Add tool to find duplicate lang tags

1 files changed, 74 insertions, 0 deletions

diff --git a/tools/global-candidates.php b/tools/global-candidates.php
new file mode 100644
index 00000000..12e900a5
--- /dev/null
+++ b/tools/global-candidates.php
@@ -0,0 +1,74 @@
+<?php
+
+$langs = $argc > 1 && $argv[1] ? $argv[1] : 'en';
+
+echo "Scanning for $langs... (pass as comma separated list, no spaces)\n";
+
+$tags = [];
+$strings = [];
+
+foreach (glob('./modules-available/*/lang/{' . $langs . '}/template-tags.json', GLOB_NOSORT | GLOB_BRACE) as $file) {
+	preg_match('#modules-available/([^/]+)/lang/(..)#', $file, $out);
+	$module = $out[1];
+	$lang = $out[2];
+	$j = json_decode(file_get_contents($file), true);
+	if (!is_array($j)) continue;
+	foreach ($j as $k => $v) {
+		if (!isset($tags[$k])) {
+			$tags[$k] = ['modules' => [], 'lang' => []];
+		}
+		$tags[$k]['modules'][$module] = true;
+		if (!isset($tags[$k]['lang'][$lang])) {
+			$tags[$k]['lang'][$lang] = [];
+		}
+		$tags[$k]['lang'][$lang][$v] = true;
+		if (!isset($strings[$v])) {
+			$strings[$v] = [];
+		}
+		if (!isset($strings[$v][$k])) {
+			$strings[$v][$k] = [];
+		}
+		$strings[$v][$k][$module] = true;
+	}
+}
+
+if ($argc > 1) {
+	$find = array_flip(array_slice($argv, 2));
+	print_r($find);
+} else $find = [];
+
+echo "\n\nDUPLICATE TAG NAME ACROSS DIFFERENT MODULES:\n";
+foreach ($tags as $k => &$tag) {
+	if (isset($find[$k])) {
+		echo "## LOOKUP: '$k'\n";
+		print_r($tag['lang']);
+	}
+	if (count($tag['modules']) < 4) continue;
+	$tag['modules'] = array_keys($tag['modules']);
+	foreach ($tag['lang'] as &$lang) {
+		$lang = array_keys($lang);
+	}
+	unset($lang);
+	echo "## Common tag '$k'\n";
+	echo "   In " . count($tag['modules']) . " modules: " . implode(', ', $tag['modules']) . "\n";
+	foreach ($tag['lang'] as $lang => $str) {
+		echo "   " . count($str) . " in $lang: '" . implode("', '", $str) . "'\n";
+		if (count($str) / count($tag['modules']) < 0.26) {
+			echo "   +++ Possible candidate +++\n";
+		}
+	}
+}
+
+echo "\n\nDUPLICATE STRINGS WITH DIFFERENT NAMES:\n";
+foreach ($strings as $text => $data) {
+	if (count($data) < 3) continue;
+	echo "## '$text' ##\n";
+	foreach ($data as $tag => $mods) {
+		echo "   As $tag in";
+		foreach($mods as $mod => $count) {
+			echo " $mod($count) ";
+		}
+		echo "\n";
+	}
+}
+