文本分类数据集加载显示进度

hankcs · Dec 16, 2017 · 28a790c · 28a790c
1 parent b8a65fe
commit 28a790c
Showing 1 changed file with 5 additions and 2 deletions.
diff --git a/src/main/java/com/hankcs/hanlp/classification/corpus/AbstractDataSet.java b/src/main/java/com/hankcs/hanlp/classification/corpus/AbstractDataSet.java
@@ -135,11 +135,14 @@ public IDataSet load(String folderPath, String charsetName, double percentage) t
                 e = files.length;
             }
 
+            int logEvery = (int) Math.ceil((e - b) / 10000f);
             for (int i = b; i < e; i++)
             {
                 add(folder.getName(), TextProcessUtility.readTxt(files[i], charsetName));
-                if (i % 100 == 0)
-                    logger.out("%.2f%%...", MathUtility.percentage(i - b, e - b));
+                if (i % logEvery == 0)
+                {
+                    logger.out("%c[%s]...%.2f%%", 13, category, MathUtility.percentage(i - b + 1, e - b));
+                }
             }
             logger.out(" %d 篇文档\n", e - b);
         }