From d00f19145627312125c593f35193f04733b4df4e Mon Sep 17 00:00:00 2001
From: Phil Burton <phil@d3r.com>
Date: Tue, 11 Jun 2019 14:17:29 +0100
Subject: First commit

---
 etym.php | 85 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 85 insertions(+)
 create mode 100644 etym.php

(limited to 'etym.php')

diff --git a/etym.php b/etym.php
new file mode 100644
index 0000000..08059fd
--- /dev/null
+++ b/etym.php
@@ -0,0 +1,85 @@
+<?php
+
+use PHPHtmlParser\Dom;
+
+$baseURL = "http://www.etymonline.com/word/";
+$htmlNode = "section[class^='word__defination']";
+$pasteCmd = "pastebinit -b http://p.of.je 2>/dev/null";
+
+function readStdin()
+{
+    $input = fgets(STDIN);
+
+    if ($input === false) {
+        echo "No input supplied!\n";
+        exit(1);
+    }
+
+    $input = rtrim($input, "\n");
+
+    return $input;
+}
+
+function cleanUpHtml($input)
+{
+    // Strip HTML Tags
+    $clear = strip_tags($input);
+    // Clean up things like &amp;
+    $clear = html_entity_decode($clear);
+    // Strip out any url-encoded stuff
+    $clear = urldecode($clear);
+    // Replace Multiple spaces with single space
+    $clear = preg_replace('/ +/', ' ', $clear);
+    // Trim the string of leading/trailing space
+    $clear = trim($clear);
+    // Capitalise the first char.
+    $clear = ucfirst($clear);
+
+    return $clear;
+}
+
+function handleTruncation($input, $definition, $url)
+{
+    $truncated = $definition;
+
+    $MAX_CHARACTERS = 350;
+    if (strlen($truncated) >= $MAX_CHARACTERS) {
+        // Create that povjee link.
+        // Capitalise the first char of the input.
+        $input    = ucfirst($input);
+        $defAndUrl        = "\"$input\"" . "\n\n" . $definition . "\n\n" . "[Original at: $url]";
+        $safeDef        = escapeshellarg($defAndUrl);
+
+        $pasteBinCmd     = "echo $safeDef | " . $pasteCmd;
+        $pasteBinLink    = exec($pasteBinCmd);
+
+        $truncated = substr($truncated, 0, $MAX_CHARACTERS) . "... [More info at $pasteBinLink]";
+    }
+
+    return $truncated;
+}
+
+function getDefinition()
+{
+
+    $input = readStdin();
+
+    $targetURL = $baseURL . $input;
+
+    $dom = new Dom;
+    $dom->load($targetURL);
+    $html = $dom->find($htmlNode)[0]->innerHtml();
+    $node = $html->find($htmlNode, 0);
+
+    if ($node === null) {
+        echo "No entry found for '$input'!\n";
+        exit(2);
+    }
+
+    $definition = cleanUpHtml($node->innertext);
+    $definition = handleTruncation($input, $definition, $targetURL);
+
+    echo $definition . "\n";
+}
+
+getDefinition();
-- 
cgit v1.2.3