Added gemtext parser

_src/gemtext.py

@@ -0,0 +1,82 @@
+import re
+
+line_patterns = {
+    'link': re.compile(r'=>(\s+)(\S+)(\s+(.*))?'),
+    'preformatted': re.compile(r'```(.*)'),
+    'list': re.compile(r'\*\s+(.*)'),
+    'head1': re.compile(r'#\s+(.*)'),
+    'head2': re.compile(r'##\s+(.*)'),
+    'head3': re.compile(r'###\s+(.*)'),
+    'quote': re.compile(r'>\s+(.*)')
+}
+
+class GemtextParser:
+    """Provides abstract representation of gemtext file, that can be used in
+    conversion to html"""
+    
+    elements = []
+    
+    def __init__(self, content: str):
+        # add blank element to refer to as last_item in loop at start
+        self.elements.append(GemtextElement(''))
+        self.elements[0].type = 'blank'
+        
+        for line in content.split('\n'):
+            last_item = self.elements[-1]
+            if last_item.type == 'preformatted' and not last_item.closed:
+                if line == '```':
+                    last_item.closed = True
+                else:
+                    last_item.add_line(line)
+            else:
+                new = GemtextElement(line)
+                if last_item.type == new.type:
+                    if new.type in ('plain', 'quote'):
+                        last_item.add_line(new.content)
+                    elif new.type == 'list':
+                        last_item.items += new.items
+                    else:
+                        self.elements.append(new)
+                else:
+                    self.elements.append(new)
+        
+        # remove blank element
+        self.elements.pop(0)
+
+class GemtextElement:
+    """Represents single Gemtext element. It is created from single line and
+    extended later.
+    """
+    type = 'plain'
+    content = None
+    
+    def __init__(self, line: str):
+        global line_pattern
+        
+        for type, pattern in line_patterns.items():
+            if match := pattern.match(line):
+                self.type = type
+                
+                if type == 'link':
+                    self.href = match.group(2)
+                    self.label = match.group(4)
+                elif type == 'list':
+                    self.items = [match.group(1)]
+                elif type == 'preformatted':
+                    self.alt = match.group(1) or None
+                    self.content = ''
+                    # to distinguish two consecutive preformatted blocks:
+                    self.closed = False
+                else:
+                    self.content = match.group(1)
+                
+                break # there is no point of testing further
+        
+        if self.type == 'plain':
+            self.content = line
+    
+    def add_line(self, line: str):
+        if self.content == None: return
+        if self.content == '':
+            self.content = line or '\n' # add newline on empty line
+        else: self.content += '\n'+line