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

[mod] tidy

Browse source
This commit is contained in:
Andy Bunce 2025-01-31 16:09:25 +00:00
parent a0cfa6d937
commit 87c0a1611e
6 changed files with 40 additions and 208 deletions
Download patch file Download diff file Expand all files Collapse all files

87
src/Pdfbox3.xqm
View file

@ -10,28 +10,19 @@ module namespace pdfbox="org.expkg_zone58.Pdfbox3";
declare namespace Loader ="java:org.apache.pdfbox.Loader";
declare namespace PDFTextStripper = "java:org.apache.pdfbox.text.PDFTextStripper";
(:~ @javadoc org/apache/pdfbox/pdmodel/PDDocument.html :)
declare namespace PDDocument ="java:org.apache.pdfbox.pdmodel.PDDocument";
declare namespace PDDocumentCatalog ="java:org.apache.pdfbox.pdmodel.PDDocumentCatalog";
declare namespace PDPageLabels ="java:org.apache.pdfbox.pdmodel.common.PDPageLabels";
(:~ @javadoc org/apache/pdfbox/multipdf/PageExtractor.html :)
declare namespace PageExtractor ="java:org.apache.pdfbox.multipdf.PageExtractor";
(:~ @javadoc org/apache/pdfbox/pdmodel/PDPageTree.html :)
declare namespace PDPageTree ="java:org.apache.pdfbox.pdmodel.PDPageTree";
(:~
@javadoc org/apache/pdfbox/pdmodel/interactive/documentnavigation/outline/PDDocumentOutline.html
:)
declare namespace PDDocumentOutline ="java:org.apache.pdfbox.pdmodel.interactive.documentnavigation.outline.PDDocumentOutline";
declare namespace PDDocumentInformation ="java:org.apache.pdfbox.pdmodel.PDDocumentInformation";
(:~
@javadoc org/apache/pdfbox/pdmodel/interactive/documentnavigation/outline/PDOutlineItem.html
:)
declare namespace PDOutlineItem="java:org.apache.pdfbox.pdmodel.interactive.documentnavigation.outline.PDOutlineItem";
declare namespace PDFRenderer="java:org.apache.pdfbox.rendering.PDFRenderer";
declare namespace RandomAccessReadBufferedFile = "java:org.apache.pdfbox.io.RandomAccessReadBufferedFile";
@ -46,15 +37,19 @@ as xs:string{
(:~ open pdf, returns pdf object :)
declare function pdfbox:open($pdfpath as xs:string)
as item(){
Loader:loadPDF( RandomAccessReadBufferedFile:new($pdfpath))
try{
Loader:loadPDF( RandomAccessReadBufferedFile:new($pdfpath))
} catch *{
error(xs:QName("pdfbox:open"),"Failed to open: " || $pdfpath)
}
};
(:~ the version of the PDF specification used by $pdf e.g "1.4"
returned as string to avoid rounding issues
returned as string to avoid float rounding issues
:)
declare function pdfbox:specification($pdf as item())
as xs:string{
PDDocument:getVersion($pdf)=>string()
PDDocument:getVersion($pdf)=>xs:decimal()=>round(4)
};
(:~ save pdf $pdf to $savepath , returns $savepath :)
@ -77,10 +72,23 @@ as xs:integer{
PDDocument:getNumberOfPages($pdf)
};
(:~ render of $pdf page to image
options.format="gif,"png" etc, options.scale= 1 is 72 dpi?? :)
declare function pdfbox:page-image($pdf as item(),$pageNo as xs:integer,$options as map(*))
as xs:base64Binary{
let $options:=map:merge(($options,map{"format":"gif","scale":1}))
let $bufferedImage:=PDFRenderer:new($pdf)=>PDFRenderer:renderImage($pageNo,$options?scale)
let $bytes:=Q{java:java.io.ByteArrayOutputStream}new()
let $_:=Q{java:javax.imageio.ImageIO}write($bufferedImage ,$options?format, $bytes)
return Q{java:java.io.ByteArrayOutputStream}toByteArray($bytes)
=>convert:integers-to-base64()
};
(:~ map with document metadata :)
declare function pdfbox:information($doc as item())
declare function pdfbox:information($pdf as item())
as map(*){
let $info:=PDDocument:getDocumentInformation($doc)
let $info:=PDDocument:getDocumentInformation($pdf)
return map{
"title": PDDocumentInformation:getTitle($info),
"creator": PDDocumentInformation:getCreator($info),
@ -92,12 +100,7 @@ as map(*){
}
};
(:~ convert date :)
declare %private
function pdfbox:gregToISO($item as item())
as xs:string{
Q{java:java.util.GregorianCalendar}toZonedDateTime($item)=>string()
};
(:~ outline for $pdf as map()* :)
declare function pdfbox:outline($pdf as item())
@ -167,13 +170,13 @@ as map(*)
{
map{
"index": PDOutlineItem:findDestinationPage($bookmark,$pdf)=>pdfbox:page-index($pdf),
"title": (# db:checkstrings #) {PDOutlineItem:getTitle($bookmark)}=>translate("<22>",""),
"title": (# db:checkstrings #) {PDOutlineItem:getTitle($bookmark)}
(:=>translate("<22>",""), :),
"hasChildren": PDOutlineItem:hasChildren($bookmark)
}
};
(:~ pageIndex of $page in $pdf :)
declare function pdfbox:page-index(
$page as item()? (: as java:org.apache.pdfbox.pdmodel.PDPage :),
@ -234,35 +237,17 @@ as map(*){
)=>map:merge()
};
(:~ java:bufferedImage for $pageNo using $scale times dpi= 72
@param $pageNo (ZERO based)
@param $scale 1=72 dpi
@return Java java.awt.image.BufferedImage object
(:~ convert date :)
declare %private
function pdfbox:gregToISO($item as item())
as xs:string{
Q{java:java.util.GregorianCalendar}toZonedDateTime($item)=>string()
};
(:~ fn:do-until shim for BaseX 9+10
if fn:do-until not found use hof:until
:)
declare function pdfbox:pageBufferedImage($pdf as item(), $pageNo as xs:integer,$scale as xs:float)
as item(){
PDFRenderer:new($pdf)=>PDFRenderer:renderImage($pageNo,$scale)
};
(:~ save bufferedimage to $dest
@param $type = "gif","png" etc:)
declare function pdfbox:imageSave($bufferedImage as item(),$dest as xs:string,$type as xs:string)
as xs:boolean{
Q{java:javax.imageio.ImageIO}write($bufferedImage , $type, File:new($dest))
};
(:~ return image
@param $type = "gif","png" etc:)
declare function pdfbox:imageBinary($bufferedImage as item(),$type as xs:string)
as xs:base64Binary{
let $bytes:=Q{java:java.io.ByteArrayOutputStream}new()
let $_:=Q{java:javax.imageio.ImageIO}write($bufferedImage , $type, $bytes)
return Q{java:java.io.ByteArrayOutputStream}toByteArray($bytes)
=>convert:integers-to-base64()
};
(:~ fn:do-until shim for BaseX 9+ :)
declare function pdfbox:do-until(
declare %private function pdfbox:do-until(
$input as item()*,
$action as function(item()*, xs:integer) as item()*,
$predicate as function(item()*, xs:integer) as xs:boolean?