public static void Main()
{
#region UserVariable
string journalUrl = @"ftp://ftp.ncbi.nih.gov/pubmed/J_Entrez.txt"; // URL for journal list text file
#endregion
if (Program.ProjectShells.Count == 0)
{
return; //no project open
}
Project activeProject = Program.ActiveProjectShell.Project;
List <Periodical> journalCollection = new List <Periodical>();
string completeList;
try
{
// Get list of journals from website
Cursor.Current = Cursors.WaitCursor;
DebugMacro.WriteLine("Reading list of journals from " + journalUrl);
WebClient client = new WebClient();
Stream stream = client.OpenRead(journalUrl);
StreamReader streamReader = new StreamReader(stream);
completeList = streamReader.ReadToEnd();
}
catch (Exception e)
{
Cursor.Current = Cursors.Default;
DebugMacro.WriteLine("Could not read file from " + journalUrl + ": " + e.Message);
return;
}
// split into individual Journals
try
{
Cursor.Current = Cursors.WaitCursor;
string[] entrySplitters = { @"--------------------------------------------------------" };
List <string> individualJournalEntries = new List <string>(completeList.Split(entrySplitters, StringSplitOptions.RemoveEmptyEntries));
DebugMacro.WriteLine("{0} journal entries found.", individualJournalEntries.Count.ToString());
DebugMacro.WriteLine("Parsing entries ...");
int counter = 0;
Regex splitEntry = new Regex(@"^(?:JrId: )(?<JournalId>\d+?)(?:\nJournalTitle: )(?<JournalTitle>.*?)(?:\nMedAbbr: )(?<Abbreviation2>.*?)(?:\nISSN \(Print\): )(?<IssnPrint>.*?)(?:\nISSN \(Online\): )(?<IssnOnline>.*?)(?:\nIsoAbbr: )(?<Abbreviation1>.*?)(?:\nNlmId: )(?<NlmId>.*?)$",
RegexOptions.CultureInvariant | RegexOptions.Compiled | RegexOptions.Multiline);
foreach (string journalEntry in individualJournalEntries)
{
counter++;
string journalTitle;
string abbreviation1; // this one should have full stops
string abbreviation2; // this one shouldn't
string abbreviation3; // this one should be any all uppercase acronym after a colon in JournalTitle
string issnPrint;
string issnOnline;
string nlmId;
// split into fields
Match m = splitEntry.Match(journalEntry);
//if (String.IsNullOrEmpty(m.Groups["JournalId"].Value)) continue; // nothing here
journalTitle = m.Groups["JournalTitle"].Value;
abbreviation1 = m.Groups["Abbreviation1"].Value;
abbreviation2 = m.Groups["Abbreviation2"].Value;
issnPrint = m.Groups["IssnPrint"].Value; // to be validated
issnOnline = m.Groups["IssnOnline"].Value;
nlmId = m.Groups["NlmId"].Value;
// check format of abbreviation1
if (String.IsNullOrEmpty(abbreviation1))
{
abbreviation1 = abbreviation2;
}
if (!abbreviation1.Contains(".") && !String.IsNullOrEmpty(abbreviation1))
{
string[] journalTitleWords = journalTitle.ToLowerInvariant().Split(new char[] { ' ', '.', ';', ',', ':', '&', '-' }, StringSplitOptions.RemoveEmptyEntries);
string[] abbreviation1Words = abbreviation1.Split(' ');
List <string> abbreviation1WithFullStops = new List <string>();
foreach (string word in abbreviation1Words)
{
if (word.StartsWith("(") || word.EndsWith(")"))
{
abbreviation1WithFullStops.Add(word);
}
else if (!Array.Exists(journalTitleWords, x => x == word.ToLowerInvariant()))
{
abbreviation1WithFullStops.Add(word + ".");
}
else
{
abbreviation1WithFullStops.Add(word);
}
}
abbreviation1 = String.Join(" ", abbreviation1WithFullStops);
}
// try to establish Abbreviation3
abbreviation3 = Regex.Match(journalTitle, @"(?:: )[A-Z]{2,6}$").ToString();
Periodical journal = new Periodical(activeProject, journalTitle);
if (!String.IsNullOrEmpty(abbreviation1))
{
journal.StandardAbbreviation = abbreviation1;
}
if (!String.IsNullOrEmpty(abbreviation2))
{
journal.UserAbbreviation1 = abbreviation2;
}
if (!String.IsNullOrEmpty(abbreviation3))
{
journal.UserAbbreviation2 = abbreviation3;
}
if (!string.IsNullOrEmpty(issnPrint) && IsValidISSN(issnPrint))
{
journal.Issn = issnPrint;
}
else if (!string.IsNullOrEmpty(issnOnline) && IsValidISSN(issnOnline))
{
journal.Issn = issnOnline;
}
if (!string.IsNullOrEmpty(issnPrint) && IsValidISSN(issnPrint) && !string.IsNullOrEmpty(issnOnline) && IsValidISSN(issnOnline))
{
journal.Notes = "ISSN (Online): " + issnOnline;
}
if (!String.IsNullOrEmpty(nlmId))
{
journal.Notes = journal.Notes + "\nNlmID: " + nlmId;
}
journalCollection.Add(journal);
}
activeProject.Periodicals.AddRange(journalCollection);
}
catch (Exception e)
{
Cursor.Current = Cursors.Default;
DebugMacro.WriteLine("An error occurred importing the journal data: " + e.Message);
return;
}
finally
{
Cursor.Current = Cursors.Default;
var saveOptions = new DesktopDatabaseSaveOptions();
saveOptions.ForceMultiUserUpdate = true;
activeProject.Save(saveOptions);
Cursor.Current = Cursors.Default;
if (journalCollection != null)
{
MessageBox.Show(journalCollection.Count.ToString() + " journal(s) were imported (not counting duplicates)", "Citavi", MessageBoxButtons.OK, MessageBoxIcon.Information);
journalCollection = null;
}
}
} //end Main()
public static void Main()
{
#region UserVariable
string journalUrl = @"http://journal-abbreviations.library.ubc.ca/dump.php"; // URL for journal list text file
#endregion
if (Program.ProjectShells.Count == 0)
{
return; //no project open
}
Project activeProject = Program.ActiveProjectShell.Project;
List <Periodical> journalCollection = new List <Periodical>();
string completeList;
try
{
// Get list of journals from website
Cursor.Current = Cursors.WaitCursor;
DebugMacro.WriteLine("Reading list of journals from " + journalUrl);
WebClient client = new WebClient();
Stream stream = client.OpenRead(journalUrl);
StreamReader streamReader = new StreamReader(stream);
completeList = streamReader.ReadToEnd();
}
catch (Exception e)
{
Cursor.Current = Cursors.Default;
DebugMacro.WriteLine("Could not read file from " + journalUrl + ": " + e.Message);
return;
}
// split into individual Journals
int counter = 0;
int refCounter = 0;
try
{
Cursor.Current = Cursors.WaitCursor;
// prepare data
completeList = Regex.Match(completeList, @"(<tbody>.*?<\\/tbody)").ToString();
completeList = Regex.Replace(completeList, @"\\/", @"/");
MatchCollection journals = Regex.Matches(completeList, "(?<=<tr>).*?(?=</tr>)");
DebugMacro.WriteLine("{0} journal entries found.", journals.Count.ToString());
DebugMacro.WriteLine("Parsing entries ...");
foreach (Match journalAndAbbrev in journals)
{
string journalTitle = String.Empty;
string abbreviation1 = String.Empty; // this one should have full stops
string abbreviation2 = String.Empty; // this one shouldn't
string abbreviation3; // this one should be any all uppercase acronym after a colon in JournalTitl
MatchCollection journalData = Regex.Matches(journalAndAbbrev.ToString(), "(?<=<td>).*?(?=</td>)");
if (journalData.Count < 2)
{
continue;
}
abbreviation1 = journalData[0].Value;
journalTitle = Regex.Replace(journalData[1].Value, @"\bfur\b", "für");
// generate abbreviation2 by removing full stops from abbreviation1
string[] abbreviation1Words = abbreviation1.Split(' ');
for (int i = 0; i < abbreviation1Words.Length; i++)
{
abbreviation1Words[i] = abbreviation1Words[i].TrimEnd('.');
}
abbreviation2 = String.Join(" ", abbreviation1Words);
// try to establish Abbreviation3
abbreviation3 = Regex.Match(journalTitle, @"(?:: )[A-Z]{2,6}$").ToString();
Periodical journal = new Periodical(activeProject, journalTitle);
if (!String.IsNullOrEmpty(abbreviation1))
{
journal.StandardAbbreviation = abbreviation1;
}
if (!String.IsNullOrEmpty(abbreviation2))
{
journal.UserAbbreviation1 = abbreviation2;
}
if (!String.IsNullOrEmpty(abbreviation3))
{
journal.UserAbbreviation2 = abbreviation3;
}
journalCollection.Add(journal);
}
DialogResult updateReferences = MessageBox.Show(
@"Would you like to update journal information for references in current selection (all references if no selection is active)?",
"Citavi Macro", MessageBoxButtons.YesNo, MessageBoxIcon.Question,
MessageBoxDefaultButton.Button2);
// update references that have identical titles
if (updateReferences == DialogResult.Yes)
{
List <Reference> references = Program.ActiveProjectShell.PrimaryMainForm.GetFilteredReferences();
foreach (Reference reference in references)
{
if (reference.Periodical == null)
{
continue;
}
if (journalCollection.Any(item => item.Name == reference.Periodical.Name) && !String.IsNullOrEmpty(reference.Periodical.Name))
{
reference.Periodical = journalCollection.Where(item => item.Name == reference.Periodical.Name).FirstOrDefault();
refCounter++;
}
}
activeProject.Periodicals.AddRange(journalCollection);
}
else
{
activeProject.Periodicals.AddRange(journalCollection);
}
}
catch (Exception e)
{
Cursor.Current = Cursors.Default;
DebugMacro.WriteLine("An error occurred importing the journal data: " + e.ToString());
return;
}
finally
{
Cursor.Current = Cursors.Default;
var saveOptions = new DesktopDatabaseSaveOptions();
saveOptions.ForceMultiUserUpdate = true;
activeProject.Save(saveOptions);
Cursor.Current = Cursors.Default;
if (journalCollection != null)
{
MessageBox.Show(journalCollection.Count.ToString() + " journal(s) were imported (not counting duplicates)\n " + refCounter.ToString() + " reference(s) were changed.", "Citavi", MessageBoxButtons.OK, MessageBoxIcon.Information);
journalCollection = null;
}
}
} //end Main()
public static void Main()
{
if (Program.ProjectShells.Count == 0)
{
return; //no project open
}
Project activeProject = Program.ActiveProjectShell.Project;
string fileName = null;
List <Periodical> journalCollection = new List <Periodical>();
try
{
// Dialog zum Auswählen des gewünschten Ordners einblenden
OpenFileDialog openFileDialog = new OpenFileDialog();
openFileDialog.Filter = "Text Dateien (*.txt)|*.txt|Alle Dateien (*.*)|*.*";
openFileDialog.Title = "Wählen Sie die zu importierende Zeitschriften-Datei aus (Text-Format)";
if (openFileDialog.ShowDialog() == DialogResult.OK)
{
fileName = openFileDialog.FileName;
}
else
{
return;
}
//Hourglass or other wait cursor
Cursor.Current = Cursors.WaitCursor;
Encoding enc = GetFileEncoding(fileName);
StreamReader streamReader = new StreamReader(fileName, enc);
string journalList = streamReader.ReadToEnd();
streamReader.Close();
//We check for certain non-printable chars to ensure we are dealing with a "text file"
Regex testRegex = new Regex("[\x00-\x1f-[\t\n\r]]", RegexOptions.CultureInvariant | RegexOptions.Compiled);
if (testRegex.IsMatch(journalList))
{ //this is most likely not a textfile
Cursor.Current = Cursors.Default;
MessageBox.Show("Diese Datei enthält ungültige Zeichen und kann daher nicht importiert werden.", "Citavi", MessageBoxButtons.OK, MessageBoxIcon.Exclamation);
return;
}
Regex matchRegex = new Regex(
@"^(?<FullName>[^#=;|\t\n]+?)(?: *[#=;|\t] *(?<Abbreviation1>[^#=;|\t\n]*))?(?: *[#=;|\t] *(?<Abbreviation2>[^#=;|\t\n]*))?(?: *[#=;|\t] *(?<Abbreviation3>[^#=;|\t\n]*))?(?: *[#=;|\t] *(?<ISSN>[^#=;|\t\n]*))??$",
RegexOptions.CultureInvariant
| RegexOptions.Compiled
| RegexOptions.Multiline //IMPORTANT!
);
//// Capture all matches in the journalList
MatchCollection ms = matchRegex.Matches(journalList);
//MessageBox.Show("Insgesamt sind " + ms.Count.ToString() + " Treffer gefunden worden.");
foreach (Match m in ms)
{
if (string.IsNullOrEmpty(m.Groups["FullName"].Value))
{
continue;
}
Periodical journal = new Periodical(activeProject, m.Groups["FullName"].Value);
if (!string.IsNullOrEmpty(m.Groups["Abbreviation1"].Value))
{
journal.StandardAbbreviation = m.Groups["Abbreviation1"].Value;
}
if (!string.IsNullOrEmpty(m.Groups["Abbreviation2"].Value))
{
journal.UserAbbreviation1 = m.Groups["Abbreviation2"].Value;
}
if (!string.IsNullOrEmpty(m.Groups["Abbreviation3"].Value))
{
journal.UserAbbreviation2 = m.Groups["Abbreviation3"].Value;
}
string sISSN = "";
if (!string.IsNullOrEmpty(m.Groups["ISSN"].Value))
{
sISSN = m.Groups["ISSN"].Value;
if (IsValidISSN(sISSN))
{
journal.Issn = sISSN;
}
}
journalCollection.Add(journal);
}
activeProject.Periodicals.AddRange(journalCollection);
//activeProject.Journals.AddRangeBound(journalCollection);
}
catch (Exception exception)
{
Cursor.Current = Cursors.Default;
MessageBox.Show(exception.ToString(), "Citavi", MessageBoxButtons.OK, MessageBoxIcon.Error);
}
finally
{
var saveOptions = new DesktopDatabaseSaveOptions();
saveOptions.ForceMultiUserUpdate = true;
activeProject.Save(saveOptions);
Cursor.Current = Cursors.Default;
if (journalCollection != null)
{
MessageBox.Show("Es wurde(n) " + journalCollection.Count.ToString() + " Zeitschrift(en) eingelesen. \n(Dubletten wurden unterdrückt.)", "Citavi", MessageBoxButtons.OK, MessageBoxIcon.Information);
journalCollection = null;
}
}
} //end Main()
