Add new decode implementation

Antares0982 · Jan 1, 2025 · c084846 · c084846
1 parent c6b478d
commit c084846
Show file tree

Hide file tree

Showing 13 changed files with 1,458 additions and 0 deletions.
diff --git a/src/decode/decode_loop_dirty/1_f_f.inl.c b/src/decode/decode_loop_dirty/1_f_f.inl.c
@@ -0,0 +1,46 @@
+/*
+ * NOTE:
+ * write_as = max(_read_state.max_char_type, COMPILE_READ_UCS_LEVEL)
+ * need_copy == <escape is met>
+ * need_check_max_char = max_char_type < COMPILE_UCS_LEVEL
+ * 
+ * additional note:
+ *  1. need_copy == false => max_char_type <= COMPILE_UCS_LEVEL
+ *      !need_copy && !need_check_max_char => max_char_type == COMPILE_UCS_LEVEL
+ *      COMPILE_UCS_LEVEL < max_char_type == write_as => need_copy == true
+ *  2. COMPILE_READ_UCS_LEVEL = COMPILE_UCS_LEVEL ? COMPILE_UCS_LEVEL : 1
+ *  3. max_char_type == 4 || COMPILE_UCS_LEVEL == 0 => need_check_max_char == false
+ */
+
+// knowledge before loop: max_char_type == COMPILE_UCS_LEVEL && max_char_type <= 1
+assert(_read_state.need_copy);
+switch (_read_state.max_char_type) {
+#if COMPILE_UCS_LEVEL == PYYJSON_STRING_TYPE_ASCII
+    case PYYJSON_STRING_TYPE_ASCII: {
+        // max char can only be greater than before
+        // so before this loop we have
+        // max_char_type == COMPILE_UCS_LEVEL <= 0
+        // therefore, need_check_max_char is false
+        // 1_t_f
+        goto loop_1_t_f;
+    }
+#endif
+    case PYYJSON_STRING_TYPE_LATIN1: {
+        // max char can only be greater than before.
+        // so "new max_char_type" == PYYJSON_STRING_TYPE_LATIN1 >= "old max_char_type" == COMPILE_UCS_LEVEL
+        // 1_t_f
+        goto loop_1_t_f;
+    }
+    case PYYJSON_STRING_TYPE_UCS2: {
+        // 2_t_f
+        goto loop_2_t_f;
+    }
+    case PYYJSON_STRING_TYPE_UCS4: {
+        // 4_t_f
+        goto loop_4_t_f;
+    }
+    default: {
+        assert(false);
+        Py_UNREACHABLE();
+    }
+}
diff --git a/src/decode/decode_loop_dirty/1_f_t.inl.c b/src/decode/decode_loop_dirty/1_f_t.inl.c
@@ -0,0 +1,48 @@
+/*
+ * NOTE:
+ * write_as = max(_read_state.max_char_type, COMPILE_READ_UCS_LEVEL)
+ * need_copy == <escape is met>
+ * need_check_max_char = max_char_type < COMPILE_UCS_LEVEL
+ * 
+ * additional note:
+ *  1. need_copy == false => max_char_type <= COMPILE_UCS_LEVEL
+ *      !need_copy && !need_check_max_char => max_char_type == COMPILE_UCS_LEVEL
+ *      COMPILE_UCS_LEVEL < max_char_type == write_as => need_copy == true
+ *  2. COMPILE_READ_UCS_LEVEL = COMPILE_UCS_LEVEL ? COMPILE_UCS_LEVEL : 1
+ *  3. max_char_type == 4 || COMPILE_UCS_LEVEL == 0 => need_check_max_char == false
+ */
+
+// knowledge before loop: max_char_type == 0 && COMPILE_UCS_LEVEL == 1
+if (_read_state.need_copy) {
+    // escaped.
+    switch (_read_state.max_char_type) {
+        case PYYJSON_STRING_TYPE_ASCII: {
+            // max_char_type not updated
+            // 1_t_t
+            goto loop_1_t_t;
+        }
+        case PYYJSON_STRING_TYPE_LATIN1: {
+            // 1_t_f
+            goto loop_1_t_f;
+        }
+        case PYYJSON_STRING_TYPE_UCS2: {
+            // 2_t_f
+            goto loop_2_t_f;
+        }
+        case PYYJSON_STRING_TYPE_UCS4: {
+            // 4_t_f
+            goto loop_4_t_f;
+        }
+        default: {
+            assert(false);
+            Py_UNREACHABLE();
+        }
+    }
+} else {
+    // "check max char" is dirty without any escape, max_char_type must have been updated
+    // so we must have "old max_char_type" == 0 and
+    // 1 == "new max_char_type" <= COMPILE_UCS_LEVEL == 1,
+    // since need_copy is false
+    assert(_read_state.max_char_type == COMPILE_UCS_LEVEL && COMPILE_UCS_LEVEL == 1);
+    goto loop_1_f_f;
+}
diff --git a/src/decode/decode_loop_dirty/1_t_f.inl.c b/src/decode/decode_loop_dirty/1_t_f.inl.c
@@ -0,0 +1,47 @@
+/*
+ * NOTE:
+ * write_as = max(_read_state.max_char_type, COMPILE_READ_UCS_LEVEL)
+ * need_copy == <escape is met>
+ * need_check_max_char = max_char_type < COMPILE_UCS_LEVEL
+ * 
+ * additional note:
+ *  1. need_copy == false => max_char_type <= COMPILE_UCS_LEVEL
+ *      !need_copy && !need_check_max_char => max_char_type == COMPILE_UCS_LEVEL
+ *      COMPILE_UCS_LEVEL < max_char_type == write_as => need_copy == true
+ *  2. COMPILE_READ_UCS_LEVEL = COMPILE_UCS_LEVEL ? COMPILE_UCS_LEVEL : 1
+ *  3. max_char_type == 4 || COMPILE_UCS_LEVEL == 0 => need_check_max_char == false
+ */
+
+// knowledge before loop: 1 >= max_char_type >= COMPILE_UCS_LEVEL
+assert(_read_state.need_copy);
+
+switch (_read_state.max_char_type) {
+#if !defined(NDEBUG) && COMPILE_UCS_LEVEL == PYYJSON_STRING_TYPE_ASCII
+    case PYYJSON_STRING_TYPE_ASCII: {
+        // max char can only be greater than before.
+        // this means max_char == 0
+        assert(false);
+        // 1_t_f (self), not really dirty.
+        goto loop_1_t_f;
+    }
+#endif
+#if !defined(NDEBUG)
+    case PYYJSON_STRING_TYPE_LATIN1: {
+        assert(false);
+        // 1_t_f (self), not really dirty.
+        goto loop_1_t_f;
+    }
+#endif
+    case PYYJSON_STRING_TYPE_UCS2: {
+        // 2_t_f
+        goto loop_2_t_f;
+    }
+    case PYYJSON_STRING_TYPE_UCS4: {
+        // 4_t_f
+        goto loop_4_t_f;
+    }
+    default: {
+        assert(false);
+        Py_UNREACHABLE();
+    }
+}
diff --git a/src/decode/decode_loop_dirty/1_t_t.inl.c b/src/decode/decode_loop_dirty/1_t_t.inl.c
@@ -0,0 +1,40 @@
+/*
+ * NOTE:
+ * write_as = max(_read_state.max_char_type, COMPILE_READ_UCS_LEVEL)
+ * need_copy == <escape is met>
+ * need_check_max_char = max_char_type < COMPILE_UCS_LEVEL
+ * 
+ * additional note:
+ *  1. need_copy == false => max_char_type <= COMPILE_UCS_LEVEL
+ *      !need_copy && !need_check_max_char => max_char_type == COMPILE_UCS_LEVEL
+ *      COMPILE_UCS_LEVEL < max_char_type == write_as => need_copy == true
+ *  2. COMPILE_READ_UCS_LEVEL = COMPILE_UCS_LEVEL ? COMPILE_UCS_LEVEL : 1
+ *  3. max_char_type == 4 || COMPILE_UCS_LEVEL == 0 => need_check_max_char == false
+ */
+
+// knowledge before loop: max_char_type == 0 && COMPILE_UCS_LEVEL == 1
+switch (_read_state.max_char_type) {
+#if !defined(NDEBUG)
+    case PYYJSON_STRING_TYPE_ASCII: {
+        assert(false);
+        // 1_t_t (self), not really dirty.
+        goto loop_1_t_t;
+    }
+#endif
+    case PYYJSON_STRING_TYPE_LATIN1: {
+        // 1_t_f
+        goto loop_1_t_f;
+    }
+    case PYYJSON_STRING_TYPE_UCS2: {
+        // 2_t_f
+        goto loop_2_t_f;
+    }
+    case PYYJSON_STRING_TYPE_UCS4: {
+        // 4_t_f
+        goto loop_4_t_f;
+    }
+    default: {
+        assert(false);
+        Py_UNREACHABLE();
+    }
+}
diff --git a/src/decode/decode_loop_dirty/2_f_f.inl.c b/src/decode/decode_loop_dirty/2_f_f.inl.c
@@ -0,0 +1,30 @@
+/*
+ * NOTE:
+ * write_as = max(_read_state.max_char_type, COMPILE_READ_UCS_LEVEL)
+ * need_copy == <escape is met>
+ * need_check_max_char = max_char_type < COMPILE_UCS_LEVEL
+ * 
+ * additional note:
+ *  1. need_copy == false => max_char_type <= COMPILE_UCS_LEVEL
+ *      !need_copy && !need_check_max_char => max_char_type == COMPILE_UCS_LEVEL
+ *      COMPILE_UCS_LEVEL < max_char_type == write_as => need_copy == true
+ *  2. COMPILE_READ_UCS_LEVEL = COMPILE_UCS_LEVEL ? COMPILE_UCS_LEVEL : 1
+ *  3. max_char_type == 4 || COMPILE_UCS_LEVEL == 0 => need_check_max_char == false
+ */
+
+// knowledge before loop: 2 == max_char_type == COMPILE_UCS_LEVEL
+assert(_read_state.need_copy);
+switch (_read_state.max_char_type) {
+    case PYYJSON_STRING_TYPE_UCS2: {
+        // 2_t_f
+        goto loop_2_t_f;
+    }
+    case PYYJSON_STRING_TYPE_UCS4: {
+        // 4_t_f
+        goto loop_4_t_f;
+    }
+    default: {
+        assert(false);
+        Py_UNREACHABLE();
+    }
+}
diff --git a/src/decode/decode_loop_dirty/2_f_t.inl.c b/src/decode/decode_loop_dirty/2_f_t.inl.c
@@ -0,0 +1,52 @@
+/*
+ * NOTE:
+ * write_as = max(_read_state.max_char_type, COMPILE_READ_UCS_LEVEL)
+ * need_copy == <escape is met>
+ * need_check_max_char = max_char_type < COMPILE_UCS_LEVEL
+ * 
+ * additional note:
+ *  1. need_copy == false => max_char_type <= COMPILE_UCS_LEVEL
+ *      !need_copy && !need_check_max_char => max_char_type == COMPILE_UCS_LEVEL
+ *      COMPILE_UCS_LEVEL < max_char_type == write_as => need_copy == true
+ *  2. COMPILE_READ_UCS_LEVEL = COMPILE_UCS_LEVEL ? COMPILE_UCS_LEVEL : 1
+ *  3. max_char_type == 4 || COMPILE_UCS_LEVEL == 0 => need_check_max_char == false
+ */
+
+// knowledge before loop: max_char_type < COMPILE_UCS_LEVEL == 2
+static_assert(COMPILE_UCS_LEVEL == 2, "COMPILE_UCS_LEVEL == 2");
+if (_read_state.need_copy) {
+    // escaped.
+    switch (_read_state.max_char_type) {
+        case PYYJSON_STRING_TYPE_ASCII: {
+            // 1_t_t
+            goto loop_2_t_t;
+        }
+        case PYYJSON_STRING_TYPE_LATIN1: {
+            // 1_t_t
+            goto loop_2_t_t;
+        }
+        case PYYJSON_STRING_TYPE_UCS2: {
+            // 2_t_f
+            goto loop_2_t_f;
+        }
+        case PYYJSON_STRING_TYPE_UCS4: {
+            // 4_t_f
+            goto loop_4_t_f;
+        }
+        default: {
+            assert(false);
+            Py_UNREACHABLE();
+        }
+    }
+} else {
+    // check max char is dirty without any escape, max_char_type must have been updated
+#if !defined(NDEBUG)
+    if (_read_state.max_char_type < COMPILE_UCS_LEVEL) {
+        assert(false);
+        // 2_f_t (self), not really dirty.
+        goto loop_2_f_t;
+    }
+#endif
+    // 2_f_f
+    goto loop_2_f_f;
+}
diff --git a/src/decode/decode_loop_dirty/2_t_f.inl.c b/src/decode/decode_loop_dirty/2_t_f.inl.c
@@ -0,0 +1,27 @@
+/*
+ * NOTE:
+ * write_as = max(_read_state.max_char_type, COMPILE_READ_UCS_LEVEL)
+ * need_copy == <escape is met>
+ * need_check_max_char = max_char_type < COMPILE_UCS_LEVEL
+ * 
+ * additional note:
+ *  1. need_copy == false => max_char_type <= COMPILE_UCS_LEVEL
+ *      !need_copy && !need_check_max_char => max_char_type == COMPILE_UCS_LEVEL
+ *      COMPILE_UCS_LEVEL < max_char_type == write_as => need_copy == true
+ *  2. COMPILE_READ_UCS_LEVEL = COMPILE_UCS_LEVEL ? COMPILE_UCS_LEVEL : 1
+ *  3. max_char_type == 4 || COMPILE_UCS_LEVEL == 0 => need_check_max_char == false
+ */
+
+// knowledge before loop: max_char_type == 2 >= COMPILE_UCS_LEVEL
+assert(_read_state.need_copy);
+
+#if !defined(NDEBUG)
+if (_read_state.max_char_type == PYYJSON_STRING_TYPE_UCS2) {
+    assert(false);
+    // 2_t_f (self), not really dirty
+    goto loop_2_t_f;
+}
+#endif
+assert(_read_state.max_char_type == PYYJSON_STRING_TYPE_UCS4);
+// 4_t_f
+goto loop_4_t_f;
diff --git a/src/decode/decode_loop_dirty/2_t_t.inl.c b/src/decode/decode_loop_dirty/2_t_t.inl.c
@@ -0,0 +1,33 @@
+/*
+ * NOTE:
+ * write_as = max(_read_state.max_char_type, COMPILE_READ_UCS_LEVEL)
+ * need_copy == <escape is met>
+ * need_check_max_char = max_char_type < COMPILE_UCS_LEVEL
+ * 
+ * additional note:
+ *  1. need_copy == false => max_char_type <= COMPILE_UCS_LEVEL
+ *      !need_copy && !need_check_max_char => max_char_type == COMPILE_UCS_LEVEL
+ *      COMPILE_UCS_LEVEL < max_char_type == write_as => need_copy == true
+ *  2. COMPILE_READ_UCS_LEVEL = COMPILE_UCS_LEVEL ? COMPILE_UCS_LEVEL : 1
+ *  3. max_char_type == 4 || COMPILE_UCS_LEVEL == 0 => need_check_max_char == false
+ */
+
+// knowledge before loop: max_char_type < COMPILE_UCS_LEVEL == 2
+
+#if !defined(NDEBUG)
+if (_read_state.max_char_type < PYYJSON_STRING_TYPE_UCS2) {
+    assert(false);
+    // 2_t_t (self), not really dirty
+    goto loop_2_t_t;
+}
+#endif
+assert(_read_state.max_char_type >= PYYJSON_STRING_TYPE_UCS2);
+
+if (_read_state.max_char_type == PYYJSON_STRING_TYPE_UCS2) {
+    // 2_t_f
+    goto loop_2_t_f;
+} else {
+    assert(_read_state.max_char_type == PYYJSON_STRING_TYPE_UCS4);
+    // 4_t_f
+    goto loop_4_t_f;
+}
diff --git a/src/decode/decode_loop_dirty/4_f_f.inl.c b/src/decode/decode_loop_dirty/4_f_f.inl.c
@@ -0,0 +1,18 @@
+/*
+ * NOTE:
+ * write_as = max(_read_state.max_char_type, COMPILE_READ_UCS_LEVEL)
+ * need_copy == <escape is met>
+ * need_check_max_char = max_char_type < COMPILE_UCS_LEVEL
+ * 
+ * additional note:
+ *  1. need_copy == false => max_char_type <= COMPILE_UCS_LEVEL
+ *      !need_copy && !need_check_max_char => max_char_type == COMPILE_UCS_LEVEL
+ *      COMPILE_UCS_LEVEL < max_char_type == write_as => need_copy == true
+ *  2. COMPILE_READ_UCS_LEVEL = COMPILE_UCS_LEVEL ? COMPILE_UCS_LEVEL : 1
+ *  3. max_char_type == 4 || COMPILE_UCS_LEVEL == 0 => need_check_max_char == false
+ */
+
+// knowledge before loop: max_char_type == COMPILE_UCS_LEVEL == 4
+assert(_read_state.need_copy);
+assert(_read_state.max_char_type == PYYJSON_STRING_TYPE_UCS4);
+goto loop_4_t_f;
diff --git a/src/decode/decode_loop_dirty/4_f_t.inl.c b/src/decode/decode_loop_dirty/4_f_t.inl.c
@@ -0,0 +1,36 @@
+/*
+ * NOTE:
+ * write_as = max(_read_state.max_char_type, COMPILE_READ_UCS_LEVEL)
+ * need_copy == <escape is met>
+ * need_check_max_char = max_char_type < COMPILE_UCS_LEVEL
+ * 
+ * additional note:
+ *  1. need_copy == false => max_char_type <= COMPILE_UCS_LEVEL
+ *      !need_copy && !need_check_max_char => max_char_type == COMPILE_UCS_LEVEL
+ *      COMPILE_UCS_LEVEL < max_char_type == write_as => need_copy == true
+ *  2. COMPILE_READ_UCS_LEVEL = COMPILE_UCS_LEVEL ? COMPILE_UCS_LEVEL : 1
+ *  3. max_char_type == 4 || COMPILE_UCS_LEVEL == 0 => need_check_max_char == false
+ */
+
+// knowledge before loop: max_char_type < COMPILE_UCS_LEVEL == 4
+if (_read_state.need_copy) {
+    // escaped.
+    if (_read_state.max_char_type == PYYJSON_STRING_TYPE_UCS4) {
+        goto loop_4_t_f;
+    } else {
+        assert(_read_state.max_char_type >= PYYJSON_STRING_TYPE_ASCII && _read_state.max_char_type <= PYYJSON_STRING_TYPE_UCS2);
+        goto loop_4_t_t;
+    }
+} else {
+    // check max char is dirty without any escape, max_char_type must have been updated
+#if !defined(NDEBUG)
+    if (_read_state.max_char_type < COMPILE_UCS_LEVEL) {
+        assert(false);
+        // loop_4_f_t (self), not really dirty.
+        goto loop_4_f_t;
+    }
+#endif
+    assert(_read_state.max_char_type == COMPILE_UCS_LEVEL);
+    // 4_f_f
+    goto loop_4_f_f;
+}