2021-01-02 13:33:53 +00:00
|
|
|
package eu.faircode.email;
|
|
|
|
|
|
|
|
/*
|
|
|
|
This file is part of FairEmail.
|
|
|
|
|
|
|
|
FairEmail is free software: you can redistribute it and/or modify
|
|
|
|
it under the terms of the GNU General Public License as published by
|
|
|
|
the Free Software Foundation, either version 3 of the License, or
|
|
|
|
(at your option) any later version.
|
|
|
|
|
|
|
|
FairEmail is distributed in the hope that it will be useful,
|
|
|
|
but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
|
|
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
|
|
GNU General Public License for more details.
|
|
|
|
|
|
|
|
You should have received a copy of the GNU General Public License
|
|
|
|
along with FairEmail. If not, see <http://www.gnu.org/licenses/>.
|
|
|
|
|
|
|
|
Copyright 2018-2021 by Marcel Bokhorst (M66B)
|
|
|
|
*/
|
|
|
|
|
|
|
|
import android.content.Context;
|
2021-01-02 14:57:24 +00:00
|
|
|
import android.content.SharedPreferences;
|
2021-01-06 20:43:26 +00:00
|
|
|
import android.os.Build;
|
2021-01-04 14:28:45 +00:00
|
|
|
import android.text.TextUtils;
|
2021-01-02 13:33:53 +00:00
|
|
|
|
2021-01-02 14:57:24 +00:00
|
|
|
import androidx.preference.PreferenceManager;
|
|
|
|
|
2021-01-02 13:33:53 +00:00
|
|
|
import org.jetbrains.annotations.NotNull;
|
2021-01-02 19:42:35 +00:00
|
|
|
import org.json.JSONArray;
|
|
|
|
import org.json.JSONException;
|
|
|
|
import org.json.JSONObject;
|
2021-01-02 13:33:53 +00:00
|
|
|
|
|
|
|
import java.io.File;
|
|
|
|
import java.io.IOException;
|
|
|
|
import java.util.ArrayList;
|
2021-01-04 14:28:45 +00:00
|
|
|
import java.util.Arrays;
|
2021-01-02 13:33:53 +00:00
|
|
|
import java.util.Collections;
|
|
|
|
import java.util.Comparator;
|
2021-01-03 07:41:29 +00:00
|
|
|
import java.util.Date;
|
2021-01-02 13:33:53 +00:00
|
|
|
import java.util.HashMap;
|
2021-01-07 08:18:29 +00:00
|
|
|
import java.util.Iterator;
|
2021-01-02 13:33:53 +00:00
|
|
|
import java.util.List;
|
|
|
|
import java.util.Map;
|
|
|
|
|
2021-01-04 14:28:45 +00:00
|
|
|
import javax.mail.Address;
|
|
|
|
import javax.mail.internet.InternetAddress;
|
|
|
|
|
2021-01-02 13:33:53 +00:00
|
|
|
public class MessageClassifier {
|
2021-01-02 19:42:35 +00:00
|
|
|
private static boolean loaded = false;
|
2021-01-02 21:52:41 +00:00
|
|
|
private static boolean dirty = false;
|
2021-01-07 18:20:17 +00:00
|
|
|
private static final Map<Long, Map<String, Integer>> classMessages = new HashMap<>();
|
2021-01-06 20:35:11 +00:00
|
|
|
private static final Map<Long, Map<String, Map<String, Frequency>>> wordClassFrequency = new HashMap<>();
|
2021-01-02 13:33:53 +00:00
|
|
|
|
2021-01-07 12:39:42 +00:00
|
|
|
private static final double CHANCE_MINIMUM = 0.20;
|
2021-01-02 13:33:53 +00:00
|
|
|
private static final double CHANCE_THRESHOLD = 2.0;
|
|
|
|
|
2021-01-05 18:51:22 +00:00
|
|
|
static void classify(EntityMessage message, EntityFolder folder, EntityFolder target, Context context) {
|
2021-01-02 19:42:35 +00:00
|
|
|
try {
|
2021-01-03 11:23:51 +00:00
|
|
|
if (!isEnabled(context))
|
|
|
|
return;
|
2021-01-03 07:27:27 +00:00
|
|
|
|
2021-01-05 18:51:22 +00:00
|
|
|
if (!canClassify(folder.type))
|
2021-01-03 11:23:51 +00:00
|
|
|
return;
|
2021-01-02 13:33:53 +00:00
|
|
|
|
2021-01-05 18:51:22 +00:00
|
|
|
if (target != null && !canClassify(target.type))
|
2021-01-03 11:23:51 +00:00
|
|
|
return;
|
2021-01-02 19:50:43 +00:00
|
|
|
|
2021-01-03 11:23:51 +00:00
|
|
|
File file = message.getFile(context);
|
|
|
|
if (!file.exists())
|
|
|
|
return;
|
2021-01-02 13:33:53 +00:00
|
|
|
|
2021-01-06 07:31:34 +00:00
|
|
|
long start = new Date().getTime();
|
|
|
|
|
2021-01-05 18:51:22 +00:00
|
|
|
// Build text to classify
|
2021-01-04 13:37:54 +00:00
|
|
|
StringBuilder sb = new StringBuilder();
|
2021-01-04 14:28:45 +00:00
|
|
|
|
|
|
|
List<Address> addresses = new ArrayList<>();
|
|
|
|
if (message.from != null)
|
|
|
|
addresses.addAll(Arrays.asList(message.from));
|
|
|
|
if (message.to != null)
|
|
|
|
addresses.addAll(Arrays.asList(message.to));
|
|
|
|
if (message.cc != null)
|
|
|
|
addresses.addAll(Arrays.asList(message.cc));
|
|
|
|
if (message.bcc != null)
|
|
|
|
addresses.addAll(Arrays.asList(message.bcc));
|
|
|
|
if (message.reply != null)
|
|
|
|
addresses.addAll(Arrays.asList(message.reply));
|
|
|
|
|
|
|
|
for (Address address : addresses) {
|
|
|
|
String email = ((InternetAddress) address).getAddress();
|
|
|
|
String name = ((InternetAddress) address).getAddress();
|
|
|
|
if (!TextUtils.isEmpty(email)) {
|
|
|
|
sb.append(email).append('\n');
|
|
|
|
int at = email.indexOf('@');
|
|
|
|
String domain = (at < 0 ? null : email.substring(at + 1));
|
|
|
|
if (!TextUtils.isEmpty(domain))
|
|
|
|
sb.append(domain).append('\n');
|
|
|
|
}
|
|
|
|
if (!TextUtils.isEmpty(name))
|
|
|
|
sb.append(name).append('\n');
|
|
|
|
}
|
|
|
|
|
2021-01-04 13:37:54 +00:00
|
|
|
if (message.subject != null)
|
|
|
|
sb.append(message.subject).append('\n');
|
2021-01-04 14:28:45 +00:00
|
|
|
|
2021-01-07 19:43:53 +00:00
|
|
|
String text = HtmlHelper.getFullText(file);
|
|
|
|
sb.append(text);
|
2021-01-04 14:28:45 +00:00
|
|
|
|
2021-01-04 13:37:54 +00:00
|
|
|
if (sb.length() == 0)
|
2021-01-03 11:23:51 +00:00
|
|
|
return;
|
2021-01-02 13:33:53 +00:00
|
|
|
|
2021-01-05 18:51:22 +00:00
|
|
|
// Load data if needed
|
2021-01-03 11:23:51 +00:00
|
|
|
load(context);
|
2021-01-02 21:52:41 +00:00
|
|
|
|
2021-01-05 18:51:22 +00:00
|
|
|
// Initialize data if needed
|
2021-01-07 18:20:17 +00:00
|
|
|
if (!classMessages.containsKey(folder.account))
|
|
|
|
classMessages.put(folder.account, new HashMap<>());
|
2021-01-05 18:51:22 +00:00
|
|
|
if (!wordClassFrequency.containsKey(folder.account))
|
|
|
|
wordClassFrequency.put(folder.account, new HashMap<>());
|
2021-01-03 11:23:51 +00:00
|
|
|
|
2021-01-05 18:51:22 +00:00
|
|
|
// Classify text
|
|
|
|
String classified = classify(folder.account, folder.name, sb.toString(), target == null, context);
|
2021-01-03 11:23:51 +00:00
|
|
|
|
2021-01-06 07:31:34 +00:00
|
|
|
long elapsed = new Date().getTime() - start;
|
2021-01-03 11:23:51 +00:00
|
|
|
EntityLog.log(context, "Classifier" +
|
|
|
|
" folder=" + folder.name +
|
|
|
|
" message=" + message.id +
|
|
|
|
"@" + new Date(message.received) +
|
|
|
|
":" + message.subject +
|
2021-01-04 11:21:51 +00:00
|
|
|
" class=" + classified +
|
2021-01-06 07:31:34 +00:00
|
|
|
" re=" + message.auto_classified +
|
|
|
|
" elapsed=" + elapsed);
|
2021-01-03 11:23:51 +00:00
|
|
|
|
2021-01-07 18:20:17 +00:00
|
|
|
Integer m = classMessages.get(folder.account).get(folder.name);
|
|
|
|
m = (m == null ? 0 : m) + (target == null ? 1 : -1);
|
|
|
|
if (m <= 0)
|
|
|
|
classMessages.get(folder.account).remove(folder.name);
|
|
|
|
else
|
|
|
|
classMessages.get(folder.account).put(folder.name, m);
|
|
|
|
Log.i("Classifier " + folder.name + "=" + m + " msgs");
|
|
|
|
|
2021-01-03 11:23:51 +00:00
|
|
|
dirty = true;
|
|
|
|
|
2021-01-05 18:51:22 +00:00
|
|
|
// Auto classify
|
|
|
|
if (classified != null &&
|
|
|
|
!classified.equals(folder.name) &&
|
|
|
|
!message.auto_classified &&
|
|
|
|
!EntityFolder.JUNK.equals(folder.type)) {
|
|
|
|
DB db = DB.getInstance(context);
|
2021-01-04 11:21:51 +00:00
|
|
|
try {
|
|
|
|
db.beginTransaction();
|
|
|
|
|
2021-01-05 18:51:22 +00:00
|
|
|
EntityFolder dest = db.folder().getFolderByName(folder.account, classified);
|
|
|
|
if (dest != null && dest.auto_classify) {
|
|
|
|
EntityOperation.queue(context, message, EntityOperation.MOVE, dest.id, false, true);
|
2021-01-04 08:27:27 +00:00
|
|
|
message.ui_hide = true;
|
|
|
|
}
|
2021-01-04 11:21:51 +00:00
|
|
|
|
|
|
|
db.setTransactionSuccessful();
|
|
|
|
|
|
|
|
} finally {
|
|
|
|
db.endTransaction();
|
|
|
|
}
|
2021-01-05 18:51:22 +00:00
|
|
|
}
|
2021-01-03 11:23:51 +00:00
|
|
|
} catch (Throwable ex) {
|
|
|
|
Log.e(ex);
|
2021-01-02 21:38:37 +00:00
|
|
|
}
|
2021-01-02 13:33:53 +00:00
|
|
|
}
|
|
|
|
|
2021-01-06 10:39:41 +00:00
|
|
|
private static String classify(long account, String currentClass, String text, boolean added, Context context) {
|
2021-01-07 09:58:56 +00:00
|
|
|
int maxMessages = 0;
|
2021-01-07 18:20:17 +00:00
|
|
|
for (String clazz : classMessages.get(account).keySet()) {
|
|
|
|
int count = classMessages.get(account).get(clazz);
|
|
|
|
if (count > maxMessages)
|
|
|
|
maxMessages = count;
|
2021-01-07 09:58:56 +00:00
|
|
|
}
|
|
|
|
|
2021-01-06 20:35:11 +00:00
|
|
|
State state = new State();
|
2021-01-07 14:20:39 +00:00
|
|
|
|
|
|
|
// First word
|
2021-01-07 13:17:38 +00:00
|
|
|
process(account, currentClass, added, null, state);
|
2021-01-02 13:33:53 +00:00
|
|
|
|
2021-01-07 14:20:39 +00:00
|
|
|
// Process words
|
2021-01-06 20:43:26 +00:00
|
|
|
if (Build.VERSION.SDK_INT < Build.VERSION_CODES.N) {
|
|
|
|
java.text.BreakIterator boundary = java.text.BreakIterator.getWordInstance();
|
|
|
|
boundary.setText(text);
|
|
|
|
int start = boundary.first();
|
|
|
|
for (int end = boundary.next(); end != java.text.BreakIterator.DONE; end = boundary.next()) {
|
2021-01-07 13:17:38 +00:00
|
|
|
String word = text.substring(start, end);
|
|
|
|
process(account, currentClass, added, word, state);
|
2021-01-06 20:43:26 +00:00
|
|
|
start = end;
|
|
|
|
}
|
|
|
|
} else {
|
2021-01-07 13:17:38 +00:00
|
|
|
// The ICU break iterator works better for Chinese texts
|
2021-01-06 20:43:26 +00:00
|
|
|
android.icu.text.BreakIterator boundary = android.icu.text.BreakIterator.getWordInstance();
|
|
|
|
boundary.setText(text);
|
|
|
|
int start = boundary.first();
|
|
|
|
for (int end = boundary.next(); end != android.icu.text.BreakIterator.DONE; end = boundary.next()) {
|
2021-01-07 13:17:38 +00:00
|
|
|
String word = text.substring(start, end);
|
|
|
|
process(account, currentClass, added, word, state);
|
2021-01-06 20:43:26 +00:00
|
|
|
start = end;
|
2021-01-02 13:33:53 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2021-01-07 14:20:39 +00:00
|
|
|
// Last word
|
2021-01-07 13:17:38 +00:00
|
|
|
process(account, currentClass, added, null, state);
|
2021-01-06 20:35:11 +00:00
|
|
|
|
2021-01-02 13:33:53 +00:00
|
|
|
if (!added)
|
|
|
|
return null;
|
|
|
|
|
2021-01-07 09:06:33 +00:00
|
|
|
if (maxMessages == 0) {
|
|
|
|
Log.e("Classifier no messages account=" + account);
|
2021-01-07 14:20:39 +00:00
|
|
|
return null;
|
2021-01-07 09:06:33 +00:00
|
|
|
}
|
|
|
|
|
2021-01-07 14:20:39 +00:00
|
|
|
// Calculate chance per class
|
2021-01-07 09:06:33 +00:00
|
|
|
DB db = DB.getInstance(context);
|
|
|
|
List<Chance> chances = new ArrayList<>();
|
|
|
|
for (String clazz : state.classStats.keySet()) {
|
2021-01-05 12:18:17 +00:00
|
|
|
EntityFolder folder = db.folder().getFolderByName(account, clazz);
|
2021-01-05 18:31:35 +00:00
|
|
|
if (folder == null) {
|
|
|
|
Log.w("Classifier no folder class=" + account + ":" + clazz);
|
2021-01-05 12:18:17 +00:00
|
|
|
continue;
|
2021-01-05 18:31:35 +00:00
|
|
|
}
|
2021-01-05 12:18:17 +00:00
|
|
|
|
2021-01-06 20:35:11 +00:00
|
|
|
Stat stat = state.classStats.get(clazz);
|
2021-01-07 09:58:56 +00:00
|
|
|
|
|
|
|
double chance = stat.totalFrequency / maxMessages / state.words.size();
|
2021-01-02 13:33:53 +00:00
|
|
|
Chance c = new Chance(clazz, chance);
|
2021-01-07 08:35:58 +00:00
|
|
|
chances.add(c);
|
2021-01-07 09:58:56 +00:00
|
|
|
EntityLog.log(context, "Classifier " + c +
|
2021-01-07 13:55:14 +00:00
|
|
|
" frequency=" + (Math.round(stat.totalFrequency * 100.0) / 100.0) + "/" + maxMessages + " msgs" +
|
2021-01-07 09:58:56 +00:00
|
|
|
" matched=" + stat.matchedWords + "/" + state.words.size() + " words" +
|
|
|
|
" text=" + TextUtils.join(", ", stat.words));
|
2021-01-02 13:33:53 +00:00
|
|
|
}
|
|
|
|
|
2021-01-04 14:28:45 +00:00
|
|
|
if (BuildConfig.DEBUG)
|
2021-01-06 20:35:11 +00:00
|
|
|
Log.i("Classifier words=" + TextUtils.join(", ", state.words));
|
2021-01-04 14:28:45 +00:00
|
|
|
|
2021-01-05 11:55:12 +00:00
|
|
|
if (chances.size() <= 1)
|
2021-01-02 13:33:53 +00:00
|
|
|
return null;
|
|
|
|
|
2021-01-07 14:20:39 +00:00
|
|
|
// Sort classes by chance
|
2021-01-02 13:33:53 +00:00
|
|
|
Collections.sort(chances, new Comparator<Chance>() {
|
|
|
|
@Override
|
|
|
|
public int compare(Chance c1, Chance c2) {
|
|
|
|
return -c1.chance.compareTo(c2.chance);
|
|
|
|
}
|
|
|
|
});
|
|
|
|
|
2021-01-07 14:20:39 +00:00
|
|
|
// Select best class
|
2021-01-02 13:33:53 +00:00
|
|
|
String classification = null;
|
2021-01-07 12:39:42 +00:00
|
|
|
if (chances.get(0).chance > CHANCE_MINIMUM &&
|
|
|
|
chances.get(0).chance / chances.get(1).chance >= CHANCE_THRESHOLD)
|
2021-01-02 13:33:53 +00:00
|
|
|
classification = chances.get(0).clazz;
|
|
|
|
|
2021-01-06 10:39:41 +00:00
|
|
|
Log.i("Classifier current=" + currentClass + " classified=" + classification);
|
2021-01-02 13:33:53 +00:00
|
|
|
|
|
|
|
return classification;
|
|
|
|
}
|
|
|
|
|
2021-01-07 13:17:38 +00:00
|
|
|
private static void process(long account, String currentClass, boolean added, String word, State state) {
|
|
|
|
if (word != null) {
|
|
|
|
word = word.trim().toLowerCase();
|
|
|
|
|
|
|
|
if (word.length() < 2 ||
|
|
|
|
state.words.contains(word) ||
|
|
|
|
word.matches(".*\\d.*"))
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
state.words.add(word);
|
|
|
|
|
2021-01-06 20:35:11 +00:00
|
|
|
if (state.words.size() < 3)
|
|
|
|
return;
|
|
|
|
|
|
|
|
String before = state.words.get(state.words.size() - 3);
|
|
|
|
String current = state.words.get(state.words.size() - 2);
|
|
|
|
String after = state.words.get(state.words.size() - 1);
|
|
|
|
|
|
|
|
Map<String, Frequency> classFrequency = wordClassFrequency.get(account).get(current);
|
|
|
|
if (added) {
|
|
|
|
if (classFrequency == null) {
|
|
|
|
classFrequency = new HashMap<>();
|
|
|
|
wordClassFrequency.get(account).put(current, classFrequency);
|
|
|
|
}
|
|
|
|
|
|
|
|
for (String clazz : classFrequency.keySet()) {
|
|
|
|
Frequency frequency = classFrequency.get(clazz);
|
2021-01-07 08:18:29 +00:00
|
|
|
if (frequency.count > 0) {
|
|
|
|
Stat stat = state.classStats.get(clazz);
|
|
|
|
if (stat == null) {
|
|
|
|
stat = new Stat();
|
|
|
|
state.classStats.put(clazz, stat);
|
|
|
|
}
|
2021-01-06 20:35:11 +00:00
|
|
|
|
2021-01-07 08:18:29 +00:00
|
|
|
int c = frequency.count;
|
|
|
|
Integer b = (before == null ? null : frequency.before.get(before));
|
|
|
|
Integer a = (after == null ? null : frequency.after.get(after));
|
2021-01-07 09:58:56 +00:00
|
|
|
double f = ((b == null ? 0 : b) + c + (a == null ? 0 : a)) / 3.0;
|
|
|
|
stat.totalFrequency += f;
|
2021-01-06 20:35:11 +00:00
|
|
|
|
2021-01-07 08:18:29 +00:00
|
|
|
stat.matchedWords++;
|
|
|
|
if (stat.matchedWords > state.maxMatchedWords)
|
|
|
|
state.maxMatchedWords = stat.matchedWords;
|
2021-01-06 20:35:11 +00:00
|
|
|
|
2021-01-07 08:18:29 +00:00
|
|
|
if (BuildConfig.DEBUG)
|
|
|
|
stat.words.add(current);
|
|
|
|
}
|
2021-01-06 20:35:11 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
Frequency c = classFrequency.get(currentClass);
|
|
|
|
if (c == null)
|
|
|
|
c = new Frequency();
|
2021-01-07 08:18:29 +00:00
|
|
|
c.add(before, after, 1);
|
2021-01-06 20:35:11 +00:00
|
|
|
classFrequency.put(currentClass, c);
|
|
|
|
} else {
|
|
|
|
Frequency c = (classFrequency == null ? null : classFrequency.get(currentClass));
|
|
|
|
if (c != null)
|
2021-01-07 08:18:29 +00:00
|
|
|
c.add(before, after, -1);
|
2021-01-06 20:35:11 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2021-01-02 19:42:35 +00:00
|
|
|
static synchronized void save(Context context) throws JSONException, IOException {
|
2021-01-02 21:52:41 +00:00
|
|
|
if (!dirty)
|
2021-01-02 21:39:30 +00:00
|
|
|
return;
|
2021-01-02 19:42:35 +00:00
|
|
|
|
|
|
|
File file = getFile(context);
|
2021-01-04 14:35:57 +00:00
|
|
|
Helper.writeText(file, toJson().toString(2));
|
2021-01-02 19:42:35 +00:00
|
|
|
|
2021-01-03 09:06:08 +00:00
|
|
|
dirty = false;
|
2021-01-04 14:35:57 +00:00
|
|
|
Log.i("Classifier data saved");
|
2021-01-02 19:42:35 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
private static synchronized void load(Context context) throws IOException, JSONException {
|
2021-01-05 12:42:43 +00:00
|
|
|
if (loaded || dirty)
|
2021-01-02 19:42:35 +00:00
|
|
|
return;
|
|
|
|
|
|
|
|
wordClassFrequency.clear();
|
|
|
|
|
|
|
|
File file = getFile(context);
|
|
|
|
if (file.exists()) {
|
|
|
|
String json = Helper.readText(file);
|
2021-01-04 14:35:57 +00:00
|
|
|
fromJson(new JSONObject(json));
|
2021-01-02 19:42:35 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
loaded = true;
|
2021-01-04 14:35:57 +00:00
|
|
|
Log.i("Classifier data loaded");
|
2021-01-02 19:42:35 +00:00
|
|
|
}
|
|
|
|
|
2021-01-03 19:21:35 +00:00
|
|
|
static synchronized void clear(Context context) {
|
|
|
|
wordClassFrequency.clear();
|
|
|
|
dirty = true;
|
2021-01-05 12:42:43 +00:00
|
|
|
Log.i("Classifier data cleared");
|
2021-01-03 19:21:35 +00:00
|
|
|
}
|
|
|
|
|
2021-01-02 21:38:37 +00:00
|
|
|
static boolean isEnabled(Context context) {
|
2021-01-02 19:42:35 +00:00
|
|
|
SharedPreferences prefs = PreferenceManager.getDefaultSharedPreferences(context);
|
2021-01-03 19:21:35 +00:00
|
|
|
return prefs.getBoolean("classification", false);
|
2021-01-02 19:42:35 +00:00
|
|
|
}
|
|
|
|
|
2021-01-03 07:27:27 +00:00
|
|
|
static boolean canClassify(String folderType) {
|
|
|
|
return EntityFolder.INBOX.equals(folderType) ||
|
|
|
|
EntityFolder.JUNK.equals(folderType) ||
|
|
|
|
EntityFolder.USER.equals(folderType);
|
|
|
|
}
|
2021-01-02 19:42:35 +00:00
|
|
|
|
2021-01-03 09:58:51 +00:00
|
|
|
static File getFile(Context context) {
|
2021-01-02 19:42:35 +00:00
|
|
|
return new File(context.getFilesDir(), "classifier.json");
|
|
|
|
}
|
|
|
|
|
2021-01-04 14:35:57 +00:00
|
|
|
static JSONObject toJson() throws JSONException {
|
2021-01-07 18:20:17 +00:00
|
|
|
JSONArray jmessages = new JSONArray();
|
|
|
|
for (Long account : classMessages.keySet())
|
|
|
|
for (String clazz : classMessages.get(account).keySet()) {
|
|
|
|
JSONObject jmessage = new JSONObject();
|
|
|
|
jmessage.put("account", account);
|
|
|
|
jmessage.put("class", clazz);
|
|
|
|
jmessage.put("count", classMessages.get(account).get(clazz));
|
|
|
|
jmessages.put(jmessage);
|
|
|
|
}
|
|
|
|
|
2021-01-04 14:35:57 +00:00
|
|
|
JSONArray jwords = new JSONArray();
|
2021-01-07 09:58:56 +00:00
|
|
|
for (Long account : wordClassFrequency.keySet())
|
2021-01-04 14:35:57 +00:00
|
|
|
for (String word : wordClassFrequency.get(account).keySet()) {
|
2021-01-06 20:35:11 +00:00
|
|
|
Map<String, Frequency> classFrequency = wordClassFrequency.get(account).get(word);
|
2021-01-04 14:35:57 +00:00
|
|
|
for (String clazz : classFrequency.keySet()) {
|
2021-01-06 20:35:11 +00:00
|
|
|
Frequency f = classFrequency.get(clazz);
|
2021-01-04 14:35:57 +00:00
|
|
|
JSONObject jword = new JSONObject();
|
|
|
|
jword.put("account", account);
|
|
|
|
jword.put("word", word);
|
|
|
|
jword.put("class", clazz);
|
2021-01-06 20:35:11 +00:00
|
|
|
jword.put("frequency", f.count);
|
|
|
|
jword.put("before", from(f.before));
|
|
|
|
jword.put("after", from(f.after));
|
2021-01-04 14:35:57 +00:00
|
|
|
jwords.put(jword);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
JSONObject jroot = new JSONObject();
|
2021-01-07 18:20:17 +00:00
|
|
|
jroot.put("version", 1);
|
|
|
|
jroot.put("messages", jmessages);
|
2021-01-04 14:35:57 +00:00
|
|
|
jroot.put("words", jwords);
|
|
|
|
|
|
|
|
return jroot;
|
|
|
|
}
|
|
|
|
|
2021-01-07 08:18:29 +00:00
|
|
|
private static JSONObject from(Map<String, Integer> map) throws JSONException {
|
|
|
|
JSONObject jmap = new JSONObject();
|
|
|
|
for (String key : map.keySet())
|
|
|
|
jmap.put(key, map.get(key));
|
|
|
|
return jmap;
|
2021-01-06 20:35:11 +00:00
|
|
|
}
|
|
|
|
|
2021-01-04 14:35:57 +00:00
|
|
|
static void fromJson(JSONObject jroot) throws JSONException {
|
2021-01-07 18:20:17 +00:00
|
|
|
int version = jroot.optInt("version");
|
|
|
|
if (version < 1)
|
|
|
|
return;
|
|
|
|
|
|
|
|
JSONArray jmessages = jroot.getJSONArray("messages");
|
|
|
|
for (int m = 0; m < jmessages.length(); m++) {
|
|
|
|
JSONObject jmessage = (JSONObject) jmessages.get(m);
|
|
|
|
long account = jmessage.getLong("account");
|
|
|
|
if (!classMessages.containsKey(account))
|
|
|
|
classMessages.put(account, new HashMap<>());
|
|
|
|
String clazz = jmessage.getString("class");
|
|
|
|
int count = jmessage.getInt("count");
|
|
|
|
classMessages.get(account).put(clazz, count);
|
|
|
|
}
|
|
|
|
|
2021-01-04 14:35:57 +00:00
|
|
|
JSONArray jwords = jroot.getJSONArray("words");
|
|
|
|
for (int w = 0; w < jwords.length(); w++) {
|
|
|
|
JSONObject jword = (JSONObject) jwords.get(w);
|
|
|
|
long account = jword.getLong("account");
|
|
|
|
if (!wordClassFrequency.containsKey(account))
|
|
|
|
wordClassFrequency.put(account, new HashMap<>());
|
|
|
|
String word = jword.getString("word");
|
2021-01-06 20:35:11 +00:00
|
|
|
Map<String, Frequency> classFrequency = wordClassFrequency.get(account).get(word);
|
2021-01-04 14:35:57 +00:00
|
|
|
if (classFrequency == null) {
|
|
|
|
classFrequency = new HashMap<>();
|
|
|
|
wordClassFrequency.get(account).put(word, classFrequency);
|
|
|
|
}
|
2021-01-06 20:35:11 +00:00
|
|
|
Frequency f = new Frequency();
|
|
|
|
f.count = jword.getInt("frequency");
|
|
|
|
if (jword.has("before"))
|
2021-01-07 08:18:29 +00:00
|
|
|
f.before = from(jword.getJSONObject("before"));
|
2021-01-06 20:35:11 +00:00
|
|
|
if (jword.has("after"))
|
2021-01-07 08:18:29 +00:00
|
|
|
f.after = from(jword.getJSONObject("after"));
|
2021-01-06 20:35:11 +00:00
|
|
|
classFrequency.put(jword.getString("class"), f);
|
2021-01-04 14:35:57 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2021-01-07 08:18:29 +00:00
|
|
|
private static Map<String, Integer> from(JSONObject jmap) throws JSONException {
|
|
|
|
Map<String, Integer> result = new HashMap<>(jmap.length());
|
|
|
|
Iterator<String> iterator = jmap.keys();
|
|
|
|
while (iterator.hasNext()) {
|
|
|
|
String key = iterator.next();
|
|
|
|
result.put(key, jmap.getInt(key));
|
|
|
|
}
|
2021-01-06 20:35:11 +00:00
|
|
|
return result;
|
|
|
|
}
|
|
|
|
|
|
|
|
private static class State {
|
2021-01-07 08:18:29 +00:00
|
|
|
private int maxMatchedWords = 0;
|
|
|
|
private List<String> words = new ArrayList<>();
|
|
|
|
private Map<String, Stat> classStats = new HashMap<>();
|
2021-01-06 20:35:11 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
private static class Frequency {
|
2021-01-07 08:18:29 +00:00
|
|
|
private int count = 0;
|
|
|
|
private Map<String, Integer> before = new HashMap<>();
|
|
|
|
private Map<String, Integer> after = new HashMap<>();
|
|
|
|
|
|
|
|
private void add(String b, String a, int c) {
|
|
|
|
if (count + c < 0)
|
|
|
|
return;
|
|
|
|
|
|
|
|
count += c;
|
|
|
|
|
|
|
|
if (b != null) {
|
|
|
|
Integer x = before.get(b);
|
|
|
|
before.put(b, (x == null ? 0 : x) + c);
|
|
|
|
}
|
|
|
|
|
|
|
|
if (a != null) {
|
|
|
|
Integer x = after.get(a);
|
|
|
|
after.put(a, (x == null ? 0 : x) + c);
|
|
|
|
}
|
|
|
|
}
|
2021-01-06 20:35:11 +00:00
|
|
|
}
|
|
|
|
|
2021-01-02 13:33:53 +00:00
|
|
|
private static class Stat {
|
2021-01-07 19:11:05 +00:00
|
|
|
private int matchedWords = 0;
|
|
|
|
private double totalFrequency = 0;
|
|
|
|
private List<String> words = new ArrayList<>();
|
2021-01-02 13:33:53 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
private static class Chance {
|
2021-01-07 08:18:29 +00:00
|
|
|
private String clazz;
|
|
|
|
private Double chance;
|
2021-01-02 13:33:53 +00:00
|
|
|
|
2021-01-07 08:18:29 +00:00
|
|
|
private Chance(String clazz, Double chance) {
|
2021-01-02 13:33:53 +00:00
|
|
|
this.clazz = clazz;
|
|
|
|
this.chance = chance;
|
|
|
|
}
|
|
|
|
|
|
|
|
@NotNull
|
|
|
|
@Override
|
|
|
|
public String toString() {
|
2021-01-07 13:55:14 +00:00
|
|
|
return clazz + "=" + Math.round(chance * 100.0 * 100.0) / 100.0 + "%";
|
2021-01-02 13:33:53 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|