/// <summary>
/// Crawl, query, and parse data from a SQL database.
/// </summary>
/// <param name="dbSettings">Database settings.</param>
/// <param name="query">Query to execute.</param>
/// <returns>Parse result.</returns>
public ParseResult ParseFromQuery(DbSettings dbSettings, string query)
{
if (dbSettings == null)
{
throw new ArgumentNullException(nameof(dbSettings));
}
if (String.IsNullOrEmpty(query))
{
throw new ArgumentNullException(nameof(query));
}
ParseResult ret = new ParseResult();
ret.Sql = new ParseResult.SqlParseResult();
SqlCrawler crawler = new SqlCrawler(dbSettings, query);
CrawlResult cr = crawler.Get();
if (!cr.Success)
{
ret.Time.End = DateTime.UtcNow;
return(ret);
}
return(ProcessSourceContent(cr.DataTable));
}
static void SqlCrawler()
{
string query = Common.InputString("Query:", null, true);
if (String.IsNullOrEmpty(query))
{
return;
}
DbSettings db = new DbSettings(
(DbType)(Enum.Parse(typeof(DbType), Common.InputString("DB type:", "Mysql", false))),
Common.InputString("Hostname:", "localhost", false),
Common.InputInteger("Port:", 3306, true, false),
Common.InputString("Username:"******"root", false),
Common.InputString("Password:"******"password", false),
Common.InputString("Instance:", null, true),
Common.InputString("Database name:", "dbname", false));
SqlCrawler sc = new SqlCrawler(db, query);
CrawlResult cr = sc.Get();
Console.WriteLine("Success : " + cr.Success);
Console.WriteLine("Start time : " + cr.Time.Start.ToString());
Console.WriteLine("End time : " + cr.Time.End.ToString());
Console.WriteLine("Total ms : " + cr.Time.TotalMs.ToString() + "ms");
Console.WriteLine("Data : ");
if (cr.DataTable != null)
{
Console.WriteLine(Common.SerializeJson(Common.DataTableToListDynamic(cr.DataTable), true));
}
else
{
Console.WriteLine(" (null)");
}
}
static void SqlCrawler()
{
string query = Common.InputString("Query:", null, true);
if (String.IsNullOrEmpty(query))
{
return;
}
DbSettings db = new DbSettings(
(DbType)(Enum.Parse(typeof(DbType), Common.InputString("DB type:", "Mysql", false))),
Common.InputString("Hostname:", "localhost", false),
Common.InputInteger("Port:", 3306, true, false),
Common.InputString("Username:"******"root", false),
Common.InputString("Password:"******"password", false),
Common.InputString("Instance:", null, true),
Common.InputString("Database name:", "dbname", false));
SqlCrawler sc = new SqlCrawler(db, query);
CrawlResult cr = sc.Get();
if (_OutputType.Equals("console"))
{
EnumerateCrawlResult(cr);
}
ParseCrawlResult(cr);
}
private static async Task PostParse(RequestMetadata md)
{
string header = "[Komodo.Server] " + md.Http.Request.Source.IpAddress + ":" + md.Http.Request.Source.Port + " PostParse ";
if (String.IsNullOrEmpty(md.Params.Type))
{
_Logging.Warn(header + "no document type supplied");
md.Http.Response.StatusCode = 400;
md.Http.Response.ContentType = "application/json";
await md.Http.Response.Send(new ErrorResponse(400, "Supply 'type' [json/xml/html/sql/text] in querystring.", null, null).ToJson(true));
return;
}
byte[] data = null;
CrawlResult crawlResult = null;
ParseResult parseResult = null;
HttpCrawler httpCrawler = null;
FileCrawler fileCrawler = null;
if (!String.IsNullOrEmpty(md.Params.Url))
{
#region Crawl-URL
switch (md.Params.Type.ToLower())
{
case "html":
httpCrawler = new HttpCrawler(md.Params.Url);
crawlResult = httpCrawler.Get();
if (!crawlResult.Success)
{
_Logging.Warn(header + "failed to crawl URL " + md.Params.Url);
md.Http.Response.StatusCode = 500;
md.Http.Response.ContentType = "application/json";
await md.Http.Response.Send(new ErrorResponse(400, "Failed to crawl supplied URL.", null, crawlResult).ToJson(true));
return;
}
data = crawlResult.Data;
HtmlParser htmlParser = new HtmlParser();
parseResult = htmlParser.ParseBytes(data);
if (!parseResult.Success)
{
_Logging.Warn(header + "failed to parse data from URL " + md.Params.Url);
md.Http.Response.StatusCode = 500;
md.Http.Response.ContentType = "application/json";
await md.Http.Response.Send(new ErrorResponse(400, "Failed to parse data from supplied URL.", null, parseResult).ToJson(true));
return;
}
break;
case "json":
httpCrawler = new HttpCrawler(md.Params.Url);
crawlResult = httpCrawler.Get();
if (!crawlResult.Success)
{
_Logging.Warn(header + "failed to crawl URL " + md.Params.Url);
md.Http.Response.StatusCode = 500;
md.Http.Response.ContentType = "application/json";
await md.Http.Response.Send(new ErrorResponse(400, "Failed to crawl supplied URL.", null, crawlResult).ToJson(true));
return;
}
data = crawlResult.Data;
JsonParser jsonParser = new JsonParser();
parseResult = jsonParser.ParseBytes(data);
if (!parseResult.Success)
{
_Logging.Warn(header + "failed to parse data from URL " + md.Params.Url);
md.Http.Response.StatusCode = 500;
md.Http.Response.ContentType = "application/json";
await md.Http.Response.Send(new ErrorResponse(400, "Failed to parse data from supplied URL.", null, parseResult).ToJson(true));
return;
}
break;
case "text":
httpCrawler = new HttpCrawler(md.Params.Url);
crawlResult = httpCrawler.Get();
if (!crawlResult.Success)
{
_Logging.Warn(header + "failed to crawl URL " + md.Params.Url);
md.Http.Response.StatusCode = 500;
md.Http.Response.ContentType = "application/json";
await md.Http.Response.Send(new ErrorResponse(400, "Failed to crawl supplied URL.", null, crawlResult).ToJson(true));
return;
}
data = crawlResult.Data;
TextParser textParser = new TextParser();
parseResult = textParser.ParseBytes(data);
if (!parseResult.Success)
{
_Logging.Warn(header + "failed to parse data from URL " + md.Params.Url);
md.Http.Response.StatusCode = 500;
md.Http.Response.ContentType = "application/json";
await md.Http.Response.Send(new ErrorResponse(400, "Failed to parse data from supplied URL.", null, parseResult).ToJson(true));
return;
}
break;
case "xml":
httpCrawler = new HttpCrawler(md.Params.Url);
crawlResult = httpCrawler.Get();
if (!crawlResult.Success)
{
_Logging.Warn(header + "failed to crawl URL " + md.Params.Url);
md.Http.Response.StatusCode = 500;
md.Http.Response.ContentType = "application/json";
await md.Http.Response.Send(new ErrorResponse(400, "Failed to crawl supplied URL.", null, crawlResult).ToJson(true));
return;
}
data = crawlResult.Data;
XmlParser xmlParser = new XmlParser();
parseResult = xmlParser.ParseBytes(data);
if (!parseResult.Success)
{
_Logging.Warn(header + "failed to parse data from URL " + md.Params.Url);
md.Http.Response.StatusCode = 500;
md.Http.Response.ContentType = "application/json";
await md.Http.Response.Send(new ErrorResponse(400, "Failed to parse data from supplied URL.", null, parseResult).ToJson(true));
return;
}
break;
default:
_Logging.Warn(header + "invalid document type for processing via URL " + md.Params.Url);
md.Http.Response.StatusCode = 400;
md.Http.Response.ContentType = "application/json";
await md.Http.Response.Send(new ErrorResponse(400, "Invalid document type.", null, null).ToJson(true));
return;
}
md.Http.Response.StatusCode = 200;
md.Http.Response.ContentType = "application/json";
await md.Http.Response.Send(Common.SerializeJson(parseResult, md.Params.Pretty));
return;
#endregion
}
else if (!String.IsNullOrEmpty(md.Params.Filename))
{
#region Filename
switch (md.Params.Type.ToLower())
{
case "html":
fileCrawler = new FileCrawler(md.Params.Filename);
crawlResult = fileCrawler.Get();
if (!crawlResult.Success)
{
_Logging.Warn(header + "failed to crawl filename " + md.Params.Filename);
md.Http.Response.StatusCode = 500;
md.Http.Response.ContentType = "application/json";
await md.Http.Response.Send(new ErrorResponse(400, "Failed to crawl supplied filename.", null, crawlResult).ToJson(true));
return;
}
data = crawlResult.Data;
HtmlParser htmlParser = new HtmlParser();
parseResult = htmlParser.ParseBytes(data);
if (!parseResult.Success)
{
_Logging.Warn(header + "failed to parse data from file " + md.Params.Filename);
md.Http.Response.StatusCode = 500;
md.Http.Response.ContentType = "application/json";
await md.Http.Response.Send(new ErrorResponse(400, "Failed to parse data from supplied filename.", null, parseResult).ToJson(true));
return;
}
break;
case "json":
fileCrawler = new FileCrawler(md.Params.Filename);
crawlResult = fileCrawler.Get();
if (!crawlResult.Success)
{
_Logging.Warn(header + "failed to crawl filename " + md.Params.Filename);
md.Http.Response.StatusCode = 500;
md.Http.Response.ContentType = "application/json";
await md.Http.Response.Send(new ErrorResponse(400, "Failed to crawl supplied filename.", null, crawlResult).ToJson(true));
return;
}
data = crawlResult.Data;
JsonParser jsonParser = new JsonParser();
parseResult = jsonParser.ParseBytes(data);
if (!parseResult.Success)
{
_Logging.Warn(header + "failed to parse data from file " + md.Params.Filename);
md.Http.Response.StatusCode = 500;
md.Http.Response.ContentType = "application/json";
await md.Http.Response.Send(new ErrorResponse(400, "Failed to parse data from supplied filename.", null, parseResult).ToJson(true));
return;
}
break;
case "text":
fileCrawler = new FileCrawler(md.Params.Filename);
crawlResult = fileCrawler.Get();
if (!crawlResult.Success)
{
_Logging.Warn(header + "failed to crawl filename " + md.Params.Filename);
md.Http.Response.StatusCode = 500;
md.Http.Response.ContentType = "application/json";
await md.Http.Response.Send(new ErrorResponse(400, "Failed to crawl supplied filename.", null, crawlResult).ToJson(true));
return;
}
data = crawlResult.Data;
TextParser textParser = new TextParser();
parseResult = textParser.ParseBytes(data);
if (!parseResult.Success)
{
_Logging.Warn(header + "failed to parse data from file " + md.Params.Filename);
md.Http.Response.StatusCode = 500;
md.Http.Response.ContentType = "application/json";
await md.Http.Response.Send(new ErrorResponse(400, "Failed to parse data from supplied filename.", null, parseResult).ToJson(true));
return;
}
break;
case "xml":
fileCrawler = new FileCrawler(md.Params.Filename);
crawlResult = fileCrawler.Get();
if (!crawlResult.Success)
{
_Logging.Warn(header + "failed to crawl filename " + md.Params.Filename);
md.Http.Response.StatusCode = 500;
md.Http.Response.ContentType = "application/json";
await md.Http.Response.Send(new ErrorResponse(400, "Failed to crawl supplied filename.", null, crawlResult).ToJson(true));
return;
}
data = crawlResult.Data;
XmlParser xmlParser = new XmlParser();
parseResult = xmlParser.ParseBytes(data);
if (!parseResult.Success)
{
_Logging.Warn(header + "failed to parse data from file " + md.Params.Filename);
md.Http.Response.StatusCode = 500;
md.Http.Response.ContentType = "application/json";
await md.Http.Response.Send(new ErrorResponse(400, "Failed to parse data from supplied filename.", null, parseResult).ToJson(true));
return;
}
break;
default:
_Logging.Warn(header + "invalid document type for processing via filename " + md.Params.Filename);
md.Http.Response.StatusCode = 400;
md.Http.Response.ContentType = "application/json";
await md.Http.Response.Send(new ErrorResponse(400, "Invalid document type.", null, null).ToJson(true));
return;
}
md.Http.Response.StatusCode = 200;
md.Http.Response.ContentType = "application/json";
await md.Http.Response.Send(Common.SerializeJson(parseResult, md.Params.Pretty));
return;
#endregion
}
else if (md.Params.Type.ToLower().Equals("sql"))
{
#region Query
if (md.Http.Request.Data == null || md.Http.Request.ContentLength < 1)
{
_Logging.Warn(header + "no query found in payload");
md.Http.Response.StatusCode = 400;
md.Http.Response.ContentType = "application/json";
await md.Http.Response.Send(new ErrorResponse(400, "No SQL query in request payload.", null, null).ToJson(true));
return;
}
DbSettings dbSettings = new DbSettings(md.Params.DbType, md.Params.DbServer, md.Params.DbPort, md.Params.DbUser, md.Params.DbPass, md.Params.DbInstance, md.Params.DbName);
SqlCrawler sqlCrawler = new SqlCrawler(dbSettings, Encoding.UTF8.GetString(Common.StreamToBytes(md.Http.Request.Data)));
crawlResult = sqlCrawler.Get();
if (!crawlResult.Success)
{
_Logging.Warn(header + "failed to crawl database " + md.Params.DbName);
md.Http.Response.StatusCode = 500;
md.Http.Response.ContentType = "application/json";
await md.Http.Response.Send(new ErrorResponse(400, "Failed to crawl specified database.", null, crawlResult).ToJson(true));
return;
}
SqlParser sqlParser = new SqlParser();
parseResult = sqlParser.Parse(crawlResult.DataTable);
if (!parseResult.Success)
{
_Logging.Warn(header + "failed to parse data from database " + md.Params.DbName);
md.Http.Response.StatusCode = 500;
md.Http.Response.ContentType = "application/json";
await md.Http.Response.Send(new ErrorResponse(400, "Failed to parse data from specified database.", null, parseResult).ToJson(true));
return;
}
md.Http.Response.StatusCode = 200;
md.Http.Response.ContentType = "application/json";
await md.Http.Response.Send(Common.SerializeJson(parseResult, md.Params.Pretty));
return;
#endregion
}
else if (md.Http.Request.Data != null && md.Http.Request.ContentLength > 0)
{
#region Supplied-Data
data = Common.StreamToBytes(md.Http.Request.Data);
switch (md.Params.Type.ToLower())
{
case "html":
HtmlParser htmlParser = new HtmlParser();
parseResult = htmlParser.ParseBytes(data);
if (!parseResult.Success)
{
_Logging.Warn(header + "failed to parse data from supplied data");
md.Http.Response.StatusCode = 500;
md.Http.Response.ContentType = "application/json";
await md.Http.Response.Send(new ErrorResponse(400, "Failed to parse data from supplied data.", null, parseResult).ToJson(true));
return;
}
break;
case "json":
JsonParser jsonParser = new JsonParser();
parseResult = jsonParser.ParseBytes(data);
if (!parseResult.Success)
{
_Logging.Warn(header + "failed to parse data from supplied data");
md.Http.Response.StatusCode = 500;
md.Http.Response.ContentType = "application/json";
await md.Http.Response.Send(new ErrorResponse(400, "Failed to parse data from supplied data.", null, parseResult).ToJson(true));
return;
}
break;
case "text":
TextParser textParser = new TextParser();
parseResult = textParser.ParseBytes(data);
if (!parseResult.Success)
{
_Logging.Warn(header + "failed to parse data from supplied data");
md.Http.Response.StatusCode = 500;
md.Http.Response.ContentType = "application/json";
await md.Http.Response.Send(new ErrorResponse(400, "Failed to parse data from supplied data.", null, parseResult).ToJson(true));
return;
}
break;
case "xml":
XmlParser xmlParser = new XmlParser();
parseResult = xmlParser.ParseBytes(data);
if (!parseResult.Success)
{
_Logging.Warn(header + "failed to parse data from supplied data");
md.Http.Response.StatusCode = 500;
md.Http.Response.ContentType = "application/json";
await md.Http.Response.Send(new ErrorResponse(400, "Failed to parse data from supplied data.", null, parseResult).ToJson(true));
return;
}
break;
default:
_Logging.Warn(header + "invalid document type for processing via data");
md.Http.Response.StatusCode = 400;
md.Http.Response.ContentType = "application/json";
await md.Http.Response.Send(new ErrorResponse(400, "Invalid document type supplied.", null, null).ToJson(true));
return;
}
md.Http.Response.StatusCode = 200;
md.Http.Response.ContentType = "application/json";
await md.Http.Response.Send(Common.SerializeJson(parseResult, md.Params.Pretty));
return;
#endregion
}
else
{
#region Unknown
_Logging.Warn(header + "unable to derive data source from request");
md.Http.Response.StatusCode = 400;
md.Http.Response.ContentType = "application/json";
await md.Http.Response.Send(new ErrorResponse(400, "Unable to derive data source from request.", null, null).ToJson(true));
return;
#endregion
}
}
