[fix] slick-router

src/lib/bookpages.xqm

@@ -0,0 +1,63 @@
+xquery version '3.1';
+(:~ describe book page numbers as sequence of ranges, similar to PDF pagelabels
+@author quodatum
+:)
+module namespace bookpages = 'urn:bookpages';
+
+(:~ Invisible-xml grammar to parse custom pagelabel representation  :)
+declare variable $bookpages:grammar:="
+book: pagecount,'#',range,(-',', range)*.
+pagecount:['0'-'9']+.
+range: s,from?,s,type,s,prefix?,s,offset?.
+@from: ['0'-'9']+. { pageIndex }
+@type: ['C'|'D'|'R'|'r'|'A'|'a'|'w'].
+@prefix: -':',~[',']+.
+@offset: -'@',['0'-'9']+.
+
+-s: ([Zs]; #9; #a; #d)*. {Optional whitespace}  
+";
+
+(:~ 
+page number range in given style 
+:)
+declare function bookpages:span($type as xs:string,$length as xs:integer,$first as xs:integer)
+as xs:string*{
+let $r:=$first to $first+$length
+return switch ($type)
+       case "D" return $r!format-integer(.,"1")
+       case "r" return $r!format-integer(.,"i")
+       case "R" return $r!format-integer(.,"I")
+       case "C" return "Cover"
+       default return $r!format-integer(.,$type)
+};
+
+(:~ pagelabels from text:)
+declare function bookpages:expand($pages as xs:string)
+as xs:string*{
+  let $x:=bookpages:parse($pages)
+  let $last:=head($x)=>xs:integer()
+  return hof:until(
+      function($m){ empty($m?ranges) or count($m?result)eq $last },  
+      function($m){ 
+          let $range:=head($m?ranges)=>trace("SS")
+          let $start:=if($range/@offset)then xs:integer($range/@offset) else 1
+          let $end:=($m?ranges[2]/xs:integer(@from)-1)  otherwise $last
+          let $length:=$end -count($m?result)-1
+          let $span:=bookpages:span($range/@type,$length,$start)
+          let $span:=if($range/@prefix)then $span!concat($range/@prefix,.) else $span
+          return map {
+            'ranges': tail($m?ranges),
+            'result': ($m?result, $span)
+      }},
+      
+       (: initial input = grammar ranges :)
+        map { 'ranges': tail($x) , 'result': () }
+    )?result
+};
+ 
+(:~ parse pagenumber description to xml :)
+declare function bookpages:parse($pages as xs:string)
+as element(range)*{
+  invisible-xml($bookpages:grammar)($pages)/*
+};
+

src/lib/pdfbox3.xqm

@@ -2,12 +2,12 @@ xquery version '3.1';
 (:~ 
 pdfbox 3.0 https://pdfbox.apache.org/ BaseX 10.7+ interface library, 
 requires pdfbox jar on classpath
-3.02 required tested with pdfbox-app-3.0.2-20240121.184204-66.jar
+3.02+ required tested with pdfbox-app-3.0.2.jar
 @see https://repository.apache.org/content/groups/snapshots/org/apache/pdfbox/pdfbox-app/3.0.2-SNAPSHOT/
 @javadoc https://javadoc.io/static/org.apache.pdfbox/pdfbox/3.0.0/
 
 :)
-module namespace pdfbox="urn:expkg-zone58:pdfbox:3";
+module namespace pdfbox="urn:expkg-zone58:pdfbox3";
 
 declare namespace Loader ="java:org.apache.pdfbox.Loader"; 
 declare namespace PDFTextStripper = "java:org.apache.pdfbox.text.PDFTextStripper";
@@ -96,7 +96,7 @@ as map(*)*{
 };
 
 (: return bookmark info for children of $outlineItem as seq of maps :)
-declare function pdfbox:outline($doc,$outlineItem )
+declare function pdfbox:outline($doc as item(),$outlineItem as item()?)
 as map(*)*
 {
   let $find:=hof:until(
@@ -143,7 +143,7 @@ as map(*)
   }
 };
 
-declare function pdfbox:outx($page,$document)
+declare function pdfbox:outx($page ,$document)
 {
   let $currentPage := PDOutlineItem:findDestinationPage($page,$document)
   let $pageNumber := pdfbox:pageIndex($currentPage,$document)

src/lib/pdfscrape.xqm

@@ -1,16 +1,16 @@
 xquery version '3.1';
 (:~ look for pagenos in pdf text
-pagenos:page-report($doc )=>pagenos:inverted-map()
+pdfscrape:page-report($doc )=>pdfscrape:inverted-map()
 :)
-module namespace pagenos = 'urn:pageno';
-import module namespace pdfbox="urn:expkg-zone58:pdfbox:3" at "pdfbox3.xqm";
+module namespace pdfscrape = 'urn:pdfscrape';
+import module namespace pdfbox="urn:expkg-zone58:pdfbox3" at "pdfbox3.xqm";
 
 (: look for possible page number in first/last line of page text
 @todo last line and roman
 1=Number system ( D=decimal, R=Roman)
 2=Side L=left,R=right
 :)
-declare variable $pagenos:pats:=map{
+declare variable $pdfscrape:pats:=map{
     "DL": "^([1-9][0-9]*).*",
     "DR": ".*[^0-9]([1-9][0-9]*)$",
     "RL": "^([ivxlc]+).*",
@@ -18,47 +18,56 @@ declare variable $pagenos:pats:=map{
 };
 
 (: page-reports for all pages :)
-declare function pagenos:page-report($doc as item())
+declare function pdfscrape:page-report($doc as item())
 as element(page)*{
   let $count:=pdfbox:page-count($doc)=>trace("Pages: ")
-  return (0 to $count -1)!pagenos:page-report($doc,.)    
+  return (1 to $count )!pdfscrape:page-report($doc,.)    
 };
 
 (: page-report for given page :)
-declare function pagenos:page-report($doc as item(), $page as xs:integer)
+declare function pdfscrape:page-report($doc as item(), $page as xs:integer)
 as element(page){
   let $txt:=pdfbox:getText($doc,$page)
   let $line1:=substring-before($txt,file:line-separator())
-  let $fn:=function($acc,$this){ $acc otherwise pagenos:line-report($this,$line1)}
-  let $found:=map:keys($pagenos:pats)=>fold-left( (),$fn)
+  let $fn:=function($acc,$this){ $acc otherwise pdfscrape:line-report($this,$line1)}
+  let $found:=map:keys($pdfscrape:pats)=>fold-left( (),$fn)
  
   return <page index="{ $page }">{ $found, $line1 }</page>
 };
 
 (: empty or attributes created by matching $style with $line1 :)
-declare function pagenos:line-report($style as xs:string, $line1 as xs:string)
+declare function pdfscrape:line-report($style as xs:string, $line1 as xs:string)
 as attribute(*)*{
-    if(matches($line1,$pagenos:pats?($style)))
+    if(matches($line1,$pdfscrape:pats?($style)))
     then (
           attribute {"style"} { substring($style,1,1) } ,(: 1st key:)
           attribute {"LR"} { substring($style,2,1) } ,(: 2nd key:)
-          attribute {"number"} { replace($line1,$pagenos:pats?($style),"$1") }
+          attribute {"number"} { replace($line1,$pdfscrape:pats?($style),"$1") }
         )                   
 };
 
 (:~ keys are parsed pageno values are pageindices where found:)
-declare function pagenos:inverted-map($pages as element(page)*) 
+declare function pdfscrape:inverted-map($pages as element(page)*) 
 as map(*) { 
   $pages[@number]!map:entry(string(@number),string(@index))
   =>map:merge(map{"duplicates":"combine"})
 };
+(:~ %match 
+$l page labels
+:)
+declare function pdfscrape:score($l as xs:string*,$report as element(page)*)  
+{ 
+  let $s:=$report!(if(@number)then string(@number) else "")
+ let $match:= for-each-pair($l,$s,function($l,$s){if($s eq "")then 0 else if ($s eq $l)then 1 else -1})
+return round(sum($match) div count($l) *100,0) 
+};
 
 (:~ convert roman to integer, zero if invalid
 @see https://joewiz.org/2021/05/30/converting-roman-numerals-with-xquery-xslt/
 :)
-declare function pagenos:decode-roman-numeral($roman-numeral as xs:string)  
+declare function pdfscrape:decode-roman-numeral($roman-numeral as xs:string)  
 as xs:integer{ 
-    $roman-numeral => upper-case() =>  pagenos:characters() 
+    $roman-numeral => upper-case() =>  characters() 
     => for-each(map { "M": 1000, "D": 500, "C": 100, "L": 50, "X": 10, "V": 5, "I": 1 })
     => fold-right([0,0], function($number,$accumulator) { 
       if ($number lt $accumulator?2) 
@@ -67,8 +76,3 @@ as xs:integer{
     => array:head() 
 };
 
-(:~ xpath 4:)
-declare function pagenos:characters($value as xs:string?) 
-as xs:string*{
-  fn:string-to-codepoints($value) ! fn:codepoints-to-string(.)
-};