Merge c048ec9faf4bfe02da004dce69471897933c3617 into d527b4dbc97235986f32accf34b246ec1a758dc2

Problem: When an exception occurs in a WITH block, __exit__ was not called,
preventing proper cleanup of context managers.

Solution:
1. Wrap WITH block body in try-except structure
2. On normal exit: call __exit__(None, None, None)
3. On exception: call __exit__ with exception info before re-raising

Changes:
- compiler.c: Wrap WITH body in try-except, ensure __exit__ called in both paths
- ceval.c: Update OP_WITH_EXIT to accept three arguments (exc_type, exc_val, exc_tb)
- tests/520_context.py: Add test to verify __exit__ called on exceptions

3rd/msgpack/src/bindings.c

@@ -56,8 +56,10 @@ static bool mpack_to_py(mpack_node_t node) {
             for(size_t i = 0; i < count; i++) {
                 mpack_node_t key_node = mpack_node_map_key_at(node, i);
                 mpack_node_t val_node = mpack_node_map_value_at(node, i);
-                if(mpack_node_type(key_node) != mpack_type_str) {
-                    return TypeError("msgpack: key must be strings");
+                mpack_type_t key_type = mpack_node_type(key_node);
+                if(key_type != mpack_type_str && key_type != mpack_type_int &&
+                   key_type != mpack_type_uint) {
+                    return TypeError("msgpack: key must be string or integer");
                 }
                 if(!mpack_to_py(key_node)) return false;
                 if(!mpack_to_py(val_node)) return false;
@@ -101,9 +103,14 @@ static bool py_to_mpack(py_Ref object, mpack_writer_t* writer);
 
 static bool mpack_write_dict_kv(py_Ref k, py_Ref v, void* ctx) {
     mpack_writer_t* writer = ctx;
-    if(k->type != tp_str) return TypeError("msgpack: key must be strings");
-    c11_sv sv = py_tosv(k);
-    mpack_write_str(writer, sv.data, (size_t)sv.size);
+    if(k->type == tp_str) {
+        c11_sv sv = py_tosv(k);
+        mpack_write_str(writer, sv.data, (size_t)sv.size);
+    } else if(k->type == tp_int) {
+        mpack_write_int(writer, py_toint(k));
+    } else {
+        return TypeError("msgpack: key must be string or integer");
+    }
     bool ok = py_to_mpack(v, writer);
     if(!ok) mpack_write_nil(writer);
     return ok;
@@ -160,7 +167,10 @@ static bool msgpack_dumps(int argc, py_Ref argv) {
     mpack_writer_init_growable(&writer, &data, &size);
     bool ok = py_to_mpack(argv, &writer);
     if(mpack_writer_destroy(&writer) != mpack_ok) { assert(false); }
-    if(!ok) return false;
+    if(!ok) {
+        MPACK_FREE(data);
+        return false;
+    }
     assert(size <= INT32_MAX);
     unsigned char* byte_data = py_newbytes(py_retval(), (int)size);
     memcpy(byte_data, data, size);

src/bindings/py_str.c

@@ -228,7 +228,8 @@ static bool str__getitem__(int argc, py_Ref argv) {
     py_Ref _1 = py_arg(1);
     if(_1->type == tp_int) {
         int index = py_toint(py_arg(1));
-        if(!pk__normalize_index(&index, self.size)) return false;
+        int u8_len = c11_sv__u8_length(self);
+        if(!pk__normalize_index(&index, u8_len)) return false;
         c11_sv res = c11_sv__u8_getitem(self, index);
         py_newstrv(py_retval(), res);
         return true;

src/common/str.c

@@ -104,8 +104,10 @@ c11_sv c11_sv__slice(c11_sv sv, int start) { return c11_sv__slice2(sv, start, sv
 
 c11_sv c11_sv__slice2(c11_sv sv, int start, int stop) {
     if(start < 0) start = 0;
-    if(stop < start) stop = start;
+    if(start > sv.size) start = sv.size;
+    if(stop < 0) stop = 0;
     if(stop > sv.size) stop = sv.size;
+    if(stop < start) stop = start;
     return (c11_sv){sv.data + start, stop - start};
 }
 
@@ -997,7 +999,6 @@ const static c11_u32_range kLoRanges[] = {
 // clang-format on
 
 bool c11__is_unicode_Lo_char(int c) {
-    if(c == 0x1f955) return true;
     const char* data =
         c11__search_u32_ranges(c, kLoRanges, sizeof(kLoRanges) / sizeof(c11_u32_range));
     return data != NULL;

src/compiler/compiler.c

@@ -2847,6 +2847,8 @@ static Error* compile_stmt(Compiler* self) {
         case TK_WITH: {
             check(EXPR(self));  // [ <expr> ]
             Ctx__s_emit_top(ctx());
+            // Save context manager for later __exit__ call
+            Ctx__emit_(ctx(), OP_DUP_TOP, BC_NOARG, prev()->line);
             Ctx__enter_block(ctx(), CodeBlockType_WITH);
             NameExpr* as_name = NULL;
             if(match(TK_AS)) {
@@ -2855,17 +2857,33 @@ static Error* compile_stmt(Compiler* self) {
                 as_name = NameExpr__new(prev()->line, name, name_scope(self));
             }
             Ctx__emit_(ctx(), OP_WITH_ENTER, BC_NOARG, prev()->line);
-            // [ <expr> <expr>.__enter__() ]
             if(as_name) {
                 bool ok = vtemit_store((Expr*)as_name, ctx());
                 vtdelete((Expr*)as_name);
                 if(!ok) return SyntaxError(self, "invalid syntax");
             } else {
-                // discard `__enter__()`'s return value
                 Ctx__emit_(ctx(), OP_POP_TOP, BC_NOARG, BC_KEEPLINE);
             }
+            // Wrap body in try-except to ensure __exit__ is called even on exception
+            Ctx__enter_block(ctx(), CodeBlockType_TRY);
+            Ctx__emit_(ctx(), OP_BEGIN_TRY, BC_NOARG, prev()->line);
             check(compile_block_body(self));
+            Ctx__emit_(ctx(), OP_END_TRY, BC_NOARG, BC_KEEPLINE);
+            // Normal exit: call __exit__(None, None, None)
+            Ctx__emit_(ctx(), OP_LOAD_NONE, BC_NOARG, prev()->line);
+            Ctx__emit_(ctx(), OP_LOAD_NONE, BC_NOARG, prev()->line);
+            Ctx__emit_(ctx(), OP_LOAD_NONE, BC_NOARG, prev()->line);
             Ctx__emit_(ctx(), OP_WITH_EXIT, BC_NOARG, prev()->line);
+            int jump_patch = Ctx__emit_(ctx(), OP_JUMP_FORWARD, BC_NOARG, BC_KEEPLINE);
+            Ctx__exit_block(ctx());
+            // Exception handler: call __exit__ with exception info, then re-raise
+            Ctx__emit_(ctx(), OP_PUSH_EXCEPTION, BC_NOARG, BC_KEEPLINE);
+            Ctx__emit_(ctx(), OP_LOAD_NONE, BC_NOARG, BC_KEEPLINE);  // exc_type
+            Ctx__emit_(ctx(), OP_ROT_TWO, BC_NOARG, BC_KEEPLINE);     // reorder: [cm, None, exc]
+            Ctx__emit_(ctx(), OP_LOAD_NONE, BC_NOARG, BC_KEEPLINE);  // exc_tb
+            Ctx__emit_(ctx(), OP_WITH_EXIT, BC_NOARG, prev()->line);
+            Ctx__emit_(ctx(), OP_RE_RAISE, BC_NOARG, BC_KEEPLINE);
+            Ctx__patch_jump(ctx(), jump_patch);
             Ctx__exit_block(ctx());
         } break;
         /*************************************************/

src/interpreter/ceval.c

@@ -1126,14 +1126,35 @@ __NEXT_STEP:
             DISPATCH();
         }
         case OP_WITH_EXIT: {
-            // [expr]
-            py_push(TOP());
+            // Stack: [cm, exc_type, exc_val, exc_tb]
+            // Call cm.__exit__(exc_type, exc_val, exc_tb)
+            py_Ref exc_tb = TOP();
+            py_Ref exc_val = SECOND();
+            py_Ref exc_type = THIRD();
+            py_Ref cm = FOURTH();
+            
+            // Save all values from stack
+            py_TValue saved_cm = *cm;
+            py_TValue saved_exc_type = *exc_type;
+            py_TValue saved_exc_val = *exc_val;
+            py_TValue saved_exc_tb = *exc_tb;
+            self->stack.sp -= 4;
+            
+            // Push cm and get __exit__ method
+            py_push(&saved_cm);
             if(!py_pushmethod(__exit__)) {
-                TypeError("'%t' object does not support the context manager protocol", TOP()->type);
+                TypeError("'%t' object does not support the context manager protocol", saved_cm.type);
                 goto __ERROR;
             }
-            if(!py_vectorcall(0, 0)) goto __ERROR;
-            POP();
+            
+            // Push arguments: exc_type, exc_val, exc_tb
+            PUSH(&saved_exc_type);
+            PUSH(&saved_exc_val);
+            PUSH(&saved_exc_tb);
+            
+            // Call __exit__(exc_type, exc_val, exc_tb)
+            if(!py_vectorcall(3, 0)) goto __ERROR;
+            py_pop();  // discard return value
             DISPATCH();
         }
         ///////////

src/modules/pickle.c

@@ -501,10 +501,10 @@ bool py_pickle_loads_body(const unsigned char* p, int memo_length, c11_smallmap_
 bool py_pickle_loads(const unsigned char* data, int size) {
     const unsigned char* p = data;
 
-    // \xf0\x9f\xa5\x95
-    if(size < 4 || p[0] != 240 || p[1] != 159 || p[2] != 165 || p[3] != 149)
+    // PK
+    if(size < 2 || p[0] != 'P' || p[1] != 'K')
         return ValueError("invalid pickle data");
-    p += 4;
+    p += 2;
 
     c11_smallmap_d2d type_mapping;
     c11_smallmap_d2d__ctor(&type_mapping);
@@ -780,7 +780,7 @@ bool py_pickle_loads_body(const unsigned char* p, int memo_length, c11_smallmap_
 static bool PickleObject__py_submit(PickleObject* self, py_OutRef out) {
     c11_sbuf cleartext;
     c11_sbuf__ctor(&cleartext);
-    c11_sbuf__write_cstr(&cleartext, "\xf0\x9f\xa5\x95");
+    c11_sbuf__write_cstr(&cleartext, "PK");
     // line 1: type mapping
     for(py_Type type = 0; type < self->used_types_length; type++) {
         if(self->used_types[type]) {

tests/041_str.py

@@ -221,6 +221,11 @@ assert chr(0x1f955) == '🥕'
 assert ord('测') == 27979
 assert chr(27979) == '测'
 
+assert '测试'[0] == '测'
+assert '测试'[1] == '试'
+assert '测试'[-1] == '试'
+assert '测试'[-2] == '测'
+
 # test format()
 assert "Hello, {}!".format("World") == "Hello, World!"
 assert "{} {} {}".format("I", "love", "Python") == "I love Python"

tests/520_context.py

@@ -27,4 +27,29 @@ assert path == ['enter', 'in', 'exit']
 
 path.clear()
 
+# Test that __exit__ is called even when an exception occurs
+class B:
+    def __init__(self):
+        self.path = []
+    
+    def __enter__(self):
+        path.append('enter')
+        return self
+    
+    def __exit__(self, exc_type, exc_val, exc_tb):
+        path.append('exit')
+        if exc_type is not None:
+            path.append('exception')
+        return False  # propagate exception
+
+try:
+    with B():
+        path.append('before_raise')
+        raise ValueError('test')
+        path.append('after_raise')  # should not be reached
+except ValueError:
+    pass
+
+assert path == ['enter', 'before_raise', 'exit', 'exception'], f"Expected ['enter', 'before_raise', 'exit', 'exception'], got {path}"
+