Merge pull request #322 from ncats/html_truncation_fix_mm

Fix and optimize HTML string truncation
ncats · Mar 19, 2024 · ff6a675 · ff6a675
2 parents 18f2b20 + 2fc8b36
commit ff6a675
Showing 1 changed file with 15 additions and 12 deletions.
diff --git a/gsrs-module-substances-core/src/main/java/gsrs/module/substance/utils/HtmlUtil.java b/gsrs-module-substances-core/src/main/java/gsrs/module/substance/utils/HtmlUtil.java
@@ -54,15 +54,20 @@ public void head(Node node, int depth) {
                         String curText = ((TextNode) node).getWholeText();
                         if (resHtmlLen + nodeHtmlLen > maxLen) {
                             StringBuilder sb = new StringBuilder(curText);
-                            int curHtmlLen = maxNodeLen + 1;
+                            int curHtmlLen = maxNodeLen;
                             sb.setLength(curHtmlLen);
-                            sb.setLength(sb.length() - (Long.valueOf(sb.chars().filter(c -> c == '&').count()).intValue() * 4));
-                            sb.setLength(sb.length() - (Long.valueOf(sb.chars().filter(c -> (c == '<' || c == '>')).count()).intValue() * 3));
+                            curHtmlLen += Long.valueOf(sb.chars().filter(c -> c == '&').count()).intValue() * 4;
+                            curHtmlLen += Long.valueOf(sb.chars().filter(c -> (c == '<' || c == '>')).count()).intValue() * 3;
                             while (curHtmlLen > maxNodeLen) {
+                                char lastChar = sb.charAt(sb.length() - 1);
+                                if (lastChar == '&') {
+                                    curHtmlLen = curHtmlLen - 5;
+                                } else if (lastChar == '<' || lastChar == '>') {
+                                    curHtmlLen = curHtmlLen - 4;
+                                } else {
+                                    curHtmlLen = curHtmlLen - 1;
+                                }
                                 sb.setLength(sb.length() - 1);
-                                curHtmlLen = sb.toString().getBytes(StandardCharsets.UTF_8).length;
-                                curHtmlLen += Long.valueOf(sb.chars().filter(c -> c == '&').count()).intValue() * 4;
-                                curHtmlLen += Long.valueOf(sb.chars().filter(c -> (c == '<' || c == '>')).count()).intValue() * 3;
                             }
                             cur.appendText(sb.toString());
                             throw new IllegalStateException();
@@ -84,6 +89,9 @@ public void tail(Node node, int depth) {
     public static String truncate(String s, int len){
         Document srcDoc = Parser.parseBodyFragment(s, "");
         srcDoc.outputSettings().prettyPrint(false);
+        if (srcDoc.body().html().length() <= len) {
+            return srcDoc.body().html();
+        }
 
         int maxLength = len-3; //for final ...
 
@@ -98,12 +106,7 @@ public static String truncate(String s, int len){
             t.traverse(v, srcDoc.body());
         } catch (IllegalStateException ex) {}
 
-        String htmlReturn = dst.html();
-        if(htmlReturn.length() < s.length() && htmlReturn.length() <= maxLength){
-            return htmlReturn + "...";
-        }else{
-            return htmlReturn;
-        }
+        return dst.html() + "...";
     }
 
     public static String clean(String content, String charset) {