[ https://issues.apache.org/jira/browse/TIKA-2618?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16417998#comment-16417998 ]
Hudson commented on TIKA-2618: ------------------------------ UNSTABLE: Integrated in Jenkins build tika-2.x-windows #224 (See [https://builds.apache.org/job/tika-2.x-windows/224/]) TIKA-2618 -- avoid overwriting labels (tallison: rev 7a9b17f478c867c7df5516b4ebb2ce3bf8b0aa36) * (edit) tika-parsers/src/main/java/org/apache/tika/parser/microsoft/ExcelExtractor.java * (edit) tika-parsers/src/test/java/org/apache/tika/parser/microsoft/ExcelParserTest.java * (add) tika-parsers/src/test/resources/test-documents/testEXCEL_labels-govdocs-515858.xls > LabelRecord and LabelSSTRecord text can be overwritten in xls > ------------------------------------------------------------- > > Key: TIKA-2618 > URL: https://issues.apache.org/jira/browse/TIKA-2618 > Project: Tika > Issue Type: Task > Reporter: Tim Allison > Priority: Major > Fix For: 1.18, 2.0.0 > > > In our regression tests, we've lost small amounts of text from quite a few > xls (standalone, but especially embedded). This is somewhat caused by > removing the {{listenForAllRecords=true}} that I accidentally left in as part > of debugging something a while ago. When that is true, we don't cache the > records in currentSheet, so they are added to the {{extraTextCells}} list. > When that is false, which is now the default, the {{LabelRecord}} and > {{LabelSSTRecord}} are sometimes being overwritten because multiple cells can > have the same x/y coordinates in the {{currentSheet}} map. > When {{listenForAllRecords=false}}, we're trying to listen for labels, but > we're often overwriting them because of the map. > Let's add labels to {{extraTextCells}} so that at least the text is processed. > As one example: "africa" in govdocs1/199/199294.ppt -- This message was sent by Atlassian JIRA (v7.6.3#76005)