diff options
| author | Brian Anderson <banderson@mozilla.com> | 2011-06-27 14:18:32 -0700 |
|---|---|---|
| committer | Brian Anderson <banderson@mozilla.com> | 2011-06-27 16:30:27 -0700 |
| commit | ba5c7a570d1bc6f28e7a2f4eb5cfd50b7c19f374 (patch) | |
| tree | 78bb67f83876f83a65464b82ae0ead676e080080 /src/comp/metadata | |
| parent | d4a5487e1c1ab641f146e2ab83409bc9b5a4ec51 (diff) | |
| download | rust-ba5c7a570d1bc6f28e7a2f4eb5cfd50b7c19f374.tar.gz rust-ba5c7a570d1bc6f28e7a2f4eb5cfd50b7c19f374.zip | |
Rename middle::metadata to metadata::cwriter. Move creader to metadata
Preparation for a lot more metadata refactoring
Diffstat (limited to 'src/comp/metadata')
| -rw-r--r-- | src/comp/metadata/creader.rs | 895 | ||||
| -rw-r--r-- | src/comp/metadata/cwriter.rs | 817 |
2 files changed, 1712 insertions, 0 deletions
diff --git a/src/comp/metadata/creader.rs b/src/comp/metadata/creader.rs new file mode 100644 index 00000000000..f85ac41297d --- /dev/null +++ b/src/comp/metadata/creader.rs @@ -0,0 +1,895 @@ + + +// -*- rust -*- +import driver::session; +import front::ast; +import lib::llvm::False; +import lib::llvm::llvm; +import lib::llvm::mk_object_file; +import lib::llvm::mk_section_iter; +import middle::resolve; +import middle::walk; +import cwriter; +import middle::trans; +import middle::ty; +import back::x86; +import util::common; +import util::common::span; +import util::common::respan; +import util::common::a_bang; +import util::common::a_ty; +import util::common::may_begin_ident; +import std::str; +import std::uint; +import std::vec; +import std::ebml; +import std::fs; +import std::io; +import std::option; +import std::option::none; +import std::option::some; +import std::os; +import std::map::hashmap; + +export get_symbol; +export get_tag_variants; +export get_type; +export read_crates; +export lookup_defs; +export get_type; +export list_file_metadata; + +// Type decoding + +// Compact string representation for ty::t values. API ty_str & parse_from_str +// (The second has to be authed pure.) Extra parameters are for converting +// to/from def_ids in the data buffer. Whatever format you choose should not +// contain pipe characters. + +// Callback to translate defs to strs or back: +type str_def = fn(str) -> ast::def_id ; + +type pstate = + rec(vec[u8] data, int crate, mutable uint pos, uint len, ty::ctxt tcx); + +type ty_or_bang = util::common::ty_or_bang[ty::t]; + +fn peek(@pstate st) -> u8 { ret st.data.(st.pos); } + +fn next(@pstate st) -> u8 { + auto ch = st.data.(st.pos); + st.pos = st.pos + 1u; + ret ch; +} + +fn parse_ident(@pstate st, str_def sd, char last) -> ast::ident { + fn is_last(char b, char c) -> bool { + ret c == b; + } + ret parse_ident_(st, sd, bind is_last(last, _)); +} + +fn parse_ident_(@pstate st, str_def sd, fn(char) -> bool is_last) + -> ast::ident { + auto rslt = ""; + while (! is_last(peek(st) as char)) { + rslt += str::unsafe_from_byte(next(st)); + } + ret rslt; +} + + +fn parse_ty_data(vec[u8] data, int crate_num, uint pos, uint len, str_def sd, + ty::ctxt tcx) -> ty::t { + auto st = + @rec(data=data, crate=crate_num, mutable pos=pos, len=len, tcx=tcx); + auto result = parse_ty(st, sd); + ret result; +} + +fn parse_ty_or_bang(@pstate st, str_def sd) -> ty_or_bang { + alt (peek(st) as char) { + case ('!') { auto ignore = next(st); ret a_bang[ty::t]; } + case (_) { ret a_ty[ty::t](parse_ty(st, sd)); } + } +} + +fn parse_constrs(@pstate st, str_def sd) -> vec[@ty::constr_def] { + let vec[@ty::constr_def] rslt = []; + alt (peek(st) as char) { + case (':') { + do { + auto ignore = next(st); + vec::push(rslt, parse_constr(st, sd)); + } while (peek(st) as char == ';') + } + case (_) { } + } + ret rslt; +} + +fn parse_path(@pstate st, str_def sd) -> ast::path { + let vec[ast::ident] idents = []; + fn is_last(char c) -> bool { + ret (c == '(' || c == ':'); + } + idents += [parse_ident_(st, sd, is_last)]; + while (true) { + alt (peek(st) as char) { + case (':') { + auto ignore = next(st); + ignore = next(st); + } + case (?c) { + if (c == '(') { + ret respan(rec(lo=0u, hi=0u), + rec(idents=idents, types=[])); + } + else { + idents += [parse_ident_(st, sd, is_last)]; + } + } + } + } + fail "parse_path: ill-formed path"; +} + +fn parse_constr(@pstate st, str_def sd) -> @ty::constr_def { + let vec[@ast::constr_arg] args = []; + auto sp = rec(lo=0u,hi=0u); // FIXME: use a real span + let ast::path pth = parse_path(st, sd); + let char ignore = next(st) as char; + assert(ignore as char == '('); + auto def = parse_def(st, sd); + do { + alt (peek(st) as char) { + case ('*') { + st.pos += 1u; + args += [@respan(sp, ast::carg_base)]; + } + case (?c) { + /* how will we disambiguate between + an arg index and a lit argument? */ + if (c >= '0' && c <= '9') { + // FIXME + args += [@respan(sp, ast::carg_ident((c as uint) - 48u))]; + ignore = next(st) as char; + } + else { + log_err("Lit args are unimplemented"); + fail; // FIXME + } + /* + else { + auto lit = parse_lit(st, sd, ','); + args += [respan(st.span, ast::carg_lit(lit))]; + } + */ + } + } + ignore = next(st) as char; + } while (ignore == ';'); + assert(ignore == ')'); + ret @respan(sp, rec(path=pth, args=args, id=def)); +} + +fn parse_ty(@pstate st, str_def sd) -> ty::t { + alt (next(st) as char) { + case ('n') { ret ty::mk_nil(st.tcx); } + case ('z') { ret ty::mk_bot(st.tcx); } + case ('b') { ret ty::mk_bool(st.tcx); } + case ('i') { ret ty::mk_int(st.tcx); } + case ('u') { ret ty::mk_uint(st.tcx); } + case ('l') { ret ty::mk_float(st.tcx); } + case ('M') { + alt (next(st) as char) { + case ('b') { ret ty::mk_mach(st.tcx, common::ty_u8); } + case ('w') { ret ty::mk_mach(st.tcx, common::ty_u16); } + case ('l') { ret ty::mk_mach(st.tcx, common::ty_u32); } + case ('d') { ret ty::mk_mach(st.tcx, common::ty_u64); } + case ('B') { ret ty::mk_mach(st.tcx, common::ty_i8); } + case ('W') { ret ty::mk_mach(st.tcx, common::ty_i16); } + case ('L') { ret ty::mk_mach(st.tcx, common::ty_i32); } + case ('D') { ret ty::mk_mach(st.tcx, common::ty_i64); } + case ('f') { ret ty::mk_mach(st.tcx, common::ty_f32); } + case ('F') { ret ty::mk_mach(st.tcx, common::ty_f64); } + } + } + case ('c') { ret ty::mk_char(st.tcx); } + case ('s') { ret ty::mk_str(st.tcx); } + case ('S') { ret ty::mk_istr(st.tcx); } + case ('t') { + assert (next(st) as char == '['); + auto def = parse_def(st, sd); + let vec[ty::t] params = []; + while (peek(st) as char != ']') { params += [parse_ty(st, sd)]; } + st.pos = st.pos + 1u; + ret ty::mk_tag(st.tcx, def, params); + } + case ('p') { ret ty::mk_param(st.tcx, parse_int(st) as uint); } + case ('@') { ret ty::mk_box(st.tcx, parse_mt(st, sd)); } + case ('*') { ret ty::mk_ptr(st.tcx, parse_mt(st, sd)); } + case ('V') { ret ty::mk_vec(st.tcx, parse_mt(st, sd)); } + case ('I') { ret ty::mk_ivec(st.tcx, parse_mt(st, sd)); } + case ('a') { ret ty::mk_task(st.tcx); } + case ('P') { ret ty::mk_port(st.tcx, parse_ty(st, sd)); } + case ('C') { ret ty::mk_chan(st.tcx, parse_ty(st, sd)); } + case ('T') { + assert (next(st) as char == '['); + let vec[ty::mt] params = []; + while (peek(st) as char != ']') { params += [parse_mt(st, sd)]; } + st.pos = st.pos + 1u; + ret ty::mk_tup(st.tcx, params); + } + case ('R') { + assert (next(st) as char == '['); + let vec[ty::field] fields = []; + while (peek(st) as char != ']') { + auto name = ""; + while (peek(st) as char != '=') { + name += str::unsafe_from_byte(next(st)); + } + st.pos = st.pos + 1u; + fields += [rec(ident=name, mt=parse_mt(st, sd))]; + } + st.pos = st.pos + 1u; + ret ty::mk_rec(st.tcx, fields); + } + case ('F') { + auto func = parse_ty_fn(st, sd); + ret ty::mk_fn(st.tcx, ast::proto_fn, func._0, func._1, func._2, + func._3); + } + case ('W') { + auto func = parse_ty_fn(st, sd); + ret ty::mk_fn(st.tcx, ast::proto_iter, func._0, func._1, func._2, + func._3); + } + case ('N') { + auto abi; + alt (next(st) as char) { + case ('r') { abi = ast::native_abi_rust; } + case ('i') { abi = ast::native_abi_rust_intrinsic; } + case ('c') { abi = ast::native_abi_cdecl; } + case ('l') { abi = ast::native_abi_llvm; } + } + auto func = parse_ty_fn(st, sd); + ret ty::mk_native_fn(st.tcx, abi, func._0, func._1); + } + case ('O') { + assert (next(st) as char == '['); + let vec[ty::method] methods = []; + while (peek(st) as char != ']') { + auto proto; + alt (next(st) as char) { + case ('W') { proto = ast::proto_iter; } + case ('F') { proto = ast::proto_fn; } + } + auto name = ""; + while (peek(st) as char != '[') { + name += str::unsafe_from_byte(next(st)); + } + auto func = parse_ty_fn(st, sd); + methods += + [rec(proto=proto, + ident=name, + inputs=func._0, + output=func._1, + cf=func._2, + constrs=func._3)]; + } + st.pos += 1u; + ret ty::mk_obj(st.tcx, methods); + } + case ('r') { + auto def = parse_def(st, sd); + auto inner = parse_ty(st, sd); + ret ty::mk_res(st.tcx, def, inner); + } + case ('X') { ret ty::mk_var(st.tcx, parse_int(st)); } + case ('E') { ret ty::mk_native(st.tcx); } + case ('Y') { ret ty::mk_type(st.tcx); } + case ('#') { + auto pos = parse_hex(st); + assert (next(st) as char == ':'); + auto len = parse_hex(st); + assert (next(st) as char == '#'); + alt (st.tcx.rcache.find(tup(st.crate, pos, len))) { + case (some(?tt)) { ret tt; } + case (none) { + auto ps = @rec(pos=pos, len=len with *st); + auto tt = parse_ty(ps, sd); + st.tcx.rcache.insert(tup(st.crate, pos, len), tt); + ret tt; + } + } + } + case (?c) { + log_err "unexpected char in type string: "; + log_err c; + fail; + } + } +} + +fn parse_mt(@pstate st, str_def sd) -> ty::mt { + auto mut; + alt (peek(st) as char) { + case ('m') { next(st); mut = ast::mut; } + case ('?') { next(st); mut = ast::maybe_mut; } + case (_) { mut = ast::imm; } + } + ret rec(ty=parse_ty(st, sd), mut=mut); +} + +fn parse_def(@pstate st, str_def sd) -> ast::def_id { + auto def = ""; + while (peek(st) as char != '|') { + def += str::unsafe_from_byte(next(st)); + } + st.pos = st.pos + 1u; + ret sd(def); +} + +fn parse_int(@pstate st) -> int { + auto n = 0; + while (true) { + auto cur = peek(st) as char; + if (cur < '0' || cur > '9') { break; } + st.pos = st.pos + 1u; + n *= 10; + n += (cur as int) - ('0' as int); + } + ret n; +} + +fn parse_hex(@pstate st) -> uint { + auto n = 0u; + while (true) { + auto cur = peek(st) as char; + if ((cur < '0' || cur > '9') && (cur < 'a' || cur > 'f')) { break; } + st.pos = st.pos + 1u; + n *= 16u; + if ('0' <= cur && cur <= '9') { + n += (cur as uint) - ('0' as uint); + } else { n += 10u + (cur as uint) - ('a' as uint); } + } + ret n; +} + +fn parse_ty_fn(@pstate st, str_def sd) -> + tup(vec[ty::arg], ty::t, ast::controlflow, vec[@ty::constr_def]) { + assert (next(st) as char == '['); + let vec[ty::arg] inputs = []; + while (peek(st) as char != ']') { + auto mode = ty::mo_val; + if (peek(st) as char == '&') { + mode = ty::mo_alias(false); + st.pos += 1u; + if (peek(st) as char == 'm') { + mode = ty::mo_alias(true); + st.pos += 1u; + } + } + inputs += [rec(mode=mode, ty=parse_ty(st, sd))]; + } + st.pos += 1u; // eat the ']' + auto cs = parse_constrs(st, sd); + alt (parse_ty_or_bang(st, sd)) { + case (a_bang) { + ret tup(inputs, ty::mk_bot(st.tcx), ast::noreturn, cs); + } + case (a_ty(?t)) { ret tup(inputs, t, ast::return, cs); } + } +} + + +// Rust metadata parsing +fn parse_def_id(vec[u8] buf) -> ast::def_id { + auto colon_idx = 0u; + auto len = vec::len[u8](buf); + while (colon_idx < len && buf.(colon_idx) != ':' as u8) { + colon_idx += 1u; + } + if (colon_idx == len) { + log_err "didn't find ':' when parsing def id"; + fail; + } + auto crate_part = vec::slice[u8](buf, 0u, colon_idx); + auto def_part = vec::slice[u8](buf, colon_idx + 1u, len); + auto crate_num = uint::parse_buf(crate_part, 10u) as int; + auto def_id = uint::parse_buf(def_part, 10u) as int; + ret tup(crate_num, def_id); +} + +fn lookup_hash(&ebml::doc d, fn(vec[u8]) -> bool eq_fn, uint hash) -> + vec[ebml::doc] { + auto index = ebml::get_doc(d, cwriter::tag_index); + auto table = ebml::get_doc(index, cwriter::tag_index_table); + auto hash_pos = table.start + hash % 256u * 4u; + auto pos = ebml::be_uint_from_bytes(d.data, hash_pos, 4u); + auto bucket = ebml::doc_at(d.data, pos); + // Awkward logic because we can't ret from foreach yet + + let vec[ebml::doc] result = []; + auto belt = cwriter::tag_index_buckets_bucket_elt; + for each (ebml::doc elt in ebml::tagged_docs(bucket, belt)) { + auto pos = ebml::be_uint_from_bytes(elt.data, elt.start, 4u); + if (eq_fn(vec::slice[u8](elt.data, elt.start + 4u, elt.end))) { + vec::push(result, ebml::doc_at(d.data, pos)); + } + } + ret result; +} + + +// Given a path and serialized crate metadata, returns the ID of the +// definition the path refers to. +fn resolve_path(vec[ast::ident] path, vec[u8] data) -> vec[ast::def_id] { + fn eq_item(vec[u8] data, str s) -> bool { + ret str::eq(str::unsafe_from_bytes(data), s); + } + auto s = str::connect(path, "::"); + auto md = ebml::new_doc(data); + auto paths = ebml::get_doc(md, cwriter::tag_paths); + auto eqer = bind eq_item(_, s); + let vec[ast::def_id] result = []; + for (ebml::doc doc in lookup_hash(paths, eqer, cwriter::hash_path(s))) { + auto did_doc = ebml::get_doc(doc, cwriter::tag_def_id); + vec::push(result, parse_def_id(ebml::doc_data(did_doc))); + } + ret result; +} + +fn maybe_find_item(int item_id, &ebml::doc items) -> option::t[ebml::doc] { + fn eq_item(vec[u8] bytes, int item_id) -> bool { + ret ebml::be_uint_from_bytes(bytes, 0u, 4u) as int == item_id; + } + auto eqer = bind eq_item(_, item_id); + auto found = lookup_hash(items, eqer, cwriter::hash_def_id(item_id)); + if (vec::len(found) == 0u) { + ret option::none[ebml::doc]; + } else { ret option::some[ebml::doc](found.(0)); } +} + +fn find_item(int item_id, &ebml::doc items) -> ebml::doc { + ret option::get(maybe_find_item(item_id, items)); +} + + +// Looks up an item in the given metadata and returns an ebml doc pointing +// to the item data. +fn lookup_item(int item_id, vec[u8] data) -> ebml::doc { + auto items = ebml::get_doc(ebml::new_doc(data), cwriter::tag_items); + ret find_item(item_id, items); +} + +fn item_kind(&ebml::doc item) -> u8 { + auto kind = ebml::get_doc(item, cwriter::tag_items_data_item_kind); + ret ebml::doc_as_uint(kind) as u8; +} + +fn item_symbol(&ebml::doc item) -> str { + auto sym = ebml::get_doc(item, cwriter::tag_items_data_item_symbol); + ret str::unsafe_from_bytes(ebml::doc_data(sym)); +} + +fn variant_tag_id(&ebml::doc d) -> ast::def_id { + auto tagdoc = ebml::get_doc(d, cwriter::tag_items_data_item_tag_id); + ret parse_def_id(ebml::doc_data(tagdoc)); +} + +fn item_type(&ebml::doc item, int this_cnum, ty::ctxt tcx) -> ty::t { + fn parse_external_def_id(int this_cnum, str s) -> ast::def_id { + // FIXME: This is completely wrong when linking against a crate + // that, in turn, links against another crate. We need a mapping + // from crate ID to crate "meta" attributes as part of the crate + // metadata: + + auto buf = str::bytes(s); + auto external_def_id = parse_def_id(buf); + ret tup(this_cnum, external_def_id._1); + } + auto tp = ebml::get_doc(item, cwriter::tag_items_data_item_type); + auto s = str::unsafe_from_bytes(ebml::doc_data(tp)); + ret parse_ty_data(item.data, this_cnum, tp.start, tp.end - tp.start, + bind parse_external_def_id(this_cnum, _), tcx); +} + +fn item_ty_param_count(&ebml::doc item, int this_cnum) -> uint { + let uint ty_param_count = 0u; + auto tp = cwriter::tag_items_data_item_ty_param_count; + for each (ebml::doc p in ebml::tagged_docs(item, tp)) { + ty_param_count = ebml::vint_at(ebml::doc_data(p), 0u)._0; + } + ret ty_param_count; +} + +fn tag_variant_ids(&ebml::doc item, int this_cnum) -> vec[ast::def_id] { + let vec[ast::def_id] ids = []; + auto v = cwriter::tag_items_data_item_variant; + for each (ebml::doc p in ebml::tagged_docs(item, v)) { + auto ext = parse_def_id(ebml::doc_data(p)); + vec::push[ast::def_id](ids, tup(this_cnum, ext._1)); + } + ret ids; +} + +fn get_metadata_section(str filename) -> option::t[vec[u8]] { + auto b = str::buf(filename); + auto mb = llvm::LLVMRustCreateMemoryBufferWithContentsOfFile(b); + if (mb as int == 0) { ret option::none[vec[u8]]; } + auto of = mk_object_file(mb); + auto si = mk_section_iter(of.llof); + while (llvm::LLVMIsSectionIteratorAtEnd(of.llof, si.llsi) == False) { + auto name_buf = llvm::LLVMGetSectionName(si.llsi); + auto name = str::str_from_cstr(name_buf); + if (str::eq(name, x86::get_meta_sect_name())) { + auto cbuf = llvm::LLVMGetSectionContents(si.llsi); + auto csz = llvm::LLVMGetSectionSize(si.llsi); + auto cvbuf = cbuf as vec::vbuf; + ret option::some[vec[u8]](vec::vec_from_vbuf[u8](cvbuf, csz)); + } + llvm::LLVMMoveToNextSection(si.llsi); + } + ret option::none[vec[u8]]; +} + +fn get_exported_metadata(&session::session sess, &str path, &vec[u8] data) -> + hashmap[str, str] { + auto meta_items = + ebml::get_doc(ebml::new_doc(data), cwriter::tag_meta_export); + auto mm = common::new_str_hash[str](); + for each (ebml::doc m in + ebml::tagged_docs(meta_items, cwriter::tag_meta_item)) { + auto kd = ebml::get_doc(m, cwriter::tag_meta_item_key); + auto vd = ebml::get_doc(m, cwriter::tag_meta_item_value); + auto k = str::unsafe_from_bytes(ebml::doc_data(kd)); + auto v = str::unsafe_from_bytes(ebml::doc_data(vd)); + log #fmt("metadata in %s: %s = %s", path, k, v); + if (!mm.insert(k, v)) { + sess.warn(#fmt("Duplicate metadata item in %s: %s", path, k)); + } + } + ret mm; +} + +fn metadata_matches(hashmap[str, str] mm, &vec[@ast::meta_item] metas) -> + bool { + log #fmt("matching %u metadata requirements against %u metadata items", + vec::len(metas), mm.size()); + for (@ast::meta_item mi in metas) { + alt (mi.node) { + case (ast::meta_key_value(?key, ?value)) { + alt (mm.find(key)) { + case (some(?v)) { + if (v == value) { + log #fmt("matched '%s': '%s'", key, + value); + } else { + log #fmt("missing '%s': '%s' (got '%s')", + key, + value, v); + ret false; + } + } + case (none) { + log #fmt("missing '%s': '%s'", + key, value); + ret false; + } + } + } + case (_) { + // FIXME (#487): Support all forms of meta_item + log_err "unimplemented meta_item variant in metadata_matches"; + ret false; + } + } + } + ret true; +} + +fn default_native_lib_naming(session::session sess) -> + rec(str prefix, str suffix) { + alt (sess.get_targ_cfg().os) { + case (session::os_win32) { ret rec(prefix="", suffix=".dll"); } + case (session::os_macos) { ret rec(prefix="lib", suffix=".dylib"); } + case (session::os_linux) { ret rec(prefix="lib", suffix=".so"); } + } +} + +fn find_library_crate(&session::session sess, &ast::ident ident, + &vec[@ast::meta_item] metas, + &vec[str] library_search_paths) -> + option::t[tup(str, vec[u8])] { + let str crate_name = ident; + for (@ast::meta_item mi in metas) { + alt (mi.node) { + case (ast::meta_key_value(?key, ?value)) { + if (key == "name") { + crate_name = value; + break; + } + } + case (_) { + // FIXME (#487) + sess.unimpl("meta_item variant") + } + } + } + auto nn = default_native_lib_naming(sess); + let str prefix = nn.prefix + crate_name; + // FIXME: we could probably use a 'glob' function in std::fs but it will + // be much easier to write once the unsafe module knows more about FFI + // tricks. Currently the glob(3) interface is a bit more than we can + // stomach from here, and writing a C++ wrapper is more work than just + // manually filtering fs::list_dir here. + + for (str library_search_path in library_search_paths) { + for (str path in fs::list_dir(library_search_path)) { + let str f = fs::basename(path); + if (!(str::starts_with(f, prefix) && + str::ends_with(f, nn.suffix))) { + log #fmt("skipping %s, doesn't look like %s*%s", path, prefix, + nn.suffix); + cont; + } + alt (get_metadata_section(path)) { + case (option::some(?cvec)) { + auto mm = get_exported_metadata(sess, path, cvec); + if (!metadata_matches(mm, metas)) { + log #fmt("skipping %s, metadata doesn't match", path); + cont; + } + log #fmt("found %s with matching metadata", path); + ret some(tup(path, cvec)); + } + case (_) { } + } + } + } + ret none; +} + +fn load_library_crate(&session::session sess, int cnum, &ast::ident ident, + &vec[@ast::meta_item] metas, + &vec[str] library_search_paths) { + alt (find_library_crate(sess, ident, metas, library_search_paths)) { + case (some(?t)) { + sess.set_external_crate(cnum, rec(name=ident, data=t._1)); + sess.add_used_crate_file(t._0); + ret; + } + case (_) { } + } + log_err #fmt("can't find crate for '%s'", ident); + fail; +} + +type env = + @rec(session::session sess, + resolve::crate_map crate_map, + @hashmap[str, int] crate_cache, + vec[str] library_search_paths, + mutable int next_crate_num); + +fn visit_view_item(env e, &@ast::view_item i) { + alt (i.node) { + case (ast::view_item_use(?ident, ?meta_items, ?id)) { + auto cnum; + if (!e.crate_cache.contains_key(ident)) { + cnum = e.next_crate_num; + load_library_crate(e.sess, cnum, ident, meta_items, + e.library_search_paths); + e.crate_cache.insert(ident, e.next_crate_num); + e.next_crate_num += 1; + } else { cnum = e.crate_cache.get(ident); } + e.crate_map.insert(id, cnum); + } + case (_) { } + } +} + +fn visit_item(env e, &@ast::item i) { + alt (i.node) { + case (ast::item_native_mod(?m)) { + auto name; + if (m.native_name == "" ) { + name = i.ident; + } else { + name = m.native_name; + } + alt (m.abi) { + case (ast::native_abi_rust) { + e.sess.add_used_library(name); + } + case (ast::native_abi_cdecl) { + e.sess.add_used_library(name); + } + case (ast::native_abi_llvm) { + } + case (ast::native_abi_rust_intrinsic) { + } + } + } + case (_) { + } + } +} + +// Reads external crates referenced by "use" directives. +fn read_crates(session::session sess, resolve::crate_map crate_map, + &ast::crate crate) { + auto e = + @rec(sess=sess, + crate_map=crate_map, + crate_cache=@common::new_str_hash[int](), + library_search_paths=sess.get_opts().library_search_paths, + mutable next_crate_num=1); + auto v = + rec(visit_view_item_pre=bind visit_view_item(e, _), + visit_item_pre=bind visit_item(e, _) + with walk::default_visitor()); + walk::walk_crate(v, crate); +} + +fn kind_has_type_params(u8 kind_ch) -> bool { + ret alt (kind_ch as char) { + case ('c') { false } + case ('f') { true } + case ('p') { true } + case ('F') { true } + case ('y') { true } + case ('t') { true } + case ('T') { false } + case ('m') { false } + case ('n') { false } + case ('v') { true } + }; +} + + +// Crate metadata queries +fn lookup_defs(session::session sess, int cnum, vec[ast::ident] path) -> + vec[ast::def] { + auto data = sess.get_external_crate(cnum).data; + ret vec::map(bind lookup_def(cnum, data, _), resolve_path(path, data)); +} + + +// FIXME doesn't yet handle re-exported externals +fn lookup_def(int cnum, vec[u8] data, &ast::def_id did_) -> ast::def { + auto item = lookup_item(did_._1, data); + auto kind_ch = item_kind(item); + auto did = tup(cnum, did_._1); + auto def = + alt (kind_ch as char) { + case ('c') { ast::def_const(did) } + case ('f') { ast::def_fn(did, ast::impure_fn) } + case ('p') { ast::def_fn(did, ast::pure_fn) } + case ('F') { ast::def_native_fn(did) } + case ('y') { ast::def_ty(did) } + case ('T') { ast::def_native_ty(did) } + // We treat references to tags as references to types. + case ('t') { ast::def_ty(did) } + case ('m') { ast::def_mod(did) } + case ('n') { ast::def_native_mod(did) } + case ('v') { + auto tid = variant_tag_id(item); + tid = tup(cnum, tid._1); + ast::def_variant(tid, did) + } + }; + ret def; +} + +fn get_type(ty::ctxt tcx, ast::def_id def) -> ty::ty_param_count_and_ty { + auto external_crate_id = def._0; + auto data = tcx.sess.get_external_crate(external_crate_id).data; + auto item = lookup_item(def._1, data); + auto t = item_type(item, external_crate_id, tcx); + auto tp_count; + auto kind_ch = item_kind(item); + auto has_ty_params = kind_has_type_params(kind_ch); + if (has_ty_params) { + tp_count = item_ty_param_count(item, external_crate_id); + } else { tp_count = 0u; } + ret tup(tp_count, t); +} + +fn get_symbol(session::session sess, ast::def_id def) -> str { + auto external_crate_id = def._0; + auto data = sess.get_external_crate(external_crate_id).data; + ret item_symbol(lookup_item(def._1, data)); +} + +fn get_tag_variants(ty::ctxt tcx, ast::def_id def) -> vec[ty::variant_info] { + auto external_crate_id = def._0; + auto data = tcx.sess.get_external_crate(external_crate_id).data; + auto items = ebml::get_doc(ebml::new_doc(data), cwriter::tag_items); + auto item = find_item(def._1, items); + let vec[ty::variant_info] infos = []; + auto variant_ids = tag_variant_ids(item, external_crate_id); + for (ast::def_id did in variant_ids) { + auto item = find_item(did._1, items); + auto ctor_ty = item_type(item, external_crate_id, tcx); + let vec[ty::t] arg_tys = []; + alt (ty::struct(tcx, ctor_ty)) { + case (ty::ty_fn(_, ?args, _, _, _)) { + for (ty::arg a in args) { arg_tys += [a.ty]; } + } + case (_) { + // Nullary tag variant. + + } + } + infos += [rec(args=arg_tys, ctor_ty=ctor_ty, id=did)]; + } + ret infos; +} + +fn list_file_metadata(str path, io::writer out) { + alt (get_metadata_section(path)) { + case (option::some(?bytes)) { list_crate_metadata(bytes, out); } + case (option::none) { + out.write_str("Could not find metadata in " + path + ".\n"); + } + } +} + +fn read_path(&ebml::doc d) -> tup(str, uint) { + auto desc = ebml::doc_data(d); + auto pos = ebml::be_uint_from_bytes(desc, 0u, 4u); + auto pathbytes = vec::slice[u8](desc, 4u, vec::len[u8](desc)); + auto path = str::unsafe_from_bytes(pathbytes); + ret tup(path, pos); +} + +fn list_crate_metadata(vec[u8] bytes, io::writer out) { + auto md = ebml::new_doc(bytes); + auto paths = ebml::get_doc(md, cwriter::tag_paths); + auto items = ebml::get_doc(md, cwriter::tag_items); + auto index = ebml::get_doc(paths, cwriter::tag_index); + auto bs = ebml::get_doc(index, cwriter::tag_index_buckets); + for each (ebml::doc bucket in + ebml::tagged_docs(bs, cwriter::tag_index_buckets_bucket)) { + auto et = cwriter::tag_index_buckets_bucket_elt; + for each (ebml::doc elt in ebml::tagged_docs(bucket, et)) { + auto data = read_path(elt); + auto def = ebml::doc_at(bytes, data._1); + auto did_doc = ebml::get_doc(def, cwriter::tag_def_id); + auto did = parse_def_id(ebml::doc_data(did_doc)); + out.write_str(#fmt("%s (%s)\n", data._0, + describe_def(items, did))); + } + } +} + +fn describe_def(&ebml::doc items, ast::def_id id) -> str { + if (id._0 != 0) { ret "external"; } + ret item_kind_to_str(item_kind(find_item(id._1, items))); +} + +fn item_kind_to_str(u8 kind) -> str { + alt (kind as char) { + case ('c') { ret "const"; } + case ('f') { ret "fn"; } + case ('p') { ret "pred"; } + case ('F') { ret "native fn"; } + case ('y') { ret "type"; } + case ('T') { ret "native type"; } + case ('t') { ret "type"; } + case ('m') { ret "mod"; } + case ('n') { ret "native mod"; } + case ('v') { ret "tag"; } + } +} +// Local Variables: +// mode: rust +// fill-column: 78; +// indent-tabs-mode: nil +// c-basic-offset: 4 +// buffer-file-coding-system: utf-8-unix +// compile-command: "make -k -C $RBUILD 2>&1 | sed -e 's/\\/x\\//x:\\//g'"; +// End: diff --git a/src/comp/metadata/cwriter.rs b/src/comp/metadata/cwriter.rs new file mode 100644 index 00000000000..e8e8654fd7f --- /dev/null +++ b/src/comp/metadata/cwriter.rs @@ -0,0 +1,817 @@ + +import std::str; +import std::uint; +import std::vec; +import std::map::hashmap; +import std::ebml; +import std::io; +import std::option; +import std::option::some; +import std::option::none; +import front::ast::*; +import middle::trans; +import middle::ty; +import middle::ty::path_to_str; +import back::x86; +import back::link; +import util::common; +import pretty::ppaux::lit_to_str; +import lib::llvm::llvm; +import lib::llvm::llvm::ValueRef; +import lib::llvm::False; + +export ac_no_abbrevs; +export def_to_str; +export encode; +export hash_def_id; +export hash_path; +export tag_def_id; +export tag_index; +export tag_index_table; +export tag_index_buckets; +export tag_index_buckets_bucket; +export tag_index_buckets_bucket_elt; +export tag_items; +export tag_items_data_item_kind; +export tag_items_data_item_symbol; +export tag_items_data_item_tag_id; +export tag_items_data_item_type; +export tag_items_data_item_ty_param_count; +export tag_items_data_item_variant; +export tag_meta_export; +export tag_meta_item; +export tag_meta_item_key; +export tag_meta_item_value; +export tag_paths; +export ty_abbrev; +export write_metadata; + +const uint tag_paths = 0x01u; + +const uint tag_items = 0x02u; + +const uint tag_paths_data = 0x03u; + +const uint tag_paths_data_name = 0x04u; + +const uint tag_paths_data_item = 0x05u; + +const uint tag_paths_data_mod = 0x06u; + +const uint tag_def_id = 0x07u; + +const uint tag_items_data = 0x08u; + +const uint tag_items_data_item = 0x09u; + +const uint tag_items_data_item_kind = 0x0au; + +const uint tag_items_data_item_ty_param_count = 0x0bu; + +const uint tag_items_data_item_type = 0x0cu; + +const uint tag_items_data_item_symbol = 0x0du; + +const uint tag_items_data_item_variant = 0x0eu; + +const uint tag_items_data_item_tag_id = 0x0fu; + +const uint tag_index = 0x11u; + +const uint tag_index_buckets = 0x12u; + +const uint tag_index_buckets_bucket = 0x13u; + +const uint tag_index_buckets_bucket_elt = 0x14u; + +const uint tag_index_table = 0x15u; + +const uint tag_meta_export = 0x16u; + +const uint tag_meta_local = 0x17u; + +const uint tag_meta_item = 0x18u; + +const uint tag_meta_item_key = 0x19u; + +const uint tag_meta_item_value = 0x20u; + + +// Type encoding + +// Compact string representation for ty.t values. API ty_str & parse_from_str. +// Extra parameters are for converting to/from def_ids in the string rep. +// Whatever format you choose should not contain pipe characters. +type ty_abbrev = rec(uint pos, uint len, str s); + +tag abbrev_ctxt { ac_no_abbrevs; ac_use_abbrevs(hashmap[ty::t, ty_abbrev]); } + +mod encode { + type ctxt = + rec(fn(&def_id) -> str ds, // Def -> str Callback: + + ty::ctxt tcx, // The type context. + + abbrev_ctxt abbrevs); + + fn cx_uses_abbrevs(&@ctxt cx) -> bool { + alt (cx.abbrevs) { + case (ac_no_abbrevs) { ret false; } + case (ac_use_abbrevs(_)) { ret true; } + } + } + fn ty_str(&@ctxt cx, &ty::t t) -> str { + assert (!cx_uses_abbrevs(cx)); + auto sw = io::string_writer(); + enc_ty(sw.get_writer(), cx, t); + ret sw.get_str(); + } + fn enc_ty(&io::writer w, &@ctxt cx, &ty::t t) { + alt (cx.abbrevs) { + case (ac_no_abbrevs) { + auto result_str; + alt (cx.tcx.short_names_cache.find(t)) { + case (some(?s)) { result_str = s; } + case (none) { + auto sw = io::string_writer(); + enc_sty(sw.get_writer(), cx, ty::struct(cx.tcx, t)); + result_str = sw.get_str(); + cx.tcx.short_names_cache.insert(t, result_str); + } + } + w.write_str(result_str); + } + case (ac_use_abbrevs(?abbrevs)) { + alt (abbrevs.find(t)) { + case (some(?a)) { w.write_str(a.s); ret; } + case (none) { + auto pos = w.get_buf_writer().tell(); + auto ss = enc_sty(w, cx, ty::struct(cx.tcx, t)); + auto end = w.get_buf_writer().tell(); + auto len = end - pos; + fn estimate_sz(uint u) -> uint { + auto n = u; + auto len = 0u; + while (n != 0u) { len += 1u; n = n >> 4u; } + ret len; + } + auto abbrev_len = + 3u + estimate_sz(pos) + estimate_sz(len); + if (abbrev_len < len) { + // I.e. it's actually an abbreviation. + + auto s = + "#" + uint::to_str(pos, 16u) + ":" + + uint::to_str(len, 16u) + "#"; + auto a = rec(pos=pos, len=len, s=s); + abbrevs.insert(t, a); + } + ret; + } + } + } + } + } + fn enc_mt(&io::writer w, &@ctxt cx, &ty::mt mt) { + alt (mt.mut) { + case (imm) { } + case (mut) { w.write_char('m'); } + case (maybe_mut) { w.write_char('?'); } + } + enc_ty(w, cx, mt.ty); + } + fn enc_sty(&io::writer w, &@ctxt cx, &ty::sty st) { + alt (st) { + case (ty::ty_nil) { w.write_char('n'); } + case (ty::ty_bot) { w.write_char('z'); } + case (ty::ty_bool) { w.write_char('b'); } + case (ty::ty_int) { w.write_char('i'); } + case (ty::ty_uint) { w.write_char('u'); } + case (ty::ty_float) { w.write_char('l'); } + case (ty::ty_machine(?mach)) { + alt (mach) { + case (common::ty_u8) { w.write_str("Mb"); } + case (common::ty_u16) { w.write_str("Mw"); } + case (common::ty_u32) { w.write_str("Ml"); } + case (common::ty_u64) { w.write_str("Md"); } + case (common::ty_i8) { w.write_str("MB"); } + case (common::ty_i16) { w.write_str("MW"); } + case (common::ty_i32) { w.write_str("ML"); } + case (common::ty_i64) { w.write_str("MD"); } + case (common::ty_f32) { w.write_str("Mf"); } + case (common::ty_f64) { w.write_str("MF"); } + } + } + case (ty::ty_char) { w.write_char('c'); } + case (ty::ty_str) { w.write_char('s'); } + case (ty::ty_istr) { w.write_char('S'); } + case (ty::ty_tag(?def, ?tys)) { + w.write_str("t["); + w.write_str(cx.ds(def)); + w.write_char('|'); + for (ty::t t in tys) { enc_ty(w, cx, t); } + w.write_char(']'); + } + case (ty::ty_box(?mt)) { w.write_char('@'); enc_mt(w, cx, mt); } + case (ty::ty_ptr(?mt)) { w.write_char('*'); enc_mt(w, cx, mt); } + case (ty::ty_vec(?mt)) { w.write_char('V'); enc_mt(w, cx, mt); } + case (ty::ty_ivec(?mt)) { w.write_char('I'); enc_mt(w, cx, mt); } + case (ty::ty_port(?t)) { w.write_char('P'); enc_ty(w, cx, t); } + case (ty::ty_chan(?t)) { w.write_char('C'); enc_ty(w, cx, t); } + case (ty::ty_tup(?mts)) { + w.write_str("T["); + for (ty::mt mt in mts) { enc_mt(w, cx, mt); } + w.write_char(']'); + } + case (ty::ty_rec(?fields)) { + w.write_str("R["); + for (ty::field field in fields) { + w.write_str(field.ident); + w.write_char('='); + enc_mt(w, cx, field.mt); + } + w.write_char(']'); + } + case (ty::ty_fn(?proto, ?args, ?out, ?cf, ?constrs)) { + enc_proto(w, proto); + enc_ty_fn(w, cx, args, out, cf, constrs); + } + case (ty::ty_native_fn(?abi, ?args, ?out)) { + w.write_char('N'); + alt (abi) { + case (native_abi_rust) { w.write_char('r'); } + case (native_abi_rust_intrinsic) { + w.write_char('i'); + } + case (native_abi_cdecl) { w.write_char('c'); } + case (native_abi_llvm) { w.write_char('l'); } + } + enc_ty_fn(w, cx, args, out, return, []); + } + case (ty::ty_obj(?methods)) { + w.write_str("O["); + for (ty::method m in methods) { + enc_proto(w, m.proto); + w.write_str(m.ident); + enc_ty_fn(w, cx, m.inputs, m.output, m.cf, m.constrs); + } + w.write_char(']'); + } + case (ty::ty_res(?def, ?ty)) { + w.write_char('r'); + w.write_str(cx.ds(def)); + w.write_char('|'); + enc_ty(w, cx, ty); + } + case (ty::ty_var(?id)) { + w.write_char('X'); + w.write_str(common::istr(id)); + } + case (ty::ty_native) { w.write_char('E'); } + case (ty::ty_param(?id)) { + w.write_char('p'); + w.write_str(common::uistr(id)); + } + case (ty::ty_type) { w.write_char('Y'); } + case (ty::ty_task) { w.write_char('a'); } + } + } + fn enc_proto(&io::writer w, proto proto) { + alt (proto) { + case (proto_iter) { w.write_char('W'); } + case (proto_fn) { w.write_char('F'); } + } + } + fn enc_ty_fn(&io::writer w, &@ctxt cx, &vec[ty::arg] args, &ty::t out, + &controlflow cf, &vec[@ty::constr_def] constrs) { + w.write_char('['); + for (ty::arg arg in args) { + alt (arg.mode) { + case (ty::mo_alias(?mut)) { + w.write_char('&'); + if (mut) { w.write_char('m'); } + } + case (ty::mo_val) { } + } + enc_ty(w, cx, arg.ty); + } + w.write_char(']'); + auto colon = true; + for (@ty::constr_def c in constrs) { + if (colon) { + w.write_char(':'); + colon = false; + } else { w.write_char(';'); } + enc_constr(w, cx, c); + } + alt (cf) { + case (noreturn) { w.write_char('!'); } + case (_) { enc_ty(w, cx, out); } + } + + } + fn enc_constr(&io::writer w, &@ctxt cx, &@ty::constr_def c) { + w.write_str(path_to_str(c.node.path)); + w.write_char('('); + w.write_str(cx.ds(c.node.id)); + w.write_char('|'); + auto semi = false; + for (@constr_arg a in c.node.args) { + if (semi) { w.write_char(';'); } else { semi = true; } + alt (a.node) { + case (carg_base) { w.write_char('*'); } + case (carg_ident(?i)) { + w.write_uint(i); + } + case (carg_lit(?l)) { w.write_str(lit_to_str(l)); } + } + } + w.write_char(')'); + } +} + + +// Returns a Plain Old LLVM String: +fn C_postr(&str s) -> ValueRef { + ret llvm::LLVMConstString(str::buf(s), str::byte_len(s), False); +} + + +// Path table encoding +fn encode_name(&ebml::writer ebml_w, &str name) { + ebml::start_tag(ebml_w, tag_paths_data_name); + ebml_w.writer.write(str::bytes(name)); + ebml::end_tag(ebml_w); +} + +fn encode_def_id(&ebml::writer ebml_w, &def_id id) { + ebml::start_tag(ebml_w, tag_def_id); + ebml_w.writer.write(str::bytes(def_to_str(id))); + ebml::end_tag(ebml_w); +} + +fn encode_tag_variant_paths(&ebml::writer ebml_w, &vec[variant] variants, + &vec[str] path, + &mutable vec[tup(str, uint)] index) { + for (variant variant in variants) { + add_to_index(ebml_w, path, index, variant.node.name); + ebml::start_tag(ebml_w, tag_paths_data_item); + encode_name(ebml_w, variant.node.name); + encode_def_id(ebml_w, local_def(variant.node.id)); + ebml::end_tag(ebml_w); + } +} + +fn add_to_index(&ebml::writer ebml_w, &vec[str] path, + &mutable vec[tup(str, uint)] index, &str name) { + auto full_path = path + [name]; + index += [tup(str::connect(full_path, "::"), ebml_w.writer.tell())]; +} + +fn encode_native_module_item_paths(&ebml::writer ebml_w, + &native_mod nmod, &vec[str] path, + &mutable vec[tup(str, uint)] index) { + for (@native_item nitem in nmod.items) { + add_to_index(ebml_w, path, index, nitem.ident); + ebml::start_tag(ebml_w, tag_paths_data_item); + encode_name(ebml_w, nitem.ident); + encode_def_id(ebml_w, local_def(nitem.id)); + ebml::end_tag(ebml_w); + } +} + +fn encode_module_item_paths(&ebml::writer ebml_w, &_mod module, + &vec[str] path, + &mutable vec[tup(str, uint)] index) { + for (@item it in module.items) { + if (!is_exported(it.ident, module)) { cont; } + alt (it.node) { + case (item_const(_, _)) { + add_to_index(ebml_w, path, index, it.ident); + ebml::start_tag(ebml_w, tag_paths_data_item); + encode_name(ebml_w, it.ident); + encode_def_id(ebml_w, local_def(it.id)); + ebml::end_tag(ebml_w); + } + case (item_fn(_, ?tps)) { + add_to_index(ebml_w, path, index, it.ident); + ebml::start_tag(ebml_w, tag_paths_data_item); + encode_name(ebml_w, it.ident); + encode_def_id(ebml_w, local_def(it.id)); + ebml::end_tag(ebml_w); + } + case (item_mod(?_mod)) { + add_to_index(ebml_w, path, index, it.ident); + ebml::start_tag(ebml_w, tag_paths_data_mod); + encode_name(ebml_w, it.ident); + encode_def_id(ebml_w, local_def(it.id)); + encode_module_item_paths(ebml_w, _mod, path + [it.ident], + index); + ebml::end_tag(ebml_w); + } + case (item_native_mod(?nmod)) { + add_to_index(ebml_w, path, index, it.ident); + ebml::start_tag(ebml_w, tag_paths_data_mod); + encode_name(ebml_w, it.ident); + encode_def_id(ebml_w, local_def(it.id)); + encode_native_module_item_paths(ebml_w, nmod, + path + [it.ident], index); + ebml::end_tag(ebml_w); + } + case (item_ty(_, ?tps)) { + add_to_index(ebml_w, path, index, it.ident); + ebml::start_tag(ebml_w, tag_paths_data_item); + encode_name(ebml_w, it.ident); + encode_def_id(ebml_w, local_def(it.id)); + ebml::end_tag(ebml_w); + } + case (item_res(_, _, ?tps, ?ctor_id)) { + add_to_index(ebml_w, path, index, it.ident); + ebml::start_tag(ebml_w, tag_paths_data_item); + encode_name(ebml_w, it.ident); + encode_def_id(ebml_w, local_def(ctor_id)); + ebml::end_tag(ebml_w); + add_to_index(ebml_w, path, index, it.ident); + ebml::start_tag(ebml_w, tag_paths_data_item); + encode_name(ebml_w, it.ident); + encode_def_id(ebml_w, local_def(it.id)); + ebml::end_tag(ebml_w); + } + case (item_tag(?variants, ?tps)) { + add_to_index(ebml_w, path, index, it.ident); + ebml::start_tag(ebml_w, tag_paths_data_item); + encode_name(ebml_w, it.ident); + encode_def_id(ebml_w, local_def(it.id)); + ebml::end_tag(ebml_w); + encode_tag_variant_paths(ebml_w, variants, path, index); + } + case (item_obj(_, ?tps, ?ctor_id)) { + add_to_index(ebml_w, path, index, it.ident); + ebml::start_tag(ebml_w, tag_paths_data_item); + encode_name(ebml_w, it.ident); + encode_def_id(ebml_w, local_def(ctor_id)); + ebml::end_tag(ebml_w); + add_to_index(ebml_w, path, index, it.ident); + ebml::start_tag(ebml_w, tag_paths_data_item); + encode_name(ebml_w, it.ident); + encode_def_id(ebml_w, local_def(it.id)); + ebml::end_tag(ebml_w); + } + } + } +} + +fn encode_item_paths(&ebml::writer ebml_w, &@crate crate) -> + vec[tup(str, uint)] { + let vec[tup(str, uint)] index = []; + let vec[str] path = []; + ebml::start_tag(ebml_w, tag_paths); + encode_module_item_paths(ebml_w, crate.node.module, path, index); + ebml::end_tag(ebml_w); + ret index; +} + + +// Item info table encoding +fn encode_kind(&ebml::writer ebml_w, u8 c) { + ebml::start_tag(ebml_w, tag_items_data_item_kind); + ebml_w.writer.write([c]); + ebml::end_tag(ebml_w); +} + +fn def_to_str(&def_id did) -> str { ret #fmt("%d:%d", did._0, did._1); } + +fn encode_type_param_count(&ebml::writer ebml_w, &vec[ty_param] tps) { + ebml::start_tag(ebml_w, tag_items_data_item_ty_param_count); + ebml::write_vint(ebml_w.writer, vec::len[ty_param](tps)); + ebml::end_tag(ebml_w); +} + +fn encode_variant_id(&ebml::writer ebml_w, &def_id vid) { + ebml::start_tag(ebml_w, tag_items_data_item_variant); + ebml_w.writer.write(str::bytes(def_to_str(vid))); + ebml::end_tag(ebml_w); +} + +fn encode_type(&@trans::crate_ctxt cx, &ebml::writer ebml_w, &ty::t typ) { + ebml::start_tag(ebml_w, tag_items_data_item_type); + auto f = def_to_str; + auto ty_str_ctxt = + @rec(ds=f, tcx=cx.tcx, abbrevs=ac_use_abbrevs(cx.type_abbrevs)); + encode::enc_ty(io::new_writer_(ebml_w.writer), ty_str_ctxt, typ); + ebml::end_tag(ebml_w); +} + +fn encode_symbol(&@trans::crate_ctxt cx, &ebml::writer ebml_w, + node_id id) { + ebml::start_tag(ebml_w, tag_items_data_item_symbol); + ebml_w.writer.write(str::bytes(cx.item_symbols.get(id))); + ebml::end_tag(ebml_w); +} + +fn encode_discriminant(&@trans::crate_ctxt cx, &ebml::writer ebml_w, + node_id id) { + ebml::start_tag(ebml_w, tag_items_data_item_symbol); + ebml_w.writer.write(str::bytes(cx.discrim_symbols.get(id))); + ebml::end_tag(ebml_w); +} + +fn encode_tag_id(&ebml::writer ebml_w, &def_id id) { + ebml::start_tag(ebml_w, tag_items_data_item_tag_id); + ebml_w.writer.write(str::bytes(def_to_str(id))); + ebml::end_tag(ebml_w); +} + +fn encode_tag_variant_info(&@trans::crate_ctxt cx, &ebml::writer ebml_w, + node_id id, &vec[variant] variants, + &mutable vec[tup(int, uint)] index, + &vec[ty_param] ty_params) { + for (variant variant in variants) { + index += [tup(variant.node.id, ebml_w.writer.tell())]; + ebml::start_tag(ebml_w, tag_items_data_item); + encode_def_id(ebml_w, local_def(variant.node.id)); + encode_kind(ebml_w, 'v' as u8); + encode_tag_id(ebml_w, local_def(id)); + encode_type(cx, ebml_w, trans::node_id_type(cx, variant.node.id)); + if (vec::len[variant_arg](variant.node.args) > 0u) { + encode_symbol(cx, ebml_w, variant.node.id); + } + encode_discriminant(cx, ebml_w, variant.node.id); + encode_type_param_count(ebml_w, ty_params); + ebml::end_tag(ebml_w); + } +} + +fn encode_info_for_item(@trans::crate_ctxt cx, &ebml::writer ebml_w, + @item item, &mutable vec[tup(int, uint)] index) { + alt (item.node) { + case (item_const(_, _)) { + ebml::start_tag(ebml_w, tag_items_data_item); + encode_def_id(ebml_w, local_def(item.id)); + encode_kind(ebml_w, 'c' as u8); + encode_type(cx, ebml_w, trans::node_id_type(cx, item.id)); + encode_symbol(cx, ebml_w, item.id); + ebml::end_tag(ebml_w); + } + case (item_fn(?fd, ?tps)) { + ebml::start_tag(ebml_w, tag_items_data_item); + encode_def_id(ebml_w, local_def(item.id)); + encode_kind(ebml_w, alt (fd.decl.purity) { + case (pure_fn) { 'p' } + case (impure_fn) { 'f' } } as u8); + encode_type_param_count(ebml_w, tps); + encode_type(cx, ebml_w, trans::node_id_type(cx, item.id)); + encode_symbol(cx, ebml_w, item.id); + ebml::end_tag(ebml_w); + } + case (item_mod(_)) { + ebml::start_tag(ebml_w, tag_items_data_item); + encode_def_id(ebml_w, local_def(item.id)); + encode_kind(ebml_w, 'm' as u8); + ebml::end_tag(ebml_w); + } + case (item_native_mod(_)) { + ebml::start_tag(ebml_w, tag_items_data_item); + encode_def_id(ebml_w, local_def(item.id)); + encode_kind(ebml_w, 'n' as u8); + ebml::end_tag(ebml_w); + } + case (item_ty(_, ?tps)) { + ebml::start_tag(ebml_w, tag_items_data_item); + encode_def_id(ebml_w, local_def(item.id)); + encode_kind(ebml_w, 'y' as u8); + encode_type_param_count(ebml_w, tps); + encode_type(cx, ebml_w, trans::node_id_type(cx, item.id)); + ebml::end_tag(ebml_w); + } + case (item_tag(?variants, ?tps)) { + ebml::start_tag(ebml_w, tag_items_data_item); + encode_def_id(ebml_w, local_def(item.id)); + encode_kind(ebml_w, 't' as u8); + encode_type_param_count(ebml_w, tps); + encode_type(cx, ebml_w, trans::node_id_type(cx, item.id)); + for (variant v in variants) { + encode_variant_id(ebml_w, local_def(v.node.id)); + } + ebml::end_tag(ebml_w); + encode_tag_variant_info(cx, ebml_w, item.id, variants, index, + tps); + } + case (item_res(_, _, ?tps, ?ctor_id)) { + auto fn_ty = trans::node_id_type(cx, item.id); + + ebml::start_tag(ebml_w, tag_items_data_item); + encode_def_id(ebml_w, local_def(item.id)); + encode_kind(ebml_w, 'y' as u8); + encode_type_param_count(ebml_w, tps); + encode_type(cx, ebml_w, ty::ty_fn_ret(cx.tcx, fn_ty)); + ebml::end_tag(ebml_w); + + index += [tup(ctor_id, ebml_w.writer.tell())]; + ebml::start_tag(ebml_w, tag_items_data_item); + encode_def_id(ebml_w, local_def(ctor_id)); + encode_kind(ebml_w, 'f' as u8); + encode_type_param_count(ebml_w, tps); + encode_type(cx, ebml_w, fn_ty); + encode_symbol(cx, ebml_w, ctor_id); + ebml::end_tag(ebml_w); + } + case (item_obj(_, ?tps, ?ctor_id)) { + auto fn_ty = trans::node_id_type(cx, ctor_id); + + ebml::start_tag(ebml_w, tag_items_data_item); + encode_def_id(ebml_w, local_def(item.id)); + encode_kind(ebml_w, 'y' as u8); + encode_type_param_count(ebml_w, tps); + encode_type(cx, ebml_w, ty::ty_fn_ret(cx.tcx, fn_ty)); + ebml::end_tag(ebml_w); + + index += [tup(ctor_id, ebml_w.writer.tell())]; + ebml::start_tag(ebml_w, tag_items_data_item); + encode_def_id(ebml_w, local_def(ctor_id)); + encode_kind(ebml_w, 'f' as u8); + encode_type_param_count(ebml_w, tps); + encode_type(cx, ebml_w, fn_ty); + encode_symbol(cx, ebml_w, ctor_id); + ebml::end_tag(ebml_w); + } + } +} + +fn encode_info_for_native_item(&@trans::crate_ctxt cx, &ebml::writer ebml_w, + &@native_item nitem) { + ebml::start_tag(ebml_w, tag_items_data_item); + alt (nitem.node) { + case (native_item_ty) { + encode_def_id(ebml_w, local_def(nitem.id)); + encode_kind(ebml_w, 'T' as u8); + encode_type(cx, ebml_w, ty::mk_native(cx.tcx)); + } + case (native_item_fn(_, _, ?tps)) { + encode_def_id(ebml_w, local_def(nitem.id)); + encode_kind(ebml_w, 'F' as u8); + encode_type_param_count(ebml_w, tps); + encode_type(cx, ebml_w, trans::node_id_type(cx, nitem.id)); + encode_symbol(cx, ebml_w, nitem.id); + } + } + ebml::end_tag(ebml_w); +} + +fn encode_info_for_items(&@trans::crate_ctxt cx, &ebml::writer ebml_w) -> + vec[tup(int, uint)] { + let vec[tup(int, uint)] index = []; + ebml::start_tag(ebml_w, tag_items_data); + for each (@tup(node_id, middle::ast_map::ast_node) kvp in cx.ast_map.items()) { + alt (kvp._1) { + case (middle::ast_map::node_item(?i)) { + index += [tup(kvp._0, ebml_w.writer.tell())]; + encode_info_for_item(cx, ebml_w, i, index); + } + case (middle::ast_map::node_native_item(?i)) { + index += [tup(kvp._0, ebml_w.writer.tell())]; + encode_info_for_native_item(cx, ebml_w, i); + } + case (_) {} + } + } + ebml::end_tag(ebml_w); + ret index; +} + + +// Path and definition ID indexing + +// djb's cdb hashes. +fn hash_def_id(&int def_id) -> uint { ret 177573u ^ (def_id as uint); } + +fn hash_path(&str s) -> uint { + auto h = 5381u; + for (u8 ch in str::bytes(s)) { h = (h << 5u) + h ^ (ch as uint); } + ret h; +} + +fn create_index[T](&vec[tup(T, uint)] index, fn(&T) -> uint hash_fn) -> + vec[vec[tup(T, uint)]] { + let vec[mutable vec[tup(T, uint)]] buckets = vec::empty_mut(); + for each (uint i in uint::range(0u, 256u)) { buckets += [mutable []]; } + for (tup(T, uint) elt in index) { + auto h = hash_fn(elt._0); + buckets.(h % 256u) += [elt]; + } + ret vec::freeze(buckets); +} + +fn encode_index[T](&ebml::writer ebml_w, &vec[vec[tup(T, uint)]] buckets, + fn(&io::writer, &T) write_fn) { + auto writer = io::new_writer_(ebml_w.writer); + ebml::start_tag(ebml_w, tag_index); + let vec[uint] bucket_locs = []; + ebml::start_tag(ebml_w, tag_index_buckets); + for (vec[tup(T, uint)] bucket in buckets) { + bucket_locs += [ebml_w.writer.tell()]; + ebml::start_tag(ebml_w, tag_index_buckets_bucket); + for (tup(T, uint) elt in bucket) { + ebml::start_tag(ebml_w, tag_index_buckets_bucket_elt); + writer.write_be_uint(elt._1, 4u); + write_fn(writer, elt._0); + ebml::end_tag(ebml_w); + } + ebml::end_tag(ebml_w); + } + ebml::end_tag(ebml_w); + ebml::start_tag(ebml_w, tag_index_table); + for (uint pos in bucket_locs) { writer.write_be_uint(pos, 4u); } + ebml::end_tag(ebml_w); + ebml::end_tag(ebml_w); +} + +fn write_str(&io::writer writer, &str s) { writer.write_str(s); } + +fn write_int(&io::writer writer, &int n) { + writer.write_be_uint(n as uint, 4u); +} + +fn encode_meta_items(&ebml::writer ebml_w, &crate crate) { + fn encode_meta_item(&ebml::writer ebml_w, &meta_item mi) { + // FIXME (#487): Support all forms of meta item + ebml::start_tag(ebml_w, tag_meta_item); + alt (mi.node) { + case (meta_key_value(?key, ?value)) { + ebml::start_tag(ebml_w, tag_meta_item_key); + ebml_w.writer.write(str::bytes(key)); + ebml::end_tag(ebml_w); + ebml::start_tag(ebml_w, tag_meta_item_value); + ebml_w.writer.write(str::bytes(value)); + ebml::end_tag(ebml_w); + } + case (_) { + log_err "unimplemented meta_item type"; + } + } + ebml::end_tag(ebml_w); + } + ebml::start_tag(ebml_w, tag_meta_export); + for each (@meta_item mi in link::crate_export_metas(crate)) { + encode_meta_item(ebml_w, *mi); + } + ebml::end_tag(ebml_w); + ebml::start_tag(ebml_w, tag_meta_local); + for each (@meta_item mi in link::crate_local_metas(crate)) { + encode_meta_item(ebml_w, *mi); + } + ebml::end_tag(ebml_w); +} + +fn encode_metadata(&@trans::crate_ctxt cx, &@crate crate) -> ValueRef { + auto string_w = io::string_writer(); + auto buf_w = string_w.get_writer().get_buf_writer(); + auto ebml_w = ebml::create_writer(buf_w); + // Encode the meta items + + encode_meta_items(ebml_w, *crate); + // Encode and index the paths. + + ebml::start_tag(ebml_w, tag_paths); + auto paths_index = encode_item_paths(ebml_w, crate); + auto str_writer = write_str; + auto path_hasher = hash_path; + auto paths_buckets = create_index[str](paths_index, path_hasher); + encode_index[str](ebml_w, paths_buckets, str_writer); + ebml::end_tag(ebml_w); + // Encode and index the items. + + ebml::start_tag(ebml_w, tag_items); + auto items_index = encode_info_for_items(cx, ebml_w); + auto int_writer = write_int; + auto item_hasher = hash_def_id; + auto items_buckets = create_index[int](items_index, item_hasher); + encode_index[int](ebml_w, items_buckets, int_writer); + ebml::end_tag(ebml_w); + // Pad this, since something (LLVM, presumably) is cutting off the + // remaining % 4 bytes. + + buf_w.write([0u8, 0u8, 0u8, 0u8]); + ret C_postr(string_w.get_str()); +} + +fn write_metadata(&@trans::crate_ctxt cx, &@crate crate) { + if (!cx.sess.get_opts().shared) { ret; } + auto llmeta = encode_metadata(cx, crate); + auto llconst = trans::C_struct([llmeta]); + auto llglobal = + llvm::LLVMAddGlobal(cx.llmod, trans::val_ty(llconst), + str::buf("rust_metadata")); + llvm::LLVMSetInitializer(llglobal, llconst); + llvm::LLVMSetSection(llglobal, str::buf(x86::get_meta_sect_name())); +} +// +// Local Variables: +// mode: rust +// fill-column: 78; +// indent-tabs-mode: nil +// c-basic-offset: 4 +// buffer-file-coding-system: utf-8-unix +// compile-command: "make -k -C $RBUILD 2>&1 | sed -e 's/\\/x\\//x:\\//g'"; +// End: +// |
