Функции для работы со строками

SELECT CRC32('ClickHouse')

┌─CRC32('ClickHouse')─┐
│          1538217360 │
└─────────────────────┘

SELECT CRC32IEEE('ClickHouse');

┌─CRC32IEEE('ClickHouse')─┐
│              3089448422 │
└─────────────────────────┘

SELECT CRC64('ClickHouse');

┌──CRC64('ClickHouse')─┐
│ 12126588151325169346 │
└──────────────────────┘

SELECT appendTrailingCharIfAbsent('https://example.com', '/');

┌─appendTraili⋯.com', '/')─┐
│ https://example.com/     │
└──────────────────────────┘

SELECT ascii('234')

┌─ascii('234')─┐
│           50 │
└──────────────┘

SELECT base32Decode('IVXGG33EMVSA====');

┌─base32Decode('IVXGG33EMVSA====')─┐
│ Encoded                          │
└──────────────────────────────────┘

SELECT base32Encode('Encoded')

┌─base32Encode('Encoded')─┐
│ IVXGG33EMVSA====        │
└─────────────────────────┘

SELECT base58Decode('JxF12TrwUP45BMd');

┌─base58Decode⋯rwUP45BMd')─┐
│ Hello World              │
└──────────────────────────┘

SELECT base58Encode('ClickHouse');

┌─base58Encode('ClickHouse')─┐
│ 4nhk8K7GHXf6zx             │
└────────────────────────────┘

SELECT base64Decode('Y2xpY2tob3VzZQ==')

┌─base64Decode('Y2xpY2tob3VzZQ==')─┐
│ clickhouse                       │
└──────────────────────────────────┘

SELECT base64Encode('clickhouse')

┌─base64Encode('clickhouse')─┐
│ Y2xpY2tob3VzZQ==           │
└────────────────────────────┘

SELECT base64URLDecode('aHR0cHM6Ly9jbGlja2hvdXNlLmNvbQ')

┌─base64URLDecode('aHR0cHM6Ly9jbGlja2hvdXNlLmNvbQ')─┐
│ https://clickhouse.com                            │
└───────────────────────────────────────────────────┘

SELECT base64URLEncode('https://clickhouse.com')

┌─base64URLEncode('https://clickhouse.com')─┐
│ aHR0cHM6Ly9jbGlja2hvdXNlLmNvbQ            │
└───────────────────────────────────────────┘

SELECT 'some/long/path/to/file' AS a, basename(a)

┌─a──────────────────────┬─basename('some/long/path/to/file')─┐
│ some/long/path/to/file │ file                               │
└────────────────────────┴────────────────────────────────────┘

SELECT 'some\\long\\path\\to\\file' AS a, basename(a)

┌─a──────────────────────┬─basename('some\\long\\path\\to\\file')─┐
│ some\long\path\to\file │ file                                   │
└────────────────────────┴────────────────────────────────────────┘

SELECT 'some-file-name' AS a, basename(a)

┌─a──────────────┬─basename('some-file-name')─┐
│ some-file-name │ some-file-name             │
└────────────────┴────────────────────────────┘

SELECT byteHammingDistance('karolin', 'kathrin')

┌─byteHammingDistance('karolin', 'kathrin')─┐
│                                         3 │
└───────────────────────────────────────────┘

SELECT caseFoldUTF8('Straße')

┌─caseFoldUTF8('Straße')─┐
│ strasse                 │
└─────────────────────────┘

SELECT compareSubstrings('Saxony', 'Anglo-Saxon', 0, 6, 5) AS result

┌─result─┐
│      0 │
└────────┘

SELECT concat('Hello, ', 'World!')

┌─concat('Hello, ', 'World!')─┐
│ Hello, World!               │
└─────────────────────────────┘

SELECT concat(42, 144)

┌─concat(42, 144)─┐
│ 42144           │
└─────────────────┘

SELECT concat(key1, key2), sum(value) FROM key_val GROUP BY concatAssumeInjective(key1, key2)

┌─concat(key1, key2)─┬─sum(value)─┐
│ Hello, World!      │          3 │
│ Hello, World!      │          2 │
│ Hello, World       │          3 │
└────────────────────┴────────────┘

SELECT concatWithSeparator('a', '1', '2', '3', '4')

┌─concatWithSeparator('a', '1', '2', '3', '4')─┐
│ 1a2a3a4                                      │
└──────────────────────────────────────────────┘

CREATE TABLE user_data (
user_id UInt32,
first_name String,
last_name String,
score UInt32
)
ENGINE = MergeTree
ORDER BY tuple();

INSERT INTO user_data VALUES
(1, 'John', 'Doe', 100),
(2, 'Jane', 'Smith', 150),
(3, 'John', 'Wilson', 120),
(4, 'Jane', 'Smith', 90);

SELECT
    concatWithSeparatorAssumeInjective('-', first_name, last_name) as full_name,
    sum(score) as total_score
FROM user_data
GROUP BY concatWithSeparatorAssumeInjective('-', first_name, last_name);

┌─full_name───┬─total_score─┐
│ Jane-Smith  │         240 │
│ John-Doe    │         100 │
│ John-Wilson │         120 │
└─────────────┴─────────────┘

SELECT conv('10', 10, 2)

1010

SELECT conv('FF', 16, 10)

255

SELECT conv('-1', 10, 16)

FFFFFFFFFFFFFFFF

SELECT conv('1010', 2, 8)

12

SELECT convertCharset('Café', 'UTF-8', 'ISO-8859-1');

┌─convertChars⋯SO-8859-1')─┐
│ Caf�                     │
└──────────────────────────┘

SELECT damerauLevenshteinDistance('clickhouse', 'mouse')

┌─damerauLevenshteinDistance('clickhouse', 'mouse')─┐
│                                                 6 │
└───────────────────────────────────────────────────┘

SELECT decodeHTMLComponent('<div>Hello & "World"</div>')

┌─decodeHTMLComponent('<div>Hello & "World"</div>')─┐
│ <div>Hello & "World"</div>                                                  │
└─────────────────────────────────────────────────────────────────────────────┘

SELECT decodeXMLComponent('<tag>Hello & World</tag>')

┌─decodeXMLCom⋯;/tag>')─┐
│ <tag>Hello & World</tag> │
└──────────────────────────┘

SELECT editDistance('clickhouse', 'mouse')

┌─editDistance('clickhouse', 'mouse')─┐
│                                   6 │
└─────────────────────────────────────┘

SELECT editDistanceUTF8('我是谁', '我是我')

┌─editDistanceUTF8('我是谁', '我是我')──┐
│                                   1 │
└─────────────────────────────────────┘

SELECT
    '<tag>Hello & "World"</tag>' AS original,
    encodeXMLComponent('<tag>Hello & "World"</tag>') AS xml_encoded;

┌─original───────────────────┬─xml_encoded──────────────────────────────────────────┐
│ <tag>Hello & "World"</tag> │ &lt;tag&gt;Hello &amp; &quot;World&quot;&lt;/tag&gt; │
└────────────────────────────┴──────────────────────────────────────────────────────┘

SELECT endsWith('ClickHouse', 'House');

┌─endsWith('Cl⋯', 'House')─┐
│                        1 │
└──────────────────────────┘

SELECT endsWithCaseInsensitive('ClickHouse', 'HOUSE');

┌─endsWithCaseInsensitive('Cl⋯', 'HOUSE')─┐
│                                       1 │
└─────────────────────────────────────────┘

SELECT endsWithCaseInsensitiveUTF8('данных', 'ых');

┌─endsWithCaseInsensitiveUTF8('данных', 'ых')─┐
│                                           1 │
└─────────────────────────────────────────────┘

SELECT endsWithUTF8('данных', 'ых');

┌─endsWithUTF8('данных', 'ых')─┐
│                            1 │
└──────────────────────────────┘

SELECT extractTextFromHTML('
<html>
    <head><title>Page Title</title></head>
    <body>
        <p>Hello <b>World</b>!</p>
        <script>alert("test");</script>
        <!-- comment -->
    </body>
</html>
');

┌─extractTextFromHTML('<html><head>...')─┐
│ Page Title Hello World!                │
└────────────────────────────────────────┘

SELECT firstLine('foo\\nbar\\nbaz')

┌─firstLine('foo\nbar\nbaz')─┐
│ foo                        │
└────────────────────────────┘

SELECT idnaDecode('xn--strae-oqa.xn--mnchen-3ya.de')

┌─idnaDecode('xn--strae-oqa.xn--mnchen-3ya.de')─┐
│ straße.münchen.de                             │
└───────────────────────────────────────────────┘

SELECT idnaEncode('straße.münchen.de')

┌─idnaEncode('straße.münchen.de')─────┐
│ xn--strae-oqa.xn--mnchen-3ya.de     │
└─────────────────────────────────────┘

SELECT initcap('building for fast')

┌─initcap('building for fast')─┐
│ Building For Fast            │
└──────────────────────────────┘

SELECT initcap('John''s cat won''t eat.');

┌─initcap('Joh⋯n\'t eat.')─┐
│ John'S Cat Won'T Eat.    │
└──────────────────────────┘

SELECT initcapUTF8('не тормозит')

┌─initcapUTF8('не тормозит')─┐
│ Не Тормозит                │
└────────────────────────────┘

SELECT isValidASCII('hello') AS is_ascii, isValidASCII('你好') AS is_not_ascii

SELECT isValidUTF8('\\xc3\\xb1') AS valid, isValidUTF8('\\xc3\\x28') AS invalid

┌─valid─┬─invalid─┐
│     1 │       0 │
└───────┴─────────┘

SELECT jaroSimilarity('clickhouse', 'click')

┌─jaroSimilarity('clickhouse', 'click')─┐
│                    0.8333333333333333 │
└───────────────────────────────────────┘

SELECT jaroWinklerSimilarity('clickhouse', 'click')

┌─jaroWinklerSimilarity('clickhouse', 'click')─┐
│                           0.8999999999999999 │
└──────────────────────────────────────────────┘

SELECT left('Hello World', 5)

Hello

SELECT left('Hello World', -6)

Hello

SELECT leftPad('abc', 7, '*'), leftPad('def', 7)

┌─leftPad('abc', 7, '*')─┬─leftPad('def', 7)─┐
│ ****abc                │     def           │
└────────────────────────┴───────────────────┘

SELECT leftPadUTF8('абвг', 7, '*'), leftPadUTF8('дежз', 7)

┌─leftPadUTF8('абвг', 7, '*')─┬─leftPadUTF8('дежз', 7)─┐
│ ***абвг                     │    дежз                │
└─────────────────────────────┴────────────────────────┘

SELECT leftUTF8('Привет', 4)

Прив

SELECT leftUTF8('Привет', -4)

Пр

SELECT lengthUTF8('Здравствуй, мир!')

┌─lengthUTF8('Здравствуй, мир!')─┐
│                             16 │
└────────────────────────────────┘

SELECT lower('CLICKHOUSE')

┌─lower('CLICKHOUSE')─┐
│ clickhouse          │
└─────────────────────┘

SELECT lowerUTF8('München') as Lowerutf8;

münchen

SELECT s FROM t ORDER BY naturalSortKey(s)

┌─s───┐
│ a1  │
| a02 │
└─────┘

SELECT
'é' AS original, -- e + комбинирующий акут (U+0065 + U+0301)
length(original),
normalizeUTF8NFC('é') AS nfc_normalized, -- é (U+00E9)
length(nfc_normalized);

┌─original─┬─length(original)─┬─nfc_normalized─┬─length(nfc_normalized)─┐
│ é        │                2 │ é              │                      2 │
└──────────┴──────────────────┴────────────────┴────────────────────────┘

SELECT
    'é' AS original, -- é (U+00E9)
    length(original),
    normalizeUTF8NFD('é') AS nfd_normalized, -- e + комбинирующий акут (U+0065 + U+0301)
    length(nfd_normalized);

┌─original─┬─length(original)─┬─nfd_normalized─┬─length(nfd_normalized)─┐
│ é        │                2 │ é              │                      3 │
└──────────┴──────────────────┴────────────────┴────────────────────────┘

SELECT
    '① ② ③' AS original,                            -- Символы цифр в кружке
    normalizeUTF8NFKC('① ② ③') AS nfkc_normalized;  -- Преобразует в 1 2 3

┌─original─┬─nfkc_normalized─┐
│ ① ② ③  │ 1 2 3           │
└──────────┴─────────────────┘

SELECT
    'Ä ① Hello' AS original,
    normalizeUTF8NFKCCasefold('Ä ① Hello') AS nfkc_cf_normalized;

┌─original───┬─nfkc_cf_normalized─┐
│ Ä ① Hello │ ä 1 hello           │
└────────────┴────────────────────┘

SELECT
    'H₂O²' AS original,                            -- H + нижний индекс 2 + O + верхний индекс 2
    normalizeUTF8NFKD('H₂O²') AS nfkd_normalized;  -- Преобразует в H 2 O 2

┌─original─┬─nfkd_normalized─┐
│ H₂O²     │ H2O2            │
└──────────┴─────────────────┘

SELECT punycodeDecode('Mnchen-3ya')

┌─punycodeDecode('Mnchen-3ya')─┐
│ München                      │
└──────────────────────────────┘

SELECT punycodeEncode('München')

┌─punycodeEncode('München')─┐
│ Mnchen-3ya                │
└───────────────────────────┘

SELECT
    regexpExtract('100-200', '(\\d+)-(\\d+)', 1),
    regexpExtract('100-200', '(\\d+)-(\\d+)', 2),
    regexpExtract('100-200', '(\\d+)-(\\d+)', 0),
    regexpExtract('100-200', '(\\d+)-(\\d+)'),
    regexpExtract('100-200', '\\d+');

┌─regexpExtract('100-200', '(\\d+)-(\\d+)', 1)─┬─regexpExtract('100-200', '(\\d+)-(\\d+)', 2)─┬─regexpExtract('100-200', '(\\d+)-(\\d+)', 0)─┬─regexpExtract('100-200', '(\\d+)-(\\d+)')─┬─regexpExtract('100-200', '\\d+')─┐
│ 100                                          │ 200                                          │ 100-200                                      │ 100                                       │ 100                              │
└──────────────────────────────────────────────┴──────────────────────────────────────────────┴──────────────────────────────────────────────┴───────────────────────────────────────────┴──────────────────────────────────┘

SELECT
    regexpPosition('hello world', 'world'),
    regexpPosition('aXbXcXd', 'X', 1, 2),
    regexpPosition('aXbXcXd', 'X', 1, 2, 1),
    regexpPosition('Hello WORLD', 'world', 1, 1, 0, 'i'),
    regexpPosition('foo123bar456', '([a-z]+)([0-9]+)', 1, 2, 0, '', 2);

┌─...─┬─...─┬─...─┬─...─┬─...─┐
│   7 │   4 │   5 │   7 │  10 │
└─────┴─────┴─────┴─────┴─────┘