Improve codebase

- BOM value are moved to an Interface - BOM functions are introduced - XML conversion is moved to its own class - MapIterator is improved and simplified
thephpleague · Mar 16, 2017 · c4412d9 · c4412d9
1 parent 473a0b9
commit c4412d9
Show file tree

Hide file tree

Showing 17 changed files with 478 additions and 179 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -9,6 +9,9 @@ All Notable changes to `Csv` will be documented in this file
 - `League\Csv\BOM`
 - `League\Csv\Statement`
 - `League\Csv\RecordSet`
+- `League\Csv\XMLEncoder`
+- `League\Csv\is_bom` function
+- `League\Csv\bom_match` function
 - `League\Csv\Exception\CsvException`
 - `League\Csv\Exception\InsertionException`
 - `League\Csv\Exception\InvalidArgumentException`

diff --git a/autoload.php b/autoload.php
@@ -1,5 +1,7 @@
 <?php
 
+require __DIR__.DIRECTORY_SEPARATOR.'src'.DIRECTORY_SEPARATOR.'functions_include.php';
+
 spl_autoload_register(function ($class) {
 
     $prefix = 'League\Csv\\';

diff --git a/composer.json b/composer.json
@@ -29,7 +29,8 @@
     "autoload": {
         "psr-4": {
             "League\\Csv\\": "src"
-        }
+        },
+        "files": ["src/functions_include.php"]
     },
     "autoload-dev": {
         "psr-4": {

diff --git a/docs/9.0/connections/bom.md b/docs/9.0/connections/bom.md
@@ -5,7 +5,6 @@ title: BOM sequence detection and addition
 
 # BOM sequences
 
-
 ## Detecting the BOM sequence
 
 ~~~php
@@ -16,12 +15,9 @@ const BOM::UTF16_BE = "\xFE\xFF";
 const BOM::UTF16_LE = "\xFF\xFE";
 const BOM::UTF32_BE = "\x00\x00\xFE\xFF";
 const BOM::UTF32_LE = "\xFF\xFE\x00\x00";
-
-public static BOM::match(string $str): string
-public static BOM::isValid(string $sequence): bool
 ~~~
 
-To improve interoperability with programs interacting with CSV, the package now provide a Enum like class `BOM` to help you detect the presence of the <abbr title="Byte Order Mark">BOM</abbr> sequence in any string.
+To improve interoperability with programs interacting with CSV, the package now provide an interface `BOM` to help you detect set the appropriate <abbr title="Byte Order Mark">BOM</abbr> sequence.
 
 ### Constants
 
@@ -33,32 +29,6 @@ To ease BOM sequence manipulation, the `BOM` class provides the following consta
 * `BOM::UTF32_BE` : `UTF-32` `BOM` with Big-Endian;
 * `BOM::UTF32_LE` : `UTF-32` `BOM` with Little-Endian;
 
-### Methods
-
-~~~php
-<?php
-
-public static BOM::match(string $str): string
-public static BOM::isValid(string $sequence): bool
-~~~
-
-If you have a string and wonders if it begins with a known BOM sequence you can use the `BOM::match` method. This method will try to detect the correct BOM sequence at the start of your string. If no BOM sequence is found, an empty string is returned.
-
-At any given time you can validate your BOM sequence against the class BOM list.
-
-~~~php
-<?php
-
-use League\Csv\BOM;
-
-$str = BOM::UTF8.'The quick brown fox jumps over the lazy dog';
-$found = BOM::match($str); //returns "\xEF\xBB\xBF"
-BOM::isValid($found); //return true
-str2 = 'The quick brown fox jumps over the lazy dog';
-$found2 = BOM::match($str2); //returns ""
-BOM::isValid($found2); //return false
-~~~
-
 ## Managing CSV documents BOM sequence
 
 ### Detecting the BOM sequence

diff --git a/docs/upgrading/upgrading-9.0.md b/docs/upgrading/upgrading-9.0.md
@@ -333,7 +333,7 @@ $records = $stmt->process($reader)->fetchAll();
 
 ### BOM handling
 
-The BOM sequence constants are now attached to a `BOM` enum like class
+The BOM sequence constants are now attached to a `BOM` interface
 
 Before:
 

diff --git a/phpunit.xml b/phpunit.xml
@@ -18,8 +18,11 @@
     </testsuites>
 
     <filter>
-        <whitelist addUncoveredFilesFromWhitelist="true">
+        <whitelist>
             <directory suffix=".php">src</directory>
+            <exclude>
+                <directory suffix="include.php">src</directory>
+            </exclude>
         </whitelist>
     </filter>
 

diff --git a/src/AbstractCsv.php b/src/AbstractCsv.php
@@ -17,6 +17,7 @@
 use League\Csv\Exception\LogicException;
 use League\Csv\Exception\RuntimeException;
 use SplFileObject;
+use function League\Csv\bom_match;
 
 /**
  *  An abstract class to enable basic CSV manipulation
@@ -218,7 +219,7 @@ public function getInputBOM(): string
         if (null === $this->input_bom) {
             $this->document->setFlags(SplFileObject::READ_CSV);
             $this->document->rewind();
-            $this->input_bom = BOM::match($this->document->fgets());
+            $this->input_bom = bom_match($this->document->fgets());
         }
 
         return $this->input_bom;

diff --git a/src/BOM.php b/src/BOM.php
@@ -14,8 +14,6 @@
 
 namespace League\Csv;
 
-use ReflectionClass;
-
 /**
  *  Defines constants for common BOM sequences
  *
@@ -24,7 +22,7 @@
  * @author  Ignace Nyamagana Butera <nyamsprod@gmail.com>
  *
  */
-final class BOM
+interface BOM
 {
     /**
      *  UTF-8 BOM sequence
@@ -50,50 +48,4 @@ final class BOM
      * UTF-32 LE BOM sequence
      */
     const UTF32_LE = "\xFF\xFE\x00\x00";
-
-    /**
-     * Returns all possible BOM sequences as an array
-     *
-     * @return string[]
-     */
-    private static function toArray(): array
-    {
-        static $cache;
-
-        $cache = $cache ?? (new ReflectionClass(BOM::class))->getConstants();
-
-        return $cache;
-    }
-
-    /**
-     * Returns the BOM sequence found at the start of the string
-     *
-     * If no valid BOM sequence is found an empty string is returned
-     *
-     * @param string $str
-     *
-     * @return string
-     */
-    public static function match(string $str): string
-    {
-        foreach (self::toArray() as $sequence) {
-            if (0 === strpos($str, $sequence)) {
-                return $sequence;
-            }
-        }
-
-        return '';
-    }
-
-    /**
-     * Tell whether the submitted sequence is a valid BOM sequence
-     *
-     * @param string $sequence
-     *
-     * @return bool
-     */
-    public static function isValid(string $sequence): bool
-    {
-        return in_array($sequence, self::toArray(), true);
-    }
 }
diff --git a/src/MapIterator.php b/src/MapIterator.php
@@ -14,8 +14,8 @@
 
 namespace League\Csv;
 
-use Iterator;
 use IteratorIterator;
+use Traversable;
 
 /**
  *  A simple MapIterator
@@ -33,15 +33,15 @@ class MapIterator extends IteratorIterator
      *
      * @var callable
      */
-    private $callable;
+    protected $callable;
 
     /**
      * The Constructor
      *
-     * @param Iterator $iterator
-     * @param callable $callable
+     * @param Traversable $iterator
+     * @param callable    $callable
      */
-    public function __construct(Iterator $iterator, callable $callable)
+    public function __construct(Traversable $iterator, callable $callable)
     {
         parent::__construct($iterator);
         $this->callable = $callable;
@@ -52,8 +52,6 @@ public function __construct(Iterator $iterator, callable $callable)
      */
     public function current()
     {
-        $iterator = $this->getInnerIterator();
-
-        return ($this->callable)($iterator->current(), $iterator->key());
+        return ($this->callable)(parent::current(), $this->key());
     }
 }
diff --git a/src/RecordSet.php b/src/RecordSet.php
@@ -17,7 +17,6 @@
 use CallbackFilterIterator;
 use Countable;
 use DOMDocument;
-use DOMElement;
 use Generator;
 use Iterator;
 use IteratorAggregate;
@@ -178,20 +177,18 @@ protected function convertToUtf8(Iterator $iterator): Iterator
             return $iterator;
         }
 
-        $convert_cell = function ($value) : string {
-            return mb_convert_encoding((string) $value, 'UTF-8', $this->conversion_input_encoding);
+        $walker = function (&$value, &$offset) {
+            $value = mb_convert_encoding((string) $value, 'UTF-8', $this->conversion_input_encoding);
+            $offset = mb_convert_encoding((string) $offset, 'UTF-8', $this->conversion_input_encoding);
         };
 
-        $convert_record = function (array $record) use ($convert_cell): array {
-            $res = [];
-            foreach ($record as $key => $value) {
-                $res[$convert_cell($key)] = $convert_cell($value);
-            }
+        $convert = function (array $record) use ($walker): array {
+            array_walk($record, $walker);
 
-            return $res;
+            return $record;
         };
 
-        return new MapIterator($iterator, $convert_record);
+        return new MapIterator($iterator, $convert);
     }
 
     /**
@@ -218,7 +215,7 @@ public function toHTML(
      * Transforms a CSV into a XML
      *
      * @param string $root_name   XML root node name
-     * @param string $row_name    XML row node name
+     * @param string $record_name XML row node name
      * @param string $cell_name   XML cell node name
      * @param string $column_attr XML column attribute name
      * @param string $offset_attr XML offset attribute name
@@ -227,66 +224,22 @@ public function toHTML(
      */
     public function toXML(
         string $root_name = 'csv',
-        string $row_name = 'row',
+        string $record_name = 'row',
         string $cell_name = 'cell',
         string $column_attr = 'name',
         string $offset_attr = 'offset'
     ): DOMDocument {
-        $doc = new DOMDocument('1.0', 'UTF-8');
-        $root = $doc->createElement($root_name);
-        foreach ($this->convertToUtf8($this->iterator) as $offset => $row) {
-            $root->appendChild($this->toDOMNode(
-                $doc,
-                $row,
-                $offset,
-                $row_name,
-                $cell_name,
-                $column_attr,
-                $offset_attr
-            ));
-        }
-        $doc->appendChild($root);
-
-        return $doc;
-    }
-
-    /**
-     * convert a Record into a DOMNode
-     *
-     * @param DOMDocument $doc         The DOMDocument
-     * @param array       $row         The CSV record
-     * @param int         $offset      The CSV record offset
-     * @param string      $row_name    XML row node name
-     * @param string      $cell_name   XML cell node name
-     * @param string      $column_attr XML header attribute name
-     * @param string      $offset_attr XML offset attribute name
-     *
-     * @return DOMElement
-     */
-    protected function toDOMNode(
-        DOMDocument $doc,
-        array $row,
-        int $offset,
-        string $row_name,
-        string $cell_name,
-        string $column_attr,
-        string $offset_attr
-    ): DOMElement {
-        $rowElement = $doc->createElement($row_name);
-        if ($this->preserve_offset) {
-            $rowElement->setAttribute($offset_attr, (string) $offset);
-        }
-        foreach ($row as $name => $value) {
-            $content = $doc->createTextNode($value);
-            $cell = $doc->createElement($cell_name);
-            if (!empty($this->column_names)) {
-                $cell->setAttribute($column_attr, $name);
-            }
-            $cell->appendChild($content);
-            $rowElement->appendChild($cell);
-        }
-
-        return $rowElement;
+        $encoder = (new XMLEncoder())
+            ->rootName($root_name)
+            ->recordName($record_name)
+            ->itemName($cell_name)
+            ->columnAttributeName($column_attr)
+            ->offsetAttributeName($offset_attr)
+            ->preserveItemOffset(!empty($this->column_names))
+            ->preserveRecordOffset($this->preserve_offset)
+        ;
+
+        return $encoder->encode($this->convertToUtf8($this->iterator));
     }
 
     /**

diff --git a/src/Statement.php b/src/Statement.php
@@ -189,20 +189,21 @@ public function process(Reader $reader): RecordSet
      */
     protected function buildColumns(array $columns): array
     {
-        $combine = null;
-        if (!empty($this->columns)) {
-            $columns_alias = $this->filterColumnAgainstCsvHeader($columns);
-            $columns = array_values($columns_alias);
-            $combine = function (array $record) use ($columns_alias): array {
-                $res = [];
-                foreach ($columns_alias as $key => $alias) {
-                    $res[$alias] = $record[$key] ?? null;
-                }
-
-                return $res;
-            };
+        if (empty($this->columns)) {
+            return [$columns, null];
         }
 
+        $columns_alias = $this->filterColumnAgainstCsvHeader($columns);
+        $columns = array_values($columns_alias);
+        $combine = function (array $record) use ($columns_alias): array {
+            $res = [];
+            foreach ($columns_alias as $key => $alias) {
+                $res[$alias] = $record[$key] ?? null;
+            }
+
+            return $res;
+        };
+
         return [$columns, $combine];
     }
 
@@ -224,7 +225,7 @@ protected function filterColumnAgainstCsvHeader(array $headers)
                 return $this->columns;
             }
 
-            throw new RuntimeException('If no header is specified the columns keys must contain only integer');
+            throw new RuntimeException('If no header is specified the columns keys must contain only positive integer or 0');
         }
 
         $columns = $this->formatColumns($this->columns);