Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
|
@@ -92,31 +92,61 @@ def extract_pptx(path):
|
|
| 92 |
return text
|
| 93 |
|
| 94 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 95 |
def mode_load(path):
|
| 96 |
choice = ""
|
| 97 |
-
file_type = path.split(".")[-1]
|
| 98 |
print(file_type)
|
|
|
|
| 99 |
if file_type in ["pdf", "txt", "py", "docx", "pptx", "json", "cpp", "md"]:
|
| 100 |
-
if file_type
|
| 101 |
content = extract_pdf(path)
|
| 102 |
-
elif file_type
|
| 103 |
content = extract_docx(path)
|
| 104 |
-
elif file_type
|
| 105 |
content = extract_pptx(path)
|
| 106 |
else:
|
| 107 |
content = extract_text(path)
|
| 108 |
choice = "doc"
|
| 109 |
print(content[:100])
|
| 110 |
return choice, content[:5000]
|
| 111 |
-
|
| 112 |
-
|
| 113 |
elif file_type in ["png", "jpg", "jpeg", "bmp", "tiff", "webp"]:
|
| 114 |
-
|
| 115 |
-
|
| 116 |
-
|
| 117 |
-
|
|
|
|
|
|
|
|
|
|
| 118 |
else:
|
| 119 |
-
raise
|
| 120 |
|
| 121 |
|
| 122 |
@spaces.GPU()
|
|
|
|
| 92 |
return text
|
| 93 |
|
| 94 |
|
| 95 |
+
# def mode_load(path):
|
| 96 |
+
# choice = ""
|
| 97 |
+
# file_type = path.split(".")[-1]
|
| 98 |
+
# print(file_type)
|
| 99 |
+
# if file_type in ["pdf", "txt", "py", "docx", "pptx", "json", "cpp", "md"]:
|
| 100 |
+
# if file_type.endswith("pdf"):
|
| 101 |
+
# content = extract_pdf(path)
|
| 102 |
+
# elif file_type.endswith("docx"):
|
| 103 |
+
# content = extract_docx(path)
|
| 104 |
+
# elif file_type.endswith("pptx"):
|
| 105 |
+
# content = extract_pptx(path)
|
| 106 |
+
# else:
|
| 107 |
+
# content = extract_text(path)
|
| 108 |
+
# choice = "doc"
|
| 109 |
+
# print(content[:100])
|
| 110 |
+
# return choice, content[:5000]
|
| 111 |
+
|
| 112 |
+
|
| 113 |
+
# elif file_type in ["png", "jpg", "jpeg", "bmp", "tiff", "webp"]:
|
| 114 |
+
# content = Image.open(path).convert('RGB')
|
| 115 |
+
# choice = "image"
|
| 116 |
+
# return choice, content
|
| 117 |
+
|
| 118 |
+
# else:
|
| 119 |
+
# raise gr.Error("Oops, unsupported files.")
|
| 120 |
+
|
| 121 |
+
|
| 122 |
def mode_load(path):
|
| 123 |
choice = ""
|
| 124 |
+
file_type = path.split(".")[-1].lower()
|
| 125 |
print(file_type)
|
| 126 |
+
|
| 127 |
if file_type in ["pdf", "txt", "py", "docx", "pptx", "json", "cpp", "md"]:
|
| 128 |
+
if file_type == "pdf":
|
| 129 |
content = extract_pdf(path)
|
| 130 |
+
elif file_type == "docx":
|
| 131 |
content = extract_docx(path)
|
| 132 |
+
elif file_type == "pptx":
|
| 133 |
content = extract_pptx(path)
|
| 134 |
else:
|
| 135 |
content = extract_text(path)
|
| 136 |
choice = "doc"
|
| 137 |
print(content[:100])
|
| 138 |
return choice, content[:5000]
|
| 139 |
+
|
|
|
|
| 140 |
elif file_type in ["png", "jpg", "jpeg", "bmp", "tiff", "webp"]:
|
| 141 |
+
try:
|
| 142 |
+
content = Image.open(path).convert('RGB')
|
| 143 |
+
choice = "image"
|
| 144 |
+
return choice, content
|
| 145 |
+
except Exception as e:
|
| 146 |
+
raise ValueError(f"Error processing image file: {e}")
|
| 147 |
+
|
| 148 |
else:
|
| 149 |
+
raise ValueError("Oops, unsupported file type.")
|
| 150 |
|
| 151 |
|
| 152 |
@spaces.GPU()
|