+ '/<(th)( [^>]*)?>(.*?)<\/th>/i', // | and |
);
/**
- * List of preg* regular expression patterns to search for in PRE body,
- * used in conjunction with $pre_replace.
+ * List of preg* regular expression patterns to search for in PRE body,
+ * used in conjunction with $pre_replace.
*
- * @var array $pre_search
- * @access public
- * @see $pre_replace
+ * @var array $pre_search
+ * @see $pre_replace
*/
- public $pre_search = array(
+ protected $pre_search = array(
"/\n/",
"/\t/",
'/ /',
@@ -201,13 +194,12 @@ class Html2Text
);
/**
- * List of pattern replacements corresponding to patterns searched for PRE body.
+ * List of pattern replacements corresponding to patterns searched for PRE body.
*
- * @var array $pre_replace
- * @access public
- * @see $pre_search
+ * @var array $pre_replace
+ * @see $pre_search
*/
- public $pre_replace = array(
+ protected $pre_replace = array(
'
',
' ',
' ',
@@ -216,47 +208,42 @@ class Html2Text
);
/**
- * Contains a list of HTML tags to allow in the resulting text.
+ * Contains a list of HTML tags to allow in the resulting text.
*
- * @var string $allowed_tags
- * @access public
- * @see set_allowed_tags()
+ * @var string $allowed_tags
+ * @see set_allowed_tags()
*/
- public $allowed_tags = '';
+ protected $allowed_tags = '';
/**
- * Contains the base URL that relative links should resolve to.
+ * Contains the base URL that relative links should resolve to.
*
- * @var string $url
- * @access public
+ * @var string $url
*/
- public $url;
+ protected $url;
/**
- * Indicates whether content in the $html variable has been converted yet.
+ * Indicates whether content in the $html variable has been converted yet.
*
- * @var boolean $_converted
- * @access private
- * @see $html, $text
+ * @var boolean $_converted
+ * @see $html, $text
*/
- private $_converted = false;
+ protected $_converted = false;
/**
- * Contains URL addresses from links to be rendered in plain text.
+ * Contains URL addresses from links to be rendered in plain text.
*
- * @var array $_link_list
- * @access private
- * @see _build_link_list()
+ * @var array $_link_list
+ * @see _build_link_list()
*/
- private $_link_list = array();
+ protected $_link_list = array();
/**
- * Various configuration options (able to be set in the constructor)
+ * Various configuration options (able to be set in the constructor)
*
- * @var array $_options
- * @access private
+ * @var array $_options
*/
- private $_options = array(
+ protected $_options = array(
// 'none'
// 'inline' (show links inline)
@@ -264,30 +251,28 @@ class Html2Text
// 'table' (if a table of link URLs should be listed after the text.
'do_links' => 'inline',
- // Maximum width of the formatted text, in columns.
- // Set this value to 0 (or less) to ignore word wrapping
- // and not constrain text to a fixed-width column.
+ // Maximum width of the formatted text, in columns.
+ // Set this value to 0 (or less) to ignore word wrapping
+ // and not constrain text to a fixed-width column.
'width' => 70,
);
/**
- * Constructor.
+ * Constructor.
*
- * If the HTML source string (or file) is supplied, the class
- * will instantiate with that source propagated, all that has
- * to be done it to call get_text().
+ * If the HTML source string (or file) is supplied, the class
+ * will instantiate with that source propagated, all that has
+ * to be done it to call get_text().
*
- * @param string $source HTML content
- * @param boolean $from_file Indicates $source is a file to pull content from
- * @param array $options Set configuration options
- * @access public
- * @return void
+ * @param string $source HTML content
+ * @param boolean $from_file Indicates $source is a file to pull content from
+ * @param array $options Set configuration options
*/
- public function __construct( $source = '', $from_file = false, $options = array() )
+ public function __construct($source = '', $from_file = false, $options = array())
{
$this->_options = array_merge($this->_options, $options);
- if ( !empty($source) ) {
+ if (!empty($source)) {
$this->set_html($source, $from_file);
}
@@ -295,28 +280,26 @@ public function __construct( $source = '', $from_file = false, $options = array(
}
/**
- * Loads source HTML into memory, either from $source string or a file.
+ * Loads source HTML into memory, either from $source string or a file.
*
- * @param string $source HTML content
- * @param boolean $from_file Indicates $source is a file to pull content from
- * @access public
- * @return void
+ * @param string $source HTML content
+ * @param boolean $from_file Indicates $source is a file to pull content from
*/
- public function set_html( $source, $from_file = false )
+ public function set_html($source, $from_file = false)
{
- if ( $from_file && file_exists($source) ) {
+ if ($from_file && file_exists($source)) {
$this->html = file_get_contents($source);
- } else
+ } else {
$this->html = $source;
+ }
$this->_converted = false;
}
/**
- * Returns the text, converted from HTML.
+ * Returns the text, converted from HTML.
*
- * @access public
- * @return string
+ * @return string
*/
public function get_text()
{
@@ -328,10 +311,7 @@ public function get_text()
}
/**
- * Prints the text, converted from HTML.
- *
- * @access public
- * @return void
+ * Prints the text, converted from HTML.
*/
public function print_text()
{
@@ -339,11 +319,9 @@ public function print_text()
}
/**
- * Alias to print_text(), operates identically.
+ * Alias to print_text(), operates identically.
*
- * @access public
- * @return void
- * @see print_text()
+ * @see print_text()
*/
public function p()
{
@@ -351,30 +329,24 @@ public function p()
}
/**
- * Sets the allowed HTML tags to pass through to the resulting text.
- *
- * Tags should be in the form "", with no corresponding closing tag.
+ * Sets the allowed HTML tags to pass through to the resulting text.
*
- * @access public
- * @return void
+ * Tags should be in the form "
", with no corresponding closing tag.
*/
- public function set_allowed_tags( $allowed_tags = '' )
+ public function set_allowed_tags($allowed_tags = '')
{
- if ( !empty($allowed_tags) ) {
+ if (!empty($allowed_tags)) {
$this->allowed_tags = $allowed_tags;
}
}
/**
- * Sets a base URL to handle relative links.
- *
- * @access public
- * @return void
+ * Sets a base URL to handle relative links.
*/
- public function set_base_url( $url = '' )
+ public function set_base_url($url = '')
{
- if ( empty($url) ) {
- if ( !empty($_SERVER['HTTP_HOST']) ) {
+ if (empty($url)) {
+ if (!empty($_SERVER['HTTP_HOST'])) {
$this->url = 'http://' . $_SERVER['HTTP_HOST'];
} else {
$this->url = '';
@@ -382,7 +354,7 @@ public function set_base_url( $url = '' )
} else {
// Strip any trailing slashes for consistency (relative
// URLs may already start with a slash like "/file.html")
- if ( substr($url, -1) == '/' ) {
+ if (substr($url, -1) == '/') {
$url = substr($url, 0, -1);
}
$this->url = $url;
@@ -390,12 +362,9 @@ public function set_base_url( $url = '' )
}
/**
- * Workhorse function that does actual conversion (calls _converter() method).
- *
- * @access private
- * @return void
+ * Workhorse function that does actual conversion (calls _converter() method).
*/
- private function _convert()
+ protected function _convert()
{
// Variables used for building the link list
$this->_link_list = array();
@@ -419,19 +388,16 @@ private function _convert()
}
/**
- * Workhorse function that does actual conversion.
- *
- * First performs custom tag replacement specified by $search and
- * $replace arrays. Then strips any remaining HTML tags, reduces whitespace
- * and newlines to a readable format, and word wraps the text to
- * $this->_options['width'] characters.
+ * Workhorse function that does actual conversion.
*
- * @param string Reference to HTML content string
+ * First performs custom tag replacement specified by $search and
+ * $replace arrays. Then strips any remaining HTML tags, reduces whitespace
+ * and newlines to a readable format, and word wraps the text to
+ * $this->_options['width'] characters.
*
- * @access private
- * @return void
+ * @param string Reference to HTML content string
*/
- private function _converter(&$text)
+ protected function _converter(&$text)
{
// Convert
(before PRE!)
$this->_convert_blockquotes($text);
@@ -477,19 +443,18 @@ private function _converter(&$text)
}
/**
- * Helper function called by preg_replace() on link replacement.
+ * Helper function called by preg_replace() on link replacement.
*
- * Maintains an internal list of links to be displayed at the end of the
- * text, with numeric indices to the original point in the text they
- * appeared. Also makes an effort at identifying and handling absolute
- * and relative links.
+ * Maintains an internal list of links to be displayed at the end of the
+ * text, with numeric indices to the original point in the text they
+ * appeared. Also makes an effort at identifying and handling absolute
+ * and relative links.
*
- * @param string $link URL of the link
- * @param string $display Part of the text to associate number with
- * @access private
- * @return string
+ * @param string $link URL of the link
+ * @param string $display Part of the text to associate number with
+ * @return string
*/
- private function _build_link_list( $link, $display, $link_override = null)
+ protected function _build_link_list($link, $display, $link_override = null)
{
$link_method = ($link_override) ? $link_override : $this->_options['do_links'];
if ($link_method == 'none')
@@ -500,6 +465,7 @@ private function _build_link_list( $link, $display, $link_override = null)
if (preg_match('!^(javascript:|mailto:|#)!i', $link)) {
return $display;
}
+
if (preg_match('!^([a-z][a-z0-9.+-]+:)!i', $link)) {
$url = $link;
} else {
@@ -526,12 +492,11 @@ private function _build_link_list( $link, $display, $link_override = null)
}
/**
- * Helper function for PRE body conversion.
+ * Helper function for PRE body conversion.
*
- * @param string HTML content
- * @access private
+ * @param string HTML content
*/
- private function _convert_pre(&$text)
+ protected function _convert_pre(&$text)
{
// get the content of PRE element
while (preg_match('/]*>(.*)<\/pre>/ismU', $text, $matches)) {
@@ -544,6 +509,7 @@ private function _convert_pre(&$text)
// convert the content
$this->pre_content = sprintf('
%s
',
preg_replace($this->pre_search, $this->pre_replace, $this->pre_content));
+
// replace the content (use callback because content can contain $0 variable)
$text = preg_replace_callback('/]*>.*<\/pre>/ismU',
array($this, '_preg_pre_callback'), $text, 1);
@@ -554,12 +520,11 @@ private function _convert_pre(&$text)
}
/**
- * Helper function for BLOCKQUOTE body conversion.
+ * Helper function for BLOCKQUOTE body conversion.
*
- * @param string HTML content
- * @access private
+ * @param string HTML content
*/
- private function _convert_blockquotes(&$text)
+ protected function _convert_blockquotes(&$text)
{
if (preg_match_all('/<\/*blockquote[^>]*>/i', $text, $matches, PREG_OFFSET_CAPTURE)) {
$start = 0;
@@ -609,12 +574,12 @@ private function _convert_blockquotes(&$text)
}
/**
- * Callback function for preg_replace_callback use.
+ * Callback function for preg_replace_callback use.
*
- * @param array PREG matches
- * @return string
+ * @param array PREG matches
+ * @return string
*/
- private function _preg_callback($matches)
+ protected function _preg_callback($matches)
{
switch (strtolower($matches[1])) {
case 'b':
@@ -638,12 +603,12 @@ private function _preg_callback($matches)
}
/**
- * Callback function for preg_replace_callback use in PRE content handler.
+ * Callback function for preg_replace_callback use in PRE content handler.
*
- * @param array PREG matches
- * @return string
+ * @param array PREG matches
+ * @return string
*/
- private function _preg_pre_callback($matches)
+ protected function _preg_pre_callback($matches)
{
return $this->pre_content;
}