Follow-up for bug 118470 SpreadsheetML parts encode most ASCII control characters (x00-x08 and x0a-x1f) with "_xhhhh_" hex sequences (e.g. "\x1f" => "_x001F_"). Additionally, if such hex sequences appear literally, they will be encoded by replacing the leading underscore character, e.g. "_001f_" => "_005F_001f_". This applies to almost all strings that can appear in an XLSX file (workbook.xml, sheet.xml, sharedStrings.xml, table.xml, etc etc). LO Calc needs to decode all these strings when loading XLSX, and furthermore *needs to encode* all strings when writing XLSX. The following contents are affected (among others): - cell content string (xl/sharedStrings.xml) - cell formula (xl/worksheets/sheetN.xml) - sheet name (xl/workbook.xml) - cell style name, font name, number format code (xl/styles.xml) - cell hyperlink ("#location" part) - named ranges (name + formula) - tables (xl/tables/table1.xml) - auto-filter/table filter: filter entries - data validation: formulas, string list, error title/text, prompt title/text - conditional formatting: formulas (comparison, color steps, databar min/max, iconset steps), text rules - cell notes (xl/commentsN.xml) - threaded comments (xl/threadedComments/threadedCommentN.xml)
Created attachment 180623 [details] test file for import
Thanks, let's set to NEW