/**************************************************************************/
private void BuildWorksheetSitemapErrors(
MacroscopeJobMaster JobMaster,
CsvWriter ws
)
{
MacroscopeDocumentCollection DocCollection = JobMaster.GetDocCollection();
MacroscopeAllowedHosts AllowedHosts = JobMaster.GetAllowedHosts();
{
ws.WriteField("Sitemap URL");
ws.WriteField("Status Code");
ws.WriteField("Robots");
ws.WriteField("URL");
ws.NextRecord();
}
foreach (MacroscopeDocument msDoc in DocCollection.IterateDocuments())
{
if (msDoc.GetIsInternal() && msDoc.IsDocumentType(Type: MacroscopeConstants.DocumentType.SITEMAPXML))
{
foreach (MacroscopeLink Outlink in msDoc.IterateOutlinks())
{
string TargetUrl = Outlink.GetTargetUrl();
MacroscopeDocument msDocLinked = DocCollection.GetDocumentByUrl(Url: TargetUrl);
bool InsertRow = false;
if (msDocLinked.GetIsInternal())
{
int StatusCode = (int)msDocLinked.GetStatusCode();
if ((StatusCode >= 400) && (StatusCode <= 599))
{
InsertRow = true;
}
if (!msDocLinked.GetAllowedByRobots())
{
InsertRow = true;
}
}
if (InsertRow)
{
this.InsertAndFormatUrlCell(ws, msDoc);
this.InsertAndFormatStatusCodeCell(ws, msDoc);
this.InsertAndFormatRobotsCell(ws, msDoc);
this.InsertAndFormatUrlCell(ws, TargetUrl);
ws.NextRecord();
}
}
}
}
}
/**************************************************************************/
private void RenderListView(
MacroscopeDocumentCollection DocCollection,
List <string> UrlList,
MacroscopeDataExtractorCssSelectors DataExtractor
)
{
MacroscopeAllowedHosts AllowedHosts = this.MainForm.GetJobMaster().GetAllowedHosts();
if (DocCollection.CountDocuments() == 0)
{
return;
}
List <ListViewItem> ListViewItems = new List <ListViewItem> ();
MacroscopeSinglePercentageProgressForm ProgressForm = new MacroscopeSinglePercentageProgressForm(this.MainForm);
decimal Count = 0;
decimal TotalDocs = ( decimal )DocCollection.CountDocuments();
decimal MajorPercentage = (( decimal )100 / TotalDocs) * Count;
if (MacroscopePreferencesManager.GetShowProgressDialogues())
{
ProgressForm.UpdatePercentages(
Title: "Preparing Display",
Message: "Processing document collection for display:",
MajorPercentage: MajorPercentage,
ProgressLabelMajor: string.Format("Document {0} / {1}", Count, TotalDocs)
);
}
foreach (string Url in UrlList)
{
MacroscopeDocument msDoc = DocCollection.GetDocumentByUrl(Url: Url);
string DocUrl = msDoc.GetUrl();
string StatusCode = (( int )msDoc.GetStatusCode()).ToString();
string Status = msDoc.GetStatusCode().ToString();
string MimeType = msDoc.GetMimeType();
if (!DataExtractor.CanApplyDataExtractorsToDocument(msDoc: msDoc))
{
continue;
}
foreach (KeyValuePair <string, string> DataExtractedPair in msDoc.IterateDataExtractedCssSelectors())
{
ListViewItem lvItem = null;
string CssSelectorLabel = DataExtractedPair.Key;
string ExtractedValue = DataExtractedPair.Value;
string PairKey = null;
if (
string.IsNullOrEmpty(CssSelectorLabel) ||
string.IsNullOrEmpty(ExtractedValue))
{
continue;
}
PairKey = string.Join(
":",
UrlToDigest(DocUrl),
UrlToDigest(Macroscope.GetStringDigest(Text: CssSelectorLabel)),
UrlToDigest(Macroscope.GetStringDigest(Text: ExtractedValue))
);
if (this.DisplayListView.Items.ContainsKey(PairKey))
{
lvItem = this.DisplayListView.Items[PairKey];
}
else
{
lvItem = new ListViewItem(PairKey);
lvItem.UseItemStyleForSubItems = false;
lvItem.Name = PairKey;
for (int i = 0; i < 6; i++)
{
lvItem.SubItems.Add("");
}
ListViewItems.Add(lvItem);
}
if (lvItem != null)
{
try
{
lvItem.SubItems[ColUrl].Text = DocUrl;
lvItem.SubItems[ColStatusCode].Text = StatusCode;
lvItem.SubItems[ColStatus].Text = Status;
lvItem.SubItems[ColMimeType].Text = MimeType;
lvItem.SubItems[ColCssSelectorLabel].Text = CssSelectorLabel;
lvItem.SubItems[ColExtractedValue].Text = ExtractedValue;
}
catch (Exception ex)
{
DebugMsg(string.Format("MacroscopeDisplayDataExtractorCssSelectors: {0}", ex.Message));
DebugMsg(string.Format("MacroscopeDisplayDataExtractorCssSelectors: {0}", ex.StackTrace));
}
}
else
{
DebugMsg(string.Format("MacroscopeDisplayDataExtractorCssSelectors MISSING: {0}", PairKey));
}
if (msDoc.GetIsInternal())
{
lvItem.SubItems[ColUrl].ForeColor = Color.Green;
}
else
{
lvItem.SubItems[ColUrl].ForeColor = Color.Gray;
}
if (Regex.IsMatch(StatusCode, "^[2]"))
{
lvItem.SubItems[ColStatusCode].ForeColor = Color.Green;
lvItem.SubItems[ColStatus].ForeColor = Color.Green;
}
else
if (Regex.IsMatch(StatusCode, "^[3]"))
{
lvItem.SubItems[ColStatusCode].ForeColor = Color.Goldenrod;
lvItem.SubItems[ColStatus].ForeColor = Color.Goldenrod;
}
else
if (Regex.IsMatch(StatusCode, "^[45]"))
{
lvItem.SubItems[ColStatusCode].ForeColor = Color.Red;
lvItem.SubItems[ColStatus].ForeColor = Color.Red;
}
else
{
lvItem.SubItems[ColStatusCode].ForeColor = Color.Blue;
lvItem.SubItems[ColStatus].ForeColor = Color.Blue;
}
}
if (MacroscopePreferencesManager.GetShowProgressDialogues())
{
Count++;
MajorPercentage = (( decimal )100 / TotalDocs) * Count;
ProgressForm.UpdatePercentages(
Title: null,
Message: null,
MajorPercentage: MajorPercentage,
ProgressLabelMajor: string.Format("Document {0} / {1}", Count, TotalDocs)
);
}
}
this.DisplayListView.Items.AddRange(ListViewItems.ToArray());
this.DisplayListView.AutoResizeColumns(ColumnHeaderAutoResizeStyle.ColumnContent);
this.DisplayListView.Columns[ColUrl].Width = 300;
this.DisplayListView.Columns[ColStatusCode].Width = 100;
this.DisplayListView.Columns[ColStatus].Width = 100;
this.DisplayListView.Columns[ColMimeType].Width = 100;
if (MacroscopePreferencesManager.GetShowProgressDialogues())
{
ProgressForm.DoClose();
}
if (ProgressForm != null)
{
ProgressForm.Dispose();
}
}
/**************************************************************************/
private void RenderListViewSearchTargetUrls(
List <ListViewItem> ListViewItems,
MacroscopeDocument msDoc,
string Url,
string UrlFragment
)
{
MacroscopeAllowedHosts AllowedHosts = this.MainForm.GetJobMaster().GetAllowedHosts();
MacroscopeHyperlinksOut HyperlinksOut = msDoc.GetHyperlinksOut();
MacroscopeDocumentCollection DocCollection = this.MainForm.GetJobMaster().GetDocCollection();
foreach (MacroscopeHyperlinkOut HyperlinkOut in HyperlinksOut.IterateLinks())
{
string UrlTarget = HyperlinkOut.GetTargetUrl();
HttpStatusCode StatusCode = HttpStatusCode.NotFound;
string StatusCodeText = "Not crawled";
string StatusText = "Not crawled";
string PairKey = string.Join(":", UrlToDigest(Url: Url), UrlToDigest(Url: UrlTarget)).ToString();
string LinkTarget = HyperlinkOut.GetLinkTarget();
string LinkText = HyperlinkOut.GetAnchorText();
string LinkTitle = HyperlinkOut.GetTitle();
string AltText = HyperlinkOut.GetAltText();
string LinkTextLabel = LinkText;
string LinkTitleLabel = LinkTitle;
string AltTextLabel = AltText;
string DoFollow = "No Follow";
try
{
if (DocCollection.ContainsDocument(Url: HyperlinkOut.GetTargetUrl()))
{
StatusCode = DocCollection.GetDocumentByUrl(Url: HyperlinkOut.GetTargetUrl()).GetStatusCode();
StatusCodeText = ((int)StatusCode).ToString();
StatusText = StatusCode.ToString();
}
else
{
DebugMsg("Not in DocCollection");
}
}
catch (Exception ex)
{
this.DebugMsg(ex.Message);
}
if (HyperlinkOut.GetDoFollow())
{
DoFollow = "Follow";
}
if (LinkText.Length == 0)
{
LinkTextLabel = "MISSING";
}
if (LinkTitle.Length == 0)
{
LinkTitleLabel = "MISSING";
}
if (AltText.Length == 0)
{
AltTextLabel = "MISSING";
}
if (
(UrlTarget != null) &&
(UrlTarget.IndexOf(UrlFragment, StringComparison.CurrentCulture) >= 0))
{
ListViewItem lvItem = null;
if (this.DisplayListView.Items.ContainsKey(PairKey))
{
try
{
lvItem = this.DisplayListView.Items[PairKey];
lvItem.SubItems[ColUrl].Text = Url;
lvItem.SubItems[ColUrlTarget].Text = UrlTarget;
lvItem.SubItems[ColStatusCode].Text = StatusCodeText;
lvItem.SubItems[ColStatus].Text = StatusText;
lvItem.SubItems[ColDoFollow].Text = DoFollow;
lvItem.SubItems[ColLinkTarget].Text = LinkTarget;
lvItem.SubItems[ColLinkAnchorTextLabel].Text = LinkTextLabel;
lvItem.SubItems[ColLinkTitleLabel].Text = LinkTitleLabel;
lvItem.SubItems[ColAltTextLabel].Text = AltTextLabel;
}
catch (Exception ex)
{
this.DebugMsg(string.Format("MacroscopeDisplayLinks 1: {0}", ex.Message));
}
}
else
{
try
{
lvItem = new ListViewItem(PairKey);
lvItem.UseItemStyleForSubItems = false;
lvItem.Name = PairKey;
lvItem.SubItems[ColUrl].Text = Url;
lvItem.SubItems.Add(UrlTarget);
lvItem.SubItems.Add(StatusCodeText);
lvItem.SubItems.Add(StatusText);
lvItem.SubItems.Add(DoFollow);
lvItem.SubItems.Add(LinkTarget);
lvItem.SubItems.Add(LinkTextLabel);
lvItem.SubItems.Add(LinkTitleLabel);
lvItem.SubItems.Add(AltTextLabel);
ListViewItems.Add(lvItem);
}
catch (Exception ex)
{
this.DebugMsg(string.Format("MacroscopeDisplayLinks 2: {0}", ex.Message));
}
}
if (lvItem != null)
{
for (int i = 0; i < lvItem.SubItems.Count; i++)
{
lvItem.SubItems[i].ForeColor = Color.Blue;
}
if (AllowedHosts.IsAllowedFromUrl(Url))
{
lvItem.SubItems[ColUrl].ForeColor = Color.Green;
}
else
{
lvItem.SubItems[ColUrl].ForeColor = Color.Gray;
}
if (AllowedHosts.IsAllowedFromUrl(UrlTarget))
{
lvItem.SubItems[ColUrlTarget].ForeColor = Color.Green;
}
else
{
lvItem.SubItems[ColUrlTarget].ForeColor = Color.Gray;
}
if (AllowedHosts.IsAllowedFromUrl(Url))
{
if (HyperlinkOut.GetDoFollow())
{
lvItem.SubItems[ColDoFollow].ForeColor = Color.Green;
}
else
{
lvItem.SubItems[ColDoFollow].ForeColor = Color.Red;
}
}
else
{
lvItem.SubItems[ColDoFollow].ForeColor = Color.Gray;
}
if (LinkText.Length == 0)
{
lvItem.SubItems[ColLinkAnchorTextLabel].ForeColor = Color.Gray;
}
if (LinkTitle.Length == 0)
{
lvItem.SubItems[ColLinkTitleLabel].ForeColor = Color.Gray;
}
if (AltText.Length == 0)
{
lvItem.SubItems[ColAltTextLabel].ForeColor = Color.Gray;
}
if (
(LinkText.Length == 0) &&
(LinkTitle.Length == 0) &&
(AltText.Length == 0))
{
lvItem.SubItems[ColLinkAnchorTextLabel].ForeColor = Color.Red;
lvItem.SubItems[ColLinkTitleLabel].ForeColor = Color.Red;
lvItem.SubItems[ColAltTextLabel].ForeColor = Color.Red;
}
}
}
}
}
/**************************************************************************/
private void RenderTreeView(
MacroscopeDocumentCollection DocCollection,
List <string> UrlList
)
{
if (DocCollection.CountDocuments() == 0)
{
return;
}
if (UrlList.Count == 0)
{
return;
}
MacroscopeSinglePercentageProgressForm ProgressForm = new MacroscopeSinglePercentageProgressForm(this.MainForm);
decimal Count = 0;
decimal TotalDocs = ( decimal )UrlList.Count;
decimal MajorPercentage = (( decimal )100 / TotalDocs) * Count;
if (MacroscopePreferencesManager.GetShowProgressDialogues())
{
ProgressForm.UpdatePercentages(
Title: "Preparing Display",
Message: "Processing document collection for display:",
MajorPercentage: MajorPercentage,
ProgressLabelMajor: string.Format("Document {0} / {1}", Count, TotalDocs)
);
}
this.tvTreeView.BeginUpdate();
DebugMsg(string.Format("HIERARCHY: {0}", "BASE"));
foreach (string Url in UrlList)
{
MacroscopeDocument msDoc = DocCollection.GetDocumentByUrl(Url: Url);
if (msDoc != null)
{
this.RenderTreeView(msDoc, Url);
}
Count++;
MajorPercentage = (( decimal )100 / TotalDocs) * Count;
ProgressForm.UpdatePercentages(
Title: null,
Message: null,
MajorPercentage: MajorPercentage,
ProgressLabelMajor: string.Format("Document {0} / {1}", Count, TotalDocs)
);
}
this.tvTreeView.ExpandAll();
this.tvTreeView.EndUpdate();
if (MacroscopePreferencesManager.GetShowProgressDialogues())
{
ProgressForm.DoClose();
}
if (ProgressForm != null)
{
ProgressForm.Dispose();
}
}
/**************************************************************************/
private void RenderListView(
MacroscopeDocumentCollection DocCollection,
List <string> UrlList,
MacroscopeCustomFilters CustomFilter
)
{
if (this.FilterColOffset == -1)
{
throw (new Exception("this.FilterColOffset invalid"));
}
if (DocCollection.CountDocuments() == 0)
{
return;
}
MacroscopeAllowedHosts AllowedHosts = this.MainForm.GetJobMaster().GetAllowedHosts();
Dictionary <string, int> FilterColsTable = new Dictionary <string, int>(CustomFilter.GetSize());
List <ListViewItem> ListViewItems = new List <ListViewItem>();
MacroscopeSinglePercentageProgressForm ProgressForm = new MacroscopeSinglePercentageProgressForm(this.MainForm);
decimal Count = 0;
decimal TotalDocs = (decimal)DocCollection.CountDocuments();
decimal MajorPercentage = ((decimal)100 / TotalDocs) * Count;
if (MacroscopePreferencesManager.GetShowProgressDialogues())
{
ProgressForm.UpdatePercentages(
Title: "Preparing Display",
Message: "Processing document collection for display:",
MajorPercentage: MajorPercentage,
ProgressLabelMajor: string.Format("Document {0} / {1}", Count, TotalDocs)
);
}
for (int Slot = 0; Slot < CustomFilter.GetSize(); Slot++)
{
string FilterPattern = CustomFilter.GetPattern(Slot).Key;
if (FilterColsTable.ContainsKey(FilterPattern))
{
FilterColsTable.Add(string.Format("EMPTY{0}", Slot + 1), Slot + 1);
}
else
{
FilterColsTable.Add(FilterPattern, Slot + 1);
}
}
foreach (string Url in UrlList)
{
MacroscopeDocument msDoc = DocCollection.GetDocumentByUrl(Url: Url);
ListViewItem lvItem = null;
string DocUrl;
string PairKey;
string StatusCode;
string Status;
string MimeType;
if (msDoc == null)
{
continue;
}
else
{
DocUrl = msDoc.GetUrl();
PairKey = DocUrl;
StatusCode = ((int)msDoc.GetStatusCode()).ToString();
Status = msDoc.GetStatusCode().ToString();
MimeType = msDoc.GetMimeType();
}
if (!CustomFilter.CanApplyCustomFiltersToDocument(msDoc: msDoc))
{
continue;
}
if (this.DisplayListView.Items.ContainsKey(PairKey))
{
lvItem = this.DisplayListView.Items[PairKey];
}
else
{
lvItem = new ListViewItem(PairKey);
lvItem.UseItemStyleForSubItems = false;
lvItem.Name = PairKey;
lvItem.SubItems.Add("");
lvItem.SubItems.Add("");
lvItem.SubItems.Add("");
lvItem.SubItems.Add("");
for (int Slot = 0; Slot < CustomFilter.GetSize(); Slot++)
{
lvItem.SubItems.Add("");
}
ListViewItems.Add(lvItem);
}
if (lvItem != null)
{
try
{
lvItem.SubItems[ColUrl].Text = DocUrl;
lvItem.SubItems[ColStatusCode].Text = StatusCode;
lvItem.SubItems[ColStatus].Text = Status;
lvItem.SubItems[ColMimeType].Text = MimeType;
for (int Slot = 0; Slot < CustomFilter.GetSize(); Slot++)
{
string FilterPattern = CustomFilter.GetPattern(Slot: Slot).Key;
KeyValuePair <string, MacroscopeConstants.TextPresence> Pair = msDoc.GetCustomFilteredItem(Text: FilterPattern);
int ColOffset = this.FilterColOffset + FilterColsTable[FilterPattern];
if ((Pair.Key != null) && (Pair.Value != MacroscopeConstants.TextPresence.UNDEFINED))
{
lvItem.SubItems[ColOffset].Text = MacroscopeConstants.TextPresenceLabels[Pair.Value];
switch (Pair.Value)
{
case MacroscopeConstants.TextPresence.CONTAINS_STRING:
lvItem.SubItems[ColOffset].ForeColor = Color.Green;
break;
case MacroscopeConstants.TextPresence.NOT_CONTAINS_STRING:
lvItem.SubItems[ColOffset].ForeColor = Color.Green;
break;
case MacroscopeConstants.TextPresence.MUST_CONTAIN_STRING:
lvItem.SubItems[ColOffset].ForeColor = Color.Red;
break;
case MacroscopeConstants.TextPresence.SHOULD_NOT_CONTAIN_STRING:
lvItem.SubItems[ColOffset].ForeColor = Color.Red;
break;
case MacroscopeConstants.TextPresence.CONTAINS_REGEX:
lvItem.SubItems[ColOffset].ForeColor = Color.Green;
break;
case MacroscopeConstants.TextPresence.NOT_CONTAINS_REGEX:
lvItem.SubItems[ColOffset].ForeColor = Color.Green;
break;
case MacroscopeConstants.TextPresence.MUST_CONTAIN_REGEX:
lvItem.SubItems[ColOffset].ForeColor = Color.Red;
break;
case MacroscopeConstants.TextPresence.SHOULD_NOT_CONTAIN_REGEX:
lvItem.SubItems[ColOffset].ForeColor = Color.Red;
break;
default:
lvItem.SubItems[ColOffset].ForeColor = Color.Gray;
break;
}
}
}
}
catch (Exception ex)
{
DebugMsg(string.Format("MacroscopeDisplayCustomFilters: {0}", ex.Message));
DebugMsg(string.Format("MacroscopeDisplayCustomFilters: {0}", ex.StackTrace));
}
}
else
{
DebugMsg(string.Format("MacroscopeDisplayCustomFilters MISSING: {0}", PairKey));
}
if (msDoc.GetIsInternal())
{
lvItem.SubItems[ColUrl].ForeColor = Color.Green;
}
else
{
lvItem.SubItems[ColUrl].ForeColor = Color.Gray;
}
if (Regex.IsMatch(StatusCode, "^[2]"))
{
lvItem.SubItems[ColStatusCode].ForeColor = Color.Green;
lvItem.SubItems[ColStatus].ForeColor = Color.Green;
}
else
if (Regex.IsMatch(StatusCode, "^[3]"))
{
lvItem.SubItems[ColStatusCode].ForeColor = Color.Goldenrod;
lvItem.SubItems[ColStatus].ForeColor = Color.Goldenrod;
}
else
if (Regex.IsMatch(StatusCode, "^[45]"))
{
lvItem.SubItems[ColStatusCode].ForeColor = Color.Red;
lvItem.SubItems[ColStatus].ForeColor = Color.Red;
}
else
{
lvItem.SubItems[ColStatusCode].ForeColor = Color.Blue;
lvItem.SubItems[ColStatus].ForeColor = Color.Blue;
}
if (MacroscopePreferencesManager.GetShowProgressDialogues())
{
Count++;
MajorPercentage = ((decimal)100 / TotalDocs) * Count;
ProgressForm.UpdatePercentages(
Title: null,
Message: null,
MajorPercentage: MajorPercentage,
ProgressLabelMajor: string.Format("Document {0} / {1}", Count, TotalDocs)
);
}
}
this.DisplayListView.Items.AddRange(ListViewItems.ToArray());
this.DisplayListView.AutoResizeColumns(ColumnHeaderAutoResizeStyle.HeaderSize);
this.DisplayListView.Columns[ColUrl].Width = 300;
this.DisplayListView.Columns[ColStatusCode].Width = 100;
this.DisplayListView.Columns[ColStatus].Width = 100;
this.DisplayListView.Columns[ColMimeType].Width = 100;
if (MacroscopePreferencesManager.GetShowProgressDialogues())
{
ProgressForm.DoClose();
}
if (ProgressForm != null)
{
ProgressForm.Dispose();
}
}
/** Render List ***********************************************************/
public void RenderListView(
MacroscopeDocumentCollection DocCollection,
List <string> UrlList
)
{
if (DocCollection.CountDocuments() == 0)
{
return;
}
if (UrlList.Count == 0)
{
return;
}
List <ListViewItem> ListViewItems = new List <ListViewItem>(UrlList.Count);
MacroscopeSinglePercentageProgressForm ProgressForm = new MacroscopeSinglePercentageProgressForm(this.MainForm);
decimal Count = 0;
decimal TotalDocs = (decimal)UrlList.Count;
decimal MajorPercentage = ((decimal)100 / TotalDocs) * Count;
if (MacroscopePreferencesManager.GetShowProgressDialogues())
{
ProgressForm.ControlBox = false;
ProgressForm.UpdatePercentages(
Title: "Preparing Display",
Message: "Processing document collection for display:",
MajorPercentage: MajorPercentage,
ProgressLabelMajor: string.Format("Document {0} / {1}", Count, TotalDocs)
);
}
lock (this.DisplayListViewLock)
{
foreach (string Url in UrlList)
{
Application.DoEvents();
MacroscopeDocument msDoc = DocCollection.GetDocumentByUrl(Url: Url);
if (msDoc != null)
{
this.RenderListView(
ListViewItems: ListViewItems,
DocCollection: DocCollection,
msDoc: msDoc,
Url: Url
);
}
if (MacroscopePreferencesManager.GetShowProgressDialogues())
{
Count++;
MajorPercentage = ((decimal)100 / TotalDocs) * Count;
ProgressForm.UpdatePercentages(
Title: null,
Message: null,
MajorPercentage: MajorPercentage,
ProgressLabelMajor: string.Format("Document {0} / {1}", Count, TotalDocs)
);
}
}
this.DisplayListView.Items.AddRange(ListViewItems.ToArray());
}
if (MacroscopePreferencesManager.GetShowProgressDialogues())
{
ProgressForm.DoClose();
}
if (ProgressForm != null)
{
ProgressForm.Dispose();
}
}
/**************************************************************************/
private void BuildWorksheetSitemapXmlErrors(
MacroscopeJobMaster JobMaster,
XLWorkbook wb,
string WorksheetLabel
)
{
var ws = wb.Worksheets.Add(WorksheetLabel);
int iRow = 1;
int iCol = 1;
int iColMax = 1;
MacroscopeDocumentCollection DocCollection = JobMaster.GetDocCollection();
MacroscopeAllowedHosts AllowedHosts = JobMaster.GetAllowedHosts();
{
ws.Cell(iRow, iCol).Value = "Sitemap URL";
iCol++;
ws.Cell(iRow, iCol).Value = "Status Code";
iCol++;
ws.Cell(iRow, iCol).Value = "Robots";
iCol++;
ws.Cell(iRow, iCol).Value = "URL";
}
iColMax = iCol;
iRow++;
foreach (MacroscopeDocument msDoc in DocCollection.IterateDocuments())
{
if (msDoc.GetIsInternal() && msDoc.IsDocumentType(Type: MacroscopeConstants.DocumentType.SITEMAPXML))
{
foreach (MacroscopeLink Outlink in msDoc.IterateOutlinks())
{
string TargetUrl = Outlink.GetTargetUrl();
MacroscopeDocument msDocLinked = DocCollection.GetDocumentByUrl(Url: TargetUrl);
bool InsertRow = false;
if (msDocLinked.GetIsInternal())
{
int StatusCode = (int)msDocLinked.GetStatusCode();
if ((StatusCode >= 400) && (StatusCode <= 599))
{
InsertRow = true;
}
if (!msDocLinked.GetAllowedByRobots())
{
InsertRow = true;
}
}
if (InsertRow)
{
iCol = 1;
this.InsertAndFormatUrlCell(ws, iRow, iCol, msDoc);
if (AllowedHosts.IsInternalUrl(Url: msDoc.GetUrl()))
{
ws.Cell(iRow, iCol).Style.Font.SetFontColor(XLColor.Green);
}
else
{
ws.Cell(iRow, iCol).Style.Font.SetFontColor(XLColor.Gray);
}
iCol++;
this.InsertAndFormatStatusCodeCell(ws, iRow, iCol, msDoc);
iCol++;
this.InsertAndFormatRobotsCell(ws, iRow, iCol, msDoc);
iCol++;
this.InsertAndFormatUrlCell(ws, iRow, iCol, TargetUrl);
if (AllowedHosts.IsInternalUrl(Url: TargetUrl))
{
ws.Cell(iRow, iCol).Style.Font.SetFontColor(XLColor.Green);
}
else
{
ws.Cell(iRow, iCol).Style.Font.SetFontColor(XLColor.Gray);
}
iRow++;
}
}
}
}
{
var rangeData = ws.Range(1, 1, iRow - 1, iColMax);
var excelTable = rangeData.CreateTable();
}
}
/**************************************************************************/
private void BuildWorksheetPageDuplicatePages(
MacroscopeJobMaster JobMaster,
XLWorkbook wb,
string WorksheetLabel
)
{
var ws = wb.Worksheets.Add(WorksheetLabel);
int iRow = 1;
int iCol = 1;
int iColMax = 1;
decimal DocCount = 0;
decimal DocListCount = 0;
decimal CountOuter = 0;
decimal CountInner = 0;
MacroscopeDocumentCollection DocCollection = JobMaster.GetDocCollection();
MacroscopeAllowedHosts AllowedHosts = JobMaster.GetAllowedHosts();
Dictionary <string, bool> CrossCheckList;
CrossCheckList = MacroscopeLevenshteinAnalysis.GetCrossCheckList(
Capacity: DocCollection.CountDocuments()
);
DocCount = ( decimal )DocCollection.CountDocuments();
{
ws.Cell(iRow, iCol).Value = "Status Code";
iCol++;
ws.Cell(iRow, iCol).Value = "Status";
iCol++;
ws.Cell(iRow, iCol).Value = "Origin URL";
iCol++;
ws.Cell(iRow, iCol).Value = "Distance";
iCol++;
ws.Cell(iRow, iCol).Value = "Similar URL";
}
iColMax = iCol;
iRow++;
foreach (string UrlLeft in DocCollection.DocumentUrls())
{
MacroscopeDocument msDocLeft = DocCollection.GetDocumentByUrl(Url: UrlLeft);
MacroscopeLevenshteinAnalysis LevenshteinAnalysis = null;
CountOuter++;
CountInner = 0;
if (DocCount > 0)
{
this.ProgressForm.UpdatePercentages(
Title: null,
Message: null,
MajorPercentage: -1,
ProgressLabelMajor: string.Format("Documents Processed: {0}", CountOuter),
MinorPercentage: (( decimal )100 / DocCount) * CountOuter,
ProgressLabelMinor: UrlLeft,
SubMinorPercentage: 0,
ProgressLabelSubMinor: ""
);
}
if (msDocLeft.GetIsExternal())
{
continue;
}
if (!msDocLeft.IsDocumentType(Type: MacroscopeConstants.DocumentType.HTML))
{
continue;
}
LevenshteinAnalysis = new MacroscopeLevenshteinAnalysis(
msDoc: msDocLeft,
SizeDifference: MacroscopePreferencesManager.GetMaxLevenshteinSizeDifference(),
Threshold: MacroscopePreferencesManager.GetMaxLevenshteinDistance(),
CrossCheckList: CrossCheckList,
IPercentageDone: this
);
Dictionary <MacroscopeDocument, int> DocList;
DocList = LevenshteinAnalysis.AnalyzeDocCollection(
DocCollection: DocCollection
);
DocListCount = ( decimal )DocList.Count;
foreach (MacroscopeDocument msDocDuplicate in DocList.Keys)
{
int StatusCode = ( int )msDocLeft.GetStatusCode();
HttpStatusCode Status = msDocLeft.GetStatusCode();
string UrlDuplicate = msDocDuplicate.GetUrl();
int Distance = DocList[msDocDuplicate];
CountInner++;
iCol = 1;
if (DocCount > 0)
{
this.ProgressForm.UpdatePercentages(
Title: null,
Message: null,
MajorPercentage: -1,
ProgressLabelMajor: string.Format("Documents Processed: {0}", CountOuter),
MinorPercentage: (( decimal )100 / DocCount) * CountOuter,
ProgressLabelMinor: UrlLeft,
SubMinorPercentage: (( decimal )100 / DocListCount) * CountInner,
ProgressLabelSubMinor: UrlDuplicate
);
}
this.InsertAndFormatStatusCodeCell(ws, iRow, iCol, StatusCode);
iCol++;
this.InsertAndFormatStatusCodeCell(ws, iRow, iCol, Status);
iCol++;
this.InsertAndFormatUrlCell(ws, iRow, iCol, UrlLeft);
if (AllowedHosts.IsInternalUrl(Url: UrlLeft))
{
ws.Cell(iRow, iCol).Style.Font.SetFontColor(XLColor.Green);
}
else
{
ws.Cell(iRow, iCol).Style.Font.SetFontColor(XLColor.Gray);
}
iCol++;
this.InsertAndFormatContentCell(ws, iRow, iCol, Distance.ToString());
if (Distance <= MacroscopePreferencesManager.GetMaxLevenshteinDistance())
{
ws.Cell(iRow, iCol).Style.Font.SetFontColor(XLColor.Red);
}
else
{
ws.Cell(iRow, iCol).Style.Font.SetFontColor(XLColor.Green);
}
iCol++;
this.InsertAndFormatUrlCell(ws, iRow, iCol, UrlDuplicate);
if (AllowedHosts.IsInternalUrl(Url: UrlDuplicate))
{
ws.Cell(iRow, iCol).Style.Font.SetFontColor(XLColor.Green);
}
else
{
ws.Cell(iRow, iCol).Style.Font.SetFontColor(XLColor.Gray);
}
iRow++;
if (this.ProgressForm.Cancelled())
{
break;
}
}
if (this.ProgressForm.Cancelled())
{
break;
}
//Thread.Yield();
}
{
var rangeData = ws.Range(1, 1, iRow - 1, iColMax);
var excelTable = rangeData.CreateTable();
}
}
/**************************************************************************/
protected override void RenderListView(
List <ListViewItem> ListViewItems,
MacroscopeDocumentCollection DocCollection,
MacroscopeDocument msDoc,
string Url
)
{
MacroscopeAllowedHosts AllowedHosts = this.MainForm.GetJobMaster().GetAllowedHosts();
foreach (MacroscopeLink Link in msDoc.IterateOutlinks())
{
ListViewItem lvItem = null;
string LinkType = Link.GetLinkType().ToString();
string UrlTarget = Link.GetTargetUrl();
HttpStatusCode StatusCode = HttpStatusCode.NotFound;
string StatusCodeText = "Not crawled";
string StatusText = "Not crawled";
string PairKey = string.Join(":", UrlToDigest(Url: Url), UrlToDigest(Url: UrlTarget));
string DoFollow = "No Follow";
string AltText = Link.GetAltText();
string AltTextLabel = AltText;
string RawSourceUrl = Link.GetRawSourceUrl();
string RawTargetUrl = Link.GetRawTargetUrl();
try
{
if (DocCollection.ContainsDocument(Url: Link.GetTargetUrl()))
{
StatusCode = DocCollection.GetDocumentByUrl(Url: Link.GetTargetUrl()).GetStatusCode();
StatusCodeText = ((int)StatusCode).ToString();
StatusText = StatusCode.ToString();
}
}
catch (Exception ex)
{
this.DebugMsg(ex.Message);
}
if (Link.GetDoFollow())
{
DoFollow = "Follow";
}
if (string.IsNullOrEmpty(AltText))
{
AltTextLabel = "";
}
if (string.IsNullOrEmpty(RawSourceUrl))
{
RawSourceUrl = "";
}
if (string.IsNullOrEmpty(RawTargetUrl))
{
RawTargetUrl = "";
}
if (this.DisplayListView.Items.ContainsKey(PairKey))
{
try
{
lvItem = this.DisplayListView.Items[PairKey];
lvItem.SubItems[ColType].Text = LinkType;
lvItem.SubItems[ColUrl].Text = Url;
lvItem.SubItems[ColUrlTarget].Text = UrlTarget;
lvItem.SubItems[ColStatusCode].Text = StatusCodeText;
lvItem.SubItems[ColStatus].Text = StatusText;
lvItem.SubItems[ColDoFollow].Text = DoFollow;
lvItem.SubItems[ColAltTextLabel].Text = AltTextLabel;
lvItem.SubItems[ColRawSourceUrl].Text = RawSourceUrl;
lvItem.SubItems[ColRawTargetUrl].Text = RawTargetUrl;
}
catch (Exception ex)
{
this.DebugMsg(string.Format("MacroscopeDisplayLinks 1: {0}", ex.Message));
}
}
else
{
try
{
lvItem = new ListViewItem(PairKey);
lvItem.UseItemStyleForSubItems = false;
lvItem.Name = PairKey;
lvItem.SubItems[ColType].Text = LinkType;
lvItem.SubItems.Add(Url);
lvItem.SubItems.Add(UrlTarget);
lvItem.SubItems.Add(StatusCodeText);
lvItem.SubItems.Add(StatusText);
lvItem.SubItems.Add(DoFollow);
lvItem.SubItems.Add(AltTextLabel);
lvItem.SubItems.Add(RawSourceUrl);
lvItem.SubItems.Add(RawTargetUrl);
ListViewItems.Add(lvItem);
}
catch (Exception ex)
{
this.DebugMsg(string.Format("MacroscopeDisplayLinks 2: {0}", ex.Message));
}
}
if (lvItem != null)
{
for (int i = 0; i < lvItem.SubItems.Count; i++)
{
lvItem.SubItems[i].ForeColor = Color.Blue;
}
if (AllowedHosts.IsAllowedFromUrl(Url))
{
lvItem.SubItems[ColUrl].ForeColor = Color.Green;
}
else
{
lvItem.SubItems[ColUrl].ForeColor = Color.Gray;
}
if (AllowedHosts.IsAllowedFromUrl(UrlTarget))
{
lvItem.SubItems[ColUrlTarget].ForeColor = Color.Green;
}
else
{
lvItem.SubItems[ColUrlTarget].ForeColor = Color.Gray;
}
if (AllowedHosts.IsAllowedFromUrl(UrlTarget))
{
if (Link.GetDoFollow())
{
lvItem.SubItems[ColDoFollow].ForeColor = Color.Green;
}
else
{
lvItem.SubItems[ColDoFollow].ForeColor = Color.Red;
}
}
else
{
lvItem.SubItems[ColDoFollow].ForeColor = Color.Gray;
}
}
}
}
/**************************************************************************/
private void BuildWorksheetPageDuplicatePages(
MacroscopeJobMaster JobMaster,
CsvWriter ws
)
{
decimal DocCount = 0;
decimal DocListCount = 0;
decimal CountOuter = 0;
decimal CountInner = 0;
MacroscopeDocumentCollection DocCollection = JobMaster.GetDocCollection();
MacroscopeAllowedHosts AllowedHosts = JobMaster.GetAllowedHosts();
Dictionary <string, bool> CrossCheckList;
CrossCheckList = MacroscopeLevenshteinAnalysis.GetCrossCheckList(
Capacity: DocCollection.CountDocuments()
);
DocCount = ( decimal )DocCollection.CountDocuments();
{
ws.WriteField("Status Code");
ws.WriteField("Status");
ws.WriteField("Origin URL");
ws.WriteField("Distance");
ws.WriteField("Similar URL");
ws.NextRecord();
}
foreach (string UrlLeft in DocCollection.DocumentUrls())
{
MacroscopeDocument msDocLeft = DocCollection.GetDocumentByUrl(Url: UrlLeft);
MacroscopeLevenshteinAnalysis LevenshteinAnalysis = null;
CountOuter++;
CountInner = 0;
if (DocCount > 0)
{
this.ProgressForm.UpdatePercentages(
Title: null,
Message: null,
MajorPercentage: -1,
ProgressLabelMajor: string.Format("Documents Processed: {0}", CountOuter),
MinorPercentage: (( decimal )100 / DocCount) * CountOuter,
ProgressLabelMinor: UrlLeft,
SubMinorPercentage: 0,
ProgressLabelSubMinor: ""
);
}
if (msDocLeft.GetIsExternal())
{
continue;
}
if (!msDocLeft.IsDocumentType(Type: MacroscopeConstants.DocumentType.HTML))
{
continue;
}
LevenshteinAnalysis = new MacroscopeLevenshteinAnalysis(
msDoc: msDocLeft,
SizeDifference: MacroscopePreferencesManager.GetMaxLevenshteinSizeDifference(),
Threshold: MacroscopePreferencesManager.GetMaxLevenshteinDistance(),
CrossCheckList: CrossCheckList,
IPercentageDone: this
);
Dictionary <MacroscopeDocument, int> DocList;
DocList = LevenshteinAnalysis.AnalyzeDocCollection(
DocCollection: DocCollection
);
DocListCount = ( decimal )DocList.Count;
foreach (MacroscopeDocument msDocDuplicate in DocList.Keys)
{
int StatusCode = ( int )msDocLeft.GetStatusCode();
HttpStatusCode Status = msDocLeft.GetStatusCode();
string UrlDuplicate = msDocDuplicate.GetUrl();
int Distance = DocList[msDocDuplicate];
CountInner++;
if (DocCount > 0)
{
this.ProgressForm.UpdatePercentages(
Title: null,
Message: null,
MajorPercentage: -1,
ProgressLabelMajor: string.Format("Documents Processed: {0}", CountOuter),
MinorPercentage: (( decimal )100 / DocCount) * CountOuter,
ProgressLabelMinor: UrlLeft,
SubMinorPercentage: (( decimal )100 / DocListCount) * CountInner,
ProgressLabelSubMinor: UrlDuplicate
);
}
this.InsertAndFormatStatusCodeCell(ws, StatusCode);
this.InsertAndFormatStatusCodeCell(ws, Status);
this.InsertAndFormatUrlCell(ws, UrlLeft);
this.InsertAndFormatContentCell(ws, Distance.ToString());
this.InsertAndFormatUrlCell(ws, UrlDuplicate);
ws.NextRecord();
if (this.ProgressForm.Cancelled())
{
break;
}
}
if (this.ProgressForm.Cancelled())
{
break;
}
//Thread.Yield();
}
}
/**************************************************************************/
private void RenderListViewSitemapErrors(MacroscopeDocumentCollection DocCollection)
{
List <ListViewItem> ListViewItems = new List <ListViewItem>(1);
List <Dictionary <string, string> > CompiledTable = DocCollection.GetSitemapErrorsAsTable();
foreach (Dictionary <string, string> Entry in CompiledTable)
{
string SitemapUrl = Entry["sitemap_url"];
string StatusCode = Entry["status_code"];
string Robots = Entry["robots"];
string TargetUrl = Entry["target_url"];
string PairKey = string.Join("::::::::", SitemapUrl, TargetUrl);
MacroscopeDocument msDoc = DocCollection.GetDocumentByUrl(Url: SitemapUrl);
MacroscopeDocument msDocLinked = DocCollection.GetDocumentByUrl(Url: TargetUrl);
ListViewItem lvItem = null;
if (this.DisplayListView.Items.ContainsKey(PairKey))
{
try
{
lvItem = this.DisplayListView.Items[PairKey];
lvItem.SubItems[0].Text = SitemapUrl;
lvItem.SubItems[1].Text = StatusCode;
lvItem.SubItems[2].Text = Robots;
lvItem.SubItems[3].Text = TargetUrl;
}
catch (Exception ex)
{
DebugMsg(string.Format("RenderListViewSitemapErrors 1: {0}", ex.Message));
}
}
else
{
try
{
lvItem = new ListViewItem(PairKey);
lvItem.UseItemStyleForSubItems = false;
lvItem.Name = PairKey;
lvItem.SubItems[0].Text = SitemapUrl;
lvItem.SubItems.Add(StatusCode);
lvItem.SubItems.Add(Robots);
lvItem.SubItems.Add(TargetUrl);
ListViewItems.Add(lvItem);
}
catch (Exception ex)
{
DebugMsg(string.Format("RenderListViewSitemapErrors 2: {0}", ex.Message));
}
}
if (lvItem != null)
{
lvItem.ForeColor = Color.Blue;
if (msDoc.GetIsInternal())
{
lvItem.SubItems[0].ForeColor = Color.Green;
}
else
{
lvItem.SubItems[0].ForeColor = Color.Gray;
}
if (!msDocLinked.GetAllowedByRobots())
{
lvItem.SubItems[2].ForeColor = Color.Red;
}
else
{
lvItem.SubItems[2].ForeColor = Color.Green;
}
if (msDocLinked.GetIsInternal())
{
lvItem.SubItems[3].ForeColor = Color.Green;
}
}
else
{
lvItem.SubItems[3].ForeColor = Color.Gray;
}
}
this.DisplayListView.Items.AddRange(ListViewItems.ToArray());
return;
}
