Am 13.05.2019 um 16:15 schrieb Zeke Steer:
Hi,
I'm using the latest version of the pdfbox command line tools
(pdfbox-app-2.0.15.jar) to extract the text from UK company annual
reports. I invoke the command line tools from a python
script, extracting each page of the company annual report .pdf
document in turn.
I've noticed that some pages of the annual reports aren't extracted
correctly. I originally observed this problem in an earlier version of
the command line tools (pdfbox-app-2.0.6.jar). However, moving to the
latest version of the tools hasn't fixed the issue.
I've attached a sample report which consistently reproduces the issue
(00054_CCH_Annual Report_2016-82-85.pdf). The report opens fine in
Adobe Reader but pdfbox is unable to extract it. The issue manifests
differently depending on whether the sequential (default) or
non-sequential parser is used.
The non-sequential parser is the only parser in 2.0.*
Using the option should bring a FileNotFoundException with 2.0.15, that
is what I get.
"Header doesn't contain versioninfo" is with empty files. I suspect one
of your calls used the PDF file as destination and you destroyed it.
Re different text extractions, please read
https://pdfbox.apache.org/2.0/faq.html#text-extraction
Your PDF file attachment didn't get through, please upload it to a
sharehoster.
Tilman
_Sequential Parser_
I was initially executing the following command with the -nonSeq flag
unset:
java -jar pdfbox-app-2.0.15.jar ExtractText -startPage 1 -endPage 1
"E:\Analyst Reports\2019-05-13 PDF Extraction Issue
Investigation\00054_CCH_Annual Report_2016-82-85\00054_CCH_Annual
Report_2016-82-85.pdf" "out\00054_CCH_Annual Report_2016-82-85\1.txt"
This would generate a large number of unicode warnings in the console,
e.g.:
May 13, 2019 10:12:18 AM org.apache.pdfbox.pdmodel.font.PDType0Font
toUnicode
WARNING: No Unicode mapping for CID+36 (36) in font
Effra-Medium-Identity-H
May 13, 2019 10:12:18 AM org.apache.pdfbox.pdmodel.font.PDType0Font
toUnicode
WARNING: No Unicode mapping for CID+88 (88) in font
Effra-Medium-Identity-H
May 13, 2019 10:12:18 AM org.apache.pdfbox.pdmodel.font.PDType0Font
toUnicode
WARNING: No Unicode mapping for CID+71 (71) in font
Effra-Medium-Identity-H
May 13, 2019 10:12:18 AM org.apache.pdfbox.pdmodel.font.PDType0Font
toUnicode
WARNING: No Unicode mapping for CID+76 (76) in font
Effra-Medium-Identity-H
May 13, 2019 10:12:18 AM org.apache.pdfbox.pdmodel.font.PDType0Font
toUnicode
WARNING: No Unicode mapping for CID+87 (87) in font
Effra-Medium-Identity-H
May 13, 2019 10:12:18 AM org.apache.pdfbox.pdmodel.font.PDType0Font
toUnicode
WARNING: No Unicode mapping for CID+3 (3) in font Effra-Medium-Identity-H
The pdfbox output was missing a large amount of text present on the
first page of the report. See the pdfbox output in the attached 1.txt
file and compare this to the first page of the company annual report,
also attached.
_Non-Sequential Parser_
I found the issue affected several of the annual reports within my
dataset. Investigating further, I read about the non-sequential
parser. You advise using this if the sequential parser fails so I
tried executing the following command instead, with the -nonSeq flag set:
java -jar pdfbox-app-2.0.15.jar ExtractText -nonSeq -startPage 1
-endPage 1 "E:\Analyst Reports\2019-05-13 PDF Extraction Issue
Investigation\00054_CCH_Annual Report_2016-82-85\00054_CCH_Annual
Report_2016-82-85.pdf" "out\00054_CCH_Annual Report_2016-82-85\1.txt"
However, this consistently fails with a 'java.io.IOException: Error:
Header doesn't contain versioninfo'. See the full exception stack
trace below:
Exception in thread "main" java.io.IOException: Error: Header doesn't
contain versioninfo
at org.apache.pdfbox.pdfparser.PDFParser.parse(PDFParser.java:221)
at org.apache.pdfbox.pdmodel.PDDocument.load(PDDocument.java:1070)
at org.apache.pdfbox.pdmodel.PDDocument.load(PDDocument.java:1008)
at
org.apache.pdfbox.tools.ExtractText.startExtraction(ExtractText.java:216)
at org.apache.pdfbox.tools.ExtractText.main(ExtractText.java:96)
at org.apache.pdfbox.tools.PDFBox.main(PDFBox.java:60)
failed to extract text from 'E:\Analyst Reports\2019-05-13 PDF
Extraction Issue Investigation\00054_
CCH_Annual Report_2016-82-85\00054_CCH_Annual Report_2016-82-85.pdf':
Command 'java -jar pdfbox-app-
2.0.15.jar ExtractText -nonSeq -startPage 1 -endPage 1 "E:\Analyst
Reports\2019-05-13 PDF Extraction
Issue Investigation\00054_CCH_Annual
Report_2016-82-85\00054_CCH_Annual Report_2016-82-85.pdf" "out
\00054_CCH_Annual Report_2016-82-85\1.txt"' returned non-zero exit
status 1.
I found a similar issue reported on your JIRA issue tracker here:
https://issues.apache.org/jira/browse/PDFBOX-4203?jql=text%20~%20%22versioninfo%22. However,
it was closed without being resolved as the original reporter failed
to provide a PDF document which reproduced the issue. Hopefully with
the information I've supplied, you'll be able to reopen the bug and
take another look.
Please can you keep me updated?
Many thanks,
Zeke
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
