py: split runtime into map, obj, builtin.

py/builtin.c

+#include <stdint.h>
+#include <stdlib.h>
+#include <stdio.h>
+#include <stdarg.h>
+#include <string.h>
+#include <assert.h>
+
+#include "nlr.h"
+#include "misc.h"
+#include "mpyconfig.h"
+#include "runtime.h"
+#include "bc.h"
+
+#include "map.h"
+#include "obj.h"
+#include "objprivate.h"
+#include "builtin.h"
+
+py_obj_t py_builtin___repl_print__(py_obj_t o) {
+    if (o != py_const_none) {
+        py_obj_print(o);
+        printf("\n");
+    }
+    return py_const_none;
+}
+
+py_obj_t py_builtin_print(int n_args, const py_obj_t* args) {
+    for (int i = 0; i < n_args; i++) {
+        if (i > 0) {
+            printf(" ");
+        }
+        if (IS_O(args[i], O_STR)) {
+            // special case, print string raw
+            printf("%s", qstr_str(((py_obj_base_t*)args[i])->u_str));
+        } else {
+            // print the object Python style
+            py_obj_print(args[i]);
+        }
+    }
+    printf("\n");
+    return py_const_none;
+}
+
+py_obj_t py_builtin_len(py_obj_t o_in) {
+    py_small_int_t len = 0;
+    if (IS_O(o_in, O_STR)) {
+        py_obj_base_t *o = o_in;
+        len = strlen(qstr_str(o->u_str));
+    } else if (IS_O(o_in, O_TUPLE) || IS_O(o_in, O_LIST)) {
+        py_obj_base_t *o = o_in;
+        len = o->u_tuple_list.len;
+    } else if (IS_O(o_in, O_MAP)) {
+        py_obj_base_t *o = o_in;
+        len = o->u_map.used;
+    } else {
+        assert(0);
+    }
+    return TO_SMALL_INT(len);
+}
+
+py_obj_t py_builtin_abs(py_obj_t o_in) {
+    if (IS_SMALL_INT(o_in)) {
+        py_small_int_t val = FROM_SMALL_INT(o_in);
+        if (val < 0) {
+            val = -val;
+        }
+        return TO_SMALL_INT(val);
+#if MICROPY_ENABLE_FLOAT
+    } else if (IS_O(o_in, O_FLOAT)) {
+        py_obj_base_t *o = o_in;
+        // TODO check for NaN etc
+        if (o->u_float < 0) {
+            return py_obj_new_float(-o->u_float);
+        } else {
+            return o_in;
+        }
+    } else if (IS_O(o_in, O_COMPLEX)) {
+        py_obj_base_t *o = o_in;
+        return py_obj_new_float(machine_sqrt(o->u_complex.real*o->u_complex.real + o->u_complex.imag*o->u_complex.imag));
+#endif
+    } else {
+        assert(0);
+        return py_const_none;
+    }
+}
+
+py_obj_t py_builtin___build_class__(py_obj_t o_class_fun, py_obj_t o_class_name) {
+    // we differ from CPython: we set the new __locals__ object here
+    py_map_t *old_locals = rt_get_map_locals();
+    py_map_t *class_locals = py_map_new(MAP_QSTR, 0);
+    rt_set_map_locals(class_locals);
+
+    // call the class code
+    rt_call_function_1(o_class_fun, (py_obj_t)0xdeadbeef);
+
+    // restore old __locals__ object
+    rt_set_map_locals(old_locals);
+
+    // create and return the new class
+    py_obj_base_t *o = m_new(py_obj_base_t, 1);
+    o->kind = O_CLASS;
+    o->u_class.locals = class_locals;
+    return o;
+}
+
+py_obj_t py_builtin_range(int n_args, const py_obj_t* args) {
+    switch (n_args) {
+        case 1: return py_obj_new_range(0, py_obj_get_int(args[0]), 1);
+        case 2: return py_obj_new_range(py_obj_get_int(args[0]), py_obj_get_int(args[1]), 1);
+        case 3: return py_obj_new_range(py_obj_get_int(args[0]), py_obj_get_int(args[1]), py_obj_get_int(args[2]));
+        default: nlr_jump(py_obj_new_exception_2(rt_q_TypeError, "range expected at most 3 arguments, got %d", (void*)(machine_int_t)n_args, NULL));
+    }
+}

py/builtin.h

+py_obj_t py_builtin___repl_print__(py_obj_t o);
+py_obj_t py_builtin_print(int n_args, const py_obj_t* args);
+py_obj_t py_builtin_len(py_obj_t o_in);
+py_obj_t py_builtin_abs(py_obj_t o_in);
+py_obj_t py_builtin___build_class__(py_obj_t o_class_fun, py_obj_t o_class_name);
+py_obj_t py_builtin_range(int n_args, const py_obj_t* args);

py/map.c

+#include <stdint.h>
+#include <stdlib.h>
+#include <assert.h>
+
+#include "misc.h"
+#include "mpyconfig.h"
+#include "runtime.h"
+
+#include "map.h"
+#include "obj.h"
+#include "objprivate.h"
+
+// approximatelly doubling primes; made with Mathematica command: Table[Prime[Floor[(1.7)^n]], {n, 3, 24}]
+static int doubling_primes[] = {7, 19, 43, 89, 179, 347, 647, 1229, 2297, 4243, 7829, 14347, 26017, 47149, 84947, 152443, 273253, 488399, 869927, 1547173, 2745121, 4861607};
+
+int get_doubling_prime_greater_or_equal_to(int x) {
+    for (int i = 0; i < sizeof(doubling_primes) / sizeof(int); i++) {
+        if (doubling_primes[i] >= x) {
+            return doubling_primes[i];
+        }
+    }
+    // ran out of primes in the table!
+    // return something sensible, at least make it odd
+    return x | 1;
+}
+
+void py_map_init(py_map_t *map, py_map_kind_t kind, int n) {
+    map->kind = kind;
+    map->used = 0;
+    map->alloc = get_doubling_prime_greater_or_equal_to(n + 1);
+    map->table = m_new0(py_map_elem_t, map->alloc);
+}
+
+py_map_t *py_map_new(py_map_kind_t kind, int n) {
+    py_map_t *map = m_new(py_map_t, 1);
+    py_map_init(map, kind, n);
+    return map;
+}
+
+py_map_elem_t* py_map_lookup_helper(py_map_t *map, py_obj_t index, bool add_if_not_found) {
+    bool is_map_py_obj = (map->kind == MAP_PY_OBJ);
+    machine_uint_t hash;
+    if (is_map_py_obj) {
+        hash = py_obj_hash(index);
+    } else {
+        hash = (machine_uint_t)index;
+    }
+    uint pos = hash % map->alloc;
+    for (;;) {
+        py_map_elem_t *elem = &map->table[pos];
+        if (elem->key == NULL) {
+            // not in table
+            if (add_if_not_found) {
+                if (map->used + 1 >= map->alloc) {
+                    // not enough room in table, rehash it
+                    int old_alloc = map->alloc;
+                    py_map_elem_t *old_table = map->table;
+                    map->alloc = get_doubling_prime_greater_or_equal_to(map->alloc + 1);
+                    map->used = 0;
+                    map->table = m_new0(py_map_elem_t, map->alloc);
+                    for (int i = 0; i < old_alloc; i++) {
+                        if (old_table[i].key != NULL) {
+                            py_map_lookup_helper(map, old_table[i].key, true)->value = old_table[i].value;
+                        }
+                    }
+                    m_free(old_table);
+                    // restart the search for the new element
+                    pos = hash % map->alloc;
+                } else {
+                    map->used += 1;
+                    elem->key = index;
+                    return elem;
+                }
+            } else {
+                return NULL;
+            }
+        } else if (elem->key == index || (is_map_py_obj && py_obj_equal(elem->key, index))) {
+            // found it
+            /* it seems CPython does not replace the index; try x={True:'true'};x[1]='one';x
+            if (add_if_not_found) {
+                elem->key = index;
+            }
+            */
+            return elem;
+        } else {
+            // not yet found, keep searching in this table
+            pos = (pos + 1) % map->alloc;
+        }
+    }
+}
+
+py_map_elem_t* py_qstr_map_lookup(py_map_t *map, qstr index, bool add_if_not_found) {
+    py_obj_t o = (py_obj_t)(machine_uint_t)index;
+    return py_map_lookup_helper(map, o, add_if_not_found);
+}
+
+py_map_elem_t* py_map_lookup(py_obj_t o, py_obj_t index, bool add_if_not_found) {
+    assert(IS_O(o, O_MAP));
+    return py_map_lookup_helper(&((py_obj_base_t *)o)->u_map, index, add_if_not_found);
+}
+
+py_obj_t py_set_lookup(py_obj_t o_in, py_obj_t index, bool add_if_not_found) {
+    assert(IS_O(o_in, O_SET));
+    py_obj_base_t *o = o_in;
+    int hash = py_obj_hash(index);
+    int pos = hash % o->u_set.alloc;
+    for (;;) {
+        py_obj_t elem = o->u_set.table[pos];
+        if (elem == NULL) {
+            // not in table
+            if (add_if_not_found) {
+                if (o->u_set.used + 1 >= o->u_set.alloc) {
+                    // not enough room in table, rehash it
+                    int old_alloc = o->u_set.alloc;
+                    py_obj_t *old_table = o->u_set.table;
+                    o->u_set.alloc = get_doubling_prime_greater_or_equal_to(o->u_set.alloc + 1);
+                    o->u_set.used = 0;
+                    o->u_set.table = m_new(py_obj_t, o->u_set.alloc);
+                    for (int i = 0; i < old_alloc; i++) {
+                        if (old_table[i] != NULL) {
+                            py_set_lookup(o, old_table[i], true);
+                        }
+                    }
+                    m_free(old_table);
+                    // restart the search for the new element
+                    pos = hash % o->u_set.alloc;
+                } else {
+                    o->u_set.used += 1;
+                    o->u_set.table[pos] = index;
+                    return index;
+                }
+            } else {
+                return NULL;
+            }
+        } else if (py_obj_equal(elem, index)) {
+            // found it
+            return elem;
+        } else {
+            // not yet found, keep searching in this table
+            pos = (pos + 1) % o->u_set.alloc;
+        }
+    }
+}

py/map.h

+typedef enum {
+    MAP_QSTR,
+    MAP_PY_OBJ,
+} py_map_kind_t;
+
+typedef struct _py_map_elem_t {
+    py_obj_t key;
+    py_obj_t value;
+} py_map_elem_t;
+
+typedef struct _py_map_t {
+    struct {
+        py_map_kind_t kind : 1;
+        machine_uint_t used : (8 * BYTES_PER_WORD - 1);
+    };
+    machine_uint_t alloc;
+    py_map_elem_t *table;
+} py_map_t;
+
+// these are defined in runtime.c
+py_map_t *rt_get_map_locals(void);
+void rt_set_map_locals(py_map_t *m);
+
+int get_doubling_prime_greater_or_equal_to(int x);
+void py_map_init(py_map_t *map, py_map_kind_t kind, int n);
+py_map_t *py_map_new(py_map_kind_t kind, int n);
+py_map_elem_t* py_map_lookup_helper(py_map_t *map, py_obj_t index, bool add_if_not_found);
+py_map_elem_t* py_qstr_map_lookup(py_map_t *map, qstr index, bool add_if_not_found);
+py_map_elem_t* py_map_lookup(py_obj_t o, py_obj_t index, bool add_if_not_found);
+py_obj_t py_set_lookup(py_obj_t o_in, py_obj_t index, bool add_if_not_found);

py/obj.c

+#include <stdint.h>
+#include <stdlib.h>
+#include <stdio.h>
+#include <stdarg.h>
+#include <assert.h>
+
+#include "nlr.h"
+#include "misc.h"
+#include "mpyconfig.h"
+#include "runtime.h"
+
+#include "map.h"
+#include "obj.h"
+#include "objprivate.h"
+
+py_obj_t py_obj_new_int(machine_int_t value) {
+    return TO_SMALL_INT(value);
+}
+
+py_obj_t py_obj_new_const(const char *id) {
+    py_obj_base_t *o = m_new(py_obj_base_t, 1);
+    o->kind = O_CONST;
+    o->id = id;
+    return (py_obj_t)o;
+}
+
+py_obj_t py_obj_new_str(qstr qstr) {
+    py_obj_base_t *o = m_new(py_obj_base_t, 1);
+    o->kind = O_STR;
+    o->u_str = qstr;
+    return (py_obj_t)o;
+}
+
+#if MICROPY_ENABLE_FLOAT
+py_obj_t py_obj_new_float(py_float_t val) {
+    py_obj_base_t *o = m_new(py_obj_base_t, 1);
+    o->kind = O_FLOAT;
+    o->u_float = val;
+    return (py_obj_t)o;
+}
+
+py_obj_t py_obj_new_complex(py_float_t real, py_float_t imag) {
+    py_obj_base_t *o = m_new(py_obj_base_t, 1);
+    o->kind = O_COMPLEX;
+    o->u_complex.real = real;
+    o->u_complex.imag = imag;
+    return (py_obj_t)o;
+}
+#endif
+
+py_obj_t py_obj_new_exception_0(qstr id) {
+    py_obj_base_t *o = m_new(py_obj_base_t, 1);
+    o->kind = O_EXCEPTION_0;
+    o->u_exc0.id = id;
+    return (py_obj_t)o;
+}
+
+py_obj_t py_obj_new_exception_2(qstr id, const char *fmt, const char *s1, const char *s2) {
+    py_obj_base_t *o = m_new(py_obj_base_t, 1);
+    o->kind = O_EXCEPTION_N;
+    o->u_exc_n.id = id;
+    o->u_exc_n.n_args = 3;
+    o->u_exc_n.args = m_new(const void*, 3);
+    o->u_exc_n.args[0] = fmt;
+    o->u_exc_n.args[1] = s1;
+    o->u_exc_n.args[2] = s2;
+    return (py_obj_t)o;
+}
+
+// range is a class and instances are immutable sequence objects
+py_obj_t py_obj_new_range(int start, int stop, int step) {
+    py_obj_base_t *o = m_new(py_obj_base_t, 1);
+    o->kind = O_RANGE;
+    o->u_range.start = start;
+    o->u_range.stop = stop;
+    o->u_range.step = step;
+    return o;
+}
+
+py_obj_t py_obj_new_range_iterator(int cur, int stop, int step) {
+    py_obj_base_t *o = m_new(py_obj_base_t, 1);
+    o->kind = O_RANGE_IT;
+    o->u_range_it.cur = cur;
+    o->u_range_it.stop = stop;
+    o->u_range_it.step = step;
+    return o;
+}
+
+py_obj_t py_obj_new_tuple_iterator(py_obj_base_t *tuple, int cur) {
+    py_obj_base_t *o = m_new(py_obj_base_t, 1);
+    o->kind = O_TUPLE_IT;
+    o->u_tuple_list_it.obj = tuple;
+    o->u_tuple_list_it.cur = cur;
+    return o;
+}
+
+py_obj_t py_obj_new_list_iterator(py_obj_base_t *list, int cur) {
+    py_obj_base_t *o = m_new(py_obj_base_t, 1);
+    o->kind = O_LIST_IT;
+    o->u_tuple_list_it.obj = list;
+    o->u_tuple_list_it.cur = cur;
+    return o;
+}
+
+py_obj_t py_obj_new_user(const py_user_info_t *info, machine_uint_t data1, machine_uint_t data2) {
+    py_obj_base_t *o = m_new(py_obj_base_t, 1);
+    o->kind = O_USER;
+    // TODO should probably parse the info to turn strings to qstr's, and wrap functions in O_FUN_N objects
+    // that'll take up some memory.  maybe we can lazily do the O_FUN_N: leave it a ptr to a C function, and
+    // only when the method is looked-up do we change that to the O_FUN_N object.
+    o->u_user.info = info;
+    o->u_user.data1 = data1;
+    o->u_user.data2 = data2;
+    return o;
+}
+
+const char *py_obj_get_type_str(py_obj_t o_in) {
+    if (IS_SMALL_INT(o_in)) {
+        return "int";
+    } else {
+        py_obj_base_t *o = o_in;
+        switch (o->kind) {
+            case O_CONST:
+                if (o == py_const_none) {
+                    return "NoneType";
+                } else {
+                    return "bool";
+                }
+            case O_STR:
+                return "str";
+#if MICROPY_ENABLE_FLOAT
+            case O_FLOAT:
+                return "float";
+#endif
+            case O_FUN_0:
+            case O_FUN_1:
+            case O_FUN_2:
+            case O_FUN_N:
+            case O_FUN_VAR:
+            case O_FUN_BC:
+                return "function";
+            case O_GEN_INSTANCE:
+                return "generator";
+            case O_TUPLE:
+                return "tuple";
+            case O_LIST:
+                return "list";
+            case O_TUPLE_IT:
+                return "tuple_iterator";
+            case O_LIST_IT:
+                return "list_iterator";
+            case O_SET:
+                return "set";
+            case O_MAP:
+                return "dict";
+            case O_OBJ:
+            {
+                py_map_elem_t *qn = py_qstr_map_lookup(o->u_obj.class->u_class.locals, qstr_from_str_static("__qualname__"), false);
+                assert(qn != NULL);
+                assert(IS_O(qn->value, O_STR));
+                return qstr_str(((py_obj_base_t*)qn->value)->u_str);
+            }
+            case O_USER:
+                return o->u_user.info->type_name;
+            default:
+                assert(0);
+                return "UnknownType";
+        }
+    }
+}
+
+void printf_wrapper(void *env, const char *fmt, ...) {
+    va_list args;
+    va_start(args, fmt);
+    vprintf(fmt, args);
+    va_end(args);
+}
+
+void py_obj_print_helper(void (*print)(void *env, const char *fmt, ...), void *env, py_obj_t o_in) {
+    if (IS_SMALL_INT(o_in)) {
+        print(env, "%d", (int)FROM_SMALL_INT(o_in));
+    } else {
+        py_obj_base_t *o = o_in;
+        switch (o->kind) {
+            case O_CONST:
+                print(env, "%s", o->id);
+                break;
+            case O_STR:
+                // TODO need to escape chars etc
+                print(env, "'%s'", qstr_str(o->u_str));
+                break;
+#if MICROPY_ENABLE_FLOAT
+            case O_FLOAT:
+                print(env, "%.8g", o->u_float);
+                break;
+            case O_COMPLEX:
+                if (o->u_complex.real == 0) {
+                    print(env, "%.8gj", o->u_complex.imag);
+                } else {
+                    print(env, "(%.8g+%.8gj)", o->u_complex.real, o->u_complex.imag);
+                }
+                break;
+#endif
+            case O_EXCEPTION_0:
+                print(env, "%s", qstr_str(o->u_exc0.id));
+                break;
+            case O_EXCEPTION_N:
+                print(env, "%s: ", qstr_str(o->u_exc_n.id));
+                print(env, o->u_exc_n.args[0], o->u_exc_n.args[1], o->u_exc_n.args[2]);
+                break;
+            case O_GEN_INSTANCE:
+                print(env, "<generator object 'fun-name' at %p>", o);
+                break;
+            case O_TUPLE:
+                print(env, "(");
+                for (int i = 0; i < o->u_tuple_list.len; i++) {
+                    if (i > 0) {
+                        print(env, ", ");
+                    }
+                    py_obj_print_helper(print, env, o->u_tuple_list.items[i]);
+                }
+                if (o->u_tuple_list.len == 1) {
+                    print(env, ",");
+                }
+                print(env, ")");
+                break;
+            case O_LIST:
+                print(env, "[");
+                for (int i = 0; i < o->u_tuple_list.len; i++) {
+                    if (i > 0) {
+                        print(env, ", ");
+                    }
+                    py_obj_print_helper(print, env, o->u_tuple_list.items[i]);
+                }
+                print(env, "]");
+                break;
+            case O_SET:
+            {
+                bool first = true;
+                print(env, "{");
+                for (int i = 0; i < o->u_set.alloc; i++) {
+                    if (o->u_set.table[i] != NULL) {
+                        if (!first) {
+                            print(env, ", ");
+                        }
+                        first = false;
+                        py_obj_print_helper(print, env, o->u_set.table[i]);
+                    }
+                }
+                print(env, "}");
+                break;
+            }
+            case O_MAP:
+            {
+                bool first = true;
+                print(env, "{");
+                for (int i = 0; i < o->u_map.alloc; i++) {
+                    if (o->u_map.table[i].key != NULL) {
+                        if (!first) {
+                            print(env, ", ");
+                        }
+                        first = false;
+                        py_obj_print_helper(print, env, o->u_map.table[i].key);
+                        print(env, ": ");
+                        py_obj_print_helper(print, env, o->u_map.table[i].value);
+                    }
+                }
+                print(env, "}");
+                break;
+            }
+            case O_USER:
+                if (o->u_user.info->print == NULL) {
+                    print(env, "<unknown user object>");
+                } else {
+                    o->u_user.info->print(o_in);
+                }
+                break;
+            default:
+                print(env, "<? %d>", o->kind);
+                assert(0);
+        }
+    }
+}
+
+void py_obj_print(py_obj_t o_in) {
+    py_obj_print_helper(printf_wrapper, NULL, o_in);
+}
+
+bool py_obj_is_callable(py_obj_t o_in) {
+    if (IS_SMALL_INT(o_in)) {
+        return false;
+    } else {
+        py_obj_base_t *o = o_in;
+        switch (o->kind) {
+            case O_FUN_0:
+            case O_FUN_1:
+            case O_FUN_2:
+            case O_FUN_VAR:
+            case O_FUN_N:
+            case O_FUN_BC:
+            case O_FUN_ASM:
+            // what about O_CLASS, and an O_OBJ that has a __call__ method?
+                return true;
+            default:
+                return false;
+        }
+    }
+}
+
+machine_int_t py_obj_hash(py_obj_t o_in) {
+    if (o_in == py_const_false) {
+        return 0; // needs to hash to same as the integer 0, since False==0
+    } else if (o_in == py_const_true) {
+        return 1; // needs to hash to same as the integer 1, since True==1
+    } else if (IS_SMALL_INT(o_in)) {
+        return FROM_SMALL_INT(o_in);
+    } else if (IS_O(o_in, O_CONST)) {
+        return (machine_int_t)o_in;
+    } else if (IS_O(o_in, O_STR)) {
+        return ((py_obj_base_t*)o_in)->u_str;
+    } else {
+        assert(0);
+        return 0;