epub converter: prettify css_reader.py

src/css_reader.py

@@ -1,4 +1,6 @@
 import re
+from typing import List
+
 import cssutils
 
 from bs4 import BeautifulSoup
@@ -18,9 +20,8 @@ sizes_pr = [-1, 0.5, 0.56, 0.63, 0.69, 0.75, 0.81, 0.88, 0.94, 1.0, 1.06, 1.13,
 
 sizes_px = ['10px', '10px', '11px', '12px', '13px', '14px', '15px', '16px', '17px', '18px', '19px', '20px', '21px',
             '22px', '23px', '24px', '25px', '26px', '27px', '28px', '29px', '30px', '31px', '32px', '33px', '34px',
-            '35px',
+            '35px', '36px', '37px', '38px', '39px', '40px', '41px', '42px', '43px', '44px', '45px', '46px', '47px',
-            '36px', '37px', '38px', '39px', '40px', '41px', '42px', '43px', '44px', '45px', '46px', '47px', '48px',
+            '48px', '49px', '50px', '64px', '72px']
-            '49px', '50px', '64px', '72px']
 
 
 def convert_font_size(value):
@@ -52,6 +53,13 @@ def convert_font_size(value):
         return ''
 
 
+""" 
+LIVECARTA_STYLE_ATTRS = { css property: value }
+
+Style properties that can be used to fit livecarta css style convention.
+If property has empty list, it means that any value can be converted.
+If property has not empty list, it means that only certain property-value combinations can be transformed.
+"""
 LIVECARTA_STYLE_ATTRS = {
     'text-indent': [],
     'font-variant': ['small-caps'],
@@ -69,6 +77,13 @@ LIVECARTA_STYLE_ATTRS = {
     'color': [],
     'background-color': [],
 }
+
+"""
+LIVECARTA_STYLE_ATTRS_MAPPING = { property: mapping function }
+
+Warning, if LIVECARTA_STYLE_ATTRS is changed, LIVECARTA_STYLE_ATTRS_MAPPING should be updated 
+to suit livecarta style convention.
+"""
 LIVECARTA_STYLE_ATTRS_MAPPING = {
     'text-indent': lambda x: LawCartaConfig.INDENT,
     'font-variant': lambda x: x,
@@ -80,6 +95,9 @@ LIVECARTA_STYLE_ATTRS_MAPPING = {
     'background-color': lambda x: LawCartaConfig.HTML42LIVECARTA_COLORS.get(str2color_name(x), ''),
 }
 
+"""
+LIVECARTA_STYLE_ATTRS_SHOULD_BE_TAG = { (property, value): tag }
+"""
 LIVECARTA_STYLE_ATTRS_SHOULD_BE_TAG = {
     ('font-weight', 'bold'): 'strong',
     ('font-weight', '600'): 'strong',
@@ -95,7 +113,9 @@ LIVECARTA_STYLE_ATTRS_SHOULD_BE_TAG = {
 }
 
 
-def check_style_to_be_tag(style):
+def check_style_to_be_tag(style) -> List[tuple]:
+    """ Some css style properties converts to tags.
+    Search for them and prepare list of properties to be removed from style string"""
     to_remove = []
     for k in LIVECARTA_STYLE_ATTRS_SHOULD_BE_TAG:
         if f'{k[0]}:{k[1]}' in style:
@@ -103,6 +123,29 @@ def check_style_to_be_tag(style):
     return to_remove
 
 
+def update_property_to_livecarta_convention(rule, property_):
+    if property_.name not in LIVECARTA_STYLE_ATTRS:
+        # property not in LIVECARTA_STYLE_ATTRS, remove from css file
+        rule.style[property_.name] = ''
+
+    elif LIVECARTA_STYLE_ATTRS.get(property_.name):
+        # check property value to decide weather to remove or not the property_
+        cleaned_property = property_.value.replace('\"', '')
+        if cleaned_property in LIVECARTA_STYLE_ATTRS[property_.name]:
+
+            if property_.name in LIVECARTA_STYLE_ATTRS_MAPPING:
+                # apply transformation
+                func = LIVECARTA_STYLE_ATTRS_MAPPING[property_.name]
+                rule.style[property_.name] = func(cleaned_property)
+        else:
+            # property + value not in LIVECARTA_STYLE_ATTRS, remove from css file
+            rule.style[property_.name] = ''
+    else:
+        # property can have any value
+        if property_.name in LIVECARTA_STYLE_ATTRS_MAPPING:
+            func = LIVECARTA_STYLE_ATTRS_MAPPING[property_.name]
+            cleaned_property = property_.value.replace('\"', '')
+            rule.style[property_.name] = func(cleaned_property)
 
 
 def clean_css(css):
@@ -110,35 +153,18 @@ def clean_css(css):
     for rule in sheet:
         if rule.type == rule.STYLE_RULE:
             for property_ in rule.style:
-
+                update_property_to_livecarta_convention(rule, property_)
-                if property_.name not in LIVECARTA_STYLE_ATTRS:
-                    rule.style[property_.name] = ''
-                # not remove based on property value
-                elif LIVECARTA_STYLE_ATTRS.get(property_.name):
-                    tmp = property_.value.replace('\"', '')
-                    if tmp in LIVECARTA_STYLE_ATTRS[property_.name]:
-                        if property_.name in LIVECARTA_STYLE_ATTRS_MAPPING:
-                            func = LIVECARTA_STYLE_ATTRS_MAPPING[property_.name]
-                            tmp = property_.value.replace('\"', '')
-                            rule.style[property_.name] = func(tmp)
-                    else:
-                        rule.style[property_.name] = ''
-                else:
-                    if property_.name in LIVECARTA_STYLE_ATTRS_MAPPING:
-                        func = LIVECARTA_STYLE_ATTRS_MAPPING[property_.name]
-                        tmp = property_.value.replace('\"', '')
-                        rule.style[property_.name] = func(tmp)
 
     css_text = sheet._getCssText().decode()
     return css_text
 
 
 def add_inline_style_to_html_soup(soup1, css_text):
-    livecarta_p_ids = []
+    livecarta_tmp_ids = []
     h_regex = f'(^h[{LawCartaConfig.SUPPORTED_LEVELS + 1}-9]$)'
-    for i, x in enumerate(soup1.find_all(re.compile('(^p$)|(^span$)|(^li$)|(^ul$)' + h_regex))):
+    for i, x in enumerate(soup1.find_all(re.compile('(^p$)|(^span$)|(^li$)|(^ul$)|(^ol$)' + h_regex))):
         x.attrs['livecarta_id'] = i
-        livecarta_p_ids.append(i)
+        livecarta_tmp_ids.append(i)
 
     html_with_inline_style = transform(str(soup1), css_text=css_text,
                                        remove_classes=False,
@@ -146,7 +172,7 @@ def add_inline_style_to_html_soup(soup1, css_text):
                                        disable_validation=True)
     soup2 = BeautifulSoup(html_with_inline_style, features='lxml')
 
-    for i in livecarta_p_ids:
+    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})

src/util/color_reader.py

@@ -1,13 +1,13 @@
 from webcolors import html4_hex_to_names, hex_to_rgb, rgb_to_name
 
 
-def closest_colour_rgb(requested_colour):
+def closest_colour_rgb(requested_color):
     min_colours = {}
     for key, name in html4_hex_to_names.items():
         r_c, g_c, b_c = hex_to_rgb(key)
-        rd = (r_c - requested_colour[0]) ** 2
+        rd = (r_c - requested_color[0]) ** 2
-        gd = (g_c - requested_colour[1]) ** 2
+        gd = (g_c - requested_color[1]) ** 2
-        bd = (b_c - requested_colour[2]) ** 2
+        bd = (b_c - requested_color[2]) ** 2
         min_colours[(rd + gd + bd)] = name
 
     return min_colours[min(min_colours.keys())]