From ba695c59c18f10e4a10d01c1f4908e24cc24fed5 Mon Sep 17 00:00:00 2001
From: Chris Truter <crisptrutski@users.noreply.github.com>
Date: Mon, 2 Sep 2024 09:56:13 +0200
Subject: [PATCH] Suppress columns coming from unknown tables (#97)

* Add failing test for phantom field from generated table
* Suppress columns from unknown tables
---
 src/macaw/collect.clj                                 | 11 +++++++++++
 .../resources/acceptance/generate_series.analysis.edn |  2 ++
 test/resources/acceptance/generate_series.sql         |  2 ++
 3 files changed, 15 insertions(+)
 create mode 100644 test/resources/acceptance/generate_series.analysis.edn
 create mode 100644 test/resources/acceptance/generate_series.sql

diff --git a/src/macaw/collect.clj b/src/macaw/collect.clj
index 81ef043..fcb2db7 100644
--- a/src/macaw/collect.clj
+++ b/src/macaw/collect.clj
@@ -170,6 +170,16 @@
                         :column))
           unqualified)))
 
+(defn- remove-phantom-table-columns
+  "If a column reference is qualified by a table that we don't know about, its probably
+   not a real field.
+   See the `:generate_series` fixture."
+  [table-map columns]
+  (let [known-table-name? (into #{} (map (comp :table :component val)) table-map)]
+    (filter (fn [{s :schema, t :table}]
+              (or (nil? t) (some? s) (known-table-name? t)))
+            columns)))
+
 (defn- infer-table-schema [columns node]
   (update node :component
           #(let [{:keys [schema table] :as element} %]
@@ -220,6 +230,7 @@
         source-columns            (->> (map :component all-columns)
                                        (remove-redundant-columns alias?)
                                        (remove literal?)
+                                       (remove-phantom-table-columns table-map)
                                        (into #{}
                                              (comp (remove (comp pseudo-table-names :table))
                                                    (remove :internal?)
diff --git a/test/resources/acceptance/generate_series.analysis.edn b/test/resources/acceptance/generate_series.analysis.edn
new file mode 100644
index 0000000..973fef9
--- /dev/null
+++ b/test/resources/acceptance/generate_series.analysis.edn
@@ -0,0 +1,2 @@
+{:tables         []
+ :source-columns []}
diff --git a/test/resources/acceptance/generate_series.sql b/test/resources/acceptance/generate_series.sql
new file mode 100644
index 0000000..def9d5c
--- /dev/null
+++ b/test/resources/acceptance/generate_series.sql
@@ -0,0 +1,2 @@
+SELECT t.day::date AS date
+FROM generate_series(timestamp '2021-01-01', now(), interval '1 day') AS t(day)