copy column order

Author: codedmonkey

.github/scripts/normalize-psql-schema.py

@@ -13,8 +13,8 @@ def filter_lines(text):
     return ''.join(lines)
 
 
-def sort_columns(match):
-    """Sort column lines in a schema dump alphabetically"""
+def sort_create_table_columns(match):
+    """Sort column lines in a CREATE TABLE block alphabetically"""
     lines = [
         line.strip().rstrip(',')
         for line in match.group(1).split('\n')
@@ -23,5 +23,32 @@ def sort_columns(match):
     return '(\n    ' + ',\n    '.join(sorted(lines)) + '\n)'
 
 
+def sort_copy_columns(text):
+    """Sort columns in COPY blocks and reorder the data rows accordingly"""
+    result = []
+    lines = text.split('\n')
+    i = 0
+    while i < len(lines):
+        copy_match = re.match(r'(COPY \S+ \()([^)]+)(\) FROM stdin;)', lines[i])
+        if copy_match:
+            cols = [c.strip() for c in copy_match.group(2).split(',')]
+            sorted_indices = sorted(range(len(cols)), key=lambda j: cols[j])
+            sorted_cols = [cols[j] for j in sorted_indices]
+            result.append(copy_match.group(1) + ', '.join(sorted_cols) + copy_match.group(3))
+            i += 1
+            while i < len(lines) and lines[i] != '\\.':
+                values = lines[i].split('\t')
+                result.append('\t'.join(values[j] for j in sorted_indices))
+                i += 1
+            if i < len(lines):
+                result.append(lines[i])  # \.
+        else:
+            result.append(lines[i])
+        i += 1
+    return '\n'.join(result)
+
+
 text = filter_lines(sys.stdin.read())
-sys.stdout.write(re.sub(r'\(((?:\n    [^\n]+)+)\n\)', sort_columns, text))
+text = re.sub(r'\(((?:\n    [^\n]+)+)\n\)', sort_create_table_columns, text)
+text = sort_copy_columns(text)
+sys.stdout.write(text)

.github/workflows/database.yaml

@@ -183,7 +183,7 @@ jobs:
             > /tmp/schema_baseline.sql
           pg_dump -U dirigent -h localhost dirigent \
             --data-only --no-comments --no-privileges --no-owner \
-            | grep -vE '^(SET |SELECT |$)' \
+            | python3 .github/scripts/normalize-psql-schema.py \
             > /tmp/data_baseline.sql
 
       - name: Install Composer dependencies for current branch
@@ -209,7 +209,7 @@ jobs:
             > /tmp/schema_reverted.sql
           pg_dump -U dirigent -h localhost dirigent \
             --data-only --no-comments --no-privileges --no-owner \
-            | grep -vE '^(SET |SELECT |$)' \
+            | python3 .github/scripts/normalize-psql-schema.py \
             > /tmp/data_reverted.sql
 
       - name: Verify schema after revert matches baseline