commit
Browse files
app.py
CHANGED
|
@@ -16,7 +16,6 @@ BUCKET = os.environ["SUPABASE_BUCKET"]
|
|
| 16 |
PDF_URL = f"{os.environ['SUPABASE_URL']}/storage/v1/object/public/{BUCKET}/pruefungsordnung.pdf"
|
| 17 |
HG_URL = f"{os.environ['SUPABASE_URL']}/storage/v1/object/public/{BUCKET}/hochschulgesetz.html"
|
| 18 |
|
| 19 |
-
|
| 20 |
# ------------------------------------------
|
| 21 |
# Viewer PDF base64
|
| 22 |
# ------------------------------------------
|
|
@@ -25,7 +24,6 @@ def encode_pdf_src():
|
|
| 25 |
b64 = base64.b64encode(pdf_bytes).decode("utf-8")
|
| 26 |
return f"data:application/pdf;base64,{b64}"
|
| 27 |
|
| 28 |
-
|
| 29 |
# ------------------------------------------
|
| 30 |
# HTML viewer
|
| 31 |
# ------------------------------------------
|
|
@@ -33,7 +31,6 @@ def encode_html():
|
|
| 33 |
html_bytes = load_file_bytes(BUCKET, "hochschulgesetz.html")
|
| 34 |
return html_bytes.decode("utf-8", errors="ignore")
|
| 35 |
|
| 36 |
-
|
| 37 |
# ------------------------------------------
|
| 38 |
# Speech-to-text FIXED
|
| 39 |
# ------------------------------------------
|
|
@@ -49,7 +46,6 @@ def transcribe(audio_path):
|
|
| 49 |
)
|
| 50 |
return (result.text or "").strip()
|
| 51 |
|
| 52 |
-
|
| 53 |
# ------------------------------------------
|
| 54 |
# MAIN CHAT FUNCTION
|
| 55 |
# ------------------------------------------
|
|
@@ -70,19 +66,29 @@ def chat_fn(text, audio, history):
|
|
| 70 |
# 2) RAG
|
| 71 |
answer, docs = rag_answer(question, history or [])
|
| 72 |
|
| 73 |
-
# 3) Build Quellen (click được)
|
| 74 |
html = "<ol>"
|
| 75 |
for i, d in enumerate(docs):
|
| 76 |
meta = d.get("metadata", {}) or {}
|
| 77 |
src = meta.get("source", "?")
|
|
|
|
|
|
|
| 78 |
|
|
|
|
| 79 |
if "Prüfungsordnung" in src:
|
| 80 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 81 |
else:
|
| 82 |
-
|
| 83 |
-
|
| 84 |
-
|
| 85 |
-
|
|
|
|
|
|
|
| 86 |
|
| 87 |
snippet = (d.get("content") or "")[:200]
|
| 88 |
|
|
@@ -96,16 +102,15 @@ def chat_fn(text, audio, history):
|
|
| 96 |
"""
|
| 97 |
html += "</ol>"
|
| 98 |
|
| 99 |
-
# 4) Gradio message history
|
| 100 |
new_history = (history or []) + [
|
| 101 |
{"role": "user", "content": question},
|
| 102 |
{"role": "assistant", "content": answer},
|
| 103 |
]
|
| 104 |
|
| 105 |
-
# Reset audio input
|
| 106 |
return new_history, html, gr.update(value=None)
|
| 107 |
|
| 108 |
-
|
| 109 |
# ------------------------------------------
|
| 110 |
# UI LAYOUT
|
| 111 |
# ------------------------------------------
|
|
@@ -116,7 +121,10 @@ with gr.Blocks() as demo:
|
|
| 116 |
with gr.Column(scale=3):
|
| 117 |
chatbot = gr.Chatbot(label="Chat (RAG)")
|
| 118 |
text_input = gr.Textbox(label="Text Eingabe")
|
| 119 |
-
audio_input = gr.Audio(
|
|
|
|
|
|
|
|
|
|
| 120 |
send_btn = gr.Button("Senden")
|
| 121 |
|
| 122 |
with gr.Column(scale=2):
|
|
|
|
| 16 |
PDF_URL = f"{os.environ['SUPABASE_URL']}/storage/v1/object/public/{BUCKET}/pruefungsordnung.pdf"
|
| 17 |
HG_URL = f"{os.environ['SUPABASE_URL']}/storage/v1/object/public/{BUCKET}/hochschulgesetz.html"
|
| 18 |
|
|
|
|
| 19 |
# ------------------------------------------
|
| 20 |
# Viewer PDF base64
|
| 21 |
# ------------------------------------------
|
|
|
|
| 24 |
b64 = base64.b64encode(pdf_bytes).decode("utf-8")
|
| 25 |
return f"data:application/pdf;base64,{b64}"
|
| 26 |
|
|
|
|
| 27 |
# ------------------------------------------
|
| 28 |
# HTML viewer
|
| 29 |
# ------------------------------------------
|
|
|
|
| 31 |
html_bytes = load_file_bytes(BUCKET, "hochschulgesetz.html")
|
| 32 |
return html_bytes.decode("utf-8", errors="ignore")
|
| 33 |
|
|
|
|
| 34 |
# ------------------------------------------
|
| 35 |
# Speech-to-text FIXED
|
| 36 |
# ------------------------------------------
|
|
|
|
| 46 |
)
|
| 47 |
return (result.text or "").strip()
|
| 48 |
|
|
|
|
| 49 |
# ------------------------------------------
|
| 50 |
# MAIN CHAT FUNCTION
|
| 51 |
# ------------------------------------------
|
|
|
|
| 66 |
# 2) RAG
|
| 67 |
answer, docs = rag_answer(question, history or [])
|
| 68 |
|
| 69 |
+
# 3) Build Quellen (click được, phân biệt PDF vs HTML)
|
| 70 |
html = "<ol>"
|
| 71 |
for i, d in enumerate(docs):
|
| 72 |
meta = d.get("metadata", {}) or {}
|
| 73 |
src = meta.get("source", "?")
|
| 74 |
+
page = meta.get("page", None)
|
| 75 |
+
anchor_id = meta.get("anchor_id") # từ ingest.py
|
| 76 |
|
| 77 |
+
# Chọn link tùy nguồn
|
| 78 |
if "Prüfungsordnung" in src:
|
| 79 |
+
# cố gắng nhảy đúng Seite
|
| 80 |
+
if page:
|
| 81 |
+
link = f"{PDF_URL}#page={page}"
|
| 82 |
+
else:
|
| 83 |
+
link = PDF_URL
|
| 84 |
+
page_info = f"(Seite {page})" if page else ""
|
| 85 |
else:
|
| 86 |
+
# Hochschulgesetz NRW – dùng anchor_id trong hochschulgesetz.html
|
| 87 |
+
if anchor_id:
|
| 88 |
+
link = f"{HG_URL}#{anchor_id}"
|
| 89 |
+
else:
|
| 90 |
+
link = HG_URL
|
| 91 |
+
page_info = "" # HTML không có page
|
| 92 |
|
| 93 |
snippet = (d.get("content") or "")[:200]
|
| 94 |
|
|
|
|
| 102 |
"""
|
| 103 |
html += "</ol>"
|
| 104 |
|
| 105 |
+
# 4) Gradio message history (kiểu messages)
|
| 106 |
new_history = (history or []) + [
|
| 107 |
{"role": "user", "content": question},
|
| 108 |
{"role": "assistant", "content": answer},
|
| 109 |
]
|
| 110 |
|
| 111 |
+
# Reset audio input (xóa sóng cũ)
|
| 112 |
return new_history, html, gr.update(value=None)
|
| 113 |
|
|
|
|
| 114 |
# ------------------------------------------
|
| 115 |
# UI LAYOUT
|
| 116 |
# ------------------------------------------
|
|
|
|
| 121 |
with gr.Column(scale=3):
|
| 122 |
chatbot = gr.Chatbot(label="Chat (RAG)")
|
| 123 |
text_input = gr.Textbox(label="Text Eingabe")
|
| 124 |
+
audio_input = gr.Audio(
|
| 125 |
+
type="filepath",
|
| 126 |
+
label="Spracheingabe (Mikrofon)"
|
| 127 |
+
)
|
| 128 |
send_btn = gr.Button("Senden")
|
| 129 |
|
| 130 |
with gr.Column(scale=2):
|