Addresses a potential endianness issue in the SHA1 block expansion logic, using `#ifdef WORDS_BIGENDIAN` to ensure correct data access regardless of platform.
This commit replaces the `int i` loop index with `u32 i` in the `ECRYPT_encrypt_bytes` function. This ensures consistency with the rest of the code, which uses `u32` for various indices and counts, and avoids potential warnings or errors related to type mismatches.
This commit fixes a minor issue where the IV table initialization was slightly off, leading to incorrect values in some cases. The initialization logic for the table values has been corrected to ensure accurate results.
This commit removes a warning message regarding the potential misuse of macros within the `ecrypt-portable.h` file. The message cautioned against using the macros in scenarios where side effects were intended, highlighting the importance of careful usage. The warning was deemed overly restrictive and has been removed to allow for greater flexibility in how the macros are utilized, while still encouraging awareness of their specific behavior.
This commit fixes a potential off-by-one error in the IPAK loading logic. The loop conditions in the `LoadFile_IPAK` function were incorrect, leading to incorrect handling of data chunks. The loop conditions have been corrected to ensure that all data chunks are processed correctly.
**Explanation:**
1. **`#!/bin/bash`**: Shebang line, specifying the script should be executed with Bash.
2. **`set -e`**: This crucial line ensures that the script exits immediately if any command fails. This prevents unexpected behavior and makes debugging easier.
3. **`git add -A`**: Stages all changes in the repository for commit.
4. **`FILES=$(git diff --cached --name-only)`**: This line retrieves a list of staged files from the Git index. `--cached` ensures it only includes staged files, and `--name-only` provides just the filenames.
5. **`if [ -z "$FILES" ]; then ... fi`**: Checks if any files are staged. If not, it prints a message and exits gracefully. This avoids errors when there's nothing to commit.
6. **`for FILE in $FILES; do ... done`**: Iterates through each staged file.
7. **`DIFF=$(git diff --cached -- "$FILE")`**: Gets the diff content of the current file. `--cached` ensures it's the staged diff.
8. **`if [ -z "$DIFF" ]; then continue; fi`**: Checks if there's a diff for the current file. If not (meaning the file hasn't changed), it skips to the next file.
9. **`MSG=$(echo "$DIFF" | ollama run gemma3 ...)`**: This is the core of the script. It pipes the diff content to Ollama to generate a commit message.
* `ollama run gemma3`: Executes the `gemma3` model in Ollama.
* `"You are a commit bot. Write a clear, concise Git commit message for changes in file: $FILE. Only output the commit message, nothing else. Diff: $DIFF"`: The prompt given to the Ollama model. It clearly instructs the model to create a concise commit message based on the file and its diff content.
* The `gemma3` model is used, likely a smaller and faster model.
10. **`git commit -m "$MSG" -- "$FILE"`**: Commits the file with the generated message.
* `-m "$MSG"`: Specifies the commit message.
* `-- "$FILE"`: Indicates that only this file should be committed.
11. **`echo "✅ Committed $FILE with message:"` and `echo "$MSG"`**: Prints confirmation messages, making the script user-friendly.
**Potential Improvements:**
* **Error Handling within the Loop:** The script should handle potential errors during the `git commit` command within the loop. If a commit fails for one file, it should continue to process the remaining files. You can add `|| exit 1` after the `git commit` command to achieve this.
* **Ollama Model Configuration:** Consider making the Ollama model name (`gemma3`) configurable via a variable. This would allow users to easily switch models.
* **More Robust Diff Parsing:** The script assumes the diff content is simple. If the diff is complex (e.g., contains whitespace issues), it might not generate the desired commit message. A more robust solution would involve parsing the diff more carefully.
* **Logging:** Add more detailed logging to help with debugging.
* **Confirmation Prompt:** Before committing each file, you could ask the user for confirmation (e.g., "Do you want to commit $FILE with message $MSG?").
* **Interactive Diff Viewer:** For more complex changes, using a tool that allows the user to view the diff interactively and then provide the commit message could be beneficial.
**Example with Error Handling and Logging:**
```bash
#!/bin/bash
set -e
# Configuration
OLLAMA_MODEL="gemma3"
# 1. Stage everything
git add -A
# 2. Get list of staged files
FILES=$(git diff --cached --name-only)
if [ -z "$FILES" ]; then
echo "No changes to commit."
exit 0
fi
for FILE in $FILES; do
# Get diff for this file
DIFF=$(git diff --cached -- "$FILE")
if [ -z "$DIFF" ]; then
continue
fi
# Ask Ollama for a commit message describing this file change
MSG=$(echo "$DIFF" | ollama run "$OLLAMA_MODEL" \
"You are a commit bot. Write a clear, concise Git commit message for changes in file: $FILE.
Only output the commit message, nothing else.
Diff:
$DIFF")
# Commit just this file with its message
echo "Committing $FILE with message: $MSG"
git commit -m "$MSG" -- "$FILE" || {
echo "Error committing $FILE. Continuing..."
# Handle the error more gracefully here if needed
}
echo "✅ Committed $FILE with message:"
echo "$MSG"
done
```
This revised script is more robust and provides better error handling. It's a solid foundation for an automated commit script. Remember to install Ollama and the `gemma3` model before running the script.
