Fixed #3214 -- Stopped parsing SQL with regex.

Avoided introducing a new regex-based SQL splitter in the migrations
framework, before we're bound by backwards compatibility.

Adapted this change to the legacy "initial SQL data" feature, even
though it's already deprecated, in order to facilitate the transition
to migrations.

sqlparse becomes mandatory for RunSQL on some databases (all but
PostgreSQL). There's no API to provide a single statement and tell
Django not to attempt splitting. Since we have a more robust splitting
implementation, that seems like a good tradeoff. It's easier to add a
new keyword argument later if necessary than to remove one.

Many people contributed to both tickets, thank you all, and especially
Claude for the review.

Refs #22401.

django/core/management/sql.py

@@ -155,6 +155,7 @@ def sql_all(app_config, style, connection):
 
 
 def _split_statements(content):
+    # Private API only called from code that emits a RemovedInDjango19Warning.
     comment_re = re.compile(r"^((?:'[^']*'|[^'])*?)--.*$")
     statements = []
     statement = []
@@ -202,9 +203,7 @@ def custom_sql_for_model(model, style, connection):
     for sql_file in sql_files:
         if os.path.exists(sql_file):
             with codecs.open(sql_file, 'r' if six.PY3 else 'U', encoding=settings.FILE_CHARSET) as fp:
-                # Some backends can't execute more than one SQL statement at a time,
+                output.extend(connection.ops.prepare_sql_script(fp.read(), _allow_fallback=True))
-                # so split into separate statements.
-                output.extend(_split_statements(fp.read()))
     return output
 
 

django/db/backends/__init__.py

@@ -1,5 +1,6 @@
 import datetime
 import time
+import warnings
 
 try:
     from django.utils.six.moves import _thread as thread
@@ -16,6 +17,7 @@ from django.db.backends.signals import connection_created
 from django.db.backends import utils
 from django.db.transaction import TransactionManagementError
 from django.db.utils import DatabaseError, DatabaseErrorWrapper, ProgrammingError
+from django.utils.deprecation import RemovedInDjango19Warning
 from django.utils.functional import cached_property
 from django.utils import six
 from django.utils import timezone
@@ -599,6 +601,10 @@ class BaseDatabaseFeatures(object):
     # Does 'a' LIKE 'A' match?
     has_case_insensitive_like = True
 
+    # Does the backend require the sqlparse library for splitting multi-line
+    # statements before executing them?
+    requires_sqlparse_for_splitting = True
+
     def __init__(self, connection):
         self.connection = connection
 
@@ -867,6 +873,34 @@ class BaseDatabaseOperations(object):
         """
         return 'DEFAULT'
 
+    def prepare_sql_script(self, sql, _allow_fallback=False):
+        """
+        Takes a SQL script that may contain multiple lines and returns a list
+        of statements to feed to successive cursor.execute() calls.
+
+        Since few databases are able to process raw SQL scripts in a single
+        cursor.execute() call and PEP 249 doesn't talk about this use case,
+        the default implementation is conservative.
+        """
+        # Remove _allow_fallback and keep only 'return ...' in Django 1.9.
+        try:
+            # This import must stay inside the method because it's optional.
+            import sqlparse
+        except ImportError:
+            if _allow_fallback:
+                # Without sqlparse, fall back to the legacy (and buggy) logic.
+                warnings.warn(
+                    "Providing intial SQL data on a %s database will require "
+                    "sqlparse in Django 1.9." % self.connection.vendor,
+                    RemovedInDjango19Warning)
+                from django.core.management.sql import _split_statements
+                return _split_statements(sql)
+            else:
+                raise
+        else:
+            return [sqlparse.format(statement, strip_comments=True)
+                    for statement in sqlparse.split(sql) if statement]
+
     def process_clob(self, value):
         """
         Returns the value of a CLOB column, for backends that return a locator

django/db/backends/postgresql_psycopg2/base.py

@@ -58,6 +58,7 @@ class DatabaseFeatures(BaseDatabaseFeatures):
     nulls_order_largest = True
     closed_cursor_error_class = InterfaceError
     has_case_insensitive_like = False
+    requires_sqlparse_for_splitting = False
 
 
 class DatabaseWrapper(BaseDatabaseWrapper):

django/db/backends/postgresql_psycopg2/operations.py

@@ -93,6 +93,9 @@ class DatabaseOperations(BaseDatabaseOperations):
     def no_limit_value(self):
         return None
 
+    def prepare_sql_script(self, sql, _allow_fallback=False):
+        return [sql]
+
     def quote_name(self, name):
         if name.startswith('"') and name.endswith('"'):
             return name  # Quoting once is enough.

django/db/migrations/operations/special.py

@@ -1,4 +1,3 @@
-import re
 from .base import Operation
 
 
@@ -43,20 +42,16 @@ class SeparateDatabaseAndState(Operation):
 
 class RunSQL(Operation):
     """
-    Runs some raw SQL - a single statement by default, but it will attempt
+    Runs some raw SQL. A reverse SQL statement may be provided.
-    to parse and split it into multiple statements if multiple=True.
-
-    A reverse SQL statement may be provided.
 
     Also accepts a list of operations that represent the state change effected
     by this SQL change, in case it's custom column/table creation/deletion.
     """
 
-    def __init__(self, sql, reverse_sql=None, state_operations=None, multiple=False):
+    def __init__(self, sql, reverse_sql=None, state_operations=None):
         self.sql = sql
         self.reverse_sql = reverse_sql
         self.state_operations = state_operations or []
-        self.multiple = multiple
 
     @property
     def reversible(self):
@@ -66,30 +61,15 @@ class RunSQL(Operation):
         for state_operation in self.state_operations:
             state_operation.state_forwards(app_label, state)
 
-    def _split_sql(self, sql):
-        regex = r"(?mx) ([^';]* (?:'[^']*'[^';]*)*)"
-        comment_regex = r"(?mx) (?:^\s*$)|(?:--.*$)"
-        # First, strip comments
-        sql = "\n".join([x.strip().replace("%", "%%") for x in re.split(comment_regex, sql) if x.strip()])
-        # Now get each statement
-        for st in re.split(regex, sql)[1:][::2]:
-            yield st
-
     def database_forwards(self, app_label, schema_editor, from_state, to_state):
-        if self.multiple:
+        statements = schema_editor.connection.ops.prepare_sql_script(self.sql)
-            statements = self._split_sql(self.sql)
-        else:
-            statements = [self.sql]
         for statement in statements:
             schema_editor.execute(statement)
 
     def database_backwards(self, app_label, schema_editor, from_state, to_state):
         if self.reverse_sql is None:
             raise NotImplementedError("You cannot reverse this operation")
-        if self.multiple:
+        statements = schema_editor.connection.ops.prepare_sql_script(self.reverse_sql)
-            statements = self._split_sql(self.reverse_sql)
-        else:
-            statements = [self.reverse_sql]
         for statement in statements:
             schema_editor.execute(statement)
 

docs/internals/contributing/writing-code/unit-tests.txt

@@ -166,6 +166,7 @@ dependencies:
 *  memcached_, plus a :ref:`supported Python binding <memcached>`
 *  gettext_ (:ref:`gettext_on_windows`)
 *  selenium_
+*  sqlparse_
 
 You can find these dependencies in `pip requirements files`_ inside the
 ``tests/requirements`` directory of the Django source tree and install them
@@ -197,6 +198,7 @@ associated tests will be skipped.
 .. _memcached: http://memcached.org/
 .. _gettext: http://www.gnu.org/software/gettext/manual/gettext.html
 .. _selenium: https://pypi.python.org/pypi/selenium
+.. _sqlparse: https://pypi.python.org/pypi/sqlparse
 .. _pip requirements files: http://www.pip-installer.org/en/latest/cookbook.html#requirements-files
 
 Code coverage

docs/ref/migration-operations.txt

@@ -167,25 +167,23 @@ Changes a field's name (and, unless ``db_column`` is set, its column name).
 Special Operations
 ==================
 
+.. _operation-run-sql:
+
 RunSQL
 ------
 
 ::
 
-    RunSQL(sql, reverse_sql=None, state_operations=None, multiple=False)
+    RunSQL(sql, reverse_sql=None, state_operations=None)
 
 Allows running of arbitrary SQL on the database - useful for more advanced
 features of database backends that Django doesn't support directly, like
 partial indexes.
 
-``sql``, and ``reverse_sql`` if provided, should be strings of SQL to run on the
+``sql``, and ``reverse_sql`` if provided, should be strings of SQL to run on
-database. They will be passed to the database as a single SQL statement unless
+the database. On most database backends (all but PostgreSQL), Django will
-``multiple`` is set to ``True``, in which case they will be split into separate
+split the SQL into individual statements prior to executing them. This
-statements manually by the operation before being passed through.
+requires installing the sqlparse_ Python library.
-
-In some extreme cases, the built-in statement splitter may not be able to split
-correctly, in which case you should manually split the SQL into multiple calls
-to ``RunSQL``.
 
 The ``state_operations`` argument is so you can supply operations that are
 equivalent to the SQL in terms of project state; for example, if you are
@@ -194,6 +192,7 @@ operation here so that the autodetector still has an up-to-date state of the
 model (otherwise, when you next run ``makemigrations``, it won't see any
 operation that adds that field and so will try to run it again).
 
+.. _sqlparse: https://pypi.python.org/pypi/sqlparse
 
 .. _operation-run-python:
 

docs/releases/1.7.txt

@@ -636,6 +636,14 @@ Management Commands
 * :djadmin:`collectstatic` command with symlink option is now supported on
   Windows NT 6 (Windows Vista and newer).
 
+* :ref:`initial-sql` now works better if the sqlparse_ Python library is
+  installed.
+
+  Note that it's deprecated in favor of the :ref:`RunSQL <operation-run-sql>`
+  operation of migrations, which benefits from the improved behavior.
+
+.. _sqlparse: https://pypi.python.org/pypi/sqlparse
+
 Models
 ^^^^^^
 

tests/initial_sql_regress/tests.py

@@ -27,7 +27,6 @@ class InitialSQLTests(TestCase):
         """
         connection = connections[DEFAULT_DB_ALIAS]
         custom_sql = custom_sql_for_model(Simple, no_style(), connection)
-        self.assertEqual(len(custom_sql), 9)
         with connection.cursor() as cursor:
             for sql in custom_sql:
                 cursor.execute(sql)

tests/migrations/test_operations.py

@@ -1,5 +1,10 @@
 import unittest
 
+try:
+    import sqlparse
+except ImportError:
+    sqlparse = None
+
 from django.db import connection, migrations, models, router
 from django.db.migrations.migration import Migration
 from django.db.migrations.state import ProjectState
@@ -640,6 +645,7 @@ class OperationTests(MigrationTestBase):
             operation.database_backwards("test_alinto", editor, new_state, project_state)
         self.assertIndexNotExists("test_alinto_pony", ["pink", "weight"])
 
+    @unittest.skipIf(sqlparse is None and connection.features.requires_sqlparse_for_splitting, "Missing sqlparse")
     def test_run_sql(self):
         """
         Tests the RunSQL operation.
@@ -647,7 +653,10 @@ class OperationTests(MigrationTestBase):
         project_state = self.set_up_test_model("test_runsql")
         # Create the operation
         operation = migrations.RunSQL(
-            "CREATE TABLE i_love_ponies (id int, special_thing int)",
+            # Use a multi-line string with a commment to test splitting on SQLite and MySQL respectively
+            "CREATE TABLE i_love_ponies (id int, special_thing int);\n"
+            "INSERT INTO i_love_ponies (id, special_thing) VALUES (1, 42); -- this is magic!\n"
+            "INSERT INTO i_love_ponies (id, special_thing) VALUES (2, 51);\n",
             "DROP TABLE i_love_ponies",
             state_operations=[migrations.CreateModel("SomethingElse", [("id", models.AutoField(primary_key=True))])],
         )
@@ -661,6 +670,10 @@ class OperationTests(MigrationTestBase):
         with connection.schema_editor() as editor:
             operation.database_forwards("test_runsql", editor, project_state, new_state)
         self.assertTableExists("i_love_ponies")
+        # Make sure all the SQL was processed
+        with connection.cursor() as cursor:
+            cursor.execute("SELECT COUNT(*) FROM i_love_ponies")
+            self.assertEqual(cursor.fetchall()[0][0], 2)
         # And test reversal
         self.assertTrue(operation.reversible)
         with connection.schema_editor() as editor:

tests/requirements/base.txt

@@ -5,3 +5,4 @@ Pillow
 PyYAML
 pytz > dev
 selenium
+sqlparse