A program that can obtain records from an EPrints server and send them to public web archiving services such as the the Wayback Machine at the Internet Archive and others.
Table of contents
- Known issues and limitations
- Relationships to other similar tools
- Getting help
- Authors and history
eprints2archives is a self-contained program to archive the web pages of an EPrints server in public web archiving sites such as the Internet Archive. It contacts an EPrints server, obtains the list of documents it serves (optionally filtered based on such things as modification date), determines the document URLs, extracts additional URLs by scraping pages under the
/view section of the public site, and finally, sends the collected URLs to web archives. Use-cases include archiving an server content ahead of migration to another system, and preserving contents in independent third-party archives.
The program is written in Python 3 and works over a network using an EPrints server's REST API and normal HTTP.
eprints2archives can work with EPrints servers that require logins as well as those that allow anonymous access. It uses parallel threads by default, transparently handles rate limits, and robustly deals with network errors.
The instructions below assume you have a Python interpreter installed on your computer; if that's not the case, please first install Python version 3 and familiarize yourself with running Python programs on your system.
On Linux, macOS, and Windows operating systems, you should be able to install
pip. To install
eprints2archives from the Python package repository (PyPI), run the following command:
python3 -m pip install eprints2archives --upgrade
As an alternative to getting it from PyPI, you can use
pip to install
eprints2archives directly from GitHub, like this:
python3 -m pip install git+https://github.com/caltechlibrary/eprints2archives.git --upgrade
After installing it, on Linux and macOS systems you should end up with a program called
eprints2archives in a location normally searched by your terminal shell. You should be able to run
eprints2archives from the shell like any other program. If
eprints2archives ended up in a location not normally searched by your terminal shell, you should still be able to invoke the program as a Python module. For example,
python3 -m eprints2archives -h
On Windows, Python 3 is usually installed as just
python instead of
eprints2archives follows the Windows convention of using
/ as the option prefix instead of a dash (
-). So, instead of the above, you would type
python -m eprints2archives /h
For help with usage at any time, run
eprints2archives with the option
/h on Windows).
eprints2archives contacts the EPrints server whose web address is given as the value to the option
/a on Windows). A typical EPrints server REST API will have a URL of the form
https://server.institution.edu/rest, but you can give it just
eprints2archives will add the
/rest part if it is missing. Note that a value for
-a is required; it cannot infer the server address on its own.
Accessing some EPrints servers via the API requires supplying a user login and password to the server. By default, this program retrieves them from your operating system's user keyring/keychain. If the login and password for a given EPrints server does not exist from a previous run of
eprints2archives, it will ask for the user name and password, and then (unless the
-K option – or
/K on Windows – is given) store them in your keyring/keychain so that it does not have to ask again in the future. It is also possible to supply the information directly on the command line using the
-p options (or
/p on Windows), but this is discouraged because it is insecure on multiuser computer systems. (However, if you need to reset the user name and/or password for some reason, use
-u with a user name and let it prompt for a password again.) If the EPrints server does not require a user name and password, do not use
-p, and supply blank values when prompted for them by
eprints2archives. (Empty user name and password are allowed values.)
How the list of records is determined
The EPrints records to be sent to the web archiving services will be limited to the records indicated by the option
/i on Windows). If no
-i option is given, this program will use all the records available at the given EPrints server. The value of
-i can be one or more integers separated by commas (e.g.,
-i 54602,54604), or a range of numbers separated by a dash (e.g.,
-i 1-100, which is interpreted as the list of numbers 1, 2, ..., 100 inclusive), or some combination thereof. The value of the option
-i can also be a file, in which case, the file is read to get a list of identifiers.
-l option (or
/l on Windows) is given, the records will be additionally filtered to return only those whose last-modified date/time stamp is no older than the given date/time description. Valid descriptors are those accepted by the Python dateparser library. Make sure to enclose descriptions within single or double quotes. Examples:
eprints2archives -l "2 weeks ago" -a .... eprints2archives -l "2014-08-29" -a .... eprints2archives -l "12 Dec 2014" -a .... eprints2archives -l "July 4, 2013" -a ....
-s option (or
/s on Windows) is given, the records will also be filtered to include only those whose
eprint_status field value is one of the listed status codes. Comparisons are done in a case-insensitive manner. Putting a caret character (
^) in front of the status (or status list) negates the sense, so that
eprints2archives will only keep those records whose
<eprint_status> value is not among those given. Examples:
eprints2archives -s archive -a ... eprints2archives -s ^inbox,buffer,deletion -a ...
--status filtering are done after the
-i argument is processed.
By default, if an error occurs when requesting a record or value from the EPrints server,
eprints2archives will keep going, moving on to the next one. Common causes of errors include missing records implied by the arguments to
-i, missing files associated with a given record, and files inaccessible due to permissions errors. If the option
/e on Windows) is given,
eprints2archives will instead stop upon encountering a missing record, or missing file within a record, or similar errors. The default is to only issue warnings because this is less frustrating for most use-cases.
How URLs are constructed
The list of EPrints URLs sent to web archiving sites depends on the command-line options given to
eprints2archives as well as the URLs that actually exist on the server. In a nutshell, if not given a list of identifiers or filtering criteria, it will send URLs for every record along with the URLs of some general pages; by contrast, if the list of records is restricted somehow (by the use of
eprints2archives will only send URLs related to the records identified.
URLs for individual EPrints records
eprints2archives always tries to construct 3 URLs for every EPrint record and verifies that they exist on the EPrints server. The URLs are as follows, where
N is the id number of the EPrint record:
- The value of the field
official_url(if any) in the EPrint record.
The first two typically go to the same page on an EPrint server, but web archiving services have no direct mechanism to indicate that a given URL is an alias or redirection for another, so they need to be sent as separate URLs. On the other hand, the value of
official_url may be an entirely different URL, which may or may not go to the same location as one of the others. For example, in the CaltechAUTHORS EPrint server, the record at
https://authors.library.caltech.edu/85447 has an
official_url value of
https://resolver.caltech.edu/CaltechAUTHORS:20180327-085537493, but the latter is a redirection back to
eprints2archives performs basic validation on the URL values, to make sure they appear to be formally valid and filter out malformed URLs, but does not verify that the locations actually exist.
Additional general URLs
If no selection or filtering is applied (i.e., none of the options
-s are given to
eprints2archives gathers additional URLs as follows (where
SERVER is the server hostname + post number, if any):
- The set of pages
https://SERVER/view/X, where each
Xis obtained by parsing the HTML of
https://SERVER/viewand extracting links to pages under
- The set of pages
https://SERVER/view/X/Y, where each
Yis obtained by parsing the HTML of
https://SERVER/view/Xand extracting links to pages under
On the other hand, if selection and/or filtering are in effect (i.e., if any of the options
-s are used), then
eprints2archives only extracts the URLs
https://SERVER/view/X/N.html for every EPrints identifier
N that is selected via
-i and left after filtering with
-s, if such URLs exist on the server. (E.g., Caltech EPrints servers have a page at
https://SERVER/view/ids/, containing every public EPrint identifier
N linked to a page of the form
https://SERVER/view/ids/N.html. Other servers may have a similar section named something other than
eprints2archives avoids hardwired assumptions and simply looks for pages ending in
The general URLs from one of these two cases (the ones used if no selection or filter is applied, or the ones used when selection and/or filtering are in effect) are combined with the URLs for individual EPrints records to produce the final set of URLs sent to web archiving destinations.
How the destination is determined
eprints2archives has a set of built-in adapters to interact with a number of known public web archiving services. To learn which services
eprints2archives knows about, use the option
/S on Windows). By default, the program will send EPrints record URLs to all the known services. The option
/d on Windows) can be used to select one or a list of destination services instead. Lists of services should be separated by commas with no spaces between them;
eprints2archives will only ask a service to archive a copy of an EPrints record if the service does not already have an archived copy. This makes sense because EPrints records usually change infrequently, and there's little point in repeatedly asking web archives to make new archives. However, if you have reason to want the web archives to re-archive EPrints records, you can use the option
/f on Windows).
eprints2archives will use parallel process threads to query the EPrints server as well as to send records to archiving services. By default, the maximum number of threads used is equal to 1/2 of the number of cores on the computer it is running on. The option
/t on Windows) can be used to change this number.
eprints2archives will always use only one thread per web archiving service (and since there are only a few services, only a few threads are usable during that phase of operation), but a higher number of threads can be helpful to speed up the initial data gathering step from the EPrints server.
Beware that there is a lag between when web archives such as Internet Archive receive a URL submission and when a saved copy is made available from the archive. (For Internet Archive, it is 3-10 hours.) If you cannot find a given EPrints page in an archive shortly after running
eprints2archives, it may be because not enough time has passed.
Other command-line arguments
To save a report of the articles sent, you can use the option
/r on Windows) followed by a file name.
eprints2archives will print informative messages as it works. To limit the messages to warnings and errors only, use the option
/q on Windows). Also, output is color-coded by default unless the
-C option (or
/C on Windows) is given; this option can be helpful if the color control signals create problems for your terminal emulator.
If given the
-@ argument (
/@ on Windows), this program will output a detailed trace of what it is doing, and will also drop into a debugger upon the occurrence of any errors. The debug trace will be written to the given destination, which can be a dash character (
-) to indicate the standard error stream (
sys.stderr), or a file path. Note, however, that if
eprints2archives is being run with Python optimization enabled, then
-@ will have no effect and will be silently ignored.
If given the
-V option (
/V on Windows), this program will print the version and other information to the console, and exit without doing anything else.
Summary of command-line options
The following table summarizes all the command line options available. (Note: on Windows computers,
/ must be used as the prefix character instead of
|Short||Long form opt||Meaning||Default|
||Use A as the server's REST API URL||⚑|
||Send to destination service(s) D||Send to all|
||Stop if encounter missing records||Keep going|
||Send each record even if copy exists||Skip already-archived records|
||Records to get (can be a file name)||Fetch all EPrints records|
||Filter by last-modified date/time||Don't filter by date/time|
||Don't print general info messages||Be chatty while working|
||Filter by status(s) in S||Don't filter by status|
||User name for EPrints server login||No user name|
||Password for EPrints proxy login||No password|
||Don't color-code the output||Color the console messages|
||Don't use a keyring/keychain||Store login info in keyring|
||Print list of known services and exit||Do other actions instead|
||Print program version info and exit||Do other actions instead|
||Debugging mode; write trace to OUT||Normal mode||⚐ ★|
⚑ Required argument.
⚐ To write to the console, use the character
- as the value of OUT; otherwise, OUT must be the name of a file where the output should be written.
★ Has no effect if
eprints2archives is being run with Python optimization enabled.
This program exits with a return code of 0 if no problems are encountered. It returns a nonzero value otherwise, following conventions used in shells such as bash which only understand return code values of 0 to 255. The following table lists the possible return values:
0 = success -- program completed normally 1 = no network detected -- cannot proceed 2 = encountered a bad or missing value for an option 3 = file error -- encountered a problem with a file 4 = the user interrupted the program's execution 5 = an exception or fatal error occurred
Known issues and limitations
Some services impose severe rate limits on URL submissions, and there is nothing that
eprints2archives can do about it. For example, at the time of this writing, Archive.Today only allows 6 URLs to be submitted every 5 minutes. If you plan on sending a large number of URLs, it may be more convenient to use a separate
eprints2archives process with the
-d option to select only one destination, and let it run in its own terminal window.
Relationships to other similar tools
Other tools exist for sending content to web archives; some are general-purpose enough that they could be used to send EPrints server contents to web archives. To the author's knowledge,
eprints2archives is the only software designed specifically to work with EPrints servers to send content to multiple archiving destinations.
The Internet Archive itself offers multiple ways of submitting content, including sending URLs via email, using a browser extension, and the simple-to-use Save Page Now. The latter offers the option of capturing pages more deeply if the user is logged in to their Internet Archive (IA) account. By contrast,
eprints2archives can send the entire content of an EPrints server in one go, without requiring the user to have an IA account.
Many institutions use IA's Archive-It service, and Archive-It can be used to crawl EPrints server pages. In principle, this can capture an EPrints server site more fully than
eprints2archives only follows specific common links and pages, and misses any custom data views or additional pages (including "About" pages) that may be present on an EPrints server. Nevertheless,
eprints2archives can be useful even for sites that use Archive-It because it asks the EPrints server itself about its records and gets URLs (such as the
official_url values) that may not be mentioned in the EPrints record views (and thus would be missed by Archive-It's "on the outside looking in" approach). Since
eprints2archives by default does not send URLs that are already present in archiving destinations, it can be used in conjunction with Archive-It as a secondary or backup scheme.
A number of third-party archiving tools exist for sending URLs to web archives. One of the few that can target other archives besides IA is Archive Now. It partly inspired the creation of
eprints2archives. Archive Now's code for interfacing to web archives also served as initial starting points for figuring out to do the same in
eprints2archives. In terms of functionality,
eprints2archives is more special-purpose (being aimed at extracting content from EPrints servers), and has more advanced capabilities such as handling rate limits, pause-and-retry handling of errors, and use of parallel threads.
Most archiving tools work only with the Internet Archive's Wayback Machine. The Ruby-based WaybackArchiver can crawl a site given a URL or a sitemap and then send URLs to IA; however, it does not appear to attempt to recover from as many error situations as
eprints2archives, and of course, lacks its EPrints-specific features. Waybackpy is a simpler tool designed to interact with IA, and while it can submit URLs to be saved, it lacks other capabilities of
eprints2archives such as automatically handling rate limits and error retries, as well as not being designed for extracting content from EPrints servers. Some other similar but much simpler and less advanced IA-specific tools include savepagenow, ia_wayback, Wayback Sitemap Archive, Save to the Wayback Machine, and the wayback api from HTTPreserve.
archive.today is one of the few tools for working with archives other than IA. It is a simple command-line tool for sending content to and downloading from Archive.Today.
If you find an issue, please submit it in the GitHub issue tracker for this repository.
We would be happy to receive your help and participation with enhancing
eprints2archives! Please visit the guidelines for contributing for some tips on getting started.
Software produced by the Caltech Library is Copyright (C) 2020, Caltech. This software is freely distributed under a BSD/MIT type license. Please see the LICENSE file for more information.
Authors and history
This program was initially written in mid-2020, in response to discussions in Caltech's Digital Library Development group.
The TimeMap parsing code in eprints2archives/services/timemap.py originally came from the Off-Topic Memento Toolkit, by Shawn M. Jones, as it existed on 2020-07-29. The OTMT code is made available according to the MIT license. Acknowledgments and additional information are provided in the file header of eprints2archives/services/timemap.py.
The algorithm and some code for interacting with Archive.Today were borrowed from ArchiveNow, a tool developed by Web Science and Digital Libraries Research Group at Old Dominion University. The authors and contributors of the specific code file used (is_handler.py), as it existed on 2020-07-17, were Mohamed Aturban, Shawn M. Jones, veloute, and evil-wayback. ArchiveNow is made available according to the MIT license. Acknowledgments and additional information are provided in the file header of eprints2archives/services/archivetoday.py.
eprints2archives makes use of numerous open-source packages, without which it would have been effectively impossible to develop
eprints2archives with the resources we had. We want to acknowledge this debt. In alphabetical order, the packages are:
- aenum – advanced enumerations for Python
- appdirs – determine the appropriate app dirs on different OSes
- cssselect –
lxmladd-on to parse CSS3 selectors
- dateparser – parse dates in almost any string format
- dateutil – additional date parsing utilities
- distro – get info about the OS distribution running the current computer
- h2 – HTTP/2 support library used by HTTPX
- httpx – Python HTTP client library that supports HTTP/2
- humanize – helps write large numbers in a more human-readable form
- ipdb – the IPython debugger
- keyring – access the system keyring service from Python
- lxml – an XML parsing library for Python
- plac – a command line argument parser
- pydash – kitchen sink of Python utility libraries for doing “stuff”
- pypubsub – a publish-and-subscribe message-passing library for Python
- rich – library for writing styled text to the terminal
- setuptools – library for
- tldextract – module to parse domains from URLs
- urllib3 – HTTP client library for Python
- validators – data validation package for Python
This work was funded by the California Institute of Technology Library.