This commit is contained in:
wangjinlei
2024-07-17 09:23:45 +08:00
parent edf9deeb1a
commit 881ac3e056
1001 changed files with 41032 additions and 5452 deletions

View File

@@ -0,0 +1,7 @@
<?php
namespace PhpOffice\Math\Element;
abstract class AbstractElement
{
}

View File

@@ -0,0 +1,35 @@
<?php
namespace PhpOffice\Math\Element;
abstract class AbstractGroupElement extends AbstractElement
{
/**
* @var AbstractElement[]
*/
protected $elements = [];
public function add(AbstractElement $element): self
{
$this->elements[] = $element;
return $this;
}
public function remove(AbstractElement $element): self
{
$this->elements = array_filter($this->elements, function ($child) use ($element) {
return $child != $element;
});
return $this;
}
/**
* @return AbstractElement[]
*/
public function getElements(): array
{
return $this->elements;
}
}

View File

@@ -0,0 +1,46 @@
<?php
namespace PhpOffice\Math\Element;
class Fraction extends AbstractElement
{
/**
* @var AbstractElement
*/
protected $denominator;
/**
* @var AbstractElement
*/
protected $numerator;
public function __construct(AbstractElement $numerator, AbstractElement $denominator)
{
$this->setNumerator($numerator);
$this->setDenominator($denominator);
}
public function getDenominator(): AbstractElement
{
return $this->denominator;
}
public function getNumerator(): AbstractElement
{
return $this->numerator;
}
public function setDenominator(AbstractElement $element): self
{
$this->denominator = $element;
return $this;
}
public function setNumerator(AbstractElement $element): self
{
$this->numerator = $element;
return $this;
}
}

View File

@@ -0,0 +1,21 @@
<?php
namespace PhpOffice\Math\Element;
class Identifier extends AbstractElement
{
/**
* @var string
*/
protected $value;
public function __construct(string $value)
{
$this->value = $value;
}
public function getValue(): string
{
return $this->value;
}
}

View File

@@ -0,0 +1,21 @@
<?php
namespace PhpOffice\Math\Element;
class Numeric extends AbstractElement
{
/**
* @var float
*/
protected $value;
public function __construct(float $value)
{
$this->value = $value;
}
public function getValue(): float
{
return $this->value;
}
}

View File

@@ -0,0 +1,21 @@
<?php
namespace PhpOffice\Math\Element;
class Operator extends AbstractElement
{
/**
* @var string
*/
protected $value;
public function __construct(string $value)
{
$this->value = $value;
}
public function getValue(): string
{
return $this->value;
}
}

View File

@@ -0,0 +1,7 @@
<?php
namespace PhpOffice\Math\Element;
class Row extends AbstractGroupElement
{
}

View File

@@ -0,0 +1,33 @@
<?php
declare(strict_types=1);
namespace PhpOffice\Math\Element;
class Semantics extends AbstractGroupElement
{
/**
* @var array<string, string>
*/
protected $annotations = [];
public function addAnnotation(string $encoding, string $annotation): self
{
$this->annotations[$encoding] = $annotation;
return $this;
}
public function getAnnotation(string $encoding): ?string
{
return $this->annotations[$encoding] ?? null;
}
/**
* @return array<string, string>
*/
public function getAnnotations(): array
{
return $this->annotations;
}
}

View File

@@ -0,0 +1,46 @@
<?php
namespace PhpOffice\Math\Element;
class Superscript extends AbstractElement
{
/**
* @var AbstractElement
*/
protected $base;
/**
* @var AbstractElement
*/
protected $superscript;
public function __construct(AbstractElement $base, AbstractElement $superscript)
{
$this->setBase($base);
$this->setSuperscript($superscript);
}
public function getBase(): AbstractElement
{
return $this->base;
}
public function getSuperscript(): AbstractElement
{
return $this->superscript;
}
public function setBase(AbstractElement $element): self
{
$this->base = $element;
return $this;
}
public function setSuperscript(AbstractElement $element): self
{
$this->superscript = $element;
return $this;
}
}

View File

@@ -0,0 +1,7 @@
<?php
namespace PhpOffice\Math\Exception;
class InvalidInputException extends MathException
{
}

View File

@@ -0,0 +1,9 @@
<?php
namespace PhpOffice\Math\Exception;
use Exception;
class MathException extends Exception
{
}

View File

@@ -0,0 +1,7 @@
<?php
namespace PhpOffice\Math\Exception;
class NotImplementedException extends MathException
{
}

View File

@@ -0,0 +1,9 @@
<?php
namespace PhpOffice\Math;
use PhpOffice\Math\Element\AbstractGroupElement;
class Math extends AbstractGroupElement
{
}

View File

@@ -0,0 +1,136 @@
<?php
namespace PhpOffice\Math\Reader;
use DOMDocument;
use DOMElement;
use DOMNode;
use DOMXPath;
use PhpOffice\Math\Element;
use PhpOffice\Math\Exception\InvalidInputException;
use PhpOffice\Math\Exception\NotImplementedException;
use PhpOffice\Math\Math;
class MathML implements ReaderInterface
{
/** @var Math */
private $math;
/** @var DOMDocument */
private $dom;
/** @var DOMXpath */
private $xpath;
public function read(string $content): ?Math
{
$content = str_replace(
[
'&InvisibleTimes;',
],
[
'<mchar name="InvisibleTimes"/>',
],
$content
);
$this->dom = new DOMDocument();
$this->dom->loadXML($content, LIBXML_DTDLOAD);
$this->math = new Math();
$this->parseNode(null, $this->math);
return $this->math;
}
/**
* @param Math|Element\AbstractGroupElement $parent
*/
protected function parseNode(?DOMNode $nodeRowElement, $parent): void
{
$this->xpath = new DOMXpath($this->dom);
foreach ($this->xpath->query('*', $nodeRowElement) ?: [] as $nodeElement) {
if ($parent instanceof Element\Semantics
&& $nodeElement instanceof DOMElement
&& $nodeElement->nodeName == 'annotation') {
$parent->addAnnotation(
$nodeElement->getAttribute('encoding'),
trim($nodeElement->nodeValue)
);
continue;
}
$element = $this->getElement($nodeElement);
$parent->add($element);
if ($element instanceof Element\AbstractGroupElement) {
$this->parseNode($nodeElement, $element);
}
}
}
protected function getElement(DOMNode $nodeElement): Element\AbstractElement
{
$nodeValue = trim($nodeElement->nodeValue);
switch ($nodeElement->nodeName) {
case 'mfrac':
$nodeList = $this->xpath->query('*', $nodeElement);
if ($nodeList && $nodeList->length == 2) {
return new Element\Fraction(
$this->getElement($nodeList->item(0)),
$this->getElement($nodeList->item(1))
);
}
throw new InvalidInputException(sprintf(
'%s : The tag `%s` has not two subelements',
__METHOD__,
$nodeElement->nodeName
));
case 'mi':
return new Element\Identifier($nodeValue);
case 'mn':
return new Element\Numeric(floatval($nodeValue));
case 'mo':
if (empty($nodeValue)) {
$nodeList = $this->xpath->query('*', $nodeElement);
if (
$nodeList
&& $nodeList->length == 1
&& $nodeList->item(0)->nodeName == 'mchar'
&& $nodeList->item(0) instanceof DOMElement
&& $nodeList->item(0)->hasAttribute('name')
) {
$nodeValue = $nodeList->item(0)->getAttribute('name');
}
}
return new Element\Operator($nodeValue);
case 'mrow':
return new Element\Row();
case 'msup':
$nodeList = $this->xpath->query('*', $nodeElement);
if ($nodeList && $nodeList->length == 2) {
return new Element\Superscript(
$this->getElement($nodeList->item(0)),
$this->getElement($nodeList->item(1))
);
}
throw new InvalidInputException(sprintf(
'%s : The tag `%s` has not two subelements',
__METHOD__,
$nodeElement->nodeName
));
case 'semantics':
return new Element\Semantics();
default:
throw new NotImplementedException(sprintf(
'%s : The tag `%s` is not implemented',
__METHOD__,
$nodeElement->nodeName
));
}
}
}

View File

@@ -0,0 +1,133 @@
<?php
namespace PhpOffice\Math\Reader;
use DOMDocument;
use DOMNode;
use DOMXPath;
use PhpOffice\Math\Element;
use PhpOffice\Math\Exception\InvalidInputException;
use PhpOffice\Math\Exception\NotImplementedException;
use PhpOffice\Math\Math;
class OfficeMathML implements ReaderInterface
{
/** @var DOMDocument */
protected $dom;
/** @var Math */
protected $math;
/** @var DOMXpath */
protected $xpath;
/** @var string[] */
protected $operators = ['+', '-', '/', ''];
public function read(string $content): ?Math
{
$nsMath = 'xmlns:m="http://schemas.openxmlformats.org/officeDocument/2006/math"';
$nsWord = 'xmlns:w="http://schemas.openxmlformats.org/wordprocessingml/2006/main"';
$content = str_replace(
$nsMath,
$nsMath . ' ' . $nsWord,
$content
);
$this->dom = new DOMDocument();
$this->dom->loadXML($content);
$this->math = new Math();
$this->parseNode(null, $this->math);
return $this->math;
}
/**
* @see https://devblogs.microsoft.com/math-in-office/officemath/
* @see https://learn.microsoft.com/fr-fr/archive/blogs/murrays/mathml-and-ecma-math-omml
*
* @param Math|Element\AbstractGroupElement $parent
*/
protected function parseNode(?DOMNode $nodeRowElement, $parent): void
{
$this->xpath = new DOMXpath($this->dom);
foreach ($this->xpath->query('*', $nodeRowElement) ?: [] as $nodeElement) {
$element = $this->getElement($nodeElement);
$parent->add($element);
if ($element instanceof Element\AbstractGroupElement) {
$this->parseNode($nodeElement, $element);
}
}
}
protected function getElement(DOMNode $nodeElement): Element\AbstractElement
{
switch ($nodeElement->nodeName) {
case 'm:f':
// Numerator
$nodeNumerator = $this->xpath->query('m:num/m:r/m:t', $nodeElement);
if ($nodeNumerator && $nodeNumerator->length == 1) {
$value = $nodeNumerator->item(0)->nodeValue;
if (is_numeric($value)) {
$numerator = new Element\Numeric(floatval($value));
} else {
$numerator = new Element\Identifier($value);
}
} else {
throw new InvalidInputException(sprintf(
'%s : The tag `%s` has no numerator defined',
__METHOD__,
$nodeElement->nodeName
));
}
// Denominator
$nodeDenominator = $this->xpath->query('m:den/m:r/m:t', $nodeElement);
if ($nodeDenominator && $nodeDenominator->length == 1) {
$value = $nodeDenominator->item(0)->nodeValue;
if (is_numeric($value)) {
$denominator = new Element\Numeric(floatval($value));
} else {
$denominator = new Element\Identifier($value);
}
} else {
throw new InvalidInputException(sprintf(
'%s : The tag `%s` has no denominator defined',
__METHOD__,
$nodeElement->nodeName
));
}
return new Element\Fraction($numerator, $denominator);
case 'm:r':
$nodeText = $this->xpath->query('m:t', $nodeElement);
if ($nodeText && $nodeText->length == 1) {
$value = trim($nodeText->item(0)->nodeValue);
if (in_array($value, $this->operators)) {
return new Element\Operator($value);
}
if (is_numeric($value)) {
return new Element\Numeric(floatval($value));
}
return new Element\Identifier($value);
}
throw new InvalidInputException(sprintf(
'%s : The tag `%s` has no tag `m:t` defined',
__METHOD__,
$nodeElement->nodeName
));
case 'm:oMath':
return new Element\Row();
default:
throw new NotImplementedException(sprintf(
'%s : The tag `%s` is not implemented',
__METHOD__,
$nodeElement->nodeName
));
}
}
}

View File

@@ -0,0 +1,10 @@
<?php
namespace PhpOffice\Math\Reader;
use PhpOffice\Math\Math;
interface ReaderInterface
{
public function read(string $content): ?Math;
}

View File

@@ -0,0 +1,131 @@
<?php
namespace PhpOffice\Math\Writer;
use PhpOffice\Math\Element;
use PhpOffice\Math\Exception\NotImplementedException;
use PhpOffice\Math\Math;
use XMLWriter;
class MathML implements WriterInterface
{
/** @var XMLWriter */
private $output;
/**
* @param Math $math
*
* @return string
*/
public function write(Math $math): string
{
$this->output = new XMLWriter();
$this->output->openMemory();
$this->output->startDocument('1.0', 'UTF-8');
$this->output->writeDtd('math', '-//W3C//DTD MathML 2.0//EN', 'http://www.w3.org/Math/DTD/mathml2/mathml2.dtd');
$this->output->startElement('math');
$this->output->writeAttribute('xmlns', 'http://www.w3.org/1998/Math/MathML');
foreach ($math->getElements() as $element) {
$this->writeElementItem($element);
}
$this->output->endElement();
$this->output->endDocument();
return $this->output->outputMemory();
}
protected function writeElementItem(Element\AbstractElement $element): void
{
$tagName = $this->getElementTagName($element);
// Element\AbstractGroupElement
if ($element instanceof Element\AbstractGroupElement) {
$this->output->startElement($tagName);
foreach ($element->getElements() as $childElement) {
$this->writeElementItem($childElement);
}
$this->output->endElement();
return;
}
// Element\Superscript
if ($element instanceof Element\Superscript) {
$this->output->startElement($tagName);
$this->writeElementItem($element->getBase());
$this->writeElementItem($element->getSuperscript());
$this->output->endElement();
return;
}
// Element\Fraction
if ($element instanceof Element\Fraction) {
$this->output->startElement($tagName);
$this->writeElementItem($element->getNumerator());
$this->writeElementItem($element->getDenominator());
$this->output->endElement();
return;
}
if ($element instanceof Element\Identifier
|| $element instanceof Element\Numeric
|| $element instanceof Element\Operator) {
$this->output->startElement($tagName);
$this->output->text((string) $element->getValue());
$this->output->endElement();
return;
}
/*
throw new NotImplementedException(sprintf(
'%s : The class `%s` is not implemented',
__METHOD__,
get_class($element)
));
*/
}
protected function getElementTagName(Element\AbstractElement $element): string
{
// Group
if ($element instanceof Element\Row) {
return 'mrow';
}
if ($element instanceof Element\AbstractGroupElement) {
/*
throw new NotImplementedException(sprintf(
'%s : The element of the class `%s` has no tag name',
__METHOD__,
get_class($element)
));
*/
}
if ($element instanceof Element\Superscript) {
return 'msup';
}
if ($element instanceof Element\Fraction) {
return 'mfrac';
}
if ($element instanceof Element\Identifier) {
return 'mi';
}
if ($element instanceof Element\Numeric) {
return 'mn';
}
if ($element instanceof Element\Operator) {
return 'mo';
}
throw new NotImplementedException(sprintf(
'%s : The element of the class `%s` has no tag name',
__METHOD__,
get_class($element)
));
}
}

View File

@@ -0,0 +1,102 @@
<?php
namespace PhpOffice\Math\Writer;
use PhpOffice\Math\Element;
use PhpOffice\Math\Exception\NotImplementedException;
use PhpOffice\Math\Math;
use XMLWriter;
class OfficeMathML implements WriterInterface
{
/** @var XMLWriter */
private $output;
/**
* @param Math $math
*
* @return string
*/
public function write(Math $math): string
{
$this->output = new XMLWriter();
$this->output->openMemory();
$this->output->startElement('m:oMathPara');
$this->output->writeAttribute('xmlns:m', 'http://schemas.openxmlformats.org/officeDocument/2006/math');
$this->output->startElement('m:oMath');
foreach ($math->getElements() as $element) {
$this->writeElementItem($element);
}
$this->output->endElement();
$this->output->endElement();
return $this->output->outputMemory();
}
protected function writeElementItem(Element\AbstractElement $element): void
{
// Element\Row
if ($element instanceof Element\Row) {
foreach ($element->getElements() as $childElement) {
$this->writeElementItem($childElement);
}
return;
}
// Element\Fraction
if ($element instanceof Element\Fraction) {
$this->output->startElement($this->getElementTagName($element));
$this->output->startElement('m:num');
$this->writeElementItem($element->getNumerator());
$this->output->endElement();
$this->output->startElement('m:den');
$this->writeElementItem($element->getDenominator());
$this->output->endElement();
$this->output->endElement();
return;
}
if ($element instanceof Element\Identifier
|| $element instanceof Element\Numeric
|| $element instanceof Element\Operator) {
$this->output->startElement('m:r');
$this->output->startElement('m:t');
$this->output->text((string) $element->getValue());
$this->output->endElement();
$this->output->endElement();
return;
}
// Check if managed
$this->getElementTagName($element);
}
protected function getElementTagName(Element\AbstractElement $element): string
{
// Group
if ($element instanceof Element\AbstractGroupElement) {
/*
throw new NotImplementedException(sprintf(
'%s : The element of the class `%s` has no tag name',
__METHOD__,
get_class($element)
));
*/
}
if ($element instanceof Element\Fraction) {
return 'm:f';
}
throw new NotImplementedException(sprintf(
'%s : The element of the class `%s` has no tag name',
__METHOD__,
get_class($element)
));
}
}

View File

@@ -0,0 +1,10 @@
<?php
namespace PhpOffice\Math\Writer;
use PhpOffice\Math\Math;
interface WriterInterface
{
public function write(Math $math): string;
}