Details
-
Bug
-
Resolution: Fixed
-
Low
-
3.0
-
None
Description
This problem occurs due to the browser sending the wrong MIME type during a file upload. It appears that Windows boxes where MS Excel handles CSV files uploads CSV files with the "application/vnd.ms-excel" MIME type. It can cause the search index to be only partially built, resulting in missing pages in search results.
Sample logs:
2010-02-22 11:09:56,038 WARN [Indexer: 2] [bonnie.search.extractor.BaseAttachmentContentExtractor] addFields Error indexing attachment (Attachment: textures-
streaming.csv v.1 (3014859) kteich)
-- url: /confluence/admin/reindex.action | userName: moise | referer: https://qix.demiurgestudios.com/confluence/admin/search-indexes.action | action: reind
ex
com.atlassian.bonnie.search.extractor.ExtractorException: Error reading content of Excel document: Invalid header signature; read 8236850760414359372, expect
ed -2226271756974174256
at com.atlassian.bonnie.search.extractor.MsExcelContentExtractor.extractText(MsExcelContentExtractor.java:101)
at com.atlassian.bonnie.search.extractor.BaseAttachmentContentExtractor.addFields(BaseAttachmentContentExtractor.java:39)
at com.atlassian.confluence.plugin.descriptor.ExtractorModuleDescriptor$BackwardsCompatibleExtractor.addFields(ExtractorModuleDescriptor.java:43)
at com.atlassian.bonnie.search.BaseDocumentBuilder.getDocument(BaseDocumentBuilder.java:104)
at com.atlassian.confluence.search.lucene.ConfluenceDocumentBuilder.getDocument(ConfluenceDocumentBuilder.java:102)
at com.atlassian.confluence.search.lucene.tasks.AddDocumentIndexTask.perform(AddDocumentIndexTask.java:41)
at com.atlassian.bonnie.index.TempIndexWriter.perform(TempIndexWriter.java:72)
at com.atlassian.confluence.search.lucene.TempIndexWriterStrategy.perform(TempIndexWriterStrategy.java:43)
at com.atlassian.confluence.search.lucene.tasks.TempIndexBackedIndexTaskPerformer.perform(TempIndexBackedIndexTaskPerformer.java:21)
at com.atlassian.confluence.search.lucene.DefaultObjectQueueWorker.indexCollection(DefaultObjectQueueWorker.java:73)
at com.atlassian.confluence.search.lucene.DefaultObjectQueueWorker$1.doInTransactionWithoutResult(DefaultObjectQueueWorker.java:61)
at org.springframework.transaction.support.TransactionCallbackWithoutResult.doInTransaction(TransactionCallbackWithoutResult.java:33)
at org.springframework.transaction.support.TransactionTemplate.execute(TransactionTemplate.java:127)
at com.atlassian.confluence.search.lucene.DefaultObjectQueueWorker.run(DefaultObjectQueueWorker.java:50)
at java.util.concurrent.ThreadPoolExecutor$Worker.runTask(ThreadPoolExecutor.java:650)
at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:675)
at java.lang.Thread.run(Thread.java:595)
Caused by: java.io.IOException: Invalid header signature; read 8236850760414359372, expected -2226271756974174256
at org.apache.poi.poifs.storage.HeaderBlockReader.<init>(HeaderBlockReader.java:103)
at org.apache.poi.poifs.filesystem.POIFSFileSystem.<init>(POIFSFileSystem.java:90)
at com.atlassian.bonnie.search.extractor.MsExcelContentExtractor.extractText(MsExcelContentExtractor.java:87)
... 16 more
Workaround
Stop Confluence, edit the confluence/WEB-INF/classes/mime.types file and add the following entry:
text/csv csv
This ensures that all files with the CSV extension are mapped to the text/csv MIME type regardless of what the browser sends.
Next, run the following query against the database and then start Confluence:
update attachments set contenttype='text/csv' where lower(title) like '%.csv';
To make the content in the CSV files searchable you will also need to run a reindex.
