Fix util/extract.php

- Captures way more L10n::t(t) instances
- Add recursive search into `src/` folder
- Fix string capture with new sprintf-free syntax

util/extract.php

@@ -1,63 +1,100 @@
+#!/usr/bin/env php
 <?php
 
-	$arr = [];
+/**
+ * @file util/extract.php
+ *
+ * Extracts translation strings from the Friendica project's files to be exported
+ * to Transifex for translation.
+ *
+ * Outputs a PHP file with language strings used by Friendica
+ */
 
-	$files = ['index.php','boot.php'];
+$s = '<?php' . PHP_EOL;
-	$files = array_merge($files,glob('mod/*'),glob('include/*'),glob('addon/*/*'));
+$s .= '
-
-
-	foreach($files as $file) {
-		$str = file_get_contents($file);
-
-		$pat = '| L10n::t\(([^\)]*)\)|';
-		$patt = '| L10n::tt\(([^\)]*)\)|';
-
-		preg_match_all($pat,$str,$matches);
-		preg_match_all($patt, $str, $matchestt);
-
-
-		if(count($matches)){
-			foreach($matches[1] as $match) {
-				if(! in_array($match,$arr))
-					$arr[] = $match;
-			}
-		}
-		if(count($matchestt)){
-			foreach($matchestt[1] as $match) {
-				$matchtkns = preg_split("|[ \t\r\n]*,[ \t\r\n]*|",$match);
-				if (count($matchtkns)==3 && !in_array($matchtkns,$arr)){
-					$arr[] = $matchtkns;
-				}
-			}
-		}
-
-	}
-
-	$s = '<?php' . "\n";
-	$s .= '
 function string_plural_select($n){
 	return ($n != 1);
 }
 
 ';
 
-	foreach($arr as $a) {
+$arr = [];
-		if (is_array($a)){
+
-			if(substr($a[1],0,1) == '$')
+$files = ['index.php', 'boot.php'];
-				continue;
+$files = array_merge(
-			$s .= '$a->strings[' . $a[0] . "] = array(\n";
+	$files,
-			$s .= "\t0 => ". $a[0]. ",\n";
+	glob('mod/*'),
-			$s .= "\t1 => ". $a[1]. ",\n";
+	glob('include/*'),
-			$s .= ");\n";
+	glob('addon/*/*'),
-		} else {
+	glob_recursive('src')
-			if(substr($a,0,1) == '$')
+);
-				continue;
+
-			$s .= '$a->strings[' . $a . '] = '. $a . ';' . "\n";
+foreach ($files as $file) {
+	$str = file_get_contents($file);
+
+	$pat = '|L10n::t\(([^\)]*+)[\)]|';
+	$patt = '|L10n::tt\(([^\)]*+)[\)]|';
+
+	preg_match_all($pat, $str, $matches);
+	preg_match_all($patt, $str, $matchestt);
+
+	if (count($matches) || count($matchestt)) {
+		$s .= '// ' . $file . PHP_EOL;
+	}
+
+	if (count($matches)) {
+		foreach ($matches[1] as $long_match) {
+			$match_arr = preg_split('/(?<=[\'"])\s*,/', $long_match);
+			$match = $match_arr[0];
+			if (!in_array($match, $arr)) {
+				if (substr($match, 0, 1) == '$') {
+					continue;
+				}
+
+				$arr[] = $match;
+
+				$s .= '$a->strings[' . $match . '] = ' . $match . ';' . "\n";
+			}
+		}
+	}
+	if (count($matchestt)) {
+		foreach ($matchestt[1] as $match) {
+			$matchtkns = preg_split("|[ \t\r\n]*,[ \t\r\n]*|", $match);
+			if (count($matchtkns) == 3 && !in_array($matchtkns[0], $arr)) {
+				if (substr($matchtkns[1], 0, 1) == '$') {
+					continue;
+				}
+
+				$arr[] = $matchtkns[0];
+
+				$s .= '$a->strings[' . $matchtkns[0] . "] = array(\n";
+				$s .= "\t0 => " . $matchtkns[0] . ",\n";
+				$s .= "\t1 => " . $matchtkns[1] . ",\n";
+				$s .= ");\n";
+			}
+		}
+	}
+}
+
+$s .= '// Timezones' . PHP_EOL;
+
+$zones = timezone_identifiers_list();
+foreach ($zones as $zone) {
+	$s .= '$a->strings[\'' . $zone . '\'] = \'' . $zone . '\';' . "\n";
+}
+
+echo $s;
+
+function glob_recursive($path) {
+	$dir_iterator = new RecursiveDirectoryIterator($path);
+	$iterator = new RecursiveIteratorIterator($dir_iterator, RecursiveIteratorIterator::SELF_FIRST);
+
+	$return = [];
+	foreach ($iterator as $file) {
+		if ($file->getBasename() != '.' && $file->getBasename() != '..') {
+			$return[] = $file->getPathname();
 		}
 	}
 
-	$zones = timezone_identifiers_list();
+	return $return;
-	foreach($zones as $zone)
+}
-		$s .= '$a->strings[\'' . $zone . '\'] = \'' . $zone . '\';' . "\n";
-
-	echo $s;