go/src/cmd/compile/internal/syntax/parser.go

2783 lines
61 KiB
Go
Raw Normal View History

// Copyright 2016 The Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
package syntax
import (
"fmt"
"io"
"strconv"
"strings"
)
const debug = false
const trace = false
type parser struct {
file *PosBase
errh ErrorHandler
mode Mode
pragh PragmaHandler
scanner
cmd/compile/internal/syntax: remove dependency on cmd/internal/src For dependency reasons, the data structure implementing source positions in the compiler is in cmd/internal/src. It contains highly compiler specific details (e.g. inlining index). This change introduces a parallel but simpler position representation, defined in the syntax package, which removes that package's dependency on cmd/internal/src, and also removes the need to deal with certain filename-specific operations (defined by the needs of the compiler) in the syntax package. As a result, the syntax package becomes again a compiler- independent, stand-alone package that at some point might replace (or augment) the existing top-level go/* syntax-related packages. Additionally, line directives that update column numbers are now correctly tracked through the syntax package, with additional tests added. (The respective changes also need to be made in cmd/internal/src; i.e., the compiler accepts but still ignores column numbers in line directives.) This change comes at the cost of a new position translation step, but that step is cheap because it only needs to do real work if the position base changed (i.e., if there is a new file, or new line directive). There is no noticeable impact on overall compiler performance measured with `compilebench -count 5 -alloc`: name old time/op new time/op delta Template 220ms ± 8% 228ms ±18% ~ (p=0.548 n=5+5) Unicode 119ms ±11% 113ms ± 5% ~ (p=0.056 n=5+5) GoTypes 684ms ± 6% 677ms ± 3% ~ (p=0.841 n=5+5) Compiler 3.19s ± 7% 3.01s ± 1% ~ (p=0.095 n=5+5) SSA 7.92s ± 8% 7.79s ± 1% ~ (p=0.690 n=5+5) Flate 141ms ± 7% 139ms ± 4% ~ (p=0.548 n=5+5) GoParser 173ms ±12% 171ms ± 4% ~ (p=1.000 n=5+5) Reflect 417ms ± 5% 411ms ± 3% ~ (p=0.548 n=5+5) Tar 205ms ± 5% 198ms ± 2% ~ (p=0.690 n=5+5) XML 232ms ± 4% 229ms ± 4% ~ (p=0.690 n=5+5) StdCmd 28.7s ± 5% 28.2s ± 2% ~ (p=0.421 n=5+5) name old user-time/op new user-time/op delta Template 269ms ± 4% 265ms ± 5% ~ (p=0.421 n=5+5) Unicode 153ms ± 7% 149ms ± 3% ~ (p=0.841 n=5+5) GoTypes 850ms ± 7% 862ms ± 4% ~ (p=0.841 n=5+5) Compiler 4.01s ± 5% 3.86s ± 0% ~ (p=0.190 n=5+4) SSA 10.9s ± 4% 10.8s ± 2% ~ (p=0.548 n=5+5) Flate 166ms ± 7% 167ms ± 6% ~ (p=1.000 n=5+5) GoParser 204ms ± 8% 206ms ± 7% ~ (p=0.841 n=5+5) Reflect 514ms ± 5% 508ms ± 4% ~ (p=0.548 n=5+5) Tar 245ms ± 6% 244ms ± 3% ~ (p=0.690 n=5+5) XML 280ms ± 4% 278ms ± 4% ~ (p=0.841 n=5+5) name old alloc/op new alloc/op delta Template 37.9MB ± 0% 37.9MB ± 0% ~ (p=0.841 n=5+5) Unicode 28.8MB ± 0% 28.8MB ± 0% ~ (p=0.841 n=5+5) GoTypes 113MB ± 0% 113MB ± 0% ~ (p=0.151 n=5+5) Compiler 468MB ± 0% 468MB ± 0% -0.01% (p=0.032 n=5+5) SSA 1.50GB ± 0% 1.50GB ± 0% ~ (p=0.548 n=5+5) Flate 24.4MB ± 0% 24.4MB ± 0% ~ (p=1.000 n=5+5) GoParser 30.7MB ± 0% 30.7MB ± 0% ~ (p=1.000 n=5+5) Reflect 76.5MB ± 0% 76.5MB ± 0% ~ (p=0.548 n=5+5) Tar 38.9MB ± 0% 38.9MB ± 0% ~ (p=0.222 n=5+5) XML 41.6MB ± 0% 41.6MB ± 0% ~ (p=0.548 n=5+5) name old allocs/op new allocs/op delta Template 382k ± 0% 382k ± 0% +0.01% (p=0.008 n=5+5) Unicode 343k ± 0% 343k ± 0% ~ (p=0.841 n=5+5) GoTypes 1.19M ± 0% 1.19M ± 0% +0.01% (p=0.008 n=5+5) Compiler 4.53M ± 0% 4.53M ± 0% +0.03% (p=0.008 n=5+5) SSA 12.4M ± 0% 12.4M ± 0% +0.00% (p=0.008 n=5+5) Flate 235k ± 0% 235k ± 0% ~ (p=0.079 n=5+5) GoParser 318k ± 0% 318k ± 0% ~ (p=0.730 n=5+5) Reflect 978k ± 0% 978k ± 0% ~ (p=1.000 n=5+5) Tar 393k ± 0% 393k ± 0% ~ (p=0.056 n=5+5) XML 405k ± 0% 405k ± 0% ~ (p=0.548 n=5+5) name old text-bytes new text-bytes delta HelloSize 672kB ± 0% 672kB ± 0% ~ (all equal) CmdGoSize 7.12MB ± 0% 7.12MB ± 0% ~ (all equal) name old data-bytes new data-bytes delta HelloSize 133kB ± 0% 133kB ± 0% ~ (all equal) CmdGoSize 390kB ± 0% 390kB ± 0% ~ (all equal) name old exe-bytes new exe-bytes delta HelloSize 1.07MB ± 0% 1.07MB ± 0% ~ (all equal) CmdGoSize 11.2MB ± 0% 11.2MB ± 0% ~ (all equal) Passes toolstash compare. For #22662. Change-Id: I19edb53dd9675af57f7122cb7dba2a6d8bdcc3da Reviewed-on: https://go-review.googlesource.com/94515 Reviewed-by: Matthew Dempsky <mdempsky@google.com>
2018-01-02 16:58:37 -08:00
base *PosBase // current position base
first error // first error encountered
errcnt int // number of errors encountered
pragma Pragma // pragmas
fnest int // function nesting level (for error handling)
xnest int // expression nesting level (for complit ambiguity resolution)
indent []byte // tracing support
}
cmd/compile/internal/syntax: remove dependency on cmd/internal/src For dependency reasons, the data structure implementing source positions in the compiler is in cmd/internal/src. It contains highly compiler specific details (e.g. inlining index). This change introduces a parallel but simpler position representation, defined in the syntax package, which removes that package's dependency on cmd/internal/src, and also removes the need to deal with certain filename-specific operations (defined by the needs of the compiler) in the syntax package. As a result, the syntax package becomes again a compiler- independent, stand-alone package that at some point might replace (or augment) the existing top-level go/* syntax-related packages. Additionally, line directives that update column numbers are now correctly tracked through the syntax package, with additional tests added. (The respective changes also need to be made in cmd/internal/src; i.e., the compiler accepts but still ignores column numbers in line directives.) This change comes at the cost of a new position translation step, but that step is cheap because it only needs to do real work if the position base changed (i.e., if there is a new file, or new line directive). There is no noticeable impact on overall compiler performance measured with `compilebench -count 5 -alloc`: name old time/op new time/op delta Template 220ms ± 8% 228ms ±18% ~ (p=0.548 n=5+5) Unicode 119ms ±11% 113ms ± 5% ~ (p=0.056 n=5+5) GoTypes 684ms ± 6% 677ms ± 3% ~ (p=0.841 n=5+5) Compiler 3.19s ± 7% 3.01s ± 1% ~ (p=0.095 n=5+5) SSA 7.92s ± 8% 7.79s ± 1% ~ (p=0.690 n=5+5) Flate 141ms ± 7% 139ms ± 4% ~ (p=0.548 n=5+5) GoParser 173ms ±12% 171ms ± 4% ~ (p=1.000 n=5+5) Reflect 417ms ± 5% 411ms ± 3% ~ (p=0.548 n=5+5) Tar 205ms ± 5% 198ms ± 2% ~ (p=0.690 n=5+5) XML 232ms ± 4% 229ms ± 4% ~ (p=0.690 n=5+5) StdCmd 28.7s ± 5% 28.2s ± 2% ~ (p=0.421 n=5+5) name old user-time/op new user-time/op delta Template 269ms ± 4% 265ms ± 5% ~ (p=0.421 n=5+5) Unicode 153ms ± 7% 149ms ± 3% ~ (p=0.841 n=5+5) GoTypes 850ms ± 7% 862ms ± 4% ~ (p=0.841 n=5+5) Compiler 4.01s ± 5% 3.86s ± 0% ~ (p=0.190 n=5+4) SSA 10.9s ± 4% 10.8s ± 2% ~ (p=0.548 n=5+5) Flate 166ms ± 7% 167ms ± 6% ~ (p=1.000 n=5+5) GoParser 204ms ± 8% 206ms ± 7% ~ (p=0.841 n=5+5) Reflect 514ms ± 5% 508ms ± 4% ~ (p=0.548 n=5+5) Tar 245ms ± 6% 244ms ± 3% ~ (p=0.690 n=5+5) XML 280ms ± 4% 278ms ± 4% ~ (p=0.841 n=5+5) name old alloc/op new alloc/op delta Template 37.9MB ± 0% 37.9MB ± 0% ~ (p=0.841 n=5+5) Unicode 28.8MB ± 0% 28.8MB ± 0% ~ (p=0.841 n=5+5) GoTypes 113MB ± 0% 113MB ± 0% ~ (p=0.151 n=5+5) Compiler 468MB ± 0% 468MB ± 0% -0.01% (p=0.032 n=5+5) SSA 1.50GB ± 0% 1.50GB ± 0% ~ (p=0.548 n=5+5) Flate 24.4MB ± 0% 24.4MB ± 0% ~ (p=1.000 n=5+5) GoParser 30.7MB ± 0% 30.7MB ± 0% ~ (p=1.000 n=5+5) Reflect 76.5MB ± 0% 76.5MB ± 0% ~ (p=0.548 n=5+5) Tar 38.9MB ± 0% 38.9MB ± 0% ~ (p=0.222 n=5+5) XML 41.6MB ± 0% 41.6MB ± 0% ~ (p=0.548 n=5+5) name old allocs/op new allocs/op delta Template 382k ± 0% 382k ± 0% +0.01% (p=0.008 n=5+5) Unicode 343k ± 0% 343k ± 0% ~ (p=0.841 n=5+5) GoTypes 1.19M ± 0% 1.19M ± 0% +0.01% (p=0.008 n=5+5) Compiler 4.53M ± 0% 4.53M ± 0% +0.03% (p=0.008 n=5+5) SSA 12.4M ± 0% 12.4M ± 0% +0.00% (p=0.008 n=5+5) Flate 235k ± 0% 235k ± 0% ~ (p=0.079 n=5+5) GoParser 318k ± 0% 318k ± 0% ~ (p=0.730 n=5+5) Reflect 978k ± 0% 978k ± 0% ~ (p=1.000 n=5+5) Tar 393k ± 0% 393k ± 0% ~ (p=0.056 n=5+5) XML 405k ± 0% 405k ± 0% ~ (p=0.548 n=5+5) name old text-bytes new text-bytes delta HelloSize 672kB ± 0% 672kB ± 0% ~ (all equal) CmdGoSize 7.12MB ± 0% 7.12MB ± 0% ~ (all equal) name old data-bytes new data-bytes delta HelloSize 133kB ± 0% 133kB ± 0% ~ (all equal) CmdGoSize 390kB ± 0% 390kB ± 0% ~ (all equal) name old exe-bytes new exe-bytes delta HelloSize 1.07MB ± 0% 1.07MB ± 0% ~ (all equal) CmdGoSize 11.2MB ± 0% 11.2MB ± 0% ~ (all equal) Passes toolstash compare. For #22662. Change-Id: I19edb53dd9675af57f7122cb7dba2a6d8bdcc3da Reviewed-on: https://go-review.googlesource.com/94515 Reviewed-by: Matthew Dempsky <mdempsky@google.com>
2018-01-02 16:58:37 -08:00
func (p *parser) init(file *PosBase, r io.Reader, errh ErrorHandler, pragh PragmaHandler, mode Mode) {
p.file = file
p.errh = errh
p.mode = mode
p.pragh = pragh
p.scanner.init(
[dev.inline] cmd/internal/src: replace src.Pos with syntax.Pos This replaces the src.Pos LineHist-based position tracking with the syntax.Pos implementation and updates all uses. The LineHist table is not used anymore - the respective code is still there but should be removed eventually. CL forthcoming. Passes toolstash -cmp when comparing to the master repo (with the exception of a couple of swapped assembly instructions, likely due to different instruction scheduling because the line-based sorting has changed; though this is won't affect correctness). The sizes of various important compiler data structures have increased significantly (see the various sizes_test.go files); this is probably the reason for an increase of compilation times (to be addressed). Here are the results of compilebench -count 5, run on a "quiet" machine (no apps running besides a terminal): name old time/op new time/op delta Template 256ms ± 1% 280ms ±15% +9.54% (p=0.008 n=5+5) Unicode 132ms ± 1% 132ms ± 1% ~ (p=0.690 n=5+5) GoTypes 891ms ± 1% 917ms ± 2% +2.88% (p=0.008 n=5+5) Compiler 3.84s ± 2% 3.99s ± 2% +3.95% (p=0.016 n=5+5) MakeBash 47.1s ± 1% 47.2s ± 2% ~ (p=0.841 n=5+5) name old user-ns/op new user-ns/op delta Template 309M ± 1% 326M ± 2% +5.18% (p=0.008 n=5+5) Unicode 165M ± 1% 168M ± 4% ~ (p=0.421 n=5+5) GoTypes 1.14G ± 2% 1.18G ± 1% +3.47% (p=0.008 n=5+5) Compiler 5.00G ± 1% 5.16G ± 1% +3.12% (p=0.008 n=5+5) Change-Id: I241c4246cdff627d7ecb95cac23060b38f9775ec Reviewed-on: https://go-review.googlesource.com/34273 Run-TryBot: Robert Griesemer <gri@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Matthew Dempsky <mdempsky@google.com>
2016-12-09 17:15:05 -08:00
r,
cmd/compile/internal/syntax: implement comment reporting in scanner R=go1.11 In order to collect comments in the AST and for error testing purposes, the scanner needs to not only recognize and skip comments, but also be able to report them if so desired. This change adds a mode flag to the scanner's init function which controls the scanner behavior around comments. In the common case where comments are not needed, there must be no significant overhead. Thus, comments are reported via a handler upcall rather than being returned as a _Comment token (which the parser would have to filter out with every scanner.next() call). Because the handlers for error messages, directives, and comments all look the same (they take a position and text), and because directives look like comments, and errors never start with a '/', this change simplifies the scanner's init call to only take one (error) handler instead of 2 or 3 different handlers with identical signature. It is trivial in the handler to determine if we have an error, directive, or general comment. Finally, because directives are comments, when reporting directives the full comment text is returned now rather than just the directive text. This simplifies the implementation and makes the scanner API more regular. Furthermore, it provides important information about the comment style used by a directive, which may matter eventually when we fully implement /*line file:line:col*/ directives. Change-Id: I2adbfcebecd615e4237ed3a832b6ceb9518bf09c Reviewed-on: https://go-review.googlesource.com/88215 Reviewed-by: Matthew Dempsky <mdempsky@google.com>
2018-01-17 21:42:51 -08:00
// Error and directive handler for scanner.
// Because the (line, col) positions passed to the
// handler is always at or after the current reading
// position, it is safe to use the most recent position
// base to compute the corresponding Pos value.
func(line, col uint, msg string) {
cmd/compile/internal/syntax: implement comment reporting in scanner R=go1.11 In order to collect comments in the AST and for error testing purposes, the scanner needs to not only recognize and skip comments, but also be able to report them if so desired. This change adds a mode flag to the scanner's init function which controls the scanner behavior around comments. In the common case where comments are not needed, there must be no significant overhead. Thus, comments are reported via a handler upcall rather than being returned as a _Comment token (which the parser would have to filter out with every scanner.next() call). Because the handlers for error messages, directives, and comments all look the same (they take a position and text), and because directives look like comments, and errors never start with a '/', this change simplifies the scanner's init call to only take one (error) handler instead of 2 or 3 different handlers with identical signature. It is trivial in the handler to determine if we have an error, directive, or general comment. Finally, because directives are comments, when reporting directives the full comment text is returned now rather than just the directive text. This simplifies the implementation and makes the scanner API more regular. Furthermore, it provides important information about the comment style used by a directive, which may matter eventually when we fully implement /*line file:line:col*/ directives. Change-Id: I2adbfcebecd615e4237ed3a832b6ceb9518bf09c Reviewed-on: https://go-review.googlesource.com/88215 Reviewed-by: Matthew Dempsky <mdempsky@google.com>
2018-01-17 21:42:51 -08:00
if msg[0] != '/' {
p.errorAt(p.posAt(line, col), msg)
[dev.inline] cmd/internal/src: replace src.Pos with syntax.Pos This replaces the src.Pos LineHist-based position tracking with the syntax.Pos implementation and updates all uses. The LineHist table is not used anymore - the respective code is still there but should be removed eventually. CL forthcoming. Passes toolstash -cmp when comparing to the master repo (with the exception of a couple of swapped assembly instructions, likely due to different instruction scheduling because the line-based sorting has changed; though this is won't affect correctness). The sizes of various important compiler data structures have increased significantly (see the various sizes_test.go files); this is probably the reason for an increase of compilation times (to be addressed). Here are the results of compilebench -count 5, run on a "quiet" machine (no apps running besides a terminal): name old time/op new time/op delta Template 256ms ± 1% 280ms ±15% +9.54% (p=0.008 n=5+5) Unicode 132ms ± 1% 132ms ± 1% ~ (p=0.690 n=5+5) GoTypes 891ms ± 1% 917ms ± 2% +2.88% (p=0.008 n=5+5) Compiler 3.84s ± 2% 3.99s ± 2% +3.95% (p=0.016 n=5+5) MakeBash 47.1s ± 1% 47.2s ± 2% ~ (p=0.841 n=5+5) name old user-ns/op new user-ns/op delta Template 309M ± 1% 326M ± 2% +5.18% (p=0.008 n=5+5) Unicode 165M ± 1% 168M ± 4% ~ (p=0.421 n=5+5) GoTypes 1.14G ± 2% 1.18G ± 1% +3.47% (p=0.008 n=5+5) Compiler 5.00G ± 1% 5.16G ± 1% +3.12% (p=0.008 n=5+5) Change-Id: I241c4246cdff627d7ecb95cac23060b38f9775ec Reviewed-on: https://go-review.googlesource.com/34273 Run-TryBot: Robert Griesemer <gri@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Matthew Dempsky <mdempsky@google.com>
2016-12-09 17:15:05 -08:00
return
}
cmd/compile/internal/syntax: implement comment reporting in scanner R=go1.11 In order to collect comments in the AST and for error testing purposes, the scanner needs to not only recognize and skip comments, but also be able to report them if so desired. This change adds a mode flag to the scanner's init function which controls the scanner behavior around comments. In the common case where comments are not needed, there must be no significant overhead. Thus, comments are reported via a handler upcall rather than being returned as a _Comment token (which the parser would have to filter out with every scanner.next() call). Because the handlers for error messages, directives, and comments all look the same (they take a position and text), and because directives look like comments, and errors never start with a '/', this change simplifies the scanner's init call to only take one (error) handler instead of 2 or 3 different handlers with identical signature. It is trivial in the handler to determine if we have an error, directive, or general comment. Finally, because directives are comments, when reporting directives the full comment text is returned now rather than just the directive text. This simplifies the implementation and makes the scanner API more regular. Furthermore, it provides important information about the comment style used by a directive, which may matter eventually when we fully implement /*line file:line:col*/ directives. Change-Id: I2adbfcebecd615e4237ed3a832b6ceb9518bf09c Reviewed-on: https://go-review.googlesource.com/88215 Reviewed-by: Matthew Dempsky <mdempsky@google.com>
2018-01-17 21:42:51 -08:00
// otherwise it must be a comment containing a line or go: directive.
// //line directives must be at the start of the line (column colbase).
// /*line*/ directives can be anywhere in the line.
cmd/compile/internal/syntax: implement comment reporting in scanner R=go1.11 In order to collect comments in the AST and for error testing purposes, the scanner needs to not only recognize and skip comments, but also be able to report them if so desired. This change adds a mode flag to the scanner's init function which controls the scanner behavior around comments. In the common case where comments are not needed, there must be no significant overhead. Thus, comments are reported via a handler upcall rather than being returned as a _Comment token (which the parser would have to filter out with every scanner.next() call). Because the handlers for error messages, directives, and comments all look the same (they take a position and text), and because directives look like comments, and errors never start with a '/', this change simplifies the scanner's init call to only take one (error) handler instead of 2 or 3 different handlers with identical signature. It is trivial in the handler to determine if we have an error, directive, or general comment. Finally, because directives are comments, when reporting directives the full comment text is returned now rather than just the directive text. This simplifies the implementation and makes the scanner API more regular. Furthermore, it provides important information about the comment style used by a directive, which may matter eventually when we fully implement /*line file:line:col*/ directives. Change-Id: I2adbfcebecd615e4237ed3a832b6ceb9518bf09c Reviewed-on: https://go-review.googlesource.com/88215 Reviewed-by: Matthew Dempsky <mdempsky@google.com>
2018-01-17 21:42:51 -08:00
text := commentText(msg)
if (col == colbase || msg[1] == '*') && strings.HasPrefix(text, "line ") {
cmd/compile/internal/syntax: remove dependency on cmd/internal/src For dependency reasons, the data structure implementing source positions in the compiler is in cmd/internal/src. It contains highly compiler specific details (e.g. inlining index). This change introduces a parallel but simpler position representation, defined in the syntax package, which removes that package's dependency on cmd/internal/src, and also removes the need to deal with certain filename-specific operations (defined by the needs of the compiler) in the syntax package. As a result, the syntax package becomes again a compiler- independent, stand-alone package that at some point might replace (or augment) the existing top-level go/* syntax-related packages. Additionally, line directives that update column numbers are now correctly tracked through the syntax package, with additional tests added. (The respective changes also need to be made in cmd/internal/src; i.e., the compiler accepts but still ignores column numbers in line directives.) This change comes at the cost of a new position translation step, but that step is cheap because it only needs to do real work if the position base changed (i.e., if there is a new file, or new line directive). There is no noticeable impact on overall compiler performance measured with `compilebench -count 5 -alloc`: name old time/op new time/op delta Template 220ms ± 8% 228ms ±18% ~ (p=0.548 n=5+5) Unicode 119ms ±11% 113ms ± 5% ~ (p=0.056 n=5+5) GoTypes 684ms ± 6% 677ms ± 3% ~ (p=0.841 n=5+5) Compiler 3.19s ± 7% 3.01s ± 1% ~ (p=0.095 n=5+5) SSA 7.92s ± 8% 7.79s ± 1% ~ (p=0.690 n=5+5) Flate 141ms ± 7% 139ms ± 4% ~ (p=0.548 n=5+5) GoParser 173ms ±12% 171ms ± 4% ~ (p=1.000 n=5+5) Reflect 417ms ± 5% 411ms ± 3% ~ (p=0.548 n=5+5) Tar 205ms ± 5% 198ms ± 2% ~ (p=0.690 n=5+5) XML 232ms ± 4% 229ms ± 4% ~ (p=0.690 n=5+5) StdCmd 28.7s ± 5% 28.2s ± 2% ~ (p=0.421 n=5+5) name old user-time/op new user-time/op delta Template 269ms ± 4% 265ms ± 5% ~ (p=0.421 n=5+5) Unicode 153ms ± 7% 149ms ± 3% ~ (p=0.841 n=5+5) GoTypes 850ms ± 7% 862ms ± 4% ~ (p=0.841 n=5+5) Compiler 4.01s ± 5% 3.86s ± 0% ~ (p=0.190 n=5+4) SSA 10.9s ± 4% 10.8s ± 2% ~ (p=0.548 n=5+5) Flate 166ms ± 7% 167ms ± 6% ~ (p=1.000 n=5+5) GoParser 204ms ± 8% 206ms ± 7% ~ (p=0.841 n=5+5) Reflect 514ms ± 5% 508ms ± 4% ~ (p=0.548 n=5+5) Tar 245ms ± 6% 244ms ± 3% ~ (p=0.690 n=5+5) XML 280ms ± 4% 278ms ± 4% ~ (p=0.841 n=5+5) name old alloc/op new alloc/op delta Template 37.9MB ± 0% 37.9MB ± 0% ~ (p=0.841 n=5+5) Unicode 28.8MB ± 0% 28.8MB ± 0% ~ (p=0.841 n=5+5) GoTypes 113MB ± 0% 113MB ± 0% ~ (p=0.151 n=5+5) Compiler 468MB ± 0% 468MB ± 0% -0.01% (p=0.032 n=5+5) SSA 1.50GB ± 0% 1.50GB ± 0% ~ (p=0.548 n=5+5) Flate 24.4MB ± 0% 24.4MB ± 0% ~ (p=1.000 n=5+5) GoParser 30.7MB ± 0% 30.7MB ± 0% ~ (p=1.000 n=5+5) Reflect 76.5MB ± 0% 76.5MB ± 0% ~ (p=0.548 n=5+5) Tar 38.9MB ± 0% 38.9MB ± 0% ~ (p=0.222 n=5+5) XML 41.6MB ± 0% 41.6MB ± 0% ~ (p=0.548 n=5+5) name old allocs/op new allocs/op delta Template 382k ± 0% 382k ± 0% +0.01% (p=0.008 n=5+5) Unicode 343k ± 0% 343k ± 0% ~ (p=0.841 n=5+5) GoTypes 1.19M ± 0% 1.19M ± 0% +0.01% (p=0.008 n=5+5) Compiler 4.53M ± 0% 4.53M ± 0% +0.03% (p=0.008 n=5+5) SSA 12.4M ± 0% 12.4M ± 0% +0.00% (p=0.008 n=5+5) Flate 235k ± 0% 235k ± 0% ~ (p=0.079 n=5+5) GoParser 318k ± 0% 318k ± 0% ~ (p=0.730 n=5+5) Reflect 978k ± 0% 978k ± 0% ~ (p=1.000 n=5+5) Tar 393k ± 0% 393k ± 0% ~ (p=0.056 n=5+5) XML 405k ± 0% 405k ± 0% ~ (p=0.548 n=5+5) name old text-bytes new text-bytes delta HelloSize 672kB ± 0% 672kB ± 0% ~ (all equal) CmdGoSize 7.12MB ± 0% 7.12MB ± 0% ~ (all equal) name old data-bytes new data-bytes delta HelloSize 133kB ± 0% 133kB ± 0% ~ (all equal) CmdGoSize 390kB ± 0% 390kB ± 0% ~ (all equal) name old exe-bytes new exe-bytes delta HelloSize 1.07MB ± 0% 1.07MB ± 0% ~ (all equal) CmdGoSize 11.2MB ± 0% 11.2MB ± 0% ~ (all equal) Passes toolstash compare. For #22662. Change-Id: I19edb53dd9675af57f7122cb7dba2a6d8bdcc3da Reviewed-on: https://go-review.googlesource.com/94515 Reviewed-by: Matthew Dempsky <mdempsky@google.com>
2018-01-02 16:58:37 -08:00
var pos Pos // position immediately following the comment
if msg[1] == '/' {
// line comment (newline is part of the comment)
cmd/compile/internal/syntax: remove dependency on cmd/internal/src For dependency reasons, the data structure implementing source positions in the compiler is in cmd/internal/src. It contains highly compiler specific details (e.g. inlining index). This change introduces a parallel but simpler position representation, defined in the syntax package, which removes that package's dependency on cmd/internal/src, and also removes the need to deal with certain filename-specific operations (defined by the needs of the compiler) in the syntax package. As a result, the syntax package becomes again a compiler- independent, stand-alone package that at some point might replace (or augment) the existing top-level go/* syntax-related packages. Additionally, line directives that update column numbers are now correctly tracked through the syntax package, with additional tests added. (The respective changes also need to be made in cmd/internal/src; i.e., the compiler accepts but still ignores column numbers in line directives.) This change comes at the cost of a new position translation step, but that step is cheap because it only needs to do real work if the position base changed (i.e., if there is a new file, or new line directive). There is no noticeable impact on overall compiler performance measured with `compilebench -count 5 -alloc`: name old time/op new time/op delta Template 220ms ± 8% 228ms ±18% ~ (p=0.548 n=5+5) Unicode 119ms ±11% 113ms ± 5% ~ (p=0.056 n=5+5) GoTypes 684ms ± 6% 677ms ± 3% ~ (p=0.841 n=5+5) Compiler 3.19s ± 7% 3.01s ± 1% ~ (p=0.095 n=5+5) SSA 7.92s ± 8% 7.79s ± 1% ~ (p=0.690 n=5+5) Flate 141ms ± 7% 139ms ± 4% ~ (p=0.548 n=5+5) GoParser 173ms ±12% 171ms ± 4% ~ (p=1.000 n=5+5) Reflect 417ms ± 5% 411ms ± 3% ~ (p=0.548 n=5+5) Tar 205ms ± 5% 198ms ± 2% ~ (p=0.690 n=5+5) XML 232ms ± 4% 229ms ± 4% ~ (p=0.690 n=5+5) StdCmd 28.7s ± 5% 28.2s ± 2% ~ (p=0.421 n=5+5) name old user-time/op new user-time/op delta Template 269ms ± 4% 265ms ± 5% ~ (p=0.421 n=5+5) Unicode 153ms ± 7% 149ms ± 3% ~ (p=0.841 n=5+5) GoTypes 850ms ± 7% 862ms ± 4% ~ (p=0.841 n=5+5) Compiler 4.01s ± 5% 3.86s ± 0% ~ (p=0.190 n=5+4) SSA 10.9s ± 4% 10.8s ± 2% ~ (p=0.548 n=5+5) Flate 166ms ± 7% 167ms ± 6% ~ (p=1.000 n=5+5) GoParser 204ms ± 8% 206ms ± 7% ~ (p=0.841 n=5+5) Reflect 514ms ± 5% 508ms ± 4% ~ (p=0.548 n=5+5) Tar 245ms ± 6% 244ms ± 3% ~ (p=0.690 n=5+5) XML 280ms ± 4% 278ms ± 4% ~ (p=0.841 n=5+5) name old alloc/op new alloc/op delta Template 37.9MB ± 0% 37.9MB ± 0% ~ (p=0.841 n=5+5) Unicode 28.8MB ± 0% 28.8MB ± 0% ~ (p=0.841 n=5+5) GoTypes 113MB ± 0% 113MB ± 0% ~ (p=0.151 n=5+5) Compiler 468MB ± 0% 468MB ± 0% -0.01% (p=0.032 n=5+5) SSA 1.50GB ± 0% 1.50GB ± 0% ~ (p=0.548 n=5+5) Flate 24.4MB ± 0% 24.4MB ± 0% ~ (p=1.000 n=5+5) GoParser 30.7MB ± 0% 30.7MB ± 0% ~ (p=1.000 n=5+5) Reflect 76.5MB ± 0% 76.5MB ± 0% ~ (p=0.548 n=5+5) Tar 38.9MB ± 0% 38.9MB ± 0% ~ (p=0.222 n=5+5) XML 41.6MB ± 0% 41.6MB ± 0% ~ (p=0.548 n=5+5) name old allocs/op new allocs/op delta Template 382k ± 0% 382k ± 0% +0.01% (p=0.008 n=5+5) Unicode 343k ± 0% 343k ± 0% ~ (p=0.841 n=5+5) GoTypes 1.19M ± 0% 1.19M ± 0% +0.01% (p=0.008 n=5+5) Compiler 4.53M ± 0% 4.53M ± 0% +0.03% (p=0.008 n=5+5) SSA 12.4M ± 0% 12.4M ± 0% +0.00% (p=0.008 n=5+5) Flate 235k ± 0% 235k ± 0% ~ (p=0.079 n=5+5) GoParser 318k ± 0% 318k ± 0% ~ (p=0.730 n=5+5) Reflect 978k ± 0% 978k ± 0% ~ (p=1.000 n=5+5) Tar 393k ± 0% 393k ± 0% ~ (p=0.056 n=5+5) XML 405k ± 0% 405k ± 0% ~ (p=0.548 n=5+5) name old text-bytes new text-bytes delta HelloSize 672kB ± 0% 672kB ± 0% ~ (all equal) CmdGoSize 7.12MB ± 0% 7.12MB ± 0% ~ (all equal) name old data-bytes new data-bytes delta HelloSize 133kB ± 0% 133kB ± 0% ~ (all equal) CmdGoSize 390kB ± 0% 390kB ± 0% ~ (all equal) name old exe-bytes new exe-bytes delta HelloSize 1.07MB ± 0% 1.07MB ± 0% ~ (all equal) CmdGoSize 11.2MB ± 0% 11.2MB ± 0% ~ (all equal) Passes toolstash compare. For #22662. Change-Id: I19edb53dd9675af57f7122cb7dba2a6d8bdcc3da Reviewed-on: https://go-review.googlesource.com/94515 Reviewed-by: Matthew Dempsky <mdempsky@google.com>
2018-01-02 16:58:37 -08:00
pos = MakePos(p.file, line+1, colbase)
} else {
// regular comment
// (if the comment spans multiple lines it's not
// a valid line directive and will be discarded
// by updateBase)
pos = MakePos(p.file, line, col+uint(len(msg)))
}
p.updateBase(pos, line, col+2+5, text[5:]) // +2 to skip over // or /*
cmd/compile/internal/syntax: implement comment reporting in scanner R=go1.11 In order to collect comments in the AST and for error testing purposes, the scanner needs to not only recognize and skip comments, but also be able to report them if so desired. This change adds a mode flag to the scanner's init function which controls the scanner behavior around comments. In the common case where comments are not needed, there must be no significant overhead. Thus, comments are reported via a handler upcall rather than being returned as a _Comment token (which the parser would have to filter out with every scanner.next() call). Because the handlers for error messages, directives, and comments all look the same (they take a position and text), and because directives look like comments, and errors never start with a '/', this change simplifies the scanner's init call to only take one (error) handler instead of 2 or 3 different handlers with identical signature. It is trivial in the handler to determine if we have an error, directive, or general comment. Finally, because directives are comments, when reporting directives the full comment text is returned now rather than just the directive text. This simplifies the implementation and makes the scanner API more regular. Furthermore, it provides important information about the comment style used by a directive, which may matter eventually when we fully implement /*line file:line:col*/ directives. Change-Id: I2adbfcebecd615e4237ed3a832b6ceb9518bf09c Reviewed-on: https://go-review.googlesource.com/88215 Reviewed-by: Matthew Dempsky <mdempsky@google.com>
2018-01-17 21:42:51 -08:00
return
}
// go: directive (but be conservative and test)
if pragh != nil && strings.HasPrefix(text, "go:") {
p.pragma = pragh(p.posAt(line, col+2), p.scanner.blank, text, p.pragma) // +2 to skip over // or /*
}
},
cmd/compile/internal/syntax: implement comment reporting in scanner R=go1.11 In order to collect comments in the AST and for error testing purposes, the scanner needs to not only recognize and skip comments, but also be able to report them if so desired. This change adds a mode flag to the scanner's init function which controls the scanner behavior around comments. In the common case where comments are not needed, there must be no significant overhead. Thus, comments are reported via a handler upcall rather than being returned as a _Comment token (which the parser would have to filter out with every scanner.next() call). Because the handlers for error messages, directives, and comments all look the same (they take a position and text), and because directives look like comments, and errors never start with a '/', this change simplifies the scanner's init call to only take one (error) handler instead of 2 or 3 different handlers with identical signature. It is trivial in the handler to determine if we have an error, directive, or general comment. Finally, because directives are comments, when reporting directives the full comment text is returned now rather than just the directive text. This simplifies the implementation and makes the scanner API more regular. Furthermore, it provides important information about the comment style used by a directive, which may matter eventually when we fully implement /*line file:line:col*/ directives. Change-Id: I2adbfcebecd615e4237ed3a832b6ceb9518bf09c Reviewed-on: https://go-review.googlesource.com/88215 Reviewed-by: Matthew Dempsky <mdempsky@google.com>
2018-01-17 21:42:51 -08:00
directives,
)
p.base = file
p.first = nil
p.errcnt = 0
p.pragma = nil
p.fnest = 0
p.xnest = 0
p.indent = nil
}
// takePragma returns the current parsed pragmas
// and clears them from the parser state.
func (p *parser) takePragma() Pragma {
prag := p.pragma
p.pragma = nil
return prag
}
// clearPragma is called at the end of a statement or
// other Go form that does NOT accept a pragma.
// It sends the pragma back to the pragma handler
// to be reported as unused.
func (p *parser) clearPragma() {
if p.pragma != nil {
p.pragh(p.pos(), p.scanner.blank, "", p.pragma)
p.pragma = nil
}
}
// updateBase sets the current position base to a new line base at pos.
// The base's filename, line, and column values are extracted from text
// which is positioned at (tline, tcol) (only needed for error messages).
func (p *parser) updateBase(pos Pos, tline, tcol uint, text string) {
i, n, ok := trailingDigits(text)
if i == 0 {
return // ignore (not a line directive)
}
// i > 0
if !ok {
// text has a suffix :xxx but xxx is not a number
p.errorAt(p.posAt(tline, tcol+i), "invalid line number: "+text[i:])
return
}
var line, col uint
i2, n2, ok2 := trailingDigits(text[:i-1])
if ok2 {
//line filename:line:col
i, i2 = i2, i
line, col = n2, n
if col == 0 || col > PosMax {
p.errorAt(p.posAt(tline, tcol+i2), "invalid column number: "+text[i2:])
return
}
text = text[:i2-1] // lop off ":col"
cmd/compile/internal/syntax: remove dependency on cmd/internal/src For dependency reasons, the data structure implementing source positions in the compiler is in cmd/internal/src. It contains highly compiler specific details (e.g. inlining index). This change introduces a parallel but simpler position representation, defined in the syntax package, which removes that package's dependency on cmd/internal/src, and also removes the need to deal with certain filename-specific operations (defined by the needs of the compiler) in the syntax package. As a result, the syntax package becomes again a compiler- independent, stand-alone package that at some point might replace (or augment) the existing top-level go/* syntax-related packages. Additionally, line directives that update column numbers are now correctly tracked through the syntax package, with additional tests added. (The respective changes also need to be made in cmd/internal/src; i.e., the compiler accepts but still ignores column numbers in line directives.) This change comes at the cost of a new position translation step, but that step is cheap because it only needs to do real work if the position base changed (i.e., if there is a new file, or new line directive). There is no noticeable impact on overall compiler performance measured with `compilebench -count 5 -alloc`: name old time/op new time/op delta Template 220ms ± 8% 228ms ±18% ~ (p=0.548 n=5+5) Unicode 119ms ±11% 113ms ± 5% ~ (p=0.056 n=5+5) GoTypes 684ms ± 6% 677ms ± 3% ~ (p=0.841 n=5+5) Compiler 3.19s ± 7% 3.01s ± 1% ~ (p=0.095 n=5+5) SSA 7.92s ± 8% 7.79s ± 1% ~ (p=0.690 n=5+5) Flate 141ms ± 7% 139ms ± 4% ~ (p=0.548 n=5+5) GoParser 173ms ±12% 171ms ± 4% ~ (p=1.000 n=5+5) Reflect 417ms ± 5% 411ms ± 3% ~ (p=0.548 n=5+5) Tar 205ms ± 5% 198ms ± 2% ~ (p=0.690 n=5+5) XML 232ms ± 4% 229ms ± 4% ~ (p=0.690 n=5+5) StdCmd 28.7s ± 5% 28.2s ± 2% ~ (p=0.421 n=5+5) name old user-time/op new user-time/op delta Template 269ms ± 4% 265ms ± 5% ~ (p=0.421 n=5+5) Unicode 153ms ± 7% 149ms ± 3% ~ (p=0.841 n=5+5) GoTypes 850ms ± 7% 862ms ± 4% ~ (p=0.841 n=5+5) Compiler 4.01s ± 5% 3.86s ± 0% ~ (p=0.190 n=5+4) SSA 10.9s ± 4% 10.8s ± 2% ~ (p=0.548 n=5+5) Flate 166ms ± 7% 167ms ± 6% ~ (p=1.000 n=5+5) GoParser 204ms ± 8% 206ms ± 7% ~ (p=0.841 n=5+5) Reflect 514ms ± 5% 508ms ± 4% ~ (p=0.548 n=5+5) Tar 245ms ± 6% 244ms ± 3% ~ (p=0.690 n=5+5) XML 280ms ± 4% 278ms ± 4% ~ (p=0.841 n=5+5) name old alloc/op new alloc/op delta Template 37.9MB ± 0% 37.9MB ± 0% ~ (p=0.841 n=5+5) Unicode 28.8MB ± 0% 28.8MB ± 0% ~ (p=0.841 n=5+5) GoTypes 113MB ± 0% 113MB ± 0% ~ (p=0.151 n=5+5) Compiler 468MB ± 0% 468MB ± 0% -0.01% (p=0.032 n=5+5) SSA 1.50GB ± 0% 1.50GB ± 0% ~ (p=0.548 n=5+5) Flate 24.4MB ± 0% 24.4MB ± 0% ~ (p=1.000 n=5+5) GoParser 30.7MB ± 0% 30.7MB ± 0% ~ (p=1.000 n=5+5) Reflect 76.5MB ± 0% 76.5MB ± 0% ~ (p=0.548 n=5+5) Tar 38.9MB ± 0% 38.9MB ± 0% ~ (p=0.222 n=5+5) XML 41.6MB ± 0% 41.6MB ± 0% ~ (p=0.548 n=5+5) name old allocs/op new allocs/op delta Template 382k ± 0% 382k ± 0% +0.01% (p=0.008 n=5+5) Unicode 343k ± 0% 343k ± 0% ~ (p=0.841 n=5+5) GoTypes 1.19M ± 0% 1.19M ± 0% +0.01% (p=0.008 n=5+5) Compiler 4.53M ± 0% 4.53M ± 0% +0.03% (p=0.008 n=5+5) SSA 12.4M ± 0% 12.4M ± 0% +0.00% (p=0.008 n=5+5) Flate 235k ± 0% 235k ± 0% ~ (p=0.079 n=5+5) GoParser 318k ± 0% 318k ± 0% ~ (p=0.730 n=5+5) Reflect 978k ± 0% 978k ± 0% ~ (p=1.000 n=5+5) Tar 393k ± 0% 393k ± 0% ~ (p=0.056 n=5+5) XML 405k ± 0% 405k ± 0% ~ (p=0.548 n=5+5) name old text-bytes new text-bytes delta HelloSize 672kB ± 0% 672kB ± 0% ~ (all equal) CmdGoSize 7.12MB ± 0% 7.12MB ± 0% ~ (all equal) name old data-bytes new data-bytes delta HelloSize 133kB ± 0% 133kB ± 0% ~ (all equal) CmdGoSize 390kB ± 0% 390kB ± 0% ~ (all equal) name old exe-bytes new exe-bytes delta HelloSize 1.07MB ± 0% 1.07MB ± 0% ~ (all equal) CmdGoSize 11.2MB ± 0% 11.2MB ± 0% ~ (all equal) Passes toolstash compare. For #22662. Change-Id: I19edb53dd9675af57f7122cb7dba2a6d8bdcc3da Reviewed-on: https://go-review.googlesource.com/94515 Reviewed-by: Matthew Dempsky <mdempsky@google.com>
2018-01-02 16:58:37 -08:00
} else {
//line filename:line
line = n
}
if line == 0 || line > PosMax {
p.errorAt(p.posAt(tline, tcol+i), "invalid line number: "+text[i:])
return
}
// If we have a column (//line filename:line:col form),
// an empty filename means to use the previous filename.
filename := text[:i-1] // lop off ":line"
trimmed := false
if filename == "" && ok2 {
filename = p.base.Filename()
trimmed = p.base.Trimmed()
}
p.base = NewLineBase(pos, filename, trimmed, line, col)
}
cmd/compile/internal/syntax: implement comment reporting in scanner R=go1.11 In order to collect comments in the AST and for error testing purposes, the scanner needs to not only recognize and skip comments, but also be able to report them if so desired. This change adds a mode flag to the scanner's init function which controls the scanner behavior around comments. In the common case where comments are not needed, there must be no significant overhead. Thus, comments are reported via a handler upcall rather than being returned as a _Comment token (which the parser would have to filter out with every scanner.next() call). Because the handlers for error messages, directives, and comments all look the same (they take a position and text), and because directives look like comments, and errors never start with a '/', this change simplifies the scanner's init call to only take one (error) handler instead of 2 or 3 different handlers with identical signature. It is trivial in the handler to determine if we have an error, directive, or general comment. Finally, because directives are comments, when reporting directives the full comment text is returned now rather than just the directive text. This simplifies the implementation and makes the scanner API more regular. Furthermore, it provides important information about the comment style used by a directive, which may matter eventually when we fully implement /*line file:line:col*/ directives. Change-Id: I2adbfcebecd615e4237ed3a832b6ceb9518bf09c Reviewed-on: https://go-review.googlesource.com/88215 Reviewed-by: Matthew Dempsky <mdempsky@google.com>
2018-01-17 21:42:51 -08:00
func commentText(s string) string {
if s[:2] == "/*" {
return s[2 : len(s)-2] // lop off /* and */
}
// line comment (does not include newline)
// (on Windows, the line comment may end in \r\n)
i := len(s)
if s[i-1] == '\r' {
i--
}
return s[2:i] // lop off //, and \r at end, if any
cmd/compile/internal/syntax: implement comment reporting in scanner R=go1.11 In order to collect comments in the AST and for error testing purposes, the scanner needs to not only recognize and skip comments, but also be able to report them if so desired. This change adds a mode flag to the scanner's init function which controls the scanner behavior around comments. In the common case where comments are not needed, there must be no significant overhead. Thus, comments are reported via a handler upcall rather than being returned as a _Comment token (which the parser would have to filter out with every scanner.next() call). Because the handlers for error messages, directives, and comments all look the same (they take a position and text), and because directives look like comments, and errors never start with a '/', this change simplifies the scanner's init call to only take one (error) handler instead of 2 or 3 different handlers with identical signature. It is trivial in the handler to determine if we have an error, directive, or general comment. Finally, because directives are comments, when reporting directives the full comment text is returned now rather than just the directive text. This simplifies the implementation and makes the scanner API more regular. Furthermore, it provides important information about the comment style used by a directive, which may matter eventually when we fully implement /*line file:line:col*/ directives. Change-Id: I2adbfcebecd615e4237ed3a832b6ceb9518bf09c Reviewed-on: https://go-review.googlesource.com/88215 Reviewed-by: Matthew Dempsky <mdempsky@google.com>
2018-01-17 21:42:51 -08:00
}
func trailingDigits(text string) (uint, uint, bool) {
// Want to use LastIndexByte below but it's not defined in Go1.4 and bootstrap fails.
i := strings.LastIndex(text, ":") // look from right (Windows filenames may contain ':')
if i < 0 {
return 0, 0, false // no ":"
}
// i >= 0
n, err := strconv.ParseUint(text[i+1:], 10, 0)
return uint(i + 1), uint(n), err == nil
}
func (p *parser) got(tok token) bool {
if p.tok == tok {
p.next()
return true
}
return false
}
func (p *parser) want(tok token) {
if !p.got(tok) {
p.syntaxError("expecting " + tokstring(tok))
p.advance()
}
}
// gotAssign is like got(_Assign) but it also accepts ":="
// (and reports an error) for better parser error recovery.
func (p *parser) gotAssign() bool {
switch p.tok {
case _Define:
p.syntaxError("expecting =")
fallthrough
case _Assign:
p.next()
return true
}
return false
}
// ----------------------------------------------------------------------------
// Error handling
// posAt returns the Pos value for (line, col) and the current position base.
cmd/compile/internal/syntax: remove dependency on cmd/internal/src For dependency reasons, the data structure implementing source positions in the compiler is in cmd/internal/src. It contains highly compiler specific details (e.g. inlining index). This change introduces a parallel but simpler position representation, defined in the syntax package, which removes that package's dependency on cmd/internal/src, and also removes the need to deal with certain filename-specific operations (defined by the needs of the compiler) in the syntax package. As a result, the syntax package becomes again a compiler- independent, stand-alone package that at some point might replace (or augment) the existing top-level go/* syntax-related packages. Additionally, line directives that update column numbers are now correctly tracked through the syntax package, with additional tests added. (The respective changes also need to be made in cmd/internal/src; i.e., the compiler accepts but still ignores column numbers in line directives.) This change comes at the cost of a new position translation step, but that step is cheap because it only needs to do real work if the position base changed (i.e., if there is a new file, or new line directive). There is no noticeable impact on overall compiler performance measured with `compilebench -count 5 -alloc`: name old time/op new time/op delta Template 220ms ± 8% 228ms ±18% ~ (p=0.548 n=5+5) Unicode 119ms ±11% 113ms ± 5% ~ (p=0.056 n=5+5) GoTypes 684ms ± 6% 677ms ± 3% ~ (p=0.841 n=5+5) Compiler 3.19s ± 7% 3.01s ± 1% ~ (p=0.095 n=5+5) SSA 7.92s ± 8% 7.79s ± 1% ~ (p=0.690 n=5+5) Flate 141ms ± 7% 139ms ± 4% ~ (p=0.548 n=5+5) GoParser 173ms ±12% 171ms ± 4% ~ (p=1.000 n=5+5) Reflect 417ms ± 5% 411ms ± 3% ~ (p=0.548 n=5+5) Tar 205ms ± 5% 198ms ± 2% ~ (p=0.690 n=5+5) XML 232ms ± 4% 229ms ± 4% ~ (p=0.690 n=5+5) StdCmd 28.7s ± 5% 28.2s ± 2% ~ (p=0.421 n=5+5) name old user-time/op new user-time/op delta Template 269ms ± 4% 265ms ± 5% ~ (p=0.421 n=5+5) Unicode 153ms ± 7% 149ms ± 3% ~ (p=0.841 n=5+5) GoTypes 850ms ± 7% 862ms ± 4% ~ (p=0.841 n=5+5) Compiler 4.01s ± 5% 3.86s ± 0% ~ (p=0.190 n=5+4) SSA 10.9s ± 4% 10.8s ± 2% ~ (p=0.548 n=5+5) Flate 166ms ± 7% 167ms ± 6% ~ (p=1.000 n=5+5) GoParser 204ms ± 8% 206ms ± 7% ~ (p=0.841 n=5+5) Reflect 514ms ± 5% 508ms ± 4% ~ (p=0.548 n=5+5) Tar 245ms ± 6% 244ms ± 3% ~ (p=0.690 n=5+5) XML 280ms ± 4% 278ms ± 4% ~ (p=0.841 n=5+5) name old alloc/op new alloc/op delta Template 37.9MB ± 0% 37.9MB ± 0% ~ (p=0.841 n=5+5) Unicode 28.8MB ± 0% 28.8MB ± 0% ~ (p=0.841 n=5+5) GoTypes 113MB ± 0% 113MB ± 0% ~ (p=0.151 n=5+5) Compiler 468MB ± 0% 468MB ± 0% -0.01% (p=0.032 n=5+5) SSA 1.50GB ± 0% 1.50GB ± 0% ~ (p=0.548 n=5+5) Flate 24.4MB ± 0% 24.4MB ± 0% ~ (p=1.000 n=5+5) GoParser 30.7MB ± 0% 30.7MB ± 0% ~ (p=1.000 n=5+5) Reflect 76.5MB ± 0% 76.5MB ± 0% ~ (p=0.548 n=5+5) Tar 38.9MB ± 0% 38.9MB ± 0% ~ (p=0.222 n=5+5) XML 41.6MB ± 0% 41.6MB ± 0% ~ (p=0.548 n=5+5) name old allocs/op new allocs/op delta Template 382k ± 0% 382k ± 0% +0.01% (p=0.008 n=5+5) Unicode 343k ± 0% 343k ± 0% ~ (p=0.841 n=5+5) GoTypes 1.19M ± 0% 1.19M ± 0% +0.01% (p=0.008 n=5+5) Compiler 4.53M ± 0% 4.53M ± 0% +0.03% (p=0.008 n=5+5) SSA 12.4M ± 0% 12.4M ± 0% +0.00% (p=0.008 n=5+5) Flate 235k ± 0% 235k ± 0% ~ (p=0.079 n=5+5) GoParser 318k ± 0% 318k ± 0% ~ (p=0.730 n=5+5) Reflect 978k ± 0% 978k ± 0% ~ (p=1.000 n=5+5) Tar 393k ± 0% 393k ± 0% ~ (p=0.056 n=5+5) XML 405k ± 0% 405k ± 0% ~ (p=0.548 n=5+5) name old text-bytes new text-bytes delta HelloSize 672kB ± 0% 672kB ± 0% ~ (all equal) CmdGoSize 7.12MB ± 0% 7.12MB ± 0% ~ (all equal) name old data-bytes new data-bytes delta HelloSize 133kB ± 0% 133kB ± 0% ~ (all equal) CmdGoSize 390kB ± 0% 390kB ± 0% ~ (all equal) name old exe-bytes new exe-bytes delta HelloSize 1.07MB ± 0% 1.07MB ± 0% ~ (all equal) CmdGoSize 11.2MB ± 0% 11.2MB ± 0% ~ (all equal) Passes toolstash compare. For #22662. Change-Id: I19edb53dd9675af57f7122cb7dba2a6d8bdcc3da Reviewed-on: https://go-review.googlesource.com/94515 Reviewed-by: Matthew Dempsky <mdempsky@google.com>
2018-01-02 16:58:37 -08:00
func (p *parser) posAt(line, col uint) Pos {
return MakePos(p.base, line, col)
}
// error reports an error at the given position.
cmd/compile/internal/syntax: remove dependency on cmd/internal/src For dependency reasons, the data structure implementing source positions in the compiler is in cmd/internal/src. It contains highly compiler specific details (e.g. inlining index). This change introduces a parallel but simpler position representation, defined in the syntax package, which removes that package's dependency on cmd/internal/src, and also removes the need to deal with certain filename-specific operations (defined by the needs of the compiler) in the syntax package. As a result, the syntax package becomes again a compiler- independent, stand-alone package that at some point might replace (or augment) the existing top-level go/* syntax-related packages. Additionally, line directives that update column numbers are now correctly tracked through the syntax package, with additional tests added. (The respective changes also need to be made in cmd/internal/src; i.e., the compiler accepts but still ignores column numbers in line directives.) This change comes at the cost of a new position translation step, but that step is cheap because it only needs to do real work if the position base changed (i.e., if there is a new file, or new line directive). There is no noticeable impact on overall compiler performance measured with `compilebench -count 5 -alloc`: name old time/op new time/op delta Template 220ms ± 8% 228ms ±18% ~ (p=0.548 n=5+5) Unicode 119ms ±11% 113ms ± 5% ~ (p=0.056 n=5+5) GoTypes 684ms ± 6% 677ms ± 3% ~ (p=0.841 n=5+5) Compiler 3.19s ± 7% 3.01s ± 1% ~ (p=0.095 n=5+5) SSA 7.92s ± 8% 7.79s ± 1% ~ (p=0.690 n=5+5) Flate 141ms ± 7% 139ms ± 4% ~ (p=0.548 n=5+5) GoParser 173ms ±12% 171ms ± 4% ~ (p=1.000 n=5+5) Reflect 417ms ± 5% 411ms ± 3% ~ (p=0.548 n=5+5) Tar 205ms ± 5% 198ms ± 2% ~ (p=0.690 n=5+5) XML 232ms ± 4% 229ms ± 4% ~ (p=0.690 n=5+5) StdCmd 28.7s ± 5% 28.2s ± 2% ~ (p=0.421 n=5+5) name old user-time/op new user-time/op delta Template 269ms ± 4% 265ms ± 5% ~ (p=0.421 n=5+5) Unicode 153ms ± 7% 149ms ± 3% ~ (p=0.841 n=5+5) GoTypes 850ms ± 7% 862ms ± 4% ~ (p=0.841 n=5+5) Compiler 4.01s ± 5% 3.86s ± 0% ~ (p=0.190 n=5+4) SSA 10.9s ± 4% 10.8s ± 2% ~ (p=0.548 n=5+5) Flate 166ms ± 7% 167ms ± 6% ~ (p=1.000 n=5+5) GoParser 204ms ± 8% 206ms ± 7% ~ (p=0.841 n=5+5) Reflect 514ms ± 5% 508ms ± 4% ~ (p=0.548 n=5+5) Tar 245ms ± 6% 244ms ± 3% ~ (p=0.690 n=5+5) XML 280ms ± 4% 278ms ± 4% ~ (p=0.841 n=5+5) name old alloc/op new alloc/op delta Template 37.9MB ± 0% 37.9MB ± 0% ~ (p=0.841 n=5+5) Unicode 28.8MB ± 0% 28.8MB ± 0% ~ (p=0.841 n=5+5) GoTypes 113MB ± 0% 113MB ± 0% ~ (p=0.151 n=5+5) Compiler 468MB ± 0% 468MB ± 0% -0.01% (p=0.032 n=5+5) SSA 1.50GB ± 0% 1.50GB ± 0% ~ (p=0.548 n=5+5) Flate 24.4MB ± 0% 24.4MB ± 0% ~ (p=1.000 n=5+5) GoParser 30.7MB ± 0% 30.7MB ± 0% ~ (p=1.000 n=5+5) Reflect 76.5MB ± 0% 76.5MB ± 0% ~ (p=0.548 n=5+5) Tar 38.9MB ± 0% 38.9MB ± 0% ~ (p=0.222 n=5+5) XML 41.6MB ± 0% 41.6MB ± 0% ~ (p=0.548 n=5+5) name old allocs/op new allocs/op delta Template 382k ± 0% 382k ± 0% +0.01% (p=0.008 n=5+5) Unicode 343k ± 0% 343k ± 0% ~ (p=0.841 n=5+5) GoTypes 1.19M ± 0% 1.19M ± 0% +0.01% (p=0.008 n=5+5) Compiler 4.53M ± 0% 4.53M ± 0% +0.03% (p=0.008 n=5+5) SSA 12.4M ± 0% 12.4M ± 0% +0.00% (p=0.008 n=5+5) Flate 235k ± 0% 235k ± 0% ~ (p=0.079 n=5+5) GoParser 318k ± 0% 318k ± 0% ~ (p=0.730 n=5+5) Reflect 978k ± 0% 978k ± 0% ~ (p=1.000 n=5+5) Tar 393k ± 0% 393k ± 0% ~ (p=0.056 n=5+5) XML 405k ± 0% 405k ± 0% ~ (p=0.548 n=5+5) name old text-bytes new text-bytes delta HelloSize 672kB ± 0% 672kB ± 0% ~ (all equal) CmdGoSize 7.12MB ± 0% 7.12MB ± 0% ~ (all equal) name old data-bytes new data-bytes delta HelloSize 133kB ± 0% 133kB ± 0% ~ (all equal) CmdGoSize 390kB ± 0% 390kB ± 0% ~ (all equal) name old exe-bytes new exe-bytes delta HelloSize 1.07MB ± 0% 1.07MB ± 0% ~ (all equal) CmdGoSize 11.2MB ± 0% 11.2MB ± 0% ~ (all equal) Passes toolstash compare. For #22662. Change-Id: I19edb53dd9675af57f7122cb7dba2a6d8bdcc3da Reviewed-on: https://go-review.googlesource.com/94515 Reviewed-by: Matthew Dempsky <mdempsky@google.com>
2018-01-02 16:58:37 -08:00
func (p *parser) errorAt(pos Pos, msg string) {
err := Error{pos, msg}
if p.first == nil {
p.first = err
}
p.errcnt++
if p.errh == nil {
panic(p.first)
}
p.errh(err)
}
// syntaxErrorAt reports a syntax error at the given position.
cmd/compile/internal/syntax: remove dependency on cmd/internal/src For dependency reasons, the data structure implementing source positions in the compiler is in cmd/internal/src. It contains highly compiler specific details (e.g. inlining index). This change introduces a parallel but simpler position representation, defined in the syntax package, which removes that package's dependency on cmd/internal/src, and also removes the need to deal with certain filename-specific operations (defined by the needs of the compiler) in the syntax package. As a result, the syntax package becomes again a compiler- independent, stand-alone package that at some point might replace (or augment) the existing top-level go/* syntax-related packages. Additionally, line directives that update column numbers are now correctly tracked through the syntax package, with additional tests added. (The respective changes also need to be made in cmd/internal/src; i.e., the compiler accepts but still ignores column numbers in line directives.) This change comes at the cost of a new position translation step, but that step is cheap because it only needs to do real work if the position base changed (i.e., if there is a new file, or new line directive). There is no noticeable impact on overall compiler performance measured with `compilebench -count 5 -alloc`: name old time/op new time/op delta Template 220ms ± 8% 228ms ±18% ~ (p=0.548 n=5+5) Unicode 119ms ±11% 113ms ± 5% ~ (p=0.056 n=5+5) GoTypes 684ms ± 6% 677ms ± 3% ~ (p=0.841 n=5+5) Compiler 3.19s ± 7% 3.01s ± 1% ~ (p=0.095 n=5+5) SSA 7.92s ± 8% 7.79s ± 1% ~ (p=0.690 n=5+5) Flate 141ms ± 7% 139ms ± 4% ~ (p=0.548 n=5+5) GoParser 173ms ±12% 171ms ± 4% ~ (p=1.000 n=5+5) Reflect 417ms ± 5% 411ms ± 3% ~ (p=0.548 n=5+5) Tar 205ms ± 5% 198ms ± 2% ~ (p=0.690 n=5+5) XML 232ms ± 4% 229ms ± 4% ~ (p=0.690 n=5+5) StdCmd 28.7s ± 5% 28.2s ± 2% ~ (p=0.421 n=5+5) name old user-time/op new user-time/op delta Template 269ms ± 4% 265ms ± 5% ~ (p=0.421 n=5+5) Unicode 153ms ± 7% 149ms ± 3% ~ (p=0.841 n=5+5) GoTypes 850ms ± 7% 862ms ± 4% ~ (p=0.841 n=5+5) Compiler 4.01s ± 5% 3.86s ± 0% ~ (p=0.190 n=5+4) SSA 10.9s ± 4% 10.8s ± 2% ~ (p=0.548 n=5+5) Flate 166ms ± 7% 167ms ± 6% ~ (p=1.000 n=5+5) GoParser 204ms ± 8% 206ms ± 7% ~ (p=0.841 n=5+5) Reflect 514ms ± 5% 508ms ± 4% ~ (p=0.548 n=5+5) Tar 245ms ± 6% 244ms ± 3% ~ (p=0.690 n=5+5) XML 280ms ± 4% 278ms ± 4% ~ (p=0.841 n=5+5) name old alloc/op new alloc/op delta Template 37.9MB ± 0% 37.9MB ± 0% ~ (p=0.841 n=5+5) Unicode 28.8MB ± 0% 28.8MB ± 0% ~ (p=0.841 n=5+5) GoTypes 113MB ± 0% 113MB ± 0% ~ (p=0.151 n=5+5) Compiler 468MB ± 0% 468MB ± 0% -0.01% (p=0.032 n=5+5) SSA 1.50GB ± 0% 1.50GB ± 0% ~ (p=0.548 n=5+5) Flate 24.4MB ± 0% 24.4MB ± 0% ~ (p=1.000 n=5+5) GoParser 30.7MB ± 0% 30.7MB ± 0% ~ (p=1.000 n=5+5) Reflect 76.5MB ± 0% 76.5MB ± 0% ~ (p=0.548 n=5+5) Tar 38.9MB ± 0% 38.9MB ± 0% ~ (p=0.222 n=5+5) XML 41.6MB ± 0% 41.6MB ± 0% ~ (p=0.548 n=5+5) name old allocs/op new allocs/op delta Template 382k ± 0% 382k ± 0% +0.01% (p=0.008 n=5+5) Unicode 343k ± 0% 343k ± 0% ~ (p=0.841 n=5+5) GoTypes 1.19M ± 0% 1.19M ± 0% +0.01% (p=0.008 n=5+5) Compiler 4.53M ± 0% 4.53M ± 0% +0.03% (p=0.008 n=5+5) SSA 12.4M ± 0% 12.4M ± 0% +0.00% (p=0.008 n=5+5) Flate 235k ± 0% 235k ± 0% ~ (p=0.079 n=5+5) GoParser 318k ± 0% 318k ± 0% ~ (p=0.730 n=5+5) Reflect 978k ± 0% 978k ± 0% ~ (p=1.000 n=5+5) Tar 393k ± 0% 393k ± 0% ~ (p=0.056 n=5+5) XML 405k ± 0% 405k ± 0% ~ (p=0.548 n=5+5) name old text-bytes new text-bytes delta HelloSize 672kB ± 0% 672kB ± 0% ~ (all equal) CmdGoSize 7.12MB ± 0% 7.12MB ± 0% ~ (all equal) name old data-bytes new data-bytes delta HelloSize 133kB ± 0% 133kB ± 0% ~ (all equal) CmdGoSize 390kB ± 0% 390kB ± 0% ~ (all equal) name old exe-bytes new exe-bytes delta HelloSize 1.07MB ± 0% 1.07MB ± 0% ~ (all equal) CmdGoSize 11.2MB ± 0% 11.2MB ± 0% ~ (all equal) Passes toolstash compare. For #22662. Change-Id: I19edb53dd9675af57f7122cb7dba2a6d8bdcc3da Reviewed-on: https://go-review.googlesource.com/94515 Reviewed-by: Matthew Dempsky <mdempsky@google.com>
2018-01-02 16:58:37 -08:00
func (p *parser) syntaxErrorAt(pos Pos, msg string) {
if trace {
p.print("syntax error: " + msg)
}
if p.tok == _EOF && p.first != nil {
return // avoid meaningless follow-up errors
}
// add punctuation etc. as needed to msg
switch {
case msg == "":
// nothing to do
case strings.HasPrefix(msg, "in "), strings.HasPrefix(msg, "at "), strings.HasPrefix(msg, "after "):
msg = " " + msg
case strings.HasPrefix(msg, "expecting "):
msg = ", " + msg
default:
// plain error - we don't care about current token
p.errorAt(pos, "syntax error: "+msg)
return
}
// determine token string
var tok string
switch p.tok {
case _Name, _Semi:
tok = p.lit
case _Literal:
tok = "literal " + p.lit
case _Operator:
tok = p.op.String()
case _AssignOp:
tok = p.op.String() + "="
case _IncOp:
tok = p.op.String()
tok += tok
default:
tok = tokstring(p.tok)
}
p.errorAt(pos, "syntax error: unexpected "+tok+msg)
}
// tokstring returns the English word for selected punctuation tokens
// for more readable error messages. Use tokstring (not tok.String())
// for user-facing (error) messages; use tok.String() for debugging
// output.
func tokstring(tok token) string {
switch tok {
case _Comma:
return "comma"
case _Semi:
return "semicolon or newline"
}
return tok.String()
}
// Convenience methods using the current token position.
cmd/compile/internal/syntax: remove dependency on cmd/internal/src For dependency reasons, the data structure implementing source positions in the compiler is in cmd/internal/src. It contains highly compiler specific details (e.g. inlining index). This change introduces a parallel but simpler position representation, defined in the syntax package, which removes that package's dependency on cmd/internal/src, and also removes the need to deal with certain filename-specific operations (defined by the needs of the compiler) in the syntax package. As a result, the syntax package becomes again a compiler- independent, stand-alone package that at some point might replace (or augment) the existing top-level go/* syntax-related packages. Additionally, line directives that update column numbers are now correctly tracked through the syntax package, with additional tests added. (The respective changes also need to be made in cmd/internal/src; i.e., the compiler accepts but still ignores column numbers in line directives.) This change comes at the cost of a new position translation step, but that step is cheap because it only needs to do real work if the position base changed (i.e., if there is a new file, or new line directive). There is no noticeable impact on overall compiler performance measured with `compilebench -count 5 -alloc`: name old time/op new time/op delta Template 220ms ± 8% 228ms ±18% ~ (p=0.548 n=5+5) Unicode 119ms ±11% 113ms ± 5% ~ (p=0.056 n=5+5) GoTypes 684ms ± 6% 677ms ± 3% ~ (p=0.841 n=5+5) Compiler 3.19s ± 7% 3.01s ± 1% ~ (p=0.095 n=5+5) SSA 7.92s ± 8% 7.79s ± 1% ~ (p=0.690 n=5+5) Flate 141ms ± 7% 139ms ± 4% ~ (p=0.548 n=5+5) GoParser 173ms ±12% 171ms ± 4% ~ (p=1.000 n=5+5) Reflect 417ms ± 5% 411ms ± 3% ~ (p=0.548 n=5+5) Tar 205ms ± 5% 198ms ± 2% ~ (p=0.690 n=5+5) XML 232ms ± 4% 229ms ± 4% ~ (p=0.690 n=5+5) StdCmd 28.7s ± 5% 28.2s ± 2% ~ (p=0.421 n=5+5) name old user-time/op new user-time/op delta Template 269ms ± 4% 265ms ± 5% ~ (p=0.421 n=5+5) Unicode 153ms ± 7% 149ms ± 3% ~ (p=0.841 n=5+5) GoTypes 850ms ± 7% 862ms ± 4% ~ (p=0.841 n=5+5) Compiler 4.01s ± 5% 3.86s ± 0% ~ (p=0.190 n=5+4) SSA 10.9s ± 4% 10.8s ± 2% ~ (p=0.548 n=5+5) Flate 166ms ± 7% 167ms ± 6% ~ (p=1.000 n=5+5) GoParser 204ms ± 8% 206ms ± 7% ~ (p=0.841 n=5+5) Reflect 514ms ± 5% 508ms ± 4% ~ (p=0.548 n=5+5) Tar 245ms ± 6% 244ms ± 3% ~ (p=0.690 n=5+5) XML 280ms ± 4% 278ms ± 4% ~ (p=0.841 n=5+5) name old alloc/op new alloc/op delta Template 37.9MB ± 0% 37.9MB ± 0% ~ (p=0.841 n=5+5) Unicode 28.8MB ± 0% 28.8MB ± 0% ~ (p=0.841 n=5+5) GoTypes 113MB ± 0% 113MB ± 0% ~ (p=0.151 n=5+5) Compiler 468MB ± 0% 468MB ± 0% -0.01% (p=0.032 n=5+5) SSA 1.50GB ± 0% 1.50GB ± 0% ~ (p=0.548 n=5+5) Flate 24.4MB ± 0% 24.4MB ± 0% ~ (p=1.000 n=5+5) GoParser 30.7MB ± 0% 30.7MB ± 0% ~ (p=1.000 n=5+5) Reflect 76.5MB ± 0% 76.5MB ± 0% ~ (p=0.548 n=5+5) Tar 38.9MB ± 0% 38.9MB ± 0% ~ (p=0.222 n=5+5) XML 41.6MB ± 0% 41.6MB ± 0% ~ (p=0.548 n=5+5) name old allocs/op new allocs/op delta Template 382k ± 0% 382k ± 0% +0.01% (p=0.008 n=5+5) Unicode 343k ± 0% 343k ± 0% ~ (p=0.841 n=5+5) GoTypes 1.19M ± 0% 1.19M ± 0% +0.01% (p=0.008 n=5+5) Compiler 4.53M ± 0% 4.53M ± 0% +0.03% (p=0.008 n=5+5) SSA 12.4M ± 0% 12.4M ± 0% +0.00% (p=0.008 n=5+5) Flate 235k ± 0% 235k ± 0% ~ (p=0.079 n=5+5) GoParser 318k ± 0% 318k ± 0% ~ (p=0.730 n=5+5) Reflect 978k ± 0% 978k ± 0% ~ (p=1.000 n=5+5) Tar 393k ± 0% 393k ± 0% ~ (p=0.056 n=5+5) XML 405k ± 0% 405k ± 0% ~ (p=0.548 n=5+5) name old text-bytes new text-bytes delta HelloSize 672kB ± 0% 672kB ± 0% ~ (all equal) CmdGoSize 7.12MB ± 0% 7.12MB ± 0% ~ (all equal) name old data-bytes new data-bytes delta HelloSize 133kB ± 0% 133kB ± 0% ~ (all equal) CmdGoSize 390kB ± 0% 390kB ± 0% ~ (all equal) name old exe-bytes new exe-bytes delta HelloSize 1.07MB ± 0% 1.07MB ± 0% ~ (all equal) CmdGoSize 11.2MB ± 0% 11.2MB ± 0% ~ (all equal) Passes toolstash compare. For #22662. Change-Id: I19edb53dd9675af57f7122cb7dba2a6d8bdcc3da Reviewed-on: https://go-review.googlesource.com/94515 Reviewed-by: Matthew Dempsky <mdempsky@google.com>
2018-01-02 16:58:37 -08:00
func (p *parser) pos() Pos { return p.posAt(p.line, p.col) }
func (p *parser) error(msg string) { p.errorAt(p.pos(), msg) }
func (p *parser) syntaxError(msg string) { p.syntaxErrorAt(p.pos(), msg) }
// The stopset contains keywords that start a statement.
// They are good synchronization points in case of syntax
// errors and (usually) shouldn't be skipped over.
const stopset uint64 = 1<<_Break |
1<<_Const |
1<<_Continue |
1<<_Defer |
1<<_Fallthrough |
1<<_For |
1<<_Go |
1<<_Goto |
1<<_If |
1<<_Return |
1<<_Select |
1<<_Switch |
1<<_Type |
1<<_Var
// Advance consumes tokens until it finds a token of the stopset or followlist.
// The stopset is only considered if we are inside a function (p.fnest > 0).
// The followlist is the list of valid tokens that can follow a production;
// if it is empty, exactly one (non-EOF) token is consumed to ensure progress.
func (p *parser) advance(followlist ...token) {
if trace {
p.print(fmt.Sprintf("advance %s", followlist))
}
// compute follow set
// (not speed critical, advance is only called in error situations)
var followset uint64 = 1 << _EOF // don't skip over EOF
if len(followlist) > 0 {
if p.fnest > 0 {
followset |= stopset
}
for _, tok := range followlist {
followset |= 1 << tok
}
}
for !contains(followset, p.tok) {
if trace {
p.print("skip " + p.tok.String())
}
p.next()
if len(followlist) == 0 {
break
}
}
if trace {
p.print("next " + p.tok.String())
}
}
// usage: defer p.trace(msg)()
func (p *parser) trace(msg string) func() {
p.print(msg + " (")
const tab = ". "
p.indent = append(p.indent, tab...)
return func() {
p.indent = p.indent[:len(p.indent)-len(tab)]
if x := recover(); x != nil {
panic(x) // skip print_trace
}
p.print(")")
}
}
func (p *parser) print(msg string) {
fmt.Printf("%5d: %s%s\n", p.line, p.indent, msg)
}
// ----------------------------------------------------------------------------
// Package files
//
// Parse methods are annotated with matching Go productions as appropriate.
// The annotations are intended as guidelines only since a single Go grammar
// rule may be covered by multiple parse methods and vice versa.
//
// Excluding methods returning slices, parse methods named xOrNil may return
// nil; all others are expected to return a valid non-nil node.
// SourceFile = PackageClause ";" { ImportDecl ";" } { TopLevelDecl ";" } .
func (p *parser) fileOrNil() *File {
if trace {
defer p.trace("file")()
}
f := new(File)
f.pos = p.pos()
// PackageClause
if !p.got(_Package) {
p.syntaxError("package statement must be first")
return nil
}
f.Pragma = p.takePragma()
f.PkgName = p.name()
p.want(_Semi)
// don't bother continuing if package clause has errors
if p.first != nil {
return nil
}
// { ImportDecl ";" }
for p.got(_Import) {
f.DeclList = p.appendGroup(f.DeclList, p.importDecl)
p.want(_Semi)
}
// { TopLevelDecl ";" }
for p.tok != _EOF {
switch p.tok {
case _Const:
p.next()
f.DeclList = p.appendGroup(f.DeclList, p.constDecl)
case _Type:
p.next()
f.DeclList = p.appendGroup(f.DeclList, p.typeDecl)
case _Var:
p.next()
f.DeclList = p.appendGroup(f.DeclList, p.varDecl)
case _Func:
p.next()
if d := p.funcDeclOrNil(); d != nil {
f.DeclList = append(f.DeclList, d)
}
default:
if p.tok == _Lbrace && len(f.DeclList) > 0 && isEmptyFuncDecl(f.DeclList[len(f.DeclList)-1]) {
// opening { of function declaration on next line
p.syntaxError("unexpected semicolon or newline before {")
} else {
p.syntaxError("non-declaration statement outside function body")
}
p.advance(_Const, _Type, _Var, _Func)
continue
}
// Reset p.pragma BEFORE advancing to the next token (consuming ';')
// since comments before may set pragmas for the next function decl.
p.clearPragma()
if p.tok != _EOF && !p.got(_Semi) {
p.syntaxError("after top level declaration")
p.advance(_Const, _Type, _Var, _Func)
}
}
// p.tok == _EOF
p.clearPragma()
f.EOF = p.pos()
return f
}
func isEmptyFuncDecl(dcl Decl) bool {
f, ok := dcl.(*FuncDecl)
return ok && f.Body == nil
}
// ----------------------------------------------------------------------------
// Declarations
// list parses a possibly empty, sep-separated list of elements, optionally
// followed by sep, and closed by close (or EOF). sep must be one of _Comma
// or _Semi, and close must be one of _Rparen, _Rbrace, or _Rbrack.
//
// For each list element, f is called. Specifically, unless we're at close
// (or EOF), f is called at least once. After f returns true, no more list
// elements are accepted. list returns the position of the closing token.
//
// list = [ f { sep f } [sep] ] close .
func (p *parser) list(context string, sep, close token, f func() bool) Pos {
if debug && (sep != _Comma && sep != _Semi || close != _Rparen && close != _Rbrace && close != _Rbrack) {
panic("invalid sep or close argument for list")
}
done := false
for p.tok != _EOF && p.tok != close && !done {
done = f()
// sep is optional before close
if !p.got(sep) && p.tok != close {
p.syntaxError(fmt.Sprintf("in %s; possibly missing %s or %s", context, tokstring(sep), tokstring(close)))
p.advance(_Rparen, _Rbrack, _Rbrace)
if p.tok != close {
// position could be better but we had an error so we don't care
return p.pos()
}
}
}
pos := p.pos()
p.want(close)
return pos
}
// appendGroup(f) = f | "(" { f ";" } ")" . // ";" is optional before ")"
func (p *parser) appendGroup(list []Decl, f func(*Group) Decl) []Decl {
if p.tok == _Lparen {
g := new(Group)
p.clearPragma()
p.next() // must consume "(" after calling clearPragma!
p.list("grouped declaration", _Semi, _Rparen, func() bool {
if x := f(g); x != nil {
list = append(list, x)
}
return false
})
} else {
if x := f(nil); x != nil {
list = append(list, x)
}
}
return list
}
// ImportSpec = [ "." | PackageName ] ImportPath .
// ImportPath = string_lit .
func (p *parser) importDecl(group *Group) Decl {
if trace {
defer p.trace("importDecl")()
}
d := new(ImportDecl)
d.pos = p.pos()
d.Group = group
d.Pragma = p.takePragma()
switch p.tok {
case _Name:
d.LocalPkgName = p.name()
case _Dot:
d.LocalPkgName = NewName(p.pos(), ".")
p.next()
}
d.Path = p.oliteral()
if d.Path == nil {
p.syntaxError("missing import path")
p.advance(_Semi, _Rparen)
return d
}
if !d.Path.Bad && d.Path.Kind != StringLit {
p.syntaxError("import path must be a string")
d.Path.Bad = true
}
// d.Path.Bad || d.Path.Kind == StringLit
return d
}
// ConstSpec = IdentifierList [ [ Type ] "=" ExpressionList ] .
func (p *parser) constDecl(group *Group) Decl {
if trace {
defer p.trace("constDecl")()
}
d := new(ConstDecl)
d.pos = p.pos()
d.Group = group
d.Pragma = p.takePragma()
d.NameList = p.nameList(p.name())
if p.tok != _EOF && p.tok != _Semi && p.tok != _Rparen {
d.Type = p.typeOrNil()
if p.gotAssign() {
d.Values = p.exprList()
}
}
return d
}
// TypeSpec = identifier [ TypeParams ] [ "=" ] Type .
func (p *parser) typeDecl(group *Group) Decl {
if trace {
defer p.trace("typeDecl")()
}
d := new(TypeDecl)
d.pos = p.pos()
d.Group = group
d.Pragma = p.takePragma()
d.Name = p.name()
if p.tok == _Lbrack {
// d.Name "[" ...
// array/slice type or type parameter list
pos := p.pos()
p.next()
switch p.tok {
case _Name:
// We may have an array type or a type parameter list.
// In either case we expect an expression x (which may
// just be a name, or a more complex expression) which
// we can analyze further.
//
// A type parameter list may have a type bound starting
// with a "[" as in: P []E. In that case, simply parsing
// an expression would lead to an error: P[] is invalid.
// But since index or slice expressions are never constant
// and thus invalid array length expressions, if we see a
// "[" following a name it must be the start of an array
// or slice constraint. Only if we don't see a "[" do we
// need to parse a full expression.
var x Expr = p.name()
if p.tok != _Lbrack {
// To parse the expression starting with name, expand
// the call sequence we would get by passing in name
// to parser.expr, and pass in name to parser.pexpr.
p.xnest++
x = p.binaryExpr(p.pexpr(x, false), 0)
p.xnest--
}
// analyze the cases
var pname *Name // pname != nil means pname is the type parameter name
var ptype Expr // ptype != nil means ptype is the type parameter type; pname != nil in this case
switch t := x.(type) {
case *Name:
// Unless we see a "]", we are at the start of a type parameter list.
if p.tok != _Rbrack {
// d.Name "[" name ...
pname = t
// no ptype
}
case *Operation:
// If we have an expression of the form name*T, and T is a (possibly
// parenthesized) type literal or the next token is a comma, we are
// at the start of a type parameter list.
if name, _ := t.X.(*Name); name != nil {
if t.Op == Mul && (isTypeLit(t.Y) || p.tok == _Comma) {
// d.Name "[" name "*" t.Y
// d.Name "[" name "*" t.Y ","
t.X, t.Y = t.Y, nil // convert t into unary *t.Y
pname = name
ptype = t
}
}
case *CallExpr:
// If we have an expression of the form name(T), and T is a (possibly
// parenthesized) type literal or the next token is a comma, we are
// at the start of a type parameter list.
if name, _ := t.Fun.(*Name); name != nil {
if len(t.ArgList) == 1 && !t.HasDots && (isTypeLit(t.ArgList[0]) || p.tok == _Comma) {
// d.Name "[" name "(" t.ArgList[0] ")"
// d.Name "[" name "(" t.ArgList[0] ")" ","
pname = name
ptype = t.ArgList[0]
}
}
}
if pname != nil {
// d.Name "[" pname ...
// d.Name "[" pname ptype ...
// d.Name "[" pname ptype "," ...
d.TParamList = p.paramList(pname, ptype, _Rbrack, true)
d.Alias = p.gotAssign()
d.Type = p.typeOrNil()
} else {
// d.Name "[" x ...
d.Type = p.arrayType(pos, x)
}
case _Rbrack:
// d.Name "[" "]" ...
p.next()
d.Type = p.sliceType(pos)
default:
// d.Name "[" ...
d.Type = p.arrayType(pos, nil)
}
} else {
d.Alias = p.gotAssign()
d.Type = p.typeOrNil()
}
if d.Type == nil {
d.Type = p.badExpr()
p.syntaxError("in type declaration")
p.advance(_Semi, _Rparen)
}
return d
}
// isTypeLit reports whether x is a (possibly parenthesized) type literal.
func isTypeLit(x Expr) bool {
switch x := x.(type) {
case *ArrayType, *StructType, *FuncType, *InterfaceType, *SliceType, *MapType, *ChanType:
return true
case *Operation:
// *T may be a pointer dereferenciation.
// Only consider *T as type literal if T is a type literal.
return x.Op == Mul && x.Y == nil && isTypeLit(x.X)
case *ParenExpr:
return isTypeLit(x.X)
}
return false
}
// VarSpec = IdentifierList ( Type [ "=" ExpressionList ] | "=" ExpressionList ) .
func (p *parser) varDecl(group *Group) Decl {
if trace {
defer p.trace("varDecl")()
}
d := new(VarDecl)
d.pos = p.pos()
d.Group = group
d.Pragma = p.takePragma()
d.NameList = p.nameList(p.name())
if p.gotAssign() {
d.Values = p.exprList()
} else {
d.Type = p.type_()
if p.gotAssign() {
d.Values = p.exprList()
}
}
return d
}
// FunctionDecl = "func" FunctionName [ TypeParams ] ( Function | Signature ) .
// FunctionName = identifier .
// Function = Signature FunctionBody .
// MethodDecl = "func" Receiver MethodName ( Function | Signature ) .
// Receiver = Parameters .
func (p *parser) funcDeclOrNil() *FuncDecl {
if trace {
defer p.trace("funcDecl")()
}
f := new(FuncDecl)
f.pos = p.pos()
f.Pragma = p.takePragma()
if p.got(_Lparen) {
rcvr := p.paramList(nil, nil, _Rparen, false)
switch len(rcvr) {
case 0:
p.error("method has no receiver")
default:
p.error("method has multiple receivers")
fallthrough
case 1:
f.Recv = rcvr[0]
}
}
if p.tok != _Name {
p.syntaxError("expecting name or (")
p.advance(_Lbrace, _Semi)
return nil
}
f.Name = p.name()
context := ""
if f.Recv != nil {
context = "method" // don't permit (method) type parameters in funcType
}
f.TParamList, f.Type = p.funcType(context)
if p.tok == _Lbrace {
f.Body = p.funcBody()
}
return f
}
func (p *parser) funcBody() *BlockStmt {
p.fnest++
errcnt := p.errcnt
body := p.blockStmt("")
p.fnest--
// Don't check branches if there were syntax errors in the function
// as it may lead to spurious errors (e.g., see test/switch2.go) or
// possibly crashes due to incomplete syntax trees.
if p.mode&CheckBranches != 0 && errcnt == p.errcnt {
checkBranches(body, p.errh)
}
return body
}
// ----------------------------------------------------------------------------
// Expressions
func (p *parser) expr() Expr {
if trace {
defer p.trace("expr")()
}
return p.binaryExpr(nil, 0)
}
// Expression = UnaryExpr | Expression binary_op Expression .
func (p *parser) binaryExpr(x Expr, prec int) Expr {
// don't trace binaryExpr - only leads to overly nested trace output
if x == nil {
x = p.unaryExpr()
}
for (p.tok == _Operator || p.tok == _Star) && p.prec > prec {
t := new(Operation)
t.pos = p.pos()
t.Op = p.op
tprec := p.prec
p.next()
t.X = x
t.Y = p.binaryExpr(nil, tprec)
x = t
}
return x
}
// UnaryExpr = PrimaryExpr | unary_op UnaryExpr .
func (p *parser) unaryExpr() Expr {
if trace {
defer p.trace("unaryExpr")()
}
switch p.tok {
case _Operator, _Star:
switch p.op {
case Mul, Add, Sub, Not, Xor, Tilde:
x := new(Operation)
x.pos = p.pos()
x.Op = p.op
p.next()
x.X = p.unaryExpr()
return x
case And:
x := new(Operation)
x.pos = p.pos()
x.Op = And
p.next()
// unaryExpr may have returned a parenthesized composite literal
// (see comment in operand) - remove parentheses if any
x.X = unparen(p.unaryExpr())
return x
}
case _Arrow:
// receive op (<-x) or receive-only channel (<-chan E)
pos := p.pos()
p.next()
// If the next token is _Chan we still don't know if it is
// a channel (<-chan int) or a receive op (<-chan int(ch)).
// We only know once we have found the end of the unaryExpr.
x := p.unaryExpr()
// There are two cases:
//
// <-chan... => <-x is a channel type
// <-x => <-x is a receive operation
//
// In the first case, <- must be re-associated with
// the channel type parsed already:
//
// <-(chan E) => (<-chan E)
// <-(chan<-E) => (<-chan (<-E))
if _, ok := x.(*ChanType); ok {
// x is a channel type => re-associate <-
dir := SendOnly
t := x
for dir == SendOnly {
c, ok := t.(*ChanType)
if !ok {
break
}
dir = c.Dir
if dir == RecvOnly {
// t is type <-chan E but <-<-chan E is not permitted
// (report same error as for "type _ <-<-chan E")
p.syntaxError("unexpected <-, expecting chan")
// already progressed, no need to advance
}
c.Dir = RecvOnly
t = c.Elem
}
if dir == SendOnly {
// channel dir is <- but channel element E is not a channel
// (report same error as for "type _ <-chan<-E")
p.syntaxError(fmt.Sprintf("unexpected %s, expecting chan", String(t)))
// already progressed, no need to advance
}
return x
}
// x is not a channel type => we have a receive op
o := new(Operation)
o.pos = pos
o.Op = Recv
o.X = x
return o
}
// TODO(mdempsky): We need parens here so we can report an
// error for "(x) := true". It should be possible to detect
// and reject that more efficiently though.
return p.pexpr(nil, true)
}
// callStmt parses call-like statements that can be preceded by 'defer' and 'go'.
func (p *parser) callStmt() *CallStmt {
if trace {
defer p.trace("callStmt")()
}
s := new(CallStmt)
s.pos = p.pos()
s.Tok = p.tok // _Defer or _Go
p.next()
x := p.pexpr(nil, p.tok == _Lparen) // keep_parens so we can report error below
if t := unparen(x); t != x {
p.errorAt(x.Pos(), fmt.Sprintf("expression in %s must not be parenthesized", s.Tok))
// already progressed, no need to advance
x = t
}
cx, ok := x.(*CallExpr)
if !ok {
p.errorAt(x.Pos(), fmt.Sprintf("expression in %s must be function call", s.Tok))
// already progressed, no need to advance
cx = new(CallExpr)
cx.pos = x.Pos()
cx.Fun = x // assume common error of missing parentheses (function invocation)
}
s.Call = cx
return s
}
// Operand = Literal | OperandName | MethodExpr | "(" Expression ")" .
// Literal = BasicLit | CompositeLit | FunctionLit .
// BasicLit = int_lit | float_lit | imaginary_lit | rune_lit | string_lit .
// OperandName = identifier | QualifiedIdent.
func (p *parser) operand(keep_parens bool) Expr {
if trace {
defer p.trace("operand " + p.tok.String())()
}
switch p.tok {
case _Name:
return p.name()
case _Literal:
return p.oliteral()
case _Lparen:
pos := p.pos()
p.next()
p.xnest++
x := p.expr()
p.xnest--
p.want(_Rparen)
// Optimization: Record presence of ()'s only where needed
// for error reporting. Don't bother in other cases; it is
// just a waste of memory and time.
//
// Parentheses are not permitted around T in a composite
// literal T{}. If the next token is a {, assume x is a
// composite literal type T (it may not be, { could be
// the opening brace of a block, but we don't know yet).
if p.tok == _Lbrace {
keep_parens = true
}
// Parentheses are also not permitted around the expression
// in a go/defer statement. In that case, operand is called
// with keep_parens set.
if keep_parens {
px := new(ParenExpr)
px.pos = pos
px.X = x
x = px
}
return x
case _Func:
pos := p.pos()
p.next()
_, ftyp := p.funcType("function type")
if p.tok == _Lbrace {
p.xnest++
f := new(FuncLit)
f.pos = pos
f.Type = ftyp
f.Body = p.funcBody()
p.xnest--
return f
}
return ftyp
case _Lbrack, _Chan, _Map, _Struct, _Interface:
return p.type_() // othertype
default:
x := p.badExpr()
p.syntaxError("expecting expression")
p.advance(_Rparen, _Rbrack, _Rbrace)
return x
}
// Syntactically, composite literals are operands. Because a complit
// type may be a qualified identifier which is handled by pexpr
// (together with selector expressions), complits are parsed there
// as well (operand is only called from pexpr).
}
// pexpr parses a PrimaryExpr.
//
// PrimaryExpr =
// Operand |
// Conversion |
// PrimaryExpr Selector |
// PrimaryExpr Index |
// PrimaryExpr Slice |
// PrimaryExpr TypeAssertion |
// PrimaryExpr Arguments .
//
// Selector = "." identifier .
// Index = "[" Expression "]" .
// Slice = "[" ( [ Expression ] ":" [ Expression ] ) |
// ( [ Expression ] ":" Expression ":" Expression )
// "]" .
// TypeAssertion = "." "(" Type ")" .
// Arguments = "(" [ ( ExpressionList | Type [ "," ExpressionList ] ) [ "..." ] [ "," ] ] ")" .
func (p *parser) pexpr(x Expr, keep_parens bool) Expr {
if trace {
defer p.trace("pexpr")()
}
if x == nil {
x = p.operand(keep_parens)
}
loop:
for {
pos := p.pos()
switch p.tok {
case _Dot:
p.next()
switch p.tok {
case _Name:
// pexpr '.' sym
t := new(SelectorExpr)
t.pos = pos
t.X = x
t.Sel = p.name()
x = t
case _Lparen:
p.next()
if p.got(_Type) {
t := new(TypeSwitchGuard)
// t.Lhs is filled in by parser.simpleStmt
t.pos = pos
t.X = x
x = t
} else {
t := new(AssertExpr)
t.pos = pos
t.X = x
t.Type = p.type_()
x = t
}
p.want(_Rparen)
default:
p.syntaxError("expecting name or (")
p.advance(_Semi, _Rparen)
}
case _Lbrack:
p.next()
if p.tok == _Rbrack {
// invalid empty instance, slice or index expression; accept but complain
p.syntaxError("expecting operand")
p.next()
break
}
var i Expr
if p.tok != _Colon {
var comma bool
i, comma = p.typeList()
if comma || p.tok == _Rbrack {
p.want(_Rbrack)
// x[i,] or x[i, j, ...]
t := new(IndexExpr)
t.pos = pos
t.X = x
t.Index = i
x = t
break
}
}
// x[i:...
// For better error message, don't simply use p.want(_Colon) here (issue #47704).
if !p.got(_Colon) {
p.syntaxError("expecting comma, : or ]")
p.advance(_Comma, _Colon, _Rbrack)
}
p.xnest++
t := new(SliceExpr)
t.pos = pos
t.X = x
t.Index[0] = i
if p.tok != _Colon && p.tok != _Rbrack {
// x[i:j...
t.Index[1] = p.expr()
}
if p.tok == _Colon {
t.Full = true
// x[i:j:...]
if t.Index[1] == nil {
p.error("middle index required in 3-index slice")
t.Index[1] = p.badExpr()
}
p.next()
if p.tok != _Rbrack {
// x[i:j:k...
t.Index[2] = p.expr()
} else {
p.error("final index required in 3-index slice")
t.Index[2] = p.badExpr()
}
}
p.xnest--
p.want(_Rbrack)
x = t
case _Lparen:
t := new(CallExpr)
t.pos = pos
p.next()
t.Fun = x
t.ArgList, t.HasDots = p.argList()
x = t
case _Lbrace:
// operand may have returned a parenthesized complit
// type; accept it but complain if we have a complit
t := unparen(x)
// determine if '{' belongs to a composite literal or a block statement
complit_ok := false
switch t.(type) {
case *Name, *SelectorExpr:
if p.xnest >= 0 {
// x is possibly a composite literal type
complit_ok = true
}
case *IndexExpr:
if p.xnest >= 0 && !isValue(t) {
// x is possibly a composite literal type
complit_ok = true
}
case *ArrayType, *SliceType, *StructType, *MapType:
// x is a comptype
complit_ok = true
}
if !complit_ok {
break loop
}
if t != x {
p.syntaxError("cannot parenthesize type in composite literal")
// already progressed, no need to advance
}
n := p.complitexpr()
n.Type = x
x = n
default:
break loop
}
}
return x
}
// isValue reports whether x syntactically must be a value (and not a type) expression.
func isValue(x Expr) bool {
switch x := x.(type) {
case *BasicLit, *CompositeLit, *FuncLit, *SliceExpr, *AssertExpr, *TypeSwitchGuard, *CallExpr:
return true
case *Operation:
return x.Op != Mul || x.Y != nil // *T may be a type
case *ParenExpr:
return isValue(x.X)
case *IndexExpr:
return isValue(x.X) || isValue(x.Index)
}
return false
}
// Element = Expression | LiteralValue .
func (p *parser) bare_complitexpr() Expr {
if trace {
defer p.trace("bare_complitexpr")()
}
if p.tok == _Lbrace {
// '{' start_complit braced_keyval_list '}'
return p.complitexpr()
}
return p.expr()
}
// LiteralValue = "{" [ ElementList [ "," ] ] "}" .
func (p *parser) complitexpr() *CompositeLit {
if trace {
defer p.trace("complitexpr")()
}
x := new(CompositeLit)
x.pos = p.pos()
p.xnest++
p.want(_Lbrace)
x.Rbrace = p.list("composite literal", _Comma, _Rbrace, func() bool {
// value
e := p.bare_complitexpr()
if p.tok == _Colon {
// key ':' value
l := new(KeyValueExpr)
l.pos = p.pos()
p.next()
l.Key = e
l.Value = p.bare_complitexpr()
e = l
x.NKeys++
}
x.ElemList = append(x.ElemList, e)
return false
})
p.xnest--
return x
}
// ----------------------------------------------------------------------------
// Types
func (p *parser) type_() Expr {
if trace {
defer p.trace("type_")()
}
typ := p.typeOrNil()
if typ == nil {
typ = p.badExpr()
p.syntaxError("expecting type")
p.advance(_Comma, _Colon, _Semi, _Rparen, _Rbrack, _Rbrace)
}
return typ
}
cmd/compile/internal/syntax: remove dependency on cmd/internal/src For dependency reasons, the data structure implementing source positions in the compiler is in cmd/internal/src. It contains highly compiler specific details (e.g. inlining index). This change introduces a parallel but simpler position representation, defined in the syntax package, which removes that package's dependency on cmd/internal/src, and also removes the need to deal with certain filename-specific operations (defined by the needs of the compiler) in the syntax package. As a result, the syntax package becomes again a compiler- independent, stand-alone package that at some point might replace (or augment) the existing top-level go/* syntax-related packages. Additionally, line directives that update column numbers are now correctly tracked through the syntax package, with additional tests added. (The respective changes also need to be made in cmd/internal/src; i.e., the compiler accepts but still ignores column numbers in line directives.) This change comes at the cost of a new position translation step, but that step is cheap because it only needs to do real work if the position base changed (i.e., if there is a new file, or new line directive). There is no noticeable impact on overall compiler performance measured with `compilebench -count 5 -alloc`: name old time/op new time/op delta Template 220ms ± 8% 228ms ±18% ~ (p=0.548 n=5+5) Unicode 119ms ±11% 113ms ± 5% ~ (p=0.056 n=5+5) GoTypes 684ms ± 6% 677ms ± 3% ~ (p=0.841 n=5+5) Compiler 3.19s ± 7% 3.01s ± 1% ~ (p=0.095 n=5+5) SSA 7.92s ± 8% 7.79s ± 1% ~ (p=0.690 n=5+5) Flate 141ms ± 7% 139ms ± 4% ~ (p=0.548 n=5+5) GoParser 173ms ±12% 171ms ± 4% ~ (p=1.000 n=5+5) Reflect 417ms ± 5% 411ms ± 3% ~ (p=0.548 n=5+5) Tar 205ms ± 5% 198ms ± 2% ~ (p=0.690 n=5+5) XML 232ms ± 4% 229ms ± 4% ~ (p=0.690 n=5+5) StdCmd 28.7s ± 5% 28.2s ± 2% ~ (p=0.421 n=5+5) name old user-time/op new user-time/op delta Template 269ms ± 4% 265ms ± 5% ~ (p=0.421 n=5+5) Unicode 153ms ± 7% 149ms ± 3% ~ (p=0.841 n=5+5) GoTypes 850ms ± 7% 862ms ± 4% ~ (p=0.841 n=5+5) Compiler 4.01s ± 5% 3.86s ± 0% ~ (p=0.190 n=5+4) SSA 10.9s ± 4% 10.8s ± 2% ~ (p=0.548 n=5+5) Flate 166ms ± 7% 167ms ± 6% ~ (p=1.000 n=5+5) GoParser 204ms ± 8% 206ms ± 7% ~ (p=0.841 n=5+5) Reflect 514ms ± 5% 508ms ± 4% ~ (p=0.548 n=5+5) Tar 245ms ± 6% 244ms ± 3% ~ (p=0.690 n=5+5) XML 280ms ± 4% 278ms ± 4% ~ (p=0.841 n=5+5) name old alloc/op new alloc/op delta Template 37.9MB ± 0% 37.9MB ± 0% ~ (p=0.841 n=5+5) Unicode 28.8MB ± 0% 28.8MB ± 0% ~ (p=0.841 n=5+5) GoTypes 113MB ± 0% 113MB ± 0% ~ (p=0.151 n=5+5) Compiler 468MB ± 0% 468MB ± 0% -0.01% (p=0.032 n=5+5) SSA 1.50GB ± 0% 1.50GB ± 0% ~ (p=0.548 n=5+5) Flate 24.4MB ± 0% 24.4MB ± 0% ~ (p=1.000 n=5+5) GoParser 30.7MB ± 0% 30.7MB ± 0% ~ (p=1.000 n=5+5) Reflect 76.5MB ± 0% 76.5MB ± 0% ~ (p=0.548 n=5+5) Tar 38.9MB ± 0% 38.9MB ± 0% ~ (p=0.222 n=5+5) XML 41.6MB ± 0% 41.6MB ± 0% ~ (p=0.548 n=5+5) name old allocs/op new allocs/op delta Template 382k ± 0% 382k ± 0% +0.01% (p=0.008 n=5+5) Unicode 343k ± 0% 343k ± 0% ~ (p=0.841 n=5+5) GoTypes 1.19M ± 0% 1.19M ± 0% +0.01% (p=0.008 n=5+5) Compiler 4.53M ± 0% 4.53M ± 0% +0.03% (p=0.008 n=5+5) SSA 12.4M ± 0% 12.4M ± 0% +0.00% (p=0.008 n=5+5) Flate 235k ± 0% 235k ± 0% ~ (p=0.079 n=5+5) GoParser 318k ± 0% 318k ± 0% ~ (p=0.730 n=5+5) Reflect 978k ± 0% 978k ± 0% ~ (p=1.000 n=5+5) Tar 393k ± 0% 393k ± 0% ~ (p=0.056 n=5+5) XML 405k ± 0% 405k ± 0% ~ (p=0.548 n=5+5) name old text-bytes new text-bytes delta HelloSize 672kB ± 0% 672kB ± 0% ~ (all equal) CmdGoSize 7.12MB ± 0% 7.12MB ± 0% ~ (all equal) name old data-bytes new data-bytes delta HelloSize 133kB ± 0% 133kB ± 0% ~ (all equal) CmdGoSize 390kB ± 0% 390kB ± 0% ~ (all equal) name old exe-bytes new exe-bytes delta HelloSize 1.07MB ± 0% 1.07MB ± 0% ~ (all equal) CmdGoSize 11.2MB ± 0% 11.2MB ± 0% ~ (all equal) Passes toolstash compare. For #22662. Change-Id: I19edb53dd9675af57f7122cb7dba2a6d8bdcc3da Reviewed-on: https://go-review.googlesource.com/94515 Reviewed-by: Matthew Dempsky <mdempsky@google.com>
2018-01-02 16:58:37 -08:00
func newIndirect(pos Pos, typ Expr) Expr {
o := new(Operation)
o.pos = pos
o.Op = Mul
o.X = typ
return o
}
// typeOrNil is like type_ but it returns nil if there was no type
// instead of reporting an error.
//
// Type = TypeName | TypeLit | "(" Type ")" .
// TypeName = identifier | QualifiedIdent .
// TypeLit = ArrayType | StructType | PointerType | FunctionType | InterfaceType |
// SliceType | MapType | Channel_Type .
func (p *parser) typeOrNil() Expr {
if trace {
defer p.trace("typeOrNil")()
}
pos := p.pos()
switch p.tok {
case _Star:
// ptrtype
p.next()
return newIndirect(pos, p.type_())
case _Arrow:
// recvchantype
p.next()
p.want(_Chan)
t := new(ChanType)
t.pos = pos
t.Dir = RecvOnly
t.Elem = p.chanElem()
return t
case _Func:
// fntype
p.next()
_, t := p.funcType("function type")
return t
case _Lbrack:
// '[' oexpr ']' ntype
// '[' _DotDotDot ']' ntype
p.next()
if p.got(_Rbrack) {
return p.sliceType(pos)
}
return p.arrayType(pos, nil)
case _Chan:
// _Chan non_recvchantype
// _Chan _Comm ntype
p.next()
t := new(ChanType)
t.pos = pos
if p.got(_Arrow) {
t.Dir = SendOnly
}
t.Elem = p.chanElem()
return t
case _Map:
// _Map '[' ntype ']' ntype
p.next()
p.want(_Lbrack)
t := new(MapType)
t.pos = pos
t.Key = p.type_()
p.want(_Rbrack)
t.Value = p.type_()
return t
case _Struct:
return p.structType()
case _Interface:
return p.interfaceType()
case _Name:
return p.qualifiedName(nil)
case _Lparen:
p.next()
t := p.type_()
p.want(_Rparen)
return t
}
return nil
}
func (p *parser) typeInstance(typ Expr) Expr {
if trace {
defer p.trace("typeInstance")()
}
pos := p.pos()
p.want(_Lbrack)
x := new(IndexExpr)
x.pos = pos
x.X = typ
if p.tok == _Rbrack {
p.syntaxError("expecting type")
x.Index = p.badExpr()
} else {
x.Index, _ = p.typeList()
}
p.want(_Rbrack)
return x
}
// If context != "", type parameters are not permitted.
func (p *parser) funcType(context string) ([]*Field, *FuncType) {
if trace {
defer p.trace("funcType")()
}
typ := new(FuncType)
typ.pos = p.pos()
var tparamList []*Field
if p.got(_Lbrack) {
if context != "" {
// accept but complain
p.syntaxErrorAt(typ.pos, context+" must have no type parameters")
}
if p.tok == _Rbrack {
p.syntaxError("empty type parameter list")
p.next()
} else {
tparamList = p.paramList(nil, nil, _Rbrack, true)
}
}
p.want(_Lparen)
typ.ParamList = p.paramList(nil, nil, _Rparen, false)
typ.ResultList = p.funcResult()
return tparamList, typ
}
// "[" has already been consumed, and pos is its position.
// If len != nil it is the already consumed array length.
func (p *parser) arrayType(pos Pos, len Expr) Expr {
if trace {
defer p.trace("arrayType")()
}
if len == nil && !p.got(_DotDotDot) {
p.xnest++
len = p.expr()
p.xnest--
}
if p.tok == _Comma {
// Trailing commas are accepted in type parameter
// lists but not in array type declarations.
// Accept for better error handling but complain.
p.syntaxError("unexpected comma; expecting ]")
p.next()
}
p.want(_Rbrack)
t := new(ArrayType)
t.pos = pos
t.Len = len
t.Elem = p.type_()
return t
}
// "[" and "]" have already been consumed, and pos is the position of "[".
func (p *parser) sliceType(pos Pos) Expr {
t := new(SliceType)
t.pos = pos
t.Elem = p.type_()
return t
}
func (p *parser) chanElem() Expr {
if trace {
defer p.trace("chanElem")()
}
typ := p.typeOrNil()
if typ == nil {
typ = p.badExpr()
p.syntaxError("missing channel element type")
// assume element type is simply absent - don't advance
}
return typ
}
// StructType = "struct" "{" { FieldDecl ";" } "}" .
func (p *parser) structType() *StructType {
if trace {
defer p.trace("structType")()
}
typ := new(StructType)
typ.pos = p.pos()
p.want(_Struct)
p.want(_Lbrace)
p.list("struct type", _Semi, _Rbrace, func() bool {
p.fieldDecl(typ)
return false
})
return typ
}
// InterfaceType = "interface" "{" { ( MethodDecl | EmbeddedElem | TypeList ) ";" } "}" .
// TypeList = "type" Type { "," Type } .
func (p *parser) interfaceType() *InterfaceType {
if trace {
defer p.trace("interfaceType")()
}
typ := new(InterfaceType)
typ.pos = p.pos()
p.want(_Interface)
p.want(_Lbrace)
p.list("interface type", _Semi, _Rbrace, func() bool {
var f *Field
if p.tok == _Name {
f = p.methodDecl()
}
if f == nil || f.Name == nil {
f = p.embeddedElem(f)
}
typ.MethodList = append(typ.MethodList, f)
return false
})
return typ
}
// Result = Parameters | Type .
func (p *parser) funcResult() []*Field {
if trace {
defer p.trace("funcResult")()
}
if p.got(_Lparen) {
return p.paramList(nil, nil, _Rparen, false)
}
pos := p.pos()
if typ := p.typeOrNil(); typ != nil {
f := new(Field)
f.pos = pos
f.Type = typ
return []*Field{f}
}
return nil
}
cmd/compile/internal/syntax: remove dependency on cmd/internal/src For dependency reasons, the data structure implementing source positions in the compiler is in cmd/internal/src. It contains highly compiler specific details (e.g. inlining index). This change introduces a parallel but simpler position representation, defined in the syntax package, which removes that package's dependency on cmd/internal/src, and also removes the need to deal with certain filename-specific operations (defined by the needs of the compiler) in the syntax package. As a result, the syntax package becomes again a compiler- independent, stand-alone package that at some point might replace (or augment) the existing top-level go/* syntax-related packages. Additionally, line directives that update column numbers are now correctly tracked through the syntax package, with additional tests added. (The respective changes also need to be made in cmd/internal/src; i.e., the compiler accepts but still ignores column numbers in line directives.) This change comes at the cost of a new position translation step, but that step is cheap because it only needs to do real work if the position base changed (i.e., if there is a new file, or new line directive). There is no noticeable impact on overall compiler performance measured with `compilebench -count 5 -alloc`: name old time/op new time/op delta Template 220ms ± 8% 228ms ±18% ~ (p=0.548 n=5+5) Unicode 119ms ±11% 113ms ± 5% ~ (p=0.056 n=5+5) GoTypes 684ms ± 6% 677ms ± 3% ~ (p=0.841 n=5+5) Compiler 3.19s ± 7% 3.01s ± 1% ~ (p=0.095 n=5+5) SSA 7.92s ± 8% 7.79s ± 1% ~ (p=0.690 n=5+5) Flate 141ms ± 7% 139ms ± 4% ~ (p=0.548 n=5+5) GoParser 173ms ±12% 171ms ± 4% ~ (p=1.000 n=5+5) Reflect 417ms ± 5% 411ms ± 3% ~ (p=0.548 n=5+5) Tar 205ms ± 5% 198ms ± 2% ~ (p=0.690 n=5+5) XML 232ms ± 4% 229ms ± 4% ~ (p=0.690 n=5+5) StdCmd 28.7s ± 5% 28.2s ± 2% ~ (p=0.421 n=5+5) name old user-time/op new user-time/op delta Template 269ms ± 4% 265ms ± 5% ~ (p=0.421 n=5+5) Unicode 153ms ± 7% 149ms ± 3% ~ (p=0.841 n=5+5) GoTypes 850ms ± 7% 862ms ± 4% ~ (p=0.841 n=5+5) Compiler 4.01s ± 5% 3.86s ± 0% ~ (p=0.190 n=5+4) SSA 10.9s ± 4% 10.8s ± 2% ~ (p=0.548 n=5+5) Flate 166ms ± 7% 167ms ± 6% ~ (p=1.000 n=5+5) GoParser 204ms ± 8% 206ms ± 7% ~ (p=0.841 n=5+5) Reflect 514ms ± 5% 508ms ± 4% ~ (p=0.548 n=5+5) Tar 245ms ± 6% 244ms ± 3% ~ (p=0.690 n=5+5) XML 280ms ± 4% 278ms ± 4% ~ (p=0.841 n=5+5) name old alloc/op new alloc/op delta Template 37.9MB ± 0% 37.9MB ± 0% ~ (p=0.841 n=5+5) Unicode 28.8MB ± 0% 28.8MB ± 0% ~ (p=0.841 n=5+5) GoTypes 113MB ± 0% 113MB ± 0% ~ (p=0.151 n=5+5) Compiler 468MB ± 0% 468MB ± 0% -0.01% (p=0.032 n=5+5) SSA 1.50GB ± 0% 1.50GB ± 0% ~ (p=0.548 n=5+5) Flate 24.4MB ± 0% 24.4MB ± 0% ~ (p=1.000 n=5+5) GoParser 30.7MB ± 0% 30.7MB ± 0% ~ (p=1.000 n=5+5) Reflect 76.5MB ± 0% 76.5MB ± 0% ~ (p=0.548 n=5+5) Tar 38.9MB ± 0% 38.9MB ± 0% ~ (p=0.222 n=5+5) XML 41.6MB ± 0% 41.6MB ± 0% ~ (p=0.548 n=5+5) name old allocs/op new allocs/op delta Template 382k ± 0% 382k ± 0% +0.01% (p=0.008 n=5+5) Unicode 343k ± 0% 343k ± 0% ~ (p=0.841 n=5+5) GoTypes 1.19M ± 0% 1.19M ± 0% +0.01% (p=0.008 n=5+5) Compiler 4.53M ± 0% 4.53M ± 0% +0.03% (p=0.008 n=5+5) SSA 12.4M ± 0% 12.4M ± 0% +0.00% (p=0.008 n=5+5) Flate 235k ± 0% 235k ± 0% ~ (p=0.079 n=5+5) GoParser 318k ± 0% 318k ± 0% ~ (p=0.730 n=5+5) Reflect 978k ± 0% 978k ± 0% ~ (p=1.000 n=5+5) Tar 393k ± 0% 393k ± 0% ~ (p=0.056 n=5+5) XML 405k ± 0% 405k ± 0% ~ (p=0.548 n=5+5) name old text-bytes new text-bytes delta HelloSize 672kB ± 0% 672kB ± 0% ~ (all equal) CmdGoSize 7.12MB ± 0% 7.12MB ± 0% ~ (all equal) name old data-bytes new data-bytes delta HelloSize 133kB ± 0% 133kB ± 0% ~ (all equal) CmdGoSize 390kB ± 0% 390kB ± 0% ~ (all equal) name old exe-bytes new exe-bytes delta HelloSize 1.07MB ± 0% 1.07MB ± 0% ~ (all equal) CmdGoSize 11.2MB ± 0% 11.2MB ± 0% ~ (all equal) Passes toolstash compare. For #22662. Change-Id: I19edb53dd9675af57f7122cb7dba2a6d8bdcc3da Reviewed-on: https://go-review.googlesource.com/94515 Reviewed-by: Matthew Dempsky <mdempsky@google.com>
2018-01-02 16:58:37 -08:00
func (p *parser) addField(styp *StructType, pos Pos, name *Name, typ Expr, tag *BasicLit) {
if tag != nil {
for i := len(styp.FieldList) - len(styp.TagList); i > 0; i-- {
styp.TagList = append(styp.TagList, nil)
}
styp.TagList = append(styp.TagList, tag)
}
f := new(Field)
f.pos = pos
f.Name = name
f.Type = typ
styp.FieldList = append(styp.FieldList, f)
if debug && tag != nil && len(styp.FieldList) != len(styp.TagList) {
panic("inconsistent struct field list")
}
}
// FieldDecl = (IdentifierList Type | AnonymousField) [ Tag ] .
// AnonymousField = [ "*" ] TypeName .
// Tag = string_lit .
func (p *parser) fieldDecl(styp *StructType) {
if trace {
defer p.trace("fieldDecl")()
}
pos := p.pos()
switch p.tok {
case _Name:
name := p.name()
if p.tok == _Dot || p.tok == _Literal || p.tok == _Semi || p.tok == _Rbrace {
// embedded type
typ := p.qualifiedName(name)
tag := p.oliteral()
p.addField(styp, pos, nil, typ, tag)
break
}
// name1, name2, ... Type [ tag ]
names := p.nameList(name)
var typ Expr
// Careful dance: We don't know if we have an embedded instantiated
// type T[P1, P2, ...] or a field T of array/slice type [P]E or []E.
if len(names) == 1 && p.tok == _Lbrack {
typ = p.arrayOrTArgs()
if typ, ok := typ.(*IndexExpr); ok {
// embedded type T[P1, P2, ...]
typ.X = name // name == names[0]
tag := p.oliteral()
p.addField(styp, pos, nil, typ, tag)
break
}
} else {
// T P
typ = p.type_()
}
tag := p.oliteral()
for _, name := range names {
p.addField(styp, name.Pos(), name, typ, tag)
}
case _Star:
p.next()
var typ Expr
if p.tok == _Lparen {
// *(T)
p.syntaxError("cannot parenthesize embedded type")
p.next()
typ = p.qualifiedName(nil)
p.got(_Rparen) // no need to complain if missing
} else {
// *T
typ = p.qualifiedName(nil)
}
tag := p.oliteral()
p.addField(styp, pos, nil, newIndirect(pos, typ), tag)
case _Lparen:
p.syntaxError("cannot parenthesize embedded type")
p.next()
var typ Expr
if p.tok == _Star {
// (*T)
pos := p.pos()
p.next()
typ = newIndirect(pos, p.qualifiedName(nil))
} else {
// (T)
typ = p.qualifiedName(nil)
}
p.got(_Rparen) // no need to complain if missing
tag := p.oliteral()
p.addField(styp, pos, nil, typ, tag)
default:
p.syntaxError("expecting field name or embedded type")
p.advance(_Semi, _Rbrace)
}
}
func (p *parser) arrayOrTArgs() Expr {
if trace {
defer p.trace("arrayOrTArgs")()
}
pos := p.pos()
p.want(_Lbrack)
if p.got(_Rbrack) {
return p.sliceType(pos)
}
// x [n]E or x[n,], x[n1, n2], ...
n, comma := p.typeList()
p.want(_Rbrack)
if !comma {
if elem := p.typeOrNil(); elem != nil {
// x [n]E
t := new(ArrayType)
t.pos = pos
t.Len = n
t.Elem = elem
return t
}
}
// x[n,], x[n1, n2], ...
t := new(IndexExpr)
t.pos = pos
// t.X will be filled in by caller
t.Index = n
return t
}
func (p *parser) oliteral() *BasicLit {
if p.tok == _Literal {
b := new(BasicLit)
b.pos = p.pos()
b.Value = p.lit
b.Kind = p.kind
b.Bad = p.bad
p.next()
return b
}
return nil
}
// MethodSpec = MethodName Signature | InterfaceTypeName .
// MethodName = identifier .
// InterfaceTypeName = TypeName .
func (p *parser) methodDecl() *Field {
if trace {
defer p.trace("methodDecl")()
}
f := new(Field)
f.pos = p.pos()
name := p.name()
const context = "interface method"
switch p.tok {
case _Lparen:
// method
f.Name = name
_, f.Type = p.funcType(context)
case _Lbrack:
// Careful dance: We don't know if we have a generic method m[T C](x T)
// or an embedded instantiated type T[P1, P2] (we accept generic methods
// for generality and robustness of parsing but complain with an error).
pos := p.pos()
p.next()
// Empty type parameter or argument lists are not permitted.
// Treat as if [] were absent.
if p.tok == _Rbrack {
// name[]
pos := p.pos()
p.next()
if p.tok == _Lparen {
// name[](
p.errorAt(pos, "empty type parameter list")
f.Name = name
_, f.Type = p.funcType(context)
} else {
p.errorAt(pos, "empty type argument list")
f.Type = name
}
break
}
// A type argument list looks like a parameter list with only
// types. Parse a parameter list and decide afterwards.
list := p.paramList(nil, nil, _Rbrack, false)
if len(list) == 0 {
// The type parameter list is not [] but we got nothing
// due to other errors (reported by paramList). Treat
// as if [] were absent.
if p.tok == _Lparen {
f.Name = name
_, f.Type = p.funcType(context)
} else {
f.Type = name
}
break
}
// len(list) > 0
if list[0].Name != nil {
// generic method
f.Name = name
_, f.Type = p.funcType(context)
p.errorAt(pos, "interface method must have no type parameters")
break
}
// embedded instantiated type
t := new(IndexExpr)
t.pos = pos
t.X = name
if len(list) == 1 {
t.Index = list[0].Type
} else {
// len(list) > 1
l := new(ListExpr)
l.pos = list[0].Pos()
l.ElemList = make([]Expr, len(list))
for i := range list {
l.ElemList[i] = list[i].Type
}
t.Index = l
}
f.Type = t
default:
// embedded type
f.Type = p.qualifiedName(name)
}
return f
}
// EmbeddedElem = MethodSpec | EmbeddedTerm { "|" EmbeddedTerm } .
func (p *parser) embeddedElem(f *Field) *Field {
if trace {
defer p.trace("embeddedElem")()
}
if f == nil {
f = new(Field)
f.pos = p.pos()
f.Type = p.embeddedTerm()
}
for p.tok == _Operator && p.op == Or {
t := new(Operation)
t.pos = p.pos()
t.Op = Or
p.next()
t.X = f.Type
t.Y = p.embeddedTerm()
f.Type = t
}
return f
}
// EmbeddedTerm = [ "~" ] Type .
func (p *parser) embeddedTerm() Expr {
if trace {
defer p.trace("embeddedTerm")()
}
if p.tok == _Operator && p.op == Tilde {
t := new(Operation)
t.pos = p.pos()
t.Op = Tilde
p.next()
t.X = p.type_()
return t
}
t := p.typeOrNil()
if t == nil {
t = p.badExpr()
p.syntaxError("expecting ~ term or type")
p.advance(_Operator, _Semi, _Rparen, _Rbrack, _Rbrace)
}
return t
}
// ParameterDecl = [ IdentifierList ] [ "..." ] Type .
func (p *parser) paramDeclOrNil(name *Name, follow token) *Field {
if trace {
defer p.trace("paramDeclOrNil")()
}
// type set notation is ok in type parameter lists
typeSetsOk := follow == _Rbrack
pos := p.pos()
if name != nil {
pos = name.pos
} else if typeSetsOk && p.tok == _Operator && p.op == Tilde {
// "~" ...
return p.embeddedElem(nil)
}
f := new(Field)
f.pos = pos
if p.tok == _Name || name != nil {
// name
if name == nil {
name = p.name()
}
if p.tok == _Lbrack {
// name "[" ...
f.Type = p.arrayOrTArgs()
if typ, ok := f.Type.(*IndexExpr); ok {
// name "[" ... "]"
typ.X = name
} else {
// name "[" n "]" E
f.Name = name
}
if typeSetsOk && p.tok == _Operator && p.op == Or {
// name "[" ... "]" "|" ...
// name "[" n "]" E "|" ...
f = p.embeddedElem(f)
}
return f
}
if p.tok == _Dot {
// name "." ...
f.Type = p.qualifiedName(name)
if typeSetsOk && p.tok == _Operator && p.op == Or {
// name "." name "|" ...
f = p.embeddedElem(f)
}
return f
}
if typeSetsOk && p.tok == _Operator && p.op == Or {
// name "|" ...
f.Type = name
return p.embeddedElem(f)
}
f.Name = name
}
if p.tok == _DotDotDot {
// [name] "..." ...
t := new(DotsType)
t.pos = p.pos()
p.next()
t.Elem = p.typeOrNil()
if t.Elem == nil {
t.Elem = p.badExpr()
p.syntaxError("... is missing type")
}
f.Type = t
return f
}
if typeSetsOk && p.tok == _Operator && p.op == Tilde {
// [name] "~" ...
f.Type = p.embeddedElem(nil).Type
return f
}
f.Type = p.typeOrNil()
if typeSetsOk && p.tok == _Operator && p.op == Or && f.Type != nil {
// [name] type "|"
f = p.embeddedElem(f)
}
if f.Name != nil || f.Type != nil {
return f
}
p.syntaxError("expecting " + tokstring(follow))
p.advance(_Comma, follow)
return nil
}
// Parameters = "(" [ ParameterList [ "," ] ] ")" .
// ParameterList = ParameterDecl { "," ParameterDecl } .
// "(" or "[" has already been consumed.
// If name != nil, it is the first name after "(" or "[".
// If typ != nil, name must be != nil, and (name, typ) is the first field in the list.
// In the result list, either all fields have a name, or no field has a name.
func (p *parser) paramList(name *Name, typ Expr, close token, requireNames bool) (list []*Field) {
if trace {
defer p.trace("paramList")()
}
// p.list won't invoke its function argument if we're at the end of the
// parameter list. If we have a complete field, handle this case here.
if name != nil && typ != nil && p.tok == close {
p.next()
par := new(Field)
par.pos = name.pos
par.Name = name
par.Type = typ
return []*Field{par}
}
var named int // number of parameters that have an explicit name and type
var typed int // number of parameters that have an explicit type
end := p.list("parameter list", _Comma, close, func() bool {
var par *Field
if typ != nil {
if debug && name == nil {
panic("initial type provided without name")
}
par = new(Field)
par.pos = name.pos
par.Name = name
par.Type = typ
} else {
par = p.paramDeclOrNil(name, close)
}
name = nil // 1st name was consumed if present
typ = nil // 1st type was consumed if present
if par != nil {
if debug && par.Name == nil && par.Type == nil {
panic("parameter without name or type")
}
if par.Name != nil && par.Type != nil {
named++
}
if par.Type != nil {
typed++
}
list = append(list, par)
}
return false
})
if len(list) == 0 {
return
}
// distribute parameter types (len(list) > 0)
if named == 0 && !requireNames {
// all unnamed => found names are named types
for _, par := range list {
if typ := par.Name; typ != nil {
par.Type = typ
par.Name = nil
}
}
} else if named != len(list) {
// some named => all must have names and types
var pos Pos // left-most error position (or unknown)
var typ Expr // current type (from right to left)
for i := len(list) - 1; i >= 0; i-- {
par := list[i]
if par.Type != nil {
typ = par.Type
if par.Name == nil {
pos = StartPos(typ)
par.Name = NewName(pos, "_")
}
} else if typ != nil {
par.Type = typ
} else {
// par.Type == nil && typ == nil => we only have a par.Name
pos = par.Name.Pos()
t := p.badExpr()
t.pos = pos // correct position
par.Type = t
}
}
if pos.IsKnown() {
var msg string
if requireNames {
if named == typed {
pos = end // position error at closing ]
msg = "missing type constraint"
} else {
msg = "type parameters must be named"
}
} else {
msg = "mixed named and unnamed parameters"
}
p.syntaxErrorAt(pos, msg)
}
}
return
}
func (p *parser) badExpr() *BadExpr {
b := new(BadExpr)
b.pos = p.pos()
return b
}
// ----------------------------------------------------------------------------
// Statements
// SimpleStmt = EmptyStmt | ExpressionStmt | SendStmt | IncDecStmt | Assignment | ShortVarDecl .
func (p *parser) simpleStmt(lhs Expr, keyword token) SimpleStmt {
if trace {
defer p.trace("simpleStmt")()
}
if keyword == _For && p.tok == _Range {
// _Range expr
if debug && lhs != nil {
panic("invalid call of simpleStmt")
}
return p.newRangeClause(nil, false)
}
if lhs == nil {
lhs = p.exprList()
}
if _, ok := lhs.(*ListExpr); !ok && p.tok != _Assign && p.tok != _Define {
// expr
pos := p.pos()
switch p.tok {
case _AssignOp:
// lhs op= rhs
op := p.op
p.next()
return p.newAssignStmt(pos, op, lhs, p.expr())
case _IncOp:
// lhs++ or lhs--
op := p.op
p.next()
return p.newAssignStmt(pos, op, lhs, nil)
case _Arrow:
// lhs <- rhs
s := new(SendStmt)
s.pos = pos
p.next()
s.Chan = lhs
s.Value = p.expr()
return s
default:
// expr
s := new(ExprStmt)
s.pos = lhs.Pos()
s.X = lhs
return s
}
}
// expr_list
switch p.tok {
case _Assign, _Define:
pos := p.pos()
var op Operator
if p.tok == _Define {
op = Def
}
p.next()
if keyword == _For && p.tok == _Range {
// expr_list op= _Range expr
return p.newRangeClause(lhs, op == Def)
}
// expr_list op= expr_list
rhs := p.exprList()
if x, ok := rhs.(*TypeSwitchGuard); ok && keyword == _Switch && op == Def {
if lhs, ok := lhs.(*Name); ok {
// switch … lhs := rhs.(type)
x.Lhs = lhs
s := new(ExprStmt)
s.pos = x.Pos()
s.X = x
return s
}
}
return p.newAssignStmt(pos, op, lhs, rhs)
default:
p.syntaxError("expecting := or = or comma")
p.advance(_Semi, _Rbrace)
// make the best of what we have
if x, ok := lhs.(*ListExpr); ok {
lhs = x.ElemList[0]
}
s := new(ExprStmt)
s.pos = lhs.Pos()
s.X = lhs
return s
}
}
func (p *parser) newRangeClause(lhs Expr, def bool) *RangeClause {
r := new(RangeClause)
r.pos = p.pos()
p.next() // consume _Range
r.Lhs = lhs
r.Def = def
r.X = p.expr()
return r
}
cmd/compile/internal/syntax: remove dependency on cmd/internal/src For dependency reasons, the data structure implementing source positions in the compiler is in cmd/internal/src. It contains highly compiler specific details (e.g. inlining index). This change introduces a parallel but simpler position representation, defined in the syntax package, which removes that package's dependency on cmd/internal/src, and also removes the need to deal with certain filename-specific operations (defined by the needs of the compiler) in the syntax package. As a result, the syntax package becomes again a compiler- independent, stand-alone package that at some point might replace (or augment) the existing top-level go/* syntax-related packages. Additionally, line directives that update column numbers are now correctly tracked through the syntax package, with additional tests added. (The respective changes also need to be made in cmd/internal/src; i.e., the compiler accepts but still ignores column numbers in line directives.) This change comes at the cost of a new position translation step, but that step is cheap because it only needs to do real work if the position base changed (i.e., if there is a new file, or new line directive). There is no noticeable impact on overall compiler performance measured with `compilebench -count 5 -alloc`: name old time/op new time/op delta Template 220ms ± 8% 228ms ±18% ~ (p=0.548 n=5+5) Unicode 119ms ±11% 113ms ± 5% ~ (p=0.056 n=5+5) GoTypes 684ms ± 6% 677ms ± 3% ~ (p=0.841 n=5+5) Compiler 3.19s ± 7% 3.01s ± 1% ~ (p=0.095 n=5+5) SSA 7.92s ± 8% 7.79s ± 1% ~ (p=0.690 n=5+5) Flate 141ms ± 7% 139ms ± 4% ~ (p=0.548 n=5+5) GoParser 173ms ±12% 171ms ± 4% ~ (p=1.000 n=5+5) Reflect 417ms ± 5% 411ms ± 3% ~ (p=0.548 n=5+5) Tar 205ms ± 5% 198ms ± 2% ~ (p=0.690 n=5+5) XML 232ms ± 4% 229ms ± 4% ~ (p=0.690 n=5+5) StdCmd 28.7s ± 5% 28.2s ± 2% ~ (p=0.421 n=5+5) name old user-time/op new user-time/op delta Template 269ms ± 4% 265ms ± 5% ~ (p=0.421 n=5+5) Unicode 153ms ± 7% 149ms ± 3% ~ (p=0.841 n=5+5) GoTypes 850ms ± 7% 862ms ± 4% ~ (p=0.841 n=5+5) Compiler 4.01s ± 5% 3.86s ± 0% ~ (p=0.190 n=5+4) SSA 10.9s ± 4% 10.8s ± 2% ~ (p=0.548 n=5+5) Flate 166ms ± 7% 167ms ± 6% ~ (p=1.000 n=5+5) GoParser 204ms ± 8% 206ms ± 7% ~ (p=0.841 n=5+5) Reflect 514ms ± 5% 508ms ± 4% ~ (p=0.548 n=5+5) Tar 245ms ± 6% 244ms ± 3% ~ (p=0.690 n=5+5) XML 280ms ± 4% 278ms ± 4% ~ (p=0.841 n=5+5) name old alloc/op new alloc/op delta Template 37.9MB ± 0% 37.9MB ± 0% ~ (p=0.841 n=5+5) Unicode 28.8MB ± 0% 28.8MB ± 0% ~ (p=0.841 n=5+5) GoTypes 113MB ± 0% 113MB ± 0% ~ (p=0.151 n=5+5) Compiler 468MB ± 0% 468MB ± 0% -0.01% (p=0.032 n=5+5) SSA 1.50GB ± 0% 1.50GB ± 0% ~ (p=0.548 n=5+5) Flate 24.4MB ± 0% 24.4MB ± 0% ~ (p=1.000 n=5+5) GoParser 30.7MB ± 0% 30.7MB ± 0% ~ (p=1.000 n=5+5) Reflect 76.5MB ± 0% 76.5MB ± 0% ~ (p=0.548 n=5+5) Tar 38.9MB ± 0% 38.9MB ± 0% ~ (p=0.222 n=5+5) XML 41.6MB ± 0% 41.6MB ± 0% ~ (p=0.548 n=5+5) name old allocs/op new allocs/op delta Template 382k ± 0% 382k ± 0% +0.01% (p=0.008 n=5+5) Unicode 343k ± 0% 343k ± 0% ~ (p=0.841 n=5+5) GoTypes 1.19M ± 0% 1.19M ± 0% +0.01% (p=0.008 n=5+5) Compiler 4.53M ± 0% 4.53M ± 0% +0.03% (p=0.008 n=5+5) SSA 12.4M ± 0% 12.4M ± 0% +0.00% (p=0.008 n=5+5) Flate 235k ± 0% 235k ± 0% ~ (p=0.079 n=5+5) GoParser 318k ± 0% 318k ± 0% ~ (p=0.730 n=5+5) Reflect 978k ± 0% 978k ± 0% ~ (p=1.000 n=5+5) Tar 393k ± 0% 393k ± 0% ~ (p=0.056 n=5+5) XML 405k ± 0% 405k ± 0% ~ (p=0.548 n=5+5) name old text-bytes new text-bytes delta HelloSize 672kB ± 0% 672kB ± 0% ~ (all equal) CmdGoSize 7.12MB ± 0% 7.12MB ± 0% ~ (all equal) name old data-bytes new data-bytes delta HelloSize 133kB ± 0% 133kB ± 0% ~ (all equal) CmdGoSize 390kB ± 0% 390kB ± 0% ~ (all equal) name old exe-bytes new exe-bytes delta HelloSize 1.07MB ± 0% 1.07MB ± 0% ~ (all equal) CmdGoSize 11.2MB ± 0% 11.2MB ± 0% ~ (all equal) Passes toolstash compare. For #22662. Change-Id: I19edb53dd9675af57f7122cb7dba2a6d8bdcc3da Reviewed-on: https://go-review.googlesource.com/94515 Reviewed-by: Matthew Dempsky <mdempsky@google.com>
2018-01-02 16:58:37 -08:00
func (p *parser) newAssignStmt(pos Pos, op Operator, lhs, rhs Expr) *AssignStmt {
a := new(AssignStmt)
a.pos = pos
a.Op = op
a.Lhs = lhs
a.Rhs = rhs
return a
}
func (p *parser) labeledStmtOrNil(label *Name) Stmt {
if trace {
defer p.trace("labeledStmt")()
}
s := new(LabeledStmt)
s.pos = p.pos()
s.Label = label
p.want(_Colon)
if p.tok == _Rbrace {
// We expect a statement (incl. an empty statement), which must be
// terminated by a semicolon. Because semicolons may be omitted before
// an _Rbrace, seeing an _Rbrace implies an empty statement.
e := new(EmptyStmt)
e.pos = p.pos()
s.Stmt = e
return s
}
s.Stmt = p.stmtOrNil()
if s.Stmt != nil {
return s
}
// report error at line of ':' token
p.syntaxErrorAt(s.pos, "missing statement after label")
// we are already at the end of the labeled statement - no need to advance
return nil // avoids follow-on errors (see e.g., fixedbugs/bug274.go)
}
// context must be a non-empty string unless we know that p.tok == _Lbrace.
func (p *parser) blockStmt(context string) *BlockStmt {
if trace {
defer p.trace("blockStmt")()
}
s := new(BlockStmt)
s.pos = p.pos()
// people coming from C may forget that braces are mandatory in Go
if !p.got(_Lbrace) {
p.syntaxError("expecting { after " + context)
p.advance(_Name, _Rbrace)
s.Rbrace = p.pos() // in case we found "}"
if p.got(_Rbrace) {
return s
}
}
s.List = p.stmtList()
s.Rbrace = p.pos()
p.want(_Rbrace)
return s
}
func (p *parser) declStmt(f func(*Group) Decl) *DeclStmt {
if trace {
defer p.trace("declStmt")()
}
s := new(DeclStmt)
s.pos = p.pos()
p.next() // _Const, _Type, or _Var
s.DeclList = p.appendGroup(nil, f)
return s
}
func (p *parser) forStmt() Stmt {
if trace {
defer p.trace("forStmt")()
}
s := new(ForStmt)
s.pos = p.pos()
s.Init, s.Cond, s.Post = p.header(_For)
s.Body = p.blockStmt("for clause")
return s
}
func (p *parser) header(keyword token) (init SimpleStmt, cond Expr, post SimpleStmt) {
p.want(keyword)
if p.tok == _Lbrace {
if keyword == _If {
p.syntaxError("missing condition in if statement")
cond = p.badExpr()
}
return
}
// p.tok != _Lbrace
outer := p.xnest
p.xnest = -1
if p.tok != _Semi {
// accept potential varDecl but complain
if p.got(_Var) {
p.syntaxError(fmt.Sprintf("var declaration not allowed in %s initializer", tokstring(keyword)))
}
init = p.simpleStmt(nil, keyword)
// If we have a range clause, we are done (can only happen for keyword == _For).
if _, ok := init.(*RangeClause); ok {
p.xnest = outer
return
}
}
var condStmt SimpleStmt
var semi struct {
cmd/compile/internal/syntax: remove dependency on cmd/internal/src For dependency reasons, the data structure implementing source positions in the compiler is in cmd/internal/src. It contains highly compiler specific details (e.g. inlining index). This change introduces a parallel but simpler position representation, defined in the syntax package, which removes that package's dependency on cmd/internal/src, and also removes the need to deal with certain filename-specific operations (defined by the needs of the compiler) in the syntax package. As a result, the syntax package becomes again a compiler- independent, stand-alone package that at some point might replace (or augment) the existing top-level go/* syntax-related packages. Additionally, line directives that update column numbers are now correctly tracked through the syntax package, with additional tests added. (The respective changes also need to be made in cmd/internal/src; i.e., the compiler accepts but still ignores column numbers in line directives.) This change comes at the cost of a new position translation step, but that step is cheap because it only needs to do real work if the position base changed (i.e., if there is a new file, or new line directive). There is no noticeable impact on overall compiler performance measured with `compilebench -count 5 -alloc`: name old time/op new time/op delta Template 220ms ± 8% 228ms ±18% ~ (p=0.548 n=5+5) Unicode 119ms ±11% 113ms ± 5% ~ (p=0.056 n=5+5) GoTypes 684ms ± 6% 677ms ± 3% ~ (p=0.841 n=5+5) Compiler 3.19s ± 7% 3.01s ± 1% ~ (p=0.095 n=5+5) SSA 7.92s ± 8% 7.79s ± 1% ~ (p=0.690 n=5+5) Flate 141ms ± 7% 139ms ± 4% ~ (p=0.548 n=5+5) GoParser 173ms ±12% 171ms ± 4% ~ (p=1.000 n=5+5) Reflect 417ms ± 5% 411ms ± 3% ~ (p=0.548 n=5+5) Tar 205ms ± 5% 198ms ± 2% ~ (p=0.690 n=5+5) XML 232ms ± 4% 229ms ± 4% ~ (p=0.690 n=5+5) StdCmd 28.7s ± 5% 28.2s ± 2% ~ (p=0.421 n=5+5) name old user-time/op new user-time/op delta Template 269ms ± 4% 265ms ± 5% ~ (p=0.421 n=5+5) Unicode 153ms ± 7% 149ms ± 3% ~ (p=0.841 n=5+5) GoTypes 850ms ± 7% 862ms ± 4% ~ (p=0.841 n=5+5) Compiler 4.01s ± 5% 3.86s ± 0% ~ (p=0.190 n=5+4) SSA 10.9s ± 4% 10.8s ± 2% ~ (p=0.548 n=5+5) Flate 166ms ± 7% 167ms ± 6% ~ (p=1.000 n=5+5) GoParser 204ms ± 8% 206ms ± 7% ~ (p=0.841 n=5+5) Reflect 514ms ± 5% 508ms ± 4% ~ (p=0.548 n=5+5) Tar 245ms ± 6% 244ms ± 3% ~ (p=0.690 n=5+5) XML 280ms ± 4% 278ms ± 4% ~ (p=0.841 n=5+5) name old alloc/op new alloc/op delta Template 37.9MB ± 0% 37.9MB ± 0% ~ (p=0.841 n=5+5) Unicode 28.8MB ± 0% 28.8MB ± 0% ~ (p=0.841 n=5+5) GoTypes 113MB ± 0% 113MB ± 0% ~ (p=0.151 n=5+5) Compiler 468MB ± 0% 468MB ± 0% -0.01% (p=0.032 n=5+5) SSA 1.50GB ± 0% 1.50GB ± 0% ~ (p=0.548 n=5+5) Flate 24.4MB ± 0% 24.4MB ± 0% ~ (p=1.000 n=5+5) GoParser 30.7MB ± 0% 30.7MB ± 0% ~ (p=1.000 n=5+5) Reflect 76.5MB ± 0% 76.5MB ± 0% ~ (p=0.548 n=5+5) Tar 38.9MB ± 0% 38.9MB ± 0% ~ (p=0.222 n=5+5) XML 41.6MB ± 0% 41.6MB ± 0% ~ (p=0.548 n=5+5) name old allocs/op new allocs/op delta Template 382k ± 0% 382k ± 0% +0.01% (p=0.008 n=5+5) Unicode 343k ± 0% 343k ± 0% ~ (p=0.841 n=5+5) GoTypes 1.19M ± 0% 1.19M ± 0% +0.01% (p=0.008 n=5+5) Compiler 4.53M ± 0% 4.53M ± 0% +0.03% (p=0.008 n=5+5) SSA 12.4M ± 0% 12.4M ± 0% +0.00% (p=0.008 n=5+5) Flate 235k ± 0% 235k ± 0% ~ (p=0.079 n=5+5) GoParser 318k ± 0% 318k ± 0% ~ (p=0.730 n=5+5) Reflect 978k ± 0% 978k ± 0% ~ (p=1.000 n=5+5) Tar 393k ± 0% 393k ± 0% ~ (p=0.056 n=5+5) XML 405k ± 0% 405k ± 0% ~ (p=0.548 n=5+5) name old text-bytes new text-bytes delta HelloSize 672kB ± 0% 672kB ± 0% ~ (all equal) CmdGoSize 7.12MB ± 0% 7.12MB ± 0% ~ (all equal) name old data-bytes new data-bytes delta HelloSize 133kB ± 0% 133kB ± 0% ~ (all equal) CmdGoSize 390kB ± 0% 390kB ± 0% ~ (all equal) name old exe-bytes new exe-bytes delta HelloSize 1.07MB ± 0% 1.07MB ± 0% ~ (all equal) CmdGoSize 11.2MB ± 0% 11.2MB ± 0% ~ (all equal) Passes toolstash compare. For #22662. Change-Id: I19edb53dd9675af57f7122cb7dba2a6d8bdcc3da Reviewed-on: https://go-review.googlesource.com/94515 Reviewed-by: Matthew Dempsky <mdempsky@google.com>
2018-01-02 16:58:37 -08:00
pos Pos
lit string // valid if pos.IsKnown()
}
if p.tok != _Lbrace {
if p.tok == _Semi {
semi.pos = p.pos()
semi.lit = p.lit
p.next()
} else {
// asking for a '{' rather than a ';' here leads to a better error message
p.want(_Lbrace)
if p.tok != _Lbrace {
p.advance(_Lbrace, _Rbrace) // for better synchronization (e.g., issue #22581)
}
}
if keyword == _For {
if p.tok != _Semi {
if p.tok == _Lbrace {
p.syntaxError("expecting for loop condition")
goto done
}
condStmt = p.simpleStmt(nil, 0 /* range not permitted */)
}
p.want(_Semi)
if p.tok != _Lbrace {
post = p.simpleStmt(nil, 0 /* range not permitted */)
if a, _ := post.(*AssignStmt); a != nil && a.Op == Def {
p.syntaxErrorAt(a.Pos(), "cannot declare in post statement of for loop")
}
}
} else if p.tok != _Lbrace {
condStmt = p.simpleStmt(nil, keyword)
}
} else {
condStmt = init
init = nil
}
done:
// unpack condStmt
switch s := condStmt.(type) {
case nil:
if keyword == _If && semi.pos.IsKnown() {
if semi.lit != "semicolon" {
p.syntaxErrorAt(semi.pos, fmt.Sprintf("unexpected %s, expecting { after if clause", semi.lit))
} else {
p.syntaxErrorAt(semi.pos, "missing condition in if statement")
}
b := new(BadExpr)
b.pos = semi.pos
cond = b
}
case *ExprStmt:
cond = s.X
default:
// A common syntax error is to write '=' instead of '==',
// which turns an expression into an assignment. Provide
// a more explicit error message in that case to prevent
// further confusion.
var str string
if as, ok := s.(*AssignStmt); ok && as.Op == 0 {
// Emphasize Lhs and Rhs of assignment with parentheses to highlight '='.
// Do it always - it's not worth going through the trouble of doing it
// only for "complex" left and right sides.
str = "assignment (" + String(as.Lhs) + ") = (" + String(as.Rhs) + ")"
} else {
str = String(s)
}
p.syntaxErrorAt(s.Pos(), fmt.Sprintf("cannot use %s as value", str))
}
p.xnest = outer
return
}
func (p *parser) ifStmt() *IfStmt {
if trace {
defer p.trace("ifStmt")()
}
s := new(IfStmt)
s.pos = p.pos()
s.Init, s.Cond, _ = p.header(_If)
s.Then = p.blockStmt("if clause")
if p.got(_Else) {
switch p.tok {
case _If:
s.Else = p.ifStmt()
case _Lbrace:
s.Else = p.blockStmt("")
default:
p.syntaxError("else must be followed by if or statement block")
p.advance(_Name, _Rbrace)
}
}
return s
}
func (p *parser) switchStmt() *SwitchStmt {
if trace {
defer p.trace("switchStmt")()
}
s := new(SwitchStmt)
s.pos = p.pos()
s.Init, s.Tag, _ = p.header(_Switch)
if !p.got(_Lbrace) {
p.syntaxError("missing { after switch clause")
p.advance(_Case, _Default, _Rbrace)
}
for p.tok != _EOF && p.tok != _Rbrace {
s.Body = append(s.Body, p.caseClause())
}
s.Rbrace = p.pos()
p.want(_Rbrace)
return s
}
func (p *parser) selectStmt() *SelectStmt {
if trace {
defer p.trace("selectStmt")()
}
s := new(SelectStmt)
s.pos = p.pos()
p.want(_Select)
if !p.got(_Lbrace) {
p.syntaxError("missing { after select clause")
p.advance(_Case, _Default, _Rbrace)
}
for p.tok != _EOF && p.tok != _Rbrace {
s.Body = append(s.Body, p.commClause())
}
s.Rbrace = p.pos()
p.want(_Rbrace)
return s
}
func (p *parser) caseClause() *CaseClause {
if trace {
defer p.trace("caseClause")()
}
c := new(CaseClause)
c.pos = p.pos()
switch p.tok {
case _Case:
p.next()
c.Cases = p.exprList()
case _Default:
p.next()
default:
p.syntaxError("expecting case or default or }")
p.advance(_Colon, _Case, _Default, _Rbrace)
}
c.Colon = p.pos()
p.want(_Colon)
c.Body = p.stmtList()
return c
}
func (p *parser) commClause() *CommClause {
if trace {
defer p.trace("commClause")()
}
c := new(CommClause)
c.pos = p.pos()
switch p.tok {
case _Case:
p.next()
c.Comm = p.simpleStmt(nil, 0)
// The syntax restricts the possible simple statements here to:
//
// lhs <- x (send statement)
// <-x
// lhs = <-x
// lhs := <-x
//
// All these (and more) are recognized by simpleStmt and invalid
// syntax trees are flagged later, during type checking.
// TODO(gri) eventually may want to restrict valid syntax trees
// here.
case _Default:
p.next()
default:
p.syntaxError("expecting case or default or }")
p.advance(_Colon, _Case, _Default, _Rbrace)
}
c.Colon = p.pos()
p.want(_Colon)
c.Body = p.stmtList()
return c
}
// stmtOrNil parses a statement if one is present, or else returns nil.
//
// Statement =
// Declaration | LabeledStmt | SimpleStmt |
// GoStmt | ReturnStmt | BreakStmt | ContinueStmt | GotoStmt |
// FallthroughStmt | Block | IfStmt | SwitchStmt | SelectStmt | ForStmt |
// DeferStmt .
func (p *parser) stmtOrNil() Stmt {
if trace {
defer p.trace("stmt " + p.tok.String())()
}
// Most statements (assignments) start with an identifier;
// look for it first before doing anything more expensive.
if p.tok == _Name {
p.clearPragma()
lhs := p.exprList()
if label, ok := lhs.(*Name); ok && p.tok == _Colon {
return p.labeledStmtOrNil(label)
}
return p.simpleStmt(lhs, 0)
}
switch p.tok {
case _Var:
return p.declStmt(p.varDecl)
case _Const:
return p.declStmt(p.constDecl)
case _Type:
return p.declStmt(p.typeDecl)
}
p.clearPragma()
switch p.tok {
case _Lbrace:
return p.blockStmt("")
case _Operator, _Star:
switch p.op {
case Add, Sub, Mul, And, Xor, Not:
return p.simpleStmt(nil, 0) // unary operators
}
case _Literal, _Func, _Lparen, // operands
_Lbrack, _Struct, _Map, _Chan, _Interface, // composite types
_Arrow: // receive operator
return p.simpleStmt(nil, 0)
case _For:
return p.forStmt()
case _Switch:
return p.switchStmt()
case _Select:
return p.selectStmt()
case _If:
return p.ifStmt()
case _Fallthrough:
s := new(BranchStmt)
s.pos = p.pos()
p.next()
s.Tok = _Fallthrough
return s
case _Break, _Continue:
s := new(BranchStmt)
s.pos = p.pos()
s.Tok = p.tok
p.next()
if p.tok == _Name {
s.Label = p.name()
}
return s
case _Go, _Defer:
return p.callStmt()
case _Goto:
s := new(BranchStmt)
s.pos = p.pos()
s.Tok = _Goto
p.next()
s.Label = p.name()
return s
case _Return:
s := new(ReturnStmt)
s.pos = p.pos()
p.next()
if p.tok != _Semi && p.tok != _Rbrace {
s.Results = p.exprList()
}
return s
case _Semi:
s := new(EmptyStmt)
s.pos = p.pos()
return s
}
return nil
}
// StatementList = { Statement ";" } .
func (p *parser) stmtList() (l []Stmt) {
if trace {
defer p.trace("stmtList")()
}
for p.tok != _EOF && p.tok != _Rbrace && p.tok != _Case && p.tok != _Default {
s := p.stmtOrNil()
p.clearPragma()
if s == nil {
break
}
l = append(l, s)
// ";" is optional before "}"
if !p.got(_Semi) && p.tok != _Rbrace {
p.syntaxError("at end of statement")
p.advance(_Semi, _Rbrace, _Case, _Default)
p.got(_Semi) // avoid spurious empty statement
}
}
return
}
// argList parses a possibly empty, comma-separated list of arguments,
// optionally followed by a comma (if not empty), and closed by ")".
// The last argument may be followed by "...".
//
// argList = [ arg { "," arg } [ "..." ] [ "," ] ] ")" .
func (p *parser) argList() (list []Expr, hasDots bool) {
if trace {
defer p.trace("argList")()
}
p.xnest++
p.list("argument list", _Comma, _Rparen, func() bool {
list = append(list, p.expr())
hasDots = p.got(_DotDotDot)
return hasDots
})
p.xnest--
return
}
// ----------------------------------------------------------------------------
// Common productions
func (p *parser) name() *Name {
// no tracing to avoid overly verbose output
if p.tok == _Name {
n := NewName(p.pos(), p.lit)
p.next()
return n
}
n := NewName(p.pos(), "_")
p.syntaxError("expecting name")
p.advance()
return n
}
// IdentifierList = identifier { "," identifier } .
// The first name must be provided.
func (p *parser) nameList(first *Name) []*Name {
if trace {
defer p.trace("nameList")()
}
if debug && first == nil {
panic("first name not provided")
}
l := []*Name{first}
for p.got(_Comma) {
l = append(l, p.name())
}
return l
}
// The first name may be provided, or nil.
func (p *parser) qualifiedName(name *Name) Expr {
if trace {
defer p.trace("qualifiedName")()
}
var x Expr
switch {
case name != nil:
x = name
case p.tok == _Name:
x = p.name()
default:
x = NewName(p.pos(), "_")
p.syntaxError("expecting name")
p.advance(_Dot, _Semi, _Rbrace)
}
if p.tok == _Dot {
s := new(SelectorExpr)
s.pos = p.pos()
p.next()
s.X = x
s.Sel = p.name()
x = s
}
if p.tok == _Lbrack {
x = p.typeInstance(x)
}
return x
}
// ExpressionList = Expression { "," Expression } .
func (p *parser) exprList() Expr {
if trace {
defer p.trace("exprList")()
}
x := p.expr()
if p.got(_Comma) {
list := []Expr{x, p.expr()}
for p.got(_Comma) {
list = append(list, p.expr())
}
t := new(ListExpr)
t.pos = x.Pos()
t.ElemList = list
x = t
}
return x
}
// typeList parses a non-empty, comma-separated list of expressions,
// optionally followed by a comma. The first list element may be any
// expression, all other list elements must be type expressions.
// If there is more than one argument, the result is a *ListExpr.
// The comma result indicates whether there was a (separating or
// trailing) comma.
//
// typeList = arg { "," arg } [ "," ] .
func (p *parser) typeList() (x Expr, comma bool) {
if trace {
defer p.trace("typeList")()
}
p.xnest++
x = p.expr()
if p.got(_Comma) {
comma = true
if t := p.typeOrNil(); t != nil {
list := []Expr{x, t}
for p.got(_Comma) {
if t = p.typeOrNil(); t == nil {
break
}
list = append(list, t)
}
l := new(ListExpr)
l.pos = x.Pos() // == list[0].Pos()
l.ElemList = list
x = l
}
}
p.xnest--
return
}
// unparen removes all parentheses around an expression.
func unparen(x Expr) Expr {
for {
p, ok := x.(*ParenExpr)
if !ok {
break
}
x = p.X
}
return x
}