jobo.dtd

网络爬虫源码

<!ELEMENT JoBo ( Robot? | 
                 URLCheck? | 
                 DownloadRuleSet? |
                 LocalizeLinks? |
                 StoreCGI )  >

<!ELEMENT Robot ( AgentName? | 
                  StartReferer? | 
                  IgnoreRobotsTxt? |
                  SleepTime? |
                  MaxDepth? |
                  WalkToOtherHosts? |
                  Proxy? |
		  Bandwidth? |
                  MaxAge? |
                  EnableCookies |
                  AllowWholeHost? |
                  AllowWholeDomain? |
                  FlexibleHostCheck? |
                  AllowCaching? |
		  AllowedURL* |
		  VisitMany* |
		  FormHandler*
		  )?  >

<!ELEMENT URLCheck ( RegExpRule* )?  >

<!ATTLIST URLCheck
           default CDATA #IMPLIED >

<!ELEMENT DownloadRuleSet ( DownloadRule* ) >

<!ATTLIST DownloadRuleSet
           default CDATA #IMPLIED >

<!ELEMENT AgentName (#PCDATA)* >

<!ELEMENT StartReferer (#PCDATA)* >

<!ELEMENT IgnoreRobotsTxt (#PCDATA)* >

<!ELEMENT SleepTime (#PCDATA)* >

<!ELEMENT MaxDepth (#PCDATA)* >

<!ELEMENT WalkToOtherHosts (#PCDATA)* >

<!ELEMENT Proxy (#PCDATA)* >

<!ELEMENT AllowWholeHost (#PCDATA)* >

<!ELEMENT AllowWholeDomain (#PCDATA)* >

<!ELEMENT AllowCaching (#PCDATA)* >

<!ELEMENT FlexibleHostCheck (#PCDATA)* >

<!ELEMENT AllowedURL (#PCDATA)* >

<!ELEMENT VisitMany (#PCDATA)* >

<!ELEMENT Bandwidth (#PCDATA)* >

<!ELEMENT FormHandler ( FormField* ) >

<!ATTLIST FormHandler 
           url CDATA #REQUIRED >

<!ELEMENT FormField EMPTY >

<!ATTLIST FormField
           name CDATA #REQUIRED
	   value CDATA #REQUIRED >

<!ELEMENT RegExpRule EMPTY >

<!ATTLIST RegExpRule 
           allow CDATA #REQUIRED
           pattern CDATA  #REQUIRED >

<!ELEMENT DownloadRule EMPTY >

<!ATTLIST DownloadRule 
           allow CDATA #REQUIRED
           minSize CDATA #IMPLIED 
           maxSize CDATA #IMPLIED 
           mimeType CDATA #REQUIRED >

<!ELEMENT LocalizeLinks ( #PCDATA ) >

<!ELEMENT StoreCGI ( #PCDATA ) >