Tarracsh is a C++20 command‑line tool for analyzing Java bytecode. It parses .class files or JARs, computes digests, builds call graphs, and can generate HTML navigation for constant‑pool entries

Features Class file parser – parse directories, JARs, or individual class files; print parse output, constant pool, filter entries, or generate an HTML navigation page

Public digest – create or compare digests of class files and JARs, run standalone or in client/server mode, and produce diff reports

Call graph – build a call‑graph database from bytecode (server/client modes currently placeholders)

Global options – output directory, logging, profiler output, and file timestamp checks configurable for all commands

Building Bazel (via Bzlmod) is the supported way to build, test, and collect coverage. The first bazel build downloads the pinned vcpkg commit, installs everything listed in vcpkg.json, and caches the result inside Bazel's external workspaces—no standalone dependency bootstrap is required.

# Build the CLI
bazel build //src:tarracsh

# Execute the GoogleTest suite
bazel test //test:tarracsh_tests

# Build the cfdiff helper
bazel build //src/interfaces/cfdiff:cfdiff

# Collect coverage (produces bazel-out/_coverage/_coverage_report.dat)
bazel coverage //test:tarracsh_tests \
  --combined_report=lcov \
  --instrumentation_filter=//src/...,//test/...

# Optional: turn the LCOV into HTML (requires lcov/genhtml)
genhtml bazel-out/_coverage/_coverage_report.dat --output-directory coverage/html

The legacy ./build.sh wrapper now just calls bazel build for compatibility. If Bazel is not installed on your machine, install bazelisk or the standard Bazel binaries first.

Usage Run tarracsh [options].

Parse tarracsh parse --input [--print-class-parse] [--print-cpool] [--cpool-filter type] [--html-nav] Parses the target and optionally prints or renders constant‑pool information.

Public Digest tarracsh public-digest --input [--diff] [--server|--port p --host h] Generates or compares digest databases, optionally communicating with a gRPC server.

Call Graph tarracsh call-graph --input Builds a call‑graph database. Server and client modes are placeholders and currently emit a “not implemented” message.

Global flags like --output-dir, --output-log-file, --print-profiler, and --use-file-timestamp apply to all commands. Use --version or --help-all for version info and comprehensive help.

Commands that support the --query option accept a compact SQL-like language defined in KapaQuery.g4.

• Statements:
  • schema prints the database tables with their columns.
  • list <table> or select <table> optionally followed by a where clause returns rows from the given table.
• Filters combine a column name, an operator, and a quoted string literal. Supported operators are:
  • = and != for equality / inequality comparisons.
  • ^* (not ^*) for starts-with (and its negation).
  • *^ (not *^) for ends-with (and its negation).
  • *^* for regular-expression matches.
• String literals can use single ('value') or double ("value") straight quotes; double the quote character to embed it inside the literal. Smart quotes copied from rich text, such as “value” or “value″, are also accepted.
• Logical composition uses parentheses together with and / or, e.g. (method ^* "com/example" and (owner = 'ClassName' or owner = 'OtherClass')).

Combine --query with --display-raw to print unformatted column values when needed.

Project Status Call‑graph server/client functionality is marked TODO and not yet implemented.

https://en.wikipedia.org/wiki/Siegbert_Tarrasch