Spaces:
Running
Running
Commit
·
c289d70
1
Parent(s):
55f9c40
progress more 17
Browse files
app.py
CHANGED
@@ -109,16 +109,21 @@ def process_file(uploaded_file):
|
|
109 |
df = pd.read_excel(uploaded_file, sheet_name='Публикации')
|
110 |
|
111 |
# Apply fuzzy deduplication
|
112 |
-
|
113 |
-
|
114 |
-
|
|
|
115 |
# Translate texts
|
116 |
translated_texts = []
|
117 |
progress_bar = st.progress(0)
|
|
|
|
|
|
|
118 |
for i, text in enumerate(df['Выдержки из текста']):
|
119 |
translated_text = translate(str(text))
|
120 |
translated_texts.append(translated_text)
|
121 |
progress_bar.progress((i + 1) / len(df))
|
|
|
122 |
|
123 |
# Perform sentiment analysis
|
124 |
vader_results = [get_vader_sentiment(text) for text in translated_texts]
|
@@ -139,9 +144,9 @@ def process_file(uploaded_file):
|
|
139 |
return df
|
140 |
|
141 |
def main():
|
142 |
-
st.title("... приступим к анализу... версия
|
143 |
|
144 |
-
uploaded_file = st.file_uploader("
|
145 |
|
146 |
if uploaded_file is not None:
|
147 |
df = process_file(uploaded_file)
|
|
|
109 |
df = pd.read_excel(uploaded_file, sheet_name='Публикации')
|
110 |
|
111 |
# Apply fuzzy deduplication
|
112 |
+
df = df.groupby('Объект').apply(
|
113 |
+
lambda x: fuzzy_deduplicate(x, 'Выдержки из текста', 65)
|
114 |
+
).reset_index(drop=True)
|
115 |
+
|
116 |
# Translate texts
|
117 |
translated_texts = []
|
118 |
progress_bar = st.progress(0)
|
119 |
+
rogress_text = st.empty()
|
120 |
+
total_news = len(df)
|
121 |
+
|
122 |
for i, text in enumerate(df['Выдержки из текста']):
|
123 |
translated_text = translate(str(text))
|
124 |
translated_texts.append(translated_text)
|
125 |
progress_bar.progress((i + 1) / len(df))
|
126 |
+
progress_text.text(f"{i + 1} из {total_news} сообщений переведено")
|
127 |
|
128 |
# Perform sentiment analysis
|
129 |
vader_results = [get_vader_sentiment(text) for text in translated_texts]
|
|
|
144 |
return df
|
145 |
|
146 |
def main():
|
147 |
+
st.title("... приступим к анализу... версия 17")
|
148 |
|
149 |
+
uploaded_file = st.file_uploader("Выбирайте Excel-файл", type="xlsx")
|
150 |
|
151 |
if uploaded_file is not None:
|
152 |
df = process_file(uploaded_file)
|