Update app.py

app.py

@@ -1,16 +1,124 @@
 import gradio as gr
 from huggingface_hub import InferenceClient
+from urllib import request
+from lxml import etree
+
+url_prefix_mapping = {
+ 'acl': 'https://aclanthology.org',
+ 'emnlp': 'https://aclanthology.org',
+ 'naacl': 'https://aclanthology.org',
+ 'tacl': 'https://aclanthology.org',
+ 'nips': 'https://papers.nips.cc',
+ 'icml': 'https://papers.nips.cc',
+ 'iclr': 'https://iclr.cc',
+}
+
+mlr_mapping = {
+ ('icml', 2020): 'v119',
+ ('icml', 2021): 'v139',
+ ('icml', 2022): 'v162',
+ ('icml', 2023): 'v202',
+ ('icml', 2024): 'v139',
+}
+
+def get_paper_home(venue, year):
+ if venue in ['acl', 'emnlp', 'naacl']:
+ return f'https://aclanthology.org/events/{venue}-{year}'
+ 
+ elif venue == 'nips':
+ return f'https://papers.{venue}.cc/paper_files/paper/{year}'
+ 
+ elif venue == 'icml':
+ return f'https://proceedings.mlr.press/{mlr_mapping[(venue, year)]}'
+ 
+ elif venue == 'iclr':
+ return f'https://iclr.cc/Downloads/{year}'
+
+
+def check_key_words(ele):
+ s = ''.join(ele.itertext()).lower()
+ url = ele.get('href')
+ for i in keywords:
+ match = re.search(i, s)
+ if match:
+ return True
+
+ return False
+
+def check_key_words_icml(ele):
+ s = ''.join(ele.find('.//p[@class="title"]').itertext()).lower()
+ url = ele.get('href')
+ for i in keywords:
+ match = re.search(i, s)
+ if match:
+ return True
+
+ return False
+
+
+def search(keywords, venues):
+ search_venues = []
+ if "NeurIPS/ICLR/ICML" in venues:
+ search_venues.extend(['nips', 'iclr', 'icml'])
+ if "*ACL" in venues:
+ search_venues.extend(['acl', 'emnlp', 'naacl', 'tacl'])
+ if "CVPR/ECCV/ICCV" in venues:
+ search_venues.extend(['nips', 'iclr', 'icml'])
+
+ results = []
+ for venue in search_venues:
+ if 'acl' in venue:
+ paper_tag_on_html = ".//a[@class='align-middle']"
+ elif venue == 'iclr':
+ paper_tag_on_html = ".//a[@class='Poster']"
+ elif venue == 'nips':
+ paper_tag_on_html = ".//a[@title='paper title']"
+ elif venue == 'icml':
+ paper_tag_on_html = ".//div[@class='paper']"
+ 
+ for year in years:
+ print(venue, year)
+ 
+ paper_home = get_paper_home(venue, year)
+ url_prefix = url_prefix_mapping[venue]
+ if venue == 'icml':
+ url_prefix = paper_home
+ 
+ try:
+ response = request.urlopen(paper_home)
+ except:
+ continue
+ 
+ html = response.read().decode()
+ tree = etree.fromstring(html, etree.HTMLParser())
+ 
+ elements = tree.findall(paper_tag_on_html)
+ if venue == 'icml':
+ elements = [i for i in elements if check_key_words_icml(i)]
+ urls = [i.find('.//p[@class="links"]').find('a').get('href') for i in elements]
+ results.extend(urls)
+ 
+ else:
+ elements = [i for i in elements if check_key_words(i)]
+ urls = [url_prefix + i.find('.//p[@class="links"]').get('href') for i in elements]
+ results.extend(urls)
+ 
+ print(len(elements))
+ print()
+
+ return results
 
-def search(a):
- return a
 
 """
 For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
 """
 demo = gr.Interface(
  search,
- inputs=gr.Textbox(lines=2, placeholder="Keywords of the paper title. Supports ReGex."),
- outputs="text"
+ inputs=[
+ gr.Textbox(lines=2, placeholder="Keywords of the paper title. Supports ReGex."),
+ gr.CheckboxGroup(["NeurIPS/ICLR/ICML", "*ACL", "CVPR/ECCV/ICCV"], label="Choose Venues to Search", value=["NeurIPS/ICLR/ICML", "*ACL", "CVPR/ECCV/ICCV"])
+ ],
+ outputs=gr.DataFrame(headers=["Paper Link", ])#"Title", "Authors"
 )