Update pages/bot.py
Browse files- pages/bot.py +6 -1
pages/bot.py
CHANGED
|
@@ -37,7 +37,12 @@ def get_pdf_text(folder_path):
|
|
| 37 |
for page in pdf_reader.pages:
|
| 38 |
text += page.extract_text()
|
| 39 |
#text += '\n'
|
| 40 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 41 |
return text
|
| 42 |
|
| 43 |
#Chunks erstellen
|
|
|
|
| 37 |
for page in pdf_reader.pages:
|
| 38 |
text += page.extract_text()
|
| 39 |
#text += '\n'
|
| 40 |
+
translator = pipeline("translation_ger_to_en", model="t5-small")
|
| 41 |
+
text=text.replace("\n", " ")
|
| 42 |
+
text=text.replace("- ", "")
|
| 43 |
+
st.text(text)
|
| 44 |
+
translator(text)
|
| 45 |
+
st.text(text)
|
| 46 |
return text
|
| 47 |
|
| 48 |
#Chunks erstellen
|