Compare commits

..
18 Commits
Author SHA1 Message Date
Adrien Loison 3a330debb3 Move ReaderCommonOptions class to Common folder (#343) 2016-10-18 16:55:05 -07:00
Adrien Loison a19231fb68 Introduce XMLProcessor to reduce ODS,XLSX readers' complexity (#342) 2016-10-18 16:28:26 -07:00
Adrien Loison 73d5d0ea17 Remove text suffix in XLSX date formats (#341)
Some date formats have a text suffix, e.g. "mm/dd/yy;@". We should remove the ";...@" part.
2016-10-18 11:55:36 -07:00
Adrien Loison 687c321363 Refactor SharedStringsHelper::extractSharedStrings (#340) 2016-10-18 00:03:15 -07:00
Adrien Loison 2fa01cd838 Remove unused SimpleXMLElement::children() method (#339) 2016-10-17 22:49:37 -07:00
Adrien Loison 752f4bf64e Add ReaderOptions for all readers (#338)
Instead of passing every single option down the chain
2016-10-17 22:41:36 -07:00
Adrien Loison b61323d7d2 Remove XLSX references in ODS Writer (#337) 2016-10-17 11:54:49 -07:00
Adrien Loison 179ab483d6 Empty rows do not need to be written for XLSX files (#336) 2016-10-17 11:40:52 -07:00
Adrien Loison 2fafb63115 ODS Reader should support num-rows-repeated for non empty rows (#335) 2016-10-17 10:51:12 -07:00
Adrien Loison 5ef5647558 Make getConcreteSheetIterator() protected (#334) 2016-10-17 10:26:37 -07:00
Adrien Loison 0978d340f0 Option to keep empty rows (#331)
* Add option to preserve empty rows when reading an XLSX file
* Add option to preserve empty rows when reading a CSV file
* Add option to preserve empty rows when reading an ODS file
2016-10-17 10:20:02 -07:00
Adrien Loison 77178122c3 Fix output file deletion after exception thrown on write (#328)
For relative paths, it would not work as the FileSystemHelper would not allow deleting a file that's not part of its base folder.
2016-10-12 11:15:07 -07:00
Adrien Loison 23f8cc4f05 Temp files should be deleted when an exception is thrown (#327)
If an exception is thrown while writing data, instead of letting the developer handle this situation gracefully, Spout can attempt to delete all the temporary files that were created so far, as well as the output file as it won't be completed and therefore corrupted.
2016-10-11 15:08:37 -07:00
Adrien Loison 442a9837f1 Removing license badge (#320)
As Github now provides this information in the project's header
2016-09-30 10:11:05 -07:00
Adrien Loison cc07072cbb Better support for Date custom format (#316)
- To determine if a style should apply a date format, the presence of "applyNumberFormat" attribute on the "cellXfs" section of styles.xml is now optional. We only look at the "numFmtId" attribute (but early return if "applyNumberFormat" is set to "0").
- The format code can contain lowercase AND now uppercase characters as its pattern.
- "General" format code used as a custom format is now supported. It seems to be used by a bunch of programs...
2016-09-24 10:46:42 -07:00
Hastegan 30aa1b87e2 Fix boolean notation in PHPDoc (#314) 2016-09-16 15:12:58 -07:00
Hastegan ddb7365a79 Add a way to disable text wrapping through the api (Fix #247) (#311)
* Add a way to disable text wrapping (Fix #247)
* Fix PHPDoc boolean notation
* Fix PHPDoc param notation
2016-09-16 11:24:36 -07:00
Adrien Loison a07a96f523 Bump dev-master version (#309)
Bump to `2.7.x-dev` as 2.6.0 just got released
2016-09-08 09:34:43 -07:00
55 changed files with 1559 additions and 464 deletions

No files matched your search

+3 -4
View File
@@ -5,7 +5,6 @@
[![Build Status](https://travis-ci.org/box/spout.svg?branch=master)](https://travis-ci.org/box/spout)
[![Code Coverage](https://scrutinizer-ci.com/g/box/spout/badges/coverage.png?b=master)](https://scrutinizer-ci.com/g/box/spout/?branch=master)
[![Total Downloads](https://poser.pugx.org/box/spout/downloads)](https://packagist.org/packages/box/spout)
[![License](https://poser.pugx.org/box/spout/license)](https://packagist.org/packages/box/spout)
Spout is a PHP library to read and write spreadsheet files (CSV, XLSX and ODS), in a fast and scalable way.
Contrary to other file readers or writers, it is capable of processing very large files while keeping the memory usage really low (less than 3MB).
@@ -209,7 +208,7 @@ Font | Bold | `StyleBuilder::setFontBold()`
| Font name | `StyleBuilder::setFontName('Arial')`
| Font size | `StyleBuilder::setFontSize(14)`
| Font color | `StyleBuilder::setFontColor(Color::BLUE)`<br>`StyleBuilder::setFontColor(Color::rgb(0, 128, 255))`
Alignment | Wrap text | `StyleBuilder::setShouldWrapText()`
Alignment | Wrap text | `StyleBuilder::setShouldWrapText(true|false)`
#### New sheet creation
@@ -259,7 +258,7 @@ $writer->setShouldUseInlineStrings(false); // will use shared strings
When reading a spreadsheet containing dates or times, Spout returns the values by default as DateTime objects.
It is possible to change this behavior and have a formatted date returned instead (e.g. "2016-11-29 1:22 AM"). The format of the date corresponds to what is specified in the spreadsheet.
```php
use Box\Spout\Reader\ReaderFactory;
use Box\Spout\Common\Type;
@@ -302,7 +301,7 @@ $sheetName = $sheet->getName();
// Customizing the sheet name when writing
$sheet = $writer->getCurrentSheet();
$sheet->setName('My custom name');
```
```
> Please note that Excel has some restrictions on the sheet's name:
> * it must not be blank
+1 -1
View File
@@ -31,7 +31,7 @@
},
"extra": {
"branch-alias": {
"dev-master": "2.6.x-dev"
"dev-master": "2.7.x-dev"
}
}
+6 -5
View File
@@ -13,15 +13,15 @@ use Box\Spout\Common\Exception\IOException;
*/
class FileSystemHelper
{
/** @var string Path of the base folder where all the I/O can occur */
protected $baseFolderPath;
/** @var string Real path of the base folder where all the I/O can occur */
protected $baseFolderRealPath;
/**
* @param string $baseFolderPath The path of the base folder where all the I/O can occur
*/
public function __construct($baseFolderPath)
{
$this->baseFolderPath = $baseFolderPath;
$this->baseFolderRealPath = realpath($baseFolderPath);
}
/**
@@ -124,9 +124,10 @@ class FileSystemHelper
*/
protected function throwIfOperationNotInBaseFolder($operationFolderPath)
{
$isInBaseFolder = (strpos($operationFolderPath, $this->baseFolderPath) === 0);
$operationFolderRealPath = realpath($operationFolderPath);
$isInBaseFolder = (strpos($operationFolderRealPath, $this->baseFolderRealPath) === 0);
if (!$isInBaseFolder) {
throw new IOException("Cannot perform I/O operation outside of the base folder: {$this->baseFolderPath}");
throw new IOException("Cannot perform I/O operation outside of the base folder: {$this->baseFolderRealPath}");
}
}
}
+25 -4
View File
@@ -19,8 +19,15 @@ abstract class AbstractReader implements ReaderInterface
/** @var \Box\Spout\Common\Helper\GlobalFunctionsHelper Helper to work with global functions */
protected $globalFunctionsHelper;
/** @var bool Whether date/time values should be returned as PHP objects or be formatted as strings */
protected $shouldFormatDates = false;
/** @var \Box\Spout\Reader\Common\ReaderOptions Reader's customized options */
protected $options;
/**
* Returns the reader's current options
*
* @return \Box\Spout\Reader\Common\ReaderOptions
*/
abstract protected function getOptions();
/**
* Returns whether stream wrappers are supported
@@ -42,7 +49,7 @@ abstract class AbstractReader implements ReaderInterface
*
* @return \Iterator To iterate over sheets
*/
abstract public function getConcreteSheetIterator();
abstract protected function getConcreteSheetIterator();
/**
* Closes the reader. To be used after reading the file.
@@ -64,12 +71,26 @@ abstract class AbstractReader implements ReaderInterface
/**
* Sets whether date/time values should be returned as PHP objects or be formatted as strings.
*
* @api
* @param bool $shouldFormatDates
* @return AbstractReader
*/
public function setShouldFormatDates($shouldFormatDates)
{
$this->shouldFormatDates = $shouldFormatDates;
$this->getOptions()->setShouldFormatDates($shouldFormatDates);
return $this;
}
/**
* Sets whether empty rows should be returned or skipped.
*
* @api
* @param bool $shouldPreserveEmptyRows
* @return AbstractReader
*/
public function setShouldPreserveEmptyRows($shouldPreserveEmptyRows)
{
$this->getOptions()->setShouldPreserveEmptyRows($shouldPreserveEmptyRows);
return $this;
}
+22 -25
View File
@@ -4,7 +4,6 @@ namespace Box\Spout\Reader\CSV;
use Box\Spout\Reader\AbstractReader;
use Box\Spout\Common\Exception\IOException;
use Box\Spout\Common\Helper\EncodingHelper;
/**
* Class Reader
@@ -20,20 +19,21 @@ class Reader extends AbstractReader
/** @var SheetIterator To iterator over the CSV unique "sheet" */
protected $sheetIterator;
/** @var string Defines the character used to delimit fields (one character only) */
protected $fieldDelimiter = ',';
/** @var string Original value for the "auto_detect_line_endings" INI value */
protected $originalAutoDetectLineEndings;
/** @var string Defines the character used to enclose fields (one character only) */
protected $fieldEnclosure = '"';
/** @var string Encoding of the CSV file to be read */
protected $encoding = EncodingHelper::ENCODING_UTF8;
/** @var string Defines the End of line */
protected $endOfLineCharacter = "\n";
/** @var string */
protected $autoDetectLineEndings;
/**
* Returns the reader's current options
*
* @return ReaderOptions
*/
protected function getOptions()
{
if (!isset($this->options)) {
$this->options = new ReaderOptions();
}
return $this->options;
}
/**
* Sets the field delimiter for the CSV.
@@ -44,7 +44,7 @@ class Reader extends AbstractReader
*/
public function setFieldDelimiter($fieldDelimiter)
{
$this->fieldDelimiter = $fieldDelimiter;
$this->getOptions()->setFieldDelimiter($fieldDelimiter);
return $this;
}
@@ -57,7 +57,7 @@ class Reader extends AbstractReader
*/
public function setFieldEnclosure($fieldEnclosure)
{
$this->fieldEnclosure = $fieldEnclosure;
$this->getOptions()->setFieldEnclosure($fieldEnclosure);
return $this;
}
@@ -70,7 +70,7 @@ class Reader extends AbstractReader
*/
public function setEncoding($encoding)
{
$this->encoding = $encoding;
$this->getOptions()->setEncoding($encoding);
return $this;
}
@@ -83,7 +83,7 @@ class Reader extends AbstractReader
*/
public function setEndOfLineCharacter($endOfLineCharacter)
{
$this->endOfLineCharacter = $endOfLineCharacter;
$this->getOptions()->setEndOfLineCharacter($endOfLineCharacter);
return $this;
}
@@ -107,7 +107,7 @@ class Reader extends AbstractReader
*/
protected function openReader($filePath)
{
$this->autoDetectLineEndings = ini_get('auto_detect_line_endings');
$this->originalAutoDetectLineEndings = ini_get('auto_detect_line_endings');
ini_set('auto_detect_line_endings', '1');
$this->filePointer = $this->globalFunctionsHelper->fopen($filePath, 'r');
@@ -117,10 +117,7 @@ class Reader extends AbstractReader
$this->sheetIterator = new SheetIterator(
$this->filePointer,
$this->fieldDelimiter,
$this->fieldEnclosure,
$this->encoding,
$this->endOfLineCharacter,
$this->getOptions(),
$this->globalFunctionsHelper
);
}
@@ -130,7 +127,7 @@ class Reader extends AbstractReader
*
* @return SheetIterator To iterate over sheets
*/
public function getConcreteSheetIterator()
protected function getConcreteSheetIterator()
{
return $this->sheetIterator;
}
@@ -147,6 +144,6 @@ class Reader extends AbstractReader
$this->globalFunctionsHelper->fclose($this->filePointer);
}
ini_set('auto_detect_line_endings', $this->autoDetectLineEndings);
ini_set('auto_detect_line_endings', $this->originalAutoDetectLineEndings);
}
}
+110
View File
@@ -0,0 +1,110 @@
<?php
namespace Box\Spout\Reader\CSV;
use Box\Spout\Common\Helper\EncodingHelper;
/**
* Class ReaderOptions
* This class is used to customize the reader's behavior
*
* @package Box\Spout\Reader\CSV
*/
class ReaderOptions extends \Box\Spout\Reader\Common\ReaderOptions
{
/** @var string Defines the character used to delimit fields (one character only) */
protected $fieldDelimiter = ',';
/** @var string Defines the character used to enclose fields (one character only) */
protected $fieldEnclosure = '"';
/** @var string Encoding of the CSV file to be read */
protected $encoding = EncodingHelper::ENCODING_UTF8;
/** @var string Defines the End of line */
protected $endOfLineCharacter = "\n";
/**
* @return string
*/
public function getFieldDelimiter()
{
return $this->fieldDelimiter;
}
/**
* Sets the field delimiter for the CSV.
* Needs to be called before opening the reader.
*
* @param string $fieldDelimiter Character that delimits fields
* @return ReaderOptions
*/
public function setFieldDelimiter($fieldDelimiter)
{
$this->fieldDelimiter = $fieldDelimiter;
return $this;
}
/**
* @return string
*/
public function getFieldEnclosure()
{
return $this->fieldEnclosure;
}
/**
* Sets the field enclosure for the CSV.
* Needs to be called before opening the reader.
*
* @param string $fieldEnclosure Character that enclose fields
* @return ReaderOptions
*/
public function setFieldEnclosure($fieldEnclosure)
{
$this->fieldEnclosure = $fieldEnclosure;
return $this;
}
/**
* @return string
*/
public function getEncoding()
{
return $this->encoding;
}
/**
* Sets the encoding of the CSV file to be read.
* Needs to be called before opening the reader.
*
* @param string $encoding Encoding of the CSV file to be read
* @return ReaderOptions
*/
public function setEncoding($encoding)
{
$this->encoding = $encoding;
return $this;
}
/**
* @return string EOL for the CSV
*/
public function getEndOfLineCharacter()
{
return $this->endOfLineCharacter;
}
/**
* Sets the EOL for the CSV.
* Needs to be called before opening the reader.
*
* @param string $endOfLineCharacter used to properly get lines from the CSV file.
* @return ReaderOptions
*/
public function setEndOfLineCharacter($endOfLineCharacter)
{
$this->endOfLineCharacter = $endOfLineCharacter;
return $this;
}
}
+46 -22
View File
@@ -40,6 +40,12 @@ class RowIterator implements IteratorInterface
/** @var string Encoding of the CSV file to be read */
protected $encoding;
/** @var string End of line delimiter, given by the user as input. */
protected $inputEOLDelimiter;
/** @var bool Whether empty rows should be returned or skipped */
protected $shouldPreserveEmptyRows;
/** @var \Box\Spout\Common\Helper\GlobalFunctionsHelper Helper to work with global functions */
protected $globalFunctionsHelper;
@@ -49,24 +55,19 @@ class RowIterator implements IteratorInterface
/** @var string End of line delimiter, encoded using the same encoding as the CSV */
protected $encodedEOLDelimiter;
/** @var string End of line delimiter, given by the user as input. */
protected $inputEOLDelimiter;
/**
* @param resource $filePointer Pointer to the CSV file to read
* @param string $fieldDelimiter Character that delimits fields
* @param string $fieldEnclosure Character that enclose fields
* @param string $encoding Encoding of the CSV file to be read
* @param string $endOfLineDelimiter End of line delimiter
* @param \Box\Spout\Reader\CSV\ReaderOptions $options
* @param \Box\Spout\Common\Helper\GlobalFunctionsHelper $globalFunctionsHelper
*/
public function __construct($filePointer, $fieldDelimiter, $fieldEnclosure, $encoding, $endOfLineDelimiter, $globalFunctionsHelper)
public function __construct($filePointer, $options, $globalFunctionsHelper)
{
$this->filePointer = $filePointer;
$this->fieldDelimiter = $fieldDelimiter;
$this->fieldEnclosure = $fieldEnclosure;
$this->encoding = $encoding;
$this->inputEOLDelimiter = $endOfLineDelimiter;
$this->fieldDelimiter = $options->getFieldDelimiter();
$this->fieldEnclosure = $options->getFieldEnclosure();
$this->encoding = $options->getEncoding();
$this->inputEOLDelimiter = $options->getEndOfLineCharacter();
$this->shouldPreserveEmptyRows = $options->shouldPreserveEmptyRows();
$this->globalFunctionsHelper = $globalFunctionsHelper;
$this->encodingHelper = new EncodingHelper($globalFunctionsHelper);
@@ -106,7 +107,7 @@ class RowIterator implements IteratorInterface
* Checks if current position is valid
* @link http://php.net/manual/en/iterator.valid.php
*
* @return boolean
* @return bool
*/
public function valid()
{
@@ -114,7 +115,7 @@ class RowIterator implements IteratorInterface
}
/**
* Move forward to next element. Empty rows are skipped.
* Move forward to next element. Reads data for the next unprocessed row.
* @link http://php.net/manual/en/iterator.next.php
*
* @return void
@@ -124,25 +125,48 @@ class RowIterator implements IteratorInterface
{
$this->hasReachedEndOfFile = $this->globalFunctionsHelper->feof($this->filePointer);
if ($this->hasReachedEndOfFile) {
return;
if (!$this->hasReachedEndOfFile) {
$this->readDataForNextRow();
}
}
/**
* @return void
* @throws \Box\Spout\Common\Exception\EncodingConversionException If unable to convert data to UTF-8
*/
protected function readDataForNextRow()
{
do {
$rowData = $this->getNextUTF8EncodedRow();
$hasNowReachedEndOfFile = $this->globalFunctionsHelper->feof($this->filePointer);
} while (($rowData === false && !$hasNowReachedEndOfFile) || $this->isEmptyLine($rowData));
} while ($this->shouldReadNextRow($rowData));
if ($rowData !== false) {
$this->rowDataBuffer = $rowData;
// str_replace will replace NULL values by empty strings
$this->rowDataBuffer = str_replace(null, null, $rowData);
$this->numReadRows++;
} else {
// If we reach this point, it means end of file was reached.
// This happens when the last lines are empty lines.
$this->hasReachedEndOfFile = $hasNowReachedEndOfFile;
$this->hasReachedEndOfFile = true;
}
}
/**
* @param array|bool $currentRowData
* @return bool Whether the data for the current row can be returned or if we need to keep reading
*/
protected function shouldReadNextRow($currentRowData)
{
$hasSuccessfullyFetchedRowData = ($currentRowData !== false);
$hasNowReachedEndOfFile = $this->globalFunctionsHelper->feof($this->filePointer);
$isEmptyLine = $this->isEmptyLine($currentRowData);
return (
(!$hasSuccessfullyFetchedRowData && !$hasNowReachedEndOfFile) ||
(!$this->shouldPreserveEmptyRows && $isEmptyLine)
);
}
/**
* Returns the next row, converted if necessary to UTF-8.
* As fgetcsv() does not manage correctly encoding for non UTF-8 data,
@@ -154,7 +178,7 @@ class RowIterator implements IteratorInterface
protected function getNextUTF8EncodedRow()
{
$encodedRowData = $this->globalFunctionsHelper->fgetcsv($this->filePointer, self::MAX_READ_BYTES_PER_LINE, $this->fieldDelimiter, $this->fieldEnclosure);
if (false === $encodedRowData) {
if ($encodedRowData === false) {
return false;
}
@@ -195,7 +219,7 @@ class RowIterator implements IteratorInterface
}
/**
* @param array $lineData Array containing the cells value for the line
* @param array|bool $lineData Array containing the cells value for the line
* @return bool Whether the given line is empty
*/
protected function isEmptyLine($lineData)
+3 -5
View File
@@ -16,14 +16,12 @@ class Sheet implements SheetInterface
/**
* @param resource $filePointer Pointer to the CSV file to read
* @param string $fieldDelimiter Character that delimits fields
* @param string $fieldEnclosure Character that enclose fields
* @param string $encoding Encoding of the CSV file to be read
* @param \Box\Spout\Reader\CSV\ReaderOptions $options
* @param \Box\Spout\Common\Helper\GlobalFunctionsHelper $globalFunctionsHelper
*/
public function __construct($filePointer, $fieldDelimiter, $fieldEnclosure, $encoding, $endOfLineCharacter, $globalFunctionsHelper)
public function __construct($filePointer, $options, $globalFunctionsHelper)
{
$this->rowIterator = new RowIterator($filePointer, $fieldDelimiter, $fieldEnclosure, $encoding, $endOfLineCharacter, $globalFunctionsHelper);
$this->rowIterator = new RowIterator($filePointer, $options, $globalFunctionsHelper);
}
/**
+4 -6
View File
@@ -20,14 +20,12 @@ class SheetIterator implements IteratorInterface
/**
* @param resource $filePointer
* @param string $fieldDelimiter Character that delimits fields
* @param string $fieldEnclosure Character that enclose fields
* @param string $encoding Encoding of the CSV file to be read
* @param \Box\Spout\Reader\CSV\ReaderOptions $options
* @param \Box\Spout\Common\Helper\GlobalFunctionsHelper $globalFunctionsHelper
*/
public function __construct($filePointer, $fieldDelimiter, $fieldEnclosure, $encoding, $endOfLineCharacter, $globalFunctionsHelper)
public function __construct($filePointer, $options, $globalFunctionsHelper)
{
$this->sheet = new Sheet($filePointer, $fieldDelimiter, $fieldEnclosure, $encoding, $endOfLineCharacter, $globalFunctionsHelper);
$this->sheet = new Sheet($filePointer, $options, $globalFunctionsHelper);
}
/**
@@ -45,7 +43,7 @@ class SheetIterator implements IteratorInterface
* Checks if current position is valid
* @link http://php.net/manual/en/iterator.valid.php
*
* @return boolean
* @return bool
*/
public function valid()
{
+58
View File
@@ -0,0 +1,58 @@
<?php
namespace Box\Spout\Reader\Common;
/**
* Class ReaderOptions
* Readers' common options
*
* @package Box\Spout\Reader\Common
*/
class ReaderOptions
{
/** @var bool Whether date/time values should be returned as PHP objects or be formatted as strings */
protected $shouldFormatDates = false;
/** @var bool Whether empty rows should be returned or skipped */
protected $shouldPreserveEmptyRows = false;
/**
* @return bool Whether date/time values should be returned as PHP objects or be formatted as strings.
*/
public function shouldFormatDates()
{
return $this->shouldFormatDates;
}
/**
* Sets whether date/time values should be returned as PHP objects or be formatted as strings.
*
* @param bool $shouldFormatDates
* @return ReaderOptions
*/
public function setShouldFormatDates($shouldFormatDates)
{
$this->shouldFormatDates = $shouldFormatDates;
return $this;
}
/**
* @return bool Whether empty rows should be returned or skipped.
*/
public function shouldPreserveEmptyRows()
{
return $this->shouldPreserveEmptyRows;
}
/**
* Sets whether empty rows should be returned or skipped.
*
* @param bool $shouldPreserveEmptyRows
* @return ReaderOptions
*/
public function setShouldPreserveEmptyRows($shouldPreserveEmptyRows)
{
$this->shouldPreserveEmptyRows = $shouldPreserveEmptyRows;
return $this;
}
}
+152
View File
@@ -0,0 +1,152 @@
<?php
namespace Box\Spout\Reader\Common;
use Box\Spout\Reader\Wrapper\XMLReader;
/**
* Class XMLProcessor
* Helps process XML files
*
* @package Box\Spout\Reader\Common
*/
class XMLProcessor
{
/* Node types */
const NODE_TYPE_START = XMLReader::ELEMENT;
const NODE_TYPE_END = XMLReader::END_ELEMENT;
/* Keys associated to reflection attributes to invoke a callback */
const CALLBACK_REFLECTION_METHOD = 'reflectionMethod';
const CALLBACK_REFLECTION_OBJECT = 'reflectionObject';
/* Values returned by the callbacks to indicate what the processor should do next */
const PROCESSING_CONTINUE = 1;
const PROCESSING_STOP = 2;
/** @var \Box\Spout\Reader\Wrapper\XMLReader The XMLReader object that will help read sheet's XML data */
protected $xmlReader;
/** @var array Registered callbacks */
private $callbacks = [];
/**
* @param \Box\Spout\Reader\Wrapper\XMLReader $xmlReader XMLReader object
*/
public function __construct($xmlReader)
{
$this->xmlReader = $xmlReader;
}
/**
* @param string $nodeName A callback may be triggered when a node with this name is read
* @param int $nodeType Type of the node [NODE_TYPE_START || NODE_TYPE_END]
* @param callable $callback Callback to execute when the read node has the given name and type
* @return XMLProcessor
*/
public function registerCallback($nodeName, $nodeType, $callback)
{
$callbackKey = $this->getCallbackKey($nodeName, $nodeType);
$this->callbacks[$callbackKey] = $this->getInvokableCallbackData($callback);
return $this;
}
/**
* @param string $nodeName Name of the node
* @param int $nodeType Type of the node [NODE_TYPE_START || NODE_TYPE_END]
* @return string Key used to store the associated callback
*/
private function getCallbackKey($nodeName, $nodeType)
{
return "$nodeName$nodeType";
}
/**
* Because the callback can be a "protected" function, we don't want to use call_user_func() directly
* but instead invoke the callback using Reflection. This allows the invocation of "protected" functions.
* Since some functions can be called a lot, we pre-process the callback to only return the elements that
* will be needed to invoke the callback later.
*
* @param callable $callback Array reference to a callback: [OBJECT, METHOD_NAME]
* @return array Associative array containing the elements needed to invoke the callback using Reflection
*/
private function getInvokableCallbackData($callback)
{
$callbackObject = $callback[0];
$callbackMethodName = $callback[1];
$reflectionMethod = new \ReflectionMethod(get_class($callbackObject), $callbackMethodName);
$reflectionMethod->setAccessible(true);
return [
self::CALLBACK_REFLECTION_METHOD => $reflectionMethod,
self::CALLBACK_REFLECTION_OBJECT => $callbackObject,
];
}
/**
* Resumes the reading of the XML file where it was left off.
* Stops whenever a callback indicates that reading should stop or at the end of the file.
*
* @return void
* @throws \Box\Spout\Reader\Exception\XMLProcessingException
*/
public function readUntilStopped()
{
while ($this->xmlReader->read()) {
$nodeType = $this->xmlReader->nodeType;
$nodeNamePossiblyWithPrefix = $this->xmlReader->name;
$nodeNameWithoutPrefix = $this->xmlReader->localName;
$callbackData = $this->getRegisteredCallbackData($nodeNamePossiblyWithPrefix, $nodeNameWithoutPrefix, $nodeType);
if ($callbackData !== null) {
$callbackResponse = $this->invokeCallback($callbackData, [$this->xmlReader]);
if ($callbackResponse === self::PROCESSING_STOP) {
// stop reading
break;
}
}
}
}
/**
* @param string $nodeNamePossiblyWithPrefix Name of the node, possibly prefixed
* @param string $nodeNameWithoutPrefix Name of the same node, un-prefixed
* @param int $nodeType Type of the node [NODE_TYPE_START || NODE_TYPE_END]
* @return array|null Callback data to be used for execution when a node of the given name/type is read or NULL if none found
*/
private function getRegisteredCallbackData($nodeNamePossiblyWithPrefix, $nodeNameWithoutPrefix, $nodeType)
{
// With prefixed nodes, we should match if (by order of preference):
// 1. the callback was registered with the prefixed node name (e.g. "x:worksheet")
// 2. the callback was registered with the un-prefixed node name (e.g. "worksheet")
$callbackKeyForPossiblyPrefixedName = $this->getCallbackKey($nodeNamePossiblyWithPrefix, $nodeType);
$callbackKeyForUnPrefixedName = $this->getCallbackKey($nodeNameWithoutPrefix, $nodeType);
$hasPrefix = ($nodeNamePossiblyWithPrefix !== $nodeNameWithoutPrefix);
$callbackKeyToUse = $callbackKeyForUnPrefixedName;
if ($hasPrefix && isset($this->callbacks[$callbackKeyForPossiblyPrefixedName])) {
$callbackKeyToUse = $callbackKeyForPossiblyPrefixedName;
}
// Using isset here because it is way faster than array_key_exists...
return isset($this->callbacks[$callbackKeyToUse]) ? $this->callbacks[$callbackKeyToUse] : null;
}
/**
* @param array $callbackData Associative array containing data to invoke the callback using Reflection
* @param array $args Arguments to pass to the callback
* @return int Callback response
*/
private function invokeCallback($callbackData, $args)
{
$reflectionMethod = $callbackData[self::CALLBACK_REFLECTION_METHOD];
$callbackObject = $callbackData[self::CALLBACK_REFLECTION_OBJECT];
return $reflectionMethod->invokeArgs($callbackObject, $args);
}
}
+15 -2
View File
@@ -19,6 +19,19 @@ class Reader extends AbstractReader
/** @var SheetIterator To iterator over the ODS sheets */
protected $sheetIterator;
/**
* Returns the reader's current options
*
* @return ReaderOptions
*/
protected function getOptions()
{
if (!isset($this->options)) {
$this->options = new ReaderOptions();
}
return $this->options;
}
/**
* Returns whether stream wrappers are supported
*
@@ -42,7 +55,7 @@ class Reader extends AbstractReader
$this->zip = new \ZipArchive();
if ($this->zip->open($filePath) === true) {
$this->sheetIterator = new SheetIterator($filePath, $this->shouldFormatDates);
$this->sheetIterator = new SheetIterator($filePath, $this->getOptions());
} else {
throw new IOException("Could not open $filePath for reading.");
}
@@ -53,7 +66,7 @@ class Reader extends AbstractReader
*
* @return SheetIterator To iterate over sheets
*/
public function getConcreteSheetIterator()
protected function getConcreteSheetIterator()
{
return $this->sheetIterator;
}
+14
View File
@@ -0,0 +1,14 @@
<?php
namespace Box\Spout\Reader\ODS;
/**
* Class ReaderOptions
* This class is used to customize the reader's behavior
*
* @package Box\Spout\Reader\ODS
*/
class ReaderOptions extends \Box\Spout\Reader\Common\ReaderOptions
{
// No extra options
}
+193 -72
View File
@@ -8,6 +8,7 @@ use Box\Spout\Reader\Exception\XMLProcessingException;
use Box\Spout\Reader\IteratorInterface;
use Box\Spout\Reader\ODS\Helper\CellValueFormatter;
use Box\Spout\Reader\Wrapper\XMLReader;
use Box\Spout\Reader\Common\XMLProcessor;
/**
* Class RowIterator
@@ -23,19 +24,26 @@ class RowIterator implements IteratorInterface
const MAX_COLUMNS_EXCEL = 16384;
/** Definition of XML attribute used to parse data */
const XML_ATTRIBUTE_NUM_ROWS_REPEATED = 'table:number-rows-repeated';
const XML_ATTRIBUTE_NUM_COLUMNS_REPEATED = 'table:number-columns-repeated';
/** @var \Box\Spout\Reader\Wrapper\XMLReader The XMLReader object that will help read sheet's XML data */
protected $xmlReader;
/** @var \Box\Spout\Reader\Common\XMLProcessor Helper Object to process XML nodes */
protected $xmlProcessor;
/** @var bool Whether empty rows should be returned or skipped */
protected $shouldPreserveEmptyRows;
/** @var Helper\CellValueFormatter Helper to format cell values */
protected $cellValueFormatter;
/** @var bool Whether the iterator has already been rewound once */
protected $hasAlreadyBeenRewound = false;
/** @var int Number of read rows */
protected $numReadRows = 0;
/** @var array Contains the data for the currently processed row (key = cell index, value = cell value) */
protected $currentlyProcessedRowData = [];
/** @var array|null Buffer used to store the row data, while checking if there are more rows to read */
protected $rowDataBuffer = null;
@@ -43,14 +51,41 @@ class RowIterator implements IteratorInterface
/** @var bool Indicates whether all rows have been read */
protected $hasReachedEndOfFile = false;
/** @var int Last row index processed (one-based) */
protected $lastRowIndexProcessed = 0;
/** @var int Row index to be processed next (one-based) */
protected $nextRowIndexToBeProcessed = 1;
/** @var mixed|null Value of the last processed cell (because when reading cell at column N+1, cell N is processed) */
protected $lastProcessedCellValue = null;
/** @var int Number of times the last processed row should be repeated */
protected $numRowsRepeated = 1;
/** @var int Number of times the last cell value should be copied to the cells on its right */
protected $numColumnsRepeated = 1;
/** @var bool Whether at least one cell has been read for the row currently being processed */
protected $hasAlreadyReadOneCellInCurrentRow = false;
/**
* @param XMLReader $xmlReader XML Reader, positioned on the "<table:table>" element
* @param bool $shouldFormatDates Whether date/time values should be returned as PHP objects or be formatted as strings
* @param \Box\Spout\Reader\ODS\ReaderOptions $options Reader's current options
*/
public function __construct($xmlReader, $shouldFormatDates)
public function __construct($xmlReader, $options)
{
$this->xmlReader = $xmlReader;
$this->cellValueFormatter = new CellValueFormatter($shouldFormatDates);
$this->shouldPreserveEmptyRows = $options->shouldPreserveEmptyRows();
$this->cellValueFormatter = new CellValueFormatter($options->shouldFormatDates());
// Register all callbacks to process different nodes when reading the XML file
$this->xmlProcessor = new XMLProcessor($this->xmlReader);
$this->xmlProcessor->registerCallback(self::XML_NODE_ROW, XMLProcessor::NODE_TYPE_START, [$this, 'processRowStartingNode']);
$this->xmlProcessor->registerCallback(self::XML_NODE_CELL, XMLProcessor::NODE_TYPE_START, [$this, 'processCellStartingNode']);
$this->xmlProcessor->registerCallback(self::XML_NODE_ROW, XMLProcessor::NODE_TYPE_END, [$this, 'processRowEndingNode']);
$this->xmlProcessor->registerCallback(self::XML_NODE_TABLE, XMLProcessor::NODE_TYPE_END, [$this, 'processTableEndingNode']);
}
/**
@@ -71,7 +106,8 @@ class RowIterator implements IteratorInterface
}
$this->hasAlreadyBeenRewound = true;
$this->numReadRows = 0;
$this->lastRowIndexProcessed = 0;
$this->nextRowIndexToBeProcessed = 1;
$this->rowDataBuffer = null;
$this->hasReachedEndOfFile = false;
@@ -82,7 +118,7 @@ class RowIterator implements IteratorInterface
* Checks if current position is valid
* @link http://php.net/manual/en/iterator.valid.php
*
* @return boolean
* @return bool
*/
public function valid()
{
@@ -99,77 +135,155 @@ class RowIterator implements IteratorInterface
*/
public function next()
{
$rowData = [];
$cellValue = null;
$numColumnsRepeated = 1;
$numCellsRead = 0;
$hasAlreadyReadOneCell = false;
if ($this->doesNeedDataForNextRowToBeProcessed()) {
$this->readDataForNextRow();
}
$this->lastRowIndexProcessed++;
}
/**
* Returns whether we need data for the next row to be processed.
* We DO need to read data if:
* - we have not read any rows yet
* OR
* - the next row to be processed immediately follows the last read row
*
* @return bool Whether we need data for the next row to be processed.
*/
protected function doesNeedDataForNextRowToBeProcessed()
{
$hasReadAtLeastOneRow = ($this->lastRowIndexProcessed !== 0);
return (
!$hasReadAtLeastOneRow ||
$this->lastRowIndexProcessed === $this->nextRowIndexToBeProcessed - 1
);
}
/**
* @return void
* @throws \Box\Spout\Reader\Exception\SharedStringNotFoundException If a shared string was not found
* @throws \Box\Spout\Common\Exception\IOException If unable to read the sheet data XML
*/
protected function readDataForNextRow()
{
$this->currentlyProcessedRowData = [];
try {
while ($this->xmlReader->read()) {
if ($this->xmlReader->isPositionedOnStartingNode(self::XML_NODE_CELL)) {
// Start of a cell description
$currentNumColumnsRepeated = $this->getNumColumnsRepeatedForCurrentNode();
$node = $this->xmlReader->expand();
$currentCellValue = $this->getCellValue($node);
// process cell N only after having read cell N+1 (see below why)
if ($hasAlreadyReadOneCell) {
for ($i = 0; $i < $numColumnsRepeated; $i++) {
$rowData[] = $cellValue;
}
}
$cellValue = $currentCellValue;
$numColumnsRepeated = $currentNumColumnsRepeated;
$numCellsRead++;
$hasAlreadyReadOneCell = true;
} else if ($this->xmlReader->isPositionedOnEndingNode(self::XML_NODE_ROW)) {
// End of the row description
$isEmptyRow = ($numCellsRead <= 1 && $this->isEmptyCellValue($cellValue));
if ($isEmptyRow) {
// skip empty rows
$this->next();
return;
}
// Only add the value if the last read cell is not a trailing empty cell repeater in Excel.
// The current count of read columns is determined by counting the values in $rowData.
// This is to avoid creating a lot of empty cells, as Excel adds a last empty "<table:table-cell>"
// with a number-columns-repeated value equals to the number of (supported columns - used columns).
// In Excel, the number of supported columns is 16384, but we don't want to returns rows with
// always 16384 cells.
if ((count($rowData) + $numColumnsRepeated) !== self::MAX_COLUMNS_EXCEL) {
for ($i = 0; $i < $numColumnsRepeated; $i++) {
$rowData[] = $cellValue;
}
$this->numReadRows++;
}
break;
} else if ($this->xmlReader->isPositionedOnEndingNode(self::XML_NODE_TABLE)) {
// The closing "</table:table>" marks the end of the file
$this->hasReachedEndOfFile = true;
break;
}
}
$this->xmlProcessor->readUntilStopped();
} catch (XMLProcessingException $exception) {
throw new IOException("The sheet's data cannot be read. [{$exception->getMessage()}]");
}
$this->rowDataBuffer = $rowData;
$this->rowDataBuffer = $this->currentlyProcessedRowData;
}
/**
* @param \Box\Spout\Reader\Wrapper\XMLReader $xmlReader XMLReader object, positioned on a "<table:table-row>" starting node
* @return int A return code that indicates what action should the processor take next
*/
protected function processRowStartingNode($xmlReader)
{
// Reset data from current row
$this->hasAlreadyReadOneCellInCurrentRow = false;
$this->lastProcessedCellValue = null;
$this->numColumnsRepeated = 1;
$this->numRowsRepeated = $this->getNumRowsRepeatedForCurrentNode($xmlReader);
return XMLProcessor::PROCESSING_CONTINUE;
}
/**
* @param \Box\Spout\Reader\Wrapper\XMLReader $xmlReader XMLReader object, positioned on a "<table:table-cell>" starting node
* @return int A return code that indicates what action should the processor take next
*/
protected function processCellStartingNode($xmlReader)
{
$currentNumColumnsRepeated = $this->getNumColumnsRepeatedForCurrentNode($xmlReader);
$node = $xmlReader->expand();
$currentCellValue = $this->getCellValue($node);
// process cell N only after having read cell N+1 (see below why)
if ($this->hasAlreadyReadOneCellInCurrentRow) {
for ($i = 0; $i < $this->numColumnsRepeated; $i++) {
$this->currentlyProcessedRowData[] = $this->lastProcessedCellValue;
}
}
$this->hasAlreadyReadOneCellInCurrentRow = true;
$this->lastProcessedCellValue = $currentCellValue;
$this->numColumnsRepeated = $currentNumColumnsRepeated;
return XMLProcessor::PROCESSING_CONTINUE;
}
/**
* @return int A return code that indicates what action should the processor take next
*/
protected function processRowEndingNode()
{
$isEmptyRow = $this->isEmptyRow($this->currentlyProcessedRowData, $this->lastProcessedCellValue);
// if the fetched row is empty and we don't want to preserve it...
if (!$this->shouldPreserveEmptyRows && $isEmptyRow) {
// ... skip it
return XMLProcessor::PROCESSING_CONTINUE;
}
// if the row is empty, we don't want to return more than one cell
$actualNumColumnsRepeated = (!$isEmptyRow) ? $this->numColumnsRepeated : 1;
// Only add the value if the last read cell is not a trailing empty cell repeater in Excel.
// The current count of read columns is determined by counting the values in "$this->currentlyProcessedRowData".
// This is to avoid creating a lot of empty cells, as Excel adds a last empty "<table:table-cell>"
// with a number-columns-repeated value equals to the number of (supported columns - used columns).
// In Excel, the number of supported columns is 16384, but we don't want to returns rows with
// always 16384 cells.
if ((count($this->currentlyProcessedRowData) + $actualNumColumnsRepeated) !== self::MAX_COLUMNS_EXCEL) {
for ($i = 0; $i < $actualNumColumnsRepeated; $i++) {
$this->currentlyProcessedRowData[] = $this->lastProcessedCellValue;
}
}
// If we are processing row N and the row is repeated M times,
// then the next row to be processed will be row (N+M).
$this->nextRowIndexToBeProcessed += $this->numRowsRepeated;
// at this point, we have all the data we need for the row
// so that we can populate the buffer
return XMLProcessor::PROCESSING_STOP;
}
/**
* @return int A return code that indicates what action should the processor take next
*/
protected function processTableEndingNode()
{
// The closing "</table:table>" marks the end of the file
$this->hasReachedEndOfFile = true;
return XMLProcessor::PROCESSING_STOP;
}
/**
* @param \Box\Spout\Reader\Wrapper\XMLReader $xmlReader XMLReader object, positioned on a "<table:table-row>" starting node
* @return int The value of "table:number-rows-repeated" attribute of the current node, or 1 if attribute missing
*/
protected function getNumRowsRepeatedForCurrentNode($xmlReader)
{
$numRowsRepeated = $xmlReader->getAttribute(self::XML_ATTRIBUTE_NUM_ROWS_REPEATED);
return ($numRowsRepeated !== null) ? intval($numRowsRepeated) : 1;
}
/**
* @param \Box\Spout\Reader\Wrapper\XMLReader $xmlReader XMLReader object, positioned on a "<table:table-cell>" starting node
* @return int The value of "table:number-columns-repeated" attribute of the current node, or 1 if attribute missing
*/
protected function getNumColumnsRepeatedForCurrentNode()
protected function getNumColumnsRepeatedForCurrentNode($xmlReader)
{
$numColumnsRepeated = $this->xmlReader->getAttribute(self::XML_ATTRIBUTE_NUM_COLUMNS_REPEATED);
$numColumnsRepeated = $xmlReader->getAttribute(self::XML_ATTRIBUTE_NUM_COLUMNS_REPEATED);
return ($numColumnsRepeated !== null) ? intval($numColumnsRepeated) : 1;
}
@@ -185,14 +299,21 @@ class RowIterator implements IteratorInterface
}
/**
* empty() replacement that honours 0 as a valid value
* After finishing processing each cell, a row is considered empty if it contains
* no cells or if the value of the last read cell is an empty string.
* After finishing processing each cell, the last read cell is not part of the
* row data yet (as we still need to apply the "num-columns-repeated" attribute).
*
* @param string|int|float|bool|\DateTime|\DateInterval|null $value The cell value
* @return bool
* @param array $rowData
* @param string|int|float|bool|\DateTime|\DateInterval|null The value of the last read cell
* @return bool Whether the row is empty
*/
protected function isEmptyCellValue($value)
protected function isEmptyRow($rowData, $lastReadCellValue)
{
return (!isset($value) || trim($value) === '');
return (
count($rowData) === 0 &&
(!isset($lastReadCellValue) || trim($lastReadCellValue) === '')
);
}
/**
@@ -214,7 +335,7 @@ class RowIterator implements IteratorInterface
*/
public function key()
{
return $this->numReadRows;
return $this->lastRowIndexProcessed;
}
+3 -3
View File
@@ -27,13 +27,13 @@ class Sheet implements SheetInterface
/**
* @param XMLReader $xmlReader XML Reader, positioned on the "<table:table>" element
* @param bool $shouldFormatDates Whether date/time values should be returned as PHP objects or be formatted as strings
* @param int $sheetIndex Index of the sheet, based on order in the workbook (zero-based)
* @param \Box\Spout\Reader\ODS\ReaderOptions $options Reader's current options
* @param string $sheetName Name of the sheet
*/
public function __construct($xmlReader, $shouldFormatDates, $sheetIndex, $sheetName)
public function __construct($xmlReader, $sheetIndex, $sheetName, $options)
{
$this->rowIterator = new RowIterator($xmlReader, $shouldFormatDates);
$this->rowIterator = new RowIterator($xmlReader, $options);
$this->index = $sheetIndex;
$this->name = $sheetName;
}
+7 -7
View File
@@ -24,8 +24,8 @@ class SheetIterator implements IteratorInterface
/** @var string $filePath Path of the file to be read */
protected $filePath;
/** @var bool Whether date/time values should be returned as PHP objects or be formatted as strings */
protected $shouldFormatDates;
/** @var \Box\Spout\Reader\ODS\ReaderOptions Reader's current options */
protected $options;
/** @var XMLReader The XMLReader object that will help read sheet's XML data */
protected $xmlReader;
@@ -41,13 +41,13 @@ class SheetIterator implements IteratorInterface
/**
* @param string $filePath Path of the file to be read
* @param bool $shouldFormatDates Whether date/time values should be returned as PHP objects or be formatted as strings
* @param \Box\Spout\Reader\ODS\ReaderOptions $options Reader's current options
* @throws \Box\Spout\Reader\Exception\NoSheetsFoundException If there are no sheets in the file
*/
public function __construct($filePath, $shouldFormatDates)
public function __construct($filePath, $options)
{
$this->filePath = $filePath;
$this->shouldFormatDates = $shouldFormatDates;
$this->options = $options;
$this->xmlReader = new XMLReader();
/** @noinspection PhpUnnecessaryFullyQualifiedNameInspection */
@@ -83,7 +83,7 @@ class SheetIterator implements IteratorInterface
* Checks if current position is valid
* @link http://php.net/manual/en/iterator.valid.php
*
* @return boolean
* @return bool
*/
public function valid()
{
@@ -116,7 +116,7 @@ class SheetIterator implements IteratorInterface
$escapedSheetName = $this->xmlReader->getAttribute(self::XML_ATTRIBUTE_TABLE_NAME);
$sheetName = $this->escaper->unescape($escapedSheetName);
return new Sheet($this->xmlReader, $this->shouldFormatDates, $sheetName, $this->currentSheetIndex);
return new Sheet($this->xmlReader, $sheetName, $this->currentSheetIndex, $this->options);
}
/**
@@ -149,22 +149,6 @@ class SimpleXMLElement
return $doesElementExist ? $this->wrapSimpleXMLElement($realElement) : null;
}
/**
* Returns the immediate children.
*
* @return array The children
*/
public function children()
{
$children = [];
foreach ($this->simpleXMLElement->children() as $child) {
$children[] = $this->wrapSimpleXMLElement($child);
}
return $children;
}
/**
* @return string
*/
@@ -228,8 +228,8 @@ class CellValueFormatter
$dateObj->setTime($hours, $minutes, $seconds);
if ($this->shouldFormatDates) {
$styleNumberFormat = $this->styleHelper->getNumberFormat($cellStyleId);
$phpDateFormat = DateFormatHelper::toPHPDateFormat($styleNumberFormat);
$styleNumberFormatCode = $this->styleHelper->getNumberFormatCode($cellStyleId);
$phpDateFormat = DateFormatHelper::toPHPDateFormat($styleNumberFormatCode);
return $dateObj->format($phpDateFormat);
} else {
return $dateObj;
@@ -257,8 +257,8 @@ class CellValueFormatter
$dateObj->modify('+' . $secondsRemainder . 'seconds');
if ($this->shouldFormatDates) {
$styleNumberFormat = $this->styleHelper->getNumberFormat($cellStyleId);
$phpDateFormat = DateFormatHelper::toPHPDateFormat($styleNumberFormat);
$styleNumberFormatCode = $this->styleHelper->getNumberFormatCode($cellStyleId);
$phpDateFormat = DateFormatHelper::toPHPDateFormat($styleNumberFormatCode);
return $dateObj->format($phpDateFormat);
} else {
return $dateObj;
@@ -62,7 +62,9 @@ class DateFormatHelper
public static function toPHPDateFormat($excelDateFormat)
{
// Remove brackets potentially present at the beginning of the format string
$dateFormat = preg_replace('/^(\[\$[^\]]+?\])/i', '', $excelDateFormat);
// and text portion of the format at the end of it (starting with ";")
// See §18.8.31 of ECMA-376 for more detail.
$dateFormat = preg_replace('/^(?:\[\$[^\]]+?\])?([^;]*).*/', '$1', $excelDateFormat);
// Double quotes are used to escape characters that must not be interpreted.
// For instance, ["Day " dd] should result in "Day 13" and we should not try to interpret "D", "a", "y"
@@ -34,7 +34,7 @@ class SharedStringsHelper
/**
* @param string $filePath Path of the XLSX file being read
* @param string|void $tempFolder Temporary folder where the temporary files to store shared strings will be stored
* @param string|null|void $tempFolder Temporary folder where the temporary files to store shared strings will be stored
*/
public function __construct($filePath, $tempFolder = null)
{
@@ -94,44 +94,19 @@ class SharedStringsHelper
$xmlReader->readUntilNodeFound('si');
while ($xmlReader->name === 'si') {
$node = $this->getSimpleXmlElementNodeFromXMLReader($xmlReader);
$node->registerXPathNamespace('ns', self::MAIN_NAMESPACE_FOR_SHARED_STRINGS_XML);
// removes nodes that should not be read, like the pronunciation of the Kanji characters
$cleanNode = $this->removeSuperfluousTextNodes($node);
// find all text nodes "t"; there can be multiple if the cell contains formatting
$textNodes = $cleanNode->xpath('//ns:t');
$textValue = '';
foreach ($textNodes as $nodeIndex => $textNode) {
if ($nodeIndex !== 0) {
// add a space between each "t" node
$textValue .= ' ';
}
if ($this->shouldPreserveWhitespace($textNode)) {
$textValue .= $textNode->__toString();
} else {
$textValue .= trim($textNode->__toString());
}
}
$unescapedTextValue = $escaper->unescape($textValue);
$this->cachingStrategy->addStringForIndex($unescapedTextValue, $sharedStringIndex);
$this->processSharedStringsItem($xmlReader, $sharedStringIndex, $escaper);
$sharedStringIndex++;
// jump to the next 'si' tag
$xmlReader->next('si');
}
$this->cachingStrategy->closeCache();
} catch (XMLProcessingException $exception) {
throw new IOException("The sharedStrings.xml file is invalid and cannot be read. [{$exception->getMessage()}]");
}
$this->cachingStrategy->closeCache();
$xmlReader->close();
}
@@ -182,6 +157,31 @@ class SharedStringsHelper
->getBestCachingStrategy($sharedStringsUniqueCount, $this->tempFolder);
}
/**
* Processes the shared strings item XML node which the given XML reader is positioned on.
*
* @param \Box\Spout\Reader\Wrapper\XMLReader $xmlReader
* @param int $sharedStringIndex Index of the processed shared strings item
* @param \Box\Spout\Common\Escaper\XLSX $escaper Helper to escape values
* @return void
*/
protected function processSharedStringsItem($xmlReader, $sharedStringIndex, $escaper)
{
$node = $this->getSimpleXmlElementNodeFromXMLReader($xmlReader);
$node->registerXPathNamespace('ns', self::MAIN_NAMESPACE_FOR_SHARED_STRINGS_XML);
// removes nodes that should not be read, like the pronunciation of the Kanji characters
$cleanNode = $this->removeSuperfluousTextNodes($node);
// find all text nodes "t"; there can be multiple if the cell contains formatting
$textNodes = $cleanNode->xpath('//ns:t');
$textValue = $this->extractTextValueForNodes($textNodes);
$unescapedTextValue = $escaper->unescape($textValue);
$this->cachingStrategy->addStringForIndex($unescapedTextValue, $sharedStringIndex);
}
/**
* Returns a SimpleXMLElement node from the current node in the given XMLReader instance.
* This is to simplify the parsing of the subtree.
@@ -225,6 +225,29 @@ class SharedStringsHelper
return $parentNode;
}
/**
* @param array $textNodes Text XML nodes ("<t>")
* @return string The value associated with the given text node(s)
*/
protected function extractTextValueForNodes($textNodes)
{
$textValue = '';
foreach ($textNodes as $nodeIndex => $textNode) {
if ($nodeIndex !== 0) {
// add a space between each "t" node
$textValue .= ' ';
}
$textNodeAsString = $textNode->__toString();
$shouldPreserveWhitespace = $this->shouldPreserveWhitespace($textNode);
$textValue .= ($shouldPreserveWhitespace) ? $textNodeAsString : trim($textNodeAsString);
}
return $textValue;
}
/**
* If the text node has the attribute 'xml:space="preserve"', then preserve whitespace.
*
+7 -7
View File
@@ -20,27 +20,27 @@ class SheetHelper
/** @var string Path of the XLSX file being read */
protected $filePath;
/** @var \Box\Spout\Reader\XLSX\ReaderOptions Reader's current options */
protected $options;
/** @var \Box\Spout\Reader\XLSX\Helper\SharedStringsHelper Helper to work with shared strings */
protected $sharedStringsHelper;
/** @var \Box\Spout\Common\Helper\GlobalFunctionsHelper Helper to work with global functions */
protected $globalFunctionsHelper;
/** @var bool Whether date/time values should be returned as PHP objects or be formatted as strings */
protected $shouldFormatDates;
/**
* @param string $filePath Path of the XLSX file being read
* @param \Box\Spout\Reader\XLSX\ReaderOptions $options Reader's current options
* @param \Box\Spout\Reader\XLSX\Helper\SharedStringsHelper Helper to work with shared strings
* @param \Box\Spout\Common\Helper\GlobalFunctionsHelper $globalFunctionsHelper
* @param bool $shouldFormatDates Whether date/time values should be returned as PHP objects or be formatted as strings
*/
public function __construct($filePath, $sharedStringsHelper, $globalFunctionsHelper, $shouldFormatDates)
public function __construct($filePath, $options, $sharedStringsHelper, $globalFunctionsHelper)
{
$this->filePath = $filePath;
$this->options = $options;
$this->sharedStringsHelper = $sharedStringsHelper;
$this->globalFunctionsHelper = $globalFunctionsHelper;
$this->shouldFormatDates = $shouldFormatDates;
}
/**
@@ -92,7 +92,7 @@ class SheetHelper
$sheetDataXMLFilePath = $this->getSheetDataXMLFilePathForSheetId($sheetId);
return new Sheet($this->filePath, $sheetDataXMLFilePath, $this->sharedStringsHelper, $this->shouldFormatDates, $sheetIndexZeroBased, $sheetName);
return new Sheet($this->filePath, $sheetDataXMLFilePath, $sheetIndexZeroBased, $sheetName, $this->options, $this->sharedStringsHelper);
}
/**
+104 -60
View File
@@ -29,6 +29,8 @@ class StyleHelper
/** By convention, default style ID is 0 */
const DEFAULT_STYLE_ID = 0;
const NUMBER_FORMAT_GENERAL = 'General';
/**
* @see https://msdn.microsoft.com/en-us/library/ff529597(v=office.12).aspx
* @var array Mapping between built-in numFmtId and the associated format - for dates only
@@ -51,18 +53,48 @@ class StyleHelper
/** @var string Path of the XLSX file being read */
protected $filePath;
/** @var array Array containing the IDs of built-in number formats indicating a date */
protected $builtinNumFmtIdIndicatingDates;
/** @var array Array containing a mapping NUM_FMT_ID => FORMAT_CODE */
protected $customNumberFormats;
/** @var array Array containing a mapping STYLE_ID => [STYLE_ATTRIBUTES] */
protected $stylesAttributes;
/** @var array Cache containing a mapping NUM_FMT_ID => IS_DATE_FORMAT. Used to avoid lots of recalculations */
protected $numFmtIdToIsDateFormatCache = [];
/**
* @param string $filePath Path of the XLSX file being read
*/
public function __construct($filePath)
{
$this->filePath = $filePath;
$this->builtinNumFmtIdIndicatingDates = array_keys(self::$builtinNumFmtIdToNumFormatMapping);
}
/**
* Returns whether the style with the given ID should consider
* numeric values as timestamps and format the cell as a date.
*
* @param int $styleId Zero-based style ID
* @return bool Whether the cell with the given cell should display a date instead of a numeric value
*/
public function shouldFormatNumericValueAsDate($styleId)
{
$stylesAttributes = $this->getStylesAttributes();
// Default style (0) does not format numeric values as timestamps. Only custom styles do.
// Also if the style ID does not exist in the styles.xml file, format as numeric value.
// Using isset here because it is way faster than array_key_exists...
if ($styleId === self::DEFAULT_STYLE_ID || !isset($stylesAttributes[$styleId])) {
return false;
}
$styleAttributes = $stylesAttributes[$styleId];
return $this->doesStyleIndicateDate($styleAttributes);
}
/**
@@ -125,9 +157,15 @@ class StyleHelper
{
while ($xmlReader->read()) {
if ($xmlReader->isPositionedOnStartingNode(self::XML_NODE_XF)) {
$numFmtId = $xmlReader->getAttribute(self::XML_ATTRIBUTE_NUM_FMT_ID);
$normalizedNumFmtId = ($numFmtId !== null) ? intval($numFmtId) : null;
$applyNumberFormat = $xmlReader->getAttribute(self::XML_ATTRIBUTE_APPLY_NUMBER_FORMAT);
$normalizedApplyNumberFormat = ($applyNumberFormat !== null) ? !!$applyNumberFormat : null;
$this->stylesAttributes[] = [
self::XML_ATTRIBUTE_NUM_FMT_ID => intval($xmlReader->getAttribute(self::XML_ATTRIBUTE_NUM_FMT_ID)),
self::XML_ATTRIBUTE_APPLY_NUMBER_FORMAT => !!($xmlReader->getAttribute(self::XML_ATTRIBUTE_APPLY_NUMBER_FORMAT)),
self::XML_ATTRIBUTE_NUM_FMT_ID => $normalizedNumFmtId,
self::XML_ATTRIBUTE_APPLY_NUMBER_FORMAT => $normalizedApplyNumberFormat,
];
} else if ($xmlReader->isPositionedOnEndingNode(self::XML_NODE_CELL_XFS)) {
// Once done reading "cellXfs" node's children
@@ -161,86 +199,92 @@ class StyleHelper
}
/**
* Returns whether the style with the given ID should consider
* numeric values as timestamps and format the cell as a date.
*
* @param int $styleId Zero-based style ID
* @return bool Whether the cell with the given cell should display a date instead of a numeric value
* @param array $styleAttributes Array containing the style attributes (2 keys: "applyNumberFormat" and "numFmtId")
* @return bool Whether the style with the given attributes indicates that the number is a date
*/
public function shouldFormatNumericValueAsDate($styleId)
protected function doesStyleIndicateDate($styleAttributes)
{
$stylesAttributes = $this->getStylesAttributes();
// Default style (0) does not format numeric values as timestamps. Only custom styles do.
// Also if the style ID does not exist in the styles.xml file, format as numeric value.
// Using isset here because it is way faster than array_key_exists...
if ($styleId === self::DEFAULT_STYLE_ID || !isset($stylesAttributes[$styleId])) {
return false;
}
$styleAttributes = $stylesAttributes[$styleId];
$applyNumberFormat = $styleAttributes[self::XML_ATTRIBUTE_APPLY_NUMBER_FORMAT];
if (!$applyNumberFormat) {
$numFmtId = $styleAttributes[self::XML_ATTRIBUTE_NUM_FMT_ID];
// A style may apply a date format if it has:
// - "applyNumberFormat" attribute not set to "false"
// - "numFmtId" attribute set
// This is a preliminary check, as having "numFmtId" set just means the style should apply a specific number format,
// but this is not necessarily a date.
if ($applyNumberFormat === false || $numFmtId === null) {
return false;
}
$numFmtId = $styleAttributes[self::XML_ATTRIBUTE_NUM_FMT_ID];
return $this->doesNumFmtIdIndicateDate($numFmtId);
}
/**
* Returns whether the number format ID indicates that the number is a date.
* The result is cached to avoid recomputing the same thing over and over, as
* "numFmtId" attributes can be shared between multiple styles.
*
* @param int $numFmtId
* @return bool Whether the number format ID indicates that the number is a timestamp
* @return bool Whether the number format ID indicates that the number is a date
*/
protected function doesNumFmtIdIndicateDate($numFmtId)
{
return (
!$this->doesNumFmtIdIndicateGeneralFormat($numFmtId) &&
(
if (!isset($this->numFmtIdToIsDateFormatCache[$numFmtId])) {
$formatCode = $this->getFormatCodeForNumFmtId($numFmtId);
$this->numFmtIdToIsDateFormatCache[$numFmtId] = (
$this->isNumFmtIdBuiltInDateFormat($numFmtId) ||
$this->isNumFmtIdCustomDateFormat($numFmtId)
)
);
$this->isFormatCodeCustomDateFormat($formatCode)
);
}
return $this->numFmtIdToIsDateFormatCache[$numFmtId];
}
/**
* @param int $numFmtId
* @return bool Whether the number format ID indicates the "General" format (0 by convention)
* @return string|null The custom number format or NULL if none defined for the given numFmtId
*/
protected function doesNumFmtIdIndicateGeneralFormat($numFmtId)
{
return ($numFmtId === 0);
}
/**
* @param int $numFmtId
* @return bool Whether the number format ID indicates that the number is a timestamp
*/
protected function isNumFmtIdBuiltInDateFormat($numFmtId)
{
$builtInDateFormatIds = array_keys(self::$builtinNumFmtIdToNumFormatMapping);
return in_array($numFmtId, $builtInDateFormatIds);
}
/**
* @param int $numFmtId
* @return bool Whether the number format ID indicates that the number is a timestamp
*/
protected function isNumFmtIdCustomDateFormat($numFmtId)
protected function getFormatCodeForNumFmtId($numFmtId)
{
$customNumberFormats = $this->getCustomNumberFormats();
// Using isset here because it is way faster than array_key_exists...
if (!isset($customNumberFormats[$numFmtId])) {
return (isset($customNumberFormats[$numFmtId])) ? $customNumberFormats[$numFmtId] : null;
}
/**
* @param int $numFmtId
* @return bool Whether the number format ID indicates that the number is a date
*/
protected function isNumFmtIdBuiltInDateFormat($numFmtId)
{
return in_array($numFmtId, $this->builtinNumFmtIdIndicatingDates);
}
/**
* @param string|null $formatCode
* @return bool Whether the given format code indicates that the number is a date
*/
protected function isFormatCodeCustomDateFormat($formatCode)
{
// if no associated format code or if using the default "General" format
if ($formatCode === null || strcasecmp($formatCode, self::NUMBER_FORMAT_GENERAL) === 0) {
return false;
}
$customNumberFormat = $customNumberFormats[$numFmtId];
return $this->isFormatCodeMatchingDateFormatPattern($formatCode);
}
/**
* @param string $formatCode
* @return bool Whether the given format code matches a date format pattern
*/
protected function isFormatCodeMatchingDateFormatPattern($formatCode)
{
// Remove extra formatting (what's between [ ], the brackets should not be preceded by a "\")
$pattern = '((?<!\\\)\[.+?(?<!\\\)\])';
$customNumberFormat = preg_replace($pattern, '', $customNumberFormat);
$formatCode = preg_replace($pattern, '', $formatCode);
// custom date formats contain specific characters to represent the date:
// e - yy - m - d - h - s
@@ -249,10 +293,10 @@ class StyleHelper
$hasFoundDateFormatCharacter = false;
foreach ($dateFormatCharacters as $dateFormatCharacter) {
// character not preceded by "\"
$pattern = '/(?<!\\\)' . $dateFormatCharacter . '/';
// character not preceded by "\" (case insensitive)
$pattern = '/(?<!\\\)' . $dateFormatCharacter . '/i';
if (preg_match($pattern, $customNumberFormat)) {
if (preg_match($pattern, $formatCode)) {
$hasFoundDateFormatCharacter = true;
break;
}
@@ -266,21 +310,21 @@ class StyleHelper
* NOTE: It is assumed that the style DOES have a number format associated to it.
*
* @param int $styleId Zero-based style ID
* @return string The number format associated with the given style
* @return string The number format code associated with the given style
*/
public function getNumberFormat($styleId)
public function getNumberFormatCode($styleId)
{
$stylesAttributes = $this->getStylesAttributes();
$styleAttributes = $stylesAttributes[$styleId];
$numFmtId = $styleAttributes[self::XML_ATTRIBUTE_NUM_FMT_ID];
if ($this->isNumFmtIdBuiltInDateFormat($numFmtId)) {
$numberFormat = self::$builtinNumFmtIdToNumFormatMapping[$numFmtId];
$numberFormatCode = self::$builtinNumFmtIdToNumFormatMapping[$numFmtId];
} else {
$customNumberFormats = $this->getCustomNumberFormats();
$numberFormat = $customNumberFormats[$numFmtId];
$numberFormatCode = $customNumberFormats[$numFmtId];
}
return $numberFormat;
return $numberFormatCode;
}
}
+17 -7
View File
@@ -14,9 +14,6 @@ use Box\Spout\Reader\XLSX\Helper\SharedStringsHelper;
*/
class Reader extends AbstractReader
{
/** @var string Temporary folder where the temporary files will be created */
protected $tempFolder;
/** @var \ZipArchive */
protected $zip;
@@ -27,13 +24,26 @@ class Reader extends AbstractReader
protected $sheetIterator;
/**
* Returns the reader's current options
*
* @return ReaderOptions
*/
protected function getOptions()
{
if (!isset($this->options)) {
$this->options = new ReaderOptions();
}
return $this->options;
}
/**
* @param string $tempFolder Temporary folder where the temporary files will be created
* @return Reader
*/
public function setTempFolder($tempFolder)
{
$this->tempFolder = $tempFolder;
$this->getOptions()->setTempFolder($tempFolder);
return $this;
}
@@ -62,14 +72,14 @@ class Reader extends AbstractReader
$this->zip = new \ZipArchive();
if ($this->zip->open($filePath) === true) {
$this->sharedStringsHelper = new SharedStringsHelper($filePath, $this->tempFolder);
$this->sharedStringsHelper = new SharedStringsHelper($filePath, $this->getOptions()->getTempFolder());
if ($this->sharedStringsHelper->hasSharedStrings()) {
// Extracts all the strings from the sheets for easy access in the future
$this->sharedStringsHelper->extractSharedStrings();
}
$this->sheetIterator = new SheetIterator($filePath, $this->sharedStringsHelper, $this->globalFunctionsHelper, $this->shouldFormatDates);
$this->sheetIterator = new SheetIterator($filePath, $this->getOptions(), $this->sharedStringsHelper, $this->globalFunctionsHelper);
} else {
throw new IOException("Could not open $filePath for reading.");
}
@@ -80,7 +90,7 @@ class Reader extends AbstractReader
*
* @return SheetIterator To iterate over sheets
*/
public function getConcreteSheetIterator()
protected function getConcreteSheetIterator()
{
return $this->sheetIterator;
}
+33
View File
@@ -0,0 +1,33 @@
<?php
namespace Box\Spout\Reader\XLSX;
/**
* Class ReaderOptions
* This class is used to customize the reader's behavior
*
* @package Box\Spout\Reader\XLSX
*/
class ReaderOptions extends \Box\Spout\Reader\Common\ReaderOptions
{
/** @var string|null Temporary folder where the temporary files will be created */
protected $tempFolder = null;
/**
* @return string|null Temporary folder where the temporary files will be created
*/
public function getTempFolder()
{
return $this->tempFolder;
}
/**
* @param string|null $tempFolder Temporary folder where the temporary files will be created
* @return ReaderOptions
*/
public function setTempFolder($tempFolder)
{
$this->tempFolder = $tempFolder;
return $this;
}
}
+218 -62
View File
@@ -9,6 +9,7 @@ use Box\Spout\Reader\Wrapper\XMLReader;
use Box\Spout\Reader\XLSX\Helper\CellHelper;
use Box\Spout\Reader\XLSX\Helper\CellValueFormatter;
use Box\Spout\Reader\XLSX\Helper\StyleHelper;
use Box\Spout\Reader\Common\XMLProcessor;
/**
* Class RowIterator
@@ -26,6 +27,7 @@ class RowIterator implements IteratorInterface
/** Definition of XML attributes used to parse data */
const XML_ATTRIBUTE_REF = 'ref';
const XML_ATTRIBUTE_SPANS = 'spans';
const XML_ATTRIBUTE_ROW_INDEX = 'r';
const XML_ATTRIBUTE_CELL_INDEX = 'r';
/** @var string Path of the XLSX file being read */
@@ -37,15 +39,24 @@ class RowIterator implements IteratorInterface
/** @var \Box\Spout\Reader\Wrapper\XMLReader The XMLReader object that will help read sheet's XML data */
protected $xmlReader;
/** @var \Box\Spout\Reader\Common\XMLProcessor Helper Object to process XML nodes */
protected $xmlProcessor;
/** @var Helper\CellValueFormatter Helper to format cell values */
protected $cellValueFormatter;
/** @var Helper\StyleHelper $styleHelper Helper to work with styles */
protected $styleHelper;
/** @var int Number of read rows */
/**
* TODO: This variable can be deleted when row indices get preserved
* @var int Number of read rows
*/
protected $numReadRows = 0;
/** @var array Contains the data for the currently processed row (key = cell index, value = cell value) */
protected $currentlyProcessedRowData = [];
/** @var array|null Buffer used to store the row data, while checking if there are more rows to read */
protected $rowDataBuffer = null;
@@ -55,16 +66,25 @@ class RowIterator implements IteratorInterface
/** @var int The number of columns the sheet has (0 meaning undefined) */
protected $numColumns = 0;
/** @var bool Whether empty rows should be returned or skipped */
protected $shouldPreserveEmptyRows;
/** @var int Last row index processed (one-based) */
protected $lastRowIndexProcessed = 0;
/** @var int Row index to be processed next (one-based) */
protected $nextRowIndexToBeProcessed = 0;
/** @var int Last column index processed (zero-based) */
protected $lastColumnIndexProcessed = -1;
/**
* @param string $filePath Path of the XLSX file being read
* @param string $sheetDataXMLFilePath Path of the sheet data XML file as in [Content_Types].xml
* @param \Box\Spout\Reader\XLSX\ReaderOptions $options Reader's current options
* @param Helper\SharedStringsHelper $sharedStringsHelper Helper to work with shared strings
* @param bool $shouldFormatDates Whether date/time values should be returned as PHP objects or be formatted as strings
*/
public function __construct($filePath, $sheetDataXMLFilePath, $sharedStringsHelper, $shouldFormatDates)
public function __construct($filePath, $sheetDataXMLFilePath, $options, $sharedStringsHelper)
{
$this->filePath = $filePath;
$this->sheetDataXMLFilePath = $this->normalizeSheetDataXMLFilePath($sheetDataXMLFilePath);
@@ -72,7 +92,17 @@ class RowIterator implements IteratorInterface
$this->xmlReader = new XMLReader();
$this->styleHelper = new StyleHelper($filePath);
$this->cellValueFormatter = new CellValueFormatter($sharedStringsHelper, $this->styleHelper, $shouldFormatDates);
$this->cellValueFormatter = new CellValueFormatter($sharedStringsHelper, $this->styleHelper, $options->shouldFormatDates());
$this->shouldPreserveEmptyRows = $options->shouldPreserveEmptyRows();
// Register all callbacks to process different nodes when reading the XML file
$this->xmlProcessor = new XMLProcessor($this->xmlReader);
$this->xmlProcessor->registerCallback(self::XML_NODE_DIMENSION, XMLProcessor::NODE_TYPE_START, [$this, 'processDimensionStartingNode']);
$this->xmlProcessor->registerCallback(self::XML_NODE_ROW, XMLProcessor::NODE_TYPE_START, [$this, 'processRowStartingNode']);
$this->xmlProcessor->registerCallback(self::XML_NODE_CELL, XMLProcessor::NODE_TYPE_START, [$this, 'processCellStartingNode']);
$this->xmlProcessor->registerCallback(self::XML_NODE_ROW, XMLProcessor::NODE_TYPE_END, [$this, 'processRowEndingNode']);
$this->xmlProcessor->registerCallback(self::XML_NODE_WORKSHEET, XMLProcessor::NODE_TYPE_END, [$this, 'processWorksheetEndingNode']);
}
/**
@@ -104,6 +134,8 @@ class RowIterator implements IteratorInterface
}
$this->numReadRows = 0;
$this->lastRowIndexProcessed = 0;
$this->nextRowIndexToBeProcessed = 0;
$this->rowDataBuffer = null;
$this->hasReachedEndOfFile = false;
$this->numColumns = 0;
@@ -115,7 +147,7 @@ class RowIterator implements IteratorInterface
* Checks if current position is valid
* @link http://php.net/manual/en/iterator.valid.php
*
* @return boolean
* @return bool
*/
public function valid()
{
@@ -123,7 +155,7 @@ class RowIterator implements IteratorInterface
}
/**
* Move forward to next element. Empty rows will be skipped.
* Move forward to next element. Reads data describing the next unprocessed row.
* @link http://php.net/manual/en/iterator.next.php
*
* @return void
@@ -132,69 +164,165 @@ class RowIterator implements IteratorInterface
*/
public function next()
{
$rowData = [];
$this->nextRowIndexToBeProcessed++;
if ($this->doesNeedDataForNextRowToBeProcessed()) {
$this->readDataForNextRow();
}
}
/**
* Returns whether we need data for the next row to be processed.
* We don't need to read data if:
* we have already read at least one row
* AND
* we need to preserve empty rows
* AND
* the last row that was read is not the row that need to be processed
* (i.e. if we need to return empty rows)
*
* @return bool Whether we need data for the next row to be processed.
*/
protected function doesNeedDataForNextRowToBeProcessed()
{
$hasReadAtLeastOneRow = ($this->lastRowIndexProcessed !== 0);
return (
!$hasReadAtLeastOneRow ||
!$this->shouldPreserveEmptyRows ||
$this->lastRowIndexProcessed < $this->nextRowIndexToBeProcessed
);
}
/**
* @return void
* @throws \Box\Spout\Reader\Exception\SharedStringNotFoundException If a shared string was not found
* @throws \Box\Spout\Common\Exception\IOException If unable to read the sheet data XML
*/
protected function readDataForNextRow()
{
$this->currentlyProcessedRowData = [];
try {
while ($this->xmlReader->read()) {
if ($this->xmlReader->isPositionedOnStartingNode(self::XML_NODE_DIMENSION)) {
// Read dimensions of the sheet
$dimensionRef = $this->xmlReader->getAttribute(self::XML_ATTRIBUTE_REF); // returns 'A1:M13' for instance (or 'A1' for empty sheet)
if (preg_match('/[A-Z\d]+:([A-Z\d]+)/', $dimensionRef, $matches)) {
$lastCellIndex = $matches[1];
$this->numColumns = CellHelper::getColumnIndexFromCellIndex($lastCellIndex) + 1;
}
} else if ($this->xmlReader->isPositionedOnStartingNode(self::XML_NODE_ROW)) {
// Start of the row description
// Reset index of the last processed column
$this->lastColumnIndexProcessed = -1;
// Read spans info if present
$numberOfColumnsForRow = $this->numColumns;
$spans = $this->xmlReader->getAttribute(self::XML_ATTRIBUTE_SPANS); // returns '1:5' for instance
if ($spans) {
list(, $numberOfColumnsForRow) = explode(':', $spans);
$numberOfColumnsForRow = intval($numberOfColumnsForRow);
}
$rowData = ($numberOfColumnsForRow !== 0) ? array_fill(0, $numberOfColumnsForRow, '') : [];
} else if ($this->xmlReader->isPositionedOnStartingNode(self::XML_NODE_CELL)) {
// Start of a cell description
$currentColumnIndex = $this->getCellIndex($this->xmlReader);
$node = $this->xmlReader->expand();
$rowData[$currentColumnIndex] = $this->getCellValue($node);
$this->lastColumnIndexProcessed = $currentColumnIndex;
} else if ($this->xmlReader->isPositionedOnEndingNode(self::XML_NODE_ROW)) {
// End of the row description
// If needed, we fill the empty cells
$rowData = ($this->numColumns !== 0) ? $rowData : CellHelper::fillMissingArrayIndexes($rowData);
$this->numReadRows++;
break;
} else if ($this->xmlReader->isPositionedOnEndingNode(self::XML_NODE_WORKSHEET)) {
// The closing "</worksheet>" marks the end of the file
$this->hasReachedEndOfFile = true;
break;
}
}
$this->xmlProcessor->readUntilStopped();
} catch (XMLProcessingException $exception) {
throw new IOException("The {$this->sheetDataXMLFilePath} file cannot be read. [{$exception->getMessage()}]");
}
$this->rowDataBuffer = $rowData;
$this->rowDataBuffer = $this->currentlyProcessedRowData;
}
/**
* @param \Box\Spout\Reader\Wrapper\XMLReader $xmlReader XMLReader object, positioned on a "<c>" tag
* @return int
* @param \Box\Spout\Reader\Wrapper\XMLReader $xmlReader XMLReader object, positioned on a "<dimension>" starting node
* @return int A return code that indicates what action should the processor take next
*/
protected function processDimensionStartingNode($xmlReader)
{
// Read dimensions of the sheet
$dimensionRef = $xmlReader->getAttribute(self::XML_ATTRIBUTE_REF); // returns 'A1:M13' for instance (or 'A1' for empty sheet)
if (preg_match('/[A-Z\d]+:([A-Z\d]+)/', $dimensionRef, $matches)) {
$lastCellIndex = $matches[1];
$this->numColumns = CellHelper::getColumnIndexFromCellIndex($lastCellIndex) + 1;
}
return XMLProcessor::PROCESSING_CONTINUE;
}
/**
* @param \Box\Spout\Reader\Wrapper\XMLReader $xmlReader XMLReader object, positioned on a "<row>" starting node
* @return int A return code that indicates what action should the processor take next
*/
protected function processRowStartingNode($xmlReader)
{
// Reset index of the last processed column
$this->lastColumnIndexProcessed = -1;
// Mark the last processed row as the one currently being read
$this->lastRowIndexProcessed = $this->getRowIndex($xmlReader);
// Read spans info if present
$numberOfColumnsForRow = $this->numColumns;
$spans = $xmlReader->getAttribute(self::XML_ATTRIBUTE_SPANS); // returns '1:5' for instance
if ($spans) {
list(, $numberOfColumnsForRow) = explode(':', $spans);
$numberOfColumnsForRow = intval($numberOfColumnsForRow);
}
$this->currentlyProcessedRowData = ($numberOfColumnsForRow !== 0) ? array_fill(0, $numberOfColumnsForRow, '') : [];
return XMLProcessor::PROCESSING_CONTINUE;
}
/**
* @param \Box\Spout\Reader\Wrapper\XMLReader $xmlReader XMLReader object, positioned on a "<cell>" starting node
* @return int A return code that indicates what action should the processor take next
*/
protected function processCellStartingNode($xmlReader)
{
$currentColumnIndex = $this->getColumnIndex($xmlReader);
$node = $xmlReader->expand();
$this->currentlyProcessedRowData[$currentColumnIndex] = $this->getCellValue($node);
$this->lastColumnIndexProcessed = $currentColumnIndex;
return XMLProcessor::PROCESSING_CONTINUE;
}
/**
* @return int A return code that indicates what action should the processor take next
*/
protected function processRowEndingNode()
{
// if the fetched row is empty and we don't want to preserve it..,
if (!$this->shouldPreserveEmptyRows && $this->isEmptyRow($this->currentlyProcessedRowData)) {
// ... skip it
return XMLProcessor::PROCESSING_CONTINUE;
}
$this->numReadRows++;
// If needed, we fill the empty cells
if ($this->numColumns === 0) {
$this->currentlyProcessedRowData = CellHelper::fillMissingArrayIndexes($this->currentlyProcessedRowData);
}
// at this point, we have all the data we need for the row
// so that we can populate the buffer
return XMLProcessor::PROCESSING_STOP;
}
/**
* @return int A return code that indicates what action should the processor take next
*/
protected function processWorksheetEndingNode()
{
// The closing "</worksheet>" marks the end of the file
$this->hasReachedEndOfFile = true;
return XMLProcessor::PROCESSING_STOP;
}
/**
* @param \Box\Spout\Reader\Wrapper\XMLReader $xmlReader XMLReader object, positioned on a "<row>" node
* @return int Row index
* @throws \Box\Spout\Common\Exception\InvalidArgumentException When the given cell index is invalid
*/
protected function getCellIndex($xmlReader)
protected function getRowIndex($xmlReader)
{
// Get "r" attribute if present (from something like <row r="3"...>
$currentRowIndex = $xmlReader->getAttribute(self::XML_ATTRIBUTE_ROW_INDEX);
return ($currentRowIndex !== null) ?
intval($currentRowIndex) :
$this->lastRowIndexProcessed + 1;
}
/**
* @param \Box\Spout\Reader\Wrapper\XMLReader $xmlReader XMLReader object, positioned on a "<c>" node
* @return int Column index
* @throws \Box\Spout\Common\Exception\InvalidArgumentException When the given cell index is invalid
*/
protected function getColumnIndex($xmlReader)
{
// Get "r" attribute if present (from something like <c r="A1"...>
$currentCellIndex = $xmlReader->getAttribute(self::XML_ATTRIBUTE_CELL_INDEX);
@@ -216,25 +344,53 @@ class RowIterator implements IteratorInterface
}
/**
* Return the current element, from the buffer.
* @param array $rowData
* @return bool Whether the given row is empty
*/
protected function isEmptyRow($rowData)
{
return (count($rowData) === 1 && $rowData[0] === '');
}
/**
* Return the current element, either an empty row or from the buffer.
* @link http://php.net/manual/en/iterator.current.php
*
* @return array|null
*/
public function current()
{
return $this->rowDataBuffer;
$rowDataForRowToBeProcessed = $this->rowDataBuffer;
if ($this->shouldPreserveEmptyRows) {
// when we need to preserve empty rows, we will either return
// an empty row or the last row read. This depends whether the
// index of last row that was read matches the index of the last
// row whose value should be returned.
if ($this->lastRowIndexProcessed !== $this->nextRowIndexToBeProcessed) {
// return empty row if mismatch between last processed row
// and the row that needs to be returned
$rowDataForRowToBeProcessed = [''];
}
}
return $rowDataForRowToBeProcessed;
}
/**
* Return the key of the current element
* Return the key of the current element. Here, the row index.
* @link http://php.net/manual/en/iterator.key.php
*
* @return int
*/
public function key()
{
return $this->numReadRows;
// TODO: This should return $this->nextRowIndexToBeProcessed
// but to avoid a breaking change, the return value for
// this function has been kept as the number of rows read.
return $this->shouldPreserveEmptyRows ?
$this->nextRowIndexToBeProcessed :
$this->numReadRows;
}
+4 -4
View File
@@ -24,14 +24,14 @@ class Sheet implements SheetInterface
/**
* @param string $filePath Path of the XLSX file being read
* @param string $sheetDataXMLFilePath Path of the sheet data XML file as in [Content_Types].xml
* @param Helper\SharedStringsHelper Helper to work with shared strings
* @param bool $shouldFormatDates Whether date/time values should be returned as PHP objects or be formatted as strings
* @param int $sheetIndex Index of the sheet, based on order in the workbook (zero-based)
* @param string $sheetName Name of the sheet
* @param \Box\Spout\Reader\XLSX\ReaderOptions $options Reader's current options
* @param Helper\SharedStringsHelper Helper to work with shared strings
*/
public function __construct($filePath, $sheetDataXMLFilePath, $sharedStringsHelper, $shouldFormatDates, $sheetIndex, $sheetName)
public function __construct($filePath, $sheetDataXMLFilePath, $sheetIndex, $sheetName, $options, $sharedStringsHelper)
{
$this->rowIterator = new RowIterator($filePath, $sheetDataXMLFilePath, $sharedStringsHelper, $shouldFormatDates);
$this->rowIterator = new RowIterator($filePath, $sheetDataXMLFilePath, $options, $sharedStringsHelper);
$this->index = $sheetIndex;
$this->name = $sheetName;
}
+4 -4
View File
@@ -22,15 +22,15 @@ class SheetIterator implements IteratorInterface
/**
* @param string $filePath Path of the file to be read
* @param \Box\Spout\Reader\XLSX\ReaderOptions $options Reader's current options
* @param \Box\Spout\Reader\XLSX\Helper\SharedStringsHelper $sharedStringsHelper
* @param \Box\Spout\Common\Helper\GlobalFunctionsHelper $globalFunctionsHelper
* @param bool $shouldFormatDates Whether date/time values should be returned as PHP objects or be formatted as strings
* @throws \Box\Spout\Reader\Exception\NoSheetsFoundException If there are no sheets in the file
*/
public function __construct($filePath, $sharedStringsHelper, $globalFunctionsHelper, $shouldFormatDates)
public function __construct($filePath, $options, $sharedStringsHelper, $globalFunctionsHelper)
{
// Fetch all available sheets
$sheetHelper = new SheetHelper($filePath, $sharedStringsHelper, $globalFunctionsHelper, $shouldFormatDates);
$sheetHelper = new SheetHelper($filePath, $options, $sharedStringsHelper, $globalFunctionsHelper);
$this->sheets = $sheetHelper->getSheets();
if (count($this->sheets) === 0) {
@@ -53,7 +53,7 @@ class SheetIterator implements IteratorInterface
* Checks if current position is valid
* @link http://php.net/manual/en/iterator.valid.php
*
* @return boolean
* @return bool
*/
public function valid()
{
+32 -1
View File
@@ -4,6 +4,8 @@ namespace Box\Spout\Writer;
use Box\Spout\Common\Exception\IOException;
use Box\Spout\Common\Exception\InvalidArgumentException;
use Box\Spout\Common\Exception\SpoutException;
use Box\Spout\Common\Helper\FileSystemHelper;
use Box\Spout\Writer\Exception\WriterAlreadyOpenedException;
use Box\Spout\Writer\Exception\WriterNotOpenedException;
use Box\Spout\Writer\Style\StyleBuilder;
@@ -199,13 +201,23 @@ abstract class AbstractWriter implements WriterInterface
* @return AbstractWriter
* @throws \Box\Spout\Writer\Exception\WriterNotOpenedException If this function is called before opening the writer
* @throws \Box\Spout\Common\Exception\IOException If unable to write data
* @throws \Box\Spout\Common\Exception\SpoutException If anything else goes wrong while writing data
*/
public function addRow(array $dataRow)
{
if ($this->isWriterOpened) {
// empty $dataRow should not add an empty line
if (!empty($dataRow)) {
$this->addRowToWriter($dataRow, $this->rowStyle);
try {
$this->addRowToWriter($dataRow, $this->rowStyle);
} catch (SpoutException $e) {
// if an exception occurs while writing data,
// close the writer and remove all files created so far.
$this->closeAndAttemptToCleanupAllFiles();
// re-throw the exception to alert developers of the error
throw $e;
}
}
} else {
throw new WriterNotOpenedException('The writer needs to be opened before adding row.');
@@ -346,5 +358,24 @@ abstract class AbstractWriter implements WriterInterface
$this->isWriterOpened = false;
}
/**
* Closes the writer and attempts to cleanup all files that were
* created during the writing process (temp files & final file).
*
* @return void
*/
private function closeAndAttemptToCleanupAllFiles()
{
// close the writer, which should remove all temp files
$this->close();
// remove output file if it was created
if ($this->globalFunctionsHelper->file_exists($this->outputFilePath)) {
$outputFolderPath = dirname($this->outputFilePath);
$fileSystemHelper = new FileSystemHelper($outputFolderPath);
$fileSystemHelper->deleteFile($this->outputFilePath);
}
}
}
@@ -122,7 +122,7 @@ abstract class AbstractStyleHelper
protected function applyWrapTextIfCellContainsNewLine($style, $dataRow)
{
// if the "wrap text" option is already set, no-op
if ($style->shouldWrapText()) {
if ($style->hasSetWrapText()) {
return $style;
}
@@ -46,6 +46,15 @@ class CellHelper
return self::$columnIndexToCellIndexCache[$originalColumnIndex];
}
/**
* @param $value
* @return bool Whether the given value is considered "empty"
*/
public static function isEmpty($value)
{
return ($value === null || $value === '');
}
/**
* @param $value
* @return bool Whether the given value is a non empty string
@@ -75,7 +75,7 @@ class FileSystemHelper extends \Box\Spout\Common\Helper\FileSystemHelper
*/
protected function createRootFolder()
{
$this->rootFolder = $this->createFolder($this->baseFolderPath, uniqid('ods'));
$this->rootFolder = $this->createFolder($this->baseFolderRealPath, uniqid('ods'));
return $this;
}
+1 -1
View File
@@ -40,7 +40,7 @@ class Worksheet implements WorksheetInterface
/**
* @param \Box\Spout\Writer\Common\Sheet $externalSheet The associated "external" sheet
* @param string $worksheetFilesFolder Temporary folder where the files to create the XLSX will be stored
* @param string $worksheetFilesFolder Temporary folder where the files to create the ODS will be stored
* @throws \Box\Spout\Common\Exception\IOException If the sheet data file cannot be opened for writing
*/
public function __construct($externalSheet, $worksheetFilesFolder)
+1 -1
View File
@@ -20,7 +20,7 @@ class Writer extends AbstractMultiSheetsWriter
/** @var string Temporary folder where the files to create the ODS will be stored */
protected $tempFolder;
/** @var Internal\Workbook The workbook for the XLSX file */
/** @var Internal\Workbook The workbook for the ODS file */
protected $book;
/**
+17 -8
View File
@@ -115,7 +115,7 @@ class Style
}
/**
* @return boolean
* @return bool
*/
public function shouldApplyBorder()
{
@@ -123,7 +123,7 @@ class Style
}
/**
* @return boolean
* @return bool
*/
public function isFontBold()
{
@@ -142,7 +142,7 @@ class Style
}
/**
* @return boolean
* @return bool
*/
public function isFontItalic()
{
@@ -161,7 +161,7 @@ class Style
}
/**
* @return boolean
* @return bool
*/
public function isFontUnderline()
{
@@ -180,7 +180,7 @@ class Style
}
/**
* @return boolean
* @return bool
*/
public function isFontStrikethrough()
{
@@ -261,7 +261,7 @@ class Style
}
/**
* @return boolean
* @return bool
*/
public function shouldWrapText()
{
@@ -269,15 +269,24 @@ class Style
}
/**
* @param bool|void $shouldWrap Should the text be wrapped
* @return Style
*/
public function setShouldWrapText()
public function setShouldWrapText($shouldWrap = true)
{
$this->shouldWrapText = true;
$this->shouldWrapText = $shouldWrap;
$this->hasSetWrapText = true;
return $this;
}
/**
* @return bool
*/
public function hasSetWrapText()
{
return $this->hasSetWrapText;
}
/**
* @return bool Whether specific font properties should be applied
*/
+5 -5
View File
@@ -109,15 +109,15 @@ class StyleBuilder
}
/**
* Makes the text wrap in the cell if it's too long or
* on multiple lines.
* Makes the text wrap in the cell if requested
*
* @api
* @param bool $shouldWrap Should the text be wrapped
* @return StyleBuilder
*/
public function setShouldWrapText()
public function setShouldWrapText($shouldWrap = true)
{
$this->style->setShouldWrapText();
$this->style->setShouldWrapText($shouldWrap);
return $this;
}
@@ -138,7 +138,7 @@ class StyleBuilder
*
* @api
* @param string $color ARGB color (@see Color)
* @return StyleBuilder
* @return StyleBuilder
*/
public function setBackgroundColor($color)
{
@@ -94,7 +94,7 @@ class FileSystemHelper extends \Box\Spout\Common\Helper\FileSystemHelper
*/
protected function createRootFolder()
{
$this->rootFolder = $this->createFolder($this->baseFolderPath, uniqid('xlsx', true));
$this->rootFolder = $this->createFolder($this->baseFolderRealPath, uniqid('xlsx', true));
return $this;
}
+32 -3
View File
@@ -131,6 +131,38 @@ EOD;
* @throws \Box\Spout\Common\Exception\InvalidArgumentException If a cell value's type is not supported
*/
public function addRow($dataRow, $style)
{
if (!$this->isEmptyRow($dataRow)) {
$this->addNonEmptyRow($dataRow, $style);
}
$this->lastWrittenRowIndex++;
}
/**
* Returns whether the given row is empty
*
* @param array $dataRow Array containing data to be written. Cannot be empty.
* Example $dataRow = ['data1', 1234, null, '', 'data5'];
* @return bool Whether the given row is empty
*/
private function isEmptyRow($dataRow)
{
$numCells = count($dataRow);
return ($numCells === 1 && CellHelper::isEmpty($dataRow[0]));
}
/**
* Adds non empty row to the worksheet.
*
* @param array $dataRow Array containing data to be written. Cannot be empty.
* Example $dataRow = ['data1', 1234, null, '', 'data5'];
* @param \Box\Spout\Writer\Style\Style $style Style to be applied to the row. NULL means use default style.
* @return void
* @throws \Box\Spout\Common\Exception\IOException If the data cannot be written
* @throws \Box\Spout\Common\Exception\InvalidArgumentException If a cell value's type is not supported
*/
private function addNonEmptyRow($dataRow, $style)
{
$cellNumber = 0;
$rowIndex = $this->lastWrittenRowIndex + 1;
@@ -149,9 +181,6 @@ EOD;
if ($wasWriteSuccessful === false) {
throw new IOException("Unable to write data in {$this->worksheetFilePath}");
}
// only update the count if the write worked
$this->lastWrittenRowIndex++;
}
/**
+47 -59
View File
@@ -115,29 +115,40 @@ class ReaderTest extends \PHPUnit_Framework_TestCase
}
/**
* @return array
* @return void
*/
public function dataProviderForTestReadShouldSkipEmptyLines()
public function testReadShouldSkipEmptyLinesIfShouldPreserveEmptyRowsNotSet()
{
return [
['csv_with_empty_line.csv'],
['csv_with_empty_last_line.csv'],
$allRows = $this->getAllRowsForFile('csv_with_multiple_empty_lines.csv');
$expectedRows = [
// skipped row here
['csv--21', 'csv--22', 'csv--23'],
// skipped row here
['csv--41', 'csv--42', 'csv--43'],
// skipped row here
// last row empty
];
$this->assertEquals($expectedRows, $allRows);
}
/**
* @dataProvider dataProviderForTestReadShouldSkipEmptyLines
*
* @param string $fileName
* @return void
*/
public function testReadShouldSkipEmptyLines($fileName)
public function testReadShouldReturnEmptyLinesIfShouldPreserveEmptyRowsSet()
{
$allRows = $this->getAllRowsForFile($fileName);
$allRows = $this->getAllRowsForFile(
'csv_with_multiple_empty_lines.csv',
',', '"', "\n", EncodingHelper::ENCODING_UTF8,
$shouldPreserveEmptyRows = true
);
$expectedRows = [
['csv--11', 'csv--12', 'csv--13'],
['csv--31', 'csv--32', 'csv--33'],
[''],
['csv--21', 'csv--22', 'csv--23'],
[''],
['csv--41', 'csv--42', 'csv--43'],
[''],
];
$this->assertEquals($expectedRows, $allRows);
}
@@ -204,6 +215,21 @@ class ReaderTest extends \PHPUnit_Framework_TestCase
$this->assertEquals('This is, a comma', $allRows[0][0]);
}
/**
* @return void
*/
public function testReadCustomEOLs()
{
$allRows = $this->getAllRowsForFile('csv_with_CR_EOL.csv', ',', '"', "\r");
$expectedRows = [
['csv--11', 'csv--12', 'csv--13'],
['csv--21', 'csv--22', 'csv--23'],
['csv--31', 'csv--32', 'csv--33'],
];
$this->assertEquals($expectedRows, $allRows);
}
/**
* @return void
*/
@@ -236,7 +262,7 @@ class ReaderTest extends \PHPUnit_Framework_TestCase
*/
public function testReadShouldSkipBom($fileName, $fileEncoding)
{
$allRows = $this->getAllRowsForFile($fileName, ',', '"', $fileEncoding);
$allRows = $this->getAllRowsForFile($fileName, ',', '"', "\n", $fileEncoding);
$expectedRows = [
['csv--11', 'csv--12', 'csv--13'],
@@ -275,6 +301,7 @@ class ReaderTest extends \PHPUnit_Framework_TestCase
$allRows = [];
$resourcePath = $this->getResourcePath($fileName);
/** @var \Box\Spout\Common\Helper\GlobalFunctionsHelper|\PHPUnit_Framework_MockObject_MockObject $helperStub */
$helperStub = $this->getMockBuilder('\Box\Spout\Common\Helper\GlobalFunctionsHelper')
->setMethods(['function_exists'])
->getMock();
@@ -405,14 +432,18 @@ class ReaderTest extends \PHPUnit_Framework_TestCase
* @param string $fileName
* @param string|void $fieldDelimiter
* @param string|void $fieldEnclosure
* @param string|void $endOfLineCharacter
* @param string|void $encoding
* @param bool|void $shouldPreserveEmptyRows
* @return array All the read rows the given file
*/
private function getAllRowsForFile(
$fileName,
$fieldDelimiter = ',',
$fieldEnclosure = '"',
$encoding = EncodingHelper::ENCODING_UTF8)
$endOfLineCharacter = "\n",
$encoding = EncodingHelper::ENCODING_UTF8,
$shouldPreserveEmptyRows = false)
{
$allRows = [];
$resourcePath = $this->getResourcePath($fileName);
@@ -422,7 +453,9 @@ class ReaderTest extends \PHPUnit_Framework_TestCase
$reader
->setFieldDelimiter($fieldDelimiter)
->setFieldEnclosure($fieldEnclosure)
->setEndOfLineCharacter($endOfLineCharacter)
->setEncoding($encoding)
->setShouldPreserveEmptyRows($shouldPreserveEmptyRows)
->open($resourcePath);
foreach ($reader->getSheetIterator() as $sheetIndex => $sheet) {
@@ -436,51 +469,6 @@ class ReaderTest extends \PHPUnit_Framework_TestCase
return $allRows;
}
/**
* @return array
*/
public function dataProviderForTestReadCustomEOL()
{
return [
['csv_with_CR_EOL.csv', "\r"],
['csv_standard.csv', "\n"],
];
}
/**
* @dataProvider dataProviderForTestReadCustomEOL
*
* @param string $fileName
* @param string $customEOL
* @return void
*/
public function testReadCustomEOLs($fileName, $customEOL)
{
$allRows = [];
$resourcePath = $this->getResourcePath($fileName);
/** @var \Box\Spout\Reader\CSV\Reader $reader */
$reader = ReaderFactory::create(Type::CSV);
$reader
->setEndOfLineCharacter($customEOL)
->open($resourcePath);
foreach ($reader->getSheetIterator() as $sheet) {
foreach ($sheet->getRowIterator() as $row) {
$allRows[] = $row;
}
}
$reader->close();
$expectedRows = [
['csv--11', 'csv--12', 'csv--13'],
['csv--21', 'csv--22', 'csv--23'],
['csv--31', 'csv--32', 'csv--33'],
];
$this->assertEquals($expectedRows, $allRows);
}
/**
* @return void
*/
+46 -6
View File
@@ -78,6 +78,19 @@ class ReaderTest extends \PHPUnit_Framework_TestCase
$this->assertEquals([['foo']], $allRows);
}
/**
* @return void
*/
public function testReadShouldSupportNumberRowsRepeated()
{
$allRows = $this->getAllRowsForFile('sheet_with_number_rows_repeated.ods');
$expectedRows = [
['foo', 10.43],
['foo', 10.43],
];
$this->assertEquals($expectedRows, $allRows);
}
/**
* @return void
*/
@@ -211,15 +224,39 @@ class ReaderTest extends \PHPUnit_Framework_TestCase
/**
* @return void
*/
public function testReadShouldSkipEmptyRow()
public function testReadShouldSkipEmptyRowsIfShouldPreserveEmptyRowsNotSet()
{
$allRows = $this->getAllRowsForFile('sheet_with_empty_row.ods');
$this->assertEquals(2, count($allRows), 'There should be only 2 rows, because the empty row is skipped');
$allRows = $this->getAllRowsForFile('sheet_with_empty_rows.ods');
$this->assertEquals(3, count($allRows), 'There should be only 3 rows, because the empty rows are skipped');
$expectedRows = [
['ods--11', 'ods--12', 'ods--13'],
// row skipped here
// skipped row here
['ods--21', 'ods--22', 'ods--23'],
// skipped row here
// skipped row here
['ods--51', 'ods--52', 'ods--53'],
['ods--61', 'ods--62', 'ods--63'],
];
$this->assertEquals($expectedRows, $allRows);
}
/**
* @return void
*/
public function testReadShouldReturnEmptyLinesIfShouldPreserveEmptyRowsSet()
{
$allRows = $this->getAllRowsForFile('sheet_with_empty_rows.ods', false, true);
$this->assertEquals(6, count($allRows), 'There should be 6 rows');
$expectedRows = [
[''],
['ods--21', 'ods--22', 'ods--23'],
[''],
[''],
['ods--51', 'ods--52', 'ods--53'],
['ods--61', 'ods--62', 'ods--63'],
];
$this->assertEquals($expectedRows, $allRows);
}
@@ -485,15 +522,18 @@ class ReaderTest extends \PHPUnit_Framework_TestCase
/**
* @param string $fileName
* @param bool|void $shouldFormatDates
* @param bool|void $shouldPreserveEmptyRows
* @return array All the read rows the given file
*/
private function getAllRowsForFile($fileName, $shouldFormatDates = false)
private function getAllRowsForFile($fileName, $shouldFormatDates = false, $shouldPreserveEmptyRows = false)
{
$allRows = [];
$resourcePath = $this->getResourcePath($fileName);
/** @var \Box\Spout\Reader\ODS\Reader $reader */
$reader = ReaderFactory::create(Type::ODS);
$reader->setShouldFormatDates($shouldFormatDates);
$reader->setShouldPreserveEmptyRows($shouldPreserveEmptyRows);
$reader->open($resourcePath);
foreach ($reader->getSheetIterator() as $sheetIndex => $sheet) {
@@ -29,6 +29,8 @@ class DateFormatHelperTest extends \PHPUnit_Framework_TestCase
['[$USD-F480]hh:mm AM/PM', 'h:i A'],
['"Day " d', '\\D\\a\\y\\ j'],
['yy "Year" m "Month"', 'y \\Y\\e\\a\\r n \\M\\o\\n\\t\\h'],
['mmm-yy;@', 'M-y'],
['[$-409]hh:mm AM/PM;"foo"@', 'h:i A'],
];
}
@@ -11,16 +11,16 @@ class StyleHelperTest extends \PHPUnit_Framework_TestCase
{
/**
* @param array $styleAttributes
* @param array|void $styleAttributes
* @param array|void $customNumberFormats
* @return StyleHelper
*/
private function getStyleHelperMock($styleAttributes, $customNumberFormats = [])
private function getStyleHelperMock($styleAttributes = [], $customNumberFormats = [])
{
/** @var StyleHelper $styleHelper */
$styleHelper = $this->getMockBuilder('\Box\Spout\Reader\XLSX\Helper\StyleHelper')
->setConstructorArgs(['/path/to/file.xlsx'])
->setMethods(['getCustomNumberFormats', 'getStylesAttributes'])
->disableOriginalConstructor()
->getMock();
$styleHelper->method('getStylesAttributes')->willReturn($styleAttributes);
@@ -34,27 +34,17 @@ class StyleHelperTest extends \PHPUnit_Framework_TestCase
*/
public function testShouldFormatNumericValueAsDateWithDefaultStyle()
{
$styleHelper = $this->getStyleHelperMock([]);
$styleHelper = $this->getStyleHelperMock();
$shouldFormatAsDate = $styleHelper->shouldFormatNumericValueAsDate(0);
$this->assertFalse($shouldFormatAsDate);
}
/**
* @return void
*/
public function testShouldFormatNumericValueAsDateWhenStyleIdNotListed()
{
$styleHelper = $this->getStyleHelperMock([['applyNumberFormat' => true]]);
$shouldFormatAsDate = $styleHelper->shouldFormatNumericValueAsDate(1);
$this->assertFalse($shouldFormatAsDate);
}
/**
* @return void
*/
public function testShouldFormatNumericValueAsDateWhenShouldNotApplyNumberFormat()
{
$styleHelper = $this->getStyleHelperMock([[], ['applyNumberFormat' => false]]);
$styleHelper = $this->getStyleHelperMock([[], ['applyNumberFormat' => false, 'numFmtId' => 14]]);
$shouldFormatAsDate = $styleHelper->shouldFormatNumericValueAsDate(1);
$this->assertFalse($shouldFormatAsDate);
}
@@ -69,6 +59,26 @@ class StyleHelperTest extends \PHPUnit_Framework_TestCase
$this->assertFalse($shouldFormatAsDate);
}
/**
* @return void
*/
public function testShouldFormatNumericValueAsDateWithNonDateBuiltinFormat()
{
$styleHelper = $this->getStyleHelperMock([[], ['applyNumberFormat' => true, 'numFmtId' => 9]]);
$shouldFormatAsDate = $styleHelper->shouldFormatNumericValueAsDate(1);
$this->assertFalse($shouldFormatAsDate);
}
/**
* @return void
*/
public function testShouldFormatNumericValueAsDateWithNoNumFmtId()
{
$styleHelper = $this->getStyleHelperMock([[], ['applyNumberFormat' => true, 'numFmtId' => null]]);
$shouldFormatAsDate = $styleHelper->shouldFormatNumericValueAsDate(1);
$this->assertFalse($shouldFormatAsDate);
}
/**
* @return void
*/
@@ -84,6 +94,28 @@ class StyleHelperTest extends \PHPUnit_Framework_TestCase
}
}
/**
* @return void
*/
public function testShouldFormatNumericValueAsDateWhenApplyNumberFormatNotSetAndUsingBuiltinDateFormat()
{
$styleHelper = $this->getStyleHelperMock([[], ['applyNumberFormat' => null, 'numFmtId' => 14]]);
$shouldFormatAsDate = $styleHelper->shouldFormatNumericValueAsDate(1);
$this->assertTrue($shouldFormatAsDate);
}
/**
* @return void
*/
public function testShouldFormatNumericValueAsDateWhenApplyNumberFormatNotSetAndUsingBuiltinNonDateFormat()
{
$styleHelper = $this->getStyleHelperMock([[], ['applyNumberFormat' => null, 'numFmtId' => 9]]);
$shouldFormatAsDate = $styleHelper->shouldFormatNumericValueAsDate(1);
$this->assertFalse($shouldFormatAsDate);
}
/**
* @return void
*/
@@ -106,6 +138,7 @@ class StyleHelperTest extends \PHPUnit_Framework_TestCase
['[$-409]d\-mmm\-yy;@', true],
['[$-409]d\-mmm\-yyyy;@', true],
['mm/dd/yy;@', true],
['MM/DD/YY;@', true],
['[$-F800]dddd\,\ mmmm\ dd\,\ yyyy', true],
['m/d;@', true],
['m/d/yy;@', true],
@@ -117,9 +150,11 @@ class StyleHelperTest extends \PHPUnit_Framework_TestCase
['[$-409]m/d/yy\ h:mm\ AM/PM;@', true],
['m/d/yy\ h:mm;@', true],
['[$-409]mmmmm;@', true],
['[$-409]MMmmM;@', true],
['[$-409]mmmmm\-yy;@', true],
['m/d/yyyy;@', true],
['[$-409]m/d/yy\--h:mm;@', true],
['General', false],
['GENERAL', false],
['\ma\yb\e', false],
['[Red]foo;', false],
+47 -6
View File
@@ -279,6 +279,21 @@ class ReaderTest extends \PHPUnit_Framework_TestCase
$this->assertEquals($expectedRows, $allRows);
}
/**
* @return void
*/
public function testReadShouldApplyCustomDateFormatNumberEvenIfApplyNumberFormatNotSpecified()
{
$shouldFormatDates = true;
$allRows = $this->getAllRowsForFile('sheet_with_custom_date_formats_and_no_apply_number_format.xlsx', $shouldFormatDates);
$expectedRows = [
// "General", "GENERAL", "MM/DD/YYYY", "MM/dd/YYYY", "H:MM:SS"
['42382', '42382', '01/13/2016', '01/13/2016', '4:43:25'],
];
$this->assertEquals($expectedRows, $allRows);
}
/**
* @return void
*/
@@ -337,16 +352,39 @@ class ReaderTest extends \PHPUnit_Framework_TestCase
/**
* @return void
*/
public function testReadShouldSkipEmptyRows()
public function testReadShouldSkipEmptyRowsIfShouldPreserveEmptyRowsNotSet()
{
$allRows = $this->getAllRowsForFile('sheet_with_empty_row.xlsx');
$allRows = $this->getAllRowsForFile('sheet_with_empty_rows_and_missing_row_index.xlsx');
$this->assertEquals(2, count($allRows), 'There should be only 2 rows, because the empty row is skipped');
$this->assertEquals(3, count($allRows), 'There should be only 3 rows, because the empty rows are skipped');
$expectedRows = [
['s1--A1', 's1--B1', 's1--C1', 's1--D1', 's1--E1'],
// skipped row here
['s1--A3', 's1--B3', 's1--C3', 's1--D3', 's1--E3'],
['s1--A2', 's1--B2', 's1--C2'],
// skipped row here
// skipped row here
['s1--A5', 's1--B5', 's1--C5'],
['s1--A6', 's1--B6', 's1--C6'],
];
$this->assertEquals($expectedRows, $allRows);
}
/**
* @return void
*/
public function testReadShouldReturnEmptyLinesIfShouldPreserveEmptyRowsSet()
{
$allRows = $this->getAllRowsForFile('sheet_with_empty_rows_and_missing_row_index.xlsx', false, true);
$this->assertEquals(6, count($allRows), 'There should be 6 rows');
$expectedRows = [
[''],
['s1--A2', 's1--B2', 's1--C2'],
[''],
[''],
['s1--A5', 's1--B5', 's1--C5'],
['s1--A6', 's1--B6', 's1--C6'],
];
$this->assertEquals($expectedRows, $allRows);
}
@@ -580,15 +618,18 @@ class ReaderTest extends \PHPUnit_Framework_TestCase
/**
* @param string $fileName
* @param bool|void $shouldFormatDates
* @param bool|void $shouldPreserveEmptyRows
* @return array All the read rows the given file
*/
private function getAllRowsForFile($fileName, $shouldFormatDates = false)
private function getAllRowsForFile($fileName, $shouldFormatDates = false, $shouldPreserveEmptyRows = false)
{
$allRows = [];
$resourcePath = $this->getResourcePath($fileName);
/** @var \Box\Spout\Reader\XLSX\Reader $reader */
$reader = ReaderFactory::create(Type::XLSX);
$reader->setShouldFormatDates($shouldFormatDates);
$reader->setShouldPreserveEmptyRows($shouldPreserveEmptyRows);
$reader->open($resourcePath);
foreach ($reader->getSheetIterator() as $sheetIndex => $sheet) {
+45
View File
@@ -18,6 +18,9 @@ trait TestUsingResource
/** @var string Path to the test resources folder, that does not have writing permissions */
private $generatedUnwritableResourcesPath = 'tests/resources/generated/unwritable';
/** @var string Path to the test temp folder */
private $tempFolderPath = 'tests/resources/generated/temp';
/**
* @param string $resourceName
* @return string|null Path of the resource who matches the given name or null if resource not found
@@ -86,6 +89,48 @@ trait TestUsingResource
}
}
/**
* @return string Path of the temp folder
*/
protected function getTempFolderPath()
{
return realpath($this->tempFolderPath);
}
/**
* @return void
*/
protected function recreateTempFolder()
{
if (file_exists($this->tempFolderPath)) {
$this->deleteFolderRecursively($this->tempFolderPath);
}
mkdir($this->tempFolderPath, 0777, true);
}
/**
* @param string $folderPath
* @return void
*/
private function deleteFolderRecursively($folderPath)
{
$itemIterator = new \RecursiveIteratorIterator(
new \RecursiveDirectoryIterator($folderPath, \RecursiveDirectoryIterator::SKIP_DOTS),
\RecursiveIteratorIterator::CHILD_FIRST
);
foreach ($itemIterator as $item) {
if ($item->isDir()) {
rmdir($item->getPathname());
} else {
unlink($item->getPathname());
}
}
rmdir($folderPath);
}
/**
* @return bool Whether the OS on which PHP is installed is Windows
*/
@@ -35,6 +35,23 @@ class CellHelperTest extends \PHPUnit_Framework_TestCase
$this->assertEquals($expectedCellIndex, CellHelper::getCellIndexFromColumnIndex($columnIndex));
}
/**
* @return array
*/
public function testIsEmpty()
{
$this->assertTrue(CellHelper::isEmpty(null));
$this->assertTrue(CellHelper::isEmpty(""));
$this->assertFalse(CellHelper::isEmpty("string"));
$this->assertFalse(CellHelper::isEmpty(0));
$this->assertFalse(CellHelper::isEmpty(1));
$this->assertFalse(CellHelper::isEmpty(true));
$this->assertFalse(CellHelper::isEmpty(false));
$this->assertFalse(CellHelper::isEmpty(["string"]));
$this->assertFalse(CellHelper::isEmpty(new \stdClass()));
}
/**
* @return array
*/
+34
View File
@@ -2,6 +2,7 @@
namespace Box\Spout\Writer\ODS;
use Box\Spout\Common\Exception\SpoutException;
use Box\Spout\Common\Type;
use Box\Spout\Reader\Wrapper\XMLReader;
use Box\Spout\TestUsingResource;
@@ -91,6 +92,39 @@ class WriterTest extends \PHPUnit_Framework_TestCase
$this->writeToODSFile($dataRows, $fileName);
}
/**
* @return void
*/
public function testAddRowShouldCleanupAllFilesIfExceptionIsThrown()
{
$fileName = 'test_add_row_should_cleanup_all_files_if_exception_thrown.ods';
$dataRows = [
['wrong'],
[new \stdClass()],
];
$this->createGeneratedFolderIfNeeded($fileName);
$resourcePath = $this->getGeneratedResourcePath($fileName);
$this->recreateTempFolder();
$tempFolderPath = $this->getTempFolderPath();
/** @var \Box\Spout\Writer\ODS\Writer $writer */
$writer = WriterFactory::create(Type::ODS);
$writer->setTempFolder($tempFolderPath);
$writer->openToFile($resourcePath);
try {
$writer->addRows($dataRows);
$this->fail('Exception should have been thrown');
} catch (SpoutException $e) {
$this->assertFalse(file_exists($fileName), 'Output file should have been deleted');
$numFiles = iterator_count(new \FilesystemIterator($tempFolderPath, \FilesystemIterator::SKIP_DOTS));
$this->assertEquals(0, $numFiles, 'All temp files should have been deleted');
}
}
/**
* @return void
*/
+58 -1
View File
@@ -2,6 +2,7 @@
namespace Box\Spout\Writer\XLSX;
use Box\Spout\Common\Exception\SpoutException;
use Box\Spout\Common\Type;
use Box\Spout\TestUsingResource;
use Box\Spout\Writer\WriterFactory;
@@ -104,6 +105,39 @@ class WriterTest extends \PHPUnit_Framework_TestCase
$this->writeToXLSXFile($dataRows, $fileName);
}
/**
* @return void
*/
public function testAddRowShouldCleanupAllFilesIfExceptionIsThrown()
{
$fileName = 'test_add_row_should_cleanup_all_files_if_exception_thrown.xlsx';
$dataRows = [
['wrong'],
[new \stdClass()],
];
$this->createGeneratedFolderIfNeeded($fileName);
$resourcePath = $this->getGeneratedResourcePath($fileName);
$this->recreateTempFolder();
$tempFolderPath = $this->getTempFolderPath();
/** @var \Box\Spout\Writer\XLSX\Writer $writer */
$writer = WriterFactory::create(Type::XLSX);
$writer->setTempFolder($tempFolderPath);
$writer->openToFile($resourcePath);
try {
$writer->addRows($dataRows);
$this->fail('Exception should have been thrown');
} catch (SpoutException $e) {
$this->assertFalse(file_exists($fileName), 'Output file should have been deleted');
$numFiles = iterator_count(new \FilesystemIterator($tempFolderPath, \FilesystemIterator::SKIP_DOTS));
$this->assertEquals(0, $numFiles, 'All temp files should have been deleted');
}
}
/**
* @return void
*/
@@ -242,7 +276,7 @@ class WriterTest extends \PHPUnit_Framework_TestCase
['foo' => 'xlsx--11', 'bar' => 'xlsx--12'],
];
$this->writeToXLSXFile($dataRows, $fileName, $shouldUseInlineStrings = true);
$this->writeToXLSXFile($dataRows, $fileName);
foreach ($dataRows as $dataRow) {
foreach ($dataRow as $cellValue) {
@@ -251,6 +285,29 @@ class WriterTest extends \PHPUnit_Framework_TestCase
}
}
/**
* @return void
*/
public function testAddRowShouldNotWriteEmptyRows()
{
$fileName = 'test_add_row_should_not_write_empty_rows.xlsx';
$dataRows = [
[''],
['xlsx--21', 'xlsx--22'],
[''],
[''],
['xlsx--51', 'xlsx--52'],
];
$this->writeToXLSXFile($dataRows, $fileName);
$this->assertInlineDataWasWrittenToSheet($fileName, 1, 'row r="2"');
$this->assertInlineDataWasWrittenToSheet($fileName, 1, 'row r="5"');
$this->assertInlineDataWasNotWrittenToSheet($fileName, 1, 'row r="1"');
$this->assertInlineDataWasNotWrittenToSheet($fileName, 1, 'row r="3"');
$this->assertInlineDataWasNotWrittenToSheet($fileName, 1, 'row r="4"');
}
/**
* @return void
*/
@@ -1,2 +0,0 @@
csv--11,csv--12,csv--13
csv--31,csv--32,csv--33
1 csv--11 csv--12 csv--13
2 csv--31 csv--32 csv--33
@@ -1,3 +0,0 @@
csv--11,csv--12,csv--13
csv--31,csv--32,csv--33
1 csv--11 csv--12 csv--13
2 csv--31 csv--32 csv--33
@@ -0,0 +1,5 @@
csv--21,csv--22,csv--23
csv--41,csv--42,csv--43
1 csv--21 csv--22 csv--23
2 csv--41 csv--42 csv--43
Binary file not shown.
Binary file not shown.
Binary file not shown.