public Form1()
{
InitializeComponent();
db = new db_adapter();
if (db.check_connection() == true) label1.Text += "\nСоединение с БД есть. \n";
else label1.Text += "\n Нет подключения к серверу БД. \n";
ya = new ya_translate();
if (ya.connected() == true) label1.Text += "\nСоединение с переводчиком есть. \n";
else label1.Text += "\n Нет подключения к серверу API Яндекс.Переводчика. \n";
}
private void handler(object sender, EventArgs e)
{
string title_ru = rf.Controls[1].Text;
string title_en = rf.Controls[4].Text;
string refs = rf.Controls[2].Text;
refs = refs.Replace("\n", "<br>");
db_adapter db = new db_adapter();
db.send_trase_refs(title_ru, title_en, refs);
rf.Close();
}
public void parse()
{
FileInfo file = new FileInfo(@"result_temp.xml");
FileStream fs = file.Create();
string text = "<?xml version=\"1.0\" encoding=\"UTF-8\"?> \n <issues> \n </issues>";
fs.Close();
FileStream fsw = file.OpenWrite();
byte[] array_in = System.Text.Encoding.Default.GetBytes(text);
fsw.Write(array_in, 0, array_in.Length);
fsw.Close();
///////////////////////////////////////////////////////////////////////
// Добавляем теги секций в статьи и разбирваем на файлы. 1 ф = 1 статья
///////////////////////////////////////////////////////////////////////
XmlDocument xDoc = new XmlDocument();
xDoc.Load(fc);
XmlElement xRoot = xDoc.DocumentElement;
XmlNodeList childnodes = xRoot.SelectNodes("issue/articles/*");
string current_section = "";
for (int i = 0; i < childnodes.Count; i++)
{
if (childnodes[i].Name == "section")
{
current_section = childnodes[i].ChildNodes[0].InnerText;
}
else
{
XmlElement cur_sec_node = xDoc.CreateElement("sect");
cur_sec_node.InnerText = current_section;
childnodes[i].AppendChild(cur_sec_node);
}
}
childnodes = xRoot.SelectNodes("issue/articles/article");
int y = 0;
string u = ".xml";
string coi = "<?xml version=\"1.0\" encoding=\"UTF-8\"?> \n";
foreach (XmlNode n in childnodes)
{
FileInfo file1 = new FileInfo(y.ToString() + u);
FileStream fs1 = file1.Create();
string text1 = coi + n.OuterXml;
fs1.Close();
FileStream fsw1 = file1.OpenWrite();
byte[] array_in1 = System.Text.Encoding.UTF8.GetBytes(text1);
fsw1.Write(array_in1, 0, array_in1.Length);
fsw1.Close();
y++;
}
///////////////////////////////////////////////////////////////////////
// Добавляем начальные данные в результирующий xml
///////////////////////////////////////////////////////////////////////
string issue_name = f.Controls[2].Text;
XDocument xdoc = XDocument.Load("result_temp.xml");
XElement root = xdoc.Element("issues");
root.Add(
new XElement("issue",
new XAttribute("published", "true"),
new XAttribute("current", "true"),
new XAttribute("identification", "title"),
new XElement("title",
new XAttribute("locale", "ru_RU"),
issue_name),
new XElement("access_date", "01-01-2001"),
new XElement("volume", "0"),
new XElement("number", "5(33)"),
new XElement("year", "0"),
new XElement("date_published", "2015-12-07")
));
xdoc.Save("result_temp.xml");
///////////////////////////////////////////////////////////////////////
// Помещаем данные в локальный объект.
///////////////////////////////////////////////////////////////////////
amount = y;
List<article> a = new List<article>();
for (int k9 = 0; k9 < amount; k9++)
{
a.Add(new article());
}
string wws = "";
y = 0;
for (int i = 0; i < amount; i++)
{
XmlDocument xDoc1 = new XmlDocument();
xDoc1.Load(y.ToString() + u);
XmlElement xRoot1 = xDoc1.DocumentElement;
XmlNodeList childnodes1;
childnodes1 = xRoot1.SelectNodes("sect");
wws = childnodes1.Item(0).InnerText;
a[y].section = wws;
try
{
childnodes1 = xRoot1.SelectNodes("artTitles/artTitle[1]");
wws = childnodes1.Item(0).InnerText;
a[y].title_rus = wws;
} catch(Exception e){
a[y].title_rus = "Для этой статьи нет заголовка";
}
try
{
childnodes1 = xRoot1.SelectNodes("artTitles/artTitle[2]");
wws = childnodes1.Item(0).InnerText;
a[y].title_en = wws;
}
catch(Exception e){
a[y].title_en = "No title for this article";
}
childnodes1 = xRoot1.SelectNodes("abstracts/abstract[1]");
if (childnodes1.Item(0) != null)
{
wws = childnodes1.Item(0).InnerText;
a[y].abstract_rus = wws;
}
else a[y].abstract_rus = "";
childnodes1 = xRoot1.SelectNodes("abstracts/abstract[2]");
if (childnodes1.Item(0) != null) wws = childnodes1.Item(0).InnerText;
a[y].abstract_en = wws;
childnodes1 = xRoot1.SelectNodes("authors/author");
int count = childnodes1.Count;
string fullname = "";
List<author> b = new List<author>();
a[y].authors = new List<author>();
for (int k3 = 0; k3 < count; k3++) { b.Add(new author()); }
for (int mm = 0; mm < count; mm++)
{
childnodes1 = xRoot1.SelectNodes("authors/author[" + (mm + 1).ToString() + "]/individInfo/surname");
wws = childnodes1.Item(0).InnerText;
string surname = wws;
childnodes1 = xRoot1.SelectNodes("authors/author[" + (mm + 1).ToString() + "]/individInfo/initials");
wws = childnodes1.Item(0).InnerText;
char[] ch = wws.ToCharArray();
string name = "";
string otchestvo = "";
bool flag = false;
for (int gh = 0; gh < ch.Length; gh++)
{
if (ch[gh] == ' ') flag = true;
if (flag) otchestvo += ch[gh]; else name += ch[gh];
}
name.Trim();
otchestvo.Trim();
fullname = name + " " + otchestvo + " " + surname;
b[mm].fullname_ru = fullname;
b[mm].firstname = name;
b[mm].middlename = otchestvo;
b[mm].lastname = surname + " " + ya.translit_toEn(name) + " " + ya.translit_toEn(otchestvo) + " " + ya.translit_toEn(surname);
b[mm].fullname_en = ya.translit_toEn(name) + " " + ya.translit_toEn(otchestvo) + " " + ya.translit_toEn(surname);
b[mm].country = "RU";
try
{
childnodes1 = xRoot1.SelectNodes("authors/author[" + (mm + 1).ToString() + "]/individInfo/email");
wws = childnodes1.Item(0).InnerText;
b[mm].email = wws;
}
catch (Exception e)
{
b[mm].email = "No e-mail";
}
childnodes1 = xRoot1.SelectNodes("authors/author[" + (mm + 1).ToString() + "]/individInfo/orgName");
wws = childnodes1.Item(0).InnerText;
b[mm].affiliation_ru = wws;
try
{
childnodes1 = xRoot1.SelectNodes("authors/author[" + (mm + 1).ToString() + "]/individInfo[@lang=\"ENG\"]/orgName");
wws = childnodes1.Item(0).InnerText;
b[mm].affiliation_en = wws;
}
catch (Exception e) { b[mm].affiliation_en = ya.translit_toEn(b[mm].affiliation_ru); }
try
{
childnodes1 = xRoot1.SelectNodes("authors/author[" + (mm + 1).ToString() + "]/individInfo/otherInfo");
wws = childnodes1.Item(0).InnerText;
b[mm].biography_ru = wws;
}
catch (Exception e) { b[mm].biography_ru = "No info"; }
try
{
childnodes1 = xRoot1.SelectNodes("authors/author[" + (mm + 1).ToString() + "]/individInfo[@lang=\"ENG\"]/otherInfo");
wws = childnodes1.Item(0).InnerText;
b[mm].biography_en = wws;
}
catch (Exception e) { b[mm].biography_en = "No info"; }
}
a[y].authors = b;
childnodes1 = xRoot1.SelectNodes("pages");
wws = childnodes1.Item(0).InnerText;
a[y].pages = wws;
if (xRoot1.SelectNodes("dates/dateReceived") == null)
{
childnodes1 = xRoot1.SelectNodes("dates/dateReceived");
wws = childnodes1.Item(0).InnerText;
a[y].date_published = wws;
}
else a[y].date_published = "2015-12-07";
a[y].copyright_holder = fullname;
a[y].copyright_year = "2015";
childnodes1 = xRoot1.SelectNodes("keywords/kwdGroup/keyword");
count = childnodes1.Count;
a[y].indexing_subject = new List<string>();
for (int kl = 0; kl < count; kl++)
{
childnodes1 = xRoot1.SelectNodes("keywords/kwdGroup/keyword[" + (kl + 1).ToString() + "]");
wws = childnodes1.Item(0).InnerText;
a[y].indexing_subject.Add(wws);
}
childnodes1 = xRoot1.SelectNodes("references/reference");
count = childnodes1.Count;
rf = new Form();
rf.Size = new Size(1370, 600);
Label l1 = new Label();
l1.Location = new Point(20,50);
l1.Size = new Size(620, 400);
Label l2 = new Label();
l2.Location = new Point(100, 5);
l2.Size = new Size(400, 20);
TextBox tb = new TextBox();
tb.Multiline = true;
tb.Size = new Size(650, 400);
tb.Location = new Point(650, 50);
Button sub = new Button();
sub.Location = new Point(100, 450);
sub.Text = "Готово -->";
sub.Click += new System.EventHandler(this.handler);
Label l3 = new Label();
l3.Visible = false;
rf.Controls.Add(l1);
rf.Controls.Add(l2);
rf.Controls.Add(tb);
rf.Controls.Add(sub);
rf.Controls.Add(l3);
for (int kl = 0; kl < count; kl++)
{
childnodes1 = xRoot1.SelectNodes("references/reference[" + (kl + 1).ToString() + "]");
wws = childnodes1.Item(0).InnerText;
a[y].citations_full += wws;
a[y].citations_full += "\n\n";
}
rf.Controls[1].Text = a[y].title_rus;
rf.Controls[0].Text = a[y].citations_full;
rf.Controls[4].Text = a[y].title_en;
a[y].citations_full = a[y].citations_full.Replace("\n", "<br>");
if ((f.Controls[0] as CheckBox).Checked)
{
string src = "ojs.ru/files/pdfs/" + issue_name + "/" + a[y].pages + ".pdf";
string title_ru = a[y].title_rus;
string title_en = a[y].title_en;
string refs = ya.translit_toEn(a[y].citations_full);
db_adapter db2 = new db_adapter();
db2.send_trase_refs(title_ru, title_en, refs);
db2.send_trase_src(title_ru, src);
db2.send_trase_src(title_en, src);
}
else rf.ShowDialog();
y++;
}
f.Controls[4].Text += "\n Парсинг окончен \n ";
List<string> sect_list = new List<string>();
string buff = a[0].section;
for (int i = 1; i < a.Count; i++)
{
if (a[i].section != buff)
{
sect_list.Add(buff);
buff = a[i].section;
}
}
sect_list.Add(buff);
// теперь удаляем все файлы
for (int vf = 0; vf < amount; vf++)
{
FileInfo file2 = new FileInfo(vf.ToString() + ".xml");
file2.Delete();
}
///////////////////////////////////////////////////////////////////////
// Из локального в таблицы БД trase_*.
///////////////////////////////////////////////////////////////////////
db_adapter db = new db_adapter();
int co = db.get_max_author_id();
for (int i = 0; i < amount; i++) {
for (int h = 0; h < a[i].authors.Count; h++) {
co++;
db.send_trase_title(a[i].title_rus, co);
db.send_trase_author(co, a[i].authors[h].fullname_ru, a[i].authors[h].biography_ru, a[i].authors[h].email, a[i].authors[h].affiliation_ru);
co++;
db.send_trase_title(a[i].title_en, co);
db.send_trase_author(co, a[i].authors[h].fullname_en, a[i].authors[h].biography_en, a[i].authors[h].email, a[i].authors[h].affiliation_en);
}
}
///////////////////////////////////////////////////////////////////////
// Из локального в результирующий.
///////////////////////////////////////////////////////////////////////
XDocument xd = XDocument.Load("result_temp.xml");
XElement ro = xd.Element("issues").Element("issue");
for (int i = 0; i < sect_list.Count; i++)
{
ro.Add(
new XElement("section",
new XElement("title",
new XAttribute("locale", "ru_RU"), sect_list[i]),
new XElement("abbrev",
new XAttribute("locale", "ru_RU"), ya.abbrev(sect_list[i]))
)
);
}
xd.Save("result_temp.xml");
XElement current_article;
for (int z = 0; z < amount; z++)
{
ya.buf = a[z].section;
IEnumerable<XElement> sections = xd.Element("issues").Element("issue").Elements("section");
Func<XElement, bool> rf = ya.fp;
XElement true_sect = sections.First<XElement>(rf);
true_sect.Add(
current_article = new XElement("article",
new XAttribute("locale", "ru_RU"),
new XAttribute("language", "ru"),
new XElement("title",
new XAttribute("locale", "ru_RU"), a[z].title_rus),
new XElement("title",
new XAttribute("locale", "en_US"), a[z].title_en),
new XElement("abstract",
new XAttribute("locale", "ru_RU"), a[z].abstract_rus),
new XElement("abstract",
new XAttribute("locale", "en_US"), a[z].abstract_en))
);
xd.Save("result_temp.xml");
for (int x = 0; x < a[z].authors.Count; x++)
{
current_article.Add(
new XElement("author",
new XAttribute("primary_contact", "true"),
new XElement("firstname", a[z].authors[x].firstname),
new XElement("middlename", a[z].authors[x].middlename),
new XElement("lastname", a[z].authors[x].lastname),
new XElement("affiliation",
new XAttribute("locale", "ru_RU"), a[z].authors[x].affiliation_ru),
new XElement("affiliation",
new XAttribute("locale", "en_US"), a[z].authors[x].affiliation_en),
new XElement("country", a[z].authors[x].country),
new XElement("email", a[z].authors[x].email),
new XElement("biography", a[z].authors[x].biography_ru)
)
);
}
string keys_ru = "";
string keys_en = "";
for (int x = 0; x < a[z].indexing_subject.Count; x++)
{
if (ya.is_eng(a[z].indexing_subject[x]))
{
keys_en += a[z].indexing_subject[x];
if (x != a[z].indexing_subject.Count - 1) keys_en += ", ";
}
else
{
keys_ru += a[z].indexing_subject[x];
if (x != a[z].indexing_subject.Count - 1) keys_ru += ", ";
}
}
current_article.Add(
new XElement("pages", a[z].pages),
new XElement("date_published", a[z].date_published),
new XElement("indexing",
new XElement("subject",
new XAttribute("locale", "en_US"), keys_en),
new XElement("subject",
new XAttribute("locale", "ru_RU"), keys_ru
)
)
);
current_article.Add(
new XElement("permissions",
new XElement("copyright_holder",
new XAttribute("locale", "ru_RU"), a[z].copyright_holder),
new XElement("copyright_year", a[z].copyright_year)
)
);
//string src = "Z:/home/ojs.ru/www/files/pdfs/" + issue_name + "/" + a[z].pages + ".pdf";
////string src = "ojs.ru/files/pdfs/" + issue_name + "/" + a[z].pages + ".pdf";
//current_article.Add(
// new XElement("galley",
// new XAttribute("locale", "ru_RU"),
// new XElement("label", "PDF"),
// new XElement("file",
// new XElement("href",
// new XAttribute("src", src),
// new XAttribute("mime_type","application/pdf")
// )
// )
// )
// );
}
xd.Save("result_temp.xml");
f.Controls[4].Text += "\n Создание нового документа окончено. Началась работа с БД \n ";
///////////////////////////////////////////////////////////////////////
SaveFileDialog saveFD = new SaveFileDialog();
saveFD.Filter = "OJS-Xml файл | *.xml";
saveFD.ShowDialog();
if (saveFD.FileName != "")
{
try
{
file.MoveTo(saveFD.FileName);
f.Controls[4].Text += "Парсинг завершен. Файл перемещен в указанный путь. \n";
}
catch (Exception e)
{
MessageBox.Show(e.Message);
}
}
else
{
f.Controls[4].Text += "Ошибка! Отменено сохранение файла.\n Нужно обработать исходный файл снова! \n";
FileInfo file3 = new FileInfo("result_xml.xml");
file3.Delete();
}
MessageBox.Show("Импортируйте полученные данные в OJS и закройте это окно");
for (int i = 0; i < amount; i++)
{
db.send_citations(a[i].title_rus, a[i].citations_full);
}
f.Controls[4].Text = "Успешно! \n";
}
