#!/bin/php -c..
<?php

/*
	HTTP file download test program
	===============================

	Basically, we produce the reply

		Content-Type: MIME_TYPE
		Content-Disposition: attachment; size=FILE_SIZE;
			filename[*]=FILE_NAME_PROPERLY_ENCODED

		BINARY_DATA_HERE

	The problems we are trying to solve are:

	- how FILE_NAME_PROPERLY_ENCODED can be encoded to support UTF-8

	- which encodings are understood by each browser

	- which are the characters that the browser misunderstand
	  for example, "File 3/4.txt" becomes "4.txt" rather that
	  being translated preserving as much chars as might be done,
	  for example giving something like "File 3_4.txt".

	References:
	-----------
	RFC 2616 (HTTP 1.1)
	RFC 2047 (MIME header)
	RFC 2183 (Content-Disposition: filename*=...)
	RFC 2231 (Content-Disposition)
*/

/*. require_module 'standard';
	require_module 'pcre';
	require_module 'mbstring';
.*/

# File di test:
define("FILE_NAME", "Caff\xC3\xA8 Brill\xC3\xAC 11/8/06 12:00");
define("FILE_NAME_ENCODING", "UTF-8");
define("FILE_TYPE", "application/pdf");
define("FILE_PATH", "/home/www.icosaedro.it/public_html/acm/BA.pdf");

# Caratteri speciali per il filtro:
define("INVALID_CHARS", ':\\/*?"<>|');
define("INVALID_CHARS_REGEX", '/[:\\x5c\\/*?"<>|]/');


/*. string .*/ function remove_ctrls(/*. string .*/ $s)
# Remove CTRLs chars ASCII codes 0-31,127.
{
	return preg_replace("/[\\000-\\037\\177]/", "", $s);
}


/*. string .*/ function escape_doublequote_backslash(/*. string .*/ $s)
{
	$s = (string) str_replace("\\", "\\\\", $s);
	$s = (string) str_replace("\"", "\\\"", $s);
	return $s;
}


/*. string .*/ function get_extension(/*. string .*/ $fn)
/*
	Returns the extension from the file name $fn. Valid extensions
	are any sequence of alphanumerical ASCII chars following the
	last period, ranging from one char up to 4. Returns NULL if no
	valid extension was found. Examples:

	"SOMETHING.XyZ"    -->  "XyZ"
	"SOME.THING.X"     -->  "X"
	"SOMETHING"        -->  NULL  (missing extension)
	"SOMETHING..."     -->  NULL  (empty extension after last period)
	"Today 12.08.2006" -->  "2006"
	"photo.album"      -->  NULL  (invalid extension: too long)
*/
{
	$i = strrpos($fn, ".");
	if( $i === FALSE )  return NULL;
	$ext = substr($fn, (int) $i + 1);
	if( preg_match("/^[a-zA-Z0-9]{1,4}\$/D", $ext) == 0 )
		return NULL;
	return $ext;
}


$extensions = array(
	# See http://www.webmaster-toolkit.com/mime-types.shtml
	# for a complete list.
	"application/octet-stream", "bin",
	"application/octet-stream", "dat",
	"text/plain", "txt",
	"text/plain", "text",
	"text/html", "htm",
	"text/html", "html",
	"image/gif", "gif",
	"image/png", "png",
	"image/jpeg", "jpg",
	"image/jpeg", "jpeg",
	"application/pdf", "pdf",
	"application/ps", "ps",
	"video/avi", "avi",
);


/*. array[int]string .*/ function extension_to_mimes(/*. string .*/ $ext)
{
	global $extensions;

	$mime_list = /*.(array[int]string).*/ array();
	$n = count($extensions);
	for($i=0; $i<$n; $i+=2)
		if( $extensions[$i+1] === $ext )
			$mime_list[] = $extensions[$i];
	return $mime_list;
}


/*. array[int]string .*/ function mime_to_extensions(/*. string .*/ $mime)
{
	global $extensions;

	$ext_list = /*.(array[int]string).*/ array();
	$n = count($extensions);
	for($i=0; $i<$n; $i+=2)
		if( $extensions[$i] === $mime )
			$ext_list[] = $extensions[$i+1];
	return $ext_list;
}


/*. string .*/ function ensure_proper_extension(
	/*. string .*/ $fn,
	/*. string .*/ $mime)
/*
	Returns the file name $fn with a proper extension corresponding to
	the MIME type $mime. If the extension isn't valid or is missing,
	add one. If the MIME type is unknown, do nothing and returns the
	filename as is.
*/
{
	$exp = mime_to_extensions($mime);
	if(	count($exp) == 0 )
		return $fn;

	$cur = strtolower( get_extension($fn) );
	if( in_array($cur, $exp) )
		return $fn;
	return $fn . "." . $exp[0];
}


$charsets = array("US-ASCII", "ISO_8859-1", "UTF-8");


class Status {
	public $file_name = FILE_NAME; /* filename, UTF-8 */
	public $charset = 0;     /* global $charsets[$this-charset] is the required
	                            charset for the filename to be sent via HTTP */
	public $filter = TRUE;  /* remove special chars */
	public $ensure_extension = TRUE;
	public $encoding = 0;

}


/*. string .*/ function build_self_url(/*. Status .*/ $s, /*. string .*/ $op)
{
	return $_SERVER["PHP_SELF"]
		. "?filename=" . rawurlencode($s->file_name)
		. "&charset=" . $s->charset
		. ($s->filter? "&filter=1":"")
		. ($s->ensure_extension? "&ensure_extension=1":"")
		. "&encoding=" . $s->encoding
		. "&op=" . $op;
}


/*. Status .*/ function get_status()
{
	$s = new Status();
	if( isset($_REQUEST['filename']) ){
		$fn = remove_ctrls( (string) $_REQUEST['filename'] );
		$fn = mb_convert_encoding($fn, "UTF-8", "UTF-8");
		$s->file_name = $fn;
	} else {
		return $s;
	}
	if( isset($_REQUEST['charset']) ){
		$i = (int) $_REQUEST['charset'];
		if( $i >= 0 and $i < count($GLOBALS['charsets']) )
			$s->charset = $i;
	}
	$s->filter = isset( $_REQUEST['filter'] );
	$s->ensure_extension = isset( $_REQUEST['ensure_extension'] );
	if( isset($_REQUEST['encoding']) ){
		$i = (int) $_REQUEST['encoding'];
		if( $i >= 0 and $i < 4 )
			$s->encoding = $i;
	}
	return $s;
}


/*. void .*/ function page_header(/*. string .*/ $title)
{
	header("Content-Type: text/html; charset=UTF-8");
	header("Cache-Control: no-cache");
	echo "<html><head><title>$title</title></head>",
		"<body link=#0000cc vlink=#0000cc alink=#cc0000>",
		"<h1>$title</h1>";
}


/*. void .*/ function page_footer()
{
	echo "<hr><p align=right><i>File Download Test Program</i></p></body></html>";
}


/*. void .*/ function echo_radio(
	/*. string .*/ $name,
	/*. int .*/ $value,
	/*. int .*/ $cur_value,
	/*. string .*/ $text)
{
	$label = $name . "_" . $value;
	echo "<input type=radio name=$name value=$value id=$label";
	if( $value === $cur_value )
		echo " checked";
	echo "> <label for=", $label, ">", $text, "</label>";
}


/*. void .*/ function echo_checkbox(
	/*. string .*/ $name,
	/*. int .*/ $value,
	/*. int .*/ $cur_value,
	/*. string .*/ $text)
{
	$label = $name . "_" . $value;
	echo "<input type=checkbox name=$name value=$value id=$label";
	if( $value === $cur_value )
		echo " checked";
	echo "> <label for=", $label, ">", $text, "</label>";
}


/*. void .*/ function echo_submit(/*. string .*/ $text, /*. string .*/ $func)
{
	echo "<input type=submit name=button_$func value=\"$text\">";
}


/*. void .*/ function page_menu(/*. Status .*/ $s)
{
	page_header("Configurazione File Download");

	echo "<form method=post enctype='multipart/form-data' action=\"",
		$_SERVER["PHP_SELF"], "\">";

	echo "<table cellspacing=0 cellpadding=10><tr><td valign=top width='50%'>";



	echo "Charset per il nome del file:<hr>",
	     "<blockquote>";
	foreach( $GLOBALS['charsets'] as $k => $v ){
		echo_radio("charset", $k, $s->charset, $v);  echo "<p>";
	}
	echo "</blockquote>";



	echo "</td><td valign=top>";



	echo "Codifica per il nome del file:<hr>",
		"<blockquote>";
	echo_radio("encoding", 0, $s->encoding, "Raw (no encoding)");  echo "<p>";
	echo_radio("encoding", 1, $s->encoding, "MIME Header (RFC 2047)");  echo "<p>";
	echo_radio("encoding", 2, $s->encoding, "Enhanced filename* parameter (RFC 2231)");  echo "<p>";
	echo_radio("encoding", 3, $s->encoding, "URL Encoded (not standard)");  echo "<p>";
	echo "</blockquote>";

	echo "</td></tr></table>";

	echo_checkbox("filter", 1, $s->filter? 1:0,
		"Filtra i caratteri <code>" . htmlspecialchars(INVALID_CHARS)
		. "</code> e convertili in <code>_</code>");  echo "<p>";

	echo_checkbox("ensure_extension", 1, $s->ensure_extension? 1:0,
		"Assicura una estensione appropriata al tipo MIME del file");
	echo "<p>";


	echo "Nome del file: <input type=text name=filename size=50 value=\"",
		htmlspecialchars($s->file_name), "\"><p>";

	echo "<center>";
	echo_submit("Download", "page_download");
	echo "</center>";

	echo "</form>";

	echo "<hr><h2>Note</h2>";


	echo "<b>Impostazioni consigliate.</b> ";
	$agent = isset($_SERVER["HTTP_USER_AGENT"])? $_SERVER["HTTP_USER_AGENT"] : "";
	if( is_int(strpos($agent, "MSIE")) ){
		echo <<< EOT
		Il browser che stai usando &egrave; stato identificato come Microsoft
		Internet Explorer. Per questo browser le impostazioni consigliate
		sono:
		charset <u>UTF-8</u>,
		codifica <u>URL Encoded</u>,
		<u>filtro abilitato</u>,
		assicura estensione... <u>abilitato</u>.
EOT;

	} else if( is_int(strpos($agent, "Gecko")) ){
		echo <<< EOT
		Il browser che stai usando &egrave; stato identificato come Gecko
		(Mozilla, Firefox).
		Per questo browser le impostazioni consigliate sono:
		charset <u>UTF-8</u>,
		codifica <u>Enhanced filename* parameter</u>,
		filtro <u>disabilitato</u>,
		assicura estensione... <u>abilitato</u>.
EOT;

	} else if( is_int(strpos($agent, "Opera")) ){
echo <<< EOT
Il browser che stai usando &egrave; stato identificato come Opera.
Per questo browser le impostazioni consigliate sono:
charset <u>UTF-8</u>,
codifica <u>Enhanced filename* parameter</u>,
filtro <u>abilitato</u>,
assicura estensione... <u>disabilitato</u>.
EOT;
	} else {
echo <<< EOT
Il tuo browser "<code>$agent</code>" non &egrave; riconosciuto da questo
programma. L'impostazione consigliata in questo caso &egrave;:
charset <u>ASCII</u>,
codifica <u>raw</u>,
filtro <u>abilitato</u>,
assicura estensione <u>abilitato</u>.
EOT;
	}

	/*
	echo "<p><i>Il bottone</i> Salva <i>qui sotto in realt&agrave;",
		" non fa nulla, aggiorna solo la pagina. Puoi passare direttamente",
		" al test pi&ugrave; sotto.</i>\n",
		"<center>";
	echo_submit("Salva", "page_save");
	echo "</center>\n<hr>\n";
	*/

	echo
		"Una volta impostata la configurazione, prova un nome di file",
		" `patologico' per eseguire il test.",
		" Spesso i browser hanno problemi con questi caratteri, ai quali",
		" attribuiscono un significato speciale:<blockquote><code>",
		"<i>SPAZIO</i> ! \" # \$ % &amp; ' () * + , - . / : ; &lt; = &gt; ? @ [ \\ ] ^ _ ` { | } ~",
		"</code></blockquote>",
		" Il file di test ti verr&agrave; inviato con questo nome",
		" per verificare se le impostazioni sono adatte al tuo browser.<p>";

echo <<< EOT
<p><b>Problemi?</b> Se non riesci ad ottenere un risultato decente, prova
le impostazioni suggerite qui sopra. L'impostazione charset <u>ASCII</u>,
filtro <u>attivo</u>, codifica <u>raw</u> dovrebbe funzionare anche con i
browser pi&ugrave; antiquati.
EOT;

	echo <<< EOT
<p><b>Sorgente di questo programma.</b>
Il sorgente di questo programma di test &egrave; scritto nel linguaggio PHP.
Il meccanismo per il download del sorgente usa... s&egrave; stesso!
basta cliccare sull'ancora qui sotto. Hai impostato bene la configurazione? ;-)
</p>
EOT;

	echo "<center><a href=\"", build_self_url($s, "download_source"),
		"\">SCARICA SORGENTE</a></center>";

	echo <<< EOT
<p><b>Riferimenti.</b> Per spiegazioni pi&ugrave; articolate sul meccanismo
di download dei file via WEB, leggi l'articolo <i>PHP File Download</i>
(<a href="http://www.icosaedro.it/articoli/php-file-download.html">www.icosaedro.it/articoli/php-file-download.html</a>).
EOT;

	page_footer();
}


/*. array[int]string .*/ function build_http_headers(
	/*. Status .*/ $s,
	/*. string .*/ $filename,
	/*. string .*/ $mime,
	/*. string .*/ $path)
{
	$fn = $filename;

	# Convert filename to the requested charset:
	$fn_charset = $GLOBALS['charsets'][$s->charset];
	$fn = mb_convert_encoding($fn, $fn_charset, FILE_NAME_ENCODING);

	# Filter invalid characters:
	if( $s->filter ){
		$fn = preg_replace(INVALID_CHARS_REGEX, '_', $fn);
	}

	# Ensure proper extension:
	if( $s->ensure_extension ){
		$fn = ensure_proper_extension($fn, $mime);
	}

	# Generate filename[*]=... HTTP parameter:
	switch($s->encoding){
		case 0:
			# Raw encoding:
			$fn = escape_doublequote_backslash($fn);
			$fn = "filename=\"$fn\"";
			break;

		case 1:
			# MIME header encoding (RFC 2047):
			$fn = "filename=\"=?$fn_charset?B?" . base64_encode($fn) . "?=\"";
			# FIXME: RFC 2047 states that there is a 75 chars length limit;
			# no problems with 100+ on Mozilla.
			break;
			
		case 2:
			# Enhanced encoding (RFC 2231):
			$fn = "filename*=$fn_charset''" . rawurlencode($fn);
			break;
			
		case 3:
			$fn = "filename=" . rawurlencode($fn);
			break;

		default: exit("unexpected encoding " . $s->encoding);
	}

	$a[0] = "Content-Type: $mime";

	$a[1] = "Content-Disposition: attachment"
		#. "; size=" . filesize($path)  # <-- FIXME: not supported (2183)
		. "; " . $fn;

	$a[2] = "Content-Length: " . filesize($path);

	$a[3] = "Cache-Control: no-cache";  # FIXME

	return $a;
}


/*. string .*/ function string_to_literal(/*. string .*/ $s)
{
	return '"' . addcslashes($s, "\000..\037\"\${\177..\377") . '"';
}


/*. void .*/ function page_download(/*. Status .*/ $s)
{
	page_header("Configurazione - Download del File di Test");

	echo "<p>Il file di test &egrave; un documento PDF di circa 5 KB.</p>";

	$a = build_http_headers($s, $s->file_name, FILE_TYPE, FILE_PATH);
	echo "Header HTTP che verr&agrave; usato: <blockquote><code>";
	foreach($a as $v)
		echo "header(" . htmlspecialchars( string_to_literal($v) ), ");<br>\n";
	echo "</code></blockquote>";

	echo "<a href=\"", build_self_url($s, "download_file"),
		"\">SCARICA IL FILE DI TEST<br></a>",
		" Alcuni browser si comportano male nella gestione dei caratteri,",
		" dei tipi o delle estensioni e propongono di salvare il file",
		" con un nome non adatto. Verifica che:<ul>",
		" <li>il nome del file corrisponda a quello impostato (<code>",
			htmlspecialchars( $s->file_name ), ")</code>",
		" <li>il tipo MIME del file venga mostrato correttamente (",
			FILE_TYPE, ")",
		" <li>l'estensione del file, se necessaria, sia quella prevista (",
			strtoupper( implode(", ", mime_to_extensions(FILE_TYPE) ) ),
			").</ul>",
		" Nota che il nome potrebbe essere intenzionalmente modificato",
		" dal programma a seconda delle opzioni di filtro che hai abilitato.</p>";

	echo "<p><a href=\"", build_self_url($s, "page_menu"),
		"\">RIPROVA CON NUOVI PARAMETRI</a><br>",
		" Se il risultato non &egrave; soddisfacente, prova con altri parametri.",
		" La maschera per la configurazione riporta in fondo",
		" le impostazioni suggerite in base al browser che stai usando.";

	page_footer();
}


/*. void .*/ function download_file(
	/*. Status .*/ $s,
	/*. string .*/ $file_name,
	/*. string .*/ $file_type,
	/*. string .*/ $file_path)
{
	$a = build_http_headers($s, $file_name, $file_type, $file_path);

	foreach($a as $v)
		header($v);

	readfile($file_path);
}


#
# MAIN
#


# Sets the char that substitute those that cannot be converted toward
# the intended charset:
mb_substitute_character(ord("_"));

# Restore info from sticky form:
$s = get_status();

#
# Request handler:
#

if( $_SERVER['REQUEST_METHOD'] === "GET" && isset($_GET['op']) ){
	$op = (string) $_GET['op'];

} else if( $_SERVER['REQUEST_METHOD'] === "POST" ){
	$op = NULL;
	foreach( /*. (array[string]string) .*/ $_POST as $k => $v ){
		if( substr($k, 0, 7) === "button_" ){
			$op = substr($k, 7);
			break;
		}
	}
	if( $op === NULL ){
		trigger_error("missing button_* in POST");
		$op = "page_menu";
	}

} else {
	$op = "page_menu";
}

switch($op){
	case "page_menu":      page_menu($s);     break;

	case "page_save":      page_menu($s);     break;

	case "page_download":  page_download($s); break;

	case "download_file":
		download_file($s, $s->file_name, FILE_TYPE, FILE_PATH);
		break;

	case "download_source":
		$fn = basename( $_SERVER["SCRIPT_FILENAME"] );
		$s->file_name = $fn;
		download_file($s, $fn, "text/plain", $fn);
		break;

	default:               page_menu($s);     break;
}


?>