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) 


2169
 
# File 'sig/types.rbs', line 2169

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) 


2136
2137
2138
 
# File 'sig/types.rbs', line 2136

def allow_subdomains
  @allow_subdomains
end

#asset_types ⇒ Array[AssetCategory]?

Returns the value of attribute asset_types.

Returns:



2153
2154
2155
 
# File 'sig/types.rbs', line 2153

def asset_types
  @asset_types
end

#auth ⇒ AuthConfig?

Returns the value of attribute auth.

Returns:



2146
2147
2148
 
# File 'sig/types.rbs', line 2146

def auth
  @auth
end

#browser ⇒ BrowserConfig?

Returns the value of attribute browser.

Returns:



2155
2156
2157
 
# File 'sig/types.rbs', line 2155

def browser
  @browser
end

#browser_profile ⇒ String?

Returns the value of attribute browser_profile.

Returns:

  • (String, nil) 


2165
2166
2167
 
# File 'sig/types.rbs', line 2165

def browser_profile
  @browser_profile
end

#capture_screenshot ⇒ Boolean?

Returns the value of attribute capture_screenshot.

Returns:

  • (Boolean, nil) 


2158
2159
2160
 
# File 'sig/types.rbs', line 2158

def capture_screenshot
  @capture_screenshot
end

#content ⇒ ContentConfig?

Returns the value of attribute content.

Returns:



2149
2150
2151
 
# File 'sig/types.rbs', line 2149

def content
  @content
end

#cookies_enabled ⇒ Boolean?

Returns the value of attribute cookies_enabled.

Returns:

  • (Boolean, nil) 


2145
2146
2147
 
# File 'sig/types.rbs', line 2145

def cookies_enabled
  @cookies_enabled
end

#custom_headers ⇒ Hash[String, String]?

Returns the value of attribute custom_headers.

Returns:

  • (Hash[String, String], nil) 


2139
2140
2141
 
# File 'sig/types.rbs', line 2139

def custom_headers
  @custom_headers
end

#document_max_size ⇒ Integer?

Returns the value of attribute document_max_size.

Returns:

  • (Integer, nil) 


2162
2163
2164
 
# File 'sig/types.rbs', line 2162

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) 


2163
2164
2165
 
# File 'sig/types.rbs', line 2163

def document_mime_types
  @document_mime_types
end

#document_url_depth ⇒ Integer?

Returns the value of attribute document_url_depth.

Returns:

  • (Integer, nil) 


2160
2161
2162
 
# File 'sig/types.rbs', line 2160

def document_url_depth
  @document_url_depth
end

#download_assets ⇒ Boolean?

Returns the value of attribute download_assets.

Returns:

  • (Boolean, nil) 


2152
2153
2154
 
# File 'sig/types.rbs', line 2152

def download_assets
  @download_assets
end

#download_documents ⇒ Boolean?

Returns the value of attribute download_documents.

Returns:

  • (Boolean, nil) 


2161
2162
2163
 
# File 'sig/types.rbs', line 2161

def download_documents
  @download_documents
end

#exclude_paths ⇒ Array[String]?

Returns the value of attribute exclude_paths.

Returns:

  • (Array[String], nil) 


2138
2139
2140
 
# File 'sig/types.rbs', line 2138

def exclude_paths
  @exclude_paths
end

#follow_document_urls ⇒ Boolean?

Returns the value of attribute follow_document_urls.

Returns:

  • (Boolean, nil) 


2159
2160
2161
 
# File 'sig/types.rbs', line 2159

def follow_document_urls
  @follow_document_urls
end

#include_paths ⇒ Array[String]?

Returns the value of attribute include_paths.

Returns:

  • (Array[String], nil) 


2137
2138
2139
 
# File 'sig/types.rbs', line 2137

def include_paths
  @include_paths
end

#map_limit ⇒ Integer?

Returns the value of attribute map_limit.

Returns:

  • (Integer, nil) 


2150
2151
2152
 
# File 'sig/types.rbs', line 2150

def map_limit
  @map_limit
end

#map_search ⇒ String?

Returns the value of attribute map_search.

Returns:

  • (String, nil) 


2151
2152
2153
 
# File 'sig/types.rbs', line 2151

def map_search
  @map_search
end

#max_asset_size ⇒ Integer?

Returns the value of attribute max_asset_size.

Returns:

  • (Integer, nil) 


2154
2155
2156
 
# File 'sig/types.rbs', line 2154

def max_asset_size
  @max_asset_size
end

#max_body_size ⇒ Integer?

Returns the value of attribute max_body_size.

Returns:

  • (Integer, nil) 


2147
2148
2149
 
# File 'sig/types.rbs', line 2147

def max_body_size
  @max_body_size
end

#max_concurrent ⇒ Integer?

Returns the value of attribute max_concurrent.

Returns:

  • (Integer, nil) 


2131
2132
2133
 
# File 'sig/types.rbs', line 2131

def max_concurrent
  @max_concurrent
end

#max_depth ⇒ Integer?

Returns the value of attribute max_depth.

Returns:

  • (Integer, nil) 


2129
2130
2131
 
# File 'sig/types.rbs', line 2129

def max_depth
  @max_depth
end

#max_pages ⇒ Integer?

Returns the value of attribute max_pages.

Returns:

  • (Integer, nil) 


2130
2131
2132
 
# File 'sig/types.rbs', line 2130

def max_pages
  @max_pages
end

#max_redirects ⇒ Integer?

Returns the value of attribute max_redirects.

Returns:

  • (Integer, nil) 


2142
2143
2144
 
# File 'sig/types.rbs', line 2142

def max_redirects
  @max_redirects
end

#proxy ⇒ ProxyConfig?

Returns the value of attribute proxy.

Returns:



2156
2157
2158
 
# File 'sig/types.rbs', line 2156

def proxy
  @proxy
end

#rate_limit_ms ⇒ Integer?

Returns the value of attribute rate_limit_ms.

Returns:

  • (Integer, nil) 


2141
2142
2143
 
# File 'sig/types.rbs', line 2141

def rate_limit_ms
  @rate_limit_ms
end

#remove_tags ⇒ Array[String]?

Returns the value of attribute remove_tags.

Returns:

  • (Array[String], nil) 


2148
2149
2150
 
# File 'sig/types.rbs', line 2148

def remove_tags
  @remove_tags
end

#request_timeout ⇒ Integer?

Returns the value of attribute request_timeout.

Returns:

  • (Integer, nil) 


2140
2141
2142
 
# File 'sig/types.rbs', line 2140

def request_timeout
  @request_timeout
end

#respect_robots_txt ⇒ Boolean?

Returns the value of attribute respect_robots_txt.

Returns:

  • (Boolean, nil) 


2132
2133
2134
 
# File 'sig/types.rbs', line 2132

def respect_robots_txt
  @respect_robots_txt
end

#retry_codes ⇒ Array[Integer]?

Returns the value of attribute retry_codes.

Returns:

  • (Array[Integer], nil) 


2144
2145
2146
 
# File 'sig/types.rbs', line 2144

def retry_codes
  @retry_codes
end

#retry_count ⇒ Integer?

Returns the value of attribute retry_count.

Returns:

  • (Integer, nil) 


2143
2144
2145
 
# File 'sig/types.rbs', line 2143

def retry_count
  @retry_count
end

#save_browser_profile ⇒ Boolean?

Returns the value of attribute save_browser_profile.

Returns:

  • (Boolean, nil) 


2166
2167
2168
 
# File 'sig/types.rbs', line 2166

def save_browser_profile
  @save_browser_profile
end

#soft_http_errors ⇒ Boolean?

Returns the value of attribute soft_http_errors.

Returns:

  • (Boolean, nil) 


2133
2134
2135
 
# File 'sig/types.rbs', line 2133

def soft_http_errors
  @soft_http_errors
end

#ssrf ⇒ SsrfPolicy?

Returns the value of attribute ssrf.

Returns:



2167
2168
2169
 
# File 'sig/types.rbs', line 2167

def ssrf
  @ssrf
end

#stay_on_domain ⇒ Boolean?

Returns the value of attribute stay_on_domain.

Returns:

  • (Boolean, nil) 


2135
2136
2137
 
# File 'sig/types.rbs', line 2135

def stay_on_domain
  @stay_on_domain
end

#user_agent ⇒ String?

Returns the value of attribute user_agent.

Returns:

  • (String, nil) 


2134
2135
2136
 
# File 'sig/types.rbs', line 2134

def user_agent
  @user_agent
end

#user_agents ⇒ Array[String]?

Returns the value of attribute user_agents.

Returns:

  • (Array[String], nil) 


2157
2158
2159
 
# File 'sig/types.rbs', line 2157

def user_agents
  @user_agents
end

#warc_output ⇒ String?

Returns the value of attribute warc_output.

Returns:

  • (String, nil) 


2164
2165
2166
 
# File 'sig/types.rbs', line 2164

def warc_output
  @warc_output
end