Fix markdown emphasis

markdown.py

@@ -10,11 +10,13 @@ def only_chars(string, chars):

                                
                                
                                
                            
                                
                                    
                                        
                                            10  
                                            
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            11  
                                            
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            12  
                                            inlineRegex = r"""
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        13 -
                                        (?P<em>[*_]{1,7}) (?P<textEm>(?:\\[*]|[^*])*) (?P=em)                              # emphasis
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        13 +
                                        (?P<imageFlag>!?) \[ (?P<urlText>[^\[\]]*) \] \((?P<urlDestination>[^\(\)]*)\)     # hyperlink or media
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            14  
                                            |
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        15 -
                                        [`] (?P<textCode>(?:\\[`]|[^`])*) [`]                                              # inline code
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        15 +
                                        (?P<em>\*{1,7}) (?P<textEm>(?:\\\*|[^*])*) (?P=em)                                 # emphasis with * not requiring space on either side
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            16  
                                            |
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        17 -
                                        (?P<imageFlag>!?) \[ (?P<urlText>[^\[\]]*) \] \((?P<urlDestination>[^\(\)]*)\)     # hyperlink or media
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        17 +
                                        (?:^|\s)(?P<em2>_{1,7}) (?P<textEm2>(?:\\.|[^*])*) (?P=em2)(?=\s|$)                # emphasis with _ requiring space on at least one side
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        18 +
                                        |
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        19 +
                                        [`] (?P<textCode>(?:\\[`]|[^`])*) [`]                                              # inline code
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            20  
                                            |
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            21  
                                            (?P<strike>~{2}) (?P<textStrike>(?:\\[~]|[^~])*) (~{2})                            # strikethrough
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            22  
                                            |
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        

@@ -272,6 +274,8 @@ def parse_line(source):

                                
                                
                                
                            
                                
                                    
                                        
                                            274  
                                            
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            275  
                                                    if i.group("em"):
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            276  
                                                        tokens.append(Emphasis(i.group("textEm"), len(i.group("em"))))
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        277 +
                                                if i.group("em2"):
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        278 +
                                                    tokens.append(Emphasis(i.group("textEm2"), len(i.group("em2"))))
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            279  
                                                    if i.group("textCode"):
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            280  
                                                        tokens.append(Code(i.group("textCode")))
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            281  
                                                    if i.group("strike"):
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        

@@ -449,7 +453,11 @@ if __name__ == '__main__':

                                
                                
                                
                            
                                
                                    
                                        
                                            453  
                                                with open("/home/vlad/roundabout/doc/changelog/0.1.0 (2024-03-31).md") as file:
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            454  
                                                    # Generate an AST from a markdown file
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            455  
                                                    ast = tokenise(
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        452 -
                                                    file.read()
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        456 +
                                                    """
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        457 +
                                                    _test_
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        458 +
                                                    
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        459 +
                                                    and this is another _test_ by the way
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        460 +
                                                    """
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            461  
                                                    )
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            462  
                                                    # for i in ast:
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            463  
                                                    #     print(repr(i))