scraping_classifying.r

# Disable Warnings (For Ratings Not Present)
options(warn=-1)

# Install Packages
if(!"rvest" %in% installed.packages()[,"Package"])install.packages("rvest")

# Load Packages
library(rvest)
library("sentimentr")
library("dplyr")
library(RTextTools)


# Variables
reviews_flag<- 0
args<- commandArgs(TRUE)
file_name<- "test_data.txt"
counter=0
acc_counter=0

# Source
src_array<- c(args[1])
#src_array<- c("https://www.amazon.com/Raspberry-Pi-Desktop-Starter-White/dp/B01CI58722/ref=sr_1_1_sspa?s=pc&ie=UTF8&qid=1515154250&sr=1-1-spons&keywords=raspberry+pi&psc=1&smid=ACZ7BCW2EXCBA")

# Code - - - -


#if(file.exists(file_name)){
#  write("",file_name,append=FALSE,sep='')
#}


actual = c()
predicted=c()

# For Number Of Sources in Array
for(k in 1:length(src_array)){


    source<- src_array[k]
    download.file(source, destfile = "scrapedpage.html", quiet=TRUE)
get_pages<- read_html("scrapedpage.html")

    # Get Reviews
    reviews<- get_pages %>% html_nodes(".review")
    get_number<- length(reviews)

    # Check If No Reviews
    if(length(reviews)==0){
      reviews_flag<- 1
      break()
    }




    # For Number Of Reviews
    for(i in 1:get_number){

      get_total_data<- reviews[i] %>% html_nodes(".a-row")

      # Check If Ratings Are Present On A Node
      node_for_rating<- 1
      check_valid<- get_total_data[node_for_rating] %>% html_node(".a-link-normal")
      if(is.na(check_valid)){
        node_for_rating<- 2
      }

      # Get Stars
      stars_title<- get_total_data[node_for_rating] %>% html_node(".a-link-normal") %>% html_attr("title")
      stars_string<-substring(stars_title,1,3)
      stars<-as.double(stars_string)
      if (is.na(stars))
        break()



      # Get Review Text
      review_data<- get_total_data[4]  %>% html_node("span") %>% html_text(trim=TRUE)
      check_read_now<- substring(review_data,nchar(review_data)-9+1,nchar(review_data))
      if (check_read_now == "Read more"){
        review_data<- substring(review_data,1,nchar(review_data)-9)
      }


      # Add To Corresponding Files

      if(stars<2.5){
        remark<- "negative"
      } else if(stars>3.5) {
        remark<- "positive"
      } else {
        remark<- "neutral"
      }

      sample_set = get_sentences(review_data)
      sentiment(sample_set) %>%
        subset(select = "sentiment") %>%
        colSums() -> result

      # *100 to identify neutral sentiments.
      result = as.integer(result*100);
      exp=""


      if(result==0){
        exp="neutral"
      }else if(result > 0){
        exp="positive"
      }else
        exp="negative"


      actual= c(actual,remark)
        predicted = c(predicted,exp)



    }

}

x=table(predicted, actual)
y=recall_accuracy(predicted,actual)

out = file(file_name, 'w')
capture.output( print(x, print.gap=3), file=out)
write(paste('\n\t\t Accuracy%: ',y,sep=''), file=out,append=TRUE)
close(out)
# End Code - - - - - -

# Enable Warnings Again
options(warn=0)
print("write finished")