add markov module and irc plugin

just listens to servers right now, doesn't speak up (yet)

hitomi/settings.py

@@ -40,6 +40,7 @@ INSTALLED_APPS = [
     'bot',
     'dice',
     'logger',
+    'markov',
     'weather',
 ]
 

markov/admin.py

@@ -0,0 +1,7 @@
+"""Display Markov data elements in the Django admin."""
+from django.contrib import admin
+
+from markov.models import MarkovContext, MarkovState
+
+admin.site.register(MarkovContext)
+admin.site.register(MarkovState)

markov/apps.py

@@ -0,0 +1,6 @@
+"""App config for the markov module."""
+from django.apps import AppConfig
+
+
+class MarkovConfig(AppConfig):
+    name = 'markov'

markov/bot.py

@@ -0,0 +1,33 @@
+"""Observe Markov chains from Discord."""
+import logging
+
+from bot import hitomi
+from markov import lib
+
+logger = logging.getLogger(__name__)
+logger.info("loading markov plugin")
+
+
+def on_message(message):
+    """Keep the observed Markov chains."""
+    # ignore self
+    if message.author == hitomi.user:
+        logger.debug("markov ignoring message authored by self")
+        return
+
+    if message.channel.is_private:
+        # DMs have a context of the author
+        context_name = message.author
+    else:
+        # channels have a context of the server, but we should ignore channels that override @everyone's read ability
+        for changed_role in message.channel.changed_roles:
+            if changed_role.is_everyone:
+                if not changed_role.permissions.read_messages:
+                    logger.debug("markov ignoring channel that @everyone can't read")
+                    return
+        context_name = message.server.id
+
+    lib.learn_line(message.content, context_name)
+
+
+hitomi.on_message_handlers.append(on_message)

markov/lib.py

@@ -0,0 +1,30 @@
+"""Shared methods for the Markov module."""
+import logging
+
+from markov.models import MarkovContext, MarkovState
+
+logger = logging.getLogger(__name__)
+
+
+def learn_line(line, context_name):
+    """Create a bunch of MarkovStates for a given line of text."""
+    logger.debug("learning %s...", line[:40])
+
+    context, created = MarkovContext.objects.get_or_create(name=context_name)
+
+    words = line.split()
+    words = [MarkovState.start1, MarkovState.start2] + words + [MarkovState.stop]
+
+    for word in words:
+        if len(word) > MarkovState._meta.get_field('k1').max_length:
+            return
+
+    for i, word in enumerate(words):
+        logger.debug("'{0:s}','{1:s}' -> '{2:s}'".format(words[i], words[i + 1], words[i + 2]))
+        state, created = MarkovState.objects.get_or_create(context=context, k1=words[i], k2=words[i + 1],
+                                                           v=words[i + 2])
+        state.count += 1
+        state.save()
+
+        if i > len(words) - 4:
+            break

markov/migrations/0001_initial.py

@@ -0,0 +1,41 @@
+# Generated by Django 2.0.1 on 2018-01-09 19:57
+
+from django.db import migrations, models
+import django.db.models.deletion
+
+
+class Migration(migrations.Migration):
+
+    initial = True
+
+    dependencies = [
+    ]
+
+    operations = [
+        migrations.CreateModel(
+            name='MarkovContext',
+            fields=[
+                ('id', models.AutoField(auto_created=True, primary_key=True, serialize=False, verbose_name='ID')),
+                ('name', models.CharField(max_length=200, unique=True)),
+            ],
+        ),
+        migrations.CreateModel(
+            name='MarkovState',
+            fields=[
+                ('id', models.AutoField(auto_created=True, primary_key=True, serialize=False, verbose_name='ID')),
+                ('k1', models.CharField(max_length=256)),
+                ('k2', models.CharField(max_length=256)),
+                ('v', models.CharField(max_length=256)),
+                ('count', models.IntegerField(default=0)),
+                ('context', models.ForeignKey(on_delete=django.db.models.deletion.CASCADE, to='markov.MarkovContext')),
+            ],
+        ),
+        migrations.AlterUniqueTogether(
+            name='markovstate',
+            unique_together={('context', 'k1', 'k2', 'v')},
+        ),
+        migrations.AlterIndexTogether(
+            name='markovstate',
+            index_together={('context', 'k1', 'k2'), ('context', 'v')},
+        ),
+    ]

markov/models.py

@@ -0,0 +1,43 @@
+"""Database objects for storing Markov chains."""
+
+import logging
+
+from django.db import models
+
+log = logging.getLogger(__name__)
+
+
+class MarkovContext(models.Model):
+    """Define contexts for Markov chains."""
+
+    name = models.CharField(max_length=200, unique=True)
+
+    def __str__(self):
+        """String representation."""
+        return "{0:s}".format(self.name)
+
+
+class MarkovState(models.Model):
+    """One element in a Markov chain, some text or something."""
+
+    start1 = '__start1'
+    start2 = '__start2'
+    stop = '__stop'
+
+    k1 = models.CharField(max_length=256)
+    k2 = models.CharField(max_length=256)
+    v = models.CharField(max_length=256)
+
+    count = models.IntegerField(default=0)
+    context = models.ForeignKey(MarkovContext, on_delete=models.CASCADE)
+
+    class Meta:
+        index_together = [
+            ['context', 'k1', 'k2'],
+            ['context', 'v'],
+        ]
+        unique_together = ('context', 'k1', 'k2', 'v')
+
+    def __str__(self):
+        """String representation."""
+        return "{0:s},{1:s} -> {2:s} (count: {3:d})".format(self.k1, self.k2, self.v, self.count)