htmldf (version 0.6.0)

html_df: Get a tabular summary of webpage content from a vector of urls

Description

From a vector of urls, html_df() will attempt to fetch the html. From the html, html_df() will attempt to look for a page title, rss feeds, images, embedded social media profile handles and other page metadata. Page language is inferred using the package cld3 which wraps Google's Compact Language Detector 3.

Usage

html_df(
  urlx,
  max_size = 5e+06,
  wait = 0,
  retry_times = 0,
  time_out = 30,
  show_progress = TRUE,
  keep_source = TRUE,
  chrome_bin = NULL,
  chrome_args = NULL,
  ...
)

Value

A tibble with columns

• url the original vector of urls provided

• title the page title, if found

• lang inferred page language

• url2 the fetched url, this may be different to the original, for example if redirected

• links a list of tibbles of hyperlinks found in <a> tags

• rss a list of embedded RSS feeds found on the page

• tables a list of tables found on the page in descending order of size, coerced to tibble wherever possible.

• images list of tibbles containing image links found on the page

• social list of tibbles containing twitter, linkedin and github user info found on page

• code_lang numeric indicating inferred code language. A negative values near -1 indicates high likelihood that the language is python, positive values near 1 indicate R. If not code tags are detected, or the language could not be inferred, value is NA.

• size the size of the downloaded page in bytes

• server the page server

• accessed datetime when the page was accessed

• published page publication or last updated date, if detected

• generator the page generator, if found

• status HTTP status code

• source character string of xml documents. These can each be coerced to xml_document for further processing using rvest using xml2:read_html().

Arguments

urlx

A character vector containing urls. Local files must be prepended with file://.

max_size

Maximum size in bytes of pages to attempt to parse, defaults to 5000000. This is to avoid reading very large pages that may cause read_html() to hang.

wait

Time in seconds to wait between successive requests. Defaults to 0.

retry_times

Number of times to retry a URL after failure.

time_out

Time in seconds to wait for httr::GET() to complete before exiting. Defaults to 30.

show_progress

Logical, defaults to TRUE. Whether to show progress during download.

keep_source

Logical argument - whether or not to retain the contents of the page source column in the output tibble. Useful to reduce memory usage when scraping many pages. Defaults to TRUE.

chrome_bin

(Optional) Path to a Chromium install to use Chrome in headless mode for scraping

chrome_args

(Optional) Vector of additional command-line arguments to pass to chrome

...

Additional arguments to `httr::GET()`.

Author

Alastair Rushworth

Examples

# Examples require an internet connection...
urlx <- c("https://github.com/alastairrushworth/htmldf", 
          "https://alastairrushworth.github.io/")
dl   <- html_df(urlx)
# preview the dataframe
head(dl)
# social tags
dl$social
# page titles
dl$title
# page language
dl$lang
# rss feeds
dl$rss
# inferred code language
dl$code_lang
# print the page source
dl$source