# Install dependencies with Homebrew
brew install srcml universal-ctags

# Install Perl modules
cpan DBI DBD::SQLite Set::Scalar HTML::FromText

apt-get install cmake libarchive-dev libxml++2.6-dev libxml2-dev \
  libcurl4-openssl-dev libxslt1-dev libboost-all-dev libantlr-dev \
  libssl-dev libxerces-c-dev exuberant-ctags libdbi-perl \
  libjgit-java libhtml-fromtext-perl libset-scalar-perl libdbd-sqlite3-perl

git clone https://github.com/Mont9165/bfg-repo-cleaner.git --branch blobexec
cd bfg-repo-cleaner
sbt "bfg/assembly"

git clone https://github.com/Mont9165/cregit.git
cd cregit

cd tokenize/srcMLtoken
make
cd ../..

# Build slickGitLog
cd slickGitLog && sbt one-jar && cd ..

# Build persons
cd persons && sbt one-jar && cd ..

# Build remapCommits
cd remapCommits && sbt one-jar && cd ..

export CREGIT_HOME=/path/to/cregit
export BFG_JAR=/path/to/bfg-repo-cleaner/bfg/target/bfg-*.jar
export BFG_MEMO_DIR=/tmp/memo
export BFG_TOKENIZE_CMD="$CREGIT_HOME/tokenize/tokenizeSrcMl.pl \
  --srcml2token=$CREGIT_HOME/tokenize/srcMLtoken/srcml2token \
  --srcml=$(which srcml) \
  --ctags=$(which ctags)"

mkdir -p "$BFG_MEMO_DIR"

# Clone the repository you want to tokenize
git clone --mirror https://github.com/example/repo.git /tmp/repo.git

java -jar "$BFG_JAR" \
  --blob-exec '$CREGIT_HOME/tokenizeByBlobId/tokenBySha.pl=\.java$' \
  --no-blob-protection \
  /tmp/repo.git

java -jar "$BFG_JAR" \
  --blob-exec '$CREGIT_HOME/tokenizeByBlobId/tokenBySha.pl=\.[ch]$' \
  --no-blob-protection \
  /tmp/repo.git

#!/bin/bash
set -e

# Configuration
REPO_NAME="camel"
REPO_URL="https://github.com/apache/camel.git"
WORK_DIR="/tmp/cregit-work"
CREGIT_HOME=/path/to/cregit
BFG_JAR=/path/to/bfg-*.jar

# Setup
mkdir -p "$WORK_DIR"
export BFG_MEMO_DIR="$WORK_DIR/memo"
export BFG_TOKENIZE_CMD="$CREGIT_HOME/tokenize/tokenizeSrcMl.pl \
  --srcml2token=$CREGIT_HOME/tokenize/srcMLtoken/srcml2token \
  --srcml=$(which srcml) \
  --ctags=$(which ctags)"
mkdir -p "$BFG_MEMO_DIR"

# 1. Clone repository
git clone --mirror "$REPO_URL" "$WORK_DIR/$REPO_NAME.git"

# 2. Keep a copy of the original
cp -r "$WORK_DIR/$REPO_NAME.git" "$WORK_DIR/$REPO_NAME-original.git"

# 3. Tokenize with BFG + cregit
java -jar "$BFG_JAR" \
  --blob-exec '$CREGIT_HOME/tokenizeByBlobId/tokenBySha.pl=\.java$' \
  --no-blob-protection \
  "$WORK_DIR/$REPO_NAME.git"

# 4. Create history database for original repo
java -jar "$CREGIT_HOME/slickGitLog/target/scala-2.11/slickgitlog_2.11-*-one-jar.jar" \
  "$WORK_DIR/$REPO_NAME-original.db" \
  "$WORK_DIR/$REPO_NAME-original.git"

# 5. Create history database for tokenized repo
java -jar "$CREGIT_HOME/slickGitLog/target/scala-2.11/slickgitlog_2.11-*-one-jar.jar" \
  "$WORK_DIR/$REPO_NAME-cregit.db" \
  "$WORK_DIR/$REPO_NAME.git"

echo "Done! Tokenized repository is at: $WORK_DIR/$REPO_NAME.git"

export BFG_TOKENIZE_CMD="..."

# macOS
brew install srcml

# Verify installation
which srcml

# macOS
brew install universal-ctags

# Verify installation
which ctags