first commit

modules/x13import/tools/Spout/Reader/CSV/Reader.php

@@ -0,0 +1,149 @@
+<?php
+
+namespace Box\Spout\Reader\CSV;
+
+use Box\Spout\Reader\AbstractReader;
+use Box\Spout\Common\Exception\IOException;
+
+/**
+ * Class Reader
+ * This class provides support to read data from a CSV file.
+ *
+ * @package Box\Spout\Reader\CSV
+ */
+class Reader extends AbstractReader
+{
+    /** @var resource Pointer to the file to be written */
+    protected $filePointer;
+
+    /** @var SheetIterator To iterator over the CSV unique "sheet" */
+    protected $sheetIterator;
+
+    /** @var string Original value for the "auto_detect_line_endings" INI value */
+    protected $originalAutoDetectLineEndings;
+
+    /**
+     * Returns the reader's current options
+     *
+     * @return ReaderOptions
+     */
+    protected function getOptions()
+    {
+        if (!isset($this->options)) {
+            $this->options = new ReaderOptions();
+        }
+        return $this->options;
+    }
+
+    /**
+     * Sets the field delimiter for the CSV.
+     * Needs to be called before opening the reader.
+     *
+     * @param string $fieldDelimiter Character that delimits fields
+     * @return Reader
+     */
+    public function setFieldDelimiter($fieldDelimiter)
+    {
+        $this->getOptions()->setFieldDelimiter($fieldDelimiter);
+        return $this;
+    }
+
+    /**
+     * Sets the field enclosure for the CSV.
+     * Needs to be called before opening the reader.
+     *
+     * @param string $fieldEnclosure Character that enclose fields
+     * @return Reader
+     */
+    public function setFieldEnclosure($fieldEnclosure)
+    {
+        $this->getOptions()->setFieldEnclosure($fieldEnclosure);
+        return $this;
+    }
+
+    /**
+     * Sets the encoding of the CSV file to be read.
+     * Needs to be called before opening the reader.
+     *
+     * @param string $encoding Encoding of the CSV file to be read
+     * @return Reader
+     */
+    public function setEncoding($encoding)
+    {
+        $this->getOptions()->setEncoding($encoding);
+        return $this;
+    }
+
+    /**
+     * Sets the EOL for the CSV.
+     * Needs to be called before opening the reader.
+     *
+     * @param string $endOfLineCharacter used to properly get lines from the CSV file.
+     * @return Reader
+     */
+    public function setEndOfLineCharacter($endOfLineCharacter)
+    {
+        $this->getOptions()->setEndOfLineCharacter($endOfLineCharacter);
+        return $this;
+    }
+
+    /**
+     * Returns whether stream wrappers are supported
+     *
+     * @return bool
+     */
+    protected function doesSupportStreamWrapper()
+    {
+        return true;
+    }
+
+    /**
+     * Opens the file at the given path to make it ready to be read.
+     * If setEncoding() was not called, it assumes that the file is encoded in UTF-8.
+     *
+     * @param  string $filePath Path of the CSV file to be read
+     * @return void
+     * @throws \Box\Spout\Common\Exception\IOException
+     */
+    protected function openReader($filePath)
+    {
+        $this->originalAutoDetectLineEndings = ini_get('auto_detect_line_endings');
+        ini_set('auto_detect_line_endings', '1');
+
+        $this->filePointer = $this->globalFunctionsHelper->fopen($filePath, 'r');
+        if (!$this->filePointer) {
+            throw new IOException("Could not open file $filePath for reading.");
+        }
+
+        $this->sheetIterator = new SheetIterator(
+            $this->filePointer,
+            $this->getOptions(),
+            $this->globalFunctionsHelper
+        );
+    }
+
+    /**
+     * Returns an iterator to iterate over sheets.
+     *
+     * @return SheetIterator To iterate over sheets
+     */
+    protected function getConcreteSheetIterator()
+    {
+        return $this->sheetIterator;
+    }
+
+
+    /**
+     * Closes the reader. To be used after reading the file.
+     *
+     * @return void
+     */
+    protected function closeReader()
+    {
+        if ($this->filePointer) {
+            $this->globalFunctionsHelper->fclose($this->filePointer);
+        }
+
+        ini_set('auto_detect_line_endings', $this->originalAutoDetectLineEndings);
+    }
+}

modules/x13import/tools/Spout/Reader/CSV/ReaderOptions.php

@@ -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;
+    }
+}

modules/x13import/tools/Spout/Reader/CSV/RowIterator.php

@@ -0,0 +1,260 @@
+<?php
+
+namespace Box\Spout\Reader\CSV;
+
+use Box\Spout\Reader\IteratorInterface;
+use Box\Spout\Common\Helper\EncodingHelper;
+
+/**
+ * Class RowIterator
+ * Iterate over CSV rows.
+ *
+ * @package Box\Spout\Reader\CSV
+ */
+class RowIterator implements IteratorInterface
+{
+    /**
+     * Value passed to fgetcsv. 0 means "unlimited" (slightly slower but accomodates for very long lines).
+     */
+    const MAX_READ_BYTES_PER_LINE = 0;
+
+    /** @var resource Pointer to the CSV file to read */
+    protected $filePointer;
+
+    /** @var int Number of read rows */
+    protected $numReadRows = 0;
+
+    /** @var array|null Buffer used to store the row data, while checking if there are more rows to read */
+    protected $rowDataBuffer = null;
+
+    /** @var bool Indicates whether all rows have been read */
+    protected $hasReachedEndOfFile = false;
+
+    /** @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;
+
+    /** @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;
+
+    /** @var \Box\Spout\Common\Helper\EncodingHelper Helper to work with different encodings */
+    protected $encodingHelper;
+
+    /** @var string End of line delimiter, encoded using the same encoding as the CSV */
+    protected $encodedEOLDelimiter;
+
+    /**
+     * @param resource $filePointer Pointer to the CSV file to read
+     * @param \Box\Spout\Reader\CSV\ReaderOptions $options
+     * @param \Box\Spout\Common\Helper\GlobalFunctionsHelper $globalFunctionsHelper
+     */
+    public function __construct($filePointer, $options, $globalFunctionsHelper)
+    {
+        $this->filePointer = $filePointer;
+        $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);
+    }
+
+    /**
+     * Rewind the Iterator to the first element
+     * @link http://php.net/manual/en/iterator.rewind.php
+     *
+     * @return void
+     */
+    public function rewind()
+    {
+        $this->rewindAndSkipBom();
+
+        $this->numReadRows = 0;
+        $this->rowDataBuffer = null;
+
+        $this->next();
+    }
+
+    /**
+     * This rewinds and skips the BOM if inserted at the beginning of the file
+     * by moving the file pointer after it, so that it is not read.
+     *
+     * @return void
+     */
+    protected function rewindAndSkipBom()
+    {
+        $byteOffsetToSkipBom = $this->encodingHelper->getBytesOffsetToSkipBOM($this->filePointer, $this->encoding);
+
+        // sets the cursor after the BOM (0 means no BOM, so rewind it)
+        $this->globalFunctionsHelper->fseek($this->filePointer, $byteOffsetToSkipBom);
+    }
+
+    /**
+     * Checks if current position is valid
+     * @link http://php.net/manual/en/iterator.valid.php
+     *
+     * @return bool
+     */
+    public function valid()
+    {
+        return ($this->filePointer && !$this->hasReachedEndOfFile);
+    }
+
+    /**
+     * Move forward to next element. Reads data for the next unprocessed row.
+     * @link http://php.net/manual/en/iterator.next.php
+     *
+     * @return void
+     * @throws \Box\Spout\Common\Exception\EncodingConversionException If unable to convert data to UTF-8
+     */
+    public function next()
+    {
+        $this->hasReachedEndOfFile = $this->globalFunctionsHelper->feof($this->filePointer);
+
+        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 = $this->getNextUTF8EncodedRow();
+        } while ($this->shouldReadNextRow($rowData));
+
+        if ($rowData !== false) {
+            // 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 = 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,
+     * we remove manually whitespace with ltrim or rtrim (depending on the order of the bytes)
+     *
+     * @return array|false The row for the current file pointer, encoded in UTF-8 or FALSE if nothing to read
+     * @throws \Box\Spout\Common\Exception\EncodingConversionException If unable to convert data to UTF-8
+     */
+    protected function getNextUTF8EncodedRow()
+    {
+        $encodedRowData = $this->globalFunctionsHelper->fgetcsv($this->filePointer, self::MAX_READ_BYTES_PER_LINE, $this->fieldDelimiter, $this->fieldEnclosure);
+        if ($encodedRowData === false) {
+            return false;
+        }
+
+        foreach ($encodedRowData as $cellIndex => $cellValue) {
+            switch($this->encoding) {
+                case EncodingHelper::ENCODING_UTF16_LE:
+                case EncodingHelper::ENCODING_UTF32_LE:
+                    // remove whitespace from the beginning of a string as fgetcsv() add extra whitespace when it try to explode non UTF-8 data
+                    $cellValue = ltrim($cellValue);
+                    break;
+
+                case EncodingHelper::ENCODING_UTF16_BE:
+                case EncodingHelper::ENCODING_UTF32_BE:
+                    // remove whitespace from the end of a string as fgetcsv() add extra whitespace when it try to explode non UTF-8 data
+                    $cellValue = rtrim($cellValue);
+                    break;
+            }
+
+            $encodedRowData[$cellIndex] = $this->encodingHelper->attemptConversionToUTF8($cellValue, $this->encoding);
+        }
+
+        return $encodedRowData;
+    }
+
+    /**
+     * Returns the end of line delimiter, encoded using the same encoding as the CSV.
+     * The return value is cached.
+     *
+     * @return string
+     */
+    protected function getEncodedEOLDelimiter()
+    {
+        if (!isset($this->encodedEOLDelimiter)) {
+            $this->encodedEOLDelimiter = $this->encodingHelper->attemptConversionFromUTF8($this->inputEOLDelimiter, $this->encoding);
+        }
+
+        return $this->encodedEOLDelimiter;
+    }
+
+    /**
+     * @param array|bool $lineData Array containing the cells value for the line
+     * @return bool Whether the given line is empty
+     */
+    protected function isEmptyLine($lineData)
+    {
+        return (is_array($lineData) && count($lineData) === 1 && $lineData[0] === null);
+    }
+
+    /**
+     * Return the current element from the buffer
+     * @link http://php.net/manual/en/iterator.current.php
+     *
+     * @return array|null
+     */
+    public function current()
+    {
+        return $this->rowDataBuffer;
+    }
+
+    /**
+     * Return the key of the current element
+     * @link http://php.net/manual/en/iterator.key.php
+     *
+     * @return int
+     */
+    public function key()
+    {
+        return $this->numReadRows;
+    }
+
+    /**
+     * Cleans up what was created to iterate over the object.
+     *
+     * @return void
+     */
+    public function end()
+    {
+        // do nothing
+    }
+}

modules/x13import/tools/Spout/Reader/CSV/Sheet.php

@@ -0,0 +1,62 @@
+<?php
+
+namespace Box\Spout\Reader\CSV;
+
+use Box\Spout\Reader\SheetInterface;
+
+/**
+ * Class Sheet
+ *
+ * @package Box\Spout\Reader\CSV
+ */
+class Sheet implements SheetInterface
+{
+    /** @var \Box\Spout\Reader\CSV\RowIterator To iterate over the CSV's rows */
+    protected $rowIterator;
+
+    /**
+     * @param resource $filePointer Pointer to the CSV file to read
+     * @param \Box\Spout\Reader\CSV\ReaderOptions $options
+     * @param \Box\Spout\Common\Helper\GlobalFunctionsHelper $globalFunctionsHelper
+     */
+    public function __construct($filePointer, $options, $globalFunctionsHelper)
+    {
+        $this->rowIterator = new RowIterator($filePointer, $options, $globalFunctionsHelper);
+    }
+
+    /**
+     * @api
+     * @return \Box\Spout\Reader\CSV\RowIterator
+     */
+    public function getRowIterator()
+    {
+        return $this->rowIterator;
+    }
+
+    /**
+     * @api
+     * @return int Index of the sheet
+     */
+    public function getIndex()
+    {
+        return 0;
+    }
+
+    /**
+     * @api
+     * @return string Name of the sheet - empty string since CSV does not support that
+     */
+    public function getName()
+    {
+        return '';
+    }
+
+    /**
+     * @api
+     * @return bool Always TRUE as there is only one sheet
+     */
+    public function isActive()
+    {
+        return true;
+    }
+}

modules/x13import/tools/Spout/Reader/CSV/SheetIterator.php

@@ -0,0 +1,95 @@
+<?php
+
+namespace Box\Spout\Reader\CSV;
+
+use Box\Spout\Reader\IteratorInterface;
+
+/**
+ * Class SheetIterator
+ * Iterate over CSV unique "sheet".
+ *
+ * @package Box\Spout\Reader\CSV
+ */
+class SheetIterator implements IteratorInterface
+{
+    /** @var \Box\Spout\Reader\CSV\Sheet The CSV unique "sheet" */
+    protected $sheet;
+
+    /** @var bool Whether the unique "sheet" has already been read */
+    protected $hasReadUniqueSheet = false;
+
+    /**
+     * @param resource $filePointer
+     * @param \Box\Spout\Reader\CSV\ReaderOptions $options
+     * @param \Box\Spout\Common\Helper\GlobalFunctionsHelper $globalFunctionsHelper
+     */
+    public function __construct($filePointer, $options, $globalFunctionsHelper)
+    {
+        $this->sheet = new Sheet($filePointer, $options, $globalFunctionsHelper);
+    }
+
+    /**
+     * Rewind the Iterator to the first element
+     * @link http://php.net/manual/en/iterator.rewind.php
+     *
+     * @return void
+     */
+    public function rewind()
+    {
+        $this->hasReadUniqueSheet = false;
+    }
+
+    /**
+     * Checks if current position is valid
+     * @link http://php.net/manual/en/iterator.valid.php
+     *
+     * @return bool
+     */
+    public function valid()
+    {
+        return (!$this->hasReadUniqueSheet);
+    }
+
+    /**
+     * Move forward to next element
+     * @link http://php.net/manual/en/iterator.next.php
+     *
+     * @return void
+     */
+    public function next()
+    {
+        $this->hasReadUniqueSheet = true;
+    }
+
+    /**
+     * Return the current element
+     * @link http://php.net/manual/en/iterator.current.php
+     *
+     * @return \Box\Spout\Reader\CSV\Sheet
+     */
+    public function current()
+    {
+        return $this->sheet;
+    }
+
+    /**
+     * Return the key of the current element
+     * @link http://php.net/manual/en/iterator.key.php
+     *
+     * @return int
+     */
+    public function key()
+    {
+        return 1;
+    }
+
+    /**
+     * Cleans up what was created to iterate over the object.
+     *
+     * @return void
+     */
+    public function end()
+    {
+        // do nothing
+    }
+}