Update syntax parser system.

include/internal/syntax/decl.h

@@ -7,7 +7,7 @@ namespace bed::internal::syntax {
 struct Token {
   uint32_t start;
   uint32_t end;
-  enum : uint8_t {
+  enum Kind : uint8_t {
     Data,
     Shebang,
     Comment,
@@ -68,9 +68,20 @@ struct Token {
   } type;
 };
 
+struct ParseEvent {
+  std::string_view name;
+  enum : uint8_t {
+    Opening,
+    Closing,
+    SymbolDef,
+    Symbol
+  } ev_type;
+  uint8_t type; // type.
+};
+
 struct Language {
   std::function<void *()> none_state;
-  std::function<void(void **, std::string_view, bool, std::vector<Token> *)> parse;
+  std::function<void(void **, std::string_view, bool, std::vector<Token> *, std::vector<ParseEvent> *)> parse;
   std::function<void *(void *)> copy;
   std::function<bool(void *, void *)> equal;
   std::function<void(void *)> destroy;
@@ -102,81 +113,40 @@ struct TreeCursor {
   ParseStateBranch *stack[64];
   uint8_t depth = 0;
   bool went_left[64];
-  TreeCursor(ParseState *root, uint64_t target_line, uint64_t *relative) {
-    ParseState *node = root;
-    while (node->is_branch()) {
-      auto *branch = (ParseStateBranch *)node;
-      auto *left = branch->left;
-      stack[depth] = branch;
-      if (target_line < left->lines()) {
-        went_left[depth] = true;
-        ++depth;
-        node = left;
-      } else {
-        target_line -= left->lines();
-        went_left[depth] = false;
-        ++depth;
-        node = branch->right;
-      }
-    }
-    *relative = target_line;
-    leaf = (ParseStateLeaf *)node;
+  TreeCursor(ParseState *root, uint64_t target_line, uint64_t *relative);
+  void next();
+  void prev();
+};
+
+struct ScopeNode {
+  static constexpr uint64_t SCOPE_BIT = 1ull << 63;
+  static constexpr uint64_t LINES_MASK = ~SCOPE_BIT;
+  uint64_t header;
+  bool is_scope() const {
+    return header & SCOPE_BIT;
   }
-  void next() {
-    while (depth > 0) {
-      auto *branch = stack[depth - 1];
-      bool from_left = went_left[depth - 1];
-      --depth;
-      if (!from_left)
-        continue;
-      ParseState *node = branch->right;
-      while (node->is_branch()) {
-        auto *b = (ParseStateBranch *)node;
-        stack[depth] = b;
-        went_left[depth] = true;
-        ++depth;
-        node = b->left;
-      }
-      leaf = (ParseStateLeaf *)node;
-      return;
-    }
-    leaf = nullptr;
-  }
-  void prev() {
-    while (depth > 0) {
-      auto *branch = stack[depth - 1];
-      bool from_left = went_left[depth - 1];
-      --depth;
-      if (from_left)
-        continue;
-      ParseState *node = branch->left;
-      while (node->is_branch()) {
-        auto *b = (ParseStateBranch *)node;
-        stack[depth] = b;
-        went_left[depth] = false;
-        ++depth;
-        node = b->right;
-      }
-      leaf = (ParseStateLeaf *)node;
-      return;
-    }
-    leaf = nullptr;
+  uint64_t lines() const {
+    return header & LINES_MASK;
   }
 };
 
 struct Symbol {
-  uint64_t definition;
-  std::vector<uint64_t> references;
+  uint32_t definition;
+  uint16_t reference_count;
+  uint8_t type;
+  uint8_t len;
+  // first chars of count len, padded to 4 bytes.
+  // then a set of references. 32bit
+  // references later as modifying at end is faster than shifting the name.
 };
 
-struct Space {
-  uint64_t len;
-};
-
-struct Scope {
-  uint64_t len;
-  uint32_t type;
-  trie::Trie<Symbol> symbols;
-  std::vector<std::variant<Space, Scope>> children;
+struct alignas(8) Scope : ScopeNode {
+  uint16_t children_count;
+  uint16_t symbol_count;
+  uint8_t type;
+  uint8_t len;
+  uint32_t : 32; // if more stuff is needed use the padding first.
+  // first chars of count len, padded to 8 bytes.
+  // followed by that many number of pointers.
 };
 } // namespace bed::internal::syntax