SNO-108-improve-search-new (#212)

SNO-108 Improve search mapping and add boost
ENCODE-DCC · Jul 23, 2019 · f6b1445 · f6b1445
1 parent 8754ae6
commit f6b1445
Show file tree

Hide file tree

Showing 3 changed files with 37 additions and 18 deletions.
diff --git a/src/snovault/elasticsearch/create_mapping.py b/src/snovault/elasticsearch/create_mapping.py
@@ -33,8 +33,7 @@
 META_MAPPING = {
     '_all': {
         'enabled': False,
-        'analyzer': 'snovault_index_analyzer',
-        'search_analyzer': 'snovault_search_analyzer'
+        'analyzer': 'snovault_search_analyzer'
     },
     'dynamic_templates': [
         {
@@ -176,6 +175,25 @@ def index_settings():
                         'type': 'nGram',
                         'min_gram': 1,
                         'max_gram': 33
+                    },
+                    'english_stop': {
+                        'type': 'stop',
+                        'stopwords': '_english_'
+                    },
+                    'english_stemmer': {
+                        'type': 'stemmer',
+                        'language': 'english'
+                    },
+                    'english_possessive_stemmer': {
+                        'type': 'stemmer',
+                        'language': 'possessive_english'
+                    },
+                    'delimiter': {
+                        'type': 'word_delimiter',
+                        'catenate_all': True,
+                        'preserve_original': True,
+                        'stem_english_possessive': True,
+                        'split_on_numerics': False
                     }
                 },
                 'analyzer': {
@@ -184,28 +202,38 @@ def index_settings():
                         'tokenizer': 'whitespace',
                         'char_filter': 'html_strip',
                         'filter': [
-                            'standard',
+                            'english_possessive_stemmer',
                             'lowercase',
+                            'english_stop',
+                            'english_stemmer',
+                            'asciifolding',
+                            'delimiter'
                         ]
                     },
                     'snovault_index_analyzer': {
                         'type': 'custom',
                         'tokenizer': 'whitespace',
                         'char_filter': 'html_strip',
                         'filter': [
-                            'standard',
+                            'english_possessive_stemmer',
                             'lowercase',
+                            'english_stop',
+                            'english_stemmer',
                             'asciifolding',
+                            'delimiter',
                             'substring'
                         ]
                     },
                     'snovault_search_analyzer': {
                         'type': 'custom',
                         'tokenizer': 'whitespace',
                         'filter': [
-                            'standard',
+                            'english_possessive_stemmer',
                             'lowercase',
-                            'asciifolding'
+                            'english_stop',
+                            'english_stemmer',
+                            'asciifolding',
+                            'delimiter'
                         ]
                     },
                     'snovault_path_analyzer': {
@@ -247,8 +275,7 @@ def es_mapping(mapping):
     return {
         '_all': {
             'enabled': True,
-            'analyzer': 'snovault_index_analyzer',
-            'search_analyzer': 'snovault_search_analyzer'
+            'analyzer': 'snovault_search_analyzer'
         },
         'dynamic_templates': [
             {

diff --git a/src/snovault/helpers/helper.py b/src/snovault/helpers/helper.py
@@ -157,15 +157,7 @@ def get_search_fields(request, doc_types):
     Returns set of columns that are being searched and highlights
     """
 
-    fields = {'uuid', 'unique_keys.*'}
-    highlights = {}
-    types = request.registry[TYPES]
-    for doc_type in doc_types:
-        type_info = types[doc_type]
-        for value in type_info.schema.get('boost_values', ()):
-            fields.add('embedded.' + value)
-            highlights['embedded.' + value] = {}
-    return list(fields), highlights
+    return [], {}
 
 
 def list_visible_columns_for_schemas(request, schemas):

diff --git a/src/snovault/viewconfigs/searchview.py b/src/snovault/viewconfigs/searchview.py
@@ -136,7 +136,7 @@ def preprocess_view(self, views=None, search_result_actions=None, preserve_order
             del query['query']['query_string']
         else:
             query['query']['query_string']['fields'].extend(
-                ['_all', '*.uuid', '*.md5sum', '*.submitted_file_name']
+                ['_all']
             )
         set_sort_order(self._request, search_term, types, doc_types, query, result)
         used_filters = set_filters(self._request, query, result)