TriBITSPub#386: - trying to avoid duplicates

tribits/doc/sphinx/sphinx_rst_generator.py

@@ -13,6 +13,8 @@
 
 
 class SphinxRstGenerator:
+    """ Changes include paths to relative to Sphinx build dir. Saves three main .rst docs files inside Sphinx dir.
+    """
     def __init__(self):
         self.paths = {
             'mainteiners_guide': {
@@ -28,6 +30,7 @@ def __init__(self):
                 'src_path': os.path.join(doc_path, 'build_ref'),
                 'final_path': os.path.join(doc_path, 'sphinx', 'build_ref_rst.rst')}}
         self.sphinx_path = os.path.abspath(os.path.join(doc_path, 'sphinx'))
+        self.already_modified_files = set()
         self.build_docs()
 
     @staticmethod
@@ -53,26 +56,17 @@ def save_rst(file_path: str, file_content: str) -> None:
         with open(file_path, 'w') as dest_file:
             dest_file.write(file_content)
 
-    @staticmethod
-    def get_src_path_from_parent() -> str:
-        pass
-
     def generate_rst(self, source_file: str, final_path: str = None, src_path: str = None,
-                     start_path: str = None) -> list:
+                     start_path: str = None) -> set:
         """ Generate corect links in .rst files, so Sphinx can find them
         """
         if final_path is None:
             overwrite_source = True
         else:
             overwrite_source = False
 
-        if src_path is None:
-            src_file_path = self.get_src_path_from_parent()
-        else:
-            src_file_path = src_path
-
-        file_content, includes = self.change_paths_and_get_includes(source_file=source_file,
-                                                                    src_file_path=src_file_path, start_path=start_path)
+        file_content, includes = self.change_paths_and_get_includes(source_file=source_file, src_file_path=src_path,
+                                                                    start_path=start_path)
 
         if overwrite_source:
             self.save_rst(file_path=source_file, file_content=file_content)
@@ -88,7 +82,7 @@ def change_paths_and_get_includes(self, source_file: str, src_file_path: str, st
         with open(source_file, 'r') as src_file:
             source_file_str = src_file.read()
             source_file_list = list()
-            include_file_list = list()
+            include_file_list = set()
             for line in source_file_str.split('\n'):
                 splitted_line = line.split()
                 if 'include::' in splitted_line:
@@ -98,7 +92,7 @@ def change_paths_and_get_includes(self, source_file: str, src_file_path: str, st
                         new_line = splitted_line[:path_index]
                         abs_path = os.path.abspath(os.path.join(src_file_path, splitted_line[path_index]))
                         if self.is_rst_file(file_path=abs_path):
-                            include_file_list.append((abs_path, src_file_path))
+                            include_file_list.add((abs_path, src_file_path))
                         rel_path_from_sphinx_dir = os.path.relpath(path=abs_path, start=start_path)
                         new_line.append(rel_path_from_sphinx_dir)
                         new_line = ' '.join(new_line)
@@ -112,24 +106,29 @@ def change_paths_and_get_includes(self, source_file: str, src_file_path: str, st
         return abs_path_str, include_file_list
 
     def main(self):
-        child_rst = list()
+        """ Main routine goes for nested .rst docs
+        """
+        child_rst = set()
         for doc_name, sources in self.paths.items():
             includes = self.generate_rst(source_file=sources.get('src'), src_path=sources.get('src_path'),
                                          final_path=sources.get('final_path'), start_path=self.sphinx_path)
-            child_rst.extend(includes)
+            child_rst.update(includes)
+        self.already_modified_files.update(child_rst)
+        child_rst_lst = list(child_rst)
 
-        grand_child_rst = list()
-        for child in child_rst:
+        grand_child_rst = set()
+        for child in child_rst_lst:
             includes_grand = self.generate_rst(source_file=child[0], src_path=os.path.split(child[0])[0],
                                                start_path=self.sphinx_path)
-            grand_child_rst.extend(includes_grand)
+            grand_child_rst.update(includes_grand)
+        grand_child_rst_lst = [gc_rst for gc_rst in grand_child_rst if gc_rst not in self.already_modified_files]
 
-        grand_grand_child_rst = list()
-        for grand_child in grand_child_rst:
+        grand_grand_child_rst = set()
+        for grand_child in grand_child_rst_lst:
             includes_grand_grand = self.generate_rst(source_file=grand_child[0],
                                                      src_path=os.path.split(grand_child[0])[0],
                                                      start_path=self.sphinx_path)
-            grand_grand_child_rst.extend(includes_grand_grand)
+            grand_grand_child_rst.update(includes_grand_grand)
 
         if not grand_grand_child_rst:
             print('DONE! ALL GOOD!')