Daniel Turcanu has proposed merging lp:~danielturcanu/zorba/my_conv_module into 
lp:zorba/data-converters-module.

Requested reviews:
  Zorba Coders (zorba-coders)

For more details, see:
https://code.launchpad.net/~danielturcanu/zorba/my_conv_module/+merge/78608

Fixed the link crawler test
-- 
https://code.launchpad.net/~danielturcanu/zorba/my_conv_module/+merge/78608
Your team Zorba Coders is requested to review the proposed merge of 
lp:~danielturcanu/zorba/my_conv_module into lp:zorba/data-converters-module.
=== modified file 'test_html/Queries/converters/html/link_crawler2.xq2'
--- test_html/Queries/converters/html/link_crawler2.xq2	2011-10-06 07:40:17 +0000
+++ test_html/Queries/converters/html/link_crawler2.xq2	2011-10-07 14:24:47 +0000
@@ -18,6 +18,7 @@
 import module namespace map = "http://www.zorba-xquery.com/modules/store/data-structures/unordered-map";;
 import module namespace html = "http://www.zorba-xquery.com/modules/converters/html";;
 import module namespace parse-xml = "http://www.zorba-xquery.com/modules/xml";;
+import module namespace file = "http://expath.org/ns/file";;
 
 declare namespace ann = "http://www.zorba-xquery.com/annotations";;
 declare namespace xhtml="http://www.w3.org/1999/xhtml";;
@@ -29,12 +30,9 @@
 declare variable $uri-host as xs:string := "http://www.zorba-xquery.com/site2/";;
 
 
-declare variable $supported-media-types as xs:string+ := ("text/xml", "application/xml", "text/xml-external-parsed-entity", "application/xml-external-parsed-entity",
-                         "application/atom+xml", "text/html");
-
-
-declare variable $local:processed-internal-links:=xs:QName("processed-internal-links");
-declare variable $local:processed-external-links  :=xs:QName("processed-external-links");
+
+declare variable $local:processed-internal-links := xs:QName("processed-internal-links");
+declare variable $local:processed-external-links := xs:QName("processed-external-links");
 
 
 declare %ann:sequential function local:create-containers()
@@ -59,9 +57,17 @@
 return  if($sb = "") then  $s1 else $sb
 };
 
-declare function local:get-real-link($href as xs:string, $start-uri as xs:string) as xs:string
+declare %ann:sequential function local:get-real-link($href as xs:string, $start-uri as xs:string) as xs:string?
 {
-   local:my-substring-before(resolve-uri($href, $start-uri), "#")
+   variable $absuri;
+   try{
+    $absuri := local:my-substring-before(resolve-uri(fn:normalize-space($href), $start-uri), "#");
+   }
+   catch *
+   { 
+     map:insert($local:processed-external-links, fn:concat("malformed, referenced in page ", $start-uri), $href);
+   }
+   $absuri
 };
 
 
@@ -70,13 +76,15 @@
    local:my-substring-before($http-call/httpsch:header[@name = 'Content-Type'][1]/string(@value), ";")
 };
 
-declare function local:alive($http-call as node()*) as xs:boolean
+declare function local:alive($http-call as item()*) as xs:boolean
 {
- if(($http-call[1]/@status eq 200)) then true() else false()
+ if((count($http-call) ge 1) and 
+    ($http-call[1]/@status eq 200)) 
+   then true() else fn:trace(false(), "alive")
 };
 
 
-declare function local:get-out-links-parsed($content as node()*, $uri as xs:string) as xs:string*
+declare %ann:sequential function local:get-out-links-parsed($content as node()*, $uri as xs:string) as xs:string*
 {  distinct-values( for $y in  ($content//*:a/string(@href),
                               $content//*:link/string(@href),
                               $content//*:script/string(@src),
@@ -87,22 +95,29 @@
 };
 
 
-declare function local:get-out-links-unparsed($content as xs:string, $uri as xs:string) as xs:string*{
+declare %ann:sequential function local:get-out-links-unparsed($content as xs:string, $uri as xs:string) as xs:string*{
 
       distinct-values( 
          let $search := fn:analyze-string($content, "(&lt;|&amp;lt;|<)(((a|link|area).+?href)|((script|img).+?src))=([""'])(.*?)\7")
          for $other-uri2 in  $search//group[@nr=8]/string()
-         let $y:= fn:normalize-space($other-uri2)
-         return local:get-real-link($y, $uri)
+         return local:get-real-link($other-uri2, $uri)
      )
 };
 
 
 
+declare %ann:sequential function local:process-link($x as xs:string, $n as xs:integer) as item()*{
+ if(local:is-internal($x))
+       then local:process-internal-link($x, $n);
+       else local:process-external-link($x);
+
+};
+
 declare  %ann:sequential function local:process-external-link($x as xs:string){
   if(not(empty(map:get($local:processed-external-links, $x))))
          then   exit returning false();
          else {}
+ fn:trace($x, "HEAD external link");
  variable $http-call:=();
   try{
         $http-call:=http:send-request(<httpsch:request method="HEAD" href="{$x}"/>, (), ());
@@ -127,39 +142,42 @@
 
 
 declare  %ann:sequential function local:process-internal-link($x as xs:string, $n as xs:integer){
-      if($n=3) then exit returning (); else {}
+      (: if($n=3) then exit returning (); else {} :)
       if(not(empty(map:get($local:processed-internal-links, $x))))
             then exit returning false();
               else {}
+      fn:trace($x, "GET internal link");
        variable $http-call:=();
        try{
              $http-call:=http:send-request(<httpsch:request method="GET" href="{$x}"/>, (), ());
        }
-       catch * {}
-      if( not(local:alive($http-call)))
+       catch * { }
+       if( not(local:alive($http-call)))
                 then { map:insert($local:processed-internal-links, "broken", $x); exit returning ();}
                else {}
-       if(not (local:get-media-type($http-call[1]) = $supported-media-types))
-                then {map:insert($local:processed-internal-links, "clean", $x);  exit returning ();}
+       
+       if(not (local:get-media-type($http-call[1]) = "text/html"))
+                then { map:insert($local:processed-internal-links, "clean", $x); exit returning ();}
                 else {}
-       variable $string-content := xs:string($http-call[2]);
+       variable $string-content := string($http-call[2]);
        variable $content:=();
 
        try{
              $content:=html:parse($string-content,local:tidy-options() );
+             map:insert($local:processed-internal-links, "clean", $x);
         }
         catch *
-             { 
-                 map:insert($local:processed-internal-links, concat("cannot tidy", $err:description), $x); 
+             {   
+                 map:insert($local:processed-internal-links, concat("cannot tidy ", $err:description), $x); 
                  try{
                        $content:=parse-xml:parse-xml-fragment ($string-content, "");
                  }
                  catch *
-                     { map:insert($local:processed-internal-links, concat("cannot parse", $err:description), $x);}
+                     { map:insert($local:processed-internal-links, concat("cannot parse ", $err:description), $x);}
             }
        variable $links :=();
        if(empty($content))
-            then $links:=local:get-out-links-unparsed($string-content, $x);
+           then $links:=local:get-out-links-unparsed($string-content, fn:trace($x, "parse with regex, because tidy failed"));
            else $links:=local:get-out-links-parsed($content, $x);
        for $l in $links
        return  local:process-link($l, $n+1);
@@ -167,13 +185,6 @@
 
 
 
-declare %ann:sequential function local:process-link($x as xs:string, $n as xs:integer) as item()*{
- if(local:is-internal($x))
-       then local:process-internal-link($x, $n);
-       else local:process-external-link($x);
-
-};
-
 
 declare function local:print-results() as element()*
 {
@@ -190,13 +201,15 @@
 
 variable $result;
 
-try {
-  local:create-containers();
-  local:process-link($uri, 1);
-  $result:=local:print-results() ;
-}
-catch * { $result:=concat("an error occurred", $err:description);}
-
+local:create-containers();
+local:process-link($uri, 1);
+$result:=local:print-results() ;
 
 local:delete-containers();
-$result
+
+file:write(fn:resolve-uri("link_crawler_result.xml"),
+            <result>{$result}</result>,
+            <output:serialization-parameters>
+                <output:indent value="yes"/>
+            </output:serialization-parameters>)
+

-- 
Mailing list: https://launchpad.net/~zorba-coders
Post to     : zorba-coders@lists.launchpad.net
Unsubscribe : https://launchpad.net/~zorba-coders
More help   : https://help.launchpad.net/ListHelp

Reply via email to