From 8cf174ecb3b3e113e0a051eea04c74272d09e675 Mon Sep 17 00:00:00 2001
From: Christophe HENRY <christophe.henry@sbgodin.fr>
Date: Sun, 4 Apr 2021 22:23:38 +0200
Subject: [PATCH] Handles line feeds: Unix, Mac, Windows

---
 lib-htmgem.inc.php                            |   2 +-
 tests/files_with_html/linefeeds-utf-16.txt    | Bin 0 -> 82 bytes
 .../files_with_html/linefeeds-utf-16.txt.html |   8 ++++++++
 tests/files_with_html/linefeeds.txt           |   4 ++++
 tests/files_with_html/linefeeds.txt.html      |   8 ++++++++
 tests/translateToGemtextTest.php              |   2 +-
 tests/translateToHtmlTest.php                 |  18 +++++++++++++++++-
 tests/utils.inc.php                           |   4 ++--
 8 files changed, 41 insertions(+), 5 deletions(-)
 create mode 100644 tests/files_with_html/linefeeds-utf-16.txt
 create mode 100644 tests/files_with_html/linefeeds-utf-16.txt.html
 create mode 100644 tests/files_with_html/linefeeds.txt
 create mode 100644 tests/files_with_html/linefeeds.txt.html

diff --git a/lib-htmgem.inc.php b/lib-htmgem.inc.php
index 6ad1ed3..f93baf4 100644
--- a/lib-htmgem.inc.php
+++ b/lib-htmgem.inc.php
@@ -12,7 +12,7 @@ mb_regex_encoding("UTF-8");
 function gemtextParser($fileContents) {
     if (empty($fileContents)) return array();
     $fileContents = rtrim($fileContents); // removes last empty line
-    $fileLines = explode("\n", $fileContents);
+    $fileLines = mb_split("\R", $fileContents); // Unix, Mac, Windows line feeds
     $mode = null;
     $current = array();
     foreach ($fileLines as $line) {
diff --git a/tests/files_with_html/linefeeds-utf-16.txt b/tests/files_with_html/linefeeds-utf-16.txt
new file mode 100644
index 0000000000000000000000000000000000000000..62a3f6f2dc38c10d03382038f3dbbc6f82af1cdd
GIT binary patch
literal 82
scmezWPm4i;p_CzyA(Nqkfr|me%VkJpNM_&#vr`!I8H#~yhzv360ADB!i~s-t

literal 0
HcmV?d00001

diff --git a/tests/files_with_html/linefeeds-utf-16.txt.html b/tests/files_with_html/linefeeds-utf-16.txt.html
new file mode 100644
index 0000000..6fe31fe
--- /dev/null
+++ b/tests/files_with_html/linefeeds-utf-16.txt.html
@@ -0,0 +1,8 @@
+<ul>
+<li>unix
+<li>mac
+<li>dos
+<li>unix
+<li>mac
+<li>dos
+</ul>
diff --git a/tests/files_with_html/linefeeds.txt b/tests/files_with_html/linefeeds.txt
new file mode 100644
index 0000000..ae9d1ae
--- /dev/null
+++ b/tests/files_with_html/linefeeds.txt
@@ -0,0 +1,4 @@
+* unix
+* mac
* dos
+* unix
+* mac
* dos
diff --git a/tests/files_with_html/linefeeds.txt.html b/tests/files_with_html/linefeeds.txt.html
new file mode 100644
index 0000000..6fe31fe
--- /dev/null
+++ b/tests/files_with_html/linefeeds.txt.html
@@ -0,0 +1,8 @@
+<ul>
+<li>unix
+<li>mac
+<li>dos
+<li>unix
+<li>mac
+<li>dos
+</ul>
diff --git a/tests/translateToGemtextTest.php b/tests/translateToGemtextTest.php
index 4c70684..4e1943a 100644
--- a/tests/translateToGemtextTest.php
+++ b/tests/translateToGemtextTest.php
@@ -51,7 +51,7 @@ final class translateToGemtextTest extends TestCase {
 
     #TODO: don't stop when problems are found, list all the faulty files
     public function test_translate_gemtext_files(): void {
-        foreach(getGmiFiles(dirname(__FILE__)."/..") as $filePathname) {
+        foreach(getFiles(dirname(__FILE__)."/..", "gmi") as $filePathname) {
             $fileContent = file_get_contents($filePathname);
             \htmgem\io\convertToUTF8($fileContent);
             $this->assertSame(
diff --git a/tests/translateToHtmlTest.php b/tests/translateToHtmlTest.php
index 0fd78aa..fb6977d 100644
--- a/tests/translateToHtmlTest.php
+++ b/tests/translateToHtmlTest.php
@@ -94,7 +94,7 @@ final class translateToHtmlTest extends TestCase {
         /** NOTE: the UTF-16 files must result in the same content as UTF-8 ones.
          * command to convert from UTF-8 to UTF-16: iconv -f utf8 -r utf16 text.gmi
          */
-        foreach(getGmiFiles(dirname(__FILE__)."/files_with_html") as $filePathname) {
+        foreach(getFiles(dirname(__FILE__)."/files_with_html", "gmi") as $filePathname) {
             $fileContentGmi = file_get_contents($filePathname);
             \htmgem\io\convertToUTF8($fileContentGmi);
             $fileContentHtml = file_get_contents($filePathname.".html");
@@ -106,4 +106,20 @@ final class translateToHtmlTest extends TestCase {
         }
     }
 
+    public function test_line_feeds(): void {
+        /** NOTE: the UTF-16 files must result in the same content as UTF-8 ones.
+         * command to convert from UTF-8 to UTF-16: iconv -f utf8 -r utf16 text.gmi
+         */
+        foreach(getFiles(dirname(__FILE__)."/files_with_html", "txt") as $filePathname) {
+            $fileContentGmi = file_get_contents($filePathname);
+            \htmgem\io\convertToUTF8($fileContentGmi);
+            $fileContentHtml = file_get_contents($filePathname.".html");
+            $this->assertSame(
+                $fileContentHtml,
+                translateHtml($fileContentGmi),
+                "Line feeds, translation to HTML: $filePathname"
+            );
+        }
+    }
+
 }
diff --git a/tests/utils.inc.php b/tests/utils.inc.php
index 52824e8..533336e 100644
--- a/tests/utils.inc.php
+++ b/tests/utils.inc.php
@@ -1,6 +1,6 @@
 <?php declare(strict_types=1);
 
-function getGmiFiles($directory): generator {
+function getFiles($directory, $targetExtension): generator {
     $flags =
         FilesystemIterator::KEY_AS_PATHNAME
       | FilesystemIterator::CURRENT_AS_FILEINFO
@@ -14,7 +14,7 @@ function getGmiFiles($directory): generator {
         $filename = $fileinfo->getFilename();
         $filePathname = $fileinfo->getPathname();
         $extension = $fileinfo->getExtension();
-        if ("gmi" == $extension) {
+        if ($targetExtension == $extension) {
             yield $filePathname;
         }
     }