feat: robust stateless hybrid architecture using base64 context handover
This commit is contained in:
cra88y
+62
-136
@@ -1,5 +1,5 @@
|
||||
import json, secrets, time, http.client, ssl, os, logging, html, urllib.parse
|
||||
from flask import Response, request, abort
|
||||
import json, http.client, ssl, os, logging, base64
|
||||
from flask import Response, request
|
||||
from searx.plugins import Plugin, PluginInfo
|
||||
from searx.result_types import EngineResults
|
||||
from flask_babel import gettext
|
||||
@@ -19,181 +19,107 @@ class SXNGPlugin(Plugin):
|
||||
preference_section="general",
|
||||
)
|
||||
self.api_key = os.getenv('GEMINI_API_KEY')
|
||||
self.model = os.getenv('GEMINI_MODEL', 'gemini-3-flash-preview')
|
||||
self.tokens = {}
|
||||
|
||||
if not self.api_key:
|
||||
logger.error(f"[{self.id}] API Key missing! Set GEMINI_API_KEY env var.")
|
||||
else:
|
||||
logger.info(f"[{self.id}] Initialized with model: {self.model}")
|
||||
self.model = os.getenv('GEMINI_MODEL', 'gemini-1.5-flash')
|
||||
|
||||
def init(self, app):
|
||||
@app.route('/gemini-stream')
|
||||
@app.route('/gemini-stream', methods=['POST'])
|
||||
def g_stream():
|
||||
t = request.args.get('token')
|
||||
q = request.args.get('q', '')
|
||||
data = request.json or {}
|
||||
context_text = data.get('context', '')
|
||||
q = data.get('q', '')
|
||||
|
||||
# Maintenance: handle dict structure
|
||||
current_time = time.time()
|
||||
self.tokens = {k: v for k, v in self.tokens.items() if v['expires'] > current_time}
|
||||
|
||||
if t not in self.tokens or not self.api_key:
|
||||
abort(403)
|
||||
|
||||
token_data = self.tokens[t]
|
||||
context_text = token_data.get('context', '')
|
||||
del self.tokens[t]
|
||||
if not self.api_key or not q:
|
||||
return Response("Error: Missing Key or Query", status=400)
|
||||
|
||||
def generate():
|
||||
host = "generativelanguage.googleapis.com"
|
||||
path = f"/v1beta/models/{self.model}:streamGenerateContent?key={self.api_key}"
|
||||
try:
|
||||
context = ssl.create_default_context()
|
||||
conn = http.client.HTTPSConnection(host, context=context)
|
||||
|
||||
# RAG PROMPT
|
||||
prompt = f"""
|
||||
You are a concise search assistant. Use the provided SEARCH RESULTS to answer the USER QUERY.
|
||||
If the results don't contain the answer, use your knowledge but prioritize the results.
|
||||
Keep the answer under 4 sentences.
|
||||
|
||||
SEARCH RESULTS:
|
||||
{context_text}
|
||||
|
||||
USER QUERY: {q}
|
||||
"""
|
||||
payload = {
|
||||
"contents": [{"parts": [{"text": prompt}]}],
|
||||
"generationConfig": {
|
||||
"maxOutputTokens": 500,
|
||||
"temperature": 0.2 # Lower temperature for better factual accuracy
|
||||
}
|
||||
}
|
||||
|
||||
conn.request("POST", path, body=json.dumps(payload),
|
||||
headers={"Content-Type": "application/json"})
|
||||
conn = http.client.HTTPSConnection(host, context=ssl.create_default_context())
|
||||
prompt = f"Using these SEARCH RESULTS, answer the USER QUERY concisely (<4 sentences). If results are irrelevant, say so.\n\nRESULTS:\n{context_text}\n\nUSER QUERY: {q}"
|
||||
payload = {"contents": [{"parts": [{"text": prompt}]}], "generationConfig": {"maxOutputTokens": 400, "temperature": 0.3}}
|
||||
conn.request("POST", path, body=json.dumps(payload), headers={"Content-Type": "application/json"})
|
||||
res = conn.getresponse()
|
||||
|
||||
buffer = ""
|
||||
for chunk in res:
|
||||
if not chunk: continue
|
||||
buffer += chunk.decode('utf-8')
|
||||
|
||||
while True:
|
||||
start = buffer.find('{')
|
||||
if start == -1:
|
||||
buffer = "" # Clear garbage
|
||||
break
|
||||
|
||||
brace_count = 0
|
||||
end = -1
|
||||
if start == -1: break
|
||||
brace_count, end = 0, -1
|
||||
for i in range(start, len(buffer)):
|
||||
if buffer[i] == '{': brace_count += 1
|
||||
elif buffer[i] == '}': brace_count -= 1
|
||||
if brace_count == 0:
|
||||
end = i + 1
|
||||
break
|
||||
|
||||
if end == -1: break # Wait for more data
|
||||
|
||||
if end == -1: break
|
||||
try:
|
||||
raw_json = buffer[start:end]
|
||||
data = json.loads(raw_json)
|
||||
parts = data.get('candidates', [{}])[0].get('content', {}).get('parts', [])
|
||||
for part in parts:
|
||||
text = part.get('text', '')
|
||||
if text:
|
||||
yield text
|
||||
except Exception:
|
||||
pass
|
||||
|
||||
data = json.loads(buffer[start:end])
|
||||
candidates = data.get('candidates', [])
|
||||
if candidates:
|
||||
text = candidates[0]['content']['parts'][0]['text']
|
||||
if text: yield text
|
||||
except: pass
|
||||
buffer = buffer[end:]
|
||||
conn.close()
|
||||
except Exception as e:
|
||||
logger.error(f"[{self.id}] Stream error: {e}")
|
||||
yield f" [Error: {str(e)}]"
|
||||
|
||||
return Response(generate(), mimetype='text/plain', headers={
|
||||
'Cache-Control': 'no-cache, no-store, must-revalidate',
|
||||
'Pragma': 'no-cache',
|
||||
'Expires': '0',
|
||||
'X-Accel-Buffering': 'no'
|
||||
})
|
||||
|
||||
@app.route('/gemini.js')
|
||||
def g_script():
|
||||
token = request.args.get('token', '')
|
||||
query = request.args.get('q', '')
|
||||
|
||||
js_query = json.dumps(query)
|
||||
js_token = json.dumps(token)
|
||||
|
||||
js_code = f"""
|
||||
(async () => {{
|
||||
const shell = document.getElementById('ai-shell');
|
||||
const out = document.getElementById('ai-out');
|
||||
if (!shell || !out) return;
|
||||
|
||||
const token = {js_token};
|
||||
const query = {js_query};
|
||||
|
||||
try {{
|
||||
const res = await fetch(`/gemini-stream?token=${{token}}&q=` + encodeURIComponent(query));
|
||||
if (!res.ok) throw new Error(res.statusText);
|
||||
|
||||
const reader = res.body.getReader();
|
||||
const decoder = new TextDecoder();
|
||||
while (true) {{
|
||||
const {{done, value}} = await reader.read();
|
||||
if (done) break;
|
||||
const chunk = decoder.decode(value);
|
||||
if (chunk.trim()) {{
|
||||
shell.style.display = 'block';
|
||||
out.innerText += chunk;
|
||||
}}
|
||||
}}
|
||||
}} catch (e) {{ console.error("Gemini Stream Failed", e); }}
|
||||
}})();
|
||||
"""
|
||||
return Response(js_code, mimetype='application/javascript')
|
||||
|
||||
return Response(generate(), mimetype='text/plain', headers={'X-Accel-Buffering': 'no'})
|
||||
return True
|
||||
|
||||
def post_search(self, request, search) -> EngineResults:
|
||||
results = EngineResults()
|
||||
if search.search_query.pageno > 1 or not self.active or not self.api_key:
|
||||
if not self.active or not self.api_key or search.search_query.pageno > 1:
|
||||
return results
|
||||
|
||||
# Extract context from top 5 search results for RAG
|
||||
context_parts = []
|
||||
raw_results = search.result_container.get_ordered_results()
|
||||
for i, res in enumerate(raw_results[:5]):
|
||||
title = res.get('title', 'No Title')
|
||||
content = res.get('content', 'No Content')
|
||||
context_parts.append(f"Source [{i+1}]: {title}\nSnippet: {content}")
|
||||
context_list = [f"[{i+1}] {r.get('title')}: {r.get('content')}" for i, r in enumerate(raw_results[:6])]
|
||||
context_str = "\n".join(context_list)
|
||||
|
||||
context_str = "\n\n".join(context_parts)
|
||||
# Base64 Encode to ensure HTML safety
|
||||
b64_context = base64.b64encode(context_str.encode('utf-8')).decode('utf-8')
|
||||
js_q = json.dumps(search.search_query.query)
|
||||
|
||||
tk = secrets.token_urlsafe(16)
|
||||
self.tokens[tk] = {
|
||||
"expires": time.time() + 90,
|
||||
"context": context_str
|
||||
}
|
||||
|
||||
logger.debug(f"[{self.id}] Prepared RAG context for query: {search.search_query.query[:20]}...")
|
||||
|
||||
# Encode query for the URL parameter in the script tag
|
||||
safe_query_param = urllib.parse.quote(search.search_query.query)
|
||||
|
||||
# HTML Payload:
|
||||
# 1. The Container (Hidden by default)
|
||||
# 2. The Script Tag (Pointing to our dynamic route with params)
|
||||
html_payload = f'''
|
||||
<div id="ai-shell" style="display:none; margin-bottom: 2rem; padding: 1.2rem; border-bottom: 1px solid var(--color-result-border);">
|
||||
<div id="ai-out" style="line-height: 1.7; white-space: pre-wrap; color: var(--color-result-description); font-size: 0.95rem;"></div>
|
||||
<div id="ai-out" style="line-height: 1.7; white-space: pre-wrap; color: var(--color-result-description); font-size: 0.95rem;">Thinking...</div>
|
||||
</div>
|
||||
<script src="/gemini.js?token={tk}&q={safe_query_param}"></script>
|
||||
'''
|
||||
<script>
|
||||
(async () => {{
|
||||
const q = {js_q};
|
||||
const b64 = "{b64_context}";
|
||||
const shell = document.getElementById('ai-shell');
|
||||
const out = document.getElementById('ai-out');
|
||||
|
||||
const container = document.getElementById('urls') || document.getElementById('main_results');
|
||||
if (container && shell) {{ container.prepend(shell); shell.style.display = 'block'; }}
|
||||
|
||||
try {{
|
||||
// Decode context client-side
|
||||
const ctx = new TextDecoder().decode(Uint8Array.from(atob(b64), c => c.charCodeAt(0)));
|
||||
|
||||
const res = await fetch('/gemini-stream', {{
|
||||
method: 'POST',
|
||||
headers: {{ 'Content-Type': 'application/json' }},
|
||||
body: JSON.stringify({{ q: q, context: ctx }})
|
||||
}});
|
||||
|
||||
const reader = res.body.getReader();
|
||||
const decoder = new TextDecoder();
|
||||
out.innerText = "";
|
||||
|
||||
while (true) {{
|
||||
const {{done, value}} = await reader.read();
|
||||
if (done) break;
|
||||
out.innerText += decoder.decode(value);
|
||||
}}
|
||||
}} catch (e) {{ console.error(e); out.innerText += " [Error]"; }}
|
||||
}})();
|
||||
</script>
|
||||
'''
|
||||
results.add(results.types.Answer(answer=Markup(html_payload)))
|
||||
return results
|
||||
Reference in New Issue
Block a user