SpreadsheetML

From Apache OpenOffice Wiki
Revision as of 08:15, 22 May 2008 by Dr (Talk | contribs)

Jump to: navigation, search


SpreadsheetML is the XML format used by Microsoft Excel 2007 and that is part of the Office Open XML specification.

SpreadsheetML Basics

workbook

A SpreadsheetML document is described at the top level by a workbook part ( /xl/workbook.xml ). The type of the work book type is

 http://schemas.openxmlformats.org/officeDocument/2006/relationships/officeDocument

The workbook part contains document's metadata and one or more sheets. Each sheet can be a worksheet, a chart sheet, or a dialog sheet.

Both strings and formulas are stored in shared tables to avoid redundant storage and speed file I/O's.

Sample Files

Convenient locations to download sample files are here

Implementation Strategy

Eventually, the code handling SpreadsheetML import will merge with the existing Excel binary filter. As such, the new code needs to be designed with this in mind. It's always desirable to understand how the existing binary filter works when implementing the XML filter to make the future merging work less painful.

Code Organization

Source files for handling the SpreadsheetML format are located in inc/oox/xls and source/xls under module oox. A good place to start tracing the code would be ExcelFilter::Import and follow the calls it makes.

A substream in the XML package is called "fragment", and each fragment has an associated *fragment.hxx header file. For instance, the code for loading of the workbook.xml fragment is found in workbookfragment.hxx, and so on.

A nested element is called "context", and, like the fragments, each context has an associated *context.hxx. For instance, the code for parsing the <sheetData> element is found in sheetdatacontext.hxx.

The term workbook in this context refers to an entire document which includes worksheets and other document metadata, whereas the term worksheet refers to each individual sheet in the workbook.

The existing binary Excel filter is located in sc/source/filter/(inc|excel)/xi(page|view).(c|h)xx. The XclImpTabViewSettings class handles importing sheet's view settings, which corresponds to worksheet/sheetViews context in the XML format.

The UNO interface code is found in sc/source/ui/uno.

Global data

Workbook-wide global data are stored in GlobalData (struct), and handled by GlobalDataHelper (class) which holds reference to the GlobalData instance. All major classes should be derived from GlobalDataHelper to ensure availability of globals in all places.

GlobalData holds reference to ImportBase instance in order to be able to create new fragment handlers.

Different buffers hold the imported data from the fragments if it is needed later, e.g. the SharedStringsBuffer (sharedstringsbuffer.hxx) and the StylesBuffer (stylesbuffer.hxx). These buffers are always part of the GlobalDataHelper.

GlobalDataHelper (class)

It's important to be familar with the member methods of this class, because they are also available in most of XLSX import classes.

Handling Fragment

WorkbookFragment (class)

Handles loading of workbook.xml fragment. It loads the associated relationship file (xl/_rels/workbook.xml.rels) in the constructor.

Handling Context

In most cases the fragment handler will handle all nested contexts by itself to increase performance. For this, some helper classes have been implemented that do all needed work to deal with nested contexts (ContextHelper, FragmentBase, and ContextBase respectively in contexthelper.hxx, excelfragmentbase.hxx, and excelcontextbase.hxx). In general, for implementing a new fragment or context handler, the interface of the ContextHelper class from contexthelper.hxx has to be implemented. The classes FragmentBase (excelfragmentbase.hxx) and ContextBase (excelcontextbase.hxx) already provide default implementations of all virtual functions, but a derived class is free to implement them as well.

Relation (class)

Holds three string data for ID, Type and Target (need more info).

AddressConverter (class)

converts strings to addresses and ranges, and tracks invalid addresses (e.g. a not-importable cell at address ZZZ1000000). Later, this information will be used to generate a "Imported document contains data outside of sheet limits" warning box after loading. Header: addressconverter.hxx

UnitConverter (class)

provides basic unit conversion, including font dependent stuff such as calculating column width from a specific number of characters. Header: unitconverter.hxx

Development

Feature Developer Status Comments/Missing
Framework, fragment handling cl/dr done
Workbook, worksheet fragment, sheet names tbe/dr done
Simple cell contents (values, strings) dr done missing Calc feature: error cells
Shared strings fragment dr done
Styles fragment dr done
Simple cell formatting (alignment, protection, borders, fill) dr done
Builtin number formats dr done
Font handling for cells tbe/dr done
Cell styles (names, formatting) dr done
Column settings (format, width, outlines) tbe/dr/kohei done missing Calc feature: outline symbol position
Row settings (format, height, outlines) tbe/dr/kohei done missing Calc feature: outline symbol position
Rich text in cells dr done
Scheme fragment, scheme colors dr done
Cell formulas, array formulas, shared formulas dr done
Conditional formatting dr done
External references/links dr/er in progress missing: mapping of Excel's '[n]Sheet' syntax to Calc sheet names
Print ranges, builtin defined names dr done
Page/print settings dr done
Header/footer kohei/dr done
Column/row breaks kohei done
Sheet/document view settings dr done
Cell hyperlinks dr/kohei done because a cell hyperlink is a in-cell text field object in Calc, as opposed to a cell property in Excel, a hyperlink will not get imported when the cell is a value cell.
Label ranges [reference] skipped MS Office 2007 has dropped support for this feature.
Data validation dr done
Web queries kohei formula parser
Pivot tables kohei in progress The geometry of Calc's data pilot and Excel's pivot table objects are different, so some overlapping problems are expected.
Drawing objects sj/dr done
Charts - chart types dr done Supported types: bar/column 2D/3D clustered/stacked/percent, line 2D/3D standard/stacked/percent, area 2D/3D, stock (4 types), line radar, pie 2D/3D, doughnut. Missing: filled radar, surface.
Charts - titles/legend dr in progress missing: legend settings
Charts - series settings dr done Including: sheet source links, primary/secondary axes, error bars, trendlines.
Charts - formatting dr in progress Including: manual formatting, chart styles.
OLE objects, form controls sj/dr not started
Auto filter, user filter kohei in progress Some incompatible bits present, so we have to be creative in importing Excel's autofilter items.
Scenarios dr not started
Change tracking dr not started

External References

  • xl/workbook.xml contains externalReference tags that references relationship items in xl/_rels/workbook.xml.rels by rID.
  • xl/_rels/workbook.xml.rels contains path(s) to external link fragments, that are usually xl/externalLinks/externalLink*.xml.
  • xl/externalLinks/externalLink*.xml contains summary content information of the external book being referenced. It should provide enough information to display external cell's content. It also contains reference to the path to the external file by rID.
  • xl/externalLinks/_rels/externalLink*.xml.rels contains the path to the external file, stored by rID.

Inside a cell that contains an external cell reference, the format is given

 <f>[1]Sheet1!A1</f>

as formula.

Data Validations

The UNO API for data validation is here.

ScConditionEntry needs new methods to allow setting pFormula1 and pFormula2 that are of type ScTokenArray. ScValidationData is an immediate child class of ScConditionEntry, and it represents the data validation attribute of a cell.

ScTableValidationObj implements UNO wrapper for ScValidationData, and this class needs to implement the css::sheet::XFormulaTokens interface so that the client code can pass formula tokens to this class, and that formula token set needs to eventually find its way into ScValidationData as ScTokenArray.

Web Queries

There are 5 relevant sections to touch.

  • /xl/worksheets/_rels/sheet?.xml.rels - contains the paths to query table properties fragments.
  • /xl/queryTables/queryTable?.xml - contains query table properties, especially their name and associated connection IDs.
  • /xl/_rels/workbook.xml.rels - contains the path to connections properties fragment.
  • /xl/connections.xml - contains connections properties, such as URL.
  • /xl/workbook.xml - defined names are used to associate a cell range to a query table by name.

The import strategy is to:

  1. Store the query table properties into query table container, with name as a key.
  2. Store the connections properties into connection container, with connection ID as a key.
  3. Parse a defined name and see if a query table for the same name exists.
  4. If yes, get the connection ID from the query table list.
  5. Get the connection properties by that ID from the connection list.
  6. Connect to the URL for data retrieval.

Pivot Table

  • /xl/workbook.xml - contains the IDs of pivot table caches and their relation IDs.
  • /xl/_rels/workbook.xml.rels - contains paths to pivot table cache fragments and their IDs.
  • /xl/worksheets/_rels/sheet?.xml.rels - contains path to pivot table xml fragment(s). Note that the source sheet fragment does not contain the relation ID to such pivot table fragment.
  • /xl/pivotTables/pivotTable?.xml - contains the pivot table definitions.
  • /xl/pivotTables/_rels/pivotTable?.xml.rels - contains the path to pivot table cache.
  • /xl/pivotCache/pivotCacheDefinition?.xml - pivot cache definitions.
  • /xl/pivotCache/_rels/pivotCacheDefinition?.xml.rels - contains the path to pivot cache records.
  • /xl/pirovtCache/pivotCacheRecords?.xml - pivot cache records.

See Also

  • sc module page for the implementation of Calc core.
Personal tools