From a55d65caf849e5b29ff20ccd62e4eebd4cfdb0fb Mon Sep 17 00:00:00 2001
From: aceforeverd <teapot@aceforeverd.com>
Date: Wed, 19 Jun 2024 06:24:17 +0000
Subject: [PATCH] fix(array_combine): behavior tweaks

- use empty string if delimiter is null
- restrict to array_combine(string, array<T> ...)
---
 cases/query/udf_query.yaml                 | 17 +++++++++++++++--
 hybridse/src/udf/default_defs/array_def.cc |  8 ++++----
 2 files changed, 19 insertions(+), 6 deletions(-)

diff --git a/cases/query/udf_query.yaml b/cases/query/udf_query.yaml
index 564d4f7594c..528ee7942a5 100644
--- a/cases/query/udf_query.yaml
+++ b/cases/query/udf_query.yaml
@@ -591,13 +591,15 @@ cases:
         array_join(array_combine("-", [1, 2], [3, 4]), ",") c0,
         array_join(array_combine("-", [1, 2], array<int64>[3], ["5", "6"]), ",") c1,
         array_join(array_combine("|", ["1"], [timestamp(1717171200000), timestamp("2024-06-02 12:00:00")]), ",") c2,
+        array_join(array_combine("|", ["1"]), ",") c3,
     expect:
       columns:
         - c0 string
         - c1 string
         - c2 string
+        - c3 string
       rows:
-        - ["1-3,1-4,2-3,2-4", "1-3-5,1-3-6,2-3-5,2-3-6", "1|2024-06-01 00:00:00,1|2024-06-02 12:00:00"]
+        - ["1-3,1-4,2-3,2-4", "1-3-5,1-3-6,2-3-5,2-3-6", "1|2024-06-01 00:00:00,1|2024-06-02 12:00:00", "1"]
   - id: array_combine_3
     desc: null values skipped
     mode: request-unsupport
@@ -605,12 +607,14 @@ cases:
       select
         array_join(array_combine("-", [1, NULL], [3, 4]), ",") c0,
         array_join(array_combine("-", ARRAY<INT>[NULL], ["9", "8"]), ",") c1,
+        array_join(array_combine(string(NULL), ARRAY<INT>[1], ["9", "8"]), ",") c2,
     expect:
       columns:
         - c0 string
         - c1 string
+        - c2 string
       rows:
-        - ["1-3,1-4", ""]
+        - ["1-3,1-4", "", "19,18"]
   - id: array_combine_4
     desc: construct array from table
     mode: request-unsupport
@@ -633,6 +637,15 @@ cases:
       rows:
         - [1, "1-foo,1-c2,10-foo,10-c2"]
         - [2, "2-bar,2-c2,10-bar,10-c2"]
+  - id: array_combine_err1
+    mode: request-unsupport
+    sql: |
+      select
+        array_join(array_combine("-"), ",") c0,
+    expect:
+      success: false
+      msg: |
+        Fail to resolve expression: array_join(array_combine(-), ,)
 
   # ================================================================
   # Map data type
diff --git a/hybridse/src/udf/default_defs/array_def.cc b/hybridse/src/udf/default_defs/array_def.cc
index 5c4a6dc7226..96cc12fc909 100644
--- a/hybridse/src/udf/default_defs/array_def.cc
+++ b/hybridse/src/udf/default_defs/array_def.cc
@@ -145,7 +145,7 @@ void DefaultUdfLibrary::InitArrayUdfs() {
     RegisterExternal("array_join")
         .args<ArrayRef<StringRef>, Nullable<StringRef>>(array_join)
         .doc(R"(
-             @brief array_join(array, delimiter) - Concatenates the elements of the given array using the delimiter and an optional string to replace nulls. Any null value is filtered.
+             @brief array_join(array, delimiter) - Concatenates the elements of the given array using the delimiter. Any null value is filtered.
 
              Example:
 
@@ -156,10 +156,10 @@ void DefaultUdfLibrary::InitArrayUdfs() {
              @since 0.9.2)");
 
     RegisterCodeGenUdf("array_combine")
-        .variadic_args<AnyArg>(
+        .variadic_args<Nullable<StringRef>>(
             [](UdfResolveContext* ctx, const ExprAttrNode& delimit, const std::vector<ExprAttrNode>& arg_attrs,
                ExprAttrNode* out) -> base::Status {
-                CHECK_TRUE(delimit.type()->IsString(), common::kCodegenError, "delimiter must be string");
+                CHECK_TRUE(!arg_attrs.empty(), common::kCodegenError, "at least one array required by array_combine");
                 for (auto & val : arg_attrs) {
                     CHECK_TRUE(val.type()->IsArray(), common::kCodegenError, "argument to array_combine must be array");
                 }
@@ -179,7 +179,7 @@ void DefaultUdfLibrary::InitArrayUdfs() {
                 @brief array_combine(delimiter, array1, array2, ...)
 
                 return array of strings for input array1, array2, ... doing cartesian product. Each product is joined with
-                {delimiter} as a string
+                {delimiter} as a string. Empty string used if {delimiter} is null.
 
                 Example: