[mod] lsp-typedefs
This commit is contained in:
parent
03e30fb082
commit
2c85aa10cc
8 changed files with 656 additions and 99 deletions
16
.vscode/lsp-proxy.json
vendored
Normal file
16
.vscode/lsp-proxy.json
vendored
Normal file
|
|
@ -0,0 +1,16 @@
|
||||||
|
[
|
||||||
|
{
|
||||||
|
"languageId": "XQuery",
|
||||||
|
"command": "docker",
|
||||||
|
"fileExtensions": [
|
||||||
|
".xqm"
|
||||||
|
],
|
||||||
|
"transport": "websocket",
|
||||||
|
"websocketUrl": "ws://localhost:3000/ws/lsp",
|
||||||
|
"args": [
|
||||||
|
"compose",
|
||||||
|
"up",
|
||||||
|
"-d"
|
||||||
|
]
|
||||||
|
}
|
||||||
|
]
|
||||||
|
|
@ -6,6 +6,7 @@ as xs:string{
|
||||||
file:resolve-path($path,file:base-dir())
|
file:resolve-path($path,file:base-dir())
|
||||||
};
|
};
|
||||||
|
|
||||||
|
(:~ :)
|
||||||
declare function local:rex($ebnf,$opts as map(*))
|
declare function local:rex($ebnf,$opts as map(*))
|
||||||
{
|
{
|
||||||
let $xq:= qform:post-form(map:put($opts,"@input",$ebnf),$REX)
|
let $xq:= qform:post-form(map:put($opts,"@input",$ebnf),$REX)
|
||||||
|
|
|
||||||
523
test/sample.docs/pdfbox.xqng
Normal file
523
test/sample.docs/pdfbox.xqng
Normal file
|
|
@ -0,0 +1,523 @@
|
||||||
|
xquery version '3.1';
|
||||||
|
(:~
|
||||||
|
requires pdfbox jars on classpath, in lib/custom or xar
|
||||||
|
@note following the java source the terms outline and bookmark
|
||||||
|
refer to the same concept. Also label and (page)range are used interchangably
|
||||||
|
@note tested with pdfbox-app-3.0.5.jar
|
||||||
|
@see https://pdfbox.apache.org/download.cgi
|
||||||
|
@javadoc https://javadoc.io/static/org.apache.pdfbox/pdfbox/3.0.5/
|
||||||
|
@author Andy Bunce 2025
|
||||||
|
:)
|
||||||
|
|
||||||
|
module namespace pdfbox="org.expkg_zone58.Pdfbox3";
|
||||||
|
|
||||||
|
declare namespace Loader ="java:org.apache.pdfbox.Loader";
|
||||||
|
declare namespace PDFTextStripper = "java:org.apache.pdfbox.text.PDFTextStripper";
|
||||||
|
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";
|
||||||
|
declare namespace PDPageLabelRange="java:org.apache.pdfbox.pdmodel.common.PDPageLabelRange";
|
||||||
|
|
||||||
|
declare namespace PageExtractor ="java:org.apache.pdfbox.multipdf.PageExtractor";
|
||||||
|
declare namespace PDPage ="java:org.apache.pdfbox.pdmodel.PDPage";
|
||||||
|
declare namespace PDPageTree ="java:org.apache.pdfbox.pdmodel.PDPageTree";
|
||||||
|
declare namespace PDDocumentOutline ="java:org.apache.pdfbox.pdmodel.interactive.documentnavigation.outline.PDDocumentOutline";
|
||||||
|
declare namespace PDDocumentInformation ="java:org.apache.pdfbox.pdmodel.PDDocumentInformation";
|
||||||
|
declare namespace PDOutlineItem="java:org.apache.pdfbox.pdmodel.interactive.documentnavigation.outline.PDOutlineItem";
|
||||||
|
declare namespace PDFRenderer="java:org.apache.pdfbox.rendering.PDFRenderer";
|
||||||
|
declare namespace PDMetadata="java:org.apache.pdfbox.pdmodel.common.PDMetadata";
|
||||||
|
declare namespace COSInputStream="java:org.apache.pdfbox.cos.COSInputStream";
|
||||||
|
|
||||||
|
|
||||||
|
declare namespace rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#";
|
||||||
|
|
||||||
|
|
||||||
|
declare namespace RandomAccessReadBuffer="java:org.apache.pdfbox.io.RandomAccessReadBuffer";
|
||||||
|
declare namespace RandomAccessReadBufferedFile = "java:org.apache.pdfbox.io.RandomAccessReadBufferedFile";
|
||||||
|
declare namespace PDRectangle="org.apache.pdfbox.pdmodel.common.PDRectangle";
|
||||||
|
|
||||||
|
declare namespace File ="java:java.io.File";
|
||||||
|
|
||||||
|
|
||||||
|
|
||||||
|
(:~ "With-document" pattern: open pdf,apply $fn function, close pdf
|
||||||
|
creates a local pdfobject and ensures it is closed after use
|
||||||
|
e.g pdfbox:with-pdf("path...",pdfbox:page-text(?,5))
|
||||||
|
:)
|
||||||
|
declare function pdfbox:with-pdf($src as xs:string,
|
||||||
|
$fn as function(item())as item()*)
|
||||||
|
as item()*{
|
||||||
|
let $pdf:=pdfbox:open($src)
|
||||||
|
return try{
|
||||||
|
$fn($pdf),pdfbox:close($pdf)
|
||||||
|
} catch *{
|
||||||
|
pdfbox:close($pdf),fn:error($err:code,$src || " " || $err:description)
|
||||||
|
}
|
||||||
|
|
||||||
|
};
|
||||||
|
|
||||||
|
|
||||||
|
(:~ open pdf using fetch:binary, returns pdf object :)
|
||||||
|
declare function pdfbox:open($pdfsrc as item())
|
||||||
|
as item(){
|
||||||
|
pdfbox:open($pdfsrc, map{})
|
||||||
|
};
|
||||||
|
|
||||||
|
(:~ open pdf from file/url/binary, opts may have password , returns pdf object
|
||||||
|
@param $pdfsrc a fetchable url or filepath, or xs:base64Binary item
|
||||||
|
@param $opts options options include map {"password":}
|
||||||
|
@note fetch:binary for https will use a lot of memory here
|
||||||
|
:)
|
||||||
|
declare function pdfbox:open($pdfsrc as item(), $opts as map(*))
|
||||||
|
as item(){
|
||||||
|
try{
|
||||||
|
|
||||||
|
if($pdfsrc instance of xs:base64Binary)
|
||||||
|
then Loader:loadPDF( $pdfsrc,string($opts?password))
|
||||||
|
else if(starts-with($pdfsrc,"http"))
|
||||||
|
then Loader:loadPDF( fetch:binary($pdfsrc),string($opts?password))
|
||||||
|
else Loader:loadPDF(RandomAccessReadBufferedFile:new($pdfsrc),string($opts?password))
|
||||||
|
|
||||||
|
} catch *{
|
||||||
|
let $loc:=if($pdfsrc instance of xs:base64Binary)
|
||||||
|
then "xs:base64Binary"
|
||||||
|
else $pdfsrc
|
||||||
|
return error(xs:QName("pdfbox:open"),"Failed PDF load " || $loc || " " || $err:description)
|
||||||
|
}
|
||||||
|
};
|
||||||
|
|
||||||
|
(:~ The version of the PDF specification used by $pdf e.g "1.4"
|
||||||
|
returned as string to avoid float rounding issues
|
||||||
|
:)
|
||||||
|
declare function pdfbox:specification($pdf as item())
|
||||||
|
as xs:string{
|
||||||
|
PDDocument:getVersion($pdf)=>xs:decimal()=>round(4)=>string()
|
||||||
|
};
|
||||||
|
|
||||||
|
(:~ Save pdf $pdf to filesystem at $savepath , returns $savepath :)
|
||||||
|
declare function pdfbox:pdf-save($pdf as item(),$savepath as xs:string)
|
||||||
|
as xs:string{
|
||||||
|
PDDocument:save($pdf, File:new($savepath)),$savepath
|
||||||
|
};
|
||||||
|
|
||||||
|
(:~ Create binary representation of $pdf object as xs:base64Binary :)
|
||||||
|
declare function pdfbox:binary($pdf as item())
|
||||||
|
as xs:base64Binary{
|
||||||
|
let $bytes:=Q{java:java.io.ByteArrayOutputStream}new()
|
||||||
|
let $_:=PDDocument:save($pdf, $bytes)
|
||||||
|
return Q{java:java.io.ByteArrayOutputStream}toByteArray($bytes)
|
||||||
|
=>convert:integers-to-base64()
|
||||||
|
};
|
||||||
|
|
||||||
|
(:~ Release any resources related to $pdf:)
|
||||||
|
declare function pdfbox:close($pdf as item())
|
||||||
|
as empty-sequence(){
|
||||||
|
(# db:wrapjava void #) {
|
||||||
|
PDDocument:close($pdf)
|
||||||
|
}
|
||||||
|
};
|
||||||
|
|
||||||
|
(:~ Number of pages in PDF:)
|
||||||
|
declare function pdfbox:number-of-pages($pdf as item())
|
||||||
|
as xs:integer{
|
||||||
|
PDDocument:getNumberOfPages($pdf)
|
||||||
|
};
|
||||||
|
|
||||||
|
(:~ Pdf page as image (zero is cover)
|
||||||
|
options.format="bmp jpg png gif" etc, options.scale= 1 is 72 dpi?? :)
|
||||||
|
declare function pdfbox:page-render($pdf as item(),$pageNo as xs:integer,$options as map(*))
|
||||||
|
as xs:base64Binary{
|
||||||
|
let $options := map:merge(($options,map{"format":"jpg","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()
|
||||||
|
|
||||||
|
};
|
||||||
|
|
||||||
|
|
||||||
|
(:~ Defines a map from property names to evaluation method.
|
||||||
|
Keys are property names,
|
||||||
|
values are sequences of functions to get property value starting from a $pdf object.
|
||||||
|
:)
|
||||||
|
declare %private variable $pdfbox:property-map:=map{
|
||||||
|
"#pages": pdfbox:number-of-pages#1,
|
||||||
|
|
||||||
|
"#bookmarks": pdfbox:number-of-bookmarks#1,
|
||||||
|
|
||||||
|
"#labels": pdfbox:number-of-labels#1,
|
||||||
|
|
||||||
|
"specification":pdfbox:specification#1,
|
||||||
|
|
||||||
|
"title": (PDDocument:getDocumentInformation#1,
|
||||||
|
PDDocumentInformation:getTitle#1) ,
|
||||||
|
|
||||||
|
"author": (PDDocument:getDocumentInformation#1,
|
||||||
|
PDDocumentInformation:getAuthor#1 ),
|
||||||
|
|
||||||
|
"creator": (PDDocument:getDocumentInformation#1,
|
||||||
|
PDDocumentInformation:getCreator#1),
|
||||||
|
|
||||||
|
"producer": (PDDocument:getDocumentInformation#1,
|
||||||
|
PDDocumentInformation:getProducer#1),
|
||||||
|
|
||||||
|
"subject": (PDDocument:getDocumentInformation#1,
|
||||||
|
PDDocumentInformation:getSubject#1),
|
||||||
|
|
||||||
|
"keywords": (PDDocument:getDocumentInformation#1,
|
||||||
|
PDDocumentInformation:getKeywords#1),
|
||||||
|
|
||||||
|
"creationDate": (PDDocument:getDocumentInformation#1,
|
||||||
|
PDDocumentInformation:getCreationDate#1,
|
||||||
|
pdfbox:gregToISO#1),
|
||||||
|
|
||||||
|
"modificationDate": (PDDocument:getDocumentInformation#1,
|
||||||
|
PDDocumentInformation:getModificationDate#1,
|
||||||
|
pdfbox:gregToISO#1),
|
||||||
|
|
||||||
|
"labels": pdfbox:labels-as-string#1
|
||||||
|
};
|
||||||
|
|
||||||
|
(:~ Defined property names, sorted :)
|
||||||
|
declare function pdfbox:property-names()
|
||||||
|
as xs:string*{
|
||||||
|
$pdfbox:property-map=>map:keys()=>sort()
|
||||||
|
};
|
||||||
|
|
||||||
|
(:~ Return the value of $property for $pdf :)
|
||||||
|
declare function pdfbox:property($pdf as item(),$property as xs:string)
|
||||||
|
as item()*{
|
||||||
|
let $fns:= $pdfbox:property-map($property)
|
||||||
|
return if(exists($fns))
|
||||||
|
then fold-left($fns,
|
||||||
|
$pdf,
|
||||||
|
function($result,$this as function(*)){$result!$this(.)})
|
||||||
|
else error(xs:QName('pdfbox:property'),concat("Property '",$property,"' not defined."))
|
||||||
|
};
|
||||||
|
|
||||||
|
(:~ summary CSV style info for all properties for $pdfpaths
|
||||||
|
:)
|
||||||
|
declare function pdfbox:report($pdfpaths as xs:string*)
|
||||||
|
as map(*){
|
||||||
|
pdfbox:report($pdfpaths,pdfbox:property-names())
|
||||||
|
};
|
||||||
|
|
||||||
|
(:~ summary CSV style info for named $properties for PDFs in $pdfpaths
|
||||||
|
@see https://docs.basex.org/main/CSV_Functions#xquery
|
||||||
|
:)
|
||||||
|
declare function pdfbox:report($pdfpaths as item()*, $properties as xs:string*)
|
||||||
|
as map(*){
|
||||||
|
map{"names": array{"path",$properties},
|
||||||
|
|
||||||
|
"records": for $path in $pdfpaths
|
||||||
|
let $name:=if($path instance of xs:base64Binary) then "binary" else $path
|
||||||
|
return try{
|
||||||
|
let $pdf:=pdfbox:open($path)
|
||||||
|
return (fold-left($properties,
|
||||||
|
array{$name},
|
||||||
|
function($result as array(*),$prop as xs:string){
|
||||||
|
array:append($result, string(pdfbox:property($pdf, $prop)))}
|
||||||
|
), pdfbox:close($pdf)
|
||||||
|
)
|
||||||
|
} catch *{
|
||||||
|
fold-left($properties,
|
||||||
|
array{$name},
|
||||||
|
function($result as array(*),$prop as xs:string){
|
||||||
|
array:append($result, "#ERROR")}
|
||||||
|
)
|
||||||
|
}
|
||||||
|
|
||||||
|
}
|
||||||
|
};
|
||||||
|
|
||||||
|
(:~ Convenience function to save report() data to file :)
|
||||||
|
declare function pdfbox:report-save($data as map(*),$dest as xs:string)
|
||||||
|
as empty-sequence(){
|
||||||
|
let $opts := map { "format":"xquery", "header":"yes", "separator" : "," }
|
||||||
|
return file:write-text($dest,csv:serialize($data,$opts))
|
||||||
|
};
|
||||||
|
|
||||||
|
(:~ The number of outline items defined in $pdf :)
|
||||||
|
declare function pdfbox:number-of-bookmarks($pdf as item())
|
||||||
|
as xs:integer{
|
||||||
|
let $xml:=pdfbox:outline-xml($pdf)
|
||||||
|
return count($xml//bookmark)
|
||||||
|
};
|
||||||
|
|
||||||
|
(:~ XMP metadata as "RDF" document
|
||||||
|
@note usually rdf:RDF root, but sometimes x:xmpmeta
|
||||||
|
:)
|
||||||
|
declare function pdfbox:metadata($pdf as item())
|
||||||
|
as document-node(element(*))?
|
||||||
|
{
|
||||||
|
let $m:=PDDocument:getDocumentCatalog($pdf)
|
||||||
|
=>PDDocumentCatalog:getMetadata()
|
||||||
|
return if(exists($m))
|
||||||
|
then
|
||||||
|
let $is:=PDMetadata:exportXMPMetadata($m)
|
||||||
|
return pdfbox:do-until(
|
||||||
|
map{"n":0,"data":""},
|
||||||
|
|
||||||
|
function($input,$pos ) { pdfbox:read-stream($is,$input?data)},
|
||||||
|
|
||||||
|
function($output,$pos) { $output?n eq -1 }
|
||||||
|
)?data=>parse-xml()
|
||||||
|
else ()
|
||||||
|
};
|
||||||
|
|
||||||
|
(:~ read next block from XMP stream :)
|
||||||
|
declare %private function pdfbox:read-stream($is,$read as xs:string)
|
||||||
|
as map(*){
|
||||||
|
let $blen:=4096
|
||||||
|
let $buff:=Q{java:java.util.Arrays}copyOf(array{xs:byte(0)},$blen)
|
||||||
|
let $n:= COSInputStream:read($is,$buff,xs:int(0),xs:int($blen))
|
||||||
|
let $data:=convert:integers-to-base64(subsequence($buff,1,$n))=>convert:binary-to-string()
|
||||||
|
return map{"n":$n, "data": $read || $data}
|
||||||
|
};
|
||||||
|
|
||||||
|
(:~ Return outline for $pdf as map()* :)
|
||||||
|
declare function pdfbox:outline($pdf as item())
|
||||||
|
as map(*)*{
|
||||||
|
(# db:wrapjava some #) {
|
||||||
|
let $outline:=
|
||||||
|
PDDocument:getDocumentCatalog($pdf)
|
||||||
|
=>PDDocumentCatalog:getDocumentOutline()
|
||||||
|
|
||||||
|
return if(exists($outline))
|
||||||
|
then pdfbox:outline($pdf,PDOutlineItem:getFirstChild($outline))
|
||||||
|
else ()
|
||||||
|
}
|
||||||
|
};
|
||||||
|
|
||||||
|
(:~ return bookmark info for children of $outlineItem as seq of maps :)
|
||||||
|
declare %private function pdfbox:outline($pdf as item(),$outlineItem as item()?)
|
||||||
|
as map(*)*{
|
||||||
|
let $find as map(*):=pdfbox:outline_($pdf ,$outlineItem)
|
||||||
|
return map:get($find,"list")
|
||||||
|
};
|
||||||
|
|
||||||
|
(:~ outline helper. BaseX bug 10.7? error if inlined in outline :)
|
||||||
|
declare %private function pdfbox:outline_($pdf as item(),$outlineItem as item()?)
|
||||||
|
as map(*){
|
||||||
|
pdfbox:do-until(
|
||||||
|
|
||||||
|
map{"list":(),"this":$outlineItem},
|
||||||
|
|
||||||
|
function($input,$pos ) {
|
||||||
|
let $bk:= pdfbox:bookmark($input?this,$pdf)
|
||||||
|
let $bk:= if($bk?hasChildren)
|
||||||
|
then let $kids:=pdfbox:outline($pdf,PDOutlineItem:getFirstChild($input?this))
|
||||||
|
return map:merge(($bk,map:entry("children",$kids)))
|
||||||
|
else $bk
|
||||||
|
return map{
|
||||||
|
"list": ($input?list, $bk),
|
||||||
|
"this": PDOutlineItem:getNextSibling($input?this)}
|
||||||
|
},
|
||||||
|
|
||||||
|
function($output,$pos) { empty($output?this) }
|
||||||
|
)
|
||||||
|
};
|
||||||
|
|
||||||
|
(:~ PDF outline in xml format :)
|
||||||
|
declare function pdfbox:outline-xml($pdf as item())
|
||||||
|
as element(outline)?{
|
||||||
|
let $outline:=pdfbox:outline($pdf)
|
||||||
|
return if(exists($outline))
|
||||||
|
then <outline>{$outline!pdfbox:bookmark-xml(.)}</outline>
|
||||||
|
else ()
|
||||||
|
};
|
||||||
|
|
||||||
|
(:~ Convert outline map to XML :)
|
||||||
|
declare %private function pdfbox:bookmark-xml($outline as map(*)*)
|
||||||
|
as element(bookmark)*
|
||||||
|
{
|
||||||
|
$outline!
|
||||||
|
<bookmark title="{?title}" index="{?index}">
|
||||||
|
{?children!pdfbox:bookmark-xml(.)}
|
||||||
|
</bookmark>
|
||||||
|
};
|
||||||
|
|
||||||
|
(:~ Return bookmark info for $bookmark
|
||||||
|
@return map{index:..,title:..,hasChildren:..}
|
||||||
|
:)
|
||||||
|
declare %private function pdfbox:bookmark($bookmark as item(),$pdf as item())
|
||||||
|
as map(*)
|
||||||
|
{
|
||||||
|
map{
|
||||||
|
"index": PDOutlineItem:findDestinationPage($bookmark,$pdf)=>pdfbox:find-page($pdf),
|
||||||
|
"title": (# db:checkstrings #) {PDOutlineItem:getTitle($bookmark)}
|
||||||
|
(:=>translate("<22>",""), :),
|
||||||
|
"hasChildren": PDOutlineItem:hasChildren($bookmark)
|
||||||
|
}
|
||||||
|
};
|
||||||
|
|
||||||
|
|
||||||
|
(:~ pageIndex of $page in $pdf :)
|
||||||
|
declare function pdfbox:find-page(
|
||||||
|
$page as item()? (: as java:org.apache.pdfbox.pdmodel.PDPage :),
|
||||||
|
$pdf as item())
|
||||||
|
as item()?
|
||||||
|
{
|
||||||
|
if(exists($page))
|
||||||
|
then PDDocument:getDocumentCatalog($pdf)
|
||||||
|
=>PDDocumentCatalog:getPages()
|
||||||
|
=>PDPageTree:indexOf($page)
|
||||||
|
else ()
|
||||||
|
};
|
||||||
|
|
||||||
|
(:~ Return new PDF doc with pages from $start to $end as xs:base64Binary, (1 based)
|
||||||
|
@param $start first page to include
|
||||||
|
@param $end last page to include
|
||||||
|
:)
|
||||||
|
declare function pdfbox:extract-range($pdf as item(),
|
||||||
|
$start as xs:integer,$end as xs:integer)
|
||||||
|
as xs:base64Binary
|
||||||
|
{
|
||||||
|
let $a:=PageExtractor:new($pdf, $start, $end) =>PageExtractor:extract()
|
||||||
|
return (pdfbox:binary($a),pdfbox:close($a))
|
||||||
|
};
|
||||||
|
|
||||||
|
(:~ The number of labels defined in PDF :)
|
||||||
|
declare function pdfbox:number-of-labels($pdf as item())
|
||||||
|
as xs:integer
|
||||||
|
{
|
||||||
|
let $labels:=PDDocument:getDocumentCatalog($pdf)
|
||||||
|
=>PDDocumentCatalog:getPageLabels()
|
||||||
|
return if(exists($labels))
|
||||||
|
then PDPageLabels:getPageRangeCount($labels)
|
||||||
|
else 0
|
||||||
|
};
|
||||||
|
|
||||||
|
(:~ pageLabel for every page from derived from page-ranges
|
||||||
|
The returned sequence will contain at MOST as much entries as the document has pages.
|
||||||
|
@see https://www.w3.org/TR/WCAG20-TECHS/PDF17.html#PDF17-examples
|
||||||
|
@see https://codereview.stackexchange.com/questions/286078/java-code-showing-page-labels-from-pdf-files
|
||||||
|
:)
|
||||||
|
declare function pdfbox:labels-by-page($pdf as item())
|
||||||
|
as xs:string*
|
||||||
|
{
|
||||||
|
PDDocument:getDocumentCatalog($pdf)
|
||||||
|
=>PDDocumentCatalog:getPageLabels()
|
||||||
|
=>PDPageLabels:getLabelsByPageIndices()
|
||||||
|
};
|
||||||
|
|
||||||
|
(:~ sequence of label ranges defined in PDF as formatted strings
|
||||||
|
@return a custom representation of the labels e.g "0-*Cover,1r,11D"
|
||||||
|
:)
|
||||||
|
declare function pdfbox:labels-as-string($pdf as item())
|
||||||
|
as xs:string{
|
||||||
|
let $pagelabels:=PDDocument:getDocumentCatalog($pdf)
|
||||||
|
=>PDDocumentCatalog:getPageLabels()
|
||||||
|
return $pagelabels
|
||||||
|
!(0 to pdfbox:number-of-pages($pdf)-1)
|
||||||
|
!pdfbox:label-as-string($pagelabels,.)=>string-join(" ")
|
||||||
|
|
||||||
|
};
|
||||||
|
|
||||||
|
(:~ get pagelabels exist :)
|
||||||
|
declare function pdfbox:page-labels($pdf)
|
||||||
|
{
|
||||||
|
PDDocument:getDocumentCatalog($pdf)
|
||||||
|
=>PDDocumentCatalog:getPageLabels()
|
||||||
|
};
|
||||||
|
|
||||||
|
(:~ label for $page formated as string, empty if none :)
|
||||||
|
declare function pdfbox:label-as-string($pagelabels,$page as xs:integer)
|
||||||
|
as xs:string?{
|
||||||
|
let $label:=PDPageLabels:getPageLabelRange($pagelabels,$page)
|
||||||
|
return if(empty($label))
|
||||||
|
then ()
|
||||||
|
else
|
||||||
|
let $start:= PDPageLabelRange:getStart($label)
|
||||||
|
let $style := PDPageLabelRange:getStyle($label)
|
||||||
|
let $prefix:= PDPageLabelRange:getPrefix($label)
|
||||||
|
return string-join(($page,
|
||||||
|
if(empty($style)) then "-" else $style,
|
||||||
|
if(($start eq 1)) then "" else $start,
|
||||||
|
if(exists($prefix)) then '*' || $prefix else () (:TODO double " :)
|
||||||
|
))
|
||||||
|
};
|
||||||
|
|
||||||
|
(:~ sequence of maps for each label/page range defined in $pdf:)
|
||||||
|
declare function pdfbox:labels-as-map($pdf as item())
|
||||||
|
as map(*)*{
|
||||||
|
let $pagelabels:=PDDocument:getDocumentCatalog($pdf)
|
||||||
|
=>PDDocumentCatalog:getPageLabels()
|
||||||
|
return $pagelabels
|
||||||
|
!(0 to pdfbox:number-of-pages($pdf)-1)
|
||||||
|
!pdfbox:label-as-map($pagelabels,.)
|
||||||
|
};
|
||||||
|
|
||||||
|
(:~ label/page-range for $page as map :)
|
||||||
|
declare function pdfbox:label-as-map($pagelabels,$page as xs:integer)
|
||||||
|
as map(*)
|
||||||
|
{
|
||||||
|
let $label:=PDPageLabels:getPageLabelRange($pagelabels,$page)
|
||||||
|
return if(empty($label))
|
||||||
|
then ()
|
||||||
|
else map{
|
||||||
|
"index": $page,
|
||||||
|
"prefix": PDPageLabelRange:getPrefix($label),
|
||||||
|
"start": PDPageLabelRange:getStart($label),
|
||||||
|
"style": PDPageLabelRange:getStyle($label)
|
||||||
|
}
|
||||||
|
};
|
||||||
|
|
||||||
|
|
||||||
|
|
||||||
|
(:~ return text on $pageNo :)
|
||||||
|
declare function pdfbox:page-text($pdf as item(), $pageNo as xs:integer)
|
||||||
|
as xs:string{
|
||||||
|
let $tStripper := (# db:wrapjava instance #) {
|
||||||
|
PDFTextStripper:new()
|
||||||
|
=> PDFTextStripper:setStartPage($pageNo)
|
||||||
|
=> PDFTextStripper:setEndPage($pageNo)
|
||||||
|
}
|
||||||
|
return (# db:checkstrings #) {PDFTextStripper:getText($tStripper,$pdf)}
|
||||||
|
};
|
||||||
|
|
||||||
|
(:~ Return size of $pageNo (zero based)
|
||||||
|
@return e.g. [0.0,0.0,168.0,239.52]
|
||||||
|
:)
|
||||||
|
declare function pdfbox:page-media-box($pdf as item(), $pageNo as xs:integer)
|
||||||
|
as xs:string{
|
||||||
|
PDDocument:getPage($pdf, $pageNo)
|
||||||
|
=>PDPage:getMediaBox()
|
||||||
|
=>PDRectangle:toString()
|
||||||
|
};
|
||||||
|
|
||||||
|
(:~ Version of Apache Pdfbox in use e.g. "3.0.4" :)
|
||||||
|
declare function pdfbox:version()
|
||||||
|
as xs:string{
|
||||||
|
Q{java:org.apache.pdfbox.util.Version}getVersion()
|
||||||
|
};
|
||||||
|
|
||||||
|
(:~ Convert date :)
|
||||||
|
declare %private
|
||||||
|
function pdfbox:gregToISO($item as item()?)
|
||||||
|
as xs:string?{
|
||||||
|
if(exists($item))
|
||||||
|
then Q{java:java.util.GregorianCalendar}toZonedDateTime($item)=>string()
|
||||||
|
else ()
|
||||||
|
};
|
||||||
|
|
||||||
|
(:~ fn:do-until shim for BaseX 9+10
|
||||||
|
if fn:do-until not found use hof:until, note: $pos always zero
|
||||||
|
:)
|
||||||
|
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?
|
||||||
|
) as item()*
|
||||||
|
{
|
||||||
|
let $fn:=function-lookup(QName('http://www.w3.org/2005/xpath-functions','do-until'), 3)
|
||||||
|
return if(exists($fn))
|
||||||
|
then $fn($input,$action,$predicate)
|
||||||
|
else let $hof:=function-lookup(QName('http://basex.org/modules/hof','until'), 3)
|
||||||
|
return if(exists($hof))
|
||||||
|
then $hof($predicate(?,0),$action(?,0),$input)
|
||||||
|
else error(xs:QName('pdfbox:do-until'),"No implementation do-until found")
|
||||||
|
|
||||||
|
};
|
||||||
|
|
@ -3,62 +3,5 @@ xquery version '4.0';
|
||||||
module namespace syms="lsp/symbols";
|
module namespace syms="lsp/symbols";
|
||||||
import module namespace pos="lsp/position" at "position.xqm";
|
import module namespace pos="lsp/position" at "position.xqm";
|
||||||
|
|
||||||
declare type syms:SymbolKind as xs:integer;
|
|
||||||
declare type syms:SymbolTag as xs:string;
|
|
||||||
|
|
||||||
declare variable $syms:SymbolKindMap :={
|
|
||||||
'File': 1 ,
|
|
||||||
'Module': 2 ,
|
|
||||||
'Namespace': 3 ,
|
|
||||||
'Package': 4 ,
|
|
||||||
'Class': 5 ,
|
|
||||||
'Method': 6 ,
|
|
||||||
'Property': 7 ,
|
|
||||||
'Field': 8 ,
|
|
||||||
'Constructor': 9 ,
|
|
||||||
'Enum': 10 ,
|
|
||||||
'Interface': 11 ,
|
|
||||||
'Function': 12 ,
|
|
||||||
'Variable': 13 ,
|
|
||||||
'Constant': 14 ,
|
|
||||||
'String': 15 ,
|
|
||||||
'Number': 16 ,
|
|
||||||
'Boolean': 17 ,
|
|
||||||
'Array': 18 ,
|
|
||||||
'Object': 19 ,
|
|
||||||
'Key': 20 ,
|
|
||||||
'Null': 21 ,
|
|
||||||
'EnumMember': 22 ,
|
|
||||||
'Struct': 23 ,
|
|
||||||
'Event': 24 ,
|
|
||||||
'Operator': 25 ,
|
|
||||||
'TypeParameter': 26
|
|
||||||
};
|
|
||||||
|
|
||||||
declare record syms:DocumentSymbol(
|
|
||||||
|
|
||||||
(: The name of this symbol. Will be displayed in the user interface and
|
|
||||||
therefore must not be an empty string or a string only consisting of white spaces. :)
|
|
||||||
name as xs:string,
|
|
||||||
|
|
||||||
(: The kind of this symbol. :)
|
|
||||||
kind as syms:SymbolKind,
|
|
||||||
|
|
||||||
(: The range enclosing this symbol not including leading/trailing whitespace
|
|
||||||
but everything else like comments. This information is typically used to
|
|
||||||
determine if the clients cursor is inside the symbol to reveal it in the UI. :)
|
|
||||||
range as pos:Range,
|
|
||||||
|
|
||||||
(:The range that should be selected and revealed when this symbol is being
|
|
||||||
picked, e.g. the name of a function. Must be contained by the `range`. :)
|
|
||||||
selectionRange as pos:Range,
|
|
||||||
|
|
||||||
(: More detail for this symbol, e.g the signature of a function.:)
|
|
||||||
detail? as xs:string,
|
|
||||||
|
|
||||||
(: Tags for this document symbol. @since 3.16.0 :)
|
|
||||||
tags? as syms:SymbolTag*,
|
|
||||||
|
|
||||||
(: Children of this symbol, e.g. properties of a class. :)
|
|
||||||
children? as syms:DocumentSymbol*
|
|
||||||
);
|
|
||||||
|
|
@ -2,7 +2,8 @@
|
||||||
@author Andy Bunce
|
@author Andy Bunce
|
||||||
:)
|
:)
|
||||||
module namespace hnd="lsp/handlers";
|
module namespace hnd="lsp/handlers";
|
||||||
import module namespace pos="lsp/position" at "position.xqm";
|
import module namespace lspt = 'lsp-typedefs' at "lsp-typedefs.xqm";
|
||||||
|
|
||||||
|
|
||||||
declare record hnd:hand(
|
declare record hnd:hand(
|
||||||
result as item()*,
|
result as item()*,
|
||||||
|
|
@ -33,15 +34,9 @@ declare function hnd:diags($parse as element(),$diags:=())
|
||||||
})
|
})
|
||||||
};
|
};
|
||||||
|
|
||||||
declare record hnd:symbol (
|
|
||||||
name as xs:string,
|
|
||||||
type as xs:string,
|
|
||||||
range-name? as pos:Range,
|
|
||||||
range-full? as pos:Range,
|
|
||||||
children? as array(hnd:symbol)
|
|
||||||
);
|
|
||||||
|
|
||||||
declare function hnd:symbols($parse as element(),$syms as hnd:symbol* :=() )
|
|
||||||
|
declare function hnd:symbols($parse as element(),$syms as lspt:symbol* :=() )
|
||||||
{
|
{
|
||||||
'todo'
|
'todo'
|
||||||
};
|
};
|
||||||
|
|
@ -49,7 +44,7 @@ declare function hnd:symbols($parse as element(),$syms as hnd:symbol* :=() )
|
||||||
declare function hnd:anotated-declaration($parse as element(),$syms)
|
declare function hnd:anotated-declaration($parse as element(),$syms)
|
||||||
as hnd:hand
|
as hnd:hand
|
||||||
{
|
{
|
||||||
let $sym:=hnd:symbol(
|
let $sym:=lspt:symbol(
|
||||||
$parse/token,
|
$parse/token,
|
||||||
"AA"
|
"AA"
|
||||||
)
|
)
|
||||||
|
|
|
||||||
|
|
@ -1,5 +1,6 @@
|
||||||
module namespace lsp-diags = 'lsp-diags';
|
module namespace lsp-diags = 'lsp-diags';
|
||||||
|
|
||||||
|
import module namespace lspt = 'lsp-typedefs' at "lsp-typedefs.xqm";
|
||||||
import module namespace pos="lsp/position" at "position.xqm";
|
import module namespace pos="lsp/position" at "position.xqm";
|
||||||
|
|
||||||
declare type lsp-diags:ParseResult as element(Module|ERROR);
|
declare type lsp-diags:ParseResult as element(Module|ERROR);
|
||||||
|
|
@ -14,7 +15,7 @@ renderMessage?: fn(view: EditorView) → Node An optional custom rendering fu
|
||||||
actions?: readonly Action[] An optional array of actions that can be taken on this diagnostic.
|
actions?: readonly Action[] An optional array of actions that can be taken on this diagnostic.
|
||||||
:)
|
:)
|
||||||
declare record lsp-diags:nostic(
|
declare record lsp-diags:nostic(
|
||||||
range as pos:Range,
|
range as lspt:Range,
|
||||||
severity as xs:integer, (: enum('error', 'hint', 'info', 'warning') :)
|
severity as xs:integer, (: enum('error', 'hint', 'info', 'warning') :)
|
||||||
message as xs:string,
|
message as xs:string,
|
||||||
code? as xs:string,
|
code? as xs:string,
|
||||||
|
|
@ -59,7 +60,7 @@ as map(*)*{
|
||||||
let $e:= number($xml/@e)-1
|
let $e:= number($xml/@e)-1
|
||||||
return (
|
return (
|
||||||
(: mark error :)
|
(: mark error :)
|
||||||
lsp-diags:nostic(pos:Range(pos:toPosition($text, $b),
|
lsp-diags:nostic(lspt:Range(pos:toPosition($text, $b),
|
||||||
pos:toPosition($text, $e)),
|
pos:toPosition($text, $e)),
|
||||||
1,
|
1,
|
||||||
$dmesg,'XPST0003'),
|
$dmesg,'XPST0003'),
|
||||||
|
|
@ -67,7 +68,7 @@ as map(*)*{
|
||||||
(:mark after error:)
|
(:mark after error:)
|
||||||
if($e ge string-length($text))
|
if($e ge string-length($text))
|
||||||
then ()
|
then ()
|
||||||
else lsp-diags:nostic(pos:Range(pos:toPosition($text, $e +1 ),
|
else lsp-diags:nostic(lspt:Range(pos:toPosition($text, $e +1 ),
|
||||||
pos:toPosition($text, $last)),
|
pos:toPosition($text, $last)),
|
||||||
2,
|
2,
|
||||||
"Unparsed due to previous parser error.",
|
"Unparsed due to previous parser error.",
|
||||||
|
|
|
||||||
97
webapp/lsp/lsp-typedefs.xqm
Normal file
97
webapp/lsp/lsp-typedefs.xqm
Normal file
|
|
@ -0,0 +1,97 @@
|
||||||
|
(:~ LSPserver type definitions
|
||||||
|
@see https://microsoft.github.io/language-server-protocol/specifications/lsp/3.17/specification/
|
||||||
|
@author Andy Bunce
|
||||||
|
:)
|
||||||
|
module namespace lspt = 'lsp-typedefs';
|
||||||
|
|
||||||
|
(:~ json numbers :)
|
||||||
|
declare type lspt:num as (xs:integer|xs:double);
|
||||||
|
|
||||||
|
(:~
|
||||||
|
@param line Line position in a document (zero-based).
|
||||||
|
@param character Character offset on a line in a document (zero-based).
|
||||||
|
:)
|
||||||
|
declare record lspt:Position(
|
||||||
|
line as lspt:num,
|
||||||
|
character as lspt:num
|
||||||
|
);
|
||||||
|
|
||||||
|
(:~
|
||||||
|
@param line Line position in a document (zero-based).
|
||||||
|
@param character Character offset on a line in a document (zero-based).
|
||||||
|
:)
|
||||||
|
declare record lspt:Range(
|
||||||
|
start as lspt:Position,
|
||||||
|
end as lspt:Position
|
||||||
|
);
|
||||||
|
|
||||||
|
declare type lspt:SymbolKind as xs:integer;
|
||||||
|
declare type lspt:SymbolTag as xs:string;
|
||||||
|
|
||||||
|
declare record lspt:symbol (
|
||||||
|
name as xs:string,
|
||||||
|
type as xs:string,
|
||||||
|
range-name? as lspt:Range,
|
||||||
|
range-full? as lspt:Range,
|
||||||
|
children? as array(lspt:symbol)
|
||||||
|
);
|
||||||
|
|
||||||
|
declare record lspt:DocumentSymbol(
|
||||||
|
|
||||||
|
(: The name of this symbol. Will be displayed in the user interface and
|
||||||
|
therefore must not be an empty string or a string only consisting of white spaces. :)
|
||||||
|
name as xs:string,
|
||||||
|
|
||||||
|
(: The kind of this symbol. :)
|
||||||
|
kind as lspt:SymbolKind,
|
||||||
|
|
||||||
|
(: The range enclosing this symbol not including leading/trailing whitespace
|
||||||
|
but everything else like comments. This information is typically used to
|
||||||
|
determine if the clients cursor is inside the symbol to reveal it in the UI. :)
|
||||||
|
range as lspt:Range,
|
||||||
|
|
||||||
|
(:The range that should be selected and revealed when this symbol is being
|
||||||
|
picked, e.g. the name of a function. Must be contained by the `range`. :)
|
||||||
|
selectionRange as lspt:Range,
|
||||||
|
|
||||||
|
(: More detail for this symbol, e.g the signature of a function.:)
|
||||||
|
detail? as xs:string,
|
||||||
|
|
||||||
|
(: Tags for this document symbol. @since 3.16.0 :)
|
||||||
|
tags? as lspt:SymbolTag*,
|
||||||
|
|
||||||
|
(: Children of this symbol, e.g. properties of a class. :)
|
||||||
|
children? as lspt:DocumentSymbol*
|
||||||
|
);
|
||||||
|
|
||||||
|
|
||||||
|
declare variable $lspt:SymbolKindMap :={
|
||||||
|
'File': 1 ,
|
||||||
|
'Module': 2 ,
|
||||||
|
'Namespace': 3 ,
|
||||||
|
'Package': 4 ,
|
||||||
|
'Class': 5 ,
|
||||||
|
'Method': 6 ,
|
||||||
|
'Property': 7 ,
|
||||||
|
'Field': 8 ,
|
||||||
|
'Constructor': 9 ,
|
||||||
|
'Enum': 10 ,
|
||||||
|
'Interface': 11 ,
|
||||||
|
'Function': 12 ,
|
||||||
|
'Variable': 13 ,
|
||||||
|
'Constant': 14 ,
|
||||||
|
'String': 15 ,
|
||||||
|
'Number': 16 ,
|
||||||
|
'Boolean': 17 ,
|
||||||
|
'Array': 18 ,
|
||||||
|
'Object': 19 ,
|
||||||
|
'Key': 20 ,
|
||||||
|
'Null': 21 ,
|
||||||
|
'EnumMember': 22 ,
|
||||||
|
'Struct': 23 ,
|
||||||
|
'Event': 24 ,
|
||||||
|
'Operator': 25 ,
|
||||||
|
'TypeParameter': 26
|
||||||
|
};
|
||||||
|
|
||||||
|
declare type lspt:TraceValue as enum( 'off' , 'messages' , 'verbose');
|
||||||
|
|
@ -2,32 +2,13 @@
|
||||||
positions in text
|
positions in text
|
||||||
:)
|
:)
|
||||||
module namespace pos="lsp/position";
|
module namespace pos="lsp/position";
|
||||||
|
import module namespace lspt = 'lsp-typedefs' at "lsp-typedefs.xqm";
|
||||||
|
|
||||||
(:~ json numbers :)
|
|
||||||
declare type pos:num as (xs:integer|xs:double);
|
|
||||||
|
|
||||||
(:~
|
|
||||||
@param line Line position in a document (zero-based).
|
|
||||||
@param character Character offset on a line in a document (zero-based).
|
|
||||||
:)
|
|
||||||
declare record pos:Position(
|
|
||||||
line as pos:num,
|
|
||||||
character as pos:num
|
|
||||||
);
|
|
||||||
|
|
||||||
(:~
|
|
||||||
@param line Line position in a document (zero-based).
|
|
||||||
@param character Character offset on a line in a document (zero-based).
|
|
||||||
:)
|
|
||||||
declare record pos:Range(
|
|
||||||
start as pos:Position,
|
|
||||||
end as pos:Position
|
|
||||||
);
|
|
||||||
|
|
||||||
|
|
||||||
(:~ find index from Position :)
|
(:~ find index from Position :)
|
||||||
declare function pos:resolvePosition($text as xs:string, $pos as pos:Position)
|
declare function pos:resolvePosition($text as xs:string, $pos as lspt:Position)
|
||||||
as pos:num
|
as lspt:num
|
||||||
{
|
{
|
||||||
let $nl:= index-of(string-to-codepoints($text),10)
|
let $nl:= index-of(string-to-codepoints($text),10)
|
||||||
let $off:=if($pos?line eq 0)
|
let $off:=if($pos?line eq 0)
|
||||||
|
|
@ -38,9 +19,9 @@ as pos:num
|
||||||
|
|
||||||
(:~ convert index into Position :)
|
(:~ convert index into Position :)
|
||||||
declare function pos:toPosition($text as xs:string,
|
declare function pos:toPosition($text as xs:string,
|
||||||
$index as pos:num
|
$index as lspt:num
|
||||||
)
|
)
|
||||||
as pos:Position {
|
as lspt:Position {
|
||||||
let $nl:= if($index=>trace("IN ") gt string-length($text)=>trace("L "))
|
let $nl:= if($index=>trace("IN ") gt string-length($text)=>trace("L "))
|
||||||
then error(xs:QName("pos:range"),`out of range: index={$index},length={string-length($text)}`)
|
then error(xs:QName("pos:range"),`out of range: index={$index},length={string-length($text)}`)
|
||||||
else index-of(string-to-codepoints($text),10)
|
else index-of(string-to-codepoints($text),10)
|
||||||
|
|
@ -48,13 +29,13 @@ as pos:Position {
|
||||||
let $off:=if($line eq 0)
|
let $off:=if($line eq 0)
|
||||||
then 0
|
then 0
|
||||||
else $nl[$line]
|
else $nl[$line]
|
||||||
return pos:Position($line, $index - $off)
|
return lspt:Position($line, $index - $off)
|
||||||
};
|
};
|
||||||
|
|
||||||
|
|
||||||
|
|
||||||
(:~ line number for $pos :)
|
(:~ line number for $pos :)
|
||||||
declare function pos:lineAt($nl as xs:integer*,$pos as pos:num)
|
declare function pos:lineAt($nl as xs:integer*,$pos as lspt:num)
|
||||||
as xs:integer
|
as xs:integer
|
||||||
{
|
{
|
||||||
if(empty($nl) or $pos le head($nl))
|
if(empty($nl) or $pos le head($nl))
|
||||||
|
|
@ -75,7 +56,7 @@ as xs:integer
|
||||||
};
|
};
|
||||||
|
|
||||||
(:~ format position for text display :)
|
(:~ format position for text display :)
|
||||||
declare function pos:ln-col($pos as pos:Position,$offset as xs:integer:=1)
|
declare function pos:ln-col($pos as lspt:Position,$offset as xs:integer:=1)
|
||||||
{
|
{
|
||||||
`Ln { $pos?line + $offset}, Col { $pos?character + $offset}`
|
`Ln { $pos?line + $offset}, Col { $pos?character + $offset}`
|
||||||
};
|
};
|
||||||
|
|
@ -99,6 +80,6 @@ as xs:string{
|
||||||
|
|
||||||
(:~ full range for $text :)
|
(:~ full range for $text :)
|
||||||
declare function pos:full-range($text as xs:string)
|
declare function pos:full-range($text as xs:string)
|
||||||
as pos:Range{
|
as lspt:Range{
|
||||||
pos:Range(pos:toPosition($text,0), pos:toPosition($text, string-length($text)))
|
lspt:Range(pos:toPosition($text,0), pos:toPosition($text, string-length($text)))
|
||||||
};
|
};
|
||||||
|
|
|
||||||
Loading…
Add table
Reference in a new issue