改进原子切分 fix hankcs#1421

hankcs · hankcs · commit e8a920cd887d · 2020-02-14T15:53:10.000-05:00
diff --git a/src/main/java/com/hankcs/hanlp/seg/Segment.java b/src/main/java/com/hankcs/hanlp/seg/Segment.java
@@ -161,7 +161,7 @@ protected static List<AtomNode> quickAtomSegment(char[] charArray, int start, in
         int offsetAtom = start;
         int preType = CharType.get(charArray[offsetAtom]);
         int curType;
-        while (++offsetAtom < end)
+        while (++offsetAtom < end || (atomNodeList.isEmpty() && offsetAtom < charArray.length))
         {
             curType = CharType.get(charArray[offsetAtom]);
             if (curType != preType)
diff --git a/src/test/java/com/hankcs/hanlp/seg/SegmentTest.java b/src/test/java/com/hankcs/hanlp/seg/SegmentTest.java
@@ -536,4 +536,11 @@ public void testIssue1172()
         CustomDictionary.insert("我的额度", "xyz");
         System.out.println(HanLP.segment("我的额度不够，需要提高额度"));
     }
+
+    public void testIssue1421()
+    {
+        System.out.println(HanLP.segment("十一中国放假吗"));
+        System.out.println(HanLP.segment("十一发展计划"));
+        System.out.println(HanLP.segment("十一通过山海关吗"));
+    }
 }

Original file line number	Diff line number	Diff line change
`@@ -161,7 +161,7 @@ protected static List<AtomNode> quickAtomSegment(char[] charArray, int start, in`
`161`	`161`	`int offsetAtom = start;`
`162`	`162`	`int preType = CharType.get(charArray[offsetAtom]);`
`163`	`163`	`int curType;`
`164`		`- while (++offsetAtom < end)`
	`164`	`+ while (++offsetAtom < end \|\| (atomNodeList.isEmpty() && offsetAtom < charArray.length))`
`165`	`165`	`{`
`166`	`166`	`curType = CharType.get(charArray[offsetAtom]);`
`167`	`167`	`if (curType != preType)`
Original file line number	Diff line number	Diff line change
`@@ -536,4 +536,11 @@ public void testIssue1172()`
`536`	`536`	`CustomDictionary.insert("我的额度", "xyz");`
`537`	`537`	`System.out.println(HanLP.segment("我的额度不够，需要提高额度"));`
`538`	`538`	`}`
	`539`	`+`
	`540`	`+ public void testIssue1421()`
	`541`	`+ {`
	`542`	`+ System.out.println(HanLP.segment("十一中国放假吗"));`
	`543`	`+ System.out.println(HanLP.segment("十一发展计划"));`
	`544`	`+ System.out.println(HanLP.segment("十一通过山海关吗"));`
	`545`	`+ }`
`539`	`546`	`}`