Skip to content

Bulk Extractor with Record Carving

What the tool does

Feature extraction and record carving

Bulk Extractor with Record Carving (bulk_extractor-rec) scans evidence bytes without relying on file-system allocation or directory structures. The supplied Windows 1.6.0-dev-rec03 build adds scanners that carve EVTX, NTFS INDX, $LogFile, $MFT, $UsnJrnl and Unix utmp records, and includes the Java Bulk Extractor Viewer for launching scans and reviewing feature reports.

  • Extract features
    Find email addresses, domains, URLs and other scanner-defined patterns with forensic paths and context.
  • Carve records
    Recover validated Windows record structures from allocated, unallocated or recursively decoded data.
  • Review reports
    Navigate feature files, histograms, carved output and source context in BEViewer.

The viewer helps launch the scanner and examine its output; it does not make every match probative. A feature can be a false positive, a carved record can be corrupt, and a forensic path is not automatically an original filename. Preserve the scan configuration, report.xml, feature row and source context.

Choose your journey

How the workflow fits together

1 Verify input2 Select scanners3 Write report4 Review forensic path5 Validate and report

Scanner output

Feature files normally contain a forensic path, extracted feature and surrounding context. Record-carving scanners also create carved directories and *_carved.txt files containing offsets and sizes.

Corrupt output is evidence too

Files labelled corrupted matched a record pattern but failed one or more integrity checks. Retain that disposition; do not silently treat them as valid.

Before you begin

Version boundary

Record the rec03 executable and Java versions; do not substitute upstream 2.x instructions.

Evidence boundary

Use a verified working image and a new output directory on separate controlled storage.

Question

Enable only scanners needed for the investigation and plan independent parsing or corroboration.

Legacy component and untrusted output

The supplied BEViewer workflow belongs to the older rec03 build. Treat its installer, Java dependency, evidence input and carved output as untrusted. Use an isolated lab, verify downloads and never execute carved content.

Analyst decisionWhich record best supports review of an extracted feature?