caddy/plugins.go

379 lines
12 KiB
Go
Raw Normal View History

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 caddy
import (
"fmt"
"log"
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
"net"
"sort"
"github.com/mholt/caddy/caddyfile"
)
// These are all the registered plugins.
var (
// serverTypes is a map of registered server types.
serverTypes = make(map[string]ServerType)
// plugins is a map of server type to map of plugin name to
// Plugin. These are the "general" plugins that may or may
// not be associated with a specific server type. If it's
// applicable to multiple server types or the server type is
// irrelevant, the key is empty string (""). But all plugins
// must have a name.
plugins = make(map[string]map[string]Plugin)
// eventHooks is a map of hook name to Hook. All hooks plugins
// must have a name.
eventHooks = make(map[string]EventHook)
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
// parsingCallbacks maps server type to map of directive
// to list of callback functions. These aren't really
// plugins on their own, but are often registered from
// plugins.
parsingCallbacks = make(map[string]map[string][]ParsingCallback)
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
// caddyfileLoaders is the list of all Caddyfile loaders
// in registration order.
caddyfileLoaders []caddyfileLoader
)
// DescribePlugins returns a string describing the registered plugins.
func DescribePlugins() string {
str := "Server types:\n"
for name := range serverTypes {
str += " " + name + "\n"
}
// List the loaders in registration order
str += "\nCaddyfile loaders:\n"
for _, loader := range caddyfileLoaders {
str += " " + loader.name + "\n"
}
if defaultCaddyfileLoader.name != "" {
str += " " + defaultCaddyfileLoader.name + "\n"
}
if len(eventHooks) > 0 {
// List the event hook plugins
str += "\nEvent hook plugins:\n"
for hookPlugin := range eventHooks {
str += " hook." + hookPlugin + "\n"
}
}
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
// Let's alphabetize the rest of these...
var others []string
for stype, stypePlugins := range plugins {
for name := range stypePlugins {
var s string
if stype != "" {
s = stype + "."
}
s += name
others = append(others, s)
}
}
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
sort.Strings(others)
str += "\nOther plugins:\n"
for _, name := range others {
str += " " + name + "\n"
}
return str
}
// ValidDirectives returns the list of all directives that are
// recognized for the server type serverType. However, not all
// directives may be installed. This makes it possible to give
// more helpful error messages, like "did you mean ..." or
// "maybe you need to plug in ...".
func ValidDirectives(serverType string) []string {
stype, err := getServerType(serverType)
if err != nil {
return nil
}
return stype.Directives()
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
}
// ServerListener pairs a server to its listener and/or packetconn.
type ServerListener struct {
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
server Server
listener net.Listener
packet net.PacketConn
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
}
// LocalAddr returns the local network address of the packetconn. It returns
// nil when it is not set.
func (s ServerListener) LocalAddr() net.Addr {
if s.packet == nil {
return nil
}
return s.packet.LocalAddr()
}
// Addr returns the listener's network address. It returns nil when it is
// not set.
func (s ServerListener) Addr() net.Addr {
if s.listener == nil {
return nil
}
return s.listener.Addr()
}
2016-06-07 08:27:54 +08:00
// Context is a type which carries a server type through
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
// the load and setup phase; it maintains the state
// between loading the Caddyfile, then executing its
// directives, then making the servers for Caddy to
// manage. Typically, such state involves configuration
// structs, etc.
type Context interface {
2016-06-07 08:27:54 +08:00
// Called after the Caddyfile is parsed into server
// blocks but before the directives are executed,
// this method gives you an opportunity to inspect
// the server blocks and prepare for the execution
// of directives. Return the server blocks (which
// you may modify, if desired) and an error, if any.
// The first argument is the name or path to the
// configuration file (Caddyfile).
//
// This function can be a no-op and simply return its
// input if there is nothing to do here.
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
InspectServerBlocks(string, []caddyfile.ServerBlock) ([]caddyfile.ServerBlock, error)
2016-06-07 08:27:54 +08:00
// This is what Caddy calls to make server instances.
// By this time, all directives have been executed and,
// presumably, the context has enough state to produce
// server instances for Caddy to start.
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
MakeServers() ([]Server, error)
}
// RegisterServerType registers a server type srv by its
// name, typeName.
func RegisterServerType(typeName string, srv ServerType) {
if _, ok := serverTypes[typeName]; ok {
panic("server type already registered")
}
serverTypes[typeName] = srv
}
// ServerType contains information about a server type.
type ServerType struct {
// Function that returns the list of directives, in
// execution order, that are valid for this server
// type. Directives should be one word if possible
// and lower-cased.
Directives func() []string
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
// DefaultInput returns a default config input if none
2016-06-07 08:27:54 +08:00
// is otherwise loaded. This is optional, but highly
// recommended, otherwise a blank Caddyfile will be
// used.
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
DefaultInput func() Input
2016-06-07 08:27:54 +08:00
// The function that produces a new server type context.
// This will be called when a new Caddyfile is being
// loaded, parsed, and executed independently of any
// startup phases before this one. It's a way to keep
// each set of server instances separate and to reduce
// the amount of global state you need.
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
NewContext func() Context
}
// Plugin is a type which holds information about a plugin.
type Plugin struct {
// ServerType is the type of server this plugin is for.
// Can be empty if not applicable, or if the plugin
// can associate with any server type.
ServerType string
// Action is the plugin's setup function, if associated
// with a directive in the Caddyfile.
Action SetupFunc
}
// RegisterPlugin plugs in plugin. All plugins should register
// themselves, even if they do not perform an action associated
// with a directive. It is important for the process to know
// which plugins are available.
//
// The plugin MUST have a name: lower case and one word.
// If this plugin has an action, it must be the name of
// the directive that invokes it. A name is always required
// and must be unique for the server type.
func RegisterPlugin(name string, plugin Plugin) {
if name == "" {
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
panic("plugin must have a name")
}
if _, ok := plugins[plugin.ServerType]; !ok {
plugins[plugin.ServerType] = make(map[string]Plugin)
}
if _, dup := plugins[plugin.ServerType][name]; dup {
panic("plugin named " + name + " already registered for server type " + plugin.ServerType)
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
}
plugins[plugin.ServerType][name] = plugin
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
}
// EventName represents the name of an event used with event hooks.
type EventName string
const (
StartupEvent EventName = "startup"
ShutdownEvent EventName = "shutdown"
)
// EventHook is a type which holds information about a startup hook plugin.
type EventHook func(eventType EventName) error
// RegisterEventHook plugs in hook. All the hooks should register themselves
// and they must have a name.
func RegisterEventHook(name string, hook EventHook) {
if name == "" {
panic("event hook must have a name")
}
if _, dup := eventHooks[name]; dup {
panic("hook named " + name + " already registered")
}
eventHooks[name] = hook
}
// EmitEvent executes the different hooks passing the EventType as an
// argument. This is a blocking function. Hook developers should
// use 'go' keyword if they don't want to block Caddy.
func EmitEvent(event EventName) {
for name, hook := range eventHooks {
err := hook(event)
if err != nil {
log.Printf("error on '%s' hook: %v", name, err)
}
}
}
// ParsingCallback is a function that is called after
// a directive's setup functions have been executed
// for all the server blocks.
type ParsingCallback func(Context) error
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
// RegisterParsingCallback registers callback to be called after
// executing the directive afterDir for server type serverType.
func RegisterParsingCallback(serverType, afterDir string, callback ParsingCallback) {
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 _, ok := parsingCallbacks[serverType]; !ok {
parsingCallbacks[serverType] = make(map[string][]ParsingCallback)
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
}
parsingCallbacks[serverType][afterDir] = append(parsingCallbacks[serverType][afterDir], callback)
}
// SetupFunc is used to set up a plugin, or in other words,
// execute a directive. It will be called once per key for
// each server block it appears in.
type SetupFunc func(c *Controller) error
// DirectiveAction gets the action for directive dir of
// server type serverType.
func DirectiveAction(serverType, dir string) (SetupFunc, error) {
if stypePlugins, ok := plugins[serverType]; ok {
if plugin, ok := stypePlugins[dir]; ok {
return plugin.Action, nil
}
}
if genericPlugins, ok := plugins[""]; ok {
if plugin, ok := genericPlugins[dir]; ok {
return plugin.Action, nil
}
}
return nil, fmt.Errorf("no action found for directive '%s' with server type '%s' (missing a plugin?)",
dir, serverType)
}
// Loader is a type that can load a Caddyfile.
// It is passed the name of the server type.
// It returns an error only if something went
// wrong, not simply if there is no Caddyfile
// for this loader to load.
//
// A Loader should only load the Caddyfile if
// a certain condition or requirement is met,
// as returning a non-nil Input value along with
// another Loader will result in an error.
// In other words, loading the Caddyfile must
// be deliberate & deterministic, not haphazard.
//
// The exception is the default Caddyfile loader,
// which will be called only if no other Caddyfile
// loaders return a non-nil Input. The default
// loader may always return an Input value.
type Loader interface {
2016-06-07 08:27:54 +08:00
Load(serverType string) (Input, error)
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
}
// LoaderFunc is a convenience type similar to http.HandlerFunc
// that allows you to use a plain function as a Load() method.
2016-06-07 08:27:54 +08:00
type LoaderFunc func(serverType string) (Input, error)
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
// Load loads a Caddyfile.
func (lf LoaderFunc) Load(serverType string) (Input, error) {
return lf(serverType)
}
// RegisterCaddyfileLoader registers loader named name.
func RegisterCaddyfileLoader(name string, loader Loader) {
caddyfileLoaders = append(caddyfileLoaders, caddyfileLoader{name: name, loader: loader})
}
// SetDefaultCaddyfileLoader registers loader by name
// as the default Caddyfile loader if no others produce
// a Caddyfile. If another Caddyfile loader has already
// been set as the default, this replaces it.
//
// Do not call RegisterCaddyfileLoader on the same
// loader; that would be redundant.
func SetDefaultCaddyfileLoader(name string, loader Loader) {
defaultCaddyfileLoader = caddyfileLoader{name: name, loader: loader}
}
// loadCaddyfileInput iterates the registered Caddyfile loaders
// and, if needed, calls the default loader, to load a Caddyfile.
// It is an error if any of the loaders return an error or if
// more than one loader returns a Caddyfile.
func loadCaddyfileInput(serverType string) (Input, error) {
var loadedBy string
var caddyfileToUse Input
for _, l := range caddyfileLoaders {
2016-08-18 07:02:01 +08:00
cdyfile, err := l.loader.Load(serverType)
if err != nil {
return nil, fmt.Errorf("loading Caddyfile via %s: %v", l.name, err)
}
if cdyfile != nil {
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 caddyfileToUse != nil {
return nil, fmt.Errorf("Caddyfile loaded multiple times; first by %s, then by %s", loadedBy, l.name)
}
loaderUsed = l
caddyfileToUse = cdyfile
loadedBy = l.name
}
}
if caddyfileToUse == nil && defaultCaddyfileLoader.loader != nil {
cdyfile, err := defaultCaddyfileLoader.loader.Load(serverType)
if err != nil {
return nil, err
}
if cdyfile != nil {
loaderUsed = defaultCaddyfileLoader
caddyfileToUse = cdyfile
}
}
return caddyfileToUse, nil
}
// caddyfileLoader pairs the name of a loader to the loader.
type caddyfileLoader struct {
name string
loader Loader
}
var (
defaultCaddyfileLoader caddyfileLoader // the default loader if all else fail
loaderUsed caddyfileLoader // the loader that was used (relevant for reloads)
)