Implementation with a preallocated list

The idea turned out to be very closely related to HPyTracker,
so I blatantly stole some code there.

hpy/devel/include/common/runtime/ctx_unicodebuilder.h

@@ -6,9 +6,9 @@
 #include "common/hpytype.h"
 
 
-_HPy_HIDDEN HPyUnicodeBuilder ctx_UnicodeBuilder_New(HPyContext *ctx);
-_HPy_HIDDEN void ctx_UnicodeBuilder_Append(HPyContext *ctx, HPyUnicodeBuilder builder,
-                                          HPy h_item);
+_HPy_HIDDEN HPyUnicodeBuilder ctx_UnicodeBuilder_New(HPyContext *ctx, HPy_ssize_t capacity);
+_HPy_HIDDEN int ctx_UnicodeBuilder_Add(HPyContext *ctx, HPyUnicodeBuilder builder,
+                                       HPy h_item);
 _HPy_HIDDEN HPy ctx_UnicodeBuilder_Build(HPyContext *ctx, HPyUnicodeBuilder builder);
 _HPy_HIDDEN void ctx_UnicodeBuilder_Cancel(HPyContext *ctx, HPyUnicodeBuilder builder);
 

hpy/devel/include/cpython/hpy.h

@@ -31,9 +31,9 @@
 
 typedef struct { PyObject *_o; } HPy;
 typedef struct { Py_ssize_t _lst; } HPyListBuilder;
-typedef struct { Py_ssize_t _lst; } HPyUnicodeBuilder;
 typedef struct { Py_ssize_t _tup; } HPyTupleBuilder;
 typedef struct { void *_o; } HPyTracker;
+typedef struct { void *_o; } HPyUnicodeBuilder;
 typedef Py_ssize_t HPy_ssize_t;
 typedef Py_hash_t HPy_hash_t;
 
@@ -380,31 +380,6 @@ HPyListBuilder_Cancel(HPyContext *ctx, HPyListBuilder builder)
     ctx_ListBuilder_Cancel(ctx, builder);
 }
 
-HPyAPI_FUNC(HPyUnicodeBuilder)
-HPyUnicodeBuilder_New(HPyContext *ctx)
-{
-    return ctx_UnicodeBuilder_New(ctx);
-}
-
-HPyAPI_FUNC(void)
-HPyUnicodeBuilder_Append(HPyContext *ctx, HPyUnicodeBuilder builder,
-                        HPy h_item)
-{
-    ctx_UnicodeBuilder_Append(ctx, builder, h_item);
-}
-
-HPyAPI_FUNC(HPy)
-HPyUnicodeBuilder_Build(HPyContext *ctx, HPyUnicodeBuilder builder)
-{
-    return ctx_UnicodeBuilder_Build(ctx, builder);
-}
-
-HPyAPI_FUNC(void)
-HPyUnicodeBuilder_Cancel(HPyContext *ctx, HPyUnicodeBuilder builder)
-{
-    ctx_UnicodeBuilder_Cancel(ctx, builder);
-}
-
 HPyAPI_FUNC(HPyTupleBuilder)
 HPyTupleBuilder_New(HPyContext *ctx, HPy_ssize_t initial_size)
 {
@@ -460,4 +435,28 @@ HPyTracker_Close(HPyContext *ctx, HPyTracker ht)
     ctx_Tracker_Close(ctx, ht);
 }
 
+HPyAPI_FUNC(HPyUnicodeBuilder)
+HPyUnicodeBuilder_New(HPyContext *ctx, HPy_ssize_t size)
+{
+    return ctx_UnicodeBuilder_New(ctx, size);
+}
+
+HPyAPI_FUNC(int)
+HPyUnicodeBuilder_Add(HPyContext *ctx, HPyUnicodeBuilder builder, HPy h_item)
+{
+    return ctx_UnicodeBuilder_Add(ctx, builder, h_item);
+}
+
+HPyAPI_FUNC(HPy)
+HPyUnicodeBuilder_Build(HPyContext *ctx, HPyUnicodeBuilder builder)
+{
+    return ctx_UnicodeBuilder_Build(ctx, builder);
+}
+
+HPyAPI_FUNC(void)
+HPyUnicodeBuilder_Cancel(HPyContext *ctx, HPyUnicodeBuilder builder)
+{
+    ctx_UnicodeBuilder_Cancel(ctx, builder);
+}
+
 #endif /* !HPy_CPYTHON_H */

hpy/devel/include/universal/hpy.h

@@ -34,9 +34,9 @@ typedef intptr_t HPy_hash_t;
    the HPy structure contains an index in a global array. */
 typedef struct _HPy_s { HPy_ssize_t _i; } HPy;
 typedef struct { HPy_ssize_t _lst; } HPyListBuilder;
-typedef struct { HPy_ssize_t _lst; } HPyUnicodeBuilder;
 typedef struct { HPy_ssize_t _tup; } HPyTupleBuilder;
 typedef struct { HPy_ssize_t _i; } HPyTracker;
+typedef struct { HPy_ssize_t _i; } HPyUnicodeBuilder;
 
 typedef struct _HPyContext_s HPyContext;
 

hpy/devel/src/runtime/ctx_unicodebuilder.c

@@ -7,57 +7,101 @@
 #  include "handles.h"
 #endif
 
-#include <stdio.h>
+static const Py_ssize_t HPYUNICODEBUILDER_INITIAL_CAPACITY = 5;
+
+typedef struct {
+    Py_ssize_t capacity;  // allocated handles
+    Py_ssize_t length;    // used handles
+    PyObject *list;
+} _PyUnicodeBuilder_s;
+
+#ifdef HPY_UNIVERSAL_ABI
+static inline _PyUnicodeBuilder_s *_hb2pb(HPyUnicodeBuilder ht) {
+    return (_PyUnicodeBuilder_s *) (ht)._i;
+}
+static inline HPyUnicodeBuilder _pb2hb(_PyUnicodeBuilder_s *bp) {
+    return (HPyUnicodeBuilder) {(HPy_ssize_t) (bp)};
+}
+#else
+static inline _PyUnicodeBuilder_s *_hb2pb(HPyUnicodeBuilder ht) {
+    return (_PyUnicodeBuilder_s *) (ht)._o;
+}
+static inline HPyUnicodeBuilder _pb2hb(_PyUnicodeBuilder_s *bp) {
+    return (HPyUnicodeBuilder) {(void *) (bp)};
+}
+#endif
 
 _HPy_HIDDEN HPyUnicodeBuilder
-ctx_UnicodeBuilder_New(HPyContext *ctx)
+ctx_UnicodeBuilder_New(HPyContext *ctx, HPy_ssize_t capacity)
 {
-    PyObject *lst = PyList_New(0);
-    if (lst == NULL)
-        PyErr_Clear();   /* delay the MemoryError */
-    return (HPyUnicodeBuilder){(HPy_ssize_t)lst };
+    _PyUnicodeBuilder_s *bp;
+    if (capacity == 0) {
+        capacity = HPYUNICODEBUILDER_INITIAL_CAPACITY;
+    }
+    capacity++; // always reserve space for an extra handle, see the docs, analogue to HPyTracker
+
+    bp = malloc(sizeof(_PyUnicodeBuilder_s));
+    if (bp == NULL) {
+        HPyErr_NoMemory(ctx);
+        return _pb2hb(0);
+    }
+
+    bp->list = PyList_New(capacity);
+    if (bp->list == NULL) {
+        free(bp);
+        HPyErr_NoMemory(ctx);
+        return _pb2hb(0);
+    }
+    bp->capacity = capacity;
+    bp->length = 0;
+    return _pb2hb(bp);
 }
 
-_HPy_HIDDEN void
-ctx_UnicodeBuilder_Append(HPyContext *ctx, HPyUnicodeBuilder builder, HPy h_item)
+_HPy_HIDDEN int
+ctx_UnicodeBuilder_Add(HPyContext *ctx, HPyUnicodeBuilder builder, HPy h_item)
 {
-    PyObject *lst = (PyObject *)builder._lst;
-    if (lst != NULL) {
-        PyObject *item = _h2py(h_item);
-        PyList_Append(lst, item);
+    if(!HPyUnicode_Check(ctx, h_item)) {
+        HPyErr_SetString(ctx, ctx->h_TypeError, "Argument must be of type HPyUnicode");
+        return -1;
+    }
+
+    _PyUnicodeBuilder_s *bp = _hb2pb(builder);
+    PyObject *item = _h2py(h_item);
+
+    // XXX: For the initial PoC we don't care about reallocation
+    if (bp->capacity <= bp->length) {
+        return -1;
     }
+    Py_INCREF(item);
+    PyList_SET_ITEM(bp->list, bp->length++, item);
+    return 0;
 }
 
 _HPy_HIDDEN HPy
 ctx_UnicodeBuilder_Build(HPyContext *ctx, HPyUnicodeBuilder builder)
 {
-    PyObject *lst = (PyObject *)builder._lst;
-    if (lst == NULL) {
-        PyErr_NoMemory();
-        return HPy_NULL;
-    }
-    builder._lst = 0;
+    _PyUnicodeBuilder_s *bp = _hb2pb(builder);
+    PyObject *list = PyList_GetSlice(bp->list, 0, bp->length);
+
     PyObject *sep = PyUnicode_FromString("");
-    PyObject *str = PyUnicode_Join(sep, lst);
+    PyObject *str = PyUnicode_Join(sep, list);
     Py_XDECREF(sep);
+
     if(str == NULL) {
         PyErr_NoMemory();
         return HPy_NULL;
     }
-    Py_XDECREF(lst);
+
+    Py_XDECREF(bp->list);
+    Py_XDECREF(list);
+    free(bp);
     return _py2h(str);
 }
 
 _HPy_HIDDEN void
 ctx_UnicodeBuilder_Cancel(HPyContext *ctx, HPyUnicodeBuilder builder)
 {
-    PyObject *lst = (PyObject *)builder._lst;
-    if (lst == NULL) {
-        // we don't report the memory error here: the builder
-        // is being cancelled (so the result of the builder is not being used)
-        // and likely it's being cancelled during the handling of another error
-        return;
-    }
-    builder._lst = 0;
-    Py_XDECREF(lst);
+    _PyUnicodeBuilder_s *bp = _hb2pb(builder);
+    Py_XDECREF(bp->list);
+    free(bp);
 }

hpy/tools/autogen/parse.py

@@ -210,10 +210,6 @@ def _visit_hpyslot_slot(self, node):
     'HPyListBuilder_Set': None,
     'HPyListBuilder_Build': None,
     'HPyListBuilder_Cancel': None,
-    'HPyUnicodeBuilder_New': None,
-    'HPyUnicodeBuilder_Append': None,
-    'HPyUnicodeBuilder_Build': None,
-    'HPyUnicodeBuilder_Cancel': None,
     'HPyTuple_FromArray': None,
     'HPyTupleBuilder_New': None,
     'HPyTupleBuilder_Set': None,
@@ -223,6 +219,10 @@ def _visit_hpyslot_slot(self, node):
     'HPyTracker_Add': None,
     'HPyTracker_ForgetAll': None,
     'HPyTracker_Close': None,
+    'HPyUnicodeBuilder_New': None,
+    'HPyUnicodeBuilder_Add': None,
+    'HPyUnicodeBuilder_Build': None,
+    'HPyUnicodeBuilder_Cancel': None,
     '_HPy_Dump': None,
     'HPy_Type': 'PyObject_Type',
     'HPy_TypeCheck': None,

hpy/tools/autogen/public_api.h

@@ -283,9 +283,8 @@ void HPyListBuilder_Set(HPyContext *ctx, HPyListBuilder builder,
 HPy HPyListBuilder_Build(HPyContext *ctx, HPyListBuilder builder);
 void HPyListBuilder_Cancel(HPyContext *ctx, HPyListBuilder builder);
 
-HPyUnicodeBuilder HPyUnicodeBuilder_New(HPyContext *ctx);
-void HPyUnicodeBuilder_Append(HPyContext *ctx, HPyUnicodeBuilder builder,
-                             HPy h_item);
+HPyUnicodeBuilder HPyUnicodeBuilder_New(HPyContext *ctx, HPy_ssize_t size);
+int HPyUnicodeBuilder_Add(HPyContext *ctx, HPyUnicodeBuilder builder, HPy h_item);
 HPy HPyUnicodeBuilder_Build(HPyContext *ctx, HPyUnicodeBuilder builder);
 void HPyUnicodeBuilder_Cancel(HPyContext *ctx, HPyUnicodeBuilder builder);