[poi-tl]转换html内容到word

  baikapala

    利用jsoup将html解析,递归解析,将不同的html标签内容转换成poi支持的内容。代码仅供参考。

    
    /**
     * 创建测试用例过程记录的渲染策略:解析对应的html并输出到word
     * @author xuwangcheng
     * @date 2019/7/26 10:10
     * @param
      * @return {@link AbstractRenderPolicy}
     */  private static AbstractRenderPolicy createHtmlRenderPolicy() {
            return new AbstractRenderPolicy() {
                @Override
      protected void afterRender(RenderContext context) {
                    // 清空模板标签所在段落
      clearPlaceholder(context, true);
                }
    
                @Override
      public void doRender(RunTemplate runTemplate, Object data, XWPFTemplate template) throws Exception {
                    if (data == null || StringUtils.isBlank(data.toString())) {
                        return;
                    }
    
                    //获得Apache POI增强类NiceXWPFDocument
      NiceXWPFDocument doc = template.getXWPFDocument();
    
                    String html = data.toString();
                    html.replaceAll(">", ">");
                    html.replaceAll("<", "<");
                    html.replaceAll(" ", " ");
                    html.replaceAll("\\n", "");
                    html.replaceAll("", "\n");
                    org.jsoup.nodes.Document htmlDoc = Jsoup.parse(html);
                    Elements nodes = htmlDoc.body().children();
                    XWPFParagraph xwpfParagraph = doc.createParagraph();
    
                    ListIterator itr = nodes.listIterator();
                    while (itr.hasNext()) {
                        Element e = itr.next();
                        xwpfParagraph = parseHtmlToWord(e, doc, xwpfParagraph, true);
                    }
                }
            };
        }
    
        /**
     * 转换整个html内容为word内容
      * @author xuwangcheng
     * @date 2019/7/29 18:46
     * @param ele ele
     * @param doc doc
     * @param xwpfParagraph xwpfParagraph
     * @return {@link XWPFParagraph}
     */  private static XWPFParagraph parseHtmlToWord(Element ele, NiceXWPFDocument doc, XWPFParagraph xwpfParagraph
                , boolean isParent) throws Exception {
    
            //处理img图片
      if ("img".equals(ele.tagName())) {
                parseImgToWord(ele.attr("src"), xwpfParagraph);
                return xwpfParagraph;
            }
    
            //处理table标签
      if ("table".equals(ele.tagName())) {
                parseTableToWord(doc, ele);
                return xwpfParagraph;
            }
    
            String text = ele.ownText();
            boolean enabledBreak = (isParent || StringUtils.isNotBlank(text))
                    && ReUtil.isMatch("(p|h[12345]|li|img)", ele.tagName());
            if (enabledBreak) {
                XWPFRun run = xwpfParagraph.createRun();
                //run.addCarriageReturn();
      run.addBreak();
            }
    
            if (StringUtils.isNotBlank(text)) {
                XWPFRun run = xwpfParagraph.createRun();
                TextRenderPolicy.Helper.renderTextRun(run, new TextRenderData(text));
            }
    
            if (ele.children().size() > 0) {
                ListIterator itr = ele.children().listIterator();
                while (itr.hasNext()) {
                    Element me = itr.next();
                    xwpfParagraph = parseHtmlToWord(me, doc, xwpfParagraph, false);
                }
            }
    
            return xwpfParagraph;
        }
    
        /**
     * 转换图片为word内容
      * @author xuwangcheng
     * @date 2019/7/29 18:45
     * @param imgUrl imgUrl
     * @param xwpfParagraph xwpfParagraph
     */  private static void parseImgToWord(String imgUrl, XWPFParagraph xwpfParagraph) throws Exception {
            //获取图片本地路径
      FileInfoService fileInfoService = SpringContextUtils.getBean(FileInfoService.class);
            FileInfoEntity fileInfoEntity = fileInfoService.queryByUrlPath(imgUrl.trim());
            if (fileInfoEntity == null || !FileUtil.exist(fileInfoEntity.getRealPath())) {
                return;
            }
    
            //插入图片
      //获取图片对象
      BufferedImage img = ImageIO.read(new File(fileInfoEntity.getRealPath()));
            //获得图片的宽
      int width = img.getWidth();
            //获得图片的高
      int height = img.getHeight();
            if (width > 600) {
                //获取比例
      int rate = (width / 600 ) + 1;
                width = width / rate - 20;
                height = height / rate;
            }
            PictureRenderData pictureRenderData = new PictureRenderData(width, height, fileInfoEntity.getRealPath());
            PictureRenderPolicy.Helper.renderPicture(xwpfParagraph.createRun(), pictureRenderData);
        }
    
        /**
     * 转换表格为word内容
      * @author xuwangcheng
     * @date 2019/7/29 18:45
     * @param doc doc
     * @param ele ele
     */  private static void parseTableToWord(NiceXWPFDocument doc, Element ele) throws Exception {
            //简化表格html
      org.jsoup.nodes.Document tableDoc = Jsoup.parse(simplifyTable(ele.outerHtml()));
            Elements trList = tableDoc.getElementsByTag("tr");
            Elements tdList = trList.get(0).getElementsByTag("td");
    
            //创建表格
      XWPFTable xwpfTable = doc.createTable(trList.size(), tdList.size());
    
            //设置样式
      TableTools.widthTable(xwpfTable, MiniTableRenderData.WIDTH_A4_FULL, tdList.size());
            TableTools.borderTable(xwpfTable, 4);
    
            //写入表格行和列内容
      Map[][] array = new Map[trList.size()][tdList.size()];
            for (int row = 0; row < trList.size(); row++) {
                Element trElement = trList.get(row);
                Elements tds = trElement.getElementsByTag("td");
                for (int col = 0; col < tds.size(); col++) {
                    Element colElement = tds.get(col);
                    String colspan = colElement.attr("colspan");
                    String rowspan = colElement.attr("rowspan");
                    String style = colElement.attr("style");
                    StringBuilder styleSB = new StringBuilder();
                    if (!StringUtils.isEmpty(colspan)) {
                        int colCount = Integer.parseInt(colspan);
                        for (int i = 0; i < colCount - 1; i++) {
                            array[row][col + i + 1] = new HashMap();
                            array[row][col + i + 1].put("mergeCol", true);
                        }
                    }
                    if (!StringUtils.isEmpty(rowspan)) {
                        int rowCount = Integer.parseInt(rowspan);
                        for (int i = 0; i < rowCount - 1; i++) {
                            array[row + i + 1][col] = new HashMap();
                            array[row + i + 1][col].put("mergeRow", true);
                        }
                    }
                    XWPFTableCell tableCell = xwpfTable.getRow(row).getCell(col);
                    if (StringUtils.isEmpty(colspan)) {
                        if (col == 0) {
                            if (tableCell.getCTTc().getTcPr() == null) {
                                tableCell.getCTTc().addNewTcPr().addNewHMerge().setVal(STMerge.RESTART);
                            } else {
                                if (tableCell.getCTTc().getTcPr().getHMerge() == null) {
                                    tableCell.getCTTc().getTcPr().addNewHMerge().setVal(STMerge.RESTART);
                                } else {
                                    tableCell.getCTTc().getTcPr().getHMerge().setVal(STMerge.RESTART);
                                }
                            }
                        } else {
                            if (array[row][col] != null && array[row][col].get("mergeCol") != null && array[row][col].get("mergeCol")) {
                                if (tableCell.getCTTc().getTcPr() == null) {
                                    tableCell.getCTTc().addNewTcPr().addNewHMerge().setVal(STMerge.CONTINUE);
                                } else {
                                    if (tableCell.getCTTc().getTcPr().getHMerge() == null) {
                                        tableCell.getCTTc().getTcPr().addNewHMerge().setVal(STMerge.CONTINUE);
                                    } else {
                                        tableCell.getCTTc().getTcPr().getHMerge().setVal(STMerge.CONTINUE);
                                    }
                                }
                                continue;
                            } else {
                                if (tableCell.getCTTc().getTcPr() == null) {
                                    tableCell.getCTTc().addNewTcPr().addNewHMerge().setVal(STMerge.RESTART);
                                } else {
                                    if (tableCell.getCTTc().getTcPr().getHMerge() == null) {
                                        tableCell.getCTTc().getTcPr().addNewHMerge().setVal(STMerge.RESTART);
                                    } else {
                                        tableCell.getCTTc().getTcPr().getHMerge().setVal(STMerge.RESTART);
                                    }
                                }
                            }
                        }
                    } else {
                        if (tableCell.getCTTc().getTcPr() == null) {
                            tableCell.getCTTc().addNewTcPr().addNewHMerge().setVal(STMerge.RESTART);
                        } else {
                            if (tableCell.getCTTc().getTcPr().getHMerge() == null) {
                                tableCell.getCTTc().getTcPr().addNewHMerge().setVal(STMerge.RESTART);
                            } else {
                                tableCell.getCTTc().getTcPr().getHMerge().setVal(STMerge.RESTART);
                            }
                        }
                    }
                    if (StringUtils.isEmpty(rowspan)) {
                        if (array[row][col] != null && array[row][col].get("mergeRow") != null && array[row][col].get("mergeRow")) {
                            if (tableCell.getCTTc().getTcPr() == null) {
                                tableCell.getCTTc().addNewTcPr().addNewVMerge().setVal(STMerge.CONTINUE);
                            } else {
                                if (tableCell.getCTTc().getTcPr().getVMerge() == null) {
                                    tableCell.getCTTc().getTcPr().addNewVMerge().setVal(STMerge.CONTINUE);
                                } else {
                                    tableCell.getCTTc().getTcPr().getVMerge().setVal(STMerge.CONTINUE);
                                }
                            }
                            continue;
                        } else {
                            if (tableCell.getCTTc().getTcPr() == null) {
                                tableCell.getCTTc().addNewTcPr().addNewVMerge().setVal(STMerge.RESTART);
                            } else {
                                if (tableCell.getCTTc().getTcPr().getVMerge() == null) {
                                    tableCell.getCTTc().getTcPr().addNewVMerge().setVal(STMerge.RESTART);
                                } else {
                                    tableCell.getCTTc().getTcPr().getVMerge().setVal(STMerge.RESTART);
                                }
                            }
                        }
                    } else {
                        if (tableCell.getCTTc().getTcPr() == null) {
                            tableCell.getCTTc().addNewTcPr().addNewVMerge().setVal(STMerge.RESTART);
                        } else {
                            if (tableCell.getCTTc().getTcPr().getVMerge() == null) {
                                tableCell.getCTTc().getTcPr().addNewVMerge().setVal(STMerge.RESTART);
                            } else {
                                tableCell.getCTTc().getTcPr().getVMerge().setVal(STMerge.RESTART);
                            }
                        }
                    }
                    tableCell.removeParagraph(0);
                    XWPFParagraph paragraph = tableCell.addParagraph();
                    paragraph.setStyle(styleSB.toString());
                    if (!StringUtils.isEmpty(style) && style.contains("text-align:center")) {
                        paragraph.setAlignment(ParagraphAlignment.CENTER);
                    }
    
                    parseHtmlToWord(colElement, doc, paragraph, true);
    
    //                XWPFRun run = paragraph.createRun();
    //                run.setText(colElement.text());
      }
            }
        }
    
        /**
     * 简化html中的表格dom
     * @author xuwangcheng
     * @date 2019/7/29 18:39
     * @param tableContent tableContent
     * @return {@link String}
     */  private static String simplifyTable(String tableContent) {
            if (StringUtils.isEmpty(tableContent)) {
                return null;
            }
            org.jsoup.nodes.Document tableDoc = Jsoup.parse(tableContent);
            Elements trElements = tableDoc.getElementsByTag("tr");
            if (trElements != null) {
                Iterator eleIterator = trElements.iterator();
                Integer rowNum = 0;
                // 针对于colspan操作
      while (eleIterator.hasNext()) {
                    rowNum++;
                    Element trElement = eleIterator.next();
                    //去除所有样式
      trElement.removeAttr("class");
                    Elements tdElements = trElement.getElementsByTag("td");
                    List tdEleList = covertElements2List(tdElements);
                    for (int i = 0; i < tdEleList.size(); i++) {
                        Element curTdElement = tdEleList.get(i);
                        //去除所有样式
      curTdElement.removeAttr("class");
                        Element ele = curTdElement.clone();
                        String colspanValStr = curTdElement.attr("colspan");
                        if (!StringUtils.isEmpty(colspanValStr)) {
                            ele.removeAttr("colspan");
                            Integer colspanVal = Integer.parseInt(colspanValStr);
                            for (int k = 0; k < colspanVal - 1; k++) {
                                curTdElement.after(ele.outerHtml());
                            }
                        }
                    }
                }
                // 针对于rowspan操作
      List trEleList = covertElements2List(trElements);
                Element firstTrEle = trElements.first();
                Elements tdElements = firstTrEle.getElementsByTag("td");
                Integer tdCount = tdElements.size();
                //获取该列下所有单元格
      for (int i = 0; i < tdElements.size(); i++) {
                    for (Element trElement : trEleList) {
                        List tdElementList = covertElements2List(trElement.getElementsByTag("td"));
                        try {
                            tdElementList.get(i);
                        } catch (Exception e) {
                            continue;
                        }
                        Node curTdNode = tdElementList.get(i);
                        Node cNode = curTdNode.clone();
                        String rowspanValStr = curTdNode.attr("rowspan");
                        if (!StringUtils.isEmpty(rowspanValStr)) {
                            cNode.removeAttr("rowspan");
                            Element nextTrElement = trElement.nextElementSibling();
                            Integer rowspanVal = Integer.parseInt(rowspanValStr);
                            for (int j = 0; j < rowspanVal - 1; j++) {
                                Node tempNode = cNode.clone();
                                List nodeList = new ArrayList();
                                nodeList.add(tempNode);
                                if (j > 0) {
                                    nextTrElement = nextTrElement.nextElementSibling();
                                }
                                Integer indexNum = i + 1;
                                if (i == 0) {
                                    indexNum = 0;
                                }
                                if (indexNum.equals(tdCount)) {
                                    nextTrElement.appendChild(tempNode);
                                } else {
                                    nextTrElement.insertChildren(indexNum, nodeList);
                                }
                            }
                        }
                    }
                }
            }
            Element tableEle = tableDoc.getElementsByTag("table").first();
            String tableHtml = tableEle.outerHtml();
    
            return tableHtml;
        }
    
        /**
     * 转换Elements为list
     * @author xuwangcheng
     * @date 2019/7/29 18:40
     * @param curElements curElements
     * @return {@link List}
     */  private static List covertElements2List(Elements curElements){
            List elementList = new ArrayList();
            Iterator eleIterator = curElements.iterator();
            while(eleIterator.hasNext()){
                Element curlement = eleIterator.next();
                elementList.add(curlement);
            }
            return elementList;
        }
    
    
    以前,我总以为自己是菜鸟,也总想着早起的鸟儿有虫吃。直到有一天我想飞,才愕然的发现自己没有翅膀和羽毛,我竟然是菜虫!早起的虫儿被鸟吃……原来,百足应该厚积薄发!