From 2eb5793d55b48782cc1c4974f957ebcce9881dba Mon Sep 17 00:00:00 2001
From: Tommy Nguyen <remyabel@gmail.com>
Date: Mon, 2 Dec 2019 07:42:33 -0500
Subject: [PATCH] LibMarkdown: Handle CRLF line endings

Previously, MDDocument only split on Unix-style line endings. This adds
a new function to StringView which handles LF, CR and CRLF.
---
 AK/StringView.cpp                    | 40 ++++++++++++++++++++++++++++
 AK/StringView.h                      |  6 +++++
 AK/Tests/TestStringView.cpp          | 21 +++++++++++++++
 Libraries/LibMarkdown/MDDocument.cpp |  2 +-
 4 files changed, 68 insertions(+), 1 deletion(-)

diff --git a/AK/StringView.cpp b/AK/StringView.cpp
index 6db1b8f144..2b79c61c28 100644
--- a/AK/StringView.cpp
+++ b/AK/StringView.cpp
@@ -40,6 +40,46 @@ Vector<StringView> StringView::split_view(const char separator, bool keep_empty)
     return v;
 }
 
+Vector<StringView> StringView::lines(bool consider_cr) const
+{
+    if (is_empty())
+        return {};
+
+    if (!consider_cr)
+        return split_view('\n', true);
+
+    Vector<StringView> v;
+    ssize_t substart = 0;
+    bool last_ch_was_cr = false;
+    bool split_view = false;
+    for (ssize_t i = 0; i < length(); ++i) {
+        char ch = characters_without_null_termination()[i];
+        if (ch == '\n') {
+            split_view = true;
+            if (last_ch_was_cr) {
+                substart = i + 1;
+                split_view = false;
+                last_ch_was_cr = false;
+            }
+        }
+        if (ch == '\r') {
+            split_view = true;
+            last_ch_was_cr = true;
+        }
+        if (split_view) {
+            ssize_t sublen = i - substart;
+            if (sublen != 0)
+                v.append(substring_view(substart, sublen));
+            substart = i + 1;
+        }
+        split_view = false;
+    }
+    ssize_t taillen = length() - substart;
+    if (taillen != 0)
+        v.append(substring_view(substart, taillen));
+    return v;
+}
+
 bool StringView::starts_with(const StringView& str) const
 {
     if (str.is_empty())
diff --git a/AK/StringView.h b/AK/StringView.h
index 00c9aaafa3..9e558947f0 100644
--- a/AK/StringView.h
+++ b/AK/StringView.h
@@ -46,6 +46,12 @@ public:
     StringView substring_view(int start, int length) const;
     Vector<StringView> split_view(char, bool keep_empty = false) const;
 
+    // Create a Vector of StringViews split by line endings. As of CommonMark
+    // 0.29, the spec defines a line ending as "a newline (U+000A), a carriage
+    // return (U+000D) not followed by a newline, or a carriage return and a
+    // following newline.".
+    Vector<StringView> lines(bool consider_cr = true) const;
+
     // FIXME: These should be shared between String and StringView somehow!
     unsigned to_uint(bool& ok) const;
     int to_int(bool& ok) const;
diff --git a/AK/Tests/TestStringView.cpp b/AK/Tests/TestStringView.cpp
index add8726a1b..d98c74138d 100644
--- a/AK/Tests/TestStringView.cpp
+++ b/AK/Tests/TestStringView.cpp
@@ -42,4 +42,25 @@ TEST_CASE(starts_with)
     EXPECT(!test_string_view.starts_with("DEF"));
 }
 
+TEST_CASE(lines)
+{
+    String test_string = "a\nb\r\nc\rd";
+    StringView test_string_view = test_string.view();
+    Vector<StringView> test_string_vector = test_string_view.lines();
+    EXPECT_EQ(test_string_vector.size(), 4);
+    EXPECT(test_string_vector.at(0) == String("a"));
+    EXPECT(test_string_vector.at(1) == String("b"));
+    EXPECT(test_string_vector.at(2) == String("c"));
+    EXPECT(test_string_vector.at(3) == String("d"));
+
+    test_string = "```\nHello there\r\nHello there\n```";
+    test_string_view = test_string.view();
+    test_string_vector = test_string_view.lines();
+    EXPECT_EQ(test_string_vector.size(), 4);
+    EXPECT(test_string_vector.at(0) == String("```"));
+    EXPECT(test_string_vector.at(1) == String("Hello there"));
+    EXPECT(test_string_vector.at(2) == String("Hello there"));
+    EXPECT(test_string_vector.at(3) == String("```"));
+}
+
 TEST_MAIN(StringView)
diff --git a/Libraries/LibMarkdown/MDDocument.cpp b/Libraries/LibMarkdown/MDDocument.cpp
index 539ad8f70d..b3161adf53 100644
--- a/Libraries/LibMarkdown/MDDocument.cpp
+++ b/Libraries/LibMarkdown/MDDocument.cpp
@@ -49,7 +49,7 @@ static bool helper(Vector<StringView>::ConstIterator& lines, NonnullOwnPtrVector
 
 bool MDDocument::parse(const StringView& str)
 {
-    const Vector<StringView> lines_vec = str.split_view('\n', true);
+    const Vector<StringView> lines_vec = str.lines();
     auto lines = lines_vec.begin();
 
     while (true) {