From 9b4ecfd63c6c361f3d21f7d3eca41190bbb136c1 Mon Sep 17 00:00:00 2001
From: Kiryl <kiryl.miatselitsa@teqniksoft.com>
Date: Mon, 27 Jun 2022 19:13:39 +0300
Subject: [PATCH] Add function - insert certain tags in parent tags

---
 src/epub_converter/html_epub_preprocessor.py | 87 ++++----------------
 1 file changed, 15 insertions(+), 72 deletions(-)

diff --git a/src/epub_converter/html_epub_preprocessor.py b/src/epub_converter/html_epub_preprocessor.py
index 60cc91e..450d776 100644
--- a/src/epub_converter/html_epub_preprocessor.py
+++ b/src/epub_converter/html_epub_preprocessor.py
@@ -210,28 +210,6 @@ def _remove_headings_content(content_tag, title_of_chapter: str):
                     break
 
 
-# todo remove
-def _process_lists(chapter_tag: BeautifulSoup):
-    """
-    Function
-    - process tags <li>.
-    - unwrap <p> tags.
-    Parameters
-    ----------
-    chapter_tag: Tag, soup object
-
-    Returns
-    -------
-    None
-
-    """
-    li_tags = chapter_tag.find_all("li")
-    for li_tag in li_tags:
-        if li_tag.p:
-            li_tag.attrs.update(li_tag.p.attrs)
-            li_tag.p.unwrap()
-
-
 def _preprocess_table(chapter_tag: BeautifulSoup):
     """Function to preprocess tables and tags(td|th|tr): style"""
     tables = chapter_tag.find_all("table")
@@ -257,53 +235,20 @@ def _preprocess_table(chapter_tag: BeautifulSoup):
             table.attrs["border"] = "1"
 
 
-def _preprocess_code_tags(chapter_tag: BeautifulSoup):
-    """
-    Function
-    - transform <code>, <kdb>, <var> tags into span
-    - add code style to this tags (if there is no)
-    Parameters
-    ----------
-    chapter_tag: Tag, soup object
-
-    Returns
-    -------
-    None
-
-    """
-    for code in chapter_tag.find_all(re.compile("code|kbd|var")):
-        if not code.parent.name == "pre":
-            code.name = "span"
-            if not code.attrs.get("style"):
-                code.attrs["style"] = "font-size: 14px; font-family: courier new,courier,monospace;"
-            continue
-
-
-
-def _preprocess_pre_tags(chapter_tag: BeautifulSoup):
-    """
-    Function preprocessing <pre> tags
-    Wrap string of the tag with <code> if its necessary
-    Parameters
-    ----------
-    chapter_tag: Tag, soup object
-
-    Returns
-    ----------
-    None
-        Modified chapter tag
-
-    """
-    for pre in chapter_tag.find_all("pre"):
-        if pre.find_all("code|kbd|var"):
-            continue
-        else:
-            code = chapter_tag.new_tag("code")
-            # insert all items that was in pre to code and remove from pre
-            for content in reversed(pre.contents):
-                code.insert(0, content.extract())
-            # wrap code with items
-            pre.append(code)
+def _insert_tags_in_parents(chapter_tag):
+    parent_tag2condition = {parent[0]: parent[1] for parent in LiveCartaConfig.INSERT_TAG_IN_PARENT_TAG.keys()}
+    for parent_tag_name, condition in parent_tag2condition.items():
+        for parent_tag in chapter_tag.select(parent_tag_name):
+            if parent_tag.select(condition):
+                continue
+            else:
+                tag_to_insert = chapter_tag.new_tag(
+                    LiveCartaConfig.INSERT_TAG_IN_PARENT_TAG[(parent_tag_name, condition)])
+                # insert all items that was in pre to code and remove from pre
+                for content in reversed(parent_tag.contents):
+                    tag_to_insert.insert(0, content.extract())
+                # wrap code with items
+                parent_tag.append(tag_to_insert)
 
 
 def _class_removing(chapter_tag):
@@ -353,10 +298,8 @@ def prepare_content(title_str: str, content_tag: BeautifulSoup, remove_title_fro
         _remove_headings_content(content_tag, title_str)
 
     # 4. processing tags (<li>, <table>, <code>, <pre>, <div>, <block>)
-    _process_lists(content_tag) # todo regex
     _preprocess_table(content_tag)
-    _preprocess_code_tags(content_tag) # todo regex
-    _preprocess_pre_tags(content_tag) # todo regex
+    _insert_tags_in_parents(content_tag)
 
     # 5. remove classes that weren't created by converter
     _class_removing(content_tag)