def analyze(self, stmt: Statement) -> LineageResult:
if stmt.get_type() == "DROP":
self._extract_from_DDL_DROP(stmt)
elif stmt.get_type() == "ALTER":
self._extract_from_DDL_ALTER(stmt)
elif (stmt.get_type() == "DELETE"
or stmt.token_first(skip_cm=True).normalized == "TRUNCATE"
or stmt.token_first(skip_cm=True).normalized.upper() == "REFRESH"
or stmt.token_first(skip_cm=True).normalized == "CACHE"):
pass
else:
# DML parsing logic also applies to CREATE DDL
self._extract_from_DML(stmt)
return self._lineage_result
def analyze(self, stmt: Statement) -> LineageResult:
"""
to analyze the Statement and store the result into :class:`LineageResult`.
:param stmt: a SQL statement parsed by `sqlparse`
"""
if stmt.get_type() == "DROP":
self._extract_from_ddl_drop(stmt)
elif stmt.get_type() == "ALTER":
self._extract_from_ddl_alter(stmt)
elif (stmt.get_type() == "DELETE"
or stmt.token_first(skip_cm=True).normalized == "TRUNCATE"
or stmt.token_first(skip_cm=True).normalized.upper() == "REFRESH"
or stmt.token_first(skip_cm=True).normalized == "CACHE"):
pass
else:
# DML parsing logic also applies to CREATE DDL
self._extract_from_dml(stmt)
return self._lineage_result
def from_statement(cls, statement: token_groups.Statement) -> SQLQuery:
"""Extract an SQLQuery object from an SQL statement token.
Params:
-------
statement: SQL token that contains the entire query.
Returns:
--------
A new SQLQuery object.
"""
first_token = statement.token_first(skip_cm=True, skip_ws=True)
tables = cls._collect_tables(statement)
if first_token.match(token_types.DML, "SELECT"):
sql_instance = cls._build_select_query(statement, tables)
if first_token.match(token_types.DML, "UPDATE"):
assert len(tables) == 1
table = tables[0]
sql_instance = cls._build_update_query(statement, table)
if first_token.match(token_types.DML, "INSERT"):
assert len(tables) == 1
table = tables[0]
sql_instance = cls._build_insert_query(statement, table)
if first_token.match(token_types.DML, "DELETE"):
assert len(tables) == 1
table = tables[0]
sql_instance = cls._build_delete_query(statement, table)
if sql_instance is None:
raise exceptions.NotSupportedError(
f"Unsupported query type {first_token}")
_, where_group = statement.token_next_by(i=(token_groups.Where))
filter_groups = sql_table.FilterGroup.from_where_group(where_group)
for filter_group in filter_groups:
sql_instance.add_filter_group(filter_group)
return sql_instance
