combine functions for statements part done

src/name_tree.cpp

@@ -0,0 +1,179 @@
+#include "name_tree.hpp"
+
+namespace names {
+
+// --- public
+
+bool NameTree::insert(const std::string &path, nodes::Statement &&statement) {
+  auto name_ids = slice_path_to_name_ids(path);
+
+  size_t node_id = add_node(get_root(), name_ids);
+
+  if (nodes_[node_id].get_statement().has_value()) {
+    return false;
+  }
+
+  nodes_[node_id].set_statement(std::move(statement));
+  return true;
+}
+
+std::optional<nodes::Statement *> NameTree::find(const std::string &path) {
+  auto name_ids = slice_path_to_existing_name_ids(path);
+
+  if (!name_ids.has_value()) {
+    return std::nullopt;
+  }
+
+  auto node_id = find_node(get_root(), name_ids.value());
+
+  if (node_id.has_value()) {
+    return nodes_[node_id.value()].get_statement();
+  }
+
+  return std::nullopt;
+}
+
+std::optional<const nodes::Statement *>
+NameTree::find(const std::string &path) const {
+  auto name_ids = slice_path_to_existing_name_ids(path);
+
+  if (!name_ids.has_value()) {
+    return std::nullopt;
+  }
+
+  auto node_id = find_node(get_root(), name_ids.value());
+
+  if (node_id.has_value()) {
+    return nodes_[node_id.value()].get_statement();
+  }
+
+  return std::nullopt;
+}
+
+// --- private
+
+size_t NameTree::add_node(size_t current_node,
+                          const std::vector<size_t> &name_ids,
+                          size_t current_name) {
+  if (current_name >= name_ids.size()) {
+    return current_node;
+  }
+
+  auto next_node = nodes_[current_node].find(name_ids[current_name]);
+
+  if (!next_node.has_value()) {
+    next_node = nodes_.size();
+    nodes_[current_node].insert(name_ids[current_name], next_node.value());
+    nodes_.emplace_back();
+  }
+
+  return add_node(next_node.value(), name_ids, current_name + 1);
+}
+
+std::optional<size_t> NameTree::find_node(size_t current_node,
+                                          const std::vector<size_t> &name_ids,
+                                          size_t current_name) const {
+  if (current_name >= name_ids.size()) {
+    return current_node;
+  }
+
+  auto next_node = nodes_[current_node].find(name_ids[current_name]);
+
+  if (!next_node.has_value()) {
+    return std::nullopt;
+  }
+
+  return find_node(next_node.value(), name_ids, current_name + 1);
+}
+
+std::vector<size_t> NameTree::slice_path_to_name_ids(const std::string &path) {
+  // check that path is / is not points only operator
+  bool points_only = true;
+  for (size_t i = 0; i < path.size(); ++i) {
+    if (path[i] != '.') {
+      points_only = true;
+      break;
+    }
+  }
+
+  if (points_only) {
+    return {add_name_id(path)};
+  }
+
+  std::vector<size_t> name_ids;
+
+  size_t last_name_start = 0;
+  for (size_t i = 0; i < path.size() + 1; ++i) {
+    if (path[i] == '.' || i == path.size()) { // name separator
+      name_ids.push_back(
+          add_name_id(path.substr(last_name_start, i - last_name_start)));
+      last_name_start = i + 1;
+    }
+  }
+
+  return name_ids;
+}
+
+std::optional<std::vector<size_t>>
+NameTree::slice_path_to_existing_name_ids(const std::string &path) const {
+  // check that path is / is not points only operator
+  bool points_only = true;
+  for (size_t i = 0; i < path.size(); ++i) {
+    if (path[i] != '.') {
+      points_only = true;
+      break;
+    }
+  }
+
+  if (points_only) {
+    auto maybe_id = find_name_id(path);
+
+    if (maybe_id.has_value()) {
+      return std::vector<size_t>{maybe_id.value()};
+    } else {
+      return std::nullopt;
+    }
+  }
+
+  std::vector<size_t> name_ids;
+
+  size_t last_name_start = 0;
+  for (size_t i = 0; i < path.size() + 1; ++i) {
+    if (path[i] == '.' || i == path.size()) { // name separator
+      auto maybe_id =
+          find_name_id(path.substr(last_name_start, i - last_name_start));
+
+      if (!maybe_id.has_value()) {
+        return std::nullopt;
+      }
+
+      name_ids.push_back(maybe_id.value());
+      last_name_start = i + 1;
+    }
+  }
+
+  return name_ids;
+}
+
+size_t NameTree::add_name_id(const std::string &name) {
+  auto name_iter = name_to_id_.find(name);
+
+  if (name_iter == name_to_id_.end()) {
+    size_t name_id = name_to_id_.size();
+    return name_to_id_[name] = name_id;
+  }
+
+  return name_iter->second;
+}
+
+std::optional<size_t> NameTree::find_name_id(const std::string &name) const {
+  auto name_iter = name_to_id_.find(name);
+
+  if (name_iter == name_to_id_.end()) {
+    return std::nullopt;
+  }
+
+  return name_iter->second;
+}
+
+} // namespace names

src/statement_nodes.cpp

@@ -1,42 +1,55 @@
 #include "statement_nodes.hpp"
 
+#include "error_handling.hpp"
+#include "utils.hpp"
+#include <algorithm>
+
 namespace nodes {
 
-// TODO: return error type instead of bool ??
-bool FunctionDefinition::combine(
-    FunctionDefinition &&other_function_definition) {
-  // should be same in all definitions of one function
+// TODO: function overloading, etc. ??
+bool FunctionDefinition::is_same_to(
+    const FunctionDefinition &other_function_definition) const {
   if (*name_.get() != *other_function_definition.name_.get()) {
     return false;
   }
+  return true;
+}
+
+// TODO: function overloading, etc. ??
+CombineResult
+FunctionDefinition::combine(FunctionDefinition &&other_function_definition) {
+  // should be same in all definitions of one function
+  if (*name_.get() != *other_function_definition.name_.get()) {
+    return CombineResult::DIFFERENT_NAME_ERROR;
+  }
   if (modifier_ != other_function_definition.modifier_) {
-    return false;
+    return CombineResult::DIFFERNENT_MODIFIER_ERROR;
   }
   if (are_annotations_same_to_names_ !=
       other_function_definition.are_annotations_same_to_names_) {
-    return false;
+    return CombineResult::ARGUMENTS_ERROR;
   }
 
   // only one definition should have constraints
   if (!constraints_.empty() &&
       !other_function_definition.constraints_.empty()) {
-    return false;
+    return CombineResult::MORE_THEN_ONE_CONSTRAINTS_ERROR;
   }
 
   // only one definition should have expression
   if (expression_.has_value() &&
       other_function_definition.expression_.has_value()) {
-    return false;
+    return CombineResult::MORE_THEN_ONE_DEFINITION_BODY_ERROR;
   }
 
   // only one definition should have documentation
   if (docs_.get_description().has_value() &&
       other_function_definition.docs_.get_description().has_value()) {
-    return false;
+    return CombineResult::MORE_THEN_ONE_DOCS_ERROR;
   }
   if (docs_.get_annotations_info_size() > 0 &&
       other_function_definition.docs_.get_annotations_info_size() > 0) {
-    return false;
+    return CombineResult::MORE_THEN_ONE_DOCS_ERROR;
   }
 
   // TODO: think about required checks
@@ -60,7 +73,7 @@ bool FunctionDefinition::combine(
                *other_function_definition.arguments_[i]
                     .get_annotation()
                     .value())) {
-        return false;
+        return CombineResult::ARGUMENTS_ERROR;
       }
 
       // check that all argument names are the same
@@ -73,14 +86,14 @@ bool FunctionDefinition::combine(
                     .get_name()
                     .value()
                     ->get())) {
-        return false;
+        return CombineResult::ARGUMENTS_ERROR;
       }
 
       // check that types are are in zero or one set of arguments (TODO: can be
       // same and in two sets ??)
       if (arguments_[i].get_type().has_value() &&
           other_function_definition.arguments_[i].get_type().has_value()) {
-        return false;
+        return CombineResult::ARGUMENTS_ERROR;
       }
 
       // argument modifiers should be same
@@ -88,21 +101,18 @@ bool FunctionDefinition::combine(
               other_function_definition.arguments_[i].get_before_modifier() ||
           arguments_[i].get_after_modifier() !=
               other_function_definition.arguments_[i].get_after_modifier()) {
-        return false;
+        return CombineResult::ARGUMENTS_ERROR;
       }
-
-      // TODO: other checks (annotations, ...)
-
     } else if (i < arguments_.size()) {
       // last arguments shouldn't be annotated (because they are not annotated
       // in other function definition statement)
       if (arguments_[i].get_annotation().has_value()) {
-        return false;
+        return CombineResult::ARGUMENTS_ERROR;
       }
 
       // last arguments can only be unnamed arguments (returned values)
       if (arguments_[i].get_name().has_value()) {
-        return false;
+        return CombineResult::ARGUMENTS_ERROR;
       }
     } else { // i < other_function_definition.size()
       // last arguments shouldn't be annotated (because they are not annotated
@@ -110,12 +120,12 @@ bool FunctionDefinition::combine(
       if (other_function_definition.arguments_[i]
               .get_annotation()
               .has_value()) {
-        return false;
+        return CombineResult::ARGUMENTS_ERROR;
       }
 
       // last arguments can only be unnamed arguments (returned values)
       if (other_function_definition.arguments_[i].get_name().has_value()) {
-        return false;
+        return CombineResult::ARGUMENTS_ERROR;
       }
     }
   }
@@ -126,7 +136,7 @@ bool FunctionDefinition::combine(
       other_function_definition.docs_.get_annotations_info_size() > 0) {
     if (docs_.get_annotations_info_size() > 0 ||
         docs_.get_description().has_value()) {
-      return false;
+      return CombineResult::MORE_THEN_ONE_DOCS_ERROR;
     }
 
     docs_ = std::move(other_function_definition.docs_);
@@ -140,28 +150,50 @@ bool FunctionDefinition::combine(
     expression_ = other_function_definition.expression_;
   }
 
-  // TODO: arguments: check and merge
+  for (size_t i = 0; i < other_function_definition.arguments_.size(); ++i) {
+    if (i < arguments_.size()) {
+      if (other_function_definition.arguments_[i].get_type().has_value()) {
+        if (arguments_[i].update_type_from(
+                other_function_definition.arguments_[i])) {
+          error_handling::handle_internal_error(
+              "Function arguments are not properly checked before merging "
+              "during combination",
+              this);
+        }
+      }
+    } else {
+      arguments_.push_back(std::move(other_function_definition.arguments_[i]));
+    }
+  }
 
-  return true;
+  return CombineResult::OK;
 }
 
-bool TypeDefinition::combine(TypeDefinition &&other_type_definition) {
-  // should be same in all definitions of one type
+bool TypeDefinition::is_same_to(
+    const TypeDefinition &other_type_definition) const {
   if (*name_.get() != *other_type_definition.name_.get()) {
     return false;
   }
+  return true;
+}
+
+CombineResult TypeDefinition::combine(TypeDefinition &&other_type_definition) {
+  // should be same in all definitions of one type
+  if (*name_.get() != *other_type_definition.name_.get()) {
+    return CombineResult::DIFFERENT_NAME_ERROR;
+  }
   if (is_on_heap_ != other_type_definition.is_on_heap_) {
-    return false;
+    return CombineResult::DIFFERNENT_MODIFIER_ERROR;
   }
 
   // only one definition should have documentation
   if (docs_.get_description().has_value() &&
       other_type_definition.docs_.get_description().has_value()) {
-    return false;
+    return CombineResult::MORE_THEN_ONE_DOCS_ERROR;
   }
   if (docs_.get_annotations_info_size() > 0 &&
       other_type_definition.docs_.get_annotations_info_size() > 0) {
-    return false;
+    return CombineResult::MORE_THEN_ONE_DOCS_ERROR;
   }
 
   // only one definition should define type and arguments (they should be
@@ -169,7 +201,7 @@ bool TypeDefinition::combine(TypeDefinition &&other_type_definition) {
   if ((other_type_definition.type_.has_value() ||
        !other_type_definition.arguments_.empty()) &&
       (type_.has_value() || !arguments_.empty())) {
-    return false;
+    return CombineResult::ARGUMENTS_ERROR;
   }
 
   // combine docs
@@ -178,7 +210,7 @@ bool TypeDefinition::combine(TypeDefinition &&other_type_definition) {
       other_type_definition.docs_.get_annotations_info_size() > 0) {
     if (docs_.get_annotations_info_size() > 0 ||
         docs_.get_description().has_value()) {
-      return false;
+      return CombineResult::MORE_THEN_ONE_DOCS_ERROR;
     }
 
     docs_ = std::move(other_type_definition.docs_);
@@ -192,38 +224,58 @@ bool TypeDefinition::combine(TypeDefinition &&other_type_definition) {
     type_ = std::move(other_type_definition.type_);
   }
 
+  // TODO: can be in incorrect state if error occure here
+  // combine methods ??
   methods_.reserve(methods_.size() + other_type_definition.methods_.size());
-
   for (size_t i = 0; i < other_type_definition.methods_.size(); ++i) {
-    methods_.push_back(std::move(other_type_definition.methods_[i]));
+    auto method_iter = methods_by_name_.find(
+        *other_type_definition.methods_[i].get_name()->get());
+
+    if (method_iter == methods_by_name_.end()) {
+      methods_by_name_[*other_type_definition.methods_[i].get_name()->get()] =
+          methods_.size();
+      methods_.push_back(std::move(other_type_definition.methods_[i]));
+    } else {
+      auto method_combine_result = methods_[method_iter->second].combine(
+          std::move(other_type_definition.methods_[i]));
+      if (method_combine_result != CombineResult::OK) {
+        return method_combine_result;
+      }
+    }
   }
 
-  // TODO: combine methods ??
+  return CombineResult::OK;
+}
 
+bool TypeclassDefinition::is_same_to(
+    const TypeclassDefinition &other_typeclass_definition) const {
+  if (*name_.get() != *other_typeclass_definition.name_.get()) {
+    return false;
+  }
   return true;
 }
 
-bool TypeclassDefinition::combine(
-    TypeclassDefinition &&other_typeclass_definition) {
+CombineResult
+TypeclassDefinition::combine(TypeclassDefinition &&other_typeclass_definition) {
   // should be same in all definitions of one typeclass
   if (*name_.get() != *other_typeclass_definition.name_.get()) {
-    return false;
+    return CombineResult::DIFFERENT_NAME_ERROR;
   }
 
   // only one definition should have documentation
   if (docs_.get_description().has_value() &&
       other_typeclass_definition.docs_.get_description().has_value()) {
-    return false;
+    return CombineResult::MORE_THEN_ONE_DOCS_ERROR;
   }
   if (docs_.get_annotations_info_size() > 0 &&
       other_typeclass_definition.docs_.get_annotations_info_size() > 0) {
-    return false;
+    return CombineResult::MORE_THEN_ONE_DOCS_ERROR;
   }
 
   // only one definition should define base typeclasses
   if (!base_typeclasses_.empty() &&
       !other_typeclass_definition.base_typeclasses_.empty()) {
-    return false;
+    return CombineResult::MORE_THEN_ONE_DEFINITION_BODY_ERROR;
   }
 
   // combine docs
@@ -232,7 +284,7 @@ bool TypeclassDefinition::combine(
       other_typeclass_definition.docs_.get_annotations_info_size() > 0) {
     if (docs_.get_annotations_info_size() > 0 ||
         docs_.get_description().has_value()) {
-      return false;
+      return CombineResult::MORE_THEN_ONE_DOCS_ERROR;
     }
 
     docs_ = std::move(other_typeclass_definition.docs_);
@@ -242,15 +294,89 @@ bool TypeclassDefinition::combine(
     base_typeclasses_ = std::move(other_typeclass_definition.base_typeclasses_);
   }
 
+  // TODO: can be in incorrect state if error occure here
+  // combine methods ??
   methods_.reserve(methods_.size() +
                    other_typeclass_definition.methods_.size());
   for (size_t i = 0; i < other_typeclass_definition.methods_.size(); ++i) {
-    methods_.push_back(std::move(other_typeclass_definition.methods_[i]));
+    auto method_iter = methods_by_name_.find(
+        *other_typeclass_definition.methods_[i].get_name()->get());
+
+    if (method_iter == methods_by_name_.end()) {
+      methods_by_name_
+          [*other_typeclass_definition.methods_[i].get_name()->get()] =
+              methods_.size();
+      methods_.push_back(std::move(other_typeclass_definition.methods_[i]));
+    } else {
+      auto method_combine_result = methods_[method_iter->second].combine(
+          std::move(other_typeclass_definition.methods_[i]));
+      if (method_combine_result != CombineResult::OK) {
+        return method_combine_result;
+      }
+    }
   }
 
-  // TODO: combine methods ??
+  return CombineResult::OK;
+}
 
-  return true;
+bool Statement::is_same_to(const Statement &other_statement) const {
+  if (expression_.index() != other_statement.expression_.index()) {
+    return false;
+  }
+
+  switch (expression_.index()) {
+  case 0: // Import
+    return false;
+  case 1: // TypeDefinition
+    return std::get<TypeDefinition>(expression_)
+        .is_same_to(
+            std::move(std::get<TypeDefinition>(other_statement.expression_)));
+  case 2: // FunctionDefinition
+    return std::get<FunctionDefinition>(expression_)
+        .is_same_to(std::move(
+            std::get<FunctionDefinition>(other_statement.expression_)));
+  case 3: // TypeclassDefinition
+    return std::get<TypeclassDefinition>(expression_)
+        .is_same_to(std::move(
+            std::get<TypeclassDefinition>(other_statement.expression_)));
+  case 4: // EmptyLines
+    return false;
+  default:
+    break;
+  }
+
+  error_handling::handle_general_error("Unreachable");
+  exit(1);
+}
+
+CombineResult Statement::combine(Statement &&other_statement) {
+  if (expression_.index() != other_statement.expression_.index()) {
+    return CombineResult::DIFFERENT_STATEMENT_TYPES;
+  }
+
+  switch (expression_.index()) {
+  case 0: // Import
+    return CombineResult::STATEMENTS_CANT_BE_COMBINED_ERROR;
+  case 1: // TypeDefinition
+    return std::get<TypeDefinition>(expression_)
+        .combine(
+            std::move(std::get<TypeDefinition>(other_statement.expression_)));
+  case 2: // FunctionDefinition
+    return std::get<FunctionDefinition>(expression_)
+        .combine(std::move(
+            std::get<FunctionDefinition>(other_statement.expression_)));
+  case 3: // TypeclassDefinition
+    return std::get<TypeclassDefinition>(expression_)
+        .combine(std::move(
+            std::get<TypeclassDefinition>(other_statement.expression_)));
+  case 4: // EmptyLines
+    return CombineResult::STATEMENTS_CANT_BE_COMBINED_ERROR;
+  default:
+    break;
+  }
+
+  error_handling::handle_general_error("Unreachable");
+  exit(1);
 }
 
 } // namespace nodes