Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -1,25 +1,17 @@
|
|
| 1 |
-
# app.py
|
| 2 |
import os
|
| 3 |
import pandas as pd
|
| 4 |
import gradio as gr
|
| 5 |
from huggingface_hub import hf_hub_download
|
| 6 |
|
| 7 |
# ----------------------------------------------------
|
| 8 |
-
#
|
| 9 |
# ----------------------------------------------------
|
| 10 |
from searchschool import (
|
| 11 |
DEFAULT_STATE_KEY,
|
| 12 |
-
on_search_web,
|
| 13 |
load_master_for_state,
|
| 14 |
update_blocks,
|
| 15 |
search_candidates,
|
| 16 |
-
|
| 17 |
-
|
| 18 |
-
from admin_patterns import (
|
| 19 |
-
load_pattern_config,
|
| 20 |
-
load_global_patterns_for_editor,
|
| 21 |
-
load_state_patterns_for_editor,
|
| 22 |
-
save_global_patterns_from_editor,
|
| 23 |
)
|
| 24 |
|
| 25 |
# ----------------------------------------------------
|
|
@@ -29,15 +21,12 @@ HF_SCHOOLS_DATASET = "Apf-AI4Good/Schools"
|
|
| 29 |
MASTER_ALL_STATES_FILE = "master_all_states.csv"
|
| 30 |
MASTER_STATE_COL = "School_State__c"
|
| 31 |
|
| 32 |
-
# ----------------------------------------------------
|
| 33 |
-
# Load ADMIN password
|
| 34 |
-
# ----------------------------------------------------
|
| 35 |
ADMIN_PASSWORD = os.getenv("ADMIN_PASSWORD", None)
|
| 36 |
if not ADMIN_PASSWORD:
|
| 37 |
-
print("β ADMIN_PASSWORD
|
| 38 |
|
| 39 |
# ----------------------------------------------------
|
| 40 |
-
#
|
| 41 |
# ----------------------------------------------------
|
| 42 |
def load_state_choices_once():
|
| 43 |
path = hf_hub_download(
|
|
@@ -45,118 +34,132 @@ def load_state_choices_once():
|
|
| 45 |
repo_type="dataset",
|
| 46 |
filename=MASTER_ALL_STATES_FILE,
|
| 47 |
)
|
| 48 |
-
|
| 49 |
df = pd.read_csv(path, dtype=str)
|
|
|
|
| 50 |
|
| 51 |
-
|
| 52 |
-
df[MASTER_STATE_COL]
|
| 53 |
-
.dropna()
|
| 54 |
-
.str.upper()
|
| 55 |
-
.unique()
|
| 56 |
-
.tolist()
|
| 57 |
-
)
|
| 58 |
|
| 59 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 60 |
|
| 61 |
|
| 62 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 63 |
|
| 64 |
# ----------------------------------------------------
|
| 65 |
# Gradio App
|
| 66 |
# ----------------------------------------------------
|
| 67 |
-
with gr.Blocks(title="State School Fuzzy Matcher
|
| 68 |
gr.Markdown(
|
| 69 |
"""
|
| 70 |
# State School Fuzzy Matcher (RapidFuzz)
|
| 71 |
|
| 72 |
-
|
| 73 |
-
|
|
|
|
| 74 |
"""
|
| 75 |
)
|
| 76 |
|
| 77 |
-
# ================= SEARCH
|
| 78 |
-
with gr.
|
| 79 |
-
|
| 80 |
-
|
| 81 |
-
|
| 82 |
-
|
| 83 |
-
value=DEFAULT_STATE_KEY if DEFAULT_STATE_KEY in STATE_CHOICES else None,
|
| 84 |
-
interactive=True,
|
| 85 |
-
)
|
| 86 |
-
|
| 87 |
-
with gr.Row():
|
| 88 |
-
district_dd = gr.Dropdown(label="District", choices=[], value=None, interactive=True)
|
| 89 |
-
block_dd = gr.Dropdown(label="Block", choices=[], value=None, interactive=True)
|
| 90 |
-
|
| 91 |
-
school_input = gr.Textbox(
|
| 92 |
-
label="Input School Name",
|
| 93 |
-
placeholder="Type school name from marksheet..."
|
| 94 |
-
)
|
| 95 |
-
|
| 96 |
-
search_btn = gr.Button("Find Candidates")
|
| 97 |
-
|
| 98 |
-
gr.Markdown("### Candidates (top matches)")
|
| 99 |
-
candidates_table = gr.Dataframe(
|
| 100 |
-
headers=["School_Name", "State", "District", "Block", "UDISE_Code", "Score"],
|
| 101 |
-
datatype=["str", "str", "str", "str", "str", "number"],
|
| 102 |
-
interactive=False,
|
| 103 |
)
|
| 104 |
|
| 105 |
-
|
| 106 |
-
|
| 107 |
-
|
| 108 |
-
datatype=["str", "str", "str", "str", "str", "number"],
|
| 109 |
-
interactive=False,
|
| 110 |
-
)
|
| 111 |
-
|
| 112 |
-
# ================= ADMIN TAB =================
|
| 113 |
-
with gr.Tab("Admin"):
|
| 114 |
-
gr.Markdown("### Admin β Patterns Editor")
|
| 115 |
|
| 116 |
-
|
| 117 |
-
|
| 118 |
-
|
| 119 |
-
|
| 120 |
-
)
|
| 121 |
-
unlock_btn = gr.Button("Unlock Admin Panel")
|
| 122 |
-
unlock_status = gr.Markdown("")
|
| 123 |
-
|
| 124 |
-
admin_panel = gr.Column(visible=False)
|
| 125 |
-
|
| 126 |
-
with admin_panel:
|
| 127 |
-
global_df = gr.Dataframe(
|
| 128 |
-
value=load_global_patterns_for_editor(),
|
| 129 |
-
headers=["pattern", "replacement"],
|
| 130 |
-
datatype=["str", "str"],
|
| 131 |
-
interactive=True,
|
| 132 |
-
row_count=5,
|
| 133 |
-
)
|
| 134 |
|
| 135 |
-
|
| 136 |
-
global_status = gr.Markdown("")
|
| 137 |
|
| 138 |
-
|
|
|
|
|
|
|
| 139 |
|
| 140 |
-
|
| 141 |
-
|
| 142 |
-
|
| 143 |
-
|
| 144 |
-
|
| 145 |
-
)
|
| 146 |
-
new_state_tb = gr.Textbox(
|
| 147 |
-
label="Or type new state key",
|
| 148 |
-
placeholder="e.g. KARNATAKA",
|
| 149 |
-
)
|
| 150 |
|
| 151 |
-
|
| 152 |
-
|
| 153 |
-
|
| 154 |
-
|
| 155 |
-
|
| 156 |
-
|
| 157 |
-
|
| 158 |
|
| 159 |
-
|
|
|
|
| 160 |
|
| 161 |
# ================= Wiring =================
|
| 162 |
state_dd.change(
|
|
@@ -165,12 +168,6 @@ with gr.Blocks(title="State School Fuzzy Matcher (with gated Admin)") as demo:
|
|
| 165 |
outputs=[district_dd, block_dd],
|
| 166 |
)
|
| 167 |
|
| 168 |
-
demo.load(
|
| 169 |
-
fn=load_master_for_state,
|
| 170 |
-
inputs=state_dd,
|
| 171 |
-
outputs=[district_dd, block_dd],
|
| 172 |
-
)
|
| 173 |
-
|
| 174 |
district_dd.change(
|
| 175 |
fn=update_blocks,
|
| 176 |
inputs=district_dd,
|
|
@@ -178,56 +175,22 @@ with gr.Blocks(title="State School Fuzzy Matcher (with gated Admin)") as demo:
|
|
| 178 |
)
|
| 179 |
|
| 180 |
search_btn.click(
|
| 181 |
-
fn=
|
| 182 |
-
inputs=[school_input, state_dd, district_dd, block_dd],
|
| 183 |
-
outputs=[candidates_table, best_table],
|
| 184 |
-
)
|
| 185 |
-
|
| 186 |
-
# ensure at top of file you have: from web_search import tavily_search
|
| 187 |
-
|
| 188 |
-
# --- inside Search tab UI, after search_btn etc. ---
|
| 189 |
-
gr.Markdown("### Web search β find UDISE codes on the web")
|
| 190 |
-
gr.Markdown("### Use Judiciously - Only if you didnt find or couldnt confirm the school in the above list")
|
| 191 |
-
search_web_btn = gr.Button("Search Web")
|
| 192 |
-
|
| 193 |
-
web_schools_table = gr.Dataframe(
|
| 194 |
-
headers=["School_Name", "State", "District", "Block", "UDISE_Code"],
|
| 195 |
-
datatype=["str", "str", "str", "str", "str"],
|
| 196 |
-
interactive=False
|
| 197 |
-
)
|
| 198 |
-
|
| 199 |
-
search_web_btn.click(
|
| 200 |
-
fn=on_search_web,
|
| 201 |
inputs=[school_input, state_dd, district_dd, block_dd],
|
| 202 |
-
outputs=
|
| 203 |
)
|
| 204 |
|
| 205 |
-
# ---------------- Admin gating ----------------
|
| 206 |
-
def try_unlock(pwd: str):
|
| 207 |
-
if ADMIN_PASSWORD is None:
|
| 208 |
-
return gr.update(visible=False), "β ADMIN_PASSWORD not configured", None
|
| 209 |
-
if pwd == ADMIN_PASSWORD:
|
| 210 |
-
return gr.update(visible=True), "π Admin panel unlocked.", True
|
| 211 |
-
return gr.update(visible=False), "β Wrong password.", False
|
| 212 |
-
|
| 213 |
unlock_btn.click(
|
| 214 |
-
fn=
|
| 215 |
inputs=unlock_pwd,
|
| 216 |
-
outputs=[
|
| 217 |
-
)
|
| 218 |
-
|
| 219 |
-
save_global_btn.click(
|
| 220 |
-
fn=lambda df, pwd: save_global_patterns_from_editor(df, pwd, ADMIN_PASSWORD),
|
| 221 |
-
inputs=[global_df, unlock_pwd],
|
| 222 |
-
outputs=global_status,
|
| 223 |
)
|
| 224 |
|
| 225 |
-
|
| 226 |
-
fn=
|
| 227 |
-
inputs=[
|
| 228 |
-
outputs=
|
| 229 |
)
|
| 230 |
|
| 231 |
-
|
| 232 |
if __name__ == "__main__":
|
| 233 |
demo.launch()
|
|
|
|
|
|
|
| 1 |
import os
|
| 2 |
import pandas as pd
|
| 3 |
import gradio as gr
|
| 4 |
from huggingface_hub import hf_hub_download
|
| 5 |
|
| 6 |
# ----------------------------------------------------
|
| 7 |
+
# Imports
|
| 8 |
# ----------------------------------------------------
|
| 9 |
from searchschool import (
|
| 10 |
DEFAULT_STATE_KEY,
|
|
|
|
| 11 |
load_master_for_state,
|
| 12 |
update_blocks,
|
| 13 |
search_candidates,
|
| 14 |
+
on_search_web,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 15 |
)
|
| 16 |
|
| 17 |
# ----------------------------------------------------
|
|
|
|
| 21 |
MASTER_ALL_STATES_FILE = "master_all_states.csv"
|
| 22 |
MASTER_STATE_COL = "School_State__c"
|
| 23 |
|
|
|
|
|
|
|
|
|
|
| 24 |
ADMIN_PASSWORD = os.getenv("ADMIN_PASSWORD", None)
|
| 25 |
if not ADMIN_PASSWORD:
|
| 26 |
+
print("β ADMIN_PASSWORD not set")
|
| 27 |
|
| 28 |
# ----------------------------------------------------
|
| 29 |
+
# Load states ONCE
|
| 30 |
# ----------------------------------------------------
|
| 31 |
def load_state_choices_once():
|
| 32 |
path = hf_hub_download(
|
|
|
|
| 34 |
repo_type="dataset",
|
| 35 |
filename=MASTER_ALL_STATES_FILE,
|
| 36 |
)
|
|
|
|
| 37 |
df = pd.read_csv(path, dtype=str)
|
| 38 |
+
return sorted(df[MASTER_STATE_COL].dropna().str.upper().unique().tolist())
|
| 39 |
|
| 40 |
+
STATE_CHOICES = load_state_choices_once()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 41 |
|
| 42 |
+
# ----------------------------------------------------
|
| 43 |
+
# UI RENDER HELPERS
|
| 44 |
+
# ----------------------------------------------------
|
| 45 |
+
def render_header(show_score=True):
|
| 46 |
+
with gr.Row():
|
| 47 |
+
gr.Markdown("**School Name**")
|
| 48 |
+
gr.Markdown("**UDISE Code**")
|
| 49 |
+
gr.Markdown("**State**")
|
| 50 |
+
gr.Markdown("**District**")
|
| 51 |
+
gr.Markdown("**Block**")
|
| 52 |
+
if show_score:
|
| 53 |
+
gr.Markdown("**Score**")
|
| 54 |
|
| 55 |
|
| 56 |
+
def render_results(rows, show_score=True):
|
| 57 |
+
comps = []
|
| 58 |
+
|
| 59 |
+
if not rows:
|
| 60 |
+
comps.append(gr.Markdown("β No results found."))
|
| 61 |
+
return comps
|
| 62 |
+
|
| 63 |
+
render_header(show_score)
|
| 64 |
+
|
| 65 |
+
for r in rows:
|
| 66 |
+
with gr.Row():
|
| 67 |
+
comps.append(
|
| 68 |
+
gr.Textbox(
|
| 69 |
+
r.get("School_Name", ""),
|
| 70 |
+
interactive=False,
|
| 71 |
+
show_copy_button=True,
|
| 72 |
+
container=False,
|
| 73 |
+
)
|
| 74 |
+
)
|
| 75 |
+
comps.append(
|
| 76 |
+
gr.Textbox(
|
| 77 |
+
r.get("UDISE_Code", ""),
|
| 78 |
+
interactive=False,
|
| 79 |
+
show_copy_button=True,
|
| 80 |
+
container=False,
|
| 81 |
+
)
|
| 82 |
+
)
|
| 83 |
+
comps.append(gr.Markdown(r.get("State", "")))
|
| 84 |
+
comps.append(gr.Markdown(r.get("District", "")))
|
| 85 |
+
comps.append(gr.Markdown(r.get("Block", "")))
|
| 86 |
+
if show_score:
|
| 87 |
+
comps.append(gr.Markdown(f"{r.get('Score', 0):.2f}"))
|
| 88 |
+
|
| 89 |
+
return comps
|
| 90 |
+
|
| 91 |
+
# ----------------------------------------------------
|
| 92 |
+
# Wrapped search functions
|
| 93 |
+
# ----------------------------------------------------
|
| 94 |
+
def rapid_search_and_render(name, state, district, block):
|
| 95 |
+
candidates, _ = search_candidates(name, state, district, block)
|
| 96 |
+
return render_results(candidates, show_score=True)
|
| 97 |
+
|
| 98 |
+
def web_search_and_render(name, state, district, block):
|
| 99 |
+
results = on_search_web(name, state, district, block)
|
| 100 |
+
return render_results(results, show_score=False)
|
| 101 |
+
|
| 102 |
+
# ----------------------------------------------------
|
| 103 |
+
# Unlock Web Search
|
| 104 |
+
# ----------------------------------------------------
|
| 105 |
+
def unlock_web_search(pwd):
|
| 106 |
+
if pwd == ADMIN_PASSWORD:
|
| 107 |
+
return gr.update(interactive=True), "π Web search unlocked."
|
| 108 |
+
return gr.update(interactive=False), "β Invalid password."
|
| 109 |
|
| 110 |
# ----------------------------------------------------
|
| 111 |
# Gradio App
|
| 112 |
# ----------------------------------------------------
|
| 113 |
+
with gr.Blocks(title="State School Fuzzy Matcher") as demo:
|
| 114 |
gr.Markdown(
|
| 115 |
"""
|
| 116 |
# State School Fuzzy Matcher (RapidFuzz)
|
| 117 |
|
| 118 |
+
**Flow**
|
| 119 |
+
1. Use Rapid search first
|
| 120 |
+
2. Unlock Web search only if required
|
| 121 |
"""
|
| 122 |
)
|
| 123 |
|
| 124 |
+
# ================= SEARCH INPUTS =================
|
| 125 |
+
with gr.Row():
|
| 126 |
+
state_dd = gr.Dropdown(
|
| 127 |
+
label="State",
|
| 128 |
+
choices=STATE_CHOICES,
|
| 129 |
+
value=DEFAULT_STATE_KEY if DEFAULT_STATE_KEY in STATE_CHOICES else None,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 130 |
)
|
| 131 |
|
| 132 |
+
with gr.Row():
|
| 133 |
+
district_dd = gr.Dropdown(label="District", interactive=True)
|
| 134 |
+
block_dd = gr.Dropdown(label="Block", interactive=True)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 135 |
|
| 136 |
+
school_input = gr.Textbox(
|
| 137 |
+
label="School Name",
|
| 138 |
+
placeholder="Enter school name from marksheet",
|
| 139 |
+
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 140 |
|
| 141 |
+
search_btn = gr.Button("π Find Candidates")
|
|
|
|
| 142 |
|
| 143 |
+
# ================= RAPID RESULTS =================
|
| 144 |
+
gr.Markdown("### π Rapid Search Results")
|
| 145 |
+
rapid_results_container = gr.Column()
|
| 146 |
|
| 147 |
+
# ================= WEB SEARCH =================
|
| 148 |
+
gr.Markdown("### π Web Search (Locked)")
|
| 149 |
+
gr.Markdown(
|
| 150 |
+
"Use only if Rapid search does not give a confident match."
|
| 151 |
+
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 152 |
|
| 153 |
+
unlock_pwd = gr.Textbox(
|
| 154 |
+
label="Admin Password",
|
| 155 |
+
type="password",
|
| 156 |
+
placeholder="Enter password to unlock web search",
|
| 157 |
+
)
|
| 158 |
+
unlock_btn = gr.Button("Unlock Web Search")
|
| 159 |
+
web_status = gr.Markdown("")
|
| 160 |
|
| 161 |
+
search_web_btn = gr.Button("Search Web", interactive=False)
|
| 162 |
+
web_results_container = gr.Column()
|
| 163 |
|
| 164 |
# ================= Wiring =================
|
| 165 |
state_dd.change(
|
|
|
|
| 168 |
outputs=[district_dd, block_dd],
|
| 169 |
)
|
| 170 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 171 |
district_dd.change(
|
| 172 |
fn=update_blocks,
|
| 173 |
inputs=district_dd,
|
|
|
|
| 175 |
)
|
| 176 |
|
| 177 |
search_btn.click(
|
| 178 |
+
fn=rapid_search_and_render,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 179 |
inputs=[school_input, state_dd, district_dd, block_dd],
|
| 180 |
+
outputs=rapid_results_container,
|
| 181 |
)
|
| 182 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 183 |
unlock_btn.click(
|
| 184 |
+
fn=unlock_web_search,
|
| 185 |
inputs=unlock_pwd,
|
| 186 |
+
outputs=[search_web_btn, web_status],
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 187 |
)
|
| 188 |
|
| 189 |
+
search_web_btn.click(
|
| 190 |
+
fn=web_search_and_render,
|
| 191 |
+
inputs=[school_input, state_dd, district_dd, block_dd],
|
| 192 |
+
outputs=web_results_container,
|
| 193 |
)
|
| 194 |
|
|
|
|
| 195 |
if __name__ == "__main__":
|
| 196 |
demo.launch()
|