Initial commit

.gitignore

@@ -0,0 +1,8 @@
+.*.swp
+*~
+*.py[co]
+.coverage
+*.egg-info
+dist
+*.egg
+db.sqlite3

manage.py

@@ -0,0 +1,10 @@
+#!/usr/bin/env python
+import os
+import sys
+
+if __name__ == "__main__":
+    os.environ.setdefault("DJANGO_SETTINGS_MODULE", "stack.settings")
+
+    from django.core.management import execute_from_command_line
+
+    execute_from_command_line(sys.argv)

qa/__init__.py

@@ -0,0 +1 @@
+default_app_config = 'qa.apps.QAConfig'

qa/admin.py

@@ -0,0 +1,3 @@
+from django.contrib import admin
+
+# Register your models here.

qa/apps.py

@@ -0,0 +1,5 @@
+from django.apps import AppConfig
+
+class QAConfig(AppConfig):
+    name = 'qa'
+    verbose_name = "Q & A"

qa/management/commands/load_dump.py

@@ -0,0 +1,199 @@
+from __future__ import print_function
+
+import time
+from os.path import join
+from xml.etree import cElementTree
+from dateutil import parser as date_parser
+
+from django.core.management.base import BaseCommand
+
+from qa.models import User, Question, Answer, QuestionComment, AnswerComment
+
+POST_TYPES = {
+    1: Question,
+    2: Answer,
+}
+
+class Command(BaseCommand):
+    def add_arguments(self, parser):
+        parser.add_argument('dump_directory',
+            help='Directory containing the XML files.')
+
+    def handle(self, **options):
+        self.dir = options['dump_directory']
+        self._answers = set()
+        self._questions = set()
+
+        for m, f in (
+                ('Users', self.parse_users),
+                ('Posts', self.parse_posts),
+                ('Comments', self.parse_comments)
+            ):
+            self.verbose_run(f, m)
+
+    def verbose_run(self, cmd_func, name, report_every=100):
+        print('Loading %s: ' % name, end='')
+        start = time.time()
+        cnt = 0
+        for u in cmd_func():
+            cnt += 1
+            if cnt % report_every:
+                print('.', end='', flush=True)
+        print('DONE\nLoaded %d %s in %d seconds'% (
+            cnt, name, time.time() - start
+        ))
+
+    def _parse_file(self, xml_file):
+        with open(join(self.dir, xml_file)) as input:
+            root = cElementTree.iterparse(input)
+
+            for event, e in root:
+                if event != 'end' or e.tag != 'row':
+                    continue
+                yield dict(
+                    (k, int(v) if v.isdigit() else v)
+                        for (k, v) in e.items()
+                )
+
+    def parse_users(self, users_file='Users.xml'):
+        """
+        Parse data into User objects
+
+            <row
+                Id="2"
+                Reputation="101"
+                CreationDate="2011-01-03T20:14:55.783"
+                DisplayName="Geoff Dalgas"
+                LastAccessDate="2012-12-19T00:28:45.110"
+                WebsiteUrl="http://stackoverflow.com"
+                Location="Corvallis, OR"
+                AboutMe="&lt;p&gt;Developer on ...."
+                Views="6"
+                UpVotes="6"
+                DownVotes="0"
+                EmailHash="b437f461b3fd27387c5d8ab47a293d35"
+                Age="36"
+            />
+        """
+        for user in self._parse_file(users_file):
+            if user['Id'] == '-1':
+                continue
+            yield User.objects.create(
+                id=user['Id'],
+                email=user.get('EmailHash', ''),
+                date_joined=date_parser.parse(user['CreationDate']),
+
+                display_name=user['DisplayName'],
+                url=user.get('WebsiteUrl', ''),
+                location=user.get('Location', ''),
+                description=user.get('AboutMe', ''),
+
+                views=user['Views'],
+                votes_up=user['UpVotes'],
+                votes_down=user['DownVotes'],
+                age=user.get('Age', 0)
+            )
+
+
+    def parse_comments(self, comments_file='Comments.xml'):
+        """
+        Comments.xml:
+
+            <row
+                Id="9"
+                PostId="9"
+                Score="3"
+                Text="Point.... "
+                CreationDate="2011-01-03T21:16:09.603"
+                UserId="60"
+            />
+        """
+        for comment in self._parse_file(comments_file):
+            if comment['PostId'] in self._answers:
+                cls = AnswerComment
+            elif comment['PostId'] in self._questions:
+                cls = QuestionComment
+            else:
+                continue
+
+            yield cls.objects.create(
+                post_id=comment['PostId'],
+                owner_id=comment['UserId'],
+                creation_date=date_parser.parse(comment['CreationDate']),
+                score=comment['Score'],
+                text=comment['Text']
+            )
+
+
+    def parse_posts(self, posts_file='Posts.xml'):
+        """
+            Posts.xml:
+            Q:  <row
+                Id="5"
+                PostTypeId="1"
+                AcceptedAnswerId="73"
+                CreationDate="2011-01-03T20:52:52.880"
+                Score="39"
+                ViewCount="5638"
+                Body="&lt;p&gt;....."
+                OwnerUserId="24"
+                LastEditorUserId="97"
+                LastEditDate="2011-01-06T11:34:27.610"
+                LastActivityDate="2012-01-27T19:12:50.900"
+                Title="What are the differences between NoSQL and a traditional RDBMS?"
+                Tags="&lt;nosql&gt;&lt;rdbms&gt;&lt;database-recommendation&gt;"
+                AnswerCount="5"
+                CommentCount="0"
+                FavoriteCount="22"
+            />
+
+            A: <row
+                Id="12"
+                PostTypeId="2"
+                ParentId="3"
+                CreationDate="2011-01-03T21:01:19.160"
+                Score="15"
+                Body="&lt;p&gt;In ..."
+                OwnerUserId="14"
+                LastActivityDate="2011-01-03T21:01:19.160"
+                CommentCount="3"
+            />
+        """
+
+        for data in self._parse_file(posts_file):
+            try:
+                cls = POST_TYPES[data['PostTypeId']]
+            except KeyError:
+                # unknown post type, ignore
+                continue
+
+            post = cls(
+                id=data['Id'],
+                owner_id=data['OwnerUserId'],
+                creation_date=date_parser.parse(data['CreationDate']),
+                last_activity_date=date_parser.parse(data['LastActivityDate']),
+                score=data['Score'],
+                body=data['Body'],
+                comment_count=data['CommentCount']
+            )
+
+            if isinstance(post, Question):
+                post.answer_count = data['AnswerCount']
+                post.tags = data['Tags']
+                post.title = data['Title']
+                post.favorite_count = data.get('FavoriteCount', 0)
+                post.view_count = data['ViewCount']
+                if 'AcceptedAnswerId' in data:
+                    post.accepted_answer_id = data['AcceptedAnswerId']
+                if 'LastEditorUserId' in data:
+                    post.last_editor_id = data['LastEditorUserId']
+                    post.last_edit_date = date_parser.parse(data['LastEditDate'])
+
+                self._questions.add(post.pk)
+            else:
+                self._answers.add(post.pk)
+                post.question_id = data['ParentId']
+
+            post.save(force_insert=True)
+
+            yield post

qa/models.py

@@ -0,0 +1,67 @@
+from django.db import models
+
+class User(models.Model):
+    email = models.CharField(max_length=200)
+    date_joined = models.DateTimeField()
+    display_name = models.CharField(max_length=200)
+    url = models.CharField(max_length=400)
+    location = models.CharField(max_length=400)
+    description = models.TextField()
+    views = models.PositiveIntegerField()
+    votes_up = models.PositiveIntegerField()
+    votes_down = models.PositiveIntegerField()
+    age = models.PositiveIntegerField()
+
+class Post(models.Model):
+    owner = models.ForeignKey(User)
+    creation_date = models.DateTimeField()
+    last_activity_date = models.DateTimeField()
+    score = models.IntegerField()
+    body = models.TextField()
+    comment_count = models.PositiveIntegerField()
+
+    def get_comments(self):
+        return self.comments.order_by('-creation_date')
+
+    class Meta:
+        abstract = True
+
+
+class Question(Post):
+    answer_count = models.PositiveIntegerField()
+    tags = models.TextField()
+    title = models.CharField(max_length=1024)
+    favorite_count = models.PositiveIntegerField()
+    view_count = models.PositiveIntegerField()
+    accepted_answer = models.ForeignKey('Answer', related_name='accepted_for',
+            null=True, blank=True)
+
+    last_editor = models.ForeignKey(User, null=True, blank=True,
+            related_name='last_edited_questions')
+    last_edit_date  = models.DateTimeField(null=True, blank=True)
+
+    def get_answers(self):
+        return self.answer_set.order_by('-creation_date')
+
+
+class Answer(Post):
+    question = models.ForeignKey(Question)
+
+
+class Comment(models.Model):
+    owner = models.ForeignKey(User)
+    creation_date = models.DateTimeField()
+    score = models.IntegerField()
+    text = models.TextField()
+
+    class Meta:
+        abstract = True
+
+
+class QuestionComment(Comment):
+    post = models.ForeignKey(Question, related_name='comments')
+
+
+class AnswerComment(Comment):
+    post = models.ForeignKey(Answer, related_name='comments')
+

qa/templates/qa/question_detail.html

@@ -0,0 +1,35 @@
+<h1>{{ object.title }}</h1>
+question by {{ object.owner.display_name }} on {{ object.creation_date|date }}, {{ object.creation_date|time }}
+{% if object.score %}
+    with score {{ object.score }}
+{% endif %}
+
+{{ object.body|safe }}
+
+{% for comment in object.get_comments %}
+    comment by {{ comment.owner.display_name }} on {{ comment.creation_date|date }}, {{ comment.creation_date|time }}
+    {% if comment.score %}
+        with score {{ comment.score }}
+    {% endif %}
+
+    <p>{{ comment.text|safe }}</p>
+{% endfor %}
+
+{% for answer in object.get_answers %}
+    answer by {{ answer.owner.display_name }} on {{ answer.creation_date|date }}, {{ answer.creation_date|time }}
+    {% if answer.score %}
+        with score {{ answer.score }}
+    {% endif %}
+
+    {{answer.body|safe}}
+
+    {% for comment in answer.get_comments %}
+        comment by {{ comment.owner.display_name }} on {{ comment.creation_date|date }}, {{ comment.creation_date|time }}
+        {% if comment.score %}
+            with score {{ comment.score }}
+        {% endif %}
+
+        <p>{{ comment.text|safe }}</p>
+    {% endfor %}
+
+{% endfor %}

qa/templates/qa/question_list.html

@@ -0,0 +1,5 @@
+{% for question in object_list %}
+  <h2><a href="{% url "qa-question" pk=question.pk %}">{{ question.title }}</a></h2>
+  {{ question.creation_date|date }}
+  {{ question.creation_date|time }}
+{% endfor %}

qa/urls.py

@@ -0,0 +1,8 @@
+from django.conf.urls import patterns, url
+
+from .views import QuestionDetail, QuestionList
+
+urlpatterns = patterns('',
+    url(r'^$', QuestionList.as_view(), name='qa-list'),
+    url(r'^(?P<pk>\d+)/$', QuestionDetail.as_view(), name='qa-question'),
+)

qa/views.py

@@ -0,0 +1,14 @@
+from django.views.generic.list import ListView
+from django.views.generic.detail import DetailView
+
+from .models import Question
+
+class QuestionList(ListView):
+    model = Question
+    ordering = '-creation_date'
+    paginate_by = 10
+
+
+class QuestionDetail(DetailView):
+    model = Question
+