From 74fb1a0cd12c9b3abe38c7c10f12211789f535e2 Mon Sep 17 00:00:00 2001
From: George Sittas <giwrgos.sittas@gmail.com>
Date: Mon, 28 Jul 2025 12:45:27 +0300
Subject: [PATCH 1/2] Chore: fail if >1 migration scripts start with the same
 version

---
 .pre-commit-config.yaml               |  6 ++++++
 scripts/check_migration_duplicates.sh | 18 ++++++++++++++++++
 2 files changed, 24 insertions(+)
 create mode 100755 scripts/check_migration_duplicates.sh

diff --git a/.pre-commit-config.yaml b/.pre-commit-config.yaml
index 96eca9b6a4..3a9a978fed 100644
--- a/.pre-commit-config.yaml
+++ b/.pre-commit-config.yaml
@@ -23,3 +23,9 @@ repos:
         files: *files
         require_serial: true
         exclude: ^(tests/fixtures)
+      - id: check-migration-duplicates
+        name: Check migration duplicates
+        language: script
+        entry: ./scripts/check_migration_duplicates.sh
+        files: ^sqlmesh/migrations/v\d{4}_.*\.py$
+        pass_filenames: false
diff --git a/scripts/check_migration_duplicates.sh b/scripts/check_migration_duplicates.sh
new file mode 100755
index 0000000000..13afd4b50f
--- /dev/null
+++ b/scripts/check_migration_duplicates.sh
@@ -0,0 +1,18 @@
+#!/bin/bash
+# Check for duplicate migration version numbers in SQLMesh
+
+# Extract version numbers from migration files and check for duplicates
+duplicates=$(ls sqlmesh/migrations/v[0-9][0-9][0-9][0-9]_*.py 2>/dev/null | \
+    sed 's/.*\/v\([0-9]\{4\}\)_.*/\1/' | \
+    sort | uniq -d)
+
+if [ -n "$duplicates" ]; then
+    echo "Error: Duplicate migration version(s) found:" >&2
+    for version in $duplicates; do
+        echo "  Version v$version appears in:" >&2
+        ls sqlmesh/migrations/v${version}_*.py | sed 's/^/    - /' >&2
+    done
+    exit 1
+fi
+
+exit 0

From 5a9cc0d8aaf0ee2746d16c6e4f711a77e201841a Mon Sep 17 00:00:00 2001
From: George Sittas <giwrgos.sittas@gmail.com>
Date: Mon, 28 Jul 2025 13:09:34 +0300
Subject: [PATCH 2/2] Refactor script to make it more comprehensive, move under
 tooling/

---
 .pre-commit-config.yaml                 |  9 ++---
 scripts/check_migration_duplicates.sh   | 18 ---------
 tooling/validating_migration_numbers.sh | 53 +++++++++++++++++++++++++
 3 files changed, 57 insertions(+), 23 deletions(-)
 delete mode 100755 scripts/check_migration_duplicates.sh
 create mode 100755 tooling/validating_migration_numbers.sh

diff --git a/.pre-commit-config.yaml b/.pre-commit-config.yaml
index 3a9a978fed..e4b61cb0f3 100644
--- a/.pre-commit-config.yaml
+++ b/.pre-commit-config.yaml
@@ -23,9 +23,8 @@ repos:
         files: *files
         require_serial: true
         exclude: ^(tests/fixtures)
-      - id: check-migration-duplicates
-        name: Check migration duplicates
-        language: script
-        entry: ./scripts/check_migration_duplicates.sh
-        files: ^sqlmesh/migrations/v\d{4}_.*\.py$
+      - id: valid migrations
+        name: valid migrations
+        entry: tooling/validating_migration_numbers.sh
+        language: system
         pass_filenames: false
diff --git a/scripts/check_migration_duplicates.sh b/scripts/check_migration_duplicates.sh
deleted file mode 100755
index 13afd4b50f..0000000000
--- a/scripts/check_migration_duplicates.sh
+++ /dev/null
@@ -1,18 +0,0 @@
-#!/bin/bash
-# Check for duplicate migration version numbers in SQLMesh
-
-# Extract version numbers from migration files and check for duplicates
-duplicates=$(ls sqlmesh/migrations/v[0-9][0-9][0-9][0-9]_*.py 2>/dev/null | \
-    sed 's/.*\/v\([0-9]\{4\}\)_.*/\1/' | \
-    sort | uniq -d)
-
-if [ -n "$duplicates" ]; then
-    echo "Error: Duplicate migration version(s) found:" >&2
-    for version in $duplicates; do
-        echo "  Version v$version appears in:" >&2
-        ls sqlmesh/migrations/v${version}_*.py | sed 's/^/    - /' >&2
-    done
-    exit 1
-fi
-
-exit 0
diff --git a/tooling/validating_migration_numbers.sh b/tooling/validating_migration_numbers.sh
new file mode 100755
index 0000000000..6dbb597dc1
--- /dev/null
+++ b/tooling/validating_migration_numbers.sh
@@ -0,0 +1,53 @@
+#!/bin/bash
+
+# Navigate to the migrations directory (modify the path if necessary)
+cd "sqlmesh/migrations" || exit 1
+
+
+# Collect all migration files matching the pattern (e.g., v0001_initial.py)
+migration_files=(v*.py)
+
+# Initialize an array to hold migration numbers
+numbers=()
+
+# Extract migration numbers from filenames
+for file in "${migration_files[@]}"; do
+  if [[ $file =~ ^v0*([0-9]+)_ ]]; then
+    num=${BASH_REMATCH[1]}
+    numbers+=("$num")
+  fi
+done
+
+# Check if any migration files were found
+if [[ ${#numbers[@]} -eq 0 ]]; then
+  echo "No migration files found matching the pattern 'v<zero-padded number>_<description>.py'."
+  exit 1
+fi
+
+# Check for duplicate migration numbers
+duplicates=$(printf "%s\n" "${numbers[@]}" | sort | uniq -d)
+if [[ -n $duplicates ]]; then
+  echo "Error: Duplicate migration numbers found: $duplicates"
+  exit 1
+fi
+
+# Sort the migration numbers
+sorted_numbers=($(printf "%s\n" "${numbers[@]}" | sort -n))
+
+# Get the first and last migration numbers
+first_number="${sorted_numbers[0]}"
+last_index=$((${#sorted_numbers[@]} - 1))
+last_number="${sorted_numbers[$last_index]}"
+
+# Check for gaps in the migration sequence
+expected_numbers=($(seq "$first_number" "$last_number"))
+
+if [[ "${sorted_numbers[*]}" != "${expected_numbers[*]}" ]]; then
+  echo "Error: Missing migration numbers in sequence."
+  echo "Expected sequence: ${expected_numbers[*]}"
+  echo "Found sequence:    ${sorted_numbers[*]}"
+  exit 1
+fi
+
+echo "All migration numbers are sequential and without overlaps."
+exit 0