merge tag

suggest-task/src/main/java/com/secoo/so/suggest/task/SuggestTask.java

@@ -243,7 +243,7 @@ public class SuggestTask {
                 esSuggestKeywordMap.put(fillKeyword, buildDefaultEsSuggestKeywordInfo(fillKeyword));
             }
         }
-
+        mergeKeywordTag(esSuggestKeywordMap);   // 处理部分keyword，合并为其他词的tag
         // 过滤词
         List<EsSuggestKeywordInfo> suggestKeywordInfoList = new ArrayList<>();
         int processCount = 0;
@@ -277,6 +277,73 @@ public class SuggestTask {
             saveSuggestKeywordToEs(suggestKeywordInfoList);
         }
     }
+    private static void mergeKeywordTag(ConcurrentHashMap<String, EsSuggestKeywordInfo> esSuggestKeywordMap) {
+        int maxCount = 100;
+        List<String> keyList = new ArrayList<>(esSuggestKeywordMap.keySet());
+        Collections.sort(keyList);  // 按照字符排序，确保扩展词都在本词后面
+        for(int i = 0;i< keyList.size(); i++) {
+            String word = keyList.get(i);
+            EsSuggestKeywordInfo suggestInfo = esSuggestKeywordMap.get(word);
+            int wordLen = StringUtils.getByteLength(word);
+            int length = word.length();
+            if (wordLen > 3 && wordLen <= 15 && StringUtils.isBlank(suggestInfo.getSuggestTags())) {
+                boolean isEN = StringUtils.isEnAndMidSpaceStr(word);
+                List<EsSuggestKeywordInfo> suggestList = new ArrayList<>();
+                int keyCount = 0;
+                for (int j = i + 1; j < keyList.size(); j++) {
+                    if (keyCount > maxCount) {
+                        break;
+                    }
+                    String fulWord = keyList.get(j);
+                    int fulLen = StringUtils.getByteLength(fulWord);
+                    if (fulWord.startsWith(word) && fulLen - wordLen > 3 && fulLen - wordLen <= 12) {
+                        String subWord = fulWord.substring(length, length + 1);
+                        String rightWord = fulWord.substring(length, fulWord.length()).trim();
+                        if (isEN && StringUtils.isEnStr(subWord)){
+                            continue;
+                        }
+                        int rightLen = StringUtils.getByteLength(rightWord);
+                        if (rightLen - wordLen <= 3 || rightLen - wordLen > 12) {
+                            break;
+                        }
+                        suggestList.add(esSuggestKeywordMap.get(fulWord));
+                        keyCount++;
+                    } else {
+                        break;
+                    }
+                }
+                if (suggestList.size() > 0) {
+                    Collections.sort(suggestList, (t1, t2) -> {
+                        Double score1 = t1.getWordABRank();
+                        Double score2 = t2.getWordABRank();
+                        if (score1 != null && score2 != null) {
+                            return score2.compareTo(score1);
+                        } else if (score1 == null && score2 == null) {
+                            return 0;
+                        } else if (score1 == null) {
+                            return 1;
+                        } else {
+                            return -1;
+                        }
+                    });
+                    StringBuffer sb = new StringBuffer("");
+                    for (int k=0, count=0; k < suggestList.size() && count < 3; k++, count++) {
+                        EsSuggestKeywordInfo info = suggestList.get(k);
+                        String fulWord = info.getKeyword();
+                        int fulLen = fulWord.length();
+                        String subWord = fulWord.substring(length, fulLen).trim();
+                        if (k > 0) {
+                            sb.append(",");
+                        }
+                        sb.append(subWord);
+                    }
+                    if (sb.length() > 0) {
+                        esSuggestKeywordMap.get(word).setSuggestTags(sb.toString());
+                    }
+                }
+            }
+        }
+    }
 
     private static EsSuggestKeywordInfo buildDefaultEsSuggestKeywordInfo(String keyword) {
         EsSuggestKeywordInfo esSuggestKeywordInfo = new EsSuggestKeywordInfo();

suggest-task/src/main/java/com/secoo/so/suggest/util/StringUtils.java

@@ -2067,6 +2067,17 @@ public abstract class StringUtils {
     }
 
     /**
+     * 判断是否是英文字符串,两边有空格认为不是英文，包含在中间的空格认为是英文
+     */
+    public static boolean isEnAndMidSpaceStr(String word) {
+        if (word.startsWith(" ") || word.endsWith(" ")) {
+            return false;
+        }
+        boolean result = word.trim().matches("[a-zA-Z\\s]+");
+        return result;
+    }
+
+    /**
      * 判断是否包含中文
      */
     public static boolean isContainChStr(String word) {
@@ -2109,6 +2120,19 @@ public abstract class StringUtils {
         return 0;
     }
 
+    public static void main(String[] arg) {
+        String word = "ab c ";
+        String word1 = "ab c 中文";
+        int wordLen = StringUtils.getByteLength(word);
+        int len = StringUtils.getByteLength(word1);
+        String aaa = word1.substring(wordLen, wordLen+1);
+        String bbb = word1.substring(wordLen, word1.length());
+        System.out.println(isEnAndMidSpaceStr(word));
+        System.out.println(isEnAndMidSpaceStr(bbb));
+        System.out.println(aaa);
+        System.out.println(bbb);
+    }
+
     /**
      * 32位md5加密
      */
@@ -2153,7 +2177,7 @@ public abstract class StringUtils {
      *
      * <p>If the stripChars String is {@code null}, whitespace is
      * stripped as defined by {@link Character#isWhitespace(char)}.
-     * Alternatively use {@link #strip(String)}.</p>
+     * Alternatively use {@link #strip(String, String)}.</p>
      *
      * <pre>
      * StringUtils.strip(null, *)          = null