public static void Run(PeriodicalList periodicalList)
{
var journalUrl = @"http://journal-abbreviations.library.ubc.ca/dump.php";
var project = periodicalList.Project;
var journalCollection = new List <Periodical>();
string completeList;
try
{
Cursor.Current = Cursors.WaitCursor;
using (var webClient = new WebClient())
{
using (var stream = webClient.OpenRead(journalUrl))
{
using (var streamReader = new StreamReader(stream))
{
completeList = streamReader.ReadToEnd();
}
}
}
}
catch (Exception e)
{
Cursor.Current = Cursors.Default;
MessageBox.Show(periodicalList, ImportJournalsResources.PubMedMacroReadErrorMessage.FormatString(journalUrl, e.Message), periodicalList.ProductName, MessageBoxButtons.OK, MessageBoxIcon.Error);
return;
}
var refCounter = 0;
try
{
Cursor.Current = Cursors.WaitCursor;
completeList = Regex.Match(completeList, @"(<tbody>.*?<\\/tbody)").ToString();
completeList = Regex.Replace(completeList, @"\\/", @"/");
var journals = Regex.Matches(completeList, "(?<=<tr>).*?(?=</tr>)");
foreach (var journalAndAbbrev in journals)
{
var journalTitle = string.Empty;
var abbreviation1 = string.Empty; // this one should have full stops
var abbreviation2 = string.Empty; // this one shouldn't
var abbreviation3 = string.Empty; // this one should be any all uppercase acronym after a colon in JournalTitl
var 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
var 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(project, 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(periodicalList, ImportJournalsResources.WoodwardMacroUpdateMessage, periodicalList.ProductName, MessageBoxButtons.YesNo, MessageBoxIcon.Question,
MessageBoxDefaultButton.Button2);
if (updateReferences == DialogResult.Yes)
{
var references = Program.ActiveProjectShell.PrimaryMainForm.GetFilteredReferences();
foreach (var 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++;
}
}
project.Periodicals.AddRange(journalCollection);
}
else
{
project.Periodicals.AddRange(journalCollection);
}
}
catch (Exception e)
{
Cursor.Current = Cursors.Default;
journalCollection = null;
MessageBox.Show(periodicalList, ImportJournalsResources.MacroImportingErrorMessage.FormatString(e.Message), periodicalList.ProductName, MessageBoxButtons.OK, MessageBoxIcon.Error);
}
finally
{
Cursor.Current = Cursors.Default;
if (journalCollection != null)
{
MessageBox.Show(periodicalList, ImportJournalsResources.WoodwardMacroResultMessage.FormatString(journalCollection.Count, refCounter), periodicalList.ProductName, MessageBoxButtons.OK, MessageBoxIcon.Information);
journalCollection = null;
}
}
}
public static void Run(PeriodicalList periodicalList)
{
var project = periodicalList.Project;
var journalUrl = @"http://ftp.ncbi.nih.gov/pubmed/J_Entrez.txt"; // URL for journal list text file
var journalCollection = new List <Periodical>();
string completeList;
try
{
Cursor.Current = Cursors.WaitCursor;
using (var webClient = new WebClient2()
{
Timeout = 60000
})
{
using (var stream = webClient.OpenRead(journalUrl))
{
using (var streamReader = new StreamReader(stream))
{
completeList = streamReader.ReadToEnd();
}
}
}
}
catch (Exception e)
{
Cursor.Current = Cursors.Default;
MessageBox.Show(periodicalList, ImportJournalsResources.PubMedMacroReadErrorMessage.FormatString(journalUrl, e.Message), periodicalList.ProductName, MessageBoxButtons.OK, MessageBoxIcon.Error);
return;
}
try
{
Cursor.Current = Cursors.WaitCursor;
var entrySplitters = new string[] { @"--------------------------------------------------------" };
var individualJournalEntries = completeList.Split(entrySplitters, StringSplitOptions.RemoveEmptyEntries).ToList();
var counter = 0;
var splitEntry = new Regex(@"^(?:JrId: )(?<JournalId>\d+?)(?:\nJournalTitle: )(?<JournalTitle>.*?)(?:\nMedAbbr: )(?<Abbreviation2>.*?)(?:\nISSN \(Print\): )(?<IssnPrint>.*?)(?:\nISSN \(Online\): )(?<IssnOnline>.*?)(?:\nIsoAbbr: )(?<Abbreviation1>.*?)(?:\nNlmId: )(?<NlmId>.*?)$",
RegexOptions.CultureInvariant | RegexOptions.Compiled | RegexOptions.Multiline);
foreach (string journalEntry in individualJournalEntries)
{
counter++;
string journalTitle;
string abbreviation1;
string abbreviation2;
string abbreviation3;
string issnPrint;
string issnOnline;
string nlmId;
var match = splitEntry.Match(journalEntry);
journalTitle = match.Groups["JournalTitle"].Value;
abbreviation1 = match.Groups["Abbreviation1"].Value;
abbreviation2 = match.Groups["Abbreviation2"].Value;
issnPrint = match.Groups["IssnPrint"].Value;
issnOnline = match.Groups["IssnOnline"].Value;
nlmId = match.Groups["NlmId"].Value;
if (string.IsNullOrEmpty(abbreviation1))
{
abbreviation1 = abbreviation2;
}
if (!abbreviation1.Contains(".") && !String.IsNullOrEmpty(abbreviation1))
{
var journalTitleWords = journalTitle.ToLowerInvariant().Split(new char[] { ' ', '.', ';', ',', ':', '&', '-' }, StringSplitOptions.RemoveEmptyEntries);
var abbreviation1Words = abbreviation1.Split(' ');
var abbreviation1WithFullStops = new List <string>();
foreach (var word in abbreviation1Words)
{
if (word.StartsWith("(") || word.EndsWith(")"))
{
abbreviation1WithFullStops.Add(word);
}
else if (!Array.Exists(journalTitleWords, x => x == word.ToLowerInvariant()))
{
abbreviation1WithFullStops.Add(word + ".");
}
else
{
abbreviation1WithFullStops.Add(word);
}
}
abbreviation1 = string.Join(" ", abbreviation1WithFullStops);
}
abbreviation3 = Regex.Match(journalTitle, @"(?:: )[A-Z]{2,6}$").ToString();
var journal = new Periodical(project, 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) && IssnValidator.IsValid(issnPrint))
{
journal.Issn = issnPrint;
}
else if (!string.IsNullOrEmpty(issnOnline) && IssnValidator.IsValid(issnOnline))
{
journal.Issn = issnOnline;
}
if (!string.IsNullOrEmpty(issnPrint) && IssnValidator.IsValid(issnPrint) && !string.IsNullOrEmpty(issnOnline) && IssnValidator.IsValid(issnOnline))
{
journal.Notes = "ISSN (Online): " + issnOnline;
}
if (!string.IsNullOrEmpty(nlmId))
{
journal.Notes = journal.Notes + "\nNlmID: " + nlmId;
}
journalCollection.Add(journal);
}
project.Periodicals.AddRange(journalCollection);
}
catch (Exception exception)
{
Cursor.Current = Cursors.Default;
journalCollection = null;
MessageBox.Show(periodicalList, ImportJournalsResources.MacroImportingErrorMessage.FormatString(exception.Message), periodicalList.ProductName, MessageBoxButtons.OK, MessageBoxIcon.Error);
}
finally
{
Cursor.Current = Cursors.Default;
if (journalCollection != null)
{
MessageBox.Show(periodicalList, ImportJournalsResources.PubMedMacroResultMessage.FormatString(journalCollection.Count.ToString()), periodicalList.ProductName, MessageBoxButtons.OK, MessageBoxIcon.Information);
journalCollection = null;
}
}
}
public static void Run(PeriodicalList periodicalList)
{
var project = periodicalList.Project;
string fileName = null;
var journalCollection = new List <Periodical>();
try
{
using (var openFileDialog = new OpenFileDialog
{
Filter = ImportJournalsResources.FileMacroOpenFileDialogFilters,
Title = ImportJournalsResources.FileMacroOpenFileDialogSubject
})
{
if (openFileDialog.ShowDialog(periodicalList) != DialogResult.OK)
{
return;
}
fileName = openFileDialog.FileName;
}
Cursor.Current = Cursors.WaitCursor;
string journalList;
var enc = GetFileEncoding(fileName);
using (var streamReader = new StreamReader(fileName, enc))
{
journalList = streamReader.ReadToEnd();
streamReader.Close();
}
var testRegex = new Regex("[\x00-\x1f-[\t\n\r]]", RegexOptions.CultureInvariant | RegexOptions.Compiled);
if (testRegex.IsMatch(journalList))
{
Cursor.Current = Cursors.Default;
MessageBox.Show(periodicalList, ImportJournalsResources.FileMacroNotSupportedCharactersMessage, periodicalList.ProductName, MessageBoxButtons.OK, MessageBoxIcon.Exclamation);
return;
}
var 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!
);
var matchCollection = matchRegex.Matches(journalList);
string sISSN = string.Empty;
foreach (Match match in matchCollection)
{
if (string.IsNullOrEmpty(match.Groups["FullName"].Value))
{
continue;
}
var journal = new Periodical(project, match.Groups["FullName"].Value);
if (!string.IsNullOrEmpty(match.Groups["Abbreviation1"].Value))
{
journal.StandardAbbreviation = match.Groups["Abbreviation1"].Value;
}
if (!string.IsNullOrEmpty(match.Groups["Abbreviation2"].Value))
{
journal.UserAbbreviation1 = match.Groups["Abbreviation2"].Value;
}
if (!string.IsNullOrEmpty(match.Groups["Abbreviation3"].Value))
{
journal.UserAbbreviation2 = match.Groups["Abbreviation3"].Value;
}
if (!string.IsNullOrEmpty(match.Groups["ISSN"].Value))
{
sISSN = match.Groups["ISSN"].Value;
if (IssnValidator.IsValid(sISSN))
{
journal.Issn = sISSN;
}
}
journalCollection.Add(journal);
}
project.Periodicals.AddRange(journalCollection);
}
catch (Exception exception)
{
Cursor.Current = Cursors.Default;
journalCollection = null;
MessageBox.Show(periodicalList, exception.ToString(), periodicalList.ProductName, MessageBoxButtons.OK, MessageBoxIcon.Error);
}
finally
{
Cursor.Current = Cursors.Default;
if (journalCollection != null)
{
MessageBox.Show(periodicalList, ImportJournalsResources.FileMacroResultMessage.FormatString(journalCollection.Count), periodicalList.ProductName, MessageBoxButtons.OK, MessageBoxIcon.Information);
journalCollection = null;
}
}
}
