pinot-query-runtime/src/test/resources/queries/StringFunctions.json

{
  "string_functions": {
    "tables": {
      "stringTbl": {
        "schema": [
          { "name": "strCol", "type": "STRING" }
        ],
        "inputs": [
          [ "HELLO" ],
          [ "Hello" ],
          [ "hello" ],
          [ " hello " ],
          [ "wiTH Mixed CaSe" ],
          [ "" ],
          [ " " ]
        ]
      }
    },
    "queries": [
      { "sql": "SELECT lower(strCol) FROM {stringTbl}" },
      { "sql": "SELECT upper(strCol) FROM {stringTbl}" },
      { "sql": "SELECT length(strCol) FROM {stringTbl}" },
      { "sql": "SELECT concat(strCol, strCol) FROM {stringTbl}" },
      { "sql": "SELECT concat_ws(',', strCol, strCol) FROM {stringTbl} WHERE strCol <> ''" },
      {
        "sql": "SELECT concat_ws(',', strCol, strCol) FROM {stringTbl}"
      },
      {
        "ignored": true,
        "comment": "standard sql concat is vararg, while we treat the third input as a separator",
        "sql": "SELECT concat(strCol, strCol, ',') FROM {stringTbl}"
      },
      { "sql": "SELECT strCol || strCol FROM {stringTbl}" },
      { "sql": "SELECT strCol || strCol || ',' FROM {stringTbl}"},
      { "sql": "SELECT trim(strCol) FROM {stringTbl}" },
      { "sql": "SELECT lower(strCol), regexp_Replace(strCol, 'e.*o', 'le') FROM {stringTbl}" },
      {
        "ignored": true,
        "comment": "split returns a MV array, we don't yet support that",
        "sql": "SELECT split(strCol, ' ') FROM {stringTbl}"
      }
    ]
  },
  "string_functions_noh2": {
    "tables": {
      "stringTbl": {
        "schema": [
          {
            "name": "strCol",
            "type": "STRING"
          }
        ],
        "inputs": [
          [ "Hello" ],
          [ "hello" ],
          [ "hello world" ],
          [ "world hello" ],
          [ "wiTH Mixed CaSe" ],
          [ "" ],
          [ " " ]
        ]
      }
    },
    "queries": [
      {
        "sql": "SELECT strpos(strCol, 'hello') FROM {stringTbl}",
        "outputs": [
          [-1],
          [0],
          [0],
          [6],
          [-1],
          [-1],
          [-1]
        ]
      },
      {
        "sql": "SELECT reverse(strCol) FROM {stringTbl}",
        "outputs": [
          ["olleH"],
          ["olleh"],
          ["dlrow olleh"],
          ["olleh dlrow"],
          ["eSaC dexiM HTiw"],
          [""],
          [" "]
        ]
      },
      {
        "sql": "SELECT starts_with(strCol, 'Hello') FROM {stringTbl}",
        "outputs": [
          [true],
          [false],
          [false],
          [false],
          [false],
          [false],
          [false]
        ]
      },
      {
        "sql": "SELECT ends_with(strCol, 'ello') FROM {stringTbl}",
        "outputs": [
          [true],
          [true],
          [false],
          [true],
          [false],
          [false],
          [false]
        ]
      },
      {
        "comment": "h2 only supports SUBSTR(input FROM x TO y) syntax, which we don't support",
        "sql": "SELECT substr(strCol, 0) FROM {stringTbl}",
        "outputs": [
          [ "Hello" ],
          [ "hello" ],
          [ "hello world" ],
          [ "world hello" ],
          [ "wiTH Mixed CaSe" ],
          [ "" ],
          [ " " ]
        ]
      },
      {
        "comment": "h2 only supports SUBSTRING(input FROM x TO y) syntax, which we don't support",
        "sql": "SELECT substring(strCol, 1) FROM {stringTbl}",
        "outputs": [
          [ "Hello" ],
          [ "hello" ],
          [ "hello world" ],
          [ "world hello" ],
          [ "wiTH Mixed CaSe" ],
          [ "" ],
          [ " " ]
        ]
      },
      {
        "sql": "SELECT substr(strCol, 3) FROM {stringTbl}",
        "outputs": [
          [ "lo" ],
          [ "lo" ],
          [ "lo world" ],
          [ "ld hello" ],
          [ "H Mixed CaSe" ],
          [ "" ],
          [ "" ]
        ]
      }
    ]
  }
}