Class: Puppeteer::PQueryHandler

Inherits:
Object
  • Object
show all
Defined in:
lib/puppeteer/p_query_handler.rb,
sig/puppeteer/p_query_handler.rbs

Overview

rbs_inline: enabled

Constant Summary collapse

POLLING_DEFAULT_SECONDS =

Returns:

  • (::Float)
0.1
CSS_QUERY_SELECTOR_JS =

Returns:

  • (::String)
<<~'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 =

Returns:

  • (::String)
<<~JAVASCRIPT
  node => {
    if (!node) {
      return [];
    }
    if (node.shadowRoot) {
      return [node.shadowRoot];
    }
    return [node];
  }
JAVASCRIPT
PIERCE_ALL_JS =

Returns:

  • (::String)
<<~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 =

Returns:

  • (::String)
<<~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

Instance Method Summary collapse

Constructor Details

#initialize(query_handler_manager: Puppeteer::QueryHandlerManager.instance) ⇒ PQueryHandler

Returns a new instance of PQueryHandler.

Parameters:

  • query_handler_manager: (Object) (defaults to: Puppeteer::QueryHandlerManager.instance)


127
128
129
# File 'lib/puppeteer/p_query_handler.rb', line 127

def initialize(query_handler_manager: Puppeteer::QueryHandlerManager.instance)
  @query_handler_manager = query_handler_manager
end

Instance Method Details

#apply_compound(elements, compound) ⇒ Object

Parameters:

  • elements (Object)
  • compound (Object)

Returns:

  • (Object)


250
251
252
253
254
255
256
257
258
259
260
# File 'lib/puppeteer/p_query_handler.rb', line 250

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

#apply_css_selector(elements, selector) ⇒ Object

Parameters:

  • elements (Object)
  • selector (Object)

Returns:

  • (Object)


262
263
264
265
266
# File 'lib/puppeteer/p_query_handler.rb', line 262

def apply_css_selector(elements, selector)
  elements.flat_map do |element|
    query_css_selector(element, selector)
  end
end

#apply_pseudo_selector(elements, pseudo) ⇒ Object

Parameters:

  • elements (Object)
  • pseudo (Object)

Returns:

  • (Object)


277
278
279
280
281
282
283
284
285
286
287
# File 'lib/puppeteer/p_query_handler.rb', line 277

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

#array_handle_to_elements(array_handle) ⇒ Object

Parameters:

  • array_handle (Object)

Returns:

  • (Object)


326
327
328
329
330
331
# File 'lib/puppeteer/p_query_handler.rb', line 326

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

#dom_sort(elements) ⇒ Object

Parameters:

  • elements (Object)

Returns:

  • (Object)


314
315
316
317
318
319
320
321
322
323
324
# File 'lib/puppeteer/p_query_handler.rb', line 314

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

#ensure_context(handle, context) ⇒ Object

Parameters:

  • handle (Object)
  • context (Object)

Returns:

  • (Object)


333
334
335
336
337
338
339
340
# File 'lib/puppeteer/p_query_handler.rb', line 333

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

#monotonic_timeObject

Returns:

  • (Object)


360
361
362
# File 'lib/puppeteer/p_query_handler.rb', line 360

def monotonic_time
  Process.clock_gettime(Process::CLOCK_MONOTONIC)
end

#pierce(elements) ⇒ Object

Parameters:

  • elements (Object)

Returns:

  • (Object)


296
297
298
299
300
301
302
303
# File 'lib/puppeteer/p_query_handler.rb', line 296

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

#pierce_all(elements) ⇒ Object

Parameters:

  • elements (Object)

Returns:

  • (Object)


305
306
307
308
309
310
311
312
# File 'lib/puppeteer/p_query_handler.rb', line 305

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

#polling_interval_seconds(polling, visible:, hidden:) ⇒ Object

Parameters:

  • polling (Object)
  • visible: (Object)
  • hidden: (Object)

Returns:

  • (Object)


350
351
352
353
354
355
356
357
358
# File 'lib/puppeteer/p_query_handler.rb', line 350

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

#query_all(element, selector) ⇒ Array[Puppeteer::ElementHandle]

Parameters:

Returns:



144
145
146
147
148
149
150
151
152
153
154
155
# File 'lib/puppeteer/p_query_handler.rb', line 144

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

#query_all_array(element, selector) ⇒ Puppeteer::JSHandle

Parameters:

Returns:



160
161
162
163
164
165
166
167
# File 'lib/puppeteer/p_query_handler.rb', line 160

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

#query_css_selector(element, selector) ⇒ Object

Parameters:

  • element (Object)
  • selector (Object)

Returns:

  • (Object)


268
269
270
271
272
273
274
275
# File 'lib/puppeteer/p_query_handler.rb', line 268

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

#query_handler_for_pseudo(name) ⇒ Object

Parameters:

  • name (Object)

Returns:

  • (Object)

Raises:



289
290
291
292
293
294
# File 'lib/puppeteer/p_query_handler.rb', line 289

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

#query_one(element, selector) ⇒ Puppeteer::ElementHandle?

Parameters:

Returns:



134
135
136
137
138
139
# File 'lib/puppeteer/p_query_handler.rb', line 134

def query_one(element, selector)
  matches = query_all(element, selector)
  return nil if matches.empty?

  matches.first
end

#raise_timeout(selector) ⇒ Object

Parameters:

  • selector (Object)

Returns:

  • (Object)

Raises:



364
365
366
# File 'lib/puppeteer/p_query_handler.rb', line 364

def raise_timeout(selector)
  raise Puppeteer::TimeoutError.new("Waiting for selector `#{selector}` failed")
end

#run_complex_selector(root, complex) ⇒ Object

Parameters:

  • root (Object)
  • complex (Object)

Returns:

  • (Object)


232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
# File 'lib/puppeteer/p_query_handler.rb', line 232

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

#transfer_handle(frame, handle) ⇒ Object

Parameters:

  • frame (Object)
  • handle (Object)

Returns:

  • (Object)


342
343
344
345
346
347
348
# File 'lib/puppeteer/p_query_handler.rb', line 342

def transfer_handle(frame, handle)
  if handle.execution_context == frame.main_world.execution_context
    handle
  else
    frame.main_world.transfer_handle(handle)
  end
end

#wait_for(element_or_frame, selector, visible:, hidden:, timeout:, polling: nil) ⇒ Puppeteer::ElementHandle?

Parameters:

  • element_or_frame (Puppeteer::ElementHandle, Puppeteer::Frame)
  • selector (String)
  • visible: (Boolean, nil)
  • hidden: (Boolean, nil)
  • timeout: (Numeric, nil)
  • polling: (String, Numeric, nil) (defaults to: nil)

Returns:



176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
# File 'lib/puppeteer/p_query_handler.rb', line 176

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