Skip to content

Speakr is a personal, self-hosted web application designed for transcribing audio recordings

Python

Fix double-quoted SQL string defaults breaking PostgreSQL, add pre-commit hook

Murtaza Nasira57611e68b7d
diff --git a/scripts/pre-commit b/scripts/pre-commit
new file mode 100755
index 00000000..692ddeeb
--- /dev/null
+++ b/scripts/pre-commit
@@ -0,0 +1,21 @@
+#!/bin/sh
+#
+# Pre-commit hook: runs PostgreSQL migration compatibility tests
+# when migration-related files are staged.
+#
+# Install: ln -sf ../../scripts/pre-commit .git/hooks/pre-commit
+
+STAGED_FILES=$(git diff --cached --name-only --diff-filter=ACM)
+
+# Only run if migration-related files are staged
+if echo "$STAGED_FILES" | grep -qE '^(src/init_db\.py|src/utils/database\.py)$'; then
+ echo "Migration files changed — running PostgreSQL compatibility checks..."
+ python tests/test_migration_compatibility.py
+ if [ $? -ne 0 ]; then
+ echo ""
+ echo "Pre-commit hook FAILED: PostgreSQL migration compatibility issues found."
+ echo "Fix the issues above before committing."
+ exit 1
+ fi
+ echo "All migration compatibility checks passed."
+fi
diff --git a/src/init_db.py b/src/init_db.py
index d52dcc91..59f448cf 100644
--- a/src/init_db.py
+++ b/src/init_db.py
@@ -71,7 +71,7 @@ def initialize_database(app):
app.logger.info("Added company column to user table")
if add_column_if_not_exists(engine, 'user', 'diarize', 'BOOLEAN'):
app.logger.info("Added diarize column to user table")
- if add_column_if_not_exists(engine, 'user', 'ui_language', 'VARCHAR(10) DEFAULT "en"'):
+ if add_column_if_not_exists(engine, 'user', 'ui_language', "VARCHAR(10) DEFAULT 'en'"):
app.logger.info("Added ui_language column to user table")
if add_column_if_not_exists(engine, 'user', 'sso_provider', 'VARCHAR(100)'):
app.logger.info("Added sso_provider column to user table")
@@ -158,7 +158,7 @@ def initialize_database(app):
app.logger.info("Added transcription_duration_seconds column to recording table")
if add_column_if_not_exists(engine, 'recording', 'summarization_duration_seconds', 'INTEGER'):
app.logger.info("Added summarization_duration_seconds column to recording table")
- if add_column_if_not_exists(engine, 'recording', 'processing_source', 'VARCHAR(50) DEFAULT "upload"'):
+ if add_column_if_not_exists(engine, 'recording', 'processing_source', "VARCHAR(50) DEFAULT 'upload'"):
app.logger.info("Added processing_source column to recording table")
if add_column_if_not_exists(engine, 'recording', 'error_message', 'TEXT'):
app.logger.info("Added error_message column to recording table")
@@ -272,7 +272,7 @@ def initialize_database(app):
# Auto speaker labelling settings
if add_column_if_not_exists(engine, 'user', 'auto_speaker_labelling', 'BOOLEAN DEFAULT 0'):
app.logger.info("Added auto_speaker_labelling column to user table")
- if add_column_if_not_exists(engine, 'user', 'auto_speaker_labelling_threshold', 'VARCHAR(10) DEFAULT "medium"'):
+ if add_column_if_not_exists(engine, 'user', 'auto_speaker_labelling_threshold', "VARCHAR(10) DEFAULT 'medium'"):
app.logger.info("Added auto_speaker_labelling_threshold column to user table")
# Auto summarization setting (per-user, default enabled)
@@ -297,7 +297,7 @@ def initialize_database(app):
app.logger.info("Added export_template_id column to folder table")
# Add source tracking columns to internal_share table
- if add_column_if_not_exists(engine, 'internal_share', 'source_type', 'VARCHAR(20) DEFAULT "manual"'):
+ if add_column_if_not_exists(engine, 'internal_share', 'source_type', "VARCHAR(20) DEFAULT 'manual'"):
app.logger.info("Added source_type column to internal_share table")
if add_column_if_not_exists(engine, 'internal_share', 'source_tag_id', 'INTEGER'):
diff --git a/src/utils/database.py b/src/utils/database.py
index 8b847592..a0ac0a31 100644
--- a/src/utils/database.py
+++ b/src/utils/database.py
@@ -42,6 +42,10 @@ def add_column_if_not_exists(engine, table_name, column_name, column_type):
# PostgreSQL uses BYTEA, not BLOB
column_type = re.sub(r'\bBLOB\b', 'BYTEA', column_type, flags=re.IGNORECASE)
+ # PostgreSQL interprets double-quoted strings as identifiers, not literals
+ # Convert DEFAULT "value" to DEFAULT 'value'
+ column_type = re.sub(r'''DEFAULT\s+"([^"]*)"''', r"DEFAULT '\1'", column_type, flags=re.IGNORECASE)
+
with engine.connect() as conn:
# Quote identifiers to handle reserved keywords (e.g., "user" in PostgreSQL)
# MySQL uses backticks, PostgreSQL/SQLite use double quotes
diff --git a/tests/test_migration_compatibility.py b/tests/test_migration_compatibility.py
index e40320c6..e8a43b52 100644
--- a/tests/test_migration_compatibility.py
+++ b/tests/test_migration_compatibility.py
@@ -188,6 +188,33 @@ class TestMigrationCompatibility(unittest.TestCase):
matches = re.findall(pattern, self.content, re.IGNORECASE)
# Just informational - these are fine because the utility converts them
+ def test_no_double_quoted_string_defaults(self):
+ """
+ Ensure no SQL DEFAULT values use double-quoted strings.
+
+ In SQL, double quotes denote identifiers (column/table names), not string
+ literals. SQLite tolerates this, but PostgreSQL will interpret DEFAULT "en"
+ as a reference to a column named "en" and fail with 'column "en" does not exist'.
+
+ String defaults must use single quotes: DEFAULT 'en'
+ """
+ # Match DEFAULT followed by a double-quoted string value
+ pattern = r'DEFAULT\s+"[^"]*"'
+
+ lines = self.content.splitlines()
+ problematic = []
+ for i, line in enumerate(lines, 1):
+ if re.search(pattern, line, re.IGNORECASE):
+ problematic.append(f" Line {i}: {line.strip()}")
+
+ self.assertEqual(
+ len(problematic), 0,
+ f"Found double-quoted string defaults in init_db.py. "
+ f"PostgreSQL interprets double quotes as column identifiers, not string literals. "
+ f"Use single quotes instead (e.g., DEFAULT 'en' not DEFAULT \"en\"):\n" +
+ "\n".join(problematic)
+ )
+
def test_create_index_uses_utility_for_user_table(self):
"""
Ensure index creation on 'user' table uses create_index_if_not_exists().