sqlite_graphrag/commands/
claude_runner.rs1use crate::errors::AppError;
12use crate::spawn::env_whitelist::apply_env_whitelist;
13use std::path::Path;
14use std::process::{Command, Stdio};
15
16const MIN_CLAUDE_VERSION: &str = "2.1.0";
18
19#[cfg(target_os = "linux")]
21const DEFAULT_SUBPROCESS_MEMORY_LIMIT_MB: u64 = 4096;
22
23#[cfg(target_os = "linux")]
48pub fn spawn_with_memory_limit(cmd: &mut Command) -> std::io::Result<std::process::Child> {
49 use std::os::unix::process::CommandExt;
50 let max_mb: u64 = crate::runtime_config::resolve_u64(
51 None,
52 "spawn.subprocess_memory_limit_mb",
53 DEFAULT_SUBPROCESS_MEMORY_LIMIT_MB,
54 );
55 let max_bytes = max_mb * 1024 * 1024;
56 unsafe {
63 cmd.pre_exec(move || {
64 let sid = libc::setsid();
65 if sid == -1 {
66 let err = std::io::Error::last_os_error();
67 if err.raw_os_error() != Some(libc::EPERM) {
68 return Err(err);
69 }
70 }
71 let limit = libc::rlimit {
72 rlim_cur: max_bytes,
73 rlim_max: max_bytes,
74 };
75 if libc::setrlimit(libc::RLIMIT_AS, &limit) != 0 {
76 return Err(std::io::Error::last_os_error());
77 }
78 Ok(())
79 });
80 }
81 tracing::debug!(
82 target: "process",
83 program = ?cmd.get_program(),
84 args = ?cmd.get_args().collect::<Vec<_>>(),
85 "spawning external process"
86 );
87 cmd.spawn()
88}
89
90#[cfg(not(target_os = "linux"))]
93pub fn spawn_with_memory_limit(cmd: &mut Command) -> std::io::Result<std::process::Child> {
94 #[cfg(unix)]
95 {
96 use std::os::unix::process::CommandExt;
97 unsafe {
100 cmd.pre_exec(|| {
101 let sid = libc::setsid();
102 if sid == -1 {
103 let err = std::io::Error::last_os_error();
104 if err.raw_os_error() != Some(libc::EPERM) {
105 return Err(err);
106 }
107 }
108 Ok(())
109 });
110 }
111 }
112 tracing::debug!(
113 target: "process",
114 program = ?cmd.get_program(),
115 args = ?cmd.get_args().collect::<Vec<_>>(),
116 "spawning external process"
117 );
118 cmd.spawn()
119}
120
121#[derive(Debug, serde::Deserialize)]
123pub struct ClaudeOutputElement {
124 pub r#type: Option<String>,
126 pub subtype: Option<String>,
128 #[serde(default)]
130 pub is_error: bool,
131 pub structured_output: Option<serde_json::Value>,
133 pub result: Option<String>,
135 pub total_cost_usd: Option<f64>,
137 pub error: Option<String>,
139 pub terminal_reason: Option<String>,
141 #[serde(rename = "apiKeySource")]
143 pub api_key_source: Option<String>,
144}
145
146#[derive(Debug)]
148pub struct ClaudeResult {
149 pub value: serde_json::Value,
151 pub cost_usd: f64,
153 pub is_oauth: bool,
155}
156
157pub fn validate_claude_version(binary: &Path) -> Result<String, AppError> {
159 let resolved = which::which(binary).map_err(|_| {
160 AppError::Validation(crate::i18n::validation::executable_not_in_path_generic(
161 &binary.display().to_string(),
162 ))
163 })?;
164 let output = Command::new(&resolved)
165 .arg("--version")
166 .stdin(Stdio::null())
167 .stdout(Stdio::piped())
168 .stderr(Stdio::piped())
169 .output()
170 .map_err(AppError::Io)?;
171
172 if !output.status.success() {
173 return Err(AppError::Validation(
174 "failed to run 'claude --version'".to_string(),
175 ));
176 }
177
178 let version_str = String::from_utf8(output.stdout)
179 .map_err(|_| AppError::Validation(crate::i18n::validation::claude_version_not_utf8()))?;
180 let version = version_str.trim().to_string();
181 let numeric = version.split([' ', '(']).next().unwrap_or("").trim();
182
183 fn parse_semver(s: &str) -> Option<(u64, u64, u64)> {
184 let parts: Vec<&str> = s.splitn(3, '.').collect();
185 if parts.len() < 2 {
186 return None;
187 }
188 let major = parts[0].parse::<u64>().ok()?;
189 let minor = parts[1].parse::<u64>().ok()?;
190 let patch = parts
191 .get(2)
192 .and_then(|p| p.parse::<u64>().ok())
193 .unwrap_or(0);
194 Some((major, minor, patch))
195 }
196
197 if let (Some(actual), Some(min)) = (parse_semver(numeric), parse_semver(MIN_CLAUDE_VERSION)) {
198 if actual < min {
199 return Err(AppError::Validation(crate::i18n::validation::version_below_minimum(
200 "Claude Code",
201 numeric,
202 MIN_CLAUDE_VERSION,
203 )));
204 }
205 }
206
207 Ok(version)
208}
209
210pub fn build_claude_command(
247 binary: &Path,
248 prompt: &str,
249 json_schema: &str,
250 model: Option<&str>,
251 max_turns: u32,
252) -> Result<Command, crate::errors::AppError> {
253 if let Ok(_key) = std::env::var("ANTHROPIC_API_KEY") {
257 let mut cmd = crate::spawn::failing_command();
262 cmd.env_clear();
263 cmd.env("PATH", "/nonexistent");
264 cmd.arg("--oauth-only-violation-anthropic-api-key-set");
265 cmd.arg("--oauth-only-resolution-use-anthropic-auth-token");
266 return Ok(cmd);
267 }
268
269 let mut cmd = Command::new(binary);
270
271 apply_env_whitelist(&mut cmd, crate::spawn::env_whitelist::is_strict_env_clear());
276 crate::spawn::apply_cwd_isolation(&mut cmd)?;
277
278 let mcp_config_path = crate::spawn::preflight::write_empty_mcp_config_tempfile()?;
287
288 cmd.arg("-p")
289 .arg(prompt)
290 .arg("--strict-mcp-config")
291 .arg("--mcp-config")
292 .arg(mcp_config_path.as_os_str())
293 .arg("--dangerously-skip-permissions")
294 .arg("--settings")
295 .arg(r#"{"hooks":{}}"#)
296 .arg("--output-format")
297 .arg("json")
298 .arg("--json-schema")
299 .arg(json_schema)
300 .arg("--max-turns")
301 .arg(max_turns.to_string())
302 .arg("--no-session-persistence");
303
304 if let Some(m) = model {
305 cmd.arg("--model").arg(m);
306 }
307
308 cmd.stdin(Stdio::null())
309 .stdout(Stdio::piped())
310 .stderr(Stdio::piped());
311
312 let argv_refs: Vec<std::ffi::OsString> = cmd.get_args().map(|s| s.to_os_string()).collect();
318 let preflight_args = crate::spawn::preflight::PreFlightArgs {
319 binary_path: binary,
320 argv: &argv_refs,
321 workspace_root: std::path::Path::new("."),
322 mcp_config_inline_json: None,
323 expected_output_bytes: 65_536,
324 spawner_name: "claude_runner",
325 };
326 if let Err(e) = crate::spawn::preflight::preflight_check(&preflight_args) {
327 return Err(crate::errors::AppError::from(e));
333 }
334
335 Ok(cmd)
336}
337
338pub fn parse_claude_output(stdout: &str) -> Result<ClaudeResult, AppError> {
343 parse_claude_output_opts(stdout, false)
344}
345
346pub fn parse_claude_output_opts(
352 stdout: &str,
353 tolerate_max_turns: bool,
354) -> Result<ClaudeResult, AppError> {
355 let elements: Vec<ClaudeOutputElement> = serde_json::from_str(stdout).map_err(|e| {
356 AppError::Validation(crate::i18n::validation::failed_to_parse_claude_json_array(&e))
357 })?;
358
359 let is_oauth = elements
360 .iter()
361 .find(|e| e.r#type.as_deref() == Some("system") && e.subtype.as_deref() == Some("init"))
362 .and_then(|e| e.api_key_source.as_deref())
363 .map(|s| s == "none")
364 .unwrap_or(false);
365
366 let result_elem = elements
367 .iter()
368 .find(|e| e.r#type.as_deref() == Some("result"))
369 .ok_or_else(|| {
370 AppError::Validation(crate::i18n::validation::claude_output_missing_result())
371 })?;
372
373 if !tolerate_max_turns && result_elem.terminal_reason.as_deref() == Some("max_turns") {
375 tracing::warn!(
376 target: "claude_runner",
377 "claude -p hit max_turns limit — hooks may have consumed turns"
378 );
379 return Err(AppError::Validation(
380 "claude -p hit max_turns: hooks may be consuming turns; increase --max-turns or disable hooks".to_string(),
381 ));
382 }
383
384 if result_elem.is_error {
385 let err_msg = result_elem
386 .error
387 .as_deref()
388 .or(result_elem.result.as_deref())
389 .unwrap_or("unknown error");
390 if err_msg.contains("rate_limit") || err_msg.contains("overloaded") {
391 return Err(AppError::RateLimited {
392 detail: err_msg.to_string(),
393 });
394 }
395 if err_msg.contains("Not logged in") || err_msg.contains("authentication") {
396 tracing::warn!(
397 target: "claude_runner",
398 "Claude Code authentication failed. Re-authenticate interactively with: claude"
399 );
400 }
401 return Err(AppError::Validation(crate::i18n::validation::claude_extraction_failed(err_msg)));
402 }
403
404 let value = if let Some(v) = result_elem.structured_output.clone() {
405 v
406 } else if let Some(text) = &result_elem.result {
407 serde_json::from_str(text).map_err(|e| {
408 AppError::Validation(crate::i18n::validation::failed_to_parse_claude_result_field(
409 &e,
410 ))
411 })?
412 } else {
413 return Err(AppError::Validation(
414 "claude result missing structured_output and result field".into(),
415 ));
416 };
417
418 let cost = result_elem.total_cost_usd.unwrap_or(0.0);
419 Ok(ClaudeResult {
420 value,
421 cost_usd: cost,
422 is_oauth,
423 })
424}
425
426pub fn run_claude(
432 binary: &Path,
433 prompt: &str,
434 json_schema: &str,
435 input_text: &str,
436 model: Option<&str>,
437 timeout_secs: u64,
438 max_turns: u32,
439) -> Result<ClaudeResult, AppError> {
440 use wait_timeout::ChildExt;
441
442 let full_prompt = format!("{prompt}\n\n{input_text}");
443 let mut cmd = build_claude_command(binary, &full_prompt, json_schema, model, max_turns)?;
444
445 let mut child = spawn_with_memory_limit(&mut cmd).map_err(|e| {
446 AppError::Io(std::io::Error::new(
447 e.kind(),
448 format!("failed to spawn claude: {e}"),
449 ))
450 })?;
451
452 let start = std::time::Instant::now();
453 let timeout = std::time::Duration::from_secs(timeout_secs);
454 let status = child.wait_timeout(timeout).map_err(AppError::Io)?;
455
456 if status.is_none() {
457 #[cfg(unix)]
463 unsafe {
464 libc::kill(child.id() as i32, libc::SIGTERM);
465 }
466 let _ = child.kill();
467 let _ = child.wait();
468 }
469
470 match status {
471 Some(exit_status) => {
472 tracing::debug!(
473 target: "process",
474 exit_code = ?exit_status.code(),
475 elapsed_ms = start.elapsed().as_millis() as u64,
476 "external process completed"
477 );
478
479 let mut stdout_buf = Vec::new();
480 let mut stderr_buf = Vec::new();
481 if let Some(mut out) = child.stdout.take() {
482 std::io::Read::read_to_end(&mut out, &mut stdout_buf).map_err(AppError::Io)?;
483 }
484 if let Some(mut err) = child.stderr.take() {
485 std::io::Read::read_to_end(&mut err, &mut stderr_buf).map_err(AppError::Io)?;
486 }
487
488 let stdout_str = String::from_utf8(stdout_buf)
489 .map_err(|_| AppError::Validation(crate::i18n::validation::claude_p_stdout_not_utf8()))?;
490
491 if !exit_status.success() {
493 if let Ok(result) = parse_claude_output(&stdout_str) {
494 return Ok(result);
495 }
496 let stderr_str = String::from_utf8_lossy(&stderr_buf);
497 if stderr_str.contains("auth") || stderr_str.contains("login") {
498 tracing::warn!(
499 target: "claude_runner",
500 "Claude Code authentication may have failed. Re-authenticate with: claude"
501 );
502 }
503 return Err(AppError::Validation(crate::i18n::validation::process_exited(
504 "claude -p",
505 exit_status.code(),
506 stderr_str.trim(),
507 )));
508 }
509
510 parse_claude_output(&stdout_str)
511 }
512 None => {
513 tracing::warn!(target: "claude_runner", timeout_secs, "claude -p timed out, terminating");
514 terminate_gracefully(&mut child, 3);
515 Err(AppError::Validation(crate::i18n::validation::process_timed_out(
516 "claude -p",
517 timeout_secs,
518 )))
519 }
520 }
521}
522
523#[cfg(unix)]
525pub fn terminate_gracefully(child: &mut std::process::Child, grace_secs: u64) {
526 use wait_timeout::ChildExt;
527 unsafe {
528 libc::kill(child.id() as i32, libc::SIGTERM);
529 }
530 match child.wait_timeout(std::time::Duration::from_secs(grace_secs)) {
531 Ok(Some(_)) => {}
532 _ => {
533 tracing::warn!(target: "process", pid = child.id(), "child ignored SIGTERM, sending SIGKILL");
534 let _ = child.kill();
535 let _ = child.wait();
536 }
537 }
538}
539
540#[cfg(not(unix))]
542pub fn terminate_gracefully(child: &mut std::process::Child, _grace_secs: u64) {
543 let _ = child.kill();
544 let _ = child.wait();
545}
546
547#[cfg(test)]
548mod tests {
549 use super::*;
550
551 #[test]
552 fn parse_output_detects_max_turns() {
553 let stdout = r#"[{"type":"system","subtype":"init","apiKeySource":"none"},{"type":"result","is_error":false,"terminal_reason":"max_turns","structured_output":{"name":"t"}}]"#;
554 let err = parse_claude_output(stdout).unwrap_err();
555 assert!(
556 format!("{err}").contains("max_turns"),
557 "must detect max_turns in output"
558 );
559 }
560
561 #[test]
562 fn parse_output_extracts_structured_value() {
563 let stdout = r#"[{"type":"system","subtype":"init","apiKeySource":"none"},{"type":"result","is_error":false,"structured_output":{"key":"val"},"total_cost_usd":0.01}]"#;
564 let result = parse_claude_output(stdout).unwrap();
565 assert_eq!(result.value["key"], "val");
566 assert!((result.cost_usd - 0.01).abs() < f64::EPSILON);
567 assert!(result.is_oauth);
568 }
569
570 #[test]
571 fn parse_output_detects_rate_limit() {
572 let stdout = r#"[{"type":"result","is_error":true,"error":"rate_limit exceeded"}]"#;
573 let err = parse_claude_output(stdout).unwrap_err();
574 assert!(
575 matches!(err, AppError::RateLimited { .. }),
576 "expected AppError::RateLimited, got: {err}"
577 );
578 }
579
580 #[test]
584 #[serial_test::serial(env)]
585 fn build_command_oauth_only_mandatory_flags() {
586 unsafe {
588 std::env::remove_var("ANTHROPIC_API_KEY");
589 std::env::remove_var("CLAUDE_CONFIG_DIR");
592 }
593 let cmd = build_claude_command(
594 std::path::Path::new("/usr/bin/false"),
595 "test prompt",
596 "{}",
597 Some("sonnet"),
598 4,
599 )
600 .expect("preflight gate accepts valid args");
601 let args: Vec<&str> = cmd.get_args().filter_map(|a| a.to_str()).collect();
602 assert!(args.contains(&"-p"), "must have -p");
604 assert!(
605 args.contains(&"--strict-mcp-config"),
606 "must have --strict-mcp-config (gaps.md:206)"
607 );
608 assert!(
609 args.contains(&"--mcp-config"),
610 "must have --mcp-config (gaps.md:207)"
611 );
612 assert!(
613 args.contains(&"--dangerously-skip-permissions"),
614 "must have --dangerously-skip-permissions (gaps.md:208)"
615 );
616 assert!(
617 args.contains(&"--settings"),
618 "must have --settings (gaps.md:209)"
619 );
620 assert!(
621 args.contains(&"--output-format"),
622 "must have --output-format json (gaps.md:213)"
623 );
624 assert!(args.contains(&"--json-schema"), "must have --json-schema");
625 assert!(
626 args.contains(&"--max-turns"),
627 "must have --max-turns (gaps.md:212)"
628 );
629 assert!(
630 args.contains(&"--no-session-persistence"),
631 "must have --no-session-persistence"
632 );
633 assert!(
634 args.contains(&"--model"),
635 "must have --model when model is Some"
636 );
637 assert!(
639 !args.contains(&"--bare"),
640 "--bare is PROHIBITED (gaps.md:49)"
641 );
642 }
643
644 #[test]
648 #[serial_test::serial(env)]
649 fn build_command_aborts_when_anthropic_api_key_set() {
650 unsafe {
652 std::env::set_var("ANTHROPIC_API_KEY", "sk-test-violation");
653 std::env::remove_var("CLAUDE_CONFIG_DIR");
657 }
658 let cmd = build_claude_command(
659 std::path::Path::new("/usr/bin/claude"),
660 "test prompt",
661 "{}",
662 Some("sonnet"),
663 4,
664 )
665 .expect("preflight gate accepts valid args");
666 let program = cmd.get_program().to_string_lossy().to_string();
667 let args: Vec<&str> = cmd.get_args().filter_map(|a| a.to_str()).collect();
668 assert_eq!(
669 program, "false",
670 "when ANTHROPIC_API_KEY is set, build_claude_command must abort"
671 );
672 assert!(
673 args.contains(&"--oauth-only-violation-anthropic-api-key-set"),
674 "aborted command must carry violation marker"
675 );
676 unsafe {
677 std::env::remove_var("ANTHROPIC_API_KEY");
678 }
679 }
680}