一.背景

項(xiàng)目中有個(gè)需求大體意思是，上傳一個(gè)word模板，根據(jù)word模板合成word文件，再將word文件轉(zhuǎn)為pdf。

二.方案選擇

Spire.Doc for Java方案
Spire.Doc for Java這個(gè)是商用收費(fèi)的，不過API文檔豐富且集成簡單，免費(fèi)版僅支持3頁轉(zhuǎn)換。類似的還有ITEXT，這個(gè)商用也是受限制的。

docx4j方案
開源可商用，僅支持docx格式的word。

jodconverter+LibreOffice 方案
開源可商用，調(diào)用本地office服務(wù)，進(jìn)行pdf轉(zhuǎn)換，類似的還有jodconverter+openOffice。

其他

至于其他的由于不支持跨平臺(tái)不做考慮。

三.實(shí)操

1.docx4j

首先嘗試了docx4j，因?yàn)閐ocx4j本身支持模板替換的操作，可一次性做替換及文檔類型轉(zhuǎn)換，而且僅支持docx類型，對于本次需求問題不大。

依賴僅需要一個(gè)即可

<dependency><groupId>org.docx4j</groupId><artifactId>docx4j-export-fo</artifactId><version>6.1.0</version></dependency>

主要代碼

@Slf4j public class PdfUtil {public static <T> void exportByLocalPath(HttpServletResponse response, String fileName, String path, Map<String,String> params){try (InputStream in = PdfUtil.class.getClassLoader().getResourceAsStream(path)) {convertDocxToPdf(in, response,fileName,params);} catch (Exception e) {log.error("docx文檔轉(zhuǎn)換為PDF失敗", e.getMessage());}}/*** docx文檔轉(zhuǎn)換為PDF* @param in* @param response* @return*/public static void convertDocxToPdf(InputStream in, HttpServletResponse response, String fileName, Map<String,String> params) throws Exception {response.setContentType("application/pdf");String fullFileName = new String(fileName.getBytes(), StandardCharsets.ISO_8859_1);response.setHeader("Content-disposition", "attachment;filename=" + fullFileName + ".pdf");WordprocessingMLPackage wmlPackage = WordprocessingMLPackage.load(in);if (params!=null&&!params.isEmpty()) {MainDocumentPart documentPart = wmlPackage.getMainDocumentPart();cleanDocumentPart(documentPart);documentPart.variableReplace(params);}setFontMapper(wmlPackage);Docx4J.toPDF(wmlPackage,response.getOutputStream());}/*** 清除文檔空白占位符* @param documentPart* @return {@link boolean}*/public static boolean cleanDocumentPart(MainDocumentPart documentPart) throws Exception {if (documentPart == null) {return false;}Document document = documentPart.getContents();String wmlTemplate =XmlUtils.marshaltoString(document, true, false, Context.jc);document = (Document) XmlUtils.unwrap(DocxVariableClearUtil.doCleanDocumentPart(wmlTemplate, Context.jc));documentPart.setContents(document);return true;}/*** 設(shè)置字體樣式* @param mlPackage*/private static void setFontMapper(WordprocessingMLPackage mlPackage) throws Exception {Mapper fontMapper = new IdentityPlusMapper();fontMapper.put("隸書", PhysicalFonts.get("LiSu"));fontMapper.put("宋體", PhysicalFonts.get("SimSun"));fontMapper.put("微軟雅黑", PhysicalFonts.get("Microsoft Yahei"));fontMapper.put("黑體", PhysicalFonts.get("SimHei"));fontMapper.put("楷體", PhysicalFonts.get("KaiTi"));fontMapper.put("新宋體", PhysicalFonts.get("NSimSun"));fontMapper.put("華文行楷", PhysicalFonts.get("STXingkai"));fontMapper.put("華文仿宋", PhysicalFonts.get("STFangsong"));fontMapper.put("宋體擴(kuò)展", PhysicalFonts.get("simsun-extB"));fontMapper.put("仿宋", PhysicalFonts.get("FangSong"));fontMapper.put("仿宋_GB2312", PhysicalFonts.get("FangSong_GB2312"));fontMapper.put("幼圓", PhysicalFonts.get("YouYuan"));fontMapper.put("華文宋體", PhysicalFonts.get("STSong"));fontMapper.put("華文中宋", PhysicalFonts.get("STZhongsong"));mlPackage.setFontMapper(fontMapper);} }

清除工具類,用于處理占位符替換不生效的問題，這里參考文章

public class DocxVariableClearUtil {/*** 去任意XML標(biāo)簽*/private static final Pattern XML_PATTERN = Pattern.compile("<[^>]*>");private DocxVariableClearUtil() {}/*** start符號*/private static final char PREFIX = '$';/*** 中包含*/private static final char LEFT_BRACE = '{';/*** 結(jié)尾*/private static final char RIGHT_BRACE = '}';/*** 未開始*/private static final int NONE_START = -1;/*** 未開始*/private static final int NONE_START_INDEX = -1;/*** 開始*/private static final int PREFIX_STATUS = 1;/*** 左括號*/private static final int LEFT_BRACE_STATUS = 2;/*** 右括號*/private static final int RIGHT_BRACE_STATUS = 3;/*** doCleanDocumentPart** @param wmlTemplate* @param jc* @return* @throws JAXBException*/public static Object doCleanDocumentPart(String wmlTemplate, JAXBContext jc) throws JAXBException {// 進(jìn)入變量塊位置int curStatus = NONE_START;// 開始位置int keyStartIndex = NONE_START_INDEX;// 當(dāng)前位置int curIndex = 0;char[] textCharacters = wmlTemplate.toCharArray();StringBuilder documentBuilder = new StringBuilder(textCharacters.length);documentBuilder.append(textCharacters);// 新文檔StringBuilder newDocumentBuilder = new StringBuilder(textCharacters.length);// 最后一次寫位置int lastWriteIndex = 0;for (char c : textCharacters) {switch (c) {case PREFIX:// 不管其何狀態(tài)直接修改指針,這也意味著變量名稱里面不能有PREFIXkeyStartIndex = curIndex;curStatus = PREFIX_STATUS;break;case LEFT_BRACE:if (curStatus == PREFIX_STATUS) {curStatus = LEFT_BRACE_STATUS;}break;case RIGHT_BRACE:if (curStatus == LEFT_BRACE_STATUS) {// 接上之前的字符newDocumentBuilder.append(documentBuilder.substring(lastWriteIndex, keyStartIndex));// 結(jié)束位置int keyEndIndex = curIndex + 1;// 替換String rawKey = documentBuilder.substring(keyStartIndex, keyEndIndex);// 干掉多余標(biāo)簽String mappingKey = XML_PATTERN.matcher(rawKey).replaceAll("");if (!mappingKey.equals(rawKey)) {char[] rawKeyChars = rawKey.toCharArray();// 保留原格式StringBuilder rawStringBuilder = new StringBuilder(rawKey.length());// 去掉變量引用字符for (char rawChar : rawKeyChars) {if (rawChar == PREFIX || rawChar == LEFT_BRACE || rawChar == RIGHT_BRACE) {continue;}rawStringBuilder.append(rawChar);}// 要求變量連在一起String variable = mappingKey.substring(2, mappingKey.length() - 1);int variableStart = rawStringBuilder.indexOf(variable);if (variableStart > 0) {rawStringBuilder = rawStringBuilder.replace(variableStart, variableStart + variable.length(), mappingKey);}newDocumentBuilder.append(rawStringBuilder.toString());} else {newDocumentBuilder.append(mappingKey);}lastWriteIndex = keyEndIndex;curStatus = NONE_START;keyStartIndex = NONE_START_INDEX;}default:break;}curIndex++;}// 余部if (lastWriteIndex < documentBuilder.length()) {newDocumentBuilder.append(documentBuilder.substring(lastWriteIndex));}return XmlUtils.unmarshalString(newDocumentBuilder.toString(), jc);}}

2.poi-tl+jodconverter+LibreOffice 方案

poi-tl這個(gè)是專門用來進(jìn)行word模板合成的開源庫，文檔很詳細(xì)。
LibreOffice 下載最新的穩(wěn)定版本即可。

maven依賴

<!-- word合成 --><!-- 這里注意版本,1.5版本依賴的poi 3.x的版本 --><dependency><groupId>com.deepoove</groupId><artifactId>poi-tl</artifactId><version>1.5.1</version></dependency><!-- jodconverter word轉(zhuǎn)pdf --><!-- jodconverter-core這個(gè)依賴，理論上不用加的，jodconverter-local已經(jīng)依賴了，但測試的時(shí)候不添加依賴找不到 --><dependency><groupId>org.jodconverter</groupId><artifactId>jodconverter-core</artifactId><version>4.2.0</version></dependency><dependency><groupId>org.jodconverter</groupId><artifactId>jodconverter-local</artifactId><version>4.2.0</version></dependency><dependency><groupId>org.jodconverter</groupId><artifactId>jodconverter-spring-boot-starter</artifactId><version>4.2.0</version></dependency><!-- 工具類,非必須 --><dependency><groupId>cn.hutool</groupId><artifactId>hutool-all</artifactId><version>5.4.3</version></dependency>

主要代碼
JodConverterConfig配置類

@Configuration public class JodConverterConfig {@Autowiredprivate OfficeManager officeManager;@Beanpublic DocumentConverter documentConverter() {return LocalConverter.builder().officeManager(officeManager).build();} }

yml配置文件

jodconverter:local:enabled: trueoffice-home: "C:\\Program Files\\LibreOffice"

PdfService合成導(dǎo)出代碼

@Slf4j @Component public class PdfService {@Autowiredprivate DocumentConverter documentConverter;public void docxToPDF(InputStream inputStream,HttpServletResponse response,String fileName) {response.setContentType("application/pdf");try {String fullFileName = new String(fileName.getBytes(), StandardCharsets.ISO_8859_1);response.setHeader("Content-disposition","attachment;filename=\\"+fullFileName+".pdf\\");documentConverter.convert(inputStream).as(DefaultDocumentFormatRegistry.DOCX).to(response.getOutputStream()).as(DefaultDocumentFormatRegistry.PDF).execute();} catch (OfficeException |IOException e) {log.error("word轉(zhuǎn)pdf失敗:{}",e.getMessage());}}public void exportByLocalPath(HttpServletResponse response, String fileName, String path, Object params) throws Exception {BufferedOutputStream outputStream = null;BufferedInputStream wordInputStream = null;try (InputStream in = PdfService.class.getClassLoader().getResourceAsStream(path)) {// 生成臨時(shí)文件String outPutWordPath = System.getProperty("java.io.tmpdir").replaceAll(File.separator + "$", "") + fileName+".docx";File tempFile = FileUtil.touch(outPutWordPath);outputStream = FileUtil.getOutputStream(tempFile);// word模板合成寫到臨時(shí)文件WordUtil.replaceWord(outputStream, in, params);// word 轉(zhuǎn)pdfwordInputStream = FileUtil.getInputStream(tempFile);docxToPDF(wordInputStream, response,fileName);// 移除臨時(shí)文件FileUtil.del(tempFile);} catch (Exception e) {log.error("docx文檔轉(zhuǎn)換為PDF失敗", e.getMessage());} finally {IoUtil.close(outputStream);IoUtil.close(wordInputStream);}}

四.結(jié)論

1.docx4j方案

• 依賴少
• 同時(shí)支持word合成及格式轉(zhuǎn)換
• 轉(zhuǎn)化效率較差
• 對于含樣式及圖片轉(zhuǎn)換不友好，容易排版混亂

2.jodconverter+LibreOffice 方案

• 操作穩(wěn)定
• 轉(zhuǎn)換效率快
• 集成依賴設(shè)置較多
• 依賴本地服務(wù)
• LibreOffice打開word可能排版樣式錯(cuò)亂
  最后考慮項(xiàng)目需求，最終選擇了jodconverter+LibreOffice方案。

總結(jié)

以上是生活随笔為你收集整理的SpringBoot word文档转pdf的全部內(nèi)容，希望文章能夠幫你解決所遇到的問題。

如果覺得生活随笔網(wǎng)站內(nèi)容還不錯(cuò)，歡迎將生活随笔推薦給好友。