/**
* Populate PHPWord element
* This recursive function processes all the elements and child elements
* from the DOM array of objects created by SimpleHTMLDom.
*
* @param object phpword_element
* PHPWord object to add in the converted html
* @param array $html_dom_array
* Array of nodes generated by simple HTML dom
* @param array $state
* Parameters for the current run
*/
function htmltodocx_insert_html_recursive(&$phpword_element, $html_dom_array, &$state = array())
{
// Go through the html_dom_array, adding bits to go in the PHPWord element
$allowed_children = htmltodocx_html_allowed_children($state['parents'][0]);
// Go through each element:
foreach ($html_dom_array as $element) {
$old_style = $state['current_style'];
$state['current_style'] = _htmltodocx_get_style($element, $state);
switch ($element->tag) {
case 'p':
case 'div':
// Treat a div as a paragraph
// Treat a div as a paragraph
case 'h1':
case 'h2':
case 'h3':
case 'h4':
case 'h5':
case 'h6':
if ($state['structure_document'] && in_array($element->tag, array('h1', 'h2', 'h3', 'h4', 'h5', 'h6')) && is_object($state['phpword_object'])) {
// If the structure_document option has been enabled, then headings
// are used to create Word heading styles. Note, in this case, any
// nested elements within the heading are displayed as text only.
// Additionally we don't now add a text break after a heading where
// sizeAfter has not been set.
$state['phpword_object']->addTitleStyle($state['structure_headings'][$element->tag], $state['current_style']);
$phpword_element->addTitle(htmltodocx_clean_text($element->innertext), $state['structure_headings'][$element->tag]);
break;
}
if ($element->tag == 'div' && $state['table_of_contents_id'] && $element->id == $state['table_of_contents_id']) {
// Replace this div with a table of contents:
$phpword_element->addTOC($state['current_style'], $state['current_style']);
break;
}
// Everything in this element should be in the same text run
// we need to initiate a text run here and pass it on. Starting one of
// these elements will cause a new line to be added in the Word
// document. In the case of divs this might not always be what is
// wanted the setting 'treat_div_as_paragraph' determines whether or
// not to add new lines for divs.
if ($element->tag != 'div' || $state['treat_div_as_paragraph'] || !isset($state['textrun'])) {
$state['textrun'] = $phpword_element->createTextRun($state['current_style']);
}
// For better usability for the end user of the Word document, we
// separate paragraphs and headings with an empty line. You can
// override this behaviour by setting the spaceAfter parameter for
// the current element.
// If the spaceAfter parameter is not set, we set it temporarily to 0
// here and record that it wasn't set in the style. Later we will add
// an empty line. Word 2007 and later have a non-zero default for
// paragraph separation, so without setting that spacing to 0 here we
// would end up with a large gap between paragraphs (the document
// template default plus the extra line).
$space_after_set = TRUE;
if (!isset($state['current_style']['spaceAfter'])) {
$state['current_style']['spaceAfter'] = 0;
$space_after_set = FALSE;
}
if (in_array($element->tag, $allowed_children)) {
array_unshift($state['parents'], $element->tag);
htmltodocx_insert_html_recursive($phpword_element, $element->nodes, $state);
array_shift($state['parents']);
} else {
$state['textrun']->addText(htmltodocx_clean_text($element->innertext), $state['current_style']);
}
unset($state['textrun']);
if (!$space_after_set) {
// Add the text break here - where the spaceAfter parameter hadn't
// been set initially - also unset the spaceAfter parameter we just
// set:
$phpword_element->addTextBreak();
unset($state['current_style']['spaceAfter']);
}
break;
case 'table':
if (in_array('table', $allowed_children)) {
$old_table_state = $state['table_allowed'];
if (!$state['table_allowed'] || in_array('td', $state['parents']) || in_array('th', $state['parents'])) {
$state['table_allowed'] = FALSE;
// This is a PHPWord constraint
} else {
$state['table_allowed'] = TRUE;
// PHPWord allows table_styles to be passed in a couple of
// different ways either using an array of properties, or by
// defining a full table style on the PHPWord object:
if (is_object($state['phpword_object']) && method_exists($state['phpword_object'], 'addTableStyle')) {
$state['phpword_object']->addTableStyle('temp_table_style', $state['current_style']);
$table_style = 'temp_table_style';
} else {
$table_style = $state['current_style'];
}
$state['table'] = $phpword_element->addTable($table_style);
}
array_unshift($state['parents'], 'table');
htmltodocx_insert_html_recursive($phpword_element, $element->nodes, $state);
array_shift($state['parents']);
// Reset table state to what it was before a table was added:
$state['table_allowed'] = $old_table_state;
$phpword_element->addTextBreak();
} else {
$state['textrun'] = $phpword_element->createTextRun();
$state['textrun']->addText(htmltodocx_clean_text($element->innertext), $state['current_style']);
}
break;
case 'tbody':
if (in_array('tbody', $allowed_children)) {
array_unshift($state['parents'], 'tbody');
htmltodocx_insert_html_recursive($phpword_element, $element->nodes, $state);
array_shift($state['parents']);
} else {
$state['textrun'] = $phpword_element->createTextRun();
$state['textrun']->addText(htmltodocx_clean_text($element->innertext), $state['current_style']);
}
break;
case 'tr':
if (in_array('tr', $allowed_children)) {
if ($state['table_allowed']) {
$state['table']->addRow();
} else {
// Simply add a new line if a table is not possible in this
// context:
$state['textrun'] = $phpword_element->createTextRun();
}
array_unshift($state['parents'], 'tr');
htmltodocx_insert_html_recursive($phpword_element, $element->nodes, $state);
array_shift($state['parents']);
} else {
$state['textrun'] = $phpword_element->createTextRun();
$state['textrun']->addText(htmltodocx_clean_text($element->innertext), $state['current_style']);
}
break;
case 'td':
case 'th':
if (in_array($element->tag, $allowed_children) && $state['table_allowed']) {
unset($state['textrun']);
if (isset($state['current_style']['width'])) {
$cell_width = $state['current_style']['width'];
} elseif (isset($element->width)) {
$cell_width = $element->width * 15;
// Converting at 15 TWIPS per pixel.
} else {
$cell_width = 800;
}
$state['table_cell'] = $state['table']->addCell($cell_width, $state['current_style']);
array_unshift($state['parents'], $element->tag);
htmltodocx_insert_html_recursive($state['table_cell'], $element->nodes, $state);
array_shift($state['parents']);
} else {
if (!isset($state['textrun'])) {
$state['textrun'] = $phpword_element->createTextRun();
}
$state['textrun']->addText(htmltodocx_clean_text($element->innertext), $state['current_style']);
}
break;
case 'a':
// Create a new text run if we aren't in one already:
if (!isset($state['textrun'])) {
$state['textrun'] = $phpword_element->createTextRun();
}
if ($state['context'] == 'section') {
if (strpos($element->href, 'http://') === 0) {
$href = $element->href;
} elseif (strpos($element->href, '/') === 0) {
$href = $state['base_root'] . $element->href;
} else {
$href = $state['base_root'] . $state['base_path'] . $element->href;
}
// Replace any spaces in url with %20 - to prevent errors in the Word
// document:
$state['textrun']->addLink(htmltodocx_url_encode_chars($href), htmltodocx_clean_text($element->innertext), $state['current_style']);
} else {
// Links can't seem to be included in headers or footers with
// PHPWord: trying to include them causes an error which stops Word
// from opening the file - in Word 2003 with the converter at least.
// So add the link styled as a link only.
$state['textrun']->addText(htmltodocx_clean_text($element->innertext), $state['current_style']);
}
break;
case 'ul':
$state['list_total_count'] = count($element->children);
// We use this to be able to add the ordered list spaceAfter onto the
// last list element. All ol children should be li elements.
_htmltodocx_add_list_start_end_spacing_style($state);
$state['list_number'] = 0;
// Reset list number.
if (in_array('ul', $allowed_children)) {
if (!isset($state['pseudo_list'])) {
// Unset any existing text run:
unset($state['textrun']);
// PHPWord lists cannot appear in a text run. If we leave a text
// run active then subsequent text will go in that text run (if it
// isn't re-initialised), which would mean that text after this
// list would appear before it in the Word document.
}
array_unshift($state['parents'], 'ul');
htmltodocx_insert_html_recursive($phpword_element, $element->nodes, $state);
array_shift($state['parents']);
} else {
$state['textrun'] = $phpword_element->createTextRun();
$state['textrun']->addText(htmltodocx_clean_text($element->innertext), $state['current_style']);
}
break;
case 'ol':
$state['list_total_count'] = count($element->children);
// We use this to be able to add the ordered list spaceAfter onto the
// last list element. All ol children should be li elements.
_htmltodocx_add_list_start_end_spacing_style($state);
$state['list_number'] = 0;
// Reset list number.
if (in_array('ol', $allowed_children)) {
if (!isset($state['pseudo_list'])) {
// Unset any existing text run:
unset($state['textrun']);
// Lists cannot appear in a text run. If we leave a text run active
// then subsequent text will go in that text run (if it isn't
// re-initialised), which would mean that text after this list
// would appear before it in the Word document.
}
array_unshift($state['parents'], 'ol');
htmltodocx_insert_html_recursive($phpword_element, $element->nodes, $state);
array_shift($state['parents']);
} else {
$state['textrun'] = $phpword_element->createTextRun();
$state['textrun']->addText(htmltodocx_clean_text($element->innertext), $state['current_style']);
}
break;
case 'li':
// You cannot style individual pieces of text in a list element so we do it
// with text runs instead. This does not allow us to indent lists at all, so
// we can't show nesting.
// Before and after spacings:
if ($state['list_number'] === 0) {
$state['current_style'] = array_merge($state['current_style'], $state['list_style_before']);
}
$last_item = FALSE;
if ($state['list_number'] == $state['list_total_count'] - 1) {
$last_item = TRUE;
if (empty($state['list_style_after'])) {
$state['current_style']['spaceAfter'] = 0;
// Set to 0 if not defined so we can add a text break without
// ending up within too much space in Word2007+.
// *Needs further testing on Word 2007+*
}
$state['current_style'] = array_merge($state['current_style'], $state['list_style_after']);
}
// We create a new text run for each element:
$state['textrun'] = $phpword_element->createTextRun($state['current_style']);
if (in_array('li', $allowed_children)) {
$state['list_number']++;
if ($state['parents'][0] == 'ol') {
$item_indicator = $state['list_number'] . '. ';
$style = $state['current_style'];
} else {
$style = $state['current_style'];
$style['name'] = $state['pseudo_list_indicator_font_name'];
$style['size'] = $state['pseudo_list_indicator_font_size'];
$item_indicator = $state['pseudo_list_indicator_character'];
}
array_unshift($state['parents'], 'li');
$state['textrun']->addText($item_indicator, $style);
htmltodocx_insert_html_recursive($phpword_element, $element->nodes, $state);
array_shift($state['parents']);
} else {
$state['textrun']->addText(htmltodocx_clean_text($element->innertext), $state['current_style']);
}
if ($last_item && empty($state['list_style_after'])) {
$phpword_element->addTextBreak();
// Add an empty line after the list if no spacing after has been
// defined.
}
unset($state['textrun']);
break;
case 'text':
// We may get some empty text nodes - containing just a space - in
// simple HTML dom - we want to exclude those, as these can cause extra
// line returns. However we don't want to exclude spaces between styling
// elements (these will be within a text run).
if (!isset($state['textrun'])) {
$text = htmltodocx_clean_text(trim($element->innertext));
} else {
$text = htmltodocx_clean_text($element->innertext);
}
if (!empty($text)) {
if (!isset($state['textrun'])) {
$state['textrun'] = $phpword_element->createTextRun();
}
$state['textrun']->addText($text, $state['current_style']);
}
break;
// Style tags:
// Style tags:
case 'strong':
case 'b':
case 'sup':
// Not working in PHPWord
// Not working in PHPWord
case 'em':
case 'i':
case 'u':
case 'span':
case 'code':
// Create a new text run if we aren't in one already:
if (!isset($state['textrun'])) {
$state['textrun'] = $phpword_element->createTextRun();
}
if (in_array($element->tag, $allowed_children)) {
array_unshift($state['parents'], $element->tag);
htmltodocx_insert_html_recursive($phpword_element, $element->nodes, $state);
array_shift($state['parents']);
} else {
$state['textrun']->addText(htmltodocx_clean_text($element->innertext), $state['current_style']);
}
break;
// NB, Simple HTML Dom might not be picking up <br> tags.
// NB, Simple HTML Dom might not be picking up <br> tags.
case 'br':
// Simply create a new text run:
$state['textrun'] = $phpword_element->createTextRun();
break;
case 'img':
$image_style = array();
if ($element->height && $element->width) {
$state['current_style']['height'] = $element->height;
$state['current_style']['width'] = $element->width;
}
if (strpos($element->src, $state['base_root']) === 0) {
// The image source is a full url, but nevertheless it is on this
// server.
$element_src = substr($element->src, strlen($state['base_root']));
} else {
$element_src = $element->src;
}
if (strpos($element_src, 'http://') === 0) {
// The image url is from another site. Most probably the image won't
// appear in the Word document.
$src = $element_src;
} elseif (strpos($element_src, '/') === 0) {
$src = htmltodocx_doc_root() . $element_src;
} else {
$src = htmltodocx_doc_root() . $state['base_path'] . $element_src;
}
$phpword_element->addImage($src, $state['current_style']);
break;
default:
$state['textrun'] = $phpword_element->createTextRun();
$state['textrun']->addText(htmltodocx_clean_text($element->innertext), $state['current_style']);
break;
}
// Reset the style back to what it was:
$state['current_style'] = $old_style;
}
}
function htmltodocx_insert_html_recursive(&$phpword_object, $phpword_element, $html_dom_array, &$state = array())
{
switch ($html_dom_array->nodeType) {
case XML_TEXT_NODE:
$state['textrun']->addText(htmltodocx_clean_text($html_dom_array->nodeValue), $state['current_style']);
break;
case XML_ELEMENT_NODE:
$a4WidthPortraitOrientationInPoints = 595;
$a4HeightPortraitOrientationInPoints = 842;
$a4WidthLandscapeOrientationInPoints = 842;
$a4HeightLandscapeOrientationInPoints = 595;
// Go through the html_dom_array, adding bits to go in the PhpWord element
$allowed_children = htmltodocx_html_allowed_children($state['parents'][0]);
foreach ($html_dom_array->childNodes as $element) {
if (!in_array($element->nodeType, [XML_ELEMENT_NODE, XML_TEXT_NODE])) {
continue;
}
/** @var DOMElement $element */
//var_dump($element->tag);
$old_style = $state['current_style'];
if ($element->nodeType == XML_ELEMENT_NODE) {
$state['current_style'] = _htmltodocx_get_style($element, $state);
}
$tag = $element->nodeName;
switch ($tag) {
case 'p':
case 'div':
// Treat a div as a paragraph
// Treat a div as a paragraph
case 'h1':
case 'h2':
case 'h3':
case 'h4':
case 'h5':
case 'h6':
case 'center':
if ($state['structure_document'] && in_array($tag, ['h1', 'h2', 'h3', 'h4', 'h5', 'h6']) && is_object($state['phpword_object'])) {
// If the structure_document option has been enabled, then headings
// are used to create Word heading styles. Note, in this case, any
// nested elements within the heading are displayed as text only.
// Additionally we don't now add a text break after a heading where
// sizeAfter has not been set.
$state['phpword_object']->addTitleStyle($state['structure_headings'][$tag], $state['current_style']);
if (empty($phpword_element) || $phpword_element->getSettings()->getMarginTop() === 0) {
$phpword_element = $phpword_object->createSection();
}
$phpword_element->addTitle(htmltodocx_clean_text($element->nodeValue), $state['structure_headings'][$tag]);
break;
}
if ($tag == 'div' && $state['table_of_contents_id'] && $element->getAttribute('id') == $state['table_of_contents_id']) {
// Replace this div with a table of contents:
if (empty($phpword_element) || $phpword_element->getSettings()->getMarginTop() === 0) {
$phpword_element = $phpword_object->createSection();
}
$phpword_element->addTOC($state['current_style'], $state['current_style']);
break;
}
// Everything in this element should be in the same text run
// we need to initiate a text run here and pass it on. Starting one of
// these elements will cause a new line to be added in the Word
// document. In the case of divs this might not always be what is
// wanted the setting 'treat_div_as_paragraph' determines whether or
// not to add new lines for divs.
if ($tag != 'div' || $state['treat_div_as_paragraph'] || !isset($state['textrun'])) {
if (empty($phpword_element) || $phpword_element->getSettings()->getMarginTop() === 0) {
$phpword_element = $phpword_object->createSection();
}
$state['textrun'] = $phpword_element->createTextRun($state['current_style']);
}
// For better usability for the end user of the Word document, we
// separate paragraphs and headings with an empty line. You can
// override this behaviour by setting the spaceAfter parameter for
// the current element.
// If the spaceAfter parameter is not set, we set it temporarily to 0
// here and record that it wasn't set in the style. Later we will add
// an empty line. Word 2007 and later have a non-zero default for
// paragraph separation, so without setting that spacing to 0 here we
// would end up with a large gap between paragraphs (the document
// template default plus the extra line).
$space_after_set = false;
if (!isset($state['current_style']['spaceAfter'])) {
$state['current_style']['spaceAfter'] = 0;
$space_after_set = false;
}
if (in_array($tag, $allowed_children)) {
array_unshift($state['parents'], $tag);
$phpword_element_returned = htmltodocx_insert_html_recursive($phpword_object, $phpword_element, $element, $state);
if (!empty($phpword_element) && $phpword_element_returned->getSettings()->getOrientation() !== $phpword_element->getSettings()->getOrientation()) {
$phpword_element = $phpword_object->createSection(array('orientation' => $phpword_element->getSettings()->getOrientation()));
} else {
$phpword_element = $phpword_element_returned;
}
array_shift($state['parents']);
} else {
$state['textrun']->addText(htmltodocx_clean_text($element->nodeValue), $state['current_style']);
}
unset($state['textrun']);
if (!$space_after_set) {
// Add the text break here - where the spaceAfter parameter hadn't
// been set initially - also unset the spaceAfter parameter we just
// set:
//$phpword_element->addTextBreak();
unset($state['current_style']['spaceAfter']);
}
break;
case 'table':
if (in_array('table', $allowed_children)) {
$old_table_state = $state['table_allowed'];
if (!$state['table_allowed'] || in_array('td', $state['parents']) || in_array('th', $state['parents'])) {
$state['table_allowed'] = false;
// This is a PhpWord constraint
} else {
$state['table_allowed'] = true;
// PhpWord allows table_styles to be passed in a couple of
// different ways either using an array of properties, or by
// defining a full table style on the PhpWord object:
if (is_object($state['phpword_object']) && method_exists($state['phpword_object'], 'addTableStyle')) {
$state['phpword_object']->addTableStyle('temp_table_style', $state['current_style']);
$table_style = 'temp_table_style';
} else {
$table_style = $state['current_style'];
}
if (empty($phpword_element) || $phpword_element->getSettings()->getMarginTop() === 0) {
$phpword_element = $phpword_object->createSection();
}
$state['table'] = $phpword_element->addTable($table_style);
}
array_unshift($state['parents'], 'table');
$phpword_element_returned = htmltodocx_insert_html_recursive($phpword_object, $phpword_element, $element, $state);
if (!empty($phpword_element) && $phpword_element_returned->getSettings()->getOrientation() !== $phpword_element->getSettings()->getOrientation()) {
$phpword_element = $phpword_object->createSection(array('orientation' => $phpword_element->getSettings()->getOrientation()));
} else {
$phpword_element = $phpword_element_returned;
}
array_shift($state['parents']);
// Reset table state to what it was before a table was added:
$state['table_allowed'] = $old_table_state;
$phpword_element->addTextBreak();
} else {
if (empty($phpword_element) || $phpword_element->getSettings()->getMarginTop() === 0) {
$phpword_element = $phpword_object->createSection();
}
$state['textrun'] = $phpword_element->createTextRun();
$state['textrun']->addText(htmltodocx_clean_text($element->nodeValue), $state['current_style']);
}
break;
case 'tbody':
if (in_array('tbody', $allowed_children)) {
array_unshift($state['parents'], 'tbody');
$phpword_element_returned = htmltodocx_insert_html_recursive($phpword_object, $phpword_element, $element, $state);
if (!empty($phpword_element) && $phpword_element_returned->getSettings()->getOrientation() !== $phpword_element->getSettings()->getOrientation()) {
$phpword_element = $phpword_object->createSection(array('orientation' => $phpword_element->getSettings()->getOrientation()));
} else {
$phpword_element = $phpword_element_returned;
}
array_shift($state['parents']);
} else {
if (empty($phpword_element) || $phpword_element->getSettings()->getMarginTop() === 0) {
$phpword_element = $phpword_object->createSection();
}
$state['textrun'] = $phpword_element->createTextRun();
$state['textrun']->addText(htmltodocx_clean_text($element->nodeValue), $state['current_style']);
}
break;
case 'tr':
if (in_array('tr', $allowed_children)) {
if ($state['table_allowed']) {
$state['table']->addRow();
} else {
// Simply add a new line if a table is not possible in this
// context:
if (empty($phpword_element) || $phpword_element->getSettings()->getMarginTop() === 0) {
$phpword_element = $phpword_object->createSection();
}
$state['textrun'] = $phpword_element->createTextRun();
}
array_unshift($state['parents'], 'tr');
$phpword_element_returned = htmltodocx_insert_html_recursive($phpword_object, $phpword_element, $element, $state);
if (!empty($phpword_element) && $phpword_element_returned->getSettings()->getOrientation() !== $phpword_element->getSettings()->getOrientation()) {
$phpword_element = $phpword_object->createSection(array('orientation' => $phpword_element->getSettings()->getOrientation()));
} else {
$phpword_element = $phpword_element_returned;
}
array_shift($state['parents']);
} else {
if (empty($phpword_element) || $phpword_element->getSettings()->getMarginTop() === 0) {
$phpword_element = $phpword_object->createSection();
}
$state['textrun'] = $phpword_element->createTextRun();
$state['textrun']->addText(htmltodocx_clean_text($element->nodeValue), $state['current_style']);
}
break;
case 'td':
case 'th':
if (in_array($tag, $allowed_children) && $state['table_allowed']) {
unset($state['textrun']);
if (isset($state['current_style']['width'])) {
$cell_width = $state['current_style']['width'];
} elseif (!empty($element->getAttribute('width'))) {
$cell_width = $element->getAttribute('width') * 15;
// Converting at 15 TWIPS per pixel.
} else {
$cell_width = 800;
}
$state['table_cell'] = $state['table']->addCell($cell_width, $state['current_style']);
array_unshift($state['parents'], $tag);
$phpword_element_returned = htmltodocx_insert_html_recursive($phpword_object, $phpword_element, $state['table_cell'], $element, $state);
if (!empty($phpword_element) && $phpword_element_returned->getSettings()->getOrientation() !== $phpword_element->getSettings()->getOrientation()) {
$phpword_element = $phpword_object->createSection(array('orientation' => $phpword_element->getSettings()->getOrientation()));
} else {
$phpword_element = $phpword_element_returned;
}
array_shift($state['parents']);
} else {
if (!isset($state['textrun'])) {
if (empty($phpword_element) || $phpword_element->getSettings()->getMarginTop() === 0) {
$phpword_element = $phpword_object->createSection();
}
$state['textrun'] = $phpword_element->createTextRun();
}
$state['textrun']->addText(htmltodocx_clean_text($element->nodeValue), $state['current_style']);
}
break;
case 'a':
/* var_dump($element->href);
var_dump(get_class($state['textrun']));
if($phpword_element instanceof \PhpOffice\PhpWord\Element\Footnote)
{
var_dump($phpword_element instanceof MyClass);
}*/
// Create a new text run if we aren't in one already:
if (!isset($state['textrun'])) {
if (empty($phpword_element) || $phpword_element->getSettings()->getMarginTop() === 0) {
$phpword_element = $phpword_object->createSection();
}
$state['textrun'] = $phpword_element->createTextRun();
}
if ($state['context'] == 'section') {
if (strpos($element->getAttribute('href'), 'http://') === 0) {
$href = $element->getAttribute('href');
} else {
if (strpos($element->getAttribute('href'), 'https://') === 0) {
$href = $element->getAttribute('href');
} elseif (strpos($element->getAttribute('href'), '/') === 0) {
$href = $state['base_root'] . $element->getAttribute('href');
} else {
$href = $state['base_root'] . $state['base_path'] . $element->getAttribute('href');
}
}
// Replace any spaces in url with %20 - to prevent errors in the Word
// document:
$state['textrun']->addLink(htmltodocx_url_encode_chars($href), htmltodocx_clean_text($element->nodeValue), $state['current_style']);
} else {
// Links can't seem to be included in headers or footers with
// PHPWord: trying to include them causes an error which stops Word
// from opening the file - in Word 2003 with the converter at least.
// So add the link styled as a link only.
$state['textrun']->addText(htmltodocx_clean_text($element->nodeValue), $state['current_style']);
}
break;
case 'ul':
$state['list_total_count'] = $element->childNodes->length;
// We use this to be able to add the ordered list spaceAfter onto the
// last list element. All ol children should be li elements.
_htmltodocx_add_list_start_end_spacing_style($state);
$state['list_number'] = 0;
// Reset list number.
if (in_array('ul', $allowed_children)) {
if (!isset($state['pseudo_list'])) {
// Unset any existing text run:
unset($state['textrun']);
// PhpWord lists cannot appear in a text run. If we leave a text
// run active then subsequent text will go in that text run (if it
// isn't re-initialised), which would mean that text after this
// list would appear before it in the Word document.
}
array_unshift($state['parents'], 'ul');
$phpword_element_returned = htmltodocx_insert_html_recursive($phpword_object, $phpword_element, $element, $state);
if (!empty($phpword_element) && $phpword_element_returned->getSettings()->getOrientation() !== $phpword_element->getSettings()->getOrientation()) {
$phpword_element = $phpword_object->createSection(array('orientation' => $phpword_element->getSettings()->getOrientation()));
} else {
$phpword_element = $phpword_element_returned;
}
array_shift($state['parents']);
} else {
if (empty($phpword_element) || $phpword_element->getSettings()->getMarginTop() === 0) {
$phpword_element = $phpword_object->createSection();
}
$state['textrun'] = $phpword_element->createTextRun();
$state['textrun']->addText(htmltodocx_clean_text($element->nodeValue), $state['current_style']);
}
break;
case 'ol':
$state['list_total_count'] = $element->childNodes->length;
// We use this to be able to add the ordered list spaceAfter onto the
// last list element. All ol children should be li elements.
_htmltodocx_add_list_start_end_spacing_style($state);
$state['list_number'] = 0;
// Reset list number.
if (in_array('ol', $allowed_children)) {
if (!isset($state['pseudo_list'])) {
// Unset any existing text run:
unset($state['textrun']);
// Lists cannot appear in a text run. If we leave a text run active
// then subsequent text will go in that text run (if it isn't
// re-initialised), which would mean that text after this list
// would appear before it in the Word document.
}
array_unshift($state['parents'], 'ol');
$phpword_element_returned = htmltodocx_insert_html_recursive($phpword_object, $phpword_element, $element, $state);
if (!empty($phpword_element) && $phpword_element_returned->getSettings()->getOrientation() !== $phpword_element->getSettings()->getOrientation()) {
$phpword_element = $phpword_object->createSection(array('orientation' => $phpword_element->getSettings()->getOrientation()));
} else {
$phpword_element = $phpword_element_returned;
}
array_shift($state['parents']);
} else {
if (empty($phpword_element) || $phpword_element->getSettings()->getMarginTop() === 0) {
$phpword_element = $phpword_object->createSection();
}
$state['textrun'] = $phpword_element->createTextRun();
$state['textrun']->addText(htmltodocx_clean_text($element->nodeValue), $state['current_style']);
}
break;
case 'li':
// You cannot style individual pieces of text in a list element so we do it
// with text runs instead. This does not allow us to indent lists at all, so
// we can't show nesting.
// Before and after spacings:
if ($state['list_number'] === 0) {
$state['current_style'] = array_merge($state['current_style'], $state['list_style_before']);
}
$last_item = false;
if ($state['list_number'] == $state['list_total_count'] - 1) {
$last_item = true;
if (empty($state['list_style_after'])) {
$state['current_style']['spaceAfter'] = 0;
// Set to 0 if not defined so we can add a text break without
// ending up within too much space in Word2007+.
// *Needs further testing on Word 2007+*
}
$state['current_style'] = array_merge($state['current_style'], $state['list_style_after']);
}
// We create a new text run for each element:
if (empty($phpword_element) || $phpword_element->getSettings()->getMarginTop() === 0) {
$phpword_element = $phpword_object->createSection();
}
$state['textrun'] = $phpword_element->createTextRun($state['current_style']);
if (in_array('li', $allowed_children)) {
$state['list_number']++;
if ($state['parents'][0] == 'ol') {
$item_indicator = $state['list_number'] . '. ';
$style = $state['current_style'];
} else {
$style = $state['current_style'];
$style['name'] = $state['pseudo_list_indicator_font_name'];
$style['size'] = $state['pseudo_list_indicator_font_size'];
$item_indicator = $state['pseudo_list_indicator_character'];
}
array_unshift($state['parents'], 'li');
$state['textrun']->addText($item_indicator, $style);
$phpword_element_returned = htmltodocx_insert_html_recursive($phpword_object, $phpword_element, $element, $state);
if (!empty($phpword_element) && $phpword_element_returned->getSettings()->getOrientation() !== $phpword_element->getSettings()->getOrientation()) {
$phpword_element = $phpword_object->createSection(array('orientation' => $phpword_element->getSettings()->getOrientation()));
} else {
$phpword_element = $phpword_element_returned;
}
array_shift($state['parents']);
} else {
$state['textrun']->addText(htmltodocx_clean_text($element->nodeValue), $state['current_style']);
}
if ($last_item && empty($state['list_style_after'])) {
if (empty($phpword_element) || $phpword_element->getSettings()->getMarginTop() === 0) {
$phpword_element = $phpword_object->createSection();
}
$phpword_element->addTextBreak();
// Add an empty line after the list if no spacing after has been
// defined.
}
unset($state['textrun']);
break;
case 'text':
// We may get some empty text nodes - containing just a space - in
// simple HTML dom - we want to exclude those, as these can cause extra
// line returns. However we don't want to exclude spaces between styling
// elements (these will be within a text run).
if (!isset($state['textrun'])) {
$text = htmltodocx_clean_text(trim($element->nodeValue));
} else {
$text = htmltodocx_clean_text($element->nodeValue);
}
if (!empty($text)) {
if (!isset($state['textrun'])) {
if (empty($phpword_element)) {
$phpword_element = $phpword_object->createSection();
}
$state['textrun'] = $phpword_element->createTextRun();
}
$state['textrun']->addText($text, $state['current_style']);
}
break;
case '#text':
// We may get some empty text nodes - containing just a space - in
// simple HTML dom - we want to exclude those, as these can cause extra
// line returns. However we don't want to exclude spaces between styling
// elements (these will be within a text run).
if (!isset($state['textrun'])) {
$text = htmltodocx_clean_text(trim($element->nodeValue));
} else {
$text = htmltodocx_clean_text($element->nodeValue);
}
if (!empty($text)) {
if (!isset($state['textrun'])) {
if (empty($phpword_element)) {
$phpword_element = $phpword_object->createSection();
}
$state['textrun'] = $phpword_element->createTextRun();
}
$state['textrun']->addText($text, $state['current_style']);
}
break;
// Style tags:
// Style tags:
case 'strong':
case 'b':
case 'sup':
// Not working in PHPWord
// Not working in PHPWord
case 'em':
case 'i':
case 'u':
case 'span':
case 'code':
// Create a new text run if we aren't in one already:
if (!isset($state['textrun'])) {
if (empty($phpword_element) || $phpword_element->getSettings()->getMarginTop() === 0) {
$phpword_element = $phpword_object->createSection();
}
$state['textrun'] = $phpword_element->createTextRun();
}
if (in_array($tag, $allowed_children)) {
array_unshift($state['parents'], $tag);
$phpword_element_returned = htmltodocx_insert_html_recursive($phpword_object, $phpword_element, $element, $state);
if (!empty($phpword_element) && $phpword_element_returned->getSettings()->getOrientation() !== $phpword_element->getSettings()->getOrientation()) {
$phpword_element = $phpword_object->createSection(array('orientation' => $phpword_element->getSettings()->getOrientation()));
} else {
$phpword_element = $phpword_element_returned;
}
array_shift($state['parents']);
} else {
$state['textrun']->addText(htmltodocx_clean_text($element->nodeValue), $state['current_style']);
}
break;
// NB, Simple HTML Dom might not be picking up <br> tags.
// NB, Simple HTML Dom might not be picking up <br> tags.
case 'br':
if ($state['textrun'] instanceof \PhpOffice\PhpWord\Element\Footnote) {
$state['textrun']->addTextBreak();
} else {
// Simply create a new text run:
if (empty($phpword_element) || $phpword_element->getSettings()->getMarginTop() === 0) {
$phpword_element = $phpword_object->createSection();
}
$state['textrun'] = $phpword_element->createTextRun();
}
break;
case 'img':
$image_style = array();
$element_src = $element->getAttribute('src');
if (strpos($element_src, 'http://') === 0) {
// The image url is from another site. Most probably the image won't
// appear in the Word document.
$src = $element_src;
} elseif (strpos($element_src, 'https://') === 0) {
// The image url is from another site. Most probably the image won't
// appear in the Word document.
$src = $element_src;
} elseif (strpos($element_src, '/') === 0) {
$src = $element_src;
} else {
$src = htmltodocx_doc_root() . $state['base_path'] . $element_src;
}
//if (file_exists($src))
if (file_get_contents($src, 0, null, 0, 1)) {
$switchFromPortraitToLandscape = false;
if ($element->getAttribute('width') && $element->getAttribute('height')) {
$aspectRatio = $element->getAttribute('width') / $element->getAttribute('height');
if ($aspectRatio > 1.0) {
$phpword_element = $phpword_object->createSection(array('orientation' => 'landscape', 'marginLeft' => 0, 'marginRight' => 0, 'marginTop' => 0, 'marginBottom' => 0));
$state['current_style']['width'] = $a4WidthLandscapeOrientationInPoints;
$state['current_style']['height'] = $a4WidthLandscapeOrientationInPoints / $aspectRatio;
$switchFromPortraitToLandscape = true;
} else {
$phpword_element = $phpword_object->createSection(array('orientation' => 'portrait', 'marginLeft' => 0, 'marginRight' => 0, 'marginTop' => 0, 'marginBottom' => 0));
$state['current_style']['height'] = $a4HeightPortraitOrientationInPoints;
$state['current_style']['width'] = $a4HeightPortraitOrientationInPoints * $aspectRatio;
}
} else {
$phpword_element = $phpword_object->createSection(array('orientation' => 'portrait', 'marginLeft' => 0, 'marginRight' => 0, 'marginTop' => 0, 'marginBottom' => 0));
$state['current_style']['height'] = $a4HeightPortraitOrientationInPoints;
$state['current_style']['width'] = $a4WidthPortraitOrientationInPoints;
}
if ($switchFromPortraitToLandscape === true) {
if (empty($phpword_element)) {
$phpword_element = $phpword_object->createSection();
}
$phpword_element->addImage($src, $state['current_style']);
} else {
if (empty($phpword_element)) {
$phpword_element = $phpword_object->createSection();
}
$phpword_element->addImage($src, $state['current_style']);
}
}
break;
case 'pb':
if (empty($phpword_element)) {
$phpword_element = $phpword_object->createSection();
}
$phpword_element->addPageBreak();
break;
case 'footnote':
if ($element->getAttribute("text")) {
$state['textrun'] = $phpword_element->createTextRun();
$state['textrun']->addText(htmltodocx_clean_text($element->getAttribute("text")), $state['current_style']);
}
$textrunBeforeFootnote = $state['textrun'];
$styleBeforeFootnote = $state['current_style'];
$state['current_style'] = array();
$state['textrun'] = $state['textrun']->addFootnote();
array_unshift($state['parents'], $tag);
$phpword_element_returned = htmltodocx_insert_html_recursive($phpword_object, $phpword_element, $element, $state);
if (!empty($phpword_element) && $phpword_element_returned->getSettings()->getOrientation() !== $phpword_element->getSettings()->getOrientation()) {
$phpword_element = $phpword_object->createSection(array('orientation' => $phpword_element->getSettings()->getOrientation()));
} else {
$phpword_element = $phpword_element_returned;
}
array_shift($state['parents']);
$state['textrun'] = $textrunBeforeFootnote;
$state['current_style'] = $styleBeforeFootnote;
break;
default:
if (empty($phpword_element)) {
$phpword_element = $phpword_object->createSection();
}
$state['textrun'] = $phpword_element->createTextRun();
$state['textrun']->addText(htmltodocx_clean_text($element->nodeValue), $state['current_style']);
break;
}
// Reset the style back to what it was:
$state['current_style'] = $old_style;
}
break;
}
if (empty($phpword_element)) {
$phpword_element = $phpword_object->createSection();
}
return $phpword_element;
}