netrunner/src/CFGFileParser.cpp

// ©2008-2017 despair <despair@netrunner.cc>
#include "CFGFileParser.h"
#include <climits>
#include "slre.h"

CFGFileParser::CFGFileParser(const char* filename){
	// Open the config file, get its size,
	// allocate the buffer, read it into
	// the buffer, close the file
	cfg_file = fopen(filename, "rb"); // On NT, opening in text mode translates \n into \r\n
	stat(filename, cfg_fileinfo);
	buffer = static_cast<char*>(tlsf_malloc(cfg_fileinfo->st_size & INT_MAX, sizeof(char) & INT_MAX));
	cfg = new BrowserConfiguration();
	bytesRead = fread(buffer, sizeof(char) & INT_MAX, cfg_fileinfo->st_size & INT_MAX, cfg_file);
    if (!bytesRead) {
        std::cout << "no config" << std::endl;
    }
	fclose(cfg_file);
}

CFGFileParser::~CFGFileParser(){
    // clean up!
    // This buffer has been written to and co-opted by
    // libc itself, but it stands that we initially allocated
    // it. So we'll free it ourselves.
    tlsf_free(buffer);
    // We've already copied the collected config directives
    // into a global class, we can safely delete the scratch
    // object, and its file info block.
    tlsf_free(cfg_fileinfo);
    delete cfg;
}

bool CFGFileParser::ParseText() {
	// Initial buffer for pass 1. tmp holds the config file (fmr. buffer)
	// token holds the actual token. pass1_length to keep track of memory,
	// reallocate as needed.
	char *tmp, *token; 
    size_t pass1_length = 0;
    struct slre regex; // final validation before adding to second-pass buffer
	// Second pass. All comments and .tags stripped out. Starts off at 512 bytes, increases if necessary.
    char* directives = static_cast<char*>(tlsf_malloc(512, sizeof(char)));
    char* tmp2;
	token = strtok_r(buffer, "\n", &tmp);
	while (tmp != nullptr) {
		if (token[0] == '#' || token[0] == '.' || token[0] == '\n'){ // Comment, Perl directive, or single <LF> found, skip over
			token = strtok_r(nullptr, "\n", &tmp);
        }
        else if (!isalpha(token[0])){ // The leading character is NOT a #, ., <LF>, or a letter. NB: Leading numbers are also invalid.
            printf("ERROR: Config file is invalid\n");
            printf("%s\n",token);
            return false;
        }
		else {
            if (!slre_compile(&regex, "([A-Z0-9_]*)(:)(\\[.*\\])")){
                printf("Internal error.\n");
                return false;
            }
            if (!slre_match(&regex, token, strlen(token), nullptr)){
                printf("Invalid directive: %s\n", token); // regex matches DIREC[_]T[I1]VE:[token] but NOT directive:[token] - the directive MUST be all caps
                return false;
            }
            // Config directive found, add to second-pass buffer
			// once we reach the 0.5KB mark, realloc exactly enough to keep going
			if (pass1_length >= 512){ 
				tlsf_realloc(directives, pass1_length+strlen(token)+2);
			}
			strcat(directives, token);
			strcat(directives, "\t");
			pass1_length = strlen(directives);
			token = strtok_r(nullptr, "\n", &tmp); // continue
		}
	}
	// First pass complete, second pass: break up into single directives and <key, value> pairs
    token = strtok_r(directives, "\t", &tmp);
    ntr::fast_string key, value;
	while (tmp != nullptr){
        key=strtok_r(token, ":", &tmp2);
        value=strtok_r(nullptr, "[]", &tmp2);
        // strip ending bracket
        cfg->Settings.insert({key,value});
        token = strtok_r(nullptr, "\t", &tmp);
    }
    #ifdef DEBUG
    std::cout << "Settings stringmap contains, in no particular order:\n";
    for ( auto it = cfg->Settings.begin(); it != cfg->Settings.end(); ++it ){
        std::cout << it->first << ":" << it->second << std::endl;
    }
    std::cout << std::endl;
    #endif
    // clean up, we're done parsing
    tlsf_free(directives);
    return true;
}

// Writes the fully parsed configuration to a BrowserConfiguration
// object supplied by the user.
void CFGFileParser::WriteConfig(BrowserConfiguration &config){
	config.Settings = cfg->Settings;
}