'T2CSV Converter - by Steven Bowdenis licensed under a Creative Commons Attribution 4.0 International License. Based on a work at tschema.org/T2CSV/index.html

'This code is just string-play, nothing fancy so anyone with a basic understanding of programming will get it and see 
'that there is no voodoo in data extraction and transformation, it's anyones game, I'm even using retro LABELS for the lulz!
'Performance-wise it is pretty impressive and overloads are unexpected, if you get an overload just modify the code to write to the .csv more regularly

Imports System.IO
Imports System.Text
Imports System.Collections
Imports Microsoft.VisualBasic




Module Module1

    Sub Main()
RESET_APP:
        'Menu variables
        Console.Title = "T2CSV Converter - by Steve Bowden"
        Dim menu_select As ConsoleKeyInfo
        Dim Op1 As String = "D1"
        Dim Op2 As String = "D2"
        Dim Op3 As String = "D3"
        Dim Op4 As String = "D4"
        Dim Op5 As String = "D5"
        Dim menu_choice As String
        'file checking variables
        Dim default_path As String = "C:\T2CSV\"
        Dim path As String = default_path
        Dim proposed_path As String = ""
        Try
            If (Not System.IO.Directory.Exists(path)) Then 'create directory
                System.IO.Directory.CreateDirectory(path)
            End If
        Catch ex As Exception
            Console.WriteLine("Well well, looks there is a permissions issue preventing directory creation")
            Console.WriteLine("You will need to set a directory using Option 2 in the main menue, press any key...")
            Console.ReadKey()
            path = "unspecified"
        End Try

        Dim line As String

        Dim discrete_path As String = ""
        Dim filename As String
        'setting up main table
        Dim maintable() As String = {"<t_created>", "<t_modified>", "<t_uuid>", "<t_name>", "<t_id_assigned>", "<t_data>"}
        Dim maintable_explicit() As String = {"<xs:element_name=""t_created"" type=""xs:dateTime"">", "<xs:element_name=""t_modified"" type=""xs:dateTime"">", "<xs:element_name=""t_uuid"" type=""xs:string"">", "<xs:element_name=""t_name"" type=""xs:string"">", "<xs:element_name=""t_id_assigned"" type=""xs:string"">", "<xs:element_name=""t_data"" type=""xs:string"">"}
        Dim uuid_tag As String = "<t_uuid>"
        Dim tag_tag As String = "<t_tag>"
        Dim uuid_tag_exp As String = "<xs:element_name=""t_uuid"" type=""xs:string"">"
        Dim tag_tag_exp As String = "<xs:element_name=""t_tag"" type=""xs:string"">"
        Dim lib_tag As String = "<t_tags_uuid>"
        Dim lib_tag_exp As String = "<xs:element_name=""t_uuid"" type=""xs:string"">"
        'extractor parameters
        Dim t_loc As Integer
        Dim t_loc_end As Integer
        Dim t_length As Integer
        Dim t_cont_loc As Integer
        Dim t_cont_length As Integer
        Dim t_contents As String
        Dim csv_line As String = ""
        Dim this_t_uuid As String = ""
        'remove comments vars
        Dim cleanse_pos As Integer
        Dim cleanse_pos_end As Integer
        Dim anchor_outcom As Integer
        Dim rebuild_line As String
        Dim line_content As String
        'n table
        Dim tag_csv_line As String
        Dim lib_csv_line As String

        

MAIN_MENU:
        If (Not System.IO.Directory.Exists(path)) Then 'create directory
            System.IO.Directory.CreateDirectory(path)
        End If
        Dim di As New IO.DirectoryInfo(path)
        Dim aryFi As IO.FileInfo()
        Dim fi As IO.FileInfo
        Console.Clear()
        Console.WriteLine("T2CSV")
        Console.WriteLine("--MAIN MENU--")
        Console.WriteLine("1. Convert files")
        Console.WriteLine("2. Set filepath")
        Console.WriteLine("3. How to")
        Console.WriteLine("4. About")
        Console.WriteLine("5. Exit")
        menu_select = Console.ReadKey()
        menu_choice = menu_select.Key.ToString

        If menu_choice = Op1 Then
            GoTo START_CONVERSION

        End If
        If menu_choice = Op2 Then
            GoTo SET_FILEPATH
        End If
        If menu_choice = Op3 Then
            GoTo GUIDE
        End If
        If menu_choice = Op4 Then
            GoTo ABOUT
        End If
        If menu_choice = Op5 Then
            Environment.Exit(True)
        End If
        GoTo MAIN_MENU




START_CONVERSION:
        Console.Clear()
        aryFi = di.GetFiles("*.t")
        If aryFi.Count < 1 Then 'Check for any files in the directory, issues would arise if the user hasn't put any files in or the directory is bogus
            Console.WriteLine("Whoops, there doesn't seem to be any files here!")
            Console.WriteLine()
            Console.WriteLine("Lets reset the directory and start over, press any key...")
            Console.ReadKey()
            GoTo RESET_APP 'pretty much starts the app over from scratch, clean slate
        End If
        For Each fi In aryFi
            Console.WriteLine("Filename: {0}", fi.Name) 'list out .t files in directory
        Next
        Console.WriteLine()
        Console.WriteLine("Are you cool with these files? (Y/N)")
        menu_select = Console.ReadKey()
        menu_choice = menu_select.Key.ToString
        If menu_choice = "Y" Then

            GoTo DO_CONVERSION

        End If
        If menu_choice = "N" Then
            GoTo MAIN_MENU
        End If
        GoTo START_CONVERSION

DO_CONVERSION:
        Console.WriteLine()
        Console.WriteLine("converting...") 'tacky console candy

        File.WriteAllText(path + "_maintable.csv", "t_created,t_modified,t_uuid,t_name,t_id_assigned,t_data") 'add column headings
        File.AppendAllText(path + "_maintable.csv", vbCrLf)
        File.WriteAllText(path + "_tagtable.csv", "t_uuid, t_tag")
        File.AppendAllText(path + "_tagtable.csv", vbCrLf)
        File.WriteAllText(path + "_libtable.csv", "t_uuid, t_tags_uuid")
        File.AppendAllText(path + "_libtable.csv", vbCrLf)

        For Each fi In aryFi  'progressing through each t-file

            csv_line = ""  'predefining the csv row to prevent data leak
            filename = fi.Name
            discrete_path = path + filename 'build the file extension
            line = File.ReadAllText(discrete_path) 'read data from file extension

            'cleansing carriage returns
            line = Replace(line, vbCrLf, "; ")

            'cleansing comments
            cleanse_pos = 0
            line_content = ""
            rebuild_line = ""
            anchor_outcom = 0
            While line.IndexOf("<!--", anchor_outcom) <> -1 'cycle through catching all the instances of comments
                cleanse_pos = line.IndexOf("<!--", anchor_outcom)
                If line.IndexOf("-->", cleanse_pos) = -1 Then
                    GoTo ESCAPE 'if there is a start comment tag but no end commment tag move along and ignore this method
                Else
                    cleanse_pos_end = line.IndexOf("-->", cleanse_pos)
                End If
                line_content = line.Substring(anchor_outcom, cleanse_pos - anchor_outcom)
                rebuild_line = rebuild_line + line_content 'rebulding the string around the comments
                anchor_outcom = cleanse_pos_end + Len("-->")

            End While
            If Len(rebuild_line) > 0 Then
                rebuild_line = rebuild_line + line.Substring(anchor_outcom, Len(line) - anchor_outcom) 'finalise stitching of string back together
                line = rebuild_line
            End If
ESCAPE:     'get out of the while loop for comments because comments closing tags could not be found

            If line.IndexOf("<xs:schema") = 0 Then 'catch an explicitately defined t-file (i.e. using the schema as a data storage
                For Each tag_explicit As String In maintable_explicit 'cycle through the maintable tags in an explicatelely defined t-file
                    t_loc = line.IndexOf(tag_explicit) 'find starting point of the current tag
                    t_length = Len(tag_explicit) 'grab the length in characters of the current tag
                    Try
                        t_loc_end = line.IndexOf("</xs:element>", t_loc)
                    Catch outOfRange As ArgumentOutOfRangeException
                        Console.WriteLine("Element ignored: {0}", outOfRange.Message) 'handle the situation when the tag is not there
                    End Try
                    t_cont_loc = t_loc + t_length 'find the starting point of the contents of the current tag
                    t_cont_length = t_loc_end - t_cont_loc 'find the length (in char) of the contents of the current tag
                    t_contents = line.Substring(t_cont_loc, t_cont_length) 'extract the contents of the current tag
                    If tag_explicit = uuid_tag_exp Then
                        this_t_uuid = t_contents
                    End If
                    If t_contents.IndexOf(",") <> -1 Then
                        t_contents = """" + t_contents + """" 'handle commas in csv data'
                    End If
                    If Len(csv_line) = 0 Then
                        csv_line = t_contents 'prevent the csv from starting with a comma
                    Else
                        csv_line = csv_line + "," + t_contents 'add a comma and the current tag contents to the csv row.
                    End If

                Next

                t_length = Len(tag_tag_exp)
                tag_csv_line = ""
                t_loc_end = 0
                While line.IndexOf(tag_tag_exp, t_loc_end) <> -1 'cycle through catching all the instances of t_tag
                    t_loc = line.IndexOf(tag_tag_exp, t_loc_end)
                    t_loc_end = line.IndexOf("</xs:element>", t_loc)
                    t_cont_loc = t_loc + t_length
                    t_cont_length = t_loc_end - t_cont_loc 'pick out the location and length of the data
                    t_contents = line.Substring(t_cont_loc, t_cont_length)
                    If t_contents.IndexOf(",") <> -1 Then
                        t_contents = """" + t_contents + """" 'handle commas in csv data'
                    End If
                    tag_csv_line = tag_csv_line + this_t_uuid + "," + t_contents + vbCrLf 'build csv rows over and over until all tags found
                End While

                t_length = Len(lib_tag_exp)
                lib_csv_line = ""
                t_loc_end = 0
                While line.IndexOf(lib_tag_exp, t_loc_end) <> -1 'cycle through catching all the instances of t_tag
                    t_loc = line.IndexOf(lib_tag_exp, t_loc_end)
                    t_loc_end = line.IndexOf("</xs:element>", t_loc)
                    t_cont_loc = t_loc + t_length
                    t_cont_length = t_loc_end - t_cont_loc 'pick out the location and length of the data
                    t_contents = line.Substring(t_cont_loc, t_cont_length)
                    If t_contents.IndexOf(",") <> -1 Then
                        t_contents = """" + t_contents + """" 'handle commas in csv data'
                    End If
                    lib_csv_line = lib_csv_line + this_t_uuid + "," + t_contents + vbCrLf 'build csv rows over and over until all lib uuids found
                End While

                File.AppendAllText(path + "_tagtable.csv", tag_csv_line) 'append line to csv file
                File.AppendAllText(path + "_libtable.csv", lib_csv_line) 'append line to csv file
            Else
                For Each tag As String In maintable 'cycle through tags
                    t_loc = line.IndexOf(tag) 'find current tag in file
                    t_length = Len(tag) 'get length of current tag
                    Dim Close_tag_stage As String = ""
                    Dim Close_tag As String = ""
                    Close_tag_stage = tag.Substring(1) 'take everything but the first character of the current tag
                    Close_tag = "</" + Close_tag_stage 'append the xml closing chevron/slash to the front of the current tag
                    Try
                        t_loc_end = line.IndexOf(Close_tag, t_loc) 'find location of current closing tag
                    Catch outOfRange As ArgumentOutOfRangeException
                        Console.WriteLine("Element ignored: {0}", outOfRange.Message) 'hand situation if it is not there
                    End Try
                    t_cont_loc = t_loc + t_length 'get location of current tag contents
                    t_cont_length = t_loc_end - t_cont_loc 'get lenth of current tag contents
                    t_contents = line.Substring(t_cont_loc, t_cont_length) 'get contents of current tag
                    If tag = "<t_uuid>" Then
                        this_t_uuid = t_contents
                    End If

                    If t_contents.IndexOf(",") <> -1 Then
                        t_contents = """" + t_contents + """" 'handle commas in csv
                    End If
                    If Len(csv_line) = 0 Then
                        csv_line = t_contents 'prevent commas from being applied to the beginning of csv row
                    Else
                        csv_line = csv_line + "," + t_contents 'build csv row by adding contents and then a comma for each tag
                    End If

                Next

                t_length = Len(tag_tag)
                tag_csv_line = ""
                t_loc_end = 0
                While line.IndexOf(tag_tag, t_loc_end) <> -1 'get the tag data out of t-file
                    t_loc = line.IndexOf(tag_tag, t_loc_end)
                    t_loc_end = line.IndexOf("</t_tag>", t_loc)
                    t_cont_loc = t_loc + t_length
                    t_cont_length = t_loc_end - t_cont_loc
                    t_contents = line.Substring(t_cont_loc, t_cont_length)
                    If t_contents.IndexOf(",") <> -1 Then
                        t_contents = """" + t_contents + """" 'handle commas in csv data'
                    End If
                    tag_csv_line = tag_csv_line + this_t_uuid + "," + t_contents + vbCrLf 'build csv rows

                End While

                t_length = Len(lib_tag)
                lib_csv_line = ""
                t_loc_end = 0
                While line.IndexOf(lib_tag, t_loc_end) <> -1
                    t_loc = line.IndexOf(lib_tag, t_loc_end)
                    t_loc_end = line.IndexOf("</t_tags_uuid>", t_loc)
                    t_cont_loc = t_loc + t_length
                    t_cont_length = t_loc_end - t_cont_loc 'pick out lib uuid data
                    t_contents = line.Substring(t_cont_loc, t_cont_length)
                    If t_contents.IndexOf(",") <> -1 Then
                        t_contents = """" + t_contents + """" 'handle commas in csv data'
                    End If
                    lib_csv_line = lib_csv_line + this_t_uuid + "," + t_contents + vbCrLf 'build csv rows

                End While
                File.AppendAllText(path + "_tagtable.csv", tag_csv_line) 'append line to csv file
                File.AppendAllText(path + "_libtable.csv", lib_csv_line) 'append line to csv file
            End If


            File.AppendAllText(path + "_maintable.csv", csv_line) 'append line to csv file
            File.AppendAllText(path + "_maintable.csv", vbCrLf) 'ad a new line feed to the csv file
        Next
        Console.WriteLine()
        Console.WriteLine("Done! csv files have been saved to " + path)
        Console.WriteLine("Press any key...")
        Console.ReadKey()


        GoTo MAIN_MENU

SET_FILEPATH:
        Console.Clear()
        Console.WriteLine("Current path - " + path)
        Console.WriteLine("Type in new directory path (enter x to cancel)")
        Console.WriteLine()
        proposed_path = Console.ReadLine()
        If proposed_path = "x" Then
            GoTo MAIN_MENU
        End If
        path = proposed_path
        Console.WriteLine("New path - " + path)
        Console.WriteLine()
        Console.WriteLine("Press any key...")
        Console.ReadKey()
        GoTo MAIN_MENU

GUIDE:
        Console.Clear()
        Console.WriteLine("Hi, this tool converts t files (i.e. files conforming to the tschema.xsd) into csv format. " & _
        "It works like this, you put your t files in a folder (e.g. C:\T2CSV\) and the app finds the files and " & _
        "converts them into three csv tables (i.e. _maintable.csv, _tagtable.csv, _libtable.csv). Some things to note:")
        Console.WriteLine()
        Console.WriteLine(" - The leading underscore in the filename is just to promote the csv files to the top of the list in the directory (since there could be thousands of t-files in there too).")
        Console.WriteLine()
        Console.WriteLine(" - You can change the conversion directory if you are not happy with the default, that's cool, no hard feelings, I'm sure you have your reasons. ")
        Console.WriteLine()
        Console.WriteLine(" - Once you have your csv files you can open and join them in visualization software and see all the powerful features of the tschema in action")
        Console.WriteLine()
        Console.WriteLine(" - If you raise an exception which I haven't handled or find a bug then please pass the info on to software.stevebowden@gmail.com")
        Console.WriteLine()
        Console.WriteLine("Press any key...")
        Console.ReadKey()
        GoTo MAIN_MENU

ABOUT:
        Console.Clear()
        Console.WriteLine("T2CSV Converter - by Steven Bowden is licensed under a Creative Commons Attribution 4.0 International License. Based on a work at tschema.org/T2CSV/index.html")
        Console.WriteLine()
        Console.WriteLine("Performance-wise it is pretty impressive and overloads are unexpected, if you get an overload just modify the code to write to the .csv more regularly. " & _
                          "The code can be downloaded from tschema.org/T2CSV/index.html")
        Console.WriteLine()
        Console.WriteLine("The code is just string-play, nothing fancy so anyone with a basic understanding can get it and see " & _
                          "that there is no voodoo in data extraction and transformation, it's anyones game.")
        Console.WriteLine("Check out tschema.org for more information on the t schema and t files etc.")
        Console.WriteLine()
        Console.WriteLine("Contact email : software.stevebowden@gmail.com")
        Console.WriteLine()
        Console.WriteLine("DISCLAIMER : no guarantees the software is perfect, use at your own risk. Always validate results to ensure the tool is working for your specific usage case")
        Console.WriteLine()
        Console.WriteLine("Press any key...")
        Console.ReadKey()

        GoTo MAIN_MENU

    End Sub

End Module