jupyter-duckdb 1.2.0.6__py3-none-any.whl → 1.2.7__py3-none-any.whl

duckdb_kernel/parser/RAParser.py

@@ -19,10 +19,14 @@ class RAParser:
             tokens = tuple(Tokenizer.tokenize(tokens[0]))
 
         # binary operators
-        for operator in RA_BINARY_OPERATORS:
+        for operator_symbols in RA_BINARY_SYMBOLS:
             # iterate tokens and match symbol
             for i in range(1, len(tokens) + 1):
-                if tokens[-i].lower() in operator.symbols():
+                lower_token = tokens[-i].lower()
+
+                if lower_token in operator_symbols:
+                    operator = operator_symbols[lower_token]
+
                     # raise error if left or right operand missing
                     if i == 1:
                         raise RAParserError(f'right operand missing after {tokens[-i]}', depth)

duckdb_kernel/parser/elements/__init__.py

@@ -9,7 +9,7 @@ from .RAOperator import RAOperator
 from .RABinaryOperator import RABinaryOperator
 from .RAUnaryOperator import RAUnaryOperator
 from .RAOperand import RAOperand
-from .binary import RA_BINARY_OPERATORS
+from .binary import RA_BINARY_OPERATORS, RA_BINARY_SYMBOLS
 from .unary import RA_UNARY_OPERATORS
 
 from .DCOperand import DCOperand

duckdb_kernel/parser/elements/binary/FullOuterJoin.py

@@ -0,0 +1,30 @@
+from typing import Tuple, Dict
+
+from duckdb_kernel.db import Table
+from ..RABinaryOperator import RABinaryOperator
+from ...util.RenamableColumnList import RenamableColumnList
+
+
+class FullOuterJoin(RABinaryOperator):
+    @staticmethod
+    def symbols() -> Tuple[str, ...]:
+        return chr(10199), 'fjoin', 'ojoin'
+
+    def to_sql(self, tables: Dict[str, Table]) -> Tuple[str, RenamableColumnList]:
+        # execute subqueries
+        lq, lcols = self.left.to_sql(tables)
+        rq, rcols = self.right.to_sql(tables)
+
+        # find matching columns
+        join_cols, all_cols = lcols.intersect(rcols)
+
+        replacements = {c1: c2 for c1, c2 in join_cols}
+        select_cols = [
+            f'COALESCE({c.current_name}, {replacements.get(c).current_name})' if c in replacements else c.current_name
+            for c in all_cols]
+        select_clause = ', '.join(select_cols)
+
+        on_clause = ' AND '.join(f'{l.current_name} = {r.current_name}' for l, r in join_cols)
+
+        # create sql
+        return f'SELECT {select_clause} FROM ({lq}) {self._name()} FULL OUTER JOIN ({rq}) {self._name()} ON {on_clause}', all_cols

duckdb_kernel/parser/elements/binary/LeftOuterJoin.py

@@ -0,0 +1,24 @@
+from typing import Tuple, Dict
+
+from duckdb_kernel.db import Table
+from ..RABinaryOperator import RABinaryOperator
+from ...util.RenamableColumnList import RenamableColumnList
+
+
+class LeftOuterJoin(RABinaryOperator):
+    @staticmethod
+    def symbols() -> Tuple[str, ...]:
+        return chr(10197), 'ljoin'
+
+    def to_sql(self, tables: Dict[str, Table]) -> Tuple[str, RenamableColumnList]:
+        # execute subqueries
+        lq, lcols = self.left.to_sql(tables)
+        rq, rcols = self.right.to_sql(tables)
+
+        # find matching columns
+        join_cols, all_cols = lcols.intersect(rcols)
+
+        on_clause = ' AND '.join(f'{l.current_name} = {r.current_name}' for l, r in join_cols)
+
+        # create sql
+        return f'SELECT {all_cols.list} FROM ({lq}) {self._name()} LEFT OUTER JOIN ({rq}) {self._name()} ON {on_clause}', all_cols

duckdb_kernel/parser/elements/binary/RightOuterJoin.py

@@ -0,0 +1,24 @@
+from typing import Tuple, Dict
+
+from duckdb_kernel.db import Table
+from ..RABinaryOperator import RABinaryOperator
+from ...util.RenamableColumnList import RenamableColumnList
+
+
+class RightOuterJoin(RABinaryOperator):
+    @staticmethod
+    def symbols() -> Tuple[str, ...]:
+        return chr(10198), 'rjoin'
+
+    def to_sql(self, tables: Dict[str, Table]) -> Tuple[str, RenamableColumnList]:
+        # execute subqueries
+        lq, lcols = self.left.to_sql(tables)
+        rq, rcols = self.right.to_sql(tables)
+
+        # find matching columns
+        join_cols, all_cols = lcols.intersect(rcols, prefer_right=True)
+
+        on_clause = ' AND '.join(f'{l.current_name} = {r.current_name}' for l, r in join_cols)
+
+        # create sql
+        return f'SELECT {all_cols.list} FROM ({lq}) {self._name()} RIGHT OUTER JOIN ({rq}) {self._name()} ON {on_clause}', all_cols

duckdb_kernel/parser/elements/binary/__init__.py

@@ -2,6 +2,9 @@ from .Cross import Cross
 from .Difference import Difference
 from .Intersection import Intersection
 from .Join import Join
+from .LeftOuterJoin import LeftOuterJoin
+from .RightOuterJoin import RightOuterJoin
+from .FullOuterJoin import FullOuterJoin
 from .Union import Union
 
 from .Add import Add
@@ -30,12 +33,22 @@ LOGIC_BINARY_OPERATORS = sorted([
 ], key=lambda x: x.order, reverse=True)
 
 RA_BINARY_OPERATORS = [
-    Difference,
-    Union,
-    Intersection,
-    Join,
-    Cross,
-    Division
+    [Difference],
+    [Union],
+    [Intersection],
+    [Join],
+    [LeftOuterJoin, RightOuterJoin, FullOuterJoin],
+    [Cross],
+    [Division]
+]
+
+RA_BINARY_SYMBOLS = [
+    {
+        symbol: operator
+        for operator in level
+        for symbol in operator.symbols()
+    }
+    for level in RA_BINARY_OPERATORS
 ]
 
 DC_SET = ConditionalSet

duckdb_kernel/parser/util/RenamableColumnList.py

@@ -73,18 +73,26 @@ class RenamableColumnList(list[RenamableColumn]):
 
         return RenamableColumnList(cols.values())
 
-    def intersect(self, other: 'RenamableColumnList') \
+    def intersect(self, other: 'RenamableColumnList', prefer_right: bool = False) \
             -> Tuple[List[Tuple[RenamableColumn, RenamableColumn]], 'RenamableColumnList']:
         self_cols: Dict[str, RenamableColumn] = {col.name: col for col in self}
         other_cols: Dict[str, RenamableColumn] = {col.name: col for col in other}
 
+        replacements: Dict[RenamableColumn, RenamableColumn] = {}
         intersection: List[Tuple[RenamableColumn, RenamableColumn]] = []
+
         for name in tuple(self_cols.keys()):
             if name in other_cols:
+                if prefer_right:
+                    replacements[self_cols[name]] = other_cols[name]
+
                 intersection.append((self_cols[name], other_cols[name]))
                 del other_cols[name]
 
         # if len(intersection) == 0:
         #     raise AssertionError('no common attributes found for join')
 
-        return intersection, RenamableColumnList(self_cols.values(), other_cols.values())
+        return intersection, RenamableColumnList(
+            (replacements.get(x, x) for x in self_cols.values()),
+            other_cols.values()
+        )

{jupyter_duckdb-1.2.0.6.dist-info → jupyter_duckdb-1.2.7.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: jupyter-duckdb
-Version: 1.2.0.6
+Version: 1.2.7
 Summary: a basic wrapper kernel for DuckDB
 Home-page: https://github.com/erictroebs/jupyter-duckdb
 Author: Eric Tröbs

{jupyter_duckdb-1.2.0.6.dist-info → jupyter_duckdb-1.2.7.dist-info}/RECORD

@@ -27,7 +27,7 @@ duckdb_kernel/magics/__init__.py,sha256=DA8gnQeRCUt1Scy3_NQ9w5CPmMEY9i8YwB-g392p
 duckdb_kernel/parser/DCParser.py,sha256=16c1mxa494KP9OreUKQHsSQKoDGZ7NNp2u_gi_D-dkw,2293
 duckdb_kernel/parser/LogicParser.py,sha256=_vZwE5OPRUEN8aEC_fSZAYKR_dpexqNthXog9OFHYRY,1233
 duckdb_kernel/parser/ParserError.py,sha256=qJQVloFtID1HgVDQ1Io247bODT1ic3oO9Z1ZrWR-2Mk,321
-duckdb_kernel/parser/RAParser.py,sha256=dL5Q5J8tmlGuKQMfadvJK6cb9KmzfZLkfcUAthqDmXI,2993
+duckdb_kernel/parser/RAParser.py,sha256=vLR2X8zJIJ3cPKdLvU3PKjNakEOvsoFuoK8pr4bA-Fg,3102
 duckdb_kernel/parser/__init__.py,sha256=nTmDm1ADvNPDHhVJQLxKYmArNJk6967EUXqn5AkT8FM,126
 duckdb_kernel/parser/elements/DCOperand.py,sha256=qEg_6Us4WV1eK4Bq6oUsmFt_L_x5pJPGce_wSapzIYA,1149
 duckdb_kernel/parser/elements/LogicElement.py,sha256=YasKHxWLDDP8UdyLIKbXzqIRA8-XaakjmvTj-1Iuzyc,280
@@ -38,7 +38,7 @@ duckdb_kernel/parser/elements/RAElement.py,sha256=d6QWO8KlWcwhA4tXtWsf8K4oT3OMMj
 duckdb_kernel/parser/elements/RAOperand.py,sha256=pghnTYCrrT6MkvynJRgVFPRoMvxIGNB3FTjaq-uCpDQ,1078
 duckdb_kernel/parser/elements/RAOperator.py,sha256=rtqMFBIBBqT-Bwg7Qm4WQwbDrE28Nb74F_7XMeR3ks4,255
 duckdb_kernel/parser/elements/RAUnaryOperator.py,sha256=XC1nphkSm88JaEu5V_HKnb_8JNoeBfE3EvNL4o0qh2c,654
-duckdb_kernel/parser/elements/__init__.py,sha256=4DA2M43hh9d1fZb5Z6YnTTI-IBkDyhChSF5L9X3EA-s,547
+duckdb_kernel/parser/elements/__init__.py,sha256=t5H6SVOm3z8r6UWRYOI7HmMIuB4Yh6TLNu4_N3jg2t0,566
 duckdb_kernel/parser/elements/binary/Add.py,sha256=XGkZMfab01huk9EaI6JUfzkd2STbV1C_-TyC2guKE8I,190
 duckdb_kernel/parser/elements/binary/And.py,sha256=0jgetTG8yo5TJSeK70Kj-PI9ERyek1eyMQXX5HBxa4Y,274
 duckdb_kernel/parser/elements/binary/ArrowLeft.py,sha256=u4fZSoyT9lfvWXBwuhUl4DdjVZAOqyVIKmMVbpElLD4,203
@@ -48,18 +48,21 @@ duckdb_kernel/parser/elements/binary/Difference.py,sha256=ZVRgJHYVMOFwnc97oPvGtK
 duckdb_kernel/parser/elements/binary/Divide.py,sha256=d7mzaOeRYSRO1F-2IHsv_C939TuYtLppbf4-5GSRJXs,265
 duckdb_kernel/parser/elements/binary/Division.py,sha256=Tq98RzGHN7QYF3R2ZVpwIiLhSIMQXzRgIuZ7l0D2fXg,1624
 duckdb_kernel/parser/elements/binary/Equal.py,sha256=xd5eMwVwM7thXqmoBg-JSux4y6osxxznNzaChmgrnGI,192
+duckdb_kernel/parser/elements/binary/FullOuterJoin.py,sha256=WwV9jVFXr0FkfOOGKNZAiEjFKO_l0ND3RAvJYKEyVzY,1133
 duckdb_kernel/parser/elements/binary/GreaterThan.py,sha256=lW7EPYjwwZ1VLZOE4ai6MBW7Zq5itsUTi_v6FxoAuU0,204
 duckdb_kernel/parser/elements/binary/GreaterThanEqual.py,sha256=gO1I3J5CZwrpqZb5Oc-WJ_fCnGylpj4YKl4s-ZAr_-4,286
 duckdb_kernel/parser/elements/binary/Intersection.py,sha256=K6NrwcLTJ1BUccifVQ2R2YHR5B71LhjuLhOalxZcM0g,757
 duckdb_kernel/parser/elements/binary/Join.py,sha256=9Z3JinEav3ulLbZZZM9DLbprloHC0oGuSm40Kf1A0mM,836
+duckdb_kernel/parser/elements/binary/LeftOuterJoin.py,sha256=qCyHLMboNjzSDK_FREtYVwZx4UYkfRcy3FMo2pEf93o,846
 duckdb_kernel/parser/elements/binary/LessThan.py,sha256=gNlyAvew0VyvU8x3ol88zbdh_t9GU51Kd2P_PmHUrAc,201
 duckdb_kernel/parser/elements/binary/LessThanEqual.py,sha256=BKKXBZ4TZQ4EBxBDSNzIpFl4Oz1IGI0tI0Oj7KZUoDo,283
 duckdb_kernel/parser/elements/binary/Minus.py,sha256=1XUTXCA0GcAcCnv9nq4B9ewT0XHJ91LlZGEW1y60POA,192
 duckdb_kernel/parser/elements/binary/Multiply.py,sha256=OctlfN71izFYxIpMo09Qwq-3BmQmSh6-PNKBljMsBrg,195
 duckdb_kernel/parser/elements/binary/Or.py,sha256=e1H_BuY7TQD7XHKGqJYLOebYQvrh1euWOUsXF0FT5dY,264
+duckdb_kernel/parser/elements/binary/RightOuterJoin.py,sha256=bDxTEQ_dJZ-MUuGbU1YL69kPFR0Hdkc1fow3wxOxfAY,867
 duckdb_kernel/parser/elements/binary/Unequal.py,sha256=1hnC1RcPMxwKKv65OL5prunGgh9cRVDmzJutmtl7gtY,269
 duckdb_kernel/parser/elements/binary/Union.py,sha256=VYTj4M2PVEhWiDwjnyP8qpVVbGvIBSVshlEt-SZYCBY,739
-duckdb_kernel/parser/elements/binary/__init__.py,sha256=PtZgC0vOQONRx0FnFvzaxK9dYobdCykbQATP20djivA,983
+duckdb_kernel/parser/elements/binary/__init__.py,sha256=BieXw51EBgIwUEK-hn4jBb1b4VKJT6sRzaxwOrSXw2A,1342
 duckdb_kernel/parser/elements/unary/Not.py,sha256=kG0a-dp3TNjPodUMPiQ6ihtsBrbvn1iWeIPCvtyAwdo,632
 duckdb_kernel/parser/elements/unary/Projection.py,sha256=CJ-MIf1-__1ewTjNZVy5hOz3Z18CWnCDNJBxUdpWXVQ,1112
 duckdb_kernel/parser/elements/unary/Rename.py,sha256=Zr2n9EJ3nA476lND0Djz2b6493nnsbSpJ9kkEgk5B_Y,1273
@@ -69,7 +72,7 @@ duckdb_kernel/parser/tokenizer/Token.py,sha256=gsCzgU_zLiA-yD0FWvd2qS9LQUXbivESY
 duckdb_kernel/parser/tokenizer/Tokenizer.py,sha256=PWGgS7gYgpULiKGDho842UbaXuqmwEkccixuF10oi5g,5081
 duckdb_kernel/parser/tokenizer/__init__.py,sha256=EOSmfc2RJwtB5cE1Hhj1JAra97tckxxS8-legybPy60,58
 duckdb_kernel/parser/util/RenamableColumn.py,sha256=LxJhFDMUv_OxYYDLwKn63QGpBRfs08jVvhuJTzRtc9c,704
-duckdb_kernel/parser/util/RenamableColumnList.py,sha256=GfhdGv4KYT64Z9YA9TCn-7hhcEcc3Gu3vI2zMZ52w-8,3015
+duckdb_kernel/parser/util/RenamableColumnList.py,sha256=5oEDbtvl4YfHbkxu_Ny2pc0EYnhCZsf7EeoNQvftbrU,3281
 duckdb_kernel/parser/util/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 duckdb_kernel/util/ResultSetComparator.py,sha256=RZDIfjJyx8-eR-HIqQlEYgZd_V1ympbszpVRF4TlA7o,2262
 duckdb_kernel/util/SQL.py,sha256=-uRfa0IwEQueZNZ7vkBPczLuvm_87y4_nnMBx3FgqNk,643
@@ -80,7 +83,7 @@ duckdb_kernel/visualization/Drawer.py,sha256=D0LkiGMvuJ2v6cQSg_axLTGaM4VXAJEQJAy
 duckdb_kernel/visualization/RATreeDrawer.py,sha256=j-Vy1zpYMzwZ3CsphyfPW-J7ou9a9tM6aXXgAlQTgDI,2128
 duckdb_kernel/visualization/SchemaDrawer.py,sha256=9K-TUUmyeGdMYMTFQJ7evIU3p8p2KyMKeizUc7-y8co,3015
 duckdb_kernel/visualization/__init__.py,sha256=5eMJmxJ01XAXcgWDn3t70eSZF2PGaXdNo6GK-x-0H3s,78
-jupyter_duckdb-1.2.0.6.dist-info/METADATA,sha256=CpGdT2u47HAdtsRNJpQJeQvQPIuBVXVLfYyybbrxQnY,9115
-jupyter_duckdb-1.2.0.6.dist-info/WHEEL,sha256=jB7zZ3N9hIM9adW7qlTAyycLYW9npaWKLRzaoVcLKcM,91
-jupyter_duckdb-1.2.0.6.dist-info/top_level.txt,sha256=KvRRPMnmkQNuhyBsXoPmwyt26LRDp0O-0HN6u0Dm5jA,14
-jupyter_duckdb-1.2.0.6.dist-info/RECORD,,
+jupyter_duckdb-1.2.7.dist-info/METADATA,sha256=g002q8FOyFbbZlRm-5qyDrf-yb_7n2bok9HqNXSvpbY,9113
+jupyter_duckdb-1.2.7.dist-info/WHEEL,sha256=jB7zZ3N9hIM9adW7qlTAyycLYW9npaWKLRzaoVcLKcM,91
+jupyter_duckdb-1.2.7.dist-info/top_level.txt,sha256=KvRRPMnmkQNuhyBsXoPmwyt26LRDp0O-0HN6u0Dm5jA,14
+jupyter_duckdb-1.2.7.dist-info/RECORD,,