scrapes official holidays from the Court's official website and inserts in our database

sdny-holiday-web-scrape.php

<?php
/** 
 * CLI script to scrape holidays from the SDNY website and store them in our database.
 * Normally you won't need this more than about once a year. And of course shit will 
 * break if the HTML changes
 */

exec('hostname',$hostname);
$host = $hostname[0];
$year = date('Y');
$command =  "curl -s http://nysd.uscourts.gov/holidays|egrep '<tr><td><strong>.+<small>'| perl -p -e \"s/<\/?[^>]+>|\(.+\)|^\s+//g\"";
exec($command, $lines, $retval);

if ("0" != $retval) {
    exit("oops, non-zero exit status from command:\n $command\n");
}

if (count($lines) < 10) {
    exit("downloaded data looks suspiciously short. please have a look and try a again\n");
}

/** the labels in our database look like: 
1	New Year's Day
2	Martin Luther King Day
3	Lincoln's Birthday
4	President's Day
5	Memorial Day
6	Independence Day
7	Labor Day
8	Columbus Day
9	Veterans' Day
10	Thanksgiving
11	Christmas
12	Election Day
* ------------------ */
/** web-scraped data should look like: 
New Year's Day - January 1st
Martin Luther King, Jr. Birthday - January 15th
Washington's Birthday / President's Day - February 19th
Memorial Day - May 28th
Independence Day - July 4th
Labor Day - September 3rd
Columbus Day - October 8th
Veteran's Day - November 12th
Thanksgiving Day - November 22nd
Friday November 23rd, 2018 - November 23rd
Christmas Eve - December 24th
Christmas Day - December 25th
New Year's Eve - December 31st 
 */

$db_params = parse_ini_file(getenv('HOME').'/.my.cnf');
$database = strstr($host,'interps') === false ? 'dev_interpreters' : 'interpreters';
$db = new PDO("mysql:host=localhost;dbname=$database", $db_params['user'], $db_params['password']);
$holidays = $db->query('SELECT * FROM holidays',PDO::FETCH_KEY_PAIR)->fetchAll();
$sql = 'INSERT INTO court_closings (date,holiday_id) VALUES (:date,:holiday_id)';
$statement = $db->prepare($sql);

foreach($lines as $line) {
    list($name,$date) = preg_split('/\s+-\s+/', $line);
    $holiday_id = array_search($name, $holidays);
    if ($holiday_id === false) {
        if (stristr($name,'Martin Luther King')) {
            $holiday_id = array_search('Martin Luther King Day', $holidays);
        } elseif (stristr($name,'President')) {
            $holiday_id = array_search('President\'s Day', $holidays);        
        } elseif (stristr($name,'Veteran')) {
            $holiday_id = array_search('Veterans\' Day', $holidays);
        } elseif (stristr($name,'Thanksgiving')) {
            $holiday_id = array_search('Thanksgiving', $holidays);
        } elseif (stristr($name,'Christmas')) {
            $holiday_id = array_search('Christmas', $holidays);
        } elseif (stristr($name,'New Year')) {
            $holiday_id = array_search('New Year\'s Day', $holidays);
        }
        if (! $holiday_id) {           
            echo "\nnot found: $name\n";
            print_r($holidays);
            while (! $holiday_id ) {
                $holiday_id = readline('Choose one of the above ids, or CTRL+C to bail: ');
                if (! key_exists($holiday_id, $holidays)) {
                    $holiday_id = false;
                    echo "not a valid id, try again\n";
                }
            }
        }
    }
    
    $timestamp = strtotime("$date, $year");
    if (false === $timestamp) {
        echo "WARNING: could not parse date (for $name) from '$date'\n";
        continue;
    }
    $formatted_date = date('Y-m-d',$timestamp);
    try {
         $statement->execute([':date'=>$formatted_date,':holiday_id'=>$holiday_id]);
        echo "added $holidays[$holiday_id] on $formatted_date\n";
    } catch (Exception $e) {
        printf("oops, caught exception %s: %s\n",get_class($e),$e->getMessage());
        echo "moving on...\n";
    }
}