sql-code-graph 1.3.0__py3-none-any.whl → 1.4.1__py3-none-any.whl

{sql_code_graph-1.3.0.dist-info → sql_code_graph-1.4.1.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: sql-code-graph
-Version: 1.3.0
+Version: 1.4.1
 Summary: SQL code graph analyzer and lineage tracer
 Project-URL: Homepage, https://github.com/Warhorze/sql-code-graph
 Project-URL: Repository, https://github.com/Warhorze/sql-code-graph
@@ -18,7 +18,7 @@ Classifier: Programming Language :: Python :: 3.13
 Classifier: Topic :: Database
 Classifier: Topic :: Software Development :: Libraries :: Python Modules
 Requires-Python: >=3.12
-Requires-Dist: kuzu==0.11.3
+Requires-Dist: duckdb<2.0,>=1.0.0
 Requires-Dist: mcp<2.0,>=1.27.0
 Requires-Dist: pathspec>=0.12.1
 Requires-Dist: pydantic>=2.0
@@ -30,8 +30,6 @@ Requires-Dist: typer>=0.9.0
 Requires-Dist: watchdog>=3.0.0
 Provides-Extra: dbt
 Requires-Dist: dbt-core>=1.7; extra == 'dbt'
-Provides-Extra: neo4j
-Requires-Dist: neo4j>=5.15.0; extra == 'neo4j'
 Provides-Extra: snowflake
 Requires-Dist: acryl-datahub<0.15.0,>=0.14.0; extra == 'snowflake'
 Description-Content-Type: text/markdown

{sql_code_graph-1.3.0.dist-info → sql_code_graph-1.4.1.dist-info}/RECORD

@@ -1,36 +1,36 @@
-sqlcg/__init__.py,sha256=Zwww9eU1OV0KsLzksvKQE65bNgEFPPwTUb0XCSW4JIE,115
+sqlcg/__init__.py,sha256=Q9RK3IHj_JNrfASKHyNaH76PD2u74uMssIwE5Mi7F7Q,115
 sqlcg/__main__.py,sha256=1YoFLcqEgTwYq1J3TbUwpkdG0zeeLIf2fJvwWI-CLFU,109
 sqlcg/cli/__init__.py,sha256=W8fD0LpMq2xm_5WKGNMvJh2WBL1ho5E8hUeAqXQYT1g,28
 sqlcg/cli/main.py,sha256=WmdTjsOlz1ozi2Y3Aq4ezR_FCRl-Lc1YOKw3_d48dlY,1650
 sqlcg/cli/commands/__init__.py,sha256=oSHtr6VD-jNubOjuCQyZj2tBppjMEpQDh-IGQ8of9eA,30
-sqlcg/cli/commands/analyze.py,sha256=xr3RHmO4eFTP4VKZn4DAx3BJzSi60_DIZmdE-QLfsHI,13601
-sqlcg/cli/commands/db.py,sha256=paW096LE8fMpxPNvoW9zHmZ9xjb-dEbwVmfHR1bcg7U,8676
-sqlcg/cli/commands/find.py,sha256=zTYN9goILalYq4R9x6lIR6MmNcydDbR17UXkx1gPRsI,2913
-sqlcg/cli/commands/gain.py,sha256=Kws76u1na2XxmbWN_YWrPaYHYmYBLC6DDDf7xqnltqc,9126
+sqlcg/cli/commands/analyze.py,sha256=_aC5ML3w7YdLi7DL3TFS9OiCEIipuNZxWR6S4peTcn4,12154
+sqlcg/cli/commands/db.py,sha256=TMhPCHRWSo8YmlNnSxmGxrR0_1r3K4e0Bn8unZTkvU4,7248
+sqlcg/cli/commands/find.py,sha256=p5Vyyx-VBk8YDWYQN16UhECh7PIeMaEyCPEcUFcRFlM,2598
+sqlcg/cli/commands/gain.py,sha256=SJU1c51a7MgNbZItqQnaBfOWGnV5xpXQctbew5Dr9BE,9062
 sqlcg/cli/commands/git.py,sha256=9a8T2FVxcAHq1H6Cslaq34t10w9fBGf4T2reiLk33t8,6135
-sqlcg/cli/commands/index.py,sha256=1ElHRPkn-CPprIz869A__98aSyf-P5E56PVA0lq7xBw,17462
+sqlcg/cli/commands/index.py,sha256=RPEpT9aXmK0rixRuHgMcmAn9FLR4JtIbZ1LM4RLkMvY,17050
 sqlcg/cli/commands/install.py,sha256=KNABvrLbamPyYnmnVdCaM_MNezbDc-pr6IkignCWI8k,9186
 sqlcg/cli/commands/mcp.py,sha256=QYaupf69lLpYzIoqsPJoCPiAggLVkYBzwpuOLRzxJDU,9140
-sqlcg/cli/commands/reindex.py,sha256=Ki5BHbI_nuM6ML0-w7qnVZqSAhELSpsFaUo6BVqzhRo,13812
+sqlcg/cli/commands/reindex.py,sha256=FqXxvQ0UrxwDS0q2V3gzYgarN3NPh9muaIZ3rgvTmHs,13810
 sqlcg/cli/commands/report.py,sha256=JU0qjyMxwOukE7bN3XvvIzOI7zMg_Gsnvk_8F6pKNpA,4915
-sqlcg/cli/commands/uninstall.py,sha256=IYwQaqnMmmzW0Nlls40wD-L3tVkMgKIMRXUkcXPMUc4,9398
+sqlcg/cli/commands/uninstall.py,sha256=WrA1FnINxnd6mmE4-_QBK0aHBnibstJeAT8swnQKG4M,8962
 sqlcg/cli/commands/watch.py,sha256=7N6c-QuvxAEGHzDZ0C3CU2BkHSraZW9YtgoFnz7SaQo,2373
-sqlcg/core/__init__.py,sha256=uNsJCrCMVWVT80sHPtI_f39BYqIf5N0i6LSq8x8HsyI,283
-sqlcg/core/config.py,sha256=qNR-yXkfYfS8Y8WX4Qo6Zkq8PPP_ZiTrvX0DLmEZkGY,14821
-sqlcg/core/freshness.py,sha256=gRb8pRPw5SdIUxAYkMXIJ00DTdQ6CegRZPAvWnv0rU0,4575
-sqlcg/core/graph_db.py,sha256=Aa85wPFg26H-Ud9SrZyxCHH-99iitAI5S3X9T_62Yyw,7957
+sqlcg/core/__init__.py,sha256=dXvLWpbQ72f5CM6sKSvBDnEGqGuIZaN5MmHyD8Vf1aA,154
+sqlcg/core/config.py,sha256=rMTYt1QsZVOyhT8W7oE65XQh5LKqdOeXCQ00sMFny4U,12375
+sqlcg/core/duckdb_backend.py,sha256=nnOGbjmjHGXR_dlrzJQpSw5otLescCW1WtltZthoURQ,29522
+sqlcg/core/freshness.py,sha256=F9jWn2cbFs60jA9ta8KrT6MghD0mzI7SqqKs_Op9AeU,4577
+sqlcg/core/graph_db.py,sha256=eLYdmiqPWrZHtKLFcKgD2aiWZHFU3wrwz6Y0A3d9NcE,8633
 sqlcg/core/jobs.py,sha256=Je-fCdSKRgiSsv1W8SgNAlp36a7t7-pJZ-qKPbka9OE,3298
-sqlcg/core/kuzu_backend.py,sha256=PHW7VqI7oCLKsHnm4OoBoNnE2XT19ohxUpQMMIJnjlY,17038
-sqlcg/core/neo4j_backend.py,sha256=AM1TncP9GBGph-rSHwalZPmGUV2kFILzaJP-PSB0UYw,8437
 sqlcg/core/queries.cypher,sha256=cvPOVe5GUOzJN4bxUvDxNI--xIIP8gm42TR-gUnea4U,4685
-sqlcg/core/queries.py,sha256=gkl4bhkZM8FsvbSA-IaK17sRFcO3hB5YlVCemkCXgWM,2064
+sqlcg/core/queries.py,sha256=jtZR6caLpF0WqqkjncYTlf5L1GN6DZweoM5dNOAB8OY,3043
+sqlcg/core/queries.sql,sha256=75EHoHA5hKz-Xs8g-CCtDXbf3lYFEw_fxghjYU48gUM,7794
 sqlcg/core/schema.cypher,sha256=rK5QMhSrzZhuj73NeNXGX6oM-rPPPvxFjex0fEyUvkQ,2859
-sqlcg/core/schema.py,sha256=JO5rkspYKjL9AEl5mt0VIJKn-IPOH3kJV_fVmAMuFCI,1467
+sqlcg/core/schema.py,sha256=7fKf314ueIV7-tIkQQUS4O6H-OhlFKFVFFvGFqw_5Xk,1476
 sqlcg/indexer/__init__.py,sha256=Wh20Unz2OHs1oIyWLrpurPAasF0BET2g4iXtNk7mh2U,56
 sqlcg/indexer/dbt_adapter.py,sha256=EB5x1WU5Z9d-I97ADDj88S_hG1C4z4nbrv8JUCzXfy8,686
 sqlcg/indexer/error_classify.py,sha256=MYjPVprwT-ARPjBCyCzu2F9DSrZfnTVtVIoBgm8s4H8,5329
 sqlcg/indexer/git_delta.py,sha256=P-QM4vnVURT2KLiE6u3cQynRUF-mTH13cbB4I20YHPQ,4468
-sqlcg/indexer/indexer.py,sha256=7TCgBLl3ml3mF8Z2q4YHZJ6HdxSuLH-rPJTibnUJUe4,66658
+sqlcg/indexer/indexer.py,sha256=NCOWwXmE7e6m42sraFJpImmtWFss-V95RzEuI-eDzb0,66400
 sqlcg/indexer/pool.py,sha256=BTYx-pBe6zwUG89MHh0X7nzGNVlsHN-GjovYKanVI1s,18553
 sqlcg/indexer/walker.py,sha256=umNaqDbuerr75VYG1TEOv0ATsbI40O3SIw35f7XJcDE,1931
 sqlcg/indexer/watcher.py,sha256=mJQq1LASRLKKwhz0WhCUWPLLqyPR2_-FD_8efYU6gE8,8442
@@ -38,30 +38,30 @@ sqlcg/lineage/__init__.py,sha256=Da1DlYwtK13WHv_RnHjAtNkHTOuFbhxqCjT1Le7DsWM,46
 sqlcg/lineage/aggregator.py,sha256=LVyNcmvLBHWbh8SrDsJJBKd7sLg3-2NhEWwEndG7Jbc,4144
 sqlcg/lineage/schema_resolver.py,sha256=iXt6LYF6UVWsGUpcfbmjmGn9wCgXl721lTGf_8AaWcc,7320
 sqlcg/metrics/__init__.py,sha256=hLJ6wm4St8qqYwKh3o9QG7lcEt1BEYM31ccqO9tGpIg,133
-sqlcg/metrics/store.py,sha256=O1UoBu4dIZYIgNBqLWIyL3vLAnSgWrJinOgSLhQigHM,10596
+sqlcg/metrics/store.py,sha256=KuDtxvyAgug9_KtiSCpvgKM2VZM7VSaI3D11uMLjJJk,10604
 sqlcg/parsers/__init__.py,sha256=AamA8wBbDZV9_zEtZCI4Hyen5UAVKHmBwjTghTt2PZE,785
 sqlcg/parsers/ansi_parser.py,sha256=mGZvijMOMQ4i1BybpwU29a8jnIGViefhy9fxzkSpsRM,17193
-sqlcg/parsers/base.py,sha256=IiOkVsm6jz9-48RqDCXiW-UXAraNxQ4pKXvSA7aolnA,49907
+sqlcg/parsers/base.py,sha256=N6uqQWHO2lpzTIWgPqzewAXmij0ikZdNmNVJkIQ8Mr0,54889
 sqlcg/parsers/bigquery_parser.py,sha256=mOnWTfXB_Dp4JwFE1PVYOB6CDPf5nYE0Dea8kJCl9uQ,2827
 sqlcg/parsers/postgres_parser.py,sha256=lYfUpQY6j4Qm7ndXBtXbgPoGzYqYddWt5YeFnWKdA6I,946
 sqlcg/parsers/registry.py,sha256=LXy1F6rqQI6VdxpRvZg_tNpoEucW3mXZHYBMlMONbX4,1496
 sqlcg/parsers/snowflake_parser.py,sha256=fovMyqfhWD2wmtEyiwTC0aoP4QWP-3XQZ8WYkXvs9hg,15511
 sqlcg/parsers/tsql_parser.py,sha256=RRj1pACtAk2tLTDaFWRYF67a0IDvaf5A1YQXWIz0bpQ,956
 sqlcg/server/__init__.py,sha256=n4wuNE7xyJIJxJZBtmtdccCMQfvTdF-IqIaZVbC4FC4,35
-sqlcg/server/control.py,sha256=v-r21npODiHlHnJHuo_6KWrKclQKq_E1QyrzIWjqgtY,4508
+sqlcg/server/control.py,sha256=qUcztb_aDhL-_X_Nq4q6uGx17cUlbLnI6vUpoZsEjbo,4506
 sqlcg/server/exceptions.py,sha256=EONw34icOByCTpppSQrvQBW6asc4hfqaGDCAFjv96II,469
 sqlcg/server/models.py,sha256=l7ORy6sbtzBW1y3qVaeLwEukbyAgBkz9S5VIm2q4b24,19378
-sqlcg/server/noise_filter.py,sha256=idSBGgdKWWccJdpOo9qgbM2350Oew-2l5W6Yc9GYQqY,6337
-sqlcg/server/read_client.py,sha256=ncoJK7UckGhWtN9bv1CgViNMNtac96zBUE7RPYQ8_WI,7783
-sqlcg/server/server.py,sha256=9SilAu18cHTZUQvSo8S8e9CxSM6CUlG8rX2OnHBUh1Y,24178
-sqlcg/server/skill.py,sha256=GE8eeimk6yiGGJ74erGypqYAviur5peSR6_2a4QQWVM,12828
-sqlcg/server/tools.py,sha256=DTaDwZQmL6jzNF8vgJeNhVMGRoIrrpcxNiXOMpWvx-A,72401
-sqlcg/server/writer.py,sha256=gagS6EVG8A4OKpf0GAb--MUielnaiIULwVVik58pT6k,24693
+sqlcg/server/noise_filter.py,sha256=Ats2FFzmzFKqvQPWvlUzK8mY9pzlUhL4m1s8P_HNnvI,6335
+sqlcg/server/read_client.py,sha256=4v1OOl12HCHp6J14HrcWQApSixvJ0wwE8UIIZ0pzLyw,7808
+sqlcg/server/server.py,sha256=A1soT-hYnAX78X5j97nIjw9oIZCxBejjV6p_tRDIK38,24335
+sqlcg/server/skill.py,sha256=EDvmEgl-LQwCxZ9Lca4lt8zCtkk2cCgh9GLYlGgDA64,12828
+sqlcg/server/tools.py,sha256=Rdl5SXAuePydk8Z12JVx-lNQeAqm3k9R-z0Q4DstCkw,69743
+sqlcg/server/writer.py,sha256=HH8pW6k1c2pbY4b6g2N87Tx4gNCezYqTLNTlAmsHg2Y,18522
 sqlcg/utils/__init__.py,sha256=--iqt5ThTXmT8Wz7da8hs3n0zDfYPl8P-z5OgRJ_77E,154
 sqlcg/utils/hashing.py,sha256=H25-sYfxHKb3_IERFnHyAIYNiXN470Oqo5sJT_D3YOA,438
 sqlcg/utils/ignore.py,sha256=wJjwa0mjnQ_xJExOUxk25y00g065XmmzJapqV3ifD5o,1151
 sqlcg/utils/logging.py,sha256=u0fCmYsLj9o81vawm3xZTHaw68GQYVm7JxG-gP81u8A,840
-sql_code_graph-1.3.0.dist-info/METADATA,sha256=2QJqn9Q606zlPKDgeYrNzq0d1QW35jIRCxmvDCvvNIE,14148
-sql_code_graph-1.3.0.dist-info/WHEEL,sha256=mffPy8wBnZQn2VnJUU5jE99KsxaSfiyMHV9Yt0aLVxs,87
-sql_code_graph-1.3.0.dist-info/entry_points.txt,sha256=Wfe49sVzV9p4eVFGo5RxcV-frr3HOP0yzzst8JBxQLQ,46
-sql_code_graph-1.3.0.dist-info/RECORD,,
+sql_code_graph-1.4.1.dist-info/METADATA,sha256=7psrSu1Vh3GiZ7PK7FED45QzazeUrHrltrRla4twAVQ,14085
+sql_code_graph-1.4.1.dist-info/WHEEL,sha256=mffPy8wBnZQn2VnJUU5jE99KsxaSfiyMHV9Yt0aLVxs,87
+sql_code_graph-1.4.1.dist-info/entry_points.txt,sha256=Wfe49sVzV9p4eVFGo5RxcV-frr3HOP0yzzst8JBxQLQ,46
+sql_code_graph-1.4.1.dist-info/RECORD,,

sqlcg/__init__.py

@@ -1,5 +1,5 @@
 """SQL Code Graph - SQL lineage and dependency analysis tool."""
 
-__version__ = "1.3.0"
+__version__ = "1.4.1"
 
 __all__ = ["__version__"]

sqlcg/cli/commands/analyze.py

@@ -9,7 +9,6 @@ from rich.console import Console
 from rich.table import Table
 
 from sqlcg.core.queries import GET_TABLE_EXTERNAL_CONSUMERS_QUERY
-from sqlcg.core.schema import NodeLabel, RelType
 from sqlcg.server.read_client import run_read_routed
 
 if TYPE_CHECKING:
@@ -19,37 +18,111 @@ app = typer.Typer(help="Lineage analysis")
 console = Console()
 
 
-def _kind_filter(node_alias: str, *, include_intermediate: bool) -> str:
-    """Build the SqlTable kind-filter clause for upstream/downstream queries.
+def _upstream_sql(depth: int, include_intermediate: bool) -> str:
+    """Build the upstream recursive-CTE SQL query.
 
-    Uses an OPTIONAL MATCH + explicit ``WITH … WHERE t.kind IS NULL OR …`` form
-    (Half B of the #38/#39 fix) so that a node-less physical source — one whose
-    SqlTable node is absent because it was only seen inside a CTE body before
-    re-index — is KEPT, not dropped.  CTE/derived intermediates carry
-    ``kind='cte'`` or ``kind='derived'`` and are excluded by the
-    ``kind IN [...]`` guard; ``IS NULL`` matches the absent-node case (t = NULL
-    → t.kind = NULL → IS NULL is TRUE).
-
-    The ``WITH {node_alias}, t WHERE`` clause is required: an OPTIONAL MATCH
-    WHERE clause in KuzuDB applies to the match attempt and does not filter the
-    surrounding row.  The subsequent WITH … WHERE is the actual row filter.
-
-    Args:
-        node_alias: The Cypher alias for the column node whose table is filtered
-                    (``"src"`` for upstream, ``"dst"`` for downstream).
-        include_intermediate: When True, return an empty string (no filtering);
-                              all intermediates including CTE nodes are kept.
-
-    Returns:
-        A Cypher fragment string (with trailing space) to embed directly in the
-        query, or an empty string when include_intermediate is True.
+    Traverses COLUMN_LINEAGE edges from dst→src (upstream direction).
+    Applies kind-filter (LEFT JOIN SqlTable) to exclude CTE/derived intermediates
+    unless include_intermediate=True.  The kind-guard mirrors the Cypher
+    OPTIONAL MATCH + WITH … WHERE t.kind IS NULL OR t.kind IN ['table','external']
+    semantics (#38/#40/19.2).
+    """
+    kind_filter = (
+        ""
+        if include_intermediate
+        else (
+            '  LEFT JOIN "SqlTable" t ON t.qualified = dr.table_qualified\n'
+            "  WHERE t.kind IS NULL OR t.kind IN ('table', 'external')\n"
+        )
+    )
+    return f"""
+WITH RECURSIVE reach(id, table_qualified, depth, path) AS (
+  SELECT
+    cl.src_key AS id,
+    c_src.table_qualified,
+    1 AS depth,
+    ARRAY[cl.dst_key, cl.src_key] AS path
+  FROM "COLUMN_LINEAGE" cl
+  JOIN "SqlColumn" c_src ON c_src.id = cl.src_key
+  WHERE cl.dst_key = ?
+  UNION ALL
+  SELECT
+    cl2.src_key,
+    c2.table_qualified,
+    reach.depth + 1,
+    array_append(reach.path, cl2.src_key)
+  FROM reach
+  JOIN "COLUMN_LINEAGE" cl2 ON cl2.dst_key = reach.id
+  JOIN "SqlColumn" c2 ON c2.id = cl2.src_key
+  WHERE reach.depth < {depth}
+    AND NOT cl2.src_key = ANY(reach.path)
+),
+distinct_reach AS (
+  SELECT DISTINCT id, table_qualified FROM reach
+)
+SELECT
+  dr.id,
+  dr.table_qualified,
+  min(q.file_path) AS file,
+  min(q.start_line) AS line
+FROM distinct_reach dr
+LEFT JOIN "COLUMN_LINEAGE" src_edge ON src_edge.src_key = dr.id
+LEFT JOIN "SqlQuery" q ON q.id = src_edge.query_id
+{kind_filter}GROUP BY dr.id, dr.table_qualified
+LIMIT 100
+"""
+
+
+def _downstream_sql(depth: int, include_intermediate: bool) -> str:
+    """Build the downstream recursive-CTE SQL query.
+
+    Traverses COLUMN_LINEAGE edges from src→dst (downstream direction).
+    Kind-filter mirrors upstream: LEFT JOIN SqlTable + IS NULL OR 'table'/'external'.
     """
-    if include_intermediate:
-        return ""
-    return (
-        f"OPTIONAL MATCH (t:SqlTable {{qualified: {node_alias}.table_qualified}}) "
-        f"WITH {node_alias}, t WHERE t.kind IS NULL OR t.kind IN ['table', 'external'] "
+    kind_filter = (
+        ""
+        if include_intermediate
+        else (
+            '  LEFT JOIN "SqlTable" t ON t.qualified = dr.table_qualified\n'
+            "  WHERE t.kind IS NULL OR t.kind IN ('table', 'external')\n"
+        )
     )
+    return f"""
+WITH RECURSIVE reach(id, table_qualified, depth, path) AS (
+  SELECT
+    cl.dst_key AS id,
+    c_dst.table_qualified,
+    1 AS depth,
+    ARRAY[cl.src_key, cl.dst_key] AS path
+  FROM "COLUMN_LINEAGE" cl
+  JOIN "SqlColumn" c_dst ON c_dst.id = cl.dst_key
+  WHERE cl.src_key = ?
+  UNION ALL
+  SELECT
+    cl2.dst_key,
+    c2.table_qualified,
+    reach.depth + 1,
+    array_append(reach.path, cl2.dst_key)
+  FROM reach
+  JOIN "COLUMN_LINEAGE" cl2 ON cl2.src_key = reach.id
+  JOIN "SqlColumn" c2 ON c2.id = cl2.dst_key
+  WHERE reach.depth < {depth}
+    AND NOT cl2.dst_key = ANY(reach.path)
+),
+distinct_reach AS (
+  SELECT DISTINCT id, table_qualified FROM reach
+)
+SELECT
+  dr.id,
+  dr.table_qualified,
+  min(q.file_path) AS file,
+  min(q.start_line) AS line
+FROM distinct_reach dr
+LEFT JOIN "COLUMN_LINEAGE" dst_edge ON dst_edge.dst_key = dr.id
+LEFT JOIN "SqlQuery" q ON q.id = dst_edge.query_id
+{kind_filter}GROUP BY dr.id, dr.table_qualified
+LIMIT 100
+"""
 
 
 @app.command("upstream")
@@ -62,36 +135,16 @@ def upstream(  # noqa: B008
     ),
 ) -> None:
     """Trace upstream column lineage."""
-    # Bounds check for depth to prevent performance DoS
     if depth < 1 or depth > 100:
         console.print("[red]Error: --depth must be between 1 and 100[/red]")
         raise typer.Exit(1)
 
-    # By default, filter out CTE/derived intermediate nodes; --include-intermediate restores them
-    kf = _kind_filter("src", include_intermediate=include_intermediate)
-
-    results = run_read_routed(
-        f"MATCH (c:{NodeLabel.COLUMN} {{id: $ref}})"
-        f"<-[:{RelType.COLUMN_LINEAGE}*1..{depth}]-(src:{NodeLabel.COLUMN}) "
-        f"{kf}"
-        f"OPTIONAL MATCH (src)-[srcedge:{RelType.COLUMN_LINEAGE}]->() "
-        "OPTIONAL MATCH (q:SqlQuery {id: srcedge.query_id}) "
-        "WITH src, min(q.start_line) AS line, min(q.file_path) AS file "
-        "RETURN src.id AS id, file AS file, line AS line LIMIT 100",
-        {"ref": ref},
-    )
+    ref = ref.lower()  # graph keys are lowercased at index time (C2 normalization)
+    sql = _upstream_sql(depth, include_intermediate)
+    results = run_read_routed(sql, {"ref": ref})
     if not results and len(ref.split(".")) >= 3:
         bare = _bare_ref(ref)
-        fallback_results = run_read_routed(
-            f"MATCH (c:{NodeLabel.COLUMN} {{id: $bare}})"
-            f"<-[:{RelType.COLUMN_LINEAGE}*1..{depth}]-(src:{NodeLabel.COLUMN}) "
-            f"{kf}"
-            f"OPTIONAL MATCH (src)-[srcedge:{RelType.COLUMN_LINEAGE}]->() "
-            "OPTIONAL MATCH (q:SqlQuery {id: srcedge.query_id}) "
-            "WITH src, min(q.start_line) AS line, min(q.file_path) AS file "
-            "RETURN src.id AS id, file AS file, line AS line LIMIT 100",
-            {"bare": bare},
-        )
+        fallback_results = run_read_routed(sql, {"bare": bare})
         if fallback_results:
             console.print(
                 f"[yellow]Hint:[/yellow] No results for '{ref}'. "
@@ -104,7 +157,7 @@ def upstream(  # noqa: B008
     if not raw:
         from sqlcg.server.noise_filter import NoiseFilter
 
-        nf = NoiseFilter.from_config()  # repo_root=None → falls back to Path.cwd()
+        nf = NoiseFilter.from_config()
         results = _filter_column_results(results, nf)
     _print_table(_add_file_line_col(results), ["id", "file:line"])
 
@@ -119,36 +172,16 @@ def downstream(  # noqa: B008
     ),
 ) -> None:
     """Trace downstream column lineage."""
-    # Bounds check for depth to prevent performance DoS
     if depth < 1 or depth > 100:
         console.print("[red]Error: --depth must be between 1 and 100[/red]")
         raise typer.Exit(1)
 
-    # By default, filter out CTE/derived intermediate nodes; --include-intermediate restores them
-    kf = _kind_filter("dst", include_intermediate=include_intermediate)
-
-    results = run_read_routed(
-        f"MATCH (c:{NodeLabel.COLUMN} {{id: $ref}})"
-        f"-[:{RelType.COLUMN_LINEAGE}*1..{depth}]->(dst:{NodeLabel.COLUMN}) "
-        f"{kf}"
-        f"OPTIONAL MATCH ()-[dstedge:{RelType.COLUMN_LINEAGE}]->(dst) "
-        "OPTIONAL MATCH (q:SqlQuery {id: dstedge.query_id}) "
-        "WITH dst, min(q.start_line) AS line, min(q.file_path) AS file "
-        "RETURN dst.id AS id, file AS file, line AS line LIMIT 100",
-        {"ref": ref},
-    )
+    ref = ref.lower()  # graph keys are lowercased at index time (C2 normalization)
+    sql = _downstream_sql(depth, include_intermediate)
+    results = run_read_routed(sql, {"ref": ref})
     if not results and len(ref.split(".")) >= 3:
         bare = _bare_ref(ref)
-        fallback_results = run_read_routed(
-            f"MATCH (c:{NodeLabel.COLUMN} {{id: $bare}})"
-            f"-[:{RelType.COLUMN_LINEAGE}*1..{depth}]->(dst:{NodeLabel.COLUMN}) "
-            f"{kf}"
-            f"OPTIONAL MATCH ()-[dstedge:{RelType.COLUMN_LINEAGE}]->(dst) "
-            "OPTIONAL MATCH (q:SqlQuery {id: dstedge.query_id}) "
-            "WITH dst, min(q.start_line) AS line, min(q.file_path) AS file "
-            "RETURN dst.id AS id, file AS file, line AS line LIMIT 100",
-            {"bare": bare},
-        )
+        fallback_results = run_read_routed(sql, {"bare": bare})
         if fallback_results:
             console.print(
                 f"[yellow]Hint:[/yellow] No results for '{ref}'. "
@@ -161,18 +194,16 @@ def downstream(  # noqa: B008
     if not raw:
         from sqlcg.server.noise_filter import NoiseFilter
 
-        nf = NoiseFilter.from_config()  # repo_root=None → falls back to Path.cwd()
+        nf = NoiseFilter.from_config()
         results = _filter_column_results(results, nf)
     _print_table(_add_file_line_col(results), ["id", "file:line"])
 
     # Append external consumer rows for terminal tables (scalar query, one per terminal).
-    # Resolve terminal tables from the column results; fall back to the root column's table.
     terminal_tables: set[str] = set()
     for r in results:
         tbl = _col_id_to_table(r["id"])
         if tbl:
             terminal_tables.add(tbl)
-    # Also check the root column's table (in case no downstream columns were found).
     root_parts = ref.rsplit(".", 1)
     if len(root_parts) == 2:
         terminal_tables.add(root_parts[0])
@@ -197,9 +228,11 @@ def impact(  # noqa: B008
 ) -> None:
     """Show all queries impacted by a table."""
     results = run_read_routed(
-        f"MATCH (t:{NodeLabel.TABLE} {{qualified: $t}})"
-        f"<-[:{RelType.SELECTS_FROM}]-(q:{NodeLabel.QUERY}) "
-        "RETURN DISTINCT q.id AS id, q.kind AS kind, q.target_table AS target LIMIT 100",
+        "SELECT DISTINCT q.id AS id, q.kind AS kind, q.target_table AS target"
+        ' FROM "SqlTable" t'
+        ' JOIN "SELECTS_FROM" sf ON sf.dst_key = t.qualified'
+        ' JOIN "SqlQuery" q ON q.id = sf.src_key'
+        " WHERE t.qualified = ? LIMIT 100",
         {"t": table},
     )
     if not raw:
@@ -222,22 +255,20 @@ def failures(
     ),
     limit: int = typer.Option(100, "--limit", help="Maximum rows to return"),  # noqa: B008
 ) -> None:
-    """List files that failed to parse, with their dominant cause (E-code bucket).
-
-    Valid --cause buckets (from highest to lowest severity):
-    timeout, E8, E3, E2, E5, E1, qualify_failed, func_fallback, pure_ddl_skip.
-
-    Requires a graph indexed with sqlcg >= v3 (schema version 3). Re-index
-    with 'sqlcg db reset && sqlcg index <path>' if the graph was built with
-    an earlier version.
-    """
-    cypher = (
-        f"MATCH (f:{NodeLabel.FILE}) WHERE f.parse_failed = true "
-        "AND ($cause IS NULL OR f.parse_cause = $cause) "
-        "RETURN f.path AS path, f.parse_cause AS cause "
-        f"ORDER BY f.parse_cause LIMIT {limit}"
-    )
-    rows = run_read_routed(cypher, {"cause": cause})
+    """List files that failed to parse, with their dominant cause (E-code bucket)."""
+    if cause is not None:
+        rows = run_read_routed(
+            'SELECT path, parse_cause AS cause FROM "File"'
+            " WHERE parse_failed = true AND parse_cause = ?"
+            f" ORDER BY parse_cause LIMIT {limit}",
+            {"cause": cause},
+        )
+    else:
+        rows = run_read_routed(
+            'SELECT path, parse_cause AS cause FROM "File"'
+            f" WHERE parse_failed = true ORDER BY parse_cause LIMIT {limit}",
+            {},
+        )
     _print_table(rows, ["path", "cause"])
 
 
@@ -248,8 +279,9 @@ def unused(
 ) -> None:
     """Find tables with no query references."""
     results = run_read_routed(
-        f"MATCH (t:{NodeLabel.TABLE}) WHERE NOT (t)<-[:{RelType.SELECTS_FROM}]-() "
-        "RETURN DISTINCT t.qualified AS qualified LIMIT 100",
+        'SELECT DISTINCT qualified FROM "SqlTable"'
+        ' WHERE qualified NOT IN (SELECT DISTINCT dst_key FROM "SELECTS_FROM")'
+        " LIMIT 100",
         {},
     )
     if not raw:
@@ -263,28 +295,18 @@ def unused(
 def _bare_ref(ref: str) -> str:
     """Strip schema prefix from a ref string, keeping table.column.
 
-    For a 3-part ref ("mart.fact_t.amount") this returns "fact_t.amount".
-    For a 2-part ref ("fact_t.amount") this returns the ref unchanged.
-    Never uses rsplit — that would yield only the column name for 3-part refs.
+    Lowercases defensively so this is safe to call even if the caller did not
+    first fold the ref — graph keys are lowercased at index time (C2 normalization).
     """
+    ref = ref.lower()
     parts = ref.split(".")
     if len(parts) >= 3:
-        return ".".join(parts[1:])  # drop schema, keep table.column
-    return ref  # already bare (no schema prefix)
+        return ".".join(parts[1:])
+    return ref
 
 
 def _col_id_to_table(col_id: str) -> str:
-    """Extract the table-qualified part from a column ID (schema.table.col → schema.table).
-
-    Column IDs follow the format: schema.table.column or table.column.
-    The table part is everything except the last component.
-
-    Args:
-        col_id: A column ID string from the graph.
-
-    Returns:
-        The table-qualified portion (all but the last dotted component).
-    """
+    """Extract the table-qualified part from a column ID."""
     parts = col_id.rsplit(".", 1)
     return parts[0] if len(parts) == 2 else col_id
 
@@ -293,16 +315,12 @@ def _filter_column_results(
     results: list[dict],
     nf: NoiseFilter,  # type: ignore[name-defined]
 ) -> list[dict]:
-    """Filter column-ID result rows by NoiseFilter, dropping rows whose table is noise."""
+    """Filter column-ID result rows by NoiseFilter."""
     return [r for r in results if not nf.is_noise(_col_id_to_table(r["id"]))]
 
 
 def _add_file_line_col(rows: list[dict]) -> list[dict]:
-    """Add a 'file:line' composite column from 'file' and 'line' fields.
-
-    Formats as 'path/to/file.sql:N' when both are present, or '?' when either
-    is absent (multi-hop upstream where file/line is not available).
-    """
+    """Add a 'file:line' composite column from 'file' and 'line' fields."""
     result = []
     for row in rows:
         new_row = dict(row)