[1.7.x] 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.

Backport of 8b5b199 from master

3 files changed, 38 insertions, 0 deletions

diff --git a/django/db/backends/__init__.py b/django/db/backends/__init__.py
index 10836f4651..4556fddd94 100644
--- a/django/db/backends/__init__.py
+++ b/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
@@ -697,6 +699,10 @@ 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
 
@@ -972,6 +978,34 @@ 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
diff --git a/django/db/backends/postgresql_psycopg2/base.py b/django/db/backends/postgresql_psycopg2/base.py
index cc44075ba7..91a183e8b5 100644
--- a/django/db/backends/postgresql_psycopg2/base.py
+++ b/django/db/backends/postgresql_psycopg2/base.py
@@ -59,6 +59,7 @@ 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):
diff --git a/django/db/backends/postgresql_psycopg2/operations.py b/django/db/backends/postgresql_psycopg2/operations.py
index 9285e6eeca..b9d0231768 100644
--- a/django/db/backends/postgresql_psycopg2/operations.py
+++ b/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.