获取word文件文本内容

获取word文件文本内容及将.doc类型文件转化成.docx类型文件

	use PhpOffice\PhpWord\IOFactory;
	public function readWord($path,$ends='docx')
    {
        if($ends == 'doc'){
            //此处是针对于doc文件，在vue预览插件中不能预览.doc类型的文件，需要先将.doc转化成.docx文件
            $pathLast =  dirname($path);
           // LibreOffice的命令行工具的路径
            //$libreofficePath = '/usr/bin/soffice';//linux环境
            $libreofficePath = "F:\www\LibreOffice\program\soffice.exe";//window环境
           // 构造命令行，将doc格式内容转化成docx格式内容，并且添加到$pathLast文件中
            $command = escapeshellcmd("$libreofficePath --headless --convert-to docx --outdir " . $pathLast . " $path");
            // 执行命令
            system($command);
            //删除doc文件
            $content1 = file_get_contents($path);
            unlink($path);
            $path .= 'x';
            $content2 = file_get_contents($path);
            if($content1 && empty($content2)){
                unlink($path);
                //识别失败
                $return['code'] = 400;
                return $return;
            }
        }
        $return['code'] = 200;
        $phpWord = IOFactory::load($path);
        //读取文件文本内容
        $return['content'] = $this->getNodeContent($phpWord);
        $return['path'] = $path;
        return $return;
    }

    /**
     * 根据word主节点获取分节点内容
     * @param $word
     * @return array
     */
    public function getNodeContent($word)
    {
        $return = [];
        //分解部分
        foreach ($word->getSections() as $section)
        {
            if ($section instanceof \PhpOffice\PhpWord\Element\Section) {
                //分解元素
                foreach ($section->getElements() as $element)
                {
                    //文本元素
                    if ($element instanceof \PhpOffice\PhpWord\Element\TextRun) {
                        $text = '';
                        foreach ($element->getElements() as $ele) {
                            $text .= $this->getTextNode($ele);
                        }
                        $return[] = $text;
                    }
                    //表格元素
                    else if ($element instanceof \PhpOffice\PhpWord\Element\Table) {
                        foreach ($element->getRows() as $ele)
                        {
                            $return[] = $this->getTableNode($ele);
                        }
                    }
                }
            }
        }
        return $return;
    }

    /**
     * 获取文档节点内容
     * @param $node
     * @return string
     */
    public function getTextNode($node)
    {
        $return = '';
        //处理文本
        if ($node instanceof \PhpOffice\PhpWord\Element\Text)
        {
            $return .= $node->getText();
        }
        //处理图片
        else if ($node instanceof \PhpOffice\PhpWord\Element\Image)
        {
            $return .= $this->pic2text($node);
        }
        //处理文本元素
        else if ($node instanceof \PhpOffice\PhpWord\Element\TextRun) {
            foreach ($node->getElements() as $ele) {
                $return .= $this->getTextNode($ele);
            }
        }
        return $return;
    }

    /**
     * 获取表格节点内容
     * @param $node
     * @return string
     */
    public function getTableNode($node)
    {
        $return = '';
        //处理行
        if ($node instanceof \PhpOffice\PhpWord\Element\Row) {
            foreach ($node->getCells() as $ele)
            {
                $return .= $this->getTableNode($ele);
            }
        }
        //处理列
        else if ($node instanceof \PhpOffice\PhpWord\Element\Cell) {
            foreach ($node->getElements() as $ele)
            {
                $return .= $this->getTextNode($ele);
            }
        }
        return $return;
    }

    /**
     * 处理word文档中base64格式图片
     * @param $node
     * @return string
     */
    public function pic2text($node)
    {
        //获取图片编码
        $imageData = $node->getImageStringData(true);
        //添加图片html显示标头
        $imageData = 'data:' . $node->getImageType() . ';base64,' . $imageData;
        $return = '<img src="'.$imageData.'">';
        return $return;
    }