Spaces:
Sleeping
Sleeping
ankur-bohra
commited on
Commit
•
52cab32
1
Parent(s):
924cad9
Add more logs
Browse files
app.py
CHANGED
@@ -80,23 +80,27 @@ def submit(input_file, old_text):
|
|
80 |
if not input_file:
|
81 |
gr.Error("Please upload a file to continue!")
|
82 |
return gr.Textbox.update()
|
83 |
-
|
|
|
84 |
# Send change to preprocessed image or to extracted text
|
85 |
if input_file.name.endswith(".pdf"):
|
86 |
text = process_pdf(Path(input_file.name), extract_only=True)
|
87 |
else:
|
88 |
text = process_image(Path(input_file.name), extract_only=True)
|
|
|
89 |
return text
|
90 |
|
91 |
|
92 |
def categorize_extracted_text(extracted_text):
|
93 |
category = categories.categorize_text(extracted_text)
|
|
|
94 |
# gr.Info(f"Recognized category: {category}")
|
95 |
return category
|
96 |
|
97 |
|
98 |
-
def
|
99 |
if not category:
|
|
|
100 |
return (
|
101 |
gr.Chatbot.update(None),
|
102 |
gr.JSON.update(None),
|
@@ -104,7 +108,7 @@ def extract_from_category(category, extracted_text):
|
|
104 |
gr.Button.update(interactive=False),
|
105 |
)
|
106 |
category = Category[category]
|
107 |
-
print("
|
108 |
chain = categories.category_modules[category].chain
|
109 |
formatted_prompt = chain.prompt.format_prompt(
|
110 |
text=extracted_text,
|
@@ -122,7 +126,9 @@ def extract_from_category(category, extracted_text):
|
|
122 |
if len(formatted_prompt.messages) > 1:
|
123 |
question += f"**System:**\n{formatted_prompt.messages[1].content}"
|
124 |
question += f"\n\n**Human:**\n{formatted_prompt.messages[0].content}"
|
|
|
125 |
answer = result.generations[0][0].text
|
|
|
126 |
try:
|
127 |
information = chain.output_parser.parse_with_prompt(answer, formatted_prompt)
|
128 |
information = information.json() if information else {}
|
@@ -276,7 +282,7 @@ with gr.Blocks(title="Automatic Reimbursement Tool Demo") as page:
|
|
276 |
[category],
|
277 |
)
|
278 |
category.change(
|
279 |
-
|
280 |
[category, extracted_text],
|
281 |
[chatbot, information, flag_incorrect_button, flag_irrelevant_button],
|
282 |
)
|
|
|
80 |
if not input_file:
|
81 |
gr.Error("Please upload a file to continue!")
|
82 |
return gr.Textbox.update()
|
83 |
+
print("-"*5)
|
84 |
+
print("New input")
|
85 |
# Send change to preprocessed image or to extracted text
|
86 |
if input_file.name.endswith(".pdf"):
|
87 |
text = process_pdf(Path(input_file.name), extract_only=True)
|
88 |
else:
|
89 |
text = process_image(Path(input_file.name), extract_only=True)
|
90 |
+
print("Extracted text")
|
91 |
return text
|
92 |
|
93 |
|
94 |
def categorize_extracted_text(extracted_text):
|
95 |
category = categories.categorize_text(extracted_text)
|
96 |
+
print("Recognized category:", category)
|
97 |
# gr.Info(f"Recognized category: {category}")
|
98 |
return category
|
99 |
|
100 |
|
101 |
+
def parse_from_category(category, extracted_text):
|
102 |
if not category:
|
103 |
+
print("Updated with no category:", category)
|
104 |
return (
|
105 |
gr.Chatbot.update(None),
|
106 |
gr.JSON.update(None),
|
|
|
108 |
gr.Button.update(interactive=False),
|
109 |
)
|
110 |
category = Category[category]
|
111 |
+
print("Parsing text from", category)
|
112 |
chain = categories.category_modules[category].chain
|
113 |
formatted_prompt = chain.prompt.format_prompt(
|
114 |
text=extracted_text,
|
|
|
126 |
if len(formatted_prompt.messages) > 1:
|
127 |
question += f"**System:**\n{formatted_prompt.messages[1].content}"
|
128 |
question += f"\n\n**Human:**\n{formatted_prompt.messages[0].content}"
|
129 |
+
print("\tConstructed prompt")
|
130 |
answer = result.generations[0][0].text
|
131 |
+
print("\tProcessed text")
|
132 |
try:
|
133 |
information = chain.output_parser.parse_with_prompt(answer, formatted_prompt)
|
134 |
information = information.json() if information else {}
|
|
|
282 |
[category],
|
283 |
)
|
284 |
category.change(
|
285 |
+
parse_from_category,
|
286 |
[category, extracted_text],
|
287 |
[chatbot, information, flag_incorrect_button, flag_irrelevant_button],
|
288 |
)
|