io_getSizeFile returns uncompressed size of given file

A bz2-file doesn't contain information about the size of its
uncompressed content. Therefore it requires reading the whole file to
obtain the filesize.

_test/tests/inc/io_getSizeFile.test.php

+<?php
+
+class io_getSizeFile_test extends DokuWikiTest {
+
+    /*
+     * dependency for tests needing zlib extension to pass
+     */
+    public function test_ext_zlib() {
+        if (!DOKU_HAS_GZIP) {
+            $this->markTestSkipped('skipping all zlib tests.  Need zlib extension');
+        }
+    }
+
+    /*
+     * dependency for tests needing zlib extension to pass
+     */
+    public function test_ext_bz2() {
+        if (!DOKU_HAS_BZIP) {
+            $this->markTestSkipped('skipping all bzip2 tests.  Need bz2 extension');
+        }
+    }
+
+    function test_plain(){
+        // since git converts line endings, we can't check in this test file but have to create it ourselves
+        $plain = TMP_DIR.'/test.txt';
+        file_put_contents($plain, "The\015\012Test\015\012");
+
+        $this->assertEquals(11, io_getSizeFile($plain));
+        $this->assertEquals(0, io_getSizeFile(__DIR__.'/io_readfile/nope.txt'));
+        $plain_mb = TMP_DIR.'/test.txt';
+        io_saveFile($plain_mb, "string with utf-8 chars åèö - doo-bee doo-bee dooh\012");
+        $this->assertEquals(54, io_getSizeFile($plain_mb));
+    }
+
+    /**
+     * @depends test_ext_zlib
+     */
+    function test_gzfiles(){
+        $this->assertEquals(11, io_getSizeFile(__DIR__.'/io_readfile/test.txt.gz'));
+        $this->assertEquals(0, io_getSizeFile(__DIR__.'/io_readfile/nope.txt.gz'));
+        $this->assertEquals(11, io_getSizeFile(__DIR__.'/io_readfile/corrupt.txt.gz'));
+        $gz_mb = TMP_DIR.'/test.txt.gz';
+        io_saveFile($gz_mb, "string with utf-8 chars åèö - doo-bee doo-bee dooh\012");
+        $this->assertEquals(54, io_getSizeFile($gz_mb));
+    }
+
+    /**
+     * @depends test_ext_bz2
+     */
+    function test_bzfiles(){
+
+        $this->assertEquals(11, io_getSizeFile(__DIR__.'/io_readfile/test.txt.bz2'));
+        $this->assertEquals(0, io_getSizeFile(__DIR__.'/io_readfile/nope.txt.bz2'));
+        $this->assertEquals(0, io_getSizeFile(__DIR__.'/io_readfile/corrupt.txt.bz2'));
+        $this->assertEquals(9720, io_getSizeFile(__DIR__.'/io_readfile/large.txt.bz2'));
+        $this->assertEquals(17780, io_getSizeFile(__DIR__.'/io_readfile/long.txt.bz2'));
+        $bz_mb = TMP_DIR.'/test.txt.bz2';
+        io_saveFile($bz_mb, "string with utf-8 chars åèö - doo-bee doo-bee dooh\012");
+        $this->assertEquals(54, io_getSizeFile($bz_mb));
+    }
+
+}

inc/io.php

@@ -776,3 +776,52 @@ function io_grep($file,$pattern,$max=0,$backref=false){
     return $matches;
 }
 
+
+/**
+ * Get size of contents of a file, for a compressed file the uncompressed size
+ * Warning: reading uncompressed size of content of bz-files requires uncompressing
+ *
+ * @author  Gerrit Uitslag <klapinklapin@gmail.com>
+ *
+ * @param string $file filename path to file
+ * @return int size of file
+ */
+function io_getSizeFile($file) {
+    if (!file_exists($file)) return 0;
+
+    if(substr($file,-3) == '.gz'){
+        if(!DOKU_HAS_GZIP) return 0;
+
+        $fp = @fopen($file, "rb");
+        if($fp === false) return 0;
+
+        fseek($fp, -4, SEEK_END);
+        $buffer = fread($fp, 4);
+        fclose($fp);
+        $array = unpack("V", $buffer);
+        $uncompressedsize = end($array);
+    }else if(substr($file,-4) == '.bz2'){
+        if(!DOKU_HAS_BZIP) return 0;
+
+        $bz = bzopen($file,"r");
+        if($bz === false) return 0;
+
+        $uncompressedsize = 0;
+        while (!feof($bz)) {
+            //8192 seems to be the maximum buffersize?
+            $buffer = bzread($bz,8192);
+            if(($buffer === false) || (bzerrno($bz) !== 0)) {
+                return 0;
+            }
+            if (UTF8_MBSTRING) {
+                $uncompressedsize += mb_strlen($buffer, '8bit');
+            } else {
+                $uncompressedsize += strlen($buffer);
+            }
+        }
+    }else{
+        $uncompressedsize = filesize($file);
+    }
+
+    return $uncompressedsize;
+ }