Spaces:
Sleeping
Sleeping
add feature for custom fields
Browse files
app.py
CHANGED
@@ -5,7 +5,8 @@ import yake
|
|
5 |
|
6 |
demo = gr.Blocks()
|
7 |
|
8 |
-
def generate(input, attempted_items, offset):
|
|
|
9 |
# Step 1: Import
|
10 |
|
11 |
pdfFileObj = open(input.name, 'rb')
|
@@ -34,6 +35,11 @@ def generate(input, attempted_items, offset):
|
|
34 |
for kw in keywords:
|
35 |
kw_list.append(kw[0])
|
36 |
|
|
|
|
|
|
|
|
|
|
|
37 |
# Step 3: Process for Assignment
|
38 |
output = {}
|
39 |
for kw in kw_list:
|
@@ -110,10 +116,16 @@ def offset_changer(offset_input):
|
|
110 |
offset: int(offset_input)
|
111 |
}
|
112 |
|
|
|
|
|
|
|
|
|
|
|
113 |
with demo:
|
114 |
|
115 |
attempted_items = gr.State(50)
|
116 |
offset = gr.State(0)
|
|
|
117 |
|
118 |
gr.Markdown("# PDF to Index")
|
119 |
|
@@ -133,7 +145,14 @@ with demo:
|
|
133 |
attempted_items_input.change(attempted_items_changer, [attempted_items_input], [attempted_items])
|
134 |
offset_input.change(offset_changer, [offset_input], [offset])
|
135 |
|
136 |
-
gr.HTML("<p><em>Attempted Items is the number of terms intended for index (output may be slightly lower), while Page Offset is a value added to each page number found in the file. In the case of invalid values, Attempted Items will default to 50 and Page Offset will default to 0.</em></p>")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
137 |
|
138 |
gr.Markdown("---")
|
139 |
|
@@ -148,7 +167,7 @@ with demo:
|
|
148 |
|
149 |
convert_button.click(
|
150 |
fn=generate,
|
151 |
-
inputs=[uploaded_file, attempted_items, offset],
|
152 |
outputs=[index],
|
153 |
)
|
154 |
|
|
|
5 |
|
6 |
demo = gr.Blocks()
|
7 |
|
8 |
+
def generate(input, attempted_items, offset, custom):
|
9 |
+
|
10 |
# Step 1: Import
|
11 |
|
12 |
pdfFileObj = open(input.name, 'rb')
|
|
|
35 |
for kw in keywords:
|
36 |
kw_list.append(kw[0])
|
37 |
|
38 |
+
if(len(custom)):
|
39 |
+
split_custom = custom.split(';')
|
40 |
+
for kw in split_custom:
|
41 |
+
kw_list.append(kw)
|
42 |
+
|
43 |
# Step 3: Process for Assignment
|
44 |
output = {}
|
45 |
for kw in kw_list:
|
|
|
116 |
offset: int(offset_input)
|
117 |
}
|
118 |
|
119 |
+
def custom_changer (custom_input):
|
120 |
+
return {
|
121 |
+
custom: custom_input
|
122 |
+
}
|
123 |
+
|
124 |
with demo:
|
125 |
|
126 |
attempted_items = gr.State(50)
|
127 |
offset = gr.State(0)
|
128 |
+
custom = gr.State("")
|
129 |
|
130 |
gr.Markdown("# PDF to Index")
|
131 |
|
|
|
145 |
attempted_items_input.change(attempted_items_changer, [attempted_items_input], [attempted_items])
|
146 |
offset_input.change(offset_changer, [offset_input], [offset])
|
147 |
|
148 |
+
gr.HTML("<p><em>Attempted Items is the number of terms intended for index (output may be slightly lower), while Page Offset is a value added to each page number found in the file. In the case of invalid values, Attempted Items will default to 50 and Page Offset will default to 0. If the fields do not produce expected values, you may be clicking too quickly -- please try again.</em></p>")
|
149 |
+
|
150 |
+
with gr.Row():
|
151 |
+
custom_input = gr.Textbox(value="", show_label=True, label="Custom")
|
152 |
+
custom_input.change(custom_changer, [custom_input], [custom])
|
153 |
+
|
154 |
+
gr.HTML("<p><em>You can add semicolon-separated values in Custom to add custom fields to index. If the fields do not produce expected values, you may be clicking too quickly -- please try again.</em></p>")
|
155 |
+
|
156 |
|
157 |
gr.Markdown("---")
|
158 |
|
|
|
167 |
|
168 |
convert_button.click(
|
169 |
fn=generate,
|
170 |
+
inputs=[uploaded_file, attempted_items, offset, custom],
|
171 |
outputs=[index],
|
172 |
)
|
173 |
|