fix: to_base64 should cast input parameter to string

Author: aestriplex

docs/en/14-reference/03-taos-sql/10-function.md

@@ -1169,9 +1169,7 @@ TO_BASE64(expr)
 
 **Return Type**: VARCHAR.
 
-**Applicable Data Types**:
-
-- `expr`: VARCHAR, NCHAR.
+**Applicable Data Types**: Suitable for any type.
 
 **Nested Subquery Support**: Applicable to both inner and outer queries.
 
@@ -1180,15 +1178,33 @@ TO_BASE64(expr)
 **Usage Notes**:
 
 - If `expr` is NULL, returns NULL.
+- if `expr` is the empty string, it returns the empty string.
+- if `expr` is a non string, it is interpreted as a string.
+- This function is multibyte safe.
 
 **Example**:
 
 ```sql
+taos> select to_base64(NULL);
+ to_base64(null) |
+==================
+ NULL            |
+
 taos> select to_base64("");
  to_base64("") |
 ================
                |
 
+taos> select to_base64(14324);
+ to_base64(14324) |
+=====================
+ MTQzMjQ=           |
+
+taos> select to_base64("14324");
+ to_base64("14324") |
+=====================
+ MTQzMjQ=           |
+
 taos> select to_base64("Hello, world!");
  to_base64("Hello, world!") |
 =============================

docs/zh/14-reference/03-taos-sql/10-function.md

@@ -1120,26 +1120,42 @@ TO_BASE64(expr)
 
 **返回结果类型**: VARCHAR
 
-**适用数据类型**:
-
-- `expr`: VARCHAR, NCHAR.
+**适用数据类型**: 任何类型。
 
 **嵌套子查询支持**：适用于内层查询和外层查询。
 
 **适用于**：表和超级表。
 
 **使用说明**:
 
-- 若 expr 为 NULL，返回 NULL。
+- 若 `expr` 为 NULL，返回 NULL。
+- 如果 `expr` 为空字符串，则返回空字符串。
+- 如果 `expr` 为非字符串，则将其解释为字符串。
+- 此函数是多字节安全的。
 
 **举例**:
 
 ```sql
+taos> select to_base64(NULL);
+ to_base64(null) |
+==================
+ NULL            |
+
 taos> select to_base64("");
  to_base64("") |
 ================
                |
 
+taos> select to_base64(14324);
+ to_base64(14324) |
+=====================
+ MTQzMjQ=           |
+
+taos> select to_base64("14324");
+ to_base64("14324") |
+=====================
+ MTQzMjQ=           |
+
 taos> select to_base64("Hello, world!");
  to_base64("Hello, world!") |
 =============================

source/libs/function/src/builtins.c

@@ -5462,7 +5462,7 @@ const SBuiltinFuncDefinition funcMgtBuiltins[] = {
                    .inputParaInfo[0][0] = {.isLastParam = true,
                                            .startParam = 1,
                                            .endParam = 1,
-                                           .validDataType = FUNC_PARAM_SUPPORT_VARCHAR_TYPE | FUNC_PARAM_SUPPORT_NCHAR_TYPE | FUNC_PARAM_SUPPORT_NULL_TYPE,
+                                           .validDataType = FUNC_PARAM_SUPPORT_ALL_TYPE,
                                            .validNodeType = FUNC_PARAM_SUPPORT_EXPR_NODE,
                                            .paramAttribute = FUNC_PARAM_NO_SPECIFIC_ATTRIBUTE,
                                            .valueRangeFlag = FUNC_PARAM_NO_SPECIFIC_VALUE,},

source/libs/scalar/src/sclfunc.c

@@ -1721,7 +1721,13 @@ static char base64Table[] = "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
                             "abcdefghijklmnopqrstuvwxyz"
                             "0123456789+/";
 
-static void base64Impl(uint8_t *base64Out, const uint8_t *inputBytes, size_t inputLen, VarDataLenT outputLen) {
+uint32_t base64BufSize(size_t inputLenBytes) {
+  return 4 * ((inputLenBytes + 2) / 3);
+}
+
+static VarDataLenT base64Impl(uint8_t *base64Out, const uint8_t *inputBytes, size_t inputLen) {
+  VarDataLenT outputLen = base64BufSize(inputLen);
+
   for (size_t i = 0, j = 0; i < inputLen;) {
     unsigned int octet_a = i < inputLen ? inputBytes[i++] : 0;
     unsigned int octet_b = i < inputLen ? inputBytes[i++] : 0;
@@ -1740,39 +1746,82 @@ static void base64Impl(uint8_t *base64Out, const uint8_t *inputBytes, size_t inp
   }
 
   base64Out[outputLen] = 0;
-}
 
-uint32_t base64BufSize(size_t inputLenBytes) {
-  return 4 * ((inputLenBytes + 2) / 3);
+  return outputLen;
 }
 
+#define BASE64_TRUE  "MQ=="
+#define BASE64_FALSE "MA=="
+
 int32_t base64Function(SScalarParam* pInput, int32_t inputNum, SScalarParam* pOutput) {
   int32_t code = TSDB_CODE_SUCCESS;
   SColumnInfoData *pInputData = pInput->columnData;
   SColumnInfoData *pOutputData = pOutput->columnData;
-  char *outputBuf = taosMemoryMalloc(TSDB_MAX_FIELD_LEN + VARSTR_HEADER_SIZE);
-  if (outputBuf == NULL) {
+  int16_t inputType = GET_PARAM_TYPE(&pInput[0]);
+  int64_t outputLen = GET_PARAM_BYTES(&pOutput[0]);
+  char *stringBuf = taosMemoryMalloc(TSDB_MAX_FIELD_LEN + VARSTR_HEADER_SIZE);
+  char *output = taosMemoryMalloc(TSDB_MAX_FIELD_LEN + VARSTR_HEADER_SIZE);
+
+  if (!stringBuf || output) {
     SCL_ERR_RET(terrno);
   }
-  
+
+  /* Preparing the output buffer with VARSTR_HEADER_SIZE
+     non-zero char at the beginning */
+  for (size_t i = 0; i < VARSTR_HEADER_SIZE; i++) {
+    output[i] = 1;
+  }
+  char *out = output + VARSTR_HEADER_SIZE;
+
   for (int32_t i = 0; i < pInput->numOfRows; ++i) {
     if (colDataIsNull_s(pInputData, i)) {
       colDataSetNULL(pOutputData, i);
       continue;
     }
 
-    char *input = colDataGetData(pInputData, i);
-    size_t inputLen = varDataLen(colDataGetData(pInputData, i));
-    char *out = outputBuf + VARSTR_HEADER_SIZE;
-    VarDataLenT outputLength = base64BufSize(inputLen);
-    base64Impl(out, varDataVal(input), inputLen, outputLength);
-    varDataSetLen(outputBuf, outputLength);
-    SCL_ERR_JRET(colDataSetVal(pOutputData, i, outputBuf, false));
+    char *input = colDataGetData(pInput[0].columnData, i);
+    size_t inputLen = varDataLen(input);
+    size_t outputSize = TSDB_MAX_FIELD_LEN;
+
+    if (inputType == TSDB_DATA_TYPE_BOOL) {
+      memcpy(out, *(int8_t *)input ? BASE64_TRUE : BASE64_FALSE, 4);
+      SCL_ERR_JRET(colDataSetVal(pOutputData, i, output, false));
+      continue;
+    }
+
+    if (IS_NUMERIC_TYPE(inputType)) {
+      if (IS_DECIMAL_TYPE(inputType)) {
+		outputSize = (outputLen - VARSTR_HEADER_SIZE) < TSDB_MAX_FIELD_LEN
+                        ? (outputLen - VARSTR_HEADER_SIZE + 1)
+                        : TSDB_MAX_FIELD_LEN;
+        uint8_t inputPrec = GET_PARAM_PRECISON(&pInput[0]), inputScale = GET_PARAM_SCALE(&pInput[0]);
+        SCL_ERR_JRET(decimalToStr(input, inputType, inputPrec, inputScale, stringBuf, outputSize));
+      } else {
+        NUM_TO_STRING(inputType, input, outputSize, stringBuf);
+      }
+      outputSize = strlen(stringBuf);
+    } else if (inputType == TSDB_DATA_TYPE_TIMESTAMP) {
+      /* The format used by MySQL in the conversion from timestamp to string */
+      char *format = "yyyy-mm-dd hh24:mi:ss";
+      SArray *formats = NULL;
+      SCL_ERR_JRET(taosTs2Char(format, &formats, *(int64_t *)input, 0, stringBuf, outputSize, pInput->tz));
+      outputSize = strlen(stringBuf);
+      taosArrayDestroy(formats);
+    } else {
+      outputSize = inputLen;
+      memcpy(stringBuf, varDataVal(input), outputSize);
+    }
+
+    VarDataLenT outputLength = base64Impl(out, stringBuf, outputSize);
+    varDataSetLen(output, outputLength);
+    SCL_ERR_JRET(colDataSetVal(pOutputData, i, output, false));
   }
 
   pOutput->numOfRows = pInput->numOfRows;
+
 _return:
-  taosMemoryFree(outputBuf);
+  taosMemoryFree(output);
+  taosMemoryFree(stringBuf);
   return code;
 }
 

tests/army/query/function/ans/base64.csv

@@ -69,3 +69,57 @@ taos> select to_base64(substring('北京涛思.com', 1, 4)) == to_base64('北京
 ===================================================================
  true                                                             |
 
+taos> select to_base64(true)
+ to_base64(true) |
+==================
+ MQ==            |
+
+taos> select to_base64(true) = to_base64(1)
+ to_base64(true) = to_base64(1) |
+=================================
+ false                          |
+
+taos> select to_base64(false)
+ to_base64(false) |
+===================
+ MA==             |
+
+taos> select to_base64(false) = to_base64(0)
+ to_base64(false) = to_base64(0) |
+==================================
+ false                           |
+
+taos> select to_base64(1234)
+ to_base64(1234) |
+==================
+ MTIzNA==        |
+
+taos> select to_base64(1234) = to_base64("1234")
+ to_base64(1234) = to_base64("1234") |
+======================================
+ true                                |
+
+taos> select to_base64(1234.567)
+ to_base64(1234.567) |
+======================
+ MTIzNC41Njc=        |
+
+taos> select to_base64(1234.567) = to_base64('1234.567')
+ to_base64(1234.567) = to_base64('1234.567') |
+==============================================
+ true                                        |
+
+taos> select to_base64(1234.567 + 0.1) = to_base64('1234.667')
+ to_base64(1234.567 + 0.1) = to_base64('1234.667') |
+====================================================
+ true                                              |
+
+taos> select to_base64(TO_TIMESTAMP("2000-01-01 00:00:00+00", "yyyy-mm-dd hh24:mi:ss")) = to_base64(TO_CHAR(TO_TIMESTAMP("2000-01-01 00:00:00+00", "yyyy-mm-dd hh24:mi:ss"), "yyyy-mm-dd hh24:mi:ss"))
+ to_base64(to_timestamp("2000-01-01 00:00:00+00", "yyyy-mm-dd hh2 |
+===================================================================
+ true                                                             |
+
+taos> select to_base64(TO_TIMESTAMP("2000-01-01 00:00:00+00", "yyyy-mm-dd hh24:mi:ss")) = to_base64("2000-01-01 00:00:00")
+ to_base64(to_timestamp("2000-01-01 00:00:00+00", "yyyy-mm-dd hh2 |
+===================================================================
+ true                                                             |

tests/army/query/function/in/base64.in

@@ -12,3 +12,14 @@ select to_base64(trim('  Hello, world!  ')) == to_base64('Hello, world!')
 select to_base64(trim('  北京涛思  '))
 select to_base64(trim('  北京涛思  ')) == to_base64('北京涛思')
 select to_base64(substring('北京涛思.com', 1, 4)) == to_base64('北京涛思')
+select to_base64(true)
+select to_base64(true) = to_base64(1)
+select to_base64(false)
+select to_base64(false) = to_base64(0)
+select to_base64(1234)
+select to_base64(1234) = to_base64("1234")
+select to_base64(1234.567)
+select to_base64(1234.567) = to_base64('1234.567')
+select to_base64(1234.567 + 0.1) = to_base64('1234.667')
+select to_base64(TO_TIMESTAMP("2000-01-01 00:00:00+00", "yyyy-mm-dd hh24:mi:ss")) = to_base64(TO_CHAR(TO_TIMESTAMP("2000-01-01 00:00:00+00", "yyyy-mm-dd hh24:mi:ss"), "yyyy-mm-dd hh24:mi:ss"))
+select to_base64(TO_TIMESTAMP("2000-01-01 00:00:00+00", "yyyy-mm-dd hh24:mi:ss")) = to_base64("2000-01-01 00:00:00")