httpserver/all: Clean up and standardize request URL handling (#1633)
* httpserver/all: Clean up and standardize request URL handling
The HTTP server now always creates a context value on the request which
is a copy of the request's URL struct. It should not be modified by
middlewares, but it is safe to get the value out of the request and make
changes to it locally-scoped. Thus, the value in the context always
stores the original request URL information as it was received. Any
rewrites that happen will be to the request's URL field directly.
The HTTP server no longer cleans /sanitizes the request URL. It made too
many strong assumptions and ended up making a lot of middleware more
complicated, including upstream proxying (and fastcgi). To alleviate
this complexity, we no longer change the request URL. Middlewares are
responsible to access the disk safely by using http.Dir or, if not
actually opening files, they can use httpserver.SafePath().
I'm hoping this will address issues with #1624, #1584, #1582, and others.
* staticfiles: Fix test on Windows
@abiosoft: I still can't figure out exactly what this is for. 😅
* Use (potentially) changed URL for browse redirects, as before
* Use filepath.ToSlash, clean up a couple proxy test cases
* Oops, fix variable name
2017-05-02 13:11:10 +08:00
|
|
|
|
// Package staticfiles provides middleware for serving static files from disk.
|
|
|
|
|
// Its handler is the default HTTP handler for the HTTP server.
|
|
|
|
|
//
|
|
|
|
|
// TODO: Should this package be rolled into the httpserver package?
|
Rewrote Caddy from the ground up; initial commit of 0.9 branch
These changes span work from the last ~4 months in an effort to make
Caddy more extensible, reduce the coupling between its components, and
lay a more robust foundation of code going forward into 1.0. A bunch of
new features have been added, too, with even higher future potential.
The most significant design change is an overall inversion of
dependencies. Instead of the caddy package knowing about the server
and the notion of middleware and config, the caddy package exposes an
interface that other components plug into. This does introduce more
indirection when reading the code, but every piece is very modular and
pluggable. Even the HTTP server is pluggable.
The caddy package has been moved to the top level, and main has been
pushed into a subfolder called caddy. The actual logic of the main
file has been pushed even further into caddy/caddymain/run.go so that
custom builds of Caddy can be 'go get'able.
The HTTPS logic was surgically separated into two parts to divide the
TLS-specific code and the HTTPS-specific code. The caddytls package can
now be used by any type of server that needs TLS, not just HTTP. I also
added the ability to customize nearly every aspect of TLS at the site
level rather than all sites sharing the same TLS configuration. Not all
of this flexibility is exposed in the Caddyfile yet, but it may be in
the future. Caddy can also generate self-signed certificates in memory
for the convenience of a developer working on localhost who wants HTTPS.
And Caddy now supports the DNS challenge, assuming at least one DNS
provider is plugged in.
Dozens, if not hundreds, of other minor changes swept through the code
base as I literally started from an empty main function, copying over
functions or files as needed, then adjusting them to fit in the new
design. Most tests have been restored and adapted to the new API,
but more work is needed there.
A lot of what was "impossible" before is now possible, or can be made
possible with minimal disruption of the code. For example, it's fairly
easy to make plugins hook into another part of the code via callbacks.
Plugins can do more than just be directives; we now have plugins that
customize how the Caddyfile is loaded (useful when you need to get your
configuration from a remote store).
Site addresses no longer need be just a host and port. They can have a
path, allowing you to scope a configuration to a specific path. There is
no inheretance, however; each site configuration is distinct.
Thanks to amazing work by Lucas Clemente, this commit adds experimental
QUIC support. Turn it on using the -quic flag; your browser may have
to be configured to enable it.
Almost everything is here, but you will notice that most of the middle-
ware are missing. After those are transferred over, we'll be ready for
beta tests.
I'm very excited to get this out. Thanks for everyone's help and
patience these last few months. I hope you like it!!
2016-06-05 07:00:29 +08:00
|
|
|
|
package staticfiles
|
2015-03-25 11:55:51 +08:00
|
|
|
|
|
|
|
|
|
import (
|
2016-04-17 02:59:24 +08:00
|
|
|
|
"math/rand"
|
2015-03-25 11:55:51 +08:00
|
|
|
|
"net/http"
|
2015-03-29 06:49:42 +08:00
|
|
|
|
"os"
|
httpserver/all: Clean up and standardize request URL handling (#1633)
* httpserver/all: Clean up and standardize request URL handling
The HTTP server now always creates a context value on the request which
is a copy of the request's URL struct. It should not be modified by
middlewares, but it is safe to get the value out of the request and make
changes to it locally-scoped. Thus, the value in the context always
stores the original request URL information as it was received. Any
rewrites that happen will be to the request's URL field directly.
The HTTP server no longer cleans /sanitizes the request URL. It made too
many strong assumptions and ended up making a lot of middleware more
complicated, including upstream proxying (and fastcgi). To alleviate
this complexity, we no longer change the request URL. Middlewares are
responsible to access the disk safely by using http.Dir or, if not
actually opening files, they can use httpserver.SafePath().
I'm hoping this will address issues with #1624, #1584, #1582, and others.
* staticfiles: Fix test on Windows
@abiosoft: I still can't figure out exactly what this is for. 😅
* Use (potentially) changed URL for browse redirects, as before
* Use filepath.ToSlash, clean up a couple proxy test cases
* Oops, fix variable name
2017-05-02 13:11:10 +08:00
|
|
|
|
"path"
|
2016-03-11 22:39:13 +08:00
|
|
|
|
"path/filepath"
|
Rewrote Caddy from the ground up; initial commit of 0.9 branch
These changes span work from the last ~4 months in an effort to make
Caddy more extensible, reduce the coupling between its components, and
lay a more robust foundation of code going forward into 1.0. A bunch of
new features have been added, too, with even higher future potential.
The most significant design change is an overall inversion of
dependencies. Instead of the caddy package knowing about the server
and the notion of middleware and config, the caddy package exposes an
interface that other components plug into. This does introduce more
indirection when reading the code, but every piece is very modular and
pluggable. Even the HTTP server is pluggable.
The caddy package has been moved to the top level, and main has been
pushed into a subfolder called caddy. The actual logic of the main
file has been pushed even further into caddy/caddymain/run.go so that
custom builds of Caddy can be 'go get'able.
The HTTPS logic was surgically separated into two parts to divide the
TLS-specific code and the HTTPS-specific code. The caddytls package can
now be used by any type of server that needs TLS, not just HTTP. I also
added the ability to customize nearly every aspect of TLS at the site
level rather than all sites sharing the same TLS configuration. Not all
of this flexibility is exposed in the Caddyfile yet, but it may be in
the future. Caddy can also generate self-signed certificates in memory
for the convenience of a developer working on localhost who wants HTTPS.
And Caddy now supports the DNS challenge, assuming at least one DNS
provider is plugged in.
Dozens, if not hundreds, of other minor changes swept through the code
base as I literally started from an empty main function, copying over
functions or files as needed, then adjusting them to fit in the new
design. Most tests have been restored and adapted to the new API,
but more work is needed there.
A lot of what was "impossible" before is now possible, or can be made
possible with minimal disruption of the code. For example, it's fairly
easy to make plugins hook into another part of the code via callbacks.
Plugins can do more than just be directives; we now have plugins that
customize how the Caddyfile is loaded (useful when you need to get your
configuration from a remote store).
Site addresses no longer need be just a host and port. They can have a
path, allowing you to scope a configuration to a specific path. There is
no inheretance, however; each site configuration is distinct.
Thanks to amazing work by Lucas Clemente, this commit adds experimental
QUIC support. Turn it on using the -quic flag; your browser may have
to be configured to enable it.
Almost everything is here, but you will notice that most of the middle-
ware are missing. After those are transferred over, we'll be ready for
beta tests.
I'm very excited to get this out. Thanks for everyone's help and
patience these last few months. I hope you like it!!
2016-06-05 07:00:29 +08:00
|
|
|
|
"runtime"
|
2016-04-17 02:59:24 +08:00
|
|
|
|
"strconv"
|
2015-03-25 11:55:51 +08:00
|
|
|
|
"strings"
|
2017-02-28 20:54:12 +08:00
|
|
|
|
|
|
|
|
|
"github.com/mholt/caddy"
|
2015-03-25 11:55:51 +08:00
|
|
|
|
)
|
|
|
|
|
|
2015-09-20 10:34:23 +08:00
|
|
|
|
// FileServer implements a production-ready file server
|
|
|
|
|
// and is the 'default' handler for all requests to Caddy.
|
Rewrote Caddy from the ground up; initial commit of 0.9 branch
These changes span work from the last ~4 months in an effort to make
Caddy more extensible, reduce the coupling between its components, and
lay a more robust foundation of code going forward into 1.0. A bunch of
new features have been added, too, with even higher future potential.
The most significant design change is an overall inversion of
dependencies. Instead of the caddy package knowing about the server
and the notion of middleware and config, the caddy package exposes an
interface that other components plug into. This does introduce more
indirection when reading the code, but every piece is very modular and
pluggable. Even the HTTP server is pluggable.
The caddy package has been moved to the top level, and main has been
pushed into a subfolder called caddy. The actual logic of the main
file has been pushed even further into caddy/caddymain/run.go so that
custom builds of Caddy can be 'go get'able.
The HTTPS logic was surgically separated into two parts to divide the
TLS-specific code and the HTTPS-specific code. The caddytls package can
now be used by any type of server that needs TLS, not just HTTP. I also
added the ability to customize nearly every aspect of TLS at the site
level rather than all sites sharing the same TLS configuration. Not all
of this flexibility is exposed in the Caddyfile yet, but it may be in
the future. Caddy can also generate self-signed certificates in memory
for the convenience of a developer working on localhost who wants HTTPS.
And Caddy now supports the DNS challenge, assuming at least one DNS
provider is plugged in.
Dozens, if not hundreds, of other minor changes swept through the code
base as I literally started from an empty main function, copying over
functions or files as needed, then adjusting them to fit in the new
design. Most tests have been restored and adapted to the new API,
but more work is needed there.
A lot of what was "impossible" before is now possible, or can be made
possible with minimal disruption of the code. For example, it's fairly
easy to make plugins hook into another part of the code via callbacks.
Plugins can do more than just be directives; we now have plugins that
customize how the Caddyfile is loaded (useful when you need to get your
configuration from a remote store).
Site addresses no longer need be just a host and port. They can have a
path, allowing you to scope a configuration to a specific path. There is
no inheretance, however; each site configuration is distinct.
Thanks to amazing work by Lucas Clemente, this commit adds experimental
QUIC support. Turn it on using the -quic flag; your browser may have
to be configured to enable it.
Almost everything is here, but you will notice that most of the middle-
ware are missing. After those are transferred over, we'll be ready for
beta tests.
I'm very excited to get this out. Thanks for everyone's help and
patience these last few months. I hope you like it!!
2016-06-05 07:00:29 +08:00
|
|
|
|
// It simply loads and serves the URI requested. FileServer
|
|
|
|
|
// is adapted from the one in net/http by the Go authors.
|
|
|
|
|
// Significant modifications have been made.
|
2015-04-13 07:44:02 +08:00
|
|
|
|
//
|
2015-09-20 10:34:23 +08:00
|
|
|
|
// Original license:
|
2015-03-25 11:55:51 +08:00
|
|
|
|
//
|
|
|
|
|
// Copyright 2009 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.
|
Rewrote Caddy from the ground up; initial commit of 0.9 branch
These changes span work from the last ~4 months in an effort to make
Caddy more extensible, reduce the coupling between its components, and
lay a more robust foundation of code going forward into 1.0. A bunch of
new features have been added, too, with even higher future potential.
The most significant design change is an overall inversion of
dependencies. Instead of the caddy package knowing about the server
and the notion of middleware and config, the caddy package exposes an
interface that other components plug into. This does introduce more
indirection when reading the code, but every piece is very modular and
pluggable. Even the HTTP server is pluggable.
The caddy package has been moved to the top level, and main has been
pushed into a subfolder called caddy. The actual logic of the main
file has been pushed even further into caddy/caddymain/run.go so that
custom builds of Caddy can be 'go get'able.
The HTTPS logic was surgically separated into two parts to divide the
TLS-specific code and the HTTPS-specific code. The caddytls package can
now be used by any type of server that needs TLS, not just HTTP. I also
added the ability to customize nearly every aspect of TLS at the site
level rather than all sites sharing the same TLS configuration. Not all
of this flexibility is exposed in the Caddyfile yet, but it may be in
the future. Caddy can also generate self-signed certificates in memory
for the convenience of a developer working on localhost who wants HTTPS.
And Caddy now supports the DNS challenge, assuming at least one DNS
provider is plugged in.
Dozens, if not hundreds, of other minor changes swept through the code
base as I literally started from an empty main function, copying over
functions or files as needed, then adjusting them to fit in the new
design. Most tests have been restored and adapted to the new API,
but more work is needed there.
A lot of what was "impossible" before is now possible, or can be made
possible with minimal disruption of the code. For example, it's fairly
easy to make plugins hook into another part of the code via callbacks.
Plugins can do more than just be directives; we now have plugins that
customize how the Caddyfile is loaded (useful when you need to get your
configuration from a remote store).
Site addresses no longer need be just a host and port. They can have a
path, allowing you to scope a configuration to a specific path. There is
no inheretance, however; each site configuration is distinct.
Thanks to amazing work by Lucas Clemente, this commit adds experimental
QUIC support. Turn it on using the -quic flag; your browser may have
to be configured to enable it.
Almost everything is here, but you will notice that most of the middle-
ware are missing. After those are transferred over, we'll be ready for
beta tests.
I'm very excited to get this out. Thanks for everyone's help and
patience these last few months. I hope you like it!!
2016-06-05 07:00:29 +08:00
|
|
|
|
type FileServer struct {
|
httpserver/all: Clean up and standardize request URL handling (#1633)
* httpserver/all: Clean up and standardize request URL handling
The HTTP server now always creates a context value on the request which
is a copy of the request's URL struct. It should not be modified by
middlewares, but it is safe to get the value out of the request and make
changes to it locally-scoped. Thus, the value in the context always
stores the original request URL information as it was received. Any
rewrites that happen will be to the request's URL field directly.
The HTTP server no longer cleans /sanitizes the request URL. It made too
many strong assumptions and ended up making a lot of middleware more
complicated, including upstream proxying (and fastcgi). To alleviate
this complexity, we no longer change the request URL. Middlewares are
responsible to access the disk safely by using http.Dir or, if not
actually opening files, they can use httpserver.SafePath().
I'm hoping this will address issues with #1624, #1584, #1582, and others.
* staticfiles: Fix test on Windows
@abiosoft: I still can't figure out exactly what this is for. 😅
* Use (potentially) changed URL for browse redirects, as before
* Use filepath.ToSlash, clean up a couple proxy test cases
* Oops, fix variable name
2017-05-02 13:11:10 +08:00
|
|
|
|
Root http.FileSystem // jailed access to the file system
|
|
|
|
|
Hide []string // list of files for which to respond with "Not Found"
|
2015-03-25 11:55:51 +08:00
|
|
|
|
}
|
|
|
|
|
|
Rewrote Caddy from the ground up; initial commit of 0.9 branch
These changes span work from the last ~4 months in an effort to make
Caddy more extensible, reduce the coupling between its components, and
lay a more robust foundation of code going forward into 1.0. A bunch of
new features have been added, too, with even higher future potential.
The most significant design change is an overall inversion of
dependencies. Instead of the caddy package knowing about the server
and the notion of middleware and config, the caddy package exposes an
interface that other components plug into. This does introduce more
indirection when reading the code, but every piece is very modular and
pluggable. Even the HTTP server is pluggable.
The caddy package has been moved to the top level, and main has been
pushed into a subfolder called caddy. The actual logic of the main
file has been pushed even further into caddy/caddymain/run.go so that
custom builds of Caddy can be 'go get'able.
The HTTPS logic was surgically separated into two parts to divide the
TLS-specific code and the HTTPS-specific code. The caddytls package can
now be used by any type of server that needs TLS, not just HTTP. I also
added the ability to customize nearly every aspect of TLS at the site
level rather than all sites sharing the same TLS configuration. Not all
of this flexibility is exposed in the Caddyfile yet, but it may be in
the future. Caddy can also generate self-signed certificates in memory
for the convenience of a developer working on localhost who wants HTTPS.
And Caddy now supports the DNS challenge, assuming at least one DNS
provider is plugged in.
Dozens, if not hundreds, of other minor changes swept through the code
base as I literally started from an empty main function, copying over
functions or files as needed, then adjusting them to fit in the new
design. Most tests have been restored and adapted to the new API,
but more work is needed there.
A lot of what was "impossible" before is now possible, or can be made
possible with minimal disruption of the code. For example, it's fairly
easy to make plugins hook into another part of the code via callbacks.
Plugins can do more than just be directives; we now have plugins that
customize how the Caddyfile is loaded (useful when you need to get your
configuration from a remote store).
Site addresses no longer need be just a host and port. They can have a
path, allowing you to scope a configuration to a specific path. There is
no inheretance, however; each site configuration is distinct.
Thanks to amazing work by Lucas Clemente, this commit adds experimental
QUIC support. Turn it on using the -quic flag; your browser may have
to be configured to enable it.
Almost everything is here, but you will notice that most of the middle-
ware are missing. After those are transferred over, we'll be ready for
beta tests.
I'm very excited to get this out. Thanks for everyone's help and
patience these last few months. I hope you like it!!
2016-06-05 07:00:29 +08:00
|
|
|
|
// ServeHTTP serves static files for r according to fs's configuration.
|
|
|
|
|
func (fs FileServer) ServeHTTP(w http.ResponseWriter, r *http.Request) (int, error) {
|
httpserver/all: Clean up and standardize request URL handling (#1633)
* httpserver/all: Clean up and standardize request URL handling
The HTTP server now always creates a context value on the request which
is a copy of the request's URL struct. It should not be modified by
middlewares, but it is safe to get the value out of the request and make
changes to it locally-scoped. Thus, the value in the context always
stores the original request URL information as it was received. Any
rewrites that happen will be to the request's URL field directly.
The HTTP server no longer cleans /sanitizes the request URL. It made too
many strong assumptions and ended up making a lot of middleware more
complicated, including upstream proxying (and fastcgi). To alleviate
this complexity, we no longer change the request URL. Middlewares are
responsible to access the disk safely by using http.Dir or, if not
actually opening files, they can use httpserver.SafePath().
I'm hoping this will address issues with #1624, #1584, #1582, and others.
* staticfiles: Fix test on Windows
@abiosoft: I still can't figure out exactly what this is for. 😅
* Use (potentially) changed URL for browse redirects, as before
* Use filepath.ToSlash, clean up a couple proxy test cases
* Oops, fix variable name
2017-05-02 13:11:10 +08:00
|
|
|
|
return fs.serveFile(w, r)
|
2017-02-19 06:52:50 +08:00
|
|
|
|
}
|
|
|
|
|
|
2015-05-25 10:52:34 +08:00
|
|
|
|
// serveFile writes the specified file to the HTTP response.
|
2015-03-25 11:55:51 +08:00
|
|
|
|
// name is '/'-separated, not filepath.Separator.
|
httpserver/all: Clean up and standardize request URL handling (#1633)
* httpserver/all: Clean up and standardize request URL handling
The HTTP server now always creates a context value on the request which
is a copy of the request's URL struct. It should not be modified by
middlewares, but it is safe to get the value out of the request and make
changes to it locally-scoped. Thus, the value in the context always
stores the original request URL information as it was received. Any
rewrites that happen will be to the request's URL field directly.
The HTTP server no longer cleans /sanitizes the request URL. It made too
many strong assumptions and ended up making a lot of middleware more
complicated, including upstream proxying (and fastcgi). To alleviate
this complexity, we no longer change the request URL. Middlewares are
responsible to access the disk safely by using http.Dir or, if not
actually opening files, they can use httpserver.SafePath().
I'm hoping this will address issues with #1624, #1584, #1582, and others.
* staticfiles: Fix test on Windows
@abiosoft: I still can't figure out exactly what this is for. 😅
* Use (potentially) changed URL for browse redirects, as before
* Use filepath.ToSlash, clean up a couple proxy test cases
* Oops, fix variable name
2017-05-02 13:11:10 +08:00
|
|
|
|
func (fs FileServer) serveFile(w http.ResponseWriter, r *http.Request) (int, error) {
|
|
|
|
|
reqPath := r.URL.Path
|
2016-12-20 00:51:09 +08:00
|
|
|
|
|
2016-03-12 06:11:21 +08:00
|
|
|
|
// Prevent absolute path access on Windows.
|
|
|
|
|
// TODO remove when stdlib http.Dir fixes this.
|
httpserver/all: Clean up and standardize request URL handling (#1633)
* httpserver/all: Clean up and standardize request URL handling
The HTTP server now always creates a context value on the request which
is a copy of the request's URL struct. It should not be modified by
middlewares, but it is safe to get the value out of the request and make
changes to it locally-scoped. Thus, the value in the context always
stores the original request URL information as it was received. Any
rewrites that happen will be to the request's URL field directly.
The HTTP server no longer cleans /sanitizes the request URL. It made too
many strong assumptions and ended up making a lot of middleware more
complicated, including upstream proxying (and fastcgi). To alleviate
this complexity, we no longer change the request URL. Middlewares are
responsible to access the disk safely by using http.Dir or, if not
actually opening files, they can use httpserver.SafePath().
I'm hoping this will address issues with #1624, #1584, #1582, and others.
* staticfiles: Fix test on Windows
@abiosoft: I still can't figure out exactly what this is for. 😅
* Use (potentially) changed URL for browse redirects, as before
* Use filepath.ToSlash, clean up a couple proxy test cases
* Oops, fix variable name
2017-05-02 13:11:10 +08:00
|
|
|
|
if runtime.GOOS == "windows" && len(reqPath) > 0 && filepath.IsAbs(reqPath[1:]) {
|
|
|
|
|
return http.StatusNotFound, nil
|
2016-03-12 06:11:21 +08:00
|
|
|
|
}
|
Rewrote Caddy from the ground up; initial commit of 0.9 branch
These changes span work from the last ~4 months in an effort to make
Caddy more extensible, reduce the coupling between its components, and
lay a more robust foundation of code going forward into 1.0. A bunch of
new features have been added, too, with even higher future potential.
The most significant design change is an overall inversion of
dependencies. Instead of the caddy package knowing about the server
and the notion of middleware and config, the caddy package exposes an
interface that other components plug into. This does introduce more
indirection when reading the code, but every piece is very modular and
pluggable. Even the HTTP server is pluggable.
The caddy package has been moved to the top level, and main has been
pushed into a subfolder called caddy. The actual logic of the main
file has been pushed even further into caddy/caddymain/run.go so that
custom builds of Caddy can be 'go get'able.
The HTTPS logic was surgically separated into two parts to divide the
TLS-specific code and the HTTPS-specific code. The caddytls package can
now be used by any type of server that needs TLS, not just HTTP. I also
added the ability to customize nearly every aspect of TLS at the site
level rather than all sites sharing the same TLS configuration. Not all
of this flexibility is exposed in the Caddyfile yet, but it may be in
the future. Caddy can also generate self-signed certificates in memory
for the convenience of a developer working on localhost who wants HTTPS.
And Caddy now supports the DNS challenge, assuming at least one DNS
provider is plugged in.
Dozens, if not hundreds, of other minor changes swept through the code
base as I literally started from an empty main function, copying over
functions or files as needed, then adjusting them to fit in the new
design. Most tests have been restored and adapted to the new API,
but more work is needed there.
A lot of what was "impossible" before is now possible, or can be made
possible with minimal disruption of the code. For example, it's fairly
easy to make plugins hook into another part of the code via callbacks.
Plugins can do more than just be directives; we now have plugins that
customize how the Caddyfile is loaded (useful when you need to get your
configuration from a remote store).
Site addresses no longer need be just a host and port. They can have a
path, allowing you to scope a configuration to a specific path. There is
no inheretance, however; each site configuration is distinct.
Thanks to amazing work by Lucas Clemente, this commit adds experimental
QUIC support. Turn it on using the -quic flag; your browser may have
to be configured to enable it.
Almost everything is here, but you will notice that most of the middle-
ware are missing. After those are transferred over, we'll be ready for
beta tests.
I'm very excited to get this out. Thanks for everyone's help and
patience these last few months. I hope you like it!!
2016-06-05 07:00:29 +08:00
|
|
|
|
|
httpserver/all: Clean up and standardize request URL handling (#1633)
* httpserver/all: Clean up and standardize request URL handling
The HTTP server now always creates a context value on the request which
is a copy of the request's URL struct. It should not be modified by
middlewares, but it is safe to get the value out of the request and make
changes to it locally-scoped. Thus, the value in the context always
stores the original request URL information as it was received. Any
rewrites that happen will be to the request's URL field directly.
The HTTP server no longer cleans /sanitizes the request URL. It made too
many strong assumptions and ended up making a lot of middleware more
complicated, including upstream proxying (and fastcgi). To alleviate
this complexity, we no longer change the request URL. Middlewares are
responsible to access the disk safely by using http.Dir or, if not
actually opening files, they can use httpserver.SafePath().
I'm hoping this will address issues with #1624, #1584, #1582, and others.
* staticfiles: Fix test on Windows
@abiosoft: I still can't figure out exactly what this is for. 😅
* Use (potentially) changed URL for browse redirects, as before
* Use filepath.ToSlash, clean up a couple proxy test cases
* Oops, fix variable name
2017-05-02 13:11:10 +08:00
|
|
|
|
// open the requested file
|
|
|
|
|
f, err := fs.Root.Open(reqPath)
|
2015-03-25 11:55:51 +08:00
|
|
|
|
if err != nil {
|
httpserver/all: Clean up and standardize request URL handling (#1633)
* httpserver/all: Clean up and standardize request URL handling
The HTTP server now always creates a context value on the request which
is a copy of the request's URL struct. It should not be modified by
middlewares, but it is safe to get the value out of the request and make
changes to it locally-scoped. Thus, the value in the context always
stores the original request URL information as it was received. Any
rewrites that happen will be to the request's URL field directly.
The HTTP server no longer cleans /sanitizes the request URL. It made too
many strong assumptions and ended up making a lot of middleware more
complicated, including upstream proxying (and fastcgi). To alleviate
this complexity, we no longer change the request URL. Middlewares are
responsible to access the disk safely by using http.Dir or, if not
actually opening files, they can use httpserver.SafePath().
I'm hoping this will address issues with #1624, #1584, #1582, and others.
* staticfiles: Fix test on Windows
@abiosoft: I still can't figure out exactly what this is for. 😅
* Use (potentially) changed URL for browse redirects, as before
* Use filepath.ToSlash, clean up a couple proxy test cases
* Oops, fix variable name
2017-05-02 13:11:10 +08:00
|
|
|
|
// TODO: remove when http.Dir handles this (Go 1.9?)
|
2017-02-11 11:02:00 +08:00
|
|
|
|
// Go issue #18984
|
|
|
|
|
err = mapFSRootOpenErr(err)
|
2015-05-01 00:14:58 +08:00
|
|
|
|
if os.IsNotExist(err) {
|
|
|
|
|
return http.StatusNotFound, nil
|
2015-05-01 01:58:38 +08:00
|
|
|
|
} else if os.IsPermission(err) {
|
|
|
|
|
return http.StatusForbidden, err
|
2015-03-29 06:49:42 +08:00
|
|
|
|
}
|
httpserver/all: Clean up and standardize request URL handling (#1633)
* httpserver/all: Clean up and standardize request URL handling
The HTTP server now always creates a context value on the request which
is a copy of the request's URL struct. It should not be modified by
middlewares, but it is safe to get the value out of the request and make
changes to it locally-scoped. Thus, the value in the context always
stores the original request URL information as it was received. Any
rewrites that happen will be to the request's URL field directly.
The HTTP server no longer cleans /sanitizes the request URL. It made too
many strong assumptions and ended up making a lot of middleware more
complicated, including upstream proxying (and fastcgi). To alleviate
this complexity, we no longer change the request URL. Middlewares are
responsible to access the disk safely by using http.Dir or, if not
actually opening files, they can use httpserver.SafePath().
I'm hoping this will address issues with #1624, #1584, #1582, and others.
* staticfiles: Fix test on Windows
@abiosoft: I still can't figure out exactly what this is for. 😅
* Use (potentially) changed URL for browse redirects, as before
* Use filepath.ToSlash, clean up a couple proxy test cases
* Oops, fix variable name
2017-05-02 13:11:10 +08:00
|
|
|
|
// otherwise, maybe the server is under load and ran out of file descriptors?
|
2016-04-17 02:59:24 +08:00
|
|
|
|
backoff := int(3 + rand.Int31()%3) // 3–5 seconds to prevent a stampede
|
|
|
|
|
w.Header().Set("Retry-After", strconv.Itoa(backoff))
|
2015-05-01 00:14:58 +08:00
|
|
|
|
return http.StatusServiceUnavailable, err
|
2015-03-25 11:55:51 +08:00
|
|
|
|
}
|
|
|
|
|
defer f.Close()
|
|
|
|
|
|
httpserver/all: Clean up and standardize request URL handling (#1633)
* httpserver/all: Clean up and standardize request URL handling
The HTTP server now always creates a context value on the request which
is a copy of the request's URL struct. It should not be modified by
middlewares, but it is safe to get the value out of the request and make
changes to it locally-scoped. Thus, the value in the context always
stores the original request URL information as it was received. Any
rewrites that happen will be to the request's URL field directly.
The HTTP server no longer cleans /sanitizes the request URL. It made too
many strong assumptions and ended up making a lot of middleware more
complicated, including upstream proxying (and fastcgi). To alleviate
this complexity, we no longer change the request URL. Middlewares are
responsible to access the disk safely by using http.Dir or, if not
actually opening files, they can use httpserver.SafePath().
I'm hoping this will address issues with #1624, #1584, #1582, and others.
* staticfiles: Fix test on Windows
@abiosoft: I still can't figure out exactly what this is for. 😅
* Use (potentially) changed URL for browse redirects, as before
* Use filepath.ToSlash, clean up a couple proxy test cases
* Oops, fix variable name
2017-05-02 13:11:10 +08:00
|
|
|
|
// get information about the file
|
2015-10-21 06:25:38 +08:00
|
|
|
|
d, err := f.Stat()
|
|
|
|
|
if err != nil {
|
2015-05-01 01:58:38 +08:00
|
|
|
|
if os.IsNotExist(err) {
|
2015-05-01 00:14:58 +08:00
|
|
|
|
return http.StatusNotFound, nil
|
2015-05-01 01:58:38 +08:00
|
|
|
|
} else if os.IsPermission(err) {
|
|
|
|
|
return http.StatusForbidden, err
|
2015-03-29 06:49:42 +08:00
|
|
|
|
}
|
httpserver/all: Clean up and standardize request URL handling (#1633)
* httpserver/all: Clean up and standardize request URL handling
The HTTP server now always creates a context value on the request which
is a copy of the request's URL struct. It should not be modified by
middlewares, but it is safe to get the value out of the request and make
changes to it locally-scoped. Thus, the value in the context always
stores the original request URL information as it was received. Any
rewrites that happen will be to the request's URL field directly.
The HTTP server no longer cleans /sanitizes the request URL. It made too
many strong assumptions and ended up making a lot of middleware more
complicated, including upstream proxying (and fastcgi). To alleviate
this complexity, we no longer change the request URL. Middlewares are
responsible to access the disk safely by using http.Dir or, if not
actually opening files, they can use httpserver.SafePath().
I'm hoping this will address issues with #1624, #1584, #1582, and others.
* staticfiles: Fix test on Windows
@abiosoft: I still can't figure out exactly what this is for. 😅
* Use (potentially) changed URL for browse redirects, as before
* Use filepath.ToSlash, clean up a couple proxy test cases
* Oops, fix variable name
2017-05-02 13:11:10 +08:00
|
|
|
|
// return a different status code than above to distinguish these cases
|
2015-05-01 00:14:58 +08:00
|
|
|
|
return http.StatusInternalServerError, err
|
2015-03-25 11:55:51 +08:00
|
|
|
|
}
|
|
|
|
|
|
httpserver/all: Clean up and standardize request URL handling (#1633)
* httpserver/all: Clean up and standardize request URL handling
The HTTP server now always creates a context value on the request which
is a copy of the request's URL struct. It should not be modified by
middlewares, but it is safe to get the value out of the request and make
changes to it locally-scoped. Thus, the value in the context always
stores the original request URL information as it was received. Any
rewrites that happen will be to the request's URL field directly.
The HTTP server no longer cleans /sanitizes the request URL. It made too
many strong assumptions and ended up making a lot of middleware more
complicated, including upstream proxying (and fastcgi). To alleviate
this complexity, we no longer change the request URL. Middlewares are
responsible to access the disk safely by using http.Dir or, if not
actually opening files, they can use httpserver.SafePath().
I'm hoping this will address issues with #1624, #1584, #1582, and others.
* staticfiles: Fix test on Windows
@abiosoft: I still can't figure out exactly what this is for. 😅
* Use (potentially) changed URL for browse redirects, as before
* Use filepath.ToSlash, clean up a couple proxy test cases
* Oops, fix variable name
2017-05-02 13:11:10 +08:00
|
|
|
|
// redirect to canonical path (being careful to preserve other parts of URL and
|
|
|
|
|
// considering cases where a site is defined with a path prefix that gets stripped)
|
2017-06-08 04:40:17 +08:00
|
|
|
|
urlCopy := *r.URL
|
|
|
|
|
pathPrefix, _ := r.Context().Value(caddy.CtxKey("path_prefix")).(string)
|
|
|
|
|
if pathPrefix != "/" {
|
|
|
|
|
urlCopy.Path = pathPrefix + urlCopy.Path
|
|
|
|
|
}
|
|
|
|
|
if urlCopy.Path == "" {
|
|
|
|
|
urlCopy.Path = "/"
|
httpserver/all: Clean up and standardize request URL handling (#1633)
* httpserver/all: Clean up and standardize request URL handling
The HTTP server now always creates a context value on the request which
is a copy of the request's URL struct. It should not be modified by
middlewares, but it is safe to get the value out of the request and make
changes to it locally-scoped. Thus, the value in the context always
stores the original request URL information as it was received. Any
rewrites that happen will be to the request's URL field directly.
The HTTP server no longer cleans /sanitizes the request URL. It made too
many strong assumptions and ended up making a lot of middleware more
complicated, including upstream proxying (and fastcgi). To alleviate
this complexity, we no longer change the request URL. Middlewares are
responsible to access the disk safely by using http.Dir or, if not
actually opening files, they can use httpserver.SafePath().
I'm hoping this will address issues with #1624, #1584, #1582, and others.
* staticfiles: Fix test on Windows
@abiosoft: I still can't figure out exactly what this is for. 😅
* Use (potentially) changed URL for browse redirects, as before
* Use filepath.ToSlash, clean up a couple proxy test cases
* Oops, fix variable name
2017-05-02 13:11:10 +08:00
|
|
|
|
}
|
2015-04-19 03:24:54 +08:00
|
|
|
|
if d.IsDir() {
|
httpserver/all: Clean up and standardize request URL handling (#1633)
* httpserver/all: Clean up and standardize request URL handling
The HTTP server now always creates a context value on the request which
is a copy of the request's URL struct. It should not be modified by
middlewares, but it is safe to get the value out of the request and make
changes to it locally-scoped. Thus, the value in the context always
stores the original request URL information as it was received. Any
rewrites that happen will be to the request's URL field directly.
The HTTP server no longer cleans /sanitizes the request URL. It made too
many strong assumptions and ended up making a lot of middleware more
complicated, including upstream proxying (and fastcgi). To alleviate
this complexity, we no longer change the request URL. Middlewares are
responsible to access the disk safely by using http.Dir or, if not
actually opening files, they can use httpserver.SafePath().
I'm hoping this will address issues with #1624, #1584, #1582, and others.
* staticfiles: Fix test on Windows
@abiosoft: I still can't figure out exactly what this is for. 😅
* Use (potentially) changed URL for browse redirects, as before
* Use filepath.ToSlash, clean up a couple proxy test cases
* Oops, fix variable name
2017-05-02 13:11:10 +08:00
|
|
|
|
// ensure there is a trailing slash
|
2017-06-08 04:40:17 +08:00
|
|
|
|
if urlCopy.Path[len(urlCopy.Path)-1] != '/' {
|
|
|
|
|
urlCopy.Path += "/"
|
|
|
|
|
http.Redirect(w, r, urlCopy.String(), http.StatusMovedPermanently)
|
2015-04-19 03:24:54 +08:00
|
|
|
|
return http.StatusMovedPermanently, nil
|
|
|
|
|
}
|
|
|
|
|
} else {
|
httpserver/all: Clean up and standardize request URL handling (#1633)
* httpserver/all: Clean up and standardize request URL handling
The HTTP server now always creates a context value on the request which
is a copy of the request's URL struct. It should not be modified by
middlewares, but it is safe to get the value out of the request and make
changes to it locally-scoped. Thus, the value in the context always
stores the original request URL information as it was received. Any
rewrites that happen will be to the request's URL field directly.
The HTTP server no longer cleans /sanitizes the request URL. It made too
many strong assumptions and ended up making a lot of middleware more
complicated, including upstream proxying (and fastcgi). To alleviate
this complexity, we no longer change the request URL. Middlewares are
responsible to access the disk safely by using http.Dir or, if not
actually opening files, they can use httpserver.SafePath().
I'm hoping this will address issues with #1624, #1584, #1582, and others.
* staticfiles: Fix test on Windows
@abiosoft: I still can't figure out exactly what this is for. 😅
* Use (potentially) changed URL for browse redirects, as before
* Use filepath.ToSlash, clean up a couple proxy test cases
* Oops, fix variable name
2017-05-02 13:11:10 +08:00
|
|
|
|
// ensure no trailing slash
|
|
|
|
|
redir := false
|
2017-06-08 04:40:17 +08:00
|
|
|
|
if urlCopy.Path[len(urlCopy.Path)-1] == '/' {
|
|
|
|
|
urlCopy.Path = urlCopy.Path[:len(urlCopy.Path)-1]
|
httpserver/all: Clean up and standardize request URL handling (#1633)
* httpserver/all: Clean up and standardize request URL handling
The HTTP server now always creates a context value on the request which
is a copy of the request's URL struct. It should not be modified by
middlewares, but it is safe to get the value out of the request and make
changes to it locally-scoped. Thus, the value in the context always
stores the original request URL information as it was received. Any
rewrites that happen will be to the request's URL field directly.
The HTTP server no longer cleans /sanitizes the request URL. It made too
many strong assumptions and ended up making a lot of middleware more
complicated, including upstream proxying (and fastcgi). To alleviate
this complexity, we no longer change the request URL. Middlewares are
responsible to access the disk safely by using http.Dir or, if not
actually opening files, they can use httpserver.SafePath().
I'm hoping this will address issues with #1624, #1584, #1582, and others.
* staticfiles: Fix test on Windows
@abiosoft: I still can't figure out exactly what this is for. 😅
* Use (potentially) changed URL for browse redirects, as before
* Use filepath.ToSlash, clean up a couple proxy test cases
* Oops, fix variable name
2017-05-02 13:11:10 +08:00
|
|
|
|
redir = true
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// if an index file was explicitly requested, strip file name from the request
|
|
|
|
|
// ("/foo/index.html" -> "/foo/")
|
|
|
|
|
for _, indexPage := range IndexPages {
|
2017-06-08 04:40:17 +08:00
|
|
|
|
if strings.HasSuffix(urlCopy.Path, indexPage) {
|
|
|
|
|
urlCopy.Path = urlCopy.Path[:len(urlCopy.Path)-len(indexPage)]
|
httpserver/all: Clean up and standardize request URL handling (#1633)
* httpserver/all: Clean up and standardize request URL handling
The HTTP server now always creates a context value on the request which
is a copy of the request's URL struct. It should not be modified by
middlewares, but it is safe to get the value out of the request and make
changes to it locally-scoped. Thus, the value in the context always
stores the original request URL information as it was received. Any
rewrites that happen will be to the request's URL field directly.
The HTTP server no longer cleans /sanitizes the request URL. It made too
many strong assumptions and ended up making a lot of middleware more
complicated, including upstream proxying (and fastcgi). To alleviate
this complexity, we no longer change the request URL. Middlewares are
responsible to access the disk safely by using http.Dir or, if not
actually opening files, they can use httpserver.SafePath().
I'm hoping this will address issues with #1624, #1584, #1582, and others.
* staticfiles: Fix test on Windows
@abiosoft: I still can't figure out exactly what this is for. 😅
* Use (potentially) changed URL for browse redirects, as before
* Use filepath.ToSlash, clean up a couple proxy test cases
* Oops, fix variable name
2017-05-02 13:11:10 +08:00
|
|
|
|
redir = true
|
|
|
|
|
break
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if redir {
|
2017-06-08 04:40:17 +08:00
|
|
|
|
http.Redirect(w, r, urlCopy.String(), http.StatusMovedPermanently)
|
2015-04-19 03:24:54 +08:00
|
|
|
|
return http.StatusMovedPermanently, nil
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
httpserver/all: Clean up and standardize request URL handling (#1633)
* httpserver/all: Clean up and standardize request URL handling
The HTTP server now always creates a context value on the request which
is a copy of the request's URL struct. It should not be modified by
middlewares, but it is safe to get the value out of the request and make
changes to it locally-scoped. Thus, the value in the context always
stores the original request URL information as it was received. Any
rewrites that happen will be to the request's URL field directly.
The HTTP server no longer cleans /sanitizes the request URL. It made too
many strong assumptions and ended up making a lot of middleware more
complicated, including upstream proxying (and fastcgi). To alleviate
this complexity, we no longer change the request URL. Middlewares are
responsible to access the disk safely by using http.Dir or, if not
actually opening files, they can use httpserver.SafePath().
I'm hoping this will address issues with #1624, #1584, #1582, and others.
* staticfiles: Fix test on Windows
@abiosoft: I still can't figure out exactly what this is for. 😅
* Use (potentially) changed URL for browse redirects, as before
* Use filepath.ToSlash, clean up a couple proxy test cases
* Oops, fix variable name
2017-05-02 13:11:10 +08:00
|
|
|
|
// use contents of an index file, if present, for directory requests
|
2015-03-25 11:55:51 +08:00
|
|
|
|
if d.IsDir() {
|
2015-09-20 10:34:23 +08:00
|
|
|
|
for _, indexPage := range IndexPages {
|
httpserver/all: Clean up and standardize request URL handling (#1633)
* httpserver/all: Clean up and standardize request URL handling
The HTTP server now always creates a context value on the request which
is a copy of the request's URL struct. It should not be modified by
middlewares, but it is safe to get the value out of the request and make
changes to it locally-scoped. Thus, the value in the context always
stores the original request URL information as it was received. Any
rewrites that happen will be to the request's URL field directly.
The HTTP server no longer cleans /sanitizes the request URL. It made too
many strong assumptions and ended up making a lot of middleware more
complicated, including upstream proxying (and fastcgi). To alleviate
this complexity, we no longer change the request URL. Middlewares are
responsible to access the disk safely by using http.Dir or, if not
actually opening files, they can use httpserver.SafePath().
I'm hoping this will address issues with #1624, #1584, #1582, and others.
* staticfiles: Fix test on Windows
@abiosoft: I still can't figure out exactly what this is for. 😅
* Use (potentially) changed URL for browse redirects, as before
* Use filepath.ToSlash, clean up a couple proxy test cases
* Oops, fix variable name
2017-05-02 13:11:10 +08:00
|
|
|
|
indexPath := path.Join(reqPath, indexPage)
|
|
|
|
|
indexFile, err := fs.Root.Open(indexPath)
|
2016-12-20 00:51:09 +08:00
|
|
|
|
if err != nil {
|
|
|
|
|
continue
|
2015-03-25 11:55:51 +08:00
|
|
|
|
}
|
2016-12-20 00:51:09 +08:00
|
|
|
|
|
httpserver/all: Clean up and standardize request URL handling (#1633)
* httpserver/all: Clean up and standardize request URL handling
The HTTP server now always creates a context value on the request which
is a copy of the request's URL struct. It should not be modified by
middlewares, but it is safe to get the value out of the request and make
changes to it locally-scoped. Thus, the value in the context always
stores the original request URL information as it was received. Any
rewrites that happen will be to the request's URL field directly.
The HTTP server no longer cleans /sanitizes the request URL. It made too
many strong assumptions and ended up making a lot of middleware more
complicated, including upstream proxying (and fastcgi). To alleviate
this complexity, we no longer change the request URL. Middlewares are
responsible to access the disk safely by using http.Dir or, if not
actually opening files, they can use httpserver.SafePath().
I'm hoping this will address issues with #1624, #1584, #1582, and others.
* staticfiles: Fix test on Windows
@abiosoft: I still can't figure out exactly what this is for. 😅
* Use (potentially) changed URL for browse redirects, as before
* Use filepath.ToSlash, clean up a couple proxy test cases
* Oops, fix variable name
2017-05-02 13:11:10 +08:00
|
|
|
|
indexInfo, err := indexFile.Stat()
|
2016-12-20 00:51:09 +08:00
|
|
|
|
if err != nil {
|
httpserver/all: Clean up and standardize request URL handling (#1633)
* httpserver/all: Clean up and standardize request URL handling
The HTTP server now always creates a context value on the request which
is a copy of the request's URL struct. It should not be modified by
middlewares, but it is safe to get the value out of the request and make
changes to it locally-scoped. Thus, the value in the context always
stores the original request URL information as it was received. Any
rewrites that happen will be to the request's URL field directly.
The HTTP server no longer cleans /sanitizes the request URL. It made too
many strong assumptions and ended up making a lot of middleware more
complicated, including upstream proxying (and fastcgi). To alleviate
this complexity, we no longer change the request URL. Middlewares are
responsible to access the disk safely by using http.Dir or, if not
actually opening files, they can use httpserver.SafePath().
I'm hoping this will address issues with #1624, #1584, #1582, and others.
* staticfiles: Fix test on Windows
@abiosoft: I still can't figure out exactly what this is for. 😅
* Use (potentially) changed URL for browse redirects, as before
* Use filepath.ToSlash, clean up a couple proxy test cases
* Oops, fix variable name
2017-05-02 13:11:10 +08:00
|
|
|
|
indexFile.Close()
|
2016-12-20 00:51:09 +08:00
|
|
|
|
continue
|
|
|
|
|
}
|
|
|
|
|
|
httpserver/all: Clean up and standardize request URL handling (#1633)
* httpserver/all: Clean up and standardize request URL handling
The HTTP server now always creates a context value on the request which
is a copy of the request's URL struct. It should not be modified by
middlewares, but it is safe to get the value out of the request and make
changes to it locally-scoped. Thus, the value in the context always
stores the original request URL information as it was received. Any
rewrites that happen will be to the request's URL field directly.
The HTTP server no longer cleans /sanitizes the request URL. It made too
many strong assumptions and ended up making a lot of middleware more
complicated, including upstream proxying (and fastcgi). To alleviate
this complexity, we no longer change the request URL. Middlewares are
responsible to access the disk safely by using http.Dir or, if not
actually opening files, they can use httpserver.SafePath().
I'm hoping this will address issues with #1624, #1584, #1582, and others.
* staticfiles: Fix test on Windows
@abiosoft: I still can't figure out exactly what this is for. 😅
* Use (potentially) changed URL for browse redirects, as before
* Use filepath.ToSlash, clean up a couple proxy test cases
* Oops, fix variable name
2017-05-02 13:11:10 +08:00
|
|
|
|
// this defer does not leak fds even though we are in a loop,
|
|
|
|
|
// because previous iterations of the loop must have had an
|
|
|
|
|
// err, so there's nothing to close from earlier iterations.
|
|
|
|
|
defer indexFile.Close()
|
|
|
|
|
|
|
|
|
|
// close previously-opened file immediately to release fd
|
2016-12-20 00:51:09 +08:00
|
|
|
|
f.Close()
|
|
|
|
|
|
httpserver/all: Clean up and standardize request URL handling (#1633)
* httpserver/all: Clean up and standardize request URL handling
The HTTP server now always creates a context value on the request which
is a copy of the request's URL struct. It should not be modified by
middlewares, but it is safe to get the value out of the request and make
changes to it locally-scoped. Thus, the value in the context always
stores the original request URL information as it was received. Any
rewrites that happen will be to the request's URL field directly.
The HTTP server no longer cleans /sanitizes the request URL. It made too
many strong assumptions and ended up making a lot of middleware more
complicated, including upstream proxying (and fastcgi). To alleviate
this complexity, we no longer change the request URL. Middlewares are
responsible to access the disk safely by using http.Dir or, if not
actually opening files, they can use httpserver.SafePath().
I'm hoping this will address issues with #1624, #1584, #1582, and others.
* staticfiles: Fix test on Windows
@abiosoft: I still can't figure out exactly what this is for. 😅
* Use (potentially) changed URL for browse redirects, as before
* Use filepath.ToSlash, clean up a couple proxy test cases
* Oops, fix variable name
2017-05-02 13:11:10 +08:00
|
|
|
|
// switch to using the index file, and we're done here
|
|
|
|
|
d = indexInfo
|
|
|
|
|
f = indexFile
|
|
|
|
|
reqPath = indexPath
|
2016-12-20 00:51:09 +08:00
|
|
|
|
break
|
2015-03-25 11:55:51 +08:00
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
httpserver/all: Clean up and standardize request URL handling (#1633)
* httpserver/all: Clean up and standardize request URL handling
The HTTP server now always creates a context value on the request which
is a copy of the request's URL struct. It should not be modified by
middlewares, but it is safe to get the value out of the request and make
changes to it locally-scoped. Thus, the value in the context always
stores the original request URL information as it was received. Any
rewrites that happen will be to the request's URL field directly.
The HTTP server no longer cleans /sanitizes the request URL. It made too
many strong assumptions and ended up making a lot of middleware more
complicated, including upstream proxying (and fastcgi). To alleviate
this complexity, we no longer change the request URL. Middlewares are
responsible to access the disk safely by using http.Dir or, if not
actually opening files, they can use httpserver.SafePath().
I'm hoping this will address issues with #1624, #1584, #1582, and others.
* staticfiles: Fix test on Windows
@abiosoft: I still can't figure out exactly what this is for. 😅
* Use (potentially) changed URL for browse redirects, as before
* Use filepath.ToSlash, clean up a couple proxy test cases
* Oops, fix variable name
2017-05-02 13:11:10 +08:00
|
|
|
|
// return Not Found if we either did not find an index file (and thus are
|
|
|
|
|
// still a directory) or if this file is supposed to be hidden
|
|
|
|
|
if d.IsDir() || fs.IsHidden(d) {
|
2015-03-29 06:49:42 +08:00
|
|
|
|
return http.StatusNotFound, nil
|
2015-03-25 11:55:51 +08:00
|
|
|
|
}
|
|
|
|
|
|
httpserver/all: Clean up and standardize request URL handling (#1633)
* httpserver/all: Clean up and standardize request URL handling
The HTTP server now always creates a context value on the request which
is a copy of the request's URL struct. It should not be modified by
middlewares, but it is safe to get the value out of the request and make
changes to it locally-scoped. Thus, the value in the context always
stores the original request URL information as it was received. Any
rewrites that happen will be to the request's URL field directly.
The HTTP server no longer cleans /sanitizes the request URL. It made too
many strong assumptions and ended up making a lot of middleware more
complicated, including upstream proxying (and fastcgi). To alleviate
this complexity, we no longer change the request URL. Middlewares are
responsible to access the disk safely by using http.Dir or, if not
actually opening files, they can use httpserver.SafePath().
I'm hoping this will address issues with #1624, #1584, #1582, and others.
* staticfiles: Fix test on Windows
@abiosoft: I still can't figure out exactly what this is for. 😅
* Use (potentially) changed URL for browse redirects, as before
* Use filepath.ToSlash, clean up a couple proxy test cases
* Oops, fix variable name
2017-05-02 13:11:10 +08:00
|
|
|
|
etag := calculateEtag(d)
|
2016-12-20 00:51:09 +08:00
|
|
|
|
|
httpserver/all: Clean up and standardize request URL handling (#1633)
* httpserver/all: Clean up and standardize request URL handling
The HTTP server now always creates a context value on the request which
is a copy of the request's URL struct. It should not be modified by
middlewares, but it is safe to get the value out of the request and make
changes to it locally-scoped. Thus, the value in the context always
stores the original request URL information as it was received. Any
rewrites that happen will be to the request's URL field directly.
The HTTP server no longer cleans /sanitizes the request URL. It made too
many strong assumptions and ended up making a lot of middleware more
complicated, including upstream proxying (and fastcgi). To alleviate
this complexity, we no longer change the request URL. Middlewares are
responsible to access the disk safely by using http.Dir or, if not
actually opening files, they can use httpserver.SafePath().
I'm hoping this will address issues with #1624, #1584, #1582, and others.
* staticfiles: Fix test on Windows
@abiosoft: I still can't figure out exactly what this is for. 😅
* Use (potentially) changed URL for browse redirects, as before
* Use filepath.ToSlash, clean up a couple proxy test cases
* Oops, fix variable name
2017-05-02 13:11:10 +08:00
|
|
|
|
// look for compressed versions of the file on disk, if the client supports that encoding
|
2016-12-20 00:51:09 +08:00
|
|
|
|
for _, encoding := range staticEncodingPriority {
|
httpserver/all: Clean up and standardize request URL handling (#1633)
* httpserver/all: Clean up and standardize request URL handling
The HTTP server now always creates a context value on the request which
is a copy of the request's URL struct. It should not be modified by
middlewares, but it is safe to get the value out of the request and make
changes to it locally-scoped. Thus, the value in the context always
stores the original request URL information as it was received. Any
rewrites that happen will be to the request's URL field directly.
The HTTP server no longer cleans /sanitizes the request URL. It made too
many strong assumptions and ended up making a lot of middleware more
complicated, including upstream proxying (and fastcgi). To alleviate
this complexity, we no longer change the request URL. Middlewares are
responsible to access the disk safely by using http.Dir or, if not
actually opening files, they can use httpserver.SafePath().
I'm hoping this will address issues with #1624, #1584, #1582, and others.
* staticfiles: Fix test on Windows
@abiosoft: I still can't figure out exactly what this is for. 😅
* Use (potentially) changed URL for browse redirects, as before
* Use filepath.ToSlash, clean up a couple proxy test cases
* Oops, fix variable name
2017-05-02 13:11:10 +08:00
|
|
|
|
// see if the client accepts a compressed encoding we offer
|
2017-02-19 06:52:50 +08:00
|
|
|
|
acceptEncoding := strings.Split(r.Header.Get("Accept-Encoding"), ",")
|
|
|
|
|
accepted := false
|
|
|
|
|
for _, acc := range acceptEncoding {
|
httpserver/all: Clean up and standardize request URL handling (#1633)
* httpserver/all: Clean up and standardize request URL handling
The HTTP server now always creates a context value on the request which
is a copy of the request's URL struct. It should not be modified by
middlewares, but it is safe to get the value out of the request and make
changes to it locally-scoped. Thus, the value in the context always
stores the original request URL information as it was received. Any
rewrites that happen will be to the request's URL field directly.
The HTTP server no longer cleans /sanitizes the request URL. It made too
many strong assumptions and ended up making a lot of middleware more
complicated, including upstream proxying (and fastcgi). To alleviate
this complexity, we no longer change the request URL. Middlewares are
responsible to access the disk safely by using http.Dir or, if not
actually opening files, they can use httpserver.SafePath().
I'm hoping this will address issues with #1624, #1584, #1582, and others.
* staticfiles: Fix test on Windows
@abiosoft: I still can't figure out exactly what this is for. 😅
* Use (potentially) changed URL for browse redirects, as before
* Use filepath.ToSlash, clean up a couple proxy test cases
* Oops, fix variable name
2017-05-02 13:11:10 +08:00
|
|
|
|
if strings.TrimSpace(acc) == encoding {
|
2017-02-19 06:52:50 +08:00
|
|
|
|
accepted = true
|
httpserver/all: Clean up and standardize request URL handling (#1633)
* httpserver/all: Clean up and standardize request URL handling
The HTTP server now always creates a context value on the request which
is a copy of the request's URL struct. It should not be modified by
middlewares, but it is safe to get the value out of the request and make
changes to it locally-scoped. Thus, the value in the context always
stores the original request URL information as it was received. Any
rewrites that happen will be to the request's URL field directly.
The HTTP server no longer cleans /sanitizes the request URL. It made too
many strong assumptions and ended up making a lot of middleware more
complicated, including upstream proxying (and fastcgi). To alleviate
this complexity, we no longer change the request URL. Middlewares are
responsible to access the disk safely by using http.Dir or, if not
actually opening files, they can use httpserver.SafePath().
I'm hoping this will address issues with #1624, #1584, #1582, and others.
* staticfiles: Fix test on Windows
@abiosoft: I still can't figure out exactly what this is for. 😅
* Use (potentially) changed URL for browse redirects, as before
* Use filepath.ToSlash, clean up a couple proxy test cases
* Oops, fix variable name
2017-05-02 13:11:10 +08:00
|
|
|
|
break
|
2017-02-19 06:52:50 +08:00
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
httpserver/all: Clean up and standardize request URL handling (#1633)
* httpserver/all: Clean up and standardize request URL handling
The HTTP server now always creates a context value on the request which
is a copy of the request's URL struct. It should not be modified by
middlewares, but it is safe to get the value out of the request and make
changes to it locally-scoped. Thus, the value in the context always
stores the original request URL information as it was received. Any
rewrites that happen will be to the request's URL field directly.
The HTTP server no longer cleans /sanitizes the request URL. It made too
many strong assumptions and ended up making a lot of middleware more
complicated, including upstream proxying (and fastcgi). To alleviate
this complexity, we no longer change the request URL. Middlewares are
responsible to access the disk safely by using http.Dir or, if not
actually opening files, they can use httpserver.SafePath().
I'm hoping this will address issues with #1624, #1584, #1582, and others.
* staticfiles: Fix test on Windows
@abiosoft: I still can't figure out exactly what this is for. 😅
* Use (potentially) changed URL for browse redirects, as before
* Use filepath.ToSlash, clean up a couple proxy test cases
* Oops, fix variable name
2017-05-02 13:11:10 +08:00
|
|
|
|
// if client doesn't support this encoding, don't even bother; try next one
|
2017-02-19 06:52:50 +08:00
|
|
|
|
if !accepted {
|
2016-12-20 00:51:09 +08:00
|
|
|
|
continue
|
|
|
|
|
}
|
|
|
|
|
|
httpserver/all: Clean up and standardize request URL handling (#1633)
* httpserver/all: Clean up and standardize request URL handling
The HTTP server now always creates a context value on the request which
is a copy of the request's URL struct. It should not be modified by
middlewares, but it is safe to get the value out of the request and make
changes to it locally-scoped. Thus, the value in the context always
stores the original request URL information as it was received. Any
rewrites that happen will be to the request's URL field directly.
The HTTP server no longer cleans /sanitizes the request URL. It made too
many strong assumptions and ended up making a lot of middleware more
complicated, including upstream proxying (and fastcgi). To alleviate
this complexity, we no longer change the request URL. Middlewares are
responsible to access the disk safely by using http.Dir or, if not
actually opening files, they can use httpserver.SafePath().
I'm hoping this will address issues with #1624, #1584, #1582, and others.
* staticfiles: Fix test on Windows
@abiosoft: I still can't figure out exactly what this is for. 😅
* Use (potentially) changed URL for browse redirects, as before
* Use filepath.ToSlash, clean up a couple proxy test cases
* Oops, fix variable name
2017-05-02 13:11:10 +08:00
|
|
|
|
// see if the compressed version of this file exists
|
|
|
|
|
encodedFile, err := fs.Root.Open(reqPath + staticEncoding[encoding])
|
2016-12-20 00:51:09 +08:00
|
|
|
|
if err != nil {
|
|
|
|
|
continue
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
encodedFileInfo, err := encodedFile.Stat()
|
|
|
|
|
if err != nil {
|
|
|
|
|
encodedFile.Close()
|
|
|
|
|
continue
|
|
|
|
|
}
|
|
|
|
|
|
httpserver/all: Clean up and standardize request URL handling (#1633)
* httpserver/all: Clean up and standardize request URL handling
The HTTP server now always creates a context value on the request which
is a copy of the request's URL struct. It should not be modified by
middlewares, but it is safe to get the value out of the request and make
changes to it locally-scoped. Thus, the value in the context always
stores the original request URL information as it was received. Any
rewrites that happen will be to the request's URL field directly.
The HTTP server no longer cleans /sanitizes the request URL. It made too
many strong assumptions and ended up making a lot of middleware more
complicated, including upstream proxying (and fastcgi). To alleviate
this complexity, we no longer change the request URL. Middlewares are
responsible to access the disk safely by using http.Dir or, if not
actually opening files, they can use httpserver.SafePath().
I'm hoping this will address issues with #1624, #1584, #1582, and others.
* staticfiles: Fix test on Windows
@abiosoft: I still can't figure out exactly what this is for. 😅
* Use (potentially) changed URL for browse redirects, as before
* Use filepath.ToSlash, clean up a couple proxy test cases
* Oops, fix variable name
2017-05-02 13:11:10 +08:00
|
|
|
|
// close the encoded file when we're done, and close the
|
|
|
|
|
// previously-opened file immediately to release the fd
|
|
|
|
|
defer encodedFile.Close()
|
2016-12-20 00:51:09 +08:00
|
|
|
|
f.Close()
|
|
|
|
|
|
httpserver/all: Clean up and standardize request URL handling (#1633)
* httpserver/all: Clean up and standardize request URL handling
The HTTP server now always creates a context value on the request which
is a copy of the request's URL struct. It should not be modified by
middlewares, but it is safe to get the value out of the request and make
changes to it locally-scoped. Thus, the value in the context always
stores the original request URL information as it was received. Any
rewrites that happen will be to the request's URL field directly.
The HTTP server no longer cleans /sanitizes the request URL. It made too
many strong assumptions and ended up making a lot of middleware more
complicated, including upstream proxying (and fastcgi). To alleviate
this complexity, we no longer change the request URL. Middlewares are
responsible to access the disk safely by using http.Dir or, if not
actually opening files, they can use httpserver.SafePath().
I'm hoping this will address issues with #1624, #1584, #1582, and others.
* staticfiles: Fix test on Windows
@abiosoft: I still can't figure out exactly what this is for. 😅
* Use (potentially) changed URL for browse redirects, as before
* Use filepath.ToSlash, clean up a couple proxy test cases
* Oops, fix variable name
2017-05-02 13:11:10 +08:00
|
|
|
|
// the encoded file is now what we're serving
|
2016-12-20 00:51:09 +08:00
|
|
|
|
f = encodedFile
|
httpserver/all: Clean up and standardize request URL handling (#1633)
* httpserver/all: Clean up and standardize request URL handling
The HTTP server now always creates a context value on the request which
is a copy of the request's URL struct. It should not be modified by
middlewares, but it is safe to get the value out of the request and make
changes to it locally-scoped. Thus, the value in the context always
stores the original request URL information as it was received. Any
rewrites that happen will be to the request's URL field directly.
The HTTP server no longer cleans /sanitizes the request URL. It made too
many strong assumptions and ended up making a lot of middleware more
complicated, including upstream proxying (and fastcgi). To alleviate
this complexity, we no longer change the request URL. Middlewares are
responsible to access the disk safely by using http.Dir or, if not
actually opening files, they can use httpserver.SafePath().
I'm hoping this will address issues with #1624, #1584, #1582, and others.
* staticfiles: Fix test on Windows
@abiosoft: I still can't figure out exactly what this is for. 😅
* Use (potentially) changed URL for browse redirects, as before
* Use filepath.ToSlash, clean up a couple proxy test cases
* Oops, fix variable name
2017-05-02 13:11:10 +08:00
|
|
|
|
etag = calculateEtag(encodedFileInfo)
|
2016-12-20 00:51:09 +08:00
|
|
|
|
w.Header().Add("Vary", "Accept-Encoding")
|
|
|
|
|
w.Header().Set("Content-Encoding", encoding)
|
2017-03-13 06:41:49 +08:00
|
|
|
|
w.Header().Set("Content-Length", strconv.FormatInt(encodedFileInfo.Size(), 10))
|
2016-12-20 00:51:09 +08:00
|
|
|
|
break
|
|
|
|
|
}
|
|
|
|
|
|
2017-02-19 06:52:50 +08:00
|
|
|
|
// Set the ETag returned to the user-agent. Note that a conditional If-None-Match
|
|
|
|
|
// request is handled in http.ServeContent below, which checks against this ETag value.
|
|
|
|
|
w.Header().Set("ETag", etag)
|
2016-04-02 05:24:04 +08:00
|
|
|
|
|
2016-03-12 06:11:21 +08:00
|
|
|
|
// Note: Errors generated by ServeContent are written immediately
|
|
|
|
|
// to the response. This usually only happens if seeking fails (rare).
|
httpserver/all: Clean up and standardize request URL handling (#1633)
* httpserver/all: Clean up and standardize request URL handling
The HTTP server now always creates a context value on the request which
is a copy of the request's URL struct. It should not be modified by
middlewares, but it is safe to get the value out of the request and make
changes to it locally-scoped. Thus, the value in the context always
stores the original request URL information as it was received. Any
rewrites that happen will be to the request's URL field directly.
The HTTP server no longer cleans /sanitizes the request URL. It made too
many strong assumptions and ended up making a lot of middleware more
complicated, including upstream proxying (and fastcgi). To alleviate
this complexity, we no longer change the request URL. Middlewares are
responsible to access the disk safely by using http.Dir or, if not
actually opening files, they can use httpserver.SafePath().
I'm hoping this will address issues with #1624, #1584, #1582, and others.
* staticfiles: Fix test on Windows
@abiosoft: I still can't figure out exactly what this is for. 😅
* Use (potentially) changed URL for browse redirects, as before
* Use filepath.ToSlash, clean up a couple proxy test cases
* Oops, fix variable name
2017-05-02 13:11:10 +08:00
|
|
|
|
// Its signature does not bubble the error up to us, so we cannot
|
|
|
|
|
// return it for any logging middleware to record. Oh well.
|
|
|
|
|
http.ServeContent(w, r, d.Name(), d.ModTime(), f)
|
2016-03-12 06:11:21 +08:00
|
|
|
|
|
|
|
|
|
return http.StatusOK, nil
|
|
|
|
|
}
|
|
|
|
|
|
2016-12-18 02:30:08 +08:00
|
|
|
|
// IsHidden checks if file with FileInfo d is on hide list.
|
|
|
|
|
func (fs FileServer) IsHidden(d os.FileInfo) bool {
|
Rewrote Caddy from the ground up; initial commit of 0.9 branch
These changes span work from the last ~4 months in an effort to make
Caddy more extensible, reduce the coupling between its components, and
lay a more robust foundation of code going forward into 1.0. A bunch of
new features have been added, too, with even higher future potential.
The most significant design change is an overall inversion of
dependencies. Instead of the caddy package knowing about the server
and the notion of middleware and config, the caddy package exposes an
interface that other components plug into. This does introduce more
indirection when reading the code, but every piece is very modular and
pluggable. Even the HTTP server is pluggable.
The caddy package has been moved to the top level, and main has been
pushed into a subfolder called caddy. The actual logic of the main
file has been pushed even further into caddy/caddymain/run.go so that
custom builds of Caddy can be 'go get'able.
The HTTPS logic was surgically separated into two parts to divide the
TLS-specific code and the HTTPS-specific code. The caddytls package can
now be used by any type of server that needs TLS, not just HTTP. I also
added the ability to customize nearly every aspect of TLS at the site
level rather than all sites sharing the same TLS configuration. Not all
of this flexibility is exposed in the Caddyfile yet, but it may be in
the future. Caddy can also generate self-signed certificates in memory
for the convenience of a developer working on localhost who wants HTTPS.
And Caddy now supports the DNS challenge, assuming at least one DNS
provider is plugged in.
Dozens, if not hundreds, of other minor changes swept through the code
base as I literally started from an empty main function, copying over
functions or files as needed, then adjusting them to fit in the new
design. Most tests have been restored and adapted to the new API,
but more work is needed there.
A lot of what was "impossible" before is now possible, or can be made
possible with minimal disruption of the code. For example, it's fairly
easy to make plugins hook into another part of the code via callbacks.
Plugins can do more than just be directives; we now have plugins that
customize how the Caddyfile is loaded (useful when you need to get your
configuration from a remote store).
Site addresses no longer need be just a host and port. They can have a
path, allowing you to scope a configuration to a specific path. There is
no inheretance, however; each site configuration is distinct.
Thanks to amazing work by Lucas Clemente, this commit adds experimental
QUIC support. Turn it on using the -quic flag; your browser may have
to be configured to enable it.
Almost everything is here, but you will notice that most of the middle-
ware are missing. After those are transferred over, we'll be ready for
beta tests.
I'm very excited to get this out. Thanks for everyone's help and
patience these last few months. I hope you like it!!
2016-06-05 07:00:29 +08:00
|
|
|
|
for _, hiddenPath := range fs.Hide {
|
httpserver/all: Clean up and standardize request URL handling (#1633)
* httpserver/all: Clean up and standardize request URL handling
The HTTP server now always creates a context value on the request which
is a copy of the request's URL struct. It should not be modified by
middlewares, but it is safe to get the value out of the request and make
changes to it locally-scoped. Thus, the value in the context always
stores the original request URL information as it was received. Any
rewrites that happen will be to the request's URL field directly.
The HTTP server no longer cleans /sanitizes the request URL. It made too
many strong assumptions and ended up making a lot of middleware more
complicated, including upstream proxying (and fastcgi). To alleviate
this complexity, we no longer change the request URL. Middlewares are
responsible to access the disk safely by using http.Dir or, if not
actually opening files, they can use httpserver.SafePath().
I'm hoping this will address issues with #1624, #1584, #1582, and others.
* staticfiles: Fix test on Windows
@abiosoft: I still can't figure out exactly what this is for. 😅
* Use (potentially) changed URL for browse redirects, as before
* Use filepath.ToSlash, clean up a couple proxy test cases
* Oops, fix variable name
2017-05-02 13:11:10 +08:00
|
|
|
|
// TODO: Could these FileInfos be stored instead of their paths, to avoid opening them all the time?
|
Rewrote Caddy from the ground up; initial commit of 0.9 branch
These changes span work from the last ~4 months in an effort to make
Caddy more extensible, reduce the coupling between its components, and
lay a more robust foundation of code going forward into 1.0. A bunch of
new features have been added, too, with even higher future potential.
The most significant design change is an overall inversion of
dependencies. Instead of the caddy package knowing about the server
and the notion of middleware and config, the caddy package exposes an
interface that other components plug into. This does introduce more
indirection when reading the code, but every piece is very modular and
pluggable. Even the HTTP server is pluggable.
The caddy package has been moved to the top level, and main has been
pushed into a subfolder called caddy. The actual logic of the main
file has been pushed even further into caddy/caddymain/run.go so that
custom builds of Caddy can be 'go get'able.
The HTTPS logic was surgically separated into two parts to divide the
TLS-specific code and the HTTPS-specific code. The caddytls package can
now be used by any type of server that needs TLS, not just HTTP. I also
added the ability to customize nearly every aspect of TLS at the site
level rather than all sites sharing the same TLS configuration. Not all
of this flexibility is exposed in the Caddyfile yet, but it may be in
the future. Caddy can also generate self-signed certificates in memory
for the convenience of a developer working on localhost who wants HTTPS.
And Caddy now supports the DNS challenge, assuming at least one DNS
provider is plugged in.
Dozens, if not hundreds, of other minor changes swept through the code
base as I literally started from an empty main function, copying over
functions or files as needed, then adjusting them to fit in the new
design. Most tests have been restored and adapted to the new API,
but more work is needed there.
A lot of what was "impossible" before is now possible, or can be made
possible with minimal disruption of the code. For example, it's fairly
easy to make plugins hook into another part of the code via callbacks.
Plugins can do more than just be directives; we now have plugins that
customize how the Caddyfile is loaded (useful when you need to get your
configuration from a remote store).
Site addresses no longer need be just a host and port. They can have a
path, allowing you to scope a configuration to a specific path. There is
no inheretance, however; each site configuration is distinct.
Thanks to amazing work by Lucas Clemente, this commit adds experimental
QUIC support. Turn it on using the -quic flag; your browser may have
to be configured to enable it.
Almost everything is here, but you will notice that most of the middle-
ware are missing. After those are transferred over, we'll be ready for
beta tests.
I'm very excited to get this out. Thanks for everyone's help and
patience these last few months. I hope you like it!!
2016-06-05 07:00:29 +08:00
|
|
|
|
if hFile, err := fs.Root.Open(hiddenPath); err == nil {
|
2016-03-13 00:47:53 +08:00
|
|
|
|
fs, _ := hFile.Stat()
|
|
|
|
|
hFile.Close()
|
|
|
|
|
if os.SameFile(d, fs) {
|
|
|
|
|
return true
|
|
|
|
|
}
|
2015-04-13 07:44:02 +08:00
|
|
|
|
}
|
|
|
|
|
}
|
2016-03-12 06:11:21 +08:00
|
|
|
|
return false
|
2015-03-25 11:55:51 +08:00
|
|
|
|
}
|
2015-04-19 03:24:54 +08:00
|
|
|
|
|
httpserver/all: Clean up and standardize request URL handling (#1633)
* httpserver/all: Clean up and standardize request URL handling
The HTTP server now always creates a context value on the request which
is a copy of the request's URL struct. It should not be modified by
middlewares, but it is safe to get the value out of the request and make
changes to it locally-scoped. Thus, the value in the context always
stores the original request URL information as it was received. Any
rewrites that happen will be to the request's URL field directly.
The HTTP server no longer cleans /sanitizes the request URL. It made too
many strong assumptions and ended up making a lot of middleware more
complicated, including upstream proxying (and fastcgi). To alleviate
this complexity, we no longer change the request URL. Middlewares are
responsible to access the disk safely by using http.Dir or, if not
actually opening files, they can use httpserver.SafePath().
I'm hoping this will address issues with #1624, #1584, #1582, and others.
* staticfiles: Fix test on Windows
@abiosoft: I still can't figure out exactly what this is for. 😅
* Use (potentially) changed URL for browse redirects, as before
* Use filepath.ToSlash, clean up a couple proxy test cases
* Oops, fix variable name
2017-05-02 13:11:10 +08:00
|
|
|
|
// calculateEtag produces a strong etag by default, although, for
|
|
|
|
|
// efficiency reasons, it does not actually consume the contents
|
|
|
|
|
// of the file to make a hash of all the bytes. ¯\_(ツ)_/¯
|
|
|
|
|
// Prefix the etag with "W/" to convert it into a weak etag.
|
|
|
|
|
// See: https://tools.ietf.org/html/rfc7232#section-2.3
|
|
|
|
|
func calculateEtag(d os.FileInfo) string {
|
|
|
|
|
t := strconv.FormatInt(d.ModTime().Unix(), 36)
|
|
|
|
|
s := strconv.FormatInt(d.Size(), 36)
|
|
|
|
|
return `"` + t + s + `"`
|
2015-04-19 03:24:54 +08:00
|
|
|
|
}
|
2015-09-20 10:34:23 +08:00
|
|
|
|
|
|
|
|
|
// IndexPages is a list of pages that may be understood as
|
|
|
|
|
// the "index" files to directories.
|
|
|
|
|
var IndexPages = []string{
|
|
|
|
|
"index.html",
|
|
|
|
|
"index.htm",
|
|
|
|
|
"index.txt",
|
|
|
|
|
"default.html",
|
|
|
|
|
"default.htm",
|
|
|
|
|
"default.txt",
|
|
|
|
|
}
|
2016-12-20 00:51:09 +08:00
|
|
|
|
|
|
|
|
|
// staticEncoding is a map of content-encoding to a file extension.
|
|
|
|
|
// If client accepts given encoding (via Accept-Encoding header) and compressed file with given extensions exists
|
|
|
|
|
// it will be served to the client instead of original one.
|
|
|
|
|
var staticEncoding = map[string]string{
|
|
|
|
|
"gzip": ".gz",
|
|
|
|
|
"br": ".br",
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// staticEncodingPriority is a list of preferred static encodings (most efficient compression to least one).
|
|
|
|
|
var staticEncodingPriority = []string{
|
|
|
|
|
"br",
|
|
|
|
|
"gzip",
|
|
|
|
|
}
|
2017-02-11 11:02:00 +08:00
|
|
|
|
|
|
|
|
|
// mapFSRootOpenErr maps the provided non-nil error
|
|
|
|
|
// to a possibly better non-nil error. In particular, it turns OS-specific errors
|
|
|
|
|
// about opening files in non-directories into os.ErrNotExist.
|
|
|
|
|
//
|
httpserver/all: Clean up and standardize request URL handling (#1633)
* httpserver/all: Clean up and standardize request URL handling
The HTTP server now always creates a context value on the request which
is a copy of the request's URL struct. It should not be modified by
middlewares, but it is safe to get the value out of the request and make
changes to it locally-scoped. Thus, the value in the context always
stores the original request URL information as it was received. Any
rewrites that happen will be to the request's URL field directly.
The HTTP server no longer cleans /sanitizes the request URL. It made too
many strong assumptions and ended up making a lot of middleware more
complicated, including upstream proxying (and fastcgi). To alleviate
this complexity, we no longer change the request URL. Middlewares are
responsible to access the disk safely by using http.Dir or, if not
actually opening files, they can use httpserver.SafePath().
I'm hoping this will address issues with #1624, #1584, #1582, and others.
* staticfiles: Fix test on Windows
@abiosoft: I still can't figure out exactly what this is for. 😅
* Use (potentially) changed URL for browse redirects, as before
* Use filepath.ToSlash, clean up a couple proxy test cases
* Oops, fix variable name
2017-05-02 13:11:10 +08:00
|
|
|
|
// TODO: remove when http.Dir handles this (slated for Go 1.9)
|
2017-02-11 11:02:00 +08:00
|
|
|
|
// Go issue #18984
|
|
|
|
|
func mapFSRootOpenErr(originalErr error) error {
|
|
|
|
|
if os.IsNotExist(originalErr) || os.IsPermission(originalErr) {
|
|
|
|
|
return originalErr
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
perr, ok := originalErr.(*os.PathError)
|
|
|
|
|
if !ok {
|
|
|
|
|
return originalErr
|
|
|
|
|
}
|
|
|
|
|
name := perr.Path
|
|
|
|
|
parts := strings.Split(name, string(filepath.Separator))
|
|
|
|
|
for i := range parts {
|
|
|
|
|
if parts[i] == "" {
|
|
|
|
|
continue
|
|
|
|
|
}
|
|
|
|
|
fi, err := os.Stat(strings.Join(parts[:i+1], string(filepath.Separator)))
|
|
|
|
|
if err != nil {
|
|
|
|
|
return originalErr
|
|
|
|
|
}
|
|
|
|
|
if !fi.IsDir() {
|
|
|
|
|
return os.ErrNotExist
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
return originalErr
|
|
|
|
|
}
|