Introduce paragraph count summary (#2761)

Co-authored-by: Justin Mayer <entroP@gmail.com>

docs/content.rst

@@ -162,7 +162,10 @@ author you can use ``author`` field.
 
 If you do not explicitly specify summary metadata for a given post, the
 ``SUMMARY_MAX_LENGTH`` setting can be used to specify how many words from the
-beginning of an article are used as the summary.
+beginning of an article are used as the summary. You can also use an article's
+first N paragraphs as its summary using the ``SUMMARY_MAX_PARAGRAPHS`` setting.
+If both settings are in use, the specified number of paragraphs will
+be used but may be truncated to respect the specified maximum length.
 
 You can also extract any metadata from the filename through a regular
 expression to be set in the ``FILENAME_METADATA`` setting. All named groups

docs/settings.rst

@@ -308,6 +308,14 @@ Basic settings
    does not otherwise specify a summary. Setting to ``None`` will cause the
    summary to be a copy of the original content.
 
+.. data:: SUMMARY_MAX_PARAGRAPHS = None
+
+   When creating a short summary of an article, this will be the number of
+   paragraphs to use as the summary. This only applies if your content
+   does not otherwise specify a summary. Setting to ``None`` will cause the
+   summary to use the whole text (up to ``SUMMARY_MAX_LENGTH``) instead of just
+   the first N paragraphs.
+
 .. data:: SUMMARY_END_SUFFIX = '…'
 
    When creating a short summary of an article and the result was truncated to

pelican/contents.py

@@ -28,6 +28,7 @@ from pelican.utils import (
     sanitised_join,
     set_date_tzinfo,
     slugify,
+    truncate_html_paragraphs,
     truncate_html_words,
 )
 
@@ -440,8 +441,13 @@ class Content:
         if "summary" in self.metadata:
             return self.metadata["summary"]
 
+        content = self.content
+        max_paragraphs = self.settings.get("SUMMARY_MAX_PARAGRAPHS")
+        if max_paragraphs is not None:
+            content = truncate_html_paragraphs(self.content, max_paragraphs)
+
         if self.settings["SUMMARY_MAX_LENGTH"] is None:
-            return self.content
+            return content
 
         return truncate_html_words(
             self.content,

pelican/tests/test_contents.py

@@ -116,6 +116,31 @@ class TestPage(TestBase):
         page = Page(**page_kwargs)
         self.assertEqual(page.summary, "")
 
+    def test_summary_paragraph(self):
+        # If SUMMARY_MAX_PARAGRAPHS is set, the generated summary should
+        # not exceed the given paragraph count.
+        page_kwargs = self._copy_page_kwargs()
+        settings = get_settings()
+        page_kwargs["settings"] = settings
+        del page_kwargs["metadata"]["summary"]
+        settings["SUMMARY_MAX_PARAGRAPHS"] = 1
+        settings["SUMMARY_MAX_LENGTH"] = None
+        page = Page(**page_kwargs)
+        self.assertEqual(page.summary, TEST_CONTENT)
+
+    def test_summary_paragraph_max_length(self):
+        # If both SUMMARY_MAX_PARAGRAPHS and SUMMARY_MAX_LENGTH are set,
+        # the generated summary should not exceed the given paragraph count and
+        # not exceed the given length.
+        page_kwargs = self._copy_page_kwargs()
+        settings = get_settings()
+        page_kwargs["settings"] = settings
+        del page_kwargs["metadata"]["summary"]
+        settings["SUMMARY_MAX_PARAGRAPHS"] = 1
+        settings["SUMMARY_MAX_LENGTH"] = 10
+        page = Page(**page_kwargs)
+        self.assertEqual(page.summary, truncate_html_words(TEST_CONTENT, 10))
+
     def test_summary_end_suffix(self):
         # If a :SUMMARY_END_SUFFIX: is set, and there is no other summary,
         # generated summary should contain the specified marker at the end.

pelican/tests/test_utils.py

@@ -401,6 +401,23 @@ class TestUtils(LoggedTestCase):
         self.assertEqual(utils.truncate_html_words("&#1234 text", 20), "&#1234 text")
         self.assertEqual(utils.truncate_html_words("&#xabc text", 20), "&#xabc text")
 
+    def test_truncate_html_paragraphs(self):
+        one = "<p>one</p>"
+
+        self.assertEqual(utils.truncate_html_paragraphs(one, 0), "")
+        self.assertEqual(utils.truncate_html_paragraphs(one, 1), one)
+        self.assertEqual(utils.truncate_html_paragraphs(one, 2), one)
+
+        two = one + "<p>two</p>"
+        self.assertEqual(utils.truncate_html_paragraphs(two, 1), one)
+        self.assertEqual(utils.truncate_html_paragraphs(two, 2), two)
+
+        three = two + "<p>three</p>"
+        self.assertEqual(utils.truncate_html_paragraphs(three, 1), one)
+        self.assertEqual(utils.truncate_html_paragraphs(three, 2), two)
+        self.assertEqual(utils.truncate_html_paragraphs(three, 3), three)
+        self.assertEqual(utils.truncate_html_paragraphs(three, 4), three)
+
     def test_process_translations(self):
         fr_articles = []
         en_articles = []

pelican/utils.py

@@ -631,6 +631,25 @@ def truncate_html_words(s: str, num: int, end_text: str = "…") -> str:
     return out
 
 
+def truncate_html_paragraphs(s, count):
+    """Truncate HTML to a certain number of paragraphs.
+
+    :param count: number of paragraphs to keep
+
+    Newlines in the HTML are preserved.
+    """
+    paragraphs = []
+    tag_stop = 0
+    substr = s[:]
+    for _ in range(count):
+        substr = substr[tag_stop:]
+        tag_start = substr.find("<p>")
+        tag_stop = substr.find("</p>") + len("</p>")
+        paragraphs.append(substr[tag_start:tag_stop])
+
+    return "".join(paragraphs)
+
+
 def process_translations(
     content_list: list[Content],
     translation_id: str | Collection[str] | None = None,