From ba1c51259c44dfdd46ef506edafa70217a717fdd Mon Sep 17 00:00:00 2001 From: Christophe HENRY Date: Thu, 11 Mar 2021 17:53:44 +0100 Subject: [PATCH 01/13] Adds Unreleased section to CHANGELOG.gmi --- CHANGELOG.gmi | 18 ++++++++++++++++++ 1 file changed, 18 insertions(+) diff --git a/CHANGELOG.gmi b/CHANGELOG.gmi index e1d3551..40df6d2 100644 --- a/CHANGELOG.gmi +++ b/CHANGELOG.gmi @@ -4,6 +4,24 @@ All notable changes to this project will be documented in this file. The format is based on [Keep a Changelog](https:/​/keepachangelog.com/en/1.0.0/), and this project adheres to [Semantic Versioning](https:/​/semver.org/spec/v2.0.0.html). +## [Unreleased] + +### Security +* Check url encoding: The filename fetched on disk may differ from that was asked by URL. +* Perform sanity checks against unauthorized access. +* Check unicode capability, UTF16 input, etc… +* Manage different type of carriage return: CR CR/LF LF + +### Development +* Command line API for script, testing, CI… +* Split codebase: library to interpret GemText, command line interface (tests, …) and web interface + +### User interface +* Page menu on upper and lower sides: [parent dir] [root dir] [raw text] +* Images +* Text localisation +* Proxy: https:/​/thewebsite.tld/htmgem/proxy/[gemini:/​/]capsule.tld/path/file.gmi + ## [1.0.0] - 2021-03-10 * Improves presentation and installation page. * Adds stylesheets and download of pages source code. From 9fb9206b7a5f48eb0b0f2e2340132cc0e71a6a71 Mon Sep 17 00:00:00 2001 From: Christophe HENRY Date: Thu, 11 Mar 2021 23:23:51 +0100 Subject: [PATCH 02/13] Adds idea about CSS file setting --- CHANGELOG.gmi | 2 ++ 1 file changed, 2 insertions(+) diff --git a/CHANGELOG.gmi b/CHANGELOG.gmi index 40df6d2..b4f4f0b 100644 --- a/CHANGELOG.gmi +++ b/CHANGELOG.gmi @@ -15,12 +15,14 @@ and this project adheres to [Semantic Versioning](https:/​/semver.org/spec/v2. ### Development * Command line API for script, testing, CI… * Split codebase: library to interpret GemText, command line interface (tests, …) and web interface +* Page caching: don’t generate twice an unmodified file. ### User interface * Page menu on upper and lower sides: [parent dir] [root dir] [raw text] * Images * Text localisation * Proxy: https:/​/thewebsite.tld/htmgem/proxy/[gemini:/​/]capsule.tld/path/file.gmi +* Search for {pages}.css from the current directory to the root. Empty css file means don’t search and go to default. ## [1.0.0] - 2021-03-10 * Improves presentation and installation page. From ef4b3b86930f7b827b1e1e4db821378d1e7fc6e1 Mon Sep 17 00:00:00 2001 From: Christophe HENRY Date: Fri, 12 Mar 2021 00:25:22 +0100 Subject: [PATCH 03/13] Adds idea in CHANGELOG --- CHANGELOG.gmi | 8 +++++--- 1 file changed, 5 insertions(+), 3 deletions(-) diff --git a/CHANGELOG.gmi b/CHANGELOG.gmi index b4f4f0b..01604f5 100644 --- a/CHANGELOG.gmi +++ b/CHANGELOG.gmi @@ -10,15 +10,17 @@ and this project adheres to [Semantic Versioning](https:/​/semver.org/spec/v2. * Check url encoding: The filename fetched on disk may differ from that was asked by URL. * Perform sanity checks against unauthorized access. * Check unicode capability, UTF16 input, etc… -* Manage different type of carriage return: CR CR/LF LF +* Manage different type of carriage return: CR CR/LF LF. ### Development * Command line API for script, testing, CI… -* Split codebase: library to interpret GemText, command line interface (tests, …) and web interface +* Split codebase: library to interpret GemText, command line interface (tests, …) and web interface. * Page caching: don’t generate twice an unmodified file. +* Properly close tags when the page exists in a non-null mode. +* Be able to move /htmgem anywhere and (for the Php part) outside the webbrowser scope. ### User interface -* Page menu on upper and lower sides: [parent dir] [root dir] [raw text] +* Page menu on upper and lower sides: [parent dir] [root dir] [raw text]. * Images * Text localisation * Proxy: https:/​/thewebsite.tld/htmgem/proxy/[gemini:/​/]capsule.tld/path/file.gmi From 64ed0a56b3957dbef29303ed3472ca88438c2531 Mon Sep 17 00:00:00 2001 From: Christophe HENRY Date: Fri, 12 Mar 2021 20:31:03 +0100 Subject: [PATCH 04/13] Adds idea in CHANGELOG --- CHANGELOG.gmi | 1 + 1 file changed, 1 insertion(+) diff --git a/CHANGELOG.gmi b/CHANGELOG.gmi index 01604f5..41218e9 100644 --- a/CHANGELOG.gmi +++ b/CHANGELOG.gmi @@ -25,6 +25,7 @@ and this project adheres to [Semantic Versioning](https:/​/semver.org/spec/v2. * Text localisation * Proxy: https:/​/thewebsite.tld/htmgem/proxy/[gemini:/​/]capsule.tld/path/file.gmi * Search for {pages}.css from the current directory to the root. Empty css file means don’t search and go to default. +* Separate project? Blog infrastructure with templates, style and RSS generator. ## [1.0.0] - 2021-03-10 * Improves presentation and installation page. From 732fd9e15a9daa554beecfe19e4ed3e2963b55e7 Mon Sep 17 00:00:00 2001 From: Christophe HENRY Date: Sat, 13 Mar 2021 10:49:19 +0100 Subject: [PATCH 05/13] Adds idea in CHANGELOG --- CHANGELOG.gmi | 2 ++ 1 file changed, 2 insertions(+) diff --git a/CHANGELOG.gmi b/CHANGELOG.gmi index 41218e9..90731a6 100644 --- a/CHANGELOG.gmi +++ b/CHANGELOG.gmi @@ -18,6 +18,8 @@ and this project adheres to [Semantic Versioning](https:/​/semver.org/spec/v2. * Page caching: don’t generate twice an unmodified file. * Properly close tags when the page exists in a non-null mode. * Be able to move /htmgem anywhere and (for the Php part) outside the webbrowser scope. +* cgi-bin? +* Chain CSS found from the root to the file. ### User interface * Page menu on upper and lower sides: [parent dir] [root dir] [raw text]. From 97e1bd9ee56ca7ff46d24e8f4df48f6a9c4f2ada Mon Sep 17 00:00:00 2001 From: Christophe HENRY Date: Sun, 14 Mar 2021 10:36:00 +0100 Subject: [PATCH 06/13] Updates CHANGELOG for v2 and v3 items --- CHANGELOG.gmi | 17 ++++++++++------- 1 file changed, 10 insertions(+), 7 deletions(-) diff --git a/CHANGELOG.gmi b/CHANGELOG.gmi index 90731a6..2e75c38 100644 --- a/CHANGELOG.gmi +++ b/CHANGELOG.gmi @@ -4,7 +4,15 @@ All notable changes to this project will be documented in this file. The format is based on [Keep a Changelog](https:/​/keepachangelog.com/en/1.0.0/), and this project adheres to [Semantic Versioning](https:/​/semver.org/spec/v2.0.0.html). -## [Unreleased] +## [Unreleased] v3 +* cgi-bin? Handle user input +* Handle client certificates? +* Page caching: don’t generate twice an unmodified file. +* Separate project? Blog infrastructure with templates, style and RSS generator. +* Proxy for remote and local servers. +* Be able to move /htmgem anywhere and (for the Php part) outside the webbrowser scope. + +## [Unreleased] v2 ### Security * Check url encoding: The filename fetched on disk may differ from that was asked by URL. @@ -15,19 +23,14 @@ and this project adheres to [Semantic Versioning](https:/​/semver.org/spec/v2. ### Development * Command line API for script, testing, CI… * Split codebase: library to interpret GemText, command line interface (tests, …) and web interface. -* Page caching: don’t generate twice an unmodified file. * Properly close tags when the page exists in a non-null mode. -* Be able to move /htmgem anywhere and (for the Php part) outside the webbrowser scope. -* cgi-bin? -* Chain CSS found from the root to the file. ### User interface * Page menu on upper and lower sides: [parent dir] [root dir] [raw text]. -* Images +* Images: click to display * Text localisation * Proxy: https:/​/thewebsite.tld/htmgem/proxy/[gemini:/​/]capsule.tld/path/file.gmi * Search for {pages}.css from the current directory to the root. Empty css file means don’t search and go to default. -* Separate project? Blog infrastructure with templates, style and RSS generator. ## [1.0.0] - 2021-03-10 * Improves presentation and installation page. From 01efc7993040e7b47dbce4a994ec2db786e4fb80 Mon Sep 17 00:00:00 2001 From: Christophe HENRY Date: Tue, 16 Mar 2021 01:08:22 +0100 Subject: [PATCH 07/13] Roadmap for v1 --- CHANGELOG.gmi | 17 ++++++++++++----- 1 file changed, 12 insertions(+), 5 deletions(-) diff --git a/CHANGELOG.gmi b/CHANGELOG.gmi index da269f5..ec1fd99 100644 --- a/CHANGELOG.gmi +++ b/CHANGELOG.gmi @@ -17,19 +17,26 @@ All notable changes to this project will be documented in this file. ### Security * Check url encoding: The filename fetched on disk may differ from that was asked by URL. * Perform sanity checks against unauthorized access. -* Check unicode capability, UTF16 input, etc… +* Check unicode capability, UTF16 input, RTL/LTR, etc… * Manage different type of carriage return: CR CR/LF LF. +### Development +∅ + +### User interface +* Page menu on upper and lower sides: [parent dir] [root dir] [raw text]. +* Images: click to display +* Proxy: https:/​/thewebsite.tld/htmgem/proxy/[gemini:/​/]capsule.tld/path/file.gmi + +## [Unreleased] v1 + ### Development * Command line API for script, testing, CI… * Split codebase: library to interpret GemText, command line interface (tests, …) and web interface. * Properly close tags when the page exists in a non-null mode. ### User interface -* Page menu on upper and lower sides: [parent dir] [root dir] [raw text]. -* Images: click to display -* Text localisation -* Proxy: https:/​/thewebsite.tld/htmgem/proxy/[gemini:/​/]capsule.tld/path/file.gmi +* Translate French texts to English. * Search for {pages}.css from the current directory to the root. Empty css file means don’t search and go to default. ## [1.1.0] - 2021-03-14 From 9726203d07004daa7c800874ea21e65695df440c Mon Sep 17 00:00:00 2001 From: Christophe HENRY Date: Tue, 16 Mar 2021 13:49:11 +0100 Subject: [PATCH 08/13] Split index.php into two files to isolate the lib --- index.php | 218 +----------------------------------------------- lib-htmgem.php | 220 +++++++++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 222 insertions(+), 216 deletions(-) create mode 100644 lib-htmgem.php diff --git a/index.php b/index.php index e30f66c..cac10d5 100644 --- a/index.php +++ b/index.php @@ -1,18 +1,7 @@ underlined. - * @param $instruction the characters to replace, ex. _ - * @param $markup the markup to replace to, ex. "u" to get … - * @param &$text where to replace. - */ -function markupPreg($instruction, $markup, &$text) { - $output = $text; - - # Replaces couples "__word__" into "word". - $output = mb_ereg_replace("${instruction}(.+?)${instruction}", "<{$markup}>\\1", $output); - - # Replaces a remaining __ into "…" to the end of the line. - $output = mb_ereg_replace("${instruction}(.+)?", "<{$markup}>\\1", $output); - - $text = $output; -} - - -/** - * Adds text attributes sucj as underline, bold, … to $line - * @param $line the line to process - */ -function addTextAttributes(&$line) { - global $textDecoration; - if (!$textDecoration) return; - markupPreg("__", "u", $line); - markupPreg("\*\*", "strong", $line); - markupPreg("//", "em", $line); - markupPreg("~~", "del", $line); -} - -/** - * Prepares the raw text to be displayed in HTML environment: - * * Escapes the HTML entities yet contained in the Gemtext. - * * Puts thin unbrakable spaces before some characters. - * @param $text1, $text2 texts to process - */ -function htmlPrepare(&$text) { - $text = htmlspecialchars($text, ENT_HTML5|ENT_NOQUOTES, "UTF-8", false); - $text = mb_ereg_replace("\ ([?!:;»€$])", NARROW_NO_BREAK_SPACE."\\1", $text); - $text = mb_ereg_replace("([«])\ ", "\\1".NARROW_NO_BREAK_SPACE, $text); # Espace fine insécable - - # Warning: using a monospace font editor may not display dashes as they should be! - # Adds no-break spaces to stick the (EM/EN dashes) to words : aaaaaa – bb – ccccc ==> aaaaaa –$bb$– ccccc - $text = mb_ereg_replace("([".DASHES."]) ([^".DASHES.".]+) ([".DASHES."])", "\\1".NARROW_NO_BREAK_SPACE."\\2".NARROW_NO_BREAK_SPACE."\\3", $text); - - # Adds no-break space to stick the (EM/EN dashes) to words : aaaaaa – bb. ==> aaaaaa –$bb. - $text = mb_ereg_replace("([—–]) ([^.]+)\.", "\\1".NARROW_NO_BREAK_SPACE."\\2.", $text); -} - -function translateGemToHtml($fileContents) { - $fileLines = preg_split("/\n/", $fileContents); - if (empty($fileLines[-1])) array_pop($fileLines); # Don't output a last empty line - ob_start(); - $mode = null; - $mode_textAttributes = true; - foreach ($fileLines as $line) { - $reDoCount = 0; - $mode_textAttributes_temp = false; - while (true) { - if ($reDoCount>2) { - error_log("HtmGem: Too many loops, mode == '$mode'"); - $mode = null; - $reDoCount = 0; - break; - } - $reDoCount += 1; - $line1 = substr($line, 0, 1); // $line can be modified - $line2 = substr($line, 0, 2); // in the meantime. - $line3 = substr($line, 0, 3); - if (is_null($mode)) { - if (empty($line)) { - echo "

 

\n"; - } elseif ('^^^' == $line3) { - $mode_textAttributes = !$mode_textAttributes; - } elseif ('^' == $line1 and !$mode_textAttributes_temp) { - if (preg_match("/^\^\s*(.+)$/", $line, $parts)) { - $line = $parts[1]; - $mode_textAttributes_temp = true; - } else { - $mode = "raw"; - } - continue; - } elseif ("#" == $line1) { - if (preg_match("/^(#{1,3})\s*(.+)/", $line, $sharps)) { - $h_level = strlen($sharps[1]); - $text = $sharps[2]; - htmlPrepare($text); - switch ($h_level) { - case 1: echo "

".$text."

\n"; break; - case 2: echo "

".$text."

\n"; break; - case 3: echo "

".$text."

\n"; break; - } - } else { - $mode = "raw"; - continue; - } - } elseif ("=>" == $line2) { - if (preg_match("/^=>\s*([^\s]+)(?:\s+(.*))?$/", $line, $linkParts)) { - $url_link = $linkParts[1]; - $url_label = @$linkParts[2]; - preg_match("/^([^:]+):/", $url_link, $matches); - $url_protocol = @$matches[1]; - if (empty($url_protocol)) $url_protocol = "local"; - if (empty(trim($url_label))) { - $url_label = $url_link; - } else { - // the label is humain-made, apply formatting - htmlPrepare($url_label); - if ($mode_textAttributes xor $mode_textAttributes_temp) addTextAttributes($url_label); - } - echo "

$url_label

\n"; - } else { - $mode = "raw"; - continue; - } - } elseif ("```" == $line3) { - preg_match("/^```\s*(.*)$/", $line, $matches); - $alt_text = trim($matches[1]); - if (empty($alt_text)) { - echo "
\n";
-                    } else {
-                        echo "
\n";
-                    }
-                    $mode="pre";
-                } elseif (">" == $line1) {
-                    echo "
\n";
-                    $mode = "quote";
-                    continue;
-                } elseif ("*" == $line1) {
-                    echo "
    \n";
-                    $mode = "ul";
-                    continue;
-                } else {
-                    $mode = "raw";
-                    continue;
-                }
-            } else {
-                if ("raw"==$mode) {
-                    if (empty($line)) {
-                        $line = " ";
-                    } else {
-                        htmlPrepare($line);
-                        if ($mode_textAttributes xor $mode_textAttributes_temp)
-                            addTextAttributes($line);
-                    }
-                    echo "
    $line
    \n";
-                    $mode = null;
-                } elseif ("pre"==$mode) {
-                    if ("```" == $line3) {
-                        echo "
\n";
-                        $mode = null;
-                    } else {
-                        htmlPrepare($line);
-                        echo $line."\n";
-                    }
-                } elseif ("quote"==$mode) {
-                    if (">" == $line1) {
-                        preg_match("/^>\s*(.*)$/", $line, $quoteParts);
-                        $quote = $quoteParts[1];
-                        if (empty($quote))
-                            echo "
 
\n";
-                        else {
-                            htmlPrepare($quote);
-                            if ($mode_textAttributes xor $mode_textAttributes_temp)
-                                addTextAttributes($line);
-                            echo "
".$quote."
\n";
-                        }
-                    } else {
-                        echo "\n";
-                        $mode = null;
-                        continue;
-                    }
-                } elseif ("ul"==$mode) {
-                    if ("*" == $line1) {
-                        preg_match("/^\*\s*(.*)$/", $line, $ulParts);
-                        $li = $ulParts[1];
-                        if (empty($li)) {
-                            echo "
  •  \n";
-                        } else {
-                            htmlPrepare($li);
-                            if ($mode_textAttributes xor $mode_textAttributes_temp) addTextAttributes($li);
-                            echo "
  • ".$li."\n";
-                        }
-                    } else {
-                        echo "\n";
-                        $mode = null;
-                        continue;
-                    }
-                } else {
-                    die("Unexpected mode: $mode!");
-                }
-            }
-            break; // exits the while(true) as no continue occured
-        } // while(true)
-    }
-    $html = ob_get_contents();
-    ob_clean();
-    return $html;
-}
-
 # Gets the page title: the first occurrence with # at the line start
 mb_ereg("#\s*([^\n]+)\n", $fileContents, $matches);
 $page_title = @$matches[1];
diff --git a/lib-htmgem.php b/lib-htmgem.php
new file mode 100644
index 0000000..0711a93
--- /dev/null
+++ b/lib-htmgem.php
@@ -0,0 +1,220 @@
+underlined.
+ * @param $instruction the characters to replace, ex. _
+ * @param $markup the markup to replace to, ex. "u" to get …
+ * @param &$text where to replace.
+ */
+function markupPreg($instruction, $markup, &$text) {
+    $output = $text;
+
+    # Replaces couples "__word__" into "word".
+    $output = mb_ereg_replace("${instruction}(.+?)${instruction}", "<{$markup}>\\1", $output);
+
+    # Replaces a remaining __ into "…" to the end of the line.
+    $output = mb_ereg_replace("${instruction}(.+)?", "<{$markup}>\\1", $output);
+
+    $text = $output;
+}
+
+/**
+ * Adds text attributes sucj as underline, bold, … to $line
+ * @param $line the line to process
+ */
+function addTextAttributes(&$line) {
+    global $textDecoration;
+    if (!$textDecoration) return;
+    markupPreg("__",   "u",   $line);
+    markupPreg("\*\*", "strong",   $line);
+    markupPreg("//",   "em",   $line);
+    markupPreg("~~",   "del", $line);
+}
+
+/**
+ * Prepares the raw text to be displayed in HTML environment:
+ * * Escapes the HTML entities yet contained in the Gemtext.
+ * * Puts thin unbrakable spaces before some characters.
+ * @param $text1, $text2 texts to process
+ */
+function htmlPrepare(&$text) {
+    $text = htmlspecialchars($text, ENT_HTML5|ENT_NOQUOTES, "UTF-8", false);
+    $text = mb_ereg_replace("\ ([?!:;»€$])", NARROW_NO_BREAK_SPACE."\\1", $text);
+    $text = mb_ereg_replace("([«])\ ", "\\1".NARROW_NO_BREAK_SPACE, $text); # Espace fine insécable
+
+    # Warning: using a monospace font editor may not display dashes as they should be!
+    # Adds no-break spaces to stick the (EM/EN dashes) to words : aaaaaa – bb – ccccc ==> aaaaaa –$bb$– ccccc
+    $text = mb_ereg_replace("([".DASHES."]) ([^".DASHES.".]+) ([".DASHES."])", "\\1".NARROW_NO_BREAK_SPACE."\\2".NARROW_NO_BREAK_SPACE."\\3", $text);
+
+    # Adds no-break space to stick the (EM/EN dashes) to words : aaaaaa – bb. ==> aaaaaa –$bb.
+    $text = mb_ereg_replace("([—–]) ([^.]+)\.", "\\1".NARROW_NO_BREAK_SPACE."\\2.", $text);
+}
+
+function translateGemToHtml($fileContents) {
+    $fileLines = preg_split("/\n/", $fileContents);
+    if (empty($fileLines[-1])) array_pop($fileLines); # Don't output a last empty line
+    ob_start();
+    $mode = null;
+    $mode_textAttributes = true;
+    foreach ($fileLines as $line) {
+        $reDoCount = 0;
+        $mode_textAttributes_temp = false;
+        while (true) {
+            if ($reDoCount>2) {
+                error_log("HtmGem: Too many loops, mode == '$mode'");
+                $mode = null;
+                $reDoCount = 0;
+                break;
+            }
+            $reDoCount += 1;
+            $line1 = substr($line, 0, 1); // $line can be modified
+            $line2 = substr($line, 0, 2); // in the meantime.
+            $line3 = substr($line, 0, 3);
+            if (is_null($mode)) {
+                if (empty($line)) {
+                    echo "
     
    \n";
+                } elseif ('^^^' == $line3) {
+                    $mode_textAttributes = !$mode_textAttributes;
+                } elseif ('^' == $line1 and !$mode_textAttributes_temp) {
+                    if (preg_match("/^\^\s*(.+)$/", $line, $parts)) {
+                        $line = $parts[1];
+                        $mode_textAttributes_temp = true;
+                    } else {
+                        $mode = "raw";
+                    }
+                    continue;
+                } elseif ("#" == $line1) {
+                    if (preg_match("/^(#{1,3})\s*(.+)/", $line, $sharps)) {
+                        $h_level = strlen($sharps[1]);
+                        $text = $sharps[2];
+                        htmlPrepare($text);
+                        switch ($h_level) {
+                            case 1: echo "
    ".$text."
    \n"; break;
+                            case 2: echo "
    ".$text."
    \n"; break;
+                            case 3: echo "
    ".$text."
    \n"; break;
+                        }
+                    } else {
+                        $mode = "raw";
+                        continue;
+                    }
+                } elseif ("=>" == $line2) {
+                    if (preg_match("/^=>\s*([^\s]+)(?:\s+(.*))?$/", $line, $linkParts)) {
+                        $url_link = $linkParts[1];
+                        $url_label = @$linkParts[2];
+                        preg_match("/^([^:]+):/", $url_link, $matches);
+                        $url_protocol = @$matches[1];
+                        if (empty($url_protocol)) $url_protocol = "local";
+                        if (empty(trim($url_label))) {
+                            $url_label = $url_link;
+                        } else {
+                            // the label is humain-made, apply formatting
+                            htmlPrepare($url_label);
+                            if ($mode_textAttributes xor $mode_textAttributes_temp) addTextAttributes($url_label);
+                        }
+                        echo "
    $url_label
    \n";
+                    } else {
+                        $mode = "raw";
+                        continue;
+                    }
+                } elseif ("```" == $line3) {
+                    preg_match("/^```\s*(.*)$/", $line, $matches);
+                    $alt_text = trim($matches[1]);
+                    if (empty($alt_text)) {
+                        echo "
    \n";
+                    } else {
+                        echo "
    \n";
+                    }
+                    $mode="pre";
+                } elseif (">" == $line1) {
+                    echo "
    \n";
+                    $mode = "quote";
+                    continue;
+                } elseif ("*" == $line1) {
+                    echo "
      \n";
+                    $mode = "ul";
+                    continue;
+                } else {
+                    $mode = "raw";
+                    continue;
+                }
+            } else {
+                if ("raw"==$mode) {
+                    if (empty($line)) {
+                        $line = " ";
+                    } else {
+                        htmlPrepare($line);
+                        if ($mode_textAttributes xor $mode_textAttributes_temp)
+                            addTextAttributes($line);
+                    }
+                    echo "
      $line
      \n";
+                    $mode = null;
+                } elseif ("pre"==$mode) {
+                    if ("```" == $line3) {
+                        echo "
    \n";
+                        $mode = null;
+                    } else {
+                        htmlPrepare($line);
+                        echo $line."\n";
+                    }
+                } elseif ("quote"==$mode) {
+                    if (">" == $line1) {
+                        preg_match("/^>\s*(.*)$/", $line, $quoteParts);
+                        $quote = $quoteParts[1];
+                        if (empty($quote))
+                            echo "
     
    \n";
+                        else {
+                            htmlPrepare($quote);
+                            if ($mode_textAttributes xor $mode_textAttributes_temp)
+                                addTextAttributes($line);
+                            echo "
    ".$quote."
    \n";
+                        }
+                    } else {
+                        echo "\n";
+                        $mode = null;
+                        continue;
+                    }
+                } elseif ("ul"==$mode) {
+                    if ("*" == $line1) {
+                        preg_match("/^\*\s*(.*)$/", $line, $ulParts);
+                        $li = $ulParts[1];
+                        if (empty($li)) {
+                            echo "
  •  \n";
+                        } else {
+                            htmlPrepare($li);
+                            if ($mode_textAttributes xor $mode_textAttributes_temp) addTextAttributes($li);
+                            echo "
  • ".$li."\n";
+                        }
+                    } else {
+                        echo "\n";
+                        $mode = null;
+                        continue;
+                    }
+                } else {
+                    die("Unexpected mode: $mode!");
+                }
+            }
+            break; // exits the while(true) as no continue occured
+        } // while(true)
+    }
+    $html = ob_get_contents();
+    ob_clean();
+    return $html;
+}
+
+?>

From 859b0aad811d730377f48abe90169b1603f95966 Mon Sep 17 00:00:00 2001
From: Christophe HENRY 
Date: Tue, 16 Mar 2021 20:54:10 +0100
Subject: [PATCH 09/13] Deep refactoring: parsing, translating, classes

* Removes "^" to disable text decoration line-wise.
* Split HTML generation in two: parsing and translating.
* Create class to handle gemtext parsing.
* Create class to translate to HTML.
* Create class to generate back gemtext (for future test cases).
* Uses generators to parse then translate.
* Fix: 404 doesn't occur for an empty file.
* Page 404 fully generated by HtmGem itself.
* CSS is no longer incorporated in the HTML page.
* Handle CSS inclusion by addCss() calls.
---
 index.gmi       |   3 +-
 index.php       | 147 ++++++---------
 lib-htmgem.php  | 467 +++++++++++++++++++++++++++++++-----------------
 tutogemtext.gmi |   4 +-
 4 files changed, 353 insertions(+), 268 deletions(-)

diff --git a/index.gmi b/index.gmi
index ba08644..5be4e06 100644
--- a/index.gmi
+++ b/index.gmi
@@ -118,8 +118,7 @@ La décoration du texte, qui interprête le **gras** par exemple, ne fait pas pa
 ### Désactiver la décoration du texte
 
 On peut :
-* commencer la ligne par **^**,
-* faire un bloc non décoré avec **^^^**,
+* désactiver et activer la décoration du texte avec une ligne **^^^**,
 * ajouter ce qui suit à la **réécriture** d’URL :
 > &textDecoration=0
 
diff --git a/index.php b/index.php
index cac10d5..5832956 100644
--- a/index.php
+++ b/index.php
@@ -1,91 +1,61 @@
 ");
     }
-?>
-
-
-
-Installation de HtmGem
-
-
-
-
-\n\n";
-    die();
+    $t = new \htmgem\GemTextTranslate_html(@file_get_contents("index.gmi"));
+    echo $t->getFullHtml();
+    exit();
 }
-######################################## /Installation page
+
+
+# to false only if textDecoration=0 in the URL
+$textDecoration = "0" != @$_REQUEST['textDecoration'];
 
 # Removes the headling and trailling slashes, to be sure there's not any.
 $filePath = rtrim($_SERVER['DOCUMENT_ROOT'], "/")."/".ltrim($url, "/");
 
-$fileContents = @file_get_contents($filePath);
-
-
-######################################## 404 page
-if (empty($fileContents)) {
+/* 404 page
+ */
+if (!file_exists($filePath)) {
     error_log("HtmGem: 404 $url $filePath");
-    http_response_code(404); ?>
-
-
-
-
-
-
-
- $url Recharger 🔄
-
 => /
 EOF;
-echo translateGemToHtml($text404);
-echo "\n";
-die();
+    $t = new \htmgem\GemTextTranslate_html($page404);
+    echo $t->getFullHtml();
+    exit();
 }
-######################################## /404 page
 
+
+$fileContents = @file_get_contents($filePath);
 # Removes the Byte Order Mark
 $fileContents = preg_replace("/\xEF\xBB\xBF/", "", $fileContents);
 
-# Gets the page title: the first occurrence with # at the line start
-mb_ereg("#\s*([^\n]+)\n", $fileContents, $matches);
-$page_title = @$matches[1];
 
-###################################### CSS Management
-/**
-* if &style=source displays the source directly and stops.
-* if there's a filename.css besides filename.gmi, use the css and stops.
-* if &style= then embbed the default style, and stops.
-* if &style= then use htmgem/word.css
-* if &style=/… then use the … as as stylesheet.
-**/
+/* CSS and special style management
+ */
 
+$style = @$_REQUEST['style'];
 if ("source" == $style) {
     $basename = basename($filePath);
     header("Cache-Control: public");
@@ -93,9 +63,12 @@ if ("source" == $style) {
     header("Content-Type: text/plain");
     header("Content-Transfer-Encoding: binary");
     header('Content-Length: ' . filesize($filePath));
-    readfile($filePath);
+    echo $fileContents;
     exit();
 } elseif ("pre" == $style) {
+    # Gets the page title: the first occurrence with # at the line start
+    mb_ereg("#\s*([^\n]+)\n", $fileContents, $matches);
+    $page_title = @$matches[1];
     $fileContents = htmlspecialchars($fileContents, ENT_HTML5|ENT_NOQUOTES, "UTF-8", false);
     echo <<
@@ -104,52 +77,32 @@ if ("source" == $style) {
 $page_title
 
 
-
    $fileContents
    
+
    +$fileContents
    
 
 
 EOL;
+    exit();
 } else {
-    $parts = pathinfo($filePath);
-    $localCss = $parts["filename"].".css";
-    $localCssFilePath = $parts["dirname"]."/".$localCss;
-    if (file_exists($localCssFilePath)) {
-        # Warning, using htmhem.php?url=… will make $localCss not found
-        # as the path is relative to htmgem.php and not / !
-        $cssContent = "";
-    } else {
-        if (empty($style)) {
-            $cssContent =
-                 "\n";
-        } else {
-            if ("none" == $style) {
-                $cssContent = "";
-            } else {
-                if ("/" == $style[0])
-                    $href = $style;
-                else
-                    $href = "/htmgem/css/$style.css";
-                $cssContent = "";
-            }
+    $t = new \htmgem\GemTextTranslate_html($fileContents);
+    if ("none" == $style) {
+        $t->addCss("");
+    } elseif ("/" == @$style[0]) {
+        $t->addCss($style);
+    } elseif (empty($style)) {
+        $parts = pathinfo($filePath);
+        $localCss = $parts["filename"].".css";
+        $localCssFilePath = $parts["dirname"]."/".$localCss;
+        if (file_exists($localCssFilePath)) {
+            # Warning, using htmhem.php?url=… will make $localCss not found
+            # as the path is relative to htmgem.php and not / !
+            $t->addCss($localCss);
         }
+    } else { #TODO: regex check for $style
+        $t->addCss("/htmgem/css/$style.css");
     }
-    echo <<
-
-
-$page_title
-
-$cssContent
-
-
-EOL;
-
-    echo "\n".translateGemToHtml($fileContents);
-    echo "\n\n";
 }
 
-
-ob_end_flush();
+echo $t->getFullHtml();
 
 ?>
diff --git a/lib-htmgem.php b/lib-htmgem.php
index 0711a93..1b40b39 100644
--- a/lib-htmgem.php
+++ b/lib-htmgem.php
@@ -5,203 +5,78 @@ namespace htmgem;
 mb_internal_encoding("UTF-8");
 mb_regex_encoding("UTF-8");
 
-$style = @$_REQUEST['style'];
-
-define("NARROW_NO_BREAK_SPACE", " ");
-define("DASHES"
-    ,"‒" # U+2012 Figure Dash
-    ."–" # U+2013 En Dash
-    ."—" # U+2014 Em Dash
-    ."⸺" # U+2E3A Two-Em Dash
-    ."⸻" # U+2E3B Three-Em Dash (Three times larger than a single char)
-);
-
 /**
- * Replaces markups things like __underlined__ to underlined.
- * @param $instruction the characters to replace, ex. _
- * @param $markup the markup to replace to, ex. "u" to get …
- * @param &$text where to replace.
+ * Parses the gemtext and generates the internal format version
+ * @param str $fileContents the gemtext to parse
  */
-function markupPreg($instruction, $markup, &$text) {
-    $output = $text;
-
-    # Replaces couples "__word__" into "word".
-    $output = mb_ereg_replace("${instruction}(.+?)${instruction}", "<{$markup}>\\1", $output);
-
-    # Replaces a remaining __ into "…" to the end of the line.
-    $output = mb_ereg_replace("${instruction}(.+)?", "<{$markup}>\\1", $output);
-
-    $text = $output;
-}
-
-/**
- * Adds text attributes sucj as underline, bold, … to $line
- * @param $line the line to process
- */
-function addTextAttributes(&$line) {
-    global $textDecoration;
-    if (!$textDecoration) return;
-    markupPreg("__",   "u",   $line);
-    markupPreg("\*\*", "strong",   $line);
-    markupPreg("//",   "em",   $line);
-    markupPreg("~~",   "del", $line);
-}
-
-/**
- * Prepares the raw text to be displayed in HTML environment:
- * * Escapes the HTML entities yet contained in the Gemtext.
- * * Puts thin unbrakable spaces before some characters.
- * @param $text1, $text2 texts to process
- */
-function htmlPrepare(&$text) {
-    $text = htmlspecialchars($text, ENT_HTML5|ENT_NOQUOTES, "UTF-8", false);
-    $text = mb_ereg_replace("\ ([?!:;»€$])", NARROW_NO_BREAK_SPACE."\\1", $text);
-    $text = mb_ereg_replace("([«])\ ", "\\1".NARROW_NO_BREAK_SPACE, $text); # Espace fine insécable
-
-    # Warning: using a monospace font editor may not display dashes as they should be!
-    # Adds no-break spaces to stick the (EM/EN dashes) to words : aaaaaa – bb – ccccc ==> aaaaaa –$bb$– ccccc
-    $text = mb_ereg_replace("([".DASHES."]) ([^".DASHES.".]+) ([".DASHES."])", "\\1".NARROW_NO_BREAK_SPACE."\\2".NARROW_NO_BREAK_SPACE."\\3", $text);
-
-    # Adds no-break space to stick the (EM/EN dashes) to words : aaaaaa – bb. ==> aaaaaa –$bb.
-    $text = mb_ereg_replace("([—–]) ([^.]+)\.", "\\1".NARROW_NO_BREAK_SPACE."\\2.", $text);
-}
-
-function translateGemToHtml($fileContents) {
-    $fileLines = preg_split("/\n/", $fileContents);
-    if (empty($fileLines[-1])) array_pop($fileLines); # Don't output a last empty line
-    ob_start();
+function gemtextParser($fileContents) {
+    $fileLines = explode("\n", $fileContents);
     $mode = null;
-    $mode_textAttributes = true;
+    $current = array();
     foreach ($fileLines as $line) {
         $reDoCount = 0;
         $mode_textAttributes_temp = false;
         while (true) {
-            if ($reDoCount>2) {
-                error_log("HtmGem: Too many loops, mode == '$mode'");
-                $mode = null;
-                $reDoCount = 0;
-                break;
+            /* The continue instruction is used to make another turn when there is a transition
+             * between two modes. */
+            if ($reDoCount>1) {
+                die("HtmGem: Too many loops, mode == '$mode'");
             }
             $reDoCount += 1;
             $line1 = substr($line, 0, 1); // $line can be modified
             $line2 = substr($line, 0, 2); // in the meantime.
             $line3 = substr($line, 0, 3);
             if (is_null($mode)) {
-                if (empty($line)) {
-                    echo "
     
    \n";
-                } elseif ('^^^' == $line3) {
-                    $mode_textAttributes = !$mode_textAttributes;
-                } elseif ('^' == $line1 and !$mode_textAttributes_temp) {
-                    if (preg_match("/^\^\s*(.+)$/", $line, $parts)) {
-                        $line = $parts[1];
-                        $mode_textAttributes_temp = true;
-                    } else {
-                        $mode = "raw";
-                    }
-                    continue;
+                if ('^^^' == $line3) {
+                    yield array("mode" => "^^^");
                 } elseif ("#" == $line1) {
-                    if (preg_match("/^(#{1,3})\s*(.+)/", $line, $sharps)) {
-                        $h_level = strlen($sharps[1]);
-                        $text = $sharps[2];
-                        htmlPrepare($text);
-                        switch ($h_level) {
-                            case 1: echo "
    ".$text."
    \n"; break;
-                            case 2: echo "
    ".$text."
    \n"; break;
-                            case 3: echo "
    ".$text."
    \n"; break;
-                        }
-                    } else {
-                        $mode = "raw";
-                        continue;
-                    }
+                    preg_match("/^(#{1,3})\s*(.+)/", $line, $matches);
+                    yield array("mode" => $matches[1], "title" => trim($matches[2]));
                 } elseif ("=>" == $line2) {
-                    if (preg_match("/^=>\s*([^\s]+)(?:\s+(.*))?$/", $line, $linkParts)) {
-                        $url_link = $linkParts[1];
-                        $url_label = @$linkParts[2];
-                        preg_match("/^([^:]+):/", $url_link, $matches);
-                        $url_protocol = @$matches[1];
-                        if (empty($url_protocol)) $url_protocol = "local";
-                        if (empty(trim($url_label))) {
-                            $url_label = $url_link;
-                        } else {
-                            // the label is humain-made, apply formatting
-                            htmlPrepare($url_label);
-                            if ($mode_textAttributes xor $mode_textAttributes_temp) addTextAttributes($url_label);
-                        }
-                        echo "
    $url_label
    \n";
-                    } else {
-                        $mode = "raw";
-                        continue;
-                    }
+                    preg_match("/^=>\s*([^\s]+)(?:\s+(.*))?$/", $line, $matches);
+                    yield array("mode" => "=>", "link" => trim($matches[1]), "text" => trim(@$matches[2]));
                 } elseif ("```" == $line3) {
                     preg_match("/^```\s*(.*)$/", $line, $matches);
-                    $alt_text = trim($matches[1]);
-                    if (empty($alt_text)) {
-                        echo "
    \n";
-                    } else {
-                        echo "
    \n";
-                    }
-                    $mode="pre";
+                    $current = array("mode" => "```", "alt" => trim($matches[1]), "texts" => array());
+                    $mode="```";
                 } elseif (">" == $line1) {
-                    echo "
    \n";
-                    $mode = "quote";
-                    continue;
+                    preg_match("/^>\s*(.*)$/", $line, $matches);
+                    $current = array("mode" => ">", "texts" => array(trim($matches[1])));
+                    $mode = ">";
                 } elseif ("*" == $line1) {
-                    echo "
      \n";
-                    $mode = "ul";
-                    continue;
+                    preg_match("/^\*\s*(.*)$/", $line, $matches);
+                    $current = array("mode" => "*", "texts" => array(trim($matches[1])));
+                    $mode = "*";
                 } else {
-                    $mode = "raw";
-                    continue;
+                    // text_line
+                    yield array("mode"=>"", "text" => trim($line));
                 }
             } else {
-                if ("raw"==$mode) {
-                    if (empty($line)) {
-                        $line = " ";
-                    } else {
-                        htmlPrepare($line);
-                        if ($mode_textAttributes xor $mode_textAttributes_temp)
-                            addTextAttributes($line);
-                    }
-                    echo "
      $line
      \n";
-                    $mode = null;
-                } elseif ("pre"==$mode) {
+                if ("```"==$mode) {
                     if ("```" == $line3) {
-                        echo "
    \n";
+                        yield $current;
+                        $current = array();
                         $mode = null;
                     } else {
-                        htmlPrepare($line);
-                        echo $line."\n";
+                        $current["texts"] []= $line; // No trim() as it’s a preformated text!
                     }
-                } elseif ("quote"==$mode) {
+                } elseif (">"==$mode) {
                     if (">" == $line1) {
-                        preg_match("/^>\s*(.*)$/", $line, $quoteParts);
-                        $quote = $quoteParts[1];
-                        if (empty($quote))
-                            echo "
     
    \n";
-                        else {
-                            htmlPrepare($quote);
-                            if ($mode_textAttributes xor $mode_textAttributes_temp)
-                                addTextAttributes($line);
-                            echo "
    ".$quote."
    \n";
-                        }
+                        preg_match("/^>\s*(.*)$/", $line, $matches);
+                        $current["texts"] []= trim($matches[1]);
                     } else {
-                        echo "\n";
+                        yield $current;
+                        $current = array();
                         $mode = null;
                         continue;
                     }
-                } elseif ("ul"==$mode) {
+                } elseif ("*"==$mode) {
                     if ("*" == $line1) {
-                        preg_match("/^\*\s*(.*)$/", $line, $ulParts);
-                        $li = $ulParts[1];
-                        if (empty($li)) {
-                            echo "
  •  \n";
-                        } else {
-                            htmlPrepare($li);
-                            if ($mode_textAttributes xor $mode_textAttributes_temp) addTextAttributes($li);
-                            echo "
  • ".$li."\n";
-                        }
+                        preg_match("/^\*\s*(.*)$/", $line, $matches);
+                        $current["texts"] []= trim($matches[1]);
                     } else {
-                        echo "\n";
+                        yield $current;
+                        $current = array();
                         $mode = null;
                         continue;
                     }
@@ -211,10 +86,266 @@ function translateGemToHtml($fileContents) {
             }
             break; // exits the while(true) as no continue occured
         } // while(true)
+    }// foreach
+    if ($current) yield $current; # File ends before the block.
+} // gemtextParser
+
+
+/**
+ * Translates the internal format into a gemtext.
+ * Uses cases:
+ *
+ * - test suites
+ * - serialisation easier with a text content
+ * - normalization (trimming spaces for instance)
+ */
+class GemtextTranslate_gemtext {
+
+    function __construct($parsedGemtext) {
+        $this->parsedGemtext = $parsedGemtext;
+        $this->translate();
     }
-    $html = ob_get_contents();
-    ob_clean();
-    return $html;
-}
+
+    protected function translate() {
+        ob_start();
+        foreach ($this->parsedGemtext as $node) {
+            $mode = $node["mode"];
+            switch($mode) {
+                case "":
+                    echo $node["text"]."\n";
+                    break;
+                case "*":
+                    foreach ($node["texts"] as $text) {
+                        echo "* $text\n";
+                    }
+                    break;
+                case "```":
+                    print("```\n");
+                    foreach ($node["texts"] as $text) {
+                        echo "$text\n";
+                    }
+                    print("```\n");
+                    break;
+                case ">":
+                    foreach ($node["texts"] as $text) {
+                        echo "> $text\n";
+                    }
+                    break;
+                case "=>":
+                    $linkText = $node["text"];
+                    if (!empty($linkText)) $linkText = " $linkText";
+                    print("=> ".$node["link"].$linkText."\n");
+                    break;
+                case "#":
+                case "##":
+                case "###":
+                    print("$mode ".$node["title"]."\n");
+                    break;
+                case "^^^":
+                    print("^^^\n");
+                    break;
+                default:
+                    die("Unknown mode: '{$node["mode"]}'\n");
+            }
+        }
+
+        $this->translatedGemtext = ob_get_contents();
+        ob_end_clean();
+    }
+
+    public function __toString() {
+        return $this->translatedGemtext;
+    }
+} // GemtextTranslate_gemtext
+
+
+/**
+ * Translates the internal format to HTML
+ */
+class GemtextTranslate_html {
+
+    protected $cssList = array();
+    protected $pageTitle = "";
+    public $translatedGemtext;
+
+    function __construct($parsedGemtext, $textDecorationEnabled=true) {
+        if (empty($parsedGemtext))
+            // So translate() will do no loop.
+            $parsedGemtext = array();
+        elseif (is_string($parsedGemtext))
+            // to delete the last empty line, 
     
     in HTML
+            $parsedGemtext = rtrim($parsedGemtext);
+            // The text must be parsed
+            $parsedGemtext = gemtextParser($parsedGemtext);
+        $this->parsedGemtext = $parsedGemtext;
+        $this->translate($textDecorationEnabled);
+    }
+
+    function addCss($css) {
+        $this->cssList []= $css;
+    }
+
+    const NARROW_NO_BREAK_SPACE = " ";
+    const DASHES
+        ="‒" # U+2012 Figure Dash
+        ."–" # U+2013 En Dash
+        ."—" # U+2014 Em Dash
+        ."⸺" # U+2E3A Two-Em Dash
+        ."⸻" # U+2E3B Three-Em Dash (Three times larger than a single char)
+    ;
+
+    /**
+     * Replaces markups things like __underlined__ to underlined.
+     * @param $instruction the characters to replace, ex. _
+     * @param $markup the markup to replace to, ex. "u" to get …
+     * @param &$text where to replace.
+     */
+    protected static function markupPreg($instruction, $markup, &$text) {
+        $output = $text;
+
+        # Replaces couples "__word__" into "word".
+        $output = mb_ereg_replace("${instruction}(.+?)${instruction}", "<{$markup}>\\1", $output);
+
+        # Replaces a remaining __ into "…" to the end of the line.
+        $output = mb_ereg_replace("${instruction}(.+)?", "<{$markup}>\\1", $output);
+
+        $text = $output;
+    }
+
+    /**
+     * Adds text attributes sucj as underline, bold, … to $line
+     * @param $line the line to process
+     */
+    protected static function addTextDecoration(&$line) {
+        self::markupPreg("__",   "u",      $line);
+        self::markupPreg("\*\*", "strong", $line);
+        self::markupPreg("//",   "em",     $line);
+        self::markupPreg("~~",   "del",    $line);
+    }
+
+    /**
+     * Prepares the raw text to be displayed in HTML environment:
+     * * Escapes the HTML entities yet contained in the Gemtext.
+     * * Puts thin unbrakable spaces before some characters.
+     * @param $text1, $text2 texts to process
+     */
+    protected static function htmlPrepare(&$text) {
+        if (empty($text)) {
+            $text = " ";
+        } else {
+            $text = htmlspecialchars($text, ENT_HTML5|ENT_NOQUOTES, "UTF-8", false);
+            $text = mb_ereg_replace("\ ([?!:;»€$])", self::NARROW_NO_BREAK_SPACE."\\1", $text);
+            $text = mb_ereg_replace("([«])\ ", "\\1".self::NARROW_NO_BREAK_SPACE, $text); # Espace fine insécable
+
+            # Warning: using a monospace font editor may not display dashes as they should be!
+            # Adds no-break spaces to stick the (EM/EN dashes) to words : aaaaaa – bb – ccccc ==> aaaaaa –$bb$– ccccc
+            $text = mb_ereg_replace("([".self::DASHES."]) ([^".self::DASHES.".]+) ([".self::DASHES."])", "\\1".self::NARROW_NO_BREAK_SPACE."\\2".self::NARROW_NO_BREAK_SPACE."\\3", $text);
+
+            # Adds no-break space to stick the (EM/EN dashes) to words : aaaaaa – bb. ==> aaaaaa –$bb.
+            $text = mb_ereg_replace("([—–]) ([^.]+)\.", "\\1".self::NARROW_NO_BREAK_SPACE."\\2.", $text);
+        }
+    }
+
+    public function translate($textDecoration=true) {
+        ob_start();
+        foreach ($this->parsedGemtext as $node) {
+            $mode = $node["mode"];
+            switch($mode) {
+                case "":
+                    $text = $node["text"];
+                    self::htmlPrepare($text);
+                    if ($textDecoration) self::addTextDecoration($text);
+                    echo "
    $text
    \n";
+                    break;
+                case "*":
+                    echo "
      \n";
+                    foreach ($node["texts"] as $text) {
+                        self::htmlPrepare($text);
+                        if ($textDecoration) self::addTextDecoration($text);
+                        print("
    • $text\n");
+                    }
+                    echo "
    \n";
+                    break;
+                case "```":
+                    $text = implode("\n", $node["texts"]);
+                    self::htmlPrepare($text);
+                    echo "
    \n$text\n
    \n";
+                    break;
+                case ">":
+                    $text = implode("\n", $node["texts"]);
+                    self::htmlPrepare($text);
+                    if ($textDecoration) self::addTextDecoration($text);
+                    echo "
    \n$text\n
    \n";
+                    break;
+                case "=>":
+                    $link = $node["link"];
+                    $linkText = $node["text"];
+                    if (empty($linkText)) {
+                        $linkText = $link;
+                        self::htmlPrepare($linkText);
+                    } else {
+                        self::htmlPrepare($linkText);
+                        if ($textDecoration) self::addTextDecoration($linkText);
+                    }
+                    preg_match("/^([^:]+):/", $link, $matches);
+                    $protocol = @$matches[1];
+                    if (empty($protocol)) $protocol = "local";
+                    echo "
    $linkText
    \n";
+                    break;
+                case "#":
+                    $title = $node["title"];
+                    self::htmlPrepare($title);
+                    if (empty($this->pageTitle)) $this->pageTitle = $title;
+                    echo "
    $title
    \n";
+                    break;
+                case "##":
+                    $title = $node["title"];
+                    self::htmlPrepare($title);
+                    echo "
    $title
    \n";
+                    break;
+                case "###":
+                    $title = $node["title"];
+                    self::htmlPrepare($title);
+                    echo "
    $title
    \n";
+                    break;
+                case "^^^":
+                    $this->textDecorationEnabled = !$this->textDecorationEnabled;
+                    break;
+                default:
+                    die("Unknown mode: '{$node["mode"]}'\n");
+            }
+        }
+
+        $this->translatedGemtext = ob_get_contents();
+        ob_end_clean();
+    }
+
+    function getFullHtml() {
+        if (!$this->cssList)
+            $css = array("/htmgem/css/htmgem.css");
+        else
+            $css = $this->cssList;
+        echo <<
+
+
+{$this->pageTitle}
+
+EOL;
+        foreach ($css as $c) {
+            echo "\n";
+        }
+        echo <<
+\n
+EOL;
+        echo $this->translatedGemtext;
+        echo "\n\n";
+    }
+
+    public function __toString() {
+        return $this->translatedGemtext;
+    }
+} // GemTextTranslate_html
 
 ?>
diff --git a/tutogemtext.gmi b/tutogemtext.gmi
index a0d6f1c..32f03c7 100644
--- a/tutogemtext.gmi
+++ b/tutogemtext.gmi
@@ -53,9 +53,11 @@ Il ne peut exister qu’un lien par ligne. Et la ligne est dédiée à ça ! Voi
 
 La décoration du texte ne fait pas partie des spécifications de Gemini.
 
-^ Cette **ligne** utilise la //décoration du texte// qui peut ~~barrer~~ ou __souligner__ des mots.
+Cette *​*ligne*​* utilise la /​/décoration du texte/​/ qui peut ~​~barrer~​~ ou _​_souligner_​_ des mots.
 Cette **ligne** utilise la //décoration du texte// qui peut ~~barrer~~ ou __souligner__ des mots.
 
+On peut désactiver et activer la décoration du texte avec **^^^** sur une ligne.
+
 Et voilà ! Vous savez tout ce qui est à savoir 🥳 Les spécifications complètes sont sur le site de Gemini :
 => https://gemini.circumlunar.space/docs/specification.gmi
 

From d280ceff94c2d805c8c908d4948cfdf9c93ae63a Mon Sep 17 00:00:00 2001
From: Christophe HENRY 
Date: Thu, 18 Mar 2021 14:43:02 +0100
Subject: [PATCH 10/13] WIP

---
 index.php      | 34 +++++++++++++++++-----------------
 lib-htmgem.php |  3 +--
 2 files changed, 18 insertions(+), 19 deletions(-)

diff --git a/index.php b/index.php
index 5832956..2942a24 100644
--- a/index.php
+++ b/index.php
@@ -83,24 +83,24 @@ $fileContents
    • 
 
 EOL;
     exit();
-} else {
-    $t = new \htmgem\GemTextTranslate_html($fileContents);
-    if ("none" == $style) {
-        $t->addCss("");
-    } elseif ("/" == @$style[0]) {
-        $t->addCss($style);
-    } elseif (empty($style)) {
-        $parts = pathinfo($filePath);
-        $localCss = $parts["filename"].".css";
-        $localCssFilePath = $parts["dirname"]."/".$localCss;
-        if (file_exists($localCssFilePath)) {
-            # Warning, using htmhem.php?url=… will make $localCss not found
-            # as the path is relative to htmgem.php and not / !
-            $t->addCss($localCss);
-        }
-    } else { #TODO: regex check for $style
-        $t->addCss("/htmgem/css/$style.css");
+}
+
+$t = new \htmgem\GemTextTranslate_html($fileContents, $textDecoration);
+if ("none" == $style) {
+    $t->addCss("");
+} elseif ("/" == @$style[0]) {
+    $t->addCss($style);
+} elseif (empty($style)) {
+    $parts = pathinfo($filePath);
+    $localCss = $parts["filename"].".css";
+    $localCssFilePath = $parts["dirname"]."/".$localCss;
+    if (file_exists($localCssFilePath)) {
+        # Warning, using htmhem.php?url=… will make $localCss not found
+        # as the path is relative to htmgem.php and not / !
+        $t->addCss($localCss);
     }
+} else { #TODO: regex check for $style
+    $t->addCss("/htmgem/css/$style.css");
 }
 
 echo $t->getFullHtml();
diff --git a/lib-htmgem.php b/lib-htmgem.php
index 1b40b39..cf2c3cc 100644
--- a/lib-htmgem.php
+++ b/lib-htmgem.php
@@ -170,8 +170,7 @@ class GemtextTranslate_html {
 
     function __construct($parsedGemtext, $textDecorationEnabled=true) {
         if (empty($parsedGemtext))
-            // So translate() will do no loop.
-            $parsedGemtext = array();
+            $parsedGemtext = "";
         elseif (is_string($parsedGemtext))
             // to delete the last empty line, 
     
     in HTML
             $parsedGemtext = rtrim($parsedGemtext);

From 365c855c004b00223dcc81ec95b364d46d25e57b Mon Sep 17 00:00:00 2001
From: Christophe HENRY 
Date: Thu, 18 Mar 2021 21:16:30 +0100
Subject: [PATCH 11/13] Adds icons in the 404 page

---
 index.php | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/index.php b/index.php
index 2942a24..e8bf014 100644
--- a/index.php
+++ b/index.php
@@ -38,8 +38,7 @@ if (!file_exists($filePath)) {
 
 ​**$url**
 
-=> $url Recharger 🔄
-=> /
+=> .. 🔄 🔄
 EOF;
     $t = new \htmgem\GemTextTranslate_html($page404);
     echo $t->getFullHtml();

From 232cecc398e5a8cca1721922f53f3b81f16fc6ff Mon Sep 17 00:00:00 2001
From: Christophe HENRY 
Date: Thu, 18 Mar 2021 21:18:49 +0100
Subject: [PATCH 12/13] Enable security on what gmi file the client asks.

* checks realPath() against no-existent files,
* checks the file suffix '.gmi'
* checks the directory belongs to that of the site.
---
 index.php | 20 ++++++++++++++++----
 1 file changed, 16 insertions(+), 4 deletions(-)

diff --git a/index.php b/index.php
index e8bf014..a551626 100644
--- a/index.php
+++ b/index.php
@@ -21,16 +21,26 @@ if (empty($url)) {
     exit();
 }
 
-
-# to false only if textDecoration=0 in the URL
-$textDecoration = "0" != @$_REQUEST['textDecoration'];
+$documentRoot = $_SERVER['DOCUMENT_ROOT'];
 
 # Removes the headling and trailling slashes, to be sure there's not any.
 $filePath = rtrim($_SERVER['DOCUMENT_ROOT'], "/")."/".ltrim($url, "/");
 
+switch(true) {
+    case false:
+    case !realPath($filePath):
+    case !preg_match("/\.gmi$/", $url): # not finishing by .gmi
+    case strpos($filePath, $documentRoot)!==0: # not in web directory
+        $go404 = true;
+        // Says 404 even if the file exists to not give any information.
+        break;
+    default:
+        $go404 = false;
+}
+
 /* 404 page
  */
-if (!file_exists($filePath)) {
+if ($go404) {
     error_log("HtmGem: 404 $url $filePath");
     http_response_code(404);
     $page404 = <<
Date: Fri, 19 Mar 2021 09:54:02 +0100
Subject: [PATCH 13/13] Updates changelog

---
 CHANGELOG.gmi | 18 +++++++++++++++---
 1 file changed, 15 insertions(+), 3 deletions(-)

diff --git a/CHANGELOG.gmi b/CHANGELOG.gmi
index ec1fd99..961d81e 100644
--- a/CHANGELOG.gmi
+++ b/CHANGELOG.gmi
@@ -16,7 +16,6 @@ All notable changes to this project will be documented in this file.
 
 ### Security
 * Check url encoding: The filename fetched on disk may differ from that was asked by URL.
-* Perform sanity checks against unauthorized access.
 * Check unicode capability, UTF16 input, RTL/LTR, etc…
 * Manage different type of carriage return: CR CR/LF LF.
 
@@ -30,15 +29,28 @@ All notable changes to this project will be documented in this file.
 
 ## [Unreleased] v1
 
+### Security
+∅
+
 ### Development
 * Command line API for script, testing, CI…
-* Split codebase: library to interpret GemText, command line interface (tests, …) and web interface.
-* Properly close tags when the page exists in a non-null mode.
+* Automated tests like phpunit
 
 ### User interface
 * Translate French texts to English.
 * Search for {pages}.css from the current directory to the root. Empty css file means don’t search and go to default.
 
+## [1.2.0] - 2021-03-19
+* Removes "^" to disable text decoration line-wise.
+* CSS is no longer incorporated in the HTML page.
+* Perform sanity checks against unauthorized file access.
+* Properly close tags when the page exists in a non-null mode.
+* Split HTML generation in two: parsing and translating.
+* Create classes to handle gemtext parsing and translating.
+* Create class to generate back gemtext (for future test cases).
+* Fix: 404 doesn't occur for an empty file.
+* Page 404 fully generated by HtmGem itself.
+
 ## [1.1.0] - 2021-03-14
 * File download when using "source" as a style.
 * Improves the regex.