feat(modules): named payload sets with streaming generation - #386
Open
TBX3D wants to merge 13 commits into
Open
Conversation
checkMatchers/checkMatcher took (resp, body), so every matcher type that needs anything else (the request url, how long the round trip took, extractor output) forces another signature change through the whole engine and all of its callers. collect the per-response state into a MatchContext built once at each of the two call sites (single request and chain step) and thread that instead. no matcher behavior changes: the classic types read Resp and Body exactly as before. extraction moves above the matcher check in executeHTTPRequest. runExtractors is side-effect-free and the finding is only built on a match, so the order is behavior-preserving, and it lets a matcher read extractor values from the context. in a chain step the context carries the running variable set, so a matcher there also sees earlier steps' values.
add the dsl matcher type with a curated helper allowlist and load-time compile validation (bad syntax, non-allowlisted functions, empty or over-length expressions all rejected before scan time). evaluation lands in a follow-up; a loaded dsl matcher currently misses at match time.
drives a live httptest server through ExecuteHTTPModule with a dsl matcher bound to status_code and body, mirroring the existing favicon integration test, and asserts exactly one finding.
parse the request url and expose host[:port] so a nuclei-style host == "..." expression matches; dedupe the header serialization.
covers the bound variables, the helper allowlist and the load-time compile, so a module author does not have to read dsl.go to know what an expression can touch.
overload the payloads field with a custom unmarshaler: a yaml sequence is the
legacy single set named "payload", a mapping is ordered named sets injectable
at {{name}}. generation moves to a lazy iter.Seq that never materializes the
product, and substitution now reaches header values. file-backed sets parse but
are rejected until the loader lands.
a named set whose yaml value is a scalar string is a local wordlist path, loaded via the existing loadWordlist at resolve time. resolveSets is the single point set resolution can fail.
per fuzzing-module-per-target request cap, default 25000, 0 = unlimited. threaded into modules.Options; enforced by the executor in a follow-up.
ExecuteHTTPModule consumes streamRequests directly instead of materializing the full product and sizing the results channel to it. a fixed worker pool pulls from an unbuffered feed, a single collector owns the findings slice (dropping the mutex), and the producer enforces -fuzz-max-requests with a one-shot truncation log.
the existing cancel test only ever passes an already-cancelled context, so the producer returns at its up-front ctx.Err() check and never reaches the send-select guarding reqCh. add a regression test that cancels while a large payload set is mid-flight against a blocking server, so the producer's and workers' send-selects both get exercised, and assert ExecuteHTTPModule still returns promptly with no goroutine leak.
substitute header values on the no-payload path too, matching the payload and chain paths; warn when a resolved payload set is empty so a misconfigured wordlist is not mistaken for a clean no-findings run. harden the mid-stream cancel test to poll for goroutine drain instead of a fixed sleep.
the payloads field grew two shapes (named sets, file-backed sets) and a scan-wide cap; none of that was discoverable without reading yaml.go.
|
Codecov Report❌ Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## main #386 +/- ##
=======================================
Coverage ? 65.56%
=======================================
Files ? 89
Lines ? 8082
Branches ? 0
=======================================
Hits ? 5299
Misses ? 2382
Partials ? 401 ☔ View full report in Codecov by Harness. 🚀 New features to boost your workflow:
|
github-actions
Bot
added
size/xl
pr summary18 files changed (+1183 -118)
|
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
stacked on #384, so only the commits above it are this pr's.
the payloads field only took one anonymous list, so a fuzzing module could vary exactly one thing, and generation materialized the whole cross product before sending anything.
a mapping now declares ordered named sets injectable at {{name}}, a scalar names a local wordlist loaded through the existing loadWordlist, and a sequence still means the legacy single set. generation becomes a lazy iter.Seq consumed by a fixed worker pool over an unbuffered feed with one collector owning the findings slice, so the mutex and the product-sized results channel both go away. -fuzz-max-requests caps a module at 25000 requests per target by default.