|
|
@@ -0,0 +1,119 @@
|
|
|
+package cn.cslg.pas.service.patentRightSplit;
|
|
|
+
|
|
|
+import cn.cslg.pas.common.model.params.PatentRightParams;
|
|
|
+import cn.cslg.pas.domain.PatentRight;
|
|
|
+import org.springframework.stereotype.Service;
|
|
|
+
|
|
|
+import java.util.ArrayList;
|
|
|
+import java.util.Arrays;
|
|
|
+import java.util.List;
|
|
|
+import java.util.regex.Matcher;
|
|
|
+import java.util.regex.Pattern;
|
|
|
+
|
|
|
+/**
|
|
|
+ * 拆分权要类（英文专利EN）
|
|
|
+ *
|
|
|
+ * @author chenyu
|
|
|
+ * @date 2023/8/30
|
|
|
+ */
|
|
|
+@Service
|
|
|
+public class PatentRightSplitENService implements PatentRightSplitService {
|
|
|
+
|
|
|
+ @Override
|
|
|
+ public List<PatentRight> formatPatentRight(PatentRightParams params) {
|
|
|
+ try {
|
|
|
+ //创建一个权要集合，用于返回结果
|
|
|
+ ArrayList<PatentRight> patentRights = new ArrayList<>();
|
|
|
+ //取出原文
|
|
|
+ String content = params.getContent();
|
|
|
+
|
|
|
+ //将原文从"1"截取到末尾（为了去除首行"What is claimed is:\n"）
|
|
|
+ if (!content.equals("") && content.contains("What is claimed is") && content.contains("1")) {
|
|
|
+ content = content.substring(content.indexOf("1"));
|
|
|
+ }
|
|
|
+
|
|
|
+ //第1步：拆分权要原文 ↓
|
|
|
+ //将原文和译文的所有换行符 "\r\n"和"\n" 替换成特殊符号 "##@@&&"，这样可以使得全文连在一起，其次再根据特殊符号##@@&&拆分权要
|
|
|
+ content = content.replace("\r\n", "##@@&&").replace("\n", "##@@&&");
|
|
|
+
|
|
|
+ //定义正则表达式
|
|
|
+ String regex = "##@@&&[0-9]+";
|
|
|
+ Matcher matcher = Pattern.compile(regex).matcher(content);
|
|
|
+ while (matcher.find()) {
|
|
|
+ String macherGroup = matcher.group();
|
|
|
+ content = content.replaceFirst(macherGroup, "\n" + macherGroup.substring(macherGroup.indexOf("##@@&&") + 6));
|
|
|
+ }
|
|
|
+ //去除所有特殊符号"##@@&&"
|
|
|
+ content = content.replace("##@@&&", "");
|
|
|
+ //最后按照换行符"\n"将权要拆成多个
|
|
|
+ String[] strs = content.split("\n");
|
|
|
+
|
|
|
+
|
|
|
+ //第2步：理出每个权要的类型（type为 1主权要还是 0附属权要）、理出权要之间的层级关系（每个权要的父级权要序号parentSort） ↓
|
|
|
+ String regex1 = "claim [0-9]+";
|
|
|
+ String regex2 = "claims [0-9]+ or [0-9]+";
|
|
|
+ String regex3 = "any preceding claim";
|
|
|
+ for (int i = 0; i < strs.length; i++) {
|
|
|
+ //创建权要对象装载当前权要
|
|
|
+ PatentRight patentRight = new PatentRight()
|
|
|
+ .setPatentId(params.getPatentId())
|
|
|
+ .setContent(strs[i])
|
|
|
+ .setSort(i);
|
|
|
+
|
|
|
+ //判断若该权要有逗号（即超过一句话）并且它第一句话中有"权利要求"4个字 或者 该权要没有逗号（即只有一句话），并且它有"权利要求"4个字，则该权要类型为附属权要
|
|
|
+ if ((strs[i].contains("，") && strs[i].substring(0, strs[i].indexOf("，")).contains("claim")) || (!strs[i].contains("，") && strs[i].contains("claim"))) {
|
|
|
+ //则该权要类型为附属权要
|
|
|
+ patentRight.setType(0);
|
|
|
+
|
|
|
+ Matcher matcher1 = Pattern.compile(regex1).matcher(strs[i]);
|
|
|
+ Matcher matcher2 = Pattern.compile(regex2).matcher(strs[i]);
|
|
|
+ Matcher matcher3 = Pattern.compile(regex3).matcher(strs[i]);
|
|
|
+ if (matcher2.find()) { //"claims 2 or 3"
|
|
|
+ String parentNum = matcher2.group().substring(matcher2.group().indexOf("claims") + 7); //"2 or 3"
|
|
|
+ String[] parentNumStrs = new String[0];
|
|
|
+ if (parentNum.contains(" or ")) {
|
|
|
+ parentNumStrs = parentNum.split(" or "); //[2, 3]
|
|
|
+ }
|
|
|
+
|
|
|
+ int[] parentSorts = new int[parentNumStrs.length]; //[1, 2]
|
|
|
+ for (int i1 = 0; i1 < parentSorts.length; i1++) {
|
|
|
+ parentSorts[i1] = Integer.parseInt(parentNumStrs[i1]) - 1;
|
|
|
+ }
|
|
|
+
|
|
|
+ StringBuilder builder = new StringBuilder();
|
|
|
+ for (int parentSort : parentSorts) {
|
|
|
+ builder.append(parentSort).append(",");
|
|
|
+ }
|
|
|
+ patentRight.setParentSort(builder.substring(0, builder.lastIndexOf(",")));
|
|
|
+
|
|
|
+ } else if (matcher1.find()) {
|
|
|
+ String parentNum = matcher1.group().substring(matcher1.group().indexOf("claim") + 6);
|
|
|
+ patentRight.setParentSort((Integer.parseInt(parentNum) - 1) + "");
|
|
|
+ } else if (matcher3.find()) {
|
|
|
+ StringBuilder builder = new StringBuilder();
|
|
|
+ for (int sort = 0; sort < i; sort++) {
|
|
|
+ builder.append(sort).append(",");
|
|
|
+ }
|
|
|
+ patentRight.setParentSort(builder.substring(0, builder.lastIndexOf(",")));
|
|
|
+ }
|
|
|
+ } else {
|
|
|
+ //否则该权要为主权要
|
|
|
+ patentRight
|
|
|
+ .setType(1)
|
|
|
+ .setParentSort("-1");
|
|
|
+ }
|
|
|
+
|
|
|
+ patentRights.add(patentRight);
|
|
|
+ }
|
|
|
+
|
|
|
+ return patentRights;
|
|
|
+ } catch (Exception e) {
|
|
|
+ e.printStackTrace();
|
|
|
+ ArrayList<PatentRight> patentRights = new ArrayList<>();
|
|
|
+ patentRights.add(new PatentRight().setContent(params.getContent()));
|
|
|
+ return patentRights;
|
|
|
+ }
|
|
|
+
|
|
|
+ }
|
|
|
+
|
|
|
+}
|
chendayu