Elasticsearch搜索查询时忽略存储数据特殊字符的实现咨询

Got it, let's figure out how to make Elasticsearch ignore special characters (like apostrophes in your "O'Ryan" example) when searching across person_name, name, and office.name. Here's the step-by-step solution that'll get this working for you:

To fix this, we need Elasticsearch to automatically strip or ignore special characters both when indexing your data and when processing search queries. This way, whether a user searches for O'Ryan or ORyan, they'll get the matching documents.

Since you're using a multi_match query for full-text text fields, the most reliable approach is to create a custom analyzer that cleans up special characters during indexing.

Step 1: Create an index with the custom analyzer #

First, delete your existing index (if it exists) and create a new one with the analyzer and field mappings configured:

PUT /your_target_index
{
  "settings": {
    "analysis": {
      "char_filter": {
        // Character filter to remove all non-alphanumeric/space characters
        "strip_special_chars": {
          "type": "pattern_replace",
          "pattern": "[^a-zA-Z0-9\\s]",
          "replacement": ""
        }
      },
      "analyzer": {
        // Custom analyzer: strip special chars → standard tokenization → lowercase
        "clean_text_analyzer": {
          "tokenizer": "standard",
          "char_filter": ["strip_special_chars"],
          "filter": ["lowercase"]
        }
      }
    }
  },
  "mappings": {
    "properties": {
      "person_name": {
        "type": "text",
        "analyzer": "clean_text_analyzer",
        "search_analyzer": "clean_text_analyzer" // Ensure queries use the same analyzer
      },
      "name": {
        "type": "text",
        "analyzer": "clean_text_analyzer",
        "search_analyzer": "clean_text_analyzer"
      },
      "office": {
        "properties": {
          "name": {
            "type": "text",
            "analyzer": "clean_text_analyzer",
            "search_analyzer": "clean_text_analyzer"
          }
        }
      }
    }
  }
}

Step 2: Reindex your existing data #

Analyzers only apply to newly indexed data, so you'll need to move your existing data into the new index using Elasticsearch's _reindex API:

POST _reindex
{
  "source": {
    "index": "your_old_index" // Replace with your original index name
  },
  "dest": {
    "index": "your_target_index"
  }
}

Step 3: Test your query #

Now run your original multi_match query—whether you search for O'Ryan or ORyan, it will match the correct documents:

{
  "query": {
    "multi_match" : {
      "query": "O'Ryan",
      "type": "best_fields",
      "fields": [ "person_name", "name", "office.name" ],
      "operator":"and"
    }
  }
}

If you need to do exact matches (like aggregations) on these fields while still ignoring special characters, add a keyword subfield with a normalizer:

Add this normalizer to your index settings:

"normalizer": {
  "clean_keyword_normalizer": {
    "type": "custom",
    "char_filter": ["strip_special_chars"],
    "filter": ["lowercase"]
  }
}

Then update the field mapping to include the keyword subfield:

"person_name": {
  "type": "text",
  "analyzer": "clean_text_analyzer",
  "fields": {
    "keyword": {
      "type": "keyword",
      "normalizer": "clean_keyword_normalizer"
    }
  }
}

This lets you use person_name for full-text searches and person_name.keyword for special-character-ignored exact matches.

If you can't reindex data right away, you can manually strip special characters from your query string. However, this is less reliable (it won't cover all edge cases):

{
  "query": {
    "multi_match" : {
      "query": "ORyan", // Use a pre-processed string without special chars
      "type": "best_fields",
      "fields": [ "person_name", "name", "office.name" ],
      "operator":"and"
    }
  }
}

Stick with the custom analyzer approach for a long-term, robust fix.

内容的提问来源于stack exchange，提问作者kgangadhar