Spaces:
Runtime error
Runtime error
Update app.py
Browse files
app.py
CHANGED
|
@@ -1,3 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
# How to use: YTVideoToText("https://www.youtube.com/watch?v=jQL0ZeHtXFc")
|
| 2 |
def YTVideoToText(video_link):
|
| 3 |
# installing & importing libraries
|
|
@@ -13,9 +18,7 @@ def YTVideoToText(video_link):
|
|
| 13 |
for i in transcript:
|
| 14 |
result += ' ' + i['text']
|
| 15 |
|
| 16 |
-
# summarize text
|
| 17 |
-
summarizerfb = pipeline("summarization", model="facebook/bart-large-cnn")
|
| 18 |
-
|
| 19 |
num_iters = int(len(result)/1000)
|
| 20 |
summarized_text = []
|
| 21 |
summarized_text2 = []
|
|
@@ -23,7 +26,7 @@ def YTVideoToText(video_link):
|
|
| 23 |
start = 0
|
| 24 |
start = i * 1000
|
| 25 |
end = (i + 1) * 1000
|
| 26 |
-
out =
|
| 27 |
out = out[0]
|
| 28 |
out = out['summary_text']
|
| 29 |
summarized_text.append(out)
|
|
@@ -40,9 +43,6 @@ def postSummaryWithBart(blog_link):
|
|
| 40 |
from bs4 import BeautifulSoup
|
| 41 |
import requests
|
| 42 |
|
| 43 |
-
# loading summarization pipeline
|
| 44 |
-
summarizer = pipeline("summarization")
|
| 45 |
-
|
| 46 |
# getting our blog post
|
| 47 |
URL = blog_link
|
| 48 |
r = requests.get(URL)
|
|
@@ -76,7 +76,7 @@ def postSummaryWithBart(blog_link):
|
|
| 76 |
chunks[chunk_id] = ' '.join(chunks[chunk_id])
|
| 77 |
|
| 78 |
# summarizing text
|
| 79 |
-
res =
|
| 80 |
text = ''.join([summ['summary_text'] for summ in res])
|
| 81 |
|
| 82 |
# returning summary
|
|
@@ -88,15 +88,14 @@ def abstractiveSummaryWithPegasus(words):
|
|
| 88 |
# importing & loading model
|
| 89 |
from transformers import PegasusForConditionalGeneration, PegasusTokenizer
|
| 90 |
tokenizer = PegasusTokenizer.from_pretrained("google/pegasus-xsum")
|
| 91 |
-
model = PegasusForConditionalGeneration.from_pretrained("google/pegasus-xsum")
|
| 92 |
|
| 93 |
# perform summarization
|
| 94 |
tokens = tokenizer(words, truncation=True, padding="longest", return_tensors="pt")
|
| 95 |
-
summary =
|
| 96 |
actual_summ = tokenizer.decode(summary[0])
|
| 97 |
|
| 98 |
# returning summary
|
| 99 |
-
|
| 100 |
|
| 101 |
|
| 102 |
# Main logic of the program
|
|
@@ -122,29 +121,38 @@ with gr.Blocks() as ui:
|
|
| 122 |
label="URI à résumer",
|
| 123 |
max_lines=1,
|
| 124 |
placeholder="https://youtube|website.ext",
|
|
|
|
| 125 |
)
|
| 126 |
TRANSCRIPT = gr.Textbox(
|
| 127 |
-
label="
|
| 128 |
lines=10,
|
| 129 |
placeholder="https://youtube|website.ext",
|
|
|
|
| 130 |
)
|
| 131 |
RESUME = gr.Textbox(
|
| 132 |
-
label="
|
| 133 |
lines=10,
|
| 134 |
interactive=False,
|
| 135 |
placeholder="https://youtube|website.ext",
|
|
|
|
| 136 |
)
|
| 137 |
with gr.Column():
|
| 138 |
-
MODE = gr.Radio(
|
|
|
|
|
|
|
|
|
|
|
|
|
| 139 |
gr.Button("Process URI").click(
|
| 140 |
fn=process,
|
| 141 |
inputs=[URI, MODE],
|
| 142 |
-
outputs=[TRANSCRIPT, RESUME]
|
|
|
|
| 143 |
)
|
| 144 |
gr.Button("Process TEXT").click(
|
| 145 |
fn=abstractiveSummaryWithPegasus,
|
| 146 |
inputs=[TRANSCRIPT],
|
| 147 |
-
outputs=[RESUME]
|
|
|
|
| 148 |
)
|
| 149 |
|
| 150 |
#translator_fr = gr.Interface.load("huggingface/Helsinki-NLP/opus-mt-fr-en")
|
|
|
|
| 1 |
+
# Initialize the space
|
| 2 |
+
summarizeryt = pipeline("summarization", model="facebook/bart-large-cnn")
|
| 3 |
+
summarizerbg = pipeline("summarization")
|
| 4 |
+
summarizertx = PegasusForConditionalGeneration.from_pretrained("google/pegasus-xsum")
|
| 5 |
+
|
| 6 |
# How to use: YTVideoToText("https://www.youtube.com/watch?v=jQL0ZeHtXFc")
|
| 7 |
def YTVideoToText(video_link):
|
| 8 |
# installing & importing libraries
|
|
|
|
| 18 |
for i in transcript:
|
| 19 |
result += ' ' + i['text']
|
| 20 |
|
| 21 |
+
# summarize text
|
|
|
|
|
|
|
| 22 |
num_iters = int(len(result)/1000)
|
| 23 |
summarized_text = []
|
| 24 |
summarized_text2 = []
|
|
|
|
| 26 |
start = 0
|
| 27 |
start = i * 1000
|
| 28 |
end = (i + 1) * 1000
|
| 29 |
+
out = summarizeryt(result[start:end], max_new_tokens=130, min_length=30, do_sample=False)
|
| 30 |
out = out[0]
|
| 31 |
out = out['summary_text']
|
| 32 |
summarized_text.append(out)
|
|
|
|
| 43 |
from bs4 import BeautifulSoup
|
| 44 |
import requests
|
| 45 |
|
|
|
|
|
|
|
|
|
|
| 46 |
# getting our blog post
|
| 47 |
URL = blog_link
|
| 48 |
r = requests.get(URL)
|
|
|
|
| 76 |
chunks[chunk_id] = ' '.join(chunks[chunk_id])
|
| 77 |
|
| 78 |
# summarizing text
|
| 79 |
+
res = summarizerbg(chunks, max_new_tokens=1024, min_length=30, do_sample=False)
|
| 80 |
text = ''.join([summ['summary_text'] for summ in res])
|
| 81 |
|
| 82 |
# returning summary
|
|
|
|
| 88 |
# importing & loading model
|
| 89 |
from transformers import PegasusForConditionalGeneration, PegasusTokenizer
|
| 90 |
tokenizer = PegasusTokenizer.from_pretrained("google/pegasus-xsum")
|
|
|
|
| 91 |
|
| 92 |
# perform summarization
|
| 93 |
tokens = tokenizer(words, truncation=True, padding="longest", return_tensors="pt")
|
| 94 |
+
summary = summarizertx.generate(**tokens)
|
| 95 |
actual_summ = tokenizer.decode(summary[0])
|
| 96 |
|
| 97 |
# returning summary
|
| 98 |
+
return actual_summ
|
| 99 |
|
| 100 |
|
| 101 |
# Main logic of the program
|
|
|
|
| 121 |
label="URI à résumer",
|
| 122 |
max_lines=1,
|
| 123 |
placeholder="https://youtube|website.ext",
|
| 124 |
+
api_name="uri"
|
| 125 |
)
|
| 126 |
TRANSCRIPT = gr.Textbox(
|
| 127 |
+
label="Transcript à résumer",
|
| 128 |
lines=10,
|
| 129 |
placeholder="https://youtube|website.ext",
|
| 130 |
+
api_name="transcript"
|
| 131 |
)
|
| 132 |
RESUME = gr.Textbox(
|
| 133 |
+
label="Résumé",
|
| 134 |
lines=10,
|
| 135 |
interactive=False,
|
| 136 |
placeholder="https://youtube|website.ext",
|
| 137 |
+
api_name="resume"
|
| 138 |
)
|
| 139 |
with gr.Column():
|
| 140 |
+
MODE = gr.Radio(
|
| 141 |
+
label="Mode pour URI",
|
| 142 |
+
choices=["Youtube", "Blog"],
|
| 143 |
+
api_name="mode"
|
| 144 |
+
)
|
| 145 |
gr.Button("Process URI").click(
|
| 146 |
fn=process,
|
| 147 |
inputs=[URI, MODE],
|
| 148 |
+
outputs=[TRANSCRIPT, RESUME],
|
| 149 |
+
api_name="process_uri"
|
| 150 |
)
|
| 151 |
gr.Button("Process TEXT").click(
|
| 152 |
fn=abstractiveSummaryWithPegasus,
|
| 153 |
inputs=[TRANSCRIPT],
|
| 154 |
+
outputs=[RESUME],
|
| 155 |
+
api_name="process_text"
|
| 156 |
)
|
| 157 |
|
| 158 |
#translator_fr = gr.Interface.load("huggingface/Helsinki-NLP/opus-mt-fr-en")
|