[TIKA-189] Text extraction from Excel files juxtaposes cells

• Type: Bug
• Status: Closed
• Priority: Minor
• Resolution: Fixed
• Affects Version/s: 0.3
• Fix Version/s: 0.3
• Component/s: general

Environment

Tika revision is svn-20090116, platform is Windows XP Pro SP3, JDK version is 1.6.0_06.

• Assignee: Jukka Zitting
• Reporter: Georger Araújo
• Created: 17/Jan/09 18:52
• Updated: 22/May/09 22:04
• Resolved: 25/Jan/09 21:15

Description

I plan on using Tika to extract text from Excel (both .xls and .xlsx) files for indexing. But, I found that Tika juxtaposes cells on output. The example worksheets are in the attached .zip file.
I took the time to run Apache POI and it does not have this bug i.e. cells are properly separated.

When I run

-begin-
java -jar tika-0.3-SNAPSHOT-standalone.jar --text no_cell_separators_when_extracted.xls
-end-

I get the following output:

-begin-
Plan1
NameEmailSanta [e-mail]
Tooth [e-mail]
-end-

Same thing with a .xlxs file:
-begin-
java -jar tika-0.3-SNAPSHOT-standalone.jar --text no_cell_separators_when_extracted.xlsx
-end-

The output is:

-begin-
[Content_Types].xml

_rels/.rels

xl/_rels/workbook.xml.rels

xl/workbook.xml

xl/theme/theme1.xml

xl/worksheets/_rels/sheet1.xml.rels

xl/worksheets/sheet2.xml

xl/worksheets/sheet3.xml

xl/sharedStrings.xml
NameEmailSanta [e-mail] [e-mail]

xl/styles.xml

xl/worksheets/sheet1.xml
012345

docProps/core.xml
GeorgerGeorger2009-01-17T15:29:04Z2009-01-17T15:30:56Z

docProps/app.xml
Microsoft Excel0falsePlanilhas3Plan1Plan2Plan3falsefalsefalse12.0000
-end-

Also note that the values from docProps/app.xml have been juxtaposed as well.

This way, after indexing these files using the output from Tika, a search engine will only find "Fairy" when substring matching is used, because "Tooth Fairy" becomes "Tooth [e-mail]". This is suboptimal and wrong.

Thanks for your attention. Best regards,

Georger

Issue Links

• relates to: TIKA-188 Automatic whitespace for block elements in XHTMLContentHandler