From 9ab592ce3765d07b6dbdb9c1d9b62dc8045e784e Mon Sep 17 00:00:00 2001
From: Richard Qin <richard.qin.x@gmail.com>
Date: Fri, 3 Apr 2026 02:46:12 +0800
Subject: feat: implement core term, symbol, and clause storage logic with
 canonicalization support

---
 include/atp/core/clause_store.h |  3 +++
 include/atp/core/symbol_table.h |  4 ++++
 include/atp/core/term.h         | 20 +++++++++++---------
 include/atp/core/term_bank.h    | 38 +++++++++++++++++++++++++++++++++-----
 4 files changed, 51 insertions(+), 14 deletions(-)

(limited to 'include')

diff --git a/include/atp/core/clause_store.h b/include/atp/core/clause_store.h
index 6e220cc..23db996 100644
--- a/include/atp/core/clause_store.h
+++ b/include/atp/core/clause_store.h
@@ -54,6 +54,9 @@ class ClauseStore {
 
     /// Iterate over all clauses with a visitor.
     void forEach(const std::function<void(const Clause&)>& visitor) const;
+    // function may have type erasure overhead, consider using a template
+    // template <typename Visitor>
+    // void forEach(Visitor&& visitor) const;
 
   private:
     std::vector<Clause> clauses_;
diff --git a/include/atp/core/symbol_table.h b/include/atp/core/symbol_table.h
index ed27a4d..856c395 100644
--- a/include/atp/core/symbol_table.h
+++ b/include/atp/core/symbol_table.h
@@ -23,8 +23,10 @@
 /// @brief Bidirectional mapping between symbol names and SymbolIds,
 ///        with kind and sort metadata.
 
+#include <deque>
 #include <string>
 #include <string_view>
+#include <unordered_map>
 
 #include "atp/core/types.h"
 
@@ -63,6 +65,8 @@ class SymbolTable {
 
   private:
     // TODO: Phase 1 implementation
+    std::deque<SymbolInfo> infos_;
+    std::unordered_map<std::string_view, SymbolId> name_to_id_;
 };
 
 }  // namespace atp
diff --git a/include/atp/core/term.h b/include/atp/core/term.h
index e1edfcf..017882a 100644
--- a/include/atp/core/term.h
+++ b/include/atp/core/term.h
@@ -16,28 +16,30 @@
  * along with this program.  If not, see <https://www.gnu.org/licenses/>.
  */
 
-
 #pragma once
 
 /// @file term.h
 /// @brief Core Term data structure — hash-consed, immutable.
+///
+/// V1 uses vector<TermId> for args.
+/// Future: flat arena with contiguous arg storage for cache locality.
 
-#include <span>
+#include <vector>
 
 #include "atp/core/types.h"
 
 namespace atp {
 
 /// A term is a function/predicate application: f(t1, t2, ..., tn).
-/// Stored in a flat arena; args are contiguous TermIds following the header.
-/// Variables are represented as terms with a special symbol range.
+/// Variables are 0-arity terms whose symbol has SymbolKind::kVariable.
 struct Term {
-    SymbolId symbol_id;  ///< Index into the SymbolTable
-    uint16_t arity;      ///< Number of arguments
+    SymbolId symbol_id;          ///< Index into the SymbolTable
+    std::vector<TermId> args;    ///< Argument sub-term IDs
 
-    /// Access the argument TermIds (stored immediately after this struct in the arena).
-    [[nodiscard]] std::span<const TermId> args() const;
-    [[nodiscard]] std::span<TermId> mutableArgs();
+    /// Number of arguments.
+    [[nodiscard]] uint16_t arity() const {
+        return static_cast<uint16_t>(args.size());
+    }
 };
 
 }  // namespace atp
diff --git a/include/atp/core/term_bank.h b/include/atp/core/term_bank.h
index a121f8a..f909e2f 100644
--- a/include/atp/core/term_bank.h
+++ b/include/atp/core/term_bank.h
@@ -16,42 +16,70 @@
  * along with this program.  If not, see <https://www.gnu.org/licenses/>.
  */
 
-
 #pragma once
 
 /// @file term_bank.h
 /// @brief Hash-consing term pool — guarantees structural sharing.
 
+#include "atp/core/symbol_table.h"
 #include "atp/core/term.h"
 #include "atp/core/types.h"
 
 #include <span>
+#include <unordered_map>
+#include <vector>
 
 namespace atp {
 
 /// Global term pool with hash-consing.
 /// Every structurally identical term maps to the same TermId.
+///
+/// Holds a reference to SymbolTable for variable detection.
 class TermBank {
   public:
-    TermBank() = default;
+    /// TermBank needs the SymbolTable to distinguish variables from constants
+    /// (both have 0 args, but SymbolKind differs).
+    explicit TermBank(const SymbolTable& symbols) : symbols_(symbols) {}
 
     /// Create or retrieve a term. Returns the canonical TermId.
     TermId makeTerm(SymbolId symbol, std::span<const TermId> args);
 
-    /// Create a variable term.
+    /// Create a variable term. Convenience for makeTerm(var, {}).
     TermId makeVar(SymbolId var_symbol);
 
     /// Look up a term by its ID.
     [[nodiscard]] const Term& getTerm(TermId id) const;
 
-    /// Check if a TermId represents a variable.
+    /// Check if a TermId represents a variable (delegates to SymbolTable).
     [[nodiscard]] bool isVariable(TermId id) const;
 
     /// Total number of unique terms stored.
     [[nodiscard]] size_t size() const;
 
+    /// Access the underlying SymbolTable.
+    [[nodiscard]] const SymbolTable& symbols() const { return symbols_; }
+
   private:
-    // TODO: Phase 1 implementation — flat arena + hash index
+    struct TermKey {
+        SymbolId symbol;
+        std::vector<TermId> args;
+
+        bool operator==(const TermKey& other) const = default;
+    };
+
+    struct TermKeyHash {
+        size_t operator()(const TermKey& key) const {
+            size_t seed = std::hash<uint32_t>{}(key.symbol);
+            for (TermId arg : key.args) {
+                seed ^= std::hash<uint32_t>{}(arg) + 0x9e3779b9 + (seed << 6) + (seed >> 2);
+            }
+            return seed;
+        }
+    };
+
+    const SymbolTable& symbols_;
+    std::vector<Term> terms_;
+    std::unordered_map<TermKey, TermId, TermKeyHash> term_to_id_;
 };
 
 }  // namespace atp
-- 
cgit v1.2.3