puppeteer-ruby 0.50.0.alpha5 → 0.50.0.alpha6

data/lib/puppeteer/p_query_handler.rb

@@ -0,0 +1,367 @@
+# rbs_inline: enabled
+
+class Puppeteer::PQueryHandler
+  POLLING_DEFAULT_SECONDS = 0.1
+
+  CSS_QUERY_SELECTOR_JS = <<~'JAVASCRIPT'
+    (element, selector) => {
+      const results = [];
+      const isQueryableNode = node => {
+        return node && typeof node.querySelectorAll === 'function';
+      };
+      if (!isQueryableNode(element)) {
+        return results;
+      }
+      const trimmed = selector.trimStart();
+      if (!trimmed) {
+        return results;
+      }
+      const IDENT_TOKEN_START = /[-\w\P{ASCII}*]/u;
+      if (IDENT_TOKEN_START.test(trimmed[0])) {
+        results.push(...element.querySelectorAll(selector));
+        return results;
+      }
+      if (!element.parentElement) {
+        results.push(...element.querySelectorAll(selector));
+        return results;
+      }
+      let index = 0;
+      for (const child of element.parentElement.children) {
+        ++index;
+        if (child === element) {
+          break;
+        }
+      }
+      results.push(
+        ...element.parentElement.querySelectorAll(
+          `:scope>:nth-child(${index})${selector}`
+        )
+      );
+      return results;
+    }
+  JAVASCRIPT
+
+  PIERCE_JS = <<~JAVASCRIPT
+    node => {
+      if (!node) {
+        return [];
+      }
+      if (node.shadowRoot) {
+        return [node.shadowRoot];
+      }
+      return [node];
+    }
+  JAVASCRIPT
+
+  PIERCE_ALL_JS = <<~JAVASCRIPT
+    root => {
+      if (!root) {
+        return [];
+      }
+      const results = [];
+      const rootNode = root.shadowRoot ? root.shadowRoot : root;
+      results.push(rootNode);
+      const walkers = [document.createTreeWalker(rootNode, NodeFilter.SHOW_ELEMENT)];
+      for (const walker of walkers) {
+        let node;
+        while ((node = walker.nextNode())) {
+          if (!node.shadowRoot) {
+            continue;
+          }
+          results.push(node.shadowRoot);
+          walkers.push(document.createTreeWalker(node.shadowRoot, NodeFilter.SHOW_ELEMENT));
+        }
+      }
+      return results;
+    }
+  JAVASCRIPT
+
+  DOM_SORT_JS = <<~JAVASCRIPT
+    (...elements) => {
+      const results = new Set(elements);
+      class DepthCalculator {
+        constructor() {
+          this.cache = new WeakMap();
+        }
+        calculate(node, depth = []) {
+          if (node === null) {
+            return depth;
+          }
+          if (node instanceof ShadowRoot) {
+            node = node.host;
+          }
+          const cached = this.cache.get(node);
+          if (cached) {
+            return [...cached, ...depth];
+          }
+          let index = 0;
+          for (let prev = node.previousSibling; prev; prev = prev.previousSibling) {
+            ++index;
+          }
+          const value = this.calculate(node.parentNode, [index]);
+          this.cache.set(node, value);
+          return [...value, ...depth];
+        }
+      }
+      const compareDepths = (a, b) => {
+        if (a.length + b.length === 0) {
+          return 0;
+        }
+        const [i = -1, ...restA] = a;
+        const [j = -1, ...restB] = b;
+        if (i === j) {
+          return compareDepths(restA, restB);
+        }
+        return i < j ? -1 : 1;
+      };
+      const calculator = new DepthCalculator();
+      return [...results]
+        .map(result => [result, calculator.calculate(result)])
+        .sort(([, a], [, b]) => compareDepths(a, b))
+        .map(([result]) => result);
+    }
+  JAVASCRIPT
+
+  # @rbs query_handler_manager: untyped -- Query handler manager
+  # @rbs return: void -- No return value
+  def initialize(query_handler_manager: Puppeteer::QueryHandlerManager.instance)
+    @query_handler_manager = query_handler_manager
+  end
+
+  # @rbs element: Puppeteer::ElementHandle -- Root element
+  # @rbs selector: String -- PSelector string
+  # @rbs return: Puppeteer::ElementHandle? -- First matching element
+  def query_one(element, selector)
+    matches = query_all(element, selector)
+    return nil if matches.empty?
+
+    matches.first
+  end
+
+  # @rbs element: Puppeteer::ElementHandle -- Root element
+  # @rbs selector: String -- PSelector string
+  # @rbs return: Array[Puppeteer::ElementHandle] -- Matching elements
+  def query_all(element, selector)
+    selectors, = Puppeteer::PSelectorParser.parse(selector)
+    return [] if selectors.empty?
+
+    results = selectors.flat_map do |complex|
+      run_complex_selector(element, complex)
+    end
+
+    return [] if results.empty?
+
+    dom_sort(results)
+  end
+
+  # @rbs element: Puppeteer::ElementHandle -- Root element
+  # @rbs selector: String -- PSelector string
+  # @rbs return: Puppeteer::JSHandle -- Handle to array of elements
+  def query_all_array(element, selector)
+    elements = query_all(element, selector)
+    return element.execution_context.evaluate_handle('() => []') if elements.empty?
+
+    element.execution_context.evaluate_handle('(...elements) => elements', *elements)
+  ensure
+    elements&.each(&:dispose)
+  end
+
+  # @rbs element_or_frame: Puppeteer::ElementHandle | Puppeteer::Frame -- Root element or frame
+  # @rbs selector: String -- PSelector string
+  # @rbs visible: bool? -- Wait for element to be visible
+  # @rbs hidden: bool? -- Wait for element to be hidden
+  # @rbs timeout: Numeric? -- Timeout in milliseconds
+  # @rbs polling: (String | Numeric | nil) -- Polling strategy
+  # @rbs return: Puppeteer::ElementHandle? -- Matching element
+  def wait_for(element_or_frame, selector, visible:, hidden:, timeout:, polling: nil)
+    case element_or_frame
+    when Puppeteer::Frame
+      frame = element_or_frame
+      root = nil
+    when Puppeteer::ElementHandle
+      frame = element_or_frame.frame
+      root = frame.puppeteer_world.adopt_handle(element_or_frame)
+    else
+      raise ArgumentError.new("element_or_frame must be a Frame or ElementHandle. #{element_or_frame.inspect}")
+    end
+
+    timeout_ms = timeout.nil? ? frame.default_timeout : timeout
+    deadline = timeout_ms && timeout_ms > 0 ? monotonic_time + (timeout_ms / 1000.0) : nil
+    polling_interval = polling_interval_seconds(polling, visible: visible, hidden: hidden)
+
+    begin
+      loop do
+        resolved_root = root || frame.puppeteer_world.document
+        handle = query_one(resolved_root, selector)
+
+        if handle
+          if visible
+            return transfer_handle(frame, handle) if handle.visible?
+          elsif hidden
+            return transfer_handle(frame, handle) if handle.hidden?
+          else
+            return transfer_handle(frame, handle)
+          end
+          handle.dispose
+        elsif hidden
+          return nil
+        end
+
+        raise_timeout(selector) if deadline && monotonic_time >= deadline
+        Puppeteer::AsyncUtils.sleep_seconds(polling_interval)
+      end
+    rescue => err
+      if err.is_a?(Puppeteer::TimeoutError)
+        raise
+      end
+
+      wait_for_selector_error =
+        if err.is_a?(Puppeteer::TimeoutError)
+          Puppeteer::TimeoutError.new("Waiting for selector `#{selector}` failed")
+        else
+          Puppeteer::Error.new("Waiting for selector `#{selector}` failed")
+        end
+      wait_for_selector_error.cause = err
+      raise wait_for_selector_error
+    ensure
+      root&.dispose
+    end
+  end
+
+  private
+    def run_complex_selector(root, complex)
+      elements = [root]
+
+      complex.each do |part|
+        case part
+        when '>>>'
+          elements = pierce_all(elements)
+        when '>>>>'
+          elements = pierce(elements)
+        else
+          elements = apply_compound(elements, part)
+        end
+        return [] if elements.empty?
+      end
+
+      elements
+    end
+
+    def apply_compound(elements, compound)
+      compound.reduce(elements) do |current, selector|
+        next [] if current.empty?
+
+        if selector.is_a?(String)
+          apply_css_selector(current, selector)
+        else
+          apply_pseudo_selector(current, selector)
+        end
+      end
+    end
+
+    def apply_css_selector(elements, selector)
+      elements.flat_map do |element|
+        query_css_selector(element, selector)
+      end
+    end
+
+    def query_css_selector(element, selector)
+      return [] if selector.strip.empty?
+
+      array_handle = element.evaluate_handle(CSS_QUERY_SELECTOR_JS, selector)
+      array_handle_to_elements(array_handle)
+    ensure
+      array_handle&.dispose
+    end
+
+    def apply_pseudo_selector(elements, pseudo)
+      handler = query_handler_for_pseudo(pseudo.name)
+      target_context = elements.first&.execution_context
+
+      elements.flat_map do |element|
+        handles = handler.query_all(element, pseudo.value) || []
+        handles.map do |handle|
+          ensure_context(handle, target_context)
+        end
+      end
+    end
+
+    def query_handler_for_pseudo(name)
+      handler = @query_handler_manager.query_handlers[name.to_sym]
+      return handler if handler
+
+      raise Puppeteer::Error.new("Unknown selector type: #{name}")
+    end
+
+    def pierce(elements)
+      elements.flat_map do |element|
+        array_handle = element.evaluate_handle(PIERCE_JS)
+        array_handle_to_elements(array_handle)
+      ensure
+        array_handle&.dispose
+      end
+    end
+
+    def pierce_all(elements)
+      elements.flat_map do |element|
+        array_handle = element.evaluate_handle(PIERCE_ALL_JS)
+        array_handle_to_elements(array_handle)
+      ensure
+        array_handle&.dispose
+      end
+    end
+
+    def dom_sort(elements)
+      return [] if elements.empty?
+
+      context = elements.first.execution_context
+      sorted_handle = context.evaluate_handle(DOM_SORT_JS, *elements)
+      sorted = array_handle_to_elements(sorted_handle)
+      elements.each(&:dispose)
+      sorted
+    ensure
+      sorted_handle&.dispose
+    end
+
+    def array_handle_to_elements(array_handle)
+      properties = array_handle.properties
+      properties.keys.sort_by(&:to_i).filter_map do |key|
+        properties[key].as_element
+      end
+    end
+
+    def ensure_context(handle, context)
+      return handle if context.nil? || handle.execution_context == context
+
+      world = context.world
+      return handle unless world
+
+      world.transfer_handle(handle)
+    end
+
+    def transfer_handle(frame, handle)
+      if handle.execution_context == frame.main_world.execution_context
+        handle
+      else
+        frame.main_world.transfer_handle(handle)
+      end
+    end
+
+    def polling_interval_seconds(polling, visible:, hidden:)
+      return polling / 1000.0 if polling.is_a?(Numeric)
+
+      if visible || hidden
+        return 0.016
+      end
+
+      POLLING_DEFAULT_SECONDS
+    end
+
+    def monotonic_time
+      Process.clock_gettime(Process::CLOCK_MONOTONIC)
+    end
+
+    def raise_timeout(selector)
+      raise Puppeteer::TimeoutError.new("Waiting for selector `#{selector}` failed")
+    end
+end

data/lib/puppeteer/p_selector_parser.rb

@@ -0,0 +1,241 @@
+# rbs_inline: enabled
+
+module Puppeteer
+  module PSelectorParser
+    class PseudoSelector
+      attr_reader :name, :value
+
+      def initialize(name:, value:)
+        @name = name
+        @value = value
+      end
+    end
+
+    PSEUDO_PREFIX = '::-p-'
+
+    # @rbs selector: String -- Selector to parse
+    # @rbs return: [Array[untyped], bool, bool, bool] -- Parsed selectors and flags
+    def self.parse(selector)
+      is_pure_css = true
+      has_pseudo_classes = false
+      has_aria = false
+
+      selectors = []
+      compound = []
+      complex = [compound]
+      storage = +''
+
+      i = 0
+      length = selector.length
+      in_quote = nil
+      escaped = false
+      paren_depth = 0
+
+      while i < length
+        char = selector[i]
+
+        if escaped
+          storage << char
+          escaped = false
+          i += 1
+          next
+        end
+
+        if char == '\\'
+          storage << char
+          escaped = true
+          i += 1
+          next
+        end
+
+        if in_quote
+          if char == in_quote
+            in_quote = nil
+          end
+          storage << char
+          i += 1
+          next
+        end
+
+        if char == '"' || char == "'"
+          in_quote = char
+          storage << char
+          i += 1
+          next
+        end
+
+        if char == '('
+          paren_depth += 1
+          storage << char
+          i += 1
+          next
+        end
+
+        if char == ')'
+          paren_depth = [paren_depth - 1, 0].max
+          storage << char
+          i += 1
+          next
+        end
+
+        if paren_depth == 0
+          if starts_with_at?(selector, i, '>>>>')
+            flush_storage(storage, compound)
+            complex << '>>>>'
+            compound = []
+            complex << compound
+            is_pure_css = false
+            i += 4
+            next
+          end
+
+          if starts_with_at?(selector, i, '>>>')
+            flush_storage(storage, compound)
+            complex << '>>>'
+            compound = []
+            complex << compound
+            is_pure_css = false
+            i += 3
+            next
+          end
+
+          if starts_with_at?(selector, i, PSEUDO_PREFIX)
+            flush_storage(storage, compound)
+            i += PSEUDO_PREFIX.length
+            name, i = parse_name(selector, i)
+            raise ArgumentError.new('Invalid PSelector name') if name.empty?
+
+            i = skip_spaces(selector, i)
+            value = ''
+            if selector[i] == '('
+              value, i = parse_argument(selector, i)
+            end
+            value = unquote(value.strip)
+
+            compound << PseudoSelector.new(name: name, value: value)
+            is_pure_css = false
+            has_aria = true if name == 'aria'
+            next
+          end
+
+          if char == ','
+            flush_storage(storage, compound)
+            selectors << complex
+            compound = []
+            complex = [compound]
+            i += 1
+            next
+          end
+
+          if char == ':' && selector[i, PSEUDO_PREFIX.length] != PSEUDO_PREFIX
+            if selector[i + 1] != ':'
+              has_pseudo_classes = true
+            end
+          end
+        end
+
+        storage << char
+        i += 1
+      end
+
+      raise ArgumentError.new('Unterminated string in PSelector') if in_quote
+      raise ArgumentError.new('Unterminated parentheses in PSelector') if paren_depth != 0
+
+      flush_storage(storage, compound)
+      selectors << complex if complex.any?
+
+      [selectors, is_pure_css, has_pseudo_classes, has_aria]
+    end
+
+    def self.flush_storage(storage, compound)
+      css = storage.strip
+      compound << css unless css.empty?
+      storage.clear
+    end
+    private_class_method :flush_storage
+
+    def self.parse_name(selector, index)
+      name = +''
+      while index < selector.length && selector[index] =~ /[A-Za-z0-9_-]/
+        name << selector[index]
+        index += 1
+      end
+      [name, index]
+    end
+    private_class_method :parse_name
+
+    def self.skip_spaces(selector, index)
+      index += 1 while index < selector.length && selector[index] =~ /\s/
+      index
+    end
+    private_class_method :skip_spaces
+
+    def self.parse_argument(selector, index)
+      raise ArgumentError.new('Expected opening parenthesis') unless selector[index] == '('
+
+      index += 1
+      start = index
+      depth = 1
+      in_quote = nil
+      escaped = false
+
+      while index < selector.length
+        char = selector[index]
+
+        if escaped
+          escaped = false
+          index += 1
+          next
+        end
+
+        if char == '\\'
+          escaped = true
+          index += 1
+          next
+        end
+
+        if in_quote
+          if char == in_quote
+            in_quote = nil
+          end
+          index += 1
+          next
+        end
+
+        if char == '"' || char == "'"
+          in_quote = char
+          index += 1
+          next
+        end
+
+        if char == '('
+          depth += 1
+        elsif char == ')'
+          depth -= 1
+          if depth == 0
+            value = selector[start...index]
+            return [value, index + 1]
+          end
+        end
+
+        index += 1
+      end
+
+      raise ArgumentError.new('Unterminated PSelector argument')
+    end
+    private_class_method :parse_argument
+
+    def self.starts_with_at?(selector, index, token)
+      selector[index, token.length] == token
+    end
+    private_class_method :starts_with_at?
+
+    def self.unquote(text)
+      if text.length > 1 && (text.start_with?('"') || text.start_with?("'")) && text.end_with?(text[0])
+        text = text[1...-1]
+      end
+      text.gsub(/\\([\s\S])/m, '\\1')
+    end
+    private_class_method :unquote
+  end
+end

data/lib/puppeteer/page.rb

@@ -459,6 +459,21 @@ class Puppeteer::Page
     @timeout_settings.default_timeout = timeout
   end
 
+  # @rbs return: Numeric -- Default timeout in milliseconds
+  def default_timeout
+    @timeout_settings.timeout
+  end
+
+  # @rbs selector_or_function: String -- Selector or JS function
+  # @rbs return: Puppeteer::Locator -- Locator for selector or function
+  def locator(selector_or_function)
+    if Puppeteer::Locator.function_string?(selector_or_function)
+      Puppeteer::FunctionLocator.create(self, selector_or_function)
+    else
+      Puppeteer::NodeLocator.create(self, selector_or_function)
+    end
+  end
+
   # `$()` in JavaScript.
   # @rbs selector: String -- CSS selector
   # @rbs return: Puppeteer::ElementHandle? -- Matching element or nil

data/lib/puppeteer/query_handler_manager.rb

@@ -45,62 +45,7 @@ class Puppeteer::QueryHandlerManager
   end
 
   private def p_query_handler
-    @p_query_handler ||= Puppeteer::CustomQueryHandler.new(
-      query_one: <<~JAVASCRIPT,
-      (element, selector) => {
-        const parts = selector.split('>>>').map((part) => part.trim()).filter(Boolean);
-        let roots = [element];
-        if (parts.length === 0) return null;
-        for (let i = 0; i < parts.length; i++) {
-          const part = parts[i];
-          const next = [];
-          for (const root of roots) {
-            const scope = root;
-            const elements = scope.querySelectorAll(part);
-            for (const node of elements) {
-              if (i === parts.length - 1) {
-                next.push(node);
-              } else if (node.shadowRoot) {
-                next.push(node.shadowRoot);
-              }
-            }
-          }
-          if (i === parts.length - 1) {
-            return next[0] || null;
-          }
-          roots = next;
-        }
-        return null;
-      }
-      JAVASCRIPT
-      query_all: <<~JAVASCRIPT,
-      (element, selector) => {
-        const parts = selector.split('>>>').map((part) => part.trim()).filter(Boolean);
-        let roots = [element];
-        if (parts.length === 0) return [];
-        for (let i = 0; i < parts.length; i++) {
-          const part = parts[i];
-          const next = [];
-          for (const root of roots) {
-            const scope = root;
-            const elements = scope.querySelectorAll(part);
-            for (const node of elements) {
-              if (i === parts.length - 1) {
-                next.push(node);
-              } else if (node.shadowRoot) {
-                next.push(node.shadowRoot);
-              }
-            }
-          }
-          if (i === parts.length - 1) {
-            return next;
-          }
-          roots = next;
-        }
-        return [];
-      }
-      JAVASCRIPT
-    )
+    @p_query_handler ||= Puppeteer::PQueryHandler.new(query_handler_manager: self)
   end
 
   private def xpath_handler
@@ -331,11 +276,20 @@ class Puppeteer::QueryHandlerManager
         raise ArgumentError.new("Query set to use \"#{name}\", but no query handler of that name was found")
       end
       polling = name == 'aria' ? 'raf' : 'mutation'
-    elsif selector.include?('>>>')
-      query_handler = p_query_handler
-      polling = 'raf'
     else
-      query_handler = default_handler
+      begin
+        _selectors, is_pure_css, has_pseudo_classes, has_aria =
+          Puppeteer::PSelectorParser.parse(selector)
+        if is_pure_css
+          polling = has_pseudo_classes ? 'raf' : 'mutation'
+          query_handler = default_handler
+        else
+          polling = has_aria ? 'raf' : 'mutation'
+          query_handler = p_query_handler
+        end
+      rescue StandardError
+        query_handler = default_handler
+      end
     end
 
     Result.new(