Intrinsify helper function PyTruffle_Unicode_FromFormat

Author: fangerer

graalpython/com.oracle.graal.python.cext/src/_warnings.c

@@ -72,15 +72,15 @@ int PyErr_WarnEx(PyObject *category, const char *text, Py_ssize_t stack_level) {
 NO_INLINE int PyErr_WarnFormat(PyObject *category, Py_ssize_t stack_level, const char *format, ...) {
     va_list args;
     va_start(args, format);
-    PyObject* result = PyTruffle_Unicode_FromFormat(format, args);
+    PyObject* result = PyUnicode_FromFormatV(format, args);
     va_end(args);
     return warn_unicode(category, result, stack_level, Py_None);
 }
 
 int PyErr_ResourceWarning(PyObject *source, Py_ssize_t stack_level, const char *format, ...) {
     va_list args;
     va_start(args, format);
-    PyObject* result = PyTruffle_Unicode_FromFormat(format, args);
+    PyObject* result = PyUnicode_FromFormatV(format, args);
     va_end(args);
     return warn_unicode(PyExc_ResourceWarning, result, stack_level, source);
 }

graalpython/com.oracle.graal.python.cext/src/capi.h

@@ -382,9 +382,6 @@ extern PyObject* wrapped_null;
 
 /* internal functions to avoid unnecessary managed <-> native conversions */
 
-/* STR */
-__attribute__((always_inline)) PyObject* PyTruffle_Unicode_FromFormat(const char *fmt, va_list va);
-
 /* BYTES, BYTEARRAY */
 int bytes_buffer_getbuffer(PyBytesObject *self, Py_buffer *view, int flags);
 int bytearray_getbuffer(PyByteArrayObject *obj, Py_buffer *view, int flags);

graalpython/com.oracle.graal.python.cext/src/errors.c

@@ -163,7 +163,7 @@ NO_INLINE
 PyObject* PyErr_Format(PyObject* exception, const char* fmt, ...) {
     va_list args;
     va_start(args, fmt);
-    PyObject* formatted_msg = PyTruffle_Unicode_FromFormat(fmt, args);
+    PyObject* formatted_msg = PyUnicode_FromFormatV(fmt, args);
     va_end(args);
     UPCALL_CEXT_VOID(_jls_PyErr_CreateAndSetException, native_to_java(exception), native_to_java(formatted_msg));
     return NULL;

graalpython/com.oracle.graal.python.cext/src/modsupport.c

@@ -96,10 +96,24 @@ static void init_upcall_PyTruffle_Arg_ParseTupleAndKeyword(void) {
     }
 
 /* just a renaming to avoid name clash with Java types */
-typedef char       char_t;
-typedef float      float_t;
-typedef double     double_t;
-
+typedef void*              void_ptr_t;
+typedef char               char_t;
+typedef float              float_t;
+typedef double             double_t;
+typedef int                int_t;
+typedef unsigned int       uint_t;
+typedef long               long_t;
+typedef unsigned long      ulong_t;
+typedef long long          longlong_t;
+typedef unsigned long long ulonglong_t;
+
+REGISTER_BASIC_TYPE(void_ptr_t);
+REGISTER_BASIC_TYPE(int_t);
+REGISTER_BASIC_TYPE(uint_t);
+REGISTER_BASIC_TYPE(long_t);
+REGISTER_BASIC_TYPE(ulong_t);
+REGISTER_BASIC_TYPE(longlong_t);
+REGISTER_BASIC_TYPE(ulonglong_t);
 REGISTER_BASIC_TYPE(int64_t);
 REGISTER_BASIC_TYPE(int32_t);
 REGISTER_BASIC_TYPE(int16_t);
@@ -114,6 +128,7 @@ REGISTER_BASIC_TYPE(PyObject);
 REGISTER_BASIC_TYPE(float_t);
 REGISTER_BASIC_TYPE(double_t);
 REGISTER_BASIC_TYPE(Py_ssize_t);
+REGISTER_BASIC_TYPE(size_t);
 
 /* For pointers, make them look like an array of size 1 such that it is
    possible to dereference the pointer by accessing element 0. */

graalpython/com.oracle.graal.python.cext/src/unicodeobject.c

@@ -230,146 +230,17 @@ PyObject * PyUnicode_FromStringAndSize(const char *u, Py_ssize_t size) {
     return to_sulong(polyglot_from_string_n(u, size, SRC_CS));
 }
 
-MUST_INLINE PyObject* PyTruffle_Unicode_FromFormat(const char *fmt, va_list va) {
-    va_list vacpy;
-    size_t fmt_size = strlen(fmt) + 1;
-    char* fmtcpy = strdup(fmt);
-    char* c = fmtcpy;
-
-    int remaining_space = 2047;
-    char* buffer = (char*)calloc(sizeof(char), remaining_space + 1);
-    char* full_buffer = buffer;
-
-    void *variable = NULL;
-    PyObject *tmp = NULL;
-    char *allocated = NULL; // points to the same as variable, if it has to be free'd
-
-    va_copy(vacpy, va);
-
-    while (c[0]) {
-        if (c[0] == '%') {
-            // we've reached the next directive, write until here
-            c[0] = '\0';
-            int bytes_written;
-            if (variable != NULL) {
-                bytes_written = snprintf(buffer, remaining_space, fmtcpy, variable);
-                if (allocated != NULL) {
-                    free(allocated);
-                    allocated = NULL;
-                }
-                variable = NULL;
-            } else {
-                bytes_written = vsnprintf(buffer, remaining_space, fmtcpy, vacpy);
-            }
-            remaining_space -= bytes_written;
-            buffer += bytes_written;
-            fmtcpy = c;
-            c[0] = '%';
-
-            // now decide if we need to do something special with this directive
-            PyObject* (*converter)(PyObject*) = NULL;
-            switch (c[1]) {
-            case 'A':
-                // The conversion cases, these all use a function to convert the
-                // PyObject* to a char* and they fall through
-                converter = PyObject_ASCII;
-            case 'U':
-                if (converter == NULL) converter = PyObject_Str;
-            case 'S':
-                if (converter == NULL) converter = PyObject_Str;
-            case 'R':
-                if (converter == NULL) converter = PyObject_Repr;
-                c[1] = 's';
-                tmp = va_arg(va, PyObject*);
-                // just advance cursor on 'vacpy'
-                va_arg(vacpy, PyObject*);
-                allocated = variable = as_char_pointer(converter(tmp));
-                break;
-            case 'c':
-                // This case should just treat it's argument as an integer
-                c[1] = 'd';
-                // intentionally fall-through
-            case 'd':
-            case 'i':
-            case 'x':
-                // just advance cursor on 'va'
-                va_arg(va, int);
-                break;
-            case 'u':
-                // just advance cursor on 'va'
-                va_arg(va, unsigned int);
-                break;
-            case 'l':
-                // 'c[2]' is guaranteed to be a correct memory access since 
-                // it will in the worst case be '\0'.
-                switch (c[2]) {
-                    // %ld, %li
-                    case 'd':
-                    case 'i':
-                        // just advance cursor on 'va'
-                        va_arg(va, long);
-                        break;
-                    // %lu
-                    case 'u':
-                        // just advance cursor on 'va'
-                        va_arg(va, unsigned long);
-                    // %ll?
-                    case 'l':
-                        // just advance cursor on 'va'
-                        va_arg(va, long long);
-                }
-                break;
-            case 'z':
-                // just advance cursor on 'va'
-                va_arg(va, Py_ssize_t);
-                break;
-            case 's':
-                // just advance cursor on 'va'
-                va_arg(va, char*);
-                break;
-            case 'p':
-                // just advance cursor on 'va'
-                va_arg(va, void*);
-                break;
-            case '%':
-                // literal %
-                break;
-            default:
-                variable = va_arg(va, PyObject*);
-                // just advance cursor on 'vacpy'
-                va_arg(vacpy, PyObject*);
-            }
-            // skip over next char, we checked it
-            c += 1;
-        }
-        c += 1;
-    }
-
-    // write the remaining buffer
-    if (variable != NULL) {
-        snprintf(buffer, remaining_space, fmtcpy, variable);
-        if (allocated) {
-            free(allocated);
-        }
-    } else {
-        vsnprintf(buffer, remaining_space, fmtcpy, vacpy);
-    }
-
-    PyObject* result = PyUnicode_FromString(full_buffer);
-    free(full_buffer);
-    va_end(vacpy);
-    return result;
-}
-
+typedef PyObject* (*unicode_fromformat_fun_t)(void* jstr, va_list va);
+UPCALL_TYPED_ID(PyTruffle_Unicode_FromFormat, unicode_fromformat_fun_t);
 PyObject* PyUnicode_FromFormatV(const char* format, va_list va) {
-    return PyTruffle_Unicode_FromFormat(format, va);
+    return _jls_PyTruffle_Unicode_FromFormat(polyglot_from_string(format, "ascii"), va);
 }
 
 NO_INLINE
 PyObject* PyUnicode_FromFormat(const char* format, ...) {
     va_list args;
     va_start(args, format);
-    PyObject* result = PyTruffle_Unicode_FromFormat(format, args);
+    PyObject* result = _jls_PyTruffle_Unicode_FromFormat(polyglot_from_string(format, "ascii"), args);
     va_end(args);
     return result;
 }

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

@@ -132,6 +132,7 @@
 import com.oracle.graal.python.builtins.objects.cext.capi.CExtNodes.ToJavaNode;
 import com.oracle.graal.python.builtins.objects.cext.capi.CExtNodes.ToNewRefNode;
 import com.oracle.graal.python.builtins.objects.cext.capi.CExtNodes.TransformExceptionToNativeNode;
+import com.oracle.graal.python.builtins.objects.cext.capi.CExtNodes.UnicodeFromFormatNode;
 import com.oracle.graal.python.builtins.objects.cext.capi.CExtNodes.VoidPtrToJavaNode;
 import com.oracle.graal.python.builtins.objects.cext.capi.CExtNodesFactory.CastToNativeLongNodeGen;
 import com.oracle.graal.python.builtins.objects.cext.capi.CExtNodesFactory.PRaiseNativeNodeGen;
@@ -3808,4 +3809,23 @@ static PBuiltinFunction asBuiltinFunction(Object methObj, AsPythonObjectNode asP
             return null;
         }
     }
+
+    // directly called without landing function
+    @Builtin(name = "PyTruffle_Unicode_FromFormat", minNumOfPositionalArgs = 2)
+    @GenerateNodeFactory
+    abstract static class PyTruffleUnicodeFromFromat extends PythonBuiltinNode {
+        @Specialization
+        static Object doGeneric(VirtualFrame frame, String format, Object vaList,
+                        @Cached UnicodeFromFormatNode unicodeFromFormatNode,
+                        @Cached CExtNodes.ToSulongNode toSulongNode,
+                        @Cached TransformExceptionToNativeNode transformExceptionToNativeNode,
+                        @Cached GetNativeNullNode getNativeNullNode) {
+            try {
+                return toSulongNode.execute(unicodeFromFormatNode.execute(format, vaList));
+            } catch (PException e) {
+                transformExceptionToNativeNode.execute(frame, e);
+                return getNativeNullNode.execute();
+            }
+        }
+    }
 }

graalpython/com.oracle.graal.python/src/com/oracle/graal/python/builtins/objects/cext/capi/CApiContext.java

@@ -732,6 +732,8 @@ public long getId() {
      * 'capi.c'.
      */
     public enum LLVMType {
+        int_t,
+        uint_t,
         int8_t,
         int16_t,
         int32_t,
@@ -740,12 +742,18 @@ public enum LLVMType {
         uint16_t,
         uint32_t,
         uint64_t,
+        long_t,
+        ulong_t,
+        longlong_t,
+        ulonglong_t,
         float_t,
         double_t,
+        size_t,
         Py_ssize_t,
         Py_complex,
         PyObject_ptr_t,
         char_ptr_t,
+        void_ptr_t,
         int8_ptr_t,
         int16_ptr_t,
         int32_ptr_t,