AI-Studio/app/MindWork AI Studio/Assistants/ERI/ERIServerFiles.cs

using System.Text;

using Markdig.Extensions.Mathematics;
using Markdig.Syntax;

namespace AIStudio.Assistants.ERI;

/// <summary>
/// Reads the files of an ERI server out of the answers of the model and writes them to disk.
/// </summary>
internal static class ERIServerFiles
{
    /// <summary>
    /// Reads the content of a file out of the answer which wrote it.
    /// </summary>
    /// <remarks>
    /// The model is asked for a heading with the path of the file, followed by a code block with its
    /// content. Only the code block counts: models write the heading as they please, e.g., with the
    /// path in backticks, with a "File:" in front, or not at all. Up to v26.10.1, a regular expression
    /// insisted on a heading without a hyphen, which left out files such as docker-compose.yml, and it
    /// read up to the last fence of the answer, which merged a usage example into the file.
    ///
    /// Markdig reads the answer the way the chat shows it. So we take the first complete code block,
    /// and a file which holds code blocks itself stays whole when the model fences it with more
    /// backticks. A block the model never closed is left out: the answer broke off, and half a file
    /// is nothing to write to disk.
    /// </remarks>
    /// <param name="answer">The answer of the model to the request for one file.</param>
    /// <returns>The content of the file, ending with a line break, or an empty string when the
    /// answer holds no complete code block.</returns>
    public static string ReadFileContent(string answer)
    {
        if (string.IsNullOrWhiteSpace(answer))
            return string.Empty;

        //
        // Markdig reads a formula between $$ as a MathBlock, which is a fenced code block as well:
        //
        var document = Markdig.Markdown.Parse(answer, Markdown.SAFE_MARKDOWN_PIPELINE);
        var codeBlock = document.Descendants<FencedCodeBlock>().FirstOrDefault(block => block is not MathBlock && block.ClosingFencedCharCount > 0);
        if (codeBlock is null)
            return string.Empty;

        var content = codeBlock.Lines.ToString();
        return content.Length is 0 ? string.Empty : $"{content}\n";
    }

    /// <summary>
    /// The path to which a file of the model is written.
    /// </summary>
    /// <param name="baseDirectory">The base directory of the ERI server.</param>
    /// <param name="file">The path of the file, as the model named it.</param>
    /// <returns>The path of the file below the base directory.</returns>
    public static string PathOf(string baseDirectory, string file) => Path.Join(baseDirectory, file);

    /// <summary>
    /// Writes a file of the model below the base directory.
    /// </summary>
    /// <remarks>
    /// A file which cannot be written ends nothing but its own attempt: the caller tells the user,
    /// and the server gets its other files. Up to v26.10.1, an error while writing ended the whole
    /// run, and the files written so far were never recorded for their deletion on the next run.
    /// </remarks>
    /// <param name="baseDirectory">The base directory of the ERI server.</param>
    /// <param name="file">The path of the file, as the model named it.</param>
    /// <param name="answer">The answer of the model to the request for this file.</param>
    /// <param name="logger">The logger which records the details.</param>
    /// <returns>What became of the file.</returns>
    public static async Task<ERIServerFileOutcome> WriteAsync(string baseDirectory, string file, string answer, ILogger logger)
    {
        var desiredFilePath = PathOf(baseDirectory, file);

        //
        // Security check: ensure that the desired file path is inside the base directory.
        // We cannot trust the beginning of the file path because it would be possible
        // to escape by using `..` in the file path.
        //
        if (!PathTools.IsInsideDirectory(baseDirectory, desiredFilePath))
        {
            logger.LogWarning("The file path '{FilePath}' is not inside the base directory '{BaseDirectory}'.", desiredFilePath, baseDirectory);
            return ERIServerFileOutcome.OUTSIDE_BASE_DIRECTORY;
        }

        var content = ReadFileContent(answer);
        if (string.IsNullOrWhiteSpace(content))
        {
            logger.LogWarning("The content of the file '{FilePath}' is empty or was not found.", desiredFilePath);
            return ERIServerFileOutcome.NO_COMPLETE_CODE_BLOCK;
        }

        try
        {
            // A path inside the base directory always has a directory:
            var fileDirectory = Path.GetDirectoryName(desiredFilePath)!;
            if (!Directory.Exists(fileDirectory))
            {
                Directory.CreateDirectory(fileDirectory);
                logger.LogInformation("The directory '{Directory}' was created.", fileDirectory);
            }

            await File.WriteAllTextAsync(desiredFilePath, content, Encoding.UTF8);
            logger.LogInformation("The file '{FilePath}' was created.", desiredFilePath);
            return ERIServerFileOutcome.WRITTEN;
        }
        catch (Exception e) when (e is IOException or UnauthorizedAccessException or ArgumentException or NotSupportedException)
        {
            logger.LogWarning(e, "The file '{FilePath}' could not be written.", desiredFilePath);
            return ERIServerFileOutcome.WRITE_FAILED;
        }
    }
}