Katie Horn has submitted this change and it was merged.
Change subject: A Streaming CSV Reader for SmashPig
......................................................................
A Streaming CSV Reader for SmashPig
These classes allow the stream processing of CSV files in a 'nicely'
wrapped way.
These were needed so that I could process Adyen reports.
Change-Id: I6cf6656bac25c557f724ea6b296a3adab0243e52
---
A Core/DataFiles/CsvReader.php
A Core/DataFiles/DataFileException.php
A Core/DataFiles/HeadedCsvReader.php
3 files changed, 149 insertions(+), 0 deletions(-)
Approvals:
Katie Horn: Verified; Looks good to me, approved
diff --git a/Core/DataFiles/CsvReader.php b/Core/DataFiles/CsvReader.php
new file mode 100644
index 0000000..d3affad
--- /dev/null
+++ b/Core/DataFiles/CsvReader.php
@@ -0,0 +1,96 @@
+<?php namespace SmashPig\Core\DataFiles;
+
+class CsvReader implements \Iterator {
+ /**
+ * @var int Maximum length of line to read from the CSV file.
+ */
+ protected $maxRowLength = 4096;
+
+ /**
+ * @var resource The pointer to the cvs file.
+ */
+ protected $filePointer = null;
+
+ /**
+ * @var array The current element, which will be returned on each
iteration.
+ */
+ protected $currentElement = null;
+
+ /**
+ * @var int Number of rows read so far.
+ */
+ protected $rowCounter = null;
+
+ /**
+ * @var string Delimiter for the csv file.
+ */
+ private $delimiter = null;
+
+ /**
+ * Create an iterative CSV file reader.
+ *
+ * @param string $file Path to file
+ * @param string $delimiter Delimiter to use between CSV fields
+ * @param int $maxRowLength Maximum length a field can take (affects
buffering)
+ *
+ * @throws DataFileException on non open-able file.
+ */
+ public function __construct( $file, $delimiter = ',', $maxRowLength =
4098 ) {
+ $this->filePointer = fopen( $file, 'r' );
+ if ( !$this->filePointer ) {
+ throw new DataFileException( "Could not open file
'{$file}' for reading." );
+ }
+
+ $this->delimiter = $delimiter;
+ $this->maxRowLength = $maxRowLength;
+
+ // Read the first row
+ $this->next();
+ }
+
+ /**
+ * Close the file pointer
+ */
+ public function __destruct() {
+ fclose( $this->filePointer );
+ }
+
+ /**
+ * Rewind to the first element.
+ */
+ public function rewind() {
+ $this->rowCounter = 0;
+ rewind( $this->filePointer );
+ }
+
+ /**
+ * @return mixed[] The currently buffered CSV row.
+ * @throws DataFileException If no data has been loaded.
+ */
+ public function current() {
+ return $this->currentElement;
+ }
+
+ /**
+ * @return int The current row number
+ */
+ public function key() {
+ return $this->rowCounter;
+ }
+
+ /**
+ * Load the next rows into memory
+ */
+ public function next() {
+ $this->currentElement = fgetcsv( $this->filePointer,
$this->maxRowLength, $this->delimiter );
+ $this->rowCounter++;
+ }
+
+ /**
+ * Check to see if we have any valid data yet to retrieve
+ * @return bool
+ */
+ public function valid() {
+ return ( $this->currentElement !== false ) && !feof(
$this->filePointer );
+ }
+}
diff --git a/Core/DataFiles/DataFileException.php
b/Core/DataFiles/DataFileException.php
new file mode 100644
index 0000000..061d9c4
--- /dev/null
+++ b/Core/DataFiles/DataFileException.php
@@ -0,0 +1,7 @@
+<?php namespace SmashPig\Core\DataFiles;
+
+use SmashPig\Core\SmashPigException;
+
+class DataFileException extends SmashPigException {
+
+}
\ No newline at end of file
diff --git a/Core/DataFiles/HeadedCsvReader.php
b/Core/DataFiles/HeadedCsvReader.php
new file mode 100644
index 0000000..1e84730
--- /dev/null
+++ b/Core/DataFiles/HeadedCsvReader.php
@@ -0,0 +1,46 @@
+<?php namespace SmashPig\Core\DataFiles;
+
+/**
+ * Iteratively reads a CSV file that contains a data header
+ *
+ * @package SmashPig\Core\DataFiles
+ */
+class HeadedCsvReader extends CsvReader {
+ /** @var string[] */
+ protected $colNames;
+
+ public function __construct( $file, $delimiter = ',', $maxRowLength =
4098 ) {
+ parent::__construct( $file, $delimiter, $maxRowLength );
+
+ // Extract the header information
+ $this->colNames = parent::current();
+ parent::next();
+ }
+
+ /**
+ * Extract the contents of the given column name.
+ *
+ * This is slightly backwards because it did not seem worth the effort
+ * to create a fully functional ArrayObject class to return from
current()
+ *
+ * @param $colName string Name of the column to extract
+ * @param $row string[] A row returned from current() **FROM THIS FILE**
+ *
+ * @throws DataFileException if the column name does not exist.
+ * @return string Contents of the column
+ */
+ public function extractCol( $colName, &$row ) {
+ $col = array_search( $colName, $this->colNames );
+ if ( $col === false ) {
+ throw new DataFileException( "Column name
{$this->colNames} not found!" );
+ }
+ return $row[$col];
+ }
+
+ /**
+ * @return string[] CSV file headers in order of columns
+ */
+ public function headers() {
+ return $this->colNames;
+ }
+}
\ No newline at end of file
--
To view, visit https://gerrit.wikimedia.org/r/90283
To unsubscribe, visit https://gerrit.wikimedia.org/r/settings
Gerrit-MessageType: merged
Gerrit-Change-Id: I6cf6656bac25c557f724ea6b296a3adab0243e52
Gerrit-PatchSet: 2
Gerrit-Project: wikimedia/fundraising/SmashPig
Gerrit-Branch: master
Gerrit-Owner: Mwalker <[email protected]>
Gerrit-Reviewer: Adamw <[email protected]>
Gerrit-Reviewer: Katie Horn <[email protected]>
_______________________________________________
MediaWiki-commits mailing list
[email protected]
https://lists.wikimedia.org/mailman/listinfo/mediawiki-commits
