test: add string function test cases (#4)

* test: add string function test cases

* test: add duckdb string functions to yaml

* fix: fix null handling in duckdb

* fix: add newline for concat

* test: add string support to velox and use i64 for substring param

* feat: add options for substring negative start

bft/testers/duckdb/runner.py

@@ -1,6 +1,5 @@
-from typing import Dict, NamedTuple
-
 import duckdb
+from typing import Dict, NamedTuple
 
 from bft.cases.runner import SqlCaseResult, SqlCaseRunner
 from bft.cases.types import Case, CaseLiteral

bft/testers/sqlite/runner.py

@@ -13,6 +13,7 @@
     "fp32": "REAL",
     "fp64": "REAL",
     "boolean": "BOOLEAN",
+    "string": "TEXT",
 }
 
 
@@ -53,7 +54,14 @@ def run_sql_case(self, case: Case, mapping: SqlMapping) -> SqlCaseResult:
             if mapping.aggregate:
                 arg_names = [arg_names[0]]
             joined_arg_names = ",".join(arg_names)
-            arg_vals = ",".join([literal_to_str(arg) for arg in case.args])
+            arg_vals = ",".join(
+                [
+                    "'" + literal_to_str(arg) + "'"
+                    if arg.type == "string" and arg.value is not None
+                    else literal_to_str(arg)
+                    for arg in case.args
+                ]
+            )
             if mapping.aggregate:
                 arg_vals_list = ", ".join(f"({val})" for val in arg_vals.split(","))
                 if arg_vals != "[]":

bft/testers/velox/runner.py

@@ -5,7 +5,7 @@
 
 
 def is_type_supported(type):
-    return type in set({"i64", "fp64", "boolean"})
+    return type in set({"i64", "fp64", "boolean", "string"})
 
 
 class VeloxRunner(SqlCaseRunner):

cases/string/concat.yaml

@@ -0,0 +1,42 @@
+function: concat
+cases:
+  - group:
+      id: basic
+      description: Basic examples without any special cases
+    args:
+      - value: 'abcd'
+        type: string
+      - value: 'efg'
+        type: string
+    result:
+      value: 'abcdefg'
+      type: string
+  - group:
+      id: null_input
+      description: Examples with null as input
+    args:
+      - value: 'abcd'
+        type: string
+      - value: null
+        type: string
+    result:
+      value: null
+      type: string
+  - group: null_input
+    args:
+      - value: null
+        type: string
+      - value: 'abcd'
+        type: string
+    result:
+      value: null
+      type: string
+  - group: null_input
+    args:
+      - value: null
+        type: string
+      - value: null
+        type: string
+    result:
+      value: null
+      type: string

cases/string/like.yaml

@@ -0,0 +1,60 @@
+function: like
+cases:
+  - group:
+      id: basic
+      description: Basic examples without any special cases
+    args:
+      - value: 'abcdefg'
+        type: string
+      - value: 'abcdefg'
+        type: string
+    result:
+      value: true
+      type: boolean
+  - group: basic
+    args:
+      - value: 'abcdefg'
+        type: string
+      - value: 'abc'
+        type: string
+    result:
+      value: false
+      type: boolean
+  - group:
+      id: wildcard
+      description: Examples using wildcards
+    args:
+      - value: 'abcdefg'
+        type: string
+      - value: 'abc%'
+        type: string
+    result:
+      value: true
+      type: boolean
+  - group: wildcard
+    args:
+      - value: 'abcdefg'
+        type: string
+      - value: '%efg'
+        type: string
+    result:
+      value: true
+      type: boolean
+  - group: wildcard
+    args:
+      - value: 'abcdefg'
+        type: string
+      - value: '_bcdefg'
+        type: string
+    result:
+      value: true
+      type: boolean
+  - group: wildcard
+    args:
+      - value: 'abcdefg'
+        type: string
+      - value: 'abc_efg'
+        type: string
+    result:
+      value: true
+      type: boolean

cases/string/substring.yaml

@@ -0,0 +1,115 @@
+function: substring
+cases:
+  - group:
+      id: basic
+      description: Basic examples without any special cases
+    args:
+      - value: 'abcdefg'
+        type: string
+      - value: 1
+        type: i32
+      - value: 5
+        type: i32
+    result:
+      value: 'abcde'
+      type: string
+  - group: basic
+    args:
+      - value: 'abcdefg'
+        type: string
+      - value: 1
+        type: i64
+      - value: 5
+        type: i64
+    result:
+      value: 'abcde'
+      type: string
+  - group:
+      id: start_greater_than_length
+      description: Example where start argument greater than the length of the string
+    args:
+      - value: 'abcdefg'
+        type: string
+      - value: 10
+        type: i32
+      - value: 2
+        type: i32
+    result:
+      value: ''
+      type: string
+  - group: start_greater_than_length
+    args:
+      - value: 'abcdefg'
+        type: string
+      - value: 10
+        type: i64
+      - value: 2
+        type: i64
+    result:
+      value: ''
+      type: string
+  - group:
+      id: multi_byte_characters
+      description: Example where multi byte characters exist in the string
+    args:
+      - value: '😊a😊b😊😊'
+        type: string
+      - value: 1
+        type: i32
+      - value: 3
+        type: i32
+    result:
+      value: '😊a😊'
+      type: string
+  - group: multi_byte_characters
+    args:
+      - value: '😊a😊b😊😊'
+        type: string
+      - value: 1
+        type: i64
+      - value: 3
+        type: i64
+    result:
+      value: '😊a😊'
+      type: string
+  - group:
+      id: negative_start
+      description: Example where start argument is a negative integer
+    args:
+      - value: 'abcdefg'
+        type: string
+      - value: -3
+        type: i64
+      - value: 2
+        type: i64
+    options:
+      negative_start: WRAP_FROM_END
+    result:
+      value: 'ef'
+      type: string
+  - group: negative_start
+    args:
+      - value: 'abcdefg'
+        type: string
+      - value: -3
+        type: i64
+      - value: 4
+        type: i64
+    options:
+      negative_start: LEFT_OF_BEGINNING
+    result:
+      value: ''
+      type: string
+  - group: negative_start
+    args:
+      - value: 'abcdefg'
+        type: string
+      - value: -3
+        type: i64
+      - value: 4
+        type: i64
+    options:
+      negative_start: LEFT_OF_BEGINNING
+    result:
+      value: 'ab'
+      type: string

dialects/duckdb.yaml

@@ -89,6 +89,14 @@ scalar_functions:
         result: fp32
   - name: extract
     extract: True
+  - name: substring
+    required_options:
+      negative_start: WRAP_FROM_END
+  - name: concat
+    local_name: "||"
+    infix: True
+  - name: like
+    infix: True
 aggregate_functions:
   - name: min
     aggregate: True

dialects/postgres.yaml

@@ -220,6 +220,20 @@ scalar_functions:
         result: fp64
   - name: extract
     extract: True
+  - name: substring
+    required_options:
+      negative_start: LEFT_OF_BEGINNING
+    unsupported_kernels:
+      - args:
+          - string
+          - i64
+          - i64
+        result: string
+  - name: concat
+    local_name: "||"
+    infix: True
+  - name: like
+    infix: True
 aggregate_functions:
   - name: min
     aggregate: True

dialects/sqlite.yaml

@@ -91,6 +91,14 @@ scalar_functions:
     unsupported: true
   - name: extract
     unsupported: True
+  - name: substring
+    required_options:
+      negative_start: WRAP_FROM_END
+  - name: concat
+    local_name: "||"
+    infix: True
+  - name: like
+    infix: True
 aggregate_functions:
   - name: min
     aggregate: True

dialects/velox_presto.yaml

@@ -287,6 +287,19 @@ scalar_functions:
         result: fp64
   - name: extract
     unsupported: True
+  - name: substring
+    local_name: substr
+    required_options:
+      negative_start: WRAP_FROM_END
+    unsupported_kernels:
+      - args:
+          - string
+          - i32
+          - i32
+        result: string
+  - name: concat
+  - name: like
+    unsupported: True
 aggregate_functions:
   - name: min
     unsupported: True