Creating xlsx files from xml template using openpyxl.

This project is intended to create xlsx files from xml api to openpyxl, supposedly generated by other tamplate engines (i.e. django, jinja).

This is a merely an xml parser translating mostly linearly to worksheet, rows and finally cells of the Excel workbook.

An xml file like this one

<workbook>
    <worksheet title="test">
        <row><cell>This</cell><cell>is</cell><cell>a TEST</cell></row>
        <row><cell>Nice, isn't it?</cell></row>
    </worksheet>
</workbook>

can be parsed to create a neat Excel workbook with two rows of data in one worksheet. Parsing can be done using command line (provided that you have your system paths set correctly:

xml2xlsx < input.xml > output.xml

or as a library call

from xml2xlsx import xml2xlsx
template = '<sheet title="test"></sheet>'
f = open('test.xlsx', 'wb')
f.write(xml2xlsx(template))
f.close()

This is mainly intended (and was developed for this purpose) to parse files generated by other templating engines, like django template system. One can generate an excel workbook from template like this:

{% for e in list %}
    <row><cell>{{ e.name }}</cell></row>
{% endfor %}

Basic features of the library include creating multiple, named sheets within one workbook and creating rows of cells in these sheets. However, there are more possibiliteis to create complex excel based reports.

Each cell can be specified to use one of the types:

• string (default)
• number
• date

Type is defined in type cell attribute. The cell value is converted appropriately to the type specified. If you insert a number in the cell value and do not specify type="number" attribute, you will find Excel complaining about storing nubers as text.

Since there are more date formats than countries, you have to be aware of current locale. The simplest way to be i18n compatible is to specify date format in date-fmt attribute and pass compatible (possibily non localized) date in the cell value, as in the following example

...
<row><cell type="date" date-fmt="%Y-%m-%d">2016-10-01</cell></row>
<row><cell type="date" date-fmt="%d.%m.%Y">01.10.2016</cell></row>
...

Generated excel file will have two rows with the same date (1st of October 2016) with date formatted according to Excel defaults (and current locale).

Columns can be tackled only in a limited way, i.e. only column widths can be changed. Column properties are defined in columns tag as one or more child of the sheet tag. It is possible to specify a range of columns using start and end atrributes. For example:

...
<sheet title="test">
    <columns start="A" end="D" width="123"/>
    <row><cell>Test</cell></row>
</sheet>
...

xml2xls can effectively create cells with formulas in them. The only limitation (as with openpyxl) is using English names of the functions.

For example:

...
<row><cell>=SUM(A1:A5)</cell></row>
...

The parser can store positions of the cell in a dictionary-like structure. It then can be referenced to create complex formulas. Each value of the cell is preprocessed using string format with stored values. This means that these values can be referenced using { and } brackets.

There are two basic values that can always be used, i.e. row and col which return current row number and column name.

<workbook>
    <sheet>
        <row><cell>{col}{row}</cell></row>
    </sheet>
</workbook>
...

would create a workbook with a text "A1" included in the A1 cell of the worksheet. Using template languages, you can create more complicated constructs, like (using django template system):

...
{% for e in list %}
<row>
    <cell type="date" date-fmt="%Y-%m-%d">{{ e|date:"Y-m-d" }}</cell>
    <cell>=TEXT(A{row}, "ddd")</cell>
</row>
{% endfor %}
...

would create a list of rows with a date in the first column and weekday names for these dates in the second column (provided list context variable contains a list of dates).

It is also possible to store cell possible to store names of specified cells in a pseudo-variable (as in a dictionary). One has to use ref-id attribute of the cell tag and then reuse the value of this attribute in the remainder of the xml input. This is very useful in formulas. A simple example would be referencing another cell in a formula like this:

...
<row><cell ref-id="mycell">This is just a test</cell></row>
...
<row><cell>={mycell}</cell></row>
...

which would create an excel formula referencing a cell with "this is just a test" text, whatever this cell address was.

A more complex example using django template engine to create summaries can look like this:

...
{% for e in list %}
    <row>
        <cell ref-id="{% if forloop.first %}start{% elsif forloop.last %}end{% endif %}">
            {{ e }}
        </cell>
    </row>
{% endfor %}
<row>
    <cell>Summary</cell>
    <cell>=SUM({start}:{end})</cell>
</row>
...

Referencing a single cell can be harsh when dealing with complex reports. Especially when creating summaries of irregularly sheet-distributed data. xml2xlsx can append a cell to a variable-like list, as in ref-id attribute, to reuse it as a comma concatenated value. Instead of ref-id, one has to use ref-append attribute.

This is a simple example to demonstrate the feature:

...
<sheet>
    <row>
        <cell ref-append="mylist">1</cell>
        <cell ref-append="mylist">2</cell>
    </row>
    <row><cell ref-append="mylist">3</cell></row>
    <row><cell>=SUM({mylist})</cell></row>
</sheet>

This will generate an Excel sheet with A3 cell containing formula to sum A1, B1 and A2 cells (=SUM(A1, B1, A2)).

It is perfectly possible to reference a cell in another sheet with both ref-id and ref-append. However, there is a limitation to that. Since xml2xslx is a linear parser, you are only allowed to reference already parsed elements. This means, you have to create sheets in a proper order (sheets referencing other sheets must be created after referenced cells are parsed).

The following example will not work:

...
<sheet title="one">
    <row><cell>{mycell}</cell></row>
</sheet>
<sheet title="two">
    <row><cell ref-id="mycell">XYZ</cell></row>
</sheet>
...

However, it is possible to make this exmaple work and retain the same worksheet ordering using index attribute:

...
<sheet title="two">
    <row><cell ref-id="mycell">XYZ</cell></row>
</sheet>
<sheet title="one" index="0">
    <row><cell>{mycell}</cell></row>
</sheet>
...

The cell format can be specified using various attributes of the cell tag. Only font formatting can be specifed for now.

A font format is specified in in font attribute. It is a semicolon separated dict like list of font formats as specified in font class of openpyxl library.

An example to create a cell with bold 10px font:

...
<cell font="bold: True; size: 10px;">Cell formatted</cell>
...

Here is the (probably incomplete) wishlist for the project

• Global font and cell styles
• Row widths and column heights
• Horizontal and vertical cell merging
• XML validation with XSD to quickly raise an error if parsing wrong xml

Parsed xml should be enclosed in a workbook tag. Each workbook tag can have multiple sheet. The hierarchy continues to row and cell tags.

Here is a complete list of available attributes of these tags.

No attributes for now.

No attributes for now

• Added documentation
• Added cell referencing with inter-sheet possibility
• Changed sheet title attribute from name to title
• Added possibility to set index for a sheet