document_urls: bool, ?document_url_depth: Integer, ?download_documents: bool, ?document_max_size: Integer, ?document_mime_types: Array[String], ?document_output_dir: String, ?document_content_encoding: DocumentContentEncoding, ?warc_output: String, ?browser_profile: String, ?save_browser_profile: bool, ?ssrf: SsrfPolicy, ?ssrf_deny_private_explicit: bool) -> void
+def initialize: (?max_depth: Integer, ?max_pages: Integer, ?max_links_per_page: Integer, ?max_concurrent: Integer, ?crawl_strategy: enum_CrawlStrategyKind, ?content_filter: enum_ContentFilterKind, ?bm25_query: String, ?bm25_threshold: Float, ?respect_robots_txt: bool, ?soft_http_errors: bool, ?user_agent: String, ?stay_on_domain: bool, ?allow_subdomains: bool, ?include_paths: Array[String], ?exclude_paths: Array[String], ?custom_headers: Hash[String, String], ?request_timeout: Integer, ?rate_limit_ms: Integer, ?max_redirects: Integer, ?retry_count: Integer, ?retry_codes: Array[Integer], ?cookies_enabled: bool, ?auth: AuthConfig, ?max_body_size: Integer, ?remove_tags: Array[String], ?content: ContentConfig, ?map_limit: Integer, ?map_search: String, ?download_assets: bool, ?asset_types: Array[enum_AssetCategory], ?max_asset_size: Integer, ?browser: BrowserConfig, ?proxy: ProxyConfig, ?user_agents: Array[String], ?capture_screenshot: bool, ?follow_document_urls: bool, ?document_url_depth: Integer, ?download_documents: bool, ?document_max_size: Integer, ?document_mime_types: Array[String], ?document_output_dir: String, ?document_content_encoding: enum_DocumentContentEncoding, ?warc_output: String, ?browser_profile: String, ?save_browser_profile: bool, ?ssrf: SsrfPolicy, ?ssrf_deny_private_explicit: bool) -> void
     def validate: () -> void
     end
 
@@ -245,10 +245,10 @@ def initialize: (?pages: Array[CrawlPageResult], ?final_url: String, ?redirect_c
         attr_reader content_hash: String
         attr_reader mime_type: String?
         attr_reader size: Integer
-        attr_reader asset_category: AssetCategory
+        attr_reader asset_category: enum_AssetCategory
         attr_reader html_tag: String?
 
-    def initialize: (?url: String, ?content_hash: String, ?mime_type: String, ?size: Integer, ?asset_category: AssetCategory, ?html_tag: String) -> void
+    def initialize: (?url: String, ?content_hash: String, ?mime_type: String, ?size: Integer, ?asset_category: enum_AssetCategory, ?html_tag: String) -> void
     end
 
     class DownloadedDocument
@@ -287,9 +287,9 @@ def initialize: (?url: String, ?mime_type: String, ?size: Integer, ?filename: St
     class FeedInfo
         attr_reader url: String
         attr_reader title: String?
-        attr_reader feed_type: FeedType
+        attr_reader feed_type: enum_FeedType
 
-    def initialize: (?url: String, ?title: String, ?feed_type: FeedType) -> void
+    def initialize: (?url: String, ?title: String, ?feed_type: enum_FeedType) -> void
     end
 
     class HeadingInfo
@@ -311,9 +311,9 @@ def initialize: (?url: String, ?mime_type: String, ?size: Integer, ?filename: St
         attr_reader alt: String?
         attr_reader width: Integer?
         attr_reader height: Integer?
-        attr_reader source: ImageSource
+        attr_reader source: enum_ImageSource
 
-    def initialize: (?url: String, ?alt: String, ?width: Integer, ?height: Integer, ?source: ImageSource) -> void
+    def initialize: (?url: String, ?alt: String, ?width: Integer, ?height: Integer, ?source: enum_ImageSource) -> void
     end
 
     class InteractionResult
@@ -336,11 +336,11 @@ def initialize: (?action_results: Array[ActionResult], ?final_html: String, ?fin
     class LinkInfo
         attr_reader url: String
         attr_reader text: String
-        attr_reader link_type: LinkType
+        attr_reader link_type: enum_LinkType
         attr_reader rel: String?
         attr_reader nofollow: bool
 
-    def initialize: (?url: String, ?text: String, ?link_type: LinkType, ?rel: String, ?nofollow: bool) -> void
+    def initialize: (?url: String, ?text: String, ?link_type: enum_LinkType, ?rel: String, ?nofollow: bool) -> void
     end
 
     class MapResult
@@ -480,61 +480,39 @@ def initialize: (?status_code: Integer, ?final_url: String, ?content_type: Strin
 def initialize: (?deny_private: bool, ?allowlist: Array[HostMatcher], ?max_redirects: Integer, ?scheme_allowlist: Array[String]) -> void
     end
 
-    class AssetCategory
-        type value = :document | :image | :audio | :video | :font | :stylesheet | :script | :archive | :data | :other
-    end
+    type enum_AssetCategory = :document | :image | :audio | :video | :font | :stylesheet | :script | :archive | :data | :other
 
     class AuthConfig
     end
 
-    class BrowserBackend
-        type value = :chromiumoxide | :native
-    end
+    type enum_BrowserBackend = :chromiumoxide | :native
 
-    class BrowserMode
-        type value = :auto | :always | :never | :stealth
-    end
+    type enum_BrowserMode = :auto | :always | :never | :stealth
 
-    class BrowserWait
-        type value = :network_idle | :selector | :fixed
-    end
+    type enum_BrowserWait = :network_idle | :selector | :fixed
 
-    class ContentFilterKind
-        type value = :bm25
-    end
+    type enum_ContentFilterKind = :bm25
 
     class CrawlEvent
     end
 
-    class CrawlStrategyKind
-        type value = :bfs | :dfs | :best_first | :adaptive
-    end
+    type enum_CrawlStrategyKind = :bfs | :dfs | :best_first | :adaptive
 
-    class DocumentContentEncoding
-        type value = :base64
-    end
+    type enum_DocumentContentEncoding = :base64
 
-    class FeedType
-        type value = :rss | :atom | :json_feed
-    end
+    type enum_FeedType = :rss | :atom | :json_feed
 
     class HostMatcher
     end
 
-    class ImageSource
-        type value = :img | :picture_source | :"og:image" | :"twitter:image"
-    end
+    type enum_ImageSource = :img | :picture_source | :"og:image" | :"twitter:image"
 
-    class LinkType
-        type value = :internal | :external | :anchor | :document
-    end
+    type enum_LinkType = :internal | :external | :anchor | :document
 
     class PageAction
     end
 
-    class ScrollDirection
-        type value = :up | :down
-    end
+    type enum_ScrollDirection = :up | :down
 
 def self.batch_crawl: (CrawlEngineHandle engine, Array[String] urls) -> BatchCrawlResults
 

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: crawlberg
 version: !ruby/object:Gem::Version
-  version: 1.7.1
+  version: 1.7.2
 platform: ruby
 authors:
 - Xberg Team
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2026-09-16 00:00:00.000000000 Z
+date: 2026-09-24 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rb_sys