html_to_text.inc in Mail System 8.2
Same filename and directory in other branches
Copy of drupal_html_to_text improvements from issue #299138.
File
html_to_text.incView source
<?php
/**
* @file
* Copy of drupal_html_to_text improvements from issue #299138.
*/
/**
* Perform format=flowed soft wrapping for mail (RFC 3676).
*
* We use delsp=yes wrapping, but only break non-spaced languages when
* absolutely necessary to avoid compatibility issues.
*
* We deliberately use variable_get('mail_line_endings', MAIL_LINE_ENDINGS)
* rather than "\r\n".
*
* @param $text
* The plain text to process.
* @param array $options
* (optional) An array containing one or more of the following keys:
* - indent: A string to indent the text with. Only '>' characters are
* repeated on subsequent wrapped lines. Others are replaced by spaces.
* - max: The maximum length at which to wrap each line. Defaults to 80.
* - stuff: Whether to space-stuff special lines. Defaults to TRUE.
* - hard: Whether to enforce the maximum line length even if no convenient
* space character is available. Defaults to FALSE.
* - pad: A string to use for padding short lines to 'max' characters. If
* more than one character, only the last will be repeated.
* - break: The line break sequence to insert. The default is one of the
* following:
* - "\r\n": Windows, when $text does not contain a space character.
* - "\n": Non-Windows, when $text does not contain a space character.
* - " \r\n": On Windows, when $text contains at least one space.
* - " \n": Non-Windows, when $text contains at least one space.
*
* @see drupal_mail()
*/
function mailsystem_wrap_mail($text, array $options = array()) {
static $defaults;
if (!isset($defaults)) {
$defaults = array(
'indent' => '',
'pad' => '',
'pad_repeat' => '',
'max' => 80,
'stuff' => TRUE,
'hard' => FALSE,
'eol' => variable_get('mail_line_endings', MAIL_LINE_ENDINGS),
);
}
$options += $defaults;
if (!isset($options['break'])) {
// Allow soft-wrap spaces only when $text contains at least one space.
$options['break'] = (strpos($text, ' ') === FALSE ? '' : ' ') . $defaults['eol'];
}
$options['wrap'] = $options['max'] - drupal_strlen($options['indent']);
if ($options['pad']) {
$options['pad_repeat'] = drupal_substr($options['pad'], -1, 1);
}
// The 'clean' indent is applied to all lines after the first one.
$options['clean'] = _mailsystem_html_to_text_clean($options['indent']);
// Wrap lines according to RFC 3676.
$lines = explode($defaults['eol'], $text);
array_walk($lines, '_mailsystem_wrap_mail_line', $options);
// Expand the lines array on newly-inserted line breaks.
$lines = explode($defaults['eol'], implode($defaults['eol'], $lines));
// Apply indentation, space-stuffing, and padding.
array_walk($lines, '_mailsystem_indent_mail_line', $options);
return implode($defaults['eol'], $lines);
}
/**
* Transform an HTML string into plain text, preserving the structure of the
* markup. Useful for preparing the body of a node to be sent by e-mail.
*
* The output will be suitable for use as 'format=flowed; delsp=yes' text
* (RFC 3676) and can be passed directly to drupal_mail() for sending.
*
* We deliberately use variable_get('mail_line_endings', MAIL_LINE_ENDINGS)
* rather than "\r\n".
*
* This function provides suitable alternatives for the following tags:
*
* <a> <address> <b> <blockquote> <br /> <caption> <cite> <dd> <div> <dl> <dt>
* <em> <h1> <h2> <h3> <h4> <h5> <h6> <hr /> <i> <li> <ol> <p> <pre> <strong>
* <table> <tbody> <td> <tfoot> <thead> <tr> <u> <ul>
*
* The following tag attributes are supported:
* - <a href=...>: Hyperlink destination urls.
* - <li value=...>: Ordered list item numbers.
* - <ol start=...>: Ordered list start number.
*
* @param $string
* The string to be transformed.
* @param $allowed_tags
* (optional) If supplied, a list of tags that will be transformed. If
* omitted, all supported tags are transformed.
*
* @return
* The transformed string.
*
* @see drupal_mail()
*/
function mailsystem_html_to_text($string, $allowed_tags = NULL) {
$eol = variable_get('mail_line_endings', MAIL_LINE_ENDINGS);
// Cache list of supported tags.
static $supported_tags;
if (!isset($supported_tags)) {
$supported_tags = array(
'a',
'address',
'b',
'blockquote',
'br',
'cite',
'dd',
'div',
'dl',
'dt',
'em',
'h1',
'h2',
'h3',
'h4',
'h5',
'h6',
'hr',
'i',
'li',
'ol',
'p',
'pre',
'strong',
'table',
'td',
'tr',
'u',
'ul',
);
}
// Make sure only supported tags are kept.
$allowed_tags = isset($allowed_tags) ? array_intersect($supported_tags, $allowed_tags) : $supported_tags;
// Parse $string into a DOM tree.
$dom = filter_dom_load($string);
$notes = array();
// Recursively convert the DOM tree into plain text.
$text = _mailsystem_html_to_text($dom->documentElement, $allowed_tags, $notes);
// Hard-wrap at 1000 characters (including the line break sequence)
// and space-stuff special lines.
$text = mailsystem_wrap_mail($text, array(
'max' => 1000 - strlen($eol),
'hard' => TRUE,
));
// Change non-breaking spaces back to regular spaces, and trim line breaks.
// chr(160) is the non-breaking space character.
$text = str_replace(chr(160), ' ', trim($text, $eol));
// Add footnotes;
if ($notes) {
// Add a blank line before the footnote list.
$text .= $eol;
foreach ($notes as $url => $note) {
$text .= $eol . '[' . $note . '] ' . $url;
}
}
return $text;
}
/**
* Helper function for drupal_html_to_text().
*
* Recursively converts $node to text, wrapping and indenting as necessary.
*
* @param $node
* The source DOMNode.
* @param $allowed_tags
* A list of tags that will be transformed.
* @param array &$notes
* A writeable array of footnote reference numbers, keyed by their
* respective hyperlink destination urls.
* @param $line_length
* The maximum length of a line, for wrapping. Defaults to 80 characters.
* @param array $parents
* The list of ancestor tags, from nearest to most distant. Defaults to an
* empty array().
* @param $count
* The number to use for the next list item within an ordered list. Defaults
* to 1.
*/
function _mailsystem_html_to_text(DOMNode $node, array $allowed_tags, array &$notes, $line_length = 80, array $parents = array(), &$count = NULL) {
if (!isset($count)) {
$count = 1;
}
$eol = variable_get('mail_line_endings', MAIL_LINE_ENDINGS);
if ($node->nodeType === XML_TEXT_NODE) {
// For text nodes, we just copy the text content.
$text = $node->textContent;
// Convert line breaks and trim trailing spaces.
$text = preg_replace('/ *\\r?\\n/', $eol, $text);
if (in_array('pre', $parents)) {
// Within <pre> tags, all spaces become non-breaking.
// chr(160) is the non-breaking space character.
$text = str_replace(' ', chr(160), $text);
}
else {
// Outside <pre> tags, collapse whitespace.
$text = preg_replace('/[[:space:]]+/', ' ', $text);
}
return $text;
}
// Non-text node.
$tag = '';
$text = '';
$child_text = '';
$child_count = 1;
$indent = '';
$prefix = '';
$suffix = '';
$pad = '';
if (isset($node->tagName) && in_array($node->tagName, $allowed_tags)) {
$tag = $node->tagName;
switch ($tag) {
// Turn links with valid hrefs into footnotes.
case 'a':
$test = !empty($node->attributes);
$test = $test && ($href = $node->attributes
->getNamedItem('href'));
$test = $test && ($url = url(preg_replace('|^' . base_path() . '|', '', $href->nodeValue), array(
'absolute' => TRUE,
)));
$test = $test && valid_url($url);
if ($test) {
// Only add links that have not already been added.
if (isset($notes[$url])) {
$note = $notes[$url];
}
else {
$note = count($notes) + 1;
$notes[$url] = $note;
}
$suffix = ' [' . $note . ']';
}
break;
// Generic block-level tags.
case 'address':
case 'caption':
case 'div':
case 'p':
case 'pre':
// Start on a new line except as the first child of a list item.
if (!isset($parents[0]) || $parents[0] !== 'li' || !$node
->isSameNode($node->parentNode->firstChild)) {
$text = $eol;
}
$suffix = $eol;
break;
// Forced line break.
case 'br':
$text = $eol;
break;
// Boldface by wrapping with "*" characters.
case 'b':
case 'strong':
$prefix = '*';
$suffix = '*';
break;
// Italicize by wrapping with "/" characters.
case 'cite':
case 'em':
case 'i':
$prefix = '/';
$suffix = '/';
break;
// Underline by wrapping with "_" characters.
case 'u':
$prefix = '_';
$suffix = '_';
break;
// Blockquotes are indented by "> " at each level.
case 'blockquote':
$text = $eol;
// chr(160) is the non-breaking space character.
$indent = '>' . chr(160);
$suffix = $eol;
break;
// Dictionary definitions are indented by four spaces.
case 'dd':
// chr(160) is the non-breaking space character.
$indent = chr(160) . chr(160) . chr(160) . chr(160);
$suffix = $eol;
break;
// Dictionary list.
case 'dl':
// Start on a new line as the first child of a list item.
if (!isset($parents[0]) || $parents[0] !== 'li' || !$node
->isSameNode($node->parentNode->firstChild)) {
$text = $eol;
}
$suffix = $eol;
break;
// Dictionary term.
case 'dt':
$suffix = $eol;
break;
// Header level 1 is prefixed by eight "=" characters.
case 'h1':
$text = "{$eol}{$eol}";
// chr(160) is the non-breaking space character.
$indent = '========' . chr(160);
$pad = chr(160) . '=';
$suffix = $eol;
break;
// Header level 2 is prefixed by six "-" characters.
case 'h2':
$text = "{$eol}{$eol}";
// chr(160) is the non-breaking space character.
$indent = '------' . chr(160);
$pad = chr(160) . '-';
$suffix = $eol;
break;
// Header level 3 is prefixed by four "." characters and a space.
case 'h3':
$text = "{$eol}{$eol}";
// chr(160) is the non-breaking space character.
$indent = '....' . chr(160);
$suffix = $eol;
break;
// Header level 4 is prefixed by three "." characters and a space.
case 'h4':
$text = "{$eol}{$eol}";
// chr(160) is the non-breaking space character.
$indent = '...' . chr(160);
$suffix = $eol;
break;
// Header level 5 is prefixed by two "." character and a space.
case 'h5':
$text = "{$eol}{$eol}";
// chr(160) is the non-breaking space character.
$indent = '..' . chr(160);
$suffix = $eol;
break;
// Header level 6 is prefixed by one "." character and a space.
case 'h6':
$text = "{$eol}{$eol}";
// chr(160) is the non-breaking space character.
$indent = '.' . chr(160);
$suffix = $eol;
break;
// Horizontal rulers become a line of "-" characters.
case 'hr':
$text = $eol;
$child_text = '-';
$pad = '-';
$suffix = $eol;
break;
// List items are treated differently depending on the parent tag.
case 'li':
// Ordered list item.
if (reset($parents) === 'ol') {
// Check the value attribute.
$test = !empty($node->attributes);
$test = $test && ($value = $node->attributes
->getNamedItem('value'));
if ($test) {
$count = $value->nodeValue;
}
// chr(160) is the non-breaking space character.
$indent = ($count < 10 ? chr(160) : '') . chr(160) . "{$count})" . chr(160);
$count++;
}
else {
// chr(160) is the non-breaking space character.
$indent = chr(160) . '*' . chr(160);
}
$suffix = $eol;
break;
// Ordered lists.
case 'ol':
// Start on a new line as the first child of a list item.
if (!isset($parents[0]) || $parents[0] !== 'li' || !$node
->isSameNode($node->parentNode->firstChild)) {
$text = $eol;
}
// Check the start attribute.
$test = !empty($node->attributes);
$test = $test && ($value = $node->attributes
->getNamedItem('start'));
if ($test) {
$child_count = $value->nodeValue;
}
break;
// Tables require special handling.
case 'table':
return _mailsystem_html_to_text_table($node, $allowed_tags, $notes, $line_length);
// Separate adjacent table cells by two non-breaking spaces.
case 'td':
if (!empty($node->nextSibling)) {
// chr(160) is the non-breaking space character.
$suffix = chr(160) . chr(160);
}
break;
// End each table row with a newline.
case 'tr':
$suffix = $eol;
break;
// Unordered lists.
case 'ul':
// Start on a new line as the first child of a list item.
if (!isset($parents[0]) || $parents[0] !== 'li' || !$node
->isSameNode($node->parentNode->firstChild)) {
$text = $eol;
}
break;
default:
// Coder review complains if there is no default case.
break;
}
// Only add allowed tags to the $parents array.
array_unshift($parents, $tag);
}
// Copy each child node to output.
if ($node
->hasChildNodes()) {
foreach ($node->childNodes as $child) {
$child_text .= _mailsystem_html_to_text($child, $allowed_tags, $notes, $line_length - drupal_strlen($indent), $parents, $child_count);
}
}
// We only add prefix and suffix if the child nodes were non-empty.
if ($child_text > '') {
// We capitalize the contents of h1 and h2 tags.
if ($tag === 'h1' || $tag === 'h2') {
$child_text = drupal_strtoupper($child_text);
}
// Don't add a newline to an existing newline.
if ($suffix === $eol && drupal_substr($child_text, -drupal_strlen($eol)) === $eol) {
$suffix = '';
}
// Trim spaces around newlines except with <pre> or inline tags.
if (!in_array($tag, array(
'a',
'b',
'cite',
'em',
'i',
'pre',
'strong',
'u',
))) {
$child_text = preg_replace('/ *' . $eol . ' */', $eol, $child_text);
}
// Soft-wrap at effective line length, but don't space-stuff.
$child_text = mailsystem_wrap_mail($prefix . $child_text, array(
// chr(160) is the non-breaking space character.
'break' => chr(160) . $eol,
'indent' => $indent,
'max' => $line_length,
'pad' => $pad,
'stuff' => FALSE,
)) . $suffix;
if ($tag === 'pre') {
// Perform RFC-3676 soft-wrapping.
// chr(160) is the non-breaking space character.
$child_text = str_replace(chr(160), ' ', $child_text);
$child_text = mailsystem_wrap_mail($child_text, array(
'max' => $line_length,
'stuff' => FALSE,
));
// chr(160) is the non-breaking space character.
$child_text = str_replace(' ', chr(160), $child_text);
}
$text .= $child_text;
}
return $text;
}
/**
* Helper function for _mailsystem_html_to_text().
*
* Renders a <table> DOM Node into plain text. Attributes such as rowspan,
* colspan, padding, border, etc. are ignored.
*
* @param DOMNode $node
* The DOMNode corresponding to the <table> tag and its contents.
* @param $allowed_tags
* The list of allowed tags passed to _mailsystem_html_to_text().
* @param array &$notes
* A writeable array of footnote reference numbers, keyed by their
* respective hyperlink destination urls.
* @param $table_width
* The desired maximum table width, after word-wrapping each table cell.
*
* @return
* A plain text representation of the table.
*
* @see _mailsystem_html_to_text()
*/
function _mailsystem_html_to_text_table(DOMNode $node, $allowed_tags = NULL, array &$notes = array(), $table_width = 80) {
$eol = variable_get('mail_line_endings', MAIL_LINE_ENDINGS);
$header = array();
$footer = array();
$body = array();
$text = $eol;
$current = $node;
while (TRUE) {
if (isset($current->tagName)) {
switch ($current->tagName) {
case 'caption':
// The table caption is added first.
$text = _mailsystem_html_to_text($current, $allowed_tags, $notes, $table_width);
break;
case 'tr':
switch ($current->parentNode->tagName) {
case 'thead':
$header[] = $current;
break;
case 'tfoot':
$footer[] = $current;
break;
default:
// Either 'tbody' or 'table'
$body[] = $current;
break;
}
break;
default:
if ($current
->hasChildNodes()) {
$current = $current->firstChild;
continue 2;
}
}
}
do {
if ($current->nextSibling) {
$current = $current->nextSibling;
continue 2;
}
$current = $current->parentNode;
} while ($current && !$current
->isSameNode($node));
break;
}
// Merge the thead, tbody, and tfoot sections together.
if ($rows = array_merge($header, $body, $footer)) {
$num_rows = count($rows);
// First just count the number of columns.
$num_cols = 0;
foreach ($rows as $row) {
$row_cols = 0;
foreach ($row->childNodes as $cell) {
if (isset($cell->tagName) && in_array($cell->tagName, array(
'td',
'th',
))) {
$row_cols++;
}
}
$num_cols = max($num_cols, $row_cols);
}
// If any columns were found, calculate each column height and width.
if ($num_cols) {
// Set up a binary search for best wrap width for each column.
$max = max($table_width - $num_cols - 1, 1);
$max_wraps = array_fill(0, $num_cols, $max);
$try = max(intval(($table_width - 1) / $num_cols - 1), 1);
$try_wraps = array_fill(0, $num_cols, $try);
$min_wraps = array_fill(0, $num_cols, 1);
// Start searching...
$change = FALSE;
do {
$change = FALSE;
$widths = array_fill(0, $num_cols, 0);
$heights = array_fill(0, $num_rows, 0);
$table = array_fill(0, $num_rows, array_fill(0, $num_cols, ''));
$breaks = array_fill(0, $num_cols, FALSE);
foreach ($rows as $i => $row) {
$j = 0;
foreach ($row->childNodes as $cell) {
if (!isset($cell->tagName) || !in_array($cell->tagName, array(
'td',
'th',
))) {
// Skip text nodes.
continue;
}
// Render the cell contents.
$cell = _mailsystem_html_to_text($cell, $allowed_tags, $notes, $try_wraps[$j]);
// Trim leading line-breaks and trailing whitespace.
// chr(160) is the non-breaking space character.
$cell = rtrim(ltrim($cell, $eol), ' ' . $eol . chr(160));
$table[$i][$j] = $cell;
if ($cell > '') {
// Split the cell into lines.
$lines = explode($eol, $cell);
// The row height is the maximum number of lines among all the
// cells in that row.
$heights[$i] = max($heights[$i], count($lines));
foreach ($lines as $line) {
$this_width = drupal_strlen($line);
// The column width is the maximum line width among all the
// lines in that column.
if ($this_width > $widths[$j]) {
$widths[$j] = $this_width;
// If the longest line in a column contains at least one
// space character, then the table can be made narrower.
$breaks[$j] = strpos(' ', $line) !== FALSE;
}
}
}
$j++;
}
}
// Calculate the total table width;
$this_width = array_sum($widths) + $num_cols + 1;
if ($this_width > $table_width) {
// Wider than desired.
if (!in_array(TRUE, $breaks)) {
// If there are no more break points, then the table is already as
// narrow as it can get, so we're done.
break;
}
foreach ($try_wraps as $i => $wrap) {
$max_wraps[$i] = min($max_wraps[$i], $wrap);
if ($breaks[$i]) {
$new_wrap = intval(($min_wraps[$i] + $max_wraps[$i]) / 2);
$new_wrap = min($new_wrap, $widths[$i] - 1);
$new_wrap = max($new_wrap, $min_wraps[$i]);
}
else {
// There's no point in trying to make the column narrower than
// the widest un-wrappable line in the column.
$min_wraps[$i] = $widths[$i];
$new_wrap = $widths[$i];
}
if ($try_wraps[$i] > $new_wrap) {
$try_wraps[$i] = $new_wrap;
$change = TRUE;
}
}
}
elseif ($this_width < $table_width) {
// Narrower than desired.
foreach ($try_wraps as $i => $wrap) {
if ($min_wraps[$i] < $wrap) {
$min_wraps[$i] = $wrap;
}
$new_wrap = intval(($min_wraps[$i] + $max_wraps[$i]) / 2);
$new_wrap = max($new_wrap, $widths[$i] + 1);
$new_wrap = min($new_wrap, $max_wraps[$i]);
if ($try_wraps[$i] < $new_wrap) {
$try_wraps[$i] = $new_wrap;
$change = TRUE;
}
}
}
} while ($change);
// Pad each cell to column width and line height.
for ($i = 0; $i < $num_rows; $i++) {
if ($heights[$i]) {
for ($j = 0; $j < $num_cols; $j++) {
$cell = $table[$i][$j];
// Pad each cell to the maximum number of lines in that row.
$lines = array_pad(explode($eol, $cell), $heights[$i], '');
foreach ($lines as $k => $line) {
// Pad each line to the maximum width in that column.
$repeat = $widths[$j] - drupal_strlen($line);
if ($repeat > 0) {
// chr(160) is the non-breaking space character.
$lines[$k] .= str_repeat(chr(160), $repeat);
}
}
$table[$i][$j] = $lines;
}
}
}
// Generate the row separator line.
$separator = '+';
for ($i = 0; $i < $num_cols; $i++) {
$separator .= str_repeat('-', $widths[$i]) . '+';
}
$separator .= $eol;
for ($i = 0; $i < $num_rows; $i++) {
$text .= $separator;
if (!$heights[$i]) {
continue;
}
$row = $table[$i];
// For each row, iterate first by lines within the row.
for ($k = 0; $k < $heights[$i]; $k++) {
// Add a vertical-bar at the beginning of each row line.
$row_line = '|';
$trimmed = '';
// Within each row line, iterate by cells within that line.
for ($j = 0; $j < $num_cols; $j++) {
// Add a vertical bar at the end of each cell line.
$row_line .= $row[$j][$k] . '|';
// chr(160) is the non-breaking space character.
$trimmed .= trim($row[$j][$k], ' ' . $eol . chr(160));
}
if ($trimmed > '') {
// Only print rows that are non-empty.
$text .= $row_line . $eol;
}
}
}
// Final output ends with a row separator.
$text .= $separator;
}
}
// Make sure formatted table content doesn't line-wrap.
// chr(160) is the non-breaking space character.
return str_replace(' ', chr(160), $text);
}
/**
* Helper function for array_walk in drupal_wrap_mail().
*
* Inserts $values['break'] sequences to break up $line into parts of no more
* than $values['wrap'] characters. Only breaks at space characters, unless
* $values['hard'] is TRUE.
*/
function _mailsystem_wrap_mail_line(&$line, $key, $values) {
$line = wordwrap($line, $values['wrap'], $values['break'], $values['hard']);
}
/**
* Helper function for array_walk in drupal_wrap_mail().
*
* If $values['pad'] is non-empty, $values['indent'] will be added at the start
* of each line, and $values['pad'] at the end, repeating the last character of
* $values['pad'] until the line length equals $values['max'].
*
* If $values['pad'] is empty, $values['indent'] will be added at the start of
* the first line, and $values['clean'] at the start of subsequent lines.
*
* If $values['stuff'] is true, then an extra space character will be added at
* the start of any line beginning with a space, a '>', or the word 'From'.
*
* @see http://www.ietf.org/rfc/rfc3676.txt
*/
function _mailsystem_indent_mail_line(&$line, $key, $values) {
if ($line == '') {
return;
}
if ($values['pad']) {
$line = $values['indent'] . $line;
$count = $values['max'] - drupal_strlen($line) - drupal_strlen($values['pad']);
if ($count >= 0) {
$line .= $values['pad'] . str_repeat($values['pad_repeat'], $count);
}
}
else {
$line = $values[$key === 0 ? 'indent' : 'clean'] . $line;
}
if ($values['stuff']) {
// chr(160) is the non-breaking space character.
$line = preg_replace('/^(' . chr(160) . '| |>|From)/', ' $1', $line);
}
}
/**
* Helper function for drupal_wrap_mail() and drupal_html_to_text().
*
* Replace all non-quotation markers from a given piece of indentation with
* non-breaking space characters.
*/
function _mailsystem_html_to_text_clean($indent) {
// chr(160) is the non-breaking space character.
return preg_replace('/[^>]/', chr(160), $indent);
}Functions
Name |
Description |
|---|---|
| mailsystem_html_to_text | Transform an HTML string into plain text, preserving the structure of the markup. Useful for preparing the body of a node to be sent by e-mail. |
| mailsystem_wrap_mail | Perform format=flowed soft wrapping for mail (RFC 3676). |
| _mailsystem_html_to_text | Helper function for drupal_html_to_text(). |
| _mailsystem_html_to_text_clean | Helper function for drupal_wrap_mail() and drupal_html_to_text(). |
| _mailsystem_html_to_text_table | Helper function for _mailsystem_html_to_text(). |
| _mailsystem_indent_mail_line | Helper function for array_walk in drupal_wrap_mail(). |
| _mailsystem_wrap_mail_line | Helper function for array_walk in drupal_wrap_mail(). |