- MFH Implement todo: SplFileObject: ability to set the CSV separator per object

6 files changed, 160 insertions, 20 deletions

diff --git a/NEWS b/NEWS
index a4ae2ba9f8..ce0ed32d08 100644
--- a/NEWS
+++ b/NEWS
@@ -50,6 +50,7 @@ PHP                                                                        NEWS
   . Added array functions to ArrayObject/ArrayIterator and made them faster.
   . Added support for reading csv and skipping empty lines in SplFileObject.
   . Added CachingIterator::TOSTRING_USE_INNER, calls inner iterator __toString.
+  . Added ability to set the CSV separator per SplFileObject.
 - Improved xmlReader: (Rob)
   . Added readInnerXml(), xmlReader::setSchema().
   . Added readInnerXML(), readOuterXML(), readString(), setSchema(). (2.6.20+)
diff --git a/ext/spl/internal/splfileobject.inc b/ext/spl/internal/splfileobject.inc
index 9d347b4b20..08b2d091dd 100755
--- a/ext/spl/internal/splfileobject.inc
+++ b/ext/spl/internal/splfileobject.inc
@@ -4,7 +4,7 @@
  * @ingroup SPL
  * @brief class FileObject
  * @author  Marcus Boerger
- * @date    2003 - 2005
+ * @date    2003 - 2006
  *
  * SPL - Standard PHP Library
  */
@@ -12,10 +12,10 @@
 /** @ingroup SPL
  * @brief   Object representation for any stream
  * @author  Marcus Boerger
- * @version 1.0
+ * @version 1.1
  * @since PHP 5.1
  */
-class SplFileObject implements RecursiveIterator, SeekableIterator
+class SplFileObject extends SplFileInfo implements RecursiveIterator, SeekableIterator
 {
 	/** Flag: wheter to suppress new lines */
 	const DROP_NEW_LINE   = 0x00000001;
@@ -26,6 +26,8 @@ class SplFileObject implements RecursiveIterator, SeekableIterator
 	private $lnum     = 0;
 	private $max_len  = 0;
 	private $flags    = 0;
+	private $delimiter= ',';
+	private $enclosure= '"';
 	
 	/**
 	 * Constructs a new file object
@@ -80,14 +82,44 @@ class SplFileObject implements RecursiveIterator, SeekableIterator
 	 * @param enclosure  end of 
 	 * @return array containing read data
 	 */
-	function fgetcsv($delimiter = ';', $enclosure = '')
+	function fgetcsv($delimiter = NULL, $enclosure = NULL)
 	{
 		$this->freeLine();
 		$this->lnum++;
+		switch(fun_num_args())
+		{
+			case 0:
+				$delimiter = $this->delimiter;
+			case 1:
+				$enclosure = $this->enclosure;
+			default:
+			case 2:
+				break;
+		}
 		return fgetcsv($this->fp, $this->max_len, $delimiter, $enclosure); 
 	}
 
 	/**
+	 * Set the delimiter and enclosure character used in fgetcsv
+	 *
+	 * @param delimiter new delimiter, defaults to ','
+	 * @param enclosure new enclosure, defaults to '"'
+	 */
+	function setCsvControl($delimiter = ';', $enclosure = '"')
+	{
+		$this->delimiter = $delimiter;
+		$this->enclosure = $enclosure;
+	}
+
+	/**
+	 * @return array(delimiter, enclosure) as used in fgetcsv
+	 */
+	function getCsvControl($delimiter = ',', $enclosure = '"')
+	{
+		return array($this->delimiter, $this->enclosure);
+	}
+
+	/**
 	 * @param operation lock operation (LOCK_SH, LOCK_EX, LOCK_UN, LOCK_NB)
 	 * @retval $wouldblock  whether the operation would block
 	 */
diff --git a/ext/spl/spl_directory.c b/ext/spl/spl_directory.c
index 23a65a52ab..f84a515dc9 100755
--- a/ext/spl/spl_directory.c
+++ b/ext/spl/spl_directory.c
@@ -232,6 +232,9 @@ static int spl_filesystem_file_open(spl_filesystem_object *intern, int use_inclu
 	/* avoid reference counting in debug mode, thus do it manually */
 	ZVAL_RESOURCE(&intern->u.file.zresource, php_stream_get_resource_id(intern->u.file.stream));
 	intern->u.file.zresource.refcount = 1;
+	
+	intern->u.file.delimiter = ',';
+	intern->u.file.enclosure = '"';
 
 	zend_hash_find(&intern->std.ce->function_table, "getcurrentline", sizeof("getcurrentline"), (void **) &intern->u.file.func_getCurr);
 
@@ -1441,17 +1444,25 @@ static int spl_filesystem_file_call(spl_filesystem_object *intern, zend_function
 	spl_filesystem_file_call(intern, func_ptr, pass_num_args, return_value, arg2 TSRMLS_CC); \
 }
 
-static void spl_filesystem_file_read_csv(zval * this_ptr, spl_filesystem_object *intern, int pass_num_args, zval *return_value TSRMLS_DC) /* {{{ */
+static int spl_filesystem_file_read_csv(spl_filesystem_object *intern, char delimiter, char enclosure, zval *return_value TSRMLS_DC) /* {{{ */
 {
-	zval *arg2 = NULL;
-	MAKE_STD_ZVAL(arg2);
-	ZVAL_LONG(arg2, intern->u.file.max_line_len);
-
-	spl_filesystem_file_free_line(intern TSRMLS_CC);
-
-	FileFunctionCall(fgetcsv, pass_num_args, arg2);
+	int ret = SUCCESS;
+	
+	do {
+		ret = spl_filesystem_file_read(intern, 1 TSRMLS_CC);
+	} while (ret == SUCCESS && !intern->u.file.current_line_len && (intern->flags & SPL_FILE_OBJECT_SKIP_EMPTY));
+	
+	if (ret == SUCCESS) {
+		size_t buf_len = intern->u.file.current_line_len;
+		char *buf = estrndup(intern->u.file.current_line, buf_len);
 
-	zval_ptr_dtor(&arg2);
+		if (Z_TYPE_P(return_value) != IS_NULL) {
+			zval_dtor(return_value);
+			ZVAL_NULL(return_value);
+		}
+		php_fgetcsv(intern->u.file.stream, delimiter, enclosure, buf_len, buf, return_value TSRMLS_CC);
+	}
+	return ret;
 }
 /* }}} */
 
@@ -1469,7 +1480,7 @@ static int spl_filesystem_file_read_line_ex(zval * this_ptr, spl_filesystem_obje
 		}
 		if (intern->flags & SPL_FILE_OBJECT_READ_CSV) {
 			MAKE_STD_ZVAL(retval);
-			spl_filesystem_file_read_csv(this_ptr, intern, 0, retval TSRMLS_CC);
+			return spl_filesystem_file_read_csv(intern, intern->u.file.delimiter, intern->u.file.enclosure, retval TSRMLS_CC);
 		} else {
 			zend_call_method_with_0_params(&this_ptr, Z_OBJCE_P(getThis()), &intern->u.file.func_getCurr, "getCurrentLine", &retval);
 		}
@@ -1794,9 +1805,86 @@ SPL_METHOD(SplFileObject, func_name) \
 SPL_METHOD(SplFileObject, fgetcsv)
 {
 	spl_filesystem_object *intern = (spl_filesystem_object*)zend_object_store_get_object(getThis() TSRMLS_CC);
+	char delimiter = intern->u.file.delimiter, enclosure = intern->u.file.enclosure;
+	char *delim, *enclo;
+	int d_len, e_len;
 	
-	spl_filesystem_file_read_csv(getThis(), intern, ZEND_NUM_ARGS(), return_value TSRMLS_CC);
-	intern->u.file.current_line_num++;
+	if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "|ss", &delim, &d_len, &enclo, &e_len) == SUCCESS) {
+		switch(ZEND_NUM_ARGS())
+		{
+			case 2:
+				if (e_len != 1) {
+					php_error_docref(NULL TSRMLS_CC, E_WARNING, "enclosure must be a character");
+					RETURN_FALSE;
+				}
+				enclosure = enclo[0];
+				/* no break */
+			case 1:
+				if (d_len != 1) {
+					php_error_docref(NULL TSRMLS_CC, E_WARNING, "delimiter must be a character");
+					RETURN_FALSE;
+				}
+				delimiter = delim[0];
+				/* no break */
+			case 0:
+				break;
+		}
+		spl_filesystem_file_read_csv(intern, delimiter, enclosure, return_value TSRMLS_CC);
+	}
+}
+/* }}} */
+
+/* {{{ proto void SplFileObject::setCsvControl([string delimiter = ',' [, string enclosure = '"']])
+   Set the delimiter and enclosure character used in fgetcsv */
+SPL_METHOD(SplFileObject, setCsvControl)
+{
+	spl_filesystem_object *intern = (spl_filesystem_object*)zend_object_store_get_object(getThis() TSRMLS_CC);
+	char delimiter = ',', enclosure = '"';
+	char *delim, *enclo;
+	int d_len, e_len;
+	
+	if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "|ss", &delim, &d_len, &enclo, &e_len) == SUCCESS) {
+		switch(ZEND_NUM_ARGS())
+		{
+			case 2:
+				if (e_len != 1) {
+					php_error_docref(NULL TSRMLS_CC, E_WARNING, "enclosure must be a character");
+					RETURN_FALSE;
+				}
+				enclosure = enclo[0];
+				/* no break */
+			case 1:
+				if (d_len != 1) {
+					php_error_docref(NULL TSRMLS_CC, E_WARNING, "delimiter must be a character");
+					RETURN_FALSE;
+				}
+				delimiter = delim[0];
+				/* no break */
+			case 0:
+				break;
+		}
+		intern->u.file.delimiter = delimiter;
+		intern->u.file.enclosure = enclosure;
+	}
+}
+/* }}} */
+
+/* {{{ proto array SplFileObject::getCsvControl()
+   Get the delimiter and enclosure character used in fgetcsv */
+SPL_METHOD(SplFileObject, getCsvControl)
+{
+	spl_filesystem_object *intern = (spl_filesystem_object*)zend_object_store_get_object(getThis() TSRMLS_CC);
+	char delimiter[2], enclosure[2];
+
+	array_init(return_value);
+	
+	delimiter[0] = intern->u.file.delimiter;
+	delimiter[1] = '\0';
+	enclosure[0] = intern->u.file.enclosure;
+	enclosure[1] = '\0';
+
+	add_next_index_string(return_value, delimiter, 1);
+	add_next_index_string(return_value, enclosure, 1);
 }
 /* }}} */
 
@@ -2057,6 +2145,8 @@ static zend_function_entry spl_SplFileObject_functions[] = {
 	SPL_ME(SplFileObject, valid,          NULL, ZEND_ACC_PUBLIC)
 	SPL_ME(SplFileObject, fgets,          NULL, ZEND_ACC_PUBLIC)
 	SPL_ME(SplFileObject, fgetcsv,        arginfo_file_object_fgetcsv,       ZEND_ACC_PUBLIC)
+	SPL_ME(SplFileObject, setCsvControl,  arginfo_file_object_fgetcsv,       ZEND_ACC_PUBLIC)
+	SPL_ME(SplFileObject, getCsvControl,  NULL,                              ZEND_ACC_PUBLIC)
 	SPL_ME(SplFileObject, flock,          arginfo_file_object_flock,         ZEND_ACC_PUBLIC)
 	SPL_ME(SplFileObject, fflush,         NULL, ZEND_ACC_PUBLIC)
 	SPL_ME(SplFileObject, ftell,          NULL, ZEND_ACC_PUBLIC)
diff --git a/ext/spl/spl_directory.h b/ext/spl/spl_directory.h
index 2ab7513a1c..e2f9de4e15 100755
--- a/ext/spl/spl_directory.h
+++ b/ext/spl/spl_directory.h
@@ -82,6 +82,8 @@ struct _spl_filesystem_object {
 			long               current_line_num;
 			zval               zresource;
 			zend_function      *func_getCurr;
+			char               delimiter;
+			char               enclosure;
 		} file;
 	} u;
 };
diff --git a/ext/standard/file.c b/ext/standard/file.c
index 744fa21d1c..0bbb23f48c 100644
--- a/ext/standard/file.c
+++ b/ext/standard/file.c
@@ -1976,17 +1976,14 @@ PHP_FUNCTION(fputcsv)
    Get line from file pointer and parse for CSV fields */
 PHP_FUNCTION(fgetcsv)
 {
-	char *temp, *tptr, *bptr, *line_end, *limit;
 	char delimiter = ',';	/* allow this to be set as parameter */
 	char enclosure = '"';	/* allow this to be set as parameter */
-	const char escape_char = '\\';
 	/* first section exactly as php_fgetss */
 
 	long len = 0;
-	size_t buf_len, temp_len, line_end_len;
+	size_t buf_len;
 	char *buf;
 	php_stream *stream;
-	int inc_len;
 
 	{
 		zval *fd, **len_zv = NULL;
@@ -2048,6 +2045,23 @@ PHP_FUNCTION(fgetcsv)
 			RETURN_FALSE;
 		}
 	}
+
+	php_fgetcsv(stream, delimiter, enclosure, buf_len, buf, return_value TSRMLS_CC);
+}
+/* }}} */
+
+
+PHPAPI void php_fgetcsv(php_stream *stream, /* {{{ */
+		char delimiter, char enclosure, 
+		size_t buf_len, char *buf,
+		zval *return_value TSRMLS_DC)
+{
+	char *temp, *tptr, *bptr, *line_end, *limit;
+	const char escape_char = '\\';
+
+	size_t temp_len, line_end_len;
+	int inc_len;
+
 	/* initialize internal state */
 	php_mblen(NULL, 0);
 
diff --git a/ext/standard/file.h b/ext/standard/file.h
index 099a6ddbcc..9d1836141d 100644
--- a/ext/standard/file.h
+++ b/ext/standard/file.h
@@ -74,6 +74,7 @@ PHPAPI int php_set_sock_blocking(int socketd, int block TSRMLS_DC);
 PHPAPI int php_copy_file(char *src, char *dest TSRMLS_DC);
 PHPAPI int php_mkdir_ex(char *dir, long mode, int options TSRMLS_DC);
 PHPAPI int php_mkdir(char *dir, long mode TSRMLS_DC);
+PHPAPI void php_fgetcsv(php_stream *stream, char delimiter, char enclosure, size_t buf_len, char *buf, zval *return_value TSRMLS_DC);
 
 #define META_DEF_BUFSIZE 8192