Change names

src/docx_converter/docx_solver.py

@@ -5,8 +5,8 @@ from threading import Event
 
 from src.book_solver import BookSolver
 from src.util.helpers import BookLogger
-from src.html_preprocessor import HtmlPreprocessor
+from src.html_presets_processor import HtmlPresetsProcessor
-from src.style_preprocessor import StylePreprocessor
+from src.style_reader import StyleReader
 from src.docx_converter.docx2libre_html import Docx2LibreHtml
 from src.docx_converter.html_docx_processor import HtmlDocxProcessor
 from src.docx_converter.libre_html2json_converter import LibreHtml2JsonConverter
@@ -49,9 +49,9 @@ class DocxBook(BookSolver):
 
         # 2. Parses and cleans html, gets list of tags, gets footnotes
         try:
-            html_preprocessor = HtmlPreprocessor(
+            html_preprocessor = HtmlPresetsProcessor(
                 logger=self.logger_object, preset_path="presets/docx_presets.json")
-            style_preprocessor = StylePreprocessor()
+            style_preprocessor = StyleReader()
             html_processor = HtmlDocxProcessor(html_soup=html_converter.html_soup,
                                                logger=self.logger_object,
                                                html_preprocessor=html_preprocessor,
@@ -80,7 +80,7 @@ class DocxBook(BookSolver):
 
 
 if __name__ == "__main__":
-    docx_file_path = "../../books/docx/Bar_Exam_MPT_2e_prepared.docx"
+    docx_file_path = "../../books/docx/AmericanGovernment3e-WEB.docx"
     logger_object = BookLogger(
         name="docx", book_id=docx_file_path.split("/")[-1])
     locker = Event()
@@ -89,9 +89,9 @@ if __name__ == "__main__":
     html_converter = Docx2LibreHtml(file_path=docx_file_path,
                                     logger=logger_object, libre_locker=locker)
 
-    html_preprocessor = HtmlPreprocessor(
+    html_preprocessor = HtmlPresetsProcessor(
         logger=logger_object, preset_path="../../presets/docx_presets.json")
-    style_preprocessor = StylePreprocessor()
+    style_preprocessor = StyleReader()
     html_processor = HtmlDocxProcessor(html_soup=html_converter.html_soup, logger=logger_object,
                                        html_preprocessor=html_preprocessor, style_preprocessor=style_preprocessor)
     content, footnotes, top_level_headers = html_processor.process_html(

src/html_presets_processor.py

@@ -7,7 +7,7 @@ from typing import List, Dict, Union
 from src.util.helpers import BookLogger
 
 
-class HtmlPreprocessor:
+class HtmlPresetsProcessor:
     def __init__(self, logger: BookLogger, preset_path):
         self.preset = json.load(open(preset_path))
         self.logger = logger
@@ -107,6 +107,8 @@ class HtmlPreprocessor:
             del kwargs["tag"][attr]
         elif attr_value_to_replace:
             kwargs["tag"].attrs[attr] = attr_value_to_replace
+        elif attr:
+            del kwargs["tag"][attr]
 
     @staticmethod
     def _unwrap_tag(**kwargs):
@@ -172,7 +174,7 @@ class HtmlPreprocessor:
                     action(body_tag=body_tag, tag=tag, rule=rule)
 
 
-def _preprocess_html(html_preprocessor: HtmlPreprocessor, html_soup: BeautifulSoup):
+def _process_presets(html_preprocessor: HtmlPresetsProcessor, html_soup: BeautifulSoup):
     for rule in html_preprocessor.preset:
         # html_preprocessor.logger.log(rule["preset_name"].title() + " process.")
         action = html_preprocessor.name2action[rule["preset_name"]]

src/inline_style_processor.py

@@ -10,7 +10,7 @@ from src.livecarta_config import LiveCartaConfig
 cssutils.log.setLevel(CRITICAL)
 
 
-class TagInlineStyleProcessor:
+class InlineStyleProcessor:
     def __init__(self, tag_inline_style: Tag):
         # tag with inline style + style parsed from css file
         self.tag_inline_style = tag_inline_style
@@ -251,6 +251,6 @@ def modify_html_soup_with_css_styles(html_soup: BeautifulSoup, css_text: str = "
 
     # go through the tags with inline style + style parsed from css file
     for tag_inline_style in tags_with_inline_style:
-        style_converter = TagInlineStyleProcessor(tag_inline_style)
+        style_converter = InlineStyleProcessor(tag_inline_style)
         style_converter.convert_initial_tag()
     return inline_soup

src/style_reader.py

@@ -7,7 +7,7 @@ from src.util.color_reader import str2hex
 from src.livecarta_config import LiveCartaConfig
 
 
-class StylePreprocessor:
+class StyleReader:
     def __init__(self):
         """
         Dictionary LIVECARTA_STYLE_ATTRS_MAPPING = { property: mapping function }