support images

src/sulguk/transformer.py

@@ -2,7 +2,6 @@
 from typing import Any, List, Optional, Tuple
 
 from sulguk.render.numbers import NumberFormat
-
 from .entities import (
     Blockquote,
     Bold,
@@ -39,7 +38,7 @@
 
 LANG_CLASS_PREFIX = "language-"
 
-NO_CLOSING_TAGS = ("br", "hr", "meta", "link")
+NO_CLOSING_TAGS = ("br", "hr", "meta", "link", "img")
 
 
 class Transformer(HTMLParser):
@@ -71,13 +70,20 @@ def _get_a(self, attrs: Attrs) -> Entity:
         if url:
             return Link(url=url)
         return Group()
-    
-    def _get_img(self, attrs: Attrs) -> Entity:
+
+    def _get_img(self, attrs: Attrs) -> Optional[Entity]:
         url = self._find_attr("src", attrs)
-        if url:
-            return Link(url=url)
-        return Group()
-
+        text = self._find_attr("alt", attrs, url)
+        if not text and not url:
+            return None
+
+        text_entity = Text(text="🖼️" + text)
+        if not url:
+            return text_entity
+        link = Link(url=url)
+        link.add(text_entity)
+        return link
+
     def _get_ul(self, attrs: Attrs) -> Entity:
         return ListGroup(numbered=False)
 
@@ -182,11 +188,14 @@ def handle_startendtag(self, tag: str, attrs: Attrs) -> None:
             entity = NewLine()
         elif tag == "hr":
             entity = HorizontalLine()
+        elif tag in ("img",):
+            entity = self._get_img(attrs)
         elif tag in ("meta", "link"):
             return  # ignored tag
         else:
             raise ValueError(f"Unsupported single tag: `{tag}`")
-        self.current.add(entity)
+        if entity:
+            self.current.add(entity)
 
     def handle_starttag(
             self,
@@ -209,10 +218,8 @@ def handle_starttag(
             nested = entity = self._get_ol(attrs)
         elif tag in ("li",):
             nested = entity = self._get_li(attrs)
-        elif tag in ("a", ):
+        elif tag in ("a",):
             nested = entity = self._get_a(attrs)
-        elif tag in ("img",):
-            nested = entity = self._get_img(attrs)
         elif tag in ("b", "strong"):
             nested = entity = Bold()
         elif tag in ("i", "em", "cite", "var"):

tests/test_img.py

@@ -0,0 +1,34 @@
+import pytest
+
+from sulguk import transform_html
+
+IMG_HTML_NOCLOSE = '0<img src="https://google.com">1'
+IMG_HTML_CLOSE = '0<img src="https://google.com" />1'
+IMG_HTML_ALT = '0<img src="https://google.com" alt="This is text" />1'
+IMG_HTML_EMPTY_ALT = '0<img src="https://google.com" alt="" />1'
+IMG_URL = "https://google.com"
+IMG_TEXT_URL = "0🖼️https://google.com1"
+IMG_TEXT_ALT = "0🖼️This is text1"
+IMG_TEXT_EMPTY_ALT = "0🖼️1"
+
+
+@pytest.mark.parametrize("html, url, text", [
+    (IMG_HTML_NOCLOSE, IMG_URL, IMG_TEXT_URL),
+    (IMG_HTML_CLOSE, IMG_URL, IMG_TEXT_URL),
+    (IMG_HTML_ALT, IMG_URL, IMG_TEXT_ALT),
+    (IMG_HTML_EMPTY_ALT, IMG_URL, IMG_TEXT_EMPTY_ALT),
+])
+def test_link_extracted(html, url, text):
+    result = transform_html(html)
+    assert result.text == text
+    assert len(result.entities) == 1
+    entity = result.entities[0]
+    assert entity['type'] == 'text_link'
+    assert entity['url'] == url
+
+
+def test_empty_image():
+    html = '0<img src="">1'
+    result = transform_html(html)
+    assert result.text == "01"
+    assert not result.entities

tests/test_spaces.py

@@ -45,8 +45,6 @@
 PRE_P_HTML = "<p>1</p>\n<pre>\n    2</pre>"
 PRE_P_PLAIN = "1\n\n    2\n\n"
 
-IMG_HTML = '<img src="https://google.com">'
-IMG_URL = "https://google.com"
 
 @pytest.mark.parametrize("html, plain, name", [
     (SPACES_SPAN_HTML, SPACES_SPAN_PLAIN, "span"),
@@ -65,16 +63,3 @@ def test_spaces(html, plain, name):
     print(repr(plain))
     print(repr(html))
     assert result.text == plain
-
-
-
-@pytest.mark.parametrize("html, url", [
-    (IMG_HTML, IMG_URL),
-])
-def test_link_extracted(html, url):
-    result = transform_html(html)
-    assert result.text == ""
-    assert len(result.entities) == 1
-    entity = result.entities[0]
-    assert entity['type'] == 'text_link'
-    assert entity['url'] == url