feat: add data elements to ES index (#1213)

* feat: add data elements to ES index * fix unit test and address comments * remove unused property * cover all TableTagSelect cases * comments
pinterest · Apr 6, 2023 · 3759cf7 · 3759cf7
1 parent 6500610
commit 3759cf7
Show file tree

Hide file tree

Showing 24 changed files with 316 additions and 24 deletions.
diff --git a/package.json b/package.json
@@ -1,6 +1,6 @@
 {
     "name": "querybook",
-    "version": "3.22.0",
+    "version": "3.23.0",
     "description": "A Big Data Webapp",
     "private": true,
     "scripts": {

diff --git a/querybook/config/elasticsearch.yaml b/querybook/config/elasticsearch.yaml
@@ -224,6 +224,15 @@ tables:
                 columns:
                     type: keyword
                     normalizer: case_insensitive
+                column_descriptions:
+                    type: text
+                    analyzer: user_content_analyzer
+                data_elements:
+                    type: keyword
+                    normalizer: case_insensitive
+                data_element_descriptions:
+                    type: text
+                    analyzer: user_content_analyzer
                 golden:
                     type: boolean
                 importance_score:

diff --git a/querybook/server/const/metastore.py b/querybook/server/const/metastore.py
@@ -70,6 +70,9 @@ class DataTable(NamedTuple):
     # Custom properties
     custom_properties: dict[str, str] = None
 
+    golden: bool = False
+    boost_score: float = 1
+
 
 class DataColumn(NamedTuple):
     name: str

diff --git a/querybook/server/datasources/data_element.py b/querybook/server/datasources/data_element.py
@@ -3,6 +3,15 @@
 from logic import data_element as logic
 
 
+@register(
+    "/data_element/keyword/",
+    methods=["GET"],
+)
+def search_data_elements_by_keyword(keyword: str) -> list[str]:
+    data_elements = logic.search_data_elements_by_keyword(keyword=keyword)
+    return [data_element.name for data_element in data_elements]
+
+
 @register("/data_element/<int:data_element_id>/metastore_link/", methods=["GET"])
 def get_data_element_metastore_link(data_element_id: int):
     data_element = logic.get_data_element_by_id(data_element_id)

diff --git a/querybook/server/lib/elasticsearch/search_table.py b/querybook/server/lib/elasticsearch/search_table.py
@@ -5,6 +5,8 @@
     combine_keyword_and_filter_query,
 )
 
+FILTERS_TO_AND = ["tags", "data_elements"]
+
 
 def _get_potential_exact_schema_table_name(keywords):
     """Get the schema and table name from a full table name.
@@ -97,7 +99,7 @@ def construct_tables_query(
         }
     }
 
-    search_filter = match_filters(filters)
+    search_filter = match_filters(filters, and_filter_names=FILTERS_TO_AND)
     query = {
         "query": {
             "bool": combine_keyword_and_filter_query(keywords_query, search_filter)
@@ -117,6 +119,10 @@ def construct_tables_query(
                     "fragment_size": 20,
                     "number_of_fragments": 5,
                 },
+                "data_elements": {
+                    "fragment_size": 20,
+                    "number_of_fragments": 5,
+                },
                 "description": {
                     "fragment_size": 60,
                     "number_of_fragments": 3,

diff --git a/querybook/server/lib/metastore/base_metastore_loader.py b/querybook/server/lib/metastore/base_metastore_loader.py
@@ -361,6 +361,8 @@ def _create_table_table(
                 location=table.location,
                 column_count=len(columns),
                 schema_id=schema_id,
+                golden=table.golden,
+                boost_score=table.boost_score,
                 commit=False,
                 session=session,
             ).id

diff --git a/querybook/server/logic/data_element.py b/querybook/server/logic/data_element.py
@@ -24,6 +24,17 @@ def get_data_element_by_name(name: str, session=None):
     return DataElement.get(name=name, session=session)
 
 
+@with_session
+def search_data_elements_by_keyword(keyword: str, limit=10, session=None):
+    return (
+        session.query(DataElement)
+        .filter(DataElement.name.like("%" + keyword + "%"))
+        .order_by(DataElement.name.asc())
+        .limit(limit)
+        .all()
+    )
+
+
 @with_session
 def get_data_element_association_by_column_id(
     column_id: int, session=None

diff --git a/querybook/server/logic/elasticsearch.py b/querybook/server/logic/elasticsearch.py
@@ -545,13 +545,26 @@ def table_to_es(table, fields=None, session=None):
     table_name = table.name
     full_name = "{}.{}".format(schema_name, table_name)
 
+    # columns may be associated with the same data element
+    data_elements = {d.name: d for c in table.columns for d in c.data_elements}.values()
+
     def get_table_description():
         return (
             richtext_to_plaintext(table.information.description, escape=True)
             if table.information
             else ""
         )
 
+    def get_column_descriptions():
+        return [
+            richtext_to_plaintext(c.description, escape=True) for c in table.columns
+        ]
+
+    def get_data_element_descriptions():
+        return [
+            richtext_to_plaintext(d.description, escape=True) for d in data_elements
+        ]
+
     weight = None
 
     def compute_weight():
@@ -583,6 +596,9 @@ def get_completion_name():
         "description": get_table_description,
         "created_at": lambda: DATETIME_TO_UTC(table.created_at),
         "columns": [c.name for c in table.columns],
+        "column_descriptions": get_column_descriptions,
+        "data_elements": [d.name for d in data_elements],
+        "data_element_descriptions": get_data_element_descriptions,
         "golden": table.golden,
         "importance_score": compute_weight,
         "tags": [tag.tag_name for tag in table.tags],

diff --git a/querybook/server/logic/metastore.py b/querybook/server/logic/metastore.py
@@ -201,6 +201,8 @@ def create_table(
     location=None,
     column_count=None,
     schema_id=None,
+    golden=False,
+    boost_score=1,
     commit=True,
     session=None,
 ):
@@ -219,6 +221,8 @@ def create_table(
         "location": location,
         "column_count": column_count,
         "schema_id": schema_id,
+        "golden": golden,
+        "boost_score": boost_score,
     }
 
     table = get_table_by_schema_id_and_name(schema_id, name, session=session)

diff --git a/querybook/server/models/metastore.py b/querybook/server/models/metastore.py
@@ -298,6 +298,10 @@ class DataTableColumn(TruncateString("name", "type", "comment"), Base):
         sql.Integer, sql.ForeignKey("data_table.id", ondelete="CASCADE")
     )
 
+    data_elements = relationship(
+        "DataElement", secondary="data_element_association", uselist=True, viewonly=True
+    )
+
     def to_dict(self, include_table=False):
         column_dict = {
             "id": self.id,

diff --git a/querybook/tests/test_lib/test_elasticsearch/test_elasticsearch.py b/querybook/tests/test_lib/test_elasticsearch/test_elasticsearch.py
@@ -339,11 +339,26 @@ def _get_data_schema_mock(self):
         mock_data_schema.name = self.SCHEMA_NAME
         return mock_data_schema
 
+    def _get_data_element_mock(self, name: str, description: str):
+        mock_de = MagicMock()
+        mock_de.name = name
+        mock_de.description = description
+        return mock_de
+
     def _get_columns_mock(self):
         mock_col_a = MagicMock()
         mock_col_a.name = "col_a"
+        mock_col_a.description = "col_a_description"
+        mock_col_a.data_elements = [
+            self._get_data_element_mock("de_a", "de_a_description"),
+            self._get_data_element_mock("de_b", "de_b_description"),
+        ]
         mock_col_b = MagicMock()
         mock_col_b.name = "col_b"
+        mock_col_b.description = "col_b_description"
+        mock_col_b.data_elements = [
+            self._get_data_element_mock("de_a", "de_a_description"),
+        ]
         return [mock_col_a, mock_col_b]
 
     def _get_table_mock(self):
@@ -393,6 +408,9 @@ def test_table_to_es(self):
             "description": self.TABLE_DESCRIPTION,
             "created_at": CREATED_AT_EPOCH,
             "columns": ["col_a", "col_b"],
+            "column_descriptions": ["col_a_description", "col_b_description"],
+            "data_elements": ["de_a", "de_b"],
+            "data_element_descriptions": ["de_a_description", "de_b_description"],
             "golden": False,
             "importance_score": self.TABLE_WEIGHT,
             "tags": ["tag_1", "tag_2"],

diff --git a/querybook/webapp/components/DataTableNavigator/DataTableNavigatorSearch.tsx b/querybook/webapp/components/DataTableNavigator/DataTableNavigatorSearch.tsx
@@ -2,7 +2,7 @@ import { startCase } from 'lodash';
 import React, { useCallback, useMemo, useRef } from 'react';
 import { useDispatch, useSelector } from 'react-redux';
 
-import { TableTagGroupSelect } from 'components/DataTableTags/TableTagGroupSelect';
+import { EntitySelect } from 'components/Search/EntitySelect';
 import { ComponentType, ElementType } from 'const/analytics';
 import { useToggleState } from 'hooks/useToggleState';
 import { trackClick } from 'lib/analytics';
@@ -12,6 +12,7 @@ import {
 } from 'redux/dataTableSearch/action';
 import { ITableSearchFilters } from 'redux/dataTableSearch/types';
 import { IStoreState } from 'redux/store/types';
+import { TableTagResource } from 'resource/table';
 import { SoftButton } from 'ui/Button/Button';
 import { IconButton } from 'ui/Button/IconButton';
 import { OrderByButton } from 'ui/OrderByButton/OrderByButton';
@@ -108,9 +109,12 @@ export const DataTableNavigatorSearch: React.FC<{
                         />
                     </SearchFilterRow>
                     <SearchFilterRow title="Tags">
-                        <TableTagGroupSelect
-                            tags={searchFilters?.tags}
-                            updateTags={updateTags}
+                        <EntitySelect
+                            selectedEntities={searchFilters?.tags || []}
+                            loadEntities={TableTagResource.search}
+                            onEntitiesChange={updateTags}
+                            placeholder="Tag name"
+                            mini
                         />
                     </SearchFilterRow>
                 </div>

diff --git a/querybook/webapp/components/DataTableTags/CreateDataTableTag.tsx b/querybook/webapp/components/DataTableTags/CreateDataTableTag.tsx
@@ -1,13 +1,14 @@
 import * as React from 'react';
 import { useDispatch } from 'react-redux';
 
+import { EntitySelect } from 'components/Search/EntitySelect';
 import { ITag } from 'const/tag';
+import { isTagValid } from 'lib/utils/tag';
 import { Dispatch } from 'redux/store/types';
 import { createTableTag } from 'redux/tag/action';
+import { TableTagResource } from 'resource/table';
 import { IconButton } from 'ui/Button/IconButton';
 
-import { TableTagSelect } from './TableTagSelect';
-
 import './CreateDataTableTag.scss';
 
 interface IProps {
@@ -43,10 +44,15 @@ export const CreateDataTableTag: React.FunctionComponent<IProps> = ({
         <div className="CreateDataTableTag flex-row">
             {showSelect ? (
                 <div className="CreateDataTableTag-input flex-row">
-                    <TableTagSelect
-                        onSelect={handleCreateTag}
-                        existingTags={existingTags}
+                    <EntitySelect
                         creatable
+                        mini
+                        selectedEntities={existingTags || []}
+                        loadEntities={TableTagResource.search}
+                        onSelect={handleCreateTag}
+                        validateEntity={isTagValid}
+                        placeholder="alphanumeric only"
+                        showSelected={false}
                     />
                 </div>
             ) : (

diff --git a/querybook/webapp/components/DataTableTags/DataTableTags.tsx b/querybook/webapp/components/DataTableTags/DataTableTags.tsx
@@ -5,6 +5,7 @@ import { useDispatch, useSelector } from 'react-redux';
 import { ITag } from 'const/tag';
 import { stopPropagationAndDefault } from 'lib/utils/noop';
 import { navigateWithinEnv } from 'lib/utils/query-string';
+import { useRankedTags } from 'lib/utils/tag';
 import { Dispatch, IStoreState } from 'redux/store/types';
 import {
     deleteTableTag,
@@ -18,7 +19,6 @@ import { HoverIconTag } from 'ui/Tag/HoverIconTag';
 
 import { CreateDataTableTag } from './CreateDataTableTag';
 import { TableTagConfigModal } from './TableTagConfigModal';
-import { useRankedTags } from './utils';
 
 import './DataTableTags.scss';
 

diff --git a/querybook/webapp/components/DataTableTags/TableTagGroupSelect.tsx b/querybook/webapp/components/DataTableTags/TableTagGroupSelect.tsx
@@ -25,7 +25,7 @@ export const TableTagGroupSelect: React.FC<{
     );
 
     const tagsListDOM = tags.length ? (
-        <div className="tables-tag-list mb8">
+        <div className="mb8">
             {tags.map((tag) => (
                 <HoverIconTag
                     key={tag}

diff --git a/querybook/webapp/components/Search/EntitySelect.scss b/querybook/webapp/components/Search/EntitySelect.scss
@@ -0,0 +1,21 @@
+.EntitySelect {
+    min-width: 180px;
+
+    .invalid {
+        border: 1px solid var(--color-false);
+        border-radius: var(--border-radius-sm);
+    }
+
+    .entity-list {
+        display: flex;
+        align-items: center;
+        flex-wrap: wrap;
+        gap: 6px;
+        margin-bottom: 8px;
+
+        .Tag {
+            margin-left: 0 !important;
+            font-size: var(--xxsmall-text-size);
+        }
+    }
+}