Changelog:
orcus 0.17.2
* ods
* fixed a bug where the state of style:cell-protect="none" was not
explicitly pushed, thereby having had the same effect as not having this
attribute. After the fix, style:cell-protect="none" will explicitly push
the hidden state to false, locked state to false, and the formula-hidden
state to false.
orcus 0.17.1
* general
* addressed a number of coverity issues.
* removed a variety of compiler warnings.
* ods
* re-generated sax parser tokens from ODF v1.3.
* revised the style import code to only push style attributes that are
actually specified in the XML.
* xls-xml
* revised the XML structure validation strategy to ignore any mis-placed
elements and their sub structures rather than aborting the import.
orcus 0.17.0
* general
* set the baseline C++ version to 17.
* cleaned up the public API to replace pstring with std::string_view, union
with std::variant, and boost::optional with std::optional. With this
change, the public API no longer has dependency on boost.
* spreadsheet document
* switched to using ixion::model_iterator for horizontal iteration of cells
instead of using mdds::mtv::collection.
* fixed a bug where exporting a spreadsheet document containing adjacent
merged cells regions to html incorrectly exported the merged cell areas.
* xlsx
* cached cell values are now correctly loaded from the file.
* sax parser
* utf-8 names are now allowed as element and attribute names.
* css parser
* unquoted utf-8 property values are now allowed.
* orcus-json
* fixed segmentation fault when using --mode structure with the Windows
build.
* added yaml output option.
* xml-map
* fixed a bug where mapping of an XML document with namespace aliases
sometimes corrupts the alias values.
* python
* added orcus.FormulaTokenOp enum type which describes type formula token
operator types in a more finer grained manner.
* documentation
* added notes to how to use orcus-xml and orcus-json to map XML and JSON
documents to spreadsheet documents.
Changelog:
0.16.1
fixed a build issue on 32-bit linux platforms, which was
indirectly caused by ixion.
fixed json parsing bug caused by an uninitialized variable,
which manifested itself on debian 32-bit platform.
removed compiler warnings on unused variables from the base
parser handlers.
0.16.0
general
full formula recalculations are now optional when loading
documents. It makes more effective use of cached formula
results.
added the option of failing on the first faulty cell, or
skipping them.
fixed a bug that caused the threaded_sax_token_parser to
deadlock.
added base parser handler classes in the public headers so
that they can be sub-classed to overwrite necessary handler
methods.
json-parser
parsing of numeric values are now more strict for better
conformance to the specs.
ods
added support for loading named expressions from ods
documents.
fixed an infinite loop when loading one of the attached
ods documents from
https://bugs.documentfoundation.org/show_bug.cgi?id=82414
xlsx
fixed a segfault when loading the xlsx document from
https://bugs.documentfoundation.org/show_bug.cgi?id=83711.
xls-xml
fixed a bug that prevented formulas from referencing cells
located in later sheets.
xml-map
adjusted the xml path expressions to be more like XPath.
Previously, an attribute was expressed as '@' in the old
expression, but XPath uses '/@'. The new expression uses
'/@' for an attribute.
added the ability to identify and import ranges from XML
documents without map file.
added the ability to generate map file from XML documents
for user customization.
added support to specify default namespace in the map file.
python
added orcus.Cell class to represent individual cell values
and attributes.
fixed several memory leaks in the python binding layer.
modified orcus.csv.read() function to take string input,
instead of bytes.
added __version__ attribute to the orcus module.
cleaned up orcus.detect_format function to only take the
stream parameter.
added named_expressions properties to Document and Sheet
class objects.
added Python API to bulk-process a number of spreadsheet
documents (orcus.tools.file_processor).
added Python API to download attachments from bugzilla
services via REST API (orcus.tools.bugzilla).
Release Notes
fixed a build error with gcc 10 with LTO. For more details, visit https://bugs.gentoo.org/715154.
removed potentially non-free specification and schema files from the package.
Changelog:
orcus 0.15.3
* xml-map
* fixed another bug related to filling of cells down the column in a linked
range with nested repeat elements. The bug would occur when the field in
a linked range is more than one level deeper than the nearest row group
element.
* xls-xml
* fixed a bug where TopCell and LeftCell attributes of the Table element
were not properly honored.
orcus 0.15.2
* xml-map
* fixed a bug that prevented filling of cells down the column in a linked
range with nested repeat elements. The bug would occur when the field in
a linked range is associated with an element content rather than an
attribute.
* xls-xml
* added code to properly pick up and pass the number format codes, including
named number format values such as 'General Date', 'Long Time, 'Currency'
etc.
* fixed a build issue on older macOS environment, related to passing an rvalue
to a tuple expecting a const reference. The root cause was a bug in libc++
of LLVM < 7.
* fixed a build issue with gcc5.
orcus 0.15.1
* switched xml_map_tree to using boost::object_pool to manage the life
cycles of the objects within xml_map_tree, to avoid memory
fragmentation.
* fixed incorrect handling of newly created elements in xml_map_tree.
* fixed segfault caused by double deletion of allocated memory for
xml_map_tree::element, which seemed to happen only on 32-bit gcc builds.
* fixed weird test failures related to equality check of two double-precision
values, caused probably by aggressive compiler optimization which only seems
to get triggered in 32-bit gcc builds.
orcus 0.15.0
* spreadsheet interface
* import_sheet::fill_down_cells() has been added as a required method, to
allow the import filter code to duplicate cell value downward in one step.
* json parser
* added test cases from JSONTestSuite.
* fixed a bug on parsing an empty array containing one or more blank
characters between the brackets.
* sax parser
* fixed a bug on parsing an attribute value with encoded character
immediately followed by a ';', such as '&;'.
* fixed a bug on parsing an assignment character '=' that either preceded or
followed by whitespaces in attribute definition.
* optionally use SSE4.2 intrinsics to speed up element name parsing.
* orcus-xml
* revised its cli interface to make use of boost's program_options.
* orcus-xml-dump's functionality has been combined into orcus-xml.
* map mode now supports nested repeat elements to be mapped as range fields.
* orcus-json
* map mode has been added to allow mapping of JSON documents to spreadsheet
document model. This mode either takes explicit mapping rule via map
file, or performs automatic mapping by auto-identifying mappable ranges by
analyzing the structure of the JSON document.
* structure mode has been added to display the logical structures of JSON
documents.
* significantly improved performance of json document tree by utilizing
object pool to manage the life cycles of json value instances.
* xls-xml
* added support for importing named color values in the ss:Color attributes.
* added support for handling UTF-16 streams that contains byte order marks.
* spreadsheet document
* significantly improved performance of flat format output generation.
* internal
* string_pool now uses boost's object_pool to manage the instances of stored
strings.
* file_content class has been added to memory-map file contents instead of
loading them in-memory.
* memory_content class has been added to map in-memory buffer with the
optional ability to perform unicode conversion.
* dom_tree has been renamed to dom::document_tree, and its interface has
been cleaned up to hide its implementation details.
libreoffice, the sole package that presently depends on orcus, requires a
minimum version of 0.14 now. (And, for that matter, the orcus change log
indicates that there have been incompatible API changes.) Bump
accordingly here. (If an older version of orcus was present, the
current version of libreoffice would simply have failed to build.)
Changelog:
orcus 0.14.1
* addressed a number of coverity issues.
* improved precision of points-to-twips measurement conversions by
reducing the number of numeric operations to be performed. This
especially helps on i386 platforms.
orcus 0.14.0
* spreadsheet interface
* import_data_table::set_range() now receives a parameter of type
range_t.
* import_sheet::set_array_formula() interface methods have been
removed and replaced with import_sheet::get_array_formula() that
returns an interface of type import_array_formula.
* import_formula interface class has been added to replace the
formula related methods of import_sheet. As a result,
set_formula(), set_shared_formula(), and set_formula_result()
methods have been removed from the import_sheet interface class.
* import_auto_filter::set_range() now receives a parameter of type
range_t, rather than a string value representing a range.
* import_sheet::set_fill_pattern_type() interface method now takes
an enum value of type fill_pattern_t, rather than a string value.
* xls-xml
* pick up the character set from the XML declaration, and pass it
to the client app via import_global_settings interface.
* support importing of array formulas.
* xlsx
* support importing of array formulas.
* fixed a bug where sheet indices being passed to the append_sheet()
interface method were incorrect.
* shared formula handling code has been re-worked.
* spreadsheet::sheet class has been de-coupled from the import and
export interfaces.
* previously known as import_styles class is now split into styles
class and import_styles factory wrapper class.
* sax_parser now gracefully ignores leading whitespace(s) if any,
rather than aborting the parsing for it's not a valid XML stream
to have leading whitespace(s). In the future we should make this
behavior configurable.
* python
* add orcus.xlsx.read() function that takes a file object to load
an xlsx file as a replacement for orcus.xlsx.read_file().
* add orcus.ods.read(), orcus.xls_xml.read(), orcus.csv.read(),
and orcus.gnumeric.read() functions.
* add orcus.Sheet.write() method which exports sheet content to
specified format. For now only the csv format type is
supported.
* xml_map_tree no longer requires the source stream persisted in
memory between the read and write.
* the sax parser now stores the offset positions of each element
rather than their memory positions, in order to make the position
values usable between duplicated stream instances.
* xml_structure_tree to support selection of an element by element
path.
* document
* correctly set the argument separator depending on the formula
grammar type. This change fixes loading of ods documents with
formula cells.
* fixed a build issue with boost 1.67.
* Fix build with boost 1.65.0
Changelog:
2018-02-26 Kohei Yoshida <kohei.yoshida@gmail.com> [ef2e27538e335583ef3ff85c4bc4f512efc72eb5]
Up the version to 0.13.4.
2018-02-21 Markus Mohrhard <markus.mohrhard@googlemail.com> [13af2fbab2cac1020d6bb840833c0e0efc231bff]
protect the self-closing xml element code against self-closing root elements
Found by Antti Levomäki and Christian Jalio from Forcepoint.
(cherry picked from commit 12e5d89cbd7101c61fbdf063322203a1590a0ef5)
2018-02-19 Kohei Yoshida <kohei.yoshida@gmail.com> [b8848ef7fc6a7d89e3f872574e36cbbab82275b0]
xls-xml: Gracefully handle formula cells without cached results.
This fixes#51.
(cherry picked from commit 32a1b05ffc6edd7d528b6760dab9035252329ab0)
Changelog:
2018-02-14 Kohei Yoshida <kohei.yoshida@gmail.com> [7ca73a7c83504a30a1d24444a27f57a86451100f]
Up the version to 0.13.3.
2018-02-13 Kohei Yoshida <kohei.yoshida@gmail.com> [66bbbd42f5d135b7e7dd57eaa7fdf6fd69c664df]
xls-xml: Import hidden row and column flags.
(cherry picked from commit 95420c1a1e8c082bb5953b2a49f0d56eef0e5f7e)
2018-02-08 Kohei Yoshida <kohei.yoshida@gmail.com> [0798d81a4c771b69b4b8eade396c88ffb5416b04]
xlsx: Remove carriage returns from multi-line strings.
Let's try to consistently only use linefeed characters for multi-
line strings.
(cherry picked from commit 0412bd269983825e5019a8a12267b54f51117aba)
2018-02-08 Kohei Yoshida <kohei.yoshida@gmail.com> [0a4e8c44fc8229818191c6b9b46e4de079d0ca3b]
xls-xml: Pick up border colors.
(cherry picked from commit e065d26dabafea465ec49e7d79775e62014ac0db)
2018-02-07 Kohei Yoshida <kohei.yoshida@gmail.com> [9662fce62ce77f87a4a8ba61f4507ec08e705b57]
xlsx: Let's not forget to apply color for diagonal borders too.
(cherry picked from commit c392ea15000b331bb6580b09c1045fd14b449b46)
2018-01-31 Kohei Yoshida <kohei.yoshida@gmail.com> [473526e1ca3a7117e2daf977e1b82a0a3977fc84]
We are supposed to use the foreground color for solid fill.
(cherry picked from commit f821995022df8dd1e580dd22cf131584b2b1ac4f)
2018-01-31 Kohei Yoshida <kohei.yoshida@gmail.com> [98d2b3377da71b713a37f9004acff3c02c22ce2b]
Alpha value of 0 means fully transparent. I'm sure 255 was intended.
(cherry picked from commit f7953a814d6a43205791b6cc01c528ef5d4b1ce3)
2018-01-26 Kohei Yoshida <kohei.yoshida@gmail.com> [5aba1df254cf4e052ad013d4b8ac886e274b74fa]
Revert "fix automake warning"
This reverts commit e4e1e3eb41755a4520a22b904a638da0770836f1.
This fixes the breakage on 'make distcheck'.
Changelog:
2016-05-11 Kohei Yoshida <kohei.yoshida@gmail.com> [d6084fe1771052e516ecfb270cb24dd9917a1895]
Up the version to 0.11.2.
2016-05-11 Kohei Yoshida <kohei.yoshida@gmail.com> [70fd8327c94b27a99e2c7800e91c13e5099cceda]
Make it buildable with mdds-1.2.
Problems found with existing distfile:
distfiles/libiconv-1.13-cp932.patch.gz
No changes made to the libiconv distinfo file.
Otherwise, existing SHA1 digests verified and found to be the same on
the machine holding the existing distfiles (morden). All existing
SHA1 digests retained for now as an audit trail.
This is a maintenance release. It primarily includes bug fixes and
build fixes since the 0.7.0 release with no new features. That
said, the most notable aspect of this release is that it is buildable
with the version 0.9.0 of the Ixion library which was just released
a week ago. So, if you are trying to package and distribute the
newly-released Ixion library but are unable to do so because of
Orcus not being buildable with it, you might be interested in this
release.
