Class: Xberg::CrawlConfig

Inherits:
Object
  • Object
show all
Defined in:
sig/types.rbs

Instance Attribute Summary collapse

Instance Method Summary collapse

Constructor Details

#initialize ⇒ CrawlConfig

Returns a new instance of CrawlConfig.

Parameters:

  • max_depth: (Integer)
  • max_pages: (Integer)
  • max_concurrent: (Integer)
  • respect_robots_txt: (Boolean)
  • soft_http_errors: (Boolean)
  • user_agent: (String)
  • stay_on_domain: (Boolean)
  • allow_subdomains: (Boolean)
  • 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: (Boolean)
  • auth: (AuthConfig)
  • max_body_size: (Integer)
  • remove_tags: (Array[String])
  • content: (ContentConfig)
  • map_limit: (Integer)
  • map_search: (String)
  • download_assets: (Boolean)
  • asset_types: (Array[AssetCategory])
  • max_asset_size: (Integer)
  • browser: (BrowserConfig)
  • proxy: (ProxyConfig)
  • user_agents: (Array[String])
  • capture_screenshot: (Boolean)
  • follow_document_urls: (Boolean)
  • document_url_depth: (Integer)
  • download_documents: (Boolean)
  • document_max_size: (Integer)
  • document_mime_types: (Array[String])
  • warc_output: (String)
  • browser_profile: (String)
  • save_browser_profile: (Boolean)
  • ssrf: (SsrfPolicy) 


2354
 
# File 'sig/types.rbs', line 2354

def initialize: (?max_depth: Integer, ?max_pages: Integer, ?max_concurrent: Integer, ?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], ?warc_output: String, ?browser_profile: String, ?save_browser_profile: bool, ?ssrf: SsrfPolicy) -> void

Instance Attribute Details

#allow_subdomains ⇒ Boolean?

Returns the value of attribute allow_subdomains.

Returns:

  • (Boolean, nil) 


2321
2322
2323
 
# File 'sig/types.rbs', line 2321

def allow_subdomains
  @allow_subdomains
end

#asset_types ⇒ Array[AssetCategory]?

Returns the value of attribute asset_types.

Returns:



2338
2339
2340
 
# File 'sig/types.rbs', line 2338

def asset_types
  @asset_types
end

#auth ⇒ AuthConfig?

Returns the value of attribute auth.

Returns:



2331
2332
2333
 
# File 'sig/types.rbs', line 2331

def auth
  @auth
end

#browser ⇒ BrowserConfig?

Returns the value of attribute browser.

Returns:



2340
2341
2342
 
# File 'sig/types.rbs', line 2340

def browser
  @browser
end

#browser_profile ⇒ String?

Returns the value of attribute browser_profile.

Returns:

  • (String, nil) 


2350
2351
2352
 
# File 'sig/types.rbs', line 2350

def browser_profile
  @browser_profile
end

#capture_screenshot ⇒ Boolean?

Returns the value of attribute capture_screenshot.

Returns:

  • (Boolean, nil) 


2343
2344
2345
 
# File 'sig/types.rbs', line 2343

def capture_screenshot
  @capture_screenshot
end

#content ⇒ ContentConfig?

Returns the value of attribute content.

Returns:



2334
2335
2336
 
# File 'sig/types.rbs', line 2334

def content
  @content
end

#cookies_enabled ⇒ Boolean?

Returns the value of attribute cookies_enabled.

Returns:

  • (Boolean, nil) 


2330
2331
2332
 
# File 'sig/types.rbs', line 2330

def cookies_enabled
  @cookies_enabled
end

#custom_headers ⇒ Hash[String, String]?

Returns the value of attribute custom_headers.

Returns:

  • (Hash[String, String], nil) 


2324
2325
2326
 
# File 'sig/types.rbs', line 2324

def custom_headers
  @custom_headers
end

#document_max_size ⇒ Integer?

Returns the value of attribute document_max_size.

Returns:

  • (Integer, nil) 


2347
2348
2349
 
# File 'sig/types.rbs', line 2347

def document_max_size
  @document_max_size
end

#document_mime_types ⇒ Array[String]?

Returns the value of attribute document_mime_types.

Returns:

  • (Array[String], nil) 


2348
2349
2350
 
# File 'sig/types.rbs', line 2348

def document_mime_types
  @document_mime_types
end

#document_url_depth ⇒ Integer?

Returns the value of attribute document_url_depth.

Returns:

  • (Integer, nil) 


2345
2346
2347
 
# File 'sig/types.rbs', line 2345

def document_url_depth
  @document_url_depth
end

#download_assets ⇒ Boolean?

Returns the value of attribute download_assets.

Returns:

  • (Boolean, nil) 


2337
2338
2339
 
# File 'sig/types.rbs', line 2337

def download_assets
  @download_assets
end

#download_documents ⇒ Boolean?

Returns the value of attribute download_documents.

Returns:

  • (Boolean, nil) 


2346
2347
2348
 
# File 'sig/types.rbs', line 2346

def download_documents
  @download_documents
end

#exclude_paths ⇒ Array[String]?

Returns the value of attribute exclude_paths.

Returns:

  • (Array[String], nil) 


2323
2324
2325
 
# File 'sig/types.rbs', line 2323

def exclude_paths
  @exclude_paths
end

#follow_document_urls ⇒ Boolean?

Returns the value of attribute follow_document_urls.

Returns:

  • (Boolean, nil) 


2344
2345
2346
 
# File 'sig/types.rbs', line 2344

def follow_document_urls
  @follow_document_urls
end

#include_paths ⇒ Array[String]?

Returns the value of attribute include_paths.

Returns:

  • (Array[String], nil) 


2322
2323
2324
 
# File 'sig/types.rbs', line 2322

def include_paths
  @include_paths
end

#map_limit ⇒ Integer?

Returns the value of attribute map_limit.

Returns:

  • (Integer, nil) 


2335
2336
2337
 
# File 'sig/types.rbs', line 2335

def map_limit
  @map_limit
end

#map_search ⇒ String?

Returns the value of attribute map_search.

Returns:

  • (String, nil) 


2336
2337
2338
 
# File 'sig/types.rbs', line 2336

def map_search
  @map_search
end

#max_asset_size ⇒ Integer?

Returns the value of attribute max_asset_size.

Returns:

  • (Integer, nil) 


2339
2340
2341
 
# File 'sig/types.rbs', line 2339

def max_asset_size
  @max_asset_size
end

#max_body_size ⇒ Integer?

Returns the value of attribute max_body_size.

Returns:

  • (Integer, nil) 


2332
2333
2334
 
# File 'sig/types.rbs', line 2332

def max_body_size
  @max_body_size
end

#max_concurrent ⇒ Integer?

Returns the value of attribute max_concurrent.

Returns:

  • (Integer, nil) 


2316
2317
2318
 
# File 'sig/types.rbs', line 2316

def max_concurrent
  @max_concurrent
end

#max_depth ⇒ Integer?

Returns the value of attribute max_depth.

Returns:

  • (Integer, nil) 


2314
2315
2316
 
# File 'sig/types.rbs', line 2314

def max_depth
  @max_depth
end

#max_pages ⇒ Integer?

Returns the value of attribute max_pages.

Returns:

  • (Integer, nil) 


2315
2316
2317
 
# File 'sig/types.rbs', line 2315

def max_pages
  @max_pages
end

#max_redirects ⇒ Integer?

Returns the value of attribute max_redirects.

Returns:

  • (Integer, nil) 


2327
2328
2329
 
# File 'sig/types.rbs', line 2327

def max_redirects
  @max_redirects
end

#proxy ⇒ ProxyConfig?

Returns the value of attribute proxy.

Returns:



2341
2342
2343
 
# File 'sig/types.rbs', line 2341

def proxy
  @proxy
end

#rate_limit_ms ⇒ Integer?

Returns the value of attribute rate_limit_ms.

Returns:

  • (Integer, nil) 


2326
2327
2328
 
# File 'sig/types.rbs', line 2326

def rate_limit_ms
  @rate_limit_ms
end

#remove_tags ⇒ Array[String]?

Returns the value of attribute remove_tags.

Returns:

  • (Array[String], nil) 


2333
2334
2335
 
# File 'sig/types.rbs', line 2333

def remove_tags
  @remove_tags
end

#request_timeout ⇒ Integer?

Returns the value of attribute request_timeout.

Returns:

  • (Integer, nil) 


2325
2326
2327
 
# File 'sig/types.rbs', line 2325

def request_timeout
  @request_timeout
end

#respect_robots_txt ⇒ Boolean?

Returns the value of attribute respect_robots_txt.

Returns:

  • (Boolean, nil) 


2317
2318
2319
 
# File 'sig/types.rbs', line 2317

def respect_robots_txt
  @respect_robots_txt
end

#retry_codes ⇒ Array[Integer]?

Returns the value of attribute retry_codes.

Returns:

  • (Array[Integer], nil) 


2329
2330
2331
 
# File 'sig/types.rbs', line 2329

def retry_codes
  @retry_codes
end

#retry_count ⇒ Integer?

Returns the value of attribute retry_count.

Returns:

  • (Integer, nil) 


2328
2329
2330
 
# File 'sig/types.rbs', line 2328

def retry_count
  @retry_count
end

#save_browser_profile ⇒ Boolean?

Returns the value of attribute save_browser_profile.

Returns:

  • (Boolean, nil) 


2351
2352
2353
 
# File 'sig/types.rbs', line 2351

def save_browser_profile
  @save_browser_profile
end

#soft_http_errors ⇒ Boolean?

Returns the value of attribute soft_http_errors.

Returns:

  • (Boolean, nil) 


2318
2319
2320
 
# File 'sig/types.rbs', line 2318

def soft_http_errors
  @soft_http_errors
end

#ssrf ⇒ SsrfPolicy?

Returns the value of attribute ssrf.

Returns:



2352
2353
2354
 
# File 'sig/types.rbs', line 2352

def ssrf
  @ssrf
end

#stay_on_domain ⇒ Boolean?

Returns the value of attribute stay_on_domain.

Returns:

  • (Boolean, nil) 


2320
2321
2322
 
# File 'sig/types.rbs', line 2320

def stay_on_domain
  @stay_on_domain
end

#user_agent ⇒ String?

Returns the value of attribute user_agent.

Returns:

  • (String, nil) 


2319
2320
2321
 
# File 'sig/types.rbs', line 2319

def user_agent
  @user_agent
end

#user_agents ⇒ Array[String]?

Returns the value of attribute user_agents.

Returns:

  • (Array[String], nil) 


2342
2343
2344
 
# File 'sig/types.rbs', line 2342

def user_agents
  @user_agents
end

#warc_output ⇒ String?

Returns the value of attribute warc_output.

Returns:

  • (String, nil) 


2349
2350
2351
 
# File 'sig/types.rbs', line 2349

def warc_output
  @warc_output
end