CBL-6156: Support Inner Unnest Query in JSON

Added tests for multiple indexes, multi-leveled index, index with N1QL expression.
couchbase · Sep 5, 2024 · cff1e8a · cff1e8a
1 parent 163b832
commit cff1e8a
Show file tree

Hide file tree

Showing 5 changed files with 84 additions and 11 deletions.
diff --git a/C/tests/c4QueryTest.cc b/C/tests/c4QueryTest.cc
@@ -779,7 +779,22 @@ N_WAY_TEST_CASE_METHOD(C4QueryTest, "C4Query UNNEST", "[Query][C]") {
             auto defaultColl = getCollection(db, kC4DefaultCollectionSpec);
             REQUIRE(c4coll_createIndex(defaultColl, C4STR("likes"), C4STR("[[\".likes\"]]"), kC4JSONQuery,
                                        kC4ArrayIndex, nullptr, nullptr));
+            REQUIRE(c4coll_createIndex(defaultColl, C4STR("phone"), C4STR("contact.phone"), kC4N1QLQuery, kC4ArrayIndex,
+                                       nullptr, nullptr));
         }
+
+        // Two UNNESTs for two array properties.
+        compileSelect(json5("{WHAT: ['.person._id', '.phone'],\
+                              FROM: [{as: 'person'}, \
+                                     {as: 'like', unnest: ['.person.likes']},\
+                                     {as: 'phone', unnest: ['.person.contact.phone']}],\
+                             WHERE: ['=', ['.like'], 'climbing'],\
+                          ORDER_BY: [['.person.name.first']]}"));
+        checkExplanation(withIndex);
+        CHECK(run2()
+              == (vector<string>{"0000021, 802-4827967", "0000017, 315-7142142", "0000017, 315-0405535",
+                                 "0000045, 501-7977106", "0000045, 501-7138486"}));
+
         compileSelect(json5("{WHAT: ['.person._id'],\
                               FROM: [{as: 'person'}, \
                                      {as: 'like', unnest: ['.person.likes']}],\
@@ -819,6 +834,8 @@ N_WAY_TEST_CASE_METHOD(NestedQueryTest, "C4Query UNNEST objects", "[Query][C]")
             C4Log("-------- Repeating with index --------");
             REQUIRE(c4db_createIndex(db, C4STR("shapes"), C4STR("[[\".shapes\"], [\".color\"]]"), kC4ArrayIndex,
                                      nullptr, nullptr));
+            REQUIRE(c4db_createIndex2(db, C4STR("shapes2"), C4STR("shapes, concat(color, to_string(size))"),
+                                      kC4N1QLQuery, kC4ArrayIndex, nullptr, nullptr));
         }
         compileSelect(json5("{WHAT: ['.shape.color'],\
                           DISTINCT: true,\
@@ -840,9 +857,48 @@ N_WAY_TEST_CASE_METHOD(NestedQueryTest, "C4Query UNNEST objects", "[Query][C]")
                              WHERE: ['=', ['.shape.color'], 'red']}"));
         checkExplanation(withIndex);
         CHECK(run() == (vector<string>{"11"}));
+
+        compileSelect(json5("{WHAT: [['sum()', ['.shape.size']]],\
+                              FROM: [{as: 'doc'}, \
+                                     {as: 'shape', unnest: ['.doc.shapes']}],\
+                             WHERE: ['=', ['concat()', ['.shape.color'], ['to_string()',['.shape.size']]], 'red3']}"));
+        checkExplanation(withIndex);
+        CHECK(run() == (vector<string>{"3"}));
     }
 }
 
+N_WAY_TEST_CASE_METHOD(NestedQueryTest, "C4Query Nested UNNEST", "[Query][C]") {
+    deleteDatabase();
+    db = c4db_openNamed(kDatabaseName, &dbConfig(), ERROR_INFO());
+    importJSONLines(sFixturesDir + "students.json");
+
+    compileSelect(json5("{WHAT: [['AS', ['.doc.name'], 'college'], ['.student.id'], ['.student.class'], ['.interest']],"
+                        " FROM: [{as: 'doc'},"
+                        "        {as: 'student', unnest: ['.doc.students']},"
+                        "        {as: 'interest', unnest: ['.student.interests']}]"
+                        "}"));
+    vector<string> results{
+            "Univ of Michigan, student_112, 3, violin",     "Univ of Michigan, student_112, 3, baseball",
+            "Univ of Michigan, student_189, 2, violin",     "Univ of Michigan, student_189, 2, tennis",
+            "Univ of Michigan, student_1209, 3, art",       "Univ of Michigan, student_1209, 3, writing",
+            "Univ of Pennsylvania, student_112, 3, piano",  "Univ of Pennsylvania, student_112, 3, swimming",
+            "Univ of Pennsylvania, student_189, 2, violin", "Univ of Pennsylvania, student_189, 2, movies"};
+
+    CHECK(run2(nullptr, 4) == results);
+
+    deleteDatabase();
+    db = c4db_openNamed(kDatabaseName, &dbConfig(), ERROR_INFO());
+    // The only difference from "students.json" is that there is an extra property from student to interests.
+    importJSONLines(sFixturesDir + "students2.json");
+
+    compileSelect(json5("{WHAT: [['AS', ['.doc.name'], 'college'], ['.student.id'], ['.student.class'], ['.interest']],"
+                        " FROM: [{as: 'doc'},"
+                        "        {as: 'student', unnest: ['.doc.students']},"
+                        "        {as: 'interest', unnest: ['.student.extra.interests']}]"
+                        "}"));
+    CHECK(run2(nullptr, 4) == results);
+}
+
 N_WAY_TEST_CASE_METHOD(C4QueryTest, "C4Query Seek", "[Query][C]") {
     compile(json5("['=', ['.', 'contact', 'address', 'state'], 'CA']"));
     C4Error error;

diff --git a/C/tests/c4QueryTest.hh b/C/tests/c4QueryTest.hh
@@ -89,15 +89,21 @@ class C4QueryTest : public C4Test {
         });
     }
 
-    // Runs query, returning vector of doc IDs
-    std::vector<std::string> run2(const char* bindings = nullptr) {
+    // Runs query, returning vector of rows. Columns are comma separated.
+    std::vector<std::string> run2(const char* bindings = nullptr, unsigned colnCount = 2) {
+        REQUIRE(colnCount >= 2);
         return runCollecting<std::string>(bindings, [&](C4QueryEnumerator* e) {
-            REQUIRE(FLArrayIterator_GetCount(&e->columns) >= 2);
-            fleece::alloc_slice c1 = FLValue_ToString(FLArrayIterator_GetValueAt(&e->columns, 0));
-            fleece::alloc_slice c2 = FLValue_ToString(FLArrayIterator_GetValueAt(&e->columns, 1));
-            if ( e->missingColumns & 1 ) c1 = "MISSING"_sl;
-            if ( e->missingColumns & 2 ) c2 = "MISSING"_sl;
-            return c1.asString() + ", " + c2.asString();
+            REQUIRE(FLArrayIterator_GetCount(&e->columns) >= colnCount);
+            std::string res;
+            for ( unsigned c = 0; c < colnCount; ++c ) {
+                if ( c > 0 ) res = res + ", ";
+                if ( e->missingColumns & (1 << c) ) res += "MISSING";
+                else {
+                    fleece::alloc_slice c1 = FLValue_ToString(FLArrayIterator_GetValueAt(&e->columns, c));
+                    res += c1.asString();
+                }
+            }
+            return res;
         });
     }
 

diff --git a/C/tests/data/students.json b/C/tests/data/students.json
@@ -0,0 +1,2 @@
+{"type":"university","name":"Univ of Michigan","students":[{"id":"student_112","class":"3","order":"1","interests":["violin","baseball"]},{"id":"student_189","class":"2","order":"5","interests":["violin","tennis"]},{"id":"student_1209","class":"3","order":"15","interests":["art","writing"]}]}
+{"type":"university","name":"Univ of Pennsylvania","students":[{"id":"student_112","class":"3","order":"1","interests":["piano","swimming"]},{"id":"student_189","class":"2","order":"5","interests":["violin","movies"]}]}
diff --git a/C/tests/data/students2.json b/C/tests/data/students2.json
@@ -0,0 +1,2 @@
+{"type":"university","name":"Univ of Michigan","students":[{"id":"student_112","class":"3","order":"1","extra":{"interests":["violin","baseball"]}},{"id":"student_189","class":"2","order":"5","extra":{"interests":["violin","tennis"]}},{"id":"student_1209","class":"3","order":"15","extra":{"interests":["art","writing"]}}]}
+{"type":"university","name":"Univ of Pennsylvania","students":[{"id":"student_112","class":"3","order":"1","extra":{"interests":["piano","swimming"]}},{"id":"student_189","class":"2","order":"5","extra":{"interests":["violin","movies"]}}]}
diff --git a/LiteCore/Query/QueryParser.cc b/LiteCore/Query/QueryParser.cc
@@ -1767,8 +1767,15 @@ namespace litecore {
             if ( !isInColumnList ) { fail("%s", verifyDbAliasError.c_str()); }
         }
 
-        if ( verifyDbAliasError.empty() && type >= kUnnestVirtualTableAlias ) {
-            // The alias is to an UNNEST. This needs to be written specially:
+        bool isNestedUnnest = fn == kEachFnName && type == kUnnestVirtualTableAlias;
+        // ex. of nested unnest: "FROM: [{as: 'doc'}, {as: 'student', unnest: ['.doc.students']},
+        //                               {as: 'interest', unnest: ['.student.interests']}]"
+        // Target SQL: "FROM kv_default AS doc JOIN fl_each(doc.body, 'students') AS student
+        //                                     JOIN fl_each(student.value, 'interests') AS interest
+
+        if ( !isNestedUnnest && verifyDbAliasError.empty() && type >= kUnnestVirtualTableAlias ) {
+            // The alias is to an UNNEST, but not inside fl_each. This needs to be written specially:
+            // The following function will reject any fn's but kValueFnName
             writeUnnestPropertyGetter(fn, property, alias, type);
             return;
         }
@@ -1844,7 +1851,7 @@ namespace litecore {
         if ( property.empty() && fn == kValueFnName ) fn = kRootFnName;
 
         // Write the function call:
-        _sql << fn << "(" << tablePrefix << _bodyColumnName;
+        _sql << fn << "(" << tablePrefix << (isNestedUnnest ? "value" : _bodyColumnName);
         if ( !property.empty() ) { _sql << ", " << sqlString(string(property)); }
         if ( param ) {
             _sql << ", ";