Markdown regex

markdown.py

@@ -1,6 +1,19 @@

                                
                                
                                
                            
                                
                                    
                                        
                                            1  
                                            import re
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            2  
                                            
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            3  
                                            
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        4 +
                                        inlineRegex = r"""
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        5 +
                                        (?P<em>[*_]{1,7}) (?P<textEm>(?:\\[*]|[^*])*) (?P=em)                              # emphasis
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        6 +
                                        |
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        7 +
                                        [`] (?P<textCode>(?:\\[`]|[^`])*) [`]                                              # inline code
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        8 +
                                        |
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        9 +
                                        (?P<imageFlag>!?) \[ (?P<urlText>[^\[\]]*) \] \((?P<urlDestination>[^\(\)]*)\)     # hyperlink or media
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        10 +
                                        |
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        11 +
                                        (?P<strike>~{2}) (?P<textStrike>(?:\\[~]|[^~])*) (~{2})                            # strikethrough
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        12 +
                                        |
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        13 +
                                        (?P<diff>\-\-|\+\+) (?P<textDiff>(?:\\[-+]|[^-+])*) (?P=diff)                      # diffs
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        14 +
                                        """
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        15 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        16 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            17  
                                            def leading(string, character):
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            18  
                                                return len(string) - len(string.lstrip(character))
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            19  
                                            
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        

@@ -40,6 +53,25 @@ class Paragraph(Element):

                                
                                
                                
                            
                                
                                    
                                        
                                            53  
                                                    return "Paragraph:\n\t" + self.content
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            54  
                                            
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            55  
                                            
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        56 +
                                        class Emphasis(Element):
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        57 +
                                            def __init__(self, content, bold, italic):
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        58 +
                                                super().__init__()
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        59 +
                                                self.content = content
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        60 +
                                                self.bold = bold
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        61 +
                                                self.italic = italic
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        62 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        63 +
                                            def __repr__(self):
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        64 +
                                                return f"Emphasis (bold {self.bold}, italic {self.italic}):\n\t" + self.content
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        65 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        66 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        67 +
                                        def _parseLine(source):
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        68 +
                                            if trailing(source, "\\") == 1:
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        69 +
                                                source = source.rstrip("\\")
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        70 +
                                                source += "\n"
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        71 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        72 +
                                            tokens = [source]
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        73 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        74 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            75  
                                            def _tokenise(source):
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            76  
                                                tokens = []
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            77  
                                            
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        

@@ -58,6 +90,8 @@ def _tokenise(source):

                                
                                
                                
                            
                                
                                    
                                        
                                            90  
                                                        currentBlock = Heading(content, leading(line, "#"))
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            91  
                                                    else:
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            92  
                                                        if not isinstance(currentBlock, Paragraph):
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        93 +
                                                        # Paragraph is default
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        94 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            95  
                                                            tokens.append(currentBlock)
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            96  
                                                            currentBlock = Paragraph("")
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            97  
                                            
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        

@@ -77,7 +111,9 @@ for i in _tokenise(

                                
                                
                                
                            
                                
                                    
                                        
                                            111  
                                            ##### Article 1
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            112  
                                            ###### Section 1
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            113  
                                            Lorem ipsum
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        80 -
                                        dolor sit amet"""
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        114 +
                                        dolor sit amet
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        115 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        116 +
                                        ..."""
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            117  
                                            ):
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            118  
                                                print(repr(i))
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            119