git.r.bdr.sh - rbdr/page/blob - src/gemini_parser.rs

   1     pub fn parse(source: &str) -> String {
   2
   3     let lines = source.split("\n");
   4     let mut is_preformatted = false;
   5
   6     let mut block_label: Option<String> = None;
   7     let mut html: String = "".to_owned();
   8     let mut current_line_type: Option<LineType> = None;
   9
  10     let mut heading_stack: Vec<u8> = Vec::new();
  11     for line in lines {
  12         let mut line_type = LineType::Blank;
  13         if line.char_indices().count() > 2 {
  14             let mut end = line.len();
  15             if line.char_indices().count() > 3 {
  16                 end = line.char_indices().map(|(i, _)| i).nth(3).unwrap();
  17             }
  18             line_type = identify_line(&line[..end], is_preformatted);
  19         }
  20         match line_type {
  21             LineType::PreformattedToggle => {
  22                 is_preformatted = !is_preformatted;
  23                 if is_preformatted && line.char_indices().count() > 3 {
  24                     block_label = Some(get_partial_line_content(&line_type, line));
  25                 } else {
  26                     block_label = None;
  27                 }
  28             },
  29             _ => {
  30                 // Close previous block if needed
  31                 if let Some(line) = &current_line_type {
  32                     if line != &line_type && is_block(line) {
  33                         html.push_str(get_line_closer(line));
  34                     }
  35                 }
  36
  37                 // Blocks
  38                 if is_block(&line_type) {
  39                     if let Some(line) = &current_line_type {
  40                         if line != &line_type  {
  41                             html.push_str(&get_line_opener(&line_type, block_label.as_ref()));
  42                         }
  43                     } else {
  44                         html.push_str(&get_line_opener(&line_type, None));
  45                     }
  46
  47                     let line_content = get_partial_line_content(&line_type, line);
  48                     html.push_str(&line_content);
  49                 } else {
  50                     html.push_str(&get_heading_wrapper(&mut heading_stack, &line_type));
  51                     html.push_str(&get_full_line_content(&line_type, line));
  52                 }
  53                 current_line_type = Some(line_type);
  54             },
  55         }
  56     }
  57     if let Some(line) = &current_line_type {
  58         if is_block(line) {
  59             html.push_str(get_line_closer(line));
  60         }
  61     }
  62     html.push_str(&close_heading_wrapper(&mut heading_stack));
  63     html
  64 }
  65
  66 fn is_block(line_type: &LineType) -> bool {
  67     return match line_type {
  68         LineType::PreformattedText | LineType::ListItem | LineType::Quote => true,
  69         _ => false,
  70     }
  71 }
  72
  73 fn get_partial_line_content(line_type: &LineType, line: &str) -> String {
  74     let encoded_line = line.replace("<", "&lt;").replace(">", "&gt;");
  75     return match line_type {
  76         LineType::ListItem => format!("<li>{}</li>", encoded_line[2..].trim()),
  77         LineType::Quote => encoded_line[1..].trim().to_string(),
  78         LineType::PreformattedText => format!("{}\n", encoded_line),
  79         LineType::PreformattedToggle => encoded_line[3..].trim().to_string(),
  80         _ => "".to_string(),
  81     }
  82 }
  83
  84 fn get_full_line_content(line_type: &LineType, line: &str) -> String {
  85     let encoded_line = line.replace("<", "&lt;").replace(">", "&gt;");
  86      match line_type {
  87         LineType::Text => format!("<p>{}</p>\n", encoded_line.trim()),
  88         LineType::Blank => "<br>\n".to_string(),
  89         LineType::Link => {
  90             let url = get_link_address(line);
  91             if url.starts_with("gemini:") {
  92                 format!("<div><a href=\"{}\">{}</a></div>\n", url, get_link_content(line))
  93             } else {
  94                 format!("<div><a href=\"{}\">{}</a></div>\n", url.replace(".gmi", ".html"), get_link_content(line))
  95             }
  96         },
  97         LineType::Heading1 => format!("<h1>{}</h1>\n", encoded_line[1..].trim()),
  98         LineType::Heading2 => format!("<h2>{}</h2>\n", encoded_line[2..].trim()),
  99         LineType::Heading3 => format!("<h3>{}</h3>\n", encoded_line[3..].trim()),
 100         _ => "".to_string(),
 101     }
 102 }
 103
 104 fn get_heading_wrapper(heading_stack: &mut Vec<u8>, line_type: &LineType) -> String {
 105     let mut string = String::new();
 106     let current_heading: u8 = match line_type {
 107         LineType::Heading1 => 1,
 108         LineType::Heading2 => 2,
 109         LineType::Heading3 => 3,
 110         _ => 255
 111     };
 112
 113     if current_heading < 255 {
 114         while let Some(open_heading) = heading_stack.pop() {
 115             // You just encountered a more important heading.
 116             // Put it back. Desist.
 117             if open_heading < current_heading {
 118                 heading_stack.push(open_heading);
 119                 break;
 120             }
 121
 122             string.push_str("</div>");
 123
 124             if open_heading == current_heading {
 125                 break;
 126             }
 127         }
 128         heading_stack.push(current_heading);
 129         string.push_str(&format!("<div class=\"h{}\">", current_heading));
 130     }
 131
 132     return string;
 133 }
 134
 135 fn close_heading_wrapper(heading_stack: &mut Vec<u8>) -> String {
 136     let mut string = String::new();
 137     while let Some(_open_heading) = heading_stack.pop() {
 138         string.push_str("</div>");
 139     }
 140     return string;
 141 }
 142
 143 fn get_line_opener(line_type: &LineType, block_label: Option<&String>) -> String {
 144     match line_type {
 145         LineType::ListItem => "<ul>".to_string(),
 146         LineType::Quote => "<blockquote>".to_string(),
 147         LineType::PreformattedText => {
 148             if let Some(label) = &block_label {
 149                 return format!("<pre role=\"img\" aria-label=\"{}\">", label);
 150             } else {
 151                 return "<pre>".to_string();
 152             }
 153         },
 154         _ => "".to_string(),
 155     }
 156 }
 157
 158 fn get_line_closer(line_type: &LineType) -> &'static str {
 159     match line_type {
 160         LineType::ListItem => "</ul>\n",
 161         LineType::Quote => "</blockquote>\n",
 162         LineType::PreformattedText => "</pre>\n",
 163         _ => "",
 164     }
 165 }
 166
 167 fn get_link_content(line: &str) -> &str {
 168     let components: Vec<&str> = line[2..].trim().splitn(2, " ").collect();
 169     if components.len() > 1 {
 170         return components[1].trim()
 171     }
 172     components[0].trim()
 173 }
 174
 175 fn get_link_address(line: &str) -> &str {
 176     let components: Vec<&str> = line[2..].trim().splitn(2, " ").collect();
 177     components[0].trim()
 178 }
 179
 180 fn identify_line(line: &str, is_preformatted: bool) -> LineType {
 181     if line.starts_with("```") {
 182         return LineType::PreformattedToggle;
 183     }
 184     if is_preformatted {
 185         return LineType::PreformattedText;
 186     }
 187     if line.is_empty() {
 188         return LineType::Blank;
 189     }
 190     if line.starts_with("=>") {
 191         return LineType::Link;
 192     }
 193     if line.starts_with("* ") {
 194         return LineType::ListItem;
 195     }
 196     if line.starts_with(">") {
 197         return LineType::Quote;
 198     }
 199     if line.starts_with("###") {
 200         return LineType::Heading3;
 201     }
 202     if line.starts_with("##") {
 203         return LineType::Heading2;
 204     }
 205     if line.starts_with("#") {
 206         return LineType::Heading1;
 207     }
 208
 209     LineType::Text
 210 }
 211
 212 #[derive(PartialEq, Eq)]
 213 enum LineType {
 214     Text,
 215     Blank,
 216     Link,
 217     PreformattedToggle,
 218     PreformattedText,
 219     Heading1,
 220     Heading2,
 221     Heading3,
 222     ListItem,
 223     Quote
 224 }