sqlspec 0.14.1__py3-none-any.whl → 0.16.0__py3-none-any.whl

sqlspec/{statement → core}/splitter.py

@@ -1,19 +1,35 @@
-"""SQL script statement splitter with dialect-aware lexer-driven state machine.
-
-This module provides a robust way to split SQL scripts into individual statements,
-handling complex constructs like PL/SQL blocks, T-SQL batches, and nested blocks.
+"""SQL statement splitter with caching and dialect support.
+
+This module provides a SQL script statement splitter with caching and
+multiple dialect support.
+
+Components:
+- StatementSplitter: SQL splitter with caching
+- DialectConfig: Dialect configuration system
+- Token/TokenType: Tokenization system
+- Caching: LRU caching for split results
+- Pattern compilation caching
+
+Features:
+- Support for multiple SQL dialects (Oracle, T-SQL, PostgreSQL, MySQL, SQLite, DuckDB, BigQuery)
+- Cached pattern compilation
+- LRU caching for split results
+- Optimized tokenization
+- Complete preservation of split_sql_script function
 """
 
 import re
+import threading
 from abc import ABC, abstractmethod
 from collections.abc import Generator
-from dataclasses import dataclass
 from enum import Enum
 from re import Pattern
-from typing import Callable, Optional, Union
+from typing import Any, Callable, Optional, Union
 
+from mypy_extensions import mypyc_attr
 from typing_extensions import TypeAlias
 
+from sqlspec.core.cache import CacheKey, UnifiedCache
 from sqlspec.utils.logging import get_logger
 
 __all__ = (
@@ -27,8 +43,33 @@ __all__ = (
     "split_sql_script",
 )
 
+logger = get_logger("sqlspec.core.splitter")
+
+DEFAULT_PATTERN_CACHE_SIZE = 1000  # Compiled regex patterns
+DEFAULT_RESULT_CACHE_SIZE = 5000  # Split results
+DEFAULT_CACHE_TTL = 3600  # 1 hour TTL
 
-logger = get_logger("sqlspec")
+DIALECT_CONFIG_SLOTS = (
+    "_block_starters",
+    "_block_enders",
+    "_statement_terminators",
+    "_batch_separators",
+    "_special_terminators",
+    "_max_nesting_depth",
+    "_name",
+)
+
+TOKEN_SLOTS = ("type", "value", "line", "column", "position")
+
+SPLITTER_SLOTS = (
+    "_dialect",
+    "_strip_trailing_semicolon",
+    "_token_patterns",
+    "_compiled_patterns",
+    "_pattern_cache_key",
+    "_result_cache",
+    "_pattern_cache",
+)
 
 
 class TokenType(Enum):
@@ -45,15 +86,21 @@ class TokenType(Enum):
     OTHER = "OTHER"
 
 
-@dataclass
+@mypyc_attr(allow_interpreted_subclasses=True)
 class Token:
-    """Represents a single token in the SQL script."""
+    """SQL token with metadata."""
+
+    __slots__ = TOKEN_SLOTS
 
-    type: TokenType
-    value: str
-    line: int
-    column: int
-    position: int  # Absolute position in the script
+    def __init__(self, type: TokenType, value: str, line: int, column: int, position: int) -> None:
+        self.type = type
+        self.value = value
+        self.line = line
+        self.column = column
+        self.position = position
+
+    def __repr__(self) -> str:
+        return f"Token({self.type.value}, {self.value!r}, {self.line}:{self.column})"
 
 
 TokenHandler: TypeAlias = Callable[[str, int, int, int], Optional[Token]]
@@ -61,9 +108,22 @@ TokenPattern: TypeAlias = Union[str, TokenHandler]
 CompiledTokenPattern: TypeAlias = Union[Pattern[str], TokenHandler]
 
 
+@mypyc_attr(allow_interpreted_subclasses=True)
 class DialectConfig(ABC):
     """Abstract base class for SQL dialect configurations."""
 
+    __slots__ = DIALECT_CONFIG_SLOTS
+
+    def __init__(self) -> None:
+        """Initialize dialect configuration."""
+        self._name: Optional[str] = None
+        self._block_starters: Optional[set[str]] = None
+        self._block_enders: Optional[set[str]] = None
+        self._statement_terminators: Optional[set[str]] = None
+        self._batch_separators: Optional[set[str]] = None
+        self._special_terminators: Optional[dict[str, Callable[[list[Token], int], bool]]] = None
+        self._max_nesting_depth: Optional[int] = None
+
     @property
     @abstractmethod
     def name(self) -> str:
@@ -87,44 +147,44 @@ class DialectConfig(ABC):
     @property
     def batch_separators(self) -> set[str]:
         """Keywords that separate batches (e.g., GO for T-SQL)."""
-        return set()
+        if self._batch_separators is None:
+            self._batch_separators = set()
+        return self._batch_separators
 
     @property
     def special_terminators(self) -> dict[str, Callable[[list[Token], int], bool]]:
         """Special terminators that need custom handling."""
-        return {}
+        if self._special_terminators is None:
+            self._special_terminators = {}
+        return self._special_terminators
 
     @property
     def max_nesting_depth(self) -> int:
         """Maximum allowed nesting depth for blocks."""
-        return 256
+        if self._max_nesting_depth is None:
+            self._max_nesting_depth = 256
+        return self._max_nesting_depth
 
     def get_all_token_patterns(self) -> list[tuple[TokenType, TokenPattern]]:
         """Assembles the complete, ordered list of token regex patterns."""
-        # 1. Start with high-precedence patterns
         patterns: list[tuple[TokenType, TokenPattern]] = [
             (TokenType.COMMENT_LINE, r"--[^\n]*"),
             (TokenType.COMMENT_BLOCK, r"/\*[\s\S]*?\*/"),
             (TokenType.STRING_LITERAL, r"'(?:[^']|'')*'"),
-            (TokenType.QUOTED_IDENTIFIER, r'"[^"]*"|\[[^\]]*\]'),  # Standard and T-SQL
+            (TokenType.QUOTED_IDENTIFIER, r'"[^"]*"|\[[^\]]*\]'),
         ]
 
-        # 2. Add dialect-specific patterns (can be overridden)
         patterns.extend(self._get_dialect_specific_patterns())
 
-        # 3. Dynamically build and insert keyword/separator patterns
         all_keywords = self.block_starters | self.block_enders | self.batch_separators
         if all_keywords:
             sorted_keywords = sorted(all_keywords, key=len, reverse=True)
             patterns.append((TokenType.KEYWORD, r"\b(" + "|".join(re.escape(kw) for kw in sorted_keywords) + r")\b"))
 
-        # 4. Add terminators
         all_terminators = self.statement_terminators | set(self.special_terminators.keys())
         if all_terminators:
-            # Escape special regex characters
             patterns.append((TokenType.TERMINATOR, "|".join(re.escape(t) for t in all_terminators)))
 
-        # 5. Add low-precedence patterns
         patterns.extend([(TokenType.WHITESPACE, r"\s+"), (TokenType.OTHER, r".")])
 
         return patterns
@@ -134,21 +194,12 @@ class DialectConfig(ABC):
         return []
 
     @staticmethod
-    def is_real_block_ender(tokens: list[Token], current_pos: int) -> bool:
-        """Check if this END keyword is actually a block ender.
-
-        Override in dialect configs to handle cases like END IF, END LOOP, etc.
-        that are not true block enders.
-        """
-        _ = tokens, current_pos  # Default implementation doesn't use these
+    def is_real_block_ender(tokens: list[Token], current_pos: int) -> bool:  # noqa: ARG004
+        """Check if this END keyword is actually a block ender."""
         return True
 
     def should_delay_semicolon_termination(self, tokens: list[Token], current_pos: int) -> bool:
-        """Check if semicolon termination should be delayed.
-
-        Override in dialect configs to handle special cases like Oracle END; /
-        """
-        _ = tokens, current_pos  # Default implementation doesn't use these
+        """Check if semicolon termination should be delayed."""
         return False
 
 
@@ -157,30 +208,36 @@ class OracleDialectConfig(DialectConfig):
 
     @property
     def name(self) -> str:
-        return "oracle"
+        if self._name is None:
+            self._name = "oracle"
+        return self._name
 
     @property
     def block_starters(self) -> set[str]:
-        return {"BEGIN", "DECLARE", "CASE"}
+        if self._block_starters is None:
+            self._block_starters = {"BEGIN", "DECLARE", "CASE"}
+        return self._block_starters
 
     @property
     def block_enders(self) -> set[str]:
-        return {"END"}
+        if self._block_enders is None:
+            self._block_enders = {"END"}
+        return self._block_enders
 
     @property
     def statement_terminators(self) -> set[str]:
-        return {";"}
+        if self._statement_terminators is None:
+            self._statement_terminators = {";"}
+        return self._statement_terminators
 
     @property
     def special_terminators(self) -> dict[str, Callable[[list[Token], int], bool]]:
-        return {"/": self._handle_slash_terminator}
+        if self._special_terminators is None:
+            self._special_terminators = {"/": self._handle_slash_terminator}
+        return self._special_terminators
 
     def should_delay_semicolon_termination(self, tokens: list[Token], current_pos: int) -> bool:
-        """Check if we should delay semicolon termination to look for a slash.
-
-        In Oracle, after END; we should check if there's a / coming up on its own line.
-        """
-        # Look backwards to see if we just processed an END token
+        """Check if we should delay semicolon termination to look for a slash."""
         pos = current_pos - 1
         while pos >= 0:
             token = tokens[pos]
@@ -188,10 +245,7 @@ class OracleDialectConfig(DialectConfig):
                 pos -= 1
                 continue
             if token.type == TokenType.KEYWORD and token.value.upper() == "END":
-                # We found END just before this semicolon
-                # Now look ahead to see if there's a / on its own line
                 return self._has_upcoming_slash(tokens, current_pos)
-            # Found something else, not an END
             break
 
         return False
@@ -209,25 +263,17 @@ class OracleDialectConfig(DialectConfig):
                 pos += 1
                 continue
             if token.type == TokenType.TERMINATOR and token.value == "/":
-                # Found a /, check if it's valid (on its own line)
                 return found_newline and self._handle_slash_terminator(tokens, pos)
             if token.type in {TokenType.COMMENT_LINE, TokenType.COMMENT_BLOCK}:
-                # Skip comments
                 pos += 1
                 continue
-            # Found non-whitespace, non-comment content
             break
 
         return False
 
     @staticmethod
     def is_real_block_ender(tokens: list[Token], current_pos: int) -> bool:
-        """Check if this END keyword is actually a block ender.
-
-        In Oracle PL/SQL, END followed by IF, LOOP, CASE etc. are not block enders
-        for BEGIN blocks - they terminate control structures.
-        """
-        # Look ahead for the next non-whitespace token(s)
+        """Check if this END keyword is actually a block ender for Oracle PL/SQL."""
         pos = current_pos + 1
         while pos < len(tokens):
             next_token = tokens[pos]
@@ -236,7 +282,6 @@ class OracleDialectConfig(DialectConfig):
                 pos += 1
                 continue
             if next_token.type == TokenType.OTHER:
-                # Collect consecutive OTHER tokens to form a word
                 word_chars = []
                 word_pos = pos
                 while word_pos < len(tokens) and tokens[word_pos].type == TokenType.OTHER:
@@ -245,26 +290,23 @@ class OracleDialectConfig(DialectConfig):
 
                 word = "".join(word_chars).upper()
                 if word in {"IF", "LOOP", "CASE", "WHILE"}:
-                    return False  # This is not a block ender
-            # Found a non-whitespace token that's not one of our special cases
+                    return False
             break
-        return True  # This is a real block ender
+        return True
 
     @staticmethod
     def _handle_slash_terminator(tokens: list[Token], current_pos: int) -> bool:
         """Oracle / must be on its own line after whitespace only."""
         if current_pos == 0:
-            return True  # / at start is valid
+            return True
 
-        # Look backwards to find the start of the line
         pos = current_pos - 1
         while pos >= 0:
             token = tokens[pos]
             if "\n" in token.value:
-                # Found newline, check if only whitespace between newline and /
                 break
             if token.type not in {TokenType.WHITESPACE, TokenType.COMMENT_LINE}:
-                return False  # Non-whitespace before / on same line
+                return False
             pos -= 1
 
         return True
@@ -275,31 +317,33 @@ class TSQLDialectConfig(DialectConfig):
 
     @property
     def name(self) -> str:
-        return "tsql"
+        if self._name is None:
+            self._name = "tsql"
+        return self._name
 
     @property
     def block_starters(self) -> set[str]:
-        return {"BEGIN", "TRY"}
+        if self._block_starters is None:
+            self._block_starters = {"BEGIN", "TRY"}
+        return self._block_starters
 
     @property
     def block_enders(self) -> set[str]:
-        return {"END", "CATCH"}
+        if self._block_enders is None:
+            self._block_enders = {"END", "CATCH"}
+        return self._block_enders
 
     @property
     def statement_terminators(self) -> set[str]:
-        return {";"}
+        if self._statement_terminators is None:
+            self._statement_terminators = {";"}
+        return self._statement_terminators
 
     @property
     def batch_separators(self) -> set[str]:
-        return {"GO"}
-
-    @staticmethod
-    def validate_batch_separator(tokens: list[Token], current_pos: int) -> bool:
-        """GO must be the only keyword on its line."""
-        # Look for non-whitespace tokens on the same line
-        # Implementation similar to Oracle slash handler
-        _ = tokens, current_pos  # Simplified implementation
-        return True  # Simplified for now
+        if self._batch_separators is None:
+            self._batch_separators = {"GO"}
+        return self._batch_separators
 
 
 class PostgreSQLDialectConfig(DialectConfig):
@@ -307,19 +351,27 @@ class PostgreSQLDialectConfig(DialectConfig):
 
     @property
     def name(self) -> str:
-        return "postgresql"
+        if self._name is None:
+            self._name = "postgresql"
+        return self._name
 
     @property
     def block_starters(self) -> set[str]:
-        return {"BEGIN", "DECLARE", "CASE", "DO"}
+        if self._block_starters is None:
+            self._block_starters = {"BEGIN", "DECLARE", "CASE", "DO"}
+        return self._block_starters
 
     @property
     def block_enders(self) -> set[str]:
-        return {"END"}
+        if self._block_enders is None:
+            self._block_enders = {"END"}
+        return self._block_enders
 
     @property
     def statement_terminators(self) -> set[str]:
-        return {";"}
+        if self._statement_terminators is None:
+            self._statement_terminators = {";"}
+        return self._statement_terminators
 
     def _get_dialect_specific_patterns(self) -> list[tuple[TokenType, TokenPattern]]:
         """Add PostgreSQL-specific patterns like dollar-quoted strings."""
@@ -328,12 +380,11 @@ class PostgreSQLDialectConfig(DialectConfig):
     @staticmethod
     def _handle_dollar_quoted_string(text: str, position: int, line: int, column: int) -> Optional[Token]:
         """Handle PostgreSQL dollar-quoted strings like $tag$...$tag$."""
-        # Match opening tag
         start_match = re.match(r"\$([a-zA-Z_][a-zA-Z0-9_]*)?\$", text[position:])
         if not start_match:
             return None
 
-        tag = start_match.group(0)  # The full opening tag, e.g., "$tag$"
+        tag = start_match.group(0)
         content_start = position + len(tag)
 
         try:
@@ -342,7 +393,6 @@ class PostgreSQLDialectConfig(DialectConfig):
 
             return Token(type=TokenType.STRING_LITERAL, value=full_value, line=line, column=column, position=position)
         except ValueError:
-            # Closing tag not found
             return None
 
 
@@ -351,19 +401,27 @@ class GenericDialectConfig(DialectConfig):
 
     @property
     def name(self) -> str:
-        return "generic"
+        if self._name is None:
+            self._name = "generic"
+        return self._name
 
     @property
     def block_starters(self) -> set[str]:
-        return {"BEGIN", "DECLARE", "CASE"}
+        if self._block_starters is None:
+            self._block_starters = {"BEGIN", "DECLARE", "CASE"}
+        return self._block_starters
 
     @property
     def block_enders(self) -> set[str]:
-        return {"END"}
+        if self._block_enders is None:
+            self._block_enders = {"END"}
+        return self._block_enders
 
     @property
     def statement_terminators(self) -> set[str]:
-        return {";"}
+        if self._statement_terminators is None:
+            self._statement_terminators = {";"}
+        return self._statement_terminators
 
 
 class MySQLDialectConfig(DialectConfig):
@@ -371,24 +429,33 @@ class MySQLDialectConfig(DialectConfig):
 
     @property
     def name(self) -> str:
-        return "mysql"
+        if self._name is None:
+            self._name = "mysql"
+        return self._name
 
     @property
     def block_starters(self) -> set[str]:
-        return {"BEGIN", "DECLARE", "CASE"}
+        if self._block_starters is None:
+            self._block_starters = {"BEGIN", "DECLARE", "CASE"}
+        return self._block_starters
 
     @property
     def block_enders(self) -> set[str]:
-        return {"END"}
+        if self._block_enders is None:
+            self._block_enders = {"END"}
+        return self._block_enders
 
     @property
     def statement_terminators(self) -> set[str]:
-        return {";"}
+        if self._statement_terminators is None:
+            self._statement_terminators = {";"}
+        return self._statement_terminators
 
     @property
     def special_terminators(self) -> dict[str, Callable[[list[Token], int], bool]]:
-        """MySQL supports DELIMITER command for changing terminators."""
-        return {"\\g": lambda _tokens, _pos: True, "\\G": lambda _tokens, _pos: True}
+        if self._special_terminators is None:
+            self._special_terminators = {"\\g": lambda _tokens, _pos: True, "\\G": lambda _tokens, _pos: True}
+        return self._special_terminators
 
 
 class SQLiteDialectConfig(DialectConfig):
@@ -396,20 +463,27 @@ class SQLiteDialectConfig(DialectConfig):
 
     @property
     def name(self) -> str:
-        return "sqlite"
+        if self._name is None:
+            self._name = "sqlite"
+        return self._name
 
     @property
     def block_starters(self) -> set[str]:
-        # SQLite has limited block support
-        return {"BEGIN", "CASE"}
+        if self._block_starters is None:
+            self._block_starters = {"BEGIN", "CASE"}
+        return self._block_starters
 
     @property
     def block_enders(self) -> set[str]:
-        return {"END"}
+        if self._block_enders is None:
+            self._block_enders = {"END"}
+        return self._block_enders
 
     @property
     def statement_terminators(self) -> set[str]:
-        return {";"}
+        if self._statement_terminators is None:
+            self._statement_terminators = {";"}
+        return self._statement_terminators
 
 
 class DuckDBDialectConfig(DialectConfig):
@@ -417,19 +491,27 @@ class DuckDBDialectConfig(DialectConfig):
 
     @property
     def name(self) -> str:
-        return "duckdb"
+        if self._name is None:
+            self._name = "duckdb"
+        return self._name
 
     @property
     def block_starters(self) -> set[str]:
-        return {"BEGIN", "CASE"}
+        if self._block_starters is None:
+            self._block_starters = {"BEGIN", "CASE"}
+        return self._block_starters
 
     @property
     def block_enders(self) -> set[str]:
-        return {"END"}
+        if self._block_enders is None:
+            self._block_enders = {"END"}
+        return self._block_enders
 
     @property
     def statement_terminators(self) -> set[str]:
-        return {";"}
+        if self._statement_terminators is None:
+            self._statement_terminators = {";"}
+        return self._statement_terminators
 
 
 class BigQueryDialectConfig(DialectConfig):
@@ -437,56 +519,97 @@ class BigQueryDialectConfig(DialectConfig):
 
     @property
     def name(self) -> str:
-        return "bigquery"
+        if self._name is None:
+            self._name = "bigquery"
+        return self._name
 
     @property
     def block_starters(self) -> set[str]:
-        return {"BEGIN", "CASE"}
+        if self._block_starters is None:
+            self._block_starters = {"BEGIN", "CASE"}
+        return self._block_starters
 
     @property
     def block_enders(self) -> set[str]:
-        return {"END"}
+        if self._block_enders is None:
+            self._block_enders = {"END"}
+        return self._block_enders
 
     @property
     def statement_terminators(self) -> set[str]:
-        return {";"}
+        if self._statement_terminators is None:
+            self._statement_terminators = {";"}
+        return self._statement_terminators
+
+
+_pattern_cache: Optional[UnifiedCache[list[tuple[TokenType, CompiledTokenPattern]]]] = None
+_result_cache: Optional[UnifiedCache[list[str]]] = None
+_cache_lock = threading.Lock()
+
 
+def _get_pattern_cache() -> UnifiedCache[list[tuple[TokenType, CompiledTokenPattern]]]:
+    """Get or create the pattern compilation cache."""
+    global _pattern_cache
+    if _pattern_cache is None:
+        with _cache_lock:
+            if _pattern_cache is None:
+                _pattern_cache = UnifiedCache[list[tuple[TokenType, CompiledTokenPattern]]](
+                    max_size=DEFAULT_PATTERN_CACHE_SIZE, ttl_seconds=DEFAULT_CACHE_TTL
+                )
+    return _pattern_cache
 
+
+def _get_result_cache() -> UnifiedCache[list[str]]:
+    """Get or create the result cache."""
+    global _result_cache
+    if _result_cache is None:
+        with _cache_lock:
+            if _result_cache is None:
+                _result_cache = UnifiedCache[list[str]](
+                    max_size=DEFAULT_RESULT_CACHE_SIZE, ttl_seconds=DEFAULT_CACHE_TTL
+                )
+    return _result_cache
+
+
+@mypyc_attr(allow_interpreted_subclasses=False)
 class StatementSplitter:
-    """Splits SQL scripts into individual statements using a lexer-driven state machine."""
+    """SQL script splitter with caching and dialect support."""
+
+    __slots__ = SPLITTER_SLOTS
 
     def __init__(self, dialect: DialectConfig, strip_trailing_semicolon: bool = False) -> None:
-        """Initialize the splitter with a specific dialect configuration.
-
-        Args:
-            dialect: The dialect configuration to use
-            strip_trailing_semicolon: If True, remove trailing semicolons from statements
-        """
-        self.dialect = dialect
-        self.strip_trailing_semicolon = strip_trailing_semicolon
-        self.token_patterns = dialect.get_all_token_patterns()
-        self._compiled_patterns = self._compile_patterns()
-
-    def _compile_patterns(self) -> list[tuple[TokenType, CompiledTokenPattern]]:
-        """Compile regex patterns for efficiency."""
+        """Initialize the splitter with caching and dialect support."""
+        self._dialect = dialect
+        self._strip_trailing_semicolon = strip_trailing_semicolon
+        self._token_patterns = dialect.get_all_token_patterns()
+
+        self._pattern_cache_key = f"{dialect.name}:{hash(tuple(str(p) for _, p in self._token_patterns))}"
+
+        self._pattern_cache = _get_pattern_cache()
+        self._result_cache = _get_result_cache()
+
+        self._compiled_patterns = self._get_or_compile_patterns()
+
+    def _get_or_compile_patterns(self) -> list[tuple[TokenType, CompiledTokenPattern]]:
+        """Get compiled patterns from cache or compile and cache them."""
+        cache_key = CacheKey(("pattern", self._pattern_cache_key))
+
+        cached_patterns = self._pattern_cache.get(cache_key)
+        if cached_patterns is not None:
+            return cached_patterns
+
         compiled: list[tuple[TokenType, CompiledTokenPattern]] = []
-        for token_type, pattern in self.token_patterns:
+        for token_type, pattern in self._token_patterns:
             if isinstance(pattern, str):
                 compiled.append((token_type, re.compile(pattern, re.IGNORECASE | re.DOTALL)))
             else:
-                # It's a callable
                 compiled.append((token_type, pattern))
+
+        self._pattern_cache.put(cache_key, compiled)
         return compiled
 
     def _tokenize(self, sql: str) -> Generator[Token, None, None]:
-        """Tokenize the SQL script into a stream of tokens.
-
-        sql: The SQL script to tokenize
-
-        Yields:
-            Token objects representing the recognized tokens in the script.
-
-        """
+        """Tokenize SQL string."""
         pos = 0
         line = 1
         line_start = 0
@@ -496,7 +619,6 @@ class StatementSplitter:
 
             for token_type, pattern in self._compiled_patterns:
                 if callable(pattern):
-                    # Call the handler function
                     column = pos - line_start + 1
                     token = pattern(sql, pos, line, column)
                     if token:
@@ -511,7 +633,6 @@ class StatementSplitter:
                         matched = True
                         break
                 else:
-                    # Use regex
                     match = pattern.match(sql, pos)
                     if match:
                         value = match.group(0)
@@ -529,12 +650,25 @@ class StatementSplitter:
                         break
 
             if not matched:
-                # This should never happen with our catch-all OTHER pattern
                 logger.error("Failed to tokenize at position %d: %s", pos, sql[pos : pos + 20])
-                pos += 1  # Skip the problematic character
+                pos += 1
 
     def split(self, sql: str) -> list[str]:
-        """Split the SQL script into individual statements."""
+        """Split SQL script with result caching."""
+        script_hash = hash(sql)
+        cache_key = CacheKey(("split", self._dialect.name, script_hash, self._strip_trailing_semicolon))
+
+        cached_result = self._result_cache.get(cache_key)
+        if cached_result is not None:
+            return cached_result
+
+        statements = self._do_split(sql)
+
+        self._result_cache.put(cache_key, statements)
+        return statements
+
+    def _do_split(self, sql: str) -> list[str]:
+        """Perform SQL script splitting."""
         statements = []
         current_statement_tokens = []
         current_statement_chars = []
@@ -543,10 +677,8 @@ class StatementSplitter:
         all_tokens = list(self._tokenize(sql))
 
         for token_idx, token in enumerate(all_tokens):
-            # Always accumulate the original text
             current_statement_chars.append(token.value)
 
-            # Skip whitespace and comments for logic (but keep in output)
             if token.type in {TokenType.WHITESPACE, TokenType.COMMENT_LINE, TokenType.COMMENT_BLOCK}:
                 current_statement_tokens.append(token)
                 continue
@@ -555,50 +687,41 @@ class StatementSplitter:
             token_upper = token.value.upper()
 
             if token.type == TokenType.KEYWORD:
-                if token_upper in self.dialect.block_starters:
+                if token_upper in self._dialect.block_starters:
                     block_stack.append(token_upper)
-                    if len(block_stack) > self.dialect.max_nesting_depth:
-                        msg = f"Maximum nesting depth ({self.dialect.max_nesting_depth}) exceeded"
+                    if len(block_stack) > self._dialect.max_nesting_depth:
+                        msg = f"Maximum nesting depth ({self._dialect.max_nesting_depth}) exceeded"
                         raise ValueError(msg)
-                elif token_upper in self.dialect.block_enders:
-                    if block_stack and self.dialect.is_real_block_ender(all_tokens, token_idx):
+                elif token_upper in self._dialect.block_enders:
+                    if block_stack and self._dialect.is_real_block_ender(all_tokens, token_idx):
                         block_stack.pop()
 
-            # Check for statement termination
             is_terminator = False
-            if not block_stack:  # Only terminate when not inside a block
+            if not block_stack:
                 if token.type == TokenType.TERMINATOR:
-                    if token.value in self.dialect.statement_terminators:
-                        should_delay = self.dialect.should_delay_semicolon_termination(all_tokens, token_idx)
+                    if token.value in self._dialect.statement_terminators:
+                        should_delay = self._dialect.should_delay_semicolon_termination(all_tokens, token_idx)
 
-                        # Also check if there's a batch separator coming up (for T-SQL GO)
-                        if not should_delay and token.value == ";" and self.dialect.batch_separators:
-                            # In dialects with batch separators, semicolons don't terminate
-                            # statements - only batch separators do
+                        if not should_delay and token.value == ";" and self._dialect.batch_separators:
                             should_delay = True
 
                         if not should_delay:
                             is_terminator = True
-                    elif token.value in self.dialect.special_terminators:
-                        # Call the handler to validate
-                        handler = self.dialect.special_terminators[token.value]
+                    elif token.value in self._dialect.special_terminators:
+                        handler = self._dialect.special_terminators[token.value]
                         if handler(all_tokens, token_idx):
                             is_terminator = True
 
-                elif token.type == TokenType.KEYWORD and token_upper in self.dialect.batch_separators:
-                    # Batch separators like GO should be included with the preceding statement
+                elif token.type == TokenType.KEYWORD and token_upper in self._dialect.batch_separators:
                     is_terminator = True
 
             if is_terminator:
-                # Save the statement
                 statement = "".join(current_statement_chars).strip()
 
                 is_plsql_block = self._is_plsql_block(current_statement_tokens)
 
-                # Optionally strip the trailing terminator
-                # For PL/SQL blocks, never strip the semicolon as it's syntactically required
                 if (
-                    self.strip_trailing_semicolon
+                    self._strip_trailing_semicolon
                     and token.type == TokenType.TERMINATOR
                     and statement.endswith(token.value)
                     and not is_plsql_block
@@ -619,29 +742,14 @@ class StatementSplitter:
 
     @staticmethod
     def _is_plsql_block(tokens: list[Token]) -> bool:
-        """Check if the token list represents a PL/SQL block.
-
-        Args:
-            tokens: List of tokens for the current statement
-
-        Returns:
-            True if this is a PL/SQL block (BEGIN...END or DECLARE...END)
-        """
+        """Check if the token list represents a PL/SQL block."""
         for token in tokens:
             if token.type == TokenType.KEYWORD:
                 return token.value.upper() in {"BEGIN", "DECLARE"}
         return False
 
     def _contains_executable_content(self, statement: str) -> bool:
-        """Check if a statement contains actual executable content (not just comments/whitespace).
-
-        Args:
-            statement: The statement string to check
-
-        Returns:
-            True if the statement contains executable SQL, False if it's only comments/whitespace
-        """
-        # Tokenize the statement to check its content
+        """Check if a statement contains actual executable content."""
         tokens = list(self._tokenize(statement))
 
         for token in tokens:
@@ -651,39 +759,61 @@ class StatementSplitter:
         return False
 
 
-def split_sql_script(script: str, dialect: str = "generic", strip_trailing_semicolon: bool = False) -> list[str]:
-    """Split a SQL script into statements using the appropriate dialect.
+def split_sql_script(script: str, dialect: Optional[str] = None, strip_trailing_terminator: bool = False) -> list[str]:
+    """Split SQL script into individual statements.
 
     Args:
         script: The SQL script to split
-        dialect: The SQL dialect name ('oracle', 'tsql', 'postgresql', etc.)
-        strip_trailing_semicolon: If True, remove trailing terminators from statements
+        dialect: The SQL dialect name
+        strip_trailing_terminator: If True, remove trailing terminators from statements
 
     Returns:
         List of individual SQL statements
     """
+    if dialect is None:
+        dialect = "generic"
+
     dialect_configs = {
-        # Standard dialects
         "generic": GenericDialectConfig(),
-        # Major databases
         "oracle": OracleDialectConfig(),
         "tsql": TSQLDialectConfig(),
-        "mssql": TSQLDialectConfig(),  # Alias for tsql
-        "sqlserver": TSQLDialectConfig(),  # Alias for tsql
+        "mssql": TSQLDialectConfig(),
+        "sqlserver": TSQLDialectConfig(),
         "postgresql": PostgreSQLDialectConfig(),
-        "postgres": PostgreSQLDialectConfig(),  # Common alias
+        "postgres": PostgreSQLDialectConfig(),
         "mysql": MySQLDialectConfig(),
         "sqlite": SQLiteDialectConfig(),
-        # Modern analytical databases
         "duckdb": DuckDBDialectConfig(),
         "bigquery": BigQueryDialectConfig(),
     }
 
     config = dialect_configs.get(dialect.lower())
     if not config:
-        # Fall back to generic config for unknown dialects
         logger.warning("Unknown dialect '%s', using generic SQL splitter", dialect)
         config = GenericDialectConfig()
 
-    splitter = StatementSplitter(config, strip_trailing_semicolon=strip_trailing_semicolon)
+    splitter = StatementSplitter(config, strip_trailing_semicolon=strip_trailing_terminator)
     return splitter.split(script)
+
+
+def clear_splitter_caches() -> None:
+    """Clear all splitter caches for memory management."""
+    pattern_cache = _get_pattern_cache()
+    result_cache = _get_result_cache()
+    pattern_cache.clear()
+    result_cache.clear()
+
+
+def get_splitter_cache_stats() -> dict[str, Any]:
+    """Get statistics from splitter caches.
+
+    Returns:
+        Dictionary containing cache statistics
+    """
+    pattern_cache = _get_pattern_cache()
+    result_cache = _get_result_cache()
+
+    return {
+        "pattern_cache": {"size": pattern_cache.size(), "stats": pattern_cache.get_stats()},
+        "result_cache": {"size": result_cache.size(), "stats": result_cache.get_stats()},
+    }