From c11af12551e44144c21e4f24628f63efcf81c188 Mon Sep 17 00:00:00 2001
From: Christophe HENRY <christophe.henry@sbgodin.fr>
Date: Sun, 11 Apr 2021 21:35:55 +0200
Subject: [PATCH] Adds the url path

* The /htmgem directory can be renamed and moved.
* Everything works without URL rewriting enabled.
---
 css/base.css       | 27 ++++++++++++--
 css/htmgem.css     | 41 ++++++++++++++------
 index.php          | 49 +++++++++++++++---------
 lib-htmgem.inc.php | 93 +++++++++++++++++++++++++++++++---------------
 lib-html.inc.php   | 79 ++++++++++++++++++++++++++++++++-------
 tests/miscTest.php | 80 ++++++++++++++++++++++++++++++---------
 6 files changed, 276 insertions(+), 93 deletions(-)

diff --git a/css/base.css b/css/base.css
index a925733..ef6993f 100644
--- a/css/base.css
+++ b/css/base.css
@@ -2,7 +2,7 @@ html {
     font-family: sans-serif;
 }
 
-body {
+#gmi {
     max-width: 1024px;
     margin: auto;
     margin-top: 0.5em;
@@ -33,13 +33,26 @@ a {
     text-decoration: none;
 }
 
-a {
-    text-decoration: none;
+.menu:nth-of-type(1) .menu-line {
+    text-align: left;
+}
+.menu:nth-of-type(3) .menu-line {
+    text-align: right;
+}
+
+.menu a, .menu a:visited {
+    #color: #888;
+}
+.menu a:hover {
+    #color: #000;
+}
+.menu hr {
+    border: 1px solid lightgrey;
 }
 
 @media only screen and (max-width: 1024px) {
     body {
-        margin: 3rem;
+        margin: 0.5rem 3rem;
     }
     h1 {
         font-size: 4rem;
@@ -53,4 +66,10 @@ a {
     p, pre, ul, blockquote {
         font-size: 2.6rem;
     }
+    .menu {
+        font-size: 2rem;
+    }
+    .menu hr {
+        border: 1px solid gray;
+    }
 }
diff --git a/css/htmgem.css b/css/htmgem.css
index d1db5fb..8949208 100644
--- a/css/htmgem.css
+++ b/css/htmgem.css
@@ -17,47 +17,66 @@ blockquote {
     padding: 1rem;
 }
 
-a {
-    margin: -1.35rem;
+.menu a, .menu a:visited {
+    color: #888;
+}
+
+.menu a:hover {
+    color: #000;
+}
+
+.menu hr {
+    color: white;
+}
+
+#gmi a {
+    margin: -0.7rem;
     color:#820;
 }
 
-a:before {
+#gmi a:before {
     content: "🔗 ";
 }
 
-a:visited {
+#gmi a:visited {
     color: #868;
 }
 
-a.local:before {
+#gmi a.local:before {
     content: "🛩️ ";
     font-weight: bold;
 }
 
-a.gemini:before {
+#gmi a.gemini:before {
     content: "🚀 ";
 }
 
-a.gopher:before {
+#gmi a.gopher:before {
     content: "📜 ";
 }
 
-a.https:before {
+#gmi a.https:before {
     content: "🕸️ ";
     font-weight: bolder;
 }
 
-a.http:before {
+#gmi a.http:before {
     content: "🕸️ ";
     font-weight: lighter;
 }
 
-a.mumble:before {
+#gmi a.mumble:before {
     content: "🎤 ";
 }
 
-a.mailto:before {
+#gmi a.mailto:before {
     content: "✉️ ";
 }
 
+@media only screen and (max-width: 1024px) {
+    
+    #gmi a {
+        margin: -2.9rem;
+    }
+}
+
diff --git a/index.php b/index.php
index cc4b7a9..2f85b0f 100644
--- a/index.php
+++ b/index.php
@@ -4,7 +4,11 @@ require_once "lib-htmgem.inc.php";
 require_once "lib-html.inc.php";
 require_once "lib-io.inc.php";
 
-# The url argument is always absolute compared to the document root.
+$documentRoot = $_SERVER['DOCUMENT_ROOT'];
+$scheme = (@$_SERVER['REQUEST_SCHEME']??"http")."://";
+$domain = $_SERVER['HTTP_HOST'];
+$php_self = $_SERVER['PHP_SELF']; // by default: /htmgem/index.php
+$php_self_dir = dirname($php_self);
 $url = @$_REQUEST["url"];
 $urlRewriting = @$_REQUEST["rw"]=="1";
 
@@ -18,18 +22,21 @@ $urlRewriting = @$_REQUEST["rw"]=="1";
 if (empty($url)) {
     if (!file_exists("index.gmi")) {
         http_response_code(403);
-        die("<!-- index.gmi missing -->");
+    } else {
+        $gt_html = new \htmgem\GemTextTranslate_html(file_get_contents("index.gmi"), true, "$php_self?url=", $php_self_dir);
+        if (empty($gt_html->getCss)) $gt_html->addCss($php_self_dir."/css/htmgem.css");
+
+        // No URL Rewritting assumed
+        echo \htmgem\html\getHtmlWithMenu($gt_html, $scheme, $domain, $php_self, "$php_self?url=");
     }
-    $gt_html = new \htmgem\GemTextTranslate_html(@file_get_contents("index.gmi"), true, "/htmgem");
-    echo \htmgem\html\getFullHtml($gt_html);
     exit();
 }
 
-$documentRoot = $_SERVER['DOCUMENT_ROOT'];
-
-/**
- * Provides index.gmi if no page given
- */
+$url = \htmgem\resolve_path(
+    // Some webservers (Apache) don't add the slash
+    // while others (Nginx) do…
+    ( $url[0] == "/" ? "" : "/" ) . $url
+);
 if (!preg_match("/\.gmi$/", $url)) {
     if ($url[-1] == "/")
         $url = $url."index.gmi";
@@ -55,9 +62,13 @@ switch(true) {
 if ($go404) {
     error_log("HtmGem: 404 $url $filePath");
     http_response_code(404);
-    $page404 = \htmgem\html\get404GmiPage("Page not found", $url);
+    $page404 = \htmgem\html\get404GmiPage($url);
     $gt_html = new \htmgem\GemTextTranslate_html($page404);
-    echo \htmgem\html\getFullHtml($gt_html);
+    if (empty($gt_html->getCss)) $gt_html->addCss($php_self_dir."/css/htmgem.css");
+    if ($urlRewriting)
+        echo \htmgem\html\getHtmlWithMenu($gt_html, $scheme, $domain, $url);
+    else
+        echo \htmgem\html\getHtmlWithMenu($gt_html, $scheme, $domain, $url, "$php_self?url=");
     exit();
 }
 
@@ -101,12 +112,12 @@ EOL;
 }
 
 if ($urlRewriting)
-    $baseUrl = null;
+    $gt_html = new \htmgem\GemTextTranslate_html($fileContents, $gt_htmlextDecoration);
 else
-    $baseUrl = dirname($url);
-$gt_html = new \htmgem\GemTextTranslate_html($fileContents, $gt_htmlextDecoration, $baseUrl);
+    $gt_html = new \htmgem\GemTextTranslate_html($fileContents, $gt_htmlextDecoration, "$php_self?url=", dirname($url));
+
 if ("none" == $style) {
-    $gt_html->addCss("");
+    #$gt_html->addCss("");
 } elseif ("/" == @$style[0]) {
     $gt_html->addCss($style);
 } elseif (empty($style)) {
@@ -119,9 +130,13 @@ if ("none" == $style) {
         $gt_html->addCss($localCss);
     }
 } else { #TODO: regex check for $style
-    $gt_html->addCss("/htmgem/css/$style.css");
+    $gt_html->addCss("$php_self_dir/css/$style.css");
 }
+if (empty($gt_html->getCss)) $gt_html->addCss($php_self_dir."/css/htmgem.css");
 
-echo \htmgem\html\getFullHtml($gt_html);
+if ($urlRewriting)
+    echo \htmgem\html\getHtmlWithMenu($gt_html, $scheme, $domain, $url);
+else
+    echo \htmgem\html\getHtmlWithMenu($gt_html, $scheme, $domain, $url, "$php_self?url=");
 
 ?>
diff --git a/lib-htmgem.inc.php b/lib-htmgem.inc.php
index cc305ec..dea91a6 100644
--- a/lib-htmgem.inc.php
+++ b/lib-htmgem.inc.php
@@ -5,6 +5,55 @@ namespace htmgem;
 mb_internal_encoding("UTF-8");
 mb_regex_encoding("UTF-8");
 
+/**
+ * Resolve $path interpretating / . and ..
+ * @param $path str
+ * @returns "/" if .. goes above the limit
+ */
+function resolve_path($path) {
+    if (empty($path)) return "";
+    $absolute = "/"==$path[0];
+    $parts = array_filter(explode("/", $path), 'strlen');
+    $chuncks = array();
+    foreach ($parts as $part) {
+        if ('.' == $part) continue;
+        if ('..' == $part) {
+            if (is_null(array_pop($chuncks))) return "/";
+        } else {
+            $chuncks[] = $part;
+        }
+    }
+    $output = implode("/", $chuncks);
+    if ($absolute) $output = "/".$output;
+    return $output;
+}
+
+/**
+ * Splits link (without .. or .) into parts along with direct url access.
+ * @param url
+ *
+ * Ex. /dir1/dir2/page.gmi
+ * --> "dir1" --> "/dir1"
+ * --> "dir2" --> "/dir1/dir2"
+ * --> "page.gmi" --> "/dir2/page.gmi"
+ */
+function split_path_links($path, $prefix="") {
+    $parts = array_filter(explode("/", $path), 'strlen');
+    if (empty($parts)) return array();
+    if ("/"==$path[0])
+        $stack = "/";
+    else
+        $stack = "";
+    $output = array();
+    $slash = "";
+    foreach ($parts as $part) {
+        $stack .= $slash.$part;
+        $output[$part] = $prefix.$stack;
+        $slash = "/";
+    }
+    return $output;
+}
+
 /**
  * Parses the gemtext and generates the internal format version
  * @param str $fileContents the gemtext to parse
@@ -12,7 +61,7 @@ mb_regex_encoding("UTF-8");
 function gemtextParser($fileContents) {
     if (empty($fileContents)) return array();
     $fileContents = rtrim($fileContents); // removes last empty line
-    $fileLines = mb_split("\R", $fileContents); // Unix, Mac, Windows line feeds
+    $fileLines = mb_split("\n|\r\n?", $fileContents); // Unix, Mac, Windows line feeds
     $mode = null;
     $current = array();
     foreach ($fileLines as $line) {
@@ -182,8 +231,15 @@ class GemtextTranslate_html {
     protected $pageTitle = "";
     public $translatedGemtext;
 
-    function __construct($parsedGemtext, $textDecoration=true, $baseUrl=Null) {
-        $this->baseUrl = $baseUrl;
+    /**
+     * @param $parsedGemtext the gemtext internal format
+     * @param $textDecoration bool to interpret or not the text decoration
+     * @param $urlPrefix the prefix to prepend if the URL rewriting is not on
+     * @param $currentPageDir the current directory, to be used without URL rewriting
+     */
+    function __construct($parsedGemtext, $textDecoration=true, $urlPrefix=null, $currentPageDir=null) {
+        $this->urlPrefix = $urlPrefix;
+        $this->currentPageDir = $currentPageDir;
         if (empty($parsedGemtext)) $parsedGemtext = "";
         // to delete the last empty lines
         $parsedGemtext = rtrim($parsedGemtext);
@@ -266,29 +322,6 @@ class GemtextTranslate_html {
         $text = preg_replace("/  +/", " ", $text);
     }
 
-    /**
-     * Resolve $path interpretating / . and ..
-     * @param $path str
-     * @returns "/" if .. goes above the limit
-     */
-    public static function resolve_path($path) {
-        if (empty($path)) return "";
-        $absolute = "/"==$path[0];
-        $parts = array_filter(explode("/", $path), 'strlen');
-        $chuncks = array();
-        foreach ($parts as $part) {
-            if ('.' == $part) continue;
-            if ('..' == $part) {
-                if (is_null(array_pop($chuncks))) return "/";
-            } else {
-                $chuncks[] = $part;
-            }
-        }
-        $output = implode("/", $chuncks);
-        if ($absolute) $output = "/".$output;
-        return $output;
-    }
-
     public function translate($textDecoration=true) {
         $output = "";
         foreach ($this->parsedGemtext as $node) {
@@ -344,10 +377,10 @@ class GemtextTranslate_html {
                     preg_match("/^([^:]+):/", $link, $matches);
                     $protocol = @$matches[1]??"local";
                     if ("local"==$protocol) {
-                        if (!is_null($this->baseUrl)) { // No URL rewriting
-                            if ($link[0]!="/") $link = "{$this->baseUrl}/$link";
-                            $link = self::resolve_path($link);
-                            $link = "/htmgem/index.php?url=$link";
+                        if (!is_null($this->urlPrefix)) { // No URL rewriting
+                            $link = $this->currentPageDir."/".$link;
+                            $link = resolve_path($link);
+                            $link = $this->urlPrefix.$link;
                         }
                         $newWindow = "";
                     } else {
diff --git a/lib-html.inc.php b/lib-html.inc.php
index ec04522..ca98040 100644
--- a/lib-html.inc.php
+++ b/lib-html.inc.php
@@ -5,12 +5,10 @@ namespace htmgem\html;
 mb_internal_encoding("UTF-8");
 mb_regex_encoding("UTF-8");
 
-/**
- * Returns a full HTML page base
- */
-function getFullHtml(\htmgem\GemtextTranslate_html $gt_html) {
+$txt_icon = "H͜͡m ";
+
+function getHeader(\htmgem\GemtextTranslate_html $gt_html) {
     $css = $gt_html->getCss();
-    if (!$css) $css = array("/htmgem/css/htmgem.css");
     $output = <<<EOL
 <!DOCTYPE html>
 <html lang="">
@@ -23,21 +21,76 @@ EOL;
     }
     $output .= <<<EOL
 </head>
-<body>\n
-EOL;
-    $output .= $gt_html->translatedGemtext;
-    $output .= "</body>\n</html>\n";
 
+EOL;
     return $output;
 }
 
-function get404Gmipage($message, $url) {
+function array_key_last_slice($array) {
+    // array_key_last() only available as of php v7.3.0
+    return key(array_slice($array, -1));
+}
+
+/**
+ * @param $url the full URL to display
+ * @param $pageLink if not null, means no URL rewritting
+ */
+function getMenu(string $scheme, string $domain, string $path, string $prefix=null) {
+    global $txt_icon;
+    $links = \htmgem\split_path_links($path, $prefix);
+
+    // Removes the last part, as it won't hold a link
+    $lastLink = array_key_last_slice($links);
+    if ("index.gmi"==$lastLink) {
+        // removes the index page
+        array_pop($links);
+        $lastLink = array_key_last_slice($links);
+    }
+    array_pop($links);
+
+    $links = array($domain => "$prefix/") + $links;
+    $linkList = array();
+    foreach ($links as $label=>$link) {
+        $linkList []= "<a href='$link'>$label</a>\n";
+    }
+    $linkList [] = $lastLink."\n"; // The last part holds no link
+    $output = "<div class='menu-line'>\n";
+    $output .= "<strong>$txt_icon</strong>$scheme\n";
+    $output .= implode(" / ", $linkList);
+    $output .= "</div>\n";
+    return $output;
+}
+
+function getFooter(\htmgem\GemtextTranslate_html $gt_html) {
+    return "</body>\n</html>\n";
+}
+
+function getHtmlWithMenu($gt_html, $scheme, $domain, $path, $prefix=null) {
+    $menu = getMenu($scheme, $domain, $path, $prefix);
+    echo getHeader($gt_html);
+    echo "<body>\n";
+    echo "<div class='menu'>\n";
+    echo $menu;
+    echo "<hr>\n";
+    echo "</div>\n";
+    echo "<div id='gmi'>\n";
+    echo $gt_html->translatedGemtext;
+    echo "</div>\n";
+    echo "<div class='menu'>\n";
+    echo "<hr>\n";
+    echo $menu;
+    echo "</div>\n";
+    echo getFooter($gt_html);
+}
+
+function get404Gmipage($url) {
     return <<<EOF
-# ⚠ $message
+# ⚠ $url ⚠
+
+* **Page non trouvée**
+* **Page not found**
 
-​**$url**
 
-=> .. 🔄 🔄
 EOF;
 }
 
diff --git a/tests/miscTest.php b/tests/miscTest.php
index beb607f..2327c7a 100644
--- a/tests/miscTest.php
+++ b/tests/miscTest.php
@@ -5,89 +5,133 @@ require_once dirname(__FILE__)."/../lib-htmgem.inc.php";
 
 final class miscTest extends TestCase {
 
-    public function test_resolveLink(): void {
+    public function test_split_path_links(): void {
         $this->assertSame(
-            \htmgem\GemtextTranslate_html::resolve_path(""),
+            array(),
+            \htmgem\split_path_links(""),
+            "empty link"
+        );
+        $this->assertSame(
+            array(
+                "noslash" => "noslash",
+            ),
+            \htmgem\split_path_links("noslash"),
+            "no slash"
+        );
+        $this->assertSame(
+            array(),
+            \htmgem\split_path_links("/"),
+            "only a slash"
+        );
+        $this->assertSame(
+            array(
+                "one" => "/one",
+            ),
+            \htmgem\split_path_links("/one"),
+            "/one"
+        );
+        $this->assertSame(
+            array(
+                "one" => "one",
+                "two" => "one/two",
+            ),
+            \htmgem\split_path_links("one/two"),
+            "one/two"
+        );
+        $this->assertSame(
+            array(
+                "one" => "/one",
+                "two" => "/one/two",
+                "file.ext" => "/one/two/file.ext",
+            ),
+            \htmgem\split_path_links("/one/two/file.ext"),
+            "/one/two/file.ext"
+        );
+    }
+
+    public function test_resolve_path(): void {
+        $this->assertSame(
+            \htmgem\resolve_path(""),
             "",
             "empty link"
         );
         $this->assertSame(
-            \htmgem\GemtextTranslate_html::resolve_path("test"),
+            \htmgem\resolve_path("test"),
             "test",
             "single word"
         );
         $this->assertSame(
-            \htmgem\GemtextTranslate_html::resolve_path(" "),
+            \htmgem\resolve_path(" "),
             " ",
             "single space"
         );
         $this->assertSame(
-            \htmgem\GemtextTranslate_html::resolve_path(" A B "),
+            \htmgem\resolve_path(" A B "),
             " A B ",
             "several space"
         );
         $this->assertSame(
-            \htmgem\GemtextTranslate_html::resolve_path("/"),
+            \htmgem\resolve_path("/"),
             "/",
             "one slash"
         );
         $this->assertSame(
-            \htmgem\GemtextTranslate_html::resolve_path("//"),
+            \htmgem\resolve_path("//"),
             "/",
             "two slashes"
         );
         $this->assertSame(
-            \htmgem\GemtextTranslate_html::resolve_path("/////"),
+            \htmgem\resolve_path("/////"),
             "/",
             "five slashes"
         );
         $this->assertSame(
-            \htmgem\GemtextTranslate_html::resolve_path("one/"),
+            \htmgem\resolve_path("one/"),
             "one",
             "strip the last slash"
         );
         $this->assertSame(
-            \htmgem\GemtextTranslate_html::resolve_path("/two"),
+            \htmgem\resolve_path("/two"),
             "/two",
             "slash at the beginning"
         );
         $this->assertSame(
-            \htmgem\GemtextTranslate_html::resolve_path("/two/"),
+            \htmgem\resolve_path("/two/"),
             "/two",
             "slash at the beginning and the end"
         );
         $this->assertSame(
-            \htmgem\GemtextTranslate_html::resolve_path("one/two/"),
+            \htmgem\resolve_path("one/two/"),
             "one/two",
             "only the last slash remains"
         );
         $this->assertSame(
-            \htmgem\GemtextTranslate_html::resolve_path("one/two/three//"),
+            \htmgem\resolve_path("one/two/three//"),
             "one/two/three",
             "strip the last slashes"
         );
         $this->assertSame(
-            \htmgem\GemtextTranslate_html::resolve_path("one/../"),
+            \htmgem\resolve_path("one/../"),
             "",
             "empty one"
         );
         $this->assertSame(
-            \htmgem\GemtextTranslate_html::resolve_path("one/two/../"),
+            \htmgem\resolve_path("one/two/../"),
             "one",
             "empty one two"
         );
         $this->assertSame(
-            \htmgem\GemtextTranslate_html::resolve_path("one/two/../.."),
+            \htmgem\resolve_path("one/two/../.."),
             "",
             "empty one two twice"
         );
         $this->assertSame(
-            \htmgem\GemtextTranslate_html::resolve_path("one/../two/./../three"),
+            \htmgem\resolve_path("one/../two/./../three"),
             "three",
             "waltz"
         );
         $this->assertSame(
-            \htmgem\GemtextTranslate_html::resolve_path("one/../.."),
+            \htmgem\resolve_path("one/../.."),
             "/",
             "directory traversal"
         );