流量选品-ES索引自定义分词器

Pyspark_job/utils/es_util.py

@@ -50,6 +50,35 @@ class EsUtils(object):
                 "number_of_shards": "3",
                 "number_of_replicas": "1",
                 "analysis": {
+                    "filter": {
+                        "en_snowball": {
+                            "type": "snowball",
+                            "language": "English"
+                        },
+                        "en_synonym": {
+                            "type": "synonym_graph",
+                            "synonyms_path": "analysis/synonyms_en.txt",
+                            "updateable": "true"
+                        }
+                    },
+                    "analyzer": {
+                        "en_analyzer": {
+                            "type": "custom",
+                            "tokenizer": "standard",
+                            "filter": [
+                                "lowercase",
+                                "en_snowball"
+                            ]
+                        },
+                        "en_search_analyzer": {
+                            "tokenizer": "standard",
+                            "filter": [
+                                "lowercase",
+                                "en_synonym",
+                                "en_snowball"
+                            ]
+                        }
+                    },
                     "normalizer": {
                         "lowercase_normalizer": {
                             "type": "custom",
@@ -71,7 +100,8 @@ class EsUtils(object):
                     },
                     "title": {
                         "type": "text",
-                        "analyzer": "standard",
+                        "analyzer": "en_analyzer",
+                        "search_analyzer": "en_search_analyzer",
                         "fields": {
                             "keyword": {
                                 "ignore_above": 32766,