Close #17694: Add minimum length to _PyUnicodeWriter

 * Add also min_char attribute to _PyUnicodeWriter structure (currently unused)
 * _PyUnicodeWriter_Init() has no more argument (except the writer itself):
   min_length and overallocate must be set explicitly
 * In error handlers, only enable overallocation if the replacement string
   is longer than 1 character
 * CJK decoders don't use overallocation anymore
 * Set min_length, instead of preallocating memory using
   _PyUnicodeWriter_Prepare(), in many decoders
 * _PyUnicode_DecodeUnicodeInternal() checks for integer overflow

Include/unicodeobject.h

@@ -898,22 +898,28 @@ typedef struct {
     Py_UCS4 maxchar;
     Py_ssize_t size;
     Py_ssize_t pos;
-    /* minimum length of the buffer when overallocation is enabled,
-       see _PyUnicodeWriter_Init() */
+
+    /* minimum number of allocated characters (default: 0) */
     Py_ssize_t min_length;
+
+    /* minimum character (default: 127, ASCII) */
+    Py_UCS4 min_char;
+
+    /* If non-zero, overallocate the buffer by 25% (default: 0). */
     unsigned char overallocate;
+
     /* If readonly is 1, buffer is a shared string (cannot be modified)
        and size is set to 0. */
     unsigned char readonly;
 } _PyUnicodeWriter ;
 
 /* Initialize a Unicode writer.
-
-   If min_length is greater than zero, _PyUnicodeWriter_Prepare()
-   overallocates the buffer and min_length is the minimum length in characters
-   of the buffer. */
+ *
+ * By default, the minimum buffer size is 0 character and overallocation is
+ * disabled. Set min_length, min_char and overallocate attributes to control
+ * the allocation of the buffer. */
 PyAPI_FUNC(void)
-_PyUnicodeWriter_Init(_PyUnicodeWriter *writer, Py_ssize_t min_length);
+_PyUnicodeWriter_Init(_PyUnicodeWriter *writer);
 
 /* Prepare the buffer to write 'length' characters
    with the specified maximum character.

Modules/cjkcodecs/multibytecodec.c

@@ -633,7 +633,8 @@ MultibyteCodec_Decode(MultibyteCodecObject *self,
         return make_tuple(PyUnicode_New(0, 0), 0);
     }
 
-    _PyUnicodeWriter_Init(&buf.writer, datalen);
+    _PyUnicodeWriter_Init(&buf.writer);
+    buf.writer.min_length = datalen;
     buf.excobj = NULL;
     buf.inbuf = buf.inbuf_top = (unsigned char *)data;
     buf.inbuf_end = buf.inbuf_top + datalen;
@@ -839,7 +840,7 @@ decoder_prepare_buffer(MultibyteDecodeBuffer *buf, const char *data,
 {
     buf->inbuf = buf->inbuf_top = (const unsigned char *)data;
     buf->inbuf_end = buf->inbuf_top + size;
-    _PyUnicodeWriter_Init(&buf->writer, size);
+    buf->writer.min_length += size;
     return 0;
 }
 
@@ -1037,7 +1038,7 @@ mbidecoder_decode(MultibyteIncrementalDecoderObject *self,
     data = pdata.buf;
     size = pdata.len;
 
-    _PyUnicodeWriter_Init(&buf.writer, 1);
+    _PyUnicodeWriter_Init(&buf.writer);
     buf.excobj = NULL;
     origpending = self->pendingsize;
 
@@ -1241,7 +1242,7 @@ mbstreamreader_iread(MultibyteStreamReaderObject *self,
     if (sizehint == 0)
         return PyUnicode_New(0, 0);
 
-    _PyUnicodeWriter_Init(&buf.writer, 1);
+    _PyUnicodeWriter_Init(&buf.writer);
     buf.excobj = NULL;
     cres = NULL;
 

Objects/complexobject.c

@@ -705,7 +705,7 @@ complex__format__(PyObject* self, PyObject* args)
     if (!PyArg_ParseTuple(args, "U:__format__", &format_spec))
         return NULL;
 
-    _PyUnicodeWriter_Init(&writer, 0);
+    _PyUnicodeWriter_Init(&writer);
     ret = _PyComplex_FormatAdvancedWriter(
         &writer,
         self,

Objects/floatobject.c

@@ -1711,7 +1711,7 @@ float__format__(PyObject *self, PyObject *args)
     if (!PyArg_ParseTuple(args, "U:__format__", &format_spec))
         return NULL;
 
-    _PyUnicodeWriter_Init(&writer, 0);
+    _PyUnicodeWriter_Init(&writer);
     ret = _PyFloat_FormatAdvancedWriter(
         &writer,
         self,

Objects/longobject.c

@@ -4379,7 +4379,7 @@ long__format__(PyObject *self, PyObject *args)
     if (!PyArg_ParseTuple(args, "U:__format__", &format_spec))
         return NULL;
 
-    _PyUnicodeWriter_Init(&writer, 0);
+    _PyUnicodeWriter_Init(&writer);
     ret = _PyLong_FormatAdvancedWriter(
         &writer,
         self,

Objects/stringlib/unicode_format.h

@@ -906,7 +906,6 @@ build_string(SubString *input, PyObject *args, PyObject *kwargs,
              int recursion_depth, AutoNumber *auto_number)
 {
     _PyUnicodeWriter writer;
-    Py_ssize_t minlen;
 
     /* check the recursion level */
     if (recursion_depth <= 0) {
@@ -915,8 +914,9 @@ build_string(SubString *input, PyObject *args, PyObject *kwargs,
         return NULL;
     }
 
-    minlen = PyUnicode_GET_LENGTH(input->str) + 100;
-    _PyUnicodeWriter_Init(&writer, minlen);
+    _PyUnicodeWriter_Init(&writer);
+    writer.overallocate = 1;
+    writer.min_length = PyUnicode_GET_LENGTH(input->str) + 100;
 
     if (!do_markup(input, args, kwargs, &writer, recursion_depth,
                    auto_number)) {