上次我们聊到 CLI 的领域交互模式。在领域交互模式中，可能存在多层次的子命令。在使用过程中如果全评记忆的话，命令少还好，多了真心记不住。频繁 --help 也是个很麻烦的事情。如果每次按 'tab' 键就可以提示或补齐命令是不是很方便呢。这一节我们就来说说 'autocommplete' 如何实现。我们还是以interactcli-rs中的实现来解说实现过程
实现过程
其实，rustyline 已经为我们提供了基本的helper功能框架,其中包括了completer。我们来看代码，文件位置src/interact/cli.rs
#[derive(Helper)]
struct MyHelper {
completer: CommandCompleter,
highlighter: MatchingBracketHighlighter,
validator: MatchingBracketValidator,
hinter: HistoryHinter,
colored_prompt: String,
}
pub fn run() {
let config = Config::builder()
.history_ignore_space(true)
.completion_type(CompletionType::List)
.output_stream(OutputStreamType::Stdout)
.build();
let h = MyHelper {
completer: get_command_completer(),
highlighter: MatchingBracketHighlighter::new(),
hinter: HistoryHinter {},
colored_prompt: "".to_owned(),
validator: MatchingBracketValidator::new(),
};
let mut rl = Editor::with_config(config);
// let mut rl = Editor::<()>::new();
rl.set_helper(Some(h));
......
}
首先定义 MyHelper 结构体， 需要实现 Completer + Hinter + Highlighter + Validator trait。然后通过rustyline的set_helper函数加载我们定义好的helper。在MyHelper 结构体中，需要我们自己来实现completer的逻辑。
Sub command autocompleter实现详解
#[derive(Debug, Clone)]
pub struct SubCmd {
pub level: usize,
pub command_name: String,
pub subcommands: Vec<String>,
}
SubCmd 结构体包含：命令级别，命令名称，以及该命令包含的子命令信息，以便在实现实现 autocomplete 时定位命令和子命令的范围
- 在程序启动时遍历所有的command，src/cmd/rootcmd.rs 中的all_subcommand函数负责收集所有命令并转换为Vec<SubCmd>
pub fn all_subcommand(app: &clap_Command, beginlevel: usize, input: &mut Vec<SubCmd>) {
let nextlevel = beginlevel + 1;
let mut subcmds = vec![];
for iterm in app.get_subcommands() {
subcmds.push(iterm.get_name().to_string());
if iterm.has_subcommands() {
all_subcommand(iterm, nextlevel, input);
} else {
if beginlevel == 0 {
all_subcommand(iterm, nextlevel, input);
}
}
}
let subcommand = SubCmd {
level: beginlevel,
command_name: app.get_name().to_string(),
subcommands: subcmds,
};
input.push(subcommand);
}
- CommandCompleter 子命令自动补充功能的核心部分
#[derive(Debug, Clone)]
pub struct CommandCompleter {
subcommands: Vec<SubCmd>,
}
impl CommandCompleter {
pub fn new(subcmds: Vec<SubCmd>) -> Self {
Self {
subcommands: subcmds,
}
}
//获取level下所有可能的子命令
pub fn level_possible_cmd(&self, level: usize) -> Vec<String> {
let mut subcmds = vec![];
let cmds = self.subcommands.clone();
for iterm in cmds {
if iterm.level == level {
subcmds.push(iterm.command_name.clone());
}
}
return subcmds;
}
//获取level下某字符串开头的子命令
pub fn level_prefix_possible_cmd(&self, level: usize, prefix: &str) -> Vec<String> {
let mut subcmds = vec![];
let cmds = self.subcommands.clone();
for iterm in cmds {
if iterm.level == level && iterm.command_name. starts_with(prefix) {
subcmds.push(iterm.command_name);
}
}
return subcmds;
}
//获取某level 下某subcommand的所有子命令
pub fn level_cmd_possible_sub_cmd(&self, level: usize, cmd: String) -> Vec<String> {
let mut subcmds = vec![];
let cmds = self.subcommands.clone();
for iterm in cmds {
if iterm.level == level && iterm.command_name == cmd {
subcmds = iterm.subcommands.clone();
}
}
return subcmds;
}
//获取某level 下某subcommand的所有prefix子命令
pub fn level_cmd_possible_prefix_sub_cmd(
&self,
level: usize,
cmd: String,
prefix: &str,
) -> Vec<String> {
let mut subcmds = vec![];
let cmds = self.subcommands.clone();
for iterm in cmds {
if iterm.level == level && iterm.command_name == cmd {
for i in iterm.subcommands {
if i.starts_with(prefix) {
subcmds.push(i);
}
}
}
}
return subcmds;
}
pub fn complete_cmd(&self, line: &str, pos: usize) -> Result<(usize, Vec<Pair>)> {
let mut entries: Vec<Pair> = Vec::new();
let d: Vec<_> = line.split(' ').collect();
if d.len() == 1 {
if d.last() == Some(&"") {
for str in self.level_possible_cmd(1) {
let mut replace = str.clone();
replace.push_str(" ");
entries.push(Pair {
display: str.clone(),
replacement: replace,
});
}
return Ok((pos, entries));
}
if let Some(last) = d.last() {
for str in self.level_prefix_possible_cmd (1, *last) {
let mut replace = str.clone();
replace.push_str(" ");
entries.push(Pair {
display: str.clone(),
replacement: replace,
});
}
return Ok((pos - last.len(), entries));
}
}
if d.last() == Some(&"") {
for str in self
.level_cmd_possible_sub_cmd(d.len() - 1, d.get(d.len() - 2).unwrap().to_string())
{
let mut replace = str.clone();
replace.push_str(" ");
entries.push(Pair {
display: str.clone(),
replacement: replace,
});
}
return Ok((pos, entries));
}
if let Some(last) = d.last() {
for str in self. level_cmd_possible_prefix_sub_cmd(
d.len() - 1,
d.get(d.len() - 2).unwrap().to_string(),
*last,
) {
let mut replace = str.clone();
replace.push_str(" ");
entries.push(Pair {
display: str.clone(),
replacement: replace,
});
}
return Ok((pos - last.len(), entries));
}
Ok((pos, entries))
}
}
impl Completer for CommandCompleter {
type Candidate = Pair;
fn complete(&self, line: &str, pos: usize, _ctx: & Context<'_>) -> Result<(usize, Vec<Pair>)> {
self.complete_cmd(line, pos)
}
}
CommandCompleter 的实现部分比较多，大致包括两个部分，前一部分包括：获取某一级别下所有可能的子命令、获取某级别下某字符串开头的子命令、获取某级别下某个命令的所有子命令，等基本功能。这部分代码中有注释就不一一累述。
函数complete_cmd用来计算行中的位置以及在该位置的替换内容。
输入项是命令行的内容以及光标所在位置，输出项为在该位置需要替换的内容。比如，我们在提示符下输入 "root cm" root 下包含 cmd1、cmd2 两个子命令，此时如果按 'tab'键，complete_cmd 函数就会返回 (7,[cmd1,cmd2])。
作者：京东科技 贾世闻
来源：京东云开发者社区 转载请注明来源
