拼写检查支持检查字符串中的内容

Author: CppCXY

CodeService/src/Spell/CodeSpellChecker.cpp

@@ -1,6 +1,7 @@
 #include "CodeService/Spell/CodeSpellChecker.h"
 #include "LuaParser/LuaTokenTypeDetail.h"
 #include "Util/format.h"
+#include "CodeService/Spell/TextParser.h"
 
 CodeSpellChecker::CodeSpellChecker()
 	: _symSpell(std::make_shared<SymSpell>(SymSpell::Strategy::LazyLoaded))
@@ -29,6 +30,10 @@ void CodeSpellChecker::Analysis(DiagnosisContext& ctx, const CustomDictionary& c
 		{
 			IdentifyAnalysis(ctx, token, customDict);
 		}
+		else if (token.TokenType == TK_STRING)
+		{
+			TextAnalysis(ctx, token, customDict);
+		}
 	}
 }
 
@@ -182,7 +187,62 @@ void CodeSpellChecker::IdentifyAnalysis(DiagnosisContext& ctx, LuaToken& token,
 			                       token.Range.StartOffset + word.Range.Start + word.Range.Count - 1
 			);
 			std::string originText(token.Text.substr(word.Range.Start, word.Range.Count));
-			ctx.PushDiagnosis(Util::format("Typo in identifier '{}'", originText), range, DiagnosisType::Spell, originText);
+			ctx.PushDiagnosis(Util::format("Typo in identifier '{}'", originText), range, DiagnosisType::Spell,
+			                  originText);
+		}
+	}
+}
+
+void CodeSpellChecker::TextAnalysis(DiagnosisContext& ctx, LuaToken& token, const CustomDictionary& customDict)
+{
+	std::shared_ptr<spell::TextParser> parser = std::make_shared<spell::TextParser>(token.Text);
+	parser->Parse();
+	auto& identifiers = parser->GetIdentifiers();
+	if (identifiers.empty())
+	{
+		return;
+	}
+
+	for (auto& identifier : identifiers)
+	{
+		auto& text = identifier.Item;
+
+		if (customDict.count(text) != 0)
+		{
+			continue;
+		}
+		std::shared_ptr<spell::IdentifyParser> identifierParser = nullptr;
+
+		auto it = _caches.find(text);
+		if (it != _caches.end())
+		{
+			identifierParser = it->second;
+		}
+		else
+		{
+			identifierParser = std::make_shared<spell::IdentifyParser>(text);
+			identifierParser->Parse();
+			_caches.insert({text, identifierParser});
+		}
+
+		auto& words = identifierParser->GetWords();
+		if (words.empty())
+		{
+			continue;
+		}
+
+		for (auto& word : words)
+		{
+			if (!word.Item.empty() && !_symSpell->IsCorrectWord(word.Item) && customDict.count(word.Item) == 0)
+			{
+				auto range = TextRange(token.Range.StartOffset + identifier.Range.Start + word.Range.Start,
+				                       token.Range.StartOffset + identifier.Range.Start + word.Range.Start + word.Range.
+				                       Count - 1
+				);
+				std::string originText(token.Text.substr(identifier.Range.Start + word.Range.Start, word.Range.Count));
+				ctx.PushDiagnosis(Util::format("Typo in string '{}'", originText), range, DiagnosisType::Spell,
+				                  originText);
+			}
 		}
 	}
 }

CodeService/src/Spell/IdentifyParser.cpp

@@ -168,6 +168,12 @@ int IdentifyParser::GetCurrentChar()
 
 void IdentifyParser::PushWords(WordRange range)
 {
+	// 因为这极大可能是缩写
+	if (range.Count <= 3)
+	{
+		return;
+	}
+
 	std::string_view wordView = _source.substr(range.Start, range.Count);
 	std::string word;
 	word.resize(wordView.size());

CodeService/src/Spell/TextParser.cpp

@@ -1,4 +1,70 @@
 #include "CodeService/Spell/TextParser.h"
+#include "CodeService/Spell/IdentifyParser.h"
 
 using namespace spell;
-// TODO implement later
+
+bool IsIdentifier(char ch)
+{
+	return ch > 0 && (std::isalnum(ch) || ch == '_');
+}
+
+TextParser::TextParser(std::string_view source)
+	: _source(source)
+{
+}
+
+void TextParser::Parse()
+{
+	enum class ParseState
+	{
+		Unknown,
+		Identify
+	} state = ParseState::Unknown;
+
+	std::size_t start = 0;
+	for (std::size_t i = 0; i != _source.size(); i++)
+	{
+		char ch = _source[i];
+		switch (state)
+		{
+		case ParseState::Unknown:
+			{
+				if (IsIdentifier(ch))
+				{
+					state = ParseState::Identify;
+					start = i;
+				}
+				break;
+			}
+		case ParseState::Identify:
+			{
+				if (!IsIdentifier(ch))
+				{
+					state = ParseState::Unknown;
+					PushIdentifier(WordRange(start, i - start));
+				}
+				break;
+			}
+		}
+	}
+	if (state == ParseState::Identify)
+	{
+		PushIdentifier(WordRange(start, _source.size() - start));
+	}
+}
+
+std::vector<Word>& TextParser::GetIdentifiers()
+{
+	return _identifiers;
+}
+
+void TextParser::PushIdentifier(spell::WordRange range)
+{
+	if (range.Count <= 3)
+	{
+		return;
+	}
+
+	std::string_view identifyView = _source.substr(range.Start, range.Count);
+	_identifiers.emplace_back(range, std::string(identifyView));
+}

include/CodeService/Spell/CodeSpellChecker.h

@@ -26,6 +26,9 @@ class CodeSpellChecker
 private:
 	void IdentifyAnalysis(DiagnosisContext& ctx, LuaToken& token, const CustomDictionary& customDict);
 
+	void TextAnalysis(DiagnosisContext& ctx, LuaToken& token, const CustomDictionary& customDict);
+
 	std::shared_ptr<SymSpell> _symSpell;
 	std::unordered_map<std::string, std::shared_ptr<spell::IdentifyParser>> _caches;
 };
+

include/CodeService/Spell/TextParser.h

@@ -8,19 +8,17 @@ namespace spell {
 class TextParser
 {
 public:
-	enum class TextType
-	{
-		Unknown,
-		End,
-	};
-
 	TextParser(std::string_view source);
 
 	void Parse();
 
-	std::vector<spell::Word>& GetWords();
+	std::vector<Word>& GetIdentifiers();
 
 private:
-	TextType Lex();
+
+
+	void PushIdentifier(spell::WordRange range);
+	std::string_view _source;
+	std::vector<Word> _identifiers;
 };
 }