Adds q.op=AND to Solr queries so that boolean syntax will work (#459)

We need to specify q.op=AND along with mm=100% to allow for boolean query syntax, which is a little surprising, but explained in https://issues.apache.org/jira/browse/SOLR-8812
monarch-initiative · Nov 2, 2023 · 9e9e7ea · 9e9e7ea
1 parent 029186a
commit 9e9e7ea
Show file tree

Hide file tree

Showing 13 changed files with 22 additions and 7 deletions.
diff --git a/backend/src/monarch_py/datamodels/solr.py b/backend/src/monarch_py/datamodels/solr.py
@@ -45,7 +45,8 @@ class SolrQuery(BaseModel):
     filter_queries: Optional[List[str]] = Field(default_factory=list)
     query_fields: str = None
     def_type: str = "edismax"
-    mm: str = "100%"  # All tokens in the query must be found in the doc, equivalent to q.op="AND"
+    q_op: str = "AND"  # See SOLR-8812, need this plus mm=100% to allow boolean operators in queries
+    mm: str = "100%"  # All tokens in the query must be found in the doc
     boost: str = None
     sort: str = None
 
@@ -83,6 +84,8 @@ def _solrize(self, value):
             return "qf"
         elif value == "def_type":
             return "defType"
+        elif value == "q_op":
+            return "q.op"
         elif value is True:
             return "true"
         elif value is False:

diff --git a/backend/tests/fixtures/association_counts_query.py b/backend/tests/fixtures/association_counts_query.py
@@ -40,6 +40,7 @@ def association_counts_query():
         ],
         "query_fields": None,
         "def_type": "edismax",
+        "q_op": "AND",
         "mm": "100%",
         "boost": None,
         "sort": None,

diff --git a/backend/tests/fixtures/association_counts_response.py b/backend/tests/fixtures/association_counts_response.py
@@ -5,7 +5,7 @@
 def association_counts_response():
     return {
         "responseHeader": {
-            "QTime": 4,
+            "QTime": 1,
             "params": {
                 "facet.query": [
                     '(category:"biolink:DiseaseToPhenotypicFeatureAssociation") AND (subject:"MONDO:0020121" OR subject_closure:"MONDO:0020121")',
@@ -38,6 +38,7 @@ def association_counts_response():
                 "defType": "edismax",
                 "facet_min_count": "1",
                 "start": "0",
+                "q.op": "AND",
                 "fq": 'subject:"MONDO\\:0020121" OR subject_closure:"MONDO\\:0020121" OR object:"MONDO\\:0020121" OR object_closure:"MONDO\\:0020121"',
                 "rows": "20",
                 "facet": "true",

diff --git a/backend/tests/fixtures/association_query_direct.py b/backend/tests/fixtures/association_query_direct.py
@@ -21,6 +21,7 @@ def association_query_direct():
         ],
         "query_fields": "subject subject_label predicate object object_label",
         "def_type": "edismax",
+        "q_op": "AND",
         "mm": "100%",
         "boost": None,
         "sort": None,

diff --git a/backend/tests/fixtures/association_query_indirect.py b/backend/tests/fixtures/association_query_indirect.py
@@ -21,6 +21,7 @@ def association_query_indirect():
         ],
         "query_fields": "subject subject_label predicate object object_label",
         "def_type": "edismax",
+        "q_op": "AND",
         "mm": "100%",
         "boost": None,
         "sort": None,

diff --git a/backend/tests/fixtures/association_response.py b/backend/tests/fixtures/association_response.py
@@ -5,13 +5,14 @@
 def association_response():
     return {
         "responseHeader": {
-            "QTime": 1,
+            "QTime": 0,
             "params": {
                 "mm": "100%",
                 "q": "*:*",
                 "defType": "edismax",
                 "facet_min_count": "1",
                 "start": "0",
+                "q.op": "AND",
                 "fq": 'subject:"MONDO\\:0020121" OR subject_closure:"MONDO\\:0020121" OR object:"MONDO\\:0020121" OR object_closure:"MONDO\\:0020121"',
                 "rows": "20",
                 "facet": "true",

diff --git a/backend/tests/fixtures/association_table_response.py b/backend/tests/fixtures/association_table_response.py
@@ -5,13 +5,14 @@
 def association_table_response():
     return {
         "responseHeader": {
-            "QTime": 1,
+            "QTime": 0,
             "params": {
                 "mm": "100%",
                 "q": "*:*",
                 "defType": "edismax",
                 "facet_min_count": "1",
                 "start": "0",
+                "q.op": "AND",
                 "fq": [
                     "category:biolink\\:DiseaseToPhenotypicFeatureAssociation",
                     'subject:"MONDO\\:0020121" OR subject_closure:"MONDO\\:0020121" OR object:"MONDO\\:0020121" OR object_closure:"MONDO\\:0020121"',

diff --git a/backend/tests/fixtures/autocomplete_query.py b/backend/tests/fixtures/autocomplete_query.py
@@ -13,6 +13,7 @@ def autocomplete_query():
         "filter_queries": [],
         "query_fields": "id^100 name^10 name_t^5 name_ac symbol^10 symbol_t^5 symbol_ac synonym synonym_t synonym_ac",
         "def_type": "edismax",
+        "q_op": "AND",
         "mm": "100%",
         "boost": 'product(if(termfreq(category,"biolink:Disease"),10.0,1),if(and(termfreq(in_taxon,"NCBITaxon:9606"),termfreq(category,"biolink:Gene")),5.0,1))',
         "sort": None,

diff --git a/backend/tests/fixtures/autocomplete_response.py b/backend/tests/fixtures/autocomplete_response.py
@@ -5,14 +5,15 @@
 def autocomplete_response():
     return {
         "responseHeader": {
-            "QTime": 1,
+            "QTime": 0,
             "params": {
                 "mm": "100%",
                 "q": "fanc",
                 "defType": "edismax",
                 "facet_min_count": "1",
                 "qf": "id^100 name^10 name_t^5 name_ac symbol^10 symbol_t^5 symbol_ac synonym synonym_t synonym_ac",
                 "start": "0",
+                "q.op": "AND",
                 "boost": 'product(if(termfreq(category,"biolink:Disease"),10.0,1),if(and(termfreq(in_taxon,"NCBITaxon:9606"),termfreq(category,"biolink:Gene")),5.0,1))',
                 "rows": "20",
                 "facet": "true",

diff --git a/backend/tests/fixtures/histopheno_query.py b/backend/tests/fixtures/histopheno_query.py
@@ -34,6 +34,7 @@ def histopheno_query():
         "filter_queries": ["subject_closure:MONDO\\:0020121"],
         "query_fields": None,
         "def_type": "edismax",
+        "q_op": "AND",
         "mm": "100%",
         "boost": None,
         "sort": None,

diff --git a/backend/tests/fixtures/histopheno_response.py b/backend/tests/fixtures/histopheno_response.py
@@ -5,7 +5,7 @@
 def histopheno_response():
     return {
         "responseHeader": {
-            "QTime": 6,
+            "QTime": 1,
             "params": {
                 "facet.query": [
                     'object_closure:"HP:0000924"',
@@ -34,6 +34,7 @@ def histopheno_response():
                 "defType": "edismax",
                 "facet_min_count": "1",
                 "start": "0",
+                "q.op": "AND",
                 "fq": "subject_closure:MONDO\\:0020121",
                 "rows": "0",
                 "facet": "true",

diff --git a/backend/tests/fixtures/search_query.py b/backend/tests/fixtures/search_query.py
@@ -13,6 +13,7 @@ def search_query():
         "filter_queries": ["name:*"],
         "query_fields": "id^100 name^10 name_t^5 name_ac symbol^10 symbol_t^5 symbol_ac synonym synonym_t synonym_ac",
         "def_type": "edismax",
+        "q_op": "AND",
         "mm": "100%",
         "boost": 'product(if(termfreq(category,"biolink:Disease"),10.0,1),if(and(termfreq(in_taxon,"NCBITaxon:9606"),termfreq(category,"biolink:Gene")),5.0,1))',
         "sort": None,

diff --git a/backend/tests/fixtures/search_response.py b/backend/tests/fixtures/search_response.py
@@ -5,14 +5,15 @@
 def search_response():
     return {
         "responseHeader": {
-            "QTime": 3,
+            "QTime": 1,
             "params": {
                 "mm": "100%",
                 "q": "fanconi",
                 "defType": "edismax",
                 "facet_min_count": "1",
                 "qf": "id^100 name^10 name_t^5 name_ac symbol^10 symbol_t^5 symbol_ac synonym synonym_t synonym_ac",
                 "start": "0",
+                "q.op": "AND",
                 "boost": 'product(if(termfreq(category,"biolink:Disease"),10.0,1),if(and(termfreq(in_taxon,"NCBITaxon:9606"),termfreq(category,"biolink:Gene")),5.0,1))',
                 "fq": "name:*",
                 "rows": "20",