Find the creation date of web pages using a combination of tree traversal, common structural patterns, text-based heuristics and robust date extraction.
Project description
- Code:
- Documentation:
- Issue tracker:
In a nutshell, with Python:
>>> from htmldate import find_date
>>> find_date('http://blog.python.org/2016/12/python-360-is-now-available.html')
'2016-12-23'
>>> find_date('https://netzpolitik.org/2016/die-cider-connection-abmahnungen-gegen-nutzer-von-creative-commons-bildern/', original_date=True)
'2016-06-23'
On the command-line:
$ htmldate -u http://blog.python.org/2016/12/python-360-is-now-available.html
'2016-12-23'
Features
htmldate finds original and updated publication dates of web pages using a combination of tree traversal, common structural patterns, text-based heuristics and robust date extraction. All the steps needed from web page download to HTML parsing, including scraping and textual analysis are handled. URLs, HTML files or HTML trees are given as input, the library outputs a date string in the desired format.
htmldate provides following ways to date a HTML document:
Markup in header: common patterns are used to identify relevant elements (e.g. link and meta elements) including Open Graph protocol attributes and a large number of CMS idiosyncracies
HTML code: The whole document is then searched for structural markers: abbr/time elements and a series of attributes (e.g. postmetadata)
Bare HTML content: A series of heuristics is run on text and markup:
in fast mode the HTML page is cleaned and precise patterns are targeted
in extensive mode date expressions are collected and the best one is chosen based on a disambiguation algorithm
The module then returns a date if a valid cue could be found in the document, per default the updated date w.r.t. the original publishing statement. The output string defaults to ISO 8601 YMD format.
Should be compatible with all common versions of Python 3 (see tests and coverage)
Safety belt included, the output is thouroughly verified with respect to its plausibility and adequateness
Designed to be computationally efficient and used in production on millions of documents
Handles batch processing of a list of URLs
Switch between original and updated date
The library currently focuses on texts written in English or German.
Installation
The package is tested on Linux, macOS and Windows systems, it is compatible with Python 3.5 upwards.
Install from package repository: pip install htmldate
For the latest version (check build status above): pip install git+https://github.com/adbar/htmldate.git
For faster processing of downloads you might consider installing the cchardet package as well (currently not working on some macOS versions).
With Python
>>> from htmldate import find_date
>>> find_date('http://blog.python.org/2016/12/python-360-is-now-available.html')
'2016-12-23'
The module can resort to a guess based on a complete screning of the document (extensive_search parameter) which can be deactivated:
>>> find_date('https://creativecommons.org/about/')
'2017-08-11' # has been updated since
>>> find_date('https://creativecommons.org/about/', extensive_search=False)
>>>
Input format
The module expects strings as shown above, it is also possible to use already parsed HTML (i.e. a LXML tree object):
# simple HTML document as string
>>> htmldoc = '<html><body><span class="entry-date">July 12th, 2016</span></body></html>'
>>> find_date(mytree)
'2016-07-12'
# parsed LXML tree
>>> from lxml import html
>>> mytree = html.fromstring('<html><body><span class="entry-date">July 12th, 2016</span></body></html>')
>>> find_date(mytree)
'2016-07-12'
Date format
The output format of the dates found can be set in a format known to Python’s datetime module, the default being %Y-%m-%d:
>>> find_date('https://www.gnu.org/licenses/gpl-3.0.en.html', outputformat='%d %B %Y')
'18 November 2016' # may have changed since
Original date
Although the time delta between the original publication and the last modified statement is usually a matter of hours or days at most, it can be useful in some contexts to prioritize the original publication date during extraction:
>>> find_date('https://netzpolitik.org/2016/die-cider-connection-abmahnungen-gegen-nutzer-von-creative-commons-bildern/') # default setting
'2019-06-24'
>>> find_date('https://netzpolitik.org/2016/die-cider-connection-abmahnungen-gegen-nutzer-von-creative-commons-bildern/', original_date=True) # modified behavior
'2016-06-23'
On the command-line
A basic command-line interface is included:
$ htmldate -u http://blog.python.org/2016/12/python-360-is-now-available.html
'2016-12-23'
For usage instructions see htmldate -h:
$ htmldate --help
htmldate [-h] [-v] [-f] [--original] [-m MAXDATE] [-i INPUTFILE] [-u URL]
- optional arguments:
- -h, --help
show this help message and exit
- -v, --verbose
increase output verbosity
- -f, --fast
fast mode: disable extensive search
- --original
original date prioritized
- -m MAXDATE, --maxdate MAXDATE
latest acceptable date (YYYY-MM-DD)
- -i INPUTFILE, --inputfile INPUTFILE
name of input file for batch processing (similar to wget -i)
- -u URL, --URL URL
custom URL download
The batch mode -i takes one URL per line as input and returns one result per line in tab-separated format:
$ htmldate -i list-of-urls.txt
Additional information
Going further
For more details check the online documentation: htmldate.readthedocs.io
If the date is nowhere to be found, it might be worth considering carbon dating the web page, however this is computationally expensive. In addition, datefinder features pattern-based date extraction for texts written in English.
Pull requests are welcome.
Context
This module is part of methods to derive metadata from web documents in order to build text corpora for computational linguistic and NLP analysis, the original problem being that there are web pages for which neither the URL nor the server response provide a reliable way to date the document, i.e. find when it was first published and/or last modified. For more information:
Barbaresi, Adrien. “The Vast and the Focused: On the need for domain-focused web corpora”, Proceedings of the 7th Workshop on Challenges in the Management of Large Corpora (CMLC-7), 2019.
Barbaresi, Adrien. “Efficient construction of metadata-enhanced web corpora”, Proceedings of the 10th Web as Corpus Workshop (WAC-X), 2016.
Kudos to…
cchardet, ciso8601, lxml, dateparser (although it is a bit slow)
A few patterns are derived from python-goose, metascraper, newspaper and articleDateExtractor. This module extends their coverage and robustness significantly.
Contact
See my contact page for details.
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.