expkg-zone58/pdfbox
1
0
Fork 0
Code Activity Actions

[add] github workflow

Browse source
This commit is contained in:
Andy Bunce 2025-01-03 16:03:13 +00:00
parent dd7b0c1d83
commit e987fcf8ba
10 changed files with 42 additions and 6 deletions
Download patch file Download diff file Expand all files Collapse all files

10
src/lib/pdfscrape.xqm
View file

@ -35,7 +35,7 @@ as element(page){
return <page index="{ $page }">{ $found, $line1 }</page>
};
(:~ empty or attributes created by matching $style with $line1 :)
(:~ attributes created by matching $style with $line1 or empty :)
declare function pdfscrape:line-report($style as xs:string, $line1 as xs:string)
as attribute(*)*{
if(matches($line1,$pdfscrape:pats?($style)))
@ -52,10 +52,12 @@ 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*,$report as element(page)*)
declare function pdfscrape:score($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 $l)then 1 else -1})
@ -76,3 +78,7 @@ as xs:integer{
=> array:head()
};
declare function pdfscrape:characters($str as xs:string)
{
};

3
src/scratch/pdfbox.xq
View file

@ -19,6 +19,5 @@ $samples?world=>file:resolve-path($base)
let $doc:=pdfbox:open($PDF)
return pdfbox:information($doc)
pdfbox:report($PDF)