[TIKA-4277] PDF parse issue for text rotated

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Duplicate
• Affects Version/s: 3.0.0-BETA, 2.9.2
• Component/s: tika-app, tika-server
• Labels: config.xml
• Flags: Important
• Assignee: Unassigned
• Reporter: ragebear
• Created: 11/Jul/24 11:48
• Updated: 12/Jul/24 07:11
• Resolved: 12/Jul/24 03:45

Description

the incorrect result parsed by Tika and Tika Server 2.9.2 and 3.0beta

The attached PDF cannot be correctly parsed by Tika 2.9.2 and 3.0beta, in server version and the standalone.

if the text is rotated 90. The parsed result will have a line break after each letter of word. It happened to symbol, English letters, and JCK characters.

In the server version, curl -g -T "sample2.pdf"
http://localhost:889/tika
--header "Accept: text/plain"

In the standalone version, java.exe -jar "C:\TikaSearch\tika-app-2.9.2.jar" --text

Both of above, deliver the the incorrect result in the attached pdf.

The output result is below

i
n
s
e
r
t
 
t
e
x
t
 
p
r
o
b
l
e
m

insert text problem

Issue Links

• duplicates: TIKA-2779 Integrate/parameterize new rotated text handling in PDFBox