Propagate non-block markup from figcaption

se/se_epub.py

@@ -5,6 +5,7 @@
 """
 
 import base64
+from copy import deepcopy
 import datetime
 import os
 from pathlib import Path
@@ -16,6 +17,7 @@
 import regex
 
 import se
+import se.css
 import se.easy_xml
 import se.formatting
 import se.images
@@ -932,19 +934,32 @@ def generate_loi(self, loi_dom: se.easy_xml.EasyXmlTree) -> None:
 			for figure in dom.xpath("/html/body//figure[@id and img]"):
 				figure_id = figure.get_attr("id")
 
-				text = (figure.xpath("./img")[0].get_attr("alt") or "").strip()
+				entry = (figure.xpath("./img")[0].get_attr("alt") or "").strip()
 
 				figcaption = figure.xpath("./figcaption")
 				if figcaption:
 					figcaption_text = figcaption[0].inner_text()
 					# The alt text is probably more useful to the reader in this case.
 					if figcaption_text and not regex.search(r"^[Ff]igure\s+\d+$", figcaption_text):
-						text = figcaption_text
+						has_block = False
+						for tag in se.css.CSS_BLOCK_ELEMENTS:
+							if figcaption[0].xpath(f"./{tag}"):
+								has_block = True
+								break
+
+						# Try to retain semantic phrasing structure.
+						if not has_block:
+							entry = deepcopy(figcaption[0])
 
 				a = se.easy_xml.EasyXmlElement("<a/>")
-				a.set_text(text or f"Unable to auto-generate LoI text for #{figure_id}.")
 				a.set_attr("href", f"{file_path.name}#{figure_id}")
 
+				if isinstance(entry, str):
+					a.set_text(entry or f"Unable to auto-generate LoI text for #{figure_id}.")
+				else:
+					a.append(entry)
+					entry.unwrap()
+
 				p = se.easy_xml.EasyXmlElement("<p/>")
 				p.append(a)
 

tests/draft_commands/build-loi/test-1/golden/src/epub/text/chapter-1.xhtml

@@ -46,6 +46,11 @@
 				<img alt="Alt is preferred over Figure N figcaption."/>
 				<figcaption> <b> Figure 8 </b> </figcaption>
 			</figure>
+
+			<figure id="f-9">
+				<img alt="Alt is preferred over figcaption with block element."/>
+				<figcaption><h1>Figure 9</h1></figcaption>
+			</figure>
 		</section>
 	</body>
 </html>

tests/draft_commands/build-loi/test-1/golden/src/epub/text/loi.xhtml

@@ -32,7 +32,7 @@
 				</li>
 				<li>
 					<p>
-						<a href="chapter-1.xhtml#f-6">Text comes from figcaption inner_text.</a>
+						<a href="chapter-1.xhtml#f-6">Text comes from <b>figcaption</b> inner_text.</a>
 					</p>
 				</li>
 				<li>
@@ -45,6 +45,11 @@
 						<a href="chapter-1.xhtml#f-8">Alt is preferred over Figure N figcaption.</a>
 					</p>
 				</li>
+				<li>
+					<p>
+						<a href="chapter-1.xhtml#f-9">Alt is preferred over figcaption with block element.</a>
+					</p>
+				</li>
 				<li>
 					<p>
 						<a href="chapter-2.xhtml#f-0">Unable to auto-generate LoI text for #f-0.</a>

tests/draft_commands/build-loi/test-1/in/src/epub/text/chapter-1.xhtml

@@ -46,6 +46,11 @@
 				<img alt="Alt is preferred over Figure N figcaption."/>
 				<figcaption> <b> Figure 8 </b> </figcaption>
 			</figure>
+
+			<figure id="f-9">
+				<img alt="Alt is preferred over figcaption with block element."/>
+				<figcaption><h1>Figure 9</h1></figcaption>
+			</figure>
 		</section>
 	</body>
 </html>