feat: initial commit

BREAKING_CHANGE

.github/workflows/tests.yml

@@ -52,8 +52,8 @@ jobs:
       - name: Run pylint
         shell: bash -l {0}
         run: |
-          pylint tests
-          pylint app
+          pylint --disable=E0401,R0801,C010 tests
+          pylint --disable=E0401,R0801,C010 app
   
   # code coverage job for ubuntu and macos
   code-cov-ubuntu-macos:

README.md

@@ -1,15 +1,9 @@
-[![TESTS](https://github.com/VirtualPatientEngine/demo/actions/workflows/tests.yml/badge.svg)](https://github.com/VirtualPatientEngine/demo/actions/workflows/tests.yml)
-[![RELEASE](https://github.com/VirtualPatientEngine/demo/actions/workflows/release.yml/badge.svg)](https://github.com/VirtualPatientEngine/demo/actions/workflows/release.yml)
+[![TESTS](https://github.com/VirtualPatientEngine/literatureSurvey/actions/workflows/tests.yml/badge.svg)](https://github.com/VirtualPatientEngine/literatureSurvey/actions/workflows/tests.yml)
+[![RELEASE](https://github.com/VirtualPatientEngine/literatureSurvey/actions/workflows/release.yml/badge.svg)](https://github.com/VirtualPatientEngine/literatureSurvey/actions/workflows/release.yml)
 
-<h1 align="center" style="border-bottom: none;">🚀 Demo</h1>
+<h1 align="center" style="border-bottom: none;">🚀 Literature Survey</h1>
 
-This repository serves as a template to develop a code repository (shareable packages). The followoing features are pre-embedded into this repository to help you kick start with your work:
-1. **Automated Workflows for testing and Semantic Release using GitHub Actions**
-2. **Automated Documentation and Hosting using MkDocs**
-3. **Automated Distributable Release (.whl package) using setup.py**
-
-In the files ```setup.py``` and ```release.config.js```, change the package name from **gurdeepdemo** to whatever you desire.
-
-View the documentation of the example code used in the repo @ https://virtualpatientengine.github.io/demo
+This repository serves as a template to develop a literature survey repository.
+View the documentation of the example code used in the repo @ https://virtualpatientengine.github.io/literatureSurvey
 
 >NOTE: Read more about these and other features in the CodeOps and DevOps documents available on the teams channel.

app/literatureFetch.py

@@ -0,0 +1,137 @@
+"""
+A class to extract the data from Uniprot
+for a given protein
+"""
+
+import requests
+from jinja2 import Environment, FileSystemLoader
+
+class Arxiv:
+    """
+    A class to extract the data from Arxiv
+    for a given arxiv id
+    """
+    def __init__(self, paper_id) -> None:
+        """
+        Initialize the class
+
+        Args:
+            arxiv_id (str): arxiv id
+            title (str): title of the paper
+            published (str): published date
+            link (str): link to the paper
+        
+        Returns:
+            None
+        """
+        self.arxiv_id = paper_id
+        self.semantic_scholar_data = {}
+    def get_semantic_scholar_data(self) -> dict:
+        """
+        Return the semantic scholar data for a given arxiv id
+
+        Args:
+            None
+        Returns:
+            dict: semantic scholar data
+        """
+        return self.semantic_scholar_data
+    def get_arxiv_id(self):
+        """
+        Return the arxiv id
+
+        Args:
+            None
+        Returns:
+            str: arxiv id
+        """
+        return self.arxiv_id
+
+def get_semantic_scholar(article_ids):
+    """
+    Return the semantic scholar data for a given arxiv ids
+
+    Args:
+        arxiv_ids (list): list of arxiv ids
+    Returns:
+        dict: dictionary of arxiv ids and their semantic scholar data
+    """
+
+    fields = 'referenceCount,citationCount,title,authors,'
+    fields += 'journal,fieldsOfStudy,publicationTypes,publicationDate,url'
+    r = requests.post(
+        'https://api.semanticscholar.org/graph/v1/paper/batch',
+        params={'fields': fields},
+        json={"ids": article_ids},
+        timeout=10
+    )
+    # print(json.dumps(r.json(), indent=2))
+    return r.json()
+
+if __name__ == "__main__":
+    # Run the script from command line
+    import markdownify
+    import feedparser
+    # Define the query
+    QUERY = 'stat.ML:causal+link+prediction'
+    # Define the base url
+    BASE_URL = 'http://export.arxiv.org/api/query?search_query='
+    # Define the search parameters
+    START = 0
+    MAX_RESULTS = 500
+    SORT_BY = 'relevance'
+    SEARCH_PARAMS = f'&start={START}&max_results={MAX_RESULTS}&sortBy={SORT_BY}'
+    # Define the url link
+    URL_LINK = BASE_URL + QUERY + SEARCH_PARAMS
+    # Parse the url link
+    feed = feedparser.parse(URL_LINK)
+    num_entries = len(feed.entries)
+    arxiv_ids = [] # List to store the arxiv ids
+    dic_arxiv_ids = {} # Dictionary to store the arxiv ids
+    for i in range(num_entries):
+        # print (feed.entries[i]['title'], feed.entries[i]['published'], feed.entries[i]['link'])
+        arxiv_id = feed.entries[i]['link'].split('/')[-1].split('v')[0]
+        dic_arxiv_ids[arxiv_id] = Arxiv(arxiv_id)
+    arxiv_ids = ["ARXIV:"+arxiv_id for arxiv_id in dic_arxiv_ids]
+    # print (arxiv_ids)
+    results = get_semantic_scholar(arxiv_ids)
+    # print (results)
+    print (len(results))
+
+    articles = []
+    for i in range(num_entries):
+        arxiv_id = arxiv_ids[i].split(':')[1]
+        print (results[i])
+        if results[i] is None:
+            continue
+        dic_arxiv_ids[arxiv_id].semantic_scholar_data = results[i]
+        # Append the articles
+        if dic_arxiv_ids[arxiv_id].semantic_scholar_data is not None:
+            if int(dic_arxiv_ids[arxiv_id].semantic_scholar_data['citationCount']) > 1000:
+                # Extract author names
+                authors = []
+                for author in dic_arxiv_ids[arxiv_id].semantic_scholar_data['authors']:
+                    authors.append(author['name'])
+                articles.append({"title": results[i]['title'],
+                                 'url': results[i]['url'],
+                                 "authors": ', '.join(authors),
+                                "citations": results[i]['citationCount'],
+                                'journal': results[i]['journal'],
+                                'fieldsOfStudy': results[i]['fieldsOfStudy'],
+                                'publicationTypes': results[i]['publicationTypes'],
+                                'publicationDate': results[i]['publicationDate']
+                             })
+    # Set the template environment
+    environment = Environment(loader=FileSystemLoader("../templates/"))
+    # Get the template
+    template = environment.get_template("message.txt")
+    # Render the template
+    content = template.render(
+        articles=articles,
+        query=QUERY,
+    )
+    # Convert the content to markdown
+    markdown_text = markdownify.markdownify(content)
+    # Write the markdown text to a file
+    with open('../docs/index.md', 'w', encoding='utf-8') as file:
+        file.write(markdown_text)

docs/index.md

@@ -1 +1,21 @@
-# Welcome to the documentation of the demo app developed by Team VPE
+
+
+
+
+
+
+### Query: stat.ML:causal+link+prediction
+
+
+
+
+| Title | Authors | PublicationDate | #Citations | Journal/Conference |
+| --- | --- | --- | --- | --- |
+| [Link Prediction in Complex Networks: A Survey](https://www.semanticscholar.org/paper/8cd9aa720a3a2f9dcb52ad9eb1bf258a80ce0648) | Linyuan Lu, Tao Zhou | 2010-10-05 | 2520 | {'name': 'ArXiv', 'volume': 'abs/1010.0725'} |
+| [Link Prediction Based on Graph Neural Networks](https://www.semanticscholar.org/paper/e4715a13f6364b1c81e64f247651c3d9e80b6808) | Muhan Zhang, Yixin Chen | 2018-02-27 | 1422 | {'pages': '5171-5181'} |
+| [Predicting missing links via local information](https://www.semanticscholar.org/paper/76c361552181f3798a3fae7485a22a333af85047) | Tao Zhou, Linyuan Lü, Yi‐Cheng Zhang | 2009-01-05 | 1495 | {'name': 'The European Physical Journal B', 'pages': '623-630', 'volume': '71'} |
+| [Variational Graph Auto-Encoders](https://www.semanticscholar.org/paper/54906484f42e871f7c47bbfe784a358b1448231f) | Thomas Kipf, M. Welling | 2016-11-21 | 2597 | {'name': 'ArXiv', 'volume': 'abs/1611.07308'} |
+| [Predicting positive and negative links in online social networks](https://www.semanticscholar.org/paper/1926bad0dc2c1d302ad9a673226f8ca56869683c) | J. Leskovec, D. Huttenlocher, J. Kleinberg | 2010-03-11 | 1601 | {'name': 'ArXiv', 'volume': 'abs/1003.2429'} |
+
+
+

docs/literatureFetch.md

@@ -0,0 +1,3 @@
+## class Arxiv
+
+:::app.literatureFetch

docs/test_literatureFetch.md

@@ -0,0 +1,3 @@
+## class Arxiv
+
+:::app.test_literatureFetch

mkdocs.yml

@@ -1,10 +1,12 @@
-site_name: Demo
+site_name: Literature Survey
 plugins:
   - mkdocstrings:
       paths: ["app", "tests"]  
 theme:
   name: material
 nav:
   - Home: index.md
-  - class UniProt: uniprot.md
-  - methods in tests/uniprot: test_uniprot.md
+  - API Reference:
+    - Arxiv: literatureFetch.md
+    - Tests: test_literatureFetch.md
+

release.config.js

@@ -6,33 +6,22 @@ const config = {
       '@semantic-release/commit-analyzer',
       '@semantic-release/release-notes-generator',
     ["@semantic-release/exec", {
-      // "publishCmd": "echo 'RELEASE_VERSION=${nextRelease.version}' >> $GITHUB_ENV && echo ${nextRelease.version} > release_version.txt",
-      // "prepareCmd": "echo ${nextRelease.version} > release_version.txt ; python3 setup.py bdist_wheel --dist-dir dist/ ; mv dist/*-${nextRelease.version}-*whl dist/gurdeep-demo.whl",
-      // "publishCmd": "echo 'RELEASE_VERSION=${nextRelease.version}' > release_version.txt",
-      // "prepareCmd": "python3 setup.py bdist_wheel --dist-dir dist/",
-      "prepareCmd": "echo ${nextRelease.version} > release_version.txt ; python3 setup.py bdist_wheel --dist-dir dist/ ; mv dist/*-${nextRelease.version}-*whl dist/"+package_name+"-latest-py3-none-any.whl",
+      // "prepareCmd": "echo ${nextRelease.version} > release_version.txt ; python3 setup.py bdist_wheel --dist-dir dist/ ; mv dist/*-${nextRelease.version}-*whl dist/"+package_name+"-latest-py3-none-any.whl",
         }],
       ["@semantic-release/git", {
-        // "assets": ["dist/*.js", "dist/*.js.map"],
-        // "assets": ["dist/gurdeep-demo.whl"],
-        // "assets": ["dist/gurdeepdemo-${nextRelease.version}-py3-none-any.whl"],
-        "assets": ["dist/"+package_name+"-latest-py3-none-any.whl"],
+        // "assets": ["dist/"+package_name+"-latest-py3-none-any.whl"],
       }],
-
-      // '@semantic-release/github'
         [
-          "@semantic-release/github",
-          {
-            "assets": [
-              {
-                // "path": "dist/gurdeep-demo.whl",
-                // "path": "dist/gurdeep-demo-${nextRelease.version}-*whl",
-                "path": "dist/"+package_name+"-latest-py3-none-any.whl",
-                "label": package_name+"-${nextRelease.version}",
-                "message": "chore(release): ${nextRelease.version} [skip ci]\n\n${nextRelease.notes}"
-              }
-            ]
-          }
+          // "@semantic-release/github",
+          // {
+          //   "assets": [
+          //     {
+          //       "path": "dist/"+package_name+"-latest-py3-none-any.whl",
+          //       "label": package_name+"-${nextRelease.version}",
+          //       "message": "chore(release): ${nextRelease.version} [skip ci]\n\n${nextRelease.notes}"
+          //     }
+          //   ]
+          // }
         ],
     ]
   };

templates/message.txt

@@ -0,0 +1,28 @@
+<!DOCTYPE html>
+<html lang="en">
+<head>
+  <meta charset="utf-8">
+</head>
+
+<body>
+  <h3>Query: {{ query }}</h3>
+  <table>
+  <tr>
+      <th>Title</th>
+      <th>Authors</th>
+      <th>PublicationDate</th>
+      <th>#Citations</th>
+      <th>Journal/Conference</th>
+  </tr>
+  {% for article in articles %}
+    <tr>
+      <td>[{{ article.title}}]({{article.url}})</a></td>
+      <td>{{ article.authors}}</td>
+      <td>{{ article.publicationDate}}</td>
+      <td>{{ article.citations }}</td>
+      <td>{{ article.journal }}</td>
+    </tr>
+  {% endfor %}
+  </table>
+</body>
+</html>