src/librustdoc/html/highlight.rs

   1 // Copyright 2014 The Rust Project Developers. See the COPYRIGHT
   2 // file at the top-level directory of this distribution and at
   3 // http://rust-lang.org/COPYRIGHT.
   4 //
   5 // Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or
   6 // http://www.apache.org/licenses/LICENSE-2.0> or the MIT license
   7 // <LICENSE-MIT or http://opensource.org/licenses/MIT>, at your
   8 // option. This file may not be copied, modified, or distributed
   9 // except according to those terms.
  10
  11 //! Basic html highlighting functionality
  12 //!
  13 //! This module uses libsyntax's lexer to provide token-based highlighting for
  14 //! the HTML documentation generated by rustdoc.
  15
  16 use std::io;
  17
  18 use syntax::parse;
  19 use syntax::parse::lexer;
  20
  21 use html::escape::Escape;
  22
  23 use t = syntax::parse::token;
  24
  25 /// Highlights some source code, returning the HTML output.
  26 pub fn highlight(src: &str, class: Option<&str>, id: Option<&str>) -> String {
  27     debug!("highlighting: ================\n{}\n==============", src);
  28     let sess = parse::new_parse_sess();
  29     let fm = parse::string_to_filemap(&sess,
  30                                       src.to_string(),
  31                                       "<stdin>".to_string());
  32
  33     let mut out = io::MemWriter::new();
  34     doit(&sess,
  35          lexer::StringReader::new(&sess.span_diagnostic, fm),
  36          class,
  37          id,
  38          &mut out).unwrap();
  39     String::from_utf8_lossy(out.unwrap().as_slice()).into_string()
  40 }
  41
  42 /// Exhausts the `lexer` writing the output into `out`.
  43 ///
  44 /// The general structure for this method is to iterate over each token,
  45 /// possibly giving it an HTML span with a class specifying what flavor of token
  46 /// it's used. All source code emission is done as slices from the source map,
  47 /// not from the tokens themselves, in order to stay true to the original
  48 /// source.
  49 fn doit(sess: &parse::ParseSess, mut lexer: lexer::StringReader,
  50         class: Option<&str>, id: Option<&str>,
  51         out: &mut Writer) -> io::IoResult<()> {
  52     use syntax::parse::lexer::Reader;
  53
  54     try!(write!(out, "<pre "));
  55     match id {
  56         Some(id) => try!(write!(out, "id='{}' ", id)),
  57         None => {}
  58     }
  59     try!(write!(out, "class='rust {}'>\n", class.unwrap_or("")));
  60     let mut is_attribute = false;
  61     let mut is_macro = false;
  62     let mut is_macro_nonterminal = false;
  63     loop {
  64         let next = lexer.next_token();
  65
  66         let snip = |sp| sess.span_diagnostic.cm.span_to_snippet(sp).unwrap();
  67
  68         if next.tok == t::EOF { break }
  69
  70         let klass = match next.tok {
  71             t::WS => {
  72                 try!(write!(out, "{}", Escape(snip(next.sp).as_slice())));
  73                 continue
  74             },
  75             t::COMMENT => {
  76                 try!(write!(out, "<span class='comment'>{}</span>",
  77                             Escape(snip(next.sp).as_slice())));
  78                 continue
  79             },
  80             t::SHEBANG(s) => {
  81                 try!(write!(out, "{}", Escape(s.as_str())));
  82                 continue
  83             },
  84             // If this '&' token is directly adjacent to another token, assume
  85             // that it's the address-of operator instead of the and-operator.
  86             // This allows us to give all pointers their own class (`Box` and
  87             // `@` are below).
  88             t::BINOP(t::AND) if lexer.peek().sp.lo == next.sp.hi => "kw-2",
  89             t::AT | t::TILDE => "kw-2",
  90
  91             // consider this as part of a macro invocation if there was a
  92             // leading identifier
  93             t::NOT if is_macro => { is_macro = false; "macro" }
  94
  95             // operators
  96             t::EQ | t::LT | t::LE | t::EQEQ | t::NE | t::GE | t::GT |
  97                 t::ANDAND | t::OROR | t::NOT | t::BINOP(..) | t::RARROW |
  98                 t::BINOPEQ(..) | t::FAT_ARROW => "op",
  99
 100             // miscellaneous, no highlighting
 101             t::DOT | t::DOTDOT | t::DOTDOTDOT | t::COMMA | t::SEMI |
 102                 t::COLON | t::MOD_SEP | t::LARROW | t::LPAREN |
 103                 t::RPAREN | t::LBRACKET | t::LBRACE | t::RBRACE | t::QUESTION => "",
 104             t::DOLLAR => {
 105                 if t::is_ident(&lexer.peek().tok) {
 106                     is_macro_nonterminal = true;
 107                     "macro-nonterminal"
 108                 } else {
 109                     ""
 110                 }
 111             }
 112
 113             // This is the start of an attribute. We're going to want to
 114             // continue highlighting it as an attribute until the ending ']' is
 115             // seen, so skip out early. Down below we terminate the attribute
 116             // span when we see the ']'.
 117             t::POUND => {
 118                 is_attribute = true;
 119                 try!(write!(out, r"<span class='attribute'>#"));
 120                 continue
 121             }
 122             t::RBRACKET => {
 123                 if is_attribute {
 124                     is_attribute = false;
 125                     try!(write!(out, "]</span>"));
 126                     continue
 127                 } else {
 128                     ""
 129                 }
 130             }
 131
 132             // text literals
 133             t::LIT_BYTE(..) | t::LIT_BINARY(..) | t::LIT_BINARY_RAW(..) |
 134                 t::LIT_CHAR(..) | t::LIT_STR(..) | t::LIT_STR_RAW(..) => "string",
 135
 136             // number literals
 137             t::LIT_INTEGER(..) | t::LIT_FLOAT(..) => "number",
 138
 139             // keywords are also included in the identifier set
 140             t::IDENT(ident, _is_mod_sep) => {
 141                 match t::get_ident(ident).get() {
 142                     "ref" | "mut" => "kw-2",
 143
 144                     "self" => "self",
 145                     "false" | "true" => "boolval",
 146
 147                     "Option" | "Result" => "prelude-ty",
 148                     "Some" | "None" | "Ok" | "Err" => "prelude-val",
 149
 150                     _ if t::is_any_keyword(&next.tok) => "kw",
 151                     _ => {
 152                         if is_macro_nonterminal {
 153                             is_macro_nonterminal = false;
 154                             "macro-nonterminal"
 155                         } else if lexer.peek().tok == t::NOT {
 156                             is_macro = true;
 157                             "macro"
 158                         } else {
 159                             "ident"
 160                         }
 161                     }
 162                 }
 163             }
 164
 165             t::LIFETIME(..) => "lifetime",
 166             t::DOC_COMMENT(..) => "doccomment",
 167             t::UNDERSCORE | t::EOF | t::INTERPOLATED(..) => "",
 168         };
 169
 170         // as mentioned above, use the original source code instead of
 171         // stringifying this token
 172         let snip = sess.span_diagnostic.cm.span_to_snippet(next.sp).unwrap();
 173         if klass == "" {
 174             try!(write!(out, "{}", Escape(snip.as_slice())));
 175         } else {
 176             try!(write!(out, "<span class='{}'>{}</span>", klass,
 177                           Escape(snip.as_slice())));
 178         }
 179     }
 180
 181     write!(out, "</pre>\n")
 182 }