1"""sqlglot expressions - string, encoding, hashing, and regex functions."""
  2
  3from __future__ import annotations
  4
  5from sqlglot.expressions.core import Expression, Func, Binary
  6
  7
  8# String basics
  9
 10
 11class Ascii(Expression, Func):
 12    pass
 13
 14
 15class BitLength(Expression, Func):
 16    pass
 17
 18
 19class ByteLength(Expression, Func):
 20    pass
 21
 22
 23class Chr(Expression, Func):
 24    arg_types = {"expressions": True, "charset": False}
 25    is_var_len_args = True
 26    _sql_names = ["CHR", "CHAR"]
 27
 28
 29class Concat(Expression, Func):
 30    arg_types = {"expressions": True, "safe": False, "coalesce": False}
 31    is_var_len_args = True
 32
 33
 34class ConcatWs(Concat):
 35    _sql_names = ["CONCAT_WS"]
 36
 37
 38class Contains(Expression, Func):
 39    arg_types = {"this": True, "expression": True, "json_scope": False}
 40
 41
 42class Elt(Expression, Func):
 43    arg_types = {"this": True, "expressions": True}
 44    is_var_len_args = True
 45
 46
 47class EndsWith(Expression, Func):
 48    _sql_names = ["ENDS_WITH", "ENDSWITH"]
 49    arg_types = {"this": True, "expression": True}
 50
 51
 52class Format(Expression, Func):
 53    arg_types = {"this": True, "expressions": False}
 54    is_var_len_args = True
 55
 56
 57class Initcap(Expression, Func):
 58    arg_types = {"this": True, "expression": False}
 59
 60
 61class IsAscii(Expression, Func):
 62    pass
 63
 64
 65class Left(Expression, Func):
 66    arg_types = {"this": True, "expression": True, "negative_length_returns_empty": False}
 67
 68
 69class Length(Expression, Func):
 70    arg_types = {"this": True, "binary": False, "encoding": False}
 71    _sql_names = ["LENGTH", "LEN", "CHAR_LENGTH", "CHARACTER_LENGTH"]
 72
 73
 74class Levenshtein(Expression, Func):
 75    arg_types = {
 76        "this": True,
 77        "expression": False,
 78        "ins_cost": False,
 79        "del_cost": False,
 80        "sub_cost": False,
 81        "max_dist": False,
 82    }
 83
 84
 85class Lower(Expression, Func):
 86    _sql_names = ["LOWER", "LCASE"]
 87
 88
 89class MatchAgainst(Expression, Func):
 90    arg_types = {"this": True, "expressions": True, "modifier": False}
 91
 92
 93class Normalize(Expression, Func):
 94    arg_types = {"this": True, "form": False, "is_casefold": False}
 95
 96
 97class NumberToStr(Expression, Func):
 98    arg_types = {"this": True, "format": True, "culture": False}
 99
100
101class Overlay(Expression, Func):
102    arg_types = {"this": True, "expression": True, "from_": True, "for_": False}
103
104
105class Pad(Expression, Func):
106    arg_types = {"this": True, "expression": True, "fill_pattern": False, "is_left": True}
107
108
109class Repeat(Expression, Func):
110    arg_types = {"this": True, "times": True}
111
112
113class Replace(Expression, Func):
114    arg_types = {"this": True, "expression": True, "replacement": False}
115
116
117class Reverse(Expression, Func):
118    pass
119
120
121class Right(Expression, Func):
122    arg_types = {"this": True, "expression": True, "negative_length_returns_empty": False}
123
124
125class RtrimmedLength(Expression, Func):
126    pass
127
128
129class Search(Expression, Func):
130    arg_types = {
131        "this": True,  # data_to_search / search_data
132        "expression": True,  # search_query / search_string
133        "json_scope": False,  # BigQuery: JSON_VALUES | JSON_KEYS | JSON_KEYS_AND_VALUES
134        "analyzer": False,  # Both: analyzer / ANALYZER
135        "analyzer_options": False,  # BigQuery: analyzer_options_values
136        "search_mode": False,  # Snowflake: OR | AND
137    }
138
139
140class SearchIp(Expression, Func):
141    arg_types = {"this": True, "expression": True}
142
143
144class Soundex(Expression, Func):
145    pass
146
147
148class SoundexP123(Expression, Func):
149    pass
150
151
152class Space(Expression, Func):
153    """
154    SPACE(n) → string consisting of n blank characters
155    """
156
157    pass
158
159
160class Split(Expression, Func):
161    arg_types = {
162        "this": True,
163        "expression": True,
164        "limit": False,
165        "null_returns_null": False,
166        "empty_delimiter_returns_whole": False,
167    }
168
169
170class SplitPart(Expression, Func):
171    arg_types = {
172        "this": True,
173        "delimiter": False,
174        "part_index": False,
175        "part_index_zero_as_one": False,  # usually part_index is 1-based, however Snowflake allows 0 and treats it as 1
176        "empty_delimiter_returns_whole": False,  # whether the whole input string should be returned if the delimiter string is empty (i.e. Snowflake)
177    }
178
179
180class Strtok(Expression, Func):
181    arg_types = {
182        "this": True,
183        "delimiter": False,
184        "part_index": False,
185    }
186
187
188class StartsWith(Expression, Func):
189    _sql_names = ["STARTS_WITH", "STARTSWITH"]
190    arg_types = {"this": True, "expression": True}
191
192
193class StrPosition(Expression, Func):
194    arg_types = {
195        "this": True,
196        "substr": True,
197        "position": False,
198        "occurrence": False,
199        "clamp_position": False,
200    }
201
202
203class StrToMap(Expression, Func):
204    arg_types = {
205        "this": True,
206        "pair_delim": False,
207        "key_value_delim": False,
208        "duplicate_resolution_callback": False,
209    }
210
211
212class String(Expression, Func):
213    arg_types = {"this": True, "zone": False}
214
215
216class Stuff(Expression, Func):
217    _sql_names = ["STUFF", "INSERT"]
218    arg_types = {"this": True, "start": True, "length": True, "expression": True}
219
220
221class Substring(Expression, Func):
222    _sql_names = ["SUBSTRING", "SUBSTR"]
223    arg_types = {"this": True, "start": False, "length": False, "zero_start": False}
224
225
226class SubstringIndex(Expression, Func):
227    """
228    SUBSTRING_INDEX(str, delim, count)
229
230    *count* > 0  → left slice before the *count*-th delimiter
231    *count* < 0  → right slice after the |count|-th delimiter
232    """
233
234    arg_types = {"this": True, "delimiter": True, "count": True}
235
236
237class Translate(Expression, Func):
238    arg_types = {"this": True, "from_": True, "to": True}
239
240
241class Trim(Expression, Func):
242    arg_types = {
243        "this": True,
244        "expression": False,
245        "position": False,
246        "collation": False,
247    }
248
249
250class Unicode(Expression, Func):
251    arg_types = {"this": True, "empty_is_zero": False}
252
253
254class Upper(Expression, Func):
255    _sql_names = ["UPPER", "UCASE"]
256
257
258# Encoding / base conversion
259
260
261class Base64DecodeBinary(Expression, Func):
262    arg_types = {"this": True, "alphabet": False}
263
264
265class Base64DecodeString(Expression, Func):
266    arg_types = {"this": True, "alphabet": False}
267
268
269class Base64Encode(Expression, Func):
270    arg_types = {"this": True, "max_line_length": False, "alphabet": False}
271
272
273class CodePointsToBytes(Expression, Func):
274    pass
275
276
277class CodePointsToString(Expression, Func):
278    pass
279
280
281class ConvertToCharset(Expression, Func):
282    arg_types = {"this": True, "dest": True, "source": False}
283
284
285class Decode(Expression, Func):
286    arg_types = {"this": True, "charset": True, "replace": False}
287
288
289class Encode(Expression, Func):
290    arg_types = {"this": True, "charset": True}
291
292
293class FromBase(Expression, Func):
294    arg_types = {"this": True, "expression": True}
295
296
297class FromBase32(Expression, Func):
298    pass
299
300
301class FromBase64(Expression, Func):
302    pass
303
304
305class Hex(Expression, Func):
306    arg_types = {"this": True, "case": False}
307
308
309class HexDecodeString(Expression, Func):
310    pass
311
312
313class LowerHex(Hex):
314    pass
315
316
317class SafeConvertBytesToString(Expression, Func):
318    pass
319
320
321class ToBase32(Expression, Func):
322    pass
323
324
325class ToBase64(Expression, Func):
326    pass
327
328
329class ToBinary(Expression, Func):
330    arg_types = {"this": True, "format": False, "safe": False}
331
332
333class ToChar(Expression, Func):
334    arg_types = {
335        "this": True,
336        "format": False,
337        "nlsparam": False,
338        "is_numeric": False,
339    }
340
341
342class ToCodePoints(Expression, Func):
343    pass
344
345
346class ToDecfloat(Expression, Func):
347    arg_types = {
348        "this": True,
349        "format": False,
350    }
351
352
353class ToDouble(Expression, Func):
354    arg_types = {
355        "this": True,
356        "format": False,
357        "safe": False,
358    }
359
360
361class ToFile(Expression, Func):
362    arg_types = {
363        "this": True,
364        "path": False,
365        "safe": False,
366    }
367
368
369class ToNumber(Expression, Func):
370    arg_types = {
371        "this": True,
372        "format": False,
373        "nlsparam": False,
374        "precision": False,
375        "scale": False,
376        "safe": False,
377        "safe_name": False,
378    }
379
380
381class TryBase64DecodeBinary(Expression, Func):
382    arg_types = {"this": True, "alphabet": False}
383
384
385class TryBase64DecodeString(Expression, Func):
386    arg_types = {"this": True, "alphabet": False}
387
388
389class TryHexDecodeBinary(Expression, Func):
390    pass
391
392
393class TryHexDecodeString(Expression, Func):
394    pass
395
396
397class TryToDecfloat(Expression, Func):
398    arg_types = {
399        "this": True,
400        "format": False,
401    }
402
403
404class Unhex(Expression, Func):
405    arg_types = {"this": True, "expression": False}
406
407
408# Regex
409
410
411class RegexpCount(Expression, Func):
412    arg_types = {
413        "this": True,
414        "expression": True,
415        "position": False,
416        "parameters": False,
417    }
418
419
420class RegexpExtract(Expression, Func):
421    arg_types = {
422        "this": True,
423        "expression": True,
424        "position": False,
425        "occurrence": False,
426        "parameters": False,
427        "group": False,
428        "null_if_pos_overflow": False,  # for transpilation target behavior
429    }
430
431
432class RegexpExtractAll(Expression, Func):
433    arg_types = {
434        "this": True,
435        "expression": True,
436        "group": False,
437        "parameters": False,
438        "position": False,
439        "occurrence": False,
440    }
441
442
443class RegexpFullMatch(Expression, Binary, Func):
444    arg_types = {"this": True, "expression": True, "options": False}
445
446
447class RegexpILike(Expression, Binary, Func):
448    arg_types = {"this": True, "expression": True, "flag": False}
449
450
451class RegexpInstr(Expression, Func):
452    arg_types = {
453        "this": True,
454        "expression": True,
455        "position": False,
456        "occurrence": False,
457        "option": False,
458        "parameters": False,
459        "group": False,
460    }
461
462
463class RegexpReplace(Expression, Func):
464    arg_types = {
465        "this": True,
466        "expression": True,
467        "replacement": False,
468        "position": False,
469        "occurrence": False,
470        "modifiers": False,
471        "single_replace": False,
472    }
473
474
475class RegexpSplit(Expression, Func):
476    arg_types = {"this": True, "expression": True, "limit": False}
477
478
479# Hashing / cryptographic
480
481
482class Compress(Expression, Func):
483    arg_types = {"this": True, "method": False}
484
485
486class Decrypt(Expression, Func):
487    arg_types = {
488        "this": True,
489        "passphrase": True,
490        "aad": False,
491        "encryption_method": False,
492        "safe": False,
493    }
494
495
496class DecryptRaw(Expression, Func):
497    arg_types = {
498        "this": True,
499        "key": True,
500        "iv": True,
501        "aad": False,
502        "encryption_method": False,
503        "aead": False,
504        "safe": False,
505    }
506
507
508class DecompressBinary(Expression, Func):
509    arg_types = {"this": True, "method": True}
510
511
512class DecompressString(Expression, Func):
513    arg_types = {"this": True, "method": True}
514
515
516class Encrypt(Expression, Func):
517    arg_types = {"this": True, "passphrase": True, "aad": False, "encryption_method": False}
518
519
520class EncryptRaw(Expression, Func):
521    arg_types = {"this": True, "key": True, "iv": True, "aad": False, "encryption_method": False}
522
523
524class CityHash64(Expression, Func):
525    arg_types = {"expressions": False}
526    is_var_len_args = True
527
528
529class FarmFingerprint(Expression, Func):
530    arg_types = {"expressions": True}
531    is_var_len_args = True
532    _sql_names = ["FARM_FINGERPRINT", "FARMFINGERPRINT64"]
533
534
535class MD5(Expression, Func):
536    _sql_names = ["MD5"]
537
538
539class MD5Digest(Expression, Func):
540    arg_types = {"this": True, "expressions": False}
541    is_var_len_args = True
542    _sql_names = ["MD5_DIGEST"]
543
544
545class MD5NumberLower64(Expression, Func):
546    pass
547
548
549class MD5NumberUpper64(Expression, Func):
550    pass
551
552
553class SHA(Expression, Func):
554    _sql_names = ["SHA", "SHA1"]
555
556
557class SHA1Digest(Expression, Func):
558    pass
559
560
561class SHA2(Expression, Func):
562    _sql_names = ["SHA2"]
563    arg_types = {"this": True, "length": False}
564
565
566class SHA2Digest(Expression, Func):
567    arg_types = {"this": True, "length": False}
568
569
570class StandardHash(Expression, Func):
571    arg_types = {"this": True, "expression": False}
572
573
574# Parse
575
576
577class ParseBignumeric(Expression, Func):
578    pass
579
580
581class ParseNumeric(Expression, Func):
582    pass
583
584
585class ParseUrl(Expression, Func):
586    arg_types = {"this": True, "part_to_extract": False, "key": False, "permissive": False}