fix multi field search (#2380)

it only checks fields with same analyzer, so we need to use same analyzer for those full text searchable fields SDESK-6567
superdesk · Sep 16, 2022 · c3b3cc4 · c3b3cc4
1 parent d6c6825
commit c3b3cc4
Show file tree

Hide file tree

Showing 3 changed files with 23 additions and 11 deletions.
diff --git a/prod_api/app/settings.py b/prod_api/app/settings.py
@@ -23,6 +23,7 @@
     MONGO_URI,
     ELASTICSEARCH_INDEX,
     ELASTICSEARCH_URL,
+    ELASTICSEARCH_SETTINGS,
     AMAZON_ACCESS_KEY_ID,
     AMAZON_SECRET_ACCESS_KEY,
     AMAZON_REGION,

diff --git a/superdesk/metadata/item.py b/superdesk/metadata/item.py
@@ -11,7 +11,15 @@
 from typing import NamedTuple
 from copy import deepcopy
 
-from superdesk.resource import Resource, not_analyzed, not_indexed, not_enabled, text_with_keyword, not_dynamic
+from superdesk.resource import (
+    Resource,
+    not_analyzed,
+    not_indexed,
+    not_enabled,
+    text_with_keyword,
+    not_dynamic,
+    string_with_analyzer,
+)
 from .packages import LINKED_IN_PACKAGES, PACKAGE
 from eve.utils import config
 from superdesk.utils import SuperdeskBaseEnum
@@ -241,14 +249,11 @@ class Formats(NamedTuple):
     "abstract": {
         "type": "string",
         "nullable": True,
+        "mapping": string_with_analyzer,
     },
     "headline": {
         "type": "string",
-        "mapping": {
-            "type": "string",
-            "analyzer": "html_field_analyzer",
-            "search_analyzer": "html_field_analyzer",
-        },
+        "mapping": string_with_analyzer,
     },
     "slugline": {
         "type": "string",
@@ -259,12 +264,12 @@ class Formats(NamedTuple):
                 "phrase": {
                     "type": "string",
                     "analyzer": "phrase_prefix_analyzer",
-                    "search_analyzer": "phrase_prefix_analyzer",
                     "fielddata": True,
                 },
                 "keyword": {
                     "type": "keyword",
                 },
+                "text": string_with_analyzer,
             },
         },
     },
@@ -288,7 +293,7 @@ class Formats(NamedTuple):
         "type": "integer",
         "nullable": True,
     },
-    "keywords": {"type": "list", "mapping": {"type": "string"}},
+    "keywords": {"type": "list", "mapping": string_with_analyzer},
     "word_count": {"type": "integer"},
     "priority": {"type": "integer", "nullable": True},
     "urgency": {"type": "integer", "nullable": True},
@@ -326,10 +331,12 @@ class Formats(NamedTuple):
     BYLINE: {
         "type": "string",
         "nullable": True,
+        "mapping": string_with_analyzer,
     },
     "ednote": {
         "type": "string",
         "nullable": True,
+        "mapping": string_with_analyzer,
     },
     "authors": {
         "type": "list",
@@ -347,7 +354,7 @@ class Formats(NamedTuple):
             },
         },
     },
-    "description_text": {"type": "string", "nullable": True},
+    "description_text": {"type": "string", "nullable": True, "mapping": string_with_analyzer},
     # This is a description of the item as recieved from its source.
     "archive_description": {"type": "string", "nullable": True},
     "groups": {
@@ -382,11 +389,12 @@ class Formats(NamedTuple):
     "body_html": {
         "type": "string",
         "nullable": True,
-        "mapping": {"type": "string", "analyzer": "html_field_analyzer", "search_analyzer": "html_field_analyzer"},
+        "mapping": string_with_analyzer,
     },
     "body_text": {
         "type": "string",
         "nullable": True,
+        "mapping": string_with_analyzer,
     },
     "dateline": {
         "type": "dict",
@@ -461,7 +469,7 @@ class Formats(NamedTuple):
         "allow_unknown": True,
         "mapping": not_enabled,
     },
-    "filemeta_json": {"type": "string"},
+    "filemeta_json": {"type": "string", "mapping": not_indexed},
     "media_file": {"type": "string"},
     "contents": {"type": "list"},
     ASSOCIATIONS: {
@@ -571,6 +579,7 @@ class Formats(NamedTuple):
     SIGN_OFF: {
         "type": "string",
         "nullable": True,
+        "mapping": string_with_analyzer,
     },
     # Desk and Stage Details
     "task": {

diff --git a/superdesk/resource.py b/superdesk/resource.py
@@ -25,8 +25,10 @@
 not_analyzed = {"type": "string", "index": "not_analyzed"}
 not_enabled = {"type": "object", "enabled": False}  # noqa
 not_dynamic = {"type": "object", "dynamic": False}  # noqa
+string_with_analyzer = {"type": "string", "analyzer": "html_field_analyzer"}  # noqa
 text_with_keyword = {
     "type": "text",
+    "analyzer": "html_field_analyzer",
     "fields": {
         "keyword": {"type": "keyword"},
     },