protected override void AfterWrite(RecordWriter writer)
{
if (Data == null)
{
Data = new KeywordData();
}
}
private void OnExtractKeywords(KeywordData keywordData, string data)
{
if (keywordData != null)
{
Log.Debug("ExampleAlchemyLanguage", "status: {0}", keywordData.status);
Log.Debug("ExampleAlchemyLanguage", "url: {0}", keywordData.url);
Log.Debug("ExampleAlchemyLanguage", "language: {0}", keywordData.language);
Log.Debug("ExampleAlchemyLanguage", "text: {0}", keywordData.text);
if (keywordData == null || keywordData.keywords.Length == 0)
{
Log.Debug("ExampleAlchemyLanguage", "No keywords found!");
}
else
{
foreach (Keyword keyword in keywordData.keywords)
{
Log.Debug("ExampleAlchemyLanguage", "text: {0}, relevance: {1}", keyword.text, keyword.relevance);
}
}
}
else
{
Log.Debug("ExampleAlchemyLanguage", "Failed to find Keywords!");
}
}
private List <string> GetSuggestions(KeywordData keyword, UserKeywordTag ukt)
{
return(new List <string>()
{
"a", "b", "c", "d"
});
}
public static XElement GetChildern(this KeywordData keyword, ListBaseColumns list = ListBaseColumns.Default)
{
ChildKeywordsFilterData keywordsDataFilter = new ChildKeywordsFilterData
{
BaseColumns = list,
};
return(Wrapper.Instance.GetListXml(keyword.Id, keywordsDataFilter));
}
/// <summary>
/// Initializes a new instance of the <see cref="Keyword"/> class.
/// </summary>
/// <param name="client"><see cref="T:TcmCoreService.Client" /></param>
/// <param name="keywordData"><see cref="T:Tridion.ContentManager.CoreService.Client.KeywordData" /></param>
protected Keyword(Client client, KeywordData keywordData) : base(client, keywordData)
{
if (keywordData == null)
{
throw new ArgumentNullException("keywordData");
}
mKeywordData = keywordData;
}
public static XElement GetClassifiedItems(this KeywordData keyword, ItemType[] itemTypes, ListBaseColumns list = ListBaseColumns.Default)
{
ClassifiedItemsFilterData filter = new ClassifiedItemsFilterData
{
ItemTypes = itemTypes,
BaseColumns = list
};
return(Wrapper.Instance.GetListXml(keyword.Id, filter));
}
/// <summary>
/// Reload the <see cref="Keyword" /> with the specified <see cref="T:Tridion.ContentManager.CoreService.Client.KeywordData" />
/// </summary>
/// <param name="keywordData"><see cref="T:Tridion.ContentManager.CoreService.Client.KeywordData" /></param>
protected void Reload(KeywordData keywordData)
{
if (keywordData == null)
{
throw new ArgumentNullException("keywordData");
}
mKeywordData = keywordData;
base.Reload(keywordData);
mParentKeywords = null;
mRelatedKeywords = null;
}
public static KeywordResult From(KeywordData item, string currentUserId)
{
var result = new KeywordResult
{
Description = TextEntry.From(item.Description, Resources.LabelDescription),
Key = TextEntry.From(item.Key, Resources.LabelKey)
};
if (item.IsAbstract == true)
{
result.IsAbstract = TextEntry.From(Resources.Yes, Resources.LabelIsAbstract);
}
AddCommonProperties(item, result);
AddPropertiesForRepositoryLocalObject(item, result, currentUserId);
return result;
}
public static KeywordResult From(KeywordData item, string currentUserId)
{
var result = new KeywordResult
{
Description = TextEntry.From(item.Description, Resources.LabelDescription),
Key = TextEntry.From(item.Key, Resources.LabelKey)
};
if (item.IsAbstract == true)
{
result.IsAbstract = TextEntry.From(Resources.Yes, Resources.LabelIsAbstract);
}
AddCommonProperties(item, result);
AddPropertiesForRepositoryLocalObject(item, result, currentUserId);
return(result);
}
/// <summary>
/// Create Keyword using KeywordData class
/// </summary>
/// <param name="coreService"></param>
/// <param name="value"></param>
/// <param name="key"></param>
/// <param name="CategoryID"></param>
/// <returns></returns>
public static string GenerateKeyword(ICoreServiceFrameworkContext coreService, string value, string key, string CategoryID)
{
try
{
KeywordData keyword = (KeywordData)coreService.Client.GetDefaultData(ItemType.Keyword, CategoryID, new ReadOptions());
keyword.Id = "tcm:0-0-0";
keyword.Title = value;
keyword.Key = key;
keyword = (KeywordData)coreService.Client.Create(keyword, new ReadOptions());
return(keyword.Id.ToString());
}
catch (Exception ex)
{
return("");
}
}
public Keyword(KeywordData keywordData, string weakDeliminator, string additionalDeliminator)
{
string defaultDeliminators = DefaultDeliminators;
StringBuilder builder = new StringBuilder();
for (int i = 0; i < defaultDeliminators.Length; ++i)
{
char c = defaultDeliminators[i];
if (weakDeliminator.IndexOf(c) == -1 && additionalDeliminator.IndexOf(c) == -1)
{
builder.Append(c);
}
}
builder.Append(additionalDeliminator);
deliminators = builder.ToString();
nodes = keywordData.nodes;
nodesCount = keywordData.nodesCount;
}
public void CreateKeywords(List <string> keywords)
{
List <string> currentKeywords = GetExistingKeywords(_keywordDirty);
foreach (var keyword in keywords)
{
if (currentKeywords.Contains(keyword))
{
continue;
}
// Must create a new one
_keywordDirty = true;
KeywordData newKeyword = (KeywordData)_client.GetDefaultData(ItemType.Keyword,
ResolveUrl(Constants.ContentCategoryUrl));
newKeyword.Title = keyword;
_client.Save(newKeyword, null);
}
}
/// <summary>
/// Create Keywirds in a given category
/// </summary>
/// <param name="model">CategoryModel</param>
/// <param name="CategoryId">tcm:xx-yy-zz</param>
/// <returns></returns>
public static string CreateKeywordsInCategory(CategoryModel model, string CategoryId)
{
var result = "true";
cs_client = CoreServiceProvider.CreateCoreService();
try
{
// open the category that is already created in Tridion
CategoryData category = (CategoryData)cs_client.Read(CategoryId, null);
var xmlCategoryKeywords = cs_client.GetListXml(CategoryId, new KeywordsFilterData());
var keywordAny = xmlCategoryKeywords.Elements()
.Where(element => element.Attribute("Key").Value == model.Key)
.Select(element => element.Attribute("ID").Value)
.Select(id => (KeywordData)cs_client.Read(id, null)).FirstOrDefault();
if (keywordAny == null)
{
// create a new keyword
KeywordData keyword = (KeywordData)cs_client.GetDefaultData(Tridion.ContentManager.CoreService.Client.ItemType.Keyword, category.Id, new ReadOptions());
// set the id to 0 to notify Tridion that it is new
keyword.Id = "tcm:0-0-0";
keyword.Title = model.Title;
keyword.Key = model.Key;
keyword.Description = model.Description;
keyword.IsAbstract = false;
// create the keyword
cs_client.Create(keyword, null);
cs_client.Close();
}
}
catch (Exception ex)
{
result = "Error: " + ex.Message;
}
finally
{
cs_client.Close();
}
return(result);
}
public KeywordData(string[] words, bool casesensitive, KeywordData next)
{
this.casesensitive = casesensitive;
this.next = next;
string[] sortedWords = new string[words.Length];
if (casesensitive)
{
Array.Copy(words, sortedWords, words.Length);
}
else
{
for (int i = 0; i < words.Length; ++i)
{
sortedWords[i] = words[i].ToLowerInvariant();
}
}
Array.Sort(sortedWords, System.StringComparer.Ordinal);
NodesAdd(new KeywordNode((char)1));
ParseNodes(sortedWords, 0, 0, sortedWords.Length);
}
public static void AddKeyword(IAlchemyCoreServiceClient client, string tcmCategory, string keywordName)
{
List <ItemInfo> keywords = GetKeywordsByCategory(client, tcmCategory);
if (keywords.All(x => x.Title != keywordName))
{
KeywordData keyword = new KeywordData
{
Title = keywordName,
Key = keywordName,
Description = keywordName,
LocationInfo = new LocationInfo {
OrganizationalItem = new LinkToOrganizationalItemData {
IdRef = tcmCategory
}
},
Id = "tcm:0-0-0"
};
keyword = (KeywordData)client.Create(keyword, new ReadOptions());
}
}
public static bool Reserve(string word)
{
if (word.Length < 3)
{
return(true);
}
HashSet <string> h = null;
if (CacheService.Contains("RESERVE_KEYWORD"))
{
h = CacheService.Get(RESERVE_KEYWORD) as HashSet <string>;
}
else
{
h = KeywordData.GetReserve(); CacheService.Add(RESERVE_KEYWORD, h);
}
if (h != null)
{
return(h.Contains(word));
}
return(false);
}
public string GetSplitKeywords(int id)
{
List <KeywordData> keywords = new List <KeywordData>();
try
{
using SqlConnection conn = new SqlConnection(connString.SQLConnection);
var sqlQuery =
"SELECT AnalysisId, Keyword, Count FROM SplitKeywords " +
"WHERE AnalysisId = @AnalysisId " +
"ORDER BY Count DESC";
conn.Open();
using SqlCommand cmd = new SqlCommand(sqlQuery, conn);
cmd.Parameters.Add("@AnalysisId", SqlDbType.Int).Value = id;
var reader = cmd.ExecuteReader();
while (reader.Read())
{
var keyword = new KeywordData();
keyword.analysisId = (int)reader["AnalysisId"];
keyword.keyword = reader["Keyword"].ToString();
keyword.count = (int)reader["Count"];
keywords.Add(keyword);
}
reader.Close();
}
catch (Exception e)
{
Debug.WriteLine(e.Message);
}
return(JsonConvert.SerializeObject(keywords));
}
private void OnExtractKeywordsText(KeywordData keywordData, string data)
{
Log.Debug("ExampleAlchemyLanguage", "Alchemy Language - Extract keywords response text: {0}", data);
Test(keywordData != null);
_getKeywordExtractionTextTested = true;
}
public void AddKeyword(Keyword keyword)
{
keywordList.Add(keyword);
toolTipInfos.Add(KeywordData.getData(keyword).info);
}
// convert keyword to tool tip public static ToolTipInfo getToolTipInfoFromKeyword(KeywordData k) => new KeywordInfo(k);
private void ParseNodes(string[] words, int position, int i0, int i1)
{
int index = nodesCount;
{
char prevC = '\0';
bool hasEnds = false;
for (int i = i0; i < i1; ++i)
{
string word = words[i];
if (position >= word.Length)
{
if (position == word.Length)
{
hasEnds = true;
}
continue;
}
char c = word[position];
if (c != prevC)
{
prevC = c;
NodesAdd(new KeywordNode(c));
if (!casesensitive)
{
char upperC = char.ToUpperInvariant(c);
if (upperC != c)
{
NodesAdd(new KeywordNode(upperC));
}
}
}
}
NodesAdd(new KeywordNode(hasEnds ? (char)1 : '\0'));
}
{
bool first = true;
char prevC = '\0';
int prevI = i0;
for (int i = i0; i < i1; ++i)
{
string word = words[i];
if (position >= word.Length)
{
continue;
}
char c = word[position];
if (first)
{
prevC = c;
first = false;
}
else if (c != prevC)
{
int next = nodesCount;
if (i - prevI == 1 && words[prevI].Length == position + 1)
{
next = 0;
}
nodes[index].next = next;
++index;
if (!casesensitive)
{
char upperC = char.ToUpperInvariant(prevC);
if (upperC != prevC)
{
nodes[index].next = next;
++index;
}
}
if (next != 0)
{
ParseNodes(words, position + 1, prevI, i);
}
prevC = c;
prevI = i;
}
}
if (!first)
{
int next = nodesCount;
if (i1 - prevI == 1 && words[prevI].Length == position + 1)
{
next = 0;
}
nodes[index].next = next;
if (!casesensitive)
{
char currentC = nodes[index].c;
char upperC = char.ToUpperInvariant(currentC);
if (upperC != currentC)
{
++index;
nodes[index].next = next;
}
}
if (next != 0)
{
ParseNodes(words, position + 1, prevI, i1);
}
}
}
}
protected override void AfterWrite(RecordWriter writer)
{
if (Data == null)
Data = new KeywordData();
}
public KeywordInfo(KeywordData k)
{
keyword = k;
}
private static bool KeywordIsValid(KeywordData k)
{
return(k != null && !String.IsNullOrEmpty(k.KeywordName) && !String.IsNullOrEmpty(k.KeywordValue));
}
public void RemoveKeyword(Keyword keyword)
{
keywordList.Remove(keyword);
toolTipInfos.Remove(KeywordData.getData(keyword).info);
}
public static XElement GetClassifiedItems(this KeywordData keyword, ItemType itemType, ListBaseColumns list = ListBaseColumns.Default)
{
ItemType[] itemList = { itemType };
return(GetClassifiedItems(keyword, itemList, list));
}
static async Task Main(string[] args)
{
#region 取得token
using (HttpClient client = new HttpClient()) //建立 HttpClient
using (HttpResponseMessage response = await client.GetAsync("https://api2.ifeel.com.tw/pro/auth?member_account=iiidsi_sa&client_secretkey=daa75a16f538c13be9e97cf91acdd9c8")) //使用 async 方法從網路 url 上取得回應
using (HttpContent content = response.Content) //將網路取得回應的內容設定給 httpcontent
{
string result = await content.ReadAsStringAsync(); // 將 httpcontent 轉為 string
if (result != "Unauthorized Access") //成功取得token
{
Console.WriteLine("Get token successfully" + "\n");
dynamic dy = JsonConvert.DeserializeObject <dynamic>(result); //將result轉換為dynamic物件，方便資料讀取
tokenType = dy.token_type;
token = dy.access_token;
Console.WriteLine("token type: " + tokenType);
Console.WriteLine("token: " + token + "\n");
}
else //取得token失敗
{
Console.WriteLine("Get token unsuccessfully" + "\n");
}
}
#endregion
#region 驗證token有效性
using (HttpClient client = new HttpClient())
{
tokenForTest = tokenType + " " + token; //串聯token type與token碼，以利之後header驗證使用
client.DefaultRequestHeaders.Add("Authorization", tokenForTest); // 指定 authorization header
using (HttpResponseMessage response = client.PostAsync("https://api2.ifeel.com.tw/pro/auth", null).Result) //使用 async 方法從網路 url 上取得回應
using (HttpContent content = response.Content) //將網路取得回應的內容設定給 httpcontent
{
string result = await content.ReadAsStringAsync(); // 將 httpcontent 轉為 string
if (result != "Unauthorized Access") //驗證成功
{
Console.WriteLine("Token accessible,Information:" + "\n");
dynamic dy = JsonConvert.DeserializeObject <dynamic>(result); //將result轉換為dynamic物件，方便資料讀取
Console.WriteLine(result);
success = true;
}
else //驗證失敗
{
Console.WriteLine("Token not accessible" + "\n");
success = false;
}
}
}
#endregion
#region 抓取新聞內容
//token驗證成功，可抓取新聞資料
if (success == true)
{
//抓取2018-04-20的資料
using (HttpClient client = new HttpClient())
{
client.DefaultRequestHeaders.Add("Authorization", tokenForTest); // 指定 authorization header
client.DefaultRequestHeaders.Accept.Add(new MediaTypeWithQualityHeaderValue("application/json")); //指定content type header
var paraBefore = new Dictionary <string, string>() //定義傳入參數
{
{
"date", "2018-04-20"
},
{
"keyword", "麻疹"
},
};
var paraAfter = new StringContent(JsonConvert.SerializeObject(paraBefore), Encoding.UTF8, "application/json"); //轉換傳入參數的編碼與資料格式
using (HttpResponseMessage response = client.PostAsync("https://api2.ifeel.com.tw/pro/document/news/all", paraAfter).Result) //使用 async 方法從網路 url 上取得回應
using (HttpContent content = response.Content) //將網路取得回應的內容設定給 httpcontent
{
string result = await content.ReadAsStringAsync(); // 將 httpcontent 轉為 string
dynamic dy = JsonConvert.DeserializeObject <dynamic>(result); //將result轉換為dynamic物件，方便資料讀取
if (dy["_status"] == "Success") //抓取成功
{
Console.WriteLine("Get " + paraBefore["date"] + " news data successfully!" + "\n");
//依序將抓到的新聞內容存進News物件，並存進news list裡
for (int i = 0; i < dy["data"].Count; i++)
{
news.Add(new News(dy["data"][i]["content"].ToString()));
//Console.WriteLine(dy["data"][i]["content"].ToString() + "\n");
}
}
else //抓取失敗
{
Console.WriteLine("Get " + paraBefore["date"] + " news data unsuccessfully!" + "\n");
}
}
//接著抓取2018-04-21的資料
paraBefore = new Dictionary <string, string>() //定義傳入參數
{
{
"date", "2018-04-21"
},
{
"keyword", "麻疹"
},
};
paraAfter = new StringContent(JsonConvert.SerializeObject(paraBefore), Encoding.UTF8, "application/json"); //轉換傳入參數的編碼與資料格式
using (HttpResponseMessage response = client.PostAsync("https://api2.ifeel.com.tw/pro/document/news/all", paraAfter).Result) //使用 async 方法從網路 url 上取得回應
using (HttpContent content = response.Content) //將網路取得回應的內容設定給 httpcontent
{
string result = await content.ReadAsStringAsync(); // 將 httpcontent 轉為 string
dynamic dy = JsonConvert.DeserializeObject <dynamic>(result); //將result轉換為dynamic物件，方便資料讀取
if (dy["_status"] == "Success") //抓取成功
{
Console.WriteLine("Get " + paraBefore["date"] + " news data successfully!" + "\n");
//依序將抓到的新聞內容存進News物件，並存進news list裡
for (int i = 0; i < dy["data"].Count; i++)
{
news.Add(new News(dy["data"][i]["content"].ToString()));
//Console.WriteLine(dy["data"][i]["content"].ToString() + "\n");
}
}
else //抓取失敗
{
Console.WriteLine("Get " + paraBefore["date"] + " news data unsuccessfully!" + "\n");
}
}
}
#endregion
#region 文章斷詞與關鍵詞分析
//合併個別文章
for (int i = 0; i < news.Count; i++)
{
newsAppend += news[i].getNews();
}
var seg = new JiebaSegmenter(); //建立Jieba分詞物件
var segments = seg.Cut(newsAppend); //進行斷詞
segCount = 0;
//計算分詞個數
foreach (var segment in segments)
{
segCount++;
}
words = new string[segCount];
int c = 0;
//將分詞集合裡的詞依序存進words array
foreach (var segment in segments)
{
words[c] = segment;
c++;
}
//從本地端讀進stopwords.txt file，並存進stopwords陣列
stopwords = File.ReadAllLines(System.IO.Directory.GetCurrentDirectory() + "\\Resources\\stopwords.txt");
for (int i = 0; i < words.Length; i++)
{
for (int j = 0; j < stopwords.Length; j++)
{
//若分詞屬於stopword，則從words list刪除該分詞
if (words[i] == stopwords[j])
{
words[i] = "Deleted!";
break;
}
}
if (words[i] != "Deleted!" && words[i] != " ")
{
wordsWithoutSw.Add(words[i]);
}
}
int count;
//計算TF值
for (int i = 0; i < wordsWithoutSw.Count; i++)
{
count = 1;
if (wordTF.ContainsKey(wordsWithoutSw[i]) == false) //wordTF dictionary裡無紀錄值才計算，以避免重複算的問題
{
for (int j = i + 1; j < wordsWithoutSw.Count; j++)
{
if (wordsWithoutSw[i].Equals(wordsWithoutSw[j]))
{
count++;
}
}
wordTF.Add(wordsWithoutSw[i], ((decimal)count / wordsWithoutSw.Count)); //將計算後的次數加入到wordTF dictionary
}
}
//取出前20大關鍵詞
var top20 = wordTF.OrderByDescending(pair => pair.Value).Take(20)
.ToDictionary(pair => pair.Key, pair => pair.Value);
#endregion
#region Json格式轉換與檔案輸出
Console.WriteLine("Top 20 most used keyword:");
//依序讀取20個關鍵字，將關鍵字資料存成KeywordData物件，並存進keydata list
foreach (KeyValuePair <string, decimal> top in top20)
{
Console.WriteLine("keyword = {0}, TF = {1}", top.Key, top.Value);
KeywordData kd = new KeywordData();
kd.text = top.Key;
kd.size = (int)(top.Value * 4000);
keydata.Add(kd);
}
var dataForOutput = JsonConvert.SerializeObject(keydata); //將list序列化為JSON字串
dataForOutput = "data = '" + dataForOutput + "';"; //轉換成HTML可讀取的資料格式
//Console.WriteLine(dataForOutput);
System.IO.File.WriteAllText(@"C:\Users\User\Desktop\資策會測驗\III program test\website\data.json", dataForOutput); //將json檔輸出至本地端
#endregion
}
}
private void OnExtractKeywordsUrl(KeywordData keywordData, string data)
{
Log.Debug("ExampleAlchemyLanguage", "Alchemy Language - Extract keywords response url: {0}", data);
_getKeywordExtractionURLTested = true;
}
private void OnExtractKeywordsText(KeywordData keywordData, Dictionary <string, object> customData)
{
Log.Debug("ExampleAlchemyLanguage.OnExtractKeywordsText()", "Alchemy Language - Extract keywords response text: {0}", customData["json"].ToString());
_getKeywordExtractionTextTested = true;
}
