[R] Scraping from different level URLs website

[Ilio Fornasero]

I am doing a research on World Bank (WB) projects on developing countries. To do so, I am scraping their website in order to collect the data I am interested in.

The structure of the webpage I want to scrape is the following:

  1.  List of countries the list of all countries in which WB has developed projects<http://projects.worldbank.org/country?lang=en&page=>

1.1. By clicking on a single country on 1. , one gets the single countries project list (that includes many webpages) it includes all the projects in a single countries <http://projects.worldbank.org/search?lang=en&searchTerm=&countrycode_exact=3A> . Of course, here I have included just one page of a single countries, but every country has a number of pages dedicated to this subject

1.1.1. By clicking on a a single project on 1.1. , one gets - among the others - the project's overview option<http://projects.worldbank.org/P155642/?lang=en&tab=overview> I am interested in.

In other words, my problem is to find out a way to create a dataframe including all the countries, a complete list of all projects for each country and an overview of any single project.


Yet, this is the code that I have (unsuccessfully) written:

WB_links <- "http://projects.worldbank.org/country?lang=en&page=projects"

 WB_proj <- function(x) {

  Sys.sleep(5)
 url <- sprintf("http://projects.worldbank.org/search?lang=en&searchTerm=&countrycode_exact=%s", x)

 html <- read_html(url)

 tibble(title = html_nodes(html, ".grid_20") %>% html_text(trim = TRUE),
     project_url = html_nodes(html, ".grid_20") %>% html_attr("href"))
    }

 WB_scrape <- map_df(1:5, WB_proj) %>%
 mutate(study_description =
       map(project_url,
           ~read_html(sprintf
     ("http://projects.worldbank.org/search?lang=en&searchTerm=&countrycode_exact=%s", .x)) %>%
            html_node() %>%
            html_text()))


Any suggestion?

Note: I am sorry if this question seems trivial, but I am quite a newbie in R and I haven't found a help on this by looking around (though I could have missed something, of course).


	[[alternative HTML version deleted]]