camel_api/redact.rs
1//! Canonical string-based URL redaction for diagnostic surfaces (ADR-0051).
2//!
3//! String surgery over raw URL bytes: userinfo in every authority window
4//! is masked as `***`, query/fragment content is replaced with
5//! `?[redacted]` / `#[redacted]` sentinels, and the result is capped at
6//! 256 bytes on a UTF-8 char boundary. Authority windows are enumerated
7//! over maximal runs of `/` and `\` characters: pure-slash runs (the
8//! landed `//` rule) open a window at two or more characters,
9//! backslash-bearing runs only behind an RFC 3986 scheme prefix. These
10//! helpers never parse — they are the strictest feasible handling for
11//! strings that may be malformed or hostile.
12//!
13//! Distinct from [`crate::endpoint_uri::EndpointUri::to_redacted_string`],
14//! which redacts the catalog-driven authored-URI layer (structured
15//! `EndpointUri` values) and is out of scope here.
16
17/// Return `input` with userinfo masked in every authority window. Windows
18/// are enumerated over maximal runs of `/` and `\` characters (see
19/// [`authority_windows`]); a window containing `@` carries userinfo, and
20/// the bytes from window start through the LAST `@` are replaced with
21/// `***` (over-masking is safe, under-masking is not). Windows are
22/// collected on the input and masked in reverse offset order (with
23/// duplicates deduped) so an edit never shifts a yet-to-be-processed
24/// window. Idempotent: an already-masked `***@host` window rewrites to
25/// itself.
26fn mask_authority_windows(input: &str) -> String {
27 let mut out = input.to_string();
28 let mut windows = authority_windows(input);
29 // The run scanner emits one window per maximal run in ascending order;
30 // sort+dedup are belt-and-braces.
31 windows.sort_unstable();
32 windows.dedup();
33 for (start, end) in windows.into_iter().rev() {
34 if let Some(at) = out[start..end].rfind('@') {
35 out.replace_range(start..start + at, "***");
36 }
37 }
38 out
39}
40
41/// Authority-window enumeration (bd rc-f05q8). Windows are derived from
42/// maximal runs of `/` and `\` characters, and a qualifying run opens
43/// exactly one window that ends at the next `/`, `?`, or `#` (or end of
44/// input). A pure-slash run qualifies at length >= 2 (the landed `//`
45/// rule) and its window starts at the run's end, keeping the slashes
46/// visible. A backslash-bearing run qualifies only behind an RFC 3986
47/// scheme prefix ending immediately before the run (see
48/// [`scheme_prefix_len`]): length >= 2 needs any scheme; a single `\`
49/// needs a scheme of two or more characters, or a one-character scheme
50/// whose candidate window content is credential-shaped (see
51/// [`single_backslash_credential_gate`]). Every qualifying window starts
52/// at the run's end — the spec's uniform rule: a window begins
53/// immediately after a maximal run of `/` and `\` — so the introducer
54/// slashes and backslashes stay visible. Scheme-less backslash runs —
55/// Windows drive paths and UNC paths — never open a window.
56fn authority_windows(input: &str) -> Vec<(usize, usize)> {
57 let bytes = input.as_bytes();
58 let mut windows = Vec::new();
59 let mut i = 0;
60 while i < bytes.len() {
61 if !matches!(bytes[i], b'/' | b'\\') {
62 i += 1;
63 continue;
64 }
65 let run_start = i;
66 let mut has_backslash = false;
67 while i < bytes.len() && matches!(bytes[i], b'/' | b'\\') {
68 has_backslash |= bytes[i] == b'\\';
69 i += 1;
70 }
71 let run_len = i - run_start;
72 let opens = if has_backslash {
73 match scheme_prefix_len(input, run_start) {
74 None => false,
75 Some(scheme_len) => {
76 run_len >= 2
77 || scheme_len >= 2
78 || (scheme_len == 1 && single_backslash_credential_gate(input, i))
79 }
80 }
81 } else {
82 run_len >= 2
83 };
84 if opens {
85 let end = input[i..]
86 .find(['/', '?', '#'])
87 .map_or(input.len(), |offset| i + offset);
88 // Every window starts at the run's end (spec: a window begins
89 // immediately after a maximal run of `/` and `\`), so the
90 // introducer slashes and backslashes stay visible.
91 let start = i;
92 windows.push((start, end));
93 }
94 }
95 windows
96}
97
98/// Length of the RFC 3986 scheme name (`[a-zA-Z][a-zA-Z0-9+.-]*`) whose
99/// `:` ends immediately before `run_start` — the LONGEST valid match, so
100/// the name is the maximal run of scheme characters directly before the
101/// colon and its first character must be alphabetic. `None` when
102/// `input[..run_start]` does not end with `scheme:`.
103fn scheme_prefix_len(input: &str, run_start: usize) -> Option<usize> {
104 let bytes = input.as_bytes();
105 if run_start < 2 || bytes[run_start - 1] != b':' {
106 return None;
107 }
108 let mut start = run_start - 1;
109 while start > 0
110 && (bytes[start - 1].is_ascii_alphanumeric()
111 || matches!(bytes[start - 1], b'+' | b'.' | b'-'))
112 {
113 start -= 1;
114 }
115 let len = run_start - 1 - start;
116 if len == 0 || !bytes[start].is_ascii_alphabetic() {
117 return None;
118 }
119 Some(len)
120}
121
122/// Credential-shaped content gate for a single `\` behind a
123/// one-character scheme (bd rc-f05q8): the candidate window content —
124/// from `content_start` (the run's end) to the next `/`, `?`, or `#` —
125/// must carry a `:` strictly before its LAST `@`. Without an `@` there is
126/// no userinfo to hide, so the gate fails.
127fn single_backslash_credential_gate(input: &str, content_start: usize) -> bool {
128 let end = input[content_start..]
129 .find(['/', '?', '#'])
130 .map_or(input.len(), |offset| content_start + offset);
131 let window = &input[content_start..end];
132 match window.rfind('@') {
133 Some(at) => window[..at].contains(':'),
134 None => false,
135 }
136}
137
138/// Truncate `s` to at most `max` bytes, walking the cut down to the nearest
139/// UTF-8 char boundary so a multibyte character straddling the cap cannot
140/// panic.
141fn truncate_utf8_safe(s: &mut String, max: usize) {
142 if s.len() <= max {
143 return;
144 }
145 let mut cut = max;
146 while !s.is_char_boundary(cut) {
147 cut -= 1;
148 }
149 s.truncate(cut);
150}
151
152/// Single left-to-right pass over `bytes` that decodes `%HH` sequences
153/// where `decode(hi, lo)` yields a byte; every other byte is copied
154/// through unchanged. `decode` is consulted only when `%` is followed by
155/// two bytes — a trailing `%`, a stray `%`, and an undecodable pair all
156/// fall through to the literal copy. The caller decides which escapes
157/// count (see [`minimal_decode_pair`] and [`decode_match_key`]).
158fn percent_scan(bytes: &[u8], decode: impl Fn(u8, u8) -> Option<u8>) -> Vec<u8> {
159 let mut out = Vec::with_capacity(bytes.len());
160 let mut i = 0;
161 while i < bytes.len() {
162 if bytes[i] == b'%'
163 && i + 2 < bytes.len()
164 && let Some(b) = decode(bytes[i + 1], bytes[i + 2])
165 {
166 out.push(b);
167 i += 3;
168 continue;
169 }
170 out.push(bytes[i]);
171 i += 1;
172 }
173 out
174}
175
176/// Single left-to-right pass over a query pair that decodes ONLY the three
177/// sequences a credential URL cannot be recognized without — `%40` → `@`,
178/// `%3a`/`%3A` → `:`, `%2f`/`%2F` → `/` — and copies every other byte
179/// through (other `%HH` escapes, stray `%`, and incomplete escapes
180/// included). Valid UTF-8 is preserved byte-for-byte; invalid bytes become
181/// `U+FFFD`, which cannot match an ASCII needle, so the shape checks below
182/// are unaffected. This "minimal decode" (bd rc-r7v8s) is deliberately
183/// weaker than a full percent-decoder: it exists only so the allowlist
184/// redactor can recognize credential-shaped values hidden behind encoding,
185/// while never giving a benign pair a reason to change. Double-encoded
186/// input (`%2540`) stays encoded — out of scope by design.
187fn minimal_decode_pair(pair: &str) -> String {
188 let decoded = percent_scan(pair.as_bytes(), |hi, lo| match (hi, lo) {
189 (b'4', b'0') => Some(b'@'),
190 (b'3', b'a' | b'A') => Some(b':'),
191 (b'2', b'f' | b'F') => Some(b'/'),
192 _ => None,
193 });
194 String::from_utf8_lossy(&decoded).into_owned()
195}
196
197/// Credential-shape test shared by both query-pair positions (ADR-0076
198/// appendix, bd rc-yvjp3): an input whose minimal decode carries `@` and
199/// also `:` or `//` is treated as an embedded `user:pass@host` credential
200/// wherever it rides — key position or value position. One predicate, one
201/// invariant; a lone `@` (an email address) is not credential-shaped.
202fn is_credential_shaped(decoded: &str) -> bool {
203 decoded.contains('@') && (decoded.contains(':') || decoded.contains("//"))
204}
205
206/// Match key for the sensitive-substring check: a single left-to-right
207/// `%HH` decode over the raw key bytes (ANY two hex digits after `%`,
208/// case-insensitive; invalid sequences copied verbatim), then lowercased.
209/// Unlike [`minimal_decode_pair`] this is a full digit decode because the
210/// denylist matches literal text (`pass%77ord` must hit `password`), not
211/// credential shape. The result is only used for substring matching and is
212/// never rendered; invalid UTF-8 from a decoded byte becomes `U+FFFD`,
213/// which cannot match an ASCII needle.
214fn decode_match_key(raw_key: &str) -> String {
215 let decoded = percent_scan(raw_key.as_bytes(), |hi, lo| {
216 let hi = (hi as char).to_digit(16)?;
217 let lo = (lo as char).to_digit(16)?;
218 Some((hi * 16 + lo) as u8)
219 });
220 String::from_utf8_lossy(&decoded).to_lowercase()
221}
222
223/// Canonical strict redaction of a raw URL string for diagnostic surfaces
224/// (ADR-0051, bd rc-eh49): authority windows are enumerated over maximal
225/// runs of `/` and `\` characters — a pure-slash run of two or more
226/// characters opens a window, a backslash-bearing run opens one only
227/// behind an RFC 3986 scheme prefix (`[a-zA-Z][a-zA-Z0-9+.-]*:`) ending
228/// immediately before the run (a single `\` additionally needs a scheme of
229/// two-plus characters, or a one-character scheme whose candidate window
230/// content is credential-shaped) — and each window ends at the next `/`,
231/// `?`, or `#`. A window containing `@` carries userinfo,
232/// and the bytes from window start through the LAST `@` are masked in
233/// place as `***@` (over-masking is safe, under-masking is not). Every
234/// window is scanned, so credentials cannot hide in a later window behind
235/// a benign first one. Everything from the earliest `?` or `#` is dropped;
236/// the sentinels
237/// compose: each distinct introducer character (`?` and/or `#`) that occurs
238/// anywhere in the URL appends its matching `?[redacted]` / `#[redacted]`
239/// sentinel in first-occurrence order — queries and fragments routinely
240/// carry tokens. The result is capped at 256 bytes on a UTF-8 char
241/// boundary. There is no URL parser here, so in-place windowed masking is
242/// the strictest feasible handling. This is the string layer;
243/// [`crate::endpoint_uri::EndpointUri::to_redacted_string`] redacts the
244/// catalog-driven authored-URI surface instead.
245pub fn redact_url(raw: &str) -> String {
246 let mut out = mask_authority_windows(raw);
247 if let Some(i) = out.find(['?', '#']) {
248 // Compose-both: one sentinel per distinct introducer found in the
249 // raw URL, in first-occurrence order.
250 let query_pos = out.find('?');
251 let fragment_pos = out.find('#');
252 out.truncate(i);
253 // Reserve the sentinel bytes before truncating so the cap never
254 // splits an appended sentinel (e_gpt stage-4).
255 let sentinel_total = match (query_pos, fragment_pos) {
256 (Some(_), Some(_)) => 22,
257 (Some(_), None) | (None, Some(_)) => 11,
258 (None, None) => 0,
259 };
260 if sentinel_total > 0 {
261 truncate_utf8_safe(&mut out, 256 - sentinel_total);
262 }
263 match (query_pos, fragment_pos) {
264 (Some(q), Some(f)) if f < q => out.push_str("#[redacted]?[redacted]"),
265 (Some(_), Some(_)) => out.push_str("?[redacted]#[redacted]"),
266 (Some(_), None) => out.push_str("?[redacted]"),
267 (None, Some(_)) => out.push_str("#[redacted]"),
268 (None, None) => {}
269 }
270 }
271 truncate_utf8_safe(&mut out, 256);
272 out
273}
274
275/// Fail-closed variant of [`redact_url`]: when any authority window of
276/// `raw` carries an `@`, the whole string is replaced with `[redacted]` —
277/// a string with an unvalidated authority marker may carry credentials
278/// nothing validated, so nothing of it is rendered (deliberate fail-closed
279/// over-redaction per ADR-0051). Otherwise identical to [`redact_url`].
280pub fn redact_url_fail_closed(raw: &str) -> String {
281 if window_has_at_sign(raw) {
282 "[redacted]".to_string()
283 } else {
284 redact_url(raw)
285 }
286}
287
288/// Broker-style redaction: window-mask userinfo like [`redact_url`], then
289/// redact sensitive query params per key while keeping benign ones. A pair
290/// whose match key — the raw key single-pass `%HH`-decoded then lowercased
291/// (bd rc-r7v8s) — contains any of `sensitive_key_substrings` renders as
292/// `{raw_key}=<redacted>` (the key keeps its original encoded bytes),
293/// EXCEPT when the raw key's own single-pass minimal decode
294/// (see `minimal_decode_pair`) is credential-shaped — contains `@` and
295/// also `:` or `//`: then the pair renders as a bare `<redacted>` and the
296/// key never echoes (bd rc-yvjp3, ADR-0076 appendix: key-position
297/// credential-shape symmetry — the same predicate the benign-key branch
298/// applies to the whole pair; a key such as `user%3Asecret%40host` embeds
299/// the credential `user:secret@host` and must not render). Well-known
300/// broker parameter names never decode to that shape, so the
301/// transport-policy diagnostic value below is preserved. Otherwise, if the
302/// pair's `minimal_decode_pair` output is
303/// credential-shaped (contains `@` and also `:` or `//`), the whole pair is
304/// replaced with a bare `<redacted>`: encoded or literal
305/// `user:secret@host` values must not survive under a benign key, while a
306/// lone `@` (an email address) keeps the pair visible. Every other pair
307/// survives byte-for-byte, because
308/// non-secret transport policy in query params (ActiveMQ failover URIs) is
309/// the sole diagnostic value of logging the broker URL. Fragments are never
310/// echoed: everything from the first `#` is dropped and replaced with the
311/// `#[redacted]` sentinel. The result is capped at 256 bytes on a UTF-8
312/// char boundary. String-based; no URL parser.
313pub fn redact_url_with_query_allowlist(raw: &str, sensitive_key_substrings: &[&str]) -> String {
314 redact_url_with_query_allowlist_impl(raw, sensitive_key_substrings, Some(256))
315}
316
317/// Whether a raw query-parameter key is itself credential-shaped: its
318/// single-pass minimal decode (see `minimal_decode_pair`) contains `@`
319/// and also `:` or `//`. This is the exact key-position predicate
320/// `redact_url_with_query_allowlist_impl` applies (bd rc-yvjp3), exposed
321/// so parsers that retain unknown keys can suppress them instead of
322/// echoing embedded `user:secret@host` material. The predicate is
323/// heuristic, and it becomes frozen public surface once camel-api
324/// publishes externally.
325pub fn is_credential_shaped_key(raw_key: &str) -> bool {
326 is_credential_shaped(&minimal_decode_pair(raw_key))
327}
328
329/// Untruncated variant of [`redact_url_with_query_allowlist`]: identical
330/// windowed userinfo masking, per-key query redaction, and fragment
331/// suppression, but WITHOUT the 256-byte diagnostic cap. Callers that use
332/// the redacted string as stable identity material (for example a
333/// broker-key hash) MUST use this variant: the 256-byte cap conflates two
334/// long URLs that share a prefix but differ in a distinct address after
335/// the cap. Never use it as a rendered diagnostic surface.
336pub fn redact_url_with_query_allowlist_untruncated(
337 raw: &str,
338 sensitive_key_substrings: &[&str],
339) -> String {
340 redact_url_with_query_allowlist_impl(raw, sensitive_key_substrings, None)
341}
342
343/// Shared implementation for the truncating and untruncated variants.
344/// `cap` is `Some(256)` for the diagnostic surface and `None` for stable
345/// identity material. When capped, the fragment sentinel is reserved before
346/// the cut so the cap never splits it (e_gpt stage-4).
347fn redact_url_with_query_allowlist_impl(
348 raw: &str,
349 sensitive_key_substrings: &[&str],
350 cap: Option<usize>,
351) -> String {
352 // 1) Mask userinfo in every authority window (see [`mask_authority_windows`]).
353 let after_userinfo = mask_authority_windows(raw);
354 // 2) Redact sensitive query params, keep the rest for diagnosability.
355 let mut out = match after_userinfo.split_once('?') {
356 Some((base, query)) => {
357 let redacted: Vec<String> = query
358 .split('&')
359 .map(|pair| {
360 let raw_key = pair.split('=').next().unwrap_or(pair);
361 let match_key = decode_match_key(raw_key);
362 if sensitive_key_substrings
363 .iter()
364 .any(|s| match_key.contains(s))
365 {
366 // Key-position symmetry (bd rc-yvjp3): a
367 // credential-shaped key is fully suppressed — the
368 // key bytes may themselves be the credential.
369 if is_credential_shaped(&minimal_decode_pair(raw_key)) {
370 "<redacted>".to_string()
371 } else {
372 format!("{raw_key}=<redacted>")
373 }
374 } else if is_credential_shaped(&minimal_decode_pair(pair)) {
375 "<redacted>".to_string()
376 } else {
377 pair.to_string()
378 }
379 })
380 .collect();
381 format!("{base}?{}", redacted.join("&"))
382 }
383 None => after_userinfo,
384 };
385 // 3) Fragments are never echoed. Reserve the sentinel bytes before
386 // truncating so the cap never splits the appended `#[redacted]`
387 // (e_gpt stage-4); the kept query content is part of the base.
388 if let Some(i) = out.find('#') {
389 out.truncate(i);
390 if let Some(cap) = cap {
391 truncate_utf8_safe(&mut out, cap - 11);
392 }
393 out.push_str("#[redacted]");
394 }
395 if let Some(cap) = cap {
396 truncate_utf8_safe(&mut out, cap);
397 }
398 out
399}
400
401/// Whether a `@` appears in any authority window of `raw`. Windows come
402/// from maximal runs of `/` and `\` characters — the same enumeration the
403/// mask uses: pure-slash runs of two or more characters always open a
404/// window; backslash-bearing runs open one only behind an RFC 3986 scheme
405/// prefix ending immediately before the run. Every run is scanned, so
406/// credentials cannot hide in a later window behind a benign first one
407/// (`http://h/a//user:pass@e/`) or behind a scheme-prefixed backslash
408/// authority (`foo:\u:p@e/`), while scheme-less drive and UNC paths stay
409/// window-free.
410pub fn window_has_at_sign(raw: &str) -> bool {
411 authority_windows(raw)
412 .into_iter()
413 .any(|(start, end)| raw[start..end].contains('@'))
414}
415
416/// Redact a bare host value for logs (ADR-0076, `lint-log-redaction`).
417/// A bare host carries no `//` authority window, so [`redact_url`] is a
418/// semantic no-op on it; this helper masks everything before the LAST
419/// `@` instead (`***@host`). The host grammar carries no userinfo, but
420/// the value can be externally influenced (a config field the URI
421/// parser can route `user:pass@host` into), so userinfo-shaped input is
422/// masked defensively — through the LAST `@`, because over-masking is
423/// safe and under-masking is not. Clean hosts pass through unchanged
424/// for diagnosability. Canonical single source: components log `host`
425/// fields through this helper, not crate-local twins (bd rc-8bxeo).
426pub fn redact_host(host: &str) -> String {
427 match host.rsplit_once('@') {
428 Some((_, after)) => format!("***@{after}"),
429 None => host.to_string(),
430 }
431}
432
433#[cfg(test)]
434mod tests {
435 use super::*;
436
437 /// Key denylist used by the camel-jms broker redactor (7 entries, same
438 /// order as the landed `redact_broker_url` list).
439 const JMS_KEYS: &[&str] = &[
440 "password",
441 "passwd",
442 "secret",
443 "credential",
444 "token",
445 "username",
446 "user",
447 ];
448
449 // ── Spec-pinned scenarios (redact2 security requirement) ─────────────
450
451 #[test]
452 fn strict_masks_windows_and_composes_sentinels() {
453 assert_eq!(
454 redact_url("https://user:pass@h/p?a=1#t=x"),
455 "https://***@h/p?[redacted]#[redacted]"
456 );
457 // A later `//` window is masked too: credentials cannot hide behind
458 // a benign first window.
459 assert_eq!(redact_url("https://h//u2:p2@evil/"), "https://h//***@evil/");
460 }
461
462 #[test]
463 fn fail_closed_suppresses_window_with_at() {
464 assert_eq!(
465 redact_url_fail_closed("http://u:secretpw@host:99999/x"),
466 "[redacted]"
467 );
468 }
469
470 #[test]
471 fn sentinel_never_splits_at_cap() {
472 // Pre-sentinel base (up to the earliest `?`) is 316 bytes; the
473 // 3-byte `日` straddles the reserved cut at byte 234, forcing the
474 // boundary walk. Both sentinels must render intact.
475 let mut url = format!("https://host/{}日{}", "x".repeat(220), "x".repeat(80));
476 url.push_str("?a=1#f");
477 assert!(url.len() > 300);
478 let redacted = redact_url(&url);
479 assert!(redacted.len() <= 256, "len={}", redacted.len());
480 assert!(
481 redacted.ends_with("?[redacted]#[redacted]"),
482 "sentinels must render intact: {redacted}"
483 );
484 assert!(std::str::from_utf8(redacted.as_bytes()).is_ok());
485 assert!(
486 !redacted.contains('日'),
487 "straddling char dropped whole: {redacted}"
488 );
489 assert!(
490 redacted.starts_with("https://host/"),
491 "host kept: {redacted}"
492 );
493 }
494
495 #[test]
496 fn idempotent_mask_composition_pin() {
497 // An already-masked window rewrites to itself and the sentinels
498 // still compose.
499 assert_eq!(
500 redact_url("https://***@host/p?a=1#f"),
501 "https://***@host/p?[redacted]#[redacted]"
502 );
503 }
504
505 #[test]
506 fn allowlist_keeps_benign_and_redacts_sensitive() {
507 assert_eq!(
508 redact_url_with_query_allowlist(
509 "tcp://host:61616?password=p&user=u&keepAlive=true",
510 JMS_KEYS
511 ),
512 "tcp://host:61616?password=<redacted>&user=<redacted>&keepAlive=true"
513 );
514 }
515
516 // ── Phase 2: minimal-decode rule (bd rc-r7v8s) ──────────────────────
517
518 /// Percent-encoded credentials under a benign key must not smuggle
519 /// through: the single-pass minimal decode of the pair carries `@` and
520 /// `//`, so the whole pair redacts. Uppercase hex form.
521 #[test]
522 fn allowlist_masks_percent_encoded_credentials_uppercase() {
523 let redacted = redact_url_with_query_allowlist(
524 "tcp://h:61616?redirect=http%3A%2F%2Fuser%3Asecret%40host",
525 JMS_KEYS,
526 );
527 assert!(
528 !redacted.contains("secret"),
529 "percent-encoded credential leaked: {redacted}"
530 );
531 assert!(
532 !redacted.contains("user%3Asecret%40"),
533 "encoded userinfo leaked: {redacted}"
534 );
535 assert_eq!(redacted, "tcp://h:61616?<redacted>");
536 }
537
538 /// Fully-lowercase hex variant of the same smuggling shape.
539 #[test]
540 fn allowlist_masks_percent_encoded_credentials_lowercase() {
541 let redacted = redact_url_with_query_allowlist(
542 "tcp://h:61616?redirect=http%3a%2f%2fuser%3asecret%40host",
543 JMS_KEYS,
544 );
545 assert_eq!(redacted, "tcp://h:61616?<redacted>");
546 assert!(
547 !redacted.contains("secret"),
548 "percent-encoded credential leaked: {redacted}"
549 );
550 }
551
552 /// bd rc-yvjp3 (ADR-0076 appendix, key-position symmetry): a pair
553 /// whose KEY embeds credentials (`user%3Asecret%40host` decodes to
554 /// `user:secret@host`) renders as a bare `<redacted>` — the key bytes
555 /// never echo, in either hex case.
556 #[test]
557 fn allowlist_suppresses_credential_shaped_key_uppercase() {
558 let redacted =
559 redact_url_with_query_allowlist("tcp://h:61616?user%3Asecret%40host=1", JMS_KEYS);
560 assert_eq!(redacted, "tcp://h:61616?<redacted>");
561 assert!(
562 !redacted.contains("user%3Asecret%40"),
563 "credential-shaped key leaked: {redacted}"
564 );
565 assert!(
566 !redacted.contains("secret"),
567 "key credential bytes leaked: {redacted}"
568 );
569 }
570
571 /// Fully-lowercase hex variant of the key-position shape.
572 #[test]
573 fn allowlist_suppresses_credential_shaped_key_lowercase() {
574 let redacted =
575 redact_url_with_query_allowlist("tcp://h:61616?user%3asecret%40host=1", JMS_KEYS);
576 assert_eq!(redacted, "tcp://h:61616?<redacted>");
577 }
578
579 /// Key-position symmetry, literal form: a fully literal
580 /// `user:pass@host` key is suppressed whole.
581 #[test]
582 fn allowlist_suppresses_literal_credential_shaped_key() {
583 let redacted = redact_url_with_query_allowlist("tcp://h:61616?user:pass@host=1", JMS_KEYS);
584 assert_eq!(redacted, "tcp://h:61616?<redacted>");
585 assert!(
586 !redacted.contains("user:pass"),
587 "literal key credentials leaked: {redacted}"
588 );
589 }
590
591 /// bd rc-yvjp3 non-regression: well-known sensitive keys that are NOT
592 /// credential-shaped keep echoing their names — the ADR-0076 exception
593 /// names transport-policy parameter visibility as the sole diagnostic
594 /// value of the broker URL, and no real parameter name decodes to an
595 /// `@`+`:`/`//` shape.
596 #[test]
597 fn allowlist_keeps_well_known_key_names_visible() {
598 let redacted = redact_url_with_query_allowlist(
599 "tcp://h:61616?password=p&jms.userName=admin&user=u&keepAlive=true",
600 JMS_KEYS,
601 );
602 assert_eq!(
603 redacted,
604 "tcp://h:61616?password=<redacted>&jms.userName=<redacted>&user=<redacted>&keepAlive=true"
605 );
606 }
607
608 /// bd rc-yvjp3 boundary pin: a denylist-matching key with a lone `@`
609 /// (no `:`, no `//`) is NOT credential-shaped — it renders
610 /// `user@host=<redacted>` and the key echoes, mirroring the lone-`@`
611 /// email rule of the benign-key branch.
612 #[test]
613 fn allowlist_keeps_lone_at_key_visible() {
614 let redacted = redact_url_with_query_allowlist("tcp://h:61616?user@host=1", JMS_KEYS);
615 assert_eq!(redacted, "tcp://h:61616?user@host=<redacted>");
616 }
617
618 /// bd rc-yvjp3 non-regression: an encoded-but-not-credential-shaped
619 /// key (`%77` is outside the minimal-decode triple) still echoes its
620 /// authored bytes when the decoded name hits the denylist — the
621 /// `pass%77ord` match rule (bd rc-r7v8s) is unchanged.
622 #[test]
623 fn allowlist_keeps_encoded_but_benign_shaped_key_visible() {
624 let redacted = redact_url_with_query_allowlist("tcp://h:61616?pass%77ord=p", JMS_KEYS);
625 assert_eq!(redacted, "tcp://h:61616?pass%77ord=<redacted>");
626 }
627
628 /// Encoded slashes plus a literal `@` decode to a `//`-bearing
629 /// credential shape under a benign key.
630 #[test]
631 fn allowlist_masks_literal_at_bypass() {
632 let redacted =
633 redact_url_with_query_allowlist("tcp://h:61616?next=%2F%2Fuser:pass@host", JMS_KEYS);
634 assert_eq!(redacted, "tcp://h:61616?<redacted>");
635 assert!(
636 !redacted.contains("pass"),
637 "literal credential leaked: {redacted}"
638 );
639 }
640
641 /// Fully literal `user:pass@host` under a benign key.
642 #[test]
643 fn allowlist_masks_fully_literal_credential_pair() {
644 let redacted =
645 redact_url_with_query_allowlist("tcp://h:61616?next=user:pass@host", JMS_KEYS);
646 assert_eq!(redacted, "tcp://h:61616?<redacted>");
647 assert!(
648 !redacted.contains("pass"),
649 "literal credential leaked: {redacted}"
650 );
651 }
652
653 /// Regression pin at the under-redaction boundary: `@` without `:` or
654 /// `//` is an email address, not a credential — the pair stays visible.
655 #[test]
656 fn allowlist_keeps_lone_email_value() {
657 assert_eq!(
658 redact_url_with_query_allowlist("tcp://h:61616?contact=admin%40corp.example", JMS_KEYS),
659 "tcp://h:61616?contact=admin%40corp.example"
660 );
661 }
662
663 /// Over-mask stance (ADR-0051): a value decoding to `user@host:port`
664 /// is credential-shaped even though no literal `//` or `:` is present.
665 #[test]
666 fn allowlist_masks_credential_shaped_userhostport() {
667 let redacted =
668 redact_url_with_query_allowlist("tcp://h:61616?next=user%40host%3Aport", JMS_KEYS);
669 assert_eq!(redacted, "tcp://h:61616?<redacted>");
670 }
671
672 /// A percent-encoded sensitive key decodes to `password` and must
673 /// redact; the rendered key keeps its original encoded bytes.
674 #[test]
675 fn allowlist_decodes_percent_encoded_sensitive_key() {
676 let redacted =
677 redact_url_with_query_allowlist("tcp://host:61616?pass%77ord=shortsecret", JMS_KEYS);
678 assert!(
679 redacted.contains("pass%77ord=<redacted>"),
680 "encoded key must redact keeping original bytes: {redacted}"
681 );
682 assert!(
683 !redacted.contains("shortsecret"),
684 "secret value leaked: {redacted}"
685 );
686 }
687
688 /// Regression pin (redact2): non-ASCII bytes in a pair are preserved
689 /// byte-for-byte (valid UTF-8), never Latin-1-transcoded, and a decoded
690 /// invalid byte becomes `U+FFFD` — the decoded output feeds only ASCII
691 /// shape checks, so neither case can flip a redaction decision.
692 #[test]
693 fn minimal_decode_preserves_non_ascii_bytes() {
694 // `é` (0xC3 0xA9) must survive as itself, not as `é`.
695 assert_eq!(minimal_decode_pair("café"), "café");
696 // Escapes still decode around non-ASCII bytes.
697 assert_eq!(minimal_decode_pair("café%40x"), "café@x");
698 // A decoded invalid byte (0xFF from `%FF`) becomes U+FFFD, which
699 // cannot match an ASCII needle (exercised via the full-hex decoder).
700 assert_eq!(decode_match_key("a%FFb"), "a\u{FFFD}b");
701 }
702
703 // ── Migrated: camel-config config_tests/url_redaction_tests.rs ──────
704
705 #[test]
706 fn redact_url_keeps_userinfo_mask_shape() {
707 assert_eq!(
708 redact_url("redis://user:secret@h:6379"),
709 "redis://***@h:6379"
710 );
711 }
712
713 /// Intentional semantic change: the query was previously echoed verbatim,
714 /// leaking `?password=...` cache credentials into Debug output.
715 #[test]
716 fn redact_url_drops_query_secrets() {
717 assert_eq!(
718 redact_url("redis://h:6379/0?password=hunter2"),
719 "redis://h:6379/0?[redacted]"
720 );
721 }
722
723 #[test]
724 fn redact_url_drops_fragment() {
725 assert_eq!(
726 redact_url("redis://h:6379/0#tok=x"),
727 "redis://h:6379/0#[redacted]"
728 );
729 }
730
731 /// Multiple `@` in the window: mask through the LAST one — over-masking is
732 /// safe, under-masking is not.
733 #[test]
734 fn redact_url_masks_through_last_at() {
735 assert_eq!(redact_url("redis://user:p@ss@h:6379"), "redis://***@h:6379");
736 }
737
738 /// Bare-host masking (bd rc-8bxeo, promoted from the camel-ws /
739 /// camel-http twins): clean hosts pass through unchanged for
740 /// diagnosability; userinfo-shaped values mask through the LAST `@`.
741 #[test]
742 fn redact_host_masks_userinfo_keeps_clean_hosts() {
743 assert_eq!(redact_host("localhost"), "localhost");
744 assert_eq!(redact_host("broker.example.com"), "broker.example.com");
745 assert_eq!(redact_host("host.example:8080"), "host.example:8080");
746 assert_eq!(redact_host("user:pass@host"), "***@host");
747 assert_eq!(redact_host("bob:p@ss@host"), "***@host");
748 assert_eq!(redact_host("a@b@c"), "***@c");
749 }
750
751 /// A slash run after `//` must not hide userinfo behind it; extra leading
752 /// slashes are kept byte-for-byte.
753 #[test]
754 fn redact_url_slash_run_evader_masked() {
755 assert_eq!(
756 redact_url("redis:////user:pass@h:6379/0"),
757 "redis:////***@h:6379/0"
758 );
759 }
760
761 /// An `@` outside any `//` window (path data here) is not userinfo.
762 #[test]
763 fn redact_url_at_outside_window_visible() {
764 assert_eq!(
765 redact_url("redis://h:6379/0/user@x"),
766 "redis://h:6379/0/user@x"
767 );
768 }
769
770 /// A `//` window after the first is scanned too: credentials cannot hide
771 /// behind a benign first window.
772 #[test]
773 fn redact_url_later_window_masked() {
774 assert_eq!(redact_url("redis://h//user:pass@x/"), "redis://h//***@x/");
775 }
776
777 /// Compose-both rule: one sentinel per distinct introducer found in the
778 /// raw URL, in first-occurrence order.
779 #[test]
780 fn redact_url_sentinels_compose_both() {
781 assert_eq!(
782 redact_url("redis://h:6379/0?password=x#tok=y"),
783 "redis://h:6379/0?[redacted]#[redacted]"
784 );
785 }
786
787 #[test]
788 fn redact_url_sentinels_compose_fragment_first() {
789 assert_eq!(
790 redact_url("redis://h:6379/0#tok=y?password=x"),
791 "redis://h:6379/0#[redacted]?[redacted]"
792 );
793 }
794
795 /// The 256-byte cap must land on a UTF-8 char boundary: a 3-byte char
796 /// straddling byte 256 forces the cut to walk back instead of panicking.
797 #[test]
798 fn redact_url_truncates_256_utf8_safe() {
799 let mut url = format!("redis://{}{}", "x".repeat(246), '日');
800 url.push_str(&"tail".repeat(20));
801 assert!(url.len() > 300);
802 let redacted = redact_url(&url);
803 assert!(redacted.len() <= 256, "len={}", redacted.len());
804 assert!(redacted.starts_with("redis://"));
805 }
806
807 /// e_gpt stage-4: the 256-byte cap must not split an appended sentinel.
808 /// The base is truncated at `256 - sentinel_len` BEFORE the sentinel is
809 /// appended, so the sentinel always renders intact and the total stays
810 /// ≤ 256.
811 #[test]
812 fn redact_url_keeps_sentinel_intact_under_256_cap() {
813 // Base (masked, cut at the `?`) is 248 bytes, so byte 256 lands inside
814 // the appended `?[redacted]` (starts at 248) pre-fix.
815 let url = format!("http://{}?x=1", "a".repeat(240));
816 let redacted = redact_url(&url);
817 assert!(redacted.len() <= 256, "len={}", redacted.len());
818 assert!(
819 redacted.ends_with("?[redacted]"),
820 "sentinel must render intact: {redacted}"
821 );
822 }
823
824 /// The untruncated variant must keep every byte that the 256-byte
825 /// diagnostic cap would otherwise drop, so it can serve as stable
826 /// identity material without conflating two long URLs that share a
827 /// prefix.
828 #[test]
829 fn redact_url_with_query_allowlist_untruncated_keeps_full_length() {
830 let long = format!(
831 "tcp://h:61616?keepAlive=true&brokerName={}",
832 "segment".repeat(60)
833 );
834 assert!(long.len() > 256);
835 let capped = redact_url_with_query_allowlist(&long, &[]);
836 let full = redact_url_with_query_allowlist_untruncated(&long, &[]);
837 assert!(capped.len() <= 256, "len={}", capped.len());
838 assert_eq!(full, long, "untruncated must preserve the whole string");
839 assert!(full.len() > capped.len());
840 // Redaction still applies without the cap.
841 let mut secret = String::from("tcp://admin:secret@h:61616?keepAlive=true&padding=");
842 secret.push_str(&"x".repeat(300));
843 let redacted = redact_url_with_query_allowlist_untruncated(&secret, &["password", "user"]);
844 assert!(!redacted.contains("secret"));
845 assert!(redacted.len() > 256);
846 }
847
848 // ── Migrated: camel-jms config.rs broker allowlist tests ────────────
849
850 /// Audit 2026-08-31, F5-3: JMS broker URLs must not leak credentials
851 /// through Debug output.
852 #[test]
853 fn redact_broker_url_masks_userinfo_and_sensitive_query() {
854 // userinfo form
855 let redacted = redact_url_with_query_allowlist(
856 "tcp://admin:secretpass@broker.example.com:61616",
857 JMS_KEYS,
858 );
859 assert!(
860 !redacted.contains("secretpass"),
861 "password masked: {redacted}"
862 );
863 assert!(
864 redacted.contains("broker.example.com"),
865 "host visible: {redacted}"
866 );
867
868 // failover + query-param form (ActiveMQ style)
869 let redacted = redact_url_with_query_allowlist(
870 "failover:(tcp://host:61616)?jms.userName=admin&jms.password=secret&keepAlive=true",
871 JMS_KEYS,
872 );
873 assert!(
874 !redacted.contains("secret"),
875 "password param masked: {redacted}"
876 );
877 assert!(
878 !redacted.contains("=admin"),
879 "username param masked: {redacted}"
880 );
881 assert!(
882 redacted.contains("keepAlive=true"),
883 "benign param kept: {redacted}"
884 );
885
886 // clean URL untouched
887 assert_eq!(
888 redact_url_with_query_allowlist("tcp://host:61616", JMS_KEYS),
889 "tcp://host:61616"
890 );
891 }
892
893 /// Exact-output pin: userinfo is fully masked by `***`, never partially
894 /// truncated, in scheme://...@ authority position.
895 #[test]
896 fn redact_exact_userinfo_mask() {
897 assert_eq!(
898 redact_url_with_query_allowlist(
899 "tcp://admin:secretpass@broker.example.com:61616",
900 JMS_KEYS
901 ),
902 "tcp://***@broker.example.com:61616"
903 );
904 }
905
906 /// Exact-output pin: sensitive query params redact to `<redacted>`,
907 /// benign params survive byte-for-byte, `&` separators preserved.
908 #[test]
909 fn redact_exact_query_join() {
910 assert_eq!(
911 redact_url_with_query_allowlist(
912 "tcp://host:61616?password=p&user=u&keepAlive=true",
913 JMS_KEYS
914 ),
915 "tcp://host:61616?password=<redacted>&user=<redacted>&keepAlive=true"
916 );
917 }
918
919 /// Exact-output pin: a bare `user@host` (no scheme) is NOT an authority
920 /// position — passthrough untouched.
921 #[test]
922 fn redact_exact_bare_at_passthrough() {
923 assert_eq!(
924 redact_url_with_query_allowlist("admin@host", JMS_KEYS),
925 "admin@host"
926 );
927 }
928
929 /// Exact-output pin: delimiter-exact query redaction on the ActiveMQ
930 /// failover form — the whole query part after '?' must match verbatim,
931 /// no dropped params, no mangled separators.
932 #[test]
933 fn redact_exact_failover_param_boundaries() {
934 let redacted = redact_url_with_query_allowlist(
935 "failover:(tcp://host:61616)?jms.userName=admin&jms.password=secret&keepAlive=true",
936 JMS_KEYS,
937 );
938 let (_, query) = redacted.split_once('?').expect("query segment after '?'");
939 assert_eq!(
940 query,
941 "jms.userName=<redacted>&jms.password=<redacted>&keepAlive=true"
942 );
943 }
944
945 /// bd rc-eh49 exact pin: an `@` riding the query is NOT userinfo — the
946 /// URL must pass through byte-for-byte. The old whole-string
947 /// `split_once('@')` mangled this to `failover:(tcp://***@b`.
948 #[test]
949 fn redact_broker_url_query_at_no_misfire() {
950 assert_eq!(
951 redact_url_with_query_allowlist("failover:(tcp://h:61616)?x=a@b", JMS_KEYS),
952 "failover:(tcp://h:61616)?x=a@b"
953 );
954 }
955
956 /// bd rc-eh49 exact pin: a slash run after `//` is preserved
957 /// byte-for-byte and the windowed mask composes with the per-key query
958 /// allowlist. The old whole-string scan rewrote from the scheme's own
959 /// `://` and swallowed the extra slashes.
960 #[test]
961 fn redact_exact_slash_run_window_composition() {
962 assert_eq!(
963 redact_url_with_query_allowlist("tcp:////user:pass@h:61616?keepAlive=true", JMS_KEYS),
964 "tcp:////***@h:61616?keepAlive=true"
965 );
966 }
967
968 /// bd rc-eh49 exact pin: the window mask consumes through the LAST `@`
969 /// of the window, so an `@` embedded in the userinfo cannot keep a
970 /// spoofable prefix alive (the old first-`@` scan left `a@` visible).
971 #[test]
972 fn redact_exact_last_at_in_window() {
973 assert_eq!(
974 redact_url_with_query_allowlist("tcp://u:p@a@h:61616", JMS_KEYS),
975 "tcp://***@h:61616"
976 );
977 }
978
979 /// bd rc-eh49: the 256-byte cap cuts on a char boundary — a multibyte
980 /// character straddling byte 256 is dropped whole, never split
981 /// mid-encode.
982 #[test]
983 fn redact_broker_url_truncate_multibyte_boundary() {
984 let mut url = String::from("tcp://broker:61616/");
985 url.push_str(&"x".repeat(236)); // 255 ASCII bytes before the multibyte char
986 url.push('日'); // 3 bytes straddling the 256-byte cap
987 url.push_str(&"y".repeat(50)); // push the total past 300 bytes
988 assert!(url.len() > 300);
989 let redacted = redact_url_with_query_allowlist(&url, JMS_KEYS);
990 assert!(redacted.len() <= 256, "len={}", redacted.len());
991 assert!(std::str::from_utf8(redacted.as_bytes()).is_ok());
992 assert!(
993 !redacted.contains('日'),
994 "straddling char dropped whole: {redacted}"
995 );
996 }
997
998 /// bd rc-eh49: a `//` window after the first is scanned too; the
999 /// benign query stays visible per the broker allowlist exception.
1000 #[test]
1001 fn redact_broker_url_later_window_masked() {
1002 assert_eq!(
1003 redact_url_with_query_allowlist("tcp://h//user:pass@x/?keepAlive=true", JMS_KEYS),
1004 "tcp://h//***@x/?keepAlive=true"
1005 );
1006 }
1007
1008 /// bd rc-eh49: fragments are never echoed — everything from the first
1009 /// `#` is dropped and the `#[redacted]` sentinel appended; processed
1010 /// query params stay visible.
1011 #[test]
1012 fn redact_broker_url_drops_fragment() {
1013 assert_eq!(
1014 redact_url_with_query_allowlist("tcp://h:61616?keepAlive=true#tok=x", JMS_KEYS),
1015 "tcp://h:61616?keepAlive=true#[redacted]"
1016 );
1017 }
1018
1019 /// bd rc-eh49: broker URLs are capped at 256 bytes on a UTF-8 char
1020 /// boundary.
1021 #[test]
1022 fn redact_broker_url_truncates() {
1023 let mut url = format!("tcp://broker:61616/{}", "x".repeat(300));
1024 url.push('日');
1025 url.push_str(&"tail".repeat(20));
1026 assert!(url.len() > 300);
1027 let redacted = redact_url_with_query_allowlist(&url, JMS_KEYS);
1028 assert!(redacted.len() <= 256, "len={}", redacted.len());
1029 assert!(redacted.starts_with("tcp://broker:61616/"));
1030 }
1031
1032 // ── Migrated: camel-jms component.rs redact_url_* block ─────────────
1033
1034 #[test]
1035 fn redact_url_strips_userinfo_with_password() {
1036 assert_eq!(
1037 redact_url("tcp://admin:s3cret@broker:61616"),
1038 "tcp://***@broker:61616"
1039 );
1040 }
1041
1042 #[test]
1043 fn redact_url_strips_userinfo_without_password() {
1044 assert_eq!(
1045 redact_url("tcp://admin@broker:61616"),
1046 "tcp://***@broker:61616"
1047 );
1048 }
1049
1050 #[test]
1051 fn redact_url_passes_clean_url_unchanged() {
1052 assert_eq!(redact_url("tcp://localhost:61616"), "tcp://localhost:61616");
1053 }
1054
1055 #[test]
1056 fn redact_url_handles_ssl_scheme() {
1057 assert_eq!(
1058 redact_url("ssl://user:pass@secure-broker:61617"),
1059 "ssl://***@secure-broker:61617"
1060 );
1061 }
1062
1063 /// bd rc-eh49: the earliest `?`/`#` introducer wins; both sentinels
1064 /// compose in first-occurrence order when the raw URL carries both
1065 /// introducers, and fragment bytes after the cut are dropped with it.
1066 #[test]
1067 fn redact_url_drops_query_and_fragment() {
1068 assert_eq!(
1069 redact_url("tcp://broker:61616?user=a#tok=x"),
1070 "tcp://broker:61616?[redacted]#[redacted]"
1071 );
1072 }
1073
1074 /// bd rc-eh49 compose-both rule: a `#` before `?` flips the sentinel
1075 /// order accordingly. (Name prefixed `jms_`: the redis fixture from
1076 /// camel-config already owns the unprefixed name above.)
1077 #[test]
1078 fn jms_redact_url_sentinels_compose_fragment_first() {
1079 assert_eq!(
1080 redact_url("tcp://broker:61616#tok=x?user=a"),
1081 "tcp://broker:61616#[redacted]?[redacted]"
1082 );
1083 }
1084
1085 /// bd rc-eh49: a `//` window after the first is scanned too —
1086 /// credentials cannot hide behind a benign first window. (Name prefixed
1087 /// `jms_`: the redis fixture from camel-config already owns the
1088 /// unprefixed name above.)
1089 #[test]
1090 fn jms_redact_url_later_window_masked() {
1091 assert_eq!(redact_url("tcp://h//user:pass@x/"), "tcp://h//***@x/");
1092 }
1093
1094 /// bd rc-eh49: a slash run after `//` cannot hide userinfo from the
1095 /// window scan.
1096 #[test]
1097 fn redact_url_slash_run_masked() {
1098 let redacted = redact_url("tcp:////user:pass@broker:61616");
1099 assert!(!redacted.contains("user:pass"), "leaked: {redacted}");
1100 assert!(
1101 redacted.contains("***@broker:61616"),
1102 "masked in place: {redacted}"
1103 );
1104 }
1105
1106 /// bd rc-eh49: the old first-`@`-anywhere scan masked through an `@`
1107 /// riding the query (`tcp://***@b`); only a window `@` is userinfo.
1108 #[test]
1109 fn redact_url_at_in_query_not_userinfo_mask() {
1110 assert_eq!(
1111 redact_url("tcp://broker:61616?q=a@b"),
1112 "tcp://broker:61616?[redacted]"
1113 );
1114 }
1115
1116 /// bd rc-eh49: the 256-byte cap cuts on a char boundary — a multibyte
1117 /// character straddling byte 256 is dropped whole, never split
1118 /// mid-encode.
1119 #[test]
1120 fn redact_url_truncate_multibyte_boundary() {
1121 let mut url = String::from("tcp://broker:61616/");
1122 url.push_str(&"x".repeat(236)); // 255 ASCII bytes before the multibyte char
1123 url.push('日'); // 3 bytes straddling the 256-byte cap
1124 url.push_str(&"y".repeat(50)); // push the total past 300 bytes
1125 assert!(url.len() > 300);
1126 let redacted = redact_url(&url);
1127 assert!(redacted.len() <= 256, "len={}", redacted.len());
1128 assert!(std::str::from_utf8(redacted.as_bytes()).is_ok());
1129 assert!(
1130 !redacted.contains('日'),
1131 "straddling char dropped whole: {redacted}"
1132 );
1133 }
1134
1135 /// e_gpt stage-4: the 256-byte cap must not split an appended sentinel.
1136 /// The base is truncated at `256 - sentinel_len` BEFORE the sentinel is
1137 /// appended, so the sentinel always renders intact and the total stays
1138 /// ≤ 256. Covers both redactors: `redact_url` (compose sentinels) and
1139 /// `redact_url_with_query_allowlist` (fragment sentinel over kept query
1140 /// content).
1141 #[test]
1142 fn redact_url_keeps_sentinels_intact_under_256_cap() {
1143 // `redact_url`: base (masked, cut at the `?`) is 255 bytes, so byte
1144 // 256 lands inside the appended `?[redacted]` (starts at 255) pre-fix.
1145 let url = format!("tcp://{}?x=1", "a".repeat(250));
1146 let redacted = redact_url(&url);
1147 assert!(redacted.len() <= 256, "len={}", redacted.len());
1148 assert!(
1149 redacted.ends_with("?[redacted]"),
1150 "redact_url sentinel must render intact: {redacted}"
1151 );
1152
1153 // `redact_url_with_query_allowlist`: base + kept query is 252 bytes,
1154 // so byte 256 lands inside the appended `#[redacted]` (starts at
1155 // 252) pre-fix.
1156 let broker = format!("tcp://{}?keep=1#frag", "a".repeat(240));
1157 let redacted = redact_url_with_query_allowlist(&broker, JMS_KEYS);
1158 assert!(redacted.len() <= 256, "len={}", redacted.len());
1159 assert!(
1160 redacted.ends_with("#[redacted]"),
1161 "allowlist sentinel must render intact: {redacted}"
1162 );
1163 }
1164
1165 // ── Migrated: camel-http lib.rs pure-string redact tests ────────────
1166 // These landed against the Err arm of `redact_url_for_diagnostics`
1167 // (unparseable inputs); that arm delegates to `redact_url` /
1168 // `redact_url_fail_closed`, so the fixtures pin the canonical helpers
1169 // byte-identically. The `url::Url::parse` precondition asserts of the
1170 // landed tests do not apply here: this module never parses.
1171
1172 #[test]
1173 fn redact_url_unparseable_fragment_credentials_dropped() {
1174 let raw = "ht tps://app.example/cb#access_token=SECRET";
1175 let redacted = redact_url(raw);
1176 assert!(
1177 !redacted.contains("SECRET"),
1178 "unparseable fragment token leaked: {redacted}"
1179 );
1180 assert!(
1181 !redacted.contains("access_token"),
1182 "unparseable fragment bytes leaked: {redacted}"
1183 );
1184 assert!(
1185 redacted.contains("#[redacted]"),
1186 "unparseable fragment must end in the sentinel: {redacted}"
1187 );
1188 }
1189
1190 #[test]
1191 fn redact_url_empty_host_userinfo_sentinel() {
1192 // The landed http test drives this through the parse-failure arm;
1193 // the canonical helper fails closed on the window `@` directly.
1194 let redacted = redact_url_fail_closed("scheme://user@");
1195 assert_eq!(
1196 redacted, "[redacted]",
1197 "empty-host userinfo must fail closed: {redacted}"
1198 );
1199 }
1200
1201 #[test]
1202 fn redact_url_unparseable_slash_run_evader_sentinel() {
1203 let raw = "schem e:////user:pass@evil/";
1204 let redacted = redact_url_fail_closed(raw);
1205 assert_eq!(
1206 redacted, "[redacted]",
1207 "unparseable slash-run evader must fail closed: {redacted}"
1208 );
1209 }
1210
1211 #[test]
1212 fn redact_url_unparseable_later_window_userinfo_sentinel() {
1213 // The first `//` window ("ho st") carries no `@`, but a later
1214 // `//user:pass@evil/` window does. The scan must consider every
1215 // `//` window, not just the first, or the credentials echo.
1216 let raw = "http://ho st/a//user:pass@evil/";
1217 let redacted = redact_url_fail_closed(raw);
1218 assert_eq!(
1219 redacted, "[redacted]",
1220 "userinfo in a later // window must fail closed: {redacted}"
1221 );
1222 }
1223
1224 #[test]
1225 fn redact_url_truncates_unparseable() {
1226 let long = "x".repeat(1000);
1227 let redacted = redact_url(&long);
1228 assert_eq!(redacted.len(), 256, "unparseable URL must be truncated");
1229 }
1230
1231 #[test]
1232 fn redact_url_suppresses_unparseable_authority_credentials() {
1233 let fixtures = [
1234 "http://u:secretpw@/x",
1235 "http://u:secretpw@host:99999/x",
1236 "http://u:secretpw@host:99999",
1237 "//u:secretpw@h/x",
1238 ];
1239 for fixture in fixtures {
1240 assert_eq!(
1241 redact_url_fail_closed(fixture),
1242 "[redacted]",
1243 "credential-bearing authority must be suppressed: {fixture}"
1244 );
1245 }
1246 }
1247
1248 #[test]
1249 fn redact_url_unparseable_query_redacted_short_and_long() {
1250 let short = "http://host:99999/path?token=shortsecret";
1251 let redacted = redact_url(short);
1252 assert_eq!(
1253 redacted, "http://host:99999/path?[redacted]",
1254 "short unparseable query must end with the suffix: {redacted}"
1255 );
1256
1257 let mut long = String::from("http://host:99999/");
1258 long.push_str(&"a".repeat(300));
1259 long.push_str("?token=longsecret");
1260 let redacted = redact_url(&long);
1261 assert!(
1262 !redacted.contains("longsecret"),
1263 "long unparseable query leaked a query byte: {redacted}"
1264 );
1265 assert!(
1266 redacted.len() <= 256,
1267 "long unparseable query must be capped: {} bytes",
1268 redacted.len()
1269 );
1270 }
1271
1272 #[test]
1273 fn redact_url_unparseable_sentinels_compose_both() {
1274 // Compose-both rule: one sentinel per distinct introducer found in
1275 // the raw string, in first-occurrence order.
1276 let raw = "ht tp://h.example/p?a=1#tok=x";
1277 assert_eq!(
1278 redact_url(raw),
1279 "ht tp://h.example/p?[redacted]#[redacted]",
1280 "query and fragment sentinels must compose: {raw}"
1281 );
1282 }
1283
1284 #[test]
1285 fn redact_url_unparseable_sentinels_compose_fragment_first() {
1286 let raw = "ht tp://h.example/p#tok=x?a=1";
1287 assert_eq!(
1288 redact_url(raw),
1289 "ht tp://h.example/p#[redacted]?[redacted]",
1290 "sentinels must follow the introducers' first-occurrence order: {raw}"
1291 );
1292 }
1293
1294 #[test]
1295 fn redact_url_unparseable_utf8_straddle_no_panic() {
1296 let fixture = format!("a{}", "é".repeat(200));
1297 let redacted = redact_url(&fixture);
1298 assert!(
1299 redacted.len() <= 256,
1300 "straddle fixture must be capped: {} bytes",
1301 redacted.len()
1302 );
1303 assert!(
1304 redacted.len() >= 253,
1305 "straddle fixture must not over-truncate: {} bytes",
1306 redacted.len()
1307 );
1308 assert!(
1309 fixture.is_char_boundary(redacted.len()),
1310 "cut must land on a UTF-8 char boundary: {} bytes",
1311 redacted.len()
1312 );
1313 }
1314
1315 #[test]
1316 fn redact_url_at_sign_outside_authority_window_visible() {
1317 let at_sign_in_path = "http://host:99999/x@y";
1318 assert_eq!(
1319 redact_url_fail_closed(at_sign_in_path),
1320 at_sign_in_path,
1321 "at-sign in path must not be suppressed"
1322 );
1323 // Opaque non-hierarchical strings pass through byte-identically.
1324 assert_eq!(
1325 redact_url("mailto:user@example.com"),
1326 "mailto:user@example.com",
1327 "at-sign in mailto must round-trip byte-identically"
1328 );
1329 }
1330
1331 // ── Phase 3: scheme-gated backslash windows (rc-f05q8) ──────────────
1332
1333 /// A backslash authority after a non-special scheme opens a window:
1334 /// `foo:\user:pass@evil/` carries no `//` run, yet the scheme-prefixed
1335 /// backslash run must mask the userinfo. The fail-closed variant
1336 /// suppresses the whole string; the clean sibling stays visible.
1337 #[test]
1338 fn backslash_run_non_special_scheme_masked() {
1339 assert_eq!(
1340 redact_url("foo:\\user:pass@evil/"),
1341 "foo:\\***@evil/",
1342 "non-special-scheme backslash authority must mask userinfo"
1343 );
1344 assert_eq!(
1345 redact_url_fail_closed("foo:\\user:pass@evil/"),
1346 "[redacted]",
1347 "non-special-scheme backslash authority must fail closed"
1348 );
1349 assert_eq!(
1350 redact_url("foo:\\clean/path"),
1351 "foo:\\clean/path",
1352 "clean backslash sibling stays visible"
1353 );
1354 }
1355
1356 /// A single backslash after a multi-character scheme opens a window
1357 /// even though the run is one character long.
1358 #[test]
1359 fn backslash_single_after_multi_char_scheme_masked() {
1360 let redacted = redact_url("http:\\user:pass@evil\\path");
1361 assert!(
1362 !redacted.contains("user:pass"),
1363 "single-backslash authority leaked: {redacted}"
1364 );
1365 assert!(
1366 redacted.contains("***@"),
1367 "single-backslash authority must mask userinfo: {redacted}"
1368 );
1369 }
1370
1371 /// A single backslash after a one-character scheme opens a window when
1372 /// the candidate window content is credential-shaped (`:` before the
1373 /// last `@`).
1374 #[test]
1375 fn backslash_single_after_one_char_scheme_credential_shaped_masked() {
1376 let redacted = redact_url("x:\\user:pass@evil");
1377 assert!(
1378 !redacted.contains("user:pass"),
1379 "one-char-scheme backslash authority leaked: {redacted}"
1380 );
1381 assert!(
1382 redacted.contains("***@"),
1383 "one-char-scheme backslash authority must mask userinfo: {redacted}"
1384 );
1385 }
1386
1387 /// Windows drive path: single backslash after a one-character scheme
1388 /// with no `:` in the candidate window — no qualifying window, the
1389 /// string renders unchanged.
1390 #[test]
1391 fn drive_path_stays_visible() {
1392 assert_eq!(
1393 redact_url("C:\\Users\\x@corp\\file"),
1394 "C:\\Users\\x@corp\\file"
1395 );
1396 }
1397
1398 /// UNC path: no scheme prefix before the backslash run — no qualifying
1399 /// window, the string renders unchanged.
1400 #[test]
1401 fn unc_path_stays_visible() {
1402 assert_eq!(redact_url("\\\\server\\x@y"), "\\\\server\\x@y");
1403 }
1404
1405 /// The fail-closed window scan must see scheme-prefixed backslash
1406 /// windows, while drive paths stay window-free.
1407 #[test]
1408 fn window_has_at_sign_sees_backslash_windows() {
1409 assert!(
1410 window_has_at_sign("foo:\\u:p@e/"),
1411 "scheme-prefixed backslash window must carry the at-sign"
1412 );
1413 assert!(
1414 !window_has_at_sign("C:\\Users\\x@corp\\file"),
1415 "drive path must not open a backslash window"
1416 );
1417 }
1418
1419 /// Branch pins for the scheme-gated backslash window rule: a len-2
1420 /// backslash run behind a scheme prefix opens (introducer stays
1421 /// visible), a mixed `/`+`\` run of length 2 behind a 1-char scheme
1422 /// opens, an invalid scheme char before the run gates it off, a
1423 /// scheme-less mixed run stays closed, and a single `\` behind a
1424 /// 1-char scheme with no `:` in the candidate content stays closed.
1425 #[test]
1426 fn backslash_gate_branches_pinned() {
1427 assert_eq!(redact_url("foo:\\\\user:pass@evil/"), "foo:\\\\***@evil/");
1428 assert_eq!(redact_url("a:/\\user:pass@evil/"), "a:/\\***@evil/");
1429 assert_eq!(
1430 redact_url("notscheme%\\user:pass@evil/"),
1431 "notscheme%\\user:pass@evil/"
1432 );
1433 assert_eq!(redact_url("/\\user:pass@evil/"), "/\\user:pass@evil/");
1434 assert_eq!(redact_url("a:\\user@evil"), "a:\\user@evil");
1435 assert!(window_has_at_sign("foo:\\\\u:p@e/"));
1436 assert!(!window_has_at_sign("a:\\user@evil"));
1437 }
1438
1439 /// Key-position credential-shape predicate (bd rc-yvjp3): a raw query
1440 /// key that decodes to `user:secret@host` is credential-shaped, while a
1441 /// benign parameter name is not.
1442 #[test]
1443 fn credential_shaped_key_predicate_flags_userinfo() {
1444 assert!(is_credential_shaped_key("user:secret@host"));
1445 assert!(!is_credential_shaped_key("mysteryBox"));
1446 }
1447
1448 /// The same predicate must see through a single-pass percent-encoded
1449 /// credential shape.
1450 #[test]
1451 fn credential_shaped_key_predicate_flags_percent_encoded() {
1452 assert!(is_credential_shaped_key("user%3Asecret%40host"));
1453 }
1454}