- parsexmltag and other hot functions now better optimized for v8
- monomorphic functions (different types -> different funcs)
- more efficient decode_range implementation when source is trusted
- regular expressions cached and simplified without breaking correctness
- more efficient utf8 techniques when available
- XLSX: large functions broken down into sub-functions (e.g. `parse_ws_xml`)
- XLSB: avoid unnecessary binds
- XLSB: assume no exotic codepage exists (no one else tries to write XLSB)
- demo exposes rABS / worker / transferable options
- more tests
- jszip updated to 2.3.0
- SSF updated to 0.8.1
- codepage updated to 1.3.1
- require cpexcel instead of full codepage library
- hardcode the unescapexml regexp
- utf8read short circuit for ASCII strings
- cellFormulae also acts on XLSX/XLSM
- bin/xlsx.njs do not process formulae unless requested
- read BOM, handle UTF16LE-encoded XML
- handle namespaces in [Content_Types].xml
- parse workbook rels to determine sheet files
- numbers OSX boolean support (apparently requires "0" or "1")
- XLSX force "General" style to be serialized, omit implied cell type and style
- updated SSF to 0.7.0 (h/t @sysarchitect)
- updated jszip to 2.2.2
- removed old tests/files path, replaced with test_files
- themes written
- ignore potential existence of thumbnail when calculating relationship ids
- very basic XLSX / XLSM write support with roundtrip tests (XLSB stubs)
- reorganized source tree
- new XLSB range check ensures that A1 is not emitted for empty sheets
- SSF table emitted in output (consistent with js-xls)
- CLI supports writing
Backwards-incompatible changes:
o new Property aliases (see CORE_PROPS and EXT_PROPS)
o FILETIME custom properties parsed as JS Dates
o `xlsx2csv` -> `xlsx` (and `bin/xlsx{2csv,}.njs`)
- all utilities are now wrapped in XLSX object
- codepage handling (h/t @xch89820)
- formula tag attributes (h/t @shaunthomas999)
- hyperlink support (h/t @sysarchitect, fixes#55)
- coverage test spinner (to prevent travis timeouts)
- opts.sheetRows limits parsing; default (0) parses all rows
- added -n mode to xlsx2csv to control number of rows
- !ref will be adjusted; !fullref holds full range
- opts.bookDeps parses calc chain; now chains are not parsed by default
- getzipdata safe option obviates some try-catch blocks
- parsing cell type 'd' (h/t @malphettes, yet again :)
- opts.WTF throws on errors
- cellHTML controls HTML generation
- cellFormula controls formula output
- sheetStubs now defaults to false
- cleaned up unnecessary CSV quotes (fixed#45)
- updated test_files to 20140211
- updated SSF to 0.5.7
- removed unused main function
- removed some dead code
Cell object stores the formatted text in the `.w` field
- Raw format and types are preserved in the `.t`, `.v` fields
- Accessors have been updated to use the field
updates for older OOXML:
- ECMA-376 makes reference to old sstItem shared string item
- [MS-XLSX] xmlns for workbook
- merged test repo with xls
- travis integration
- shifted to Apache 2.0 LICENSE
- mocha tests
The shift to Apache 2.0 was long overdue. It's becoming clear that compliance
requires deeper integration with the xls libraries and other implementations
whose specifications are covered under the OSP. It was a grave oversight for
the ISO and ECMA people to not try to implement parsers themselves, for if they
did I suspect the outcome would have been much different
