(Crudely) collect Amazon ASINs from URLs

asins.php

<?php
/**
 * Quick script to extract ASINs from Amazon URLs.
 *
 * @see http://stackoverflow.com/a/12827734/70876
 *
 * beporter at users dot sourceforge dot net
 * 2016-05-18
 */

//--------------------------------------
function findAsins($i) {
	return array_filter(
		explode('/', parse_url($i, PHP_URL_PATH)),
		function ($s) { return preg_match('/^(B\d{2}\w{7}|\d{9}(X|\d))$/', $s); }
	);
}

//--------------------------------------
// main()
$inputs = [
	'http://www.amazon.com/gp/product/B00H00OG7G/ref=s9_simh_gw_g23_i2_r?ie=UTF8&fpl=fresh&pf_rd_m=ATVPDKIKX0DER&pf_rd_s=&pf_rd_r=1T4A8079XNP9A8ZH73RT&pf_rd_t=36701&pf_rd_p=5d23eaf6-6278-49c1-b6df-7de0cb9b3a26&pf_rd_i=desktop',
	'/Rust-Oleum-260357-Touch-Polyurethane-Matte/dp/B00714ZO22/more-shit/so-much/shit/',
	'Touch-Polyurethane-Matte/dp/B00714ZO22',
	'www.amazon.cn/电脑-it-办公/dp/B00YARCGHO/',
	'/B00YCCCCCC/ALSO10NUMS/?something-else',
	'Polyurethane/dp/B00RZYBZ9Y?foo=bar',
];

foreach ($inputs as $i) {
	// array_pop() is us assuming the "last" of multiple matches in the URL is correct.
	echo array_pop((findAsins($i))) . PHP_EOL;
}

/*--------------------------------------
Output:

$ php asins.php
B00H00OG7G
B00714ZO22
B00714ZO22
B00YARCGHO
B00YCCCCCC
B00RZYBZ9Y

*/

extract-csv-asins.php

#!/usr/bin/env php
<?php
/**
 * This script takes two arguments: A path to a CSV file and a column name
 * from which to extract ASINs. It will return a quoted, comma separated
 * string of all found ASINs, suitable for use in an SQL query's `IN (...)`
 * clause.
 *
 * Example:
 *    $ ./asin-extract.php a_file.csv columnName
 */

/**
 * Returns an array of all path-separated ASINs matched in a provided URL.
 *
 * @param string $s A URL possibly containing one or more `/ASIN/` path segment.
 * @return array All matched ASINs, in order they were found in the URL (left to right).
 */
function findAsins($s) {
  return array_filter(
    explode('/', parse_url($s, PHP_URL_PATH)),
    function ($s) { return preg_match('/^(B\d{2}\w{7}|\d{9}(X|\d))$/', $s); }
  );
}

/**
 * main()
 */

if ($argc < 3) {
  die('Must supply CSV path as first argument and column name containing URLs with ASINs as second argument.');
}

$path = $argv[1];
$column = $argv[2];

if (!is_readable($path)) {
  die('Supplied CSV path is not readable.');
}

ini_set('auto_detect_line_endings', true);
$csv = new \SplFileObject($path);
$csv->setFlags(SplFileObject::READ_CSV);
$headers = $csv->current();

if (!in_array($column, $headers)) {
  die(sprintf('First row of CSV file does not contain a column named `%s`.', $column));
}

$columnPosition = array_search($column, $headers);
$foundAsins = [];

foreach ($csv as $row) {
  $foundAsins = array_merge($foundAsins, findAsins($row[$columnPosition]));
}

foreach (array_chunk($foundAsins, 8) as $chunk) {
  echo '"' . implode('","', $chunk) . '"' . PHP_EOL;
}