tv_rename/functions.cpp

#include <algorithm>
#include <array>
#include <cctype>
#include <iomanip>
#include <sstream>
#include <vector>

#ifdef _WIN32

#include <codecvt>
#include <iostream>
#include <shlobj.h>

#else // UNIX

#ifndef GUI

#include <iostream>
#include <stdlib.h>
#include <vector>

#else // UNIX and GUI

#include <pwd.h>
#include <unistd.h>

#endif // GUI

#endif // UNIX

#include "filesystem.hpp"
#include "functions.hpp"

#ifdef _WIN32

#define cout std::wcout
#define cerr std::wcerr
#define cin std::wcin

constexpr const char_t *dir_divider = L"\\";

#else // UNIX

#define TEXT( a ) a

#define cout std::cout
#define cerr std::cerr
#define cin std::cin

constexpr const char_t *dir_divider = "/";

#endif // UNIX

#ifndef GUI

constexpr std::array< const char_t *, 46 > languages{
    TEXT( "en" ), TEXT( "English" ),   TEXT( "sv" ), TEXT( "Svenska" ),
    TEXT( "no" ), TEXT( "Norsk" ),     TEXT( "da" ), TEXT( "Dansk" ),
    TEXT( "fi" ), TEXT( "Suomeksi" ),  TEXT( "nl" ), TEXT( "Nederlands" ),
    TEXT( "de" ), TEXT( "Deutsch" ),   TEXT( "it" ), TEXT( "Italiano" ),
    TEXT( "es" ), TEXT( "Español" ),   TEXT( "fr" ), TEXT( "Français" ),
    TEXT( "pl" ), TEXT( "Polski" ),    TEXT( "hu" ), TEXT( "Magyar" ),
    TEXT( "el" ), TEXT( "Greek" ),     TEXT( "tr" ), TEXT( "Turkish" ),
    TEXT( "ru" ), TEXT( "Russian" ),   TEXT( "he" ), TEXT( "Hebrew" ),
    TEXT( "ja" ), TEXT( "Japanese" ),  TEXT( "pt" ), TEXT( "Portuguese" ),
    TEXT( "zh" ), TEXT( "Chinese" ),   TEXT( "cs" ), TEXT( "Czech" ),
    TEXT( "sl" ), TEXT( "Slovenian" ), TEXT( "hr" ), TEXT( "Croatian" ),
    TEXT( "ko" ), TEXT( "Korea" )
};

#endif // not GUI

#ifdef _WIN32

// functions to convert between string and wstring

std::string wstring_to_utf8( const std::wstring &wstring ) {
    std::wstring_convert< std::codecvt_utf8_utf16< wchar_t > > wconv;
    return wconv.to_bytes( wstring );
}

std::wstring utf8_to_wstring( const std::string &utf8 ) {
    std::wstring_convert< std::codecvt_utf8_utf16< wchar_t > > wconv;
    return wconv.from_bytes( utf8 );
}

#endif // _WIN32

// encode url so it's valid even with UTF-8 characters
string encodeUrl( const string &url ) {
    // stolen from here -
    // https://stackoverflow.com/questions/154536/encode-decode-urls-in-c
#ifdef _WIN32
    std::wstringstream encoded;
    auto url_c = wstring_to_utf8( url );
#else
    std::stringstream encoded;
    const auto &url_c = url;
#endif
    encoded.fill( '0' );
    encoded << std::hex;
    for ( const auto &x : url_c ) {
        if ( isalnum( static_cast< unsigned char >( x ) ) || x == '-' ||
             x == '_' || x == '.' || x == '~' ) {
            encoded << x;
            continue;
        }
        encoded << std::uppercase << '%' << std::setw( 2 );
        encoded << int( static_cast< unsigned char >( x ) ) << std::nouppercase;
    }
    return encoded.str();
}

// return true if filename has specified season
// set ep_pos to position where episode number starts
bool searchSpecificSeason( const char_t *const path, size_t &ep_pos,
                           const string &number ) {
    size_t cur_pos{};
#ifdef _WIN32
    auto ncompare = wcsncmp;
#else
    auto ncompare = strncmp;
#endif
    // search for S[0-9]+E[0-9]+
    while ( path[cur_pos] != '\0' ) {
        if ( ( path[cur_pos] == 's' || path[cur_pos] == 'S' ) &&
             iswdigit( path[cur_pos + 1] ) ) {
            cur_pos++;
            while ( path[cur_pos] == '0' )
                cur_pos++;
            // if season number is 0, move back because previous while skipped it
            if( number == TEXT("0") )
                cur_pos--;
            // make sure season's number is the same as provided in argument
            // `number`
            if ( !ncompare( path + cur_pos, number.c_str(), number.size() ) ) {
                cur_pos += number.size();
                if ( ( path[cur_pos] == 'e' || path[cur_pos] == 'E' ) &&
                     iswdigit( path[cur_pos + 1] ) ) {
                    ep_pos = cur_pos + 1;
                    return true;
                }
            }
        }
        cur_pos++;
    }
    return false;
}

bool searchSpecificSeason( const char_t *const p, const string &number ) {
    size_t tmp;
    return searchSpecificSeason( p, tmp, number );
}

// return true if file contains S[0-9]+E[0-9]+ nad set
// season_pos to start of season number
bool searchSeason( const char_t *const path, size_t &season_pos ) {
    size_t cur_pos{};
    while ( path[cur_pos] != '\0' ) {
        if ( ( path[cur_pos] == 's' || path[cur_pos] == 'S' ) &&
             iswdigit( path[cur_pos + 1] ) ) {
            cur_pos++;
            season_pos = cur_pos; // after ++ because we want the first pos to
                                  // point to season's number
            while ( iswdigit( path[cur_pos] ) )
                cur_pos++;
            if ( ( path[cur_pos] == 'e' || path[cur_pos] == 'E' ) &&
                 iswdigit( path[cur_pos + 1] ) ) {
                return true;
            }
        }
        cur_pos++;
    }
    return false;
}

bool searchSeason( const char_t *const path ) {
    size_t tmp{};
    return searchSeason( path, tmp );
}

void iterateFS( std::map< int, std::set< string > > &seasons,
                const string &path ) {
    // season_pos - position of first digit of the season
    size_t season_pos{ string::npos };
    for ( const auto p : FSLib::Directory( path ) ) {
        // if p is directory, iterate through it
        if ( FSLib::isDirectory( path + dir_divider + p ) ) {
            iterateFS( seasons, path + dir_divider + p );
            continue;
        }

        // if file is a correct format, add it to file list
        // for its season
        if ( searchSeason( p, season_pos ) )
            seasons[std::stoi( p + season_pos )].insert(
                    path + dir_divider + p );
    }
}

#ifndef GUI
// following functions are only needed for CLI version

// find all files for provided season in `path` and store it in `files`
void findSeason( std::set< string > &files, int season, const string &path ) {
#ifdef _WIN32
    auto number = std::to_wstring( season );
#else
    auto number = std::to_string( season );
#endif

    for ( const auto p : FSLib::Directory( path ) ) {
        // if p is directory, iterate through it
        if ( FSLib::isDirectory( path + dir_divider + p ) ) {
            findSeason( files, season, path + dir_divider + p );
            continue;
        }

        if ( searchSpecificSeason( p, number ) )
            files.insert( path + dir_divider + p );
    }
}

// find all files that comply with the S[0-9]+E[0-9]+ pattern
// and their season is in season_numbers and store tem in `seasons`
void findSeasons( std::map< int, std::set< string > > &seasons,
                  const string &path, const std::set< int > &season_numbers ) {
    // season_pos - position of first digit of the season
    size_t season_pos{ string::npos };
    for ( const auto p : FSLib::Directory( path ) ) {
        // if p is directory, iterate through it
        if ( FSLib::isDirectory( path + dir_divider + p ) ) {
            findSeasons( seasons, path + dir_divider + p, season_numbers );
            continue;
        }

        if ( searchSeason( p, season_pos ) ) {
            auto num = std::stoi( p + season_pos );
            if ( season_numbers.find( num ) != season_numbers.end() )
                seasons[num].insert( path + dir_divider + p );
        }
    }
}

string getDefUrl( string &show, const string &language, Curl &c ) {
    string base_url = TEXT( "https://www.thetvdb.com" );

#ifdef _WIN32
    string source_code = utf8_to_wstring(
        c.execute( TEXT( "https://www.thetvdb.com/search?q=" ) +
                   encodeUrl( show ) + TEXT( "&l=" ) + language ) );
#else
    string source_code =
        c.execute( TEXT( "https://www.thetvdb.com/search?q=" ) +
                   encodeUrl( show ) + TEXT( "&l=" ) + language );
#endif

    size_t order{}, pos{};
    std::vector< std::pair< string, string > > urls;

    // find all possible shows
    while ( true ) {
        pos = source_code.find( TEXT( "/ser" ), pos );
        if ( pos != string::npos ) {
            auto end = source_code.find( TEXT( ">" ), pos );
            end--;
            auto end2 = source_code.find( TEXT( "<" ), end + 2 );
            // store shows in urls, first is name, second is url
            urls.emplace_back( source_code.substr( end + 2, end2 - end - 2 ),
                               source_code.substr( pos, end - pos ) );
            cout << ++order << ". " << urls.back().first << std::endl;
            pos = end2;
        } else {
            break;
        }
    }
    cout << "Which TV Show is the right one? ";
    cin >> pos;
    cin.clear();
    cin.ignore( 1, '\n' );
    show = urls[pos - 1].first;
    return base_url + urls[pos - 1].second;
}

void printHelp() {
    cout << "usage: tv_rename [--help] [--show show name] [--season season "
            "number]"
         << std::endl;
    cout
        << "                 [--correct-path] [--show-path show path] [--trust]"
        << std::endl;
    cout << "                 [--linux] [--lang language] [--print-langs]"
         << std::endl;
    cout << std::endl
         << "Rename TV episodes" << std::endl
         << std::endl
         << "optional arguments:" << std::endl;
    cout << "  -h, --help\t\tshow this help message and exit" << std::endl;
    cout << "  --show show name, -s show name" << std::endl;
    cout << "\t\t\tTV show from which you want episode names (needs to be"
         << std::endl;
    cout << "\t\t\tin quotation marks if it has more than one word)"
         << std::endl;
    cout << "  --season season number, -n season number" << std::endl;
    cout << "\t\t\tSeason number/s (if multiple seasons, put them in"
         << std::endl;
    cout << "\t\t\tquotation marks and seperate by one space)" << std::endl;
    cout << "\t\t\tor 'all' for all seasons in selected subdirectory"
         << std::endl;
    cout << "  --show-path show path, -p show path" << std::endl;
    cout << "\t\t\tPath of the directory with episodes" << std::endl;
    cout << "  --correct-path, -c\tThis is the correct path, stop asking me!"
         << std::endl;
    cout << "  --name-pattern pattern" << std::endl;
    cout << "\t\t\tPattern to which change the file name. Possible sequences "
            "are:"
         << std::endl;
    cout << "\t\t\t\t%filename - original filename (without filetype extension)"
         << std::endl;
    cout << "\t\t\t\t%show - show name from thetvdb" << std::endl;
    cout << "\t\t\t\t%epname - episode name from thetvdb" << std::endl;
    cout << "\t\t\t\t%season - season number" << std::endl;
    cout << "\t\t\t\ttpossible to specify leading 0 like this: %2season "
            "(number means how many leading zeros)"
         << std::endl;
    cout << "\t\t\t\t%episode - episode number" << std::endl;
    cout << "\t\t\t\t\tpossible to specify leading 0 like this: %2episode "
            "(number means how many leading zeros)"
         << std::endl;
    cout << "\t\t\tDefault pattern is \"$filename - $epname\"" << std::endl;
    cout << "  --trust, -t\t\tDon't ask whether the names are correct"
         << std::endl;
    cout << "  --linux, -x\t\tDon't replace characters characters that are "
            "illegal in Windows"
         << std::endl;
    cout << "  --lang language, -l language" << std::endl;
    cout << "\t\t\tSelect which language the episode names shoud be in"
         << std::endl;
    cout << "  --print-langs\t\tPring available language" << std::endl;
}

// parse command line argument --seasons (e.g. '1 2 3 4 5')
// and store season numbers as integers in seasons_num
void parseSeasonNumbers( std::set< int > &seasons_num,
                         const char_t *argument ) {
    size_t pos{ 0 };

    while ( !iswdigit( argument[pos] ) && argument[pos] != '\0' )
        pos++;

    if ( argument[pos] == '\0' ) {
        seasons_num.clear();
        return;
    }

    int temp;

#ifdef _WIN32
    std::wstringstream iss( argument + pos );
#else
    std::stringstream iss( argument + pos );
#endif

    while ( iss >> temp ) {
        seasons_num.insert( temp );
    }
}

// print possible language codes and their corresponding language
void printLangs() {
    for ( size_t i = 0; i < languages.size(); i += 2 ) {
        cout << languages[i] << " - " << languages[i + 1] << std::endl;
    }
}

// make sure language is a valide language code
bool findLanguage( const char_t *language ) {
    for ( size_t i = 0; i < languages.size(); i += 2 ) {
#ifdef _WIN32
        if ( !wcscmp( language, languages[i] ) )
#else
        if ( !strcmp( language, languages[i] ) )
#endif
            return true;
    }
    return false;
}

#else // GUI
// functions that are needed for GUI but not for CLI

// get possible shows for search query in `show`
std::vector< std::pair< string, string > >
getPossibleShows( string show, const string &language, Curl &c ) {
    // encode show name so it can be resolved as url
    show = encodeUrl( show );
#ifdef _WIN32
    auto source_code = utf8_to_wstring(
        c.execute( TEXT( "https://www.thetvdb.com/search?q=" ) +
                   show + TEXT( "&l=" ) + language ) );
#else
    auto source_code =
        c.execute( TEXT( "https://www.thetvdb.com/search?q=" ) +
                   show + TEXT( "&l=" ) + language );
#endif
    size_t pos{};
    std::vector< std::pair< string, string > > urls;
    while ( true ) {
        pos = source_code.find( TEXT( "/ser" ), pos );
        if ( pos != string::npos ) {
            auto end = source_code.find( TEXT( ">" ), pos );
            auto end2 = source_code.find( TEXT( "<" ), end + 1 );
            end--;
            urls.emplace_back(
                source_code.substr( end + 2, end2 - ( end + 2 ) ),
                source_code.substr( pos, end - pos ) );
            pos = end + 2;
        } else {
            break;
        }
    }
    return urls;
}

#ifndef _WIN32

// get user's home directory
string userHome() {
    uid_t user_uid; // current user's uid
    {
        uid_t eid;
        uid_t sid;
        getresuid( &user_uid, &eid, &sid ); // don't need eid and sid
    }
    // password file entry
    auto user_passwd = getpwuid( user_uid );

    if ( user_passwd == nullptr )
        throw std::runtime_error(
            "User with uid " + std::to_string( user_uid ) + " doesn't exist!" );
    return user_passwd->pw_dir;
}

#else // UNIX

// get user's %APPDATA% folder location
string userHome() {
    wchar_t *dir = static_cast< wchar_t * >( CoTaskMemAlloc( MAX_PATH ) );
    auto res = SHGetKnownFolderPath( FOLDERID_RoamingAppData, 0, NULL, &dir );
    if ( res == S_OK ) {
        string dir_s = dir;
        CoTaskMemFree( dir );
        return dir_s;
    }
    return L"";
}

#endif // UNIX

#endif // ndef GUI

// create file name based on given pattern
string compilePattern( const string &pattern, int season, int episode,
                       const string &filename, const string &episodeName,
                       const string &showName ) {
    string output;

#ifdef _WIN32
    auto season_num = std::to_wstring( season );
    auto ep_num = std::to_wstring( episode );
#else
    auto season_num = std::to_string( season );
    auto ep_num = std::to_string( episode );
#endif

    for ( size_t i = 0; i < pattern.size(); i++ ) {
        // if current character is % check if a pattern follows, otherwise
        // put %
        if ( pattern[i] == '%' ) {
            // check for numbers right after % indicating size of zero
            // padding for numbers
            auto pos = pattern.find_first_not_of( TEXT( "0123456789" ), i + 1 );

            if ( pattern.find( TEXT( "season" ), pos - 1 ) == pos &&
                 pos != i + 1 ) {
                // if season is AFTER numbers, put season number padded
                // with zeros

                // get number of leading zeros
                auto leading = std::stoi( pattern.c_str() + i + 1 );
                // move i to the last char of 'season'
                i = pos + 5;

                // get number of zeros to be put before the season number
                leading -= season_num.size();
                if ( leading < 0 )
                    leading = 0;

                // add padded season to output
                output += string( leading, '0' ) + season_num;
            } else if ( pattern.find( TEXT( "season" ), i ) == i + 1 ) {
                // if season isn't after numbers, just put season number to
                // output
                i += 6;
                output += season_num;
            } else if ( pattern.find( TEXT( "episode" ), pos - 1 ) == pos &&
                        pos != i + 1 ) {
                // same principle as with season after number
                auto leading = std::stoi( pattern.c_str() + i + 1 );

                i = pos + 6;

                leading -= ep_num.size();
                if ( leading < 0 )
                    leading = 0;

                output += string( leading, '0' ) + ep_num;
            } else if ( pattern.find( TEXT( "episode" ), i ) == i + 1 ) {
                // if episode isn't after number, just put the episode number to
                // output
                i += 7;
                output += ep_num;
            } else if ( pattern.find( TEXT( "epname" ), i ) == i + 1 ) {
                // episode name from thetvdb
                i += 6;
                output += episodeName;
            } else if ( pattern.find( TEXT( "show" ), i ) == i + 1 ) {
                // show name from thetvdb
                i += 4;
                output += showName;
            } else if ( pattern.find( TEXT( "filename" ), i ) == i + 1 ) {
                // original file name
                i += 8;
                output += filename;
            } else {
                // output % if no escape sequence was found
                output += '%';
            }
        } else if ( pattern[i] == '\\' ) {
            // possibility to escape %
            if ( pattern[i + 1] == '%' ) {
                output += '%';
                i++;
            } else if ( pattern[i + 1] == '\\' ) {
                output += '\\';
                i++;
            } else {
                output += '\\';
            }
        } else {
            // if char isn't % or / just add it to the output string
            output += pattern[i];
        }
    }

    return output;
}