sqlglot.dialects.snowflake

 17class Snowflake(Dialect):
 18    # https://docs.snowflake.com/en/sql-reference/identifiers-syntax
 19    NORMALIZATION_STRATEGY = NormalizationStrategy.UPPERCASE
 20    # https://docs.snowflake.com/en/sql-reference/data-types-text#escape-sequences
 21    UNESCAPED_SEQUENCES = {"\\a": "a", "\\v": "v"}
 22    NULL_ORDERING = "nulls_are_large"
 23    TIME_FORMAT = "'YYYY-MM-DD HH24:MI:SS'"
 24    SUPPORTS_USER_DEFINED_TYPES = False
 25    PREFER_CTE_ALIAS_COLUMN = True
 26    TABLESAMPLE_SIZE_IS_PERCENT = True
 27    COPY_PARAMS_ARE_CSV = False
 28    ARRAY_AGG_INCLUDES_NULLS = None
 29    ARRAY_FUNCS_PROPAGATES_NULLS = True
 30    ALTER_TABLE_ADD_REQUIRED_FOR_EACH_COLUMN = False
 31    TRY_CAST_REQUIRES_STRING = True
 32    SUPPORTS_ALIAS_REFS_IN_JOIN_CONDITIONS = True
 33    LEAST_GREATEST_IGNORES_NULLS = False
 34    UUID_IS_STRING_TYPE = True
 35    STAR_ILIKE_BACKSLASH_ESCAPE = True
 36
 37    EXPRESSION_METADATA = EXPRESSION_METADATA.copy()
 38
 39    # https://docs.snowflake.com/en/en/sql-reference/functions/initcap
 40    INITCAP_DEFAULT_DELIMITER_CHARS = ' \t\n\r\f\v!?@"^#$&~_,.:;+\\-*%/|\\[\\](){}<>'
 41
 42    INVERSE_TIME_MAPPING = {
 43        "T": "T",  # in TIME_MAPPING we map '"T"' with the double quotes to 'T', and we want to prevent 'T' from being mapped back to '"T"' so that 'AUTO' doesn't become 'AU"T"O'
 44    }
 45
 46    TIME_MAPPING = {
 47        "YYYY": "%Y",
 48        "yyyy": "%Y",
 49        "YY": "%y",
 50        "yy": "%y",
 51        "MMMM": "%B",
 52        "mmmm": "%B",
 53        "MON": "%b",
 54        "mon": "%b",
 55        "MM": "%m",
 56        "mm": "%m",
 57        "DD": "%d",
 58        "dd": "%-d",
 59        "DY": "%a",
 60        "dy": "%w",
 61        "HH24": "%H",
 62        "hh24": "%H",
 63        "HH12": "%I",
 64        "hh12": "%I",
 65        "MI": "%M",
 66        "mi": "%M",
 67        "SS": "%S",
 68        "ss": "%S",
 69        "FF": "%f_nine",  # %f_ internal representation with precision specified
 70        "ff": "%f_nine",
 71        "FF0": "%f_zero",
 72        "ff0": "%f_zero",
 73        "FF1": "%f_one",
 74        "ff1": "%f_one",
 75        "FF2": "%f_two",
 76        "ff2": "%f_two",
 77        "FF3": "%f_three",
 78        "ff3": "%f_three",
 79        "FF4": "%f_four",
 80        "ff4": "%f_four",
 81        "FF5": "%f_five",
 82        "ff5": "%f_five",
 83        "FF6": "%f",
 84        "ff6": "%f",
 85        "FF7": "%f_seven",
 86        "ff7": "%f_seven",
 87        "FF8": "%f_eight",
 88        "ff8": "%f_eight",
 89        "FF9": "%f_nine",
 90        "ff9": "%f_nine",
 91        "TZHTZM": "%z",
 92        "tzhtzm": "%z",
 93        "TZH:TZM": "%:z",  # internal representation for ±HH:MM
 94        "tzh:tzm": "%:z",
 95        "TZH": "%-z",  # internal representation ±HH
 96        "tzh": "%-z",
 97        '"T"': "T",  # remove the optional double quotes around the separator between the date and time
 98        # Seems like Snowflake treats AM/PM in the format string as equivalent,
 99        # only the time (stamp) value's AM/PM affects the output
100        "AM": "%p",
101        "am": "%p",
102        "PM": "%p",
103        "pm": "%p",
104    }
105
106    DATE_PART_MAPPING = {
107        **Dialect.DATE_PART_MAPPING,
108        "ISOWEEK": "WEEKISO",
109        # The base Dialect maps EPOCH_SECOND -> EPOCH, but we need to preserve
110        # EPOCH_SECOND as a distinct value for two reasons:
111        # 1. Type annotation: EPOCH_SECOND returns BIGINT, while EPOCH returns DOUBLE
112        # 2. Transpilation: DuckDB's EPOCH() returns float, so we cast EPOCH_SECOND
113        #    to BIGINT to match Snowflake's integer behavior
114        # Without this override, EXTRACT(EPOCH_SECOND FROM ts) would be normalized
115        # to EXTRACT(EPOCH FROM ts) and lose the integer semantics.
116        "EPOCH_SECOND": "EPOCH_SECOND",
117        "EPOCH_SECONDS": "EPOCH_SECOND",
118    }
119
120    PSEUDOCOLUMNS = {"LEVEL"}
121
122    def can_quote(self, identifier: exp.Identifier, identify: str | bool = "safe") -> bool:
123        # This disables quoting DUAL in SELECT ... FROM DUAL, because Snowflake treats an
124        # unquoted DUAL keyword in a special way and does not map it to a user-defined table
125        return super().can_quote(identifier, identify) and not (
126            isinstance(identifier.parent, exp.Table)
127            and not identifier.quoted
128            and identifier.name.lower() == "dual"
129        )
130
131    class JSONPathTokenizer(jsonpath.JSONPathTokenizer):
132        SINGLE_TOKENS = jsonpath.JSONPathTokenizer.SINGLE_TOKENS.copy()
133        SINGLE_TOKENS.pop("$")
134
135    Parser = SnowflakeParser
136
137    class Tokenizer(tokens.Tokenizer):
138        STRING_ESCAPES = ["\\", "'"]
139        HEX_STRINGS = [("x'", "'"), ("X'", "'")]
140        RAW_STRINGS = ["$$"]
141        COMMENTS = ["--", "//", ("/*", "*/")]
142        NESTED_COMMENTS = False
143
144        KEYWORDS = {
145            **tokens.Tokenizer.KEYWORDS,
146            "BYTEINT": TokenType.INT,
147            "FILE://": TokenType.URI_START,
148            "FILE FORMAT": TokenType.FILE_FORMAT,
149            "GET": TokenType.GET,
150            "INTEGRATION": TokenType.INTEGRATION,
151            "MATCH_CONDITION": TokenType.MATCH_CONDITION,
152            "MATCH_RECOGNIZE": TokenType.MATCH_RECOGNIZE,
153            "MINUS": TokenType.EXCEPT,
154            "NCHAR VARYING": TokenType.VARCHAR,
155            "PACKAGE": TokenType.PACKAGE,
156            "POLICY": TokenType.POLICY,
157            "POOL": TokenType.POOL,
158            "PUT": TokenType.PUT,
159            "UNDROP": TokenType.UNDROP,
160            "REMOVE": TokenType.COMMAND,
161            "RM": TokenType.COMMAND,
162            "ROLE": TokenType.ROLE,
163            "RULE": TokenType.RULE,
164            "SAMPLE": TokenType.TABLE_SAMPLE,
165            "SEMANTIC VIEW": TokenType.SEMANTIC_VIEW,
166            "SQL_DOUBLE": TokenType.DOUBLE,
167            "SQL_VARCHAR": TokenType.VARCHAR,
168            "STAGE": TokenType.STAGE,
169            "STORAGE INTEGRATION": TokenType.STORAGE_INTEGRATION,
170            "STREAMLIT": TokenType.STREAMLIT,
171            "TAG": TokenType.TAG,
172            "TIMESTAMP_TZ": TokenType.TIMESTAMPTZ,
173            "TOP": TokenType.TOP,
174            "VOLUME": TokenType.VOLUME,
175            "WAREHOUSE": TokenType.WAREHOUSE,
176            # https://docs.snowflake.com/en/sql-reference/data-types-numeric#float
177            # FLOAT is a synonym for DOUBLE in Snowflake
178            "FLOAT": TokenType.DOUBLE,
179        }
180        KEYWORDS.pop("/*+")
181
182        SINGLE_TOKENS = {
183            **tokens.Tokenizer.SINGLE_TOKENS,
184            "$": TokenType.PARAMETER,
185            "!": TokenType.EXCLAMATION,
186        }
187
188        VAR_SINGLE_TOKENS = {"$"}
189
190        COMMANDS = tokens.Tokenizer.COMMANDS - {TokenType.SHOW}
191
192    Generator = SnowflakeGenerator

131    class JSONPathTokenizer(jsonpath.JSONPathTokenizer):
132        SINGLE_TOKENS = jsonpath.JSONPathTokenizer.SINGLE_TOKENS.copy()
133        SINGLE_TOKENS.pop("$")

137    class Tokenizer(tokens.Tokenizer):
138        STRING_ESCAPES = ["\\", "'"]
139        HEX_STRINGS = [("x'", "'"), ("X'", "'")]
140        RAW_STRINGS = ["$$"]
141        COMMENTS = ["--", "//", ("/*", "*/")]
142        NESTED_COMMENTS = False
143
144        KEYWORDS = {
145            **tokens.Tokenizer.KEYWORDS,
146            "BYTEINT": TokenType.INT,
147            "FILE://": TokenType.URI_START,
148            "FILE FORMAT": TokenType.FILE_FORMAT,
149            "GET": TokenType.GET,
150            "INTEGRATION": TokenType.INTEGRATION,
151            "MATCH_CONDITION": TokenType.MATCH_CONDITION,
152            "MATCH_RECOGNIZE": TokenType.MATCH_RECOGNIZE,
153            "MINUS": TokenType.EXCEPT,
154            "NCHAR VARYING": TokenType.VARCHAR,
155            "PACKAGE": TokenType.PACKAGE,
156            "POLICY": TokenType.POLICY,
157            "POOL": TokenType.POOL,
158            "PUT": TokenType.PUT,
159            "UNDROP": TokenType.UNDROP,
160            "REMOVE": TokenType.COMMAND,
161            "RM": TokenType.COMMAND,
162            "ROLE": TokenType.ROLE,
163            "RULE": TokenType.RULE,
164            "SAMPLE": TokenType.TABLE_SAMPLE,
165            "SEMANTIC VIEW": TokenType.SEMANTIC_VIEW,
166            "SQL_DOUBLE": TokenType.DOUBLE,
167            "SQL_VARCHAR": TokenType.VARCHAR,
168            "STAGE": TokenType.STAGE,
169            "STORAGE INTEGRATION": TokenType.STORAGE_INTEGRATION,
170            "STREAMLIT": TokenType.STREAMLIT,
171            "TAG": TokenType.TAG,
172            "TIMESTAMP_TZ": TokenType.TIMESTAMPTZ,
173            "TOP": TokenType.TOP,
174            "VOLUME": TokenType.VOLUME,
175            "WAREHOUSE": TokenType.WAREHOUSE,
176            # https://docs.snowflake.com/en/sql-reference/data-types-numeric#float
177            # FLOAT is a synonym for DOUBLE in Snowflake
178            "FLOAT": TokenType.DOUBLE,
179        }
180        KEYWORDS.pop("/*+")
181
182        SINGLE_TOKENS = {
183            **tokens.Tokenizer.SINGLE_TOKENS,
184            "$": TokenType.PARAMETER,
185            "!": TokenType.EXCLAMATION,
186        }
187
188        VAR_SINGLE_TOKENS = {"$"}
189
190        COMMANDS = tokens.Tokenizer.COMMANDS - {TokenType.SHOW}