fix a bug about %s -> %v

This reverts commit b03feb0294c984002992b5b226df02d0d53103f5.

include/pocketpy/common/utils.h

@@ -12,9 +12,9 @@
 #define PK_HEX_TABLE "0123456789abcdef"
 
 #ifdef _MSC_VER
-#define c11__unreachedable() __assume(0)
+#define c11__unreachable() __assume(0)
 #else
-#define c11__unreachedable() __builtin_unreachable()
+#define c11__unreachable() __builtin_unreachable()
 #endif
 
 #define c11__abort(...)                                                                            \

scripts/gen_primes.py

@@ -0,0 +1,45 @@
+import numba
+from typing import List
+
+@numba.jit(nopython=True)
+def sieve_of_eratosthenes(n: int) -> List[int]:
+    assert n >= 2
+    is_prime = [True] * (n + 1)
+    is_prime[0] = is_prime[1] = False  # 0 和 1 不是素数
+    
+    for start in range(2, int(n**0.5) + 1):
+        if is_prime[start]:
+            for multiple in range(start*start, n + 1, start):
+                is_prime[multiple] = False
+    
+    primes = [num for num, prime in enumerate(is_prime) if prime]
+    return primes
+
+all_primes = sieve_of_eratosthenes(2**30)
+print(len(all_primes), all_primes[:10], all_primes[-10:])
+
+index = 3
+caps = [all_primes[index]]
+
+while True:
+    for i in range(index+1, len(all_primes)):
+        last_cap = caps[-1]
+        if last_cap < 1000:
+            min_cap = last_cap * 2
+        else:
+            min_cap = last_cap * 1.5
+        if all_primes[i] >= min_cap:
+            caps.append(all_primes[i])
+            index = i
+            break
+    else:
+        break
+
+print('-'*20)
+print(caps)
+
+print('switch(cap) {')
+for i in range(len(caps)-1):
+    print(f'    case {caps[i]}:', f'return {caps[i+1]};')
+print('    default: c11__unreachable();')
+print('}')

src/compiler/compiler.c

@@ -128,7 +128,7 @@ bool NameExpr__emit_del(Expr* self_, Ctx* ctx) {
             break;
         case NAME_GLOBAL: Ctx__emit_(ctx, OP_DELETE_GLOBAL, self->name, self->line); break;
         case NAME_GLOBAL_UNKNOWN: Ctx__emit_(ctx, OP_DELETE_NAME, self->name, self->line); break;
-        default: c11__unreachedable();
+        default: c11__unreachable();
     }
     return true;
 }
@@ -327,7 +327,7 @@ void LiteralExpr__emit_(Expr* self_, Ctx* ctx) {
             Ctx__emit_(ctx, OP_LOAD_CONST, index, self->line);
             break;
         }
-        default: c11__unreachedable();
+        default: c11__unreachable();
     }
 }
 
@@ -355,7 +355,7 @@ void Literal0Expr__emit_(Expr* self_, Ctx* ctx) {
         case TK_TRUE: opcode = OP_LOAD_TRUE; break;
         case TK_FALSE: opcode = OP_LOAD_FALSE; break;
         case TK_DOTDOTDOT: opcode = OP_LOAD_ELLIPSIS; break;
-        default: c11__unreachedable();
+        default: c11__unreachable();
     }
     Ctx__emit_(ctx, opcode, BC_NOARG, self->line);
 }
@@ -1245,7 +1245,7 @@ static void Ctx__emit_store_name(Ctx* self, NameScope scope, py_Name name, int l
         case NAME_LOCAL: Ctx__emit_(self, OP_STORE_FAST, Ctx__add_varname(self, name), line); break;
         case NAME_GLOBAL: Ctx__emit_(self, OP_STORE_GLOBAL, name, line); break;
         case NAME_GLOBAL_UNKNOWN: Ctx__emit_(self, OP_STORE_NAME, name, line); break;
-        default: c11__unreachedable();
+        default: c11__unreachable();
     }
 }
 
@@ -2182,7 +2182,7 @@ static Error* read_literal(Compiler* self, py_Ref out) {
             } else if(value->index == TokenValue_F64) {
                 py_newfloat(out, negated ? -value->_f64 : value->_f64);
             } else {
-                c11__unreachedable();
+                c11__unreachable();
             }
             return NULL;
         }

src/interpreter/ceval.c

@@ -60,7 +60,7 @@ static bool stack_format_object(VM* self, c11_sv spec);
             case RES_RETURN: PUSH(&self->last_retval); break;                                      \
             case RES_CALL: frame = self->top_frame; goto __NEXT_FRAME;                             \
             case RES_ERROR: goto __ERROR;                                                          \
-            default: c11__unreachedable();                                                         \
+            default: c11__unreachable();                                                         \
         }                                                                                          \
     } while(0)
 
@@ -1101,10 +1101,10 @@ FrameResult VM__run_top_frame(VM* self) {
                 if(!ok) goto __ERROR;
                 DISPATCH();
             }
-            default: c11__unreachedable();
+            default: c11__unreachable();
         }
 
-        c11__unreachedable();
+        c11__unreachable();
 
     __ERROR:
         py_BaseException__stpush(&self->curr_exception,
@@ -1340,7 +1340,7 @@ static bool stack_format_object(VM* self, c11_sv spec) {
                 c11_sbuf__write_pad(&buf, pad_right, pad_c);
                 break;
             }
-            default: c11__unreachedable();
+            default: c11__unreachable();
         }
     } else {
         c11_sbuf__write_sv(&buf, c11_string__sv(body));

src/interpreter/vm.c

@@ -502,10 +502,10 @@ FrameResult VM__vectorcall(VM* self, uint16_t argc, uint16_t kwargc, bool opcall
                 self->stack.sp = p0;    // reset the stack
                 return RES_RETURN;
             }
-            default: c11__unreachedable();
+            default: c11__unreachable();
         };
 
-        c11__unreachedable();
+        c11__unreachable();
         /*****************_py_call*****************/
     }
 

src/modules/dis.c

@@ -61,7 +61,7 @@ static bool disassemble(CodeObject* co) {
                 case OP_LOAD_CONST: {
                     py_Ref value = c11__at(py_TValue, &co->consts, byte.arg);
                     if(py_repr(value)) {
-                        pk_sprintf(&ss, " (%s)", py_tosv(py_retval()));
+                        pk_sprintf(&ss, " (%v)", py_tosv(py_retval()));
                     } else {
                         return false;
                     }

src/modules/linalg.c

@@ -298,21 +298,32 @@ DEF_VECTOR_OPS(3)
             sum += a.data[i] * b.data[i];                                                          \
         py_newint(py_retval(), sum);                                                               \
         return true;                                                                               \
-    }                                                                                              \
-    static bool vec##D##i##__hash__(int argc, py_Ref argv) {                                       \
-        PY_CHECK_ARGC(1);                                                                          \
-        const uint32_t C = 2654435761;                                                             \
-        c11_vec##D##i v = py_tovec##D##i(argv);                                                    \
-        uint64_t hash = 0;                                                                         \
-        for(int i = 0; i < D; i++)                                                                 \
-            hash = hash * 31 + (uint32_t)v.data[i] * C;                                            \
-        py_newint(py_retval(), (py_i64)hash);                                                      \
-        return true;                                                                               \
     }
 
 DEF_VECTOR_INT_OPS(2)
 DEF_VECTOR_INT_OPS(3)
 
+static bool vec2i__hash__(int argc, py_Ref argv) {
+    PY_CHECK_ARGC(1);
+    c11_vec2i v = py_tovec2i(argv);
+    uint64_t x_part = (uint32_t)v.x & 0xFFFFFFFF;
+    uint64_t y_part = (uint32_t)v.y & 0xFFFFFFFF;
+    uint64_t hash = (x_part << 32) | y_part;
+    py_newint(py_retval(), (py_i64)hash);
+    return true;
+}
+
+static bool vec3i__hash__(int argc, py_Ref argv) {
+    PY_CHECK_ARGC(1);
+    c11_vec3i v = py_tovec3i(argv);
+    uint64_t x_part = (uint32_t)v.x & 0xFFFFFF;
+    uint64_t y_part = (uint32_t)v.y & 0xFFFFFF;
+    uint64_t z_part = (uint32_t)v.z & 0xFFFF;
+    uint64_t hash = (x_part << 40) | (y_part << 16) | z_part;
+    py_newint(py_retval(), (py_i64)hash);
+    return true;
+}
+
 static bool vec2__repr__(int argc, py_Ref argv) {
     PY_CHECK_ARGC(1);
     char buf[64];

src/public/exec.c

@@ -64,7 +64,7 @@ bool pk_exec(CodeObject* co, py_Ref module) {
     FrameResult res = VM__run_top_frame(vm);
     if(res == RES_ERROR) return false;
     if(res == RES_RETURN) return true;
-    c11__unreachedable();
+    c11__unreachable();
 }
 
 bool py_exec(const char* source, const char* filename, enum py_CompileMode mode, py_Ref module) {

src/public/internal.c

@@ -210,7 +210,7 @@ bool pk_loadmethod(py_StackRef self, py_Name name) {
                 self[0] = *py_getslot(cls_var, 0);
                 self[1] = pk__type_info(type)->self;
                 break;
-            default: c11__unreachedable();
+            default: c11__unreachable();
         }
         return true;
     }

src/public/py_dict.c

@@ -7,8 +7,54 @@
 
 #define PK_DICT_MAX_COLLISION 4
 
+static uint32_t Dict__next_cap(uint32_t cap) {
+    switch(cap) {
+        case 7: return 17;
+        case 17: return 37;
+        case 37: return 79;
+        case 79: return 163;
+        case 163: return 331;
+        case 331: return 673;
+        case 673: return 1361;
+        case 1361: return 2053;
+        case 2053: return 3083;
+        case 3083: return 4637;
+        case 4637: return 6959;
+        case 6959: return 10453;
+        case 10453: return 15683;
+        case 15683: return 23531;
+        case 23531: return 35311;
+        case 35311: return 52967;
+        case 52967: return 79451;
+        case 79451: return 119179;
+        case 119179: return 178781;
+        case 178781: return 268189;
+        case 268189: return 402299;
+        case 402299: return 603457;
+        case 603457: return 905189;
+        case 905189: return 1357787;
+        case 1357787: return 2036687;
+        case 2036687: return 3055043;
+        case 3055043: return 4582577;
+        case 4582577: return 6873871;
+        case 6873871: return 10310819;
+        case 10310819: return 15466229;
+        case 15466229: return 23199347;
+        case 23199347: return 34799021;
+        case 34799021: return 52198537;
+        case 52198537: return 78297827;
+        case 78297827: return 117446801;
+        case 117446801: return 176170229;
+        case 176170229: return 264255353;
+        case 264255353: return 396383041;
+        case 396383041: return 594574583;
+        case 594574583: return 891861923;
+        default: c11__unreachable();
+    }
+}
+
 typedef struct {
-    py_i64 hash;
+    uint64_t hash;
     py_TValue key;
     py_TValue val;
 } DictEntry;
@@ -19,7 +65,7 @@ typedef struct {
 
 typedef struct {
     int length;
-    int capacity;
+    uint32_t capacity;
     DictIndex* indices;
     c11_vector /*T=DictEntry*/ entries;
 } Dict;
@@ -29,13 +75,13 @@ typedef struct {
     DictEntry* end;
 } DictIterator;
 
-static void Dict__ctor(Dict* self, int capacity) {
+static void Dict__ctor(Dict* self, uint32_t capacity, int entries_capacity) {
     self->length = 0;
     self->capacity = capacity;
     self->indices = malloc(self->capacity * sizeof(DictIndex));
     memset(self->indices, -1, self->capacity * sizeof(DictIndex));
     c11_vector__ctor(&self->entries, sizeof(DictEntry));
-    c11_vector__reserve(&self->entries, capacity);
+    c11_vector__reserve(&self->entries, entries_capacity);
 }
 
 static void Dict__dtor(Dict* self) {
@@ -48,7 +94,7 @@ static void Dict__dtor(Dict* self) {
 static bool Dict__try_get(Dict* self, py_TValue* key, DictEntry** out) {
     py_i64 hash;
     if(!py_hash(key, &hash)) return false;
-    int idx = hash & (self->capacity - 1);
+    int idx = (uint64_t)hash % self->capacity;
     for(int i = 0; i < PK_DICT_MAX_COLLISION; i++) {
         int idx2 = self->indices[idx]._[i];
         if(idx2 == -1) continue;
@@ -72,16 +118,18 @@ static void Dict__clear(Dict* self) {
 
 static void Dict__rehash_2x(Dict* self) {
     Dict old_dict = *self;
+    uint32_t new_capacity = self->capacity;
 
-    int new_capacity = self->capacity * 2;
+__RETRY:
-
+    // use next capacity
-    do {
+    new_capacity = Dict__next_cap(new_capacity);
-        Dict__ctor(self, new_capacity);
+    // create a new dict with new capacity
-
+    Dict__ctor(self, new_capacity, old_dict.entries.capacity);
+    // move entries from old dict to new dict
     for(int i = 0; i < old_dict.entries.length; i++) {
         DictEntry* entry = c11__at(DictEntry, &old_dict.entries, i);
         if(py_isnil(&entry->key)) continue;
-            int idx = entry->hash & (new_capacity - 1);
+        int idx = entry->hash % new_capacity;
         bool success = false;
         for(int i = 0; i < PK_DICT_MAX_COLLISION; i++) {
             int idx2 = self->indices[idx]._[i];
@@ -96,14 +144,11 @@ static void Dict__rehash_2x(Dict* self) {
         }
         if(!success) {
             Dict__dtor(self);
-                new_capacity *= 2;
+            goto __RETRY;
-                continue;
         }
     }
-        // resize complete
+    // done
     Dict__dtor(&old_dict);
-        return;
-    } while(1);
 }
 
 static void Dict__compact_entries(Dict* self) {
@@ -135,13 +180,13 @@ static void Dict__compact_entries(Dict* self) {
 static bool Dict__set(Dict* self, py_TValue* key, py_TValue* val) {
     py_i64 hash;
     if(!py_hash(key, &hash)) return false;
-    int idx = hash & (self->capacity - 1);
+    int idx = (uint64_t)hash % self->capacity;
     for(int i = 0; i < PK_DICT_MAX_COLLISION; i++) {
         int idx2 = self->indices[idx]._[i];
         if(idx2 == -1) {
             // insert new entry
             DictEntry* new_entry = c11_vector__emplace(&self->entries);
-            new_entry->hash = hash;
+            new_entry->hash = (uint64_t)hash;
             new_entry->key = *key;
             new_entry->val = *val;
             self->indices[idx]._[i] = self->entries.length - 1;
@@ -159,7 +204,11 @@ static bool Dict__set(Dict* self, py_TValue* key, py_TValue* val) {
     }
     // no empty slot found
     if(self->capacity >= self->entries.length * 10) {
-        return RuntimeError("dict has too much collision: %d/%d", self->entries.length, self->capacity);
+        // raise error if we reach the minimum load factor (10%)
+        return RuntimeError("dict has too much collision: %d/%d/%d",
+                            self->entries.length,
+                            self->entries.capacity,
+                            self->capacity);
     }
     Dict__rehash_2x(self);
     return Dict__set(self, key, val);
@@ -170,7 +219,7 @@ static bool Dict__set(Dict* self, py_TValue* key, py_TValue* val) {
 static int Dict__pop(Dict* self, py_Ref key) {
     py_i64 hash;
     if(!py_hash(key, &hash)) return -1;
-    int idx = hash & (self->capacity - 1);
+    int idx = (uint64_t)hash % self->capacity;
     for(int i = 0; i < PK_DICT_MAX_COLLISION; i++) {
         int idx2 = self->indices[idx]._[i];
         if(idx2 == -1) continue;
@@ -208,13 +257,13 @@ static bool dict__new__(int argc, py_Ref argv) {
     py_Type cls = py_totype(argv);
     int slots = cls == tp_dict ? 0 : -1;
     Dict* ud = py_newobject(py_retval(), cls, slots, sizeof(Dict));
-    Dict__ctor(ud, 8);
+    Dict__ctor(ud, 7, 8);
     return true;
 }
 
 void py_newdict(py_Ref out) {
     Dict* ud = py_newobject(out, tp_dict, 0, sizeof(Dict));
-    Dict__ctor(ud, 8);
+    Dict__ctor(ud, 7, 8);
 }
 
 static bool dict__init__(int argc, py_Ref argv) {

src/public/py_number.c

@@ -244,9 +244,7 @@ static py_i64 c11_8bytes__hash(union c11_8bytes u) {
 
 static bool int__hash__(int argc, py_Ref argv) {
     PY_CHECK_ARGC(1);
-    py_i64 val = py_toint(&argv[0]);
+    py_assign(py_retval(), argv);
-    union c11_8bytes u = {._i64 = val};
-    py_newint(py_retval(), c11_8bytes__hash(u));
     return true;
 }
 

tests/08_dict.py

@@ -138,3 +138,13 @@ for i in range(n):
 #     if i % 3 == 0:
 #         y = b.pop()
 #         delattr(a, y)
+
+d = {}
+for i in range(-1000, 1000):
+    d[i] = i
+    assert d[i] == i
+
+e = {}
+for i in range(-10000, 10000, 3):
+    e[i] = i
+    assert e[i] == i

tests/80_linalg.py

@@ -408,12 +408,14 @@ assert x == 3.0 and y == 4.0
 x, y, z = vec3(1.0, 2.0, 3.0)
 assert x == 1.0 and y == 2.0 and z == 3.0
 
+assert hash(vec2i(11, -1)) == 51539607551
+assert hash(vec3i(11, -1, 0)) == 13194139467776
 
 d = {vec2i(12, 12): 1035.313708305359, vec2i(12, 11): 2059.313708305359, vec2i(12, 13): 2059.313708305359, vec2i(11, 12): 2059.313708305359, vec2i(13, 12): 2059.313708305359, vec2i(13, 11): 3083.313708305359, vec2i(13, 13): 3083.313708305359, vec2i(14, 12): 3083.313708305359, vec2i(12, 14): 3083.313708305359, vec2i(11, 13): 3083.313708305359, vec2i(12, 10): 3083.313708305359, vec2i(11, 11): 3083.313708305359, vec2i(10, 12): 3083.313708305359, vec2i(13, 14): 4107.313708305359, vec2i(14, 13): 4107.313708305359, vec2i(14, 11): 4107.313708305359, vec2i(15, 12): 4107.313708305359, vec2i(12, 15): 4107.313708305359, vec2i(11, 14): 4107.313708305359, vec2i(13, 10): 4107.313708305359, vec2i(10, 13): 4107.313708305359, vec2i(11, 10): 4107.313708305359, vec2i(10, 11): 4107.313708305359, vec2i(12, 9): 4107.313708305359, vec2i(9, 12): 4107.313708305359, vec2i(14, 14): 5131.313708305359, vec2i(15, 13): 5131.313708305359, vec2i(13, 15): 5131.313708305359, vec2i(15, 11): 5131.313708305359, vec2i(16, 12): 5131.313708305359, vec2i(12, 16): 5131.313708305359, vec2i(11, 15): 5131.313708305359, vec2i(14, 10): 5131.313708305359, vec2i(10, 14): 5131.313708305359, vec2i(13, 9): 5131.313708305359, vec2i(9, 13): 5131.313708305359}
 d[vec2i(11, 9)] = 1
 
 e = {}
-for i in range(10000):
+for i in range(-1000, 10000):
     e[vec2i(12, i)] = i
     e[vec2i(11, i)] = i
     e[vec2i(13, i)] = i