THIS WILL LIKELY BE UNDERGOING MAJOR CHANGES
See: https://github.com/hrbrmstr/jwatr for a new spin on the WARC ecosystem in R
The current thought is to move higher order functions to here (read/write WARC) and expose lower-level stuff via jwatr, but jwatr may also just become warc for a CRAN release.
warc : Tools to Work with the Web Archive Ecosystem
WARC files (and the metadata files that usually follow them) are the de facto method of archiving web content. There are tools in Python & Java to work with this data and there are many "big data" tools that make working with large-scale data from sites like Common Crawl and The Internet Archive very straightforward.
Now there are tools to create and work with the WARC ecosystem in R.
Possible use-cases:
- If you need to scrape data from many URLs and would like to make the analyses on that data reproducible but are concerned that the sites may change format or may be offline but also don't want to manage individual HTML (etc) files
- Analyzing Common Crawl data (etc) natively in R
- Saving the entire state of an
httrrequest (warccan turnhttrresponses into WARC files and turns WARCresponserecords intohttr::responseobjects)
warc can work with WARC files that are composed of individual gzip streams or on plaintext WARC files and can also read & generate CDX files. Support for more file types (e.g. WET, WAT, etc) are planned.
Since I ended up making some gz file functions for this package, it only seemed appropriate to expose them.
The following functions are implemented:
as_warc: Convert an ‘httr::respone’ object to WARC response objectscreate_cdx: Takes as input an optionally compressed WARC file and creates acreate_warc_wget: Newer versions of ‘wget’ are designed to support capturing of orgz_close: Close the gz filegz_eof: Test for end of filegz_flush: This will flush all zlib output buffers for the current file andgz_fseek: Sets the starting position for the next ‘gz_read()’ orgz_gets: Read a line from a gz filegz_gets_raw: Read a line from a gz filegz_offset: Return the current raw compressed offset in the filegz_open: Open a gzip file for reading or writinggz_read_char: Read from a gz file into a character vectorgz_read_raw: Read from a gz file into a raw vectorgz_seek: Sets the starting position for the next ‘gz_read()’ orgz_tell: Return the current raw uncompressedf offset in the filegz_write_char: Write an atomic character vector to a filegz_write_raw: Write a raw vector to a gz filegzip_inflate_from_pos: Given a gzip file that was built with concatenated individual gzipread_cdx: CDX files are used to index the content of WARC files.read_warc_entry: Given the path to a WARC file (compressed or uncompressed) and thewarc_headers: Extract WARC headers from a WARC response objectwrite_warc_record: Write a WARC record to a file
You need wget on your system PATH. Folks on real operating systems can do the apt-get, yum install or brew install (et al) dance for your particular system. Version 1.18+ is recommended, but any version with support for WARC extensions should do.
Windows folks will need to grab the statically linked 32-bit or 64-bit binaries from here and put them on your system PATH somewhere if you want to create WARC files in bulk using wget.
devtools::install_git("https://gitlab.com/hrbrmstr/warc.git")library(warc)
library(httr)
# current verison
packageVersion("warc")## [1] '0.1.0'
cdx <- read_cdx(system.file("extdata", "20160901.cdx", package="warc"))
i <- 5
path <- file.path(cdx$warc_path[i], cdx$file_name[i])
start <- cdx$compressed_arc_file_offset[i]
entry <- read_warc_entry(path, start)
print(entry)## Response [https://r-project.org/]
## Date: 2016-09-01 18:03
## Status: 301
## Content-Type: text/html; charset=iso-8859-1
## Size: 314 B
## <!DOCTYPE HTML PUBLIC "-//IETF//DTD HTML 2.0//EN">
## <html><head>
## <title>301 Moved Permanently</title>
## </head><body>
## <h1>Moved Permanently</h1>
## <p>The document has moved <a href="https://www.r-project.org/">here</a>.</p>
## <hr>
## <address>Apache/2.4.10 (Debian) Server at r-project.org Port 443</address>
## </body></html>
print(warc_headers(entry))## $`warc-type`
## [1] "response"
##
## $`warc-record-id`
## [1] "<urn:uuid:9EB1B94E-A929-4B5E-AE10-FB54C2E7308B>"
##
## $`warc-warcinfo-id`
## [1] "<urn:uuid:7D7DC0CA-8FC7-4FFB-9D5A-4187EB50ED74>"
##
## $`warc-concurrent-to`
## [1] "<urn:uuid:592019F2-C037-4899-8DB2-CC04266B9E29>"
##
## $`warc-target-uri`
## [1] "https://r-project.org/"
##
## $`warc-date`
## [1] "2016-09-01 18:03:46 UTC"
##
## $`warc-ip-address`
## [1] "137.208.57.37"
##
## $`warc-block-digest`
## [1] "\020"
##
## $`warc-payload-digest`
## [1] ":\001"
##
## $`content-type`
## [1] "application/http;msgtype=response"
##
## $`content-length`
## [1] 578
##
## attr(,"class")
## [1] "insensitive" "list"
print(status_code(entry))## [1] 301
print(http_type(entry))## [1] "text/html"
library(warc)
library(purrr)
library(rvest)
warc_dir <- file.path(tempdir(), "rfolks")
dir.create(warc_dir)
urls <- c("http://rud.is/",
"http://hadley.nz/",
"http://dirk.eddelbuettel.com/",
"https://jeroenooms.github.io/",
"https://ironholds.org/")
create_warc_wget(urls, warc_dir, warc_file="rfolks-warc")
cdx <- read_cdx(file.path(warc_dir, "rfolks-warc.cdx"))
sites <- map(1:nrow(cdx),
~read_warc_entry(file.path(cdx$warc_path[.],
cdx$file_name[.]),
cdx$compressed_arc_file_offset[.]))
map(sites, ~read_html(content(., as="text", encoding="UTF-8"))) %>%
map_chr(~html_text(html_nodes(., "title")))## [1] "rud.is" "Hadley Wickham" "Dirk Eddelbuettel" "About Jeroen..." "Oliver Keyes - Home"
unlink(warc_dir)library(warc)
library(testthat)
date()## [1] "Mon Sep 12 15:27:08 2016"
test_dir("tests/")## testthat results ========================================================================================================
## OK: 0 SKIPPED: 0 FAILED: 0
##
## DONE ===================================================================================================================