crawlberg 1.4.0 → 1.4.2

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 8cf439052944de6e1b0749a460cbedabf63f50d76e41225872af8faebc34b222
-  data.tar.gz: 1096be9c99bf1c6df8089d5d24a914c560f90d2d3666ea410a60ec7cfa850059
+  metadata.gz: f597e0e18bfca18b073ac6647f09704f8f811d3083175ef73a01e1d0a467355a
+  data.tar.gz: 034ab10f2cd798d6bf32c457b7b5abbae47737332c43c492b80345fa85eef7ac
 SHA512:
-  metadata.gz: 289f0fe593287a255c9203cb192075d25a440f5ada41d67cb8b608c81f9e7c26897d9478f787ae732891cf71896ed3e4a0a14e46723e6ff1bee1d981cb6206d0
-  data.tar.gz: 243fc300050abbd78a31aff30d8845498cd35dd8e3f0e83524b4d7a7b58e60ea10b53fbaa847f42cb4f80d2e94e37bcd5d277cc4e721e4475e4bb838d6063c6b
+  metadata.gz: bf52c93eee44c22122a61090b0a7df4824674f24e7718308f24f9c0d535e7f589ddbc903f1baa915842923e460f4dcd5b9995cbf185d629d6151571a9c431947
+  data.tar.gz: 8e7e088921ac8276371fa54c61429b746d2bce838bb37e643f12e2ef579e515e08fa526eaabf6a0e0aa9b27e01c6cfa9edb97d0eff70de1771fe632c1bb36780

data/README.md

@@ -1,5 +1,5 @@
 <!-- This file is auto-generated by alef — DO NOT EDIT. -->
-<!-- alef:hash:b2d11257f40cf60f8f51c4f4352c2bee60792489ae695464d458d193c1f6871a -->
+<!-- alef:hash:442342be79ebc6986eda548e4fe12a6ce7adfbf0ae84fde27ddcc064f25ff0be -->
 <!-- To regenerate: alef readme -->
 <!-- To verify freshness: alef verify -->
 

data/ext/crawlberg_rb/native/Cargo.lock

@@ -99,7 +99,7 @@ checksum = "82f6aeea286b8eb4dd3431a1be1b59d290ace00f5bfd8e2a159bc2a05e2c1667"
 dependencies = [
  "proc-macro2",
  "quote",
- "syn 3.0.3",
+ "syn 3.0.4",
 ]
 
 [[package]]
@@ -549,9 +549,9 @@ dependencies = [
 
 [[package]]
 name = "crawlberg"
-version = "1.4.0"
+version = "1.4.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "cd756a90c129c58039e58485ecd1a054940305a5f7a374d5a522421cf597c878"
+checksum = "723dbd626a68331c8b235e0c1878d38630fc827b9df6e86bf176250e87a54520"
 dependencies = [
  "ahash",
  "aho-corasick",
@@ -593,7 +593,7 @@ dependencies = [
 
 [[package]]
 name = "crawlberg-rb"
-version = "1.4.0"
+version = "1.4.2"
 dependencies = [
  "crawlberg",
  "futures",
@@ -606,9 +606,9 @@ dependencies = [
 
 [[package]]
 name = "crc32fast"
-version = "1.5.0"
+version = "1.5.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "9481c1c90cbf2ac953f07c8d4a58aa3945c425b7185c9154d67a65e4230da511"
+checksum = "8498c871161e1742aaa9d52551b2d6ebdd4c3d45a3be423e3728f33b955be550"
 dependencies = [
  "cfg-if",
 ]
@@ -720,7 +720,7 @@ checksum = "c6232dd377dcc64799954cbd3a9bb882e9cdc1308ccd87b1c098f1fb2eaf82a8"
 dependencies = [
  "proc-macro2",
  "quote",
- "syn 3.0.3",
+ "syn 3.0.4",
 ]
 
 [[package]]
@@ -886,7 +886,7 @@ checksum = "9fb9654ba8355388abeb8dcb4fc62f511300867002afc858860463bdd9fe0c44"
 dependencies = [
  "proc-macro2",
  "quote",
- "syn 3.0.3",
+ "syn 3.0.4",
 ]
 
 [[package]]
@@ -1036,9 +1036,9 @@ checksum = "c9356095b4b41197bba32173600e1582792cda618f65d12f68e2e77d273413c5"
 
 [[package]]
 name = "html-to-markdown-rs"
-version = "3.11.4"
+version = "3.11.6"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "6331a4c21bc99309e5863092660c08505c426eb1158e0472013487ecff85412b"
+checksum = "c5fe2583aad3e87439e40d841fed327a60892bb9946da1e3a5788ee7f0f4fe3c"
 dependencies = [
  "ahash",
  "astral-tl",
@@ -1527,9 +1527,9 @@ dependencies = [
 
 [[package]]
 name = "log"
-version = "0.4.33"
+version = "0.4.34"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "0ceec5bc11778974d1bcb055b18002eba7f4b3518b6a0081b3af5f21666da9ad"
+checksum = "f9f8bd3e56ce4dfc153cf470fffbfa98c7620958b312ca5c3a4b8d5181fd13c6"
 
 [[package]]
 name = "lru-slab"
@@ -2377,7 +2377,7 @@ checksum = "e7a5d71263a5a7d47b41f6b3f06ba276f10cc18b0931f1799f710578e2309348"
 dependencies = [
  "proc-macro2",
  "quote",
- "syn 3.0.3",
+ "syn 3.0.4",
 ]
 
 [[package]]
@@ -2551,9 +2551,9 @@ dependencies = [
 
 [[package]]
 name = "syn"
-version = "3.0.3"
+version = "3.0.4"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "53e9bae58849f64dfa4f5d5ae372c8341f7305f82a3868709269343628b659a3"
+checksum = "e6275cddf4610d1775e6d1fe9469b2e77d0f39fd98fb7450901b821e0c53649f"
 dependencies = [
  "proc-macro2",
  "quote",
@@ -2627,7 +2627,7 @@ checksum = "bc04cd3e1236dd4a98afca4569f2deb3f120e5422a4023be2cb683f8486292af"
 dependencies = [
  "proc-macro2",
  "quote",
- "syn 3.0.3",
+ "syn 3.0.4",
 ]
 
 [[package]]
@@ -2718,7 +2718,7 @@ checksum = "78773a2a397f451582ce068015985c33193cf6dea8b74d2a639fe457b2f07b0e"
 dependencies = [
  "proc-macro2",
  "quote",
- "syn 3.0.3",
+ "syn 3.0.4",
 ]
 
 [[package]]
@@ -3447,7 +3447,7 @@ checksum = "34df6fc39dbd26ddc9c10e6a2984476e13acce22e64e4487636ef494369225da"
 dependencies = [
  "proc-macro2",
  "quote",
- "syn 3.0.3",
+ "syn 3.0.4",
 ]
 
 [[package]]

data/ext/crawlberg_rb/native/Cargo.toml

@@ -1,11 +1,11 @@
 # This file is auto-generated by alef — DO NOT EDIT.
-# alef:hash:5890f3fa9fbc0338c6eb7af1d95fa187d4152d63b319069bfd2d5b67dca06442
+# alef:hash:7cb6aa801d5f904cfc9a0079f817dfe7816c1d5b05bfd24c2eec864119ad7ee6
 # To regenerate: alef generate
 # To verify freshness: alef verify
 
 [package]
 name = "crawlberg-rb"
-version = "1.4.0"
+version = "1.4.2"
 edition = "2024"
 license = "MIT"
 description = "High-performance web crawling engine"
@@ -22,7 +22,7 @@ path = "../src/lib.rs"
 crate-type = ["cdylib"]
 
 [dependencies]
-crawlberg = { version = "1.4.0", features = ["interact", "browser-chromiumoxide"] }
+crawlberg = { version = "1.4.2", features = ["interact", "browser-chromiumoxide"] }
 futures = "0.3"
 magnus = "0.8"
 rb-sys = ">=0.9, <0.9.128"

data/ext/crawlberg_rb/native/extconf.rb

@@ -1,5 +1,5 @@
 # This file is auto-generated by alef — DO NOT EDIT.
-# alef:hash:3553e8a3571c57268a7a40eac8287287ea943334492411222c0c3ac94e5a191e
+# alef:hash:5326e03e26baf331d1f85eb82de9b52903317cf694aa9d920b1b22dc6c1f009f
 # To regenerate: alef generate
 # To verify freshness: alef verify
 

data/ext/crawlberg_rb/src/lib.rs

@@ -1,5 +1,5 @@
 // This file is auto-generated by alef. DO NOT EDIT.
-// alef:hash:3c4509c0d17fbbc9f751520beae401c94c0dc67c713271806c4220ba7b2a1ad6
+// alef:hash:c8bd1f23995c2c54aa56bd9e9dd95391dff81fd7eefe7a65306513caf8564400
 // Re-generate with: alef generate
 #![allow(dead_code, unused_imports, unused_variables)]
 #![allow(
@@ -4482,6 +4482,7 @@ pub struct SsrfPolicy {
     deny_private: bool,
     allowlist: Vec<HostMatcher>,
     max_redirects: u8,
+    scheme_allowlist: Vec<String>,
 }
 
 unsafe impl IntoValueFromNative for SsrfPolicy {}
@@ -4537,6 +4538,14 @@ impl SsrfPolicy {
                 .get(ruby.to_symbol("max_redirects"))
                 .and_then(|v| u8::try_convert(v).ok())
                 .unwrap_or(5),
+            scheme_allowlist: kwargs
+                .get(ruby.to_symbol("scheme_allowlist"))
+                .and_then(|v| <Vec<String>>::try_convert(v).ok())
+                .unwrap_or(
+                    serde_json::from_str::<crawlberg::SsrfPolicy>(r#"{}"#)
+                        .expect("alef-generated default JSON for `SsrfPolicy` failed to deserialize")
+                        .scheme_allowlist,
+                ),
         })
     }
 
@@ -4551,6 +4560,10 @@ impl SsrfPolicy {
     fn max_redirects(&self) -> u8 {
         self.max_redirects
     }
+
+    fn scheme_allowlist(&self) -> Vec<String> {
+        self.scheme_allowlist.clone()
+    }
 }
 
 #[derive(Clone, Copy, PartialEq, Eq, Debug, serde::Serialize, serde::Deserialize)]
@@ -6510,7 +6523,7 @@ impl From<MarkdownResult> for crawlberg::MarkdownResult {
             tables: val
                 .tables
                 .into_iter()
-                .filter_map(|s| serde_json::from_str(&s).ok())
+                .map(|s| serde_json::from_str(&s).unwrap_or_default())
                 .collect(),
             warnings: val.warnings.into_iter().collect(),
             citations: val.citations,
@@ -6801,6 +6814,7 @@ impl From<SsrfPolicy> for crawlberg::SsrfPolicy {
             deny_private: val.deny_private,
             allowlist: val.allowlist.into_iter().map(Into::into).collect(),
             max_redirects: val.max_redirects,
+            scheme_allowlist: val.scheme_allowlist.into_iter().collect(),
             ..Default::default()
         }
     }
@@ -6813,6 +6827,7 @@ impl From<crawlberg::SsrfPolicy> for SsrfPolicy {
             deny_private: val.deny_private,
             allowlist: val.allowlist.into_iter().map(Into::into).collect(),
             max_redirects: val.max_redirects,
+            scheme_allowlist: val.scheme_allowlist.into_iter().collect(),
         }
     }
 }
@@ -6854,15 +6869,9 @@ impl From<crawlberg::AssetCategory> for AssetCategory {
 impl From<AuthConfig> for crawlberg::AuthConfig {
     fn from(val: AuthConfig) -> Self {
         match val {
-            AuthConfig::Basic { username, password } => Self::Basic {
-                username: username,
-                password: password,
-            },
-            AuthConfig::Bearer { token } => Self::Bearer { token: token },
-            AuthConfig::Header { name, value } => Self::Header {
-                name: name,
-                value: value,
-            },
+            AuthConfig::Basic { username, password } => Self::Basic { username, password },
+            AuthConfig::Bearer { token } => Self::Bearer { token },
+            AuthConfig::Header { name, value } => Self::Header { name, value },
         }
     }
 }
@@ -6971,9 +6980,7 @@ impl From<crawlberg::CrawlEvent> for CrawlEvent {
                 url: url.to_string(),
                 error: error.to_string(),
             },
-            crawlberg::CrawlEvent::Complete { pages_crawled } => Self::Complete {
-                pages_crawled: pages_crawled,
-            },
+            crawlberg::CrawlEvent::Complete { pages_crawled } => Self::Complete { pages_crawled },
         }
     }
 }
@@ -7039,9 +7046,9 @@ impl From<crawlberg::FeedType> for FeedType {
 impl From<HostMatcher> for crawlberg::HostMatcher {
     fn from(val: HostMatcher) -> Self {
         match val {
-            HostMatcher::Exact { value } => Self::Exact { value: value },
-            HostMatcher::Suffix { value } => Self::Suffix { value: value },
-            HostMatcher::Cidr { value } => Self::Cidr { value: value },
+            HostMatcher::Exact { value } => Self::Exact { value },
+            HostMatcher::Suffix { value } => Self::Suffix { value },
+            HostMatcher::Cidr { value } => Self::Cidr { value },
         }
     }
 }
@@ -7109,27 +7116,21 @@ impl From<crawlberg::LinkType> for LinkType {
 impl From<PageAction> for crawlberg::PageAction {
     fn from(val: PageAction) -> Self {
         match val {
-            PageAction::Click { selector } => Self::Click { selector: selector },
-            PageAction::TypeText { selector, text } => Self::TypeText {
-                selector: selector,
-                text: text,
-            },
-            PageAction::Press { key } => Self::Press { key: key },
+            PageAction::Click { selector } => Self::Click { selector },
+            PageAction::TypeText { selector, text } => Self::TypeText { selector, text },
+            PageAction::Press { key } => Self::Press { key },
             PageAction::Scroll {
                 direction,
                 selector,
                 amount,
             } => Self::Scroll {
                 direction: direction.into(),
-                selector: selector,
-                amount: amount,
-            },
-            PageAction::Wait { milliseconds, selector } => Self::Wait {
-                milliseconds: milliseconds,
-                selector: selector,
+                selector,
+                amount,
             },
-            PageAction::Screenshot { full_page } => Self::Screenshot { full_page: full_page },
-            PageAction::ExecuteJs { script } => Self::ExecuteJs { script: script },
+            PageAction::Wait { milliseconds, selector } => Self::Wait { milliseconds, selector },
+            PageAction::Screenshot { full_page } => Self::Screenshot { full_page },
+            PageAction::ExecuteJs { script } => Self::ExecuteJs { script },
             PageAction::Scrape => Self::Scrape,
         }
     }
@@ -7153,13 +7154,13 @@ impl From<crawlberg::PageAction> for PageAction {
             } => Self::Scroll {
                 direction: direction.into(),
                 selector: selector.map(|v| v.to_string()),
-                amount: amount,
+                amount,
             },
             crawlberg::PageAction::Wait { milliseconds, selector } => Self::Wait {
-                milliseconds: milliseconds,
+                milliseconds,
                 selector: selector.map(|v| v.to_string()),
             },
-            crawlberg::PageAction::Screenshot { full_page } => Self::Screenshot { full_page: full_page },
+            crawlberg::PageAction::Screenshot { full_page } => Self::Screenshot { full_page },
             crawlberg::PageAction::ExecuteJs { script } => Self::ExecuteJs {
                 script: script.to_string(),
             },
@@ -8109,6 +8110,8 @@ fn ruby_init(ruby: &Ruby) -> Result<(), Error> {
 
     class.define_method("max_redirects", method!(SsrfPolicy::max_redirects, 0))?;
 
+    class.define_method("scheme_allowlist", method!(SsrfPolicy::scheme_allowlist, 0))?;
+
     let class = module.define_class("CrawlStreamIterator", ruby.class_object())?;
     class.define_method("next_chunk", method!(CrawlStreamIterator::next_chunk, 0))?;
     class.define_method("each", method!(CrawlStreamIterator::each, 0))?;

data/lib/crawlberg/native.rb

@@ -1,5 +1,5 @@
 # This file is auto-generated by alef — DO NOT EDIT.
-# alef:hash:4c0568b31b53eb28d5e56237159cb0d1c93079e2bbae63713fbc6b85e9683aa7
+# alef:hash:925532ba2f606ce36a0d3444d5c96407ca56a1c55bcd256372a1ea2a383d11e5
 # To regenerate: alef generate
 # To verify freshness: alef verify
 # frozen_string_literal: true
@@ -18,7 +18,7 @@ library_candidates = dlexts.flat_map do |dlext|
     File.join(__dir__, "..", extension_name, ruby_abi, "#{extension_name}.#{dlext}"),
     File.join(__dir__, "..", extension_name, ruby_abi, "lib#{extension_name}.#{dlext}"),
     File.join(__dir__, "..", "#{extension_name}.#{dlext}"),
-    File.join(__dir__, "..", "lib#{extension_name}.#{dlext}"),
+    File.join(__dir__, "..", "lib#{extension_name}.#{dlext}")
   ]
 end
 
@@ -34,6 +34,7 @@ unless native_extension
 end
 
 require native_extension
+
 module Crawlberg
   # Authentication configuration.
   module AuthConfig
@@ -49,10 +50,14 @@ module Crawlberg
     def self.from_hash(hash)
       discriminator = hash[:type] || hash["type"]
       case discriminator
-      when "basic" then AuthConfigBasic.from_hash(hash)
-      when "bearer" then AuthConfigBearer.from_hash(hash)
-      when "header" then AuthConfigHeader.from_hash(hash)
-      else raise "Unknown discriminator: #{discriminator}"
+      when "basic"
+        AuthConfigBasic.from_hash(hash)
+      when "bearer"
+        AuthConfigBearer.from_hash(hash)
+      when "header"
+        AuthConfigHeader.from_hash(hash)
+      else
+        raise "Unknown discriminator: #{discriminator}"
       end
     end
   end
@@ -63,10 +68,12 @@ module Crawlberg
 
     # Username sent in the `Authorization: Basic` header.
     sig { returns(String) }
-    def username = super # rubocop:disable Lint/UselessMethodDefinition
+    # rubocop:disable Lint/UselessMethodDefinition
+    def username = super
     # Password sent in the `Authorization: Basic` header.
     sig { returns(String) }
-    def password = super # rubocop:disable Lint/UselessMethodDefinition
+    # rubocop:disable Lint/UselessMethodDefinition
+    def password = super
     sig { returns(T::Boolean) }
     def basic? = true
     sig { returns(T::Boolean) }
@@ -87,7 +94,8 @@ module Crawlberg
 
     # Token sent in the `Authorization: Bearer` header.
     sig { returns(String) }
-    def token = super # rubocop:disable Lint/UselessMethodDefinition
+    # rubocop:disable Lint/UselessMethodDefinition
+    def token = super
     sig { returns(T::Boolean) }
     def basic? = false
     sig { returns(T::Boolean) }
@@ -108,10 +116,12 @@ module Crawlberg
 
     # HTTP header name to set on each request.
     sig { returns(String) }
-    def name = super # rubocop:disable Lint/UselessMethodDefinition
+    # rubocop:disable Lint/UselessMethodDefinition
+    def name = super
     # HTTP header value to send.
     sig { returns(String) }
-    def value = super # rubocop:disable Lint/UselessMethodDefinition
+    # rubocop:disable Lint/UselessMethodDefinition
+    def value = super
     sig { returns(T::Boolean) }
     def basic? = false
     sig { returns(T::Boolean) }
@@ -147,10 +157,14 @@ module Crawlberg
     def self.from_hash(hash)
       discriminator = hash[:type] || hash["type"]
       case discriminator
-      when "page" then CrawlEventPage.from_hash(hash)
-      when "error" then CrawlEventError.from_hash(hash)
-      when "complete" then CrawlEventComplete.from_hash(hash)
-      else raise "Unknown discriminator: #{discriminator}"
+      when "page"
+        CrawlEventPage.from_hash(hash)
+      when "error"
+        CrawlEventError.from_hash(hash)
+      when "complete"
+        CrawlEventComplete.from_hash(hash)
+      else
+        raise "Unknown discriminator: #{discriminator}"
       end
     end
   end
@@ -161,7 +175,8 @@ module Crawlberg
 
     # The crawled page result.
     sig { returns(CrawlPageResult) }
-    def result = super # rubocop:disable Lint/UselessMethodDefinition
+    # rubocop:disable Lint/UselessMethodDefinition
+    def result = super
     sig { returns(T::Boolean) }
     def page? = true
     sig { returns(T::Boolean) }
@@ -182,10 +197,12 @@ module Crawlberg
 
     # The URL that failed.
     sig { returns(String) }
-    def url = super # rubocop:disable Lint/UselessMethodDefinition
+    # rubocop:disable Lint/UselessMethodDefinition
+    def url = super
     # The error message.
     sig { returns(String) }
-    def error = super # rubocop:disable Lint/UselessMethodDefinition
+    # rubocop:disable Lint/UselessMethodDefinition
+    def error = super
     sig { returns(T::Boolean) }
     def page? = false
     sig { returns(T::Boolean) }
@@ -206,7 +223,8 @@ module Crawlberg
 
     # Total number of pages crawled.
     sig { returns(Integer) }
-    def pages_crawled = super # rubocop:disable Lint/UselessMethodDefinition
+    # rubocop:disable Lint/UselessMethodDefinition
+    def pages_crawled = super
     sig { returns(T::Boolean) }
     def page? = false
     sig { returns(T::Boolean) }
@@ -251,10 +269,14 @@ module Crawlberg
     def self.from_hash(hash)
       discriminator = hash[:type] || hash["type"]
       case discriminator
-      when "exact" then HostMatcherExact.from_hash(hash)
-      when "suffix" then HostMatcherSuffix.from_hash(hash)
-      when "cidr" then HostMatcherCidr.from_hash(hash)
-      else raise "Unknown discriminator: #{discriminator}"
+      when "exact"
+        HostMatcherExact.from_hash(hash)
+      when "suffix"
+        HostMatcherSuffix.from_hash(hash)
+      when "cidr"
+        HostMatcherCidr.from_hash(hash)
+      else
+        raise "Unknown discriminator: #{discriminator}"
       end
     end
   end
@@ -265,7 +287,8 @@ module Crawlberg
 
     # The hostname to match.
     sig { returns(String) }
-    def value = super # rubocop:disable Lint/UselessMethodDefinition
+    # rubocop:disable Lint/UselessMethodDefinition
+    def value = super
     sig { returns(T::Boolean) }
     def exact? = true
     sig { returns(T::Boolean) }
@@ -286,7 +309,8 @@ module Crawlberg
 
     # The dot-prefixed suffix to match. A leading dot is optional.
     sig { returns(String) }
-    def value = super # rubocop:disable Lint/UselessMethodDefinition
+    # rubocop:disable Lint/UselessMethodDefinition
+    def value = super
     sig { returns(T::Boolean) }
     def exact? = false
     sig { returns(T::Boolean) }
@@ -308,7 +332,8 @@ module Crawlberg
     # The CIDR block. Validated when built through [`HostMatcher::cidr`] or
     # deserialization.
     sig { returns(String) }
-    def value = super # rubocop:disable Lint/UselessMethodDefinition
+    # rubocop:disable Lint/UselessMethodDefinition
+    def value = super
     sig { returns(T::Boolean) }
     def exact? = false
     sig { returns(T::Boolean) }
@@ -342,15 +367,24 @@ module Crawlberg
     def self.from_hash(hash)
       discriminator = hash[:type] || hash["type"]
       case discriminator
-      when "click" then PageActionClick.from_hash(hash)
-      when "type" then PageActionTypeText.from_hash(hash)
-      when "press" then PageActionPress.from_hash(hash)
-      when "scroll" then PageActionScroll.from_hash(hash)
-      when "wait" then PageActionWait.from_hash(hash)
-      when "screenshot" then PageActionScreenshot.from_hash(hash)
-      when "executeJs" then PageActionExecuteJs.from_hash(hash)
-      when "scrape" then PageActionScrape.from_hash(hash)
-      else raise "Unknown discriminator: #{discriminator}"
+      when "click"
+        PageActionClick.from_hash(hash)
+      when "type"
+        PageActionTypeText.from_hash(hash)
+      when "press"
+        PageActionPress.from_hash(hash)
+      when "scroll"
+        PageActionScroll.from_hash(hash)
+      when "wait"
+        PageActionWait.from_hash(hash)
+      when "screenshot"
+        PageActionScreenshot.from_hash(hash)
+      when "executeJs"
+        PageActionExecuteJs.from_hash(hash)
+      when "scrape"
+        PageActionScrape.from_hash(hash)
+      else
+        raise "Unknown discriminator: #{discriminator}"
       end
     end
   end
@@ -361,7 +395,8 @@ module Crawlberg
 
     # CSS selector for the element to click.
     sig { returns(String) }
-    def selector = super # rubocop:disable Lint/UselessMethodDefinition
+    # rubocop:disable Lint/UselessMethodDefinition
+    def selector = super
     sig { returns(T::Boolean) }
     def click? = true
     sig { returns(T::Boolean) }
@@ -392,10 +427,12 @@ module Crawlberg
 
     # CSS selector for the input element.
     sig { returns(String) }
-    def selector = super # rubocop:disable Lint/UselessMethodDefinition
+    # rubocop:disable Lint/UselessMethodDefinition
+    def selector = super
     # Text to type into the element.
     sig { returns(String) }
-    def text = super # rubocop:disable Lint/UselessMethodDefinition
+    # rubocop:disable Lint/UselessMethodDefinition
+    def text = super
     sig { returns(T::Boolean) }
     def click? = false
     sig { returns(T::Boolean) }
@@ -426,7 +463,8 @@ module Crawlberg
 
     # Key name to press.
     sig { returns(String) }
-    def key = super # rubocop:disable Lint/UselessMethodDefinition
+    # rubocop:disable Lint/UselessMethodDefinition
+    def key = super
     sig { returns(T::Boolean) }
     def click? = false
     sig { returns(T::Boolean) }
@@ -457,13 +495,16 @@ module Crawlberg
 
     # Direction to scroll.
     sig { returns(ScrollDirection) }
-    def direction = super # rubocop:disable Lint/UselessMethodDefinition
+    # rubocop:disable Lint/UselessMethodDefinition
+    def direction = super
     # Optional CSS selector for a scrollable element. Scrolls the page if absent.
     sig { returns(T.nilable(String)) }
-    def selector = super # rubocop:disable Lint/UselessMethodDefinition
+    # rubocop:disable Lint/UselessMethodDefinition
+    def selector = super
     # Optional pixel amount to scroll. Uses a default if absent.
     sig { returns(T.nilable(Integer)) }
-    def amount = super # rubocop:disable Lint/UselessMethodDefinition
+    # rubocop:disable Lint/UselessMethodDefinition
+    def amount = super
     sig { returns(T::Boolean) }
     def click? = false
     sig { returns(T::Boolean) }
@@ -484,7 +525,11 @@ module Crawlberg
     # @return [self]
     sig { params(hash: T::Hash[T.untyped, T.untyped]).returns(T.attached_class) }
     def self.from_hash(hash)
-      new(direction: hash[:direction] || hash["direction"], selector: hash[:selector] || hash["selector"], amount: hash[:amount] || hash["amount"])
+      new(
+        direction: hash[:direction] || hash["direction"],
+        selector: hash[:selector] || hash["selector"],
+        amount: hash[:amount] || hash["amount"]
+      )
     end
   end
   ## Wait for a duration or for an element to appear.
@@ -494,10 +539,12 @@ module Crawlberg
 
     # Milliseconds to wait. Ignored if `selector` is provided.
     sig { returns(T.nilable(Integer)) }
-    def milliseconds = super # rubocop:disable Lint/UselessMethodDefinition
+    # rubocop:disable Lint/UselessMethodDefinition
+    def milliseconds = super
     # CSS selector to wait for.
     sig { returns(T.nilable(String)) }
-    def selector = super # rubocop:disable Lint/UselessMethodDefinition
+    # rubocop:disable Lint/UselessMethodDefinition
+    def selector = super
     sig { returns(T::Boolean) }
     def click? = false
     sig { returns(T::Boolean) }
@@ -532,7 +579,8 @@ module Crawlberg
     # `full_page` (snake_case) alias so language bindings and fixtures can
     # use either convention without error.
     sig { returns(T.nilable(T::Boolean)) }
-    def full_page = super # rubocop:disable Lint/UselessMethodDefinition
+    # rubocop:disable Lint/UselessMethodDefinition
+    def full_page = super
     sig { returns(T::Boolean) }
     def click? = false
     sig { returns(T::Boolean) }
@@ -568,7 +616,8 @@ module Crawlberg
 
     # JavaScript source code to execute. Max 1 MB.
     sig { returns(String) }
-    def script = super # rubocop:disable Lint/UselessMethodDefinition
+    # rubocop:disable Lint/UselessMethodDefinition
+    def script = super
     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:0a9a0a6f7c1b865f2c375eef53bab2695c7dbf595519ca3bd7f5d258c2936ef7
+# alef:hash:9ea9a14d259e4988466100a061689d975727c16c521c140aef2205c9b97bf831
 # To regenerate: alef generate
 # To verify freshness: alef verify
 # frozen_string_literal: true
 
 module Crawlberg
   ## The version string for this package.
-  VERSION = "1.4.0"
+  VERSION = "1.4.2"
 end