Add The Globe and Mail

docs/supported_publishers.md

@@ -114,6 +114,21 @@
  <td>&#160;</td>
  <td>&#160;</td>
  </tr>
+ <tr>
+ <td>
+ <code>TheGlobeAndMail</code>
+ </td>
+ <td>
+ <div>The Globe and Mail</div>
+ </td>
+ <td>
+ <a href="https://www.theglobeandmail.com">
+ <span>www.theglobeandmail.com</span>
+ </a>
+ </td>
+ <td>&#160;</td>
+ <td>&#160;</td>
+ </tr>
  </tbody>
 </table>
 

src/fundus/publishers/ca/__init__.py

@@ -1,5 +1,6 @@
 from fundus.publishers.base_objects import Publisher, PublisherGroup
 from fundus.publishers.ca.cbc_news import CBCNewsParser
+from fundus.publishers.ca.globe_and_mail import TheGlobeAndMailParser
 from fundus.scraping.url import NewsMap, RSSFeed, Sitemap
 
 # noinspection PyPep8Naming
@@ -16,3 +17,12 @@ class CA(metaclass=PublisherGroup):
  RSSFeed("https://www.cbc.ca/webfeed/rss/rss-canada"),
  ],
  )
+ TheGlobeAndMail = Publisher(
+ name="The Globe and Mail",
+ domain="https://www.theglobeandmail.com",
+ parser=TheGlobeAndMailParser,
+ sources=[
+ NewsMap("https://www.theglobeandmail.com/arc/outboundfeeds/news-sitemap-index/?outputType=xml"),
+ Sitemap("https://www.theglobeandmail.com/arc/outboundfeeds/sitemap-index/?outputType=xml"),
+ ],
+ )

src/fundus/publishers/ca/globe_and_mail.py

@@ -0,0 +1,49 @@
+import datetime
+from typing import List, Optional
+
+from lxml.cssselect import CSSSelector
+
+from fundus.parser import ArticleBody, BaseParser, ParserProxy, attribute
+from fundus.parser.utility import (
+ extract_article_body_with_selector,
+ generic_author_parsing,
+ generic_date_parsing,
+ generic_topic_parsing,
+)
+
+
+class TheGlobeAndMailParser(ParserProxy):
+ class V1(BaseParser):
+ _subheadline_selector = CSSSelector("article > h4")
+ _paragraph_selector = CSSSelector("article > p")
+
+ @attribute
+ def body(self) -> ArticleBody:
+ return extract_article_body_with_selector(
+ self.precomputed.doc,
+ subheadline_selector=self._subheadline_selector,
+ paragraph_selector=self._paragraph_selector,
+ )
+
+ @attribute
+ def authors(self) -> List[str]:
+ return generic_author_parsing(self.precomputed.ld.bf_search("author"))
+
+ @attribute
+ def publishing_date(self) -> Optional[datetime.datetime]:
+ return generic_date_parsing(self.precomputed.ld.bf_search("datePublished"))
+
+ @attribute
+ def title(self) -> Optional[str]:
+ return self.precomputed.meta.get("og:title")
+
+ @attribute
+ def topics(self) -> List[str]:
+ topic_list = [topic.lower() for topic in generic_topic_parsing(self.precomputed.meta.get("keywords"))]
+ topic_set = set(topic_list)
+ topic_duplicates = list(topic_list)
+ for element in topic_set:
+ topic_duplicates.remove(element)
+ for duplicate in topic_duplicates:
+ topic_list.remove(duplicate)
+ return [topic.title() for topic in topic_list if "news" not in topic]

tests/resources/parser/test_data/ca/TheGlobeAndMail.json

@@ -0,0 +1,37 @@
+{
+ "V1": {
+ "authors": [
+ "Reuters"
+ ],
+ "body": {
+ "summary": [],
+ "sections": [
+ {
+ "headline": [],
+ "paragraphs": [
+ "Cannabis firm Canopy Growth WEED-T said on Friday its CEO, David Klein, will retire at the end of the company’s current fiscal year on March 31.",
+ "Klein was appointed as Canopy Growth’s chairman when the then co-CEO and founder Bruce Linton was fired in 2019.",
+ "The announcement comes just days after the company reported a smaller-than-expected core loss and pointed to easing doubts around its ability to continue as a going concern.",
+ "The company had raised doubts about its ability to continue as a going concern last year as it struggled to turn profitable. It later sold its sports nutrition products unit, Biosteel, after seeking bankruptcy protection for it.",
+ "Canopy Growth said it is in the process of securing a search firm to initiate a comprehensive CEO selection process."
+ ]
+ }
+ ]
+ },
+ "publishing_date": "2024-08-16 12:27:08.062000+00:00",
+ "title": "Canopy Growth CEO David Klein to retire at end of March",
+ "topics": [
+ "Company",
+ "Process",
+ "Ability",
+ "Concern",
+ "Ceo",
+ "Klein",
+ "End",
+ "Co-Ceo",
+ "David Klein",
+ "Canopy Growth Weed-T",
+ "Bruce Linton"
+ ]
+ }
+}

tests/resources/parser/test_data/ca/meta.info

@@ -2,5 +2,9 @@
  "CBCNews_2024_08_08.html.gz": {
  "url": "https://www.cbc.ca/news/world/gaza-israel-ceasefire-negotiations-sinwar-1.7287711?cmp=rss",
  "crawl_date": "2024-08-08 23:53:17.604667"
+ },
+ "TheGlobeAndMail_2024_08_16.html.gz": {
+ "url": "https://www.theglobeandmail.com/business/article-canopy-growth-ceo-david-klein-to-retire-at-end-of-march/",
+ "crawl_date": "2024-08-16 14:45:35.165117"
  }
 }