public AinunuDownloadDTO GetMovieDownloadDetail(string downloadPageUrl, HtmlNodeCollection downloadNodes, string movieName, int MovieId)
{
HtmlNodeCollection nodes = null;
if (downloadPageUrl != null && downloadPageUrl != "")
{
var helper = new CrawlerHelper();
var html = helper.DownloadHtml(downloadPageUrl, Encoding.GetEncoding("UTF-8"));
HtmlDocument document = new HtmlDocument();
if (html == null || html == "")
{
return(null);
}
document.LoadHtml(html);
nodes = document.DocumentNode.SelectNodes("/html/body/div[1]/div[1]/div[1]/div[1]/article[1]/div[1]/div");
}
else
{
nodes = downloadNodes;
}
if (nodes == null || nodes.Count < 2)
{
return(null);
}
var Resources = new List <ResourceDTO>();
for (int i = 0; i < nodes.Count; i = i + 2)//第一个是说明内容的div
{
HtmlNode node = nodes[i];
if (i < nodes.Count)
{
HtmlNode node2 = null;
if (i + 1 < nodes.Count)
{
node2 = nodes[i + 1];
}
string formatName = node.InnerHtml;
var resourceLinks = new List <ResourceLinkDTO>();
HtmlNodeCollection resourceNodes = node2 == null ? null : node2.SelectNodes("child::a");
HtmlNodeCollection resourceOthersNodes = node2 == null ? null : node2.SelectNodes("child::text()");
if (resourceNodes != null)
{
for (int j = 0; j < resourceNodes.Count; j++)
{
string resourceLinkName = resourceNodes[j].InnerHtml;
string resourceLinkUrl = resourceNodes[j].GetAttributeValue("href", "");
string resourceLinkOther = resourceOthersNodes != null ? resourceOthersNodes[j].InnerHtml : "";
if (resourceLinkUrl != null)
{
var entity = new ResourceLinkDTO()
{
Name = resourceLinkName,
Type = "",
Url = resourceLinkUrl,
Others = resourceLinkOther
};
resourceLinks.Add(entity);
}
}
}
var resource = new ResourceDTO()
{
Id = i.ToString(),
FormatName = formatName,
ResourceLinks = resourceLinks
};
Resources.Add(resource);
}
}
var download = new AinunuDownloadDTO()
{
id = MovieId.ToString(),
MovieId = MovieId,
MovieName = movieName,
Resources = Resources,
UpdateDate = "",
CreateDate = DateTimeOffset.UtcNow.ToString("yyyy-MM-dd HH:mm:ss"),
partitionKey = "download"
};
return(download);
}
public async Task <string> GetAinunuContent(int switcher)
{
this.cosmosClient = new CosmosClient(EndpointUri, PrimaryKey, new CosmosClientOptions()
{
ApplicationName = "CosmosDBDotnetQuickstart"
});
// Create a new database
this.database = await this.cosmosClient.CreateDatabaseIfNotExistsAsync(databaseId);
this.detailContainer = await this.database.CreateContainerIfNotExistsAsync(detailContainerId, "/partitionKey");
this.downloadContainer = await this.database.CreateContainerIfNotExistsAsync(downloadContainerId, "/partitionKey");
var JsonFileHelper = new JsonFileHelper();
string exsistMoives = "";
string exsistDownloads = "";
var tempMoives = new List <AinunuMovieDTO>();
var tempDownloads = new List <AinunuDownloadDTO>();
//增量更新
if (switcher == 2)
{
exsistMoives = JsonFileHelper.ReadJsonFile("Data.json");
exsistDownloads = JsonFileHelper.ReadJsonFile("Download.json");
tempMoives = JsonConvert.DeserializeObject <List <AinunuMovieDTO> >(exsistMoives);
tempDownloads = JsonConvert.DeserializeObject <List <AinunuDownloadDTO> >(exsistDownloads);
}
Encoding.RegisterProvider(CodePagesEncodingProvider.Instance);
var movieList = new List <AinunuMovieDTO>();
var downloadList = new List <AinunuDownloadDTO>();
var totalPageUrl = baseUrl + "/c/movie/";
var helper = new CrawlerHelper();
var html = helper.DownloadHtml(totalPageUrl, Encoding.GetEncoding("GB2312"));
HtmlDocument document = new HtmlDocument();
document.LoadHtml(html);
HtmlNode totalPage = document.DocumentNode.SelectSingleNode("/html/body/div[4]/div[1]/div[3]/div/ul/li[15]/span/strong[1]");
var totalPageCount = int.Parse(totalPage.InnerText);
HtmlNode totalMovies = document.DocumentNode.SelectSingleNode("/html/body/div[4]/div[1]/div[3]/div/ul/li[15]/span/strong[2]");
//Start Id 12286 total -- 2021/1/2
var Id = int.Parse(totalMovies.InnerText);
Id = Id > 0 ? Id : 15000; //起始Id
for (int i = 1; i <= totalPageCount; i++) //410 total -- 2022/1/2 TODO:动态读取总页数
{
var breakFlag = false;
var url = baseUrl + "/c/movie/list_" + i.ToString() + ".html";
html = helper.DownloadHtml(url, Encoding.GetEncoding("GB2312"));
if (html == "")
{
break;
}
HtmlNodeCollection nodes = GetMovieList(html);
for (int j = 0; j < nodes.Count; j++)
{
HtmlNode node = nodes[j].SelectSingleNode("child::a[2]");
string movieCategory = nodes[j].SelectSingleNode("child::a[1]").InnerHtml;
string movieUpdateTime = nodes[j].SelectSingleNode("child::span").InnerHtml;
string detaiRelativelUrl = node.GetAttributeValue("href", "");
string movieDetailPageUrl = baseUrl + detaiRelativelUrl;
var urlName = node.InnerHtml;
Console.WriteLine(urlName);
if (tempMoives.Exists(x => x.UrlName == urlName) && switcher == 2)
{
breakFlag = true;
break;
}
try
{
var detail = GetMovieDetail(movieDetailPageUrl, Id);
if (detail.movie != null)
{
detail.movie.UpdateDate = movieUpdateTime;
detail.movie.Category = movieCategory;
detail.movie.partitionKey = movieCategory;
detail.movie.UrlName = urlName;
await InsertMovieIntoCosmos(detail.movie);
movieList.Add(detail.movie);
var download = GetMovieDownloadDetail(detail.movie.DownloadUrl, detail.downloadNodes, detail.movie.Name, Id);
if (download != null)
{
downloadList.Add(download);
await InsertDownloadIntoCosmos(download);
}
}
}
catch (Exception ex)
{
LogHelper.Error(string.Format(ex.Message + " " + node.InnerHtml), ex);
}
Id -= 1;
}
if (breakFlag)
{
break;
}
}
var newMovieList = movieList.Concat(tempMoives).ToList();
var newDownloadList = downloadList.Concat(tempDownloads).ToList();
JsonFileHelper.WriteJsonFile("Data.json", JsonConvert.SerializeObject(newMovieList));
JsonFileHelper.WriteJsonFile("Download.json", JsonConvert.SerializeObject(newDownloadList));
// JsonFileHelper.WriteJsonFile("../FrontEnd/src/data/movie.json", JsonConvert.SerializeObject(newMovieList));
// JsonFileHelper.WriteJsonFile("../FrontEnd/src/data/download.json", JsonConvert.SerializeObject(newDownloadList));
return("grab done Total:" + movieList.Count.ToString());
}
public virtual Recipe CreateRecipe(string recipeURL, string recipeName, int recipeSourceID)
{
string html = getHTML(recipeURL);
if (string.IsNullOrEmpty(html))
{
return(null);
}
var doc = new HtmlDocument();
doc.LoadHtml(html);
if (string.IsNullOrEmpty(recipeName))
{
var recipeNameNode = doc.DocumentNode.SelectSingleNode(recipeNameXPath);
if (recipeNameNode == null)
{
throw new NotImplementedException();
}
recipeName = CrawlerHelper.ChildSafeName(HttpUtility.HtmlDecode(Common.StripHTML(recipeNameNode.InnerText))).Trim();
}
Recipe rec = new Recipe();
rec.RecipeSourceID = recipeSourceID;
rec.RecipeName = recipeName;
rec.RecipeURL = recipeURL;
var directions = doc.DocumentNode.SelectNodes(directionsXPath);
if (directions == null || !directions.Any())
{
directions = doc.DocumentNode.SelectNodes(directions2XPath);
}
rec.Directions = string.Empty;
if (directions != null)
{
foreach (HtmlNode d in directions)
{
rec.Directions += Common.StripHTML(d.InnerText) + "\r\n";
}
rec.Directions = rec.Directions.Trim();
}
rec.Rating = getRating(doc.DocumentNode);
var servingsNode = doc.DocumentNode.SelectSingleNode(servingsXPath);
if (servingsNode == null)
{
servingsNode = doc.DocumentNode.SelectSingleNode(servings2XPath);
}
if (servingsNode != null)
{
int tempInt = -1;
if (!int.TryParse(Common.StripHTML(servingsNode.InnerText), out tempInt))
{
Match servingsMatch2 = Regex.Match(Common.StripHTML(servingsNode.InnerText), "(\\d+)");
if (!int.TryParse(servingsMatch2.Groups[1].Value, out tempInt))
{
}
}
rec.NumberOfServings = tempInt;
}
var ingredients = getIngredients(doc.DocumentNode);
if (ingredients == null)
{
return(null);
}
foreach (var ing in ingredients)
{
rec.RecipeIngredientMeasurements.Add(ing);
}
foreach (var img in getRecipeImages(doc.DocumentNode))
{
rec.RecipeImages.Add(img);
}
DbContext.Recipes.Add(rec);
DbContext.SaveChanges();
return(rec);
}
