/// <summary>Gets a page from a Url.</summary>
public EntryPage GetFrom(LiveJournalTarget url, ILJClientData clientData)
{
// Gets the string.
string content = Client.GetContent(url, clientData);
// Parses the string as an entry page.
EntryPage p = _parser.ParseAsAnEntryPage(content);
return(p);
}
public EntryPage[] LoadOtherCommentPages(CommentPages commentPages, ILJClientData clientData)
{
int initialIndex = commentPages.Current;
int total = commentPages.Total;
log.Info(
string.Format(
"Loading other comment pages given page №{0} out of {1}."
, commentPages.Current
, commentPages.Total
)
);
// We need to download these.
int[] need = Enumerable.Range(1, total)
.Where(i => i != initialIndex)
.ToArray();
IDictionary <int, LiveJournalTarget> targets = new SortedDictionary <int, LiveJournalTarget>();
IDictionary <int, EntryPage> pages = new SortedDictionary <int, EntryPage>();
EntryPage p;
CommentPages latest = commentPages;
while (pages.Count < need.Length)
{
int cur = latest.Current;
if (cur != 1 && !string.IsNullOrWhiteSpace(latest.FirstUrl))
{
targets[1] = LiveJournalTarget.FromString(latest.FirstUrl);
}
if (cur != total && !string.IsNullOrWhiteSpace(latest.LastUrl))
{
targets[total] = LiveJournalTarget.FromString(latest.LastUrl);
}
if (!string.IsNullOrWhiteSpace(latest.PrevUrl))
{
targets[cur - 1] = LiveJournalTarget.FromString(latest.PrevUrl);
}
if (!string.IsNullOrWhiteSpace(latest.NextUrl))
{
targets[cur + 1] = LiveJournalTarget.FromString(latest.NextUrl);
}
// First target without a page.
int keyToDownload = targets.Keys.First(z => z != initialIndex && !pages.ContainsKey(z));
log.Info(string.Format("Will download page №{0}.", keyToDownload));
LiveJournalTarget targetToDownload = targets[keyToDownload];
// Download the page.
string content = _client.GetContent(targetToDownload, clientData);
p = _parser.ParseAsAnEntryPage(content);
latest = p.CommentPages;
pages[keyToDownload] = p;
log.Info(string.Format("Parsed page №{0}.", keyToDownload));
}
EntryPage[] ret = pages.Values.ToArray();
return(ret);
}
public EntryPage Work(string URI, string rootLocation, IFolderNamingStrategy subFolderGetter, string cookie)
{
LiveJournalTarget t = LiveJournalTarget.FromString(URI);
ILJClientData cookieData = _ext.Client.CreateDataObject(cookie);
// Get fresh version.
log.InfoFormat("Extracting {0}...", t);
EntryPage freshSource = _ext.GetFrom(t, cookieData);
string innerFolder;
IEntryBase freshSourceEntry = freshSource.Entry;
if (!subFolderGetter.TryGetSubfolderByEntry(freshSourceEntry, out innerFolder))
{
string error = string.Format(
"Cannot extract number from entry {0}, \"{1}\"."
, freshSourceEntry.Id
, freshSourceEntry.Subject
);
throw new NotSupportedException(error);
}
string subFolder = string.Format("{0}\\{1}", freshSource.Entry.Date.Value.Year, innerFolder);
string workLocation = _fs.Path.Combine(rootLocation, subFolder);
log.Info("Will work from " + workLocation);
EntryPage ep = null;
string dumpFile = _fs.Path.Combine(workLocation, DumpFileName);
if (_fs.File.Exists(dumpFile))
{
log.Info("File " + DumpFileName + " exists, will load it...");
ep = _lp.ParseAsAnEntryPage(_fs.File.ReadAllText(dumpFile));
}
else
{
log.Info("File " + DumpFileName + " does not exist.");
}
bool needsSaving = _ext.AbsorbAllData(freshSource, cookieData, ref ep);
log.Info("Will save changes: " + needsSaving + ".");
if (needsSaving)
{
// Save the content as is.
string content = _lp.Serialize(ep);
_fs.Directory.CreateDirectory(workLocation);
UTF8Encoding enc = new UTF8Encoding(true);
_fs.File.WriteAllText(dumpFile, content, enc);
// Pick usable comments.
List <Comment[]> comments = _scp.Pick(ep);
log.Info("Picked threads: " + comments.Count + ".");
// Everything we want to store.
var allData = new List <EntryBase>();
allData.Add(ep.Entry);
allData.AddRange(comments.SelectMany(a => a));
log.Info("Making sure everything is saved.");
_rds.EnsureAllIsSaved(allData, rootLocation, workLocation);
}
log.Info("Finished.");
return(ep);
}
