refactor: separate parsing classes

This commit is contained in:
3lswear
2022-01-22 13:58:44 +03:00
parent 7e82ea0f86
commit 51a3050beb
12 changed files with 609 additions and 515 deletions

View File

@@ -2,6 +2,8 @@
#define TOMLNODE_HPP
#include "webserv.hpp"
#include "tomlstuff.hpp"
#include <map>
#include <vector>
#include <fstream>
@@ -48,8 +50,9 @@ class toml_node
void setObject(TOMLMap *obj);
void setMapArray(TOMLMapArray *map_array);
std::string *TOMLMap_to_string(TOMLMap *map) const;
std::string *toString(void) const;
static std::string *TOMLMap_to_string(TOMLMap *map);
};
#endif

43
includes/TOMLParser.hpp Normal file
View File

@@ -0,0 +1,43 @@
#ifndef TOMLPARSER_HPP
#define TOMLPARSER_HPP
#include "webserv.hpp"
#include "tomlstuff.hpp"
#include "Tokenizer.hpp"
#include "TOMLNode.hpp"
#include <string>
namespace config
{
class TOMLParser
{
private:
std::fstream file;
TOMLMap *root; //root of TOML tree
/* toml_node *current; //node currently being parsed */
Tokenizer tokenizer;
public:
TOMLParser(const std::string filename);
TOMLMap *parse(void);
toml_node *parseMap(void);
void parseMapArray(void);
toml_node *parseString(void);
toml_node *parseNumber(void);
toml_node *parseArray(void);
toml_node *parseBool(void);
toml_node *parseNil(void);
};
/* parse tha root ! */
/* TOMLMap *TOMLParser::parse(void); */
}
#endif

74
includes/Tokenizer.hpp Normal file
View File

@@ -0,0 +1,74 @@
#ifndef TOKENIZER_HPP
#define TOKENIZER_HPP
#include "webserv.hpp"
#include "tomlstuff.hpp"
#include <map>
#include <vector>
#include <fstream>
#include <cstdlib>
#include <iostream>
#include <exception>
namespace config
{
enum e_token
{
KEY,
NEWLINE,
ASSIGN,
STRING,
NUMBER,
COMMA,
BOOL,
NIL,
ARR_OPEN,
ARR_CLOSE,
MAP_OPEN,
MAP_CLOSE,
MAPARRAY_DECL
};
struct s_token
{
std::string value;
e_token type;
/* std::string to_string(void); */
};
bool isspace(char c);
bool istomlkey(char c);
class Tokenizer
{
private:
std::fstream file;
size_t prev_pos;
e_token last_token;
public:
Tokenizer(std::string filename);
char getWithoutWhiteSpace();
struct s_token getToken();
bool hasMoreTokens();
bool firstToken();
void rollBackToken();
};
/* struct s_token Tokenizer::getKey(void) */
/* { */
/* char c; */
/* struct s_token token; */
/* if (file.eof()) */
/* { */
/* std::cout << "Tokens exhausted" << std::endl; */
/* } */
/* } */
}
#endif

13
includes/parse.hpp Normal file
View File

@@ -0,0 +1,13 @@
#ifndef PARSE_HPP
#define PARSE_HPP
#include "tomlstuff.hpp"
#include "Tokenizer.hpp"
#include "TOMLNode.hpp"
#include "TOMLParser.hpp"
void parse(void);
void display(TOMLMap *config);
#endif

11
includes/tomlstuff.hpp Normal file
View File

@@ -0,0 +1,11 @@
#ifndef TOMLSTUFF_HPP
#define TOMLSTUFF_HPP
class toml_node;
typedef std::map<std::string, toml_node *> TOMLMap; // = JSONObject
typedef std::vector<TOMLMap *> TOMLMapArray;
typedef std::vector<toml_node *> TOMLArray;
#endif

View File

@@ -19,16 +19,4 @@
#include <map>
#include <vector>
void parse(void);
class toml_node;
/* typedef std::vector<TOMLMap *> TOMLArray; */
/* typedef std::vector<TOMLArray *> TOMLArrayOfMap; */
typedef std::map<std::string, toml_node *> TOMLMap; // = JSONObject
typedef std::vector<TOMLMap *> TOMLMapArray;
typedef std::vector<toml_node *> TOMLArray;
void display(TOMLMap *config);
#endif

View File

@@ -1,4 +1,3 @@
#include "webserv.hpp"
#include "TOMLNode.hpp"
/* toml_node::toml_node(void) */
@@ -143,7 +142,7 @@ std::string *toml_node::toString(void) const
}
}
std::string *TOMLMap_to_string(TOMLMap *map)
std::string *toml_node::TOMLMap_to_string(TOMLMap *map)
{
std::stringstream ss;
std::string *result = new std::string();

403
src/config/TOMLParser.cpp Normal file
View File

@@ -0,0 +1,403 @@
#include "TOMLParser.hpp"
/* -Template::-Template(void) */
/* { */
/* return; */
/* } */
/* -Template::-Template(const -Template &src) */
/* { */
/* *this = src; */
/* return; */
/* } */
/* -Template::~-Template(void) */
/* { */
/* return; */
/* } */
/* -Template &-Template::operator=(const -Template &rhs) */
/* { */
/* //code */
/* return (*this); */
/* } */
config::TOMLParser::TOMLParser(const std::string filename) : tokenizer(filename)
{};
toml_node *config::TOMLParser::parseMap(void)
{
std::cerr << "Parsing map" << std::endl;
toml_node *node = new toml_node;
TOMLMap *mapObject = new TOMLMap;
bool completed = false;
while (!completed)
{
if (tokenizer.hasMoreTokens())
{
s_token nextToken;
try { nextToken = tokenizer.getToken(); }
catch (std::logic_error e)
{
std::cerr << e.what() << std::endl;
break;
}
if (nextToken.type == MAPARRAY_DECL)
{
tokenizer.rollBackToken();
break;
}
std::string key = nextToken.value;
std::cerr << key << std::endl;
if (tokenizer.getToken().type != ASSIGN)
throw std::logic_error("EXPECTED assign!");
nextToken = tokenizer.getToken();
switch (nextToken.type)
{
case STRING:
{
tokenizer.rollBackToken();
(*mapObject)[key] = parseString();
break;
}
case ARR_OPEN:
{
(*mapObject)[key] = parseArray();
break;
}
case NUMBER:
{
tokenizer.rollBackToken();
(*mapObject)[key] = parseNumber();
break;
}
case BOOL:
{
tokenizer.rollBackToken();
(*mapObject)[key] = parseBool();
break;
}
case MAPARRAY_DECL:
{
std::cerr << "reached MAPARRAY_DECL in parseMap" << std::endl;
completed = true;
break;
}
default:
{
/* throw std::logic_error("jopa in parseMap"); */
std::cerr << "Unknown token, marking as complete" << std::endl;
completed = true;
break;
}
}
if (tokenizer.hasMoreTokens())
nextToken = tokenizer.getToken();
else
break;
if (nextToken.type != NEWLINE)
{
throw std::logic_error("EXPECTED newline");
}
}
else
{
throw std::logic_error("parseMap: no more tokens");
}
}
node->setObject(mapObject);
return (node);
}
void config::TOMLParser::parseMapArray(void)
{
std::cerr << "Parsing MapArray" << std::endl;
toml_node *map_node;
toml_node *maparr_node;
s_token current;
try { current = tokenizer.getToken(); }
catch (std::logic_error e)
{
std::cerr << e.what() << std::endl;
return;
}
if (current.type == MAPARRAY_DECL)
{
if (tokenizer.getToken().type != NEWLINE)
throw std::logic_error("no newline after map_array declaration");
map_node = parseMap();
}
else
throw std::logic_error("unexpected token in parseMapArray");
TOMLMap::iterator it;
std::cout << current.value << std::endl;
std::string name = current.value;
std::vector<std::string> full_name;
size_t dot;
while (1)
{
dot = name.find('.');
if (dot == std::string::npos)
break;
/* std::cout << dot << std::endl; */
full_name.push_back(name.substr(0, dot));
name.erase(0, dot + 1);
}
full_name.push_back(name);
for (size_t i = 0; i < full_name.size(); i++)
std::cout << full_name[i] << ", ";
std::cout << std::endl;
/* throw std::logic_error("tha end"); */
TOMLMap *local_root;
std::vector<std::string>::iterator subname = full_name.begin();
local_root = root;
while (1)
{
if (subname + 1 == full_name.end())
{
it = local_root->find(*subname);
if (it == local_root->end())
{
maparr_node = new toml_node;
TOMLMapArray *map_array = new TOMLMapArray;
map_array->push_back(map_node->getMap());
maparr_node->setMapArray(map_array);
(*local_root)[*subname] = maparr_node;
}
else
(it->second)->getMapArray()->push_back(map_node->getMap());
break;
}
else
{
it = local_root->find(*subname);
toml_node *map_node2;
map_node2 = new toml_node;
TOMLMap *map = new TOMLMap;
map_node2->setObject(map);
/* subname not found in local_root */
if (it == local_root->end())
{
(*local_root)[*subname] = map_node2;
local_root = map;
}
/* subname found in local_root */
else
{
local_root = *((it->second)->getMapArray()->end() - 1);
}
}
++subname;
}
}
toml_node *config::TOMLParser::parseString(void)
{
/* toml_node *node; */
toml_node *node = new toml_node;
std::string *sValue;
std::cerr << "Parsing string" << std::endl;
s_token token = tokenizer.getToken();
sValue = new std::string(token.value);
node->setString(sValue);
return (node);
}
toml_node *config::TOMLParser::parseNumber(void)
{
toml_node *node = new toml_node;
int value;
std::cerr << "Parsing number" << std::endl;
s_token token = tokenizer.getToken();
value = std::atoi(token.value.c_str());
node->setNumber(value);
return (node);
}
toml_node *config::TOMLParser::parseArray(void)
{
std::cerr << "Parsing array" << std::endl;
toml_node *node;
toml_node *result = new toml_node;
TOMLArray *array = new TOMLArray;
bool completed = false;
s_token current;
while (!completed)
{
if (!tokenizer.hasMoreTokens())
throw std::logic_error("No more tokens");
else
{
current = tokenizer.getToken();
switch (current.type)
{
case ARR_OPEN:
{
node = parseArray();
break;
}
case STRING:
{
tokenizer.rollBackToken();
node = parseString();
break;
}
case NUMBER:
{
tokenizer.rollBackToken();
node = parseNumber();
break;
}
case BOOL:
{
tokenizer.rollBackToken();
node = parseBool();
break;
}
case NIL:
{
node = parseNil();
break;
}
default:
{
throw std::logic_error("unkown token in parseArray");
}
}
array->push_back(node);
current = tokenizer.getToken();
if (current.type == COMMA)
continue;
else if (current.type == ARR_CLOSE)
completed = true;
else
throw std::invalid_argument("Unexpected token in array!");
}
}
result->setArr(array);
return (result);
}
toml_node *config::TOMLParser::parseBool(void)
{
toml_node *node = new toml_node;
bool value;
std::cerr << "Parsing bool" << std::endl;
s_token token = tokenizer.getToken();
if (token.value == "true")
value = true;
else if (token.value == "false")
value = false;
else
throw std::invalid_argument("Unexpected bool value");
node->setBool(value);
return (node);
}
toml_node *config::TOMLParser::parseNil(void)
{
toml_node *node = new toml_node;
std::cerr << "Parsing NIL" << std::endl;
node->setNil();
return (node);
}
/* parse tha root ! */
TOMLMap *config::TOMLParser::parse(void)
{
std::cerr << "Parsing ROOT" << std::endl;
root = new TOMLMap;
bool completed = false;
while (!completed)
{
if (tokenizer.hasMoreTokens())
{
s_token current;
try { current = tokenizer.getToken(); }
catch (std::logic_error e)
{
std::cerr << e.what() << std::endl;
break;
}
if (current.type == MAPARRAY_DECL)
{
/* parseMapArray(); */
tokenizer.rollBackToken();
parseMapArray();
}
else
{
std::string key = current.value;
std::cerr << key << std::endl;
if (tokenizer.getToken().type != ASSIGN)
throw std::logic_error("EXPECTED assign!");
current = tokenizer.getToken();
switch (current.type)
{
case STRING:
{
tokenizer.rollBackToken();
(*root)[key] = parseString();
break;
}
case ARR_OPEN:
{
(*root)[key] = parseArray();
break;
}
case NUMBER:
{
tokenizer.rollBackToken();
(*root)[key] = parseNumber();
break;
}
case BOOL:
{
tokenizer.rollBackToken();
(*root)[key] = parseBool();
break;
}
default:
{
/* throw std::logic_error("jopa in parseMap"); */
std::cerr << "Unknown token, marking as complete" << std::endl;
completed = true;
break;
}
}
if (tokenizer.hasMoreTokens())
current = tokenizer.getToken();
else
break;
if (current.type != NEWLINE)
{
throw std::logic_error("EXPECTED newline");
}
}
}
else
{
completed = true;
break;
}
}
return (root);
}

View File

@@ -1,400 +0,0 @@
#ifndef TOMLPARSER_HPP
#define TOMLPARSER_HPP
#include "TOMLNode.hpp"
#include "Tokenizer.hpp"
#include <string>
namespace config
{
class TOMLParser
{
private:
std::fstream file;
TOMLMap *root; //root of TOML tree
/* toml_node *current; //node currently being parsed */
Tokenizer tokenizer;
public:
TOMLParser(const std::string filename) : tokenizer(filename) {}
TOMLMap *parse(void);
toml_node *parseMap(void)
{
std::cerr << "Parsing object" << std::endl;
toml_node *node = new toml_node;
TOMLMap *mapObject = new TOMLMap;
bool completed = false;
while (!completed)
{
if (tokenizer.hasMoreTokens())
{
s_token nextToken;
try { nextToken = tokenizer.getToken(); }
catch (std::logic_error e)
{
std::cerr << e.what() << std::endl;
break;
}
if (nextToken.type == MAPARRAY_DECL)
{
tokenizer.rollBackToken();
break;
}
std::string key = nextToken.value;
std::cerr << key << std::endl;
if (tokenizer.getToken().type != ASSIGN)
throw std::logic_error("EXPECTED assign!");
nextToken = tokenizer.getToken();
switch (nextToken.type)
{
case STRING:
{
tokenizer.rollBackToken();
(*mapObject)[key] = parseString();
break;
}
case ARR_OPEN:
{
(*mapObject)[key] = parseArray();
break;
}
case NUMBER:
{
tokenizer.rollBackToken();
(*mapObject)[key] = parseNumber();
break;
}
case BOOL:
{
tokenizer.rollBackToken();
(*mapObject)[key] = parseBool();
break;
}
case MAPARRAY_DECL:
{
std::cerr << "reached MAPARRAY_DECL in parseMap" << std::endl;
completed = true;
break;
}
default:
{
/* throw std::logic_error("jopa in parseMap"); */
std::cerr << "Unknown token, marking as complete" << std::endl;
completed = true;
break;
}
}
if (tokenizer.hasMoreTokens())
nextToken = tokenizer.getToken();
else
break;
if (nextToken.type != NEWLINE)
{
throw std::logic_error("EXPECTED newline");
}
}
else
{
throw std::logic_error("parseMap: no more tokens");
}
}
node->setObject(mapObject);
return (node);
}
void parseMapArray(void)
{
std::cerr << "Parsing MapArray" << std::endl;
toml_node *map_node;
toml_node *maparr_node;
s_token current;
try { current = tokenizer.getToken(); }
catch (std::logic_error e)
{
std::cerr << e.what() << std::endl;
return;
}
if (current.type == MAPARRAY_DECL)
{
if (tokenizer.getToken().type != NEWLINE)
throw std::logic_error("no newline after map_array declaration");
map_node = parseMap();
}
else
throw std::logic_error("unexpected token in parseMapArray");
TOMLMap::iterator it;
std::cout << current.value << std::endl;
std::string name = current.value;
std::vector<std::string> full_name;
size_t dot;
while (1)
{
dot = name.find('.');
if (dot == std::string::npos)
break;
/* std::cout << dot << std::endl; */
full_name.push_back(name.substr(0, dot));
name.erase(0, dot + 1);
}
full_name.push_back(name);
for (size_t i = 0; i < full_name.size(); i++)
std::cout << full_name[i] << ", ";
std::cout << std::endl;
/* throw std::logic_error("tha end"); */
TOMLMap *local_root;
std::vector<std::string>::iterator subname = full_name.begin();
local_root = root;
while (1)
{
if (subname + 1 == full_name.end())
{
it = local_root->find(*subname);
if (it == local_root->end())
{
maparr_node = new toml_node;
TOMLMapArray *map_array = new TOMLMapArray;
map_array->push_back(map_node->getMap());
maparr_node->setMapArray(map_array);
(*local_root)[*subname] = maparr_node;
}
else
(it->second)->getMapArray()->push_back(map_node->getMap());
break;
}
else
{
it = local_root->find(*subname);
toml_node *map_node2;
map_node2 = new toml_node;
TOMLMap *map = new TOMLMap;
map_node2->setObject(map);
/* subname not found in local_root */
if (it == local_root->end())
{
(*local_root)[*subname] = map_node2;
local_root = map;
}
/* subname found in local_root */
else
{
local_root = *((it->second)->getMapArray()->end() - 1);
}
}
++subname;
}
}
toml_node *parseString(void)
{
/* toml_node *node; */
toml_node *node = new toml_node;
std::string *sValue;
std::cerr << "Parsing string" << std::endl;
s_token token = tokenizer.getToken();
sValue = new std::string(token.value);
node->setString(sValue);
return (node);
}
toml_node *parseNumber(void)
{
toml_node *node = new toml_node;
int value;
std::cerr << "Parsing number" << std::endl;
s_token token = tokenizer.getToken();
value = std::atoi(token.value.c_str());
node->setNumber(value);
return (node);
}
toml_node *parseArray(void)
{
std::cerr << "Parsing array" << std::endl;
toml_node *node;
toml_node *result = new toml_node;
TOMLArray *array = new TOMLArray;
bool completed = false;
s_token current;
while (!completed)
{
if (!tokenizer.hasMoreTokens())
throw std::logic_error("No more tokens");
else
{
current = tokenizer.getToken();
switch (current.type)
{
case ARR_OPEN:
{
node = parseArray();
break;
}
case STRING:
{
tokenizer.rollBackToken();
node = parseString();
break;
}
case NUMBER:
{
tokenizer.rollBackToken();
node = parseNumber();
break;
}
case BOOL:
{
tokenizer.rollBackToken();
node = parseBool();
break;
}
case NIL:
{
node = parseNil();
break;
}
default:
{
throw std::logic_error("unkown token in parseArray");
}
}
array->push_back(node);
current = tokenizer.getToken();
if (current.type == COMMA)
continue;
else if (current.type == ARR_CLOSE)
completed = true;
else
throw std::invalid_argument("Unexpected token in array!");
}
}
result->setArr(array);
return (result);
}
toml_node *parseBool(void)
{
toml_node *node = new toml_node;
bool value;
std::cerr << "Parsing bool" << std::endl;
s_token token = tokenizer.getToken();
if (token.value == "true")
value = true;
else if (token.value == "false")
value = false;
else
throw std::invalid_argument("Unexpected bool value");
node->setBool(value);
return (node);
}
toml_node *parseNil(void)
{
toml_node *node = new toml_node;
std::cerr << "Parsing NIL" << std::endl;
node->setNil();
return (node);
}
};
/* parse tha root ! */
TOMLMap *TOMLParser::parse(void)
{
std::cerr << "Parsing ROOT" << std::endl;
root = new TOMLMap;
bool completed = false;
while (!completed)
{
if (tokenizer.hasMoreTokens())
{
s_token current;
try { current = tokenizer.getToken(); }
catch (std::logic_error e)
{
std::cerr << e.what() << std::endl;
break;
}
if (current.type == MAPARRAY_DECL)
{
/* parseMapArray(); */
tokenizer.rollBackToken();
parseMapArray();
}
else
{
std::string key = current.value;
std::cerr << key << std::endl;
if (tokenizer.getToken().type != ASSIGN)
throw std::logic_error("EXPECTED assign!");
current = tokenizer.getToken();
switch (current.type)
{
case STRING:
{
tokenizer.rollBackToken();
(*root)[key] = parseString();
break;
}
case ARR_OPEN:
{
(*root)[key] = parseArray();
break;
}
case NUMBER:
{
tokenizer.rollBackToken();
(*root)[key] = parseNumber();
break;
}
case BOOL:
{
tokenizer.rollBackToken();
(*root)[key] = parseBool();
break;
}
default:
{
/* throw std::logic_error("jopa in parseMap"); */
std::cerr << "Unknown token, marking as complete" << std::endl;
completed = true;
break;
}
}
if (tokenizer.hasMoreTokens())
current = tokenizer.getToken();
else
break;
if (current.type != NEWLINE)
{
throw std::logic_error("EXPECTED newline");
}
}
}
else
{
completed = true;
break;
}
}
return (root);
}
}
#endif

View File

@@ -1,40 +1,28 @@
#ifndef TOKENIZER_HPP
#define TOKENIZER_HPP
#include "Tokenizer.hpp"
#include "webserv.hpp"
#include <map>
#include <vector>
#include <fstream>
#include <cstdlib>
#include <iostream>
#include <exception>
/* -Template::-Template(void) */
/* { */
/* return; */
/* } */
/* -Template::-Template(const -Template &src) */
/* { */
/* *this = src; */
/* return; */
/* } */
/* -Template::~-Template(void) */
/* { */
/* return; */
/* } */
/* -Template &-Template::operator=(const -Template &rhs) */
/* { */
/* //code */
/* return (*this); */
/* } */
namespace config
{
enum e_token
{
KEY,
NEWLINE,
ASSIGN,
STRING,
NUMBER,
COMMA,
BOOL,
NIL,
ARR_OPEN,
ARR_CLOSE,
MAP_OPEN,
MAP_CLOSE,
MAPARRAY_DECL
};
struct s_token
{
std::string value;
e_token type;
/* std::string to_string(void); */
};
bool isspace(char c)
{
if (c == ' ' || c == '\t')
@@ -50,74 +38,23 @@ namespace config
else
return (false);
}
class Tokenizer
Tokenizer::Tokenizer(std::string filename)
{
private:
std::fstream file;
size_t prev_pos;
e_token last_token;
public:
Tokenizer(std::string filename)
{
file.open(filename.c_str(), std::ios::in);
if (!file.good())
{
std::cerr << "file didn't open" << std::endl;
}
}
char getWithoutWhiteSpace();
struct s_token getToken();
bool hasMoreTokens();
bool firstToken()
{
// doesn't account for indent!
if (file.tellg() == 0 || file.tellg() == 1 || (last_token == NEWLINE))
return (true);
else
return (false);
}
void rollBackToken();
};
char Tokenizer::getWithoutWhiteSpace(void)
{
char c = ' ';
while (config::isspace(c))
file.open(filename.c_str(), std::ios::in);
if (!file.good())
{
file.get(c);
if ((c == ' ') && !file.good())
{
throw std::logic_error("No more tokens!");
}
else if (!file.good())
return (c);
std::cerr << "file didn't open" << std::endl;
}
return (c);
}
bool Tokenizer::hasMoreTokens(void)
bool Tokenizer::firstToken()
{
return (!file.eof());
// doesn't account for indent!
if (file.tellg() == 0 || file.tellg() == 1 || (last_token == NEWLINE))
return (true);
else
return (false);
}
void Tokenizer::rollBackToken(void)
{
if (file.eof())
file.clear();
file.seekg(prev_pos);
}
/* struct s_token Tokenizer::getKey(void) */
/* { */
/* char c; */
/* struct s_token token; */
/* if (file.eof()) */
/* { */
/* std::cout << "Tokens exhausted" << std::endl; */
/* } */
/* } */
struct s_token Tokenizer::getToken(void)
{
char c;
@@ -240,6 +177,31 @@ namespace config
last_token = token.type;
return (token);
}
}
char Tokenizer::getWithoutWhiteSpace(void)
{
char c = ' ';
while (config::isspace(c))
{
file.get(c);
if ((c == ' ') && !file.good())
{
throw std::logic_error("No more tokens!");
}
else if (!file.good())
return (c);
}
return (c);
}
#endif
bool Tokenizer::hasMoreTokens(void)
{
return (!file.eof());
}
void Tokenizer::rollBackToken(void)
{
if (file.eof())
file.clear();
file.seekg(prev_pos);
}
}

View File

@@ -6,10 +6,7 @@
#include <iostream>
#include <exception>
#include "TOMLNode.hpp"
#include "TOMLParser.hpp"
#include "Tokenizer.hpp"
#include "parse.hpp"
/* struct location */
/* { */
/* std::string location; */

View File

@@ -1,4 +1,5 @@
#include "webserv.hpp"
#include "parse.hpp"
int main(int argc, char **argv)
{