From e5783d41132d353a40fe1cf1d546ccd325d10ea9 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?J=C3=A9r=C3=B4me=20BOULMIER?= <jerome.boulmier@outlook.fr>
Date: Thu, 21 Nov 2019 22:26:53 +0100
Subject: [PATCH] Include empty tokens when splitting if necessary (#1893)

Fixes #1881
---
 include/utils/string.hpp          |  2 +-
 src/modules/bspwm.cpp             |  8 ++----
 src/modules/i3.cpp                |  2 +-
 src/modules/xkeyboard.cpp         |  6 ++--
 src/modules/xworkspaces.cpp       |  2 +-
 src/utils/string.cpp              | 47 ++++++++++++++++++++-----------
 tests/unit_tests/utils/string.cpp | 44 +++++++++++++++++++++--------
 7 files changed, 71 insertions(+), 40 deletions(-)

diff --git a/include/utils/string.hpp b/include/utils/string.hpp
index abc073be..40130c44 100644
--- a/include/utils/string.hpp
+++ b/include/utils/string.hpp
@@ -88,8 +88,8 @@ namespace string_util {
   string utf8_truncate(string&& value, size_t len);
 
   string join(const vector<string>& strs, const string& delim);
-  vector<string>& split_into(const string& s, char delim, vector<string>& container);
   vector<string> split(const string& s, char delim);
+  std::vector<std::string> tokenize(const string& str, char delimiters);
 
   size_t find_nth(const string& haystack, size_t pos, const string& needle, size_t nth);
 
diff --git a/src/modules/bspwm.cpp b/src/modules/bspwm.cpp
index cf6817cd..3d145f64 100644
--- a/src/modules/bspwm.cpp
+++ b/src/modules/bspwm.cpp
@@ -131,7 +131,7 @@ namespace modules {
     m_icons->add(DEFAULT_ICON, factory_util::shared<label>(m_conf.get(name(), DEFAULT_ICON, ""s)));
 
     for (const auto& workspace : m_conf.get_list<string>(name(), "ws-icon", {})) {
-      auto vec = string_util::split(workspace, ';');
+      auto vec = string_util::tokenize(workspace, ';');
       if (vec.size() == 2) {
         m_icons->add(vec[0], factory_util::shared<label>(vec[1]));
       }
@@ -224,11 +224,7 @@ namespace modules {
     size_t workspace_n{0U};
 
     for (auto&& tag : string_util::split(data, ':')) {
-      if (tag.empty()) {
-        continue;
-      }
-
-      auto value = !tag.empty() ? tag.substr(1) : "";
+      auto value = tag.substr(1);
       auto mode_flag = mode::NONE;
       unsigned int workspace_mask{0U};
 
diff --git a/src/modules/i3.cpp b/src/modules/i3.cpp
index b01cf5c8..383a39fc 100644
--- a/src/modules/i3.cpp
+++ b/src/modules/i3.cpp
@@ -58,7 +58,7 @@ namespace modules {
     m_icons->add(DEFAULT_WS_ICON, factory_util::shared<label>(m_conf.get(name(), DEFAULT_WS_ICON, ""s)));
 
     for (const auto& workspace : m_conf.get_list<string>(name(), "ws-icon", {})) {
-      auto vec = string_util::split(workspace, ';');
+      auto vec = string_util::tokenize(workspace, ';');
       if (vec.size() == 2) {
         m_icons->add(vec[0], factory_util::shared<label>(vec[1]));
       }
diff --git a/src/modules/xkeyboard.cpp b/src/modules/xkeyboard.cpp
index 23b686bd..7f7b22dd 100644
--- a/src/modules/xkeyboard.cpp
+++ b/src/modules/xkeyboard.cpp
@@ -49,7 +49,7 @@ namespace modules {
     m_layout_icons->add(DEFAULT_LAYOUT_ICON, load_optional_label(m_conf, name(), DEFAULT_LAYOUT_ICON, ""s));
 
     for (const auto& it : m_conf.get_list<string>(name(), "layout-icon", {})) {
-      auto vec = string_util::split(it, ';');
+      auto vec = string_util::tokenize(it, ';');
       if (vec.size() == 2) {
         m_layout_icons->add(vec[0], factory_util::shared<label>(vec[1]));
       }
@@ -80,7 +80,7 @@ namespace modules {
       m_indicator_icons_off = factory_util::shared<iconset>();
       m_indicator_icons_on = factory_util::shared<iconset>();
 
-      auto icon_pair = string_util::split(m_conf.get(name(), DEFAULT_INDICATOR_ICON, ""s), ';');
+      auto icon_pair = string_util::tokenize(m_conf.get(name(), DEFAULT_INDICATOR_ICON, ""s), ';');
       if(icon_pair.size() == 2) {
         m_indicator_icons_off->add(DEFAULT_INDICATOR_ICON, factory_util::shared<label>(icon_pair[0]));
         m_indicator_icons_on->add(DEFAULT_INDICATOR_ICON, factory_util::shared<label>(icon_pair[1]));
@@ -90,7 +90,7 @@ namespace modules {
       }
 
       for (const auto& it : m_conf.get_list<string>(name(), "indicator-icon", {})) {
-        auto icon_triple = string_util::split(it, ';');
+        auto icon_triple = string_util::tokenize(it, ';');
         if (icon_triple.size() == 3) {
           auto const indicator_str = string_util::lower(icon_triple[0]);
           m_indicator_icons_off->add(indicator_str, factory_util::shared<label>(icon_triple[1]));
diff --git a/src/modules/xworkspaces.cpp b/src/modules/xworkspaces.cpp
index 1515ca4e..47c24b7a 100644
--- a/src/modules/xworkspaces.cpp
+++ b/src/modules/xworkspaces.cpp
@@ -72,7 +72,7 @@ namespace modules {
     m_icons->add(DEFAULT_ICON, factory_util::shared<label>(m_conf.get(name(), DEFAULT_ICON, ""s)));
 
     for (const auto& workspace : m_conf.get_list<string>(name(), "icon", {})) {
-      auto vec = string_util::split(workspace, ';');
+      auto vec = string_util::tokenize(workspace, ';');
       if (vec.size() == 2) {
         m_icons->add(vec[0], factory_util::shared<label>(vec[1]));
       }
diff --git a/src/utils/string.cpp b/src/utils/string.cpp
index cf6d2622..27fad0ee 100644
--- a/src/utils/string.cpp
+++ b/src/utils/string.cpp
@@ -222,27 +222,42 @@ namespace string_util {
   }
 
   /**
-   * Explode string by delim into container
-   */
-  vector<string>& split_into(const string& s, char delim, vector<string>& container) {
-    string str;
-    std::stringstream buffer(s);
-    while (getline(buffer, str, delim)) {
-      container.emplace_back(str);
-    }
-    return container;
-  }
-
-  /**
-   * Explode string by delim
+   * Explode string by delim, ignore empty tokens
    */
   vector<string> split(const string& s, char delim) {
-    vector<string> vec;
-    return split_into(s, delim, vec);
+    std::string::size_type pos = 0;
+    std::vector<std::string> result;
+
+    while ((pos = s.find_first_not_of(delim, pos)) != std::string::npos) {
+      auto nextpos = s.find_first_of(delim, pos);
+      result.emplace_back(s.substr(pos, nextpos - pos));
+      pos = nextpos;
+    }
+
+    return result;
   }
 
   /**
-   * Find the nth occurence of needle in haystack starting from pos
+   * Explode string by delim, include empty tokens
+   */
+  std::vector<std::string> tokenize(const string& str, char delimiters) {
+    std::vector<std::string> tokens;
+    std::string::size_type lastPos = 0;
+    auto pos = str.find_first_of(delimiters, lastPos);
+
+    while (pos != std::string::npos && lastPos != std::string::npos) {
+      tokens.emplace_back(str.substr(lastPos, pos - lastPos));
+
+      lastPos = pos + 1;
+      pos = str.find_first_of(delimiters, lastPos);
+    }
+
+    tokens.emplace_back(str.substr(lastPos, pos - lastPos));
+    return tokens;
+  }
+
+  /**
+   * Find the nth occurrence of needle in haystack starting from pos
    */
   size_t find_nth(const string& haystack, size_t pos, const string& needle, size_t nth) {
     size_t found_pos = haystack.find(needle, pos);
diff --git a/tests/unit_tests/utils/string.cpp b/tests/unit_tests/utils/string.cpp
index b74f3296..09f48f8f 100644
--- a/tests/unit_tests/utils/string.cpp
+++ b/tests/unit_tests/utils/string.cpp
@@ -64,20 +64,40 @@ TEST(String, join) {
   EXPECT_EQ("A, B, C", string_util::join({"A", "B", "C"}, ", "));
 }
 
-TEST(String, splitInto) {
-  vector<string> strings;
-  string_util::split_into("A,B,C", ',', strings);
-  EXPECT_EQ(3, strings.size());
-  EXPECT_EQ("A", strings[0]);
-  EXPECT_EQ("C", strings[2]);
+TEST(String, split) {
+  {
+    vector<string> strings = string_util::split("A,B,C", ',');
+    EXPECT_EQ(3, strings.size());
+    EXPECT_EQ("A", strings[0]);
+    EXPECT_EQ("B", strings[1]);
+    EXPECT_EQ("C", strings[2]);
+  }
+
+  {
+    vector<string> strings = string_util::split(",A,,B,,C,", ',');
+    EXPECT_EQ(3, strings.size());
+    EXPECT_EQ("A", strings[0]);
+    EXPECT_EQ("B", strings[1]);
+    EXPECT_EQ("C", strings[2]);
+  }
 }
 
-TEST(String, split) {
-  vector<string> strings{"foo", "bar"};
-  vector<string> result{string_util::split("foo,bar", ',')};
-  EXPECT_EQ(strings.size(), result.size());
-  EXPECT_EQ(strings[0], result[0]);
-  EXPECT_EQ("bar", result[1]);
+TEST(String, tokenize) {
+  {
+    vector<string> strings = string_util::tokenize("A,B,C", ',');
+    EXPECT_EQ(3, strings.size());
+    EXPECT_EQ("A", strings[0]);
+    EXPECT_EQ("B", strings[1]);
+    EXPECT_EQ("C", strings[2]);
+  }
+
+  {
+    using namespace std::string_literals;
+    vector<string> strings = string_util::tokenize(",A,,B,,C,", ',');
+    vector<string> result{""s, "A"s, ""s, "B"s, ""s, "C"s, ""s};
+
+    EXPECT_TRUE(strings == result);
+  }
 }
 
 TEST(String, findNth) {