diff --git a/laws-modules/src/main/java/com/jero/modules/split/controller/SarFileSplitInfoController.java b/laws-modules/src/main/java/com/jero/modules/split/controller/SarFileSplitInfoController.java index a617cf43..549c9458 100644 --- a/laws-modules/src/main/java/com/jero/modules/split/controller/SarFileSplitInfoController.java +++ b/laws-modules/src/main/java/com/jero/modules/split/controller/SarFileSplitInfoController.java @@ -161,6 +161,22 @@ public class SarFileSplitInfoController extends JeroController add1(@RequestBody SarFileSplitInfoEO sarFileSplitInfoEO) { + sarFileSplitInfoEO.setStartNumber(0); + sarFileSplitInfoEO.setStopNumber(0); + sarFileSplitInfoService.add1(sarFileSplitInfoEO); + return Result.OK("添加成功！"); + } + /** * 更改权限 * diff --git a/laws-modules/src/main/java/com/jero/modules/split/service/ISarFileSplitInfoService.java b/laws-modules/src/main/java/com/jero/modules/split/service/ISarFileSplitInfoService.java index 9b059876..bab0262d 100644 --- a/laws-modules/src/main/java/com/jero/modules/split/service/ISarFileSplitInfoService.java +++ b/laws-modules/src/main/java/com/jero/modules/split/service/ISarFileSplitInfoService.java @@ -26,7 +26,9 @@ public interface ISarFileSplitInfoService extends IService { */ void add(SarFileSplitInfoEO sarFileSplitInfoEO); - /** + void add1(SarFileSplitInfoEO sarFileSplitInfoEO); + + /** * 更新 * * @param sarFileSplitInfoEO diff --git a/laws-modules/src/main/java/com/jero/modules/split/service/impl/FileSpiltService.java b/laws-modules/src/main/java/com/jero/modules/split/service/impl/FileSpiltService.java index de204861..ad1f4d1e 100644 --- a/laws-modules/src/main/java/com/jero/modules/split/service/impl/FileSpiltService.java +++ b/laws-modules/src/main/java/com/jero/modules/split/service/impl/FileSpiltService.java @@ -82,7 +82,353 @@ public class FileSpiltService { @Value("${jero.splitUrl}") private String splitUrl; + + /** + * 中国标准数据拆分 + * + * @param sarFileSplitInfoEO + * @param enumByValue + * @return + */ + public int fileCHN1(SarFileSplitInfoEO sarFileSplitInfoEO, SplitFileTypeTypeEnum enumByValue) { + if (sarFileSplitInfoEO.getStopNumber() <= 0) { + sarFileSplitInfoEO.setStopNumber(200); + } + if (sarFileSplitInfoEO.getStartNumber() <= 0) { + sarFileSplitInfoEO.setStartNumber(1); + } + + // 拆分文件条款表中最终插入数据 + List messageList = new ArrayList<>(); +// List messageList = new ArrayList<>(); + int messageNumber = 0; + // 拆分文件目录表中最终插入数据 + List treeList = new ArrayList<>(); + // 拆分文件条款详情表中最终插入数据 + List itemValList = new ArrayList<>(); + Long treeListDisplay = 1L; + + String generalCatalogueId = UUIDUtils.randomUUID20(); + SarFileSplitMenuEO documentTreeEO = new SarFileSplitMenuEO(); +// SarFileSplitMenuEO documentTreeEO = new SarFileSplitMenuEO(generalCatalogueId,"总目录","",sarFileSplitInfoEO.getId(),treeListDisplay++,""); + documentTreeEO.setId(generalCatalogueId); + documentTreeEO.setName("总目录"); + documentTreeEO.setInfoId(sarFileSplitInfoEO.getId()); + documentTreeEO.setDisplaySeq(1L); + documentTreeEO.setDisplaySeq(treeListDisplay++); + documentTreeEO.setValidFlag(0); + + + treeList.add(documentTreeEO); + + List nowStart = new ArrayList<>(); +// if (sarFileSplitInfoEO.getStartNumber()<4){ + //从第二个开始拆解 + nowStart.add("2"); +// } else { +// nowStart.add(String.valueOf(sarFileSplitInfoEO.getStartNumber())); +// } +//// 获取文件地址 +// String fileId = sarFileSplitInfoEO.getFileId(); +// LambdaQueryWrapper queryWrapper = new LambdaQueryWrapper(); +// queryWrapper.eq(OSSFile::getId, fileId); +// queryWrapper.orderByDesc(OSSFile::getCreateTime); +// List ossFileList = iossFileService.list(queryWrapper); +// if (CollectionUtils.isEmpty(ossFileList)) { +// throw new JeroBootException("当前文件未找到，请重新选择！"); +// } +// OSSFile ossFile = ossFileList.get(0); +// String readFilePath = ossFile.getUrl(); +// if (StringUtils.isEmpty(readFilePath) || !MinioUtil.doesObjectExist(readFilePath)) { +// throw new JeroBootException("当前文件未找到，请重新选择！"); +// } +// try (InputStream is = MinioUtil.download(readFilePath);InputStream is1 = MinioUtil.download(readFilePath)) { + String readFilePath = "E:\\Data\\Downloads\\wordtest\\前言.docx"; + try (InputStream is = new FileInputStream(readFilePath); InputStream is1 = new FileInputStream(readFilePath)) {//需要将文件路更改为word文档所在路径。 + ZipSecureFile.setMinInflateRatio(-1.0d); + XWPFDocument doc = new XWPFDocument(is); + + TitleNumberVO titleNumberVO = AsposeUtil.getNumber(is1); + //TitleNumberVO titleNumberVO = new TitleNumberVO(); + Map titleNumberMap = titleNumberVO.getTitleNumberMap(); + List emptyTitleList = titleNumberVO.getEmptyTitleList(); + + Pattern ptest = Pattern.compile("^1范围"); + Matcher matcher; + + // 记录拆分后的每一段文字 + List elements = doc.getBodyElements(); + // 记录word自动生成的编号数据 + Map numberMap = new HashMap(); + boolean menuCanAdd = false; // 判断当前目录下是否可以正常填写数据 + boolean firstFlag = true; + List startDigitList = getStartDigit(elements, enumByValue); + String regex = "^[0-9]*$"; + String preName = ""; + int preNameNum = 0; +// for (IBodyElement element : elements) { + for (int g=0; g< elements.size(); g++) { + IBodyElement element = elements.get(g); + + // 段落 + if (element instanceof XWPFParagraph) { + XWPFParagraph p = (XWPFParagraph) element; + StringBuilder stringBuilder = new StringBuilder(); + + ParagraphChildOrderManager.parseParagraph(p); + ParagraphChildOrderManager runOrMaths = new ParagraphChildOrderManager(p); + List childList = runOrMaths.getChildList(); + + for (Object child : childList) { + if (child instanceof CTOMath) { + // 处理word中存在的公式成mathML格式 + handleParagraphOMath(stringBuilder, (CTOMath) child); + if (messageList.size() > 0) { + addItermsConditionsMath(messageList.get(messageList.size() - 1), itemValList, stringBuilder.toString()); + } + } else if (child instanceof CTOMathPara) { + //处理word中存在的公式 + handleParagraphOMath(stringBuilder, (CTOMathPara) child); + if (messageList.size() > 0) { + addItermsConditionsMath(messageList.get(messageList.size() - 1), itemValList, stringBuilder.toString()); + } + } + } + + for (XWPFRun xwrun : p.getRuns()) { + VerticalAlign subscript = xwrun.getSubscript(); + String smalltext = xwrun.getText(0); + if (smalltext==null){ + continue; + } + switch (subscript) { + case BASELINE: + stringBuilder.append(smalltext); + break; + case SUBSCRIPT: + stringBuilder.append("").append(smalltext).append(""); + break; + case SUPERSCRIPT: + stringBuilder.append("").append(smalltext).append(""); + break; + default: + stringBuilder.append(smalltext); + } + } + // 处理段落生成编号不识别问题 + String paragraphString = stringBuilder.toString(); + if (StringUtils.isNotBlank(paragraphString)) { + paragraphString = paragraphString.replace((char) 12288, ' '); + paragraphString = paragraphString.trim(); + } + if(StringUtils.isEmpty(paragraphString)){ + // 处理当前段落仅为图片的情况 + List imageBundleList = readImageInParagraph((XWPFParagraph) p); + if (CollectionUtils.isNotEmpty(imageBundleList)) { + if (messageList.size() > 0) { + addItermsConditionsImage(messageList.get(messageList.size() - 1), itemValList, p, imageBundleList); + } + continue; + } + + } + paragraphString = paragraphString.replaceAll("\r\n",""); + paragraphString = paragraphString.replaceAll("\n",""); + paragraphString = paragraphString.trim(); + + if ("前言".equals(getChinese(paragraphString)) + && (paragraphString + .replaceAll("\t", "") + .replaceAll("\b", "") + .replaceAll(" ", "") + .length() == 2)) { // 前言后面有文本，判定为是目录中的前言 + firstFlag = false; + } + if(firstFlag) { + continue; + } + + if(StringUtils.isEmpty(paragraphString)){ + // 处理当前段落仅为图片的情况 + List imageBundleList = readImageInParagraph((XWPFParagraph) p); + if (CollectionUtils.isNotEmpty(imageBundleList)) { + if (messageList.size() > 0) { + addItermsConditionsImage(messageList.get(messageList.size() - 1), itemValList, p, imageBundleList); + } + continue; + } + } + if (StringUtils.isNotEmpty(p.getNumLevelText()) && p.getNumLevelText().indexOf(".") > 0) { + String numberParent = p.getNumLevelText().substring(0, p.getNumLevelText().lastIndexOf(".")); + if (numberMap.containsKey(numberParent)) { + Integer numbernow = (Integer) numberMap.get(numberParent) + 1; + if (messageList != null && !messageList.isEmpty() && messageList.get(messageList.size() - 1).getItemNum().equals(numberParent + "." + String.valueOf(numbernow))) { + numberMap.put(numberParent, numbernow + 1); + paragraphString = numberParent + "." + String.valueOf(numbernow + 1) + paragraphString; + } else { + numberMap.put(numberParent, numbernow); + paragraphString = numberParent + "." + String.valueOf(numbernow) + paragraphString; + } + } else { + numberMap.put(numberParent, 1); + paragraphString = numberParent + ".1" + paragraphString; + } + } + if (messageNumber == 0) { + ptest = Pattern.compile("^1\\s{0,50}范围"); + matcher = ptest.matcher(paragraphString); + ptest = Pattern.compile("^\\s{0,50}范围"); + Matcher matcher1 = ptest.matcher(paragraphString);; + while (matcher.find() || matcher1.find()) { + List clauseContent = new ArrayList<>(); + SarFileSplitItemsEO message = getSarFileSplitItemsEO(sarFileSplitInfoEO.getId(), "1", "范围", clauseContent); + SarFileSplitMenuEO documentTreeEO1 = new SarFileSplitMenuEO(); +// SarFileSplitMenuEO documentTreeEO1 = new SarFileSplitMenuEO(UUIDUtils.randomUUID20(),message.getItemsNum(),generalCatalogueId,sarFileSplitInfoEO.getId(),treeListDisplay++,"范围"); + documentTreeEO1.setId(UUIDUtils.randomUUID20()); + documentTreeEO1.setItemName("范围"); + documentTreeEO1.setPId(generalCatalogueId); + documentTreeEO1.setInfoId(sarFileSplitInfoEO.getId()); + documentTreeEO1.setDisplaySeq(treeListDisplay++); + documentTreeEO1.setName(message.getItemNum()); + documentTreeEO1.setValidFlag(0); +// treeList.add(documentTreeEO1); + + message.setMenuId(documentTreeEO1.getId()); + if (sarFileSplitInfoEO.getStartNumber() <= 1 && sarFileSplitInfoEO.getStopNumber() >= 1) { + messageList.add(message); + treeList.add(documentTreeEO1); + preName = getOneStartDigit(paragraphString, enumByValue);; + preNameNum = g; + menuCanAdd = true; + } else { + menuCanAdd = false; + } + messageNumber++; + } + } + else if (messageNumber >= 1) { + ptest = Pattern.compile("^[\\d.]*$"); + boolean hasGoIf = false; + String nextParagraphString = getNextParagraphString(g, elements, enumByValue); + + //替换序号 + boolean replaceNumberFlag = false; + String replaceNumber = ""; + if (!emptyTitleList.isEmpty() && StringUtils.isBlank(paragraphString) && StringUtils.isNotBlank(nextParagraphString)){ + paragraphString = emptyTitleList.get(0); + emptyTitleList.remove(0); + }else if (StringUtils.isNotBlank(paragraphString) && titleNumberMap.containsKey(paragraphString)){ + replaceNumberFlag = true; + replaceNumber = titleNumberMap.get(paragraphString); + paragraphString = replaceNumber + paragraphString; + } + String startDigit; + if (replaceNumberFlag){ + startDigit = replaceNumber; + }else { + startDigit = getOneStartDigit(paragraphString, enumByValue); + } + for (int i = 0; i < nowStart.size(); i++) { +// if (paragraphString.startsWith(nowStart.get(i))) { + if (startDigit.equals(nowStart.get(i))) { + // 判断真的是标题还是干扰项 TODO + if (nowStart.get(i).matches(regex)) { + boolean result = fileSplitPdfService.judge(preName, preNameNum, startDigit, g, startDigitList, "."); + if (result) { + continue; + } + } + // 判断名称，如果大于10位，取十位， + String itemName = paragraphString.substring(nowStart.get(i).length()).length() > 10 ? paragraphString.substring(nowStart.get(i).length(), nowStart.get(i).length() + 10) : paragraphString.substring(nowStart.get(i).length()); + List clauseContent = new ArrayList<>(); + clauseContent.add(paragraphString); + SarFileSplitItemsEO message = getSarFileSplitItemsEO(sarFileSplitInfoEO.getId(), nowStart.get(i), itemName, clauseContent); + + String[] numberStr = nowStart.get(i).split("\\."); + if (Integer.valueOf(numberStr[0]) > sarFileSplitInfoEO.getStopNumber()) { + menuCanAdd = false; + } else { + // 增加一条数据，需要在数结构中加一条数据 + SarFileSplitMenuEO sarFileSplitMenuEO = getTreeList(nowStart.get(i), treeList, generalCatalogueId, sarFileSplitInfoEO, treeListDisplay++, itemName, enumByValue); + treeList.add(sarFileSplitMenuEO); + preName = startDigit; + preNameNum = g; + message.setMenuId(sarFileSplitMenuEO.getId()); + if(!StringUtils.isEmpty(itemName)){ + addItermsConditionsText(message, itemValList, paragraphString.substring(nowStart.get(i).length())); + } + messageList.add(message); + nowStart = getNewNowStart(nowStart.get(i)); + menuCanAdd = true; + } + hasGoIf = true; + break; + } + } + if (!hasGoIf && menuCanAdd) { + // 判断是否是附录 + ptest = Pattern.compile("^附录\\s{0,3}[A-Z]{1}"); + matcher = ptest.matcher(paragraphString); + // 获取当前最后一项num + String appendixNum = messageList.get(messageList.size() - 1).getItemNum().split("\\.")[0]; + if (matcher.find()&& !messageList.get(messageList.size() - 1).getItemNum().equals("附录")) { + if (Integer.valueOf(appendixNum) + 1 <= sarFileSplitInfoEO.getStopNumber()) { + String itemName = "附录"; + List clauseContent = new ArrayList<>(); + clauseContent.add(paragraphString); + SarFileSplitItemsEO message = getSarFileSplitItemsEO(sarFileSplitInfoEO.getId(), "附录", itemName, clauseContent); + // 增加一条数据，需要在数结构中加一条数据 + SarFileSplitMenuEO sarFileSplitMenuEO = getTreeList("附录", treeList, generalCatalogueId, sarFileSplitInfoEO, treeListDisplay++, "", enumByValue); + addItermsConditionsText(message, itemValList, paragraphString); + treeList.add(sarFileSplitMenuEO); + message.setMenuId(sarFileSplitMenuEO.getId()); + messageList.add(message); + nowStart = new ArrayList<>(); + } else { + break; + } + } else { + if (!paragraphString.equals("")) { + addItermsConditionsText(messageList.get(messageList.size() - 1), itemValList, paragraphString); + } + List imageBundleList = readImageInParagraph((XWPFParagraph) p); + if (CollectionUtils.isNotEmpty(imageBundleList)) { + addItermsConditionsImage(messageList.get(messageList.size() - 1), itemValList, p, imageBundleList); + } + } + } + // 判断是否直接跳出所有循环 + if (!menuCanAdd && messageList.size() > 0) { + break; + } + } + } + else if (element instanceof XWPFTable) { + if (!messageList.isEmpty()) { + addItermsConditionsOfTable(messageList.get(messageList.size() - 1), itemValList, (XWPFTable) element); + } + } + } + + } catch (Exception e) { + e.printStackTrace(); + return -1; + } + if (treeList.size() > 0 && messageList.size() > 0) { + sarFileSplitMenuEOMapper.insertForeach(treeList); + for (int i = 0; i < messageList.size(); i++) { + messageList.get(i).setZhan3Shi4Shun4Xu4(String.valueOf(i+1)); + messageList.get(i).setInfoId(sarFileSplitInfoEO.getId()); + } + iSarFileSplitItemsEOService.insertForeach(messageList); + iSarFileSplitItemsValEOService.insertForeach(itemValList); + } else { + return -1; + } + return treeList.size(); + + } /** * 中国标准数据拆分 * diff --git a/laws-modules/src/main/java/com/jero/modules/split/service/impl/SarFileSplitInfoServiceImpl.java b/laws-modules/src/main/java/com/jero/modules/split/service/impl/SarFileSplitInfoServiceImpl.java index 01b4f9e2..52c3a8db 100644 --- a/laws-modules/src/main/java/com/jero/modules/split/service/impl/SarFileSplitInfoServiceImpl.java +++ b/laws-modules/src/main/java/com/jero/modules/split/service/impl/SarFileSplitInfoServiceImpl.java @@ -139,6 +139,42 @@ public class SarFileSplitInfoServiceImpl extends ServiceImpl