commit 4d0cc0f2fc863efde75c3772c36bac1159f02131 · zenfyr.dev/xpost

+80 -101

util/html_util.py

···

       5
        
           def __init__(self) -> None:

     

       6
        
               super().__init__()

     

       7
        
               self.tokens: list[cross.Token] = []

     

       8
       -
               self.status: dict

     

       9
        
               

     

       10
        
               self.mentions: list[tuple[str, str]]

     

       11
        
               self.tags: list[str]

     
···

       28
        
                       if isinstance(last_token, cross.TextToken) and not last_token.text.endswith('\n'):

     

       29
        
                           self.tokens.append(cross.TextToken('\n'))

     

       30
        
               

     

       31
       -
               if tag == 'br':

     

       32
       -
                   self.tokens.append(cross.TextToken('  \n'))

     

       33
       -
               

     

       34
       -
               elif tag == 'a':

     

       35
       -
                   href = attrs_dict.get('href', '')

     

       36
       -
                   self.anchor_stack.append(href)

     

       37
       -
               

     

       38
       -
               elif tag == 'strong' or tag == 'b':

     

       39
       -
                   self.tokens.append(cross.TextToken('**'))

     

       40
       -
               

     

       41
       -
               elif tag == 'em' or tag == 'i':

     

       42
       -
                   self.tokens.append(cross.TextToken('*'))

     

       43
       -
                   

     

       44
       -
               elif tag == 'del' or tag == 's':

     

       45
       -
                   self.tokens.append(cross.TextToken('~~'))

     

       46
       -
                   

     

       47
       -
               elif tag == 'code':

     

       48
       -
                   if not self.in_pre:

     

       49
       -
                       self.tokens.append(cross.TextToken('`'))

     

       50
       -
                       self.in_code = True

     

       51
       -
               

     

       52
       -
               elif tag == 'pre':

     

       53
       -
                   append_newline()

     

       54
       -
                   self.tokens.append(cross.TextToken('```\n'))

     

       55
       -
                   self.in_pre = True

     

       56
       -
                       

     

       57
       -
               elif tag == 'blockquote':

     

       58
       -
                   append_newline()

     

       59
       -
                   self.tokens.append(cross.TextToken('\n> '))

     

       60
       -
               

     

       61
       -
               elif tag == 'ul':

     

       62
       -
                   self.list_stack.append('ul')

     

       63
       -
                   append_newline()

     

       64
       -
                   

     

       65
       -
               elif tag == 'ol':

     

       66
       -
                   self.list_stack.append('ol')

     

       67
       -
                   append_newline()

     

       68
       -
               

     

       69
       -
               elif tag == 'li':

     

       70
       -
                   indent = '  ' * (len(self.list_stack) - 1)

     

       71
       -
                   if self.list_stack and self.list_stack[-1] == 'ul':

     

       72
       -
                       self.tokens.append(cross.TextToken(f'{indent}- '))

     

       73
       -
                   elif self.list_stack and self.list_stack[-1] == 'ol':

     

       74
       -
                       self.tokens.append(cross.TextToken(f'{indent}1. '))

     

       75
       -
               

     

       76
       -
               elif tag == {'h1', 'h2', 'h3', 'h4', 'h5', 'h6'}:

     

       77
       -
                   level = int(tag[1])

     

       78
       -
                   self.tokens.append(cross.TextToken("\n" + "#" * level + " "))

     

       79
        
               

     

       80
        
               self.current_tag_stack.append(tag)

     

       81
        
           

     
···

       92
        
               if tag in self.current_tag_stack:

     

       93
        
                   self.current_tag_stack.remove(tag)

     

       94
        
               

     

       95
       -
               if tag == 'p':

     

       96
       -
                   self.tokens.append(cross.TextToken('\n\n'))

     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       97
        
                   

     

       98
       -
               elif tag == 'a':

     

       99
       -
                   href = self.anchor_stack.pop()

     

       100
       -
                   anchor_data = ''.join(self.anchor_data)

     

       101
       -
                   self.anchor_data = []

     

       102
       -
                   

     

       103
       -
                   if anchor_data.startswith('#'):

     

       104
       -
                       as_tag = anchor_data[1:].lower()

     

       105
       -
                       if any(as_tag == block for block in self.tags):

     

       106
       -
                           self.tokens.append(cross.TagToken(anchor_data[1:]))

     

       107
       -
                   elif anchor_data.startswith('@'):

     

       108
       -
                       match = next(

     

       109
       -
                           (pair for pair in self.mentions if anchor_data in pair),

     

       110
       -
                           None

     

       111
       -
                       )

     

       112
        
                       

     

       113
       -
                       if match:

     

       114
       -
                           self.tokens.append(cross.MentionToken(match[1], ''))

     

       115
       -
                   else:

     

       116
       -
                       self.tokens.append(cross.LinkToken(href, anchor_data))

     

       117
       -
               

     

       118
       -
               elif tag == 'strong' or tag == 'b':

     

       119
       -
                   self.tokens.append(cross.TextToken('**'))

     

       120
       -
               

     

       121
       -
               elif tag == 'em' or tag == 'i':

     

       122
       -
                   self.tokens.append(cross.TextToken('*'))

     

       123
       -
               

     

       124
       -
               elif tag == 'del' or tag == 's':

     

       125
       -
                   self.tokens.append(cross.TextToken('~~'))

     

       126
       -
               

     

       127
       -
               elif tag == 'code':

     

       128
       -
                   if not self.in_pre and self.in_code:

     

       129
       -
                       self.tokens.append(cross.TextToken('`'))

     

       130
       -
                       self.in_code = False

     

       131
       -
               

     

       132
       -
               elif tag == 'pre':

     

       133
       -
                   self.tokens.append(cross.TextToken('\n```\n'))

     

       134
       -
                   self.in_pre = False

     

       135
       -
       

     

       136
       -
               elif tag == 'blockquote':

     

       137
       -
                   self.tokens.append(cross.TextToken('\n'))

     

       138
       -
               

     

       139
       -
               elif tag == 'ul' or tag == 'ol':

     

       140
       -
                   if self.list_stack:

     

       141
       -
                       self.list_stack.pop()

     

       142
       -
                   self.tokens.append(cross.TextToken('\n'))

     

       143
       -
               

     

       144
       -
               elif tag == 'li':

     

       145
       -
                   self.tokens.append(cross.TextToken('\n'))

     

       146
       -
               

     

       147
       -
               elif tag in ['h1', 'h2', 'h3', 'h4', 'h5', 'h6']:

     

       148
       -
                   self.tokens.append(cross.TextToken('\n'))

     

       149
        
           

     

       150
        
           def get_tokens(self) -> list[cross.Token]:

     

       151
        
               if not self.tokens:

···

       5
        
           def __init__(self) -> None:

     

       6
        
               super().__init__()

     

       7
        
               self.tokens: list[cross.Token] = []

     

       0
        
       
     

       8
        
               

     

       9
        
               self.mentions: list[tuple[str, str]]

     

       10
        
               self.tags: list[str]

     
···

       27
        
                       if isinstance(last_token, cross.TextToken) and not last_token.text.endswith('\n'):

     

       28
        
                           self.tokens.append(cross.TextToken('\n'))

     

       29
        
               

     

       30
       +
               match tag:

     

       31
       +
                   case 'br':

     

       32
       +
                       self.tokens.append(cross.TextToken('  \n'))

     

       33
       +
                   case 'a':

     

       34
       +
                       href = attrs_dict.get('href', '')

     

       35
       +
                       self.anchor_stack.append(href)

     

       36
       +
                   case 'strong', 'b':

     

       37
       +
                       self.tokens.append(cross.TextToken('**'))

     

       38
       +
                   case 'em', 'i':

     

       39
       +
                       self.tokens.append(cross.TextToken('*'))

     

       40
       +
                   case 'del', 's':

     

       41
       +
                       self.tokens.append(cross.TextToken('~~'))

     

       42
       +
                   case 'code':

     

       43
       +
                       if not self.in_pre:

     

       44
       +
                           self.tokens.append(cross.TextToken('`'))

     

       45
       +
                           self.in_code = True

     

       46
       +
                   case 'pre':

     

       47
       +
                       append_newline()

     

       48
       +
                       self.tokens.append(cross.TextToken('```\n'))

     

       49
       +
                       self.in_pre = True

     

       50
       +
                   case 'blockquote':

     

       51
       +
                       append_newline()

     

       52
       +
                       self.tokens.append(cross.TextToken('> '))

     

       53
       +
                   case 'ul', 'ol':

     

       54
       +
                       self.list_stack.append(tag)

     

       55
       +
                       append_newline()

     

       56
       +
                   case 'li':

     

       57
       +
                       indent = '  ' * (len(self.list_stack) - 1)

     

       58
       +
                       if self.list_stack and self.list_stack[-1] == 'ul':

     

       59
       +
                           self.tokens.append(cross.TextToken(f'{indent}- '))

     

       60
       +
                       elif self.list_stack and self.list_stack[-1] == 'ol':

     

       61
       +
                           self.tokens.append(cross.TextToken(f'{indent}1. '))

     

       62
       +
                   case _:

     

       63
       +
                       if tag in {'h1', 'h2', 'h3', 'h4', 'h5', 'h6'}:

     

       64
       +
                           level = int(tag[1])

     

       65
       +
                           self.tokens.append(cross.TextToken("\n" + "#" * level + " "))

     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       66
        
               

     

       67
        
               self.current_tag_stack.append(tag)

     

       68
        
           

     
···

       79
        
               if tag in self.current_tag_stack:

     

       80
        
                   self.current_tag_stack.remove(tag)

     

       81
        
               

     

       82
       +
               match tag:

     

       83
       +
                   case 'p':

     

       84
       +
                       self.tokens.append(cross.TextToken('\n\n'))

     

       85
       +
                   case 'a':

     

       86
       +
                       href = self.anchor_stack.pop()

     

       87
       +
                       anchor_data = ''.join(self.anchor_data)

     

       88
       +
                       self.anchor_data = []

     

       89
        
                   

     

       90
       +
                       if anchor_data.startswith('#'):

     

       91
       +
                           as_tag = anchor_data[1:].lower()

     

       92
       +
                           if any(as_tag == block for block in self.tags):

     

       93
       +
                               self.tokens.append(cross.TagToken(anchor_data[1:]))

     

       94
       +
                       elif anchor_data.startswith('@'):

     

       95
       +
                           match = next(

     

       96
       +
                              (pair for pair in self.mentions if anchor_data in pair),

     

       97
       +
                               None

     

       98
       +
                           )

     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       99
        
                       

     

       100
       +
                           if match:

     

       101
       +
                               self.tokens.append(cross.MentionToken(match[1], ''))

     

       102
       +
                       else:

     

       103
       +
                           self.tokens.append(cross.LinkToken(href, anchor_data))

     

       104
       +
                   case 'strong', 'b':

     

       105
       +
                       self.tokens.append(cross.TextToken('**'))

     

       106
       +
                   case 'em', 'i':

     

       107
       +
                       self.tokens.append(cross.TextToken('*'))

     

       108
       +
                   case 'del', 's':

     

       109
       +
                       self.tokens.append(cross.TextToken('~~'))

     

       110
       +
                   case 'code':

     

       111
       +
                       if not self.in_pre and self.in_code:

     

       112
       +
                           self.tokens.append(cross.TextToken('`'))

     

       113
       +
                           self.in_code = False

     

       114
       +
                   case 'pre':

     

       115
       +
                       self.tokens.append(cross.TextToken('\n```\n'))

     

       116
       +
                       self.in_pre = False

     

       117
       +
                   case 'blockquote':

     

       118
       +
                       self.tokens.append(cross.TextToken('\n'))

     

       119
       +
                   case 'ul', 'ol':

     

       120
       +
                       if self.list_stack:

     

       121
       +
                           self.list_stack.pop()

     

       122
       +
                       self.tokens.append(cross.TextToken('\n'))

     

       123
       +
                   case 'li':

     

       124
       +
                       self.tokens.append(cross.TextToken('\n'))

     

       125
       +
                   case _:

     

       126
       +
                       if tag in ['h1', 'h2', 'h3', 'h4', 'h5', 'h6']:

     

       127
       +
                           self.tokens.append(cross.TextToken('\n'))

     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       128
        
           

     

       129
        
           def get_tokens(self) -> list[cross.Token]:

     

       130
        
               if not self.tokens: