[ https://issues.apache.org/jira/browse/TIKA-3544?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17412012#comment-17412012 ]
Tim Allison edited comment on TIKA-3544 at 9/8/21, 3:40 PM: ------------------------------------------------------------ In TIKA-2025 (which is nearly exactly this issue), we added a custom TikaExcelDataFormatter that allowed us to inject TikaExcelGeneralFormat. This broke Excel and POI's default handling to allow up to 15 digits to be extracted. When I look at the underlying xml of the attached file, 6480195344642780 is, in fact, stored there. If we bump our custom handling to 16 digits this problem would be solved _for this file_ and for numbers with 16 digits. As Tilman and Nick note, though, Excel is really bad for numbers that might start with leading zeros, like credit card #s, etc. You have to be really careful to enter them as strings or, better yet, use an actual database. was (Author: talli...@mitre.org): In TIKA-2025 (which is nearly exactly this issue), we added a custom TikaExcelDataFormatter that allowed us to inject TikaExcelGeneralFormat. This broke Excel and POI's default handling to allow up to 15 digits to be extracted. When I look at the underlying xml, 6480195344642780 is, in fact, stored there. If we bump our custom handling to 16 digits this problem would be solved _for this file_ and for numbers with 16 digits. As Tilman and Nick note, though, Excel is really bad for numbers that might start with leading zeros, like credit card #s, etc. You have to be really careful to enter them as strings or, better yet, use an actual database. > Extraction of long sequences of digits from Excel spreadsheets using Tika > 1.20 doesn’t yield the expected results > ----------------------------------------------------------------------------------------------------------------- > > Key: TIKA-3544 > URL: https://issues.apache.org/jira/browse/TIKA-3544 > Project: Tika > Issue Type: Bug > Components: parser > Affects Versions: 1.20 > Reporter: Jitin Jindal > Priority: Major > Attachments: Credit Card Numbers.xlsx > > > If an Excel spreadsheet contains a long sequence of digits, such as a credit > card number, Tika 1.13 will emit the said sequence in scientific notation. > For example, the credit card number “6011799905775830” is extracted from the > attached spreadsheet as 6.480195344642784E15, which clearly is not the > desired output. > I think the impact of this issue is significant. There’s plenty of > information that can no longer be reliably extracted from spreadsheets. Think > credit card numbers, telephone numbers and product identifiers to name a few. -- This message was sent by Atlassian Jira (v8.3.4#803005)