core-agent-ide/codex-rs/shell-command/src/bash.rs

591 lines
19 KiB
Rust
Raw Normal View History

use std::path::PathBuf;
use tree_sitter::Node;
feat: expand the set of commands that can be safely identified as "trusted" (#1668) This PR updates `is_known_safe_command()` to account for "safe operators" to expand the set of commands that can be run without approval. This concept existed in the TypeScript CLI, and we are [finally!] porting it to the Rust one: https://github.com/openai/codex/blob/c9e2def49487585cfe6f8bb7b2be442e8c0b5e1b/codex-cli/src/approvals.ts#L531-L541 The idea is that if we have `EXPR1 SAFE_OP EXPR2` and `EXPR1` and `EXPR2` are considered safe independently, then `EXPR1 SAFE_OP EXPR2` should be considered safe. Currently, `SAFE_OP` includes `&&`, `||`, `;`, and `|`. In the TypeScript implementation, we relied on https://www.npmjs.com/package/shell-quote to parse the string of Bash, as it could provide a "lightweight" parse tree, parsing `'beep || boop > /byte'` as: ``` [ 'beep', { op: '||' }, 'boop', { op: '>' }, '/byte' ] ``` Though in this PR, we introduce the use of https://crates.io/crates/tree-sitter-bash for parsing (which incidentally we were already using in [`codex-apply-patch`](https://github.com/openai/codex/blob/c9e2def49487585cfe6f8bb7b2be442e8c0b5e1b/codex-rs/apply-patch/Cargo.toml#L18)), which gives us a richer parse tree. (Incidentally, if you have never played with tree-sitter, try the [playground](https://tree-sitter.github.io/tree-sitter/7-playground.html) and select **Bash** from the dropdown to see how it parses various expressions.) As a concrete example, prior to this change, our implementation of `is_known_safe_command()` could verify things like: ``` ["bash", "-lc", "grep -R \"Cargo.toml\" -n"] ``` but not: ``` ["bash", "-lc", "grep -R \"Cargo.toml\" -n || true"] ``` With this change, the version with `|| true` is also accepted. Admittedly, this PR does not expand the safety check to support subshells, so it would reject, e.g. `bash -lc 'ls || (pwd && echo hi)'`, but that can be addressed in a subsequent PR.
2025-07-24 14:13:30 -07:00
use tree_sitter::Parser;
use tree_sitter::Tree;
use tree_sitter_bash::LANGUAGE as BASH;
# Split command parsing/safety out of `codex-core` into new `codex-command` (#11361) `codex-core` had accumulated command parsing and command safety logic (`bash`, `powershell`, `parse_command`, and `command_safety`) that is logically cohesive but orthogonal to most core session/runtime logic. Keeping this code in `codex-core` made the crate increasingly monolithic and raised iteration cost for unrelated core changes. This change extracts that surface into a dedicated crate, `codex-command`, while preserving existing `codex_core::...` call sites via re-exports. ## Why this refactor During analysis, command parsing/safety stood out as a good first split because it has: - a clear domain boundary (shell parsing + safety classification) - relatively self-contained dependencies (notably `tree-sitter` / `tree-sitter-bash`) - a meaningful standalone test surface (`134` tests moved with the crate) - many downstream uses that benefit from independent compilation and caching The practical problem was build latency from a large `codex-core` compile/test graph. Clean-build timings before and after this split showed measurable wins: - `cargo check -p codex-core`: `57.08s` -> `53.54s` (~`6.2%` faster) - `cargo test -p codex-core --no-run`: `2m39.9s` -> `2m20s` (~`12.4%` faster) - `codex-core lib` compile unit: `57.18s` -> `49.67s` (~`13.1%` faster) - `codex-core lib(test)` compile unit: `60.87s` -> `53.21s` (~`12.6%` faster) This gives a concrete reduction in core build overhead without changing behavior. ## What changed ### New crate - Added `codex-rs/command` as workspace crate `codex-command`. - Added: - `command/src/lib.rs` - `command/src/bash.rs` - `command/src/powershell.rs` - `command/src/parse_command.rs` - `command/src/command_safety/*` - `command/src/shell_detect.rs` - `command/BUILD.bazel` ### Code moved out of `codex-core` - Moved modules from `core/src` into `command/src`: - `bash.rs` - `powershell.rs` - `parse_command.rs` - `command_safety/*` ### Dependency graph updates - Added workspace member/dependency entries for `codex-command` in `codex-rs/Cargo.toml`. - Added `codex-command` dependency to `codex-rs/core/Cargo.toml`. - Removed `tree-sitter` and `tree-sitter-bash` from `codex-core` direct deps (now owned by `codex-command`). ### API compatibility for callers To avoid immediate downstream churn, `codex-core` now re-exports the moved modules/functions: - `codex_command::bash` - `codex_command::powershell` - `codex_command::parse_command` - `codex_command::is_safe_command` - `codex_command::is_dangerous_command` This keeps existing `codex_core::...` paths working while enabling gradual migration to direct `codex-command` usage. ### Internal decoupling detail - Added `command::shell_detect` so moved `bash`/`powershell` logic no longer depends on core shell internals. - Adjusted PowerShell helper visibility in `codex-command` for existing core test usage (`UTF8` prefix helper + executable discovery functions). ## Validation - `just fmt` - `just fix -p codex-command -p codex-core` - `cargo test -p codex-command` (`134` passed) - `cargo test -p codex-core --no-run` - `cargo test -p codex-core shell_command_handler` ## Notes / follow-up This commit intentionally prioritizes boundary extraction and compatibility. A follow-up can migrate downstream crates to depend directly on `codex-command` (instead of through `codex-core` re-exports) to realize additional incremental build wins.
2026-02-10 14:43:16 -08:00
use crate::shell_detect::ShellType;
use crate::shell_detect::detect_shell_type;
feat: expand the set of commands that can be safely identified as "trusted" (#1668) This PR updates `is_known_safe_command()` to account for "safe operators" to expand the set of commands that can be run without approval. This concept existed in the TypeScript CLI, and we are [finally!] porting it to the Rust one: https://github.com/openai/codex/blob/c9e2def49487585cfe6f8bb7b2be442e8c0b5e1b/codex-cli/src/approvals.ts#L531-L541 The idea is that if we have `EXPR1 SAFE_OP EXPR2` and `EXPR1` and `EXPR2` are considered safe independently, then `EXPR1 SAFE_OP EXPR2` should be considered safe. Currently, `SAFE_OP` includes `&&`, `||`, `;`, and `|`. In the TypeScript implementation, we relied on https://www.npmjs.com/package/shell-quote to parse the string of Bash, as it could provide a "lightweight" parse tree, parsing `'beep || boop > /byte'` as: ``` [ 'beep', { op: '||' }, 'boop', { op: '>' }, '/byte' ] ``` Though in this PR, we introduce the use of https://crates.io/crates/tree-sitter-bash for parsing (which incidentally we were already using in [`codex-apply-patch`](https://github.com/openai/codex/blob/c9e2def49487585cfe6f8bb7b2be442e8c0b5e1b/codex-rs/apply-patch/Cargo.toml#L18)), which gives us a richer parse tree. (Incidentally, if you have never played with tree-sitter, try the [playground](https://tree-sitter.github.io/tree-sitter/7-playground.html) and select **Bash** from the dropdown to see how it parses various expressions.) As a concrete example, prior to this change, our implementation of `is_known_safe_command()` could verify things like: ``` ["bash", "-lc", "grep -R \"Cargo.toml\" -n"] ``` but not: ``` ["bash", "-lc", "grep -R \"Cargo.toml\" -n || true"] ``` With this change, the version with `|| true` is also accepted. Admittedly, this PR does not expand the safety check to support subshells, so it would reject, e.g. `bash -lc 'ls || (pwd && echo hi)'`, but that can be addressed in a subsequent PR.
2025-07-24 14:13:30 -07:00
/// Parse the provided bash source using tree-sitter-bash, returning a Tree on
/// success or None if parsing failed.
pub fn try_parse_shell(shell_lc_arg: &str) -> Option<Tree> {
feat: expand the set of commands that can be safely identified as "trusted" (#1668) This PR updates `is_known_safe_command()` to account for "safe operators" to expand the set of commands that can be run without approval. This concept existed in the TypeScript CLI, and we are [finally!] porting it to the Rust one: https://github.com/openai/codex/blob/c9e2def49487585cfe6f8bb7b2be442e8c0b5e1b/codex-cli/src/approvals.ts#L531-L541 The idea is that if we have `EXPR1 SAFE_OP EXPR2` and `EXPR1` and `EXPR2` are considered safe independently, then `EXPR1 SAFE_OP EXPR2` should be considered safe. Currently, `SAFE_OP` includes `&&`, `||`, `;`, and `|`. In the TypeScript implementation, we relied on https://www.npmjs.com/package/shell-quote to parse the string of Bash, as it could provide a "lightweight" parse tree, parsing `'beep || boop > /byte'` as: ``` [ 'beep', { op: '||' }, 'boop', { op: '>' }, '/byte' ] ``` Though in this PR, we introduce the use of https://crates.io/crates/tree-sitter-bash for parsing (which incidentally we were already using in [`codex-apply-patch`](https://github.com/openai/codex/blob/c9e2def49487585cfe6f8bb7b2be442e8c0b5e1b/codex-rs/apply-patch/Cargo.toml#L18)), which gives us a richer parse tree. (Incidentally, if you have never played with tree-sitter, try the [playground](https://tree-sitter.github.io/tree-sitter/7-playground.html) and select **Bash** from the dropdown to see how it parses various expressions.) As a concrete example, prior to this change, our implementation of `is_known_safe_command()` could verify things like: ``` ["bash", "-lc", "grep -R \"Cargo.toml\" -n"] ``` but not: ``` ["bash", "-lc", "grep -R \"Cargo.toml\" -n || true"] ``` With this change, the version with `|| true` is also accepted. Admittedly, this PR does not expand the safety check to support subshells, so it would reject, e.g. `bash -lc 'ls || (pwd && echo hi)'`, but that can be addressed in a subsequent PR.
2025-07-24 14:13:30 -07:00
let lang = BASH.into();
let mut parser = Parser::new();
#[expect(clippy::expect_used)]
parser.set_language(&lang).expect("load bash grammar");
let old_tree: Option<&Tree> = None;
parser.parse(shell_lc_arg, old_tree)
feat: expand the set of commands that can be safely identified as "trusted" (#1668) This PR updates `is_known_safe_command()` to account for "safe operators" to expand the set of commands that can be run without approval. This concept existed in the TypeScript CLI, and we are [finally!] porting it to the Rust one: https://github.com/openai/codex/blob/c9e2def49487585cfe6f8bb7b2be442e8c0b5e1b/codex-cli/src/approvals.ts#L531-L541 The idea is that if we have `EXPR1 SAFE_OP EXPR2` and `EXPR1` and `EXPR2` are considered safe independently, then `EXPR1 SAFE_OP EXPR2` should be considered safe. Currently, `SAFE_OP` includes `&&`, `||`, `;`, and `|`. In the TypeScript implementation, we relied on https://www.npmjs.com/package/shell-quote to parse the string of Bash, as it could provide a "lightweight" parse tree, parsing `'beep || boop > /byte'` as: ``` [ 'beep', { op: '||' }, 'boop', { op: '>' }, '/byte' ] ``` Though in this PR, we introduce the use of https://crates.io/crates/tree-sitter-bash for parsing (which incidentally we were already using in [`codex-apply-patch`](https://github.com/openai/codex/blob/c9e2def49487585cfe6f8bb7b2be442e8c0b5e1b/codex-rs/apply-patch/Cargo.toml#L18)), which gives us a richer parse tree. (Incidentally, if you have never played with tree-sitter, try the [playground](https://tree-sitter.github.io/tree-sitter/7-playground.html) and select **Bash** from the dropdown to see how it parses various expressions.) As a concrete example, prior to this change, our implementation of `is_known_safe_command()` could verify things like: ``` ["bash", "-lc", "grep -R \"Cargo.toml\" -n"] ``` but not: ``` ["bash", "-lc", "grep -R \"Cargo.toml\" -n || true"] ``` With this change, the version with `|| true` is also accepted. Admittedly, this PR does not expand the safety check to support subshells, so it would reject, e.g. `bash -lc 'ls || (pwd && echo hi)'`, but that can be addressed in a subsequent PR.
2025-07-24 14:13:30 -07:00
}
/// Parse a script which may contain multiple simple commands joined only by
/// the safe logical/pipe/sequencing operators: `&&`, `||`, `;`, `|`.
///
/// Returns `Some(Vec<command_words>)` if every command is a plain wordonly
/// command and the parse tree does not contain disallowed constructs
/// (parentheses, redirections, substitutions, control flow, etc.). Otherwise
/// returns `None`.
pub fn try_parse_word_only_commands_sequence(tree: &Tree, src: &str) -> Option<Vec<Vec<String>>> {
if tree.root_node().has_error() {
return None;
}
// List of allowed (named) node kinds for a "word only commands sequence".
// If we encounter a named node that is not in this list we reject.
const ALLOWED_KINDS: &[&str] = &[
// top level containers
"program",
"list",
"pipeline",
// commands & words
"command",
"command_name",
"word",
"string",
"string_content",
"raw_string",
"number",
"concatenation",
feat: expand the set of commands that can be safely identified as "trusted" (#1668) This PR updates `is_known_safe_command()` to account for "safe operators" to expand the set of commands that can be run without approval. This concept existed in the TypeScript CLI, and we are [finally!] porting it to the Rust one: https://github.com/openai/codex/blob/c9e2def49487585cfe6f8bb7b2be442e8c0b5e1b/codex-cli/src/approvals.ts#L531-L541 The idea is that if we have `EXPR1 SAFE_OP EXPR2` and `EXPR1` and `EXPR2` are considered safe independently, then `EXPR1 SAFE_OP EXPR2` should be considered safe. Currently, `SAFE_OP` includes `&&`, `||`, `;`, and `|`. In the TypeScript implementation, we relied on https://www.npmjs.com/package/shell-quote to parse the string of Bash, as it could provide a "lightweight" parse tree, parsing `'beep || boop > /byte'` as: ``` [ 'beep', { op: '||' }, 'boop', { op: '>' }, '/byte' ] ``` Though in this PR, we introduce the use of https://crates.io/crates/tree-sitter-bash for parsing (which incidentally we were already using in [`codex-apply-patch`](https://github.com/openai/codex/blob/c9e2def49487585cfe6f8bb7b2be442e8c0b5e1b/codex-rs/apply-patch/Cargo.toml#L18)), which gives us a richer parse tree. (Incidentally, if you have never played with tree-sitter, try the [playground](https://tree-sitter.github.io/tree-sitter/7-playground.html) and select **Bash** from the dropdown to see how it parses various expressions.) As a concrete example, prior to this change, our implementation of `is_known_safe_command()` could verify things like: ``` ["bash", "-lc", "grep -R \"Cargo.toml\" -n"] ``` but not: ``` ["bash", "-lc", "grep -R \"Cargo.toml\" -n || true"] ``` With this change, the version with `|| true` is also accepted. Admittedly, this PR does not expand the safety check to support subshells, so it would reject, e.g. `bash -lc 'ls || (pwd && echo hi)'`, but that can be addressed in a subsequent PR.
2025-07-24 14:13:30 -07:00
];
// Allow only safe punctuation / operator tokens; anything else causes reject.
const ALLOWED_PUNCT_TOKENS: &[&str] = &["&&", "||", ";", "|", "\"", "'"];
let root = tree.root_node();
let mut cursor = root.walk();
let mut stack = vec![root];
let mut command_nodes = Vec::new();
while let Some(node) = stack.pop() {
let kind = node.kind();
if node.is_named() {
if !ALLOWED_KINDS.contains(&kind) {
return None;
}
if kind == "command" {
command_nodes.push(node);
}
} else {
// Reject any punctuation / operator tokens that are not explicitly allowed.
if kind.chars().any(|c| "&;|".contains(c)) && !ALLOWED_PUNCT_TOKENS.contains(&kind) {
return None;
}
if !(ALLOWED_PUNCT_TOKENS.contains(&kind) || kind.trim().is_empty()) {
// If it's a quote token or operator it's allowed above; we also allow whitespace tokens.
// Any other punctuation like parentheses, braces, redirects, backticks, etc are rejected.
return None;
}
}
for child in node.children(&mut cursor) {
stack.push(child);
}
}
// Walk uses a stack (LIFO), so re-sort by position to restore source order.
command_nodes.sort_by_key(Node::start_byte);
feat: expand the set of commands that can be safely identified as "trusted" (#1668) This PR updates `is_known_safe_command()` to account for "safe operators" to expand the set of commands that can be run without approval. This concept existed in the TypeScript CLI, and we are [finally!] porting it to the Rust one: https://github.com/openai/codex/blob/c9e2def49487585cfe6f8bb7b2be442e8c0b5e1b/codex-cli/src/approvals.ts#L531-L541 The idea is that if we have `EXPR1 SAFE_OP EXPR2` and `EXPR1` and `EXPR2` are considered safe independently, then `EXPR1 SAFE_OP EXPR2` should be considered safe. Currently, `SAFE_OP` includes `&&`, `||`, `;`, and `|`. In the TypeScript implementation, we relied on https://www.npmjs.com/package/shell-quote to parse the string of Bash, as it could provide a "lightweight" parse tree, parsing `'beep || boop > /byte'` as: ``` [ 'beep', { op: '||' }, 'boop', { op: '>' }, '/byte' ] ``` Though in this PR, we introduce the use of https://crates.io/crates/tree-sitter-bash for parsing (which incidentally we were already using in [`codex-apply-patch`](https://github.com/openai/codex/blob/c9e2def49487585cfe6f8bb7b2be442e8c0b5e1b/codex-rs/apply-patch/Cargo.toml#L18)), which gives us a richer parse tree. (Incidentally, if you have never played with tree-sitter, try the [playground](https://tree-sitter.github.io/tree-sitter/7-playground.html) and select **Bash** from the dropdown to see how it parses various expressions.) As a concrete example, prior to this change, our implementation of `is_known_safe_command()` could verify things like: ``` ["bash", "-lc", "grep -R \"Cargo.toml\" -n"] ``` but not: ``` ["bash", "-lc", "grep -R \"Cargo.toml\" -n || true"] ``` With this change, the version with `|| true` is also accepted. Admittedly, this PR does not expand the safety check to support subshells, so it would reject, e.g. `bash -lc 'ls || (pwd && echo hi)'`, but that can be addressed in a subsequent PR.
2025-07-24 14:13:30 -07:00
let mut commands = Vec::new();
for node in command_nodes {
if let Some(words) = parse_plain_command_from_node(node, src) {
commands.push(words);
} else {
return None;
}
}
Some(commands)
}
pub fn extract_bash_command(command: &[String]) -> Option<(&str, &str)> {
let [shell, flag, script] = command else {
return None;
};
if !matches!(flag.as_str(), "-lc" | "-c")
|| !matches!(
detect_shell_type(&PathBuf::from(shell)),
Some(ShellType::Zsh) | Some(ShellType::Bash) | Some(ShellType::Sh)
)
{
return None;
}
Some((shell, script))
}
/// Returns the sequence of plain commands within a `bash -lc "..."` or
/// `zsh -lc "..."` invocation when the script only contains word-only commands
/// joined by safe operators.
pub fn parse_shell_lc_plain_commands(command: &[String]) -> Option<Vec<Vec<String>>> {
let (_, script) = extract_bash_command(command)?;
let tree = try_parse_shell(script)?;
try_parse_word_only_commands_sequence(&tree, script)
}
/// Returns the parsed argv for a single shell command in a here-doc style
/// script (`<<`), as long as the script contains exactly one command node.
pub fn parse_shell_lc_single_command_prefix(command: &[String]) -> Option<Vec<String>> {
let (_, script) = extract_bash_command(command)?;
let tree = try_parse_shell(script)?;
let root = tree.root_node();
if root.has_error() {
return None;
}
if !has_named_descendant_kind(root, "heredoc_redirect") {
return None;
}
let command_node = find_single_command_node(root)?;
parse_heredoc_command_words(command_node, script)
}
feat: expand the set of commands that can be safely identified as "trusted" (#1668) This PR updates `is_known_safe_command()` to account for "safe operators" to expand the set of commands that can be run without approval. This concept existed in the TypeScript CLI, and we are [finally!] porting it to the Rust one: https://github.com/openai/codex/blob/c9e2def49487585cfe6f8bb7b2be442e8c0b5e1b/codex-cli/src/approvals.ts#L531-L541 The idea is that if we have `EXPR1 SAFE_OP EXPR2` and `EXPR1` and `EXPR2` are considered safe independently, then `EXPR1 SAFE_OP EXPR2` should be considered safe. Currently, `SAFE_OP` includes `&&`, `||`, `;`, and `|`. In the TypeScript implementation, we relied on https://www.npmjs.com/package/shell-quote to parse the string of Bash, as it could provide a "lightweight" parse tree, parsing `'beep || boop > /byte'` as: ``` [ 'beep', { op: '||' }, 'boop', { op: '>' }, '/byte' ] ``` Though in this PR, we introduce the use of https://crates.io/crates/tree-sitter-bash for parsing (which incidentally we were already using in [`codex-apply-patch`](https://github.com/openai/codex/blob/c9e2def49487585cfe6f8bb7b2be442e8c0b5e1b/codex-rs/apply-patch/Cargo.toml#L18)), which gives us a richer parse tree. (Incidentally, if you have never played with tree-sitter, try the [playground](https://tree-sitter.github.io/tree-sitter/7-playground.html) and select **Bash** from the dropdown to see how it parses various expressions.) As a concrete example, prior to this change, our implementation of `is_known_safe_command()` could verify things like: ``` ["bash", "-lc", "grep -R \"Cargo.toml\" -n"] ``` but not: ``` ["bash", "-lc", "grep -R \"Cargo.toml\" -n || true"] ``` With this change, the version with `|| true` is also accepted. Admittedly, this PR does not expand the safety check to support subshells, so it would reject, e.g. `bash -lc 'ls || (pwd && echo hi)'`, but that can be addressed in a subsequent PR.
2025-07-24 14:13:30 -07:00
fn parse_plain_command_from_node(cmd: tree_sitter::Node, src: &str) -> Option<Vec<String>> {
if cmd.kind() != "command" {
return None;
}
let mut words = Vec::new();
let mut cursor = cmd.walk();
for child in cmd.named_children(&mut cursor) {
match child.kind() {
"command_name" => {
let word_node = child.named_child(0)?;
if word_node.kind() != "word" {
return None;
}
words.push(word_node.utf8_text(src.as_bytes()).ok()?.to_owned());
}
"word" | "number" => {
words.push(child.utf8_text(src.as_bytes()).ok()?.to_owned());
}
"string" => {
let parsed = parse_double_quoted_string(child, src)?;
words.push(parsed);
feat: expand the set of commands that can be safely identified as "trusted" (#1668) This PR updates `is_known_safe_command()` to account for "safe operators" to expand the set of commands that can be run without approval. This concept existed in the TypeScript CLI, and we are [finally!] porting it to the Rust one: https://github.com/openai/codex/blob/c9e2def49487585cfe6f8bb7b2be442e8c0b5e1b/codex-cli/src/approvals.ts#L531-L541 The idea is that if we have `EXPR1 SAFE_OP EXPR2` and `EXPR1` and `EXPR2` are considered safe independently, then `EXPR1 SAFE_OP EXPR2` should be considered safe. Currently, `SAFE_OP` includes `&&`, `||`, `;`, and `|`. In the TypeScript implementation, we relied on https://www.npmjs.com/package/shell-quote to parse the string of Bash, as it could provide a "lightweight" parse tree, parsing `'beep || boop > /byte'` as: ``` [ 'beep', { op: '||' }, 'boop', { op: '>' }, '/byte' ] ``` Though in this PR, we introduce the use of https://crates.io/crates/tree-sitter-bash for parsing (which incidentally we were already using in [`codex-apply-patch`](https://github.com/openai/codex/blob/c9e2def49487585cfe6f8bb7b2be442e8c0b5e1b/codex-rs/apply-patch/Cargo.toml#L18)), which gives us a richer parse tree. (Incidentally, if you have never played with tree-sitter, try the [playground](https://tree-sitter.github.io/tree-sitter/7-playground.html) and select **Bash** from the dropdown to see how it parses various expressions.) As a concrete example, prior to this change, our implementation of `is_known_safe_command()` could verify things like: ``` ["bash", "-lc", "grep -R \"Cargo.toml\" -n"] ``` but not: ``` ["bash", "-lc", "grep -R \"Cargo.toml\" -n || true"] ``` With this change, the version with `|| true` is also accepted. Admittedly, this PR does not expand the safety check to support subshells, so it would reject, e.g. `bash -lc 'ls || (pwd && echo hi)'`, but that can be addressed in a subsequent PR.
2025-07-24 14:13:30 -07:00
}
"raw_string" => {
let parsed = parse_raw_string(child, src)?;
words.push(parsed);
feat: expand the set of commands that can be safely identified as "trusted" (#1668) This PR updates `is_known_safe_command()` to account for "safe operators" to expand the set of commands that can be run without approval. This concept existed in the TypeScript CLI, and we are [finally!] porting it to the Rust one: https://github.com/openai/codex/blob/c9e2def49487585cfe6f8bb7b2be442e8c0b5e1b/codex-cli/src/approvals.ts#L531-L541 The idea is that if we have `EXPR1 SAFE_OP EXPR2` and `EXPR1` and `EXPR2` are considered safe independently, then `EXPR1 SAFE_OP EXPR2` should be considered safe. Currently, `SAFE_OP` includes `&&`, `||`, `;`, and `|`. In the TypeScript implementation, we relied on https://www.npmjs.com/package/shell-quote to parse the string of Bash, as it could provide a "lightweight" parse tree, parsing `'beep || boop > /byte'` as: ``` [ 'beep', { op: '||' }, 'boop', { op: '>' }, '/byte' ] ``` Though in this PR, we introduce the use of https://crates.io/crates/tree-sitter-bash for parsing (which incidentally we were already using in [`codex-apply-patch`](https://github.com/openai/codex/blob/c9e2def49487585cfe6f8bb7b2be442e8c0b5e1b/codex-rs/apply-patch/Cargo.toml#L18)), which gives us a richer parse tree. (Incidentally, if you have never played with tree-sitter, try the [playground](https://tree-sitter.github.io/tree-sitter/7-playground.html) and select **Bash** from the dropdown to see how it parses various expressions.) As a concrete example, prior to this change, our implementation of `is_known_safe_command()` could verify things like: ``` ["bash", "-lc", "grep -R \"Cargo.toml\" -n"] ``` but not: ``` ["bash", "-lc", "grep -R \"Cargo.toml\" -n || true"] ``` With this change, the version with `|| true` is also accepted. Admittedly, this PR does not expand the safety check to support subshells, so it would reject, e.g. `bash -lc 'ls || (pwd && echo hi)'`, but that can be addressed in a subsequent PR.
2025-07-24 14:13:30 -07:00
}
"concatenation" => {
// Handle concatenated arguments like -g"*.py"
let mut concatenated = String::new();
let mut concat_cursor = child.walk();
for part in child.named_children(&mut concat_cursor) {
match part.kind() {
"word" | "number" => {
concatenated
.push_str(part.utf8_text(src.as_bytes()).ok()?.to_owned().as_str());
}
"string" => {
let parsed = parse_double_quoted_string(part, src)?;
concatenated.push_str(&parsed);
}
"raw_string" => {
let parsed = parse_raw_string(part, src)?;
concatenated.push_str(&parsed);
}
_ => return None,
}
}
if concatenated.is_empty() {
return None;
}
words.push(concatenated);
}
feat: expand the set of commands that can be safely identified as "trusted" (#1668) This PR updates `is_known_safe_command()` to account for "safe operators" to expand the set of commands that can be run without approval. This concept existed in the TypeScript CLI, and we are [finally!] porting it to the Rust one: https://github.com/openai/codex/blob/c9e2def49487585cfe6f8bb7b2be442e8c0b5e1b/codex-cli/src/approvals.ts#L531-L541 The idea is that if we have `EXPR1 SAFE_OP EXPR2` and `EXPR1` and `EXPR2` are considered safe independently, then `EXPR1 SAFE_OP EXPR2` should be considered safe. Currently, `SAFE_OP` includes `&&`, `||`, `;`, and `|`. In the TypeScript implementation, we relied on https://www.npmjs.com/package/shell-quote to parse the string of Bash, as it could provide a "lightweight" parse tree, parsing `'beep || boop > /byte'` as: ``` [ 'beep', { op: '||' }, 'boop', { op: '>' }, '/byte' ] ``` Though in this PR, we introduce the use of https://crates.io/crates/tree-sitter-bash for parsing (which incidentally we were already using in [`codex-apply-patch`](https://github.com/openai/codex/blob/c9e2def49487585cfe6f8bb7b2be442e8c0b5e1b/codex-rs/apply-patch/Cargo.toml#L18)), which gives us a richer parse tree. (Incidentally, if you have never played with tree-sitter, try the [playground](https://tree-sitter.github.io/tree-sitter/7-playground.html) and select **Bash** from the dropdown to see how it parses various expressions.) As a concrete example, prior to this change, our implementation of `is_known_safe_command()` could verify things like: ``` ["bash", "-lc", "grep -R \"Cargo.toml\" -n"] ``` but not: ``` ["bash", "-lc", "grep -R \"Cargo.toml\" -n || true"] ``` With this change, the version with `|| true` is also accepted. Admittedly, this PR does not expand the safety check to support subshells, so it would reject, e.g. `bash -lc 'ls || (pwd && echo hi)'`, but that can be addressed in a subsequent PR.
2025-07-24 14:13:30 -07:00
_ => return None,
}
}
Some(words)
}
fn parse_heredoc_command_words(cmd: Node<'_>, src: &str) -> Option<Vec<String>> {
if cmd.kind() != "command" {
return None;
}
let mut words = Vec::new();
let mut cursor = cmd.walk();
for child in cmd.named_children(&mut cursor) {
match child.kind() {
"command_name" => {
let word_node = child.named_child(0)?;
if !matches!(word_node.kind(), "word" | "number")
|| !is_literal_word_or_number(word_node)
{
return None;
}
words.push(word_node.utf8_text(src.as_bytes()).ok()?.to_owned());
}
"word" | "number" => {
if !is_literal_word_or_number(child) {
return None;
}
words.push(child.utf8_text(src.as_bytes()).ok()?.to_owned());
}
// Allow shell constructs that attach IO to a single command without
// changing argv matching semantics for the executable prefix.
"variable_assignment" | "comment" => {}
kind if is_allowed_heredoc_attachment_kind(kind) => {}
_ => return None,
}
}
if words.is_empty() { None } else { Some(words) }
}
fn is_literal_word_or_number(node: Node<'_>) -> bool {
if !matches!(node.kind(), "word" | "number") {
return false;
}
let mut cursor = node.walk();
node.named_children(&mut cursor).next().is_none()
}
fn is_allowed_heredoc_attachment_kind(kind: &str) -> bool {
matches!(
kind,
"heredoc_body"
| "simple_heredoc_body"
| "heredoc_redirect"
| "herestring_redirect"
| "file_redirect"
| "redirected_statement"
)
}
fn find_single_command_node(root: Node<'_>) -> Option<Node<'_>> {
let mut stack = vec![root];
let mut single_command = None;
while let Some(node) = stack.pop() {
if node.kind() == "command" {
if single_command.is_some() {
return None;
}
single_command = Some(node);
}
let mut cursor = node.walk();
for child in node.named_children(&mut cursor) {
stack.push(child);
}
}
single_command
}
fn has_named_descendant_kind(node: Node<'_>, kind: &str) -> bool {
let mut stack = vec![node];
while let Some(current) = stack.pop() {
if current.kind() == kind {
return true;
}
let mut cursor = current.walk();
for child in current.named_children(&mut cursor) {
stack.push(child);
}
}
false
}
fn parse_double_quoted_string(node: Node, src: &str) -> Option<String> {
if node.kind() != "string" {
return None;
}
let mut cursor = node.walk();
for part in node.named_children(&mut cursor) {
if part.kind() != "string_content" {
return None;
}
}
let raw = node.utf8_text(src.as_bytes()).ok()?;
let stripped = raw
.strip_prefix('"')
.and_then(|text| text.strip_suffix('"'))?;
Some(stripped.to_string())
}
fn parse_raw_string(node: Node, src: &str) -> Option<String> {
if node.kind() != "raw_string" {
return None;
}
let raw_string = node.utf8_text(src.as_bytes()).ok()?;
let stripped = raw_string
.strip_prefix('\'')
.and_then(|s| s.strip_suffix('\''));
stripped.map(str::to_owned)
}
feat: expand the set of commands that can be safely identified as "trusted" (#1668) This PR updates `is_known_safe_command()` to account for "safe operators" to expand the set of commands that can be run without approval. This concept existed in the TypeScript CLI, and we are [finally!] porting it to the Rust one: https://github.com/openai/codex/blob/c9e2def49487585cfe6f8bb7b2be442e8c0b5e1b/codex-cli/src/approvals.ts#L531-L541 The idea is that if we have `EXPR1 SAFE_OP EXPR2` and `EXPR1` and `EXPR2` are considered safe independently, then `EXPR1 SAFE_OP EXPR2` should be considered safe. Currently, `SAFE_OP` includes `&&`, `||`, `;`, and `|`. In the TypeScript implementation, we relied on https://www.npmjs.com/package/shell-quote to parse the string of Bash, as it could provide a "lightweight" parse tree, parsing `'beep || boop > /byte'` as: ``` [ 'beep', { op: '||' }, 'boop', { op: '>' }, '/byte' ] ``` Though in this PR, we introduce the use of https://crates.io/crates/tree-sitter-bash for parsing (which incidentally we were already using in [`codex-apply-patch`](https://github.com/openai/codex/blob/c9e2def49487585cfe6f8bb7b2be442e8c0b5e1b/codex-rs/apply-patch/Cargo.toml#L18)), which gives us a richer parse tree. (Incidentally, if you have never played with tree-sitter, try the [playground](https://tree-sitter.github.io/tree-sitter/7-playground.html) and select **Bash** from the dropdown to see how it parses various expressions.) As a concrete example, prior to this change, our implementation of `is_known_safe_command()` could verify things like: ``` ["bash", "-lc", "grep -R \"Cargo.toml\" -n"] ``` but not: ``` ["bash", "-lc", "grep -R \"Cargo.toml\" -n || true"] ``` With this change, the version with `|| true` is also accepted. Admittedly, this PR does not expand the safety check to support subshells, so it would reject, e.g. `bash -lc 'ls || (pwd && echo hi)'`, but that can be addressed in a subsequent PR.
2025-07-24 14:13:30 -07:00
#[cfg(test)]
mod tests {
use super::*;
use pretty_assertions::assert_eq;
feat: expand the set of commands that can be safely identified as "trusted" (#1668) This PR updates `is_known_safe_command()` to account for "safe operators" to expand the set of commands that can be run without approval. This concept existed in the TypeScript CLI, and we are [finally!] porting it to the Rust one: https://github.com/openai/codex/blob/c9e2def49487585cfe6f8bb7b2be442e8c0b5e1b/codex-cli/src/approvals.ts#L531-L541 The idea is that if we have `EXPR1 SAFE_OP EXPR2` and `EXPR1` and `EXPR2` are considered safe independently, then `EXPR1 SAFE_OP EXPR2` should be considered safe. Currently, `SAFE_OP` includes `&&`, `||`, `;`, and `|`. In the TypeScript implementation, we relied on https://www.npmjs.com/package/shell-quote to parse the string of Bash, as it could provide a "lightweight" parse tree, parsing `'beep || boop > /byte'` as: ``` [ 'beep', { op: '||' }, 'boop', { op: '>' }, '/byte' ] ``` Though in this PR, we introduce the use of https://crates.io/crates/tree-sitter-bash for parsing (which incidentally we were already using in [`codex-apply-patch`](https://github.com/openai/codex/blob/c9e2def49487585cfe6f8bb7b2be442e8c0b5e1b/codex-rs/apply-patch/Cargo.toml#L18)), which gives us a richer parse tree. (Incidentally, if you have never played with tree-sitter, try the [playground](https://tree-sitter.github.io/tree-sitter/7-playground.html) and select **Bash** from the dropdown to see how it parses various expressions.) As a concrete example, prior to this change, our implementation of `is_known_safe_command()` could verify things like: ``` ["bash", "-lc", "grep -R \"Cargo.toml\" -n"] ``` but not: ``` ["bash", "-lc", "grep -R \"Cargo.toml\" -n || true"] ``` With this change, the version with `|| true` is also accepted. Admittedly, this PR does not expand the safety check to support subshells, so it would reject, e.g. `bash -lc 'ls || (pwd && echo hi)'`, but that can be addressed in a subsequent PR.
2025-07-24 14:13:30 -07:00
fn parse_seq(src: &str) -> Option<Vec<Vec<String>>> {
let tree = try_parse_shell(src)?;
feat: expand the set of commands that can be safely identified as "trusted" (#1668) This PR updates `is_known_safe_command()` to account for "safe operators" to expand the set of commands that can be run without approval. This concept existed in the TypeScript CLI, and we are [finally!] porting it to the Rust one: https://github.com/openai/codex/blob/c9e2def49487585cfe6f8bb7b2be442e8c0b5e1b/codex-cli/src/approvals.ts#L531-L541 The idea is that if we have `EXPR1 SAFE_OP EXPR2` and `EXPR1` and `EXPR2` are considered safe independently, then `EXPR1 SAFE_OP EXPR2` should be considered safe. Currently, `SAFE_OP` includes `&&`, `||`, `;`, and `|`. In the TypeScript implementation, we relied on https://www.npmjs.com/package/shell-quote to parse the string of Bash, as it could provide a "lightweight" parse tree, parsing `'beep || boop > /byte'` as: ``` [ 'beep', { op: '||' }, 'boop', { op: '>' }, '/byte' ] ``` Though in this PR, we introduce the use of https://crates.io/crates/tree-sitter-bash for parsing (which incidentally we were already using in [`codex-apply-patch`](https://github.com/openai/codex/blob/c9e2def49487585cfe6f8bb7b2be442e8c0b5e1b/codex-rs/apply-patch/Cargo.toml#L18)), which gives us a richer parse tree. (Incidentally, if you have never played with tree-sitter, try the [playground](https://tree-sitter.github.io/tree-sitter/7-playground.html) and select **Bash** from the dropdown to see how it parses various expressions.) As a concrete example, prior to this change, our implementation of `is_known_safe_command()` could verify things like: ``` ["bash", "-lc", "grep -R \"Cargo.toml\" -n"] ``` but not: ``` ["bash", "-lc", "grep -R \"Cargo.toml\" -n || true"] ``` With this change, the version with `|| true` is also accepted. Admittedly, this PR does not expand the safety check to support subshells, so it would reject, e.g. `bash -lc 'ls || (pwd && echo hi)'`, but that can be addressed in a subsequent PR.
2025-07-24 14:13:30 -07:00
try_parse_word_only_commands_sequence(&tree, src)
}
#[test]
fn accepts_single_simple_command() {
let cmds = parse_seq("ls -1").unwrap();
assert_eq!(cmds, vec![vec!["ls".to_string(), "-1".to_string()]]);
}
#[test]
fn accepts_multiple_commands_with_allowed_operators() {
let src = "ls && pwd; echo 'hi there' | wc -l";
let cmds = parse_seq(src).unwrap();
let expected: Vec<Vec<String>> = vec![
vec!["ls".to_string()],
vec!["pwd".to_string()],
vec!["echo".to_string(), "hi there".to_string()],
vec!["wc".to_string(), "-l".to_string()],
feat: expand the set of commands that can be safely identified as "trusted" (#1668) This PR updates `is_known_safe_command()` to account for "safe operators" to expand the set of commands that can be run without approval. This concept existed in the TypeScript CLI, and we are [finally!] porting it to the Rust one: https://github.com/openai/codex/blob/c9e2def49487585cfe6f8bb7b2be442e8c0b5e1b/codex-cli/src/approvals.ts#L531-L541 The idea is that if we have `EXPR1 SAFE_OP EXPR2` and `EXPR1` and `EXPR2` are considered safe independently, then `EXPR1 SAFE_OP EXPR2` should be considered safe. Currently, `SAFE_OP` includes `&&`, `||`, `;`, and `|`. In the TypeScript implementation, we relied on https://www.npmjs.com/package/shell-quote to parse the string of Bash, as it could provide a "lightweight" parse tree, parsing `'beep || boop > /byte'` as: ``` [ 'beep', { op: '||' }, 'boop', { op: '>' }, '/byte' ] ``` Though in this PR, we introduce the use of https://crates.io/crates/tree-sitter-bash for parsing (which incidentally we were already using in [`codex-apply-patch`](https://github.com/openai/codex/blob/c9e2def49487585cfe6f8bb7b2be442e8c0b5e1b/codex-rs/apply-patch/Cargo.toml#L18)), which gives us a richer parse tree. (Incidentally, if you have never played with tree-sitter, try the [playground](https://tree-sitter.github.io/tree-sitter/7-playground.html) and select **Bash** from the dropdown to see how it parses various expressions.) As a concrete example, prior to this change, our implementation of `is_known_safe_command()` could verify things like: ``` ["bash", "-lc", "grep -R \"Cargo.toml\" -n"] ``` but not: ``` ["bash", "-lc", "grep -R \"Cargo.toml\" -n || true"] ``` With this change, the version with `|| true` is also accepted. Admittedly, this PR does not expand the safety check to support subshells, so it would reject, e.g. `bash -lc 'ls || (pwd && echo hi)'`, but that can be addressed in a subsequent PR.
2025-07-24 14:13:30 -07:00
];
assert_eq!(cmds, expected);
}
#[test]
fn extracts_double_and_single_quoted_strings() {
let cmds = parse_seq("echo \"hello world\"").unwrap();
assert_eq!(
cmds,
vec![vec!["echo".to_string(), "hello world".to_string()]]
);
let cmds2 = parse_seq("echo 'hi there'").unwrap();
assert_eq!(
cmds2,
vec![vec!["echo".to_string(), "hi there".to_string()]]
);
}
#[test]
fn accepts_double_quoted_strings_with_newlines() {
let cmds = parse_seq("git commit -m \"line1\nline2\"").unwrap();
assert_eq!(
cmds,
vec![vec![
"git".to_string(),
"commit".to_string(),
"-m".to_string(),
"line1\nline2".to_string(),
]]
);
}
#[test]
fn accepts_mixed_quote_concatenation() {
assert_eq!(
parse_seq(r#"echo "/usr"'/'"local"/bin"#).unwrap(),
vec![vec!["echo".to_string(), "/usr/local/bin".to_string()]]
);
assert_eq!(
parse_seq(r#"echo '/usr'"/"'local'/bin"#).unwrap(),
vec![vec!["echo".to_string(), "/usr/local/bin".to_string()]]
);
}
#[test]
fn rejects_double_quoted_strings_with_expansions() {
assert!(parse_seq(r#"echo "hi ${USER}""#).is_none());
assert!(parse_seq(r#"echo "$HOME""#).is_none());
}
feat: expand the set of commands that can be safely identified as "trusted" (#1668) This PR updates `is_known_safe_command()` to account for "safe operators" to expand the set of commands that can be run without approval. This concept existed in the TypeScript CLI, and we are [finally!] porting it to the Rust one: https://github.com/openai/codex/blob/c9e2def49487585cfe6f8bb7b2be442e8c0b5e1b/codex-cli/src/approvals.ts#L531-L541 The idea is that if we have `EXPR1 SAFE_OP EXPR2` and `EXPR1` and `EXPR2` are considered safe independently, then `EXPR1 SAFE_OP EXPR2` should be considered safe. Currently, `SAFE_OP` includes `&&`, `||`, `;`, and `|`. In the TypeScript implementation, we relied on https://www.npmjs.com/package/shell-quote to parse the string of Bash, as it could provide a "lightweight" parse tree, parsing `'beep || boop > /byte'` as: ``` [ 'beep', { op: '||' }, 'boop', { op: '>' }, '/byte' ] ``` Though in this PR, we introduce the use of https://crates.io/crates/tree-sitter-bash for parsing (which incidentally we were already using in [`codex-apply-patch`](https://github.com/openai/codex/blob/c9e2def49487585cfe6f8bb7b2be442e8c0b5e1b/codex-rs/apply-patch/Cargo.toml#L18)), which gives us a richer parse tree. (Incidentally, if you have never played with tree-sitter, try the [playground](https://tree-sitter.github.io/tree-sitter/7-playground.html) and select **Bash** from the dropdown to see how it parses various expressions.) As a concrete example, prior to this change, our implementation of `is_known_safe_command()` could verify things like: ``` ["bash", "-lc", "grep -R \"Cargo.toml\" -n"] ``` but not: ``` ["bash", "-lc", "grep -R \"Cargo.toml\" -n || true"] ``` With this change, the version with `|| true` is also accepted. Admittedly, this PR does not expand the safety check to support subshells, so it would reject, e.g. `bash -lc 'ls || (pwd && echo hi)'`, but that can be addressed in a subsequent PR.
2025-07-24 14:13:30 -07:00
#[test]
fn accepts_numbers_as_words() {
let cmds = parse_seq("echo 123 456").unwrap();
assert_eq!(
cmds,
vec![vec![
"echo".to_string(),
"123".to_string(),
"456".to_string()
]]
);
}
#[test]
fn rejects_parentheses_and_subshells() {
assert!(parse_seq("(ls)").is_none());
assert!(parse_seq("ls || (pwd && echo hi)").is_none());
}
#[test]
fn rejects_redirections_and_unsupported_operators() {
assert!(parse_seq("ls > out.txt").is_none());
assert!(parse_seq("echo hi & echo bye").is_none());
}
#[test]
fn rejects_command_and_process_substitutions_and_expansions() {
assert!(parse_seq("echo $(pwd)").is_none());
assert!(parse_seq("echo `pwd`").is_none());
assert!(parse_seq("echo $HOME").is_none());
assert!(parse_seq("echo \"hi $USER\"").is_none());
}
#[test]
fn rejects_variable_assignment_prefix() {
assert!(parse_seq("FOO=bar ls").is_none());
}
#[test]
fn rejects_trailing_operator_parse_error() {
assert!(parse_seq("ls &&").is_none());
}
#[test]
fn rejects_empty_command_position_with_leading_operator() {
assert!(parse_seq("&& ls").is_none());
}
#[test]
fn rejects_empty_command_position_with_double_separator() {
assert!(parse_seq("ls ;; pwd").is_none());
}
#[test]
fn rejects_empty_command_position_with_empty_pipeline_segment() {
assert!(parse_seq("ls | | wc").is_none());
}
#[test]
fn parse_zsh_lc_plain_commands() {
let command = vec!["zsh".to_string(), "-lc".to_string(), "ls".to_string()];
let parsed = parse_shell_lc_plain_commands(&command).unwrap();
assert_eq!(parsed, vec![vec!["ls".to_string()]]);
}
#[test]
fn accepts_concatenated_flag_and_value() {
// Test case: -g"*.py" (flag directly concatenated with quoted value)
let cmds = parse_seq("rg -n \"foo\" -g\"*.py\"").unwrap();
assert_eq!(
cmds,
vec![vec![
"rg".to_string(),
"-n".to_string(),
"foo".to_string(),
"-g*.py".to_string(),
]]
);
}
#[test]
fn accepts_concatenated_flag_with_single_quotes() {
let cmds = parse_seq("grep -n 'pattern' -g'*.txt'").unwrap();
assert_eq!(
cmds,
vec![vec![
"grep".to_string(),
"-n".to_string(),
"pattern".to_string(),
"-g*.txt".to_string(),
]]
);
}
#[test]
fn rejects_concatenation_with_variable_substitution() {
// Environment variables in concatenated strings should be rejected
assert!(parse_seq("rg -g\"$VAR\" pattern").is_none());
assert!(parse_seq("rg -g\"${VAR}\" pattern").is_none());
}
#[test]
fn rejects_concatenation_with_command_substitution() {
// Command substitution in concatenated strings should be rejected
assert!(parse_seq("rg -g\"$(pwd)\" pattern").is_none());
assert!(parse_seq("rg -g\"$(echo '*.py')\" pattern").is_none());
}
#[test]
fn parse_shell_lc_single_command_prefix_supports_heredoc() {
let command = vec![
"zsh".to_string(),
"-lc".to_string(),
"python3 <<'PY'\nprint('hello')\nPY".to_string(),
];
let parsed = parse_shell_lc_single_command_prefix(&command);
assert_eq!(parsed, Some(vec!["python3".to_string()]));
let command_unquoted = vec![
"zsh".to_string(),
"-lc".to_string(),
"python3 << PY\nprint('hello')\nPY".to_string(),
];
let parsed_unquoted = parse_shell_lc_single_command_prefix(&command_unquoted);
assert_eq!(parsed_unquoted, Some(vec!["python3".to_string()]));
}
#[test]
fn parse_shell_lc_single_command_prefix_rejects_multi_command_scripts() {
let command = vec![
"bash".to_string(),
"-lc".to_string(),
"python3 <<'PY'\nprint('hello')\nPY\necho done".to_string(),
];
assert_eq!(parse_shell_lc_single_command_prefix(&command), None);
}
#[test]
fn parse_shell_lc_single_command_prefix_rejects_non_heredoc_redirects() {
let command = vec![
"bash".to_string(),
"-lc".to_string(),
"echo hello > /tmp/out.txt".to_string(),
];
assert_eq!(parse_shell_lc_single_command_prefix(&command), None);
}
#[test]
fn parse_shell_lc_single_command_prefix_accepts_heredoc_with_extra_redirect() {
let command = vec![
"bash".to_string(),
"-lc".to_string(),
"python3 <<'PY' > /tmp/out.txt\nprint('hello')\nPY".to_string(),
];
assert_eq!(
parse_shell_lc_single_command_prefix(&command),
Some(vec!["python3".to_string()])
);
}
#[test]
fn parse_shell_lc_single_command_prefix_rejects_herestring_with_chaining() {
let command = vec![
"bash".to_string(),
"-lc".to_string(),
r#"echo hello > /tmp/out.txt && cat /tmp/out.txt"#.to_string(),
];
assert_eq!(parse_shell_lc_single_command_prefix(&command), None);
}
#[test]
fn parse_shell_lc_single_command_prefix_rejects_herestring_with_substitution() {
let command = vec![
"bash".to_string(),
"-lc".to_string(),
r#"python3 <<< "$(rm -rf /)""#.to_string(),
];
assert_eq!(parse_shell_lc_single_command_prefix(&command), None);
}
#[test]
fn parse_shell_lc_single_command_prefix_rejects_arithmetic_shift_non_heredoc_script() {
let command = vec![
"bash".to_string(),
"-lc".to_string(),
"echo $((1<<2))".to_string(),
];
assert_eq!(parse_shell_lc_single_command_prefix(&command), None);
}
#[test]
fn parse_shell_lc_single_command_prefix_rejects_heredoc_command_with_word_expansion() {
let command = vec![
"bash".to_string(),
"-lc".to_string(),
"python3 $((1<<2)) <<'PY'\nprint('hello')\nPY".to_string(),
];
assert_eq!(parse_shell_lc_single_command_prefix(&command), None);
}
feat: expand the set of commands that can be safely identified as "trusted" (#1668) This PR updates `is_known_safe_command()` to account for "safe operators" to expand the set of commands that can be run without approval. This concept existed in the TypeScript CLI, and we are [finally!] porting it to the Rust one: https://github.com/openai/codex/blob/c9e2def49487585cfe6f8bb7b2be442e8c0b5e1b/codex-cli/src/approvals.ts#L531-L541 The idea is that if we have `EXPR1 SAFE_OP EXPR2` and `EXPR1` and `EXPR2` are considered safe independently, then `EXPR1 SAFE_OP EXPR2` should be considered safe. Currently, `SAFE_OP` includes `&&`, `||`, `;`, and `|`. In the TypeScript implementation, we relied on https://www.npmjs.com/package/shell-quote to parse the string of Bash, as it could provide a "lightweight" parse tree, parsing `'beep || boop > /byte'` as: ``` [ 'beep', { op: '||' }, 'boop', { op: '>' }, '/byte' ] ``` Though in this PR, we introduce the use of https://crates.io/crates/tree-sitter-bash for parsing (which incidentally we were already using in [`codex-apply-patch`](https://github.com/openai/codex/blob/c9e2def49487585cfe6f8bb7b2be442e8c0b5e1b/codex-rs/apply-patch/Cargo.toml#L18)), which gives us a richer parse tree. (Incidentally, if you have never played with tree-sitter, try the [playground](https://tree-sitter.github.io/tree-sitter/7-playground.html) and select **Bash** from the dropdown to see how it parses various expressions.) As a concrete example, prior to this change, our implementation of `is_known_safe_command()` could verify things like: ``` ["bash", "-lc", "grep -R \"Cargo.toml\" -n"] ``` but not: ``` ["bash", "-lc", "grep -R \"Cargo.toml\" -n || true"] ``` With this change, the version with `|| true` is also accepted. Admittedly, this PR does not expand the safety check to support subshells, so it would reject, e.g. `bash -lc 'ls || (pwd && echo hi)'`, but that can be addressed in a subsequent PR.
2025-07-24 14:13:30 -07:00
}