Shell+LibLine: Handle escaped characters correctly

This patchset fixes incorrect handling of escaped tokens (`a\ b`) in
Shell autocompletion and LibLine.
The users of LibLine can now choose between two token splitting modes,
either taking into account escapes, or ignoring them.

Libraries/LibLine/Editor.cpp

@@ -35,9 +35,10 @@
 
 namespace Line {
 
-Editor::Editor(bool always_refresh)
+Editor::Editor(Configuration configuration)
+ : m_configuration(configuration)
 {
- m_always_refresh = always_refresh;
+ m_always_refresh = configuration.refresh_behaviour == Configuration::RefreshBehaviour::Eager;
  m_pending_chars = ByteBuffer::create_uninitialized(0);
  struct winsize ws;
  if (ioctl(STDOUT_FILENO, TIOCGWINSZ, &ws) < 0) {
@@ -352,21 +353,34 @@ String Editor::get_line(const String& prompt)
  if (!on_tab_complete_first_token || !on_tab_complete_other_token)
  continue;
 
- bool is_empty_token = m_cursor == 0 || m_buffer[m_cursor - 1] == ' ';
+ auto should_break_token = [mode = m_configuration.split_mechanism](auto& buffer, size_t index) {
+ switch (mode) {
+ case Configuration::TokenSplitMechanism::Spaces:
+ return buffer[index] == ' ';
+ case Configuration::TokenSplitMechanism::UnescapedSpaces:
+ return buffer[index] == ' ' && (index == 0 || buffer[index - 1] != '\\');
+ }
+
+ ASSERT_NOT_REACHED();
+ return true;
+ };
+
+ bool is_empty_token = m_cursor == 0 || should_break_token(m_buffer, m_cursor - 1);
 
  // reverse tab can count as regular tab here
  m_times_tab_pressed++;
 
  int token_start = m_cursor - 1;
+
  if (!is_empty_token) {
- while (token_start >= 0 && m_buffer[token_start] != ' ')
+ while (token_start >= 0 && !should_break_token(m_buffer, token_start))
  --token_start;
  ++token_start;
  }
 
  bool is_first_token = true;
  for (int i = token_start - 1; i >= 0; --i) {
- if (m_buffer[i] != ' ') {
+ if (should_break_token(m_buffer, i)) {
  is_first_token = false;
  break;
  }
@@ -651,7 +665,7 @@ String Editor::get_line(const String& prompt)
  for (auto ch : m_buffer)
  m_pre_search_buffer.append(ch);
  m_pre_search_cursor = m_cursor;
- m_search_editor = make<Editor>(true); // Has anyone seen 'Inception'?
+ m_search_editor = make<Editor>(Configuration { Configuration::Eager, m_configuration.split_mechanism }); // Has anyone seen 'Inception'?
  m_search_editor->on_display_refresh = [this](Editor& search_editor) {
  search(StringView { search_editor.buffer().data(), search_editor.buffer().size() });
  refresh_display();

Libraries/LibLine/Editor.h

@@ -75,9 +75,37 @@ struct CompletionSuggestion {
  String trailing_trivia;
 };
 
+struct Configuration {
+ enum TokenSplitMechanism {
+ Spaces,
+ UnescapedSpaces,
+ };
+ enum RefreshBehaviour {
+ Lazy,
+ Eager,
+ };
+
+ Configuration()
+ {
+ }
+
+ template<typename Arg, typename... Rest>
+ Configuration(Arg arg, Rest... rest)
+ : Configuration(rest...)
+ {
+ set(arg);
+ }
+
+ void set(RefreshBehaviour refresh) { refresh_behaviour = refresh; }
+ void set(TokenSplitMechanism split) { split_mechanism = split; }
+
+ RefreshBehaviour refresh_behaviour { RefreshBehaviour::Lazy };
+ TokenSplitMechanism split_mechanism { TokenSplitMechanism::Spaces };
+};
+
 class Editor {
 public:
- explicit Editor(bool always_refresh = false);
+ explicit Editor(Configuration configuration = {});
  ~Editor();
 
  String get_line(const String& prompt);
@@ -308,6 +336,8 @@ class Editor {
  bool m_refresh_needed { false };
 
  bool m_is_editing { false };
+
+ Configuration m_configuration;
 };
 
 }

Shell/main.cpp

@@ -50,7 +50,7 @@
 //#define SH_DEBUG
 
 GlobalState g;
-static Line::Editor editor {};
+static Line::Editor editor { Line::Configuration { Line::Configuration::UnescapedSpaces } };
 
 static int run_command(const String&);
 void cache_path();
@@ -580,7 +580,7 @@ static bool handle_builtin(int argc, const char** argv, int& retval)
 
 class FileDescriptionCollector {
 public:
- FileDescriptionCollector() {}
+ FileDescriptionCollector() { }
  ~FileDescriptionCollector() { collect(); }
 
  void collect()
@@ -1003,6 +1003,62 @@ void save_history()
  }
 }
 
+String escape_token(const String& token)
+{
+ StringBuilder builder;
+
+ for (auto c : token) {
+ switch (c) {
+ case '\'':
+ case '"':
+ case '$':
+ case '|':
+ case '>':
+ case '<':
+ case '&':
+ case '\\':
+ case ' ':
+ builder.append('\\');
+ break;
+ default:
+ break;
+ }
+ builder.append(c);
+ }
+
+ return builder.build();
+}
+
+String unescape_token(const String& token)
+{
+ StringBuilder builder;
+
+ enum {
+ Free,
+ Escaped
+ } state { Free };
+
+ for (auto c : token) {
+ switch (state) {
+ case Escaped:
+ builder.append(c);
+ state = Free;
+ break;
+ case Free:
+ if (c == '\\')
+ state = Escaped;
+ else
+ builder.append(c);
+ break;
+ }
+ }
+
+ if (state == Escaped)
+ builder.append('\\');
+
+ return builder.build();
+}
+
 Vector<String, 256> cached_path;
 void cache_path()
 {
@@ -1020,7 +1076,7 @@ void cache_path()
  auto program = programs.next_path();
  String program_path = String::format("%s/%s", directory.characters(), program.characters());
  if (access(program_path.characters(), X_OK) == 0)
- cached_path.append(program.characters());
+ cached_path.append(escape_token(program.characters()));
  }
  }
 
@@ -1041,15 +1097,16 @@ int main(int argc, char** argv)
  g.termios = editor.termios();
  g.default_termios = editor.default_termios();
 
- editor.on_tab_complete_first_token = [&](const String& token) -> Vector<Line::CompletionSuggestion> {
+ editor.on_tab_complete_first_token = [&](const String& token_to_complete) -> Vector<Line::CompletionSuggestion> {
+ auto token = unescape_token(token_to_complete);
+
  auto match = binary_search(cached_path.data(), cached_path.size(), token, [](const String& token, const String& program) -> int {
  return strncmp(token.characters(), program.characters(), token.length());
  });
 
  if (!match) {
  // There is no executable in the $PATH starting with $token
  // Suggest local executables and directories
- auto mut_token = token; // copy it :(
  String path;
  Vector<Line::CompletionSuggestion> local_suggestions;
  bool suggest_executables = true;
@@ -1061,11 +1118,11 @@ int main(int argc, char** argv)
  if (last_slash >= 0) {
  // Split on the last slash. We'll use the first part as the directory
  // to search and the second part as the token to complete.
- path = mut_token.substring(0, last_slash + 1);
+ path = token.substring(0, last_slash + 1);
  if (path[0] != '/')
  path = String::format("%s/%s", g.cwd.characters(), path.characters());
  path = canonicalized_path(path);
- mut_token = mut_token.substring(last_slash + 1, mut_token.length() - last_slash - 1);
+ token = token.substring(last_slash + 1, token.length() - last_slash - 1);
  } else {
  // We have no slashes, so the directory to search is the current
  // directory and the token to complete is just the original token.
@@ -1078,19 +1135,19 @@ int main(int argc, char** argv)
  // e.g. in `cd /foo/bar', 'bar' is the invariant
  // since we are not suggesting anything starting with
  // `/foo/', but rather just `bar...'
- editor.suggest(mut_token.length(), 0);
+ editor.suggest(token_to_complete.length(), 0);
 
  // only suggest dot-files if path starts with a dot
  Core::DirIterator files(path,
- mut_token.starts_with('.') ? Core::DirIterator::NoFlags : Core::DirIterator::SkipDots);
+ token.starts_with('.') ? Core::DirIterator::NoFlags : Core::DirIterator::SkipDots);
 
  while (files.has_next()) {
  auto file = files.next_path();
  // manually skip `.' and `..'
  if (file == "." || file == "..")
  continue;
  auto trivia = " ";
- if (file.starts_with(mut_token)) {
+ if (file.starts_with(token)) {
  String file_path = String::format("%s/%s", path.characters(), file.characters());
  struct stat program_status;
  int stat_error = stat(file_path.characters(), &program_status);
@@ -1105,7 +1162,7 @@ int main(int argc, char** argv)
  trivia = "/";
  }
 
- local_suggestions.append({ file, trivia });
+ local_suggestions.append({ escape_token(file), trivia });
  }
  }
 
@@ -1128,12 +1185,12 @@ int main(int argc, char** argv)
  }
  suggestions.append({ cached_path[index], " " });
 
- editor.suggest(token.length(), 0);
+ editor.suggest(token_to_complete.length(), 0);
 
  return suggestions;
  };
- editor.on_tab_complete_other_token = [&](const String& vtoken) -> Vector<Line::CompletionSuggestion> {
- auto token = vtoken; // copy it :(
+ editor.on_tab_complete_other_token = [&](const String& token_to_complete) -> Vector<Line::CompletionSuggestion> {
+ auto token = unescape_token(token_to_complete);
  String path;
  Vector<Line::CompletionSuggestion> suggestions;
 
@@ -1159,7 +1216,7 @@ int main(int argc, char** argv)
  // e.g. in `cd /foo/bar', 'bar' is the invariant
  // since we are not suggesting anything starting with
  // `/foo/', but rather just `bar...'
- editor.suggest(token.length(), 0);
+ editor.suggest(token_to_complete.length(), 0);
 
  // only suggest dot-files if path starts with a dot
  Core::DirIterator files(path,
@@ -1176,9 +1233,9 @@ int main(int argc, char** argv)
  int stat_error = stat(file_path.characters(), &program_status);
  if (!stat_error) {
  if (S_ISDIR(program_status.st_mode))
- suggestions.append({ file, "/" });
+ suggestions.append({ escape_token(file), "/" });
  else
- suggestions.append({ file, " " });
+ suggestions.append({ escape_token(file), " " });
  }
  }
  }