Make clear the differentiation between char pos and byte pos in filemaps. Fix up error printing for files with multi-byte characters.

Author: jdm

src/comp/driver/session.rs

@@ -103,7 +103,7 @@ obj session(@config targ_cfg,
     fn unimpl(str msg) -> ! { self.bug("unimplemented " + msg); }
     fn get_codemap() -> codemap::codemap { ret parse_sess.cm; }
     fn lookup_pos(uint pos) -> codemap::loc {
-        ret codemap::lookup_pos(parse_sess.cm, pos);
+        ret codemap::lookup_char_pos(parse_sess.cm, pos);
     }
     fn get_parse_sess() -> parse_sess { ret parse_sess; }
     fn next_node_id() -> ast::node_id {

src/comp/syntax/codemap.rs

@@ -9,50 +9,69 @@ import std::option::none;
 
 type filename = str;
 
+type file_pos = rec(uint ch, uint byte);
+
 /* A codemap is a thing that maps uints to file/line/column positions
  * in a crate. This to make it possible to represent the positions
  * with single-word things, rather than passing records all over the
  * compiler.
  */
-type filemap = @rec(filename name, uint start_pos, mutable uint[] lines);
+type filemap = @rec(filename name, file_pos start_pos,
+                    mutable file_pos[] lines);
 
 type codemap = @rec(mutable filemap[] files);
 
 type loc = rec(filename filename, uint line, uint col);
 
 fn new_codemap() -> codemap {
-    let filemap[] files = ~[];
-    ret @rec(mutable files=files);
+    ret @rec(mutable files=~[]);
+}
+
+fn new_filemap(filename filename, uint start_pos_ch, uint start_pos_byte)
+        -> filemap {
+    ret @rec(name=filename, start_pos=rec(ch=start_pos_ch,
+                                          byte=start_pos_byte),
+             mutable lines=~[rec(ch=start_pos_ch, byte=start_pos_byte)]);
 }
 
-fn new_filemap(filename filename, uint start_pos) -> filemap {
-    ret @rec(name=filename, start_pos=start_pos, mutable lines=[start_pos]);
+fn next_line(filemap file, uint chpos, uint byte_pos) {
+    file.lines += ~[rec(ch=chpos, byte=byte_pos)];
 }
 
-fn next_line(filemap file, uint pos) { file.lines += ~[pos]; }
+type lookup_fn = fn (file_pos pos) -> uint;
 
-fn lookup_pos(codemap map, uint pos) -> loc {
+fn lookup_pos(codemap map, uint pos, lookup_fn lookup) -> loc {
     auto a = 0u;
-    auto b = ivec::len[filemap](map.files);
+    auto b = ivec::len(map.files);
     while (b - a > 1u) {
         auto m = (a + b) / 2u;
-        if (map.files.(m).start_pos > pos) { b = m; } else { a = m; }
+        if (lookup(map.files.(m).start_pos) > pos) { b = m; } else { a = m; }
     }
     auto f = map.files.(a);
     a = 0u;
-    b = ivec::len[uint](f.lines);
+    b = ivec::len(f.lines);
     while (b - a > 1u) {
         auto m = (a + b) / 2u;
-        if (f.lines.(m) > pos) { b = m; } else { a = m; }
+        if (lookup(f.lines.(m)) > pos) { b = m; } else { a = m; }
     }
-    ret rec(filename=f.name, line=a + 1u, col=pos - f.lines.(a));
+    ret rec(filename=f.name, line=a + 1u, col=pos - lookup(f.lines.(a)));
+}
+
+fn lookup_char_pos(codemap map, uint pos) -> loc {
+    fn lookup(file_pos pos) -> uint { ret pos.ch; }
+    ret lookup_pos(map, pos, lookup);
+}
+
+fn lookup_byte_pos(codemap map, uint pos) -> loc {
+    fn lookup(file_pos pos) -> uint { ret pos.byte; }
+    ret lookup_pos(map, pos, lookup);
 }
 
 type span = rec(uint lo, uint hi);
 
 fn span_to_str(&span sp, &codemap cm) -> str {
-    auto lo = lookup_pos(cm, sp.lo);
-    auto hi = lookup_pos(cm, sp.hi);
+    auto lo = lookup_char_pos(cm, sp.lo);
+    auto hi = lookup_char_pos(cm, sp.hi);
     ret #fmt("%s:%u:%u:%u:%u", lo.filename, lo.line, lo.col, hi.line, hi.col);
 }
 
@@ -115,8 +134,7 @@ fn emit_diagnostic(&option::t[span] sp, &str msg, &str kind, u8 color,
 
             // If there's one line at fault we can easily point to the problem
             if (ivec::len(lines.lines) == 1u) {
-                auto lo = codemap::lookup_pos(cm, option::get(sp).lo);
-                auto lo = lookup_pos(cm, option::get(sp).lo);
+                auto lo = lookup_char_pos(cm, option::get(sp).lo);
                 auto digits = 0u;
                 auto num = lines.lines.(0) / 10u;
 
@@ -129,7 +147,7 @@ fn emit_diagnostic(&option::t[span] sp, &str msg, &str kind, u8 color,
                 while (left > 0u) { str::push_char(s, ' '); left -= 1u; }
 
                 s += "^";
-                auto hi = lookup_pos(cm, option::get(sp).hi);
+                auto hi = lookup_char_pos(cm, option::get(sp).hi);
                 if (hi.col != lo.col) {
                     // the ^ already takes up one space
                     auto width = hi.col - lo.col - 1u;
@@ -158,8 +176,8 @@ fn emit_note(&option::t[span] sp, &str msg, &codemap cm) {
 type file_lines = rec(str name, uint[] lines);
 
 fn span_to_lines(span sp, codemap::codemap cm) -> @file_lines {
-    auto lo = codemap::lookup_pos(cm, sp.lo);
-    auto hi = codemap::lookup_pos(cm, sp.hi);
+    auto lo = lookup_char_pos(cm, sp.lo);
+    auto hi = lookup_char_pos(cm, sp.hi);
     auto lines = ~[];
     for each (uint i in uint::range(lo.line - 1u, hi.line as uint)) {
         lines += ~[i];
@@ -168,10 +186,10 @@ fn span_to_lines(span sp, codemap::codemap cm) -> @file_lines {
 }
 
 fn get_line(filemap fm, int line, &str file) -> str {
-    let uint begin = fm.lines.(line) - fm.start_pos;
+    let uint begin = fm.lines.(line).byte - fm.start_pos.byte;
     let uint end;
     if (line as uint < ivec::len(fm.lines) - 1u) {
-        end = fm.lines.(line + 1) - fm.start_pos;
+        end = fm.lines.(line + 1).byte - fm.start_pos.byte;
     } else {
         // If we're not done parsing the file, we're at the limit of what's
         // parsed. If we just slice the rest of the string, we'll print out

src/comp/syntax/parse/eval.rs

@@ -21,6 +21,7 @@ type ctx =
          mutable str[] deps,
          parser::parse_sess sess,
          mutable uint chpos,
+         mutable uint byte_pos,
          ast::crate_cfg cfg);
 
 fn eval_crate_directives(ctx cx, &(@ast::crate_directive)[] cdirs,
@@ -56,7 +57,8 @@ fn eval_crate_directive(ctx cx, @ast::crate_directive cdir, str prefix,
             };
             if (cx.mode == mode_depend) { cx.deps += ~[full_path]; ret; }
             auto p0 =
-                new_parser_from_file(cx.sess, cx.cfg, full_path, cx.chpos);
+                new_parser_from_file(cx.sess, cx.cfg, full_path, cx.chpos,
+                                     cx.byte_pos);
             auto inner_attrs = parse_inner_attrs_and_next(p0);
             auto mod_attrs = attrs + inner_attrs._0;
             auto first_item_outer_attrs = inner_attrs._1;
@@ -65,8 +67,9 @@ fn eval_crate_directive(ctx cx, @ast::crate_directive cdir, str prefix,
             auto i = syntax::parse::parser::mk_item
                 (p0, cdir.span.lo, cdir.span.hi, id, ast::item_mod(m0),
                  mod_attrs);
-            // Thread defids and chpos through the parsers
+            // Thread defids, chpos and byte_pos through the parsers
             cx.chpos = p0.get_chpos();
+            cx.byte_pos = p0.get_byte_pos();
             items += ~[i];
         }
         case (ast::cdir_dir_mod(?id, ?dir_opt, ?cdirs, ?attrs)) {

src/comp/syntax/parse/lexer.rs

@@ -24,6 +24,7 @@ type reader =
         fn get_mark_str() -> str ;
         fn get_interner() -> @interner::interner[str] ;
         fn get_chpos() -> uint ;
+        fn get_byte_pos() -> uint ;
         fn get_col() -> uint ;
         fn get_filemap() -> codemap::filemap ;
         fn err(str) ;
@@ -53,6 +54,7 @@ fn new_reader(&codemap::codemap cm, str src, codemap::filemap filemap,
         }
         fn get_mark_chpos() -> uint { ret mark_chpos; }
         fn get_chpos() -> uint { ret chpos; }
+        fn get_byte_pos() -> uint { ret pos; }
         fn curr() -> char { ret ch; }
         fn next() -> char {
             if (pos < len) {
@@ -70,7 +72,10 @@ fn new_reader(&codemap::codemap cm, str src, codemap::filemap filemap,
             if (pos < len) {
                 col += 1u;
                 chpos += 1u;
-                if (ch == '\n') { codemap::next_line(fm, chpos); col = 0u; }
+                if (ch == '\n') {
+                    codemap::next_line(fm, chpos, pos + fm.start_pos.byte);
+                    col = 0u;
+                }
                 auto next = str::char_range_at(src, pos);
                 pos = next._1;
                 ch = next._0;
@@ -86,7 +91,8 @@ fn new_reader(&codemap::codemap cm, str src, codemap::filemap filemap,
     let str[] strs = ~[];
     auto rd =
         reader(cm, src, str::byte_len(src), 0u, 0u, -1 as char, 0u,
-               filemap.start_pos, filemap.start_pos, strs, filemap, itr);
+               filemap.start_pos.ch, filemap.start_pos.ch, strs, filemap,
+               itr);
     rd.init();
     ret rd;
 }
@@ -737,7 +743,7 @@ fn gather_comments_and_literals(&codemap::codemap cm, str path)
     auto srdr = ioivec::file_reader(path);
     auto src = str::unsafe_from_bytes_ivec(srdr.read_whole_stream());
     auto itr = @interner::mk[str](str::hash, str::eq);
-    auto rdr = new_reader(cm, src, codemap::new_filemap(path, 0u), itr);
+    auto rdr = new_reader(cm, src, codemap::new_filemap(path, 0u, 0u), itr);
     let cmnt[] comments = ~[];
     let lit[] literals = ~[];
     let bool first_read = true;

src/comp/syntax/parse/parser.rs

@@ -50,17 +50,18 @@ type parser =
         fn get_filemap() -> codemap::filemap ;
         fn get_bad_expr_words() -> hashmap[str, ()] ;
         fn get_chpos() -> uint ;
+        fn get_byte_pos() -> uint ;
         fn get_id() -> ast::node_id ;
         fn get_sess() -> parse_sess;
     };
 
 fn new_parser_from_file(parse_sess sess, ast::crate_cfg cfg,
-                        str path, uint pos) -> parser {
+                        str path, uint chpos, uint byte_pos) -> parser {
     auto ftype = SOURCE_FILE;
     if (str::ends_with(path, ".rc")) { ftype = CRATE_FILE; }
     auto srdr = ioivec::file_reader(path);
     auto src = str::unsafe_from_bytes_ivec(srdr.read_whole_stream());
-    auto filemap = codemap::new_filemap(path, pos);
+    auto filemap = codemap::new_filemap(path, chpos, byte_pos);
     sess.cm.files += ~[filemap];
     auto itr = @interner::mk(str::hash, str::eq);
     auto rdr = lexer::new_reader(sess.cm, src, filemap, itr);
@@ -114,6 +115,7 @@ fn new_parser(parse_sess sess, ast::crate_cfg cfg, lexer::reader rdr,
         fn get_filemap() -> codemap::filemap { ret rdr.get_filemap(); }
         fn get_bad_expr_words() -> hashmap[str, ()] { ret bad_words; }
         fn get_chpos() -> uint { ret rdr.get_chpos(); }
+        fn get_byte_pos() -> uint { ret rdr.get_byte_pos(); }
         fn get_id() -> ast::node_id { ret next_node_id(sess); }
         fn get_sess() -> parse_sess { ret sess; }
     }
@@ -2378,15 +2380,15 @@ fn parse_native_view(&parser p) -> (@ast::view_item)[] {
 
 fn parse_crate_from_source_file(&str input, &ast::crate_cfg cfg,
                                 &parse_sess sess) -> @ast::crate {
-    auto p = new_parser_from_file(sess, cfg, input, 0u);
+    auto p = new_parser_from_file(sess, cfg, input, 0u, 0u);
     ret parse_crate_mod(p, cfg, sess);
 }
 
 fn parse_crate_from_source_str(&str name, &str source, &ast::crate_cfg cfg,
                                &codemap::codemap cm) -> @ast::crate {
     auto sess = @rec(cm=cm, mutable next_id=0);
     auto ftype = SOURCE_FILE;
-    auto filemap = codemap::new_filemap(name, 0u);
+    auto filemap = codemap::new_filemap(name, 0u, 0u);
     sess.cm.files += ~[filemap];
     auto itr = @interner::mk(str::hash, str::eq);
     auto rdr = lexer::new_reader(sess.cm, source, filemap, itr);
@@ -2504,7 +2506,7 @@ fn parse_crate_directives(&parser p, token::token term,
 
 fn parse_crate_from_crate_file(&str input, &ast::crate_cfg cfg,
                                &parse_sess sess) -> @ast::crate {
-    auto p = new_parser_from_file(sess, cfg, input, 0u);
+    auto p = new_parser_from_file(sess, cfg, input, 0u, 0u);
     auto lo = p.get_lo_pos();
     auto prefix = std::fs::dirname(p.get_filemap().name);
     auto leading_attrs = parse_inner_attrs_and_next(p);
@@ -2517,6 +2519,7 @@ fn parse_crate_from_crate_file(&str input, &ast::crate_cfg cfg,
                    mutable deps=deps,
                    sess=sess,
                    mutable chpos=p.get_chpos(),
+                   mutable byte_pos=p.get_byte_pos(),
                    cfg = p.get_cfg());
     auto m = eval::eval_crate_directives_to_mod(cx, cdirs, prefix);
     auto hi = p.get_hi_pos();

src/comp/syntax/print/pprust.rs

@@ -1345,8 +1345,8 @@ fn maybe_print_trailing_comment(&ps s, codemap::span span,
     alt (next_comment(s)) {
         case (some(?cmnt)) {
             if (cmnt.style != lexer::trailing) { ret; }
-            auto span_line = codemap::lookup_pos(cm, span.hi);
-            auto comment_line = codemap::lookup_pos(cm, cmnt.pos);
+            auto span_line = codemap::lookup_char_pos(cm, span.hi);
+            auto comment_line = codemap::lookup_char_pos(cm, cmnt.pos);
             auto next = cmnt.pos + 1u;
             alt (next_pos) { case (none) { } case (some(?p)) { next = p; } }
             if (span.hi < cmnt.pos && cmnt.pos < next &&