Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -142,48 +142,20 @@ def compute_pdf_hash(pdf_path):
|
|
| 142 |
return hashlib.md5(f.read()).hexdigest()
|
| 143 |
|
| 144 |
def get_total_pages(pdf_path):
|
| 145 |
-
"""Get total pages
|
| 146 |
reader = PdfReader(pdf_path)
|
| 147 |
-
# Handle PDFs that are flagged as encrypted but openable without a password
|
| 148 |
-
try:
|
| 149 |
-
if getattr(reader, "is_encrypted", False):
|
| 150 |
-
try:
|
| 151 |
-
reader.decrypt("")
|
| 152 |
-
except Exception:
|
| 153 |
-
pass
|
| 154 |
-
except Exception:
|
| 155 |
-
pass
|
| 156 |
-
|
| 157 |
-
try:
|
| 158 |
-
num_pages = len(reader.pages)
|
| 159 |
-
except Exception:
|
| 160 |
-
# Fallback: iterate until failure (very rare)
|
| 161 |
-
num_pages = 0
|
| 162 |
-
try:
|
| 163 |
-
while True:
|
| 164 |
-
_ = reader.pages[num_pages]
|
| 165 |
-
num_pages += 1
|
| 166 |
-
except Exception:
|
| 167 |
-
pass
|
| 168 |
-
|
| 169 |
documents = []
|
| 170 |
-
for
|
| 171 |
-
|
| 172 |
-
|
| 173 |
-
|
| 174 |
-
text = page.extract_text() or ""
|
| 175 |
-
except Exception:
|
| 176 |
-
text = ""
|
| 177 |
-
|
| 178 |
-
class Doc:
|
| 179 |
def __init__(self, content, page_number):
|
| 180 |
self.content = content
|
| 181 |
self.page_number = page_number
|
| 182 |
-
|
| 183 |
-
|
| 184 |
-
|
| 185 |
-
|
| 186 |
-
return num_pages, documents
|
| 187 |
|
| 188 |
def get_embeddings(text):
|
| 189 |
"""Generate embeddings for text."""
|
|
@@ -928,4 +900,4 @@ if __name__ == '__main__':
|
|
| 928 |
os.makedirs(app.config['UPLOAD_FOLDER'], exist_ok=True)
|
| 929 |
os.makedirs(FAISS_INDEX_DIR, exist_ok=True)
|
| 930 |
port = int(os.getenv("PORT", "7860"))
|
| 931 |
-
app.run(host="0.0.0.0", port=port)
|
|
|
|
| 142 |
return hashlib.md5(f.read()).hexdigest()
|
| 143 |
|
| 144 |
def get_total_pages(pdf_path):
|
| 145 |
+
"""Get total pages and extract text using PyPDF2 for fast processing."""
|
| 146 |
reader = PdfReader(pdf_path)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 147 |
documents = []
|
| 148 |
+
for page_num, page in enumerate(reader.pages, 1):
|
| 149 |
+
text = page.extract_text() or ""
|
| 150 |
+
|
| 151 |
+
class Doc:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 152 |
def __init__(self, content, page_number):
|
| 153 |
self.content = content
|
| 154 |
self.page_number = page_number
|
| 155 |
+
|
| 156 |
+
doc = Doc(text, page_num)
|
| 157 |
+
documents.append(doc)
|
| 158 |
+
return len(documents), documents
|
|
|
|
| 159 |
|
| 160 |
def get_embeddings(text):
|
| 161 |
"""Generate embeddings for text."""
|
|
|
|
| 900 |
os.makedirs(app.config['UPLOAD_FOLDER'], exist_ok=True)
|
| 901 |
os.makedirs(FAISS_INDEX_DIR, exist_ok=True)
|
| 902 |
port = int(os.getenv("PORT", "7860"))
|
| 903 |
+
app.run(host="0.0.0.0", port=port, debug=True)
|