<?php
############################################################################
#
#   Project:    World Extremes
#   Module:     worldextremes.php
#               (based on Michael's work on usaextremes.php)
#   Purpose:    Provides World/Country/State Extremes for Web page display
#   Authors:    Michael (michael at relayweather.com, website shut down)
#               Ken True (webmaster@saratoga-weather.org) V2.01, V2.02, V4+
#
#   Note: this script replaces the prior usaextremes.php and stateextremes.php
#         scripts as their data source has been deprecated by the NWS.
#
############################################################################
#      Usage:  Place the following on your webpage
#
#      include_once('worldextremes.php');
#      
#      Then echo/print these variables within the page where you would like them displayed:
#
#      $omittedCountry (text list of countries excluded from $world high/low/precip scans)
#         Note: the setting $ignoreCountrys is the array of country names to to exclude
#      $worldhigh
#      $worldlow
#      $worldprecip
# 
#      $selectedCountry   (setting: country name for selected country high/low/precip)
#         Note:  run http://your.site.com/worldextremes.php?list to see the list of names to use
#      $countryhigh
#      $countrylow
#      $countryprecip 
#
#      $usahigh     (Note: for lower-48 USA states)
#      $usalow      (Note: for lower-48 USA states)
#      $usaprecip   (Note: for lower-48 USA states)
# 
#      $selectedState  (setting: USA state 2-character name abbreviation in settings area)
#      $selectState    (a copy of $selectedState for compatibility with old stateextremes.php)
#      $statehigh
#      $statelow
#      $stateprecip
# 
#      $reportDate (nicely formatted date of the report)
#      $stateReportDate  (a copy of $reportDate for compatibility with old stateextremes.php)
#
############################################################################
// version
// Version 2.00 - 14-Sep-2010 - initial release
// Version 2.01 - 21-Sep-2012 - adapted for WeatherUnderground use - K. True
// Version 2.02 - 23-Aug-2016 - adapted for www.wpc.ncep.noaa.gov - K. True
//
// Version 4.00 - 11-Sep-2016 - rewrite to use new URL:
//       http://www.cpc.noaa.gov/products/analysis_monitoring/cdus/prcp_temp_tables/dly_glob1.txt
// Version 4.01 - 14-Sep-2016 - added fix for malformed data record
// Version 4.02 - 17-Sep-2016 - added fixes for more malformed data records
// Version 4.03 - 13-Oct-2016 - added IgnoreStations feature
// Version 4.03c - 06-Feb-2019 - use https for www.cpc.noaa.gov
// Version 4.03d - 07-Feb-2019 - add curl_setopt($ch,CURLOPT_SSL_VERIFYHOST,0); for bad cert
// Version 4.03e - 09-Feb-2019 - corrected url to www.cpc.ncep.noaa.gov

//  rewrite of usaextremes for worldextremes to provide World/Country/State variables for display 02-Apr-2020
// Version 5.00 - 04-Apr-2020 - use CSV files from https://ftp.cpc.ncep.noaa.gov/cadb_v2/YYYY/MM/daily_summary_YYYYMMDD_v2.csv	
// Version 5.01 - 05-Apr-2020 - replace array_key_first() with array_keys()/array_pop for PHP 5x compatibility
// Version 5.02 - 25-Jun-2020 - update for station call entry in data
// Version 5.03 - 26-Jun-2020 - add $ignoreStations to remove stn_id with bogus data from list
// Version 5.04 - 28-Jul-2020 - add filter to remove reporting for stations with Tmax-Tmin < 2.0

$worldextremesverion = "5.04 - 28-Jul-2020";
/////////////////////////////////////////////////////////////////////////////
//SETTINGS START HERE////////////////////////////////////////////////////////
/////////////////////////////////////////////////////////////////////////////
// the daily files seem to be updated by 13:00 each day for the prior day
// Name of cache file  --  This file must be set to CHMOD-666
$cacheFileDir = './'; // directory to store cache file in.
$cacheFile2 = "worldextremesCache.txt";  
// Age of cache file before re-fetch caching time, in seconds (3600 = 1 hour)
$cache_life = 3600*6;
$reportDateFormat = "l, F j, Y"; // Day, Month d, yyyy 
$tUOM = '&deg;F'; // or ='' for no temperature unit display (display in C is default)
$rUOM = 'in';     // or ='' for no rain unit display (display in mm is default)
#
$ignoreCountrys = array('Antarctica','Greenland');     // for world extremes - exclude these countrie(s)
$ignoreStations = array('99KLRJ','99KWVI','99KGAD','99KGDJ');     // list of stn_id (field 0) to ignore for bogus data
$tempDiffAllowed = 40; // max difference Tmax-Tmin (C) for valid data
$selectedCountry = 'Canada'; // for country max/min/precip in $country* variables
// note: the $usa* variables will have the min/max/precip for the lower-48 states
$selectedState   = 'CA';  // for USA state max/min/precip in $state* variables
/////////////////////////////////////////////////////////////////////////////
//END SETTINGS///////////////////////////////////////////////////////////////
/////////////////////////////////////////////////////////////////////////////
if (isset($_REQUEST['sce']) && strtolower($_REQUEST['sce']) == 'view' ) {
   //--self downloader --
   $filenameReal = __FILE__;
   $download_size = filesize($filenameReal);
   header('Pragma: public');
   header('Cache-Control: private');
   header('Cache-Control: no-cache, must-revalidate');
   header("Content-type: text/plain,charset=ISO-8859-1");
   header("Accept-Ranges: bytes");
   header("Content-Length: $download_size");
   header('Connection: close');

   readfile($filenameReal);
   exit;
}
list($utcYMD,$utcNowhm) = explode(' ',gmdate('Y-m-d hi'));
$utcYMD1 = gmdate('Y-m-d',strtotime('-1 day'));
$utcYMD2 = gmdate('Y-m-d',strtotime('-2 day'));
$utcUse = $utcYMD1;

list($YY,$MM,$DD) = explode('-',$utcUse);

if(isset($_REQUEST['gmdate']) and preg_match('!^(\d{8})$!',$_REQUEST['gmdate'])) {
	$utcD = $_REQUEST['gmdate'];
	$YY = substr($utcD,0,4);
	$MM = substr($utcD,4,2);
	$DD = substr($utcD,6,2);
	$_REQUEST['force'] = '1';
}
// Site to Parse
//       https://ftp.cpc.ncep.noaa.gov/cadb_v2/daily/2020/03/daily_summary_20200331_v2.csv

echo "<!-- World Extremes Script Version $worldextremesverion saratoga-weather.org -->\n";
echo "<!-- utcYMD=$utcYMD utcNowhm=$utcNowhm utcYMD1=$utcYMD1 utcWMD2=$utcYMD2 -->\n";
echo "<!-- utcUse=$utcUse using YY=$YY MM=$MM DD=$DD date for data file -->\n"; 
# configuration from Saratoga template if available
global $SITE;
if(file_exists("Settings.php")) {include_once("Settings.php"); }
if (isset($SITE['cacheFileDir']))   {$cacheFileDir = $SITE['cacheFileDir']; }
if (isset($SITE['uomTemp']))        {$tUOM = $SITE['uomTemp']; }
if (isset($SITE['uomRain']))        {$rUOM = trim($SITE['uomRain']);}

$cacheFile2 = $cacheFileDir . $cacheFile2;

if (file_exists($cacheFile2)) {
    $filemtime = filemtime($cacheFile2);
    $filesize = filesize($cacheFile2);
    if (0 == $filesize){
        $filemtime = 0;
    }
} else {
    $filemtime = 0;
}   
//   open the cache file and write the new data and then close the file.
$forceRefresh = (isset($_REQUEST['force']))?true:false;
$current_time = time();
$cache_age = $current_time - $filemtime;
print "<!-- Cache refresh rate = $cache_life seconds.  Cache age = $cache_age seconds. -->\n";
if ($cache_age >= $cache_life){
	$url2 = gen_url($utcUse);
  print "<!-- fetching from '$url2' -->\n";
  list($html2,$RC) = curl_get_contents($url2);
	if($html2 === false or $RC <> '200') {
		print "<!-- RC='$RC' .. trying prior date=$utcYMD2 instead. -->\n";
		$utcUse = $utcYMD2; // use day before
		$url2 = gen_url($utcUse);
    print "<!-- fetching from '$url2' -->\n";
    list($html2,$RC) = curl_get_contents($url2);
	}
	if($html2 !== false and $RC == '200') {
    file_put_contents($cacheFile2,$html2);
    echo "<!-- cache file saved to $cacheFile2 -->\n";
	} else {
		echo "<!-- cache not written due to curl error -->\n";
    $filesize = filesize($cacheFile2);
    $html2 = file_get_contents($cacheFile2);
		echo "<!-- cache $cacheFile2 data reloaded and used instead -->\n";
	}
		 
} else {
   echo "<!-- The cache is current .. using $cacheFile2 for data -->\n";
   $filesize = filesize($cacheFile2);
   $html2 = file_get_contents($cacheFile2);
}


/* 
NOW looking for stuff in CSV like:
stn_id,stn_call,city,state,country,date,lat,lon,elev,tmax,tmin,report_p,final_p,p_flag,num_6hr_p,wxchars,trace,vp,vp_def,slp_6,slp_12,slp_18,slp_0,max_rh,min_rh,at,wc,wspd_3,wspd_6,wspd_9,wspd_12,wspd_15,wspd_18,wspd_21,wspd_24

The CSV Fields are:

0	stn_id
1	stn_call
2	city
3	state
4	country
5	date
6	lat
7	lon
8	elev
9	tmax
10	tmin
11	report_p
12	final_p
13	p_flag
14	num_6hr_p
15	wxchars
16	trace
17	vp
18	vp_def
19	slp_6
20	slp_12
21	slp_18
22	slp_0
23	max_rh
24	min_rh
25	at
26	wc
27	wspd_3
28	wspd_6
29	wspd_9
30	wspd_12
31	wspd_15
32	wspd_18
33	wspd_21
34	wspd_24

Units are C,kts,mb(hPa),mm for temp, wind, baro, rain
See: https://ftp.cpc.ncep.noaa.gov/cadb_v2/docs/CADB_Daily_v2_Output_Documentation.pdf for docs

*/

$worldStations = array();
$reportDate = '';

list($headers2,$content2) = explode("\r\n\r\n",$html2);

$rawrecs = explode("\n",$content2); // process the file

$idx=0;

$datesSeen = array();
$countrySeen = array();

# load up the raw data for further processing .. reject the ones with no country or no city
foreach ($rawrecs as $n => $rec) {
	$V = explode(',',$rec);
	if(!isset($V[9])) {continue; }
	$tC =substr($V[4],0,1);
	if($tC == '-' or $tC == '*') { continue; }
	if($V[0] == 'stn_id') { continue; }
	if($V[9] < -999 or $V[10] < -999 or $V[16] < -999) { continue; }
	if(abs($V[9]-$V[10]) < 2.0) { continue;}
	if(in_array($V[0],$ignoreStations)) { continue; }

	$tCountry = $V[4];
	$tCountry = strtolower( trim($tCountry) );
	$tCountry = str_replace('/','| ',$tCountry);
	
	$tCountry = ucwords( str_replace('_',' ', $tCountry) );
	$tCountry = str_replace('| ','/',$tCountry);
  if(!isset($countrySeen[$tCountry])) {
		$countrySeen[$tCountry] = 1;
	} else {
		$countrySeen[$tCountry]++;
	}
	if(in_array($tCountry,$ignoreCountrys)) { continue; }
	
	if(abs($V[9]-$V[10]) > $tempDiffAllowed) { continue; } 
	
	$recmarker = $V[13];
	if($recmarker > '0') {
		$tHigh = cvtCtemp($V[9],$tUOM);
		$tLow  = cvtCtemp($V[10],$tUOM);
		$tPrecip=cvtMrain($V[12],$rUOM);
		$tPrecipEstim=cvtMrain($V[11],$rUOM);
		$tData=$V[15];
		$tState=$V[3];
		if(substr($tState,0,1) == '-') {$tState = ''; }
		$tStation=$V[2]; 
		$tStation = strtolower( trim($tStation) );
		$tStation = str_replace('/','| ',$tStation);
		
		$tStation = ucwords( str_replace('_',' ', $tStation) );
		$tStation = str_replace('| ','/',$tStation);
		if(!isset($datesSeen[$V[5]])) {
			$datesSeen[$V[5]] = 1;
		} else {
			$datesSeen[$V[5]]++;
		}
			
		if(isset($_REQUEST['debug2'])) {
		  print "<!-- rec='$rec' -->\n";
          print "<!-- tHigh='$tHigh' tLow=$tLow, tPrecip='$tPrecip', tPrecipEstim='$tPrecipEstim',";
		  print " tData='$tData', tState='$tState', tStation='$tStation' tCountry='$tCountry'-->\n";
		}
		$worldStations[$idx] = join("\t",array($tHigh,$tLow,$tPrecip,$tPrecipEstim,$tData,$tState,$tStation,$tCountry));
		$idx++;
	}
	
}
//print "<!-- records by date: \n".var_export($datesSeen,true)."\n-->\n";
//$tDate = array_key_first($datesSeen);  // php 7 only
$rDates = array_keys($datesSeen); // get the date(s) seen
$tDate  = array_pop($rDates);     // get first one
$utcUse = substr($tDate,0,4).'-'.substr($tDate,4,2).'-'.substr($tDate,6,2);
$reportDate = gmdate($reportDateFormat,strtotime($utcUse.'T1200'));
print "<!-- reportDate='$reportDate' -->\n";


// now scan for the highs/lows 
list($worldhigh,$worldlow,$worldprecip) = scan_for($worldStations,'','',$tUOM,$rUOM);

$omittedCountry = join(',',$ignoreCountrys);
print "<!-- world values omitting \$omittedCountry='$omittedCountry' from list -->\n";
print "<!-- \$worldhigh  ='$worldhigh' -->\n";
print "<!-- \$worldlow   ='$worldlow' -->\n";
print "<!-- \$worldprecip='$worldprecip' -->\n";


list($countryhigh,$countrylow,$countryprecip) = scan_for($worldStations,$selectedCountry,'',$tUOM,$rUOM);

print "<!-- \$selectedCountry='$selectedCountry' -->\n";
print "<!-- \$countryhigh  ='$countryhigh' -->\n";
print "<!-- \$countrylow   ='$countrylow' -->\n";
print "<!-- \$countryprecip='$countryprecip' -->\n";

list($usahigh,$usalow,$usaprecip) = scan_for($worldStations,'United States','',$tUOM,$rUOM);

print "<!-- country='United States' -->\n";
print "<!-- \$usahigh  ='$usahigh' -->\n";
print "<!-- \$usalow   ='$usalow' -->\n";
print "<!-- \$usaprecip='$usaprecip' -->\n";


list($statehigh,$statelow,$stateprecip) = scan_for($worldStations,'United States',$selectedState,$tUOM,$rUOM);

print "<!-- country='United States' \$selectedState='$selectedState' -->\n";
print "<!-- \$statehigh  ='$statehigh' -->\n";
print "<!-- \$statelow   ='$statelow' -->\n";
print "<!-- \$stateprecip='$stateprecip' -->\n";

$stateReportDate = $reportDate;    // for compatibility with stateextremes.php
$selectState     = $selectedState; // for compatibility with stateextremes.php

if(isset($_REQUEST['list'])) {
	print "<pre>List of available country names in CPS data\n";
	print "Use in this script as \$selectedCountry='name'; in settings area\n\n";
	ksort($countrySeen);
	foreach ($countrySeen as $name => $obs) {
	  print "'$name'\t($obs observations)\n";
	}
	print "</pre>\n";
	
}
# end of main program

# ---------------------------------------------------------------------
# functions
# ---------------------------------------------------------------------

function scan_for($worldStations,$country,$state,$tUOM,$rUOM) {
	$worldhighStation = '';
	$worldhighValue = -999;
	$worldlowStation = '';
	$worldlowValue = 999;
	$worldprecipStation = '';
	$worldprecipValue = 0;

	if (count($worldStations) > 0) {
		$worldhigh = '';
		$worldlow = '';
		$worldprecip = '';
	} else {
		$worldhigh = 'N/A';
		$worldlow = 'N/A';
		$worldprecip = 'N/A';
	}

  $selectLower48 = false;
	if($country == 'United States' and $state == '') {
		$selectLower48 = true;
	}
	
	foreach ($worldStations as $n => $vals) {
		list($tHigh,$tLow,$tPrecip,$tPrecipEstim,$tData,$tState,$tStation,$tCountry) = explode("\t",$vals);
		if(strlen($country) > 0 and strcasecmp($tCountry,$country) !== 0) { continue; }
		if(strlen($state) > 0 and strcasecmp($tState,$state) !== 0) { continue; }
		if($selectLower48 and in_array($tState,array('AK','HI')) )  { continue; }

		if($tHigh < -90 or $tLow < -90 ) { continue; }
		
		if($tHigh > $worldhighValue) {
			if(isset($_REQUEST['debug'])) {
				print "<!-- new HIGH: '$tHigh' $n - $tCountry $tState $tStation -->\n";
			}
			$worldhighValue = $tHigh;
			$worldhighStation = $n; // remember for later
		}
		if($tLow < $worldlowValue) {
			if(isset($_REQUEST['debug'])) {
				print "<!-- new LOW: '$tLow' $n - $tCountry $tState $tStation -->\n";
			}
			$worldlowValue = $tLow;
			$worldlowStation = $n; // remember for later
		}
		if($tPrecip > $worldprecipValue) {
			if(isset($_REQUEST['debug'])) {
				print "<!-- new Precip: '$tPrecip' $n - $tCountry $tState $tStation -->\n";
			}
			$worldprecipValue = $tPrecip;
			$worldprecipStation = $n;
			
		}
		
	}
	if(isset($_REQUEST['debug'])) {
	  print "<!-- selecting for country='$country' state='$state' selectLower48='$selectLower48' -->\n";
		print "<!-- highValue=$worldhighValue station=$worldhighStation ".$worldStations[$worldhighStation]." -->\n";
		print "<!-- lowValue=$worldlowValue station=$worldlowStation ".$worldStations[$worldlowStation]." -->\n";
		if($worldprecipValue > 0.0) {
		  print "<!-- precipValue=$worldprecipValue station=$worldprecipStation ".$worldStations[$worldprecipStation]." -->\n";
		} else {
			print "<!-- precipValue=0.0 (no rain found) -->\n";
		}
		
	}
	
	$showState = (strlen($state)>0)?true:false;
	$showCountry = (strlen($country) == 0)?true:false;
	if(strcasecmp('United States',$country) == 0 and strlen($state) < 1) {
		$showState = true;
		$showCountry = false;
	}
	
	// now pass through to see if any duplicate high, low, high precip exists and format the strings
	foreach ($worldStations as $n => $vals) {
		list($tHigh,$tLow,$tPrecip,$tPrecipEstim,$tData,$tState,$tStation,$tCountry) = explode("\t",$vals);
		if(strlen($country) > 0 and strcasecmp($tCountry,$country) !== 0) {continue; }
		if(strlen($state) > 0 and strcasecmp($tState,$state) !== 0) {continue; }
		if($tHigh < -90 or $tLow < -90 ) { continue; }
		if($selectLower48 and in_array($tState,array('AK','HI')) )  { continue; }
		
		if($tHigh == $worldhighValue) {
			$worldhigh .= "<b>$tHigh$tUOM at $tStation";
			$worldhigh .= $showState?", $tState":"";
			$worldhigh .= $showCountry?", $tCountry":"";
			$worldhigh .= "</b>\n";
		}
		if($tLow == $worldlowValue) {
			$worldlow .= "<b>$tLow$tUOM at $tStation";
			$worldlow .= $showState?", $tState":"";
			$worldlow .= $showCountry?", $tCountry":"";
			$worldlow .= "</b>\n";
		}
		if($tPrecip == $worldprecipValue and $tPrecip > 0) {
			$tPrecipFmt = $tPrecip;
			$worldprecip .= "<b>$tPrecipFmt$rUOM at $tStation";
			$worldprecip .= $showState?", $tState":"";
			$worldprecip .= $showCountry?", $tCountry":"";
			$worldprecip .= "</b>\n";
		}
	}
	// Make results HTML pretty
	
	if(strlen($worldprecip) < 10) {
		$worldprecip = "<b>none</b>\n";
	}
	
	$worldhigh   = str_replace("\n","<br/>\n",$worldhigh);
	$worldlow    = str_replace("\n","<br/>\n",$worldlow);
	$worldprecip = str_replace("\n","<br/>\n",$worldprecip);
	
	return(array($worldhigh,$worldlow,$worldprecip));
}


# ---------------------------------------------------------------------
function check_ignore_station($str, array $arr)
{
    foreach($arr as $a) {
        if (stripos($str,$a) !== false) {return true;}
    }
    return false;
}

# ---------------------------------------------------------------------
function cvtCtemp($val,$uom) {
	//return($val);
	if(strpos($uom,'F') !== false) {
		$out = round(1.8 * $val + 32);
	} else {
		$out = $val;
	}
	return($out);
}

# ---------------------------------------------------------------------
function cvtMrain($val,$uom) {
	//return($val);
	if(strpos($uom,'in') !== false) {
		$out = sprintf("%01.2f",$val / 25.4,2);
	} else {
		$out = $val;
	}
	return($out);
}

# ---------------------------------------------------------------------
function curl_get_contents($url){
		$ch = curl_init();
		curl_setopt($ch, CURLOPT_HEADER, 1);
		curl_setopt($ch, CURLOPT_RETURNTRANSFER, 1);
		curl_setopt($ch, CURLOPT_URL, $url);
		curl_setopt($ch, CURLOPT_SSL_VERIFYPEER,0);
		curl_setopt($ch, CURLOPT_SSL_VERIFYHOST,0);
		curl_setopt($ch, CURLOPT_USERAGENT, "Mozilla/5.0 (worldextremes.php - saratoga-weather.org)");
		$data = curl_exec($ch);
    $cinfo = curl_getinfo($ch); // get info on curl exec.
		$RC = $cinfo['http_code'];

    $Status = "<!-- HTTP stats: " . " RC=" . $cinfo['http_code'];
		if (isset($cinfo['primary_ip'])) {
			$Status .= " dest=" . $cinfo['primary_ip'];
		}
    if (isset($cinfo['primary_port'])) {
      $Status .= " port=" . $cinfo['primary_port'];
    }

    if (isset($cinfo['local_ip'])) {
      $Status.= " (from sce=" . $cinfo['local_ip'] . ")";
    }

    $Status.= "\n      Times:" . 
		" dns=" . sprintf("%01.3f", round($cinfo['namelookup_time'], 3)) . 
		" conn=" . sprintf("%01.3f", round($cinfo['connect_time'], 3)) . 
		" pxfer=" . sprintf("%01.3f", round($cinfo['pretransfer_time'], 3));
    if ($cinfo['total_time'] - $cinfo['pretransfer_time'] > 0.0000) {
      $Status.= " get=" . sprintf("%01.3f", round($cinfo['total_time'] - $cinfo['pretransfer_time'], 3));
    }

    $Status.= " total=" . sprintf("%01.3f", round($cinfo['total_time'], 3)) . " secs -->\n";
		echo $Status;
		if($data === false or $RC <> '200') {
      echo "<!-- Error: RC=$RC " . curl_error($ch) . " -->\n"; 
			} else {
      echo "<!-- curl completed successfully -->\n";
    }
		curl_close($ch);
		
		return (array($data,$RC));
}

# ---------------------------------------------------------------------
function gen_url($date) {
	$urlTemplate = "https://ftp.cpc.ncep.noaa.gov/cadb_v2/daily/%s/%s/daily_summary_%s%s%s_v2.csv";
  list($YY,$MM,$DD) = explode('-',$date);
	
	$outUrl = sprintf($urlTemplate,$YY,$MM,$YY,$MM,$DD);
	
	return($outUrl);
}