|
@@ -34,7 +34,7 @@ public class BeiLunDifficultCaseDiscussHtmlAnalysis implements BeiLunHtmlAnalysi
|
|
|
Document doc = Jsoup.parse(html);
|
|
|
String htmlContent = BeiLunHtmlAnalysisUtil.blockDivToStr(doc.selectFirst("body").child(0), true);
|
|
|
if (StringUtil.isNotBlank(htmlContent)) {
|
|
|
- htmlContent = htmlContent.replaceAll("[ ]", " ");
|
|
|
+ htmlContent = htmlContent.replaceAll("[ ]", " ").replace("第一页","");
|
|
|
CommonAnalysisUtil.html2StructureMap(titles, htmlContent, structureMap);
|
|
|
}
|
|
|
CommonAnalysisUtil.extractDateByTitle(structureMap, "告知时间");
|
|
@@ -66,11 +66,13 @@ public class BeiLunDifficultCaseDiscussHtmlAnalysis implements BeiLunHtmlAnalysi
|
|
|
}
|
|
|
|
|
|
private String findLast(String value) {
|
|
|
- if (value.lastIndexOf(",") != value.length() - 1) {
|
|
|
- return value;
|
|
|
+ if (StringUtil.isNotBlank(value)) {
|
|
|
+ if (value.lastIndexOf(",") != value.length() - 1) {
|
|
|
+ return value;
|
|
|
+ }
|
|
|
+ value = value.substring(0, value.length() - 1);
|
|
|
+ findLast(value);
|
|
|
}
|
|
|
- value = value.substring(0, value.length() - 1);
|
|
|
- findLast(value);
|
|
|
return value;
|
|
|
}
|
|
|
}
|