Shell: Include some metadata in parsed tokens and ask for continuation

This patchset adds some metadata to Parser::parse() which allows the
Shell to ask for the rest of a command, if it is not complete.
A command is considered complete if it has no trailing pipe, or
unterminated string.

Shell/Parser.cpp

@@ -29,7 +29,7 @@
 #include <stdio.h>
 #include <unistd.h>
 
-void Parser::commit_token(AllowEmptyToken allow_empty)
+void Parser::commit_token(Token::Type type, AllowEmptyToken allow_empty)
 {
     if (allow_empty == AllowEmptyToken::No && m_token.is_empty())
         return;
@@ -38,7 +38,7 @@ void Parser::commit_token(AllowEmptyToken allow_empty)
         m_token.clear_with_capacity();
         return;
     }
-    m_tokens.append(String::copy(m_token));
+    m_tokens.append({ String::copy(m_token), m_position, m_token.size(), type });
     m_token.clear_with_capacity();
 };
 
@@ -82,22 +82,22 @@ bool Parser::in_state(State state) const
 
 Vector<Command> Parser::parse()
 {
-    for (size_t i = 0; i < m_input.length(); ++i) {
+    for (size_t i = 0; i < m_input.length(); ++i, m_position = i) {
         char ch = m_input.characters()[i];
         switch (state()) {
         case State::Free:
             if (ch == ' ') {
-                commit_token();
+                commit_token(Token::Bare);
                 break;
             }
             if (ch == ';') {
-                commit_token();
+                commit_token(Token::Special);
                 commit_subcommand();
                 commit_command();
                 break;
             }
             if (ch == '|') {
-                commit_token();
+                commit_token(Token::Special);
                 if (m_tokens.is_empty()) {
                     fprintf(stderr, "Syntax error: Nothing before pipe (|)\n");
                     return {};
@@ -106,7 +106,7 @@ Vector<Command> Parser::parse()
                 break;
             }
             if (ch == '>') {
-                commit_token();
+                commit_token(Token::Special);
                 begin_redirect_write(STDOUT_FILENO);
 
                 // Search for another > for append.
@@ -114,7 +114,7 @@ Vector<Command> Parser::parse()
                 break;
             }
             if (ch == '<') {
-                commit_token();
+                commit_token(Token::Special);
                 begin_redirect_read(STDIN_FILENO);
                 push_state(State::InRedirectionPath);
                 break;
@@ -163,7 +163,7 @@ Vector<Command> Parser::parse()
                 }
 
                 if (is_multi_fd_redirection) {
-                    commit_token();
+                    commit_token(Token::Special);
 
                     int fd = atoi(&m_input.characters()[i + 1]);
 
@@ -186,7 +186,7 @@ Vector<Command> Parser::parse()
                 if (i != m_input.length() - 1) {
                     char next_ch = m_input.characters()[i + 1];
                     if (next_ch == '>') {
-                        commit_token();
+                        commit_token(Token::Special);
                         begin_redirect_write(ch - '0');
                         ++i;
 
@@ -195,7 +195,7 @@ Vector<Command> Parser::parse()
                         break;
                     }
                     if (next_ch == '<') {
-                        commit_token();
+                        commit_token(Token::Special);
                         begin_redirect_read(ch - '0');
                         ++i;
 
@@ -208,7 +208,7 @@ Vector<Command> Parser::parse()
             break;
         case State::InWriteAppendOrRedirectionPath:
             if (ch == '>') {
-                commit_token();
+                commit_token(Token::Special);
                 pop_state();
                 push_state(State::InRedirectionPath);
                 ASSERT(m_redirections.size());
@@ -222,21 +222,21 @@ Vector<Command> Parser::parse()
             [[fallthrough]];
         case State::InRedirectionPath:
             if (ch == '<') {
-                commit_token();
+                commit_token(Token::Special);
                 begin_redirect_read(STDIN_FILENO);
                 pop_state();
                 push_state(State::InRedirectionPath);
                 break;
             }
             if (ch == '>') {
-                commit_token();
+                commit_token(Token::Special);
                 begin_redirect_read(STDOUT_FILENO);
                 pop_state();
                 push_state(State::InRedirectionPath);
                 break;
             }
             if (ch == '|') {
-                commit_token();
+                commit_token(Token::Special);
                 if (m_tokens.is_empty()) {
                     fprintf(stderr, "Syntax error: Nothing before pipe (|)\n");
                     return {};
@@ -260,7 +260,7 @@ Vector<Command> Parser::parse()
         case State::InSingleQuotes:
             if (ch == '\'') {
                 if (!in_state(State::InRedirectionPath))
-                    commit_token(AllowEmptyToken::Yes);
+                    commit_token(Token::SingleQuoted, AllowEmptyToken::Yes);
                 pop_state();
                 break;
             }
@@ -269,7 +269,7 @@ Vector<Command> Parser::parse()
         case State::InDoubleQuotes:
             if (ch == '\"') {
                 if (!in_state(State::InRedirectionPath))
-                    commit_token(AllowEmptyToken::Yes);
+                    commit_token(Token::DoubleQuoted, AllowEmptyToken::Yes);
                 pop_state();
                 break;
             }
@@ -294,15 +294,18 @@ Vector<Command> Parser::parse()
     }
 
     while (m_state_stack.size() > 1) {
-        auto allow_empty = AllowEmptyToken::No;
-        if (state() == State::InDoubleQuotes || state() == State::InSingleQuotes)
-            allow_empty = AllowEmptyToken::Yes;
-        commit_token(allow_empty);
+        if (state() == State::InDoubleQuotes) {
+            commit_token(Token::UnterminatedDoubleQuoted, AllowEmptyToken::Yes);
+        } else if (state() == State::InSingleQuotes) {
+            commit_token(Token::UnterminatedSingleQuoted, AllowEmptyToken::Yes);
+        } else {
+            commit_token(Token::Bare, AllowEmptyToken::No);
+        }
         pop_state();
     }
     ASSERT(state() == State::Free);
 
-    commit_token();
+    commit_token(Token::Bare);
     commit_subcommand();
     commit_command();
 

Shell/Parser.h

@@ -29,6 +29,21 @@
 #include <AK/String.h>
 #include <AK/Vector.h>
 
+struct Token {
+    enum Type {
+        Bare,
+        SingleQuoted,
+        DoubleQuoted,
+        UnterminatedSingleQuoted,
+        UnterminatedDoubleQuoted,
+        Special,
+    };
+    String text;
+    size_t end;
+    size_t length;
+    Type type;
+};
+
 struct Redirection {
     enum Type {
         Pipe,
@@ -48,7 +63,7 @@ struct Rewiring {
 };
 
 struct Subcommand {
-    Vector<String> args;
+    Vector<Token> args;
     Vector<Redirection> redirections;
     Vector<Rewiring> rewirings;
 };
@@ -71,7 +86,7 @@ private:
         No,
         Yes,
     };
-    void commit_token(AllowEmptyToken = AllowEmptyToken::No);
+    void commit_token(Token::Type, AllowEmptyToken = AllowEmptyToken::No);
     void commit_subcommand();
     void commit_command();
     void do_pipe();
@@ -105,7 +120,8 @@ private:
 
     Vector<Command> m_commands;
     Vector<Subcommand> m_subcommands;
-    Vector<String> m_tokens;
+    Vector<Token> m_tokens;
     Vector<Redirection> m_redirections;
     Vector<char> m_token;
+    size_t m_position { 0 };
 };

Shell/main.cpp

@@ -53,11 +53,42 @@
 GlobalState g;
 static Line::Editor editor { Line::Configuration { Line::Configuration::UnescapedSpaces } };
 
-static int run_command(const String&);
+struct ExitCodeOrContinuationRequest {
+    enum ContinuationRequest {
+        Nothing,
+        Pipe,
+        DoubleQuotedString,
+        SingleQuotedString,
+    };
+
+    ExitCodeOrContinuationRequest(ContinuationRequest continuation)
+        : continuation(continuation)
+    {
+    }
+
+    ExitCodeOrContinuationRequest(int exit)
+        : exit_code(exit)
+    {
+    }
+
+    bool has_value() const { return exit_code.has_value(); }
+    int value() const
+    {
+        ASSERT(has_value());
+        return exit_code.value();
+    }
+
+    Optional<int> exit_code;
+    ContinuationRequest continuation { Nothing };
+};
+
+static ExitCodeOrContinuationRequest run_command(const StringView&);
 void cache_path();
+static ExitCodeOrContinuationRequest::ContinuationRequest s_should_continue { ExitCodeOrContinuationRequest::Nothing };
 
 static String prompt()
 {
+    auto build_prompt = []() -> String {
         auto* ps1 = getenv("PROMPT");
         if (!ps1) {
             if (g.uid == 0)
@@ -110,6 +141,25 @@ static String prompt()
             builder.append(*ptr);
         }
         return builder.to_string();
+    };
+
+    auto the_prompt = build_prompt();
+    auto prompt_length = editor.actual_rendered_string_length(the_prompt);
+
+    if (s_should_continue != ExitCodeOrContinuationRequest::Nothing) {
+        const auto format_string = "\033[34m%.*-s\033[m";
+        switch (s_should_continue) {
+        case ExitCodeOrContinuationRequest::Pipe:
+            return String::format(format_string, prompt_length, "pipe> ");
+        case ExitCodeOrContinuationRequest::DoubleQuotedString:
+            return String::format(format_string, prompt_length, "dquote> ");
+        case ExitCodeOrContinuationRequest::SingleQuotedString:
+            return String::format(format_string, prompt_length, "squote> ");
+        default:
+            break;
+        }
+    }
+    return the_prompt;
 }
 
 static int sh_pwd(int, const char**)
@@ -306,9 +356,13 @@ static int sh_time(int argc, const char** argv)
     }
     Core::ElapsedTimer timer;
     timer.start();
-    int exit_code = run_command(builder.to_string());
+    auto exit_code = run_command(builder.string_view());
+    if (!exit_code.has_value()) {
+        printf("Shell: Incomplete command: %s\n", builder.to_string().characters());
+        exit_code = 1;
+    }
     printf("Time: %d ms\n", timer.elapsed());
-    return exit_code;
+    return exit_code.value();
 }
 
 static int sh_umask(int argc, const char** argv)
@@ -581,7 +635,7 @@ static bool handle_builtin(int argc, const char** argv, int& retval)
 
 class FileDescriptionCollector {
 public:
-    FileDescriptionCollector() {}
+    FileDescriptionCollector() { }
     ~FileDescriptionCollector() { collect(); }
 
     void collect()
@@ -724,13 +778,13 @@ static Vector<String> expand_parameters(const StringView& param)
     return res;
 }
 
-static Vector<String> process_arguments(const Vector<String>& args)
+static Vector<String> process_arguments(const Vector<Token>& args)
 {
     Vector<String> argv_string;
     for (auto& arg : args) {
         // This will return the text passed in if it wasn't a variable
         // This lets us just loop over its values
-        auto expanded_parameters = expand_parameters(arg);
+        auto expanded_parameters = expand_parameters(arg.text);
 
         for (auto& exp_arg : expanded_parameters) {
             if (exp_arg.starts_with('~'))
@@ -748,7 +802,29 @@ static Vector<String> process_arguments(const Vector<String>& args)
     return argv_string;
 }
 
-static int run_command(const String& cmd)
+static ExitCodeOrContinuationRequest::ContinuationRequest is_complete(const Vector<Command>& commands)
+{
+    // check if the last command ends with a pipe, or an unterminated string
+    auto& last_command = commands.last();
+    auto& subcommands = last_command.subcommands;
+    if (subcommands.size() == 0)
+        return ExitCodeOrContinuationRequest::Nothing;
+
+    auto& last_subcommand = subcommands.last();
+
+    if (!last_subcommand.redirections.find([](auto& redirection) { return redirection.type == Redirection::Pipe; }).is_end())
+        return ExitCodeOrContinuationRequest::Pipe;
+
+    if (!last_subcommand.args.find([](auto& token) { return token.type == Token::UnterminatedSingleQuoted; }).is_end())
+        return ExitCodeOrContinuationRequest::SingleQuotedString;
+
+    if (!last_subcommand.args.find([](auto& token) { return token.type == Token::UnterminatedDoubleQuoted; }).is_end())
+        return ExitCodeOrContinuationRequest::DoubleQuotedString;
+
+    return ExitCodeOrContinuationRequest::Nothing;
+}
+
+static ExitCodeOrContinuationRequest run_command(const StringView& cmd)
 {
     if (cmd.is_empty())
         return 0;
@@ -758,13 +834,36 @@ static int run_command(const String& cmd)
 
     auto commands = Parser(cmd).parse();
 
+    auto needs_more = is_complete(commands);
+    if (needs_more != ExitCodeOrContinuationRequest::Nothing)
+        return needs_more;
+
 #ifdef SH_DEBUG
     for (auto& command : commands) {
         for (size_t i = 0; i < command.subcommands.size(); ++i) {
             for (size_t j = 0; j < i; ++j)
                 dbgprintf("    ");
             for (auto& arg : command.subcommands[i].args) {
-                dbgprintf("<%s> ", arg.characters());
+                switch (arg.type) {
+                case Token::Bare:
+                    dbgprintf("<%s> ", arg.text.characters());
+                    break;
+                case Token::SingleQuoted:
+                    dbgprintf("'<%s>' ", arg.text.characters());
+                    break;
+                case Token::DoubleQuoted:
+                    dbgprintf("\"<%s>\" ", arg.text.characters());
+                    break;
+                case Token::UnterminatedSingleQuoted:
+                    dbgprintf("\'<%s> ", arg.text.characters());
+                    break;
+                case Token::UnterminatedDoubleQuoted:
+                    dbgprintf("\"<%s> ", arg.text.characters());
+                    break;
+                case Token::Special:
+                    dbgprintf("<%s> ", arg.text.characters());
+                    break;
+                }
             }
             dbgprintf("\n");
             for (auto& redirecton : command.subcommands[i].redirections) {
@@ -1320,12 +1419,27 @@ int main(int argc, char** argv)
 
     cache_path();
 
+    StringBuilder complete_line_builder;
+
     for (;;) {
         auto line = editor.get_line(prompt());
         if (line.is_empty())
             continue;
-        run_command(line);
-        editor.add_to_history(line);
+
+        // FIXME: This might be a bit counter-intuitive, since we put nothing
+        //        between the two lines, even though the user has pressed enter
+        //        but since the LineEditor cannot yet handle literal newlines
+        //        inside the text, we opt to do this the wrong way (for the time being)
+        complete_line_builder.append(line);
+
+        auto complete_or_exit_code = run_command(complete_line_builder.string_view());
+        s_should_continue = complete_or_exit_code.continuation;
+
+        if (!complete_or_exit_code.has_value())
+            continue;
+
+        editor.add_to_history(complete_line_builder.build());
+        complete_line_builder.clear();
     }
 
     return 0;