add table processing to converter

src/book.py

@@ -48,6 +48,7 @@ class Book:
         self.footnotes = list()
         self.images = list()
         self.content_dict = dict()
+        self.tables_amount = 0
 
         assert self.SUPPORTED_LEVELS == len(self.SUPPORTED_HEADERS), \
             "Length of headers doesn't match allowd levels."
@@ -266,7 +267,11 @@ class Book:
                 u[0].unwrap()
 
     @classmethod
-    def convert_pt_to_px(cls, style):
+    def convert_pt_to_px(cls, value):
+        return round(cls.FONT_CONVERT_RATIO * float(value))
+
+    @classmethod
+    def convert_font_pt_to_px(cls, style):
         """
         Method converts point in the font-size to pixels.
 
@@ -279,7 +284,7 @@ class Book:
             return style
 
         size = size.group(1)
-        new_size = round(cls.FONT_CONVERT_RATIO * float(size))
+        new_size = cls.convert_pt_to_px(size)
 
         if new_size == cls.LAWCARTA_DEFAULT_FONT_SIZE:
             return ""
@@ -298,7 +303,7 @@ class Book:
             font.attrs = {}
             font.name = "span"
             if style:
-                style = self.convert_pt_to_px(style)
+                style = self.convert_font_pt_to_px(style)
                 if style != "":
                     font.attrs["style"] = style
             if face is not None:
@@ -402,6 +407,42 @@ class Book:
                     child["class"] = "columns2"
             div.unwrap()
 
+    def _process_tables(self):
+        """
+        Function to process tables. Set "border" attribute.
+        """
+        tables = self.body_tag.find_all("table")
+        for table in tables:
+            tds = table.find_all("td")
+
+            sizes = []
+            for td in tds:
+                style = td.get('style')
+
+                if style:
+                    match = re.search(r"border: ?(\d+\.?\d*)(p[tx])", style)
+
+                    if match:
+                        size = match.group(1)
+                        units = match.group(2)
+
+                        if units == "pt":
+                            size = self.convert_pt_to_px(size)
+
+                        sizes.append(float(size))
+
+                width = td.get('width')
+
+                td.attrs = {}
+                if width:
+                    td.attrs['width'] = width
+
+            if sizes:
+                border_size = sum(sizes)/len(sizes)
+                table.attrs['border'] = f'{border_size:.2}'
+
+        self.tables_amount = len(tables)
+
     # def _process_quotes(self):
     #     """
     #     Function to process <dl> tags. All tags will be replaced with <blockquote> tags.
@@ -603,6 +644,10 @@ class Book:
             self._process_two_columns()
             # self._process_quotes()
 
+            self.log('Tables processing.')
+            self._process_tables()
+            self.log(f'{self.tables_amount} tables have been processed.')
+
             self.log('Footnotes processing.')
             self._process_footnotes()
             self.log(f'{len(self.footnotes)} footnotes have been processed.')