Spaces:
Sleeping
Sleeping
first commit
Browse files
app.py
CHANGED
|
@@ -125,37 +125,38 @@ with tab_jsonl:
|
|
| 125 |
st.json(obj)
|
| 126 |
|
| 127 |
# 上傳按鈕:會在送出前幫每一筆加上 metadata
|
| 128 |
-
if st.button("
|
| 129 |
if BACKEND_URL is None:
|
| 130 |
st.warning("尚未設定 BACKEND_URL,無法實際送出,請在 `st.secrets` 中配置。")
|
| 131 |
else:
|
| 132 |
-
|
| 133 |
-
|
| 134 |
-
|
| 135 |
-
|
| 136 |
-
|
| 137 |
-
|
| 138 |
-
|
| 139 |
-
# 重新組一份帶 metadata 的 jsonl 內容
|
| 140 |
-
enriched_lines = []
|
| 141 |
-
for obj in parsed_lines:
|
| 142 |
-
obj_with_meta = {
|
| 143 |
-
**obj,
|
| 144 |
-
"metadata": meta,
|
| 145 |
}
|
| 146 |
-
enriched_lines.append(json.dumps(obj_with_meta, ensure_ascii=False))
|
| 147 |
|
| 148 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 149 |
|
| 150 |
-
|
| 151 |
-
|
| 152 |
-
|
| 153 |
-
|
| 154 |
-
|
| 155 |
-
|
| 156 |
-
|
| 157 |
-
|
| 158 |
-
|
|
|
|
|
|
|
| 159 |
|
| 160 |
|
| 161 |
# ---------- Tab 2: PDF ----------
|
|
@@ -209,32 +210,36 @@ with tab_pdf:
|
|
| 209 |
|
| 210 |
any_valid_pdf = any(tlen >= 100 for _, tlen in pdf_results) if pdf_results else False
|
| 211 |
|
| 212 |
-
if st.button("
|
| 213 |
if BACKEND_URL is None:
|
| 214 |
st.warning("尚未設定 BACKEND_URL,無法實際送出,請在 `st.secrets` 中配置。")
|
| 215 |
else:
|
| 216 |
-
|
| 217 |
-
|
| 218 |
-
if text_len < 100:
|
| 219 |
-
continue # 跳過疑似掃描檔
|
| 220 |
-
pdf.seek(0)
|
| 221 |
-
files.append(("files", (pdf.name, pdf.getvalue(), "application/pdf")))
|
| 222 |
-
|
| 223 |
-
if not files:
|
| 224 |
-
st.warning("沒有通過文字檢查的 PDF 檔案可送出。")
|
| 225 |
else:
|
| 226 |
-
|
| 227 |
-
|
| 228 |
-
|
| 229 |
-
|
| 230 |
-
|
| 231 |
-
|
| 232 |
-
|
| 233 |
-
|
| 234 |
-
|
| 235 |
-
|
| 236 |
-
st.success("已成功送交後端伺服器,等待後端進一步檢查與處理。")
|
| 237 |
else:
|
| 238 |
-
|
| 239 |
-
|
| 240 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 125 |
st.json(obj)
|
| 126 |
|
| 127 |
# 上傳按鈕:會在送出前幫每一筆加上 metadata
|
| 128 |
+
if st.button("上傳對話資料", disabled=not (jsonl_file and jsonl_valid or BACKEND_URL is None)):
|
| 129 |
if BACKEND_URL is None:
|
| 130 |
st.warning("尚未設定 BACKEND_URL,無法實際送出,請在 `st.secrets` 中配置。")
|
| 131 |
else:
|
| 132 |
+
with st.spinner("正在上傳對話資料並檢查,請稍候…"):
|
| 133 |
+
# 準備 metadata(會附加在每一行 JSON 物件上)
|
| 134 |
+
meta = {
|
| 135 |
+
"uploaded_at": uploaded_at, # UTC+8 ISO 字串
|
| 136 |
+
"contributor_email": contributor_email if contributor_email.strip() else None,
|
| 137 |
+
"share_permission": bool(share_permission),
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 138 |
}
|
|
|
|
| 139 |
|
| 140 |
+
# 重新組一份帶 metadata 的 jsonl 內容
|
| 141 |
+
enriched_lines = []
|
| 142 |
+
for obj in parsed_lines:
|
| 143 |
+
obj_with_meta = {
|
| 144 |
+
**obj,
|
| 145 |
+
"metadata": meta,
|
| 146 |
+
}
|
| 147 |
+
enriched_lines.append(json.dumps(obj_with_meta, ensure_ascii=False))
|
| 148 |
|
| 149 |
+
payload = "\n".join(enriched_lines).encode("utf-8")
|
| 150 |
+
|
| 151 |
+
files = {"file": ("contrib.jsonl", payload, "application/jsonl")}
|
| 152 |
+
try:
|
| 153 |
+
resp = requests.post(f"{BACKEND_URL}/upload-jsonl", files=files)
|
| 154 |
+
if resp.ok:
|
| 155 |
+
st.success("已成功送交後端伺服器,等待後端進一步檢查與處理。")
|
| 156 |
+
else:
|
| 157 |
+
st.error(f"後端回傳錯誤:{resp.status_code} {resp.text}")
|
| 158 |
+
except Exception as e:
|
| 159 |
+
st.error(f"送出時發生錯誤:{e}")
|
| 160 |
|
| 161 |
|
| 162 |
# ---------- Tab 2: PDF ----------
|
|
|
|
| 210 |
|
| 211 |
any_valid_pdf = any(tlen >= 100 for _, tlen in pdf_results) if pdf_results else False
|
| 212 |
|
| 213 |
+
if st.button("上傳 PDF 檔案", disabled=not (pdf_files and any_valid_pdf or BACKEND_URL is None)):
|
| 214 |
if BACKEND_URL is None:
|
| 215 |
st.warning("尚未設定 BACKEND_URL,無法實際送出,請在 `st.secrets` 中配置。")
|
| 216 |
else:
|
| 217 |
+
if not pdf_results:
|
| 218 |
+
st.warning("沒有可上傳的 PDF 檔案。")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 219 |
else:
|
| 220 |
+
with st.spinner("正在上傳 PDF 並進行檢查,請稍候…"):
|
| 221 |
+
files = []
|
| 222 |
+
for pdf, text_len in pdf_results:
|
| 223 |
+
if text_len < 100:
|
| 224 |
+
continue # 跳過疑似掃描檔
|
| 225 |
+
pdf.seek(0)
|
| 226 |
+
files.append(("files", (pdf.name, pdf.getvalue(), "application/pdf")))
|
| 227 |
+
|
| 228 |
+
if not files:
|
| 229 |
+
st.warning("沒有通過文字檢查的 PDF 檔案可送出。")
|
|
|
|
| 230 |
else:
|
| 231 |
+
# PDF 部分的 metadata 用 form data 一起送出,讓後端可以記錄
|
| 232 |
+
meta = {
|
| 233 |
+
"uploaded_at": uploaded_at,
|
| 234 |
+
"contributor_email": contributor_email if contributor_email.strip() else "",
|
| 235 |
+
"share_permission": json.dumps(bool(share_permission)),
|
| 236 |
+
}
|
| 237 |
+
|
| 238 |
+
try:
|
| 239 |
+
resp = requests.post(f"{BACKEND_URL}/upload-pdf", files=files, data=meta)
|
| 240 |
+
if resp.ok:
|
| 241 |
+
st.success("已成功送交後端伺服器,等待後端進一步檢查與處理。")
|
| 242 |
+
else:
|
| 243 |
+
st.error(f"後端回傳錯誤:{resp.status_code} {resp.text}")
|
| 244 |
+
except Exception as e:
|
| 245 |
+
st.error(f"送出時發生錯誤:{e}")
|