From b5a838247bd7d66037ba48378c34ba4460747deb Mon Sep 17 00:00:00 2001 From: Veikka Tuominen Date: Thu, 14 Jul 2022 21:26:01 +0300 Subject: stage2: point to error location using spans --- src/Compilation.zig | 59 ++++++++++++++++++++++++++++------------------------- 1 file changed, 31 insertions(+), 28 deletions(-) (limited to 'src/Compilation.zig') diff --git a/src/Compilation.zig b/src/Compilation.zig index b3ae4e787d..edca150988 100644 --- a/src/Compilation.zig +++ b/src/Compilation.zig @@ -338,7 +338,7 @@ pub const AllErrors = struct { src_path: []const u8, line: u32, column: u32, - byte_offset: u32, + span: Module.SrcLoc.Span, /// Usually one, but incremented for redundant messages. count: u32 = 1, /// Does not include the trailing newline. @@ -429,7 +429,10 @@ pub const AllErrors = struct { try stderr.writeByte('\n'); try stderr.writeByteNTimes(' ', src.column); ttyconf.setColor(stderr, .Green); - try stderr.writeAll("^\n"); + try stderr.writeByte('^'); + // TODO basic unicode code point monospace width + try stderr.writeByteNTimes('~', src.span.end - src.span.start - 1); + try stderr.writeByte('\n'); ttyconf.setColor(stderr, .Reset); } } @@ -469,7 +472,8 @@ pub const AllErrors = struct { hasher.update(src.src_path); std.hash.autoHash(&hasher, src.line); std.hash.autoHash(&hasher, src.column); - std.hash.autoHash(&hasher, src.byte_offset); + std.hash.autoHash(&hasher, src.span.start); + std.hash.autoHash(&hasher, src.span.end); }, .plain => |plain| { hasher.update(plain.msg); @@ -488,7 +492,8 @@ pub const AllErrors = struct { mem.eql(u8, a_src.src_path, b_src.src_path) and a_src.line == b_src.line and a_src.column == b_src.column and - a_src.byte_offset == b_src.byte_offset; + a_src.span.start == b_src.span.start and + a_src.span.end == b_src.span.end; }, .plain => return false, }, @@ -527,20 +532,20 @@ pub const AllErrors = struct { std.hash_map.default_max_load_percentage, ).init(allocator); const err_source = try module_err_msg.src_loc.file_scope.getSource(module.gpa); - const err_byte_offset = try module_err_msg.src_loc.byteOffset(module.gpa); - const err_loc = std.zig.findLineColumn(err_source.bytes, err_byte_offset); + const err_span = try module_err_msg.src_loc.span(module.gpa); + const err_loc = std.zig.findLineColumn(err_source.bytes, err_span.start); for (module_err_msg.notes) |module_note| { const source = try module_note.src_loc.file_scope.getSource(module.gpa); - const byte_offset = try module_note.src_loc.byteOffset(module.gpa); - const loc = std.zig.findLineColumn(source.bytes, byte_offset); + const span = try module_note.src_loc.span(module.gpa); + const loc = std.zig.findLineColumn(source.bytes, span.start); const file_path = try module_note.src_loc.file_scope.fullPath(allocator); const note = ¬es_buf[note_i]; note.* = .{ .src = .{ .src_path = file_path, .msg = try allocator.dupe(u8, module_note.msg), - .byte_offset = byte_offset, + .span = span, .line = @intCast(u32, loc.line), .column = @intCast(u32, loc.column), .source_line = if (err_loc.eql(loc)) null else try allocator.dupe(u8, loc.source_line), @@ -566,7 +571,7 @@ pub const AllErrors = struct { .src = .{ .src_path = file_path, .msg = try allocator.dupe(u8, module_err_msg.msg), - .byte_offset = err_byte_offset, + .span = err_span, .line = @intCast(u32, err_loc.line), .column = @intCast(u32, err_loc.column), .notes = notes_buf[0..note_i], @@ -593,16 +598,15 @@ pub const AllErrors = struct { while (item_i < items_len) : (item_i += 1) { const item = file.zir.extraData(Zir.Inst.CompileErrors.Item, extra_index); extra_index = item.end; - const err_byte_offset = blk: { - const token_starts = file.tree.tokens.items(.start); + const err_span = blk: { if (item.data.node != 0) { - const main_tokens = file.tree.nodes.items(.main_token); - const main_token = main_tokens[item.data.node]; - break :blk token_starts[main_token]; + break :blk Module.SrcLoc.nodeToSpan(&file.tree, item.data.node); } - break :blk token_starts[item.data.token] + item.data.byte_offset; + const token_starts = file.tree.tokens.items(.start); + const start = token_starts[item.data.token] + item.data.byte_offset; + break :blk Module.SrcLoc.Span{ .start = start, .end = start + 1 }; }; - const err_loc = std.zig.findLineColumn(file.source, err_byte_offset); + const err_loc = std.zig.findLineColumn(file.source, err_span.start); var notes: []Message = &[0]Message{}; if (item.data.notes != 0) { @@ -612,22 +616,21 @@ pub const AllErrors = struct { for (notes) |*note, i| { const note_item = file.zir.extraData(Zir.Inst.CompileErrors.Item, body[i]); const msg = file.zir.nullTerminatedString(note_item.data.msg); - const byte_offset = blk: { - const token_starts = file.tree.tokens.items(.start); + const span = blk: { if (note_item.data.node != 0) { - const main_tokens = file.tree.nodes.items(.main_token); - const main_token = main_tokens[note_item.data.node]; - break :blk token_starts[main_token]; + break :blk Module.SrcLoc.nodeToSpan(&file.tree, note_item.data.node); } - break :blk token_starts[note_item.data.token] + note_item.data.byte_offset; + const token_starts = file.tree.tokens.items(.start); + const start = token_starts[note_item.data.token] + note_item.data.byte_offset; + break :blk Module.SrcLoc.Span{ .start = start, .end = start + 1 }; }; - const loc = std.zig.findLineColumn(file.source, byte_offset); + const loc = std.zig.findLineColumn(file.source, span.start); note.* = .{ .src = .{ .src_path = try file.fullPath(arena), .msg = try arena.dupe(u8, msg), - .byte_offset = byte_offset, + .span = span, .line = @intCast(u32, loc.line), .column = @intCast(u32, loc.column), .notes = &.{}, // TODO rework this function to be recursive @@ -642,7 +645,7 @@ pub const AllErrors = struct { .src = .{ .src_path = try file.fullPath(arena), .msg = try arena.dupe(u8, msg), - .byte_offset = err_byte_offset, + .span = err_span, .line = @intCast(u32, err_loc.line), .column = @intCast(u32, err_loc.column), .notes = notes, @@ -688,7 +691,7 @@ pub const AllErrors = struct { .src_path = try arena.dupe(u8, src.src_path), .line = src.line, .column = src.column, - .byte_offset = src.byte_offset, + .span = src.span, .source_line = if (src.source_line) |s| try arena.dupe(u8, s) else null, .notes = try dupeList(src.notes, arena), } }, @@ -2662,7 +2665,7 @@ pub fn getAllErrorsAlloc(self: *Compilation) !AllErrors { .msg = try std.fmt.allocPrint(arena_allocator, "unable to build C object: {s}", .{ err_msg.msg, }), - .byte_offset = 0, + .span = .{ .start = 0, .end = 1 }, .line = err_msg.line, .column = err_msg.column, .source_line = null, // TODO -- cgit v1.2.3 From 1463144fc88550ba1dca3888acdcdd4903781222 Mon Sep 17 00:00:00 2001 From: Veikka Tuominen Date: Fri, 15 Jul 2022 11:38:16 +0300 Subject: Compilation: point caret in error message at the main token --- src/Compilation.zig | 32 ++++++++++++---------- src/Module.zig | 71 +++++++++++++++++++++++-------------------------- src/main.zig | 14 +++++++--- src/print_zir.zig | 2 +- test/compile_errors.zig | 9 +++++++ 5 files changed, 72 insertions(+), 56 deletions(-) (limited to 'src/Compilation.zig') diff --git a/src/Compilation.zig b/src/Compilation.zig index edca150988..0e4d76889d 100644 --- a/src/Compilation.zig +++ b/src/Compilation.zig @@ -427,11 +427,15 @@ pub const AllErrors = struct { else => try stderr.writeByte(b), }; try stderr.writeByte('\n'); - try stderr.writeByteNTimes(' ', src.column); + // TODO basic unicode code point monospace width + const before_caret = src.span.main - src.span.start; + // -1 since span.main includes the caret + const after_caret = src.span.end - src.span.main -| 1; + try stderr.writeByteNTimes(' ', src.column - before_caret); ttyconf.setColor(stderr, .Green); + try stderr.writeByteNTimes('~', before_caret); try stderr.writeByte('^'); - // TODO basic unicode code point monospace width - try stderr.writeByteNTimes('~', src.span.end - src.span.start - 1); + try stderr.writeByteNTimes('~', after_caret); try stderr.writeByte('\n'); ttyconf.setColor(stderr, .Reset); } @@ -472,8 +476,7 @@ pub const AllErrors = struct { hasher.update(src.src_path); std.hash.autoHash(&hasher, src.line); std.hash.autoHash(&hasher, src.column); - std.hash.autoHash(&hasher, src.span.start); - std.hash.autoHash(&hasher, src.span.end); + std.hash.autoHash(&hasher, src.span.main); }, .plain => |plain| { hasher.update(plain.msg); @@ -492,8 +495,7 @@ pub const AllErrors = struct { mem.eql(u8, a_src.src_path, b_src.src_path) and a_src.line == b_src.line and a_src.column == b_src.column and - a_src.span.start == b_src.span.start and - a_src.span.end == b_src.span.end; + a_src.span.main == b_src.span.main; }, .plain => return false, }, @@ -533,12 +535,12 @@ pub const AllErrors = struct { ).init(allocator); const err_source = try module_err_msg.src_loc.file_scope.getSource(module.gpa); const err_span = try module_err_msg.src_loc.span(module.gpa); - const err_loc = std.zig.findLineColumn(err_source.bytes, err_span.start); + const err_loc = std.zig.findLineColumn(err_source.bytes, err_span.main); for (module_err_msg.notes) |module_note| { const source = try module_note.src_loc.file_scope.getSource(module.gpa); const span = try module_note.src_loc.span(module.gpa); - const loc = std.zig.findLineColumn(source.bytes, span.start); + const loc = std.zig.findLineColumn(source.bytes, span.main); const file_path = try module_note.src_loc.file_scope.fullPath(allocator); const note = ¬es_buf[note_i]; note.* = .{ @@ -604,9 +606,10 @@ pub const AllErrors = struct { } const token_starts = file.tree.tokens.items(.start); const start = token_starts[item.data.token] + item.data.byte_offset; - break :blk Module.SrcLoc.Span{ .start = start, .end = start + 1 }; + const end = start + @intCast(u32, file.tree.tokenSlice(item.data.token).len); + break :blk Module.SrcLoc.Span{ .start = start, .end = end, .main = start }; }; - const err_loc = std.zig.findLineColumn(file.source, err_span.start); + const err_loc = std.zig.findLineColumn(file.source, err_span.main); var notes: []Message = &[0]Message{}; if (item.data.notes != 0) { @@ -622,9 +625,10 @@ pub const AllErrors = struct { } const token_starts = file.tree.tokens.items(.start); const start = token_starts[note_item.data.token] + note_item.data.byte_offset; - break :blk Module.SrcLoc.Span{ .start = start, .end = start + 1 }; + const end = start + @intCast(u32, file.tree.tokenSlice(note_item.data.token).len); + break :blk Module.SrcLoc.Span{ .start = start, .end = end, .main = start }; }; - const loc = std.zig.findLineColumn(file.source, span.start); + const loc = std.zig.findLineColumn(file.source, span.main); note.* = .{ .src = .{ @@ -2665,7 +2669,7 @@ pub fn getAllErrorsAlloc(self: *Compilation) !AllErrors { .msg = try std.fmt.allocPrint(arena_allocator, "unable to build C object: {s}", .{ err_msg.msg, }), - .span = .{ .start = 0, .end = 1 }, + .span = .{ .start = 0, .end = 1, .main = 0 }, .line = err_msg.line, .column = err_msg.column, .source_line = null, // TODO diff --git a/src/Module.zig b/src/Module.zig index 08625406ab..35b48406cb 100644 --- a/src/Module.zig +++ b/src/Module.zig @@ -2085,20 +2085,21 @@ pub const SrcLoc = struct { pub const Span = struct { start: u32, end: u32, + main: u32, }; pub fn span(src_loc: SrcLoc, gpa: Allocator) !Span { switch (src_loc.lazy) { .unneeded => unreachable, - .entire_file => return Span{ .start = 0, .end = 1 }, + .entire_file => return Span{ .start = 0, .end = 1, .main = 0 }, - .byte_abs => |byte_index| return Span{ .start = byte_index, .end = byte_index + 1 }, + .byte_abs => |byte_index| return Span{ .start = byte_index, .end = byte_index + 1, .main = byte_index }, .token_abs => |tok_index| { const tree = try src_loc.file_scope.getTree(gpa); const start = tree.tokens.items(.start)[tok_index]; const end = start + @intCast(u32, tree.tokenSlice(tok_index).len); - return Span{ .start = start, .end = end }; + return Span{ .start = start, .end = end, .main = start }; }, .node_abs => |node| { const tree = try src_loc.file_scope.getTree(gpa); @@ -2109,14 +2110,14 @@ pub const SrcLoc = struct { const tok_index = src_loc.declSrcToken(); const start = tree.tokens.items(.start)[tok_index] + byte_off; const end = start + @intCast(u32, tree.tokenSlice(tok_index).len); - return Span{ .start = start, .end = end }; + return Span{ .start = start, .end = end, .main = start }; }, .token_offset => |tok_off| { const tree = try src_loc.file_scope.getTree(gpa); const tok_index = src_loc.declSrcToken() + tok_off; const start = tree.tokens.items(.start)[tok_index]; const end = start + @intCast(u32, tree.tokenSlice(tok_index).len); - return Span{ .start = start, .end = end }; + return Span{ .start = start, .end = end, .main = start }; }, .node_offset => |traced_off| { const node_off = traced_off.x; @@ -2137,7 +2138,7 @@ pub const SrcLoc = struct { const tok_index = tree.firstToken(node) - 2; const start = tree.tokens.items(.start)[tok_index]; const end = start + @intCast(u32, tree.tokenSlice(tok_index).len); - return Span{ .start = start, .end = end }; + return Span{ .start = start, .end = end, .main = start }; }, .node_offset_var_decl_ty => |node_off| { const tree = try src_loc.file_scope.getTree(gpa); @@ -2158,7 +2159,7 @@ pub const SrcLoc = struct { }; const start = tree.tokens.items(.start)[tok_index]; const end = start + @intCast(u32, tree.tokenSlice(tok_index).len); - return Span{ .start = start, .end = end }; + return Span{ .start = start, .end = end, .main = start }; }, .node_offset_builtin_call_arg0 => |n| return src_loc.byteOffsetBuiltinCallArg(gpa, n, 0), .node_offset_builtin_call_arg1 => |n| return src_loc.byteOffsetBuiltinCallArg(gpa, n, 1), @@ -2186,16 +2187,13 @@ pub const SrcLoc = struct { .slice_sentinel => tree.sliceSentinel(node), else => unreachable, }; - const main_tokens = tree.nodes.items(.main_token); - const part_node = main_tokens[ - switch (src_loc.lazy) { - .node_offset_slice_ptr => full.ast.sliced, - .node_offset_slice_start => full.ast.start, - .node_offset_slice_end => full.ast.end, - .node_offset_slice_sentinel => full.ast.sentinel, - else => unreachable, - } - ]; + const part_node = switch (src_loc.lazy) { + .node_offset_slice_ptr => full.ast.sliced, + .node_offset_slice_start => full.ast.start, + .node_offset_slice_end => full.ast.end, + .node_offset_slice_sentinel => full.ast.sentinel, + else => unreachable, + }; return nodeToSpan(tree, part_node); }, .node_offset_call_func => |node_off| { @@ -2231,7 +2229,7 @@ pub const SrcLoc = struct { }; const start = tree.tokens.items(.start)[tok_index]; const end = start + @intCast(u32, tree.tokenSlice(tok_index).len); - return Span{ .start = start, .end = end }; + return Span{ .start = start, .end = end, .main = start }; }, .node_offset_deref_ptr => |node_off| { const tree = try src_loc.file_scope.getTree(gpa); @@ -2422,7 +2420,7 @@ pub const SrcLoc = struct { const tok_index = full.lib_name.?; const start = tree.tokens.items(.start)[tok_index]; const end = start + @intCast(u32, tree.tokenSlice(tok_index).len); - return Span{ .start = start, .end = end }; + return Span{ .start = start, .end = end, .main = start }; }, .node_offset_array_type_len => |node_off| { @@ -2495,28 +2493,25 @@ pub const SrcLoc = struct { pub fn nodeToSpan(tree: *const Ast, node: u32) Span { const token_starts = tree.tokens.items(.start); + const main_token = tree.nodes.items(.main_token)[node]; const start = tree.firstToken(node); const end = tree.lastToken(node); - if (tree.tokensOnSameLine(start, end)) { - const start_off = token_starts[start]; - const end_off = token_starts[end] + @intCast(u32, tree.tokenSlice(end).len); - return Span{ .start = start_off, .end = end_off }; - } + var start_tok = start; + var end_tok = end; - const main_token = tree.nodes.items(.main_token)[node]; - if (tree.tokensOnSameLine(start, main_token)) { - const start_off = token_starts[start]; - const end_off = token_starts[main_token] + @intCast(u32, tree.tokenSlice(main_token).len); - return Span{ .start = start_off, .end = end_off }; - } - if (tree.tokensOnSameLine(main_token, end)) { - const start_off = token_starts[main_token]; - const end_off = token_starts[end] + @intCast(u32, tree.tokenSlice(end).len); - return Span{ .start = start_off, .end = end_off }; + if (tree.tokensOnSameLine(start, end)) { + // do nothing + } else if (tree.tokensOnSameLine(start, main_token)) { + end_tok = main_token; + } else if (tree.tokensOnSameLine(main_token, end)) { + start_tok = main_token; + } else { + start_tok = main_token; + end_tok = main_token; } - const start_off = token_starts[main_token]; - const end_off = token_starts[main_token] + @intCast(u32, tree.tokenSlice(main_token).len); - return Span{ .start = start_off, .end = end_off }; + const start_off = token_starts[start_tok]; + const end_off = token_starts[end_tok] + @intCast(u32, tree.tokenSlice(end_tok).len); + return Span{ .start = start_off, .end = end_off, .main = token_starts[main_token] }; } }; @@ -3283,7 +3278,7 @@ pub fn astGenFile(mod: *Module, file: *File) !void { .lazy = if (extra_offset == 0) .{ .token_abs = parse_err.token, } else .{ - .byte_abs = token_starts[parse_err.token], + .byte_abs = token_starts[parse_err.token] + extra_offset, }, }, .msg = msg.toOwnedSlice(), diff --git a/src/main.zig b/src/main.zig index 7187dac3e7..a95c3c5dd7 100644 --- a/src/main.zig +++ b/src/main.zig @@ -4381,7 +4381,7 @@ fn printErrsMsgToStdErr( .msg = try std.fmt.allocPrint(arena, "invalid byte: '{'}'", .{ std.zig.fmtEscapes(tree.source[byte_offset..][0..1]), }), - .span = .{ .start = byte_offset, .end = byte_offset + 1 }, + .span = .{ .start = byte_offset, .end = byte_offset + 1, .main = byte_offset }, .line = @intCast(u32, start_loc.line), .column = @intCast(u32, start_loc.column) + bad_off, .source_line = source_line, @@ -4401,7 +4401,11 @@ fn printErrsMsgToStdErr( .src = .{ .src_path = path, .msg = try arena.dupe(u8, text_buf.items), - .span = .{ .start = byte_offset, .end = byte_offset + @intCast(u32, tree.tokenSlice(note.token).len) }, + .span = .{ + .start = byte_offset, + .end = byte_offset + @intCast(u32, tree.tokenSlice(note.token).len), + .main = byte_offset, + }, .line = @intCast(u32, note_loc.line), .column = @intCast(u32, note_loc.column), .source_line = tree.source[note_loc.line_start..note_loc.line_end], @@ -4417,7 +4421,11 @@ fn printErrsMsgToStdErr( .src = .{ .src_path = path, .msg = text, - .span = .{ .start = byte_offset, .end = byte_offset + @intCast(u32, tree.tokenSlice(lok_token).len) }, + .span = .{ + .start = byte_offset, + .end = byte_offset + @intCast(u32, tree.tokenSlice(lok_token).len), + .main = byte_offset, + }, .line = @intCast(u32, start_loc.line), .column = @intCast(u32, start_loc.column) + extra_offset, .source_line = source_line, diff --git a/src/print_zir.zig b/src/print_zir.zig index 3b622a70fd..48ee8894e8 100644 --- a/src/print_zir.zig +++ b/src/print_zir.zig @@ -2386,7 +2386,7 @@ const Writer = struct { const end = std.zig.findLineColumn(tree.source, src_span.end); try stream.print("{s}:{d}:{d} to :{d}:{d}", .{ @tagName(src), start.line + 1, start.column + 1, - end.line + 1, end.column + 1, + end.line + 1, end.column + 1, }); } } diff --git a/test/compile_errors.zig b/test/compile_errors.zig index 1337917323..fe6145cc12 100644 --- a/test/compile_errors.zig +++ b/test/compile_errors.zig @@ -174,6 +174,15 @@ pub fn addCases(ctx: *TestContext) !void { }); } + { + const case = ctx.obj("missing semicolon at EOF", .{}); + case.addError( + \\const foo = 1 + , &[_][]const u8{ + \\:1:14: error: expected ';' after declaration + }); + } + // TODO test this in stage2, but we won't even try in stage1 //ctx.objErrStage1("inline fn calls itself indirectly", // \\export fn foo() void { -- cgit v1.2.3