Update app.py
Browse files
app.py
CHANGED
@@ -1,58 +1,59 @@
|
|
1 |
# app.py
|
2 |
import json
|
3 |
import traceback
|
4 |
-
import tempfile
|
5 |
import os
|
6 |
import gradio as gr
|
7 |
-
from typing import Tuple, Optional, Any
|
8 |
|
9 |
# Import config first to ensure JAVA_HOME is set early
|
10 |
import config
|
11 |
|
12 |
from main_analyzer import analyze_pdf
|
13 |
-
#
|
14 |
-
|
15 |
|
16 |
-
def process_upload(file_data_binary: Optional[Any]) -> Tuple[str, Optional[str]]: # Use Optional[Any] for Gradio File type="binary"
|
17 |
-
if file_data_binary is None or not hasattr(file_data_binary, 'read'): # Check if it's a file-like object
|
18 |
-
# Gradio's binary type for gr.File returns a tempfile._TemporaryFileWrapper object
|
19 |
-
# If it's None, no file was uploaded.
|
20 |
-
# If it's not None but doesn't have 'read', it's an unexpected type.
|
21 |
-
# However, gradio usually passes the bytes directly if type="binary" was used in older versions
|
22 |
-
# or a TemporaryFileWrapper which is file-like.
|
23 |
-
# For robustness, let's check if it's bytes.
|
24 |
-
if isinstance(file_data_binary, bytes):
|
25 |
-
pass # Good, it's bytes
|
26 |
-
elif file_data_binary is None:
|
27 |
-
return json.dumps({"error": "No file uploaded or file data is None"}, indent=2), None
|
28 |
-
elif not hasattr(file_data_binary, 'read'): # It's not None, not bytes, not file-like
|
29 |
-
return json.dumps({"error": f"Unexpected file data type: {type(file_data_binary)}"}), None
|
30 |
-
# If it has 'read', it's a file-like object, proceed.
|
31 |
|
32 |
-
|
33 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
34 |
try:
|
35 |
-
|
36 |
-
#
|
37 |
-
#
|
38 |
-
|
39 |
-
# Gradio with type="binary" gives a tempfile._TemporaryFileWrapper.
|
40 |
-
# This object is already file-like and can be passed directly.
|
41 |
-
# No need to create another temp file here in app.py if main_analyzer handles it.
|
42 |
|
43 |
-
|
44 |
-
results_dict, _ = analyze_pdf(file_data_binary)
|
45 |
|
|
|
|
|
|
|
|
|
|
|
|
|
46 |
results_json = json.dumps(results_dict, indent=2, ensure_ascii=False)
|
47 |
return results_json, None
|
48 |
|
49 |
except Exception as e:
|
50 |
-
|
51 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
52 |
return error_message, None
|
53 |
-
# No
|
54 |
-
#
|
55 |
-
|
56 |
|
57 |
def create_interface():
|
58 |
with gr.Blocks(title="PDF Analyzer") as interface:
|
@@ -67,8 +68,9 @@ def create_interface():
|
|
67 |
with gr.Row():
|
68 |
file_input = gr.File(
|
69 |
label="Upload PDF",
|
70 |
-
file_types=[".pdf"]
|
71 |
-
# type
|
|
|
72 |
)
|
73 |
|
74 |
with gr.Row():
|
@@ -76,7 +78,7 @@ def create_interface():
|
|
76 |
|
77 |
with gr.Row():
|
78 |
results_output = gr.JSON(
|
79 |
-
label="Analysis Results",
|
80 |
show_label=True
|
81 |
)
|
82 |
|
@@ -84,7 +86,7 @@ def create_interface():
|
|
84 |
pdf_output = gr.File(
|
85 |
label="Annotated PDF (Placeholder - View Coordinates in JSON)",
|
86 |
show_label=True,
|
87 |
-
interactive=False
|
88 |
)
|
89 |
|
90 |
analyze_btn.click(
|
@@ -98,18 +100,15 @@ if __name__ == "__main__":
|
|
98 |
print("\n--- Launching Gradio Interface ---")
|
99 |
# config.set_java_home() is called when config.py is imported.
|
100 |
|
101 |
-
# Optional: Test LanguageTool initialization
|
102 |
try:
|
103 |
-
import language_tool_python # Import here for the test
|
104 |
lt_test = language_tool_python.LanguageTool('en-US')
|
105 |
lt_test.close()
|
106 |
print("App: LanguageTool initialized successfully for test.")
|
107 |
except Exception as lt_e:
|
108 |
-
print(f"App: Warning: Could not initialize LanguageTool for test. Language checks might fail: {lt_e}")
|
109 |
print("Please ensure Java is installed and JAVA_HOME is correctly set (see config.py).")
|
110 |
|
111 |
app_interface = create_interface()
|
112 |
app_interface.launch(
|
113 |
-
share=False,
|
114 |
-
# server_port=7860 # Optionally specify a port
|
115 |
)
|
|
|
1 |
# app.py
|
2 |
import json
|
3 |
import traceback
|
4 |
+
import tempfile # Not strictly needed by process_upload anymore, but good to keep if other parts use it.
|
5 |
import os
|
6 |
import gradio as gr
|
7 |
+
from typing import Tuple, Optional, Any
|
8 |
|
9 |
# Import config first to ensure JAVA_HOME is set early
|
10 |
import config
|
11 |
|
12 |
from main_analyzer import analyze_pdf
|
13 |
+
# language_tool_python needed for the test in __main__
|
14 |
+
import language_tool_python
|
15 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
16 |
|
17 |
+
def process_upload(uploaded_file_input: Optional[Any]) -> Tuple[str, Optional[str]]:
|
18 |
+
"""
|
19 |
+
Processes the uploaded file from Gradio.
|
20 |
+
'uploaded_file_input' is expected to be an object from gr.File(),
|
21 |
+
which could be a path string (like gradio.utils.NamedString) or a file-like object.
|
22 |
+
"""
|
23 |
+
if uploaded_file_input is None:
|
24 |
+
print("App: No file uploaded.")
|
25 |
+
return json.dumps({"error": "No file uploaded."}, indent=2), None
|
26 |
+
|
27 |
try:
|
28 |
+
# Pass the uploaded_file_input directly to analyze_pdf.
|
29 |
+
# analyze_pdf is responsible for determining if it's a path or a stream
|
30 |
+
# and handling it accordingly (including creating a temp file for streams if needed).
|
31 |
+
print(f"App: Received file input of type: {type(uploaded_file_input)}. Passing to analyzer.")
|
|
|
|
|
|
|
32 |
|
33 |
+
results_dict, _ = analyze_pdf(uploaded_file_input)
|
|
|
34 |
|
35 |
+
# Check if analyze_pdf returned an error (it returns a dict with "error" key in that case)
|
36 |
+
if isinstance(results_dict, dict) and "error" in results_dict:
|
37 |
+
print(f"App: Analysis returned an error: {results_dict['error']}")
|
38 |
+
# Return the error JSON directly
|
39 |
+
return json.dumps(results_dict, indent=2, ensure_ascii=False), None
|
40 |
+
|
41 |
results_json = json.dumps(results_dict, indent=2, ensure_ascii=False)
|
42 |
return results_json, None
|
43 |
|
44 |
except Exception as e:
|
45 |
+
# This catches unexpected errors during the call to analyze_pdf or JSON dumping.
|
46 |
+
print(f"App: Error in process_upload: {e}\n{traceback.format_exc()}")
|
47 |
+
error_detail = str(e)
|
48 |
+
# If the error is due to an invalid input type that analyze_pdf couldn't handle:
|
49 |
+
if "Invalid PDF input type" in error_detail: # Check for specific error from analyze_pdf
|
50 |
+
error_detail = f"Invalid PDF input type received from uploader: {type(uploaded_file_input)}. Details: {str(e)}"
|
51 |
+
|
52 |
+
error_message = json.dumps({"error": error_detail, "traceback": traceback.format_exc()}, indent=2)
|
53 |
return error_message, None
|
54 |
+
# No explicit temp file cleanup needed here in process_upload for the Gradio-provided file object.
|
55 |
+
# Gradio manages its own temporary files.
|
56 |
+
# analyze_pdf manages any internal temporary files it creates.
|
57 |
|
58 |
def create_interface():
|
59 |
with gr.Blocks(title="PDF Analyzer") as interface:
|
|
|
68 |
with gr.Row():
|
69 |
file_input = gr.File(
|
70 |
label="Upload PDF",
|
71 |
+
file_types=[".pdf"]
|
72 |
+
# Default type: Gradio provides a path-like object (e.g., NamedString)
|
73 |
+
# or a TemporaryFileWrapper. Both should be acceptable by analyze_pdf.
|
74 |
)
|
75 |
|
76 |
with gr.Row():
|
|
|
78 |
|
79 |
with gr.Row():
|
80 |
results_output = gr.JSON(
|
81 |
+
label="Analysis Results",
|
82 |
show_label=True
|
83 |
)
|
84 |
|
|
|
86 |
pdf_output = gr.File(
|
87 |
label="Annotated PDF (Placeholder - View Coordinates in JSON)",
|
88 |
show_label=True,
|
89 |
+
interactive=False
|
90 |
)
|
91 |
|
92 |
analyze_btn.click(
|
|
|
100 |
print("\n--- Launching Gradio Interface ---")
|
101 |
# config.set_java_home() is called when config.py is imported.
|
102 |
|
|
|
103 |
try:
|
|
|
104 |
lt_test = language_tool_python.LanguageTool('en-US')
|
105 |
lt_test.close()
|
106 |
print("App: LanguageTool initialized successfully for test.")
|
107 |
except Exception as lt_e:
|
108 |
+
print(f"App: Warning: Could not initialize LanguageTool for test. Language checks might fail: {lt_e}\n{traceback.format_exc(limit=1)}")
|
109 |
print("Please ensure Java is installed and JAVA_HOME is correctly set (see config.py).")
|
110 |
|
111 |
app_interface = create_interface()
|
112 |
app_interface.launch(
|
113 |
+
share=False,
|
|
|
114 |
)
|