How to use post within rvest's html_session

As @chinsoon12 mentioned in the comment; it's against the term of use of TripAdvisor to scrape the information. But I just would like to know how to use POST method in rvest with this example. I've searched on google and stackoverflow but the answers are not very helpful. Any general suggestions are also appreciated!

Here is the website: https://www.tripadvisor.com/Hotel_Review-g33657-d85704-Reviews-Hotel_Bristol-Steamboat_Springs_Colorado.html

So, I need to click "more" button to view full reviews. Otherwise it only gives partial reviews.

I've succeeded using Rselenium to simulate clicks and get full reviews, but I want to know how to do it with rvest and httr.

After observing the network traffic, I discovered that after I click the "more" button, I sent two POST requests listed below:

I tried this code below, but the body is empty.

library(rvest)
library(httr)
url <- "https://www.tripadvisor.com/Hotel_Review-g33657-d85704-Reviews-Hotel_Bristol-Steamboat_Springs_Colorado.html"
post_to_url <-"https://www.tripadvisor.com/OverlayWidgetAjax?Mode=EXPANDED_HOTEL_REVIEWS_RESP&metaReferer="
user_agent_table <- read.csv("https://raw.githubusercontent.com/yusuzech/top-50-user-agents/master/user_agent.csv",stringsAsFactors = F)
post_body <- "reviews=556957481%2C511497076%2C556144452%2C554686822%2C548218482&contextChoice=DETAIL_HR&haveJses=earlyRequireDefine%2Camdearly%2Cglobal_error%2Clong_lived_global%2Capg-Hotel_Review%2Capg-Hotel_Review-in%2Cbootstrap%2Cdesktop-rooms-guests-dust-en_US%2Cresponsive-calendar-templates-dust-en_US%2Ctaevents&haveCsses=apg-Hotel_Review-in&Action=install"
user_agent_list <- user_agent_table$User.agent
random_agent <- function(){user_agent(sample(user_agent_list,1))}

mysession <- html_session(url,random_agent())
result <- POST(url=post_to_url,
               config=list(referer = mysession$url),
               user_agent(mysession$config$options$useragent),
               body=post_body,
               encode="raw")
result

> result
Response [https://www.tripadvisor.com/OverlayWidgetAjaxMode=EXPANDED_HOTEL_REVIEWS_RESP&metaReferer=]
Date: 2018-05-10 01:49
Status: 200
Content-Type: text/html;charset=UTF-8
<EMPTY BODY>

I know I need to use POST method, but I don't know how to set body and other configurations. I'm also not sure if I have to send two post requests simultaneously and how to achieve this in httr and rvest.

Any help is appreciated!

Solution

I tried to post rvest:::request_POST like you did but failed. It got 'Method Not Allowed (HTTP 405)' error message. But it turns out we don't have to post rvest:::request_POST like that, the full reviews are already in the source, they're just not displayed by default. Here is the code to crawl all reviews on this hotel:

library(rvest)
library(stringr)

reviews_df <- data.frame(reviewers = character(),
                        review_dates = character(),
                        stars = integer(),
                        contributions = integer(),
                        helpful_votes = integer(),
                        review_titles = character(),
                        reviews = character())
pages_url <- character();
pages_url[1] <- "https://www.tripadvisor.com/Hotel_Review-g33657-d85704-Reviews-The_Bristol_by_Magnuson_Worldwide-Steamboat_Springs_Colorado.html#REVIEWS";
for (i in 2:42) pages_url[i] <- paste("https://www.tripadvisor.com/Hotel_Review-g33657-d85704-Reviews-or", as.character(5 * (i-1)),"-The_Bristol_by_Magnuson_Worldwide-Steamboat_Springs_Colorado.html#REVIEWS", sep = "");
for (i in 1:42){
  reviewers <- character();
  review_dates <- character();
  stars <- integer();
  contributions <- integer();
  helpful_votes <- integer();
  review_titles <- character();
  reviews <- character();
  page <- read_html(pages_url[i]);
  review_nodes <- page %>% html_nodes(xpath = "//div[@data-test-target='reviews-tab']/div[@data-test-target='HR_CC_CARD']");
  reviewers <- review_nodes %>% html_nodes(xpath = "./descendant::a[contains(@class, 'social-member-event-MemberEventOnObjectBlock__member')]") %>% html_text(trim = TRUE);
  review_dates <- review_nodes %>% html_nodes(xpath = "./descendant::a[contains(@class, 'social-member-event-MemberEventOnObjectBlock__member')]/following-sibling::text()[1]") %>% html_text(trim = TRUE);
  review_dates <- str_sub(review_dates, start = 16);
  star_classes <- review_nodes %>% html_nodes(xpath = "./descendant::div[@data-test-target='review-rating']/span[contains(@class, 'ui_bubble_rating')]") %>% html_attr("class");
  for (j in 1:length(review_nodes)){
    if (grepl("bubble_10", star_classes[j], fixed = TRUE)) stars[j] <- 1
    else if (grepl("bubble_20", star_classes[j], fixed = TRUE)) stars[j] <- 2
    else if (grepl("bubble_30", star_classes[j], fixed = TRUE)) stars[j] <- 3
    else if (grepl("bubble_40", star_classes[j], fixed = TRUE)) stars[j] <- 4
    else if (grepl("bubble_50", star_classes[j], fixed = TRUE)) stars[j] <- 5;
    if (length(review_nodes[j] %>% html_nodes(xpath = "./descendant::span[contains(@class, 'social-member-MemberHeaderStats__stat_item') and contains(., 'contribution')]/span")) == 0) contributions[j] <- 0
    else contributions[j] <- as.numeric(review_nodes[j] %>% html_nodes(xpath = "./descendant::span[contains(@class, 'social-member-MemberHeaderStats__stat_item') and contains(., 'contribution')]/descendant::span[contains(@class, 'social-member-MemberHeaderStats__bold')]") %>% html_text());
    if (length(review_nodes[j] %>% html_nodes(xpath = "./descendant::span[contains(@class, 'social-member-MemberHeaderStats__stat_item') and contains(., 'helpful vote')]/span")) == 0) helpful_votes[j] <- 0
    else helpful_votes[j] <- as.numeric(review_nodes[j] %>% html_nodes(xpath = "./descendant::span[contains(@class, 'social-member-MemberHeaderStats__stat_item') and contains(., 'helpful vote')]/descendant::span[contains(@class, 'social-member-MemberHeaderStats__bold')]") %>% html_text());
  }
  review_titles <- review_nodes %>% html_nodes(xpath = "./descendant::div[@data-test-target='review-title']") %>% html_text(trim = TRUE);
  reviews <- review_nodes %>% html_nodes(xpath = "./descendant::q[contains(@class, 'location-review-review-list-parts-ExpandableReview__reviewText')]") %>% html_text(trim = TRUE);
  pgreviews_df <- data.frame(reviewers, review_dates, stars, contributions, helpful_votes, review_titles, reviews)
  reviews_df <- rbind(reviews_df, pgreviews_df);
}