Start markdown parser

markdown.py

@@ -0,0 +1,89 @@

                                
                                
                                
                            
                                
                                    
                                        1 +
                                        import re
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        2 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        3 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        4 +
                                        def leading(string, character):
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        5 +
                                            return len(string) - len(string.lstrip(character))
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        6 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        7 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        8 +
                                        def trailing(string, character):
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        9 +
                                            return len(string) - len(string.rstrip(character))
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        10 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        11 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        12 +
                                        class Element:
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        13 +
                                            def __init__(self):
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        14 +
                                                pass
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        15 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        16 +
                                            def __repr__(self):
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        17 +
                                                return "Void block"
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        18 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        19 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        20 +
                                        class Heading(Element):
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        21 +
                                            def __init__(self, content, level):
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        22 +
                                                super().__init__()
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        23 +
                                                self.content = content
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        24 +
                                                self.level = level
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        25 +
                                                pass
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        26 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        27 +
                                            def __repr__(self):
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        28 +
                                                return f"Heading level {self.level}:\n\t" + self.content
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        29 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        30 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        31 +
                                        class Paragraph(Element):
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        32 +
                                            def __init__(self, content):
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        33 +
                                                super().__init__()
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        34 +
                                                self.content = content
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        35 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        36 +
                                            def addLine(self, content):
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        37 +
                                                self.content += content.strip() + " "
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        38 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        39 +
                                            def __repr__(self):
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        40 +
                                                return "Paragraph:\n\t" + self.content
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        41 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        42 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        43 +
                                        def _tokenise(source):
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        44 +
                                            tokens = []
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        45 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        46 +
                                            currentBlock = Element
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        47 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        48 +
                                            for line in source.split("\n"):
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        49 +
                                                if not line.strip():
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        50 +
                                                    # Void block
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        51 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        52 +
                                                    tokens.append(currentBlock)
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        53 +
                                                    currentBlock = Element()
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        54 +
                                                elif line.startswith("#") and leading(line.lstrip("#"), " ") == 1:
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        55 +
                                                    tokens.append(currentBlock)
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        56 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        57 +
                                                    content = line.lstrip("#").strip()
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        58 +
                                                    currentBlock = Heading(content, leading(line, "#"))
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        59 +
                                                else:
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        60 +
                                                    if not isinstance(currentBlock, Paragraph):
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        61 +
                                                        tokens.append(currentBlock)
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        62 +
                                                        currentBlock = Paragraph("")
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        63 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        64 +
                                                    currentBlock.addLine(line)
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        65 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        66 +
                                            tokens.append(currentBlock)
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        67 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        68 +
                                            return tokens
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        69 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        70 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        71 +
                                        for i in _tokenise(
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        72 +
                                            """
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        73 +
                                        # Hello World!
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        74 +
                                        ## Title 1
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        75 +
                                        ### Part 1
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        76 +
                                        #### Chapter 1
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        77 +
                                        ##### Article 1
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        78 +
                                        ###### Section 1
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        79 +
                                        Lorem ipsum
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        80 +
                                        dolor sit amet"""
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        81 +
                                        ):
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        82 +
                                            print(repr(i))
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        83 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        84 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        85 +
                                        def parseMarkdown(source):
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        86 +
                                            tokens = _tokenise(source)
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        87 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        88 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        89 +
                                        parseMarkdown("")