Dokumenten-Check/dc-backend/target/classes/de/frigosped/dc/service/DocumentProcessingService.class

<EFBFBD><EFBFBD><EFBFBD><EFBFBD>AF

java/lang/Object
<init>
()V
	

1de/frigosped/dc/service/DocumentProcessingService
resolveMimeType
8(Ljava/lang/String;Ljava/lang/String;)Ljava/lang/String;	
LOG
Lorg/jboss/logging/Logger;
Extrahiere Text: %s (%s)

org/jboss/logging/Logger
debugf
9(Ljava/lang/String;Ljava/lang/Object;Ljava/lang/Object;)V

java/lang/String
hashCode
()I 
application/pdf
"#$
equals
(Ljava/lang/Object;)Z&
Gapplication/vnd.openxmlformats-officedocument.wordprocessingml.document(
application/msword*
'application/vnd.oasis.opendocument.text,

text/plain.

text/markdown
012
extractFromPdf
([B)Ljava/lang/String;
452
extractFromDocx
782
extractFromOdt	:;<=>
!java/nio/charset/StandardCharsets
UTF_8
Ljava/nio/charset/Charset;
@A
([BLjava/nio/charset/Charset;)VC
/Unbekannter MIME-Typ '%s' – versuche als Text
EFG
warnf
'(Ljava/lang/String;Ljava/lang/Object;)VI
java/lang/ExceptionK
Fehler bei Textextraktion (%s)
MNO
errorf
<(Ljava/lang/Throwable;Ljava/lang/String;Ljava/lang/Object;)VQ
java/lang/RuntimeException
HSTU

getMessage
()Ljava/lang/String;WXY
makeConcatWithConstants
&(Ljava/lang/String;)Ljava/lang/String;
P[\
*(Ljava/lang/String;Ljava/lang/Throwable;)V^
Starte PDF-OCR...
`ab
debug
(Ljava/lang/Object;)Vd
java/lang/StringBuilder
c
ghijk
org/apache/pdfbox/Loader
loadPDF
*([B)Lorg/apache/pdfbox/pdmodel/PDDocument;m
'org/apache/pdfbox/rendering/PDFRenderer
lop
)(Lorg/apache/pdfbox/pdmodel/PDDocument;)V
rstu
$org/apache/pdfbox/pdmodel/PDDocument
getNumberOfPagesw
PDF hat %d Seite(n)
yz
(Ljava/lang/String;I)V|
OCR Seite %d/%d
~
(Ljava/lang/String;II)VCH	<00><00><00><00><00>
%org/apache/pdfbox/rendering/ImageType
RGB
'Lorg/apache/pdfbox/rendering/ImageType;
l<00><00><00>
renderImageWithDPI
I(IFLorg/apache/pdfbox/rendering/ImageType;)Ljava/awt/image/BufferedImage;<00>
java/io/ByteArrayOutputStream
<00><00>
PNG
<00><00><00><00><00>
javax/imageio/ImageIO
write
I(Ljava/awt/image/RenderedImage;Ljava/lang/String;Ljava/io/OutputStream;)Z
<00><00><00><00><00>
java/util/Base64

getEncoder
()Ljava/util/Base64$Encoder;
<00><00><00><00>
toByteArray
()[B
<00><00><00><00>2
java/util/Base64$Encoder
encodeToString
<00><00><00>
ocrPage
((Ljava/lang/String;II)Ljava/lang/String;
c<00><00><00>
append
-(Ljava/lang/String;)Ljava/lang/StringBuilder;<00>



r<00><00>
close<00>
java/lang/Throwable
<00><00><00><00>

addSuppressed
(Ljava/lang/Throwable;)V
c<00><00>U
toString
<00><00>U
trim<00>
$dev/langchain4j/data/message/Content<00>
	image/png
<00><00><00><00><00>
)dev/langchain4j/data/message/ImageContent
from
Q(Ljava/lang/String;Ljava/lang/String;)Ldev/langchain4j/data/message/ImageContent;
<00>X<00>
(II)Ljava/lang/String;
<00><00><00><00><00>
(dev/langchain4j/data/message/TextContent
>(Ljava/lang/String;)Ldev/langchain4j/data/message/TextContent;
<00><00><00><00><00>
(dev/langchain4j/data/message/UserMessage
S([Ldev/langchain4j/data/message/Content;)Ldev/langchain4j/data/message/UserMessage;	<00><00><00>
ocrModel
.Ldev/langchain4j/model/chat/ChatLanguageModel;<00><00><00><00><00>
java/util/List
of
$(Ljava/lang/Object;)Ljava/util/List;<00><00><00><00><00>
,dev/langchain4j/model/chat/ChatLanguageModel
generate
9(Ljava/util/List;)Ldev/langchain4j/model/output/Response;
<00><00><00><00><00>
%dev/langchain4j/model/output/Response
content
()Ljava/lang/Object;<00>
&dev/langchain4j/data/message/AiMessage
<00><00><00>U
text<00>
 Konvertiere DOCX → Markdown...<00>
*org/apache/poi/xwpf/usermodel/XWPFDocument<00>
java/io/ByteArrayInputStream
<00><00><00>
([B)V
<00><00><00>
(Ljava/io/InputStream;)V
<00><00><00><00>
getBodyElements
()Ljava/util/List;<00>




iterator
()Ljava/util/Iterator;




	
java/util/Iterator
hasNext
()Z


<00>
next

*org/apache/poi/xwpf/usermodel/IBodyElement

+org/apache/poi/xwpf/usermodel/XWPFParagraph




paragraphToMarkdown
A(Lorg/apache/poi/xwpf/usermodel/XWPFParagraph;)Ljava/lang/String;





'org/apache/poi/xwpf/usermodel/XWPFTable




tableToMarkdown
=(Lorg/apache/poi/xwpf/usermodel/XWPFTable;)Ljava/lang/String;
<00><00>



 U
getText

"
#
	
isBlank
%



'
(U
getStyle
*
Heading1

,
-
.
equalsIgnoreCase
(Ljava/lang/String;)Z
0
berschrift1W
3
Heading2
5
berschrift2W
8
Heading3
:
berschrift3W
=
Heading4
?
berschrift4W


B
C
D

getNumIlvl
()Ljava/math/BigInteger;W


G
H<00>
getRows<00>
J
K
	
isEmpty<00>
M
N
O
get
(I)Ljava/lang/Object;
Q
*org/apache/poi/xwpf/usermodel/XWPFTableRow
S
| 

P
U
V<00>

getTableCells
X
Y
Z
accept
8(Ljava/lang/StringBuilder;)Ljava/util/function/Consumer;<00>
\
]
^
forEach
 (Ljava/util/function/Consumer;)V
`

|
X<00>
c
d
size	
X
g
Konvertiere ODT → Markdown...

i
j
k
l
m
)org/odftoolkit/odfdom/doc/OdfTextDocument
loadDocument
B(Ljava/io/InputStream;)Lorg/odftoolkit/odfdom/doc/OdfTextDocument;

i
o
p
q

getContentDom
+()Lorg/odftoolkit/odfdom/dom/OdfContentDom;
s
text:p

u
v
w
x
y
'org/odftoolkit/odfdom/dom/OdfContentDom
getElementsByTagName
*(Ljava/lang/String;)Lorg/w3c/dom/NodeList;
{
|
}
~
org/w3c/dom/NodeList
	getLength
{
<01>
<01>
<01>
item
(I)Lorg/w3c/dom/Node;
<01>
<01>
<01>
<01>U
org/w3c/dom/Node
getTextContent
<01>
<01>
<01>
<01>

getAttributes
()Lorg/w3c/dom/NamedNodeMap;
<01>
text:style-name
<01>
<01>
<01>
<01>
<01>
org/w3c/dom/NamedNodeMap
getNamedItem
&(Ljava/lang/String;)Lorg/w3c/dom/Node;
<01>
<01>
<01>U
getNodeValue
<01>
Heading_20_1

<01>
<01>
.

startsWith
<01>
# 
<01>
Heading_20_2
<01>
## 
<01>
Heading_20_3
<01>
### 

i<00>
<01>
application/octet-stream

<01>
<01>U
toLowerCase
<01>
.pdf

<01>
<01>
.
endsWith
<01>
.docx
<01>
.doc
<01>
.odt
<01>
.txt
<01>
.md

<01>

<01>
+org/apache/poi/xwpf/usermodel/XWPFTableCell
<01>
 | 
<01>
---|

<01>
<01>
<01>
	getLogger
-(Ljava/lang/Class;)Lorg/jboss/logging/Logger;
OCR_DPI

F

ConstantValue
RuntimeVisibleAnnotations
Ljakarta/inject/Inject;
Ljakarta/inject/Named;
value
ocr
Code
LineNumberTable
LocalVariableTable
this
3Lde/frigosped/dc/service/DocumentProcessingService;
extractText
:([BLjava/lang/String;Ljava/lang/String;)Ljava/lang/String;

e
Ljava/lang/Exception;
	fileBytes
[B
mimeType
Ljava/lang/String;
filename

effectiveMime

StackMapTable
<01>
MethodParameters
image
Ljava/awt/image/BufferedImage;
baos
Ljava/io/ByteArrayOutputStream;
base64Image
pageText

i

I
renderer
)Lorg/apache/pdfbox/rendering/PDFRenderer;
	pageCount
pdf
&Lorg/apache/pdfbox/pdmodel/PDDocument;
result
Ljava/lang/StringBuilder;

Exceptions
pageNum

totalPages
message
*Ldev/langchain4j/data/message/UserMessage;
para
-Lorg/apache/poi/xwpf/usermodel/XWPFParagraph;
table
)Lorg/apache/poi/xwpf/usermodel/XWPFTable;
element
,Lorg/apache/poi/xwpf/usermodel/IBodyElement;
doc
,Lorg/apache/poi/xwpf/usermodel/XWPFDocument;
md
style
sb
rows
Ljava/util/List;
header
,Lorg/apache/poi/xwpf/usermodel/XWPFTableRow;
LocalVariableTypeTable
>Ljava/util/List<Lorg/apache/poi/xwpf/usermodel/XWPFTableRow;>;
	styleAttr
Lorg/w3c/dom/Node;
node

paragraphs
Lorg/w3c/dom/NodeList;
odfDoc
+Lorg/odftoolkit/odfdom/doc/OdfTextDocument;
lower
lambda$tableToMarkdown$2
I(Ljava/lang/StringBuilder;Lorg/apache/poi/xwpf/usermodel/XWPFTableCell;)V

c
-Lorg/apache/poi/xwpf/usermodel/XWPFTableCell;
lambda$tableToMarkdown$1
lambda$tableToMarkdown$0
<clinit>

SourceFile
DocumentProcessingService.java
.Ljakarta/enterprise/context/ApplicationScoped;
BootstrapMethods
 Textextraktion fehlgeschlagen: 


yDu bist ein OCR-Spezialist. Extrahiere den gesamten Text aus diesem Dokument-Bild (Seite 
 von 
).

Regeln:
- Formatiere den Text als Markdown
- Überschriften als # / ## / ###
- Tabellen als Markdown-Tabellen
- Listen als - oder 1. 2. 3.
- Behalte den Originaltext exakt (keine Korrekturen, keine Zusammenfassungen)
- Antworte NUR mit dem extrahierten Markdown, ohne Präambel
# 

## 
!
### 
#
#### 
%
- 
b(
)+
0(Lorg/apache/poi/xwpf/usermodel/XWPFTableCell;)V-
.0
1
3
456X7
$java/lang/invoke/StringConcatFactory
<00>(Ljava/lang/invoke/MethodHandles$Lookup;Ljava/lang/String;Ljava/lang/invoke/MethodType;Ljava/lang/String;[Ljava/lang/Object;)Ljava/lang/invoke/CallSite;9
:;<=>
"java/lang/invoke/LambdaMetafactory
metafactory
<00>(Ljava/lang/invoke/MethodHandles$Lookup;Ljava/lang/String;Ljava/lang/invoke/MethodType;Ljava/lang/invoke/MethodType;Ljava/lang/invoke/MethodHandle;Ljava/lang/invoke/MethodType;)Ljava/lang/invoke/CallSite;
InnerClasses
EncoderB
%java/lang/invoke/MethodHandles$LookupD
java/lang/invoke/MethodHandles
Lookup!
<01>
<01>

<01><00><00><00>

<01>
<01>
<01>

<01>s
<01>



<01>/

*<2A>
<01>
<01>
&
<01>

<01>
<01>

<01>
<01>
<01>
9*,-<2D>:<04>
-<04>:6<05><19><00><06><>k2i<><69><EFBFBD><EFBFBD>9<>\<5C><>I<>8<EFBFBD><38><00>0<EFBFBD><30>hy5<79><35>OY<1F>!<21>V6<06>P%<25>!<21>F6<06>@'<27>!<21>66<06>0)<29>!<21>&6<06> +<2B>!<21>6<06>-<2D>!<21>6<06>K%--5==*+<2B>/<2F>6*+<2B>3<>.*+<2B>6<>&<26>Y+<2B>9<>?<3F><18>
B<04>D<>Y+<2B>9<>?<3F>:<05>
J-<2D>L<>PY<05>R<>V<05>Z<>


H
<01>6
>?B<00>D<00>G<00>I<00>K
M
N
B
Q
R
%S
<01>>

<01>
<01>
9
<01>
<01>
9
<01>
<01>

9
<01>
<01>
9
<01>
<01>
1
<01>
<01>
<01>4<0E>X
&
<0A>
<01>
@H
<01>

<01>
<01>
<01>12
<01>
<01><00><>
]<5D>_<>cY<63>eM+<2B>fN<66>lY-<2D>n:-<2D>q6<05>
v<05>x6<05>_<>
{`<05>}<12><><00><><00>:<07><00>Y<EFBFBD><00>:<12><08><00>W<EFBFBD><00><08><00><><00>:	*	`<05><00>:
,
<EFBFBD><00><12><><00>W<EFBFBD>
<01><><EFBFBD>-<2D>#-<2D><00><>:-<2D>-<2D><00><>:<05><00><04>,<2C><00><><00><><00><00><00><00><00><00><00>
<01>J\]_`a%b/d9eGfUi^jhkum<00>n<00>d<00>p<00>_<00>r
<01>pU:
<01>
<01>^1
<01>
<01>u
<01>
<01>	<00>
<01>
<01>
2c
<01>
<01>v
<01>
<01>%p
<01>
<01><00>
<01>
<01><00>
<01>
<01><00>
<01>
<01>
<00>
<01>
<01>
<01>><06>2
<01>crl

<00>bJ<00><>
<01>cr<00>
<00><08>
<01>
H
<01>

<01><00><00>
<01><00>7<05><00>Y+<12><><00>SY<1D><00><00><00>S<EFBFBD><00>:*<2A><00><04>ٹ<00><00><00><><00><><00><>
<01>y	z{y<00>
<01>47
<01>
<01>7
<01>
<01>
7
<01>
<01>7
<01>
<01>
<01>
<01>
<01>

<01>
<01>
<01>52
<01>
<01><00><>
<12><>_<>cY<63>eM<65><00>Y<EFBFBD><00>Y+<2B><00><><00>N-<2D><00><>

:<04>

<00>U<04>


<00>

:<05>
<0F><05>
:,*<06>
<11><00>
<15><00>W<EFBFBD>#<05>
<17><05>
:,*<07>
<19><00>
<15><00>W<EFBFBD><57><EFBFBD>-<2D>
<1D>:-<2D>
<1D>:<05><00><04>,<2C><00><><00><> <00><00><00><00><00><00><00>
<01>2<00><00><00> <00>A<00>P<00>d<00>s<00><00><00><00><00><00><00><00><00>
<01>HP
<01>
<01>s
<01>
<01>AC
<01>
<01> <00>
<01>
<01><00>
<01>
<01><00>
<01>
<01>
<00>
<01>
<01>
<01>;<08>+c<00>
<05>8

<0A><1F>F<00><>
<01>c<00><00>
<00><08>
<01>
H
<01>

<01>


<01>
9<00>+<2B>
M,<2C>
,<2C>
!<21>
$<24>+<2B>
&N-<2D>,<2C>-
)<29>
+<2B>
-
/<2F>
+<2B>
,<2C>
1<00>-
2<>
+<2B>
-
4<>
+<2B>
,<2C>
6<00>-
7<>
+<2B>
-
9<>
+<2B>
,<2C>
;<00>-
<<3C>
+<2B>
-
><3E>
+<2B>
,<2C>
@<00>+<2B>
A<>
,<2C>
E<00>,<2C>
<01>><00><00><00><00><00>3<00>:<00>N<00>U<00>i<00>p<00><00><00><00><00><00><00><00><00>
<01>*<00>
<01>
<01><00>
<01>
<01>
<00><00>
<01><00>
<01>
<01>
<01><0C><03>


<01>

<01>


<01>
a<00><>cY<63>eM+<2B>
FN-<2D>
I
<00>
$<24>-<03>
L<00>
P:,
R<><00>W<04>
T,<2C>
W<00>
[,
_<><00>W<04>
T,<2C>
a<00>
[,
<15><00>W6-<2D>
b
<00>2,
R<><00>W-<05>
L<00>
P<>
T,<2C>
e<00>
[,
<15><00>W<EFBFBD>
<01><><EFBFBD>,<2C><00><>
<01>><00><00>
<00><00>&<00>.<00>><00>F<00>V<00>^<00>l<00>t<00><00><00><00><00><00><00>
<01>>a:
<01>
<01><00>
<01>
<01><00>
<01>
<01>
<00>
<01>
<01>
<00>
<01>&z


<00>
<01>
<01><03>c<00><>F
P
<01>9
<01>

<01>82
<01><02>

A<>

f<>_<>cY<63>eM<65><00>Y+<2B><00><>
hN-<2D>
n
r<>
t:6<04>
z
<00><00><05>
:<06>
<01>
:<07><00><07>
!<21><00><06>
<01>

<01><>
<01>:<08>
<08>
<01>
<00>
$:		
<01><>
<01><>	
)<29>
<01><>,
<01><><00><07><00>
<15><00>W<EFBFBD>h	
<01><>
<01><>	
2<>
<01><>,
<01><><00><07><00>
<15><00>W<EFBFBD><	
<01><>
<01><>	
7<>
<01><>,
<01><><00><07><00>
<15><00>W<EFBFBD>,<07><00>
<15><00>W<EFBFBD>
<01><>-<2D>#-<2D>
<01><>:-<2D>-<2D>
<01><>:<05><00><04>,<2C><00><><00><>

 <00>
&
*
-<00>
<01>^<00>	<00><00><00>$<00>)<00>8<00>C<00>L<00>Y<00>c<00>j<00>~<00><00><00><00><00><00><00><00><00><00><00>
<00>
<00>
<00>
 <00>
9<00>
<01>f
j<00>~<00>
<01>
<01>	C<00>L<00><00>
<01>,<00>
<01>
<01>)<00>	


A
<01>
<01>
A
<01>
<01>

0
<01>
<01>
<01>o<0F>,
<01>c
i
{
<00>L
<01>
<01>B<1A><15>
<01>c
i
{
<00>J<00><>
<01>c
i<00>
<00><08>
<01>
H
<01>

<01>
<01><00>v+<2B>+<2B>
!<21>+
<01><>!<21>+<2B>,<2C>
<01><>,<2C>
<01>N-
<01><>
<01><><1F>-
<01><>
<01><>%<25>-
<01><>
<01><>'<27>-
<01><>
<01><>)<29>-
<01><>
<01><>+<2B>-
<01><>
<01><>-<2D>
<01><>
<01>2





$
1
>
K
	X

e
r

<01>*v
<01>
<01>v
<01>
<01>
v
<01>
<01>$R
<01>
<01><07>
<01>	
<01>
<01>



<01>D*+<2B>
<01><><00>
<01><><00>W<EFBFBD>
<01>
<00>
<01>
<01>
<01>



<01>=	*
<01><><00>W<EFBFBD>
<01>
<00>
<01>	
<01>
<01>	



<01>D*+<2B>
<01><><00>
<01><><00>W<EFBFBD>
<01>
<00>
<01>
<01>
<01>


<01>!
	<08>
³
<0A>

<01>
(
<01>
J
2
2
2
2
2
 2
"2
$8&'*8&,*8&/*?<00><00>@	ACE