epub converter: fix css styles processing

src/css_reader.py

@@ -179,21 +179,14 @@ def clean_css(css):
     return css_text
 
 
-def add_inline_style_to_html_soup(soup1, css_text):
+class TagStyleConverter:
-    css_text = css_text.replace('@namespace epub "http://www.idpf.org/2007/ops";', '')
+    def __init__(self, tag, tag_with_style):
-    livecarta_tmp_ids = []
+        self.tag = tag
-    h_regex = f'(^h[1-9]$)'
+        self.tag_initial_name = tag.name
-    could_have_style_in_livecarta_regexp = re.compile('(^p$)|(^span$)|(^li$)|(^ul$)|(^ol$)|(^td$)|(^th$)|' + h_regex)
+        self.tag_with_style = tag_with_style
-    elements_with_possible_style_attr = soup1.find_all(could_have_style_in_livecarta_regexp)
+        self.style = self.preprocess_style()
-    for i, x in enumerate(elements_with_possible_style_attr):
-        x.attrs['livecarta_id'] = i
-        livecarta_tmp_ids.append(i)
-    html_with_inline_style = transform(str(soup1), css_text=css_text,
-                                       remove_classes=False,
-                                       external_styles=False,
-                                       disable_validation=True)
-    soup2 = BeautifulSoup(html_with_inline_style, features='lxml')
 
+    @staticmethod
     def remove_white_if_no_bgcolor(style_, tag):
         if 'background' in style_:
             return style_
@@ -221,57 +214,98 @@ def add_inline_style_to_html_soup(soup1, css_text):
             style_ = style_.replace('color:white;', '')
         return style_
 
+    def preprocess_style(self):
+        style = self.tag_with_style.attrs.get('style') + ';'
+        style = self.remove_white_if_no_bgcolor(style, self.tag_with_style)
+        style = style.replace('background:', 'background-color:')
+        return style
+
+    def change_attrs_with_corresponding_tags(self):
+        # adds <b>, <u>, <sup>, etc
+        to_remove = check_style_to_be_tag(self.style)
+        new_tags = []
+        for i, (p, v) in enumerate(to_remove):
+            s = f'{p}:{v};'
+            self.style = self.style.replace(s, '')
+            self.style = self.style.strip()
+            if i == 0:
+                self.tag.name = LIVECARTA_STYLE_ATTRS_SHOULD_BE_TAG[(p, v)]
+                new_tags.append(self.tag)
+            else:
+                name = LIVECARTA_STYLE_ATTRS_SHOULD_BE_TAG[(p, v)]
+                new_tag = BeautifulSoup(features='lxml').new_tag(name)
+                new_tags[-1].wrap(new_tag)
+                new_tags.append(new_tag)
+
+        top_tag = self.tag
+
+        if new_tags:
+            tmp_attrs = top_tag.attrs.copy()
+            top_tag = BeautifulSoup(features='lxml').new_tag(self.tag_initial_name)
+            top_tag.attrs = tmp_attrs
+            if self.style:
+                top_tag.attrs['style'] = self.style
+            new_tags[-1].wrap(top_tag)
+        else:
+            top_tag.attrs['style'] = self.style
+
+        return top_tag
+
+    @staticmethod
     def wrap_p_to_save_style_attrs(t):
-        styles_cant_be_in_p = [attr for attr in LIVECARTA_STYLE_ATTRS if attr not in ['text-align', 'text-indent']]
+        styles_cant_be_in_p = [attr for attr in LIVECARTA_STYLE_ATTRS
+                               if attr not in ['text-align', 'text-indent']]
 
         if t.name == 'p' and t.attrs.get('style'):
             check = [attr in t.attrs.get('style') for attr in styles_cant_be_in_p]
             if any(check):
                 t.name = 'span'
-                t.wrap( BeautifulSoup(features='lxml').new_tag('p'))
+                p_tag = BeautifulSoup(features='lxml').new_tag('p')
+                old_style = t.attrs['style']
+                new_style = ''
+                possible_p_attrs_regexp = re.compile(r'(text-align:(\w+);)|(text-indent:(\w+);)')
+                has_p_style_attrs = re.search(possible_p_attrs_regexp, old_style)
+                if has_p_style_attrs:
+                    if has_p_style_attrs.group(1):
+                        new_style += has_p_style_attrs.group(1)
+                        old_style = old_style.replace(has_p_style_attrs.group(1), '')
+                    if has_p_style_attrs.group(3):
+                        new_style += has_p_style_attrs.group(3)
+                        old_style = old_style.replace(has_p_style_attrs.group(3), '')
+
+                    p_tag.attrs['style'] = new_style
+
+                t.attrs['style'] = old_style
+                t.wrap(p_tag)
+
+    def convert_initial_tag(self):
+        del self.tag.attrs['livecarta_id']
+        self.tag = self.change_attrs_with_corresponding_tags()
+        self.wrap_p_to_save_style_attrs(self.tag)
+        return self.tag
+
+
+def add_inline_style_to_html_soup(soup1, css_text):
+    css_text = css_text.replace('@namespace epub "http://www.idpf.org/2007/ops";', '')
+    livecarta_tmp_ids = []
+    h_regex = f'(^h[1-9]$)'
+    could_have_style_in_livecarta_regexp = re.compile('(^p$)|(^span$)|(^li$)|(^ul$)|(^ol$)|(^td$)|(^th$)|' + h_regex)
+    elements_with_possible_style_attr = soup1.find_all(could_have_style_in_livecarta_regexp)
+    for i, x in enumerate(elements_with_possible_style_attr):
+        x.attrs['livecarta_id'] = i
+        livecarta_tmp_ids.append(i)
+    html_with_inline_style = transform(str(soup1), css_text=css_text,
+                                       remove_classes=False,
+                                       external_styles=False,
+                                       disable_validation=True)
+    soup2 = BeautifulSoup(html_with_inline_style, features='lxml')
 
     for i in livecarta_tmp_ids:
         tag = soup1.find(attrs={'livecarta_id': i})
-        tag_initial_name = tag.name
         tag_with_style = soup2.find(attrs={'livecarta_id': i})
         if tag_with_style.attrs.get('style'):
-            style = tag_with_style.attrs.get('style') + ';'
+            style_converter = TagStyleConverter(tag, tag_with_style)
-            style = remove_white_if_no_bgcolor(style, tag_with_style)
+            style_converter.convert_initial_tag()
-            style = style.replace('background:', 'background-color:')
-            to_remove = check_style_to_be_tag(style)
-            new_tags = []
-
-            for i, (p, v) in enumerate(to_remove):
-                s = f'{p}:{v};'
-                style = style.replace(s, '')
-                if i == 0:
-                    tag.name = LIVECARTA_STYLE_ATTRS_SHOULD_BE_TAG[(p, v)]
-                    new_tags.append(tag)
-                else:
-                    name = LIVECARTA_STYLE_ATTRS_SHOULD_BE_TAG[(p, v)]
-                    new_tag = BeautifulSoup(features='lxml').new_tag(name)
-                    new_tags[-1].wrap(new_tag)
-                    new_tags.append(new_tag)
-
-            top_tag = tag
-            if to_remove:
-                style = style.strip()
-                tmp_attrs = tag.attrs.copy()
-                tag.attrs = {}
-
-                top_tag = BeautifulSoup(features='lxml').new_tag(tag_initial_name)
-                top_tag.attrs = tmp_attrs
-                if style:
-                    top_tag.attrs['style'] = style
-                del top_tag.attrs['livecarta_id']
-
-                new_tags[-1].wrap(top_tag)
-            else:
-                tag.attrs['style'] = style
-                del tag.attrs['livecarta_id']
-
-            wrap_p_to_save_style_attrs(top_tag)
-
         else:
             del tag.attrs['livecarta_id']
     return soup1