#!/bin/bash

# Create accurate tools database for REMnux-based container
# This script only includes tools that actually exist in the container
# with their correct command names and practical usage examples

set -e

DOCS_DIR="/opt/remnux-docs"
TOOLS_DB="/opt/remnux-docs/tools.db"

echo "🔧 Creating accurate tools database..."

# Create documentation directory
mkdir -p "$DOCS_DIR"

# Create the tools database with verified tools
cat > "$TOOLS_DB" << 'EOF'
# REMnux Enhanced Tools Database
# Format: TOOL_NAME|DESCRIPTION|CATEGORY|USAGE_EXAMPLE|AVAILABLE
# Only includes tools verified to be available in container

# PDF Analysis Tools
pdf-parser.py|Parse and analyze PDF structure and objects|analyze documents pdf|pdf-parser.py suspicious.pdf|✓
pdfid.py|Identify suspicious elements in PDF files|analyze documents pdf|pdfid.py document.pdf|✓
peepdf|Interactive PDF analysis framework with JavaScript detection|analyze documents pdf|peepdf -i malicious.pdf|✓
pdftk|Manipulate PDF files (merge, split, encrypt)|analyze documents pdf|pdftk input.pdf output output.pdf flatten|✓
qpdf|PDF manipulation tool (merge, convert, transform)|analyze documents pdf|qpdf --decrypt encrypted.pdf output.pdf|✓
pdfresurrect|Extract previous versions from PDF files|analyze documents pdf|pdfresurrect document.pdf|✓
pdftool.py|Analyze PDF incremental updates|analyze documents pdf|pdftool.py document.pdf|✓

# Malware Analysis Tools  
capa|Detect malware capabilities using MITRE ATT&CK framework|malware analysis|capa malware.exe|✓
box-js|JavaScript sandbox for malware analysis|malware analysis|box-js --output-dir=/tmp suspicious.js|✓
oletools|Office document analysis suite (oledump.py, rtfdump.py, etc)|malware analysis|oledump.py document.doc|✓
oledump.py|Analyze OLE files (Office documents)|malware analysis|oledump.py document.xls|✓
rtfdump.py|Analyze RTF documents|malware analysis|rtfdump.py document.rtf|✓
emldump.py|Analyze EML email files|malware analysis|emldump.py message.eml|✓
base64dump.py|Extract and decode base64 strings|malware analysis|base64dump.py document.txt|✓
strings|Extract printable strings from files|malware analysis|strings -n 10 binary.exe|✓
hexdump|Display file content in hexadecimal|malware analysis|hexdump -C binary.dat|✓
xxd|Create hex dump or reverse it|malware analysis|xxd binary.exe|✓

# Data Analysis Tools
visidata|Interactive data exploration and analysis|data analysis|vd data.csv|✓
unfurl|URL analysis and extraction tool|data analysis|echo "http://example.com/path" | unfurl|✓

# Origami PDF Tools (Ruby-based)
pdfcop|Examine PDF file structure using Origami|analyze documents pdf|ruby -e "require 'origami'; puts Origami::PDF.read('file.pdf').info"|⚠
pdfextract|Extract objects from PDF using Origami|analyze documents pdf|pdfextract document.pdf|⚠
pdfmetadata|Extract PDF metadata using Origami|analyze documents pdf|pdfmetadata document.pdf|⚠

# File Format Tools
exiftool|Extract metadata from files|general utilities|exiftool document.pdf|✓
file|Determine file type|general utilities|file suspicious.dat|✓
binwalk|Analyze and extract firmware images|malware analysis|binwalk firmware.bin|✓
foremost|File carving tool|forensics|foremost -t pdf,jpg -i disk.img|✓

# Text Processing
catdoc|Convert MS Word documents to text|view or edit files|catdoc document.doc|✓
docx2txt|Extract text from DOCX files|view or edit files|docx2txt document.docx|✓
unrtf|Convert RTF to other formats|view or edit files|unrtf --html document.rtf|✓

# Archive Tools
7z|7-Zip archiver with high compression|general utilities|7z x archive.7z|✓
unzip|Extract ZIP archives|general utilities|unzip archive.zip|✓
tar|Archive files|general utilities|tar -xzf archive.tar.gz|✓

# Network Analysis  
tshark|Network protocol analyzer (Wireshark CLI)|network analysis|tshark -r capture.pcap|✓
tcpdump|Command-line packet analyzer|network analysis|tcpdump -r capture.pcap|✓

# System Tools
mc|Midnight Commander file manager|general utilities|mc|✓
htop|Interactive process viewer|general utilities|htop|✓
tree|Display directory structure as tree|general utilities|tree /path/to/directory|✓

# Image Processing
convert|ImageMagick convert tool|general utilities|convert input.pdf output.png|✓
identify|ImageMagick identify tool|general utilities|identify image.jpg|✓

# OCR Tools
tesseract|OCR text extraction|general utilities|tesseract image.png output|✓

# Container-Added Tools
curl|Transfer data from servers|network analysis|curl -L http://example.com|✓
wget|Download files from web|network analysis|wget http://example.com/file.pdf|✓
jq|JSON processor|data analysis|cat data.json | jq '.field'|✓
yq|YAML processor|data analysis|cat data.yaml | yq '.field'|✓

# Text Editors
vim|Vi/Vim text editor|view or edit files|vim filename.txt|✓
nano|Simple text editor|view or edit files|nano filename.txt|✓

# Python Tools Available
python3|Python 3 interpreter|general utilities|python3 script.py|✓
pip3|Python package installer|general utilities|pip3 install package|✓

EOF

echo "✅ Accurate tools database created at: $TOOLS_DB"

# Count tools
TOOL_COUNT=$(grep -c "^[^#]" "$TOOLS_DB" 2>/dev/null || echo "0")
echo "🔧 Indexed $TOOL_COUNT verified tools"

echo ""
echo "📝 Notes:"
echo "  ✓ = Tool verified available"
echo "  ⚠ = Tool may need different invocation or setup"
echo ""
echo "Usage: find-tool [search_term] to search through available tools"