Removed peg::parse_error and introduced predicate action

README.md

@@ -216,17 +216,22 @@ peg::parser parser(R"(
 )");
 ```
 
-*Semantic predicate* support is available. We can do it by throwing a `peg::parse_error` exception in a semantic action.
+*Semantic predicate* support is available with a *predicate* action.
 
 ```cpp
 peg::parser parser("NUMBER  <-  [0-9]+");
 
-parser["NUMBER"] = [](const SemanticValues& vs) {
+parser["NUMBER"] = [](const SemanticValues &vs) {
-  auto val = vs.token_to_number<long>();
+  return vs.token_to_number<long>();
-  if (val != 100) {
+};
-    throw peg::parse_error("value error!!");
+
+parser["NUMBER"].predicate = [](const SemanticValues &vs,
+                                const std::any & /*dt*/, std::string &msg) {
+  if (vs.token_to_number<long>() != 100) {
+    msg = "value error!!";
+    return false;
   }
-  return val;
+  return true;
 };
 
 long val;

example/CMakeLists.txt

@@ -20,3 +20,6 @@ target_link_libraries(calc5 ${add_link_deps})
 
 add_executable(indent indent.cc)
 target_link_libraries(indent ${add_link_deps})
+
+add_executable(symbol_check symbol_check.cc)
+target_link_libraries(symbol_check ${add_link_deps})

example/indent.cc

@@ -39,8 +39,13 @@ Block <- Statements {}
                               size_t /*matchlen*/, std::any & /*value*/,
                               std::any & /*dt*/) { indent -= 2; };
 
-  parser["Samedent"] = [&](const SemanticValues &vs, std::any & /*dt*/) {
+  parser["Samedent"].predicate =
-    if (indent != vs.sv().size()) { throw parse_error("different indent..."); }
+      [&](const SemanticValues &vs, const std::any & /*dt*/, std::string &msg) {
+        if (indent != vs.sv().size()) {
+          msg = "different indent...";
+          return false;
+        }
+        return true;
       };
 
   parser.enable_ast();

example/symbol_check.cc

@@ -0,0 +1,69 @@
+//
+//  symbol_check.cc
+//
+//  Copyright (c) 2022 Yuji Hirose. All rights reserved.
+//  MIT License
+//
+
+#include <cstdlib>
+#include <iostream>
+#include <peglib.h>
+#include <set>
+
+using namespace peg;
+
+int main(void) {
+  parser parser(R"(
+S           <- (Decl / Ref)*
+Decl        <- 'decl' symbol(Name)
+Ref         <- 'ref' symbol_reference(Name)
+Name        <- < [a-zA-Z]+ >
+%whitespace <- [ \t\r\n]*
+
+symbol(s)           <- < s >
+symbol_reference(s) <- < s >
+)");
+
+  std::set<std::string> dic;
+
+  parser[R"(symbol)"].predicate =
+      [&](const SemanticValues &vs, const std::any & /*dt*/, std::string &msg) {
+        auto tok = vs.token_to_string();
+        if (dic.find(tok) != dic.end()) {
+          msg = "'" + tok + "' already exists...";
+          return false;
+        }
+        dic.insert(tok);
+        return true;
+      };
+
+  parser[R"(symbol_reference)"].predicate =
+      [&](const SemanticValues &vs, const std::any & /*dt*/, std::string &msg) {
+        auto tok = vs.token_to_string();
+        if (dic.find(tok) == dic.end()) {
+          msg = "'" + tok + "' doesn't exists...";
+          return false;
+        }
+        return true;
+      };
+
+  parser.enable_ast();
+
+  parser.log = [](size_t line, size_t col, const std::string &msg) {
+    std::cerr << line << ":" << col << ": " << msg << "\n";
+  };
+
+  const auto source = R"(decl aaa
+ref aaa
+ref bbb
+)";
+
+  std::shared_ptr<Ast> ast;
+  if (parser.parse(source, ast)) {
+    ast = parser.optimize_ast(ast);
+    std::cout << ast_to_s(ast);
+    return 0;
+  }
+
+  return -1;
+}

peglib.h

@@ -631,16 +631,6 @@ private:
   Fty fn_;
 };
 
-/*
- * Semantic predicate
- */
-// Note: 'parse_error' exception class should be be used in sematic action
-// handlers to reject the rule.
-class parse_error : public std::runtime_error {
-public:
-  parse_error(const char *what_arg) : std::runtime_error(what_arg) {}
-};
-
 /*
  * Parse result helper
  */
@@ -996,29 +986,29 @@ public:
 
   size_t parse_core(const char *s, size_t n, SemanticValues &vs, Context &c,
                     std::any &dt) const override {
-    auto &chldsv = c.push();
+    auto &chldvs = c.push();
     auto pop_se = scope_exit([&]() { c.pop(); });
     size_t i = 0;
     for (const auto &ope : opes_) {
       const auto &rule = *ope;
-      auto len = rule.parse(s + i, n - i, chldsv, c, dt);
+      auto len = rule.parse(s + i, n - i, chldvs, c, dt);
       if (fail(len)) { return len; }
       i += len;
     }
-    if (!chldsv.empty()) {
+    if (!chldvs.empty()) {
-      for (size_t j = 0; j < chldsv.size(); j++) {
+      for (size_t j = 0; j < chldvs.size(); j++) {
-        vs.emplace_back(std::move(chldsv[j]));
+        vs.emplace_back(std::move(chldvs[j]));
       }
     }
-    if (!chldsv.tags.empty()) {
+    if (!chldvs.tags.empty()) {
-      for (size_t j = 0; j < chldsv.tags.size(); j++) {
+      for (size_t j = 0; j < chldvs.tags.size(); j++) {
-        vs.tags.emplace_back(std::move(chldsv.tags[j]));
+        vs.tags.emplace_back(std::move(chldvs.tags[j]));
       }
     }
-    vs.sv_ = chldsv.sv_;
+    vs.sv_ = chldvs.sv_;
-    if (!chldsv.tokens.empty()) {
+    if (!chldvs.tokens.empty()) {
-      for (size_t j = 0; j < chldsv.tokens.size(); j++) {
+      for (size_t j = 0; j < chldvs.tokens.size(); j++) {
-        vs.tokens.emplace_back(std::move(chldsv.tokens[j]));
+        vs.tokens.emplace_back(std::move(chldvs.tokens[j]));
       }
     }
     return i;
@@ -1049,7 +1039,7 @@ public:
     for (const auto &ope : opes_) {
       if (!c.cut_stack.empty()) { c.cut_stack.back() = false; }
 
-      auto &chldsv = c.push();
+      auto &chldvs = c.push();
       c.push_capture_scope();
 
       c.error_info.keep_previous_token = id > 0;
@@ -1059,25 +1049,25 @@ public:
         c.error_info.keep_previous_token = false;
       });
 
-      len = ope->parse(s, n, chldsv, c, dt);
+      len = ope->parse(s, n, chldvs, c, dt);
 
       if (success(len)) {
-        if (!chldsv.empty()) {
+        if (!chldvs.empty()) {
-          for (size_t i = 0; i < chldsv.size(); i++) {
+          for (size_t i = 0; i < chldvs.size(); i++) {
-            vs.emplace_back(std::move(chldsv[i]));
+            vs.emplace_back(std::move(chldvs[i]));
           }
         }
-        if (!chldsv.tags.empty()) {
+        if (!chldvs.tags.empty()) {
-          for (size_t i = 0; i < chldsv.tags.size(); i++) {
+          for (size_t i = 0; i < chldvs.tags.size(); i++) {
-            vs.tags.emplace_back(std::move(chldsv.tags[i]));
+            vs.tags.emplace_back(std::move(chldvs.tags[i]));
           }
         }
-        vs.sv_ = chldsv.sv_;
+        vs.sv_ = chldvs.sv_;
         vs.choice_count_ = opes_.size();
         vs.choice_ = id;
-        if (!chldsv.tokens.empty()) {
+        if (!chldvs.tokens.empty()) {
-          for (size_t i = 0; i < chldsv.tokens.size(); i++) {
+          for (size_t i = 0; i < chldvs.tokens.size(); i++) {
-            vs.tokens.emplace_back(std::move(chldsv.tokens[i]));
+            vs.tokens.emplace_back(std::move(chldvs.tokens[i]));
           }
         }
         c.shift_capture_values();
@@ -1184,14 +1174,14 @@ public:
 
   size_t parse_core(const char *s, size_t n, SemanticValues & /*vs*/,
                     Context &c, std::any &dt) const override {
-    auto &chldsv = c.push();
+    auto &chldvs = c.push();
     c.push_capture_scope();
     auto se = scope_exit([&]() {
       c.pop();
       c.pop_capture_scope();
     });
     const auto &rule = *ope_;
-    auto len = rule.parse(s, n, chldsv, c, dt);
+    auto len = rule.parse(s, n, chldvs, c, dt);
     if (success(len)) {
       return 0;
     } else {
@@ -1210,13 +1200,13 @@ public:
 
   size_t parse_core(const char *s, size_t n, SemanticValues & /*vs*/,
                     Context &c, std::any &dt) const override {
-    auto &chldsv = c.push();
+    auto &chldvs = c.push();
     c.push_capture_scope();
     auto se = scope_exit([&]() {
       c.pop();
       c.pop_capture_scope();
     });
-    auto len = ope_->parse(s, n, chldsv, c, dt);
+    auto len = ope_->parse(s, n, chldvs, c, dt);
     if (success(len)) {
       c.set_error_pos(s);
       return static_cast<size_t>(-1);
@@ -1429,9 +1419,9 @@ public:
   size_t parse_core(const char *s, size_t n, SemanticValues & /*vs*/,
                     Context &c, std::any &dt) const override {
     const auto &rule = *ope_;
-    auto &chldsv = c.push();
+    auto &chldvs = c.push();
     auto se = scope_exit([&]() { c.pop(); });
-    return rule.parse(s, n, chldsv, c, dt);
+    return rule.parse(s, n, chldvs, c, dt);
   }
 
   void accept(Visitor &v) override;
@@ -2408,6 +2398,10 @@ public:
   std::string name;
   const char *s_ = nullptr;
 
+  std::function<bool(const SemanticValues &vs, const std::any &dt,
+                     std::string &msg)>
+      predicate;
+
   size_t id = 0;
   Action action;
   std::function<void(const char *s, size_t n, std::any &dt)> enter;
@@ -2695,6 +2689,15 @@ inline size_t Holder::parse_core(const char *s, size_t n, SemanticValues &vs,
     c.rule_stack.push_back(outer_);
     auto len = ope_->parse(s, n, vs, c, dt);
     c.rule_stack.pop_back();
+
+    std::string msg;
+    if (outer_->predicate && !outer_->predicate(vs, dt, msg)) {
+      if (c.log && !msg.empty() && c.error_info.message_pos < s) {
+        c.error_info.message_pos = s;
+        c.error_info.message = msg;
+      }
+      len = static_cast<size_t>(-1);
+    }
     return len;
   }
 
@@ -2709,34 +2712,31 @@ inline size_t Holder::parse_core(const char *s, size_t n, SemanticValues &vs,
       if (outer_->leave) { outer_->leave(s, n, len, a_val, dt); }
     });
 
-    auto &chldsv = c.push();
+    auto &chldvs = c.push();
 
     c.rule_stack.push_back(outer_);
-    len = ope_->parse(s, n, chldsv, c, dt);
+    len = ope_->parse(s, n, chldvs, c, dt);
     c.rule_stack.pop_back();
 
     // Invoke action
     if (success(len)) {
-      chldsv.sv_ = std::string_view(s, len);
+      chldvs.sv_ = std::string_view(s, len);
-      chldsv.name_ = outer_->name;
+      chldvs.name_ = outer_->name;
 
       if (!dynamic_cast<const peg::PrioritizedChoice *>(ope_.get())) {
-        chldsv.choice_count_ = 0;
+        chldvs.choice_count_ = 0;
-        chldsv.choice_ = 0;
+        chldvs.choice_ = 0;
       }
 
-      try {
+      std::string msg;
-        a_val = reduce(chldsv, dt);
+      if (outer_->predicate && !outer_->predicate(chldvs, dt, msg)) {
-      } catch (const parse_error &e) {
+        if (c.log && !msg.empty() && c.error_info.message_pos < s) {
-        if (c.log) {
-          if (e.what()) {
-            if (c.error_info.message_pos < s) {
           c.error_info.message_pos = s;
-              c.error_info.message = e.what();
+          c.error_info.message = msg;
-            }
-          }
         }
         len = static_cast<size_t>(-1);
+      } else {
+        a_val = reduce(chldvs, dt);
       }
     }
   });

test/test1.cc

@@ -226,14 +226,17 @@ TEST(GeneralTest, enter_leave_handlers_test) {
 
   auto message = "should be upper case string...";
 
-  parser["TOKEN"] = [&](const SemanticValues &vs, std::any &dt) {
+  parser["TOKEN"].predicate = [&](const SemanticValues &vs, const std::any &dt,
+                                  std::string &msg) {
     auto &require_upper_case = *std::any_cast<bool *>(dt);
     if (require_upper_case) {
       const auto &s = vs.sv();
       if (!std::all_of(s.begin(), s.end(), ::isupper)) {
-        throw parse_error(message);
+        msg = message;
+        return false;
       }
     }
+    return true;
   };
 
   bool require_upper_case = false;

test/test2.cc

@@ -837,9 +837,16 @@ TEST(PredicateTest, Semantic_predicate_test) {
   parser parser("NUMBER  <-  [0-9]+");
 
   parser["NUMBER"] = [](const SemanticValues &vs) {
-    auto val = vs.token_to_number<long>();
+    return vs.token_to_number<long>();
-    if (val != 100) { throw parse_error("value error!!"); }
+  };
-    return val;
+
+  parser["NUMBER"].predicate = [](const SemanticValues &vs,
+                                  const std::any & /*dt*/, std::string &msg) {
+    if (vs.token_to_number<long>() != 100) {
+      msg = "value error!!";
+      return false;
+    }
+    return true;
   };
 
   long val;