add crawl_google_scholar_citation.py (#3879)

* add crawl_google_scholar_citation.py

* pass flack8

* pass isort

* pass isort

* change comment in main

* modify main code

* delete file

* change how to build url

* add a key 'hl' in params dict

* Update crawl_google_scholar_citation.py

* Create crawl_google_results.py

* codespell: Mater

Co-authored-by: Christian Clauss <[email protected]>

Author: boyuuuun

Diff for: .pre-commit-config.yaml

@@ -42,7 +42,7 @@ repos:
     hooks:
       - id: codespell
         args:
-          - --ignore-words-list=ans,fo,followings,hist,iff,secant,som,tim
+          - --ignore-words-list=ans,fo,followings,hist,iff,mater,secant,som,tim
           - --skip="./.*,./other/dictionary.txt,./other/words,./project_euler/problem_022/p022_names.txt"
           - --quiet-level=2
         exclude: |

Diff for: web_programming/crawl_google_scholar_citation.py

@@ -0,0 +1,32 @@
+"""
+Get the citation from google scholar
+using title and year of publication, and volume and pages of journal.
+"""
+
+import requests
+from bs4 import BeautifulSoup
+
+
+def get_citation(base_url: str, params: dict) -> str:
+    """
+    Return the citation number.
+    """
+    soup = BeautifulSoup(requests.get(base_url, params=params).content, "html.parser")
+    div = soup.find("div", attrs={"class": "gs_ri"})
+    anchors = div.find("div", attrs={"class": "gs_fl"}).find_all("a")
+    return anchors[2].get_text()
+
+
+if __name__ == "__main__":
+    params = {
+        "title": (
+            "Precisely geometry controlled microsupercapacitors for ultrahigh areal "
+            "capacitance, volumetric capacitance, and energy density"
+        ),
+        "journal": "Chem. Mater.",
+        "volume": 30,
+        "pages": "3979-3990",
+        "year": 2018,
+        "hl": "en",
+    }
+    print(get_citation("http://scholar.google.com/scholar_lookup", params=params))