public string Save(RequestResponse log, FolderRelativePath relativeFolder = null, FileName fileName = null, int howManyFilesToKeep = 0)
{
var finalFolder = baseDirectory.AppendPath(relativeFolder);
if (!Directory.Exists(finalFolder))
{
Directory.CreateDirectory(finalFolder);
}
if (fileName == null)
{
fileName = log.Response.Status.ToString();
}
var finalFileName = GetFinalFileName(finalFolder, fileName, howManyFilesToKeep);
if (finalFileName == null)
{
return(null); // limit reached
}
SaveToFile(Path.Combine(finalFolder, finalFileName), log);
return(finalFileName);
}
public void SaveToFile(string fullFileName, RequestResponse log)
{
_fileSystem.CreateFile(fullFileName, RecordingFormatter.Format(log));
}
/// <summary>
///
/// </summary>
/// <param name="log"></param>
/// <returns>created file name</returns>
public string Write(RequestResponse log)
{
Interlocked.Increment(ref _callsCounter);
var content = new StringBuilder();
// details about request
content.AppendLine($"METADATA");
content.AppendLine($"Observations: !! EXPLAIN WHY THIS EXAMPLE IS IMPORTANT HERE !!");
content.AppendLine($"Date: {log.Metadata.DateTime.ToString("yyyy-MM-dd HH:mm:ss.fff")} {log.Metadata.Timezone}");
content.AppendLine($"Recorded from: {log.Metadata.RecordedFrom}");
content.AppendLine($"Local machine: {log.Metadata.LocalMachine}");
content.AppendLine($"User: {log.Metadata.User}");
content.AppendLine($"Using tool: {log.Metadata.ToolNameAndVersion} ({log.Metadata.ToolUrl})");
content.AppendLine();
content.AppendLine($"REQUEST");
content.AppendLine($"{log.Request.Method.ToString().ToLower()} {log.Request.Url}");
AddHeadersAndBody(log.Request);
// lots of thought went into the bellow lines, because a separator needed to be created to split the response that
// the method ResponseFactory.FromFiddlerLikeResponseFile can read, from (optional) comments
// it has been decided that a new line + "--!?@Divider: some text here" + new line would be an acceptable separator
// as we want to keep files in human readable/editable format
// we still have the risk (very low) that such string could be found in real use case, it seemed like an acceptable
// risk/downside comparing to the upsides
// we could have put the metadata + request details in a different file, but that would risk the 2 files from getting
// separated, which defeats the purpose of a high quality documentation file of what request lead to what response
content.AppendLine();
content.AppendLine("--!?@Divider: Any text BEFORE this line = comments, AFTER = response in Fiddler like format");
content.AppendLine();
// details about response
content.AppendLine($"HTTP/{log.Response.HttpVersion} {(int)log.Response.Status} {log.Response.Status}");
AddHeadersAndBody(log.Response);
var filename = $"{_callsCounter}_{log.Response.Status}";
_fileSystem.CreateNewTextFile(folder, filename, content.ToString().Trim());
return(filename);
void AddHeadersAndBody(RequestResponseInfo requestOrResponse)
{
var contentType = Helper.GetKnownContentType(requestOrResponse.Headers.GetValueOrDefault("Content-Type"));
bool IsResponseXml = requestOrResponse is ResponseInfo && contentType == Helper.KnownContentTypes.Xml;
foreach (var header in requestOrResponse.Headers.ToList())
{
// we skip Content-Length for XML responses (liked WCF), because for some reason when trying to read XML responses (that have been changed by formatting)
// the content-length being smaller seems to throw WCF readers out and cause exceptions
if (header.Key.ToLower() == "content-length" && IsResponseXml)
{
continue;
}
content.AppendLine($"{header.Key}:{header.Value}");
}
content.AppendLine();
if (requestOrResponse.Body == null)
{
content.AppendLine(); // a NoContent (204) could return this, we mark an empty file
return;
}
switch (contentType)
{
case Helper.KnownContentTypes.Json:
content.AppendLine(requestOrResponse.Body.FormatJson());
break;
case Helper.KnownContentTypes.Xml:
content.AppendLine(requestOrResponse.Body.FormatXml());
break;
case Helper.KnownContentTypes.Other:
content.AppendLine(requestOrResponse.Body);
break;
}
}
}
