create_manpage_completions.py: introduce TypeScdocManParser which is capable of parsing scdoc manpages

This greatly improves generated completions for scdoc man pages, see #7187.

share/tools/create_manpage_completions.py

@@ -502,6 +502,56 @@ class Type4ManParser(ManParser):
             options_section = options_section[options_matched.end() - 3 :]
             options_matched = re.search(options_parts_regex, options_section)
 
+class TypeScdocManParser(ManParser):
+    def is_my_type(self, manpage):
+        return compile_and_search(r"\.(\\)(\") Generated by scdoc(.*?)\.SH OPTIONS(.*?)", manpage) != None
+
+    def parse_man_page(self, manpage):
+        options_section_regex = re.compile("\.SH OPTIONS(.*?)\.SH", re.DOTALL)
+        options_section_matched = re.search(options_section_regex, manpage)
+        if options_section_matched == None:
+            return False
+        options_section = options_section_matched.group(1)
+
+        options_parts_regex = re.compile("(.*?).RE", re.DOTALL)
+        options_matched = re.match(options_parts_regex, options_section)
+        add_diagnostic("Command is %r" % CMDNAME)
+
+        if options_matched == None:
+            add_diagnostic("%r: Unable to find options" % self)
+            return False
+
+        while options_matched != None:
+            # Get first option and move options_section
+            option = options_matched.group(1)
+            options_section = options_section[options_matched.end()::]
+            options_matched = re.match(options_parts_regex, options_section)
+
+            option = remove_groff_formatting(option)
+            option = option.split("\n")
+
+            # Crean option list
+            option_clean = []
+            for line in option:
+                if line not in ["", ".P", ".RS 4"]:
+                    option_clean.append(line)
+
+            # Shold be at least two lines
+            if len(option_clean) < 2:
+                add_diagnostic("Unable to split option from description")
+                continue
+            
+            # Name and description, others lines are ignored
+            option_name = option_clean[0]
+            option_description = option_clean[1]
+
+            if "-" not in option_name:
+                add_diagnostic("%r doesn't contain '-'" % option_name)
+
+            option_name = unquote_double_quotes(option_name)
+            option_name = unquote_single_quotes(option_name)
+            built_command(option_name, option_description)
+
         return True
 
 
@@ -780,6 +830,7 @@ def parse_manpage_at_path(manpage_path, output_directory):
         parsers = [TypeDeroffManParser()]
     else:
         parsers = [
+            TypeScdocManParser(),
             Type1ManParser(),
             Type2ManParser(),
             Type4ManParser(),