1use std::ffi::{OsStr, OsString};
11use std::fmt;
12
13const REGISTERED_NAMES: &str = include_str!("environment_registry_names.txt");
14const UNKNOWN_CODE: &str = "HARN-ENV-001";
15const INVALID_VALUE_CODE: &str = "HARN-ENV-002";
16
17#[derive(Clone, Copy, Debug, PartialEq, Eq)]
19pub enum EnvironmentConsumer {
20 Runtime,
21 Cli,
22 BuildTooling,
23 TestHarness,
24 EmbedderExtension,
25}
26
27#[derive(Clone, Copy, Debug, PartialEq, Eq)]
32pub enum EnvironmentValueShape {
33 OwnerValidated,
34 Boolean,
35 UnsignedInteger,
36 NonNegativeNumber,
37 UnitInterval,
38}
39
40#[derive(Clone, Copy, Debug, PartialEq, Eq)]
42pub enum EnvironmentSensitivity {
43 Public,
44 Credential,
45}
46
47#[derive(Clone, Debug, PartialEq, Eq)]
48pub struct EnvironmentVariableSpec {
49 pub name: String,
50 pub consumer: EnvironmentConsumer,
51 pub value_shape: EnvironmentValueShape,
52 pub sensitivity: EnvironmentSensitivity,
53}
54
55#[derive(Clone, Debug, PartialEq, Eq)]
56pub enum EnvironmentDiagnosticKind {
57 UnknownName { suggestion: Option<String> },
58 InvalidValue { expected: EnvironmentValueShape },
59}
60
61#[derive(Clone, Debug, PartialEq, Eq)]
64pub struct EnvironmentDiagnostic {
65 pub code: &'static str,
66 pub key: String,
67 pub kind: EnvironmentDiagnosticKind,
68}
69
70impl fmt::Display for EnvironmentDiagnostic {
71 fn fmt(&self, formatter: &mut fmt::Formatter<'_>) -> fmt::Result {
72 match &self.kind {
73 EnvironmentDiagnosticKind::UnknownName { suggestion } => {
74 write!(
75 formatter,
76 "{}: unknown Harn environment variable `{}`",
77 self.code, self.key
78 )?;
79 if let Some(suggestion) = suggestion {
80 write!(formatter, "; did you mean `{suggestion}`?")?;
81 }
82 Ok(())
83 }
84 EnvironmentDiagnosticKind::InvalidValue { expected } => write!(
85 formatter,
86 "{}: environment variable `{}` must be {}",
87 self.code,
88 self.key,
89 expected.description()
90 ),
91 }
92 }
93}
94
95#[derive(Clone, Debug, PartialEq, Eq)]
96pub struct EnvironmentValidationError {
97 diagnostics: Vec<EnvironmentDiagnostic>,
98}
99
100impl EnvironmentValidationError {
101 pub fn diagnostics(&self) -> &[EnvironmentDiagnostic] {
102 &self.diagnostics
103 }
104}
105
106impl fmt::Display for EnvironmentValidationError {
107 fn fmt(&self, formatter: &mut fmt::Formatter<'_>) -> fmt::Result {
108 for (index, diagnostic) in self.diagnostics.iter().enumerate() {
109 if index > 0 {
110 formatter.write_str("\n")?;
111 }
112 diagnostic.fmt(formatter)?;
113 }
114 Ok(())
115 }
116}
117
118impl std::error::Error for EnvironmentValidationError {}
119
120impl EnvironmentValueShape {
121 fn description(self) -> &'static str {
122 match self {
123 Self::OwnerValidated => "valid for its owning subsystem",
124 Self::Boolean => "a boolean (`true`, `false`, `yes`, `no`, `on`, `off`, `1`, or `0`)",
125 Self::UnsignedInteger => "an unsigned integer",
126 Self::NonNegativeNumber => "a non-negative number",
127 Self::UnitInterval => "a number between 0 and 1",
128 }
129 }
130
131 fn accepts(self, value: &OsStr) -> bool {
132 let Some(value) = value.to_str() else {
133 return matches!(self, Self::OwnerValidated);
134 };
135 match self {
136 Self::OwnerValidated => true,
137 Self::Boolean => matches!(
138 value.trim().to_ascii_lowercase().as_str(),
139 "1" | "0" | "true" | "false" | "yes" | "no" | "on" | "off"
140 ),
141 Self::UnsignedInteger => value.trim().parse::<u64>().is_ok(),
142 Self::NonNegativeNumber => value
143 .trim()
144 .parse::<f64>()
145 .is_ok_and(|parsed| parsed.is_finite() && parsed >= 0.0),
146 Self::UnitInterval => value
147 .trim()
148 .parse::<f64>()
149 .is_ok_and(|parsed| parsed.is_finite() && (0.0..=1.0).contains(&parsed)),
150 }
151 }
152}
153
154pub fn variable_spec(name: &str) -> Option<EnvironmentVariableSpec> {
156 if registered_names().binary_search(&name).is_ok() {
157 return Some(spec_for_registered_name(name));
158 }
159 if is_extension_name(name) {
160 return Some(EnvironmentVariableSpec {
161 name: name.to_string(),
162 consumer: EnvironmentConsumer::EmbedderExtension,
163 value_shape: EnvironmentValueShape::OwnerValidated,
164 sensitivity: sensitivity_for(name),
165 });
166 }
167 if is_structured_runtime_name(name) {
168 return Some(spec_for_registered_name(name));
169 }
170 None
171}
172
173pub fn validate_startup_environment() -> Result<(), EnvironmentValidationError> {
176 validate_environment(std::env::vars_os())
177}
178
179pub fn validate_environment<I, K, V>(vars: I) -> Result<(), EnvironmentValidationError>
182where
183 I: IntoIterator<Item = (K, V)>,
184 K: Into<OsString>,
185 V: Into<OsString>,
186{
187 let mut diagnostics = Vec::new();
188 for (key, value) in vars {
189 let key = key.into();
190 let Some(key) = key.to_str() else {
191 continue;
192 };
193 if !key.starts_with("HARN_") {
194 continue;
195 }
196 let Some(spec) = variable_spec(key) else {
197 diagnostics.push(EnvironmentDiagnostic {
198 code: UNKNOWN_CODE,
199 key: key.to_string(),
200 kind: EnvironmentDiagnosticKind::UnknownName {
201 suggestion: nearest_registered_name(key),
202 },
203 });
204 continue;
205 };
206 let value = value.into();
207 if !spec.value_shape.accepts(&value) {
208 diagnostics.push(EnvironmentDiagnostic {
209 code: INVALID_VALUE_CODE,
210 key: key.to_string(),
211 kind: EnvironmentDiagnosticKind::InvalidValue {
212 expected: spec.value_shape,
213 },
214 });
215 }
216 }
217 diagnostics.sort_by(|left, right| left.key.cmp(&right.key));
218 if diagnostics.is_empty() {
219 Ok(())
220 } else {
221 Err(EnvironmentValidationError { diagnostics })
222 }
223}
224
225fn registered_names() -> &'static [&'static str] {
226 static NAMES: std::sync::OnceLock<Vec<&'static str>> = std::sync::OnceLock::new();
227 NAMES
228 .get_or_init(|| {
229 REGISTERED_NAMES
230 .lines()
231 .filter(|name| !name.is_empty())
232 .collect()
233 })
234 .as_slice()
235}
236
237fn spec_for_registered_name(name: &str) -> EnvironmentVariableSpec {
238 EnvironmentVariableSpec {
239 name: name.to_string(),
240 consumer: consumer_for(name),
241 value_shape: value_shape_for(name),
242 sensitivity: sensitivity_for(name),
243 }
244}
245
246fn consumer_for(name: &str) -> EnvironmentConsumer {
247 if name.starts_with("HARN_TEST_") || name.contains("_TEST_") || name.starts_with("HARN_E2E_") {
248 EnvironmentConsumer::TestHarness
249 } else if [
250 "HARN_BUILD_",
251 "HARN_CARGO_",
252 "HARN_CHECK_",
253 "HARN_CI_",
254 "HARN_CODEGEN_",
255 "HARN_DEV_",
256 "HARN_RELEASE_",
257 ]
258 .iter()
259 .any(|prefix| name.starts_with(prefix))
260 {
261 EnvironmentConsumer::BuildTooling
262 } else if name.starts_with("HARN_CLI_")
263 || name.starts_with("HARN_DOCTOR_")
264 || name.starts_with("HARN_INIT_")
265 {
266 EnvironmentConsumer::Cli
267 } else {
268 EnvironmentConsumer::Runtime
269 }
270}
271
272fn value_shape_for(name: &str) -> EnvironmentValueShape {
273 match name {
274 "HARN_LLM_TIMEOUT"
275 | "HARN_LLM_IDLE_TIMEOUT"
276 | "HARN_LLM_FIRST_TOKEN_TIMEOUT"
277 | "HARN_MAX_CONCURRENCY"
278 | "HARN_RETENTION_DAYS"
279 | "HARN_TOKEN_BUDGET"
280 | "HARN_EVENT_LOG_QUEUE_DEPTH" => EnvironmentValueShape::UnsignedInteger,
281 "HARN_BUDGET_USD" => EnvironmentValueShape::NonNegativeNumber,
282 "HARN_OTEL_SAMPLE_RATIO" => EnvironmentValueShape::UnitInterval,
283 "HARN_BYTECODE_CACHE"
284 | harn_parser::HARN_LEGACY_AMBIENT_CAPABILITIES_ENV
285 | "HARN_LLM_STREAM"
286 | "HARN_REPLAY_ENABLED"
287 | "HARN_REQUIRE_SIGNED_SKILLS"
288 | "HARN_TRACE"
289 | "HARN_VERBOSE_CONFIG" => EnvironmentValueShape::Boolean,
290 _ => EnvironmentValueShape::OwnerValidated,
291 }
292}
293
294fn sensitivity_for(name: &str) -> EnvironmentSensitivity {
295 if [
296 "TOKEN",
297 "SECRET",
298 "PASSWORD",
299 "API_KEY",
300 "OAUTH_KEY",
301 "HEADERS",
302 "PRIVATE_KEY",
303 ]
304 .iter()
305 .any(|fragment| name.contains(fragment))
306 {
307 EnvironmentSensitivity::Credential
308 } else {
309 EnvironmentSensitivity::Public
310 }
311}
312
313fn is_extension_name(name: &str) -> bool {
317 name.strip_prefix("HARN_EXT_")
318 .is_some_and(is_upper_identifier)
319}
320
321fn is_structured_runtime_name(name: &str) -> bool {
325 is_secret_name(name)
326 || is_rate_limit_name(name)
327 || is_model_role_name(name)
328 || is_agent_model_option_name(name)
329}
330
331fn is_secret_name(name: &str) -> bool {
332 name.strip_prefix("HARN_SECRET_")
333 .is_some_and(is_upper_identifier)
334}
335
336fn is_rate_limit_name(name: &str) -> bool {
337 let Some(suffix) = name.strip_prefix("HARN_RATE_LIMIT_") else {
338 return false;
339 };
340 let Some((provider, field)) = suffix.rsplit_once('_') else {
341 return false;
342 };
343 is_upper_identifier(provider) && matches!(field, "QUEUE" | "RPM" | "TPM" | "CONCURRENCY")
344}
345
346fn is_model_role_name(name: &str) -> bool {
347 let suffix = name
348 .strip_prefix("HARN_LLM_ROLE_")
349 .or_else(|| name.strip_prefix("HARN_LLM_"));
350 let Some(suffix) = suffix else {
351 return false;
352 };
353 ["_MODEL", "_PROVIDER", "_ROUTE_POLICY"]
354 .iter()
355 .find_map(|ending| suffix.strip_suffix(ending))
356 .is_some_and(is_upper_identifier)
357}
358
359fn is_agent_model_option_name(name: &str) -> bool {
363 const SUFFIXES: &[&str] = &[
364 "_EFFORT",
365 "_MODEL",
366 "_MODEL_ROLE",
367 "_PROVIDER",
368 "_REASONING_TASK",
369 "_TOOL_FORMAT",
370 ];
371 let Some(prefix) = SUFFIXES.iter().find_map(|suffix| name.strip_suffix(suffix)) else {
372 return false;
373 };
374 let Some(prefix) = prefix.strip_prefix("HARN_") else {
375 return false;
376 };
377 let role = prefix
378 .strip_prefix("AGENT_")
379 .or_else(|| prefix.strip_prefix("LLM_"))
380 .unwrap_or(prefix);
381 matches!(prefix, "AGENT" | "LLM") || is_upper_identifier(role)
382}
383
384fn is_upper_identifier(value: &str) -> bool {
385 !value.is_empty()
386 && value
387 .bytes()
388 .all(|byte| byte.is_ascii_uppercase() || byte.is_ascii_digit() || byte == b'_')
389 && !value.starts_with('_')
390 && !value.ends_with('_')
391}
392
393fn nearest_registered_name(name: &str) -> Option<String> {
394 registered_names()
395 .iter()
396 .copied()
397 .filter(|candidate| !candidate.ends_with('_'))
398 .map(|candidate| (strsim::levenshtein(name, candidate), candidate))
399 .min_by_key(|(distance, candidate)| (*distance, *candidate))
400 .filter(|(distance, _)| *distance <= 3)
401 .map(|(_, candidate)| candidate.to_string())
402}
403
404#[cfg(test)]
405mod tests {
406 use super::*;
407
408 #[test]
409 fn typo_is_typed_and_suggests_registered_name() {
410 let error = validate_environment([("HARN_LLM_TIMOUT", "30")]).unwrap_err();
411 assert_eq!(
412 error.diagnostics(),
413 &[EnvironmentDiagnostic {
414 code: UNKNOWN_CODE,
415 key: "HARN_LLM_TIMOUT".to_string(),
416 kind: EnvironmentDiagnosticKind::UnknownName {
417 suggestion: Some("HARN_LLM_TIMEOUT".to_string()),
418 },
419 }]
420 );
421 }
422
423 #[test]
424 fn known_and_structured_extension_names_are_accepted() {
425 validate_environment([
426 ("HARN_LLM_TIMEOUT", "30"),
427 ("HARN_EXT_ACME_MODE", "custom"),
428 ("HARN_LLM_ROLE_REVIEW_MODEL", "reviewer"),
429 ("HARN_SECRET_ACME_TOKEN", "credential"),
430 ])
431 .unwrap();
432 }
433
434 #[test]
435 fn malformed_extension_name_is_not_a_prefix_bypass() {
436 let error = validate_environment([("HARN_EXT_", "anything")]).unwrap_err();
437 assert!(matches!(
438 error.diagnostics()[0].kind,
439 EnvironmentDiagnosticKind::UnknownName { .. }
440 ));
441 }
442
443 #[test]
444 fn dynamic_agent_role_options_follow_a_closed_structural_grammar() {
445 for name in [
446 "HARN_AGENT_MODEL",
447 "HARN_LLM_TOOL_FORMAT",
448 "HARN_AGENT_REVIEW_PROVIDER",
449 "HARN_LLM_PLANNER_REASONING_TASK",
450 "HARN_RELEASE_EFFORT",
451 ] {
452 assert!(variable_spec(name).is_some(), "{name}");
453 }
454 for name in [
455 "HARN_AGENT_REVIEW_UNKNOWN",
456 "HARN_LLM_TIMOUT",
457 "HARN_RELEASE_",
458 ] {
459 assert!(variable_spec(name).is_none(), "{name}");
460 }
461 }
462
463 #[test]
464 fn credential_metadata_covers_non_api_oauth_keys() {
465 assert_eq!(
466 variable_spec("HARN_OAUTH_KEY").unwrap().sensitivity,
467 EnvironmentSensitivity::Credential
468 );
469 }
470
471 #[test]
472 fn invalid_known_value_is_rejected_at_startup() {
473 let error = validate_environment([("HARN_LLM_TIMEOUT", "soon")]).unwrap_err();
474 assert_eq!(
475 error.diagnostics()[0].kind,
476 EnvironmentDiagnosticKind::InvalidValue {
477 expected: EnvironmentValueShape::UnsignedInteger
478 }
479 );
480 }
481
482 #[test]
483 fn diagnostics_cannot_render_values_even_for_credentials() {
484 let secret = "must-never-appear";
485 let error = validate_environment([("HARN_CLOUD_API_KEZ", secret)]).unwrap_err();
486 let rendered = error.to_string();
487 assert!(rendered.contains("HARN_CLOUD_API_KEZ"));
488 assert!(!rendered.contains(secret));
489 }
490
491 #[test]
492 fn registry_is_sorted_unique_and_contains_metadata() {
493 let names = registered_names();
494 assert!(
495 names.windows(2).all(|pair| pair[0] < pair[1]),
496 "environment registry must remain sorted and unique"
497 );
498 let timeout = variable_spec("HARN_LLM_TIMEOUT").unwrap();
499 assert_eq!(timeout.consumer, EnvironmentConsumer::Runtime);
500 assert_eq!(timeout.value_shape, EnvironmentValueShape::UnsignedInteger);
501 let token = variable_spec("HARN_PACKAGE_REGISTRY_TOKEN").unwrap();
502 assert_eq!(token.sensitivity, EnvironmentSensitivity::Credential);
503 }
504
505 #[test]
506 fn embedded_runtime_bootstrap_accepts_registered_process_environment() {
507 crate::initialize_runtime().expect("registered process environment");
508 }
509
510 #[test]
511 fn every_compiled_harn_name_is_registered_or_structurally_owned() {
512 let crates_dir = std::path::Path::new(env!("CARGO_MANIFEST_DIR"))
513 .parent()
514 .expect("harn-vm lives below crates");
515 let mut missing = std::collections::BTreeSet::new();
516 for entry in walkdir::WalkDir::new(crates_dir)
517 .into_iter()
518 .filter_map(Result::ok)
519 .filter(|entry| {
520 entry.file_type().is_file()
521 && entry.path().extension().and_then(OsStr::to_str) == Some("rs")
522 && entry
523 .path()
524 .components()
525 .any(|component| component.as_os_str() == "src")
526 && entry.file_name() != "environment_registry.rs"
527 })
528 {
529 let source = std::fs::read_to_string(entry.path()).expect("read Rust source");
530 for token in harn_name_tokens(&source) {
531 if variable_spec(token).is_none()
532 && !matches!(token, "HARN_LLM_" | "HARN_LLM_ROLE_" | "HARN_SECRET_")
533 {
534 missing.insert(format!("{}: {token}", entry.path().display()));
535 }
536 }
537 }
538 assert!(
539 missing.is_empty(),
540 "compiled HARN_* names missing from environment_registry_names.txt:\n{}",
541 missing.into_iter().collect::<Vec<_>>().join("\n")
542 );
543 }
544
545 #[test]
546 fn every_harn_script_owned_name_is_registered_or_structurally_owned() {
547 let workspace_root = std::path::Path::new(env!("CARGO_MANIFEST_DIR"))
548 .ancestors()
549 .nth(2)
550 .expect("harn-vm lives below workspace/crates");
551 let source_roots = [
552 "benchmarks",
553 "conformance",
554 "crates",
555 "evals",
556 "examples",
557 "experiments",
558 "perf",
559 "personas",
560 "scripts",
561 "tests",
562 ];
563 let mut missing = std::collections::BTreeSet::new();
564 for source_root in source_roots {
565 let source_root = workspace_root.join(source_root);
566 if !source_root.exists() {
567 continue;
568 }
569 for entry in walkdir::WalkDir::new(source_root)
570 .into_iter()
571 .filter_map(Result::ok)
572 .filter(|entry| {
573 entry.file_type().is_file()
574 && entry.path().extension().and_then(OsStr::to_str) == Some("harn")
575 })
576 {
577 let source = std::fs::read_to_string(entry.path()).expect("read Harn source");
578 for token in harn_name_tokens(&source) {
579 if variable_spec(token).is_none()
580 && !matches!(
581 token,
582 "HARN_AGENT"
583 | "HARN_AGENT_"
584 | "HARN_LLM"
585 | "HARN_LLM_"
586 | "HARN_PLANNER"
587 | "HARN_RELEASE"
588 )
589 {
590 missing.insert(format!("{}: {token}", entry.path().display()));
591 }
592 }
593 }
594 }
595 assert!(
596 missing.is_empty(),
597 "Harn-script HARN_* names missing from environment_registry_names.txt:\n{}",
598 missing.into_iter().collect::<Vec<_>>().join("\n")
599 );
600 }
601
602 fn harn_name_tokens(source: &str) -> impl Iterator<Item = &str> {
603 source.match_indices("\"HARN_").filter_map(|(quote, _)| {
604 let start = quote + 1;
605 let bytes = source.as_bytes();
606 let mut end = start + "HARN_".len();
607 while end < bytes.len()
608 && (bytes[end].is_ascii_uppercase()
609 || bytes[end].is_ascii_digit()
610 || bytes[end] == b'_')
611 {
612 end += 1;
613 }
614 (end > start + "HARN_".len()).then(|| &source[start..end])
615 })
616 }
617}