sqlglot.optimizer.qualify_tables

  1from __future__ import annotations
  2
  3import itertools
  4import typing as t
  5
  6from sqlglot import alias, exp
  7from sqlglot.dialects.dialect import DialectType
  8from sqlglot.helper import csv_reader, name_sequence
  9from sqlglot.optimizer.scope import Scope, traverse_scope
 10from sqlglot.schema import Schema
 11
 12if t.TYPE_CHECKING:
 13    from sqlglot._typing import E
 14
 15
 16def qualify_tables(
 17    expression: E,
 18    db: t.Optional[str | exp.Identifier] = None,
 19    catalog: t.Optional[str | exp.Identifier] = None,
 20    schema: t.Optional[Schema] = None,
 21    dialect: DialectType = None,
 22) -> E:
 23    """
 24    Rewrite sqlglot AST to have fully qualified tables. Join constructs such as
 25    (t1 JOIN t2) AS t will be expanded into (SELECT * FROM t1 AS t1, t2 AS t2) AS t.
 26
 27    Examples:
 28        >>> import sqlglot
 29        >>> expression = sqlglot.parse_one("SELECT 1 FROM tbl")
 30        >>> qualify_tables(expression, db="db").sql()
 31        'SELECT 1 FROM db.tbl AS tbl'
 32        >>>
 33        >>> expression = sqlglot.parse_one("SELECT 1 FROM (t1 JOIN t2) AS t")
 34        >>> qualify_tables(expression).sql()
 35        'SELECT 1 FROM (SELECT * FROM t1 AS t1, t2 AS t2) AS t'
 36
 37    Args:
 38        expression: Expression to qualify
 39        db: Database name
 40        catalog: Catalog name
 41        schema: A schema to populate
 42        dialect: The dialect to parse catalog and schema into.
 43
 44    Returns:
 45        The qualified expression.
 46    """
 47    next_alias_name = name_sequence("_q_")
 48    db = exp.parse_identifier(db, dialect=dialect) if db else None
 49    catalog = exp.parse_identifier(catalog, dialect=dialect) if catalog else None
 50
 51    def _qualify(table: exp.Table) -> None:
 52        if isinstance(table.this, exp.Identifier):
 53            if not table.args.get("db"):
 54                table.set("db", db)
 55            if not table.args.get("catalog") and table.args.get("db"):
 56                table.set("catalog", catalog)
 57
 58    if not isinstance(expression, exp.Subqueryable):
 59        for node, *_ in expression.walk(prune=lambda n, *_: isinstance(n, exp.Unionable)):
 60            if isinstance(node, exp.Table):
 61                _qualify(node)
 62
 63    for scope in traverse_scope(expression):
 64        for derived_table in itertools.chain(scope.ctes, scope.derived_tables):
 65            if isinstance(derived_table, exp.Subquery):
 66                unnested = derived_table.unnest()
 67                if isinstance(unnested, exp.Table):
 68                    joins = unnested.args.pop("joins", None)
 69                    derived_table.this.replace(exp.select("*").from_(unnested.copy(), copy=False))
 70                    derived_table.this.set("joins", joins)
 71
 72            if not derived_table.args.get("alias"):
 73                alias_ = next_alias_name()
 74                derived_table.set("alias", exp.TableAlias(this=exp.to_identifier(alias_)))
 75                scope.rename_source(None, alias_)
 76
 77            pivots = derived_table.args.get("pivots")
 78            if pivots and not pivots[0].alias:
 79                pivots[0].set("alias", exp.TableAlias(this=exp.to_identifier(next_alias_name())))
 80
 81        for name, source in scope.sources.items():
 82            if isinstance(source, exp.Table):
 83                _qualify(source)
 84
 85                pivots = pivots = source.args.get("pivots")
 86                if not source.alias:
 87                    # Don't add the pivot's alias to the pivoted table, use the table's name instead
 88                    if pivots and pivots[0].alias == name:
 89                        name = source.name
 90
 91                    # Mutates the source by attaching an alias to it
 92                    alias(source, name or source.name or next_alias_name(), copy=False, table=True)
 93
 94                if pivots and not pivots[0].alias:
 95                    pivots[0].set(
 96                        "alias", exp.TableAlias(this=exp.to_identifier(next_alias_name()))
 97                    )
 98
 99                if schema and isinstance(source.this, exp.ReadCSV):
100                    with csv_reader(source.this) as reader:
101                        header = next(reader)
102                        columns = next(reader)
103                        schema.add_table(
104                            source,
105                            {k: type(v).__name__ for k, v in zip(header, columns)},
106                            match_depth=False,
107                        )
108            elif isinstance(source, Scope) and source.is_udtf:
109                udtf = source.expression
110                table_alias = udtf.args.get("alias") or exp.TableAlias(
111                    this=exp.to_identifier(next_alias_name())
112                )
113                udtf.set("alias", table_alias)
114
115                if not table_alias.name:
116                    table_alias.set("this", exp.to_identifier(next_alias_name()))
117                if isinstance(udtf, exp.Values) and not table_alias.columns:
118                    for i, e in enumerate(udtf.expressions[0].expressions):
119                        table_alias.append("columns", exp.to_identifier(f"_col_{i}"))
120            else:
121                for node, parent, _ in scope.walk():
122                    if (
123                        isinstance(node, exp.Table)
124                        and not node.alias
125                        and isinstance(parent, (exp.From, exp.Join))
126                    ):
127                        # Mutates the table by attaching an alias to it
128                        alias(node, node.name, copy=False, table=True)
129
130    return expression