protected virtual IEnumerable <Clipping> ParseStreamFromPath(string path, FormatType format)
{
//Open stream via path to the .txt file.
var stream = new FileStream(path, FileMode.Open);
using (var sr = new StreamReader(stream)) {
int lineNumber = 0;
string line = null;
int clippingLineNumber = 0;
Clipping clipping = new Clipping();
while ((line = sr.ReadLine()) != null)
{
lineNumber++;
if (line == ClippingSeparator)
{
yield return(clipping);
clippingLineNumber = 0;
clipping = new Clipping();
}
else
{
clippingLineNumber++;
}
ParseClipping(clippingLineNumber, line, clipping, format);
}
}
}
//TODO Too many parameters, simplify
protected virtual void ParseDateExact(string[] splitLine, Clipping clipping, int dateIndex,
string[] dateFormats, CultureInfo culture)
{
//Formatting help: https://msdn.microsoft.com/en-us/library/8kb3ddd4.aspx
string dateAdded = String.Join(" ", splitLine[dateIndex], splitLine[dateIndex + 1],
splitLine[dateIndex + 3], splitLine[dateIndex + 5], splitLine[dateIndex + 6]);
//Removing single quotes. There might be other "noise" characters, this one is especially important in typeRub.
string input = dateAdded.Replace("'", string.Empty);
try {
/*Dates have to be parsed and converted to a dateTime format. TryParseExact should do the
* trick as long as the proper format is added to the dateFormats array. */
DateTime dt;
if (DateTime.TryParseExact(input, dateFormats, culture, DateTimeStyles.None, out dt))
{
if (dt < DateTime.Now)
{
clipping.DateAdded = dt;
}
}
}
catch (Exception ex) {
clipping.DateAdded = Defaults.DateAdded;
new Exception("Error encountered adding date: " + ex.Message, ex);
}
}
/// <summary>
/// Adds a clipping to the storage and returns a unique identifier for later retrieval.
/// </summary>
/// <param name="clipping">The clipping to add to the storage</param>
/// <returns>The unique identifier corresponding to this clipping</returns>
public int AddClipping(Clipping clipping)
{
int id = numberedClippings.Keys.Count;
numberedClippings.Add(id, clipping);
return(id);
}
public void GenerateClippingList(IEnumerable <Clipping> clippings)
{
rawClippingCount = 0;
foreach (var item in clippings)
{
//Adding clippings to the currently used, dictionary database.
if (!Clipping.IsNullOrEmpty(item) ||
(Clipping.IsNullOrEmpty(item) && Clipping.IsBookMark(item)))
{
ClippingStorage.AddClipping(item);
}
++rawClippingCount;
}
//Now adding clippings to the layout'ed, list database.
int numberOfClippings = ClippingStorage.numberedClippings.Count;
if (numberOfClippings > 0)
{
for (int i = 0; i < numberOfClippings; i++)
{
Clipping clippingToAdd = ClippingStorage.GetClipping(i);
ClippingStorage.finalClippingsList.Add(clippingToAdd);
}
}
else
{
//TODO What if there is no valid clippings at all?
}
}
protected virtual void ParseLine4(string line, Clipping clipping)
{
try {
clipping.Text = line.Trim();
}
catch (Exception) {
clipping.Text = Defaults.Text;
}
}
protected override void InitDefaults()
{
Defaults = new Clipping();
Defaults.BookName = "Título desconocido";
Defaults.Author = "Autor desconocido";
Defaults.Text = "";
Defaults.Location = "";
Defaults.Page = "";
Defaults.DateAdded = new DateTime();
}
protected override void InitDefaults()
{
Defaults = new Clipping();
Defaults.BookName = "Unknown book";
Defaults.Author = "Unknown author";
Defaults.Location = "";
Defaults.Text = "";
Defaults.Page = "";
Defaults.DateAdded = new DateTime();
}
//TODO quick hardcoded solution in order to exclude bookmarks from empty clipping removal.
//Quick and dirty, can be way better.
public static bool IsBookMark(Clipping item)
{
if (item.ClippingType == ClippingTypeEnum.Bookmark ||
item.ClippingType == ClippingTypeEnum.Marcador)
{
return(true);
}
else
{
return(false);
}
}
protected virtual IEnumerable <Clipping> ParseFromString(string content, FormatType format)
{
Clipping clipping = new Clipping();
//TODO Hardcoded line break to solve?
string[] result = content.Split(new string[] { "\n", "\r\n" }, StringSplitOptions.None);
string lines = null;
int lineNumber = 0;
int clippingLineNumber = 0;
while (lineNumber < result.Length)
{
lineNumber++;
clippingLineNumber++;
if (result[lineNumber - 1] == ClippingSeparator)
{
yield return(clipping);
clippingLineNumber = 0;
clipping = new Clipping();
}
else
{
//TODO Simplify this foul smelling conditional logic
if (!(clippingLineNumber == 1 || clippingLineNumber == 2 || clippingLineNumber == 3) && result[lineNumber] != ClippingSeparator)
{
lines += result[lineNumber - 1];
}
else if (clippingLineNumber == 3)
{
lines = null;
}
else
{
if (lines == null)
{
lines = result[lineNumber - 1];
}
ParseClipping(clippingLineNumber, lines, clipping, format);
lines = null;
}
}
}
}
protected virtual void ParseLine1(string line, Clipping clipping)
{
try {
Match match = Regex.Match(line, Line1RegexPattern);
if (match.Success)
{
string bookName = match.Groups[1].Value.Trim();
string author = match.Groups[2].Value.Trim();
clipping.BookName = !string.IsNullOrEmpty(bookName) ? bookName : Defaults.BookName;
clipping.Author = !string.IsNullOrEmpty(author) ? author : Defaults.Author;
}
}
catch (Exception) {
clipping.BookName = Defaults.BookName;
clipping.Author = Defaults.Author;
System.Diagnostics.Debug.WriteLine("Clipping Line 1 did not match regex pattern, using default values for Author and Bookname.");
}
}
private void SetClippingType(string clippingType, Clipping clipping)
{
switch (clippingType.ToLower())
{
case "subrayado":
clipping.ClippingType = ClippingTypeEnum.Subrayado;
break;
case "nota":
clipping.ClippingType = ClippingTypeEnum.Notas;
break;
case "marcador":
clipping.ClippingType = ClippingTypeEnum.Marcador;
break;
default:
clipping.ClippingType = ClippingTypeEnum.NoReconocido;
break;
}
}
/* Calling to the different methods parsing the different lines. Line 3 is irrelevant
* (just white space acting as a separator) and thus is not included in the logic. */
public virtual void ParseClipping(int lineNumber, string line, Clipping clipping, FormatType format)
{
try {
switch (lineNumber)
{
case 1:
ParseLine1(line, clipping);
break;
case 2:
ParseLine2(line, clipping, format);
break;
case 4:
ParseLine4(line, clipping);
break;
}
} catch (Exception ex) {
new Exception("Error encountered parsing line " + lineNumber + ": " + ex.Message, ex);
}
}
protected virtual void ParseDate(string[] splitLine, Clipping clipping, int dateIndex)
{
try {
string[] filteredLine = splitLine.Where(item => !item.Contains("GMT")).ToArray();
string dateAddedString;
//Hackish, removing GMT to simplify parse. Indexes change and problems. Can be improved.
if (splitLine.Length != filteredLine.Length)
{
dateAddedString = String.Join(" ", splitLine[dateIndex], splitLine[dateIndex + 1], splitLine[dateIndex + 2], splitLine[dateIndex + 3], splitLine[dateIndex + 4]);
}
else
{
dateAddedString = String.Join(" ", splitLine[dateIndex], splitLine[dateIndex + 1], splitLine[dateIndex + 2], splitLine[dateIndex + 3], splitLine[dateIndex + 4], splitLine[dateIndex + 5]);
}
DateTime dateAdded = DateTime.Parse(dateAddedString);
clipping.DateAdded = dateAdded;
} catch (Exception ex) {
clipping.DateAdded = Defaults.DateAdded;
new Exception("Error encountered adding date: " + ex.Message, ex);
}
}
protected override void ParseLine2(string line, Clipping clipping, FormatType format)
{
var split = line.Split(' ');
string fileType = null;
bool hasPageNumber = false;
bool hasLocation = false;
bool hasInstapaper = false;
try {
if (!String.IsNullOrEmpty(format.ID))
{
fileType = format.ID;
}
} catch (Exception ex) {
System.Diagnostics.Debug.WriteLine(ex.Message, "Can't identify TXT format.");
}
var clippingType = split[format.clippingTypePosition];
switch (clippingType.ToLower())
{
case "highlight":
clipping.ClippingType = ClippingTypeEnum.Highlight;
break;
case "note":
clipping.ClippingType = ClippingTypeEnum.Note;
break;
case "bookmark":
clipping.ClippingType = ClippingTypeEnum.Bookmark;
break;
default:
clipping.ClippingType = ClippingTypeEnum.NotRecognized;
break;
}
//Check if line contains any of the critical strings stored in keywords arrays.
hasPageNumber = format.pageWording.Any(line.Contains);
hasLocation = format.locationWording.Any(line.Contains);
/* Indexes are different in Spanish and English version (answers "where to cut" for the different variables).
* It also depends on particular formats for each language.*/
var dateIndex = format.dateIndex;
var locationIndex = format.locationIndex;
var pageIndex = format.pageIndex;
try {
if (hasPageNumber)
{
var pageNumber = split[pageIndex];
clipping.Page = pageNumber;
locationIndex = format.hasPageLocationIndex;
dateIndex = hasLocation ? format.hasPageHasLocationDateIndex : format.hasPageDateIndex;
}
} catch (Exception) {
clipping.Page = Defaults.Page;
}
try {
if (hasLocation)
{
var location = split[locationIndex];
clipping.Location = location;
}
} catch (Exception) {
clipping.Location = Defaults.Location;
}
/*Indexes are different in Spanish and English version (answers "where to cut" for the different variables).
* It also depends on particular formats for each language. If any format exceptions occur or it is better to
* manually look for something for any reason, add logic below. */
if (fileType == "typeRick")
{
if (split[1] == "Clip")
{
hasInstapaper = true;
}
if (hasInstapaper)
{
locationIndex = 6;
dateIndex = 10;
if (hasLocation)
{
var location = split[locationIndex];
clipping.Location = location;
}
}
}
ParseDate(split, clipping, dateIndex);
}
protected override void ParseLine2(string line, Clipping clipping, FormatType format)
{
var split = line.Split(' ');
var fileType = "";
bool hasPageNumber = false;
bool hasLocation = false;
//Detect type of file.
try {
if (!String.IsNullOrEmpty(format.ID))
{
fileType = format.ID;
}
}
catch (Exception ex) {
System.Diagnostics.Debug.WriteLine(ex.Message, "Can't identify TXT format.");
}
string clippingType = split[format.clippingTypePosition];
SetClippingType(clippingType, clipping);
hasPageNumber = format.pageWording.Any(line.Contains);
hasLocation = format.locationWording.Any(line.Contains);
var dateIndex = format.dateIndex;
var locationIndex = format.locationIndex;
var pageIndex = format.pageIndex;
var hasPageDateIndex = format.hasPageDateIndex;
var hasPageHasLocationDateIndex = format.hasPageHasLocationDateIndex;
var hasPageLocationIndex = format.hasPageLocationIndex;
bool isSubtypeKyuni = false;
if (split[1] == "Tu")
{
isSubtypeKyuni = true;
}
if (isSubtypeKyuni)
{
dateIndex = 10;
locationIndex = 6;
hasPageDateIndex = 10;
}
try {
if (hasPageNumber)
{
var pageNumber = split[pageIndex];
clipping.Page = pageNumber;
locationIndex = hasPageLocationIndex;
dateIndex = hasLocation ? hasPageHasLocationDateIndex : hasPageDateIndex;
}
}
catch (Exception) {
clipping.Page = Defaults.Page;
}
try {
if (hasLocation)
{
var location = split[locationIndex];
clipping.Location = location;
}
}
catch (Exception) {
clipping.Location = Defaults.Location;
}
ParseDateExact(split, clipping, dateIndex, dateFormats, spaCulture);
}
public static bool IsNullOrEmpty(Clipping item)
{
return(item == null || string.IsNullOrEmpty(item.Text));
}
protected abstract void ParseLine2(string line, Clipping clipping, FormatType format);
