pubmedText.R

library(ff)
pubmedText <- function(url) {
    source("D:/Program Files/RStudio/RFile/textMining/regexp.R")
    # get the article's pmid, title, abstract, journal name and
    # publish date. url url 输入的网址链接 e.g.
    # http://eutils.ncbi.nlm.nih.gov/entrez/eutils/efetch.fcgi?db=pubmed&id=24961184,24961183,24961148,24960992,24960298
    # &retmode=xml&rettype=abstract
    text <- readLines(url, encoding = "UTF-8")
    # 读取网页内容，encoding
    # 不设置会在使用substring()时出现multibite 读取错误
    text <- paste(text, collapse = "")  #多行文本连成一行，有利于使用正则表达式
    # 将 文本按照 ariticle 进行分割，对每一个ariticle进行处理
    articleStr <- "<PubmedArticle>.*?</PubmedArticle>"
    article <- gregexpResult(articleStr, text)  #自写函数
    
    # pmid
    pmidStr <- "(?<=<PMID Version=\"1\">)[0-9]*?(?=</PMID>)"
    pmid <- regexpResult(pmidStr, article)
    
    # <ArticleTitle>
    titleStr <- "(?<=<ArticleTitle>).*?(?=</ArticleTitle>)"
    title <- regexpResult(titleStr, article)
    
    # </Abstract>
    abstractStr <- "(?<=<Abstract>).*?(?=</Abstract>)"
    abstract <- regexpResult(abstractStr, article)
    abstract <- gsub(pattern = "(<AbstractText.*?>)*(</AbstractText>)*", 
        replacement = "", x = abstract, perl = T)
    abstract <- gsub(pattern = "\\s+", replacement = " ", x = abstract, 
        perl = T)
    # 多个空格替换成一个空格
    
    # <Title>
    journalStr <- "(?<=<ISOAbbreviation>).*?(?=</ISOAbbreviation>)"
    journal <- regexpResult(journalStr, article)
    
    # <PubDate>\n<Year>
    yearStr <- "(?<=<PubDate>).*?(?=</PubDate>)"
    year <- regexpResult(yearStr, article)
    # 去除杂质 此处分多次剔除，写在一起出错了，错因不明待查
    year <- gsub(pattern = "(<Year>)*(<Month>)*", replacement = "", 
        x = year, perl = T)
    
    year <- gsub(pattern = "(<Year>)*(</Year>)*(<Month>)*(</Month>)*(<Day>.*</Day>)*(\\s+)", 
        replacement = "", x = year, perl = T)
    year <- gsub(pattern = "(<MedlineDate>)*(</MedlineDate>)*", 
        replacement = "", x = year, perl = T)
    
    pubmed <- data.frame(pmid, title, abstract, journal, year)  #存储
    bgtext <- as.ffdf(pubmed)  #转为ffdf格式 （ff package）
}