crawlberg 1.12.2 → 1.12.3

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: d94c2b28c7e7bc04835352b1e68dd17e01e73435438d896869793e8205dda557
-  data.tar.gz: af40fde2be40012801c809a9a29aa588f7b5e173f06c7c97a553b8edabf38fc4
+  metadata.gz: 30a868ed0481017eb9b39c702345e1400e344c6a5344e032c3373c2639c4d899
+  data.tar.gz: e3c0a251a23a696d8c50f43dac142b125d39a8c63aea27198df4195b5ebe62c9
 SHA512:
-  metadata.gz: 40b098a85ecdbead4feb24b6364064e2ce2efcc5bb0feb8b9c2df00cef4a1de529fb3a9b2c5496568aaee0b03b5e9f3913732b0c9cdc882561c91e811d017bca
-  data.tar.gz: d6d13a3632dbf1f84a21c53dfd1432de7f79be1ed3f818dd19883af9fd4fbc3258a13fb918de4db8b390320b642b5c19543aad67f6ce62928311d675d3c667cc
+  metadata.gz: 31908f8286aecb43890904b744b3b7ef377965d3db07e0bab5e84c3384314d2b9eb77cda839a21ec9ed76d656de192c63b1afb1c807b3bfaf957518937b4e704
+  data.tar.gz: '028a9b3167894b417bb5c5a3cf47a268123e9f5f749be60f7a04347ae47e19791db3ae185289140cd379851fc93e954883be62228c229b1fddb9461836658744'

data/ext/crawlberg_rb/native/Cargo.lock

@@ -537,9 +537,9 @@ dependencies = [
 
 [[package]]
 name = "crawlberg"
-version = "1.12.2"
+version = "1.12.3"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "8949315433622c191e4f83b23041a46f02f6096839149706af09715f75280971"
+checksum = "e9cece970a1c583dc55de73dcbc4d85242c565cd1ab5223eb090aa36632c8f05"
 dependencies = [
  "ahash",
  "aho-corasick",
@@ -607,9 +607,9 @@ dependencies = [
 
 [[package]]
 name = "crawlberg-chromiumoxide"
-version = "0.9.4"
+version = "0.9.5"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "937d12e969f7090c02cb64dff52375f3e26e669cbdda7bc49ce364fb8ce7bb68"
+checksum = "28a8fa6f75a2a698c099f1defa98fe1dd45367e0574d7f52306a1e54d66bf4df"
 dependencies = [
  "async-tungstenite",
  "base64",
@@ -635,7 +635,7 @@ dependencies = [
 
 [[package]]
 name = "crawlberg-rb"
-version = "1.12.2"
+version = "1.12.3"
 dependencies = [
  "crawlberg",
  "futures",
@@ -648,9 +648,9 @@ dependencies = [
 
 [[package]]
 name = "crawlberg-robots"
-version = "1.12.2"
+version = "1.12.3"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "30cb48557123dd4e13fd756b83490faf9aaeb7d70e1bcd74a410c4f12a9b6320"
+checksum = "3a38455515bc47cf5e05e8a02a761d24e2b80970af157bcc828d22a99225962f"
 
 [[package]]
 name = "crc32fast"
@@ -1077,9 +1077,9 @@ checksum = "491194d601b4fa9fac3c9571e014070a0ad1bc3392d1bc10419695196a8b3cc3"
 
 [[package]]
 name = "html-to-markdown-rs"
-version = "3.18.7"
+version = "3.19.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "259eff8de02b67e436fa4ccd377dbf6105d530fc405f20489677f417cf673a7a"
+checksum = "ddada45be32a036999fbda2be5df286dc2001816e60099ec3fd445fc6c3bb69b"
 dependencies = [
  "ahash",
  "astral-tl",

data/ext/crawlberg_rb/native/Cargo.toml

@@ -1,11 +1,11 @@
 # This file is auto-generated by alef — DO NOT EDIT.
-# alef:hash:b3f271455dce968c83cbdf03e1934902e2ad21295928a2c5f148e7dddc67ac2c
+# alef:hash:a86b7727e26a9e336e600fe8a57e0a117bf4bfa5b149da399b9648824aaa6c73
 # To regenerate: alef generate
 # To verify freshness: alef verify
 
 [package]
 name = "crawlberg-rb"
-version = "1.12.2"
+version = "1.12.3"
 edition = "2024"
 authors = ["Na'aman Hirschfeld <naaman@xberg.io>", "Tobias Perlstein <tobias@xberg.io>"]
 license = "MIT"
@@ -23,7 +23,7 @@ path = "../src/lib.rs"
 crate-type = ["cdylib"]
 
 [dependencies]
-crawlberg = { version = "1.12.2", features = ["interact", "browser-chromiumoxide"] }
+crawlberg = { version = "1.12.3", features = ["interact", "browser-chromiumoxide"] }
 futures = "0.3"
 magnus = "0.9"
 rb-sys = ">=0.9.130, <0.10"

data/lib/crawlberg/version.rb

@@ -1,10 +1,10 @@
 # This file is auto-generated by alef — DO NOT EDIT.
-# alef:hash:b402d1b3bc989de04db5241785f4bc16c0b5e0f76714438d3687648c64d8c10f
+# alef:hash:4b49e60e4c3bdd958dfccc4aee6deec2a78c41bafced6ee25ec8b66aa0b049ca
 # To regenerate: alef generate
 # To verify freshness: alef verify
 # frozen_string_literal: true
 
 module Crawlberg
   ## The version string for this package.
-  VERSION = "1.12.2"
+  VERSION = "1.12.3"
 end

data/sig/types.rbs

@@ -1,5 +1,5 @@
 # This file is auto-generated by alef — DO NOT EDIT.
-# alef:hash:9aca68325287aae22a855afd34597e3d8bd90c6b7774f3b305591fde4fb051cd
+# alef:hash:4cdc9eb4c9b67c7adb9baf33b70eb0c6ee54e28b0c0973121fbef2934ff8a431
 # To regenerate: alef generate
 # To verify freshness: alef verify
 
@@ -530,7 +530,46 @@ def initialize: (?deny_private: bool, ?allowlist: Array[HostMatcher], ?denylist:
 
     type enum_AssetCategory = :document | :image | :audio | :video | :font | :stylesheet | :script | :archive | :data | :other
 
-    class AuthConfig
+    module AuthConfig
+    def self.from_hash: (untyped hash) -> untyped
+    end
+
+    class AuthConfigBasic
+        include AuthConfig
+        attr_reader username: String
+        attr_reader password: String
+
+    def basic?: () -> bool
+    def bearer?: () -> bool
+    def header?: () -> bool
+
+    def self.from_hash: (untyped hash) -> untyped
+    def self.new: (username: String, password: String) -> AuthConfigBasic
+    end
+
+    class AuthConfigBearer
+        include AuthConfig
+        attr_reader token: String
+
+    def basic?: () -> bool
+    def bearer?: () -> bool
+    def header?: () -> bool
+
+    def self.from_hash: (untyped hash) -> untyped
+    def self.new: (token: String) -> AuthConfigBearer
+    end
+
+    class AuthConfigHeader
+        include AuthConfig
+        attr_reader name: String
+        attr_reader value: String
+
+    def basic?: () -> bool
+    def bearer?: () -> bool
+    def header?: () -> bool
+
+    def self.from_hash: (untyped hash) -> untyped
+    def self.new: (name: String, value: String) -> AuthConfigHeader
     end
 
     type enum_BrowserBackend = :chromiumoxide | :native
@@ -541,7 +580,45 @@ def initialize: (?deny_private: bool, ?allowlist: Array[HostMatcher], ?denylist:
 
     type enum_ContentFilterKind = :bm25
 
-    class CrawlEvent
+    module CrawlEvent
+    def self.from_hash: (untyped hash) -> untyped
+    end
+
+    class CrawlEventPage
+        include CrawlEvent
+        attr_reader result: CrawlPageResult
+
+    def page?: () -> bool
+    def error?: () -> bool
+    def complete?: () -> bool
+
+    def self.from_hash: (untyped hash) -> untyped
+    def self.new: (result: CrawlPageResult) -> CrawlEventPage
+    end
+
+    class CrawlEventError
+        include CrawlEvent
+        attr_reader url: String
+        attr_reader error: String
+
+    def page?: () -> bool
+    def error?: () -> bool
+    def complete?: () -> bool
+
+    def self.from_hash: (untyped hash) -> untyped
+    def self.new: (url: String, error: String) -> CrawlEventError
+    end
+
+    class CrawlEventComplete
+        include CrawlEvent
+        attr_reader pages_crawled: Integer
+
+    def page?: () -> bool
+    def error?: () -> bool
+    def complete?: () -> bool
+
+    def self.from_hash: (untyped hash) -> untyped
+    def self.new: (pages_crawled: Integer) -> CrawlEventComplete
     end
 
     type enum_CrawlStrategyKind = :bfs | :dfs | :best_first | :adaptive
@@ -550,14 +627,190 @@ def initialize: (?deny_private: bool, ?allowlist: Array[HostMatcher], ?denylist:
 
     type enum_FeedType = :rss | :atom | :json_feed
 
-    class HostMatcher
+    module HostMatcher
+    def self.from_hash: (untyped hash) -> untyped
+    end
+
+    class HostMatcherExact
+        include HostMatcher
+        attr_reader value: String
+
+    def exact?: () -> bool
+    def suffix?: () -> bool
+    def cidr?: () -> bool
+
+    def self.from_hash: (untyped hash) -> untyped
+    def self.new: (value: String) -> HostMatcherExact
+    end
+
+    class HostMatcherSuffix
+        include HostMatcher
+        attr_reader value: String
+
+    def exact?: () -> bool
+    def suffix?: () -> bool
+    def cidr?: () -> bool
+
+    def self.from_hash: (untyped hash) -> untyped
+    def self.new: (value: String) -> HostMatcherSuffix
+    end
+
+    class HostMatcherCidr
+        include HostMatcher
+        attr_reader value: String
+
+    def exact?: () -> bool
+    def suffix?: () -> bool
+    def cidr?: () -> bool
+
+    def self.from_hash: (untyped hash) -> untyped
+    def self.new: (value: String) -> HostMatcherCidr
     end
 
     type enum_ImageSource = :img | :picture_source | :"og:image" | :"twitter:image"
 
     type enum_LinkType = :internal | :external | :anchor | :document
 
-    class PageAction
+    module PageAction
+    def self.from_hash: (untyped hash) -> untyped
+    end
+
+    class PageActionClick
+        include PageAction
+        attr_reader selector: String
+
+    def click?: () -> bool
+    def type_text?: () -> bool
+    def press?: () -> bool
+    def scroll?: () -> bool
+    def wait?: () -> bool
+    def screenshot?: () -> bool
+    def execute_js?: () -> bool
+    def scrape?: () -> bool
+
+    def self.from_hash: (untyped hash) -> untyped
+    def self.new: (selector: String) -> PageActionClick
+    end
+
+    class PageActionTypeText
+        include PageAction
+        attr_reader selector: String
+        attr_reader text: String
+
+    def click?: () -> bool
+    def type_text?: () -> bool
+    def press?: () -> bool
+    def scroll?: () -> bool
+    def wait?: () -> bool
+    def screenshot?: () -> bool
+    def execute_js?: () -> bool
+    def scrape?: () -> bool
+
+    def self.from_hash: (untyped hash) -> untyped
+    def self.new: (selector: String, text: String) -> PageActionTypeText
+    end
+
+    class PageActionPress
+        include PageAction
+        attr_reader key: String
+
+    def click?: () -> bool
+    def type_text?: () -> bool
+    def press?: () -> bool
+    def scroll?: () -> bool
+    def wait?: () -> bool
+    def screenshot?: () -> bool
+    def execute_js?: () -> bool
+    def scrape?: () -> bool
+
+    def self.from_hash: (untyped hash) -> untyped
+    def self.new: (key: String) -> PageActionPress
+    end
+
+    class PageActionScroll
+        include PageAction
+        attr_reader direction: enum_ScrollDirection
+        attr_reader selector: String?
+        attr_reader amount: Integer?
+
+    def click?: () -> bool
+    def type_text?: () -> bool
+    def press?: () -> bool
+    def scroll?: () -> bool
+    def wait?: () -> bool
+    def screenshot?: () -> bool
+    def execute_js?: () -> bool
+    def scrape?: () -> bool
+
+    def self.from_hash: (untyped hash) -> untyped
+    def self.new: (direction: enum_ScrollDirection, selector: String, amount: Integer) -> PageActionScroll
+    end
+
+    class PageActionWait
+        include PageAction
+        attr_reader milliseconds: Integer?
+        attr_reader selector: String?
+
+    def click?: () -> bool
+    def type_text?: () -> bool
+    def press?: () -> bool
+    def scroll?: () -> bool
+    def wait?: () -> bool
+    def screenshot?: () -> bool
+    def execute_js?: () -> bool
+    def scrape?: () -> bool
+
+    def self.from_hash: (untyped hash) -> untyped
+    def self.new: (milliseconds: Integer, selector: String) -> PageActionWait
+    end
+
+    class PageActionScreenshot
+        include PageAction
+        attr_reader full_page: bool?
+
+    def click?: () -> bool
+    def type_text?: () -> bool
+    def press?: () -> bool
+    def scroll?: () -> bool
+    def wait?: () -> bool
+    def screenshot?: () -> bool
+    def execute_js?: () -> bool
+    def scrape?: () -> bool
+
+    def self.from_hash: (untyped hash) -> untyped
+    def self.new: (full_page: bool) -> PageActionScreenshot
+    end
+
+    class PageActionExecuteJs
+        include PageAction
+        attr_reader script: String
+
+    def click?: () -> bool
+    def type_text?: () -> bool
+    def press?: () -> bool
+    def scroll?: () -> bool
+    def wait?: () -> bool
+    def screenshot?: () -> bool
+    def execute_js?: () -> bool
+    def scrape?: () -> bool
+
+    def self.from_hash: (untyped hash) -> untyped
+    def self.new: (script: String) -> PageActionExecuteJs
+    end
+
+    class PageActionScrape
+        include PageAction
+    def click?: () -> bool
+    def type_text?: () -> bool
+    def press?: () -> bool
+    def scroll?: () -> bool
+    def wait?: () -> bool
+    def screenshot?: () -> bool
+    def execute_js?: () -> bool
+    def scrape?: () -> bool
+
+    def self.from_hash: (untyped hash) -> untyped
+        def self.new:  -> PageActionScrape
     end
 
     type enum_ScrollDirection = :up | :down

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: crawlberg
 version: !ruby/object:Gem::Version
-  version: 1.12.2
+  version: 1.12.3
 platform: ruby
 authors:
 - Na'aman Hirschfeld