ngrams-plain.php

<?php
#
# Ad-hoc oblique service for ngram databases with compatible interface
#

header('Content-Type: text/plain;charset=UtF-8');

if (@!isset($_REQUEST['q']) || @!isset($_REQUEST['nick'])) {
  printf("I need a q! I need a nick! Sonst functioneren ik niet.");
  exit();
}

$lang2code["en-us"]      = 17;
$lang2code["en-gb"]      = 18;
$lang2code["en"]         = 15;
$lang2code["zh"]         = 23;
$lang2code["fr"]         = 19;
$lang2code["de"]         = 20;
$lang2code["he"]         = 24;
$lang2code["it"]         = 22;
$lang2code["ru"]         = 25;
$lang2code["es"]         = 21;
$lang2code["en-us-2009"] =  5;
$lang2code["en-gb-2009"] =  6;
$lang2code["zh-2009"]    = 11;
$lang2code["en-2009"]    =  0;
$lang2code["fr-2009"]    =  7;
$lang2code["de-2009"]    =  8;
$lang2code["he-2009"]    =  9;
$lang2code["ru-2009"]    = 12;
$lang2code["es-2009"]    = 10;

if (!!preg_match_all("/:(\S+)/", $_REQUEST['q'], $mq)) {
  # TODO: ...
}

foreach ($mq[1] as $v) {
  if (array_key_exists(strtolower($v), $lang2code)) {
    $corpus = $lang2code[strtolower($v)];

  } else if (is_numeric($v) && !isset($to_year)) {
    $to_year = $v;

  } else if (is_numeric($v) && !isset($from_year)) {
    $from_year = $v;

  } else {
    # TODO: ...

  }
}

# TODO: This is hopefully not the right defined-or idiom...
$from_year = isset($from_year) ? $from_year : "1800";
$to_year = isset($to_year) ? $to_year : "2000";
$corpus = isset($corpus) ? $corpus : 0;

$query = preg_replace("/^(\s*:(\S+)\s*)*/", "", $_REQUEST['q']);

if (strlen($query) < 1) {
  printf(":(");
  exit();
}

$fetch = "http://books.example.com/ngrams/graph?content="
           . urlencode($query)
           . "&year_start=" . urlencode($from_year)
           . "&year_end=" . urlencode($to_year)
           . "&corpus=" . urlencode($corpus)
           . "&smoothing=3";

$ch = curl_init($fetch);
curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
curl_setopt($ch, CURLOPT_HEADER, 0);
$data = curl_exec($ch);
curl_close($ch);

if (!!preg_match_all("/data\\.addColumn\\('number',\\s*'(.*?)'\\)/",
        $data, $mcol)) {
  # ...
}

if (!!preg_match("/data\\.addRows\\((.*?)\\);/s",
        $data, $mjson)) {
  # ...
}

if (@!$mjson[1]) {
  printf(":(");
  exit();
}

$decoded = json_decode($mjson[1]);

if (@!$decoded) {
  printf(":(");
  exit();
}

$result = $_REQUEST['nick'] . ": ";

$max = max(array_slice(array_values(end($decoded)), 1));
$max = $max ? $max : 1;

foreach (end($decoded) as $k => $v) {
  if ($k > 0) {
    $result .= sprintf("[%s] %.03f%% ", $mcol[1][$k], $v/$max);
  } else {
    $result .= sprintf("[%s] %s ", $mcol[1][$k], $v);
  }
}

print($result);
?>