From 2e521c8413c3273795552cf26cd968d6eccfad8f Mon Sep 17 00:00:00 2001
From: KongQun Yang <kqyang@google.com>
Date: Wed, 5 May 2021 10:58:17 -0700
Subject: [PATCH] Remove another use of regex library

It is not working correctly in gcc 4.8 or earlier, which is still
popular (bundled by default in CentOS 7).

Issue #865, #929.

Change-Id: I136446a70831bd0237cd29646dd349fe7558176b
---
 packager/media/formats/webvtt/webvtt_utils.cc | 31 +++++++++++++++++--
 .../formats/webvtt/webvtt_utils_unittest.cc   | 15 +++++++++
 2 files changed, 43 insertions(+), 3 deletions(-)

diff --git a/packager/media/formats/webvtt/webvtt_utils.cc b/packager/media/formats/webvtt/webvtt_utils.cc
index bb0bea3481..3a7af3f6a7 100644
--- a/packager/media/formats/webvtt/webvtt_utils.cc
+++ b/packager/media/formats/webvtt/webvtt_utils.cc
@@ -9,11 +9,12 @@
 #include <ctype.h>
 #include <inttypes.h>
 
-#include <regex>
+#include <algorithm>
 #include <unordered_set>
 
 #include "packager/base/logging.h"
 #include "packager/base/strings/string_number_conversions.h"
+#include "packager/base/strings/string_util.h"
 #include "packager/base/strings/stringprintf.h"
 
 namespace shaka {
@@ -67,6 +68,31 @@ std::string GetCloseTag(StyleTagKind tag) {
   return "";  // Not reached, but Windows doesn't like NOTREACHED.
 }
 
+bool IsWhitespace(char c) {
+  return c == '\t' || c == '\r' || c == '\n' || c == ' ';
+}
+
+// Replace consecutive whitespaces with a single whitespace.
+std::string CollapseWhitespace(const std::string& data) {
+  std::string output;
+  output.resize(data.size());
+  size_t chars_written = 0;
+  bool in_whitespace = false;
+  for (char c : data) {
+    if (IsWhitespace(c)) {
+      if (!in_whitespace) {
+        in_whitespace = true;
+        output[chars_written++] = ' ';
+      }
+    } else {
+      in_whitespace = false;
+      output[chars_written++] = c;
+    }
+  }
+  output.resize(chars_written);
+  return output;
+}
+
 std::string WriteFragment(const TextFragment& fragment,
                           std::list<StyleTagKind>* tags) {
   std::string ret;
@@ -115,8 +141,7 @@ std::string WriteFragment(const TextFragment& fragment,
       // Replace newlines and consecutive whitespace with a single space.  If
       // the user wanted an explicit newline, they should use the "newline"
       // field.
-      std::regex whitespace("\\s+", std::regex_constants::ECMAScript);
-      ret += std::regex_replace(fragment.body, whitespace, std::string(" "));
+      ret += CollapseWhitespace(fragment.body);
     } else {
       for (const auto& frag : fragment.sub_fragments) {
         ret += WriteFragment(frag, tags);
diff --git a/packager/media/formats/webvtt/webvtt_utils_unittest.cc b/packager/media/formats/webvtt/webvtt_utils_unittest.cc
index eb4281ef96..a56d5d08c6 100644
--- a/packager/media/formats/webvtt/webvtt_utils_unittest.cc
+++ b/packager/media/formats/webvtt/webvtt_utils_unittest.cc
@@ -179,6 +179,21 @@ TEST(WebVttUtilsTest, FragmentToString_PreservesTags) {
   EXPECT_EQ(WebVttFragmentToString(frag), "<i>Foobar</i>");
 }
 
+TEST(WebVttUtilsTest, FragmentToString_ConsecutiveLeadingWhitespaces) {
+  TextFragment frag(kNoStyle, "\r\n\t \r\nFoobar");
+  EXPECT_EQ(WebVttFragmentToString(frag), " Foobar");
+}
+
+TEST(WebVttUtilsTest, FragmentToString_ConsecutiveTrailingWhitespaces) {
+  TextFragment frag(kNoStyle, "Foobar\r\n\t \r\n");
+  EXPECT_EQ(WebVttFragmentToString(frag), "Foobar ");
+}
+
+TEST(WebVttUtilsTest, FragmentToString_ConsecutiveInternalWhitespaces) {
+  TextFragment frag(kNoStyle, "Hello\r\n\t \r\nWorld");
+  EXPECT_EQ(WebVttFragmentToString(frag), "Hello World");
+}
+
 TEST(WebVttUtilsTest, FragmentToString_HandlesNestedFragments) {
   TextFragment frag;
   frag.sub_fragments.emplace_back(kNoStyle, "Hello ");