Grabber time weather updates files (tvpogoda.com)

getData4Statistic.php

#!/usr/bin/php -q
<?php
#...//.../*...*/
/* develop: Alexey Kovalev
    e-mail: [email protected] */

# --- GLOBAL VARIABLES ---------------------------------------------------------
# string error prefix & message
$str_errorPrx = "[ php.error ] ";
$str_errorMsg = "";

# --- GET HTML CODE ------------------------------------------------------------
# path parser webPage
$str_sourceNamePage = "http://tvpogoda.com/fobmail/Mos24/";

# get html-code my web-site
$str_sourceHtmlPage = file_get_contents( $str_sourceNamePage );

#                                                                                     WRITE *LOG

# check return value
if ( $str_sourceHtmlPage == FALSE ) {
    # generate error

    $str_errorMsg = "#01
Return value is not string.
'file_get_contents( $str_sourceNamePage )'
";
    # exit script             
    exit( $str_errorPrx . $str_errorMsg );
}

# --- FIND DATA ----------------------------------------------------------------
# pattern ( regular expression ) for find "value"
$str_patternFileName = "/>[\w\s]*.(txt|xml|doc)</";

$str_date = "[0-3][0-9]-[A-Z][a-z]{2,}-20[1-9][0-9]";
$str_time = "[0-2][0-9]:[0-5][0-9]\s*";
$str_patternDatetime = "/>" . $str_date . "\s" . $str_time . "</";
# "/>[0-3][0-9]-[A-Z][a-z]{2}-20[1-9][0-9]\s[0-2][0-9]:[0-5][0-9]\s*</";

# array replace pair string ">" & "<" - null
$arr_badSymbolReplacePair = array( ">" => "", "<" => "" );

# output array
$arr_fileName = array();
$arr_dateTime = array();

# get all element $arr_fileName
$arr_fileName = getValueFromStringUsePattern( $str_sourceHtmlPage, 
                                              $str_patternFileName, 
                                              $arr_badSymbolReplacePair );
# get all element $arr_dateTime
$arr_dateTime = getValueFromStringUsePattern( $str_sourceHtmlPage, 
                                              $str_patternDatetime, 
                                              $arr_badSymbolReplacePair );

# --- DATETIME -----------------------------------------------------------------
# format date
$str_formatDatetime = "Y-m-d H:i:s O";

foreach( $arr_dateTime as &$str_dateTime )
{
    # create object type DateTime
    $dt_value = date_create( $str_dateTime .":00", timezone_open('UTC'));
#                                                                                     WRITE *LOG
    # check return value
    if( $dt_value == FALSE ) {
        # generate error. [ log ]

        $str_errorMsg = "#03
Return value is not datetime ('$str_dateTime' => '$dt_value').
date_create( $str_dateTime .':00', timezone_open('UTC'))'
";
        # add error to log             
        # log( $str_errorPrx . $str_errorMsg );

        continue;
    }
    # get datetime(string)
    $str_dateTime = date_format( $dt_value, $str_formatDatetime );
#                                                                                     WRITE *LOG
    # check return value
    if( $str_dateTime == FALSE ) {
        # generate error. [ log ]

        $str_errorMsg = "#04
Return value is not srting ('$dt_value' => '$str_dateTime')
date_format( $dt_value, $str_formatDatetime )'
";
        # add error to log             
        # log( $str_errorPrx . $str_errorMsg );

        continue;
    }
}

# current datetime
$str_datetimeNow = date( $str_formatDatetime );

# --- STRUCTURING DATA ---------------------------------------------------------
#output array
$asArr_data = array();

# create array ('fileName': 'dateTime')
for( $i = 0; $i < count( $arr_fileName ); $i++) {
     $asArr_data[ $arr_fileName[$i] ] = 
        array( $arr_dateTime[$i], $str_datetimeNow ); }

#                                                                                     WRITE *LOG

# check count element (array)
if ( count( $arr_fileName ) != count( $arr_dateTime )) {
    # generate error
    
    $str_listValue = getStringFromAssociativeArray( $asArr_data,
                                                    "Name",
                                                    "Date" );
    $str_errorMsg = "#02
Discrepancy count element in array.\n'\n$str_listValue'
";
    # exit script             
    exit( $str_errorPrx . $str_errorMsg );
}

/*# print result
echo getStringFromAssociativeArray( $asArr_data,
                                    "Name",
                                    "Date",
                                     30 );*/
# print result
echo getStringFromAssociativeArrayWhereValueArray( $asArr_data,
                                                   "Name",
                                                   "Date",
                                                    30 );
# print current datetime
echo "\n " .str_pad( "Datetime.now ", 30, ".", STR_PAD_RIGHT).
       " " .$str_datetimeNow;

# --- EXIST DATA (JSON) --------------------------------------------------------
# set current date +.json
$str_currentDate_json = substr( $str_datetimeNow, 0, 10 ) . ".json";

# root folder 4 weather data
$str_pathRootWeatherFolder = dirName(__FILE__) . "/../weather/";

# full path 2 file
$str_fullPath2FileCurrentDate_json = $str_pathRootWeatherFolder .
                                     $str_currentDate_json;

# check exist folder 'weather'?
if( !is_dir( $str_pathRootWeatherFolder ) ){
      mkdir( $str_pathRootWeatherFolder );

#                                                                                     WRITE *LOG
}

#find current file (date)
if( !file_exists( $str_fullPath2FileCurrentDate_json ) ){
    $fl =  fopen( $str_fullPath2FileCurrentDate_json, "w" );
    fclose( $fl );
#                                                                                     WRITE *LOG
}

# --- READ DATA (JSON) ---------------------------------------------------------
# open file 2 read
$f_currentDate_json = fopen( $str_fullPath2FileCurrentDate_json, "r" );

# content file (currentDate.json)
$str_contentCurrentDate_json = null;

# check exist file
if ( $f_currentDate_json ) {
    # get filesize
    $i_fileSize = filesize ( $str_fullPath2FileCurrentDate_json );

    if( $i_fileSize != 0) {
        # get string file
        $str_contentCurrentDate_json = fread( $f_currentDate_json, $i_fileSize);
    }
}

# close file
fclose( $f_currentDate_json);   

# print result
#echo "\n\n" . $str_contentCurrentDate_json . "\n";

# --- COMPARE & CHANGE DATA (JSON) ---------------------------------------------
# get assiciative array from string
$asArr_currentDate_json = json_decode( $str_contentCurrentDate_json, true );

# exist element
$bl_existElementInArray = FALSE;

# add/create new data
foreach ( $asArr_data as $str_fileName => $arr_date ) {
    # add new datetime
    if( isset( $asArr_currentDate_json[ $str_fileName ] )) {
        foreach ( $asArr_currentDate_json[ $str_fileName ] as $arr_value ) {
            if( $arr_value[0] == $arr_date[0] ) {
                $bl_existElementInArray = TRUE;
                break;
            }
        }
        if ( !$bl_existElementInArray ) {
            # add element in start array
             array_unshift( $asArr_currentDate_json[ $str_fileName ],
                            $arr_date );
        }

        # return default flag
        $bl_existElementInArray = FALSE;
    }
    # create new datetime
    else { $asArr_currentDate_json[ $str_fileName ] = array( $arr_date ); }
}

# print result
#print_r( $asArr_currentDate_json );

# --- SAVE DATA (JSON) ---------------------------------------------------------
# encode asArr2json
$json_currentDate_json = json_encode( $asArr_currentDate_json );

# set pretty print (json)
$json_currentDate_json = prettyPrint( $json_currentDate_json );

# print result
#echo "\n\n". $json_currentDate_json ."\n\n";

# open file 2 write
$f_currentDate_json = fopen( $str_fullPath2FileCurrentDate_json, "w" );

# write data
fwrite( $f_currentDate_json, $json_currentDate_json);

# close file
fclose( $f_currentDate_json);

# ------------------------------------------------------------------------------
# ==============================================================================
# ------------------------------------------------------------------------------

# --- FUNCTION -----------------------------------------------------------------
function getValueFromStringUsePattern( $str_string, 
                                       $str_pattern,
                                       $arr_replacePair )
{
    # array from output values
    $arr_values = array();
    $arr_result = array();
    $str_value  = null;

    # get all string equal pattern
    preg_match_all( $str_pattern, $str_string, $arr_values, PREG_SET_ORDER );

    # replace all bad symbol
    foreach( $arr_values as $arr_value )
    {
        # delete "space"...
        $str_value =  trim( $arr_value[0] );
        # delete bad symbol
        $str_value = strtr( $str_value, $arr_replacePair );
        # delete "space"...
        $str_value =  trim( $str_value );

        # push in array - $str_value
        $arr_result[] = $str_value;
    }

    # return result array
    return $arr_result;
}

function getStringFromAssociativeArray( $asArr_associative,
                                        $str_key,
                                        $str_value,
                                        $i_lenghtString )
{
    # top
    $str_joinValueArray = 
        str_pad( "[$str_key]",   ($i_lenghtString +1), " ", STR_PAD_RIGHT) .
                 "[$str_value]" ."\n";
        #str_pad( "[$str_value]", (25 +1), " ", STR_PAD_LEFT)  ."\n";

    # body (pair value: 'name' => 'date')
    foreach( $asArr_associative as $key => $value ) {
        $str_joinValueArray .= " ". 
            str_pad( $key . " ", $i_lenghtString, ".", STR_PAD_RIGHT) ." ".
                     $value . "\n";
            #str_pad((" " . $value), $i_lenghtString, ".", STR_PAD_LEFT)  ." \n";
    }

    return $str_joinValueArray;
}

function getStringFromAssociativeArrayWhereValueArray( $asArr_associative,
                                                       $str_key,
                                                       $str_value,
                                                       $i_lenghtString )
{
    # top
    $str_joinValueArray = 
        str_pad( "[$str_key]",   ($i_lenghtString +1), " ", STR_PAD_RIGHT) .
                 "[$str_value]" ."\n";
        #str_pad( "[$str_value]", (25 +1), " ", STR_PAD_LEFT)  ."\n";

    # body (pair value: 'name' => 'date')
    foreach( $asArr_associative as $key => $value ) {
        $str_joinValueArray .= " ". 
            str_pad( $key . " ", $i_lenghtString, ".", STR_PAD_RIGHT) ." ".
                     $value[0] . "\n  "
                    .str_repeat( " ", $i_lenghtString ).
                     $value[1] . "\n";
    }

    return $str_joinValueArray;
}

# http://stackoverflow.com/questions/6054033/pretty-printing-json-with-php
# edit: alexey kovalev
function prettyPrint( $json )
{
    $result = '';
    $level = 0;
    $prev_char = '';
    $in_quotes = false;
    $ends_line_level = NULL;
    $json_length = strlen( $json );

    for( $i = 0; $i < $json_length; $i++ ) {
        $char = $json[$i];
        $new_line_level = NULL;
        $post = "";
        if( $ends_line_level !== NULL ) {
            $new_line_level = $ends_line_level;
            $ends_line_level = NULL;
        }
        if( $char === '"' && $prev_char != '\\' ) {
            $in_quotes = !$in_quotes;
        } else if( ! $in_quotes ) {
            switch( $char ) {
                case '}': case ']':
                    $level--;
                    $ends_line_level = NULL;
                    $new_line_level = $level;
                    break;

                case '{': case '[':
                    $level++;
                case ',':
                    $ends_line_level = $level;
                    break;

                case ':':
                    $post = " ";
                    break;

                case " ": case "\t": case "\n": case "\r":
                    $char = "";
                    $ends_line_level = $new_line_level;
                    $new_line_level = NULL;
                    break;
            }
        }
        if( $new_line_level !== NULL ) {
            $result .= "\r\n".str_repeat( "\t", $new_line_level );
        }
        $result .= $char.$post;
        $prev_char = $char;
    }

    # breaty array: alexey kovalev
    $str_startArray = "/\[\r\n\t{2,}\"/";
    $result = preg_replace( $str_startArray, '[ "', $result );

    $str_middleArray = "/,\r\n\t{2,}\"/";
    $result = preg_replace( $str_middleArray, ', "', $result );

    $str_endArray = "/\"\r\n\t{2,}\]/";
    $result = preg_replace( $str_endArray, '" ]', $result );

    return $result;
}

?>