aboutsummaryrefslogtreecommitdiffstats
path: root/vendor/pear/text_languagedetect/docs/example_web.php
diff options
context:
space:
mode:
authorMario Vavti <mario@mariovavti.com>2020-08-22 20:01:59 +0200
committerMario Vavti <mario@mariovavti.com>2020-08-22 20:01:59 +0200
commit646dce7765b36afab5c1688c51ebab48a214864e (patch)
tree5f017403ba8902507a0f1203db4a8954f04b48fa /vendor/pear/text_languagedetect/docs/example_web.php
parent88a68f96da303893d911f09c25088d4f8288b5fb (diff)
downloadvolse-hubzilla-646dce7765b36afab5c1688c51ebab48a214864e.tar.gz
volse-hubzilla-646dce7765b36afab5c1688c51ebab48a214864e.tar.bz2
volse-hubzilla-646dce7765b36afab5c1688c51ebab48a214864e.zip
composer update pear/text_languagedetect
Diffstat (limited to 'vendor/pear/text_languagedetect/docs/example_web.php')
-rw-r--r--vendor/pear/text_languagedetect/docs/example_web.php72
1 files changed, 72 insertions, 0 deletions
diff --git a/vendor/pear/text_languagedetect/docs/example_web.php b/vendor/pear/text_languagedetect/docs/example_web.php
new file mode 100644
index 000000000..bee8f51a4
--- /dev/null
+++ b/vendor/pear/text_languagedetect/docs/example_web.php
@@ -0,0 +1,72 @@
+<?php
+
+/**
+ * example usage (web)
+ *
+ * @package Text_LanguageDetect
+ * @version CVS: $Id$
+ */
+
+// browsers will encode multi-byte characters wrong unless they think the page is utf8-encoded
+header('Content-type: text/html; charset=utf-8', true);
+
+require_once 'Text/LanguageDetect.php';
+
+$l = new Text_LanguageDetect;
+if (isset($_REQUEST['q'])) {
+ $q = stripslashes($_REQUEST['q']);
+}
+
+?>
+<html>
+<head>
+<title>Text_LanguageDetect demonstration</title>
+</head>
+<body>
+<h2>Text_LanguageDetect</h2>
+<?
+echo "<small>Supported languages:\n";
+$langs = $l->getLanguages();
+sort($langs);
+foreach ($langs as $lang) {
+ echo ucfirst($lang), ', ';
+ $i++;
+}
+
+echo "<br />total $i</small><br /><br />";
+
+?>
+<form method="post">
+Enter text to identify language (at least a couple of sentences):<br />
+<textarea name="q" wrap="virtual" cols="80" rows="8"><?= $q ?></textarea>
+<br />
+<input type="submit" value="Submit" />
+</form>
+<?
+if (isset($q) && strlen($q)) {
+ $len = $l->utf8strlen($q);
+ if ($len < 20) { // this value picked somewhat arbitrarily
+ echo "Warning: string not very long ($len chars)<br />\n";
+ }
+
+ $result = $l->detectConfidence($q);
+
+ if ($result == null) {
+ echo "Text_LanguageDetect cannot identify this piece of text. <br /><br />\n";
+ } else {
+ echo "Text_LanguageDetect thinks this text is written in <b>{$result['language']}</b> ({$result['similarity']}, {$result['confidence']})<br /><br />\n";
+ }
+
+ $result = $l->detectUnicodeBlocks($q, false);
+ if (!empty($result)) {
+ arsort($result);
+ echo "Unicode blocks present: ", join(', ', array_keys($result)), "\n<br /><br />";
+ }
+}
+
+unset($l);
+
+/* vim: set expandtab tabstop=4 shiftwidth=4 softtabstop=4: */
+
+?>
+</body></html>