priority: high

MIME Detection & Routing

Detection Flow

Extension → EXT_TO_MIME map → validate → Registry lookup → Extractor

Key Functions

detect_mime_type(path, inspect)

core/mime.rs

detect_mime_type_from_bytes(bytes)

core/mime.rs

validate_mime_type(mime)

core/mime.rs

Extension Mapping

118+ extensions mapped in

EXT_TO_MIME

core/mime.rs

Key mappings:

.pdf

application/pdf

.docx

application/vnd.openxmlformats-officedocument.wordprocessingml.document

.xlsx

.png

.jpg

image/*

Registry Selection

// In core/extractor/bytes.rs
fn select_extractor_for_mime(mime_type: &str) -> Result<Arc<dyn DocumentExtractor>> {
    let registry = get_document_extractor_registry();
    let registry_guard = registry.read()?;
    registry_guard.get_for_mime_type(mime_type)
        .ok_or_else(|| KreuzbergError::UnsupportedFormat(mime_type.into()))
}

Selects highest-priority extractor registered for that MIME type.

Adding New MIME Types

1. Add extension mapping:

   m.insert("ext", "application/x-new");

   in

   core/mime.rs

2. Implement

   DocumentExtractor

   with

   supported_mime_types()

   returning the MIME
3. Register in

   register_default_extractors()

Wildcard Support

Extractors can register for MIME type families:

"image/*"

image/png

image/jpeg

Critical Rules

1. Always

   validate_mime_type()

   before extraction
2. Extension mapping is case-insensitive
3. Content inspection (infer crate) is fallback for extension-less files
4. Registry validation is final authority on supported types