Top Level Namespace

Defined in:

Constant Summary

AI_CONTEXT_FEATURES = NoirAIContext::ACCEPTED_FEATURES

--ai-context accepts an optional comma-separated feature list. Crystal's OptionParser cannot express "optional positional value", so we rewrite the few well-defined ambiguous forms upfront:

--ai-context → --ai-context= (bare, all features) --ai-context=guards,sinks → unchanged (explicit value) --ai-context guards,sinks → --ai-context=guards,sinks (heuristic) --ai-context ./app → --ai-context= (next token is a path)

The heuristic for "is the next token a feature list?": lowercase words joined by commas, where either (a) there's more than one comma-separated word, or (b) the single word matches the fixed vocabulary below exactly. A real filesystem path essentially never contains a literal comma, so any multi-word comma list — typo'd feature names included — is routed to --ai-context=... and left for the vocabulary check in apply_ai_context to reject with a precise "unknown feature" error, rather than silently falling through to "Base path does not exist: ". A single bare word still has to match a known feature exactly, since that shape is genuinely ambiguous with a real one-word directory name (noir scan --ai-context myapp must keep scanning myapp). Both this and the flag's own validator used to spell the vocabulary out by hand, and both omitted sources.

ALLOWED_HTTP_METHODS = ["GET", "POST", "PUT", "DELETE", "PATCH", "OPTIONS", "HEAD", "TRACE", "CONNECT", "QUERY"]
ANDROID_EMBEDDED_SERVER_MARKER = /io\.ktor\.server\.|embeddedServer|\brouting\s*\{|\bfun\s+Route\.|org\.http4k\.routing|@RestController|@RequestMapping|@(?:Get|Post|Put|Delete|Patch)Mapping|\bRouterFunction\b/

Strong server-routing constructs. A .kt / .java file that sits inside an Android app's source set is normally scoped to mobile detectors only (an incidental import ...SpringApplication must not flag the project as a server). But an Android app can legitimately embed an on-device HTTP server — e.g. plain-app runs a local Ktor web server whose routes live under app/src/main/java/.... When a file carries one of these markers it is a real server, so the Ktor / http4k / Spring detectors are allowed to run on it. Kept as a single precompiled constant — recompiling it per file would recreate the PCRE2 program on every read.

ANDROID_SOURCE_SUBDIRS = Set {"aidl", "assets", "cpp", "java", "jni", "kotlin", "res"}
CFML_FRAMEWORK_TECHS = Set {"cfml_taffy", "cfml_coldbox", "cfml_wheels", "cfml_fw1"}

CFML frameworks that own their application's route table.

CLI_ENDPOINT_METHOD = "CLI"

The synthetic verb CLI entry points carry (cli://<binary>/<subcommand>).

DETECTOR_IGNORED_DIR_NAMES = Set {".git", ".idea", ".vscode", ".claude", "node_modules", "vendor", "__pycache__", ".venv", "venv", ".pytest_cache", ".tox", ".gradle", ".bundle", ".dart_tool", ".cargo", ".terraform", ".zig-cache", "zig-cache", ".zig-out", "zig-out", "dist", "build", "target", "out", "tmp", ".cache", ".next", ".nuxt", ".svelte-kit", ".turbo", ".parcel-cache", ".serverless", ".expo", "coverage", ".coverage", "Pods", "__MACOSX"}
DETECTOR_IGNORED_DIR_SUFFIXES = Set {".xcassets"}
DETECTOR_PATH_SEGMENT_PROBES = ["proj/grails-app/conf/application", "proj/supabase/migrations/001.sql", "proj/migrations/001.sql", "proj/supabase/config.toml", "proj/server/api/hello.js", "proj/server/routes/hello.js", "proj/pages/api/hello.js", "proj/app/api/hello/route.ts", "proj/routes/+server.ts", "proj/routes/index.dart", "proj/directus/snapshots/snap.json", "proj/wp-content/plugins/x.php", "proj/metadata/databases/tables.yaml", "proj/Magento/module.xml"]

Paths that embed directory segments real detectors gate on (/grails-app/, /migrations/, /server/api/, …). Used only to classify path-sensitive detectors — not as production file samples.

DETECTOR_SPECIAL_BASENAMES = Set {"package.json", "tsconfig.json", "composer.json", "composer.lock", "vercel.json", "now.json", "netlify.toml", "wrangler.toml", "Gemfile", "Gemfile.lock", "Package.swift", "Cargo.toml", "go.mod", "mix.exs", "pubspec.yaml", "pubspec.lock", "shard.yml", "shard.lock", "build.sbt", "pom.xml", "AndroidManifest.xml", "config.toml", "rebar.config", "erlang.mk", "project.clj", "deps.edn", "stack.yaml", "package.yaml", "gleam.toml", "manifest.toml", "paket.dependencies", "Caddyfile", "Dockerfile", "Makefile", "Rakefile", "serverless.yml", "serverless.yaml", "app.yaml", "openapi.yaml", "openapi.json", "swagger.json", "swagger.yaml"}

Filenames that detectors match by exact basename (often with path constraints like "must sit at the project root"). These must not share an extension-only cache bucket — e.g. vercel.json is not "any .json".

ENDPOINT_METHODS = (ALLOWED_HTTP_METHODS + SYNTHETIC_ENDPOINT_METHODS) + [CLI_ENDPOINT_METHOD]

Every verb that can appear as Endpoint#method. Consumers that have to tell "this token names a method" from "this token is part of a URL" (the probe matchers) read this instead of keeping their own list.

INCLUDE_TARGETS = {"path" => "include_path", "techs" => "include_techs", "callee" => "include_callee"}
LEGACY_INCLUDE_TARGETS = {"--include-path" => "include_path", "--include-techs" => "include_techs", "--include-callee" => "include_callee"}
LEGACY_PVALUE_TARGETS = {"--set-pvalue" => "set_pvalue", "--set-pvalue-header" => "set_pvalue_header", "--set-pvalue-cookie" => "set_pvalue_cookie", "--set-pvalue-query" => "set_pvalue_query", "--set-pvalue-form" => "set_pvalue_form", "--set-pvalue-json" => "set_pvalue_json", "--set-pvalue-path" => "set_pvalue_path"}

Silently translate v0 flag spellings to their v1 storage. Keeping the work here (instead of parser.on "--include-path") means the legacy names no longer clutter noir scan -h, while every v0 script keeps working untouched in v1.x.

MOBILE_DETECTOR_NAMES = Set {"android", "ios", "well_known_applinks"}
PVALUE_TYPE_KEYS = {"any" => "set_pvalue", "all" => "set_pvalue", "header" => "set_pvalue_header", "cookie" => "set_pvalue_cookie", "query" => "set_pvalue_query", "form" => "set_pvalue_form", "json" => "set_pvalue_json", "path" => "set_pvalue_path"}

Split TYPE=VAL (or bare VAL → all-types) and route it into the right slot in noir_options. --pvalue may be repeated to set multiple values across different types.

SYNTHETIC_ANY_METHODS = ["ANY", "ALL", "*"]
SYNTHETIC_ENDPOINT_METHODS = ["ANY", "PUBLISH", "SUBSCRIBE", "SEND", "RECEIVE"]

Verbs an endpoint can carry that are not HTTP methods: the wildcard ANY, and the AsyncAPI / messaging verbs the optimizer allow-lists so event-driven endpoints aren't downgraded to GET.

WILDCARD_HTTP_METHODS = ["GET", "POST", "PUT", "PATCH", "DELETE", "OPTIONS", "HEAD", "TRACE"]

Method Summary

Method Detail

def analysis_endpoints(options : Hash(String, YAML::Any), techs, logger : NoirLogger) #

[View source]
def any_to_bool(any) : Bool #

[View source]
def apply_ai_context(noir_options : Hash(String, YAML::Any), spec : String) #

--ai-context[=LIST] always enables AI context output. An empty LIST means "every category"; a non-empty LIST narrows the output to the named categories.


[View source]
def apply_include_list(noir_options : Hash(String, YAML::Any), spec : String) #

[View source]
def banner(io : IO = STDERR) #

[View source]
def build_detector_list(options : Hash(String, YAML::Any)) : Array(Detector) #

The complete detector registry.

Derived from the classes themselves rather than a hand-maintained list. Every detector under the Detector:: namespace is registered by existing; there is no second place to add it to and therefore no way to write a detector that silently never runs. That was a real failure mode — it is how zap_sites_tree ended up detectable but unnameable (see spec/unit_test/techs/registry_integrity_spec.cr, which still checks the remaining hand-maintained lists it cannot yet be derived from).

The Detector:: filter is the production contract. crystal spec compiles the suite into one binary, so all_subclasses also sees the throwaway subclasses spec/unit_test/detector/detector_for_spec.cr defines to exercise the base class; those must never join a real scan.

Sorted by class name so the order is a property of the source, not of whatever sequence require "./detectors/**" happened to produce. Order is not load-bearing — techs is accumulated concurrently across worker fibers and the optimizer sorts endpoints before output — but a stable order keeps debug logs and detector_list indices comparable between runs.


[View source]
def detect_techs(base_paths : Array(String), options : Hash(String, YAML::Any), passive_scans : Array(PassiveScan), logger : NoirLogger) #

[View source]
def detector_add_android_source_prefixes_from_dir(dir : String, prefixes : Array(String)) #

[View source]
def detector_android_source_file?(path : String, prefixes : Array(String)) : Bool #

[View source]
def detector_android_source_prefixes_for_manifest(manifest_path : String) : Array(String) #

[View source]
def detector_build_applicable_lookup(detectors : Array(Detector)) : Proc(String, Array(Int32)) #

Build a lookup that turns a path into the list of detector indices whose applicable? returns true — without re-walking every detector on every file. Most detectors only inspect extension / basename, so their answers are memoized by basename. Detectors that look at path segments or root placement are classified as path-sensitive and always evaluated against the real path.


[View source]
def detector_mobile_detector?(name : String) : Bool #

[View source]
def detector_path_sensitive?(detector : Detector) : Bool #

Whether applicable? depends on more than the basename (root placement, directory segments, multi-hop path layout). Those detectors must always see the real path in the hot loop.


[View source]
def endpoint_method_token?(token : String) : Bool #

[View source]
def escape_glob_path(path : String) : String #

Escapes glob metacharacters in a path string. This is necessary when the path contains characters like { } [ ] * ?
which would otherwise be interpreted as glob patterns. Example: "/path/{{cookiecutter}}/file" -> "/path/\{\{cookiecutter\}\}/file"


[View source]
def expand_synthetic_http_methods(method : String) : Array(String) #

[View source]
def extract_hidden_prompt_flags(noir_options : Hash(String, YAML::Any)) : Array(String) #

[View source]
def extract_legacy_aliases(args : Array(String), noir_options : Hash(String, YAML::Any)) : Array(String) #

[View source]
def filter_redundant_generic_techs(techs : Array(String)) : Array(String) #

[View source]
def generate_bash_completion_script #

[View source]
def generate_elvish_completion_script #

Native Elvish (https://elv.sh) completion. Wires the noir verb surface into $edit:completion:arg-completer so noir <Tab> lists the subcommands, noir <verb> <Tab> lists that verb's sub-actions, and noir scan <Tab> falls back to filesystem path completion.

Install: noir completion elvish > ~/.config/elvish/lib/noir.elv then add use noir to ~/.config/elvish/rc.elv.


[View source]
def generate_fish_completion_script #

[View source]
def generate_zsh_completion_script #

[View source]
def get_allowed_methods #

[View source]
def get_home #

[View source]
def get_relative_path(base_path : String, path : String) : String #

[View source]
def get_symbol(method : String) #

[View source]
def handle_pvalue(noir_options : Hash(String, YAML::Any), spec : String) #

[View source]
def initialize_analyzers(logger : NoirLogger) #

[View source]
def join_path(*segments : String) : String #

[View source]
def join_paths(*paths : String) : String #

[View source]
def json_any?(content : String) : JSON::Any | Nil #

Strict JSON.parse-or-nil. Replaces the valid_json?(content) + JSON.parse(content) idiom in detectors, which parsed the same content twice per file.


[View source]
def normalize_ai_context_flag(args : Array(String)) : Array(String) #

[View source]
def parse_yaml(content : String) : YAML::Any #

Parses YAML, recovering from a stray-tab failure that libyaml (Crystal's YAML backend) is stricter about than most other parsers.

Real-world OpenAPI/Swagger documents occasionally carry a TAB character on an otherwise-blank line inside a block scalar (descriptions, examples, embedded code). libyaml rejects it with "found a tab character where an indentation space is expected", which drops the entire document — and with it every endpoint noir would have found — even though PyYAML, JS, and Go parsers accept it. As a last resort we blank out lines that consist solely of whitespace and retry. That transformation never touches real indentation, keys, or values, so a document that already parses is returned unchanged.


[View source]
def regex_matches_with_timeout?(regex : Regex, input : String, timeout : Time::Span = 500.milliseconds) : Bool #

Safely checks if a regex matches a string within a given timeout. This helps mitigate ReDoS (Regular Expression Denial of Service) attacks.


[View source]
def remove_start_slash(input_path : String) : String #

[View source]
def requestable_http_methods(method : String) : Array(String) #

[View source]
def run_options_parser #

[View source]
def synthetic_any_method?(method : String) : Bool #

[View source]
def valid_json?(content : String) : Bool #

[View source]
def valid_yaml?(content : String) : Bool #

[View source]
def validate_ai_context_features(spec : String, origin : String) #

Rejects AI-context bucket names outside the accepted vocabulary. origin names the source in the error so a config-file typo doesn't read as a command-line one.


[View source]
def yaml_any?(content : String) : YAML::Any | Nil #

Strict YAML.parse-or-nil. Replaces the valid_yaml?(content) + YAML.parse(content) idiom in detectors, which paid for two full libyaml passes over the same content per file.


[View source]