Upload 2 files
Browse files- app.py +215 -50
- database.py +156 -95
app.py
CHANGED
|
@@ -10,14 +10,33 @@ Two-step flow:
|
|
| 10 |
summarizes the likely symptoms or intent for the provider.
|
| 11 |
2. Provider submits a reply in English. The system translates it back
|
| 12 |
into the patient's language and generates spoken audio for it.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 13 |
"""
|
| 14 |
|
| 15 |
import os
|
| 16 |
import uuid
|
| 17 |
-
|
|
|
|
|
|
|
| 18 |
|
| 19 |
from dotenv import load_dotenv
|
| 20 |
-
from flask import Flask, request, jsonify, render_template, session
|
| 21 |
|
| 22 |
import database
|
| 23 |
import stt
|
|
@@ -28,6 +47,7 @@ import tts
|
|
| 28 |
load_dotenv("env")
|
| 29 |
|
| 30 |
AUDIO_DIR = os.path.join(os.path.dirname(os.path.abspath(__file__)), "static", "audio")
|
|
|
|
| 31 |
|
| 32 |
|
| 33 |
def create_app():
|
|
@@ -48,6 +68,100 @@ database.init_db()
|
|
| 48 |
tts.preload_models()
|
| 49 |
|
| 50 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 51 |
def get_current_user_id(language):
|
| 52 |
"""
|
| 53 |
Returns the user_id for the current browser session, creating a new
|
|
@@ -67,16 +181,28 @@ def get_current_user_id(language):
|
|
| 67 |
return session["user_id"]
|
| 68 |
|
| 69 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 70 |
@app.route("/")
|
| 71 |
def index():
|
| 72 |
return render_template("patient.html", languages=translation.SUPPORTED_LANGUAGES)
|
| 73 |
|
| 74 |
|
| 75 |
-
@app.route("/provider")
|
| 76 |
-
def provider_page():
|
| 77 |
-
return render_template("provider.html")
|
| 78 |
-
|
| 79 |
-
|
| 80 |
@app.route("/transcribe", methods=["POST"])
|
| 81 |
def transcribe_only():
|
| 82 |
"""
|
|
@@ -148,15 +274,51 @@ def patient_input():
|
|
| 148 |
})
|
| 149 |
|
| 150 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 151 |
@app.route("/preview-response", methods=["POST"])
|
|
|
|
| 152 |
def preview_response():
|
| 153 |
"""
|
| 154 |
-
Runs the exact same translate + synthesize pipeline as
|
| 155 |
-
|
| 156 |
-
|
| 157 |
-
|
| 158 |
-
|
| 159 |
-
translation + speech synthesis call, same cost as an actual send.
|
| 160 |
"""
|
| 161 |
interaction_id = request.form.get("interaction_id")
|
| 162 |
response_text = request.form.get("response_text")
|
|
@@ -182,16 +344,10 @@ def preview_response():
|
|
| 182 |
|
| 183 |
|
| 184 |
@app.route("/provider-response", methods=["POST"])
|
|
|
|
| 185 |
def provider_response():
|
| 186 |
interaction_id = request.form.get("interaction_id")
|
| 187 |
response_text = request.form.get("response_text")
|
| 188 |
-
# Optional cache-reuse fields: if the provider already generated a
|
| 189 |
-
# preview of this exact reply via /preview-response and didn't edit
|
| 190 |
-
# it afterward, the frontend passes the already-computed translation
|
| 191 |
-
# back here so this route can skip re-translating and re-synthesizing
|
| 192 |
-
# speech. Both are best-effort - if the preview audio isn't actually
|
| 193 |
-
# on disk, this quietly falls back to doing the full work itself,
|
| 194 |
-
# exactly as if these fields had never been sent.
|
| 195 |
cached_translation = request.form.get("translated_response")
|
| 196 |
reuse_audio = request.form.get("reuse_audio") == "true"
|
| 197 |
|
|
@@ -210,9 +366,14 @@ def provider_response():
|
|
| 210 |
if cached_translation and reuse_audio and os.path.exists(preview_path):
|
| 211 |
translated_response = cached_translation
|
| 212 |
os.replace(preview_path, audio_path)
|
|
|
|
|
|
|
| 213 |
else:
|
| 214 |
translated_response = translation.translate(response_text, "english", patient_language)
|
| 215 |
tts.synthesize_speech(translated_response, patient_language, audio_path)
|
|
|
|
|
|
|
|
|
|
| 216 |
|
| 217 |
database.update_interaction_response(interaction_id, response_text, translated_response)
|
| 218 |
|
|
@@ -222,54 +383,58 @@ def provider_response():
|
|
| 222 |
})
|
| 223 |
|
| 224 |
|
| 225 |
-
@app.route("/interaction/<interaction_id>")
|
| 226 |
-
def get_interaction(interaction_id):
|
| 227 |
-
interaction = database.get_interaction(interaction_id)
|
| 228 |
-
if not interaction:
|
| 229 |
-
return jsonify({"error": "interaction not found"}), 404
|
| 230 |
-
|
| 231 |
-
audio_filename = f"response_{interaction_id}.wav"
|
| 232 |
-
audio_path = os.path.join(AUDIO_DIR, audio_filename)
|
| 233 |
-
interaction["audio_url"] = f"/static/audio/{audio_filename}" if os.path.exists(audio_path) else None
|
| 234 |
-
|
| 235 |
-
return jsonify(interaction)
|
| 236 |
-
|
| 237 |
-
|
| 238 |
@app.route("/pending-interactions")
|
|
|
|
| 239 |
def pending_interactions():
|
| 240 |
return jsonify(database.get_pending_interactions())
|
| 241 |
|
| 242 |
|
| 243 |
@app.route("/completed-interactions")
|
|
|
|
| 244 |
def completed_interactions():
|
| 245 |
return jsonify(database.get_completed_interactions())
|
| 246 |
|
| 247 |
|
| 248 |
-
@app.route("/history")
|
| 249 |
-
def history():
|
| 250 |
-
if "user_id" not in session:
|
| 251 |
-
return jsonify([])
|
| 252 |
-
return jsonify(database.get_interactions_for_user(session["user_id"]))
|
| 253 |
-
|
| 254 |
-
|
| 255 |
@app.route("/patient-history/<user_id>")
|
|
|
|
| 256 |
def patient_history(user_id):
|
| 257 |
"""
|
| 258 |
All of one patient's interactions, for the provider workspace's
|
| 259 |
-
|
| 260 |
-
|
| 261 |
-
|
| 262 |
-
up a specific patient's history by id instead.
|
| 263 |
"""
|
| 264 |
return jsonify(database.get_interactions_for_user(user_id))
|
| 265 |
|
| 266 |
|
| 267 |
-
@app.route("/
|
| 268 |
-
|
| 269 |
-
|
| 270 |
-
|
| 271 |
-
|
| 272 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 273 |
|
| 274 |
|
| 275 |
if __name__ == "__main__":
|
|
|
|
| 10 |
summarizes the likely symptoms or intent for the provider.
|
| 11 |
2. Provider submits a reply in English. The system translates it back
|
| 12 |
into the patient's language and generates spoken audio for it.
|
| 13 |
+
|
| 14 |
+
Provider authentication: all provider-side routes check for a
|
| 15 |
+
PROVIDER_TOKEN in the env file. The provider enters this token once
|
| 16 |
+
per browser session via the /provider-login page. It is stored in a
|
| 17 |
+
signed Flask session cookie, same as the patient's user_id. This is
|
| 18 |
+
not a full authentication system - it is a single shared credential
|
| 19 |
+
that closes the unauthenticated IDOR exposure on provider routes
|
| 20 |
+
without introducing a user accounts table or an auth framework.
|
| 21 |
+
Set PROVIDER_TOKEN to any long random string in the env file before
|
| 22 |
+
deploying with real patient data.
|
| 23 |
+
|
| 24 |
+
Audio cleanup: preview_*.wav files are cleaned up automatically after
|
| 25 |
+
/provider-response either promotes them (os.replace) or falls back to
|
| 26 |
+
a fresh synthesis. response_*.wav files are retained because the
|
| 27 |
+
patient needs to be able to play them back from /interaction/<id>.
|
| 28 |
+
A separate /cleanup-audio admin route deletes response audio older
|
| 29 |
+
than 30 days for disk management on long-running deployments.
|
| 30 |
"""
|
| 31 |
|
| 32 |
import os
|
| 33 |
import uuid
|
| 34 |
+
import glob
|
| 35 |
+
import functools
|
| 36 |
+
from datetime import datetime, timezone, timedelta
|
| 37 |
|
| 38 |
from dotenv import load_dotenv
|
| 39 |
+
from flask import Flask, request, jsonify, render_template, render_template_string, session, redirect, url_for
|
| 40 |
|
| 41 |
import database
|
| 42 |
import stt
|
|
|
|
| 47 |
load_dotenv("env")
|
| 48 |
|
| 49 |
AUDIO_DIR = os.path.join(os.path.dirname(os.path.abspath(__file__)), "static", "audio")
|
| 50 |
+
PROVIDER_TOKEN = os.environ.get("PROVIDER_TOKEN", "")
|
| 51 |
|
| 52 |
|
| 53 |
def create_app():
|
|
|
|
| 68 |
tts.preload_models()
|
| 69 |
|
| 70 |
|
| 71 |
+
# ── Provider authentication ───────────────────────────────────────────────────
|
| 72 |
+
|
| 73 |
+
def provider_required(f):
|
| 74 |
+
"""
|
| 75 |
+
Decorator for provider-side routes. If PROVIDER_TOKEN is not set in
|
| 76 |
+
the env file, the route is unprotected (preserving the existing
|
| 77 |
+
behaviour for local development and demos). If it is set, the request
|
| 78 |
+
must carry a matching token in the provider_session cookie, set by
|
| 79 |
+
/provider-login. Returns 401 JSON for API routes, redirects to the
|
| 80 |
+
login page for the /provider HTML route.
|
| 81 |
+
"""
|
| 82 |
+
@functools.wraps(f)
|
| 83 |
+
def decorated(*args, **kwargs):
|
| 84 |
+
if not PROVIDER_TOKEN:
|
| 85 |
+
# Token not configured: unprotected, existing behaviour preserved.
|
| 86 |
+
return f(*args, **kwargs)
|
| 87 |
+
if session.get("provider_authenticated"):
|
| 88 |
+
return f(*args, **kwargs)
|
| 89 |
+
# HTML route gets a redirect; JSON routes get a 401.
|
| 90 |
+
if request.accept_mimetypes.accept_html and not request.is_json:
|
| 91 |
+
return redirect(url_for("provider_login"))
|
| 92 |
+
return jsonify({"error": "provider authentication required"}), 401
|
| 93 |
+
return decorated
|
| 94 |
+
|
| 95 |
+
|
| 96 |
+
@app.route("/provider-login", methods=["GET", "POST"])
|
| 97 |
+
def provider_login():
|
| 98 |
+
"""
|
| 99 |
+
Simple token entry page. The provider enters the PROVIDER_TOKEN value
|
| 100 |
+
from the env file. On success, sets provider_authenticated in the
|
| 101 |
+
signed session cookie and redirects to /provider.
|
| 102 |
+
"""
|
| 103 |
+
error = None
|
| 104 |
+
if request.method == "POST":
|
| 105 |
+
token = request.form.get("token", "").strip()
|
| 106 |
+
if token and token == PROVIDER_TOKEN:
|
| 107 |
+
session["provider_authenticated"] = True
|
| 108 |
+
return redirect(url_for("provider_page"))
|
| 109 |
+
error = "Incorrect token. Please try again."
|
| 110 |
+
|
| 111 |
+
# Inline template so this fix does not require a new file in templates/.
|
| 112 |
+
return render_template_string("""
|
| 113 |
+
<!DOCTYPE html>
|
| 114 |
+
<html lang="en">
|
| 115 |
+
<head>
|
| 116 |
+
<meta charset="UTF-8">
|
| 117 |
+
<meta name="viewport" content="width=device-width, initial-scale=1.0">
|
| 118 |
+
<title>TalkToDoc — Provider Login</title>
|
| 119 |
+
<link rel="stylesheet" href="{{ url_for('static', filename='css/style.css') }}">
|
| 120 |
+
</head>
|
| 121 |
+
<body>
|
| 122 |
+
<div class="app-shell">
|
| 123 |
+
<div class="app-header provider">
|
| 124 |
+
<div class="brand">
|
| 125 |
+
<div class="brand-mark">
|
| 126 |
+
<svg viewBox="0 0 24 24" fill="none" stroke="white" stroke-width="2"
|
| 127 |
+
stroke-linecap="round" stroke-linejoin="round">
|
| 128 |
+
<path d="M21 11.5a8.38 8.38 0 0 1-.9 3.8 8.5 8.5 0 0 1-7.6 4.7
|
| 129 |
+
8.38 8.38 0 0 1-3.8-.9L3 21l1.9-5.7a8.38 8.38 0 0 1-.9-3.8
|
| 130 |
+
8.5 8.5 0 0 1 4.7-7.6 8.38 8.38 0 0 1 3.8-.9h.5a8.48 8.48
|
| 131 |
+
0 0 1 8 8v.5z"/>
|
| 132 |
+
</svg>
|
| 133 |
+
</div>
|
| 134 |
+
<div>
|
| 135 |
+
<h1>TalkToDoc</h1>
|
| 136 |
+
<span class="role-tag">Provider</span>
|
| 137 |
+
</div>
|
| 138 |
+
</div>
|
| 139 |
+
</div>
|
| 140 |
+
<main class="app-main" id="main-content">
|
| 141 |
+
<div class="card" style="max-width:400px;margin:0 auto;">
|
| 142 |
+
<h2>Provider access</h2>
|
| 143 |
+
<p class="helper-text">Enter the provider token to continue.</p>
|
| 144 |
+
{% if error %}
|
| 145 |
+
<div class="status-banner error">{{ error }}</div>
|
| 146 |
+
{% endif %}
|
| 147 |
+
<form method="post">
|
| 148 |
+
<label for="token">Access token</label>
|
| 149 |
+
<input type="password" id="token" name="token"
|
| 150 |
+
placeholder="Enter provider token" autocomplete="current-password">
|
| 151 |
+
<button type="submit" class="btn btn-provider" style="margin-top:16px;width:100%">
|
| 152 |
+
Continue
|
| 153 |
+
</button>
|
| 154 |
+
</form>
|
| 155 |
+
</div>
|
| 156 |
+
</main>
|
| 157 |
+
</div>
|
| 158 |
+
</body>
|
| 159 |
+
</html>
|
| 160 |
+
""", error=error)
|
| 161 |
+
|
| 162 |
+
|
| 163 |
+
# ── Helpers ───────────────────────────────────────────────────────────────────
|
| 164 |
+
|
| 165 |
def get_current_user_id(language):
|
| 166 |
"""
|
| 167 |
Returns the user_id for the current browser session, creating a new
|
|
|
|
| 181 |
return session["user_id"]
|
| 182 |
|
| 183 |
|
| 184 |
+
def _cleanup_preview(interaction_id):
|
| 185 |
+
"""
|
| 186 |
+
Deletes the preview audio file for an interaction if it still exists.
|
| 187 |
+
Called after /provider-response either promotes the file (os.replace)
|
| 188 |
+
or falls back to fresh synthesis - in the fresh-synthesis path the
|
| 189 |
+
preview file is left behind unused, so we clean it here.
|
| 190 |
+
"""
|
| 191 |
+
preview_path = os.path.join(AUDIO_DIR, f"preview_{interaction_id}.wav")
|
| 192 |
+
try:
|
| 193 |
+
if os.path.exists(preview_path):
|
| 194 |
+
os.remove(preview_path)
|
| 195 |
+
except OSError:
|
| 196 |
+
pass # If the remove fails, it is not worth surfacing as an error.
|
| 197 |
+
|
| 198 |
+
|
| 199 |
+
# ── Patient routes ────────────────────────────────────────────────────────────
|
| 200 |
+
|
| 201 |
@app.route("/")
|
| 202 |
def index():
|
| 203 |
return render_template("patient.html", languages=translation.SUPPORTED_LANGUAGES)
|
| 204 |
|
| 205 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 206 |
@app.route("/transcribe", methods=["POST"])
|
| 207 |
def transcribe_only():
|
| 208 |
"""
|
|
|
|
| 274 |
})
|
| 275 |
|
| 276 |
|
| 277 |
+
@app.route("/interaction/<interaction_id>")
|
| 278 |
+
def get_interaction(interaction_id):
|
| 279 |
+
interaction = database.get_interaction(interaction_id)
|
| 280 |
+
if not interaction:
|
| 281 |
+
return jsonify({"error": "interaction not found"}), 404
|
| 282 |
+
|
| 283 |
+
audio_filename = f"response_{interaction_id}.wav"
|
| 284 |
+
audio_path = os.path.join(AUDIO_DIR, audio_filename)
|
| 285 |
+
interaction["audio_url"] = f"/static/audio/{audio_filename}" if os.path.exists(audio_path) else None
|
| 286 |
+
|
| 287 |
+
return jsonify(interaction)
|
| 288 |
+
|
| 289 |
+
|
| 290 |
+
@app.route("/history")
|
| 291 |
+
def history():
|
| 292 |
+
if "user_id" not in session:
|
| 293 |
+
return jsonify([])
|
| 294 |
+
return jsonify(database.get_interactions_for_user(session["user_id"]))
|
| 295 |
+
|
| 296 |
+
|
| 297 |
+
@app.route("/end-session", methods=["POST"])
|
| 298 |
+
def end_session_route():
|
| 299 |
+
if "session_id" in session:
|
| 300 |
+
database.end_session(session["session_id"], datetime.now(timezone.utc).isoformat())
|
| 301 |
+
session.clear()
|
| 302 |
+
return jsonify({"status": "ended"})
|
| 303 |
+
|
| 304 |
+
|
| 305 |
+
# ── Provider routes (all protected by @provider_required) ─────────────────────
|
| 306 |
+
|
| 307 |
+
@app.route("/provider")
|
| 308 |
+
@provider_required
|
| 309 |
+
def provider_page():
|
| 310 |
+
return render_template("provider.html")
|
| 311 |
+
|
| 312 |
+
|
| 313 |
@app.route("/preview-response", methods=["POST"])
|
| 314 |
+
@provider_required
|
| 315 |
def preview_response():
|
| 316 |
"""
|
| 317 |
+
Runs the exact same translate + synthesize pipeline as /provider-response,
|
| 318 |
+
so the provider can read and hear precisely what the patient will receive
|
| 319 |
+
before committing. Unlike /provider-response, this never touches the
|
| 320 |
+
database. The preview audio file is cleaned up automatically by
|
| 321 |
+
/provider-response regardless of which code path it takes.
|
|
|
|
| 322 |
"""
|
| 323 |
interaction_id = request.form.get("interaction_id")
|
| 324 |
response_text = request.form.get("response_text")
|
|
|
|
| 344 |
|
| 345 |
|
| 346 |
@app.route("/provider-response", methods=["POST"])
|
| 347 |
+
@provider_required
|
| 348 |
def provider_response():
|
| 349 |
interaction_id = request.form.get("interaction_id")
|
| 350 |
response_text = request.form.get("response_text")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 351 |
cached_translation = request.form.get("translated_response")
|
| 352 |
reuse_audio = request.form.get("reuse_audio") == "true"
|
| 353 |
|
|
|
|
| 366 |
if cached_translation and reuse_audio and os.path.exists(preview_path):
|
| 367 |
translated_response = cached_translation
|
| 368 |
os.replace(preview_path, audio_path)
|
| 369 |
+
# preview file was promoted via os.replace so it no longer exists;
|
| 370 |
+
# _cleanup_preview is a no-op here but called for consistency.
|
| 371 |
else:
|
| 372 |
translated_response = translation.translate(response_text, "english", patient_language)
|
| 373 |
tts.synthesize_speech(translated_response, patient_language, audio_path)
|
| 374 |
+
# preview file (if one existed from a prior Preview click) was not
|
| 375 |
+
# promoted and is now orphaned; delete it.
|
| 376 |
+
_cleanup_preview(interaction_id)
|
| 377 |
|
| 378 |
database.update_interaction_response(interaction_id, response_text, translated_response)
|
| 379 |
|
|
|
|
| 383 |
})
|
| 384 |
|
| 385 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 386 |
@app.route("/pending-interactions")
|
| 387 |
+
@provider_required
|
| 388 |
def pending_interactions():
|
| 389 |
return jsonify(database.get_pending_interactions())
|
| 390 |
|
| 391 |
|
| 392 |
@app.route("/completed-interactions")
|
| 393 |
+
@provider_required
|
| 394 |
def completed_interactions():
|
| 395 |
return jsonify(database.get_completed_interactions())
|
| 396 |
|
| 397 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 398 |
@app.route("/patient-history/<user_id>")
|
| 399 |
+
@provider_required
|
| 400 |
def patient_history(user_id):
|
| 401 |
"""
|
| 402 |
All of one patient's interactions, for the provider workspace's
|
| 403 |
+
conversation history panel. Distinct from /history: that route looks
|
| 404 |
+
up the current browser session, which only works for patients viewing
|
| 405 |
+
their own messages. The provider needs to look up by explicit user_id.
|
|
|
|
| 406 |
"""
|
| 407 |
return jsonify(database.get_interactions_for_user(user_id))
|
| 408 |
|
| 409 |
|
| 410 |
+
@app.route("/cleanup-audio", methods=["POST"])
|
| 411 |
+
@provider_required
|
| 412 |
+
def cleanup_audio():
|
| 413 |
+
"""
|
| 414 |
+
Deletes response_*.wav files older than 30 days. Intended for
|
| 415 |
+
long-running deployments where response audio accumulates on disk.
|
| 416 |
+
Call this from a cron job or a scheduled task; it does not run
|
| 417 |
+
automatically. Returns the number of files deleted and their
|
| 418 |
+
total size freed in bytes.
|
| 419 |
+
"""
|
| 420 |
+
cutoff = datetime.now(timezone.utc) - timedelta(days=30)
|
| 421 |
+
deleted_count = 0
|
| 422 |
+
freed_bytes = 0
|
| 423 |
+
|
| 424 |
+
for path in glob.glob(os.path.join(AUDIO_DIR, "response_*.wav")):
|
| 425 |
+
try:
|
| 426 |
+
mtime = datetime.fromtimestamp(os.path.getmtime(path), tz=timezone.utc)
|
| 427 |
+
if mtime < cutoff:
|
| 428 |
+
freed_bytes += os.path.getsize(path)
|
| 429 |
+
os.remove(path)
|
| 430 |
+
deleted_count += 1
|
| 431 |
+
except OSError:
|
| 432 |
+
pass
|
| 433 |
+
|
| 434 |
+
return jsonify({
|
| 435 |
+
"deleted": deleted_count,
|
| 436 |
+
"freed_bytes": freed_bytes,
|
| 437 |
+
})
|
| 438 |
|
| 439 |
|
| 440 |
if __name__ == "__main__":
|
database.py
CHANGED
|
@@ -1,147 +1,208 @@
|
|
| 1 |
"""
|
| 2 |
-
Database layer for TalkToDoc.
|
| 3 |
-
|
| 4 |
-
|
| 5 |
-
|
| 6 |
-
|
| 7 |
-
|
| 8 |
-
|
| 9 |
-
|
| 10 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 11 |
"""
|
| 12 |
|
| 13 |
import os
|
| 14 |
-
|
| 15 |
|
| 16 |
-
import
|
| 17 |
|
| 18 |
-
|
| 19 |
-
SCHEMA_PATH = Path(__file__).parent / "schema.sql"
|
| 20 |
-
SCHEMA_PATH_POSTGRES = Path(__file__).parent / "schema_postgres.sql"
|
| 21 |
|
| 22 |
DATABASE_URL = os.environ.get("DATABASE_URL")
|
| 23 |
|
| 24 |
-
if DATABASE_URL:
|
| 25 |
-
import psycopg2
|
| 26 |
-
import psycopg2.extras
|
| 27 |
-
|
| 28 |
|
| 29 |
def get_connection():
|
|
|
|
|
|
|
|
|
|
|
|
|
| 30 |
if DATABASE_URL:
|
| 31 |
-
|
| 32 |
-
|
| 33 |
-
|
| 34 |
-
|
| 35 |
-
|
| 36 |
-
|
| 37 |
-
|
| 38 |
-
|
| 39 |
-
|
| 40 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 41 |
cursor = connection.cursor()
|
| 42 |
-
cursor.execute(
|
| 43 |
return cursor
|
| 44 |
|
| 45 |
|
| 46 |
-
def
|
| 47 |
-
|
| 48 |
-
|
| 49 |
-
|
| 50 |
-
|
| 51 |
-
return
|
| 52 |
|
| 53 |
|
| 54 |
def init_db():
|
| 55 |
-
|
| 56 |
-
|
| 57 |
-
with
|
|
|
|
|
|
|
| 58 |
if DATABASE_URL:
|
| 59 |
-
|
|
|
|
|
|
|
|
|
|
| 60 |
else:
|
| 61 |
-
|
| 62 |
|
| 63 |
|
| 64 |
def add_user(name, preferred_language, role):
|
| 65 |
-
|
| 66 |
-
|
| 67 |
-
|
| 68 |
-
|
| 69 |
-
|
| 70 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 71 |
|
| 72 |
|
| 73 |
def add_session(user_id, start_time):
|
| 74 |
-
|
| 75 |
-
|
| 76 |
-
|
| 77 |
-
|
| 78 |
-
|
| 79 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 80 |
|
| 81 |
|
| 82 |
def end_session(session_id, end_time):
|
| 83 |
-
|
| 84 |
-
|
| 85 |
-
|
| 86 |
-
|
| 87 |
-
|
| 88 |
-
|
| 89 |
-
|
| 90 |
-
|
| 91 |
-
|
| 92 |
-
|
| 93 |
-
|
| 94 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 95 |
)
|
| 96 |
|
| 97 |
|
| 98 |
def get_interactions_for_user(user_id):
|
| 99 |
-
|
|
|
|
| 100 |
cursor = _run(
|
| 101 |
-
|
| 102 |
-
"SELECT * FROM interaction WHERE user_id =
|
| 103 |
(user_id,),
|
| 104 |
)
|
| 105 |
return [dict(row) for row in cursor.fetchall()]
|
| 106 |
|
| 107 |
|
| 108 |
def get_pending_interactions():
|
| 109 |
-
with
|
| 110 |
cursor = _run(
|
| 111 |
-
|
| 112 |
"SELECT * FROM interaction WHERE provider_response IS NULL ORDER BY timestamp",
|
| 113 |
)
|
| 114 |
return [dict(row) for row in cursor.fetchall()]
|
| 115 |
|
| 116 |
|
| 117 |
def get_completed_interactions():
|
| 118 |
-
with
|
| 119 |
cursor = _run(
|
| 120 |
-
|
| 121 |
"SELECT * FROM interaction WHERE provider_response IS NOT NULL ORDER BY timestamp DESC",
|
| 122 |
)
|
| 123 |
return [dict(row) for row in cursor.fetchall()]
|
| 124 |
-
|
| 125 |
-
|
| 126 |
-
def get_interaction(interaction_id):
|
| 127 |
-
with get_connection() as connection:
|
| 128 |
-
cursor = _run(connection, "SELECT * FROM interaction WHERE id = ?", (interaction_id,))
|
| 129 |
-
row = cursor.fetchone()
|
| 130 |
-
return dict(row) if row else None
|
| 131 |
-
|
| 132 |
-
|
| 133 |
-
def update_interaction_response(interaction_id, provider_response, translated_response):
|
| 134 |
-
with get_connection() as connection:
|
| 135 |
-
_run(
|
| 136 |
-
connection,
|
| 137 |
-
"UPDATE interaction SET provider_response = ?, translated_response = ? WHERE id = ?",
|
| 138 |
-
(provider_response, translated_response, interaction_id),
|
| 139 |
-
)
|
| 140 |
-
|
| 141 |
-
|
| 142 |
-
if __name__ == "__main__":
|
| 143 |
-
init_db()
|
| 144 |
-
if DATABASE_URL:
|
| 145 |
-
print("Database initialized (Postgres)")
|
| 146 |
-
else:
|
| 147 |
-
print("Database created at:", DB_PATH)
|
|
|
|
| 1 |
"""
|
| 2 |
+
Database access layer for TalkToDoc.
|
| 3 |
+
Handles both SQLite (local development) and PostgreSQL (production) through
|
| 4 |
+
a single shared query interface. The only difference between the two is how
|
| 5 |
+
get_connection() builds the connection object; every function above that is
|
| 6 |
+
identical regardless of which database is running underneath.
|
| 7 |
+
|
| 8 |
+
Connection management: every public function in this module uses the
|
| 9 |
+
_connection() context manager (not get_connection() directly). That
|
| 10 |
+
context manager commits on success, rolls back on exception, and always
|
| 11 |
+
closes the connection before returning control. The bare
|
| 12 |
+
`with get_connection() as conn:` pattern that psycopg2 and sqlite3 both
|
| 13 |
+
support commits/rolls back correctly but never closes the connection,
|
| 14 |
+
which leaks file handles against SQLite and exhausts the Postgres
|
| 15 |
+
connection pool in production over time.
|
| 16 |
"""
|
| 17 |
|
| 18 |
import os
|
| 19 |
+
import contextlib
|
| 20 |
|
| 21 |
+
from dotenv import load_dotenv
|
| 22 |
|
| 23 |
+
load_dotenv("env")
|
|
|
|
|
|
|
| 24 |
|
| 25 |
DATABASE_URL = os.environ.get("DATABASE_URL")
|
| 26 |
|
|
|
|
|
|
|
|
|
|
|
|
|
| 27 |
|
| 28 |
def get_connection():
|
| 29 |
+
"""
|
| 30 |
+
Returns a raw database connection. Callers should use _connection()
|
| 31 |
+
instead, which wraps this and guarantees the connection is closed.
|
| 32 |
+
"""
|
| 33 |
if DATABASE_URL:
|
| 34 |
+
import psycopg2
|
| 35 |
+
import psycopg2.extras
|
| 36 |
+
conn = psycopg2.connect(DATABASE_URL, cursor_factory=psycopg2.extras.RealDictCursor)
|
| 37 |
+
return conn
|
| 38 |
+
else:
|
| 39 |
+
import sqlite3
|
| 40 |
+
conn = sqlite3.connect(
|
| 41 |
+
os.path.join(os.path.dirname(os.path.abspath(__file__)), "talktodoc.db")
|
| 42 |
+
)
|
| 43 |
+
conn.row_factory = sqlite3.Row
|
| 44 |
+
return conn
|
| 45 |
+
|
| 46 |
+
|
| 47 |
+
@contextlib.contextmanager
|
| 48 |
+
def _connection():
|
| 49 |
+
"""
|
| 50 |
+
Context manager that opens a connection, yields it, commits on clean
|
| 51 |
+
exit, rolls back on exception, and always closes. Use this everywhere
|
| 52 |
+
instead of `with get_connection() as conn:`.
|
| 53 |
+
"""
|
| 54 |
+
conn = get_connection()
|
| 55 |
+
try:
|
| 56 |
+
yield conn
|
| 57 |
+
conn.commit()
|
| 58 |
+
except Exception:
|
| 59 |
+
conn.rollback()
|
| 60 |
+
raise
|
| 61 |
+
finally:
|
| 62 |
+
conn.close()
|
| 63 |
+
|
| 64 |
+
|
| 65 |
+
def _run(connection, sql, params=()):
|
| 66 |
cursor = connection.cursor()
|
| 67 |
+
cursor.execute(sql, params)
|
| 68 |
return cursor
|
| 69 |
|
| 70 |
|
| 71 |
+
def _placeholder():
|
| 72 |
+
"""
|
| 73 |
+
SQLite uses ? for bind parameters; psycopg2 uses %s.
|
| 74 |
+
Returns the right one for whichever database is active.
|
| 75 |
+
"""
|
| 76 |
+
return "%s" if DATABASE_URL else "?"
|
| 77 |
|
| 78 |
|
| 79 |
def init_db():
|
| 80 |
+
schema_file = "schema_postgres.sql" if DATABASE_URL else "schema.sql"
|
| 81 |
+
schema_path = os.path.join(os.path.dirname(os.path.abspath(__file__)), schema_file)
|
| 82 |
+
with open(schema_path) as f:
|
| 83 |
+
schema = f.read()
|
| 84 |
+
with _connection() as conn:
|
| 85 |
if DATABASE_URL:
|
| 86 |
+
for statement in schema.split(";"):
|
| 87 |
+
stmt = statement.strip()
|
| 88 |
+
if stmt:
|
| 89 |
+
_run(conn, stmt)
|
| 90 |
else:
|
| 91 |
+
conn.executescript(schema)
|
| 92 |
|
| 93 |
|
| 94 |
def add_user(name, preferred_language, role):
|
| 95 |
+
p = _placeholder()
|
| 96 |
+
with _connection() as conn:
|
| 97 |
+
if DATABASE_URL:
|
| 98 |
+
cursor = _run(
|
| 99 |
+
conn,
|
| 100 |
+
f"INSERT INTO app_user (name, preferred_language, role) VALUES ({p}, {p}, {p}) RETURNING id",
|
| 101 |
+
(name, preferred_language, role),
|
| 102 |
+
)
|
| 103 |
+
return cursor.fetchone()["id"]
|
| 104 |
+
else:
|
| 105 |
+
cursor = _run(
|
| 106 |
+
conn,
|
| 107 |
+
f"INSERT INTO app_user (name, preferred_language, role) VALUES ({p}, {p}, {p})",
|
| 108 |
+
(name, preferred_language, role),
|
| 109 |
+
)
|
| 110 |
+
return cursor.lastrowid
|
| 111 |
|
| 112 |
|
| 113 |
def add_session(user_id, start_time):
|
| 114 |
+
p = _placeholder()
|
| 115 |
+
with _connection() as conn:
|
| 116 |
+
if DATABASE_URL:
|
| 117 |
+
cursor = _run(
|
| 118 |
+
conn,
|
| 119 |
+
f"INSERT INTO session (user_id, start_time) VALUES ({p}, {p}) RETURNING id",
|
| 120 |
+
(user_id, start_time),
|
| 121 |
+
)
|
| 122 |
+
return cursor.fetchone()["id"]
|
| 123 |
+
else:
|
| 124 |
+
cursor = _run(
|
| 125 |
+
conn,
|
| 126 |
+
f"INSERT INTO session (user_id, start_time) VALUES ({p}, {p})",
|
| 127 |
+
(user_id, start_time),
|
| 128 |
+
)
|
| 129 |
+
return cursor.lastrowid
|
| 130 |
|
| 131 |
|
| 132 |
def end_session(session_id, end_time):
|
| 133 |
+
p = _placeholder()
|
| 134 |
+
with _connection() as conn:
|
| 135 |
+
_run(
|
| 136 |
+
conn,
|
| 137 |
+
f"UPDATE session SET end_time = {p} WHERE id = {p}",
|
| 138 |
+
(end_time, session_id),
|
| 139 |
+
)
|
| 140 |
+
|
| 141 |
+
|
| 142 |
+
def add_interaction(user_id, input_text, detected_language, translated_text,
|
| 143 |
+
nlu_summary, timestamp):
|
| 144 |
+
p = _placeholder()
|
| 145 |
+
sql = (
|
| 146 |
+
f"INSERT INTO interaction "
|
| 147 |
+
f"(user_id, input_text, detected_language, translated_text, nlu_summary, timestamp) "
|
| 148 |
+
f"VALUES ({p}, {p}, {p}, {p}, {p}, {p})"
|
| 149 |
+
)
|
| 150 |
+
params = (user_id, input_text, detected_language, translated_text, nlu_summary, timestamp)
|
| 151 |
+
with _connection() as conn:
|
| 152 |
+
if DATABASE_URL:
|
| 153 |
+
cursor = _run(conn, sql.replace(")", " RETURNING id)", 1), params)
|
| 154 |
+
return cursor.fetchone()["id"]
|
| 155 |
+
else:
|
| 156 |
+
cursor = _run(conn, sql, params)
|
| 157 |
+
return cursor.lastrowid
|
| 158 |
+
|
| 159 |
+
|
| 160 |
+
def get_interaction(interaction_id):
|
| 161 |
+
p = _placeholder()
|
| 162 |
+
with _connection() as conn:
|
| 163 |
+
cursor = _run(
|
| 164 |
+
conn,
|
| 165 |
+
f"SELECT * FROM interaction WHERE id = {p}",
|
| 166 |
+
(interaction_id,),
|
| 167 |
+
)
|
| 168 |
+
row = cursor.fetchone()
|
| 169 |
+
return dict(row) if row else None
|
| 170 |
+
|
| 171 |
+
|
| 172 |
+
def update_interaction_response(interaction_id, provider_response, translated_response):
|
| 173 |
+
p = _placeholder()
|
| 174 |
+
with _connection() as conn:
|
| 175 |
+
_run(
|
| 176 |
+
conn,
|
| 177 |
+
f"UPDATE interaction SET provider_response = {p}, translated_response = {p} WHERE id = {p}",
|
| 178 |
+
(provider_response, translated_response, interaction_id),
|
| 179 |
)
|
| 180 |
|
| 181 |
|
| 182 |
def get_interactions_for_user(user_id):
|
| 183 |
+
p = _placeholder()
|
| 184 |
+
with _connection() as conn:
|
| 185 |
cursor = _run(
|
| 186 |
+
conn,
|
| 187 |
+
f"SELECT * FROM interaction WHERE user_id = {p} ORDER BY timestamp",
|
| 188 |
(user_id,),
|
| 189 |
)
|
| 190 |
return [dict(row) for row in cursor.fetchall()]
|
| 191 |
|
| 192 |
|
| 193 |
def get_pending_interactions():
|
| 194 |
+
with _connection() as conn:
|
| 195 |
cursor = _run(
|
| 196 |
+
conn,
|
| 197 |
"SELECT * FROM interaction WHERE provider_response IS NULL ORDER BY timestamp",
|
| 198 |
)
|
| 199 |
return [dict(row) for row in cursor.fetchall()]
|
| 200 |
|
| 201 |
|
| 202 |
def get_completed_interactions():
|
| 203 |
+
with _connection() as conn:
|
| 204 |
cursor = _run(
|
| 205 |
+
conn,
|
| 206 |
"SELECT * FROM interaction WHERE provider_response IS NOT NULL ORDER BY timestamp DESC",
|
| 207 |
)
|
| 208 |
return [dict(row) for row in cursor.fetchall()]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|