py: Start implementing "struct" module.

Only calcsize() and unpack() functions provided so far, for little-endian
byte order. Format strings don't support repition spec (like "2b3i").

Unfortunately, dealing with all the various binary type sizes and alignments
will lead to quite a bloated "binary" helper functions  - if optimizing for
speed. Need to think if using dynamic parametrized algos makes more sense.

py/binary.c

@@ -77,6 +77,52 @@ mp_obj_t mp_binary_get_val(char typecode, void *p, int index) {
     return MP_OBJ_NEW_SMALL_INT(val);
 }
 
+mp_obj_t mp_binary_get_val_unaligned_le(char typecode, byte **ptr) {
+    machine_int_t val = 0;
+    byte *p = *ptr;
+    switch (typecode) {
+        case 'b':
+            val = (int8_t)*p++;
+            break;
+        case BYTEARRAY_TYPECODE:
+        case 'B':
+            val = *p++;
+            break;
+        case 'h':
+            val = (int16_t)((p[1] << 8) | p[0]);
+            break;
+        case 'H':
+            val = (p[1] << 8) | p[0];
+            break;
+        case 'i':
+        case 'l':
+            val = (p[3] << 24) | (p[2] << 16) | (p[1] << 8) | p[0];
+            *ptr = p + 4;
+            return mp_obj_new_int(val);
+        case 'I':
+        case 'L':
+            val = (p[3] << 24) | (p[2] << 16) | (p[1] << 8) | p[0];
+            *ptr = p + 4;
+            return mp_obj_new_int_from_uint(val);
+#if 0 //TODO
+#if MICROPY_LONGINT_IMPL != MICROPY_LONGINT_IMPL_NONE
+        case 'q':
+        case 'Q':
+            // TODO: Explode API more to cover signedness
+            return mp_obj_new_int_from_ll(((long long*)p)[index]);
+#endif
+#if MICROPY_ENABLE_FLOAT
+        case 'f':
+            return mp_obj_new_float(((float*)p)[index]);
+        case 'd':
+            return mp_obj_new_float(((double*)p)[index]);
+#endif
+#endif
+    }
+    *ptr = p;
+    return MP_OBJ_NEW_SMALL_INT(val);
+}
+
 void mp_binary_set_val(char typecode, void *p, int index, mp_obj_t val_in) {
     machine_int_t val = 0;
     if (MP_OBJ_IS_INT(val_in)) {

py/binary.h

@@ -4,4 +4,5 @@
 
 int mp_binary_get_size(char typecode);
 mp_obj_t mp_binary_get_val(char typecode, void *p, int index);
+mp_obj_t mp_binary_get_val_unaligned_le(char typecode, byte **ptr);
 void mp_binary_set_val(char typecode, void *p, int index, mp_obj_t val_in);

py/builtin.h

@@ -41,3 +41,4 @@ extern const mp_obj_module_t mp_module_collections;
 extern const mp_obj_module_t mp_module_io;
 extern const mp_obj_module_t mp_module_math;
 extern const mp_obj_module_t mp_module_micropython;
+extern const mp_obj_module_t mp_module_struct;

py/builtintables.c

@@ -127,6 +127,9 @@ STATIC const mp_map_elem_t mp_builtin_module_table[] = {
     { MP_OBJ_NEW_QSTR(MP_QSTR_io), (mp_obj_t)&mp_module_io },
 #endif
     { MP_OBJ_NEW_QSTR(MP_QSTR_collections), (mp_obj_t)&mp_module_collections },
+#if MICROPY_ENABLE_MOD_STRUCT
+    { MP_OBJ_NEW_QSTR(MP_QSTR_struct), (mp_obj_t)&mp_module_struct },
+#endif
 
 #if MICROPY_ENABLE_FLOAT
     { MP_OBJ_NEW_QSTR(MP_QSTR_math), (mp_obj_t)&mp_module_math },

py/modstruct.c

@@ -0,0 +1,94 @@
+#include <assert.h>
+#include <string.h>
+#include "misc.h"
+#include "mpconfig.h"
+#include "qstr.h"
+#include "obj.h"
+#include "builtin.h"
+#include "objtuple.h"
+#include "binary.h"
+
+#if MICROPY_ENABLE_MOD_STRUCT
+
+STATIC char get_fmt_type(const char **fmt) {
+    char t = **fmt;
+    switch (t) {
+        case '!':
+            t = '>';
+            break;
+        case '@':
+        case '=':
+        case '<':
+        case '>':
+            break;
+        default:
+            return '@';
+    }
+    // Skip type char
+    (*fmt)++;
+    return t;
+}
+
+STATIC uint calcsize_items(const char *fmt) {
+    // TODO
+    return strlen(fmt);
+}
+
+STATIC mp_obj_t struct_calcsize(mp_obj_t fmt_in) {
+    const char *fmt = mp_obj_str_get_str(fmt_in);
+    char fmt_type = get_fmt_type(&fmt);
+    assert(fmt_type == '<'); (void)fmt_type;
+    uint size;
+    for (size = 0; *fmt; fmt++) {
+        int sz = mp_binary_get_size(*fmt);
+        // TODO
+        assert(sz != -1);
+        size += sz;
+    }
+    return MP_OBJ_NEW_SMALL_INT(size);
+}
+MP_DEFINE_CONST_FUN_OBJ_1(struct_calcsize_obj, struct_calcsize);
+
+STATIC mp_obj_t struct_unpack(mp_obj_t fmt_in, mp_obj_t data_in) {
+    // TODO: "The buffer must contain exactly the amount of data required by the format (len(bytes) must equal calcsize(fmt))."
+    const char *fmt = mp_obj_str_get_str(fmt_in);
+    char fmt_type = get_fmt_type(&fmt);
+    assert(fmt_type == '<'); (void)fmt_type;
+    uint size = calcsize_items(fmt);
+    mp_obj_tuple_t *res = mp_obj_new_tuple(size, NULL);
+    buffer_info_t bufinfo;
+    mp_get_buffer_raise(data_in, &bufinfo);
+    byte *p = bufinfo.buf;
+
+    for (uint i = 0; i < size; i++) {
+        mp_obj_t item = mp_binary_get_val_unaligned_le(*fmt++, &p);
+        res->items[i] = item;
+    }
+    return res;
+}
+MP_DEFINE_CONST_FUN_OBJ_2(struct_unpack_obj, struct_unpack);
+
+STATIC const mp_map_elem_t mp_module_struct_globals_table[] = {
+    { MP_OBJ_NEW_QSTR(MP_QSTR___name__), MP_OBJ_NEW_QSTR(MP_QSTR_struct) },
+    { MP_OBJ_NEW_QSTR(MP_QSTR_calcsize), (mp_obj_t)&struct_calcsize_obj },
+    { MP_OBJ_NEW_QSTR(MP_QSTR_unpack), (mp_obj_t)&struct_unpack_obj },
+};
+
+STATIC const mp_obj_dict_t mp_module_struct_globals = {
+    .base = {&mp_type_dict},
+    .map = {
+        .all_keys_are_qstrs = 1,
+        .table_is_fixed_array = 1,
+        .used = sizeof(mp_module_struct_globals_table) / sizeof(mp_map_elem_t),
+        .alloc = sizeof(mp_module_struct_globals_table) / sizeof(mp_map_elem_t),
+        .table = (mp_map_elem_t*)mp_module_struct_globals_table,
+    },
+};
+
+const mp_obj_module_t mp_module_struct = {
+    .base = { &mp_type_module },
+    .name = MP_QSTR_struct,
+    .globals = (mp_obj_dict_t*)&mp_module_struct_globals,
+};
+
+#endif

py/mpconfig.h

@@ -115,6 +115,11 @@ typedef double mp_float_t;
 #define MICROPY_ENABLE_MOD_IO (1)
 #endif
 
+// Whether to provide "struct" module
+#ifndef MICROPY_ENABLE_MOD_STRUCT
+#define MICROPY_ENABLE_MOD_STRUCT (1)
+#endif
+
 // Whether to support slice object and correspondingly
 // slice subscript operators
 #ifndef MICROPY_ENABLE_SLICE

py/py.mk

@@ -81,6 +81,7 @@ PY_O_BASENAME = \
 	modio.o \
 	modmath.o \
 	modmicropython.o \
+	modstruct.o \
 	vm.o \
 	showbc.o \
 	repl.o \

py/qstrdefs.h

@@ -84,6 +84,9 @@ Q(bool)
 Q(bytearray)
 Q(bytes)
 Q(callable)
+#if MICROPY_ENABLE_MOD_STRUCT
+Q(calcsize)
+#endif
 Q(chr)
 Q(classmethod)
 Q(collections)
@@ -127,10 +130,16 @@ Q(staticmethod)
 Q(sum)
 Q(super)
 Q(str)
+#if MICROPY_ENABLE_MOD_STRUCT
+Q(struct)
+#endif
 Q(sys)
 Q(to_bytes)
 Q(tuple)
 Q(type)
+#if MICROPY_ENABLE_MOD_STRUCT
+Q(unpack)
+#endif
 Q(value)
 Q(zip)
 

tests/basics/struct1.py

@@ -0,0 +1,3 @@
+import struct
+print(struct.calcsize("<bI"))
+print(struct.unpack("<bI", b"\x80\0\0\x01\0"))