Force split on Pattern and Scanner regexp entry points

Author: jirkamarsik

graalpython/com.oracle.graal.python/src/com/oracle/graal/python/builtins/modules/GraalPythonModuleBuiltins.java

@@ -504,6 +504,16 @@ public Object doIt(PFunction func,
         }
     }
 
+    @Builtin(name = "force_split_direct_calls", minNumOfPositionalArgs = 1)
+    @GenerateNodeFactory
+    public abstract static class ForceSplitDirectCallsNode extends PythonUnaryBuiltinNode {
+        @Specialization
+        public Object doIt(PFunction func) {
+            func.setForceSplitDirectCalls(true);
+            return func;
+        }
+    }
+
     @Builtin(name = "get_toolchain_tools_for_venv")
     @TypeSystemReference(PythonArithmeticTypes.class)
     @GenerateNodeFactory

graalpython/com.oracle.graal.python/src/com/oracle/graal/python/builtins/objects/function/PFunction.java

@@ -56,6 +56,7 @@
 public final class PFunction extends PythonObject {
     private TruffleString name;
     private TruffleString qualname;
+    private boolean forceSplitDirectCalls;
     private final Assumption codeStableAssumption;
     private final Assumption defaultsStableAssumption;
     private final PythonObject globals;
@@ -89,6 +90,7 @@ public PFunction(PythonLanguage lang, TruffleString name, TruffleString qualname
         this.closure = closure;
         this.codeStableAssumption = codeStableAssumption;
         this.defaultsStableAssumption = defaultsStableAssumption;
+        this.forceSplitDirectCalls = false;
     }
 
     public Assumption getCodeStableAssumption() {
@@ -131,6 +133,14 @@ public void setBuiltin(boolean builtin) {
         isBuiltin = builtin;
     }
 
+    public void setForceSplitDirectCalls(boolean forceSplitDirectCalls) {
+        this.forceSplitDirectCalls = forceSplitDirectCalls;
+    }
+
+    public boolean forceSplitDirectCalls() {
+        return forceSplitDirectCalls;
+    }
+
     @Override
     public final String toString() {
         CompilerAsserts.neverPartOfCompilation();

graalpython/com.oracle.graal.python/src/com/oracle/graal/python/nodes/call/FunctionInvokeNode.java

@@ -72,10 +72,10 @@ public abstract class FunctionInvokeNode extends DirectInvokeNode {
     private final PCell[] closure;
     protected final boolean isBuiltin;
 
-    protected FunctionInvokeNode(PFunction callee, CallTarget callTarget, PythonObject globals, PCell[] closure, boolean isBuiltin, boolean isGenerator) {
+    protected FunctionInvokeNode(PFunction callee, CallTarget callTarget, PythonObject globals, PCell[] closure, boolean isBuiltin, boolean isGenerator, boolean split) {
         this.callee = callee;
         this.callNode = Truffle.getRuntime().createDirectCallNode(callTarget);
-        if (isBuiltin && forceSplitBuiltins()) {
+        if (split) {
             callNode.cloneCallTarget();
         }
         if (isGenerator && shouldInlineGenerators()) {
@@ -117,15 +117,15 @@ public final RootNode getCurrentRootNode() {
     @TruffleBoundary
     public static FunctionInvokeNode create(PFunction callee) {
         RootCallTarget callTarget = getCallTarget(callee);
-        boolean builtin = isBuiltin(callee);
-        return FunctionInvokeNodeGen.create(callee, callTarget, callee.getGlobals(), callee.getClosure(), builtin, callTarget.getRootNode() instanceof PBytecodeGeneratorFunctionRootNode);
+        return FunctionInvokeNodeGen.create(callee, callTarget, callee.getGlobals(), callee.getClosure(), false, callTarget.getRootNode() instanceof PBytecodeGeneratorFunctionRootNode,
+                        callee.forceSplitDirectCalls());
     }
 
     @TruffleBoundary
     public static FunctionInvokeNode create(PBuiltinFunction callee) {
         RootCallTarget callTarget = getCallTarget(callee);
-        boolean builtin = isBuiltin(callee);
-        return FunctionInvokeNodeGen.create(null, callTarget, null, null, builtin, false);
+        boolean split = forceSplitBuiltins();
+        return FunctionInvokeNodeGen.create(null, callTarget, null, null, true, false, split);
     }
 
     /**
@@ -135,6 +135,6 @@ public static FunctionInvokeNode create(PBuiltinFunction callee) {
      */
     @TruffleBoundary
     public static FunctionInvokeNode createBuiltinFunction(RootCallTarget callTarget) {
-        return FunctionInvokeNodeGen.create(null, callTarget, null, null, true, false);
+        return FunctionInvokeNodeGen.create(null, callTarget, null, null, true, false, forceSplitBuiltins());
     }
 }

graalpython/lib-graalpython/_sre.py

@@ -327,12 +327,15 @@ def __class_getitem__(cls, item):
     def _search(self, string, pos, endpos, method=_METHOD_SEARCH, must_advance=False):
         return tregex_search(self, string, pos, endpos, method, must_advance)
 
+    @__graalpython__.force_split_direct_calls
     def search(self, string, pos=0, endpos=maxsize):
         return self._search(string, pos, endpos, method=_METHOD_SEARCH)
 
+    @__graalpython__.force_split_direct_calls
     def match(self, string, pos=0, endpos=maxsize):
         return self._search(string, pos, endpos, method=_METHOD_MATCH)
 
+    @__graalpython__.force_split_direct_calls
     def fullmatch(self, string, pos=0, endpos=maxsize):
         return self._search(string, pos, endpos, method=_METHOD_FULLMATCH)
 
@@ -346,6 +349,7 @@ def __sanitize_out_type(self, elem):
         else:
             return str(elem)
 
+    @__graalpython__.force_split_direct_calls
     def finditer(self, string, pos=0, endpos=maxsize):
         for must_advance in [False, True]:
             if tregex_compile(self, _METHOD_SEARCH, must_advance) is None:
@@ -368,6 +372,7 @@ def __finditer_gen(self, string, substring, pos, endpos):
             must_advance = (result.getStart(0) == result.getEnd(0))
         return
 
+    @__graalpython__.force_split_direct_calls
     def findall(self, string, pos=0, endpos=maxsize):
         for must_advance in [False, True]:
             if tregex_compile(self, _METHOD_SEARCH, must_advance) is None:
@@ -392,9 +397,11 @@ def findall(self, string, pos=0, endpos=maxsize):
             must_advance = (result.getStart(0) == result.getEnd(0))
         return matchlist
 
+    @__graalpython__.force_split_direct_calls
     def sub(self, repl, string, count=0):
         return self.subn(repl, string, count)[0]
 
+    @__graalpython__.force_split_direct_calls
     def subn(self, repl, string, count=0):
         for must_advance in [False, True]:
             if tregex_compile(self, _METHOD_SEARCH, must_advance) is None:
@@ -440,6 +447,7 @@ def subn(self, repl, string, count=0):
         else:
             return "".join(result), n
 
+    @__graalpython__.force_split_direct_calls
     def split(self, string, maxsplit=0):
         for must_advance in [False, True]:
             if tregex_compile(self, _METHOD_SEARCH, must_advance) is None:
@@ -500,9 +508,11 @@ def __match_search(self, method):
             self.__must_advance = match.start() == self.__start
         return match
 
+    @__graalpython__.force_split_direct_calls
     def match(self):
         return self.__match_search(_METHOD_MATCH)
 
+    @__graalpython__.force_split_direct_calls
     def search(self):
         return self.__match_search(_METHOD_SEARCH)