aboutsummaryrefslogtreecommitdiff
path: root/src/Conf.cpp
diff options
context:
space:
mode:
authordec05eba <dec05eba@protonmail.com>2017-12-09 01:14:41 +0100
committerdec05eba <dec05eba@protonmail.com>2017-12-09 01:14:47 +0100
commitfb2072deb3e50afdb062570a3a80ec1afb5bfb56 (patch)
tree44baddd61a9aac3ead96ac8a6fab95d087d0bdd7 /src/Conf.cpp
parentbf24f6fc48b4eebb06cdcd7029d1d31d4c6028dd (diff)
Finished project config file parsing
Diffstat (limited to 'src/Conf.cpp')
-rw-r--r--src/Conf.cpp336
1 files changed, 330 insertions, 6 deletions
diff --git a/src/Conf.cpp b/src/Conf.cpp
index d535c60..cea1b01 100644
--- a/src/Conf.cpp
+++ b/src/Conf.cpp
@@ -1,18 +1,342 @@
#include "../include/Conf.hpp"
#include "../include/FileUtil.hpp"
+#include "../external/utf8/unchecked.h"
+#include <stdexcept>
using namespace std;
+using u8string = utf8::unchecked::iterator<char*>;
namespace sibs
{
- Result<string> readConf(const char *filepath)
+ u32 min(u32 a, u32 b) { return a < b ? a : b; }
+
+ class UnexpectedTokenException : public std::runtime_error
+ {
+ public:
+ UnexpectedTokenException(const string &errMsg) : runtime_error(errMsg)
+ {
+
+ }
+ };
+
+ enum class Token
+ {
+ NONE,
+ END_OF_FILE,
+ IDENTIFIER,
+ OPEN_BRACKET,
+ CLOSING_BRACKET,
+ EQUALS,
+ STRING
+ };
+
+ const char *getTokenName(Token token)
+ {
+ switch(token)
+ {
+ case Token::NONE: return "NONE";
+ case Token::END_OF_FILE: return "<EOF>";
+ case Token::IDENTIFIER: return "identifier";
+ case Token::OPEN_BRACKET: return "[";
+ case Token::CLOSING_BRACKET: return "]";
+ case Token::EQUALS: return "=";
+ case Token::STRING: return "string";
+ default: return "Unknown";
+ }
+ }
+
+ class Tokenizer
+ {
+ public:
+ Tokenizer(const char *_code) :
+ currentToken(Token::NONE),
+ code((char*)_code)
+ {
+
+ }
+
+ Token nextToken()
+ {
+ u32 c = *code;
+ while(isWhitespace(c))
+ {
+ ++code;
+ c = *code;
+ }
+
+ if(isAlpha(c) || c == '_')
+ {
+ char *startOfIdentifier = code.base();
+ ++code;
+ c = *code;
+ while(isAlpha(c) || c == '_' || isDigit(c))
+ {
+ ++code;
+ c = *code;
+ }
+ char *endOfIdentifier = code.base();
+
+ identifier = StringView(startOfIdentifier, endOfIdentifier - startOfIdentifier);
+ return Token::IDENTIFIER;
+ }
+ else if(c == '[')
+ {
+ ++code;
+ return Token::OPEN_BRACKET;
+ }
+ else if(c == ']')
+ {
+ ++code;
+ return Token::CLOSING_BRACKET;
+ }
+ else if(c == '=')
+ {
+ ++code;
+ return Token::EQUALS;
+ }
+ else if(c == '"')
+ {
+ u32 escapeCount = 0;
+
+ ++code;
+ char *startOfStr = code.base();
+ while(escapeCount > 0 || *code != '"')
+ {
+ c = *code;
+ if(c == '\0')
+ return Token::END_OF_FILE;
+ else if(c == '\\')
+ ++escapeCount;
+ else
+ escapeCount = min(0, escapeCount - 1);
+ ++code;
+ }
+
+ str = StringView(startOfStr, code.base() - startOfStr);
+ ++code;
+ return Token::STRING;
+ }
+ else if(c == '\0')
+ {
+ return Token::END_OF_FILE;
+ }
+ else
+ {
+ string errMsg = "Unexpected token: ";
+ errMsg += (char)c;
+ throw UnexpectedTokenException(errMsg);
+ }
+ }
+
+ StringView getIdentifier() const { return identifier; }
+ StringView getString() const { return str; }
+ private:
+ bool isWhitespace(u32 c)
+ {
+ return c == ' ' || c == '\t' || c == '\n' || c == '\r';
+ }
+
+ bool isAlpha(u32 c)
+ {
+ return (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z');
+ }
+
+ bool isDigit(u32 c)
+ {
+ return c >= '0' && c <= '9';
+ }
+ private:
+ Token currentToken;
+ u8string code;
+
+ union
+ {
+ StringView identifier;
+ StringView str;
+ };
+ };
+
+ class ParserException : public std::runtime_error
{
- Result<string> fileContentResult = getFileContent(filepath);
+ public:
+ ParserException(const string &errMsg) : runtime_error(errMsg)
+ {
+
+ }
+ };
+
+ class Parser
+ {
+ public:
+ static Result<bool> parse(const char *code, const ConfigCallback &callback)
+ {
+ try
+ {
+ Parser parser(code, (ConfigCallback*)&callback);
+ parser.parse();
+ return Result<bool>::Ok(true);
+ }
+ catch (const UnexpectedTokenException &e)
+ {
+ return Result<bool>::Err(e.what());
+ }
+ catch (const ParserException &e)
+ {
+ return Result<bool>::Err(e.what());
+ }
+ }
+ private:
+ Parser(const char *code, ConfigCallback *_callback) :
+ tokenizer(code),
+ callback(_callback),
+ objectDefined(false)
+ {
+
+ }
+
+ void parse()
+ {
+ while(true)
+ {
+ Token token = tokenizer.nextToken();
+ switch(token)
+ {
+ case Token::IDENTIFIER:
+ {
+ parseConfigField();
+ break;
+ }
+ case Token::OPEN_BRACKET:
+ {
+ parseConfigObject();
+ break;
+ }
+ case Token::END_OF_FILE:
+ {
+ callback->finished();
+ return;
+ }
+ default:
+ {
+ string errMsg = "Expected identifier or object, got: ";
+ errMsg += getTokenName(token);
+ throw ParserException(errMsg);
+ }
+ }
+ }
+ }
+
+ void parseConfigField()
+ {
+ StringView fieldName = tokenizer.getIdentifier();
+ if (!objectDefined)
+ {
+ string errMsg = "An object has to be the first element defined in a config file";
+ throw ParserException(errMsg);
+ }
+
+ Token token = tokenizer.nextToken();
+ if(token == Token::EQUALS)
+ {
+ parseConfigFieldRhs(fieldName);
+ }
+ else
+ {
+ string errMsg = "Expected '=' after identifier, got: ";
+ errMsg += getTokenName(token);
+ throw ParserException(errMsg);
+ }
+ }
+
+ void parseConfigFieldRhs(StringView fieldName)
+ {
+ Token token = tokenizer.nextToken();
+ if(token == Token::STRING)
+ {
+ callback->processField(fieldName, tokenizer.getString());
+ }
+ else if(token == Token::OPEN_BRACKET)
+ {
+ token = tokenizer.nextToken();
+ if(token == Token::STRING)
+ {
+ StringView str = tokenizer.getString();
+ token = tokenizer.nextToken();
+ if(token == Token::CLOSING_BRACKET)
+ {
+ vector<StringView> values;
+ values.push_back(str);
+ callback->processField(fieldName, values);
+ }
+ else
+ {
+ string errMsg = "Expected ']' to close value list, got: ";
+ errMsg += getTokenName(token);
+ throw ParserException(errMsg);
+ }
+ }
+ else
+ {
+ string errMsg = "Expected string value inside list in field definition, got: ";
+ errMsg += getTokenName(token);
+ throw ParserException(errMsg);
+ }
+ }
+ else
+ {
+ string errMsg = "Expected string on right-hand side of field '";
+ errMsg += string(fieldName.data, fieldName.size);
+ errMsg += "', got: ";
+ errMsg += getTokenName(token);
+ throw ParserException(errMsg);
+ }
+ }
+
+ void parseConfigObject()
+ {
+ Token token = tokenizer.nextToken();
+ if(token == Token::IDENTIFIER)
+ {
+ StringView objectName = tokenizer.getIdentifier();
+ token = tokenizer.nextToken();
+ if(token == Token::CLOSING_BRACKET)
+ {
+ objectDefined = true;
+ callback->processObject(objectName);
+ }
+ else
+ {
+ string errMsg = "Expected ']' after identifier to close object definition, got: ";
+ errMsg += getTokenName(token);
+ throw ParserException(errMsg);
+ }
+ }
+ else
+ {
+ string errMsg = "Expected identifier after '[', got: ";
+ errMsg += getTokenName(token);
+ throw ParserException(errMsg);
+ }
+ }
+ private:
+ Tokenizer tokenizer;
+ ConfigCallback *callback;
+ bool objectDefined;
+ };
+
+ Result<bool> Config::readFromFile(const char *filepath, const ConfigCallback &callback)
+ {
+ Result<StringView> fileContentResult = getFileContent(filepath);
if(fileContentResult.isErr())
- return fileContentResult;
+ return Result<bool>::Err(fileContentResult.getErrMsg());
+
+ const char *code = fileContentResult.unwrap().data;
+ if(!utf8::is_valid(code, code + fileContentResult.unwrap().size))
+ return Result<bool>::Err("File is not in valid utf8 format");
+
+ if(fileContentResult.unwrap().size >= 3 && utf8::is_bom(code))
+ code += 3;
- string configData;
- printf("file content:\n%s\n", fileContentResult.unwrap().c_str());
- return Result<string>::Ok(std::move(configData));
+ return Parser::parse(code, callback);
}
} \ No newline at end of file