Add attr replacer & svg -> img

src/epub_converter/html_epub_processor.py

@@ -11,6 +11,7 @@ class HtmlEpubPreprocessor:
         self.name2function = {
             "table_wrapper": self._wrap_tags_with_table,
             "replacer": self._tags_to_correspond_livecarta_tag,
+            "attr_replacer": self._replace_attrs_in_tags,
             "unwrapper": self._unwrap_tags,
             "inserter": self._insert_tags_into_correspond_tags
         }
@@ -190,6 +191,30 @@ class HtmlEpubPreprocessor:
                     # todo can cause appearance of \n <p><p>...</p></p> -> <p>\n</p> <p>...</p> <p>\n</p> (section)
                     tag.name = tag_to_replace
 
+    @staticmethod
+    def _replace_attrs_in_tags(chapter_tag: BeautifulSoup, rules: list):
+        """
+        Function to replace all tags to correspond LiveCarta tags
+        Parameters
+        ----------
+        chapter_tag: BeautifulSoup
+            Tag & contents of the chapter tag
+
+        Returns
+        -------
+        None
+            Chapter Tag with all tags replaced with LiveCarta tags
+
+        """
+        for rule in rules:
+            attr = rule["attr"]
+            tags = rule["condition"]["tags"]
+            attr_to_replace = rule["attr_to_replace"]
+            for tag in chapter_tag.find_all([re.compile(tag) for tag in tags],
+                                                            {attr: re.compile(r".*")}):
+                tag[attr_to_replace] = tag[attr]
+                del tag[attr]
+
     def _unwrap_tags(self, chapter_tag: BeautifulSoup, rules: dict):
         """
         Function unwrap tags and moves id to span
@@ -353,7 +378,7 @@ class HtmlEpubPreprocessor:
                     and (tag.attrs.get("class") not in ["link-anchor", "footnote-element"]):
                 del tag.attrs["class"]
 
-    def prepare_content(self, title_str: str, content_tag: BeautifulSoup, remove_title_from_chapter: bool) -> str:
+    def prepare_content(self, title_str: str, content_tag: BeautifulSoup, remove_title_from_chapter: bool) -> Tag:
         """
         Function finalise processing/cleaning content
         Parameters
@@ -378,7 +403,7 @@ class HtmlEpubPreprocessor:
 
         Returns
         -------
-        content_tag: str
+        content_tag: Tag
             prepared content
 
         """
@@ -397,4 +422,4 @@ class HtmlEpubPreprocessor:
         self._process_tables(content_tag)
         # 9. remove classes that weren't created by converter
         self._class_removing(content_tag)
-        return str(content_tag)
+        return content_tag