private bool GenerateNewPage(IListSheet listSheet)
{
bool succeed = true;
string exportDir = this.RunPage.GetExportDir();
string pageSourceDir = this.RunPage.GetDetailSourceFileDir();
string outputTitleTextDir = Path.Combine(exportDir, "titleText");
Dictionary <string, int> subjectColumnDic = CommonUtil.InitStringIndexDic(new string[] {
"detailPageUrl",
"detailPageName",
"cookie",
"grabStatus",
"giveUpGrab",
"subjectIndex",
"id",
"title",
"googleUrl",
"creator",
"createDate",
"messageCount",
"authorCount"
});
string subjectFileExcelPath = Path.Combine(exportDir, this.RunPage.Project.Name + "_List.xlsx");
ExcelWriter subjectEW = new ExcelWriter(subjectFileExcelPath, "List", subjectColumnDic);
Dictionary <string, int> subjectImportColumnDic = CommonUtil.InitStringIndexDic(new string[] {
"index",
"id",
"creator",
"createDate",
"commentPublisher",
"commentPublished",
"googleUrl",
"title",
"url"
});
string subjectFileXmlPath = Path.Combine(exportDir, this.RunPage.Project.Name + "_List.xml");
XmlWriter subjectXW = new XmlWriter(subjectFileXmlPath, subjectImportColumnDic);
string[] commentUsers = new string[] { "sunhua", "shizhengzhong", "liyuzhu" };
int sujectIndex = 1;
for (int i = listSheet.RowCount - 1; i >= 0; i--)
{
Dictionary <string, string> row = listSheet.GetRow(i);
bool giveUp = "Y".Equals(row[SysConfig.GiveUpGrabFieldName]);
if (!giveUp)
{
string url = row[SysConfig.DetailPageUrlFieldName];
string localFilePath = this.RunPage.GetFilePath(url, pageSourceDir);
TextReader tr = null;
try
{
tr = new StreamReader(localFilePath);
string webPageHtml = tr.ReadToEnd();
HtmlAgilityPack.HtmlDocument htmlDoc = new HtmlAgilityPack.HtmlDocument();
htmlDoc.LoadHtml(webPageHtml);
HtmlNodeCollection subjectNodes = htmlDoc.DocumentNode.SelectNodes("//body[1]/table[1]/tr");
if (subjectNodes != null)
{
for (int j = subjectNodes.Count - 1; j >= 1; j--)
{
HtmlNode subjectNode = subjectNodes[j];
HtmlNode titleNode = subjectNode.SelectSingleNode("./td[@class=\"subject\"]/a");
string title = CommonUtil.HtmlDecode(CommonUtil.ReplaceAsciiByString(titleNode.InnerText.Trim()));
string detailPageUrl = titleNode.Attributes["href"].Value.Trim();
string detailPageName = detailPageUrl.Substring(detailPageUrl.LastIndexOf("/") + 1);
HtmlNode authorNode = subjectNode.SelectSingleNode("./td[@class=\"author\"]");
string author = authorNode == null ? "" : CommonUtil.HtmlDecode(CommonUtil.ReplaceAsciiByString(authorNode.InnerText.Trim()));
HtmlNode lastPostDateNode = subjectNode.SelectSingleNode("./td[@class=\"lastPostDate\"]");
string lastPostDate = lastPostDateNode == null ? "" : lastPostDateNode.InnerText.Trim();
Dictionary <string, string> commentF2vs = new Dictionary <string, string>();
commentF2vs.Add("subjectIndex", sujectIndex.ToString());
commentF2vs.Add("detailPageUrl", detailPageUrl);
commentF2vs.Add("detailPageName", detailPageName);
commentF2vs.Add("id", detailPageName);
commentF2vs.Add("title", title);
commentF2vs.Add("googleUrl", detailPageUrl);
commentF2vs.Add("creator", author);
commentF2vs.Add("createDate", lastPostDate);
subjectEW.AddRow(commentF2vs);
for (int u = 0; u < commentUsers.Length; u++)
{
string user = commentUsers[u];
Dictionary <string, string> commentXF2vs = new Dictionary <string, string>();
commentXF2vs.Add("index", sujectIndex.ToString());
commentXF2vs.Add("id", user + "_" + detailPageName);
commentXF2vs.Add("title", title);
commentXF2vs.Add("url", detailPageName + ".html");
commentXF2vs.Add("googleUrl", detailPageUrl);
commentXF2vs.Add("creator", author);
commentXF2vs.Add("createDate", lastPostDate);
commentXF2vs.Add("commentPublisher", user);
commentXF2vs.Add("commentPublished", "No");
subjectXW.AddRow(commentXF2vs);
}
sujectIndex++;
}
}
}
catch (Exception ex)
{
if (tr != null)
{
tr.Close();
tr.Dispose();
}
throw ex;
}
}
}
subjectXW.SaveToDisk();
subjectEW.SaveToDisk();
return(succeed);
}
private bool GenerateNewPage(IListSheet listSheet)
{
bool succeed = true;
string exportDir = this.RunPage.GetExportDir();
string pageSourceDir = this.RunPage.GetDetailSourceFileDir();
string outputTitleTextDir = Path.Combine(exportDir, "titleText");
Dictionary <string, int> subjectColumnDic = CommonUtil.InitStringIndexDic(new string[] {
"index",
"id",
"title",
"url",
"googleUrl",
"creator",
"createDate",
"messageCount",
"authorCount",
"commentPublisher",
"commentPublished"
});
string subjectFileExcelPath = Path.Combine(exportDir, this.RunPage.Project.Name + "_List.xlsx");
ExcelWriter subjectEW = new ExcelWriter(subjectFileExcelPath, "List", subjectColumnDic);
string subjectFileXmlPath = Path.Combine(exportDir, this.RunPage.Project.Name + "_List.xml");
XmlWriter subjectXW = new XmlWriter(subjectFileXmlPath, subjectColumnDic);
Dictionary <string, int> allCommentsColumnDic = new Dictionary <string, int>();
allCommentsColumnDic.Add("subjectIndex", 0);
allCommentsColumnDic.Add("googleUrl", 1);
allCommentsColumnDic.Add("creator", 2);
allCommentsColumnDic.Add("author", 3);
allCommentsColumnDic.Add("lastPostDate", 4);
string allCommentsFilePath = Path.Combine(exportDir, this.RunPage.Project.Name + "_AllComments.xlsx");
ExcelWriter allCommentsListEW = new ExcelWriter(allCommentsFilePath, "List", allCommentsColumnDic);
for (int i = 0; i < listSheet.RowCount; i++)
{
Dictionary <string, string> row = listSheet.GetRow(i);
bool giveUp = "Y".Equals(row[SysConfig.GiveUpGrabFieldName]);
if (!giveUp)
{
string index = row[SysConfig.ListPageIndexFieldName].PadLeft(4, '0');
string googleUrl = row[SysConfig.DetailPageUrlFieldName];
string title = row["title"];
string creator = row["creator"];
string createDate = row["createDate"];
string localFilePath = this.RunPage.GetFilePath(googleUrl, pageSourceDir);
TextReader tr = null;
try
{
tr = new StreamReader(localFilePath);
string webPageHtml = tr.ReadToEnd();
HtmlAgilityPack.HtmlDocument htmlDoc = new HtmlAgilityPack.HtmlDocument();
htmlDoc.LoadHtml(webPageHtml);
string messagesNumStr = htmlDoc.DocumentNode.SelectSingleNode("//body/i").InnerText;
int ofIndex = messagesNumStr.IndexOf("of");
int messagesIndex = messagesNumStr.IndexOf("messages");
string messageCount = messagesNumStr.Substring(ofIndex + 2, messagesIndex - ofIndex - 2).Trim();
List <string> authorNames = new List <string>();
HtmlNodeCollection messageNodes = htmlDoc.DocumentNode.SelectNodes("//body/table/tr");
if (messageNodes != null)
{
for (int j = 0; j < messageNodes.Count; j++)
{
HtmlNode messageNode = messageNodes[j];
HtmlNode authorNode = messageNode.SelectSingleNode("./td[2]");
HtmlNode lastPostDateNode = messageNode.SelectSingleNode("./td[3]");
if (authorNode != null)
{
string author = HttpUtility.HtmlDecode(authorNode.InnerText).Trim();
if (j == 0)
{
creator = author;
}
if (!authorNames.Contains(author))
{
authorNames.Add(author);
}
string lastPostDate = lastPostDateNode == null ? "" : lastPostDateNode.InnerText;
Dictionary <string, string> commentF2vs = new Dictionary <string, string>();
commentF2vs.Add("subjectIndex", index);
commentF2vs.Add("googleUrl", googleUrl);
commentF2vs.Add("creator", creator);
commentF2vs.Add("author", author);
commentF2vs.Add("lastPostDate", lastPostDate);
allCommentsListEW.AddRow(commentF2vs);
}
}
}
//修改html内容，增加线框
HtmlNode tableNode = htmlDoc.DocumentNode.SelectSingleNode("//body/table");
tableNode.Attributes["border"].Value = "1";
int localHtmlUrlStartIndex = googleUrl.IndexOf("/idempiere/") + "/idempiere/".Length;
string htmlLocalName = CommonUtil.ProcessFileName(googleUrl.Substring(localHtmlUrlStartIndex), "_") + ".html";
string htmlLocalUrl = Path.Combine(Path.GetDirectoryName(pageSourceDir), "export\\html\\" + htmlLocalName);
CommonUtil.CreateFileDirectory(htmlLocalUrl);
htmlDoc.Save(htmlLocalUrl);
Dictionary <string, string> f2vs = new Dictionary <string, string>();
f2vs.Add("index", index);
f2vs.Add("title", title);
f2vs.Add("googleUrl", googleUrl);
f2vs.Add("url", htmlLocalName);
f2vs.Add("creator", creator);
f2vs.Add("createDate", createDate);
f2vs.Add("messageCount", messageNodes.Count.ToString());
f2vs.Add("authorCount", authorNames.Count.ToString());
f2vs.Add("commentPublisher", "");
f2vs.Add("commentPublished", "");
f2vs.Add("id", "");
IRow newPageListRow = subjectEW.AddRow(f2vs);
f2vs["commentPublished"] = "No";
f2vs["commentPublisher"] = "sunhua";
f2vs["id"] = "sunhua" + index;
subjectXW.AddRow(f2vs);
f2vs["commentPublisher"] = "shizhengzhong";
f2vs["id"] = "shizhengzhong" + index;
subjectXW.AddRow(f2vs);
f2vs["commentPublisher"] = "liyuzhu";
f2vs["id"] = "liyuzhu" + index;
subjectXW.AddRow(f2vs);
ICell localUrlCell = subjectEW.GetCell(newPageListRow, "url", true);
IHyperlink hyperlink = new XSSFHyperlink(HyperlinkType.File);
hyperlink.Address = "html/" + htmlLocalName;
localUrlCell.Hyperlink = hyperlink;
//保存message文本
string msg = Encoding.UTF8.GetString(Encoding.Convert(Encoding.ASCII, Encoding.UTF8, Encoding.ASCII.GetBytes(title))).Trim();
if (msg.Length > 0)
{
string titleTextFilePath = Path.Combine(outputTitleTextDir, index + ".txt");
CommonUtil.CreateFileDirectory(titleTextFilePath);
TextWriter tw = null;
try
{
tw = new StreamWriter(titleTextFilePath, false, new UTF8Encoding(false));
tw.Write(msg);
tw.Flush();
}
catch (Exception ee)
{
throw ee;
}
finally
{
if (tw != null)
{
tw.Close();
tw.Dispose();
}
}
}
}
catch (Exception ex)
{
if (tr != null)
{
tr.Dispose();
tr = null;
}
this.RunPage.InvokeAppendLogText("读取出错. " + ex.Message + " LocalPath = " + localFilePath, LogLevelType.Error, true);
}
}
}
subjectXW.SaveToDisk();
subjectEW.SaveToDisk();
allCommentsListEW.SaveToDisk();
return(succeed);
}
