Struct VendorClassifier 

pub struct VendorClassifier { /* private fields */ }

Vendor-classification engine.

Construct with VendorClassifier::with_builtin_defaults to load the four baseline Tier 1 vendor definitions shipped in crates/stygian-charon/data/vendors/, or VendorClassifier::new for an empty / custom registry.

The classifier is stateless and Send + Sync so it can be shared across threads and requests without locking.

Example

use stygian_charon::vendor_classifier::{VendorClassifier, VendorId};
use std::collections::BTreeMap;

let empty = VendorClassifier::new(Vec::new());
let cookies: Vec<String> = Vec::new();
let headers: BTreeMap<String, String> = BTreeMap::new();
let classification = empty.classify(&cookies, &headers, None, "https://example.com/");
assert_eq!(classification.top_vendor, VendorId::Unknown);
assert!(classification.is_unknown());

Implementations

impl VendorClassifier

pub const fn new(definitions: Vec<VendorDefinition>) -> Self

Build a classifier from a pre-loaded list of VendorDefinition entries.

The threshold defaults to DEFAULT_HIGH_CONFIDENCE_THRESHOLD. Override with with_threshold.

pub fn with_builtin_defaults() -> Self

Build a classifier seeded with the four baseline Tier 1 vendor definitions embedded at compile time from crates/stygian-charon/data/vendors/.

The compile-time check compile_check_builtin_vendors guarantees that every embedded TOML is valid; if it regresses, the build will fail.

Example

use stygian_charon::vendor_classifier::{VendorClassifier, VendorId};

let classifier = VendorClassifier::with_builtin_defaults();
assert!(classifier.contains(VendorId::DataDome));
assert!(classifier.contains(VendorId::PerimeterX));
assert!(classifier.contains(VendorId::Akamai));
assert!(classifier.contains(VendorId::Cloudflare));

pub fn with_threshold(self, threshold: f64) -> Self

Override the high-confidence threshold. The supplied value is clamped to (0.0, 1.0]. Non-finite values (NaN, ±∞) fall back to DEFAULT_HIGH_CONFIDENCE_THRESHOLD.

Example

use stygian_charon::vendor_classifier::{VendorClassifier, DEFAULT_HIGH_CONFIDENCE_THRESHOLD};

let classifier = VendorClassifier::new(Vec::new()).with_threshold(0.85);
assert!((classifier.threshold() - 0.85).abs() < 1e-9);

// Out-of-range values clamp to the default.
let reset = VendorClassifier::new(Vec::new()).with_threshold(f64::NAN);
assert!((reset.threshold() - DEFAULT_HIGH_CONFIDENCE_THRESHOLD).abs() < 1e-9);

pub const fn threshold(&self) -> f64

Configured high-confidence threshold.

pub fn contains(&self, vendor: VendorId) -> bool

true when the registry contains a definition for the given VendorId.

pub const fn len(&self) -> usize

Number of vendor definitions currently registered.

pub const fn is_empty(&self) -> bool

true when the registry has no definitions.

pub fn classify( &self, cookies: &[String], headers: &BTreeMap<String, String>, body: Option<&str>, url: &str, ) -> VendorClassification

Classify a single set of input strings (cookies, headers, optional body, request URL) into a ranked vendor classification.

The classifier scans every registered VendorDefinition’s signal catalogue and computes a per-vendor weighted score. The match is case-insensitive (definitions are lower-cased at load time, and the input strings are lower-cased at the match site). strings are lower-cased at the match site).

Determinism

• Signals are matched in (source, pattern) lex order.
• Ties on the top score are broken by VendorId discriminant order (see module docs).
• The output is Send + Sync and contains no HashMap/HashSet so the JSON form is byte-stable.

pub fn classify_view(&self, tx: &TransactionView) -> VendorClassification

Convenience wrapper around classify that pulls the inputs out of a TransactionView.

Cookies are extracted from the set-cookie / cookie response header (everything else is treated as a generic header). The body is the response_body_snippet. The URL is tx.url.

pub fn classify_har( &self, har_json: &str, ) -> Result<VendorClassification, HarError>

Classify every transaction in a HAR payload and return the top vendor’s classification. Cookies, headers, and body snippets are pulled from each HAR entry directly.

Errors

Returns har::HarError when the HAR JSON is invalid or exceeds a configured safety limit.

Trait Implementations

impl Clone for VendorClassifier

fn clone(&self) -> VendorClassifier

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for VendorClassifier

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.