* Extract character set, so we can convert to UTF-8 if required
* Set column width and row height when defined on tr/td
* Parse align and valign on td
* Specify number format of cell via html attribute
* Formatting of b, strong, i and em tags
* Inserting image in cell when using img tag in html
* Add applying inline styles: border, fonts, alignment, dimensions
* Add tests for applying inline styles
`master` is now the only permanent branch. Features and fixes should be merged
into `master` when stable. Pull requests should be forked from `master`.
`develop` branch disappear entirely in favor of temporary features/fixes branches.
In case we generate Spreadsheet from html file and the code
in file have text color in css "color:#FF00FF" it will showing
as black color because it will render like rgb content with } "FF00FF}"
So, we fix it by adding missing bracket "{".
Closes#831
Using an operator is significantly faster than calling the max function.
As this method is called more than once per cell the difference adds up.
Closes#824
Removing the duplicate strtoupper call has a meaningful impact on
performance since this method is called at least once per cell.
`Worksheet::getCells` currently calls `strtoupper` twice. `strtoupper`
is kind of expensive and this method is called at least once for every
cell in the spreadsheet. By removing the unnecessary second call the
runtime decreases by 18% when importing a ~100K cell spreadsheet.
Closes#825
For large XLSX files `Reader/Xlsx::readColumnsAndRowsAttributes()` performs
a lot of calls to `$this->getReadFilter()` and `$this->getReadFilter()->readCell()`
as `readCell()` is called twice for each (possibbly filled) cell.
By ignoring calls to the DefaultReadFilter implementation (which always returns true),
using no custom read filter will not incur any runtime penalty.
The runtime penaltiy when using a custom read filter is reduced by a third by
caching the read filter into a variable instead of using the getter method.
Fixes issue #772.
Due to a limitation in Mpdf, the HTML string passed to its WriteHTML method
must not exceed a particular length. PhpSpreadsheet produces one HTML string
containing all spreadsheet data when writing to HTML, which can easily exceed
Mpdf's size limit. Thus, it was impossible to write large spreadsheets to PDF
using the Mpdf writer - this change fixes that issue.
Fixes#637Fixes#706
Commit 8dddf56 inadvertently removed the ability to omit the width
and height arguments to the OFFSET function. And #REF! is returned
because the function is validating that the new $pCell argument
is present. It is present, but it has been passed in the $height position.
We fixed this by always passing $pCell at the last position and filling
missing arguments with NULL values.
Fixes#561Fixes#565
When extracting sheet title from string reference (like `"Work!sheet1!A1"`), PHP function `explode()` divide this string into three parts: `['Work', 'sheet1', 'A1']`. And then these wrong values are used in formulas, ranges, etc.
This change fix that problem by using special function `Worksheet::extractSheetTitle()`. This function also has been changed to make sure that worksheet title can contain "!" character. So, that function search last position of "!" in reference string and divide it to 2 parts correctly: `['Work!sheet1', 'A1']`.
Fixes#325Fixes#662
When a formatting string has a locale in it an error can occur when outputting. For example when the format string with a locale such as `[$-1010409]#,##0.00;-#,##0.00` appears, a value of 9.98 comes back as $9.98. This is because at https://github.com/PHPOffice/PhpSpreadsheet/blob/1.4.0/src/PhpSpreadsheet/Style/NumberFormat.php#L711 the numberFormat regex will match to the zeros inside the locale ([$-1010409]). Attempts to adjust the numberFormat regex caused regressions in other tests. Adding another step to filter out the locale caused no regression.
Iterators prev() behavior is now consistent with next(), meaning
that it can go out of bounds and it must be validated with valid()
before using it.
Fixes#587Fixes#627
Rowspans/colspans are now respected for each HTML document added to an existing
spreadsheet as a new worksheet. The protected $rowspan class property should
be emptied on each call to `loadIntoExisting`.
Fixes#619Fixes#620
Fräntz Miccoli