Spaces:
Sleeping
Sleeping
Upload folder using huggingface_hub
Browse files- README.md +8 -11
- app.py +11 -0
- data/auth/users.json +1 -0
- requirements.txt +14 -0
- src/app/__init__.py +15 -0
- src/app/auth.py +89 -0
- src/app/config.py +44 -0
- src/app/conversation_core.py +357 -0
- src/app/flashcards_tools.py +241 -0
- src/app/main_app.py +1235 -0
- src/app/ocr_tools.py +64 -0
- src/app/quiz_tools.py +59 -0
- src/app/viewers.py +272 -0
- src/generate_flashcard_viewer.py +32 -0
- src/generate_quiz.py +2 -0
- src/generate_quiz_viewer.py +2 -0
README.md
CHANGED
|
@@ -1,12 +1,9 @@
|
|
| 1 |
-
|
| 2 |
-
title: Agentic Language Partner
|
| 3 |
-
emoji: 🔥
|
| 4 |
-
colorFrom: green
|
| 5 |
-
colorTo: yellow
|
| 6 |
-
sdk: gradio
|
| 7 |
-
sdk_version: 6.0.1
|
| 8 |
-
app_file: app.py
|
| 9 |
-
pinned: false
|
| 10 |
-
---
|
| 11 |
|
| 12 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# Agentic Language Partner
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2 |
|
| 3 |
+
Deployed automatically from Google Colab for `mastefan`.
|
| 4 |
+
This space runs a full Streamlit app including:
|
| 5 |
+
|
| 6 |
+
- Audio conversation partner (Qwen + Whisper)
|
| 7 |
+
- OCR → flashcards
|
| 8 |
+
- Flashcard viewer + quizzes
|
| 9 |
+
- User login system (local JSON)
|
app.py
ADDED
|
@@ -0,0 +1,11 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import sys
|
| 2 |
+
from pathlib import Path
|
| 3 |
+
|
| 4 |
+
ROOT = Path(__file__).resolve().parent
|
| 5 |
+
SRC = ROOT / "src"
|
| 6 |
+
sys.path.append(str(SRC))
|
| 7 |
+
|
| 8 |
+
from app.main_app import main
|
| 9 |
+
|
| 10 |
+
if __name__ == "__main__":
|
| 11 |
+
main()
|
data/auth/users.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{}
|
requirements.txt
ADDED
|
@@ -0,0 +1,14 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
streamlit==1.28.0
|
| 2 |
+
tornado==6.3.3
|
| 3 |
+
deep-translator
|
| 4 |
+
pytesseract
|
| 5 |
+
pillow
|
| 6 |
+
gTTS
|
| 7 |
+
pydub
|
| 8 |
+
soundfile
|
| 9 |
+
transformers==4.40.0
|
| 10 |
+
accelerate
|
| 11 |
+
sentencepiece
|
| 12 |
+
faster-whisper==1.0.3
|
| 13 |
+
ctranslate2==4.5.0
|
| 14 |
+
streamlit-audiorecorder
|
src/app/__init__.py
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
|
| 2 |
+
"""
|
| 3 |
+
App package initializer for the Agentic Language Partner project.
|
| 4 |
+
Makes core modules importable and organizes app-wide namespaces.
|
| 5 |
+
"""
|
| 6 |
+
|
| 7 |
+
__all__ = [
|
| 8 |
+
"auth",
|
| 9 |
+
"config",
|
| 10 |
+
"conversation_core",
|
| 11 |
+
"flashcards_tools",
|
| 12 |
+
"ocr_tools",
|
| 13 |
+
"quiz_tools",
|
| 14 |
+
"viewers",
|
| 15 |
+
]
|
src/app/auth.py
ADDED
|
@@ -0,0 +1,89 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
|
| 2 |
+
# src/app/auth.py
|
| 3 |
+
|
| 4 |
+
import json
|
| 5 |
+
from pathlib import Path
|
| 6 |
+
from typing import Any, Dict, List, Optional
|
| 7 |
+
|
| 8 |
+
|
| 9 |
+
from app.config import get_user_dir
|
| 10 |
+
|
| 11 |
+
|
| 12 |
+
def _get_users_json() -> Path:
|
| 13 |
+
"""
|
| 14 |
+
Returns: Path to the main users.json file.
|
| 15 |
+
"""
|
| 16 |
+
root = Path(__file__).resolve().parents[2]
|
| 17 |
+
auth_dir = root / "data" / "auth"
|
| 18 |
+
auth_dir.mkdir(parents=True, exist_ok=True)
|
| 19 |
+
users_file = auth_dir / "users.json"
|
| 20 |
+
|
| 21 |
+
if not users_file.exists():
|
| 22 |
+
users_file.write_text("{}", encoding="utf-8")
|
| 23 |
+
|
| 24 |
+
return users_file
|
| 25 |
+
|
| 26 |
+
|
| 27 |
+
def _load_users() -> Dict[str, Dict]:
|
| 28 |
+
users_file = _get_users_json()
|
| 29 |
+
try:
|
| 30 |
+
return json.loads(users_file.read_text(encoding="utf-8"))
|
| 31 |
+
except Exception:
|
| 32 |
+
return {}
|
| 33 |
+
|
| 34 |
+
|
| 35 |
+
def _save_users(data: Dict[str, Dict]) -> None:
|
| 36 |
+
users_file = _get_users_json()
|
| 37 |
+
users_file.write_text(json.dumps(data, indent=2), encoding="utf-8")
|
| 38 |
+
|
| 39 |
+
|
| 40 |
+
# ------------------------------------------------------------
|
| 41 |
+
# AUTH FUNCTIONS
|
| 42 |
+
# ------------------------------------------------------------
|
| 43 |
+
|
| 44 |
+
def register_user(username: str, password: str) -> bool:
|
| 45 |
+
if not username or not password:
|
| 46 |
+
return False
|
| 47 |
+
|
| 48 |
+
users = _load_users()
|
| 49 |
+
if username in users:
|
| 50 |
+
return False
|
| 51 |
+
|
| 52 |
+
users[username] = {
|
| 53 |
+
"password": password,
|
| 54 |
+
"prefs": {
|
| 55 |
+
"target_language": "english",
|
| 56 |
+
"native_language": "english",
|
| 57 |
+
"cefr_level": "B1",
|
| 58 |
+
"topic": "general conversation",
|
| 59 |
+
},
|
| 60 |
+
}
|
| 61 |
+
_save_users(users)
|
| 62 |
+
|
| 63 |
+
# create the user folder
|
| 64 |
+
get_user_dir(username)
|
| 65 |
+
|
| 66 |
+
return True
|
| 67 |
+
|
| 68 |
+
|
| 69 |
+
def authenticate_user(username: str, password: str) -> bool:
|
| 70 |
+
users = _load_users()
|
| 71 |
+
entry = users.get(username)
|
| 72 |
+
if not entry:
|
| 73 |
+
return False
|
| 74 |
+
return entry.get("password") == password
|
| 75 |
+
|
| 76 |
+
|
| 77 |
+
def get_user_prefs(username: str) -> Dict:
|
| 78 |
+
users = _load_users()
|
| 79 |
+
entry = users.get(username, {})
|
| 80 |
+
return entry.get("prefs", {})
|
| 81 |
+
|
| 82 |
+
|
| 83 |
+
def update_user_prefs(username: str, prefs: Dict) -> None:
|
| 84 |
+
users = _load_users()
|
| 85 |
+
if username not in users:
|
| 86 |
+
return
|
| 87 |
+
users[username]["prefs"] = prefs
|
| 88 |
+
_save_users(users)
|
| 89 |
+
|
src/app/config.py
ADDED
|
@@ -0,0 +1,44 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
|
| 2 |
+
# src/app/config.py
|
| 3 |
+
|
| 4 |
+
from pathlib import Path
|
| 5 |
+
|
| 6 |
+
|
| 7 |
+
def get_project_root() -> Path:
|
| 8 |
+
"""
|
| 9 |
+
Returns the project root directory.
|
| 10 |
+
|
| 11 |
+
Assumes this file is located at:
|
| 12 |
+
<project_root>/src/app/config.py
|
| 13 |
+
so we go up two parents.
|
| 14 |
+
"""
|
| 15 |
+
return Path(__file__).resolve().parents[2]
|
| 16 |
+
|
| 17 |
+
|
| 18 |
+
def get_data_dir() -> Path:
|
| 19 |
+
"""
|
| 20 |
+
Root data directory for all persisted user content.
|
| 21 |
+
"""
|
| 22 |
+
root = get_project_root()
|
| 23 |
+
data_dir = root / "data"
|
| 24 |
+
data_dir.mkdir(parents=True, exist_ok=True)
|
| 25 |
+
return data_dir
|
| 26 |
+
|
| 27 |
+
|
| 28 |
+
def get_user_dir(username: str) -> Path:
|
| 29 |
+
"""
|
| 30 |
+
Returns the directory for a given user and ensures that
|
| 31 |
+
its subfolders exist.
|
| 32 |
+
"""
|
| 33 |
+
data_dir = get_data_dir()
|
| 34 |
+
user_dir = data_dir / "users" / username
|
| 35 |
+
user_dir.mkdir(parents=True, exist_ok=True)
|
| 36 |
+
|
| 37 |
+
# create standard subfolders
|
| 38 |
+
(user_dir / "decks").mkdir(parents=True, exist_ok=True)
|
| 39 |
+
(user_dir / "viewers").mkdir(parents=True, exist_ok=True)
|
| 40 |
+
(user_dir / "chats").mkdir(parents=True, exist_ok=True)
|
| 41 |
+
(user_dir / "quizzes").mkdir(parents=True, exist_ok=True)
|
| 42 |
+
|
| 43 |
+
return user_dir
|
| 44 |
+
|
src/app/conversation_core.py
ADDED
|
@@ -0,0 +1,357 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
|
| 2 |
+
###############################################################
|
| 3 |
+
# conversation_core.py — Agentic Partner Core (Qwen 1.5B + Whisper)
|
| 4 |
+
###############################################################
|
| 5 |
+
|
| 6 |
+
import io
|
| 7 |
+
import re
|
| 8 |
+
import tempfile
|
| 9 |
+
from dataclasses import dataclass
|
| 10 |
+
from typing import List, Optional, Tuple
|
| 11 |
+
|
| 12 |
+
import torch
|
| 13 |
+
from gtts import gTTS
|
| 14 |
+
from faster_whisper import WhisperModel
|
| 15 |
+
from transformers import (
|
| 16 |
+
AutoTokenizer,
|
| 17 |
+
AutoModelForCausalLM,
|
| 18 |
+
)
|
| 19 |
+
|
| 20 |
+
###############################################################
|
| 21 |
+
# MODEL / LANGUAGE CONSTANTS
|
| 22 |
+
###############################################################
|
| 23 |
+
|
| 24 |
+
QWEN_MODEL_NAME = "Qwen/Qwen2.5-1.5B-Instruct"
|
| 25 |
+
WHISPER_MODEL_SIZE = "base" # you can change to "large-v3" if GPU budget allows
|
| 26 |
+
|
| 27 |
+
_QWEN_TOKENIZER = None
|
| 28 |
+
_QWEN_MODEL = None
|
| 29 |
+
_WHISPER = None
|
| 30 |
+
|
| 31 |
+
# Spoken language -> Whisper hint
|
| 32 |
+
WHISPER_LANG_MAP = {
|
| 33 |
+
"english": "en",
|
| 34 |
+
"german": "de",
|
| 35 |
+
"spanish": "es",
|
| 36 |
+
"russian": "ru",
|
| 37 |
+
"french": "fr",
|
| 38 |
+
"italian": "it",
|
| 39 |
+
"japanese": "ja",
|
| 40 |
+
"chinese": "zh",
|
| 41 |
+
"korean": "ko",
|
| 42 |
+
"arabic": "ar",
|
| 43 |
+
"hindi": "hi",
|
| 44 |
+
}
|
| 45 |
+
|
| 46 |
+
# Spoken language -> gTTS language code
|
| 47 |
+
GTTS_LANG = {
|
| 48 |
+
"english": "en",
|
| 49 |
+
"spanish": "es",
|
| 50 |
+
"german": "de",
|
| 51 |
+
"russian": "ru",
|
| 52 |
+
"japanese": "ja",
|
| 53 |
+
"chinese": "zh-cn",
|
| 54 |
+
"korean": "ko",
|
| 55 |
+
"french": "fr",
|
| 56 |
+
"italian": "it",
|
| 57 |
+
}
|
| 58 |
+
|
| 59 |
+
CONTROL_PROMPTS = {
|
| 60 |
+
"A1": "Use extremely short, simple sentences and very basic vocabulary.",
|
| 61 |
+
"A2": "Use simple sentences and common everyday vocabulary.",
|
| 62 |
+
"B1": "Use moderately complex sentences and conversational vocabulary.",
|
| 63 |
+
"B2": "Use natural, fluent sentences with richer vocabulary.",
|
| 64 |
+
"C1": "Use complex, advanced sentences with nuanced expressions.",
|
| 65 |
+
"C2": "Use highly sophisticated, near-native language and style.",
|
| 66 |
+
}
|
| 67 |
+
|
| 68 |
+
|
| 69 |
+
###############################################################
|
| 70 |
+
# GLOBAL LOADERS
|
| 71 |
+
###############################################################
|
| 72 |
+
|
| 73 |
+
def load_partner_lm() -> Tuple[AutoTokenizer, AutoModelForCausalLM]:
|
| 74 |
+
global _QWEN_TOKENIZER, _QWEN_MODEL
|
| 75 |
+
if _QWEN_TOKENIZER is not None and _QWEN_MODEL is not None:
|
| 76 |
+
return _QWEN_TOKENIZER, _QWEN_MODEL
|
| 77 |
+
|
| 78 |
+
print("[conversation_core] Loading partner LM:", QWEN_MODEL_NAME)
|
| 79 |
+
tok = AutoTokenizer.from_pretrained(QWEN_MODEL_NAME, trust_remote_code=True)
|
| 80 |
+
model = AutoModelForCausalLM.from_pretrained(
|
| 81 |
+
QWEN_MODEL_NAME,
|
| 82 |
+
torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32,
|
| 83 |
+
device_map="auto",
|
| 84 |
+
trust_remote_code=True,
|
| 85 |
+
)
|
| 86 |
+
_QWEN_TOKENIZER = tok
|
| 87 |
+
_QWEN_MODEL = model
|
| 88 |
+
return tok, model
|
| 89 |
+
|
| 90 |
+
|
| 91 |
+
def load_whisper() -> WhisperModel:
|
| 92 |
+
global _WHISPER
|
| 93 |
+
if _WHISPER is not None:
|
| 94 |
+
return _WHISPER
|
| 95 |
+
|
| 96 |
+
device = "cuda" if torch.cuda.is_available() else "cpu"
|
| 97 |
+
compute_type = "float16" if device == "cuda" else "int8"
|
| 98 |
+
print(f"[conversation_core] Loading Whisper {WHISPER_MODEL_SIZE} on {device} ({compute_type})")
|
| 99 |
+
_WHISPER = WhisperModel(WHISPER_MODEL_SIZE, device=device, compute_type=compute_type)
|
| 100 |
+
return _WHISPER
|
| 101 |
+
|
| 102 |
+
|
| 103 |
+
###############################################################
|
| 104 |
+
# DATA STRUCTURE
|
| 105 |
+
###############################################################
|
| 106 |
+
|
| 107 |
+
@dataclass
|
| 108 |
+
class ConversationTurn:
|
| 109 |
+
role: str
|
| 110 |
+
text: str
|
| 111 |
+
|
| 112 |
+
|
| 113 |
+
###############################################################
|
| 114 |
+
# CLEANING
|
| 115 |
+
###############################################################
|
| 116 |
+
|
| 117 |
+
def clean_assistant_reply(text: str) -> str:
|
| 118 |
+
"""Strip meta, identity, and obvious junk from LM output."""
|
| 119 |
+
if not text:
|
| 120 |
+
return ""
|
| 121 |
+
|
| 122 |
+
# Remove labels
|
| 123 |
+
text = re.sub(r"(?i)\b(user|assistant|system)\s*:\s*", "", text)
|
| 124 |
+
|
| 125 |
+
# Remove numbered / bullet lists (not wanted in casual chat)
|
| 126 |
+
text = re.sub(r"(?m)^\s*[-•*]\s+.*$", "", text)
|
| 127 |
+
text = re.sub(r"(?m)^\s*\d+\.\s+.*$", "", text)
|
| 128 |
+
|
| 129 |
+
# Remove obvious identity / HR / meta nonsense
|
| 130 |
+
identity_patterns = [
|
| 131 |
+
r"(?i)i am (an?|the)? ?(ai|assistant|speaker|model|natural person).*",
|
| 132 |
+
r"(?i)my name is [A-Za-zäöüÄÖÜß]+.*",
|
| 133 |
+
r"(?i)i was created.*",
|
| 134 |
+
r"(?i)human resources manager.*",
|
| 135 |
+
r"(?i)job description.*",
|
| 136 |
+
r"(?i)i am a large language model.*",
|
| 137 |
+
]
|
| 138 |
+
for pat in identity_patterns:
|
| 139 |
+
text = re.sub(pat, "", text)
|
| 140 |
+
|
| 141 |
+
# Trim hanging word fragments at the end
|
| 142 |
+
text = re.sub(r"[A-Za-zÄÖÜäöüß]+$", "", text)
|
| 143 |
+
|
| 144 |
+
# Collapse whitespace
|
| 145 |
+
text = re.sub(r"\s{2,}", " ", text)
|
| 146 |
+
return text.strip()
|
| 147 |
+
|
| 148 |
+
|
| 149 |
+
###############################################################
|
| 150 |
+
# CONVERSATION MANAGER
|
| 151 |
+
###############################################################
|
| 152 |
+
|
| 153 |
+
class ConversationManager:
|
| 154 |
+
def __init__(
|
| 155 |
+
self,
|
| 156 |
+
target_language: str = "german",
|
| 157 |
+
native_language: str = "english",
|
| 158 |
+
cefr_level: str = "B1",
|
| 159 |
+
topic: str = "general conversation",
|
| 160 |
+
):
|
| 161 |
+
self.target_language = (target_language or "english").strip().lower()
|
| 162 |
+
self.native_language = (native_language or "english").strip().lower()
|
| 163 |
+
self.cefr_level = cefr_level or "B1"
|
| 164 |
+
self.topic = topic or "general conversation"
|
| 165 |
+
self.history: List[ConversationTurn] = []
|
| 166 |
+
|
| 167 |
+
# Warm-load models once per session
|
| 168 |
+
load_partner_lm()
|
| 169 |
+
load_whisper()
|
| 170 |
+
|
| 171 |
+
###########################################################
|
| 172 |
+
# PROMPT + GENERATION
|
| 173 |
+
###########################################################
|
| 174 |
+
|
| 175 |
+
def _build_system_prompt(self) -> str:
|
| 176 |
+
base = (
|
| 177 |
+
f"You are a friendly conversation partner speaking {self.target_language}. "
|
| 178 |
+
f"Reply ONLY in {self.target_language}. "
|
| 179 |
+
f"Do NOT explain grammar, vocabulary, or translations unless the user explicitly asks. "
|
| 180 |
+
f"Do NOT describe what the sentence means, do NOT say 'the sentence translates to...', "
|
| 181 |
+
f"and do NOT mention that you are explaining anything. "
|
| 182 |
+
f"Adapt your language to CEFR level {self.cefr_level}. "
|
| 183 |
+
f"{CONTROL_PROMPTS.get(self.cefr_level, '')} "
|
| 184 |
+
"Keep your replies natural and conversational, usually 1–3 short sentences. "
|
| 185 |
+
"Ask exactly ONE natural follow-up question related to what the user said. "
|
| 186 |
+
"Never end the conversation unless the user explicitly ends it. "
|
| 187 |
+
"Do NOT say goodbye or conclude unless the user does. "
|
| 188 |
+
"Never talk about being an AI, model, or assistant. "
|
| 189 |
+
"Do not mention job descriptions, resumes, or HR responsibilities unless the user clearly asks. "
|
| 190 |
+
)
|
| 191 |
+
if self.topic.strip():
|
| 192 |
+
base += f"The main topic of conversation is: {self.topic.strip()}. "
|
| 193 |
+
return base
|
| 194 |
+
|
| 195 |
+
def _generate_lm(self, user_text: str) -> str:
|
| 196 |
+
tok, model = load_partner_lm()
|
| 197 |
+
|
| 198 |
+
system_prompt = self._build_system_prompt()
|
| 199 |
+
messages = [
|
| 200 |
+
{"role": "system", "content": system_prompt},
|
| 201 |
+
{
|
| 202 |
+
"role": "user",
|
| 203 |
+
"content": f"The user (who speaks {self.native_language}) said: {user_text}",
|
| 204 |
+
},
|
| 205 |
+
]
|
| 206 |
+
|
| 207 |
+
prompt = tok.apply_chat_template(
|
| 208 |
+
messages,
|
| 209 |
+
tokenize=False,
|
| 210 |
+
add_generation_prompt=True,
|
| 211 |
+
)
|
| 212 |
+
|
| 213 |
+
enc = tok(prompt, return_tensors="pt").to(model.device)
|
| 214 |
+
|
| 215 |
+
with torch.no_grad():
|
| 216 |
+
out = model.generate(
|
| 217 |
+
**enc,
|
| 218 |
+
max_new_tokens=160, # enough space for natural replies
|
| 219 |
+
temperature=0.8,
|
| 220 |
+
top_p=0.95,
|
| 221 |
+
top_k=50,
|
| 222 |
+
repetition_penalty=1.15,
|
| 223 |
+
pad_token_id=tok.eos_token_id,
|
| 224 |
+
do_sample=True,
|
| 225 |
+
)
|
| 226 |
+
|
| 227 |
+
raw = tok.decode(out[0], skip_special_tokens=True).strip()
|
| 228 |
+
|
| 229 |
+
# If the user text is echoed, strip it
|
| 230 |
+
if user_text in raw:
|
| 231 |
+
raw = raw.split(user_text)[-1].strip()
|
| 232 |
+
|
| 233 |
+
# Remove "assistant" label echoes
|
| 234 |
+
lines = [
|
| 235 |
+
ln for ln in raw.splitlines()
|
| 236 |
+
if ln.strip().lower() not in ("assistant", "assistant:")
|
| 237 |
+
]
|
| 238 |
+
raw = "\n".join(lines).strip()
|
| 239 |
+
|
| 240 |
+
return clean_assistant_reply(raw)
|
| 241 |
+
|
| 242 |
+
###########################################################
|
| 243 |
+
# PUBLIC REPLY API
|
| 244 |
+
###########################################################
|
| 245 |
+
|
| 246 |
+
def reply(self, user_text: str, input_lang: str = "german"):
|
| 247 |
+
"""Generate a reply + explanation + TTS audio."""
|
| 248 |
+
self.history.append(ConversationTurn("user", user_text))
|
| 249 |
+
|
| 250 |
+
assistant_text = self._generate_lm(user_text)
|
| 251 |
+
self.history.append(ConversationTurn("assistant", assistant_text))
|
| 252 |
+
|
| 253 |
+
explanation = self._generate_explanation(assistant_text)
|
| 254 |
+
audio = self.text_to_speech(assistant_text)
|
| 255 |
+
|
| 256 |
+
return {
|
| 257 |
+
"reply_text": assistant_text,
|
| 258 |
+
"explanation": explanation,
|
| 259 |
+
"audio": audio,
|
| 260 |
+
}
|
| 261 |
+
|
| 262 |
+
###########################################################
|
| 263 |
+
# SHORT EXPLANATION (EN / native language)
|
| 264 |
+
###########################################################
|
| 265 |
+
|
| 266 |
+
def _generate_explanation(self, assistant_text: str) -> str:
|
| 267 |
+
"""Return exactly ONE simple native-language sentence, no meta, no logic."""
|
| 268 |
+
if not assistant_text:
|
| 269 |
+
return ""
|
| 270 |
+
|
| 271 |
+
tok, model = load_partner_lm()
|
| 272 |
+
prompt = (
|
| 273 |
+
f"Rewrite the meaning of this {self.target_language} sentence "
|
| 274 |
+
f"in ONE very short {self.native_language} sentence. "
|
| 275 |
+
f"Do NOT explain what you are doing, do NOT say 'the sentence means', "
|
| 276 |
+
f"do NOT describe tone, and do NOT provide multiple versions.\n"
|
| 277 |
+
f"Sentence: \"{assistant_text}\""
|
| 278 |
+
)
|
| 279 |
+
|
| 280 |
+
enc = tok(prompt, return_tensors="pt").to(model.device)
|
| 281 |
+
with torch.no_grad():
|
| 282 |
+
out = model.generate(
|
| 283 |
+
**enc,
|
| 284 |
+
max_new_tokens=40,
|
| 285 |
+
temperature=0.6,
|
| 286 |
+
top_p=0.9,
|
| 287 |
+
pad_token_id=tok.eos_token_id,
|
| 288 |
+
)
|
| 289 |
+
|
| 290 |
+
raw = tok.decode(out[0], skip_special_tokens=True)
|
| 291 |
+
raw = raw.replace(prompt, "").strip()
|
| 292 |
+
|
| 293 |
+
# keep first sentence only
|
| 294 |
+
parts = re.split(r"(?<=[.!?])\s+", raw)
|
| 295 |
+
if parts:
|
| 296 |
+
raw = parts[0].strip()
|
| 297 |
+
|
| 298 |
+
# remove meta leftovers
|
| 299 |
+
raw = re.sub(r"(?i)the sentence.*$", "", raw)
|
| 300 |
+
raw = re.sub(r"(?i)this means.*$", "", raw)
|
| 301 |
+
|
| 302 |
+
return raw.strip()
|
| 303 |
+
|
| 304 |
+
|
| 305 |
+
|
| 306 |
+
###########################################################
|
| 307 |
+
# AUDIO TRANSCRIPTION
|
| 308 |
+
###########################################################
|
| 309 |
+
|
| 310 |
+
def transcribe(self, audio_segment, spoken_lang: str = "english"):
|
| 311 |
+
"""
|
| 312 |
+
Faster-Whisper transcription with optional language hint.
|
| 313 |
+
Returns (text, detected_lang_or_hint, dummy_confidence).
|
| 314 |
+
"""
|
| 315 |
+
whisper = load_whisper()
|
| 316 |
+
|
| 317 |
+
lang_key = (spoken_lang or "english").strip().lower()
|
| 318 |
+
lang_hint = WHISPER_LANG_MAP.get(lang_key)
|
| 319 |
+
|
| 320 |
+
with tempfile.NamedTemporaryFile(suffix=".wav", delete=False) as tmp:
|
| 321 |
+
# Normalize and export pydub AudioSegment
|
| 322 |
+
audio_segment.set_channels(1).set_frame_rate(16000).export(
|
| 323 |
+
tmp.name, format="wav"
|
| 324 |
+
)
|
| 325 |
+
|
| 326 |
+
decode_opts = {"beam_size": 5}
|
| 327 |
+
if lang_hint:
|
| 328 |
+
decode_opts["language"] = lang_hint
|
| 329 |
+
|
| 330 |
+
segments, info = whisper.transcribe(tmp.name, **decode_opts)
|
| 331 |
+
|
| 332 |
+
full_text = " ".join(getattr(seg, "text", "") for seg in segments)
|
| 333 |
+
return full_text.strip(), (lang_hint or "auto"), 1.0
|
| 334 |
+
|
| 335 |
+
###########################################################
|
| 336 |
+
# TEXT → SPEECH
|
| 337 |
+
###########################################################
|
| 338 |
+
|
| 339 |
+
def text_to_speech(self, text: str) -> Optional[bytes]:
|
| 340 |
+
"""Return MP3 bytes for the assistant text, or None on failure."""
|
| 341 |
+
if not text:
|
| 342 |
+
return None
|
| 343 |
+
try:
|
| 344 |
+
lang_code = GTTS_LANG.get(self.target_language, "en")
|
| 345 |
+
tts = gTTS(text=text, lang=lang_code)
|
| 346 |
+
buf = io.BytesIO()
|
| 347 |
+
tts.write_to_fp(buf)
|
| 348 |
+
return buf.getvalue()
|
| 349 |
+
except Exception:
|
| 350 |
+
return None
|
| 351 |
+
|
| 352 |
+
|
| 353 |
+
###############################################################
|
| 354 |
+
# END OF FILE
|
| 355 |
+
###############################################################
|
| 356 |
+
|
| 357 |
+
|
src/app/flashcards_tools.py
ADDED
|
@@ -0,0 +1,241 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
|
| 2 |
+
# src/app/flashcards_tools.py
|
| 3 |
+
|
| 4 |
+
import json
|
| 5 |
+
import re
|
| 6 |
+
from pathlib import Path
|
| 7 |
+
from typing import Dict, List, Tuple, Optional
|
| 8 |
+
|
| 9 |
+
from deep_translator import GoogleTranslator
|
| 10 |
+
|
| 11 |
+
from app.config import get_user_dir
|
| 12 |
+
|
| 13 |
+
|
| 14 |
+
def _get_decks_dir(username: str) -> Path:
|
| 15 |
+
"""
|
| 16 |
+
Returns the directory where all of a user's decks are stored.
|
| 17 |
+
"""
|
| 18 |
+
user_dir = get_user_dir(username)
|
| 19 |
+
decks_dir = user_dir / "decks"
|
| 20 |
+
decks_dir.mkdir(parents=True, exist_ok=True)
|
| 21 |
+
return decks_dir
|
| 22 |
+
|
| 23 |
+
|
| 24 |
+
def list_user_decks(username: str) -> Dict[str, Path]:
|
| 25 |
+
"""
|
| 26 |
+
Returns a mapping of deck name -> deck json path.
|
| 27 |
+
Deck name is taken from the deck's "name" field if present,
|
| 28 |
+
otherwise the filename stem.
|
| 29 |
+
"""
|
| 30 |
+
decks_dir = _get_decks_dir(username)
|
| 31 |
+
deck_files = sorted(decks_dir.glob("*.json"))
|
| 32 |
+
decks: Dict[str, Path] = {}
|
| 33 |
+
|
| 34 |
+
for path in deck_files:
|
| 35 |
+
try:
|
| 36 |
+
data = json.loads(path.read_text(encoding="utf-8"))
|
| 37 |
+
name = data.get("name") or path.stem
|
| 38 |
+
except Exception:
|
| 39 |
+
name = path.stem
|
| 40 |
+
|
| 41 |
+
# ensure uniqueness by appending stem if needed
|
| 42 |
+
if name in decks and decks[name] != path:
|
| 43 |
+
name = f"{name} ({path.stem})"
|
| 44 |
+
decks[name] = path
|
| 45 |
+
|
| 46 |
+
return decks
|
| 47 |
+
|
| 48 |
+
|
| 49 |
+
def _ensure_card_stats(card: Dict) -> None:
|
| 50 |
+
"""
|
| 51 |
+
Ensure that a card has simple spaced-repetition stats.
|
| 52 |
+
"""
|
| 53 |
+
if "score" not in card: # learning strength
|
| 54 |
+
card["score"] = 0
|
| 55 |
+
if "reviews" not in card:
|
| 56 |
+
card["reviews"] = 0
|
| 57 |
+
|
| 58 |
+
|
| 59 |
+
def load_deck(path: Path) -> Dict:
|
| 60 |
+
"""
|
| 61 |
+
Loads a deck from JSON, ensuring 'cards' exists and that
|
| 62 |
+
each card has basic stats for spaced repetition.
|
| 63 |
+
"""
|
| 64 |
+
try:
|
| 65 |
+
data = json.loads(path.read_text(encoding="utf-8"))
|
| 66 |
+
except Exception:
|
| 67 |
+
data = {}
|
| 68 |
+
if "cards" not in data or not isinstance(data["cards"], list):
|
| 69 |
+
data["cards"] = []
|
| 70 |
+
if "name" not in data:
|
| 71 |
+
data["name"] = path.stem
|
| 72 |
+
if "tags" not in data or not isinstance(data["tags"], list):
|
| 73 |
+
data["tags"] = []
|
| 74 |
+
|
| 75 |
+
for card in data["cards"]:
|
| 76 |
+
_ensure_card_stats(card)
|
| 77 |
+
|
| 78 |
+
return data
|
| 79 |
+
|
| 80 |
+
|
| 81 |
+
def save_deck(path: Path, deck: Dict) -> None:
|
| 82 |
+
"""
|
| 83 |
+
Saves deck to JSON.
|
| 84 |
+
"""
|
| 85 |
+
if "cards" not in deck:
|
| 86 |
+
deck["cards"] = []
|
| 87 |
+
if "name" not in deck:
|
| 88 |
+
deck["name"] = path.stem
|
| 89 |
+
if "tags" not in deck or not isinstance(deck["tags"], list):
|
| 90 |
+
deck["tags"] = []
|
| 91 |
+
|
| 92 |
+
# make sure stats are present
|
| 93 |
+
for card in deck["cards"]:
|
| 94 |
+
_ensure_card_stats(card)
|
| 95 |
+
|
| 96 |
+
path.write_text(json.dumps(deck, indent=2, ensure_ascii=False), encoding="utf-8")
|
| 97 |
+
|
| 98 |
+
|
| 99 |
+
# ------------------------------------------------------------
|
| 100 |
+
# Shared tokenization
|
| 101 |
+
# ------------------------------------------------------------
|
| 102 |
+
|
| 103 |
+
def _extract_candidate_words(text: str) -> List[str]:
|
| 104 |
+
"""
|
| 105 |
+
Simple tokenizer & filter for candidate vocab words.
|
| 106 |
+
"""
|
| 107 |
+
tokens = re.findall(r"\b\w+\b", text, flags=re.UNICODE)
|
| 108 |
+
out = []
|
| 109 |
+
seen = set()
|
| 110 |
+
for t in tokens:
|
| 111 |
+
t_norm = t.strip()
|
| 112 |
+
if len(t_norm) < 2:
|
| 113 |
+
continue
|
| 114 |
+
if any(ch.isdigit() for ch in t_norm):
|
| 115 |
+
continue
|
| 116 |
+
lower = t_norm.lower()
|
| 117 |
+
if lower in seen:
|
| 118 |
+
continue
|
| 119 |
+
seen.add(lower)
|
| 120 |
+
out.append(t_norm)
|
| 121 |
+
return out
|
| 122 |
+
|
| 123 |
+
|
| 124 |
+
# ------------------------------------------------------------
|
| 125 |
+
# OCR → Flashcards
|
| 126 |
+
# ------------------------------------------------------------
|
| 127 |
+
|
| 128 |
+
def generate_flashcards_from_ocr_results(
|
| 129 |
+
username: str,
|
| 130 |
+
ocr_results: List[Dict],
|
| 131 |
+
deck_name: str = "ocr",
|
| 132 |
+
target_lang: str = "en",
|
| 133 |
+
tags: Optional[List[str]] = None,
|
| 134 |
+
) -> Path:
|
| 135 |
+
"""
|
| 136 |
+
Takes OCR results (as produced by ocr_tools.ocr_and_translate_batch)
|
| 137 |
+
and constructs a simple vocab deck.
|
| 138 |
+
|
| 139 |
+
ocr_results: list of dict with keys:
|
| 140 |
+
- "text": original text
|
| 141 |
+
- optionally other fields (ignored)
|
| 142 |
+
"""
|
| 143 |
+
all_text = []
|
| 144 |
+
for res in ocr_results:
|
| 145 |
+
t = res.get("text") or res.get("raw_text") or ""
|
| 146 |
+
if t:
|
| 147 |
+
all_text.append(t)
|
| 148 |
+
joined = "\n".join(all_text)
|
| 149 |
+
|
| 150 |
+
words = _extract_candidate_words(joined)
|
| 151 |
+
if not words:
|
| 152 |
+
raise ValueError("No candidate words found in OCR results.")
|
| 153 |
+
|
| 154 |
+
translator = GoogleTranslator(source="auto", target=target_lang)
|
| 155 |
+
cards = []
|
| 156 |
+
for w in words:
|
| 157 |
+
try:
|
| 158 |
+
trans = translator.translate(w)
|
| 159 |
+
except Exception:
|
| 160 |
+
continue
|
| 161 |
+
if not trans:
|
| 162 |
+
continue
|
| 163 |
+
if trans.strip().lower() == w.strip().lower():
|
| 164 |
+
continue
|
| 165 |
+
card = {
|
| 166 |
+
"front": w,
|
| 167 |
+
"back": trans,
|
| 168 |
+
"content_type": "ocr_vocab",
|
| 169 |
+
"language": target_lang,
|
| 170 |
+
}
|
| 171 |
+
_ensure_card_stats(card)
|
| 172 |
+
cards.append(card)
|
| 173 |
+
|
| 174 |
+
if not cards:
|
| 175 |
+
raise ValueError("No translatable words found to build cards.")
|
| 176 |
+
|
| 177 |
+
decks_dir = _get_decks_dir(username)
|
| 178 |
+
deck_path = decks_dir / f"{deck_name}.json"
|
| 179 |
+
|
| 180 |
+
deck = {
|
| 181 |
+
"name": deck_name,
|
| 182 |
+
"cards": cards,
|
| 183 |
+
"tags": tags or [],
|
| 184 |
+
}
|
| 185 |
+
save_deck(deck_path, deck)
|
| 186 |
+
return deck_path
|
| 187 |
+
|
| 188 |
+
|
| 189 |
+
# ------------------------------------------------------------
|
| 190 |
+
# Conversation/Text → Flashcards
|
| 191 |
+
# ------------------------------------------------------------
|
| 192 |
+
|
| 193 |
+
def generate_flashcards_from_text(
|
| 194 |
+
username: str,
|
| 195 |
+
text: str,
|
| 196 |
+
deck_name: str = "conversation",
|
| 197 |
+
target_lang: str = "en",
|
| 198 |
+
tags: Optional[List[str]] = None,
|
| 199 |
+
) -> Path:
|
| 200 |
+
"""
|
| 201 |
+
Build a vocab deck from raw conversation text.
|
| 202 |
+
"""
|
| 203 |
+
words = _extract_candidate_words(text)
|
| 204 |
+
if not words:
|
| 205 |
+
raise ValueError("No candidate words found in text.")
|
| 206 |
+
|
| 207 |
+
translator = GoogleTranslator(source="auto", target=target_lang)
|
| 208 |
+
cards = []
|
| 209 |
+
for w in words:
|
| 210 |
+
try:
|
| 211 |
+
trans = translator.translate(w)
|
| 212 |
+
except Exception:
|
| 213 |
+
continue
|
| 214 |
+
if not trans:
|
| 215 |
+
continue
|
| 216 |
+
if trans.strip().lower() == w.strip().lower():
|
| 217 |
+
continue
|
| 218 |
+
card = {
|
| 219 |
+
"front": w,
|
| 220 |
+
"back": trans,
|
| 221 |
+
"content_type": "conversation_vocab",
|
| 222 |
+
"language": target_lang,
|
| 223 |
+
}
|
| 224 |
+
_ensure_card_stats(card)
|
| 225 |
+
cards.append(card)
|
| 226 |
+
|
| 227 |
+
if not cards:
|
| 228 |
+
raise ValueError("No translatable words found to build cards.")
|
| 229 |
+
|
| 230 |
+
decks_dir = _get_decks_dir(username)
|
| 231 |
+
deck_path = decks_dir / f"{deck_name}.json"
|
| 232 |
+
|
| 233 |
+
deck = {
|
| 234 |
+
"name": deck_name,
|
| 235 |
+
"cards": cards,
|
| 236 |
+
"tags": tags or ["conversation"],
|
| 237 |
+
}
|
| 238 |
+
save_deck(deck_path, deck)
|
| 239 |
+
return deck_path
|
| 240 |
+
|
| 241 |
+
|
src/app/main_app.py
ADDED
|
@@ -0,0 +1,1235 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
|
| 2 |
+
###############################################################
|
| 3 |
+
# main_app.py — Agentic Language Partner UI (Streamlit)
|
| 4 |
+
###############################################################
|
| 5 |
+
import pandas as pd
|
| 6 |
+
import json
|
| 7 |
+
import random
|
| 8 |
+
import re
|
| 9 |
+
from datetime import datetime
|
| 10 |
+
from pathlib import Path
|
| 11 |
+
from typing import Dict, List, Any
|
| 12 |
+
|
| 13 |
+
import streamlit as st
|
| 14 |
+
import streamlit.components.v1 as components
|
| 15 |
+
from audiorecorder import audiorecorder
|
| 16 |
+
from deep_translator import GoogleTranslator
|
| 17 |
+
|
| 18 |
+
from app.auth import (
|
| 19 |
+
authenticate_user,
|
| 20 |
+
register_user,
|
| 21 |
+
get_user_prefs,
|
| 22 |
+
update_user_prefs,
|
| 23 |
+
)
|
| 24 |
+
from app.config import get_user_dir
|
| 25 |
+
from app.conversation_core import ConversationManager
|
| 26 |
+
from app.flashcards_tools import (
|
| 27 |
+
list_user_decks,
|
| 28 |
+
load_deck,
|
| 29 |
+
_get_decks_dir,
|
| 30 |
+
save_deck,
|
| 31 |
+
generate_flashcards_from_text,
|
| 32 |
+
generate_flashcards_from_ocr_results,
|
| 33 |
+
)
|
| 34 |
+
from app.ocr_tools import ocr_and_translate_batch
|
| 35 |
+
from app.viewers import generate_flashcard_viewer_for_user
|
| 36 |
+
|
| 37 |
+
|
| 38 |
+
###############################################################
|
| 39 |
+
# PAGE + GLOBAL STYLE
|
| 40 |
+
###############################################################
|
| 41 |
+
|
| 42 |
+
st.set_page_config(
|
| 43 |
+
page_title="Agentic Language Partner",
|
| 44 |
+
layout="wide",
|
| 45 |
+
page_icon="🌐",
|
| 46 |
+
)
|
| 47 |
+
|
| 48 |
+
st.markdown(
|
| 49 |
+
"""
|
| 50 |
+
<style>
|
| 51 |
+
|
| 52 |
+
.chat-column {
|
| 53 |
+
display: flex;
|
| 54 |
+
flex-direction: column;
|
| 55 |
+
}
|
| 56 |
+
|
| 57 |
+
|
| 58 |
+
|
| 59 |
+
/* Input bar at the top */
|
| 60 |
+
.chat-input-bar {
|
| 61 |
+
margin-bottom: 0.5rem;
|
| 62 |
+
background-color: #111;
|
| 63 |
+
padding: 0.75rem 0.5rem 0.5rem;
|
| 64 |
+
border: 1px solid #333;
|
| 65 |
+
border-radius: 0.5rem;
|
| 66 |
+
}
|
| 67 |
+
|
| 68 |
+
/* Scrollable chat messages below input */
|
| 69 |
+
.chat-window {
|
| 70 |
+
max-height: 65vh;
|
| 71 |
+
overflow-y: auto;
|
| 72 |
+
padding-right: .75rem;
|
| 73 |
+
padding-bottom: 0.5rem;
|
| 74 |
+
}
|
| 75 |
+
|
| 76 |
+
/* Chat bubbles */
|
| 77 |
+
.chat-row-user { justify-content:flex-end; display:flex; margin-bottom:.4rem; }
|
| 78 |
+
.chat-row-assistant { justify-content:flex-start; display:flex; margin-bottom:.4rem; }
|
| 79 |
+
|
| 80 |
+
.chat-bubble {
|
| 81 |
+
border-radius:14px;
|
| 82 |
+
padding:.55rem .95rem;
|
| 83 |
+
max-width:80%;
|
| 84 |
+
line-height:1.4;
|
| 85 |
+
box-shadow:0 2px 5px rgba(0,0,0,0.4);
|
| 86 |
+
font-size:1.05rem; /* larger for readability */
|
| 87 |
+
}
|
| 88 |
+
.chat-bubble-user { background:#3a3b3c; color:white; }
|
| 89 |
+
.chat-bubble-assistant { background:#1a73e8; color:white; }
|
| 90 |
+
|
| 91 |
+
.chat-aux {
|
| 92 |
+
font-size:1.0rem; /* larger translation/explanation */
|
| 93 |
+
color:#ccc;
|
| 94 |
+
margin:0.1rem 0.25rem 0.5rem 0.25rem;
|
| 95 |
+
}
|
| 96 |
+
|
| 97 |
+
/* Lock viewport height and avoid infinite page scrolling */
|
| 98 |
+
html, body {
|
| 99 |
+
height: 100%;
|
| 100 |
+
overflow: hidden !important;
|
| 101 |
+
}
|
| 102 |
+
|
| 103 |
+
.block-container {
|
| 104 |
+
height: 100vh !important;
|
| 105 |
+
overflow-y: auto !important;
|
| 106 |
+
}
|
| 107 |
+
|
| 108 |
+
.saved-conv-panel { max-width: 360px; }
|
| 109 |
+
|
| 110 |
+
</style>
|
| 111 |
+
""",
|
| 112 |
+
unsafe_allow_html=True,
|
| 113 |
+
)
|
| 114 |
+
|
| 115 |
+
|
| 116 |
+
###############################################################
|
| 117 |
+
# HELPERS / GLOBALS
|
| 118 |
+
###############################################################
|
| 119 |
+
# ------------------------------------------------------------
|
| 120 |
+
# Model preload / Conversation manager
|
| 121 |
+
# ------------------------------------------------------------
|
| 122 |
+
|
| 123 |
+
def preload_models():
|
| 124 |
+
"""
|
| 125 |
+
Preloads Whisper, Qwen, and NLLB models only one time.
|
| 126 |
+
This prevents UI freeze or repeated loading during conversation.
|
| 127 |
+
"""
|
| 128 |
+
if st.session_state.get("models_loaded"):
|
| 129 |
+
return
|
| 130 |
+
|
| 131 |
+
with st.spinner("Loading language & speech models (one-time)…"):
|
| 132 |
+
try:
|
| 133 |
+
load_whisper()
|
| 134 |
+
except Exception:
|
| 135 |
+
pass
|
| 136 |
+
try:
|
| 137 |
+
load_partner_lm()
|
| 138 |
+
except Exception:
|
| 139 |
+
pass
|
| 140 |
+
try:
|
| 141 |
+
load_nllb()
|
| 142 |
+
except Exception:
|
| 143 |
+
pass
|
| 144 |
+
|
| 145 |
+
st.session_state["models_loaded"] = True
|
| 146 |
+
|
| 147 |
+
|
| 148 |
+
|
| 149 |
+
def get_conv_manager() -> ConversationManager:
|
| 150 |
+
if "conv_manager" not in st.session_state:
|
| 151 |
+
prefs = st.session_state["prefs"]
|
| 152 |
+
st.session_state["conv_manager"] = ConversationManager(
|
| 153 |
+
target_language=prefs.get("target_language", "english"),
|
| 154 |
+
native_language=prefs.get("native_language", "english"),
|
| 155 |
+
cefr_level=prefs.get("cefr_level", "B1"),
|
| 156 |
+
topic=prefs.get("topic", "general conversation"),
|
| 157 |
+
)
|
| 158 |
+
return st.session_state["conv_manager"]
|
| 159 |
+
|
| 160 |
+
|
| 161 |
+
def ensure_default_decks(username: str):
|
| 162 |
+
decks_dir = _get_decks_dir(username)
|
| 163 |
+
|
| 164 |
+
alpha = decks_dir / "alphabet.json"
|
| 165 |
+
if not alpha.exists():
|
| 166 |
+
save_deck(alpha, {
|
| 167 |
+
"name": "Alphabet (A–Z)",
|
| 168 |
+
"cards": [{"front": chr(65+i), "back": f"Letter {chr(65+i)}"} for i in range(26)],
|
| 169 |
+
"tags": ["starter"],
|
| 170 |
+
})
|
| 171 |
+
|
| 172 |
+
nums = decks_dir / "numbers_1_10.json"
|
| 173 |
+
if not nums.exists():
|
| 174 |
+
save_deck(nums, {
|
| 175 |
+
"name": "Numbers 1–10",
|
| 176 |
+
"cards": [{"front": str(i), "back": f"Number {i}"} for i in range(1, 11)],
|
| 177 |
+
"tags": ["starter"],
|
| 178 |
+
})
|
| 179 |
+
|
| 180 |
+
greetings = decks_dir / "greetings_intros.json"
|
| 181 |
+
if not greetings.exists():
|
| 182 |
+
save_deck(greetings, {
|
| 183 |
+
"name": "Greetings & Introductions",
|
| 184 |
+
"cards": [
|
| 185 |
+
{"front": "Hallo!", "back": "Hello!"},
|
| 186 |
+
{"front": "Wie geht's?", "back": "How are you?"},
|
| 187 |
+
{"front": "Ich heiße …", "back": "My name is …"},
|
| 188 |
+
{"front": "Freut mich!", "back": "Nice to meet you!"},
|
| 189 |
+
],
|
| 190 |
+
"tags": ["starter"],
|
| 191 |
+
})
|
| 192 |
+
|
| 193 |
+
|
| 194 |
+
def ui_clean_assistant_text(text: str) -> str:
|
| 195 |
+
if not text:
|
| 196 |
+
return ""
|
| 197 |
+
text = re.sub(r"(?i)\b(user|assistant|system):\s*", "", text)
|
| 198 |
+
text = re.sub(r"\s{2,}", " ", text)
|
| 199 |
+
return text.strip()
|
| 200 |
+
|
| 201 |
+
|
| 202 |
+
def save_current_conversation(username: str, name: str) -> Path:
|
| 203 |
+
"""Save chat_history as JSON, stripping non-serializable fields (audio bytes)."""
|
| 204 |
+
user_dir = get_user_dir(username)
|
| 205 |
+
save_dir = user_dir / "chats" / "saved"
|
| 206 |
+
save_dir.mkdir(parents=True, exist_ok=True)
|
| 207 |
+
|
| 208 |
+
cleaned_messages = []
|
| 209 |
+
for m in st.session_state.get("chat_history", []):
|
| 210 |
+
cleaned_messages.append(
|
| 211 |
+
{
|
| 212 |
+
"role": m.get("role"),
|
| 213 |
+
"text": m.get("text"),
|
| 214 |
+
"explanation": m.get("explanation"),
|
| 215 |
+
# store only a flag for audio, not raw bytes
|
| 216 |
+
"audio_present": bool(m.get("audio")),
|
| 217 |
+
}
|
| 218 |
+
)
|
| 219 |
+
|
| 220 |
+
payload = {
|
| 221 |
+
"name": name,
|
| 222 |
+
"timestamp": datetime.utcnow().isoformat(),
|
| 223 |
+
"messages": cleaned_messages,
|
| 224 |
+
}
|
| 225 |
+
fname = datetime.utcnow().strftime("%Y%m%d_%H%M%S") + ".json"
|
| 226 |
+
path = save_dir / fname
|
| 227 |
+
path.write_text(json.dumps(payload, ensure_ascii=False, indent=2), encoding="utf-8")
|
| 228 |
+
return path
|
| 229 |
+
|
| 230 |
+
|
| 231 |
+
|
| 232 |
+
###############################################################
|
| 233 |
+
# CHAT HANDLING
|
| 234 |
+
###############################################################
|
| 235 |
+
|
| 236 |
+
def handle_user_message(username: str, text: str):
|
| 237 |
+
text = text.strip()
|
| 238 |
+
if not text:
|
| 239 |
+
return
|
| 240 |
+
|
| 241 |
+
conv = get_conv_manager()
|
| 242 |
+
|
| 243 |
+
st.session_state["chat_history"].append(
|
| 244 |
+
{"role": "user", "text": text, "audio": None, "explanation": None}
|
| 245 |
+
)
|
| 246 |
+
|
| 247 |
+
with st.spinner("Thinking…"):
|
| 248 |
+
result = conv.reply(text)
|
| 249 |
+
|
| 250 |
+
reply_text = ui_clean_assistant_text(result.get("reply_text", ""))
|
| 251 |
+
reply_audio = result.get("audio", None)
|
| 252 |
+
explanation = ui_clean_assistant_text(result.get("explanation", ""))
|
| 253 |
+
|
| 254 |
+
st.session_state["chat_history"].append(
|
| 255 |
+
{
|
| 256 |
+
"role": "assistant",
|
| 257 |
+
"text": reply_text,
|
| 258 |
+
"audio": reply_audio,
|
| 259 |
+
"explanation": explanation,
|
| 260 |
+
}
|
| 261 |
+
)
|
| 262 |
+
|
| 263 |
+
|
| 264 |
+
###############################################################
|
| 265 |
+
# AUTH
|
| 266 |
+
###############################################################
|
| 267 |
+
|
| 268 |
+
def login_view():
|
| 269 |
+
st.title("🌐 Agentic Language Partner")
|
| 270 |
+
tab1, tab2 = st.tabs(["Login", "Register"])
|
| 271 |
+
|
| 272 |
+
with tab1:
|
| 273 |
+
u = st.text_input("Username")
|
| 274 |
+
p = st.text_input("Password", type="password")
|
| 275 |
+
if st.button("Login"):
|
| 276 |
+
if authenticate_user(u, p):
|
| 277 |
+
st.session_state["user"] = u
|
| 278 |
+
st.session_state["prefs"] = get_user_prefs(u)
|
| 279 |
+
st.experimental_rerun()
|
| 280 |
+
else:
|
| 281 |
+
st.error("Invalid login.")
|
| 282 |
+
|
| 283 |
+
with tab2:
|
| 284 |
+
u = st.text_input("New username")
|
| 285 |
+
p = st.text_input("New password", type="password")
|
| 286 |
+
if st.button("Register"):
|
| 287 |
+
if register_user(u, p):
|
| 288 |
+
st.success("Registered! Please log in.")
|
| 289 |
+
else:
|
| 290 |
+
st.error("Username already exists.")
|
| 291 |
+
|
| 292 |
+
|
| 293 |
+
###############################################################
|
| 294 |
+
# SIDEBAR SETTINGS
|
| 295 |
+
###############################################################
|
| 296 |
+
|
| 297 |
+
def sidebar_settings(username: str):
|
| 298 |
+
st.sidebar.header("⚙ Settings")
|
| 299 |
+
|
| 300 |
+
prefs = st.session_state["prefs"]
|
| 301 |
+
langs = ["english", "spanish", "german", "russian", "japanese", "chinese", "korean"]
|
| 302 |
+
|
| 303 |
+
tgt = st.sidebar.selectbox(
|
| 304 |
+
"Target language",
|
| 305 |
+
langs,
|
| 306 |
+
index=langs.index(prefs.get("target_language", "english")),
|
| 307 |
+
key="sidebar_target",
|
| 308 |
+
)
|
| 309 |
+
nat = st.sidebar.selectbox(
|
| 310 |
+
"Native language",
|
| 311 |
+
langs,
|
| 312 |
+
index=langs.index(prefs.get("native_language", "english")),
|
| 313 |
+
key="sidebar_native",
|
| 314 |
+
)
|
| 315 |
+
|
| 316 |
+
cefr_levels = ["A1", "A2", "B1", "B2", "C1", "C2"]
|
| 317 |
+
level = st.sidebar.selectbox(
|
| 318 |
+
"CEFR Level",
|
| 319 |
+
cefr_levels,
|
| 320 |
+
index=cefr_levels.index(prefs.get("cefr_level", "B1")),
|
| 321 |
+
key="sidebar_cefr",
|
| 322 |
+
)
|
| 323 |
+
|
| 324 |
+
topic = st.sidebar.text_input(
|
| 325 |
+
"Conversation Topic",
|
| 326 |
+
prefs.get("topic", "general conversation"),
|
| 327 |
+
key="sidebar_topic",
|
| 328 |
+
)
|
| 329 |
+
|
| 330 |
+
show_exp = st.sidebar.checkbox(
|
| 331 |
+
"Show Explanations",
|
| 332 |
+
value=prefs.get("show_explanations", True),
|
| 333 |
+
key="sidebar_show_exp",
|
| 334 |
+
)
|
| 335 |
+
|
| 336 |
+
if st.sidebar.button("Save Settings"):
|
| 337 |
+
new = {
|
| 338 |
+
"target_language": tgt,
|
| 339 |
+
"native_language": nat,
|
| 340 |
+
"cefr_level": level,
|
| 341 |
+
"topic": topic,
|
| 342 |
+
"show_explanations": show_exp,
|
| 343 |
+
}
|
| 344 |
+
st.session_state["prefs"] = new
|
| 345 |
+
update_user_prefs(username, new)
|
| 346 |
+
if "conv_manager" in st.session_state:
|
| 347 |
+
del st.session_state["conv_manager"]
|
| 348 |
+
st.sidebar.success("Settings saved!")
|
| 349 |
+
|
| 350 |
+
|
| 351 |
+
###############################################################
|
| 352 |
+
# DASHBOARD TAB
|
| 353 |
+
###############################################################
|
| 354 |
+
|
| 355 |
+
def dashboard_tab(username: str):
|
| 356 |
+
st.title("Agentic Language Partner — Dashboard")
|
| 357 |
+
|
| 358 |
+
prefs = st.session_state["prefs"]
|
| 359 |
+
langs = ["english", "spanish", "german", "russian", "japanese", "chinese", "korean"]
|
| 360 |
+
|
| 361 |
+
st.subheader("Language Settings")
|
| 362 |
+
|
| 363 |
+
col1, col2, col3 = st.columns(3)
|
| 364 |
+
|
| 365 |
+
with col1:
|
| 366 |
+
native = st.selectbox(
|
| 367 |
+
"Native language",
|
| 368 |
+
langs,
|
| 369 |
+
index=langs.index(prefs.get("native_language", "english")),
|
| 370 |
+
key="dash_native_language",
|
| 371 |
+
)
|
| 372 |
+
with col2:
|
| 373 |
+
target = st.selectbox(
|
| 374 |
+
"Target language",
|
| 375 |
+
langs,
|
| 376 |
+
index=langs.index(prefs.get("target_language", "english")),
|
| 377 |
+
key="dash_target_language",
|
| 378 |
+
)
|
| 379 |
+
with col3:
|
| 380 |
+
cefr_levels = ["A1", "A2", "B1", "B2", "C1", "C2"]
|
| 381 |
+
level = st.selectbox(
|
| 382 |
+
"CEFR Level",
|
| 383 |
+
cefr_levels,
|
| 384 |
+
index=cefr_levels.index(prefs.get("cefr_level", "B1")),
|
| 385 |
+
key="dash_cefr_level",
|
| 386 |
+
)
|
| 387 |
+
|
| 388 |
+
topic = st.text_input(
|
| 389 |
+
"Conversation Topic",
|
| 390 |
+
prefs.get("topic", "general conversation"),
|
| 391 |
+
key="dash_topic",
|
| 392 |
+
)
|
| 393 |
+
|
| 394 |
+
if st.button("Save Language Settings", key="dash_save_lang"):
|
| 395 |
+
new = {
|
| 396 |
+
"native_language": native,
|
| 397 |
+
"target_language": target,
|
| 398 |
+
"cefr_level": level,
|
| 399 |
+
"topic": topic,
|
| 400 |
+
"show_explanations": prefs.get("show_explanations", True),
|
| 401 |
+
}
|
| 402 |
+
st.session_state["prefs"] = new
|
| 403 |
+
update_user_prefs(username, new)
|
| 404 |
+
if "conv_manager" in st.session_state:
|
| 405 |
+
del st.session_state["conv_manager"]
|
| 406 |
+
st.success("Language settings saved!")
|
| 407 |
+
|
| 408 |
+
st.markdown("---")
|
| 409 |
+
|
| 410 |
+
###########################################################
|
| 411 |
+
# MICROPHONE & TRANSCRIPTION CALIBRATION (native phrase)
|
| 412 |
+
###########################################################
|
| 413 |
+
# ---- Microphone & transcription calibration (restored original version) ----
|
| 414 |
+
st.subheader("Microphone & Transcription Calibration")
|
| 415 |
+
|
| 416 |
+
st.write(
|
| 417 |
+
"To verify that audio recording and transcription are working, "
|
| 418 |
+
"please repeat this phrase in your native language:\n\n"
|
| 419 |
+
"> \"Hello, my name is [Your Name], and I am here to practice languages.\"\n\n"
|
| 420 |
+
"Record a short clip and then run transcription to check accuracy."
|
| 421 |
+
)
|
| 422 |
+
|
| 423 |
+
calib_col1, calib_col2 = st.columns([2, 1])
|
| 424 |
+
|
| 425 |
+
with calib_col1:
|
| 426 |
+
calib_audio = audiorecorder(
|
| 427 |
+
"🎤 Start calibration",
|
| 428 |
+
"⏹ Stop",
|
| 429 |
+
key="calibration_audio",
|
| 430 |
+
)
|
| 431 |
+
|
| 432 |
+
if len(calib_audio) > 0:
|
| 433 |
+
st.caption("Calibration audio recorded. Click 'Transcribe sample' to test.")
|
| 434 |
+
|
| 435 |
+
if st.button("Transcribe sample", key="calibration_transcribe"):
|
| 436 |
+
if len(calib_audio) == 0:
|
| 437 |
+
st.warning("Please record a short calibration clip first.")
|
| 438 |
+
else:
|
| 439 |
+
conv = get_conv_manager()
|
| 440 |
+
try:
|
| 441 |
+
seg = calib_audio.set_frame_rate(16000).set_channels(1)
|
| 442 |
+
with st.spinner("Transcribing calibration audio…"):
|
| 443 |
+
text_out, det_lang, det_prob = conv.transcribe(
|
| 444 |
+
seg,
|
| 445 |
+
spoken_lang=st.session_state["prefs"]["native_language"]
|
| 446 |
+
)
|
| 447 |
+
st.session_state["calibration_result"] = {
|
| 448 |
+
"text": text_out,
|
| 449 |
+
"det_lang": det_lang,
|
| 450 |
+
"det_prob": det_prob,
|
| 451 |
+
}
|
| 452 |
+
st.success("Calibration transcript updated.")
|
| 453 |
+
except Exception as e:
|
| 454 |
+
st.error(f"Calibration error: {e}")
|
| 455 |
+
|
| 456 |
+
with calib_col2:
|
| 457 |
+
if st.session_state.get("calibration_result"):
|
| 458 |
+
res = st.session_state["calibration_result"]
|
| 459 |
+
st.markdown("**Calibration transcript:**")
|
| 460 |
+
st.info(res.get("text", ""))
|
| 461 |
+
st.caption(
|
| 462 |
+
f"Detected lang: {res.get('det_lang','?')} · Confidence ~ {res.get('det_prob', 0):.2f}"
|
| 463 |
+
)
|
| 464 |
+
else:
|
| 465 |
+
st.caption("No calibration transcript yet.")
|
| 466 |
+
|
| 467 |
+
st.markdown("---")
|
| 468 |
+
|
| 469 |
+
|
| 470 |
+
|
| 471 |
+
|
| 472 |
+
###########################################################
|
| 473 |
+
# TOOL OVERVIEW
|
| 474 |
+
###########################################################
|
| 475 |
+
st.subheader("Tools Overview")
|
| 476 |
+
|
| 477 |
+
c1, c2, c3 = st.columns(3)
|
| 478 |
+
with c1:
|
| 479 |
+
st.markdown("### 🎙️ Conversation Partner")
|
| 480 |
+
st.write("Real-time language practice with microphone support.")
|
| 481 |
+
|
| 482 |
+
with c2:
|
| 483 |
+
st.markdown("### 🃏 Flashcards & Quizzes")
|
| 484 |
+
st.write("Starter decks: Alphabet, Numbers, Greetings.")
|
| 485 |
+
|
| 486 |
+
with c3:
|
| 487 |
+
st.markdown("### 📷 OCR Helper")
|
| 488 |
+
st.write("Upload images to extract and translate text.")
|
| 489 |
+
|
| 490 |
+
# ------------------------------------------------------------
|
| 491 |
+
# Settings tab (restore missing function)
|
| 492 |
+
# ------------------------------------------------------------
|
| 493 |
+
def settings_tab(username: str):
|
| 494 |
+
"""Minimal settings tab so main() can call it safely."""
|
| 495 |
+
st.header("Settings")
|
| 496 |
+
|
| 497 |
+
st.subheader("User Preferences")
|
| 498 |
+
prefs = st.session_state.get("prefs", {})
|
| 499 |
+
st.json(prefs)
|
| 500 |
+
|
| 501 |
+
st.markdown("---")
|
| 502 |
+
|
| 503 |
+
st.subheader("System Status")
|
| 504 |
+
st.write("Models preloaded:", st.session_state.get("models_loaded", False))
|
| 505 |
+
|
| 506 |
+
st.markdown(
|
| 507 |
+
"This is a placeholder settings panel. "
|
| 508 |
+
"You can customize this later with user-specific configuration."
|
| 509 |
+
)
|
| 510 |
+
|
| 511 |
+
|
| 512 |
+
|
| 513 |
+
###############################################################
|
| 514 |
+
# CONVERSATION TAB
|
| 515 |
+
###############################################################
|
| 516 |
+
|
| 517 |
+
def conversation_tab(username: str):
|
| 518 |
+
import re
|
| 519 |
+
from datetime import datetime
|
| 520 |
+
from deep_translator import GoogleTranslator
|
| 521 |
+
|
| 522 |
+
st.header("Conversation")
|
| 523 |
+
|
| 524 |
+
# ------------------------------------------
|
| 525 |
+
# INITIAL STATE
|
| 526 |
+
# ------------------------------------------
|
| 527 |
+
if "chat_history" not in st.session_state:
|
| 528 |
+
st.session_state["chat_history"] = []
|
| 529 |
+
|
| 530 |
+
if "pending_transcript" not in st.session_state:
|
| 531 |
+
st.session_state["pending_transcript"] = ""
|
| 532 |
+
|
| 533 |
+
if "speech_state" not in st.session_state:
|
| 534 |
+
st.session_state["speech_state"] = "idle" # idle | pending_speech
|
| 535 |
+
|
| 536 |
+
if "recorder_key" not in st.session_state:
|
| 537 |
+
st.session_state["recorder_key"] = 0
|
| 538 |
+
|
| 539 |
+
conv = get_conv_manager()
|
| 540 |
+
prefs = st.session_state.get("prefs", {})
|
| 541 |
+
show_exp = prefs.get("show_explanations", True)
|
| 542 |
+
|
| 543 |
+
# ------------------------------------------
|
| 544 |
+
# RESET BUTTON (ONLY ONE)
|
| 545 |
+
# ------------------------------------------
|
| 546 |
+
if st.button("🔄 Reset Conversation"):
|
| 547 |
+
st.session_state["chat_history"] = []
|
| 548 |
+
st.session_state["pending_transcript"] = ""
|
| 549 |
+
st.session_state["speech_state"] = "idle"
|
| 550 |
+
st.session_state["recorder_key"] += 1
|
| 551 |
+
st.experimental_rerun()
|
| 552 |
+
|
| 553 |
+
|
| 554 |
+
# ------------------------------------------
|
| 555 |
+
# FIRST MESSAGE GREETING
|
| 556 |
+
# ------------------------------------------
|
| 557 |
+
if len(st.session_state["chat_history"]) == 0:
|
| 558 |
+
lang = conv.target_language.lower()
|
| 559 |
+
topic = prefs.get("topic", "").strip()
|
| 560 |
+
|
| 561 |
+
default_greetings = {
|
| 562 |
+
"english": "Hello! I heard you want to practice with me. How is your day going?",
|
| 563 |
+
"german": "Hallo! Ich habe gehört, dass du üben möchtest. Wie geht dein Tag bisher?",
|
| 564 |
+
"spanish": "¡Hola! Escuché que querías practicar conmigo. ¿Cómo va tu día?",
|
| 565 |
+
"japanese":"こんにちは!練習したいと聞きました。今日はどんな一日ですか?",
|
| 566 |
+
}
|
| 567 |
+
|
| 568 |
+
intro = default_greetings.get(lang, default_greetings["english"])
|
| 569 |
+
|
| 570 |
+
if topic and topic.lower() != "general conversation":
|
| 571 |
+
try:
|
| 572 |
+
intro = GoogleTranslator(source="en", target=lang).translate(
|
| 573 |
+
f"Hello! Let's talk about {topic}. What do you think about it?"
|
| 574 |
+
)
|
| 575 |
+
except Exception:
|
| 576 |
+
pass
|
| 577 |
+
|
| 578 |
+
st.session_state["chat_history"].append(
|
| 579 |
+
{"role":"assistant","text":intro,"audio":None,"explanation":None}
|
| 580 |
+
)
|
| 581 |
+
|
| 582 |
+
# ------------------------------------------
|
| 583 |
+
# LAYOUT
|
| 584 |
+
# ------------------------------------------
|
| 585 |
+
col_chat, col_saved = st.columns([3,1])
|
| 586 |
+
|
| 587 |
+
# ===========================
|
| 588 |
+
# LEFT: CHAT WINDOW
|
| 589 |
+
# ===========================
|
| 590 |
+
with col_chat:
|
| 591 |
+
|
| 592 |
+
st.markdown('<div class="chat-window">', unsafe_allow_html=True)
|
| 593 |
+
|
| 594 |
+
for msg in st.session_state["chat_history"]:
|
| 595 |
+
role = msg["role"]
|
| 596 |
+
bubble = "chat-bubble-user" if role == "user" else "chat-bubble-assistant"
|
| 597 |
+
row = "chat-row-user" if role == "user" else "chat-row-assistant"
|
| 598 |
+
|
| 599 |
+
st.markdown(
|
| 600 |
+
f'<div class="{row}"><div class="chat-bubble {bubble}">{msg["text"]}</div></div>',
|
| 601 |
+
unsafe_allow_html=True,
|
| 602 |
+
)
|
| 603 |
+
|
| 604 |
+
if role == "assistant" and msg.get("audio"):
|
| 605 |
+
st.audio(msg["audio"], format="audio/mp3")
|
| 606 |
+
|
| 607 |
+
if role == "assistant":
|
| 608 |
+
try:
|
| 609 |
+
tr = GoogleTranslator(source="auto", target=conv.native_language).translate(msg["text"])
|
| 610 |
+
st.markdown(f'<div class="chat-aux">{tr}</div>', unsafe_allow_html=True)
|
| 611 |
+
except: pass
|
| 612 |
+
|
| 613 |
+
if show_exp and msg.get("explanation"):
|
| 614 |
+
exp = msg["explanation"]
|
| 615 |
+
|
| 616 |
+
# Force EXACTLY ONE sentence
|
| 617 |
+
exp = re.split(r"(?<=[.!?])\s+", exp)[0].strip()
|
| 618 |
+
|
| 619 |
+
# Remove any meta nonsense ("version:", "meaning:", "this sentence", etc)
|
| 620 |
+
exp = re.sub(r"(?i)(english version|the meaning|this sentence|the german sentence).*", "", exp).strip()
|
| 621 |
+
|
| 622 |
+
if exp:
|
| 623 |
+
st.markdown(f'<div class="chat-aux">{exp}</div>', unsafe_allow_html=True)
|
| 624 |
+
|
| 625 |
+
# scroll
|
| 626 |
+
st.markdown("""
|
| 627 |
+
<script>
|
| 628 |
+
setTimeout(() => {
|
| 629 |
+
let w = window.parent.document.getElementsByClassName('chat-window')[0];
|
| 630 |
+
if (w) w.scrollTop = w.scrollHeight;
|
| 631 |
+
}, 200);
|
| 632 |
+
</script>
|
| 633 |
+
""", unsafe_allow_html=True)
|
| 634 |
+
|
| 635 |
+
# -------------------------------
|
| 636 |
+
# AUDIO RECORDER
|
| 637 |
+
# -------------------------------
|
| 638 |
+
st.markdown('<div class="sticky-input-bar">', unsafe_allow_html=True)
|
| 639 |
+
|
| 640 |
+
audio = audiorecorder("🎤 Speak", "⏹ Stop",
|
| 641 |
+
key=f"chat_audio_{st.session_state['recorder_key']}")
|
| 642 |
+
|
| 643 |
+
# ------------------------------------------
|
| 644 |
+
# STATE: idle → record → transcribe
|
| 645 |
+
# ------------------------------------------
|
| 646 |
+
if st.session_state["speech_state"] == "idle":
|
| 647 |
+
if audio and len(audio) > 0:
|
| 648 |
+
seg = audio.set_frame_rate(16000).set_channels(1)
|
| 649 |
+
with st.spinner("Transcribing…"):
|
| 650 |
+
txt, lang, conf = conv.transcribe(seg, spoken_lang=conv.target_language)
|
| 651 |
+
|
| 652 |
+
st.session_state["pending_transcript"] = txt.strip()
|
| 653 |
+
st.session_state["speech_state"] = "pending_speech"
|
| 654 |
+
st.session_state["recorder_key"] += 1
|
| 655 |
+
st.experimental_rerun()
|
| 656 |
+
|
| 657 |
+
# ------------------------------------------
|
| 658 |
+
# STATE: pending_speech → confirm
|
| 659 |
+
# ------------------------------------------
|
| 660 |
+
if st.session_state["speech_state"] == "pending_speech":
|
| 661 |
+
|
| 662 |
+
st.write("### Confirm your spoken message:")
|
| 663 |
+
st.info(st.session_state["pending_transcript"])
|
| 664 |
+
|
| 665 |
+
c1, c2 = st.columns([1,1])
|
| 666 |
+
with c1:
|
| 667 |
+
if st.button("Send message", key="send_pending"):
|
| 668 |
+
txt = st.session_state["pending_transcript"]
|
| 669 |
+
|
| 670 |
+
with st.spinner("Partner is responding…"):
|
| 671 |
+
handle_user_message(username, txt)
|
| 672 |
+
|
| 673 |
+
# cleanup
|
| 674 |
+
st.session_state["speech_state"] = "idle"
|
| 675 |
+
st.session_state["pending_transcript"] = ""
|
| 676 |
+
st.session_state["recorder_key"] += 1
|
| 677 |
+
st.experimental_rerun()
|
| 678 |
+
|
| 679 |
+
with c2:
|
| 680 |
+
if st.button("Discard", key="discard_pending"):
|
| 681 |
+
st.session_state["speech_state"] = "idle"
|
| 682 |
+
st.session_state["pending_transcript"] = ""
|
| 683 |
+
st.session_state["recorder_key"] += 1
|
| 684 |
+
st.experimental_rerun()
|
| 685 |
+
|
| 686 |
+
# -------------------------------
|
| 687 |
+
# TYPED TEXT INPUT
|
| 688 |
+
# -------------------------------
|
| 689 |
+
typed = st.text_input("Type your message:", key="typed_input")
|
| 690 |
+
if typed.strip() and st.button("Send typed message"):
|
| 691 |
+
handle_user_message(username, typed.strip())
|
| 692 |
+
st.session_state["typed_input"] = ""
|
| 693 |
+
st.experimental_rerun()
|
| 694 |
+
|
| 695 |
+
st.markdown("</div>", unsafe_allow_html=True)
|
| 696 |
+
|
| 697 |
+
# ======================================================
|
| 698 |
+
# RIGHT: SAVED CONVERSATIONS (RESTORED)
|
| 699 |
+
# ======================================================
|
| 700 |
+
with col_saved:
|
| 701 |
+
from pathlib import Path
|
| 702 |
+
import json
|
| 703 |
+
|
| 704 |
+
st.markdown("### Saved Conversations")
|
| 705 |
+
|
| 706 |
+
default_name = datetime.utcnow().strftime("Session %Y-%m-%d %H:%M")
|
| 707 |
+
name_box = st.text_input("Name conversation", value=default_name)
|
| 708 |
+
|
| 709 |
+
if st.button("Save conversation"):
|
| 710 |
+
if not st.session_state["chat_history"]:
|
| 711 |
+
st.warning("Nothing to save.")
|
| 712 |
+
else:
|
| 713 |
+
safe = re.sub(r"[^0-9A-Za-z_-]", "_", name_box)
|
| 714 |
+
|
| 715 |
+
path = save_current_conversation(username, safe)
|
| 716 |
+
st.success(f"Saved as {path.name}")
|
| 717 |
+
|
| 718 |
+
saved_dir = get_user_dir(username) / "chats" / "saved"
|
| 719 |
+
saved_dir.mkdir(parents=True, exist_ok=True)
|
| 720 |
+
|
| 721 |
+
files = sorted(saved_dir.glob("*.json"), key=lambda p: p.stat().st_mtime, reverse=True)
|
| 722 |
+
|
| 723 |
+
for f in files:
|
| 724 |
+
data = json.loads(f.read_text())
|
| 725 |
+
sess_name = data.get("name", f.stem)
|
| 726 |
+
msgs = data.get("messages", [])
|
| 727 |
+
|
| 728 |
+
with st.expander(f"{sess_name} ({len(msgs)} msgs)"):
|
| 729 |
+
deck_name = st.text_input(f"Deck name for {sess_name}", value=f"deck_{f.stem}")
|
| 730 |
+
|
| 731 |
+
if st.button(f"Export {f.stem}", key=f"export_{f.stem}"):
|
| 732 |
+
body = "\n".join(m["text"] for m in msgs if m["role"] == "assistant")
|
| 733 |
+
deck_path = generate_flashcards_from_text(
|
| 734 |
+
username=username,
|
| 735 |
+
text=body,
|
| 736 |
+
deck_name=deck_name,
|
| 737 |
+
target_lang=prefs["native_language"],
|
| 738 |
+
tags=["conversation"],
|
| 739 |
+
)
|
| 740 |
+
st.success(f"Deck exported: {deck_path.name}")
|
| 741 |
+
|
| 742 |
+
if st.button(f"Delete {f.stem}", key=f"delete_{f.stem}"):
|
| 743 |
+
f.unlink()
|
| 744 |
+
st.experimental_rerun()
|
| 745 |
+
|
| 746 |
+
###############################################################
|
| 747 |
+
# OCR TAB
|
| 748 |
+
###############################################################
|
| 749 |
+
|
| 750 |
+
def ocr_tab(username: str):
|
| 751 |
+
st.header("OCR → Flashcards")
|
| 752 |
+
|
| 753 |
+
imgs = st.file_uploader("Upload images", ["png", "jpg", "jpeg"], accept_multiple_files=True)
|
| 754 |
+
tgt = st.selectbox("Translate to", ["en", "de", "ja", "zh-cn", "es"])
|
| 755 |
+
deck_name = st.text_input("Deck name", "ocr_vocab")
|
| 756 |
+
|
| 757 |
+
if st.button("Create Deck from OCR"):
|
| 758 |
+
if not imgs:
|
| 759 |
+
st.warning("Upload at least one image.")
|
| 760 |
+
return
|
| 761 |
+
|
| 762 |
+
with st.spinner("Running OCR…"):
|
| 763 |
+
results = ocr_and_translate_batch([f.read() for f in imgs], target_lang=tgt)
|
| 764 |
+
|
| 765 |
+
deck_path = generate_flashcards_from_ocr_results(
|
| 766 |
+
username=username,
|
| 767 |
+
ocr_results=results,
|
| 768 |
+
deck_name=deck_name,
|
| 769 |
+
target_lang=tgt,
|
| 770 |
+
tags=["ocr"],
|
| 771 |
+
)
|
| 772 |
+
st.success(f"Deck saved: {deck_path}")
|
| 773 |
+
|
| 774 |
+
|
| 775 |
+
###############################################################
|
| 776 |
+
# FLASHCARDS TAB
|
| 777 |
+
###############################################################
|
| 778 |
+
|
| 779 |
+
def flashcards_tab(username: str):
|
| 780 |
+
import pandas as pd
|
| 781 |
+
import re
|
| 782 |
+
|
| 783 |
+
# ---------------------------------------------------------
|
| 784 |
+
# Helpers
|
| 785 |
+
# ---------------------------------------------------------
|
| 786 |
+
|
| 787 |
+
def normalize(s: str) -> str:
|
| 788 |
+
"""lowercase + strip non-alphanumerics for loose grading."""
|
| 789 |
+
s = s.lower()
|
| 790 |
+
s = re.sub(r"[^a-z0-9]+", "", s)
|
| 791 |
+
return s
|
| 792 |
+
|
| 793 |
+
def card_front_html(text: str) -> str:
|
| 794 |
+
return f"""
|
| 795 |
+
<div style="
|
| 796 |
+
background:#1a73e8;
|
| 797 |
+
color:white;
|
| 798 |
+
border-radius:18px;
|
| 799 |
+
padding:50px;
|
| 800 |
+
font-size:2.2rem;
|
| 801 |
+
text-align:center;
|
| 802 |
+
width:70%;
|
| 803 |
+
margin-left:auto;
|
| 804 |
+
margin-right:auto;
|
| 805 |
+
box-shadow:0 4px 12px rgba(0,0,0,0.3);
|
| 806 |
+
">
|
| 807 |
+
{text}
|
| 808 |
+
</div>
|
| 809 |
+
"""
|
| 810 |
+
|
| 811 |
+
def card_back_html(front: str, back: str) -> str:
|
| 812 |
+
return f"""
|
| 813 |
+
<div style="margin-bottom:20px;">
|
| 814 |
+
<div style="
|
| 815 |
+
background:#1a73e8;
|
| 816 |
+
color:white;
|
| 817 |
+
border-radius:18px;
|
| 818 |
+
padding:35px;
|
| 819 |
+
font-size:1.8rem;
|
| 820 |
+
text-align:center;
|
| 821 |
+
width:70%;
|
| 822 |
+
margin-left:auto;
|
| 823 |
+
margin-right:auto;
|
| 824 |
+
box-shadow:0 4px 12px rgba(0,0,0,0.25);
|
| 825 |
+
">{front}</div>
|
| 826 |
+
|
| 827 |
+
<div style="
|
| 828 |
+
background:#2b2b2b;
|
| 829 |
+
color:#f5f5f5;
|
| 830 |
+
border-radius:18px;
|
| 831 |
+
padding:40px;
|
| 832 |
+
margin-top:18px;
|
| 833 |
+
font-size:2rem;
|
| 834 |
+
text-align:center;
|
| 835 |
+
width:70%;
|
| 836 |
+
margin-left:auto;
|
| 837 |
+
margin-right:auto;
|
| 838 |
+
box-shadow:0 4px 12px rgba(0,0,0,0.25);
|
| 839 |
+
">{back}</div>
|
| 840 |
+
</div>
|
| 841 |
+
"""
|
| 842 |
+
|
| 843 |
+
# ---------------------------------------------------------
|
| 844 |
+
# Load deck
|
| 845 |
+
# ---------------------------------------------------------
|
| 846 |
+
|
| 847 |
+
st.header("Flashcards")
|
| 848 |
+
|
| 849 |
+
decks = list_user_decks(username)
|
| 850 |
+
if not decks:
|
| 851 |
+
st.info("No decks available yet.")
|
| 852 |
+
return
|
| 853 |
+
|
| 854 |
+
deck_name = st.selectbox("Select deck", sorted(decks.keys()))
|
| 855 |
+
deck_path = decks[deck_name]
|
| 856 |
+
deck = load_deck(deck_path)
|
| 857 |
+
cards = deck.get("cards", [])
|
| 858 |
+
tags = deck.get("tags", [])
|
| 859 |
+
|
| 860 |
+
if not cards:
|
| 861 |
+
st.warning("Deck is empty.")
|
| 862 |
+
return
|
| 863 |
+
|
| 864 |
+
st.write(f"Total cards: **{len(cards)}**")
|
| 865 |
+
if tags:
|
| 866 |
+
st.caption("Tags: " + ", ".join(tags))
|
| 867 |
+
|
| 868 |
+
# Delete deck button
|
| 869 |
+
if st.button("Delete deck"):
|
| 870 |
+
deck_path.unlink()
|
| 871 |
+
st.experimental_rerun()
|
| 872 |
+
|
| 873 |
+
# ---------------------------------------------------------
|
| 874 |
+
# Session state setup
|
| 875 |
+
# ---------------------------------------------------------
|
| 876 |
+
|
| 877 |
+
key = f"fc_{deck_name}_"
|
| 878 |
+
ss = st.session_state
|
| 879 |
+
|
| 880 |
+
if key + "init" not in ss:
|
| 881 |
+
ss[key + "mode"] = "Study"
|
| 882 |
+
ss[key + "idx"] = 0
|
| 883 |
+
ss[key + "show_back"] = False
|
| 884 |
+
|
| 885 |
+
# test state
|
| 886 |
+
ss[key + "test_active"] = False
|
| 887 |
+
ss[key + "test_order"] = []
|
| 888 |
+
ss[key + "test_pos"] = 0
|
| 889 |
+
ss[key + "test_results"] = []
|
| 890 |
+
|
| 891 |
+
ss[key + "init"] = True
|
| 892 |
+
|
| 893 |
+
conv = get_conv_manager()
|
| 894 |
+
|
| 895 |
+
mode = st.radio("Mode", ["Study", "Test"], horizontal=True, key=key + "mode")
|
| 896 |
+
|
| 897 |
+
st.markdown("---")
|
| 898 |
+
|
| 899 |
+
# =======================================================
|
| 900 |
+
# CENTER PANEL
|
| 901 |
+
# =======================================================
|
| 902 |
+
with st.container():
|
| 903 |
+
|
| 904 |
+
# ---------------------------------------------------
|
| 905 |
+
# STUDY MODE
|
| 906 |
+
# ---------------------------------------------------
|
| 907 |
+
if mode == "Study":
|
| 908 |
+
idx = ss[key + "idx"] % len(cards)
|
| 909 |
+
card = cards[idx]
|
| 910 |
+
show_back = ss[key + "show_back"]
|
| 911 |
+
|
| 912 |
+
st.markdown("### Study Mode")
|
| 913 |
+
st.markdown("---")
|
| 914 |
+
|
| 915 |
+
# CARD DISPLAY
|
| 916 |
+
if not show_back:
|
| 917 |
+
st.markdown(card_front_html(card["front"]), unsafe_allow_html=True)
|
| 918 |
+
if st.button("🔊 Pronounce", key=key + f"tts_front_{idx}"):
|
| 919 |
+
audio = conv.text_to_speech(card["front"])
|
| 920 |
+
if audio:
|
| 921 |
+
st.audio(audio, format="audio/mp3")
|
| 922 |
+
else:
|
| 923 |
+
st.markdown(card_back_html(card["front"], card["back"]), unsafe_allow_html=True)
|
| 924 |
+
if st.button("🔊 Pronounce", key=key + f"tts_back_{idx}"):
|
| 925 |
+
audio = conv.text_to_speech(card["back"])
|
| 926 |
+
if audio:
|
| 927 |
+
st.audio(audio, format="audio/mp3")
|
| 928 |
+
|
| 929 |
+
# FLIPBOOK CONTROLS
|
| 930 |
+
st.markdown("---")
|
| 931 |
+
c1, c2, c3 = st.columns(3)
|
| 932 |
+
with c1:
|
| 933 |
+
if st.button("Flip", key=key + "flip"):
|
| 934 |
+
ss[key + "show_back"] = not show_back
|
| 935 |
+
st.experimental_rerun()
|
| 936 |
+
with c2:
|
| 937 |
+
if st.button("Shuffle deck", key=key + "shuf"):
|
| 938 |
+
random.shuffle(cards)
|
| 939 |
+
deck["cards"] = cards
|
| 940 |
+
save_deck(deck_path, deck)
|
| 941 |
+
ss[key + "idx"] = 0
|
| 942 |
+
ss[key + "show_back"] = False
|
| 943 |
+
st.experimental_rerun()
|
| 944 |
+
with c3:
|
| 945 |
+
if st.button("Next →", key=key + "next"):
|
| 946 |
+
ss[key + "idx"] = (idx + 1) % len(cards)
|
| 947 |
+
ss[key + "show_back"] = False
|
| 948 |
+
st.experimental_rerun()
|
| 949 |
+
|
| 950 |
+
# DIFFICULTY GRADING (centered)
|
| 951 |
+
st.markdown("### Rate this card")
|
| 952 |
+
cA, cB, cC, cD, cE = st.columns(5)
|
| 953 |
+
|
| 954 |
+
def apply_grade(delta):
|
| 955 |
+
card["score"] = max(0, card.get("score", 0) + delta)
|
| 956 |
+
card["reviews"] = card.get("reviews", 0) + 1
|
| 957 |
+
save_deck(deck_path, deck)
|
| 958 |
+
ss[key + "idx"] = _choose_next_card_index(cards)
|
| 959 |
+
ss[key + "show_back"] = False
|
| 960 |
+
st.experimental_rerun()
|
| 961 |
+
|
| 962 |
+
with cA:
|
| 963 |
+
if st.button("🔥 Very Difficult", key=key+"g_vd"):
|
| 964 |
+
apply_grade(-2)
|
| 965 |
+
with cB:
|
| 966 |
+
if st.button("😣 Hard", key=key+"g_h"):
|
| 967 |
+
apply_grade(-1)
|
| 968 |
+
with cC:
|
| 969 |
+
if st.button("😐 Neutral", key=key+"g_n"):
|
| 970 |
+
apply_grade(0)
|
| 971 |
+
with cD:
|
| 972 |
+
if st.button("🙂 Easy", key=key+"g_e"):
|
| 973 |
+
apply_grade(1)
|
| 974 |
+
with cE:
|
| 975 |
+
if st.button("🏆 Mastered", key=key+"g_m"):
|
| 976 |
+
apply_grade(3)
|
| 977 |
+
|
| 978 |
+
# ---------------------------------------------------
|
| 979 |
+
# TEST MODE
|
| 980 |
+
# ---------------------------------------------------
|
| 981 |
+
else:
|
| 982 |
+
# Initial test setup
|
| 983 |
+
if not ss[key + "test_active"]:
|
| 984 |
+
st.markdown("### Test Setup")
|
| 985 |
+
num_q = st.slider("Number of questions", 3, min(20, len(cards)), min(5, len(cards)), key=key+"nq")
|
| 986 |
+
|
| 987 |
+
if st.button("Start Test", key=key+"begin"):
|
| 988 |
+
order = list(range(len(cards)))
|
| 989 |
+
random.shuffle(order)
|
| 990 |
+
order = order[:num_q]
|
| 991 |
+
|
| 992 |
+
ss[key + "test_active"] = True
|
| 993 |
+
ss[key + "test_order"] = order
|
| 994 |
+
ss[key + "test_pos"] = 0
|
| 995 |
+
ss[key + "test_results"] = []
|
| 996 |
+
st.experimental_rerun()
|
| 997 |
+
|
| 998 |
+
else:
|
| 999 |
+
order = ss[key + "test_order"]
|
| 1000 |
+
pos = ss[key + "test_pos"]
|
| 1001 |
+
results = ss[key + "test_results"]
|
| 1002 |
+
|
| 1003 |
+
# Test Complete
|
| 1004 |
+
if pos >= len(order):
|
| 1005 |
+
correct = sum(r["correct"] for r in results)
|
| 1006 |
+
st.markdown(f"### Test Complete — Score: {correct}/{len(results)} ({correct/len(results)*100:.1f}%)")
|
| 1007 |
+
st.markdown("---")
|
| 1008 |
+
|
| 1009 |
+
for i, r in enumerate(results, 1):
|
| 1010 |
+
emoji = "✅" if r["correct"] else "❌"
|
| 1011 |
+
st.write(f"**{i}.** {r['front']} → expected **{r['back']}**, you answered *{r['user_answer']}* {emoji}")
|
| 1012 |
+
|
| 1013 |
+
if st.button("Restart Test", key=key+"restart"):
|
| 1014 |
+
ss[key + "test_active"] = False
|
| 1015 |
+
ss[key + "test_pos"] = 0
|
| 1016 |
+
ss[key + "test_results"] = []
|
| 1017 |
+
ss[key + "test_order"] = []
|
| 1018 |
+
st.experimental_rerun()
|
| 1019 |
+
return
|
| 1020 |
+
|
| 1021 |
+
# Current question
|
| 1022 |
+
cid = order[pos]
|
| 1023 |
+
card = cards[cid]
|
| 1024 |
+
|
| 1025 |
+
st.progress(pos / len(order))
|
| 1026 |
+
st.caption(f"Question {pos+1} / {len(order)}")
|
| 1027 |
+
|
| 1028 |
+
st.markdown(card_front_html(card["front"]), unsafe_allow_html=True)
|
| 1029 |
+
|
| 1030 |
+
# TTS
|
| 1031 |
+
if st.button("🔊 Pronounce", key=key+f"tts_test_{pos}"):
|
| 1032 |
+
audio = conv.text_to_speech(card["front"])
|
| 1033 |
+
if audio:
|
| 1034 |
+
st.audio(audio, format="audio/mp3")
|
| 1035 |
+
|
| 1036 |
+
user_answer = st.text_input("Your answer:", key=key+f"ans_{pos}")
|
| 1037 |
+
|
| 1038 |
+
if st.button("Submit Answer", key=key+f"submit_{pos}"):
|
| 1039 |
+
ua = user_answer.strip()
|
| 1040 |
+
correct = normalize(ua) == normalize(card["back"])
|
| 1041 |
+
|
| 1042 |
+
# Flash feedback
|
| 1043 |
+
if correct:
|
| 1044 |
+
st.success("Correct!")
|
| 1045 |
+
else:
|
| 1046 |
+
st.error(f"Incorrect — expected: {card['back']}")
|
| 1047 |
+
|
| 1048 |
+
results.append({
|
| 1049 |
+
"front": card["front"],
|
| 1050 |
+
"back": card["back"],
|
| 1051 |
+
"user_answer": ua,
|
| 1052 |
+
"correct": correct,
|
| 1053 |
+
})
|
| 1054 |
+
ss[key + "test_results"] = results
|
| 1055 |
+
ss[key + "test_pos"] = pos + 1
|
| 1056 |
+
st.experimental_rerun()
|
| 1057 |
+
|
| 1058 |
+
# =======================================================
|
| 1059 |
+
# DECK AT A GLANCE (FULL WIDTH)
|
| 1060 |
+
# =======================================================
|
| 1061 |
+
st.markdown("---")
|
| 1062 |
+
st.subheader("Deck at a glance")
|
| 1063 |
+
|
| 1064 |
+
df_rows = []
|
| 1065 |
+
for i, c in enumerate(cards, start=1):
|
| 1066 |
+
df_rows.append({
|
| 1067 |
+
"#": i,
|
| 1068 |
+
"Front": c.get("front", ""),
|
| 1069 |
+
"Back": c.get("back", ""),
|
| 1070 |
+
"Score": c.get("score", 0),
|
| 1071 |
+
"Reviews": c.get("reviews", 0),
|
| 1072 |
+
})
|
| 1073 |
+
|
| 1074 |
+
st.dataframe(pd.DataFrame(df_rows), height=500, use_container_width=True)
|
| 1075 |
+
|
| 1076 |
+
|
| 1077 |
+
|
| 1078 |
+
|
| 1079 |
+
###############################################################
|
| 1080 |
+
# QUIZ TAB
|
| 1081 |
+
###############################################################
|
| 1082 |
+
|
| 1083 |
+
def quiz_tab(username: str):
|
| 1084 |
+
st.header("Quiz")
|
| 1085 |
+
ensure_default_decks(username)
|
| 1086 |
+
|
| 1087 |
+
user_dir = get_user_dir(username)
|
| 1088 |
+
quiz_dir = user_dir / "quizzes"
|
| 1089 |
+
quiz_dir.mkdir(exist_ok=True)
|
| 1090 |
+
|
| 1091 |
+
decks = list_user_decks(username)
|
| 1092 |
+
if not decks:
|
| 1093 |
+
st.info("No decks.")
|
| 1094 |
+
return
|
| 1095 |
+
|
| 1096 |
+
selected = st.multiselect("Use decks", sorted(decks.keys()))
|
| 1097 |
+
if not selected:
|
| 1098 |
+
return
|
| 1099 |
+
|
| 1100 |
+
num_q = st.slider("Questions", 3, 20, 6)
|
| 1101 |
+
|
| 1102 |
+
if st.button("Generate quiz"):
|
| 1103 |
+
pool = []
|
| 1104 |
+
for name in selected:
|
| 1105 |
+
pool.extend(load_deck(decks[name])["cards"])
|
| 1106 |
+
|
| 1107 |
+
questions = []
|
| 1108 |
+
for _ in range(num_q):
|
| 1109 |
+
c = random.choice(pool)
|
| 1110 |
+
qtype = random.choice(["mc", "fill"])
|
| 1111 |
+
if qtype == "mc":
|
| 1112 |
+
others = random.sample(pool, min(3, len(pool) - 1))
|
| 1113 |
+
opts = [c["back"]] + [x["back"] for x in others]
|
| 1114 |
+
random.shuffle(opts)
|
| 1115 |
+
questions.append(
|
| 1116 |
+
{"type": "mc", "prompt": c["front"], "options": opts, "answer": c["back"]}
|
| 1117 |
+
)
|
| 1118 |
+
else:
|
| 1119 |
+
questions.append(
|
| 1120 |
+
{"type": "fill", "prompt": c["front"], "answer": c["back"]}
|
| 1121 |
+
)
|
| 1122 |
+
|
| 1123 |
+
qid = datetime.utcnow().strftime("quiz_%Y%m%d_%H%M%S")
|
| 1124 |
+
quiz = {"id": qid, "questions": questions}
|
| 1125 |
+
(quiz_dir / f"{qid}.json").write_text(json.dumps(quiz, indent=2))
|
| 1126 |
+
|
| 1127 |
+
st.session_state["quiz"] = quiz
|
| 1128 |
+
st.session_state["quiz_idx"] = 0
|
| 1129 |
+
st.session_state["quiz_answers"] = {}
|
| 1130 |
+
st.success("Quiz created!")
|
| 1131 |
+
|
| 1132 |
+
if "quiz" not in st.session_state:
|
| 1133 |
+
return
|
| 1134 |
+
|
| 1135 |
+
quiz = st.session_state["quiz"]
|
| 1136 |
+
qs = quiz["questions"]
|
| 1137 |
+
idx = st.session_state["quiz_idx"]
|
| 1138 |
+
|
| 1139 |
+
if idx >= len(qs):
|
| 1140 |
+
correct = sum(1 for v in st.session_state["quiz_answers"].values() if v["correct"])
|
| 1141 |
+
st.success(f"Score: {correct}/{len(qs)}")
|
| 1142 |
+
if st.button("New quiz"):
|
| 1143 |
+
del st.session_state["quiz"]
|
| 1144 |
+
del st.session_state["quiz_idx"]
|
| 1145 |
+
del st.session_state["quiz_answers"]
|
| 1146 |
+
return
|
| 1147 |
+
|
| 1148 |
+
q = qs[idx]
|
| 1149 |
+
st.subheader(f"Question {idx+1}/{len(qs)}")
|
| 1150 |
+
st.markdown(f"**{q['prompt']}**")
|
| 1151 |
+
|
| 1152 |
+
if q["type"] == "mc":
|
| 1153 |
+
choice = st.radio("Choose:", q["options"], key=f"mc_{idx}")
|
| 1154 |
+
if st.button("Submit", key=f"sub_{idx}"):
|
| 1155 |
+
st.session_state["quiz_answers"][idx] = {
|
| 1156 |
+
"given": choice,
|
| 1157 |
+
"correct": choice == q["answer"],
|
| 1158 |
+
}
|
| 1159 |
+
st.session_state["quiz_idx"] += 1
|
| 1160 |
+
st.experimental_rerun()
|
| 1161 |
+
else:
|
| 1162 |
+
ans = st.text_input("Your answer", key=f"fill_{idx}")
|
| 1163 |
+
if st.button("Submit", key=f"sub_{idx}"):
|
| 1164 |
+
st.session_state["quiz_answers"][idx] = {
|
| 1165 |
+
"given": ans,
|
| 1166 |
+
"correct": ans.strip().lower() == q["answer"].lower(),
|
| 1167 |
+
}
|
| 1168 |
+
st.session_state["quiz_idx"] += 1
|
| 1169 |
+
st.experimental_rerun()
|
| 1170 |
+
|
| 1171 |
+
|
| 1172 |
+
###############################################################
|
| 1173 |
+
# MAIN
|
| 1174 |
+
###############################################################
|
| 1175 |
+
|
| 1176 |
+
def main():
|
| 1177 |
+
# ---------- AUTH ----------
|
| 1178 |
+
if "user" not in st.session_state:
|
| 1179 |
+
login_view()
|
| 1180 |
+
return
|
| 1181 |
+
|
| 1182 |
+
username = st.session_state["user"]
|
| 1183 |
+
st.sidebar.write(f"Logged in as **{username}**")
|
| 1184 |
+
|
| 1185 |
+
if st.sidebar.button("Log out"):
|
| 1186 |
+
st.session_state.clear()
|
| 1187 |
+
st.experimental_rerun()
|
| 1188 |
+
|
| 1189 |
+
# ---------- LOAD MODELS + PREFS ----------
|
| 1190 |
+
preload_models()
|
| 1191 |
+
sidebar_settings(username)
|
| 1192 |
+
ensure_default_decks(username)
|
| 1193 |
+
|
| 1194 |
+
# ---------- TAB PERSISTENCE ----------
|
| 1195 |
+
if "active_tab" not in st.session_state:
|
| 1196 |
+
st.session_state["active_tab"] = 0
|
| 1197 |
+
|
| 1198 |
+
tab_labels = ["Dashboard", "Conversation", "OCR", "Flashcards", "Quiz", "Settings"]
|
| 1199 |
+
tabs = st.tabs(tab_labels)
|
| 1200 |
+
|
| 1201 |
+
tab_dash, tab_conv, tab_ocr, tab_flash, tab_quiz, tab_settings = tabs
|
| 1202 |
+
|
| 1203 |
+
# restore active tab (required so Streamlit shows correct tab on rerun)
|
| 1204 |
+
_ = tabs[st.session_state["active_tab"]]
|
| 1205 |
+
|
| 1206 |
+
with tab_dash:
|
| 1207 |
+
st.session_state["active_tab"] = 0
|
| 1208 |
+
dashboard_tab(username)
|
| 1209 |
+
|
| 1210 |
+
with tab_conv:
|
| 1211 |
+
st.session_state["active_tab"] = 1
|
| 1212 |
+
conversation_tab(username)
|
| 1213 |
+
|
| 1214 |
+
with tab_ocr:
|
| 1215 |
+
st.session_state["active_tab"] = 2
|
| 1216 |
+
ocr_tab(username)
|
| 1217 |
+
|
| 1218 |
+
with tab_flash:
|
| 1219 |
+
st.session_state["active_tab"] = 3
|
| 1220 |
+
flashcards_tab(username)
|
| 1221 |
+
|
| 1222 |
+
with tab_quiz:
|
| 1223 |
+
st.session_state["active_tab"] = 4
|
| 1224 |
+
quiz_tab(username)
|
| 1225 |
+
|
| 1226 |
+
with tab_settings:
|
| 1227 |
+
st.session_state["active_tab"] = 5
|
| 1228 |
+
settings_tab(username)
|
| 1229 |
+
|
| 1230 |
+
|
| 1231 |
+
|
| 1232 |
+
if __name__ == "__main__":
|
| 1233 |
+
main()
|
| 1234 |
+
|
| 1235 |
+
|
src/app/ocr_tools.py
ADDED
|
@@ -0,0 +1,64 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
|
| 2 |
+
# src/app/ocr_tools.py
|
| 3 |
+
|
| 4 |
+
import io
|
| 5 |
+
from typing import Any, Dict, List, Optional
|
| 6 |
+
|
| 7 |
+
from PIL import Image
|
| 8 |
+
import pytesseract
|
| 9 |
+
from deep_translator import GoogleTranslator
|
| 10 |
+
|
| 11 |
+
|
| 12 |
+
def _simple_ocr(image_bytes: bytes) -> str:
|
| 13 |
+
"""
|
| 14 |
+
Fallback OCR using pytesseract.
|
| 15 |
+
"""
|
| 16 |
+
img = Image.open(io.BytesIO(image_bytes)).convert("RGB")
|
| 17 |
+
text = pytesseract.image_to_string(img)
|
| 18 |
+
return text.strip()
|
| 19 |
+
|
| 20 |
+
|
| 21 |
+
def ocr_and_translate_batch(
|
| 22 |
+
images: List[bytes],
|
| 23 |
+
target_lang: str = "en",
|
| 24 |
+
prefer_ocr_local: bool = True,
|
| 25 |
+
) -> List[Dict]:
|
| 26 |
+
"""
|
| 27 |
+
Runs OCR on a batch of images. For now, we always use the
|
| 28 |
+
simple pytesseract-based OCR, but the 'prefer_ocr_local'
|
| 29 |
+
flag is kept for compatibility with previous versions that
|
| 30 |
+
used a local PaddleOCR pipeline.
|
| 31 |
+
|
| 32 |
+
Returns: list of dicts with keys:
|
| 33 |
+
- "text": original OCR text
|
| 34 |
+
- "translation": translation into target_lang
|
| 35 |
+
- "target_lang": target_lang
|
| 36 |
+
"""
|
| 37 |
+
translator = GoogleTranslator(source="auto", target=target_lang)
|
| 38 |
+
|
| 39 |
+
results: List[Dict] = []
|
| 40 |
+
for img_bytes in images:
|
| 41 |
+
text = _simple_ocr(img_bytes)
|
| 42 |
+
if text:
|
| 43 |
+
try:
|
| 44 |
+
translated = translator.translate(text)
|
| 45 |
+
except Exception:
|
| 46 |
+
translated = ""
|
| 47 |
+
|
| 48 |
+
results.append(
|
| 49 |
+
{
|
| 50 |
+
"text": text,
|
| 51 |
+
"translation": translated,
|
| 52 |
+
"target_lang": target_lang,
|
| 53 |
+
}
|
| 54 |
+
)
|
| 55 |
+
else:
|
| 56 |
+
results.append(
|
| 57 |
+
{
|
| 58 |
+
"text": "",
|
| 59 |
+
"translation": "",
|
| 60 |
+
"target_lang": target_lang,
|
| 61 |
+
}
|
| 62 |
+
)
|
| 63 |
+
return results
|
| 64 |
+
|
src/app/quiz_tools.py
ADDED
|
@@ -0,0 +1,59 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
|
| 2 |
+
# src/app/quiz_tools.py
|
| 3 |
+
|
| 4 |
+
# Placeholder restored because modifications moved to main_app.
|
| 5 |
+
# This keeps the file present so import does not fail.
|
| 6 |
+
|
| 7 |
+
import json
|
| 8 |
+
import random
|
| 9 |
+
from datetime import datetime
|
| 10 |
+
|
| 11 |
+
def create_semantic_quiz_for_user(username: str, topic: str, num_questions: int = 5):
|
| 12 |
+
reading_passages = [
|
| 13 |
+
f"{topic.capitalize()} is important in daily life. Many people enjoy talking about it.",
|
| 14 |
+
f"Here is a short story based on the topic '{topic}'.",
|
| 15 |
+
f"In this short description, you will learn more about {topic}.",
|
| 16 |
+
]
|
| 17 |
+
|
| 18 |
+
questions = []
|
| 19 |
+
for i in range(num_questions):
|
| 20 |
+
passage = random.choice(reading_passages)
|
| 21 |
+
q_type = random.choice(["translate_phrase", "summarize", "interpret"])
|
| 22 |
+
|
| 23 |
+
if q_type == "translate_phrase":
|
| 24 |
+
questions.append({
|
| 25 |
+
"type": "semantic_translate_phrase",
|
| 26 |
+
"prompt": f"Translate:
|
| 27 |
+
|
| 28 |
+
'{passage}'",
|
| 29 |
+
"answer": "(model evaluated)",
|
| 30 |
+
"explanation": f"Checks ability to translate topic '{topic}'."
|
| 31 |
+
})
|
| 32 |
+
elif q_type == "summarize":
|
| 33 |
+
questions.append({
|
| 34 |
+
"type": "semantic_summarize",
|
| 35 |
+
"prompt": f"Summarize:
|
| 36 |
+
|
| 37 |
+
{passage}",
|
| 38 |
+
"answer": "(model evaluated)",
|
| 39 |
+
"explanation": f"Checks comprehension of topic '{topic}'."
|
| 40 |
+
})
|
| 41 |
+
elif q_type == "interpret":
|
| 42 |
+
questions.append({
|
| 43 |
+
"type": "semantic_interpret",
|
| 44 |
+
"prompt": f"Interpret meaning:
|
| 45 |
+
|
| 46 |
+
{passage}",
|
| 47 |
+
"answer": "(model evaluated)",
|
| 48 |
+
"explanation": f"Checks conceptual understanding of '{topic}'."
|
| 49 |
+
})
|
| 50 |
+
|
| 51 |
+
ts = datetime.utcnow().strftime("%Y-%m-%dT%H-%M-%SZ")
|
| 52 |
+
quiz_id = f"semantic_quiz_{ts}"
|
| 53 |
+
|
| 54 |
+
return {
|
| 55 |
+
"id": quiz_id,
|
| 56 |
+
"created_at": ts,
|
| 57 |
+
"topic": topic,
|
| 58 |
+
"questions": questions,
|
| 59 |
+
}
|
src/app/viewers.py
ADDED
|
@@ -0,0 +1,272 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
|
| 2 |
+
# src/app/viewers.py
|
| 3 |
+
|
| 4 |
+
import json
|
| 5 |
+
from pathlib import Path
|
| 6 |
+
from typing import Dict, List
|
| 7 |
+
|
| 8 |
+
from app.config import get_user_dir
|
| 9 |
+
from app.flashcards_tools import load_deck
|
| 10 |
+
|
| 11 |
+
|
| 12 |
+
def _build_flipbook_html(deck_name: str, cards: List[Dict]) -> str:
|
| 13 |
+
"""
|
| 14 |
+
Builds a simple HTML+JS flip-style viewer for a deck of cards.
|
| 15 |
+
Front = card['front'], Back = card['back'].
|
| 16 |
+
"""
|
| 17 |
+
js_cards = json.dumps(
|
| 18 |
+
[
|
| 19 |
+
{"front": c.get("front", ""), "back": c.get("back", "")}
|
| 20 |
+
for c in cards
|
| 21 |
+
],
|
| 22 |
+
ensure_ascii=False,
|
| 23 |
+
)
|
| 24 |
+
|
| 25 |
+
html = f"""<!DOCTYPE html>
|
| 26 |
+
<html lang="en">
|
| 27 |
+
<head>
|
| 28 |
+
<meta charset="UTF-8" />
|
| 29 |
+
<title>Flashcards — {deck_name}</title>
|
| 30 |
+
<style>
|
| 31 |
+
body {{
|
| 32 |
+
background: #111;
|
| 33 |
+
color: #eee;
|
| 34 |
+
font-family: system-ui, -apple-system, BlinkMacSystemFont, "Segoe UI", sans-serif;
|
| 35 |
+
padding: 1.5rem;
|
| 36 |
+
}}
|
| 37 |
+
|
| 38 |
+
.wrapper {{
|
| 39 |
+
max-width: 700px;
|
| 40 |
+
margin: 0 auto;
|
| 41 |
+
}}
|
| 42 |
+
|
| 43 |
+
h1 {{
|
| 44 |
+
text-align: center;
|
| 45 |
+
margin-bottom: 1rem;
|
| 46 |
+
}}
|
| 47 |
+
|
| 48 |
+
.card-container {{
|
| 49 |
+
perspective: 1000px;
|
| 50 |
+
margin-bottom: 1rem;
|
| 51 |
+
}}
|
| 52 |
+
|
| 53 |
+
.card {{
|
| 54 |
+
position: relative;
|
| 55 |
+
width: 100%;
|
| 56 |
+
height: 250px;
|
| 57 |
+
border-radius: 16px;
|
| 58 |
+
background: #222;
|
| 59 |
+
box-shadow: 0 8px 15px rgba(0,0,0,0.4);
|
| 60 |
+
transition: transform 0.6s;
|
| 61 |
+
transform-style: preserve-3d;
|
| 62 |
+
cursor: pointer;
|
| 63 |
+
}}
|
| 64 |
+
|
| 65 |
+
.card.flipped {{
|
| 66 |
+
transform: rotateY(180deg);
|
| 67 |
+
}}
|
| 68 |
+
|
| 69 |
+
.card-face {{
|
| 70 |
+
position: absolute;
|
| 71 |
+
inset: 0;
|
| 72 |
+
border-radius: 16px;
|
| 73 |
+
backface-visibility: hidden;
|
| 74 |
+
display: flex;
|
| 75 |
+
align-items: center;
|
| 76 |
+
justify-content: center;
|
| 77 |
+
padding: 1rem;
|
| 78 |
+
}}
|
| 79 |
+
|
| 80 |
+
.card-face.front {{
|
| 81 |
+
background: #333;
|
| 82 |
+
}}
|
| 83 |
+
|
| 84 |
+
.card-face.back {{
|
| 85 |
+
background: #1a73e8;
|
| 86 |
+
transform: rotateY(180deg);
|
| 87 |
+
}}
|
| 88 |
+
|
| 89 |
+
.card-text {{
|
| 90 |
+
font-size: 2.2rem;
|
| 91 |
+
text-align: center;
|
| 92 |
+
word-wrap: break-word;
|
| 93 |
+
}}
|
| 94 |
+
|
| 95 |
+
.controls {{
|
| 96 |
+
display: flex;
|
| 97 |
+
justify-content: center;
|
| 98 |
+
gap: 0.75rem;
|
| 99 |
+
margin-bottom: 0.5rem;
|
| 100 |
+
}}
|
| 101 |
+
|
| 102 |
+
button {{
|
| 103 |
+
background: #333;
|
| 104 |
+
color: #eee;
|
| 105 |
+
border-radius: 999px;
|
| 106 |
+
border: none;
|
| 107 |
+
padding: 0.5rem 1rem;
|
| 108 |
+
font-size: 0.95rem;
|
| 109 |
+
cursor: pointer;
|
| 110 |
+
transition: background 0.2s ease, transform 0.1s ease;
|
| 111 |
+
}}
|
| 112 |
+
|
| 113 |
+
button:hover {{
|
| 114 |
+
background: #444;
|
| 115 |
+
transform: translateY(-1px);
|
| 116 |
+
}}
|
| 117 |
+
|
| 118 |
+
.meta {{
|
| 119 |
+
text-align: center;
|
| 120 |
+
margin-top: 0.25rem;
|
| 121 |
+
font-size: 0.9rem;
|
| 122 |
+
color: #ccc;
|
| 123 |
+
}}
|
| 124 |
+
|
| 125 |
+
.badge {{
|
| 126 |
+
display: inline-block;
|
| 127 |
+
padding: 0.1rem 0.75rem;
|
| 128 |
+
border-radius: 999px;
|
| 129 |
+
font-size: 0.8rem;
|
| 130 |
+
background: #555;
|
| 131 |
+
margin-left: 0.5rem;
|
| 132 |
+
}}
|
| 133 |
+
</style>
|
| 134 |
+
</head>
|
| 135 |
+
<body>
|
| 136 |
+
<div class="wrapper">
|
| 137 |
+
<h1>{deck_name}</h1>
|
| 138 |
+
<div class="card-container">
|
| 139 |
+
<div class="card" id="card">
|
| 140 |
+
<div class="card-face front">
|
| 141 |
+
<div class="card-text" id="cardFront"></div>
|
| 142 |
+
</div>
|
| 143 |
+
<div class="card-face back">
|
| 144 |
+
<div class="card-text" id="cardBack"></div>
|
| 145 |
+
</div>
|
| 146 |
+
</div>
|
| 147 |
+
</div>
|
| 148 |
+
|
| 149 |
+
<div class="controls">
|
| 150 |
+
<button id="prevBtn">⏮ Prev</button>
|
| 151 |
+
<button id="flipBtn">🔁 Flip</button>
|
| 152 |
+
<button id="nextBtn">Next ⏭</button>
|
| 153 |
+
<button id="shuffleBtn">🔀 Shuffle</button>
|
| 154 |
+
</div>
|
| 155 |
+
|
| 156 |
+
<div class="meta">
|
| 157 |
+
<span id="cardIndex">Card 0 / 0</span>
|
| 158 |
+
<span class="badge" id="sideLabel">Front</span>
|
| 159 |
+
</div>
|
| 160 |
+
</div>
|
| 161 |
+
|
| 162 |
+
<script>
|
| 163 |
+
const cards = {js_cards};
|
| 164 |
+
let currentIndex = 0;
|
| 165 |
+
let isFlipped = false;
|
| 166 |
+
|
| 167 |
+
const cardEl = document.getElementById('card');
|
| 168 |
+
const frontEl = document.getElementById('cardFront');
|
| 169 |
+
const backEl = document.getElementById('cardBack');
|
| 170 |
+
const indexEl = document.getElementById('cardIndex');
|
| 171 |
+
const sideLabelEl = document.getElementById('sideLabel');
|
| 172 |
+
|
| 173 |
+
function renderCard() {{
|
| 174 |
+
if (!cards.length) {{
|
| 175 |
+
frontEl.textContent = "(No cards)";
|
| 176 |
+
backEl.textContent = "";
|
| 177 |
+
indexEl.textContent = "Card 0 / 0";
|
| 178 |
+
sideLabelEl.textContent = "Front";
|
| 179 |
+
cardEl.classList.remove('flipped');
|
| 180 |
+
return;
|
| 181 |
+
}}
|
| 182 |
+
|
| 183 |
+
const card = cards[currentIndex];
|
| 184 |
+
frontEl.textContent = card.front || "";
|
| 185 |
+
backEl.textContent = card.back || "";
|
| 186 |
+
|
| 187 |
+
indexEl.textContent = "Card " + (currentIndex + 1) + " / " + cards.length;
|
| 188 |
+
sideLabelEl.textContent = isFlipped ? "Back" : "Front";
|
| 189 |
+
}}
|
| 190 |
+
|
| 191 |
+
function flipCard() {{
|
| 192 |
+
if (!cards.length) return;
|
| 193 |
+
isFlipped = !isFlipped;
|
| 194 |
+
if (isFlipped) {{
|
| 195 |
+
cardEl.classList.add('flipped');
|
| 196 |
+
}} else {{
|
| 197 |
+
cardEl.classList.remove('flipped');
|
| 198 |
+
}}
|
| 199 |
+
sideLabelEl.textContent = isFlipped ? "Back" : "Front";
|
| 200 |
+
}}
|
| 201 |
+
|
| 202 |
+
function nextCard() {{
|
| 203 |
+
if (!cards.length) return;
|
| 204 |
+
currentIndex = (currentIndex + 1) % cards.length;
|
| 205 |
+
isFlipped = false;
|
| 206 |
+
cardEl.classList.remove('flipped');
|
| 207 |
+
renderCard();
|
| 208 |
+
}}
|
| 209 |
+
|
| 210 |
+
function prevCard() {{
|
| 211 |
+
if (!cards.length) return;
|
| 212 |
+
currentIndex = (currentIndex - 1 + cards.length) % cards.length;
|
| 213 |
+
isFlipped = false;
|
| 214 |
+
cardEl.classList.remove('flipped');
|
| 215 |
+
renderCard();
|
| 216 |
+
}}
|
| 217 |
+
|
| 218 |
+
function shuffleCards() {{
|
| 219 |
+
for (let i = cards.length - 1; i > 0; i--) {{
|
| 220 |
+
const j = Math.floor(Math.random() * (i + 1));
|
| 221 |
+
[cards[i], cards[j]] = [cards[j], cards[i]];
|
| 222 |
+
}}
|
| 223 |
+
currentIndex = 0;
|
| 224 |
+
isFlipped = false;
|
| 225 |
+
cardEl.classList.remove('flipped');
|
| 226 |
+
renderCard();
|
| 227 |
+
}}
|
| 228 |
+
|
| 229 |
+
cardEl.addEventListener('click', flipCard);
|
| 230 |
+
document.getElementById('flipBtn').addEventListener('click', flipCard);
|
| 231 |
+
document.getElementById('nextBtn').addEventListener('click', nextCard);
|
| 232 |
+
document.getElementById('prevBtn').addEventListener('click', prevCard);
|
| 233 |
+
document.getElementById('shuffleBtn').addEventListener('click', shuffleCards);
|
| 234 |
+
|
| 235 |
+
document.addEventListener('keydown', (e) => {{
|
| 236 |
+
if (!cards.length) return;
|
| 237 |
+
if (e.code === "ArrowRight") nextCard();
|
| 238 |
+
else if (e.code === "ArrowLeft") prevCard();
|
| 239 |
+
else if (e.code === "Space") {{
|
| 240 |
+
e.preventDefault();
|
| 241 |
+
flipCard();
|
| 242 |
+
}}
|
| 243 |
+
}});
|
| 244 |
+
|
| 245 |
+
renderCard();
|
| 246 |
+
</script>
|
| 247 |
+
</body>
|
| 248 |
+
</html>
|
| 249 |
+
"""
|
| 250 |
+
return html
|
| 251 |
+
|
| 252 |
+
|
| 253 |
+
def generate_flashcard_viewer_for_user(username: str, deck_path: Path) -> Path:
|
| 254 |
+
"""
|
| 255 |
+
Generates an HTML flipbook viewer for the given deck in the user's
|
| 256 |
+
/viewers directory, and returns the path to the HTML file.
|
| 257 |
+
"""
|
| 258 |
+
deck = load_deck(deck_path)
|
| 259 |
+
deck_name = deck.get("name", deck_path.stem)
|
| 260 |
+
cards = deck.get("cards", [])
|
| 261 |
+
|
| 262 |
+
html_str = _build_flipbook_html(deck_name, cards)
|
| 263 |
+
|
| 264 |
+
user_dir = get_user_dir(username)
|
| 265 |
+
viewer_dir = user_dir / "viewers"
|
| 266 |
+
viewer_dir.mkdir(parents=True, exist_ok=True)
|
| 267 |
+
|
| 268 |
+
safe_name = deck_path.stem
|
| 269 |
+
out_path = viewer_dir / f"{safe_name}_viewer.html"
|
| 270 |
+
out_path.write_text(html_str, encoding="utf-8")
|
| 271 |
+
return out_path
|
| 272 |
+
|
src/generate_flashcard_viewer.py
ADDED
|
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
|
| 2 |
+
# src/generate_flashcard_viewer.py
|
| 3 |
+
|
| 4 |
+
"""
|
| 5 |
+
Utility script to generate a standalone HTML viewer for a given
|
| 6 |
+
user's flashcard deck. Typically, the Streamlit app calls
|
| 7 |
+
app.viewers.generate_flashcard_viewer_for_user directly.
|
| 8 |
+
"""
|
| 9 |
+
|
| 10 |
+
import argparse
|
| 11 |
+
from pathlib import Path
|
| 12 |
+
|
| 13 |
+
from app.viewers import generate_flashcard_viewer_for_user
|
| 14 |
+
|
| 15 |
+
|
| 16 |
+
def main():
|
| 17 |
+
parser = argparse.ArgumentParser()
|
| 18 |
+
parser.add_argument("--user", required=True, help="Username")
|
| 19 |
+
parser.add_argument("--deck", required=True, help="Path to deck JSON")
|
| 20 |
+
args = parser.parse_args()
|
| 21 |
+
|
| 22 |
+
deck_path = Path(args.deck)
|
| 23 |
+
if not deck_path.exists():
|
| 24 |
+
raise FileNotFoundError(deck_path)
|
| 25 |
+
|
| 26 |
+
out_path = generate_flashcard_viewer_for_user(args.user, deck_path)
|
| 27 |
+
print(f"Viewer written to: {out_path}")
|
| 28 |
+
|
| 29 |
+
|
| 30 |
+
if __name__ == "__main__":
|
| 31 |
+
main()
|
| 32 |
+
|
src/generate_quiz.py
ADDED
|
@@ -0,0 +1,2 @@
|
|
|
|
|
|
|
|
|
|
| 1 |
+
|
| 2 |
+
print("Deprecated — use quiz inside Streamlit UI")
|
src/generate_quiz_viewer.py
ADDED
|
@@ -0,0 +1,2 @@
|
|
|
|
|
|
|
|
|
|
| 1 |
+
|
| 2 |
+
print("Viewer now handled by app.viewers")
|