PubMedの新着論文のSlackアラート (勝手改造2021年5月版)

function PubMedSlack() {
  // 基本設定
  var rss_sheet_name = "RSS";
  var webhook_sheet_name = "webhooks";
  var notify_text = "Here are {ITEM_NUMBER} new papers for *{KEYWORD}* :eyes:";
  var N_IN_BLOCK = 5;
  
  var rw_length = "{ITEM_NUMBER}";   //アイテムの数
  var rw_keyword = "{KEYWORD}"; //キーワード（RSSリストのkeywordに対応）
  var ns_dc = XmlService.getNamespace("http://purl.org/dc/elements/1.1/");
  var pm_rss_legacy = "https://eutils.ncbi.nlm.nih.gov/entrez/eutils/";
  var pm_rss_new = "https://pubmed.ncbi.nlm.nih.gov/rss/search/";
  
  //Scopusを使ってCiteScoreなどを表示させる時に設定する。PubMedとScopus両方のAPI Keyが必要。
  var enable_scopus_database = 0;
  var display_pmid = 0;
  var display_citescore = 0;
  var display_snip = 0;
  var display_sjr = 0;
  var pubmed_api_key = '';
  var scopus_api_key = '';
  
  var pm_base_url = 'https://eutils.ncbi.nlm.nih.gov/entrez/eutils/esummary.fcgi?db=pubmed&version=2.0&id={ID}&api_key='+pubmed_api_key;
  var pm_reptxt = '{ID}';
  var pm_params = {
    "method": "get",
    "muteHttpExceptions": false,
    "headers": {
      "accept": "text/xml",
      "timeout": "20000"
    }
  };
  var sc_base_url = 'https://api.elsevier.com/content/serial/title/issn/{ISSN}?apiKey='+scopus_api_key;
  var sc_reptxt = '{ISSN}';
  var sc_params = {
    "method": "get",
    "muteHttpExceptions": false,
    "headers": {
      "accept": "text/xml",
      "timeout": "20000"
    }
  };
  var mrkdwnEscape = function (string) {
    if(typeof string !== 'string') {
      return string;
    }
    return string.replace(/[&<>]/g, function(match) {
      return {
        '&': '&',
        '<': '<',
        '>': '>',
      }[match]
    });
  }
  
  
  // スプレッドシートからRSSとWebhookのURLを取得
  var ss = SpreadsheetApp.getActiveSpreadsheet();
  // rss_list[row:1-max][col:1-4] = "value"
  // col=1: keyword
  // col=2: RSS URL
  // col=3: Target ID
  // col=4: History Data
  var rss_sheet = ss.getSheetByName(rss_sheet_name);
  var rss_total_num = rss_sheet.getLastRow() - 1;
  var rss_list = rss_sheet.getRange(2, 1, rss_total_num+1, 4).getValues();
  // webhook_list["target_id"] = "Webhook URL" となるように格納する
  var webhook_sheet = ss.getSheetByName(webhook_sheet_name);
  var webhook_total_num = webhook_sheet.getLastRow() - 1;
  var webhook_table = webhook_sheet.getRange(2, 1, webhook_total_num+1, 2).getValues();
  var webhook_list = {};
  for (var i=0; i<webhook_total_num; i++) {
    webhook_list[webhook_table[i][0].toString().trim()] = webhook_table[i][1].toString().trim();
  }
  
  for (var i=0; i<rss_total_num; i++) {
    
    var keyword = rss_list[i][0].toString().trim();
    var rss_url = rss_list[i][1].toString().trim();
    var target = rss_list[i][2].toString().trim();
    try {
      var history = JSON.parse(rss_list[i][3]);
    } catch (e) {
      var history = [''];
    }
    var webhook_url = webhook_list[target];
    if (!webhook_url) {
      Logger.log("webhook url not found: "+target);
      continue;
    }
    
    // Fetch
    try {
      var rss_text = UrlFetchApp.fetch(rss_url).getContentText(); // たまにPubMedが404エラーを出すことがあるのでtryで回避
    } catch (e) {
      continue; // もしエラーが出てもとりあえず無視(次回fetchできれば良しとする)
    }
    
    // Parse
    // 新しい順に並んでいるので逆転させる
    var rss = XmlService.parse(rss_text);
    var items = rss.getRootElement().getChildren('channel')[0].getChildren('item').reverse();
    
    // Check new items
    var news = [];
    for (var item of items) {
      if (history.indexOf(item.getChild('guid').getText()) < 0) {
        news.push(item);
      }
    }
    
    if (news.length > 0) {
      // Prepare blocks
      var blocks = [{
        type: "section",
        text: {
          type:"mrkdwn",
          text:notify_text.replace(rw_length, news.length).replace(rw_keyword, keyword)
        }
      }];
      
      var txt = '';
      
      var j = 1;
      for (var item of news) {
        
        // rss_urlの形式によってデータの抽出方法を変えていたが、書式が統一されたためルーチンも統一した。 (2021-02-27)
        if (rss_url.indexOf(pm_rss_legacy) === 0 |rss_url.indexOf(pm_rss_new) === 0) {
          var link = item.getChild('link').getText();
          var title = item.getChild('title', ns_dc).getText();
          var title_escaped = mrkdwnEscape(title);
          var title_ja = LanguageApp.translate(title,'en','ja');
          var authors = item.getChildren('creator', ns_dc);
          if (authors.length > 0) {
            var author = authors[0].getText();
            if (authors.length>1) {
              author += ", _et al._";
            } 
          } else {
            var author = "No Author";
          }

          var journal = item.getChild('source', ns_dc).getText();
          var date = item.getChild('date', ns_dc).getText();
          var pmid = item.getChild('guid').getText().substr(7);
          var journal_issn = '';
          var sjr = '?';
          var snip = '?';
          var citescore = '?';
          if (enable_scopus_database) {
            try {
              var pm_xmlText = UrlFetchApp.fetch(pm_base_url.replace(pm_reptxt,pmid), pm_params).getContentText();
              var pm_xml = XmlService.parse(pm_xmlText).getRootElement().getChild('DocumentSummarySet');
              if (pm_xml.getAttribute('status').getValue() == 'OK') {
                var article = pm_xml.getChild('DocumentSummary');
                try {
                  journal_issn = article.getChild('ISSN').getText();
                } catch (e) {
                  journal_issn = '';
                }
                if (journal_issn == '') {
                  journal_issn = article.getChild('ESSN').getText();
                }
                if (journal_issn != '') {
                  var sc_xmlText = UrlFetchApp.fetch(sc_base_url.replace(sc_reptxt,journal_issn), sc_params).getContentText();
                  var sc_xml = XmlService.parse(sc_xmlText).getRootElement();
                  var sc_journal = sc_xml.getChild('entry');
                  try {
                    journal = sc_journal.getChild('title', ns_dc).getText();
                  } catch (e) { }
                  try {
                    sjr = sc_journal.getChild('SJRList').getChild('SJR').getText();
                  } catch (e) {
                    sjr = '?';
                  }
                  try {
                    snip = sc_journal.getChild('SNIPList').getChild('SNIP').getText();
                  } catch (e) {
                    snip = '?';
                  }
                  try {
                    citescore = sc_journal.getChild('citeScoreYearInfoList').getChild('citeScoreCurrentMetric').getText();
                  } catch (e) {
                    citescore = '?';
                  }
                }
              }
            } catch (e) {
              journal_issn = '';
            }
          }
        } else {
          Logger.log("RSS URL pattern did not match.: "+rss_url);
          continue;
        };
        
        var additional_data = [];
        if (display_pmid) { additional_data.push('PMID=' + pmid); }
        if (display_snip) { additional_data.push('SNIP=' + snip); }
        if (display_citescore) { additional_data.push('CiteScore=' + citescore); }
        if (display_sjr) { additional_data.push('SJR='+ sjr); }
        
        txt += '>*<'+link+'|'+title_escaped+'>*\n>'+title_ja+'\n>'+ author + ' ' + '_*' + journal + '*_ [' + additional_data.join(',') + '] ' + date + '\n  \n';
        
        //5記事ごとに１block生成する
        //1記事ごとにブロックを生成するとブロック数の上限50にひっかかることがある
        //maximum length for the 'text' in a block is 3,000 characters.
        //messageあたりは100,000characters
        //1記事1メッセージしないのは無料版slackでメッセージ数を節約するため
        if ((j%N_IN_BLOCK)==0){
          var l = txt.length;
          blocks.push({
            type: "section",
            text: {
              type:"mrkdwn",
              text: txt
            }
          });
          txt = '';          
        };
        
        j += 1;
      }; 
      
      if ((j%N_IN_BLOCK)!=1){
        blocks.push({
          type: "section",
          text: {
            type:"mrkdwn",
            text: txt
          }
        });       
      };
      
      var payload = {
        text: notify_text.replace(rw_length, news.length).replace(rw_keyword, keyword), //failback sentence
        blocks: blocks
      };
      
      // Post Slack on Incoming Hook
      var options = {
        method: "post",
        headers: {"Content-type": "application/json"},
        payload: JSON.stringify(payload)
      };
      UrlFetchApp.fetch(webhook_url, options);
      
      // Record fetched items
      var new_history = [];
      for (var item of items) {
        new_history.push(item.getChild('guid').getText());
      };
      rss_list[i][3] = JSON.stringify(new_history);
      
    };
    
  };
  
  //historyデータ書き込み
  var history_column = rss_sheet.getRange(2, 4, rss_total_num+1, 4);
  for (i=0; i<rss_total_num; i++) {
    history_column.getCell(i+1, 1).setValue(rss_list[i][3]);
  };
  
};