pyhtml2md 1.6.1__tar.gz → 1.6.2__tar.gz

{pyhtml2md-1.6.1 → pyhtml2md-1.6.2}/CHANGELOG.md

@@ -2,6 +2,9 @@
 
 [TOC]
 
+## 1.6.2
+- Fix HTML entities not converted (see #131)
+
 ## 1.6.0
 
 - Add option for soft line break

{pyhtml2md-1.6.1 → pyhtml2md-1.6.2}/CMakeLists.txt

@@ -1,5 +1,5 @@
 cmake_minimum_required(VERSION 3.8)
-project(html2md VERSION 1.6.1 LANGUAGES CXX)
+project(html2md VERSION 1.6.2 LANGUAGES CXX)
 
 set(PROJECT_HOMEPAGE_URL "https://tim-gromeyer.github.io/html2md/")
 set(html2md_HOMEPAGE_URL "${PROJECT_HOMEPAGE_URL}")

{pyhtml2md-1.6.1 → pyhtml2md-1.6.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: pyhtml2md
-Version: 1.6.1
+Version: 1.6.2
 Summary: Transform your HTML into clean, easy-to-read markdown with pyhtml2md.
 Keywords: html,markdown,html-to-markdown,python3,cpp17,cpp-library,html2markdown,html2md
 Author-Email: Tim Gromeyer <sakul8826@gmail.com>

{pyhtml2md-1.6.1 → pyhtml2md-1.6.2}/pyproject.toml

@@ -11,7 +11,7 @@ description = "Transform your HTML into clean, easy-to-read markdown with pyhtml
 readme = "python/README.md"
 requires-python = ">=3.7"
 license = { text = "MIT" }
-version = "1.6.1"
+version = "1.6.2"
 classifiers = [
     "Intended Audience :: Developers",
     "License :: OSI Approved :: MIT License",

{pyhtml2md-1.6.1 → pyhtml2md-1.6.2}/src/html2md.cpp

@@ -155,9 +155,9 @@ void Converter::CleanUpMarkdown() {
   ReplaceAll(&md_, """, '"');
   ReplaceAll(&md_, "&lt;", "<");
   ReplaceAll(&md_, "&gt;", ">");
-  ReplaceAll(&html_, "&amp;", '&');
-  ReplaceAll(&html_, "&nbsp;", ' ');
-  ReplaceAll(&html_, "&rarr;", "→");
+  ReplaceAll(&md_, "&amp;", '&');
+  ReplaceAll(&md_, "&nbsp;", ' ');
+  ReplaceAll(&md_, "&rarr;", "→");
 
   ReplaceAll(&md_, "\t\t  ", "\t\t");
 }

{pyhtml2md-1.6.1 → pyhtml2md-1.6.2}/tests/test_advanced.py

@@ -68,7 +68,7 @@ def test_table_formatting():
     result = converter.convert()
 
     assert "|" in result
-    # assert "Header 1" in result  # BUG: The generated table is wrong
+    # assert "Header 1" in result  # BUG: The generated table is wrong, don't uncomment this
     assert "Data 1" in result
 
 def test_line_breaks():
@@ -106,6 +106,18 @@ def test_special_characters():
     assert '"characters"' in result
     assert "&" in result
 
+def test_html_entities():
+    html = """
+    <p>&quot;Double quotes&quot; &lt;less than&gt; &gt;greater than&lt; &amp; ampersand &nbsp; non-breaking space &rarr; right arrow</p>
+    """
+    result = pyhtml2md.convert(html)
+    assert '"Double quotes"' in result
+    assert "<less than>" in result
+    assert ">greater than<" in result
+    assert "& ampersand" in result
+    assert " non-breaking space" in result
+    assert "→ right arrow" in result
+
 def test_nested_structures():
     html = """
     <blockquote>