Merge pull request #52 from earwig/feat/auth

Require OAuth for search engine checks

.gitignore

@@ -7,3 +7,4 @@ __pycache__
 .earwigbot
 logs/*
 !logs/.gitinclude
+config.py

README.md

@@ -13,6 +13,7 @@ Dependencies
 * [flask-mako](https://pythonhosted.org/Flask-Mako/) >= 0.3
 * [mako](https://www.makotemplates.org/) >= 0.7.2
 * [mwparserfromhell](https://github.com/earwig/mwparserfromhell) >= 0.3
+* [mwoauth](https://github.com/mediawiki-utilities/python-mwoauth) == 0.3.8
 * [oursql](https://pythonhosted.org/oursql/) >= 0.9.3.1
 * [requests](https://requests.readthedocs.io/) >= 2.9.1
 * [SQLAlchemy](https://www.sqlalchemy.org/) >= 0.9.6

app.py

@@ -1,18 +1,19 @@
 #! /usr/bin/env python
 # -*- coding: utf-8  -*-
-
 from functools import wraps
 from hashlib import md5
 from json import dumps
 from logging import DEBUG, INFO, getLogger
 from logging.handlers import TimedRotatingFileHandler
+from multiprocessing import Value
 from os import path
 from time import asctime
 from traceback import format_exc
+from urllib import quote_plus, quote
 
 from earwigbot.bot import Bot
 from earwigbot.wiki.copyvios import globalize
-from flask import Flask, g, make_response, request
+from flask import Flask, g, make_response, request, redirect, session
 from flask_mako import MakoTemplates, render_template, TemplateError
 
 from copyvios.api import format_api_error, handle_api_request
@@ -21,12 +22,14 @@ from copyvios.cookies import parse_cookies
 from copyvios.misc import cache, get_notice
 from copyvios.settings import process_settings
 from copyvios.sites import update_sites
+from copyvios.auth import oauth_login_start, oauth_login_end, clear_login_session
 
 app = Flask(__name__)
 MakoTemplates(app)
 
 hand = TimedRotatingFileHandler("logs/app.log", when="midnight", backupCount=7)
 hand.setLevel(DEBUG)
+app.config.from_pyfile("config.py", True)
 app.logger.addHandler(hand)
 app.logger.info(u"Flask server started " + asctime())
 app._hash_cache = {}
@@ -52,6 +55,12 @@ def setup_app():
     cache.background_data = {}
     cache.last_background_updates = {}
 
+    oauth_config = cache.bot.config.wiki.get('copyvios', {}).get('oauth', {})
+    if oauth_config.get('consumer_token') is None:
+        raise ValueError("No OAuth consumer token is configured (config.wiki.copyvios.oauth.consumer_token).")
+    if oauth_config.get('consumer_secret') is None:
+        raise ValueError("No OAuth consumer secret is configured (config.wiki.copyvios.oauth.consumer_secret).")
+
     globalize(num_workers=8)
 
 @app.before_request
@@ -101,10 +110,50 @@ def index():
     notice = get_notice()
     update_sites()
     query = do_check()
+    if query.submitted and query.error == "not logged in":
+        return redirect("/login?next=" + quote("/?" + request.query_string), 302)
+
     return render_template(
         "index.mako", notice=notice, query=query, result=query.result,
         turnitin_result=query.turnitin_result)
 
+@app.route("/login", methods=["GET", "POST"])
+@catch_errors
+def login():
+    try:
+        redirect_url = oauth_login_start() if request.method == "POST" else None
+        if redirect_url:
+            return redirect(redirect_url, 302)
+    except Exception as e:
+        app.log_exception(e)
+        print e.message
+        kwargs = {"error": e.message}
+    else:
+        if session.get("username") is not None:
+            return redirect("/", 302)
+        kwargs = {"error": request.args.get("error")}
+    return render_template("login.mako", **kwargs)
+
+@app.route("/logout", methods=["GET", "POST"])
+@catch_errors
+def logout():
+    if request.method == "POST":
+        clear_login_session()
+        return redirect("/", 302)
+    else:
+        return render_template("logout.mako")
+
+@app.route("/oauth-callback")
+@catch_errors
+def oauth_callback():
+    try:
+        next_url = oauth_login_end()
+    except Exception as e:
+        app.log_exception(e)
+        return redirect("/login?error=" + quote_plus(e.message), 302)
+    else:
+        return redirect(next_url, 302)
+
 @app.route("/settings", methods=["GET", "POST"])
 @catch_errors
 def settings():

copyvios/api.py

@@ -10,6 +10,8 @@ from .sites import update_sites
 __all__ = ["format_api_error", "handle_api_request"]
 
 _CHECK_ERRORS = {
+    "not logged in": "You are required to log in with your Wikipedia account "
+                     "to perform checks with the search engine",
     "no search method": "Either 'use_engine' or 'use_links' must be true",
     "bad oldid": "The revision ID is invalid",
     "no URL": "The parameter 'url' is required for URL comparisons",
@@ -116,6 +118,7 @@ _HOOKS = {
 
 def handle_api_request():
     query = Query()
+    query.api = True
     if query.version:
         try:
             query.version = int(query.version)

copyvios/auth.py

@@ -0,0 +1,48 @@
+import mwoauth
+from flask import session, request
+from .misc import cache
+
+__all__ = ["oauth_login_start", "oauth_login_end", "clear_login_session"]
+
+def oauth_login_start():
+    consumer_token = mwoauth.ConsumerToken(
+        cache.bot.config.wiki["copyvios"]["oauth"]["consumer_token"],
+        cache.bot.config.wiki["copyvios"]["oauth"]["consumer_secret"])
+
+    redirect, request_token = mwoauth.initiate(
+        "https://meta.wikimedia.org/w/index.php", consumer_token)
+    session["request_token"] = dict(zip(request_token._fields, request_token))
+
+    # Take note of where to send the user after logging in
+    next_url = (request.form if request.method == "POST" else request.args).get("next", "/")
+    if next_url[0] == "/":
+        # Only allow internal redirects
+        session["next"] = next_url
+
+    return redirect
+
+def oauth_login_end():
+    if "request_token" not in session:
+        raise ValueError("OAuth request token not found in session.")
+
+    consumer_token = mwoauth.ConsumerToken(
+        cache.bot.config.wiki["copyvios"]["oauth"]["consumer_token"],
+        cache.bot.config.wiki["copyvios"]["oauth"]["consumer_secret"])
+
+    access_token = mwoauth.complete(
+        "https://meta.wikimedia.org/w/index.php",
+        consumer_token,
+        mwoauth.RequestToken(**session["request_token"]),
+        request.query_string)
+    identity = mwoauth.identify(
+        "https://meta.wikimedia.org/w/index.php",
+        consumer_token,
+        access_token)
+
+    session["access_token"] = dict(zip(access_token._fields, access_token))
+    session["username"] = identity["username"]
+
+    return session.get("next", "/")
+
+def clear_login_session():
+    session.clear()

copyvios/checker.py

@@ -144,6 +144,10 @@ def _perform_check(query, page, use_engine, use_links):
             _LOGGER.exception("Failed to retrieve cached results")
 
     if not query.result:
+        if use_engine and not query.requester_username and not query.api:
+            query.error = "not logged in"
+            return
+
         try:
             query.result = page.copyvio_check(
                 min_confidence=T_SUSPECT, max_queries=8, max_time=30,

copyvios/misc.py

@@ -5,7 +5,7 @@ import datetime
 from os.path import expanduser, join
 
 import apsw
-from flask import g, request
+from flask import g, request, session
 import oursql
 from sqlalchemy.pool import manage
 
@@ -19,6 +19,7 @@ class Query(object):
         data = request.form if method == "POST" else request.args
         for key in data:
             self.query[key] = data.getlist(key)[-1]
+        self.query["requester_username"] = session.get("username")
 
     def __getattr__(self, key):
         return self.query.get(key)

static/script.js

@@ -148,5 +148,44 @@ $(document).ready(function() {
         });
     }
 
+    if ($(".login-link").length >= 0) {
+        $(".login-link").click(function(e) {
+            e.preventDefault();
+            var $loginForm = $("<form>")
+                .attr("action", "/login")
+                .attr("method", "POST");
+
+            // Tell `/login` where to go after logging in
+            $loginForm.append(
+                $("<input>")
+                    .attr("type", "hidden")
+                    .attr("name", "next")
+                    .attr("value", window.location.pathname + window.location.search)
+            )
+
+            $("body").after($loginForm);
+            $loginForm.trigger("submit");
+            $loginForm.remove();
+            return false;
+        });
+    }
+
+    if ($(".logout-link").length >= 0) {
+        $(".logout-link").click(function(e) {
+            e.preventDefault();
+            if (!confirm("Are you sure you want to log out?")) {
+                return;
+            }
+
+            var $logoutForm = $("<form>")
+                .attr("action", "/logout")
+                .attr("method", "POST");
+            $("body").after($logoutForm);
+            $logoutForm.trigger("submit");
+            $logoutForm.remove();
+            return false;
+        });
+    }
+
     install_notice();
 });

static/script.min.js

@@ -1 +1 @@
-function update_screen_size(){var cache=cache_cookie(),data={width:window.screen.availWidth,height:window.screen.availHeight};cache&&cache.width==data.width&&cache.height==data.height||set_cookie("CopyviosScreenCache",JSON.stringify(data),1095)}function cache_cookie(){var cookie=get_cookie("CopyviosScreenCache");if(cookie)try{var width=(data=JSON.parse(cookie)).width,height=data.height;if(width&&height)return{width:width,height:height}}catch(SyntaxError){}return!1}function get_cookie(name){for(var nameEQ=name+"=",ca=document.cookie.split(";"),i=0;i<ca.length;i++){for(var c=ca[i];" "==c.charAt(0);)c=c.substring(1,c.length);if(0==c.indexOf(nameEQ)){var value=window.atob(c.substring(nameEQ.length,c.length));if(0==value.indexOf("--cpv2"))return value.substring("--cpv2".length,value.length)}}return null}function set_cookie_with_date(name,value,date){value=window.btoa("--cpv2"+value);var path=window.location.pathname.split("/",2)[1];date=date?"; expires="+date.toUTCString():"",document.cookie=name+"="+value+date+"; path=/"+path}function set_cookie(name,value,days){var date;days?((date=new Date).setTime(date.getTime()+24*days*60*60*1e3),set_cookie_with_date(name,value,date)):set_cookie_with_date(name,value)}function delete_cookie(name){set_cookie(name,"",-1)}function toggle_notice(){var details=$("#notice-collapse-box"),trigger=$("#notice-collapse-trigger");details.is(":hidden")?(details.show(),trigger.text("[hide]")):(details.hide(),trigger.text("[show]"))}function install_notice(){var details=$("#notice-collapse-box"),trigger=$("#notice-collapse-trigger");0<=details.length&&0<=trigger.length&&(trigger.replaceWith($("<a/>",{id:"notice-collapse-trigger",href:"#",text:"[show]",click:function(){return toggle_notice(),!1}})),details.hide())}$(document).ready(function(){$("#action-search").change(function(){$(".cv-search").prop("disabled",!1),$(".cv-compare").prop("disabled",!0),$(".cv-search-oo-ui").addClass("oo-ui-widget-enabled").removeClass("oo-ui-widget-disabled"),$(".cv-compare-oo-ui").addClass("oo-ui-widget-disabled").removeClass("oo-ui-widget-enabled")}),$("#action-compare").change(function(){$(".cv-search").prop("disabled",!0),$(".cv-compare").prop("disabled",!1),$(".cv-search-oo-ui").addClass("oo-ui-widget-disabled").removeClass("oo-ui-widget-enabled"),$(".cv-compare-oo-ui").addClass("oo-ui-widget-enabled").removeClass("oo-ui-widget-disabled")}),$("#action-search").is(":checked")&&$("#action-search").change(),$("#action-compare").is(":checked")&&$("#action-compare").change(),$("#cv-form").submit(function(){$("#action-search").is(":checked")&&$.each([["engine","use_engine"],["links","use_links"],["turnitin","turnitin"]],function(i,val){$("#cv-cb-"+val[0]).is(":checked")&&$("#cv-form input[type='hidden'][name='"+val[1]+"']").prop("disabled",!0)}),$("#cv-form button[type='submit']").prop("disabled",!0).css("cursor","progress").parent().addClass("oo-ui-widget-disabled").removeClass("oo-ui-widget-enabled")}),0<=$("#cv-additional").length&&($("#cv-additional").css("display","block"),$(".source-default-hidden").css("display","none"),$("#show-additional-sources").click(function(){return $(".source-default-hidden").css("display",""),$("#cv-additional").css("display","none"),!1})),install_notice()});
+function update_screen_size(){var cache=cache_cookie(),data={width:window.screen.availWidth,height:window.screen.availHeight};cache&&cache.width==data.width&&cache.height==data.height||set_cookie("CopyviosScreenCache",JSON.stringify(data),1095)}function cache_cookie(){var cookie=get_cookie("CopyviosScreenCache");if(cookie)try{var width=(data=JSON.parse(cookie)).width,height=data.height;if(width&&height)return{width:width,height:height}}catch(SyntaxError){}return!1}function get_cookie(name){for(var nameEQ=name+"=",ca=document.cookie.split(";"),i=0;i<ca.length;i++){for(var c=ca[i];" "==c.charAt(0);)c=c.substring(1,c.length);if(0==c.indexOf(nameEQ)){var value=window.atob(c.substring(nameEQ.length,c.length));if(0==value.indexOf("--cpv2"))return value.substring("--cpv2".length,value.length)}}return null}function set_cookie_with_date(name,value,date){value=window.btoa("--cpv2"+value);var path=window.location.pathname.split("/",2)[1];date=date?"; expires="+date.toUTCString():"",document.cookie=name+"="+value+date+"; path=/"+path}function set_cookie(name,value,days){var date;days?((date=new Date).setTime(date.getTime()+24*days*60*60*1e3),set_cookie_with_date(name,value,date)):set_cookie_with_date(name,value)}function delete_cookie(name){set_cookie(name,"",-1)}function toggle_notice(){var details=$("#notice-collapse-box"),trigger=$("#notice-collapse-trigger");details.is(":hidden")?(details.show(),trigger.text("[hide]")):(details.hide(),trigger.text("[show]"))}function install_notice(){var details=$("#notice-collapse-box"),trigger=$("#notice-collapse-trigger");0<=details.length&&0<=trigger.length&&(trigger.replaceWith($("<a/>",{id:"notice-collapse-trigger",href:"#",text:"[show]",click:function(){return toggle_notice(),!1}})),details.hide())}$(document).ready(function(){$("#action-search").change(function(){$(".cv-search").prop("disabled",!1),$(".cv-compare").prop("disabled",!0),$(".cv-search-oo-ui").addClass("oo-ui-widget-enabled").removeClass("oo-ui-widget-disabled"),$(".cv-compare-oo-ui").addClass("oo-ui-widget-disabled").removeClass("oo-ui-widget-enabled")}),$("#action-compare").change(function(){$(".cv-search").prop("disabled",!0),$(".cv-compare").prop("disabled",!1),$(".cv-search-oo-ui").addClass("oo-ui-widget-disabled").removeClass("oo-ui-widget-enabled"),$(".cv-compare-oo-ui").addClass("oo-ui-widget-enabled").removeClass("oo-ui-widget-disabled")}),$("#action-search").is(":checked")&&$("#action-search").change(),$("#action-compare").is(":checked")&&$("#action-compare").change(),$("#cv-form").submit(function(){$("#action-search").is(":checked")&&$.each([["engine","use_engine"],["links","use_links"],["turnitin","turnitin"]],function(i,val){$("#cv-cb-"+val[0]).is(":checked")&&$("#cv-form input[type='hidden'][name='"+val[1]+"']").prop("disabled",!0)}),$("#cv-form button[type='submit']").prop("disabled",!0).css("cursor","progress").parent().addClass("oo-ui-widget-disabled").removeClass("oo-ui-widget-enabled")}),0<=$("#cv-additional").length&&($("#cv-additional").css("display","block"),$(".source-default-hidden").css("display","none"),$("#show-additional-sources").click(function(){return $(".source-default-hidden").css("display",""),$("#cv-additional").css("display","none"),!1})),0<=$(".login-link").length&&$(".login-link").click(function(e){e.preventDefault();e=$("<form>").attr("action","/login").attr("method","POST");return e.append($("<input>").attr("type","hidden").attr("name","next").attr("value",window.location.pathname+window.location.search)),$("body").after(e),e.trigger("submit"),e.remove(),!1}),0<=$(".logout-link").length&&$(".logout-link").click(function(e){if(e.preventDefault(),confirm("Are you sure you want to log out?"))return e=$("<form>").attr("action","/logout").attr("method","POST"),$("body").after(e),e.trigger("submit"),e.remove(),!1}),install_notice()});

static/style.css

@@ -86,6 +86,21 @@ header h1 {
     }
 }
 
+header .login-link, header .logout-link {
+    margin-right: 1em;
+}
+
+header .login-link::before, header .logout-link::before {
+    content: ' ';
+    font-size: 0.85em;
+    color: black;
+    opacity: 0.6;
+    padding-left: 1.67em;
+    background-image: linear-gradient(transparent,transparent), url("data:image/svg+xml,%3Csvg%20xmlns%3D%22http%3A%2F%2Fwww.w3.org%2F2000%2Fsvg%22%20width%3D%2220%22%20height%3D%2220%22%20viewBox%3D%220%200%2020%2020%22%3E%3Ctitle%3Euser%20avatar%3C%2Ftitle%3E%3Cpath%20d%3D%22M10%2011c-5.92%200-8%203-8%205v3h16v-3c0-2-2.08-5-8-5%22%2F%3E%3Ccircle%20cx%3D%2210%22%20cy%3D%225.5%22%20r%3D%224.5%22%2F%3E%3C%2Fsvg%3E");
+    background-repeat: no-repeat;
+    background-size: contain;
+}
+
 #settings-link::before {
     content: ' ';
     font-size: 0.85em;

templates/login.mako

@@ -0,0 +1,39 @@
+<%!
+    from json import dumps, loads
+    from flask import g, request
+    from copyvios.misc import cache
+    from urlparse import parse_qsl
+%>\
+<%include file="/support/header.mako" args="title='Login | Earwig\'s Copyvio Detector', splash=True"/>
+% if error:
+    <div id="info-box" class="red-box">
+        <p>Error trying to log in: ${error | h}</p>
+    </div>
+% endif
+<h2>Login</h2>
+<p>You are required to log in with your Wikimedia account to perform checks with the search engine.</p>
+%if request.args.get('next'):
+    <p>
+        After logging in,
+        % if request.args["next"][0:2] == "/?" and dict(parse_qsl(request.args["next"])).get("action") == "search":
+            your check will be run.
+        % else:
+            you will be redirected to: ${request.args.get('next') | h}
+        % endif
+    </p>
+%endif
+<form action="${request.script_root}/login" method="post">
+    <input type="hidden" name="next" value="${request.args.get('next', '/') | h}" />
+    <div class="oo-ui-layout oo-ui-fieldLayout oo-ui-fieldLayout-align-left">
+        <div class="oo-ui-fieldLayout-body">
+            <span class="oo-ui-fieldLayout-field">
+                <span class="oo-ui-widget oo-ui-widget-enabled oo-ui-inputWidget oo-ui-buttonElement oo-ui-buttonElement-framed oo-ui-labelElement oo-ui-flaggedElement-primary oo-ui-flaggedElement-progressive oo-ui-labelElement oo-ui-buttonInputWidget">
+                    <button type="submit" class="oo-ui-inputWidget-input oo-ui-buttonElement-button">
+                        <span class="oo-ui-labelElement-label">Login</span>
+                    </button>
+                </span>
+            </span>
+        </div>
+    </div>
+</form>
+<%include file="/support/footer.mako"/>

templates/logout.mako

@@ -0,0 +1,22 @@
+<%!
+    from json import dumps, loads
+    from flask import g, request
+    from copyvios.misc import cache
+%>\
+<%include file="/support/header.mako" args="title='Logout | Earwig\'s Copyvio Detector', splash=True"/>
+<h2>Logout</h2>
+<p>Logging out will prevent you from making search engine checks.</p>
+<form action="${request.script_root}/logout" method="post">
+    <div class="oo-ui-layout oo-ui-fieldLayout oo-ui-fieldLayout-align-left">
+        <div class="oo-ui-fieldLayout-body">
+            <span class="oo-ui-fieldLayout-field">
+                <span class="oo-ui-widget oo-ui-widget-enabled oo-ui-inputWidget oo-ui-buttonElement oo-ui-buttonElement-framed oo-ui-labelElement oo-ui-flaggedElement-primary oo-ui-flaggedElement-progressive oo-ui-labelElement oo-ui-buttonInputWidget">
+                    <button type="submit" class="oo-ui-inputWidget-input oo-ui-buttonElement-button">
+                        <span class="oo-ui-labelElement-label">Logout</span>
+                    </button>
+                </span>
+            </span>
+        </div>
+    </div>
+</form>
+<%include file="/support/footer.mako"/>

templates/support/header.mako

@@ -1,6 +1,6 @@
 <%page args="title, splash=False"/>\
 <%!
-    from flask import g, request, url_for
+    from flask import g, request, session, url_for
     from copyvios.background import set_background
 %>\
 <!DOCTYPE html>
@@ -25,6 +25,12 @@
         <div id="content">
             <header>
                 <h1><a href="/">Earwig&apos;s <strong>Copyvio Detector</strong></a></h1>
+                <% logged_in_user = session.get("username") %>\
+                % if logged_in_user:
+                    <a class="logout-link" href="/logout" title="Log out">${logged_in_user | h}</a>
+                % else:
+                    <a class="login-link" href="/login" title="Log in">Login</a>
+                % endif
                 <a id="settings-link" href="/settings">Settings</a>
             </header>
             <main>