1
var draftFolderId = 'XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX';
2

3
function getAlreadyPostedList(postedList) {
4
  // declare for prepare.
5
  var alreadyPostedList = [], alreadyPostedChr = '済';
6

7
  // get list of aticle already posted.
8
  postedList.forEach(function(value) {
9
    if(value[1] == alreadyPostedChr) {
10
      alreadyPostedList.push([value[0], value[2]]);
11
    }
12
  });
13

14
  return alreadyPostedList;
15
}
16

17
function readGssColumns() {
18
  // declare for prepare.
19
  var ss,
20
    sheetName = 'ブログ_記事',
21
    sheet,
22
    postedNumber,
23
    row_for_postedNumber = 6,
24
    column_for_postedNumber = 13;
25
  ss = SpreadsheetApp.getActive();
26
  sheet = ss.getSheetByName(sheetName);
27
  postedNumber = Number(sheet.getRange(row_for_postedNumber, column_for_postedNumber).getValue());
28

29
  // declare for get list from GSS.
30
  var dateList   = [],
31
    dateList_formated = [],
32
    permalinkList   = [],
33
    permalinkList_formated = [],
34
    postedList = [],
35
    i = 2, // index of row to start reading sheet.
36
    column_for_date      = 3, // index of 「起票日」column.
37
    column_for_posted    = 6, // index og 「投稿済」column.
38
    column_for_permalink = 7, // index of 「パーマリンク」column.
39
    returnList = [];
40

41
  // get dateList and cleansing.
42
  dateList   = sheet.getRange(i, column_for_date, postedNumber, 1).getValues();
43
  for (let j = 0; j < postedNumber; j++) {
44
    dateList_formated.push(Utilities.formatDate(dateList[j][0], 'JST', 'yyyyMMdd'));
45
  }
46

47
  // get permalinkList and cleansing.
48
  permalinkList   = sheet.getRange(i, column_for_permalink, postedNumber, 1).getValues();
49
  for (let j = 0; j < postedNumber; j++) {
50
    permalinkList_formated.push(permalinkList[j][0]);
51
  }
52

53
  // get postedList.
54
  postedList = sheet.getRange(i, column_for_posted, postedNumber, 1).getValues();
55

56
  // create returnList
57
  for (let k = 0; k < postedNumber; k++) {
58
    returnList.push([dateList_formated[k], postedList[k][0], permalinkList_formated[k]]);
59
  }
60

61
  return returnList;
62
}
63

64
function readYetReadArticles(postedList) {
65
  // declare for prepare.
66
  var alreadyPostedList = getAlreadyPostedList(postedList);
67

68
  // declare for prepare.
69
  var yetReadList = getYetReadList(alreadyPostedList);
70

71
  return yetReadList;
72
}
73

74
function exportYetReadArticles(yetReadArticlesList) {
75
  // declare.
76
  var targetUrl = 'https://www.endorphinbath.com/',
77
    getUrl,
78
    html,
79
    articleTitle,
80
    articleText,
81
    editFile,
82
    editFileId,
83
    docFile,
84
    body_docFile,
85
    docFileName,
86
    endOfText = '<p>以上になります！</p>\n      </div>',
87
    errorDocFile,
88
    body_errorDocFile,
89
    today = new Date();;
90

91
  // テンプレートファイル（「yyyyMMdd(E)」）
92
  var templateFile = DriveApp.getFileById('YYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYY');
93
  // 出力先フォルダ
94
  var outputFolder = DriveApp.getFolderById(draftFolderId);
95

96
  // 「ErrorLog_Batch」Docファイル
97
  errorDocFile = DocumentApp.openById('ZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZ');
98
  body_errorDocFile = errorDocFile.getBody();
99
  var paragraph = body_errorDocFile.appendParagraph(Utilities.formatDate(today, 'JST', 'yyyyMMdd') + '---ErrorLog---\n');
100

101
  // Read articles not read yet and which draft is nothing.
102
  yetReadArticlesList.some(function(value) {
103
    getUrl = targetUrl + value[1];
104
    html = UrlFetchApp.fetch(getUrl).getContentText('UTF-8');
105

106
    // Get articleTitle and cleansing.
107
    articleTitle = String(Parser.data(html).from('<h1 class="entry-title" itemprop="headline">').to('</h1>').iterate());
108
    articleTitle = articleTitle.replace(/\n          /g, '');
109
    articleTitle = articleTitle.replace(/        /g, '');
110

111
    // Get articleText and cleansing.
112
    articleText = String(Parser.data(html).from('<div class="entry-content cf" itemprop="mainEntityOfPage">').to(endOfText).iterate());
113
    articleText = articleText.replace(/p>\n\n\n\n<p>/g, 'p>\n<p>');
114

115
    // Check whether scraping is correctly.
116
    docFileName = value[0] + '_' + articleTitle;
117
    if (articleText.indexOf('<h2><span id="toc1">はじまり</span></h2>') != -1) {
118
      // Copy document with articleTitle.
119
      editFile = templateFile.makeCopy(docFileName, outputFolder);
120
      editFileId = editFile.getId();
121
      docFile = DocumentApp.openById(editFileId);
122
      body_docFile = docFile.getBody();
123

124
      // And write articleText.
125
      body_docFile.clear(); // 全消去
126
      var paragraph = body_docFile.appendParagraph(articleText);
127
      console.log(editFileId);
128
    }else{
129
      var paragraph = body_errorDocFile.appendParagraph(docFileName + '\n');
130
    }
131
  });
132
  var paragraph = body_errorDocFile.appendParagraph('\n\n');
133
}
134

135
function getYetReadList(alreadyPostedList) {
136
  // declare for prepare.
137
  var yetReadList = [];
138

139
  // declare for prepare.
140
  var folder_id = draftFolderId, // Folder ID of draft for article.
141
    folder,
142
    files,
143
    file_now,
144
    date_format_8 = "yyyyMMdd";
145

146
  // get list of folders.
147
  folder = DriveApp.getFolderById(folder_id);
148

149
  // get list of aticle not read yet.
150
  alreadyPostedList.forEach(function(value) {
151
    files = folder.getFiles();
152
    while(files.hasNext()) {
153
      file_now = files.next();
154
      // If draft of article exitsts, break while loop.
155
      if (value[0] == file_now.getName().substring(0, date_format_8.length)) {
156
        break;
157
      }
158

159
      // If draft of article is nothing, ...
160
      if (files.hasNext() == false) {
161
        // If Permalink is not null, ...
162
        if (value[1] != "") {
163
          // Push Article into list.
164
          yetReadList.push([value[0], value[1]]);
165
        }
166
      }
167
    }
168
  });
169

170
  return yetReadList;
171
}
172

173
function exportPostedArticles()  {
174
  // declare for execute.
175
  var postedList;
176
  postedList = readGssColumns();
177
  yetReadArticlesList = readYetReadArticles(postedList);
178
  exportYetReadArticles(yetReadArticlesList);
179
}