ores_assessment.py

import sys
import pywikibot
import json, urllib
import datetime
from language_dict import language_dict

# TLDR: The main function, get_ores_assessment, takes in a list of article names and
# a language (English, Russian, or French). It then prints out each article's name,
# along with an automatically generated (machine learning oooohh) assessment of that
# article (Featured Article, Good Article, A-Class Article, B-Class Article, etc)
# This serves as one possible criterion by which users can sort articles.

# ORES Wiki: https://www.mediawiki.org/wiki/ORES
# ORES Documentation: https://ores.wikimedia.org/v3/

# Base ORES API URL
url = "http://ores.wmflabs.org/v3/scores/"
# url_example = "https://ores.wmflabs.org/v3/scores/enwiki/?models=wp10&revids=34854345"

# Note: We only use wp10 because this is what will give us the current article quality.
# We are not extremely concerned with draftquality

# Unfortunately, only three languages currently support automatic ORES wp10 assessment
# We hope that the impact of this function for sorting by article quality will increase
# as more languages begin to support ORES
ORES_SUPPORTED_WIKIS = {"en": "enwiki",
                    "ru": "ruwiki",
                    "fr": "frwiki"
}

# def generate_language_dict():
#     with open("list_of_wiki_languages.txt", "r") as file:
#         lines = file.read().split(",")
#         for i in range(len(lines)):
#             lines[i] = lines[i].strip()
#             lines[i] = lines[i].strip("\'")
#         dictionary = {lines[i+1]:lines[i] for i in range(0, len(lines), 2)}
#     return dictionary


###############
#MAIN FUNCTION#
###############

# Returns the ORES automatic article quality assessment for the given article_name and language_code
def get_ores_assessment(article_names, language_code):
    assert language_code in ORES_SUPPORTED_WIKIS, "ORES not supported in " + language_code

    built_url, rev_ids, = build_ores_url(article_names, language_code)
    with urllib.request.urlopen(built_url) as url:
        data = json.loads(url.read().decode())
        all_assessments = get_article_assessments(data)
        ores_rating_results = {}
        for i in range(len(article_names)):
            print("Article Name: " + article_names[i])
            print("ORES Assessment: " + all_assessments[rev_ids[i]])
            ores_rating_results[article_names[i]] = all_assessments[rev_ids[i]]
        return ores_rating_results

##################
#HELPER FUNCTIONS#
##################

# Builds the JSON request URL
def build_ores_url(article_names, language_code):
    # Get Wikipedia's revision id for the articles' most recent revision (aka the current version)
    site = pywikibot.getSite(language_code)
    rev_ids = []
    for article_name in article_names:
        page = pywikibot.Page(site, article_name)
        rev_id = page.latest_revision.hist_entry().revid
        rev_ids.append(str(rev_id))
    # The context is essentially the same as the language (only three languages/contexts are supported at this time)
    context = ORES_SUPPORTED_WIKIS[language_code]

    result = url + context + "/?models=wp10&revids=" + "|".join(rev_ids)

    return result, rev_ids


# Returns the ORES automatic article assessment (see following link for table of assessments:
# https://www.mediawiki.org/wiki/ORES#/media/File:Article_quality_and_importance.wp10bot.enwiki.png)
def get_article_assessments(data):
    print(data)
    scores = {}
    # first index 0 is the wiki (enwiki, ruwiki, or frwiki)
    # second index 0 is the rev_id
    scores_json = list(data.values())[0]["scores"]
    for article in scores_json:
        score = scores_json[article]["wp10"]["score"]
        scores[article] = score["prediction"]
    return scores

def scale_article_assessments(article_rating_results):
    article_rating_rules = {
        "FA" : 5,
        "GA" : 4,
        "B" : 3,
        "C" : 2,
        "START" : 1,
        "STUB" : 0,

        "ADQ" : 6,
        "BA" : 5,
        "A" : 4,
        "B" : 3,
        "BD" : 2,
        "E" : 1,

        "ХС" : 7,
        "ИС" : 6,
        "ДС" : 5,
        "IV" : 4,
        "III" : 3,
        "II" : 2,
        "I" : 1

    }
    for name in article_rating_results:
        letter_grade = article_rating_results[name].upper()
        article_rating_results[name] = article_rating_rules[letter_grade]
    return article_rating_results

######################
#EXAMPLE / HOW TO USE#
######################

if __name__ == "__main__":

    get_ores_assessment(["List of tallest buildings in Las Vegas", "Rhode Island", "Car"], language_dict["Spanish"])

# Performance notes:
# After an article has been queried, it seems that it is cached automatically, so if it is queried again,
# the process is much faster.