Bash script to check on keywords in codebase per month across time range and exports to csv

script.sh

#!/bin/bash
# Make sure its executable with chmod +x script.sh
# Ideally run in an isolated env like a codespace
# Make sure no files already known to git are modified
# --- Configuration ---
REPO_PATH="./" # Path to your Git repository. ./ is current directory.
KEYWORDS=("Checkbox" "AvatarDropdown" "Button") # Array of keywords
PACKAGE_NAME="antd" # Package name to filter imports
START_DATE="2021-07-01" # YYYY-MM-DD
END_DATE="2023-07-31"   # YYYY-MM-DD
OUTPUT_CSV_FILE="keyword_analysis.csv" #optionally add path to save the output CSV file before the filename
# ---------------------

echo "Starting Git keyword analysis for repository: $REPO_PATH"
echo "Keywords: ${KEYWORDS[*]}"
echo "Timespan: $START_DATE to $END_DATE"

# --- Validate repository path ---
if [ ! -d "$REPO_PATH" ]; then
    echo "Error: Repository path '$REPO_PATH' does not exist." >&2
    exit 1
fi
if [ ! -d "$REPO_PATH/.git" ]; then
    echo "Error: '$REPO_PATH' is not a Git repository." >&2
    exit 1
fi

# Store the current working directory to return to later
ORIGINAL_CWD=$(pwd)

# Change to the repository directory
cd "$REPO_PATH" || { echo "Error: Could not change directory to $REPO_PATH. Exiting." >&2; exit 1; }

# --- Get current git state before modification ---
ORIGINAL_BRANCH=$(git rev-parse --abbrev-ref HEAD 2>/dev/null)
ORIGINAL_HEAD=$(git rev-parse HEAD 2>/dev/null)
if [ $? -ne 0 ]; then
    echo "Could not get original Git state. Exiting." >&2
    exit 1
fi


# --- Prepare CSV header ---
MONTHS=()
CURRENT_DATE="$START_DATE"
END_MONTH_DATE=$(date -d "$END_DATE" +%Y-%m)

while true; do
    MONTH_YEAR_STR=$(date -d "$CURRENT_DATE" +%Y-%m)
    MONTHS+=("$MONTH_YEAR_STR")
    if [[ "$MONTH_YEAR_STR" == "$END_MONTH_DATE" ]]; then
        break
    fi
    # Move to the next month
    CURRENT_DATE=$(date -d "$CURRENT_DATE +1 month" +%Y-%m-01)
done


# Write CSV header
echo -n "Keyword" > "$OUTPUT_CSV_FILE"
for month in "${MONTHS[@]}"; do
    echo -n ",$month" >> "$OUTPUT_CSV_FILE"
done
echo "" >> "$OUTPUT_CSV_FILE"

# --- Prepare array of monthly commits ---
MONTHLY_COMMITS=()
for month_str in "${MONTHS[@]}"; do
    END_OF_MONTH_DATETIME=$(date -d "$month_str-01 +1 month -1 second" -Iseconds)
    # Find the last commit in or before this month
    LAST_COMMIT_HASH=$(git log --pretty=format:%H --before="$END_OF_MONTH_DATETIME" --max-count=1 2>/dev/null)
    MONTHLY_COMMITS+=("$LAST_COMMIT_HASH")
    if [ -z "$LAST_COMMIT_HASH" ]; then
        echo "No commit found for $month_str (up to $END_OF_MONTH_DATETIME)."
    else
        echo "Commit for $month_str: $LAST_COMMIT_HASH"
    fi
    # This array will have empty string for months with no commit
    # The index matches the MONTHS array
    # This allows us to later iterate and know which commit to use for each month
    # and which months to skip
done

# done # End of keyword loop
# --- Main analysis loop ---
for keyword in "${KEYWORDS[@]}"; do

    echo "Processing React component import: '$keyword'"
    ROW_DATA=("$keyword") # Start with the keyword for this row

    for i in "${!MONTHS[@]}"; do
        month_str="${MONTHS[$i]}"
        commit_hash="${MONTHLY_COMMITS[$i]}"
        echo "  Analyzing month: $month_str"
        if [ -z "$commit_hash" ]; then
            echo "    No commit found for $month_str. Count for '$keyword' will be 0."
            ROW_DATA+=(0)
            continue
        fi
        echo "    Checking out commit: $commit_hash..."
        if ! git checkout -q "$commit_hash" &>/dev/null; then
            echo "      Error checking out commit $commit_hash. Skipping." >&2
            ROW_DATA+=(0)
            continue
        fi
        echo "    Counting import of '$keyword' from package '$PACKAGE_NAME'..."
        # Only count if the keyword is imported from the specific PACKAGE_NAME
        IMPORT_COUNT=0
        # Named import: import { ...keyword... } from 'PACKAGE_NAME';
        NAMED_IMPORTS=$(git grep -h -I -E "^import[[:space:]]+\{[^}]*\b$keyword\b[^}]*\}[[:space:]]+from[[:space:]]+[\'\"]$PACKAGE_NAME[\'\"]" -- '*.js' '*.jsx' '*.ts' '*.tsx' 2>/dev/null | wc -l)
        # Default import: import Keyword from 'PACKAGE_NAME';
        DEFAULT_IMPORTS=$(git grep -h -I -E "^import[[:space:]]+$keyword([[:space:]]+|,|$)from[[:space:]]+[\'\"]$PACKAGE_NAME[\'\"]" -- '*.js' '*.jsx' '*.ts' '*.tsx' 2>/dev/null | wc -l)
        # Aliased named import: import { ... as ... } from 'PACKAGE_NAME';
        ALIAS_IMPORTS=$(git grep -h -I -E "^import[[:space:]]+\{[^}]*\b$keyword[[:space:]]+as[[:space:]]+[^}]+\}[[:space:]]+from[[:space:]]+[\'\"]$PACKAGE_NAME[\'\"]" -- '*.js' '*.jsx' '*.ts' '*.tsx' 2>/dev/null | wc -l)
        IMPORT_COUNT=$((NAMED_IMPORTS + DEFAULT_IMPORTS + ALIAS_IMPORTS))
        echo "      Import count for '$keyword' from '$PACKAGE_NAME': $IMPORT_COUNT"
        ROW_DATA+=("$IMPORT_COUNT")
    done # End of month loop

    # Append keyword's row data to CSV
    IFS=',' # Set Internal Field Separator for echo
    echo "${ROW_DATA[*]}" >> "$OUTPUT_CSV_FILE"
    unset IFS # Unset it to avoid unexpected behavior later
done # End of keyword loop

# --- Revert to original git state ---
echo "Reverting to original Git state..."
# First, try to go back to the original branch
if ! git checkout -q "$ORIGINAL_BRANCH" &>/dev/null; then
    echo "Warning: Could not checkout original branch '$ORIGINAL_BRANCH'. Attempting to checkout original HEAD." >&2
    # If branch checkout fails, go back to the exact commit hash
    if ! git checkout -q "$ORIGINAL_HEAD" &>/dev/null; then
        echo "Error: Could not revert to original Git state. Please manually checkout '$ORIGINAL_HEAD' or your desired branch." >&2
    else
        echo "Successfully reverted to original HEAD: $ORIGINAL_HEAD"
    fi
else
    echo "Successfully reverted to original branch: $ORIGINAL_BRANCH"
fi

# Change back to the original working directory
cd "$ORIGINAL_CWD" || { echo "Error: Could not change back to original directory $ORIGINAL_CWD." >&2; exit 1; }


echo -e "\nAnalysis complete. Results saved to '$OUTPUT_CSV_FILE'"