Add Typogrify/SmartyPants dashes config parameter (#2615)

docs/settings.rst

@@ -279,6 +279,17 @@ Basic settings
    ``pre`` and ``code`` tags. This requires that Typogrify version 2.0.4 or
    later is installed
 
+.. data:: TYPOGRIFY_DASHES = 'default'
+
+   This setting controls how Typogrify sets up the Smartypants filter to
+   interpret multiple dash/hyphen/minus characters. A single ASCII dash
+   character (``-``) is always rendered as a hyphen. The ``default`` setting
+   does not handle en-dashes and converts double-hyphens into em-dashes. The
+   ``oldschool`` setting renders both en-dashes and em-dashes when it sees two
+   (``--``) and three (``---``) hyphen characters, respectively. The
+   ``oldschool_inverted`` setting turns two hyphens into an em-dash and three
+   hyphens into an en-dash.
+
 .. data:: SUMMARY_MAX_LENGTH = 50
 
    When creating a short summary of an article, this will be the default length

pelican/readers.py

@@ -585,6 +585,14 @@ class Readers(FileStampDataCacher):
             from typogrify.filters import typogrify
             import smartypants
 
+            typogrify_dashes = self.settings['TYPOGRIFY_DASHES']
+            if typogrify_dashes == 'oldschool':
+                smartypants.Attr.default = smartypants.Attr.set2
+            elif typogrify_dashes == 'oldschool_inverted':
+                smartypants.Attr.default = smartypants.Attr.set3
+            else:
+                smartypants.Attr.default = smartypants.Attr.set1
+
             # Tell `smartypants` to also replace " HTML entities with
             # smart quotes. This is necessary because Docutils has already
             # replaced double quotes with said entities by the time we run

pelican/settings.py

@@ -138,6 +138,7 @@ DEFAULT_CONFIG = {
     'ARTICLE_PERMALINK_STRUCTURE': '',
     'TYPOGRIFY': False,
     'TYPOGRIFY_IGNORE_TAGS': [],
+    'TYPOGRIFY_DASHES': 'default',
     'SUMMARY_END_MARKER': '…',
     'SUMMARY_MAX_LENGTH': 50,
     'PLUGIN_PATHS': [],

pelican/tests/content/article_with_typogrify_dashes.md

@@ -0,0 +1,3 @@
+Title: One -, two --, three --- dashes!
+
+One: -; Two: --; Three: ---

pelican/tests/content/article_with_typogrify_dashes.rst

@@ -0,0 +1,4 @@
+One -, two --, three --- dashes!
+################################
+
+One: -; Two: --; Three: ---

pelican/tests/test_generators.py

@@ -248,6 +248,10 @@ class TestArticlesGenerator(unittest.TestCase):
             ['Article with template', 'published', 'Default', 'custom'],
             ['Metadata tags as list!', 'published', 'Default', 'article'],
             ['Rst with filename metadata', 'published', 'yeah', 'article'],
+            ['One -, two --, three --- dashes!', 'published', 'Default',
+             'article'],
+            ['One -, two --, three --- dashes!', 'published', 'Default',
+             'article'],
             ['Test Markdown extensions', 'published', 'Default', 'article'],
             ['Test markdown File', 'published', 'test', 'article'],
             ['Test md File', 'published', 'test', 'article'],
@@ -562,6 +566,8 @@ class TestArticlesGenerator(unittest.TestCase):
             'Article with markdown containing footnotes',
             'Article with template',
             'Metadata tags as list!',
+            'One -, two --, three --- dashes!',
+            'One -, two --, three --- dashes!',
             'Rst with filename metadata',
             'Test Markdown extensions',
             'Test markdown File',

pelican/tests/test_readers.py

@@ -443,6 +443,40 @@ class RstReaderTest(ReaderTest):
         with self.assertRaisesRegex(Exception, "underline too short"):
             self.read_file(path='../parse_error/parse_error.rst')
 
+    def test_typogrify_dashes_config(self):
+        # Test default config
+        page = self.read_file(
+            path='article_with_typogrify_dashes.rst',
+            TYPOGRIFY=True,
+            TYPOGRIFY_DASHES='default')
+        expected = "<p>One: -; Two: &#8212;; Three:&nbsp;&#8212;-</p>\n"
+        expected_title = "One -, two &#8212;, three &#8212;-&nbsp;dashes!"
+
+        self.assertEqual(page.content, expected)
+        self.assertEqual(page.title, expected_title)
+
+        # Test 'oldschool' variant
+        page = self.read_file(
+            path='article_with_typogrify_dashes.rst',
+            TYPOGRIFY=True,
+            TYPOGRIFY_DASHES='oldschool')
+        expected = "<p>One: -; Two: &#8211;; Three:&nbsp;&#8212;</p>\n"
+        expected_title = "One -, two &#8211;, three &#8212;&nbsp;dashes!"
+
+        self.assertEqual(page.content, expected)
+        self.assertEqual(page.title, expected_title)
+
+        # Test 'oldschool_inverted' variant
+        page = self.read_file(
+            path='article_with_typogrify_dashes.rst',
+            TYPOGRIFY=True,
+            TYPOGRIFY_DASHES='oldschool_inverted')
+        expected = "<p>One: -; Two: &#8212;; Three:&nbsp;&#8211;</p>\n"
+        expected_title = "One -, two &#8212;, three &#8211;&nbsp;dashes!"
+
+        self.assertEqual(page.content, expected)
+        self.assertEqual(page.title, expected_title)
+
 
 @unittest.skipUnless(readers.Markdown, "markdown isn't installed")
 class MdReaderTest(ReaderTest):
@@ -674,6 +708,40 @@ class MdReaderTest(ReaderTest):
         self.assertEqual(metadata, {})
         self.assertEqual(content, '')
 
+    def test_typogrify_dashes_config(self):
+        # Test default config
+        page = self.read_file(
+            path='article_with_typogrify_dashes.md',
+            TYPOGRIFY=True,
+            TYPOGRIFY_DASHES='default')
+        expected = "<p>One: -; Two: &#8212;; Three:&nbsp;&#8212;-</p>"
+        expected_title = "One -, two &#8212;, three &#8212;-&nbsp;dashes!"
+
+        self.assertEqual(page.content, expected)
+        self.assertEqual(page.title, expected_title)
+
+        # Test 'oldschool' variant
+        page = self.read_file(
+            path='article_with_typogrify_dashes.md',
+            TYPOGRIFY=True,
+            TYPOGRIFY_DASHES='oldschool')
+        expected = "<p>One: -; Two: &#8211;; Three:&nbsp;&#8212;</p>"
+        expected_title = "One -, two &#8211;, three &#8212;&nbsp;dashes!"
+
+        self.assertEqual(page.content, expected)
+        self.assertEqual(page.title, expected_title)
+
+        # Test 'oldschool_inverted' variant
+        page = self.read_file(
+            path='article_with_typogrify_dashes.md',
+            TYPOGRIFY=True,
+            TYPOGRIFY_DASHES='oldschool_inverted')
+        expected = "<p>One: -; Two: &#8212;; Three:&nbsp;&#8211;</p>"
+        expected_title = "One -, two &#8212;, three &#8211;&nbsp;dashes!"
+
+        self.assertEqual(page.content, expected)
+        self.assertEqual(page.title, expected_title)
+
 
 class HTMLReaderTest(ReaderTest):
     def test_article_with_comments(self):