|
| 1 | +import typing as t |
| 2 | +import logging |
| 3 | + |
| 4 | +logger = logging.getLogger(__name__) |
| 5 | + |
| 6 | + |
| 7 | +def to_sqlmesh(dbt_select: t.Collection[str], dbt_exclude: t.Collection[str]) -> t.Optional[str]: |
| 8 | + """ |
| 9 | + Given selectors defined in the format of the dbt cli --select and --exclude arguments, convert them into a selector expression that |
| 10 | + the SQLMesh selector engine can understand. |
| 11 | +
|
| 12 | + The main things being mapped are: |
| 13 | + - set union (" " between items within the same selector string OR multiple --select arguments) is mapped to " | " |
| 14 | + - set intersection ("," between items within the same selector string) is mapped to " & " |
| 15 | + - `--exclude`. The SQLMesh selector engine does not treat this as a separate parameter and rather treats exclusion as a normal selector |
| 16 | + that just happens to contain negation syntax, so we generate these by negating each expression and then intersecting the result |
| 17 | + with any --select expressions |
| 18 | +
|
| 19 | + Things that are *not* currently being mapped include: |
| 20 | + - selectors based on file paths |
| 21 | + - selectors based on partially qualified names like "model_a". The SQLMesh selector engine requires either: |
| 22 | + - wildcards, eg "*model_a*" |
| 23 | + - the full model name qualified with the schema, eg "staging.model_a" |
| 24 | +
|
| 25 | + Examples: |
| 26 | + --select "model_a" |
| 27 | + -> "model_a" |
| 28 | + --select "main.model_a" |
| 29 | + -> "main.model_a" |
| 30 | + --select "main.model_a" --select "main.model_b" |
| 31 | + -> "main.model_a | main.model_b" |
| 32 | + --select "main.model_a main.model_b" |
| 33 | + -> "main.model_a | main.model_b" |
| 34 | + --select "(main.model_a+ & ^main.model_b)" |
| 35 | + -> "(main.model_a+ & ^main.model_b)" |
| 36 | + --select "+main.model_a" --exclude "raw.src_data" |
| 37 | + -> "+main.model_a & ^(raw.src_data)" |
| 38 | + --select "+main.model_a" --select "main.*b+" --exclude "raw.src_data" |
| 39 | + -> "(+main.model_a | main.*b+) & ^(raw.src_data)" |
| 40 | + """ |
| 41 | + if not dbt_select and not dbt_exclude: |
| 42 | + return None |
| 43 | + |
| 44 | + select_expr = " | ".join(_to_sqlmesh(expr) for expr in dbt_select) |
| 45 | + select_expr = _wrap(select_expr) if dbt_exclude and len(dbt_select) > 1 else select_expr |
| 46 | + |
| 47 | + exclude_expr = " | ".join(_to_sqlmesh(expr, negate=True) for expr in dbt_exclude) |
| 48 | + exclude_expr = _wrap(exclude_expr) if dbt_select and len(dbt_exclude) > 1 else exclude_expr |
| 49 | + |
| 50 | + main_expr = " & ".join([expr for expr in [select_expr, exclude_expr] if expr]) |
| 51 | + |
| 52 | + logger.debug( |
| 53 | + f"Expanded dbt select: {dbt_select}, exclude: {dbt_exclude} into SQLMesh: {main_expr}" |
| 54 | + ) |
| 55 | + |
| 56 | + return main_expr |
| 57 | + |
| 58 | + |
| 59 | +def _to_sqlmesh(selector_str: str, negate: bool = False) -> str: |
| 60 | + unions, intersections = _split_unions_and_intersections(selector_str) |
| 61 | + |
| 62 | + if negate: |
| 63 | + unions = [_negate(u) for u in unions] |
| 64 | + intersections = [_negate(i) for i in intersections] |
| 65 | + |
| 66 | + union_expr = " | ".join(unions) |
| 67 | + intersection_expr = " & ".join(intersections) |
| 68 | + |
| 69 | + if len(unions) > 1 and intersections: |
| 70 | + union_expr = f"({union_expr})" |
| 71 | + |
| 72 | + if len(intersections) > 1 and unions: |
| 73 | + intersection_expr = f"({intersection_expr})" |
| 74 | + |
| 75 | + return " | ".join([expr for expr in [union_expr, intersection_expr] if expr]) |
| 76 | + |
| 77 | + |
| 78 | +def _split_unions_and_intersections(selector_str: str) -> t.Tuple[t.List[str], t.List[str]]: |
| 79 | + # break space-separated items like: "my_first_model my_second_model" into a list of selectors to union |
| 80 | + # and comma-separated items like: "my_first_model,my_second_model" into a list of selectors to intersect |
| 81 | + # but, take into account brackets, eg "(my_first_model & my_second_model)" should not be split |
| 82 | + |
| 83 | + def _split_by(input: str, delimiter: str) -> t.Iterator[str]: |
| 84 | + buf = "" |
| 85 | + depth = 0 |
| 86 | + |
| 87 | + for char in input: |
| 88 | + if char == delimiter and depth <= 0: |
| 89 | + # only split on a space if we are not within parenthesis |
| 90 | + yield buf |
| 91 | + buf = "" |
| 92 | + continue |
| 93 | + elif char == "(": |
| 94 | + depth += 1 |
| 95 | + elif char == ")": |
| 96 | + depth -= 1 |
| 97 | + |
| 98 | + buf += char |
| 99 | + |
| 100 | + if buf: |
| 101 | + yield buf |
| 102 | + |
| 103 | + # first, break up based on spaces |
| 104 | + segments = list(_split_by(selector_str, " ")) |
| 105 | + |
| 106 | + # then, within each segment, identify the unions and intersections |
| 107 | + unions = [] |
| 108 | + intersections = [] |
| 109 | + |
| 110 | + for segment in segments: |
| 111 | + maybe_intersections = list(_split_by(segment, ",")) |
| 112 | + if len(maybe_intersections) > 1: |
| 113 | + intersections.extend(maybe_intersections) |
| 114 | + else: |
| 115 | + unions.append(segment) |
| 116 | + |
| 117 | + return unions, intersections |
| 118 | + |
| 119 | + |
| 120 | +def _negate(expr: str) -> str: |
| 121 | + return f"^{_wrap(expr)}" |
| 122 | + |
| 123 | + |
| 124 | +def _wrap(expr: str) -> str: |
| 125 | + already_wrapped = expr.strip().startswith("(") and expr.strip().endswith(")") |
| 126 | + |
| 127 | + if expr and not already_wrapped: |
| 128 | + return f"({expr})" |
| 129 | + |
| 130 | + return expr |
0 commit comments