This commit was manufactured by cvs2svn to create tag 'luasocket-2-0-2'.

Sprout from master 2007-10-11 21:16:28 UTC Diego Nehab <diego@tecgraf.puc-rio.br> 'Tested each sample.'
Cherrypick from master 2007-05-31 22:27:40 UTC Diego Nehab <diego@tecgraf.puc-rio.br> 'Before sending to Roberto.':
    gem/ltn012.tex
    gem/makefile
This commit is contained in:
cvs2git convertor 2007-10-11 21:16:29 +00:00
parent 52ac60af81
commit 81ebe649f0
2 changed files with 163 additions and 197 deletions

View File

@ -6,10 +6,7 @@
\DefineVerbatimEnvironment{mime}{Verbatim}{fontsize=\small,commandchars=\$\#\%}
\newcommand{\stick}[1]{\vbox{\setlength{\parskip}{0pt}#1}}
\newcommand{\bl}{\ensuremath{\mathtt{\backslash}}}
\newcommand{\CR}{\texttt{CR}}
\newcommand{\LF}{\texttt{LF}}
\newcommand{\CRLF}{\texttt{CR~LF}}
\newcommand{\nil}{\texttt{nil}}
\title{Filters, sources, sinks, and pumps\\
{\large or Functional programming for the rest of us}}
@ -20,31 +17,30 @@
\maketitle
\begin{abstract}
Certain data processing operations can be implemented in the
form of filters. A filter is a function that can process
data received in consecutive invocations, returning partial
results each time it is called. Examples of operations that
can be implemented as filters include the end-of-line
normalization for text, Base64 and Quoted-Printable transfer
content encodings, the breaking of text into lines, SMTP
dot-stuffing, and there are many others. Filters become
even more powerful when we allow them to be chained together
to create composite filters. In this context, filters can be
seen as the internal links in a chain of data transformations.
Sources and sinks are the corresponding end points in these
chains. A source is a function that produces data, chunk by
chunk, and a sink is a function that takes data, chunk by
chunk. Finally, pumps are procedures that actively drive
data from a source to a sink, and indirectly through all
intervening filters. In this article, we describe the design of an
elegant interface for filters, sources, sinks, chains, and
pumps, and we illustrate each step with concrete examples.
Certain data processing operations can be implemented in the
form of filters. A filter is a function that can process data
received in consecutive function calls, returning partial
results after each invocation. Examples of operations that can be
implemented as filters include the end-of-line normalization
for text, Base64 and Quoted-Printable transfer content
encodings, the breaking of text into lines, SMTP dot-stuffing,
and there are many others. Filters become even
more powerful when we allow them to be chained together to
create composite filters. In this context, filters can be seen
as the middle links in a chain of data transformations. Sources an sinks
are the corresponding end points of these chains. A source
is a function that produces data, chunk by chunk, and a sink
is a function that takes data, chunk by chunk. In this
article, we describe the design of an elegant interface for filters,
sources, sinks, and chaining, and illustrate each step
with concrete examples.
\end{abstract}
\section{Introduction}
Within the realm of networking applications, we are often
required to apply transformations to streams of data. Examples
required apply transformations to streams of data. Examples
include the end-of-line normalization for text, Base64 and
Quoted-Printable transfer content encodings, breaking text
into lines with a maximum number of columns, SMTP
@ -54,10 +50,11 @@ transfer coding, and the list goes on.
Many complex tasks require a combination of two or more such
transformations, and therefore a general mechanism for
promoting reuse is desirable. In the process of designing
\texttt{LuaSocket~2.0}, we repeatedly faced this problem.
The solution we reached proved to be very general and
convenient. It is based on the concepts of filters, sources,
sinks, and pumps, which we introduce below.
\texttt{LuaSocket~2.0}, David Burgess and I were forced to deal with
this problem. The solution we reached proved to be very
general and convenient. It is based on the concepts of
filters, sources, sinks, and pumps, which we introduce
below.
\emph{Filters} are functions that can be repeatedly invoked
with chunks of input, successively returning processed
@ -65,33 +62,34 @@ chunks of output. More importantly, the result of
concatenating all the output chunks must be the same as the
result of applying the filter to the concatenation of all
input chunks. In fancier language, filters \emph{commute}
with the concatenation operator. More importantly, filters
must handle input data correctly no matter how the stream
has been split into chunks.
with the concatenation operator. As a result, chunk
boundaries are irrelevant: filters correctly handle input
data no matter how it is split.
A \emph{chain} is a function that transparently combines the
effect of one or more filters. The interface of a chain is
indistinguishable from the interface of its component
filters. This allows a chained filter to be used wherever
an atomic filter is accepted. In particular, chains can be
A \emph{chain} transparently combines the effect of one or
more filters. The interface of a chain is
indistinguishable from the interface of its components.
This allows a chained filter to be used wherever an atomic
filter is expected. In particular, chains can be
themselves chained to create arbitrarily complex operations.
Filters can be seen as internal nodes in a network through
which data will flow, potentially being transformed many
times along the way. Chains connect these nodes together.
The initial and final nodes of the network are
\emph{sources} and \emph{sinks}, respectively. Less
abstractly, a source is a function that produces new data
every time it is invoked. Conversely, sinks are functions
that give a final destination to the data they receive.
Naturally, sources and sinks can also be chained with
filters to produce filtered sources and sinks.
times along its way. Chains connect these nodes together.
To complete the picture, we need \emph{sources} and
\emph{sinks}. These are the initial and final nodes of the
network, respectively. Less abstractly, a source is a
function that produces new data every time it is called.
Conversely, sinks are functions that give a final
destination to the data they receive. Naturally, sources
and sinks can also be chained with filters to produce
filtered sources and sinks.
Finally, filters, chains, sources, and sinks are all passive
entities: they must be repeatedly invoked in order for
anything to happen. \emph{Pumps} provide the driving force
that pushes data through the network, from a source to a
sink, and indirectly through all intervening filters.
sink.
In the following sections, we start with a simplified
interface, which we later refine. The evolution we present
@ -101,28 +99,27 @@ concepts within our application domain.
\subsection{A simple example}
The end-of-line normalization of text is a good
Let us use the end-of-line normalization of text as an
example to motivate our initial filter interface.
Assume we are given text in an unknown end-of-line
convention (including possibly mixed conventions) out of the
commonly found Unix (\LF), Mac OS (\CR), and
DOS (\CRLF) conventions. We would like to be able to
use the folowing code to normalize the end-of-line markers:
commonly found Unix (LF), Mac OS (CR), and DOS (CRLF)
conventions. We would like to be able to write code like the
following:
\begin{quote}
\begin{lua}
@stick#
local CRLF = "\013\010"
local input = source.chain(source.file(io.stdin), normalize(CRLF))
local output = sink.file(io.stdout)
pump.all(input, output)
local in = source.chain(source.file(io.stdin), normalize("\r\n"))
local out = sink.file(io.stdout)
pump.all(in, out)
%
\end{lua}
\end{quote}
This program should read data from the standard input stream
and normalize the end-of-line markers to the canonic
\CRLF\ marker, as defined by the MIME standard.
Finally, the normalized text should be sent to the standard output
and normalize the end-of-line markers to the canonic CRLF
marker, as defined by the MIME standard. Finally, the
normalized text should be sent to the standard output
stream. We use a \emph{file source} that produces data from
standard input, and chain it with a filter that normalizes
the data. The pump then repeatedly obtains data from the
@ -130,28 +127,27 @@ source, and passes it to the \emph{file sink}, which sends
it to the standard output.
In the code above, the \texttt{normalize} \emph{factory} is a
function that creates our normalization filter, which
replaces any end-of-line marker with the canonic marker.
The initial filter interface is
function that creates our normalization filter. This filter
will replace any end-of-line marker with the canonic
`\verb|\r\n|' marker. The initial filter interface is
trivial: a filter function receives a chunk of input data,
and returns a chunk of processed data. When there are no
more input data left, the caller notifies the filter by invoking
it with a \nil\ chunk. The filter responds by returning
the final chunk of processed data (which could of course be
the empty string).
it with a \texttt{nil} chunk. The filter responds by returning
the final chunk of processed data.
Although the interface is extremely simple, the
implementation is not so obvious. A normalization filter
respecting this interface needs to keep some kind of context
between calls. This is because a chunk boundary may lie between
the \CR\ and \LF\ characters marking the end of a single line. This
the CR and LF characters marking the end of a line. This
need for contextual storage motivates the use of
factories: each time the factory is invoked, it returns a
filter with its own context so that we can have several
independent filters being used at the same time. For
efficiency reasons, we must avoid the obvious solution of
concatenating all the input into the context before
producing any output chunks.
producing any output.
To that end, we break the implementation into two parts:
a low-level filter, and a factory of high-level filters. The
@ -171,10 +167,10 @@ end-of-line normalization filters:
\begin{quote}
\begin{lua}
@stick#
function filter.cycle(lowlevel, context, extra)
function filter.cycle(low, ctx, extra)
return function(chunk)
local ret
ret, context = lowlevel(context, chunk, extra)
ret, ctx = low(ctx, chunk, extra)
return ret
end
end
@ -182,30 +178,27 @@ end
@stick#
function normalize(marker)
return filter.cycle(eol, 0, marker)
return cycle(eol, 0, marker)
end
%
\end{lua}
\end{quote}
The \texttt{normalize} factory simply calls a more generic
factory, the \texttt{cycle}~factory, passing the low-level
filter~\texttt{eol}. The \texttt{cycle}~factory receives a
factory, the \texttt{cycle} factory. This factory receives a
low-level filter, an initial context, and an extra
parameter, and returns a new high-level filter. Each time
the high-level filer is passed a new chunk, it invokes the
low-level filter with the previous context, the new chunk,
and the extra argument. It is the low-level filter that
does all the work, producing the chunk of processed data and
a new context. The high-level filter then replaces its
a new context. The high-level filter then updates its
internal context, and returns the processed chunk of data to
the user. Notice that we take advantage of Lua's lexical
scoping to store the context in a closure between function
calls.
\subsection{The C part of the filter}
As for the low-level filter, we must first accept
Concerning the low-level filter code, we must first accept
that there is no perfect solution to the end-of-line marker
normalization problem. The difficulty comes from an
inherent ambiguity in the definition of empty lines within
@ -215,39 +208,39 @@ mixed input. It also does a reasonable job with empty lines
and serves as a good example of how to implement a low-level
filter.
The idea is to consider both \CR\ and~\LF\ as end-of-line
The idea is to consider both CR and~LF as end-of-line
\emph{candidates}. We issue a single break if any candidate
is seen alone, or if it is followed by a different
candidate. In other words, \CR~\CR~and \LF~\LF\ each issue
two end-of-line markers, whereas \CR~\LF~and \LF~\CR\ issue
only one marker each. It is easy to see that this method
correctly handles the most common end-of-line conventions.
is seen alone, or followed by a different candidate. In
other words, CR~CR~and LF~LF each issue two end-of-line
markers, whereas CR~LF~and LF~CR issue only one marker each.
This method correctly handles the Unix, DOS/MIME, VMS, and Mac
OS conventions.
With this in mind, we divide the low-level filter into two
simple functions. The inner function~\texttt{pushchar} performs the
normalization itself. It takes each input character in turn,
deciding what to output and how to modify the context. The
context tells if the last processed character was an
end-of-line candidate, and if so, which candidate it was.
For efficiency, we use Lua's auxiliary library's buffer
interface:
\subsection{The C part of the filter}
Our low-level filter is divided into two simple functions.
The inner function performs the normalization itself. It takes
each input character in turn, deciding what to output and
how to modify the context. The context tells if the last
processed character was an end-of-line candidate, and if so,
which candidate it was. For efficiency, it uses
Lua's auxiliary library's buffer interface:
\begin{quote}
\begin{C}
@stick#
@#define candidate(c) (c == CR || c == LF)
static int pushchar(int c, int last, const char *marker,
static int process(int c, int last, const char *marker,
luaL_Buffer *buffer) {
if (candidate(c)) {
if (candidate(last)) {
if (c == last)
luaL_addstring(buffer, marker);
if (c == last) luaL_addstring(buffer, marker);
return 0;
} else {
luaL_addstring(buffer, marker);
return c;
}
} else {
luaL_pushchar(buffer, c);
luaL_putchar(buffer, c);
return 0;
}
}
@ -255,20 +248,15 @@ static int pushchar(int c, int last, const char *marker,
\end{C}
\end{quote}
The outer function~\texttt{eol} simply interfaces with Lua.
It receives the context and input chunk (as well as an
optional custom end-of-line marker), and returns the
transformed output chunk and the new context.
Notice that if the input chunk is \nil, the operation
is considered to be finished. In that case, the loop will
not execute a single time and the context is reset to the
initial state. This allows the filter to be reused many
times:
The outer function simply interfaces with Lua. It receives the
context and input chunk (as well as an optional
custom end-of-line marker), and returns the transformed
output chunk and the new context:
\begin{quote}
\begin{C}
@stick#
static int eol(lua_State *L) {
int context = luaL_checkint(L, 1);
int ctx = luaL_checkint(L, 1);
size_t isize = 0;
const char *input = luaL_optlstring(L, 2, NULL, &isize);
const char *last = input + isize;
@ -281,18 +269,24 @@ static int eol(lua_State *L) {
return 2;
}
while (input < last)
context = pushchar(*input++, context, marker, &buffer);
ctx = process(*input++, ctx, marker, &buffer);
luaL_pushresult(&buffer);
lua_pushnumber(L, context);
lua_pushnumber(L, ctx);
return 2;
}
%
\end{C}
\end{quote}
Notice that if the input chunk is \texttt{nil}, the operation
is considered to be finished. In that case, the loop will
not execute a single time and the context is reset to the
initial state. This allows the filter to be reused many
times.
When designing your own filters, the challenging part is to
decide what will be in the context. For line breaking, for
instance, it could be the number of bytes that still fit in the
instance, it could be the number of bytes left in the
current line. For Base64 encoding, it could be a string
with the bytes that remain after the division of the input
into 3-byte atoms. The MIME module in the \texttt{LuaSocket}
@ -300,22 +294,19 @@ distribution has many other examples.
\section{Filter chains}
Chains greatly increase the power of filters. For example,
Chains add a lot to the power of filters. For example,
according to the standard for Quoted-Printable encoding,
text should be normalized to a canonic end-of-line marker
prior to encoding. After encoding, the resulting text must
be broken into lines of no more than 76 characters, with the
use of soft line breaks (a line terminated by the \texttt{=}
sign). To help specifying complex transformations like
this, we define a chain factory that creates a composite
filter from one or more filters. A chained filter passes
data through all its components, and can be used wherever a
primitive filter is accepted.
text must be normalized to a canonic end-of-line marker
prior to encoding. To help specifying complex
transformations like this, we define a chain factory that
creates a composite filter from one or more filters. A
chained filter passes data through all its components, and
can be used wherever a primitive filter is accepted.
The chaining factory is very simple. The auxiliary
function~\texttt{chainpair} chains two filters together,
taking special care if the chunk is the last. This is
because the final \nil\ chunk notification has to be
because the final \texttt{nil} chunk notification has to be
pushed through both filters in turn:
\begin{quote}
\begin{lua}
@ -331,9 +322,9 @@ end
@stick#
function filter.chain(...)
local f = select(1, ...)
for i = 2, select('@#', ...) do
f = chainpair(f, select(i, ...))
local f = arg[1]
for i = 2, @#arg do
f = chainpair(f, arg[i])
end
return f
end
@ -346,11 +337,11 @@ define the Quoted-Printable conversion as such:
\begin{quote}
\begin{lua}
@stick#
local qp = filter.chain(normalize(CRLF), encode("quoted-printable"),
wrap("quoted-printable"))
local input = source.chain(source.file(io.stdin), qp)
local output = sink.file(io.stdout)
pump.all(input, output)
local qp = filter.chain(normalize("\r\n"),
encode("quoted-printable"))
local in = source.chain(source.file(io.stdin), qp)
local out = sink.file(io.stdout)
pump.all(in, out)
%
\end{lua}
\end{quote}
@ -369,14 +360,14 @@ gives a final destination to the data.
\subsection{Sources}
A source returns the next chunk of data each time it is
invoked. When there is no more data, it simply returns~\nil.
In the event of an error, the source can inform the
caller by returning \nil\ followed by the error message.
invoked. When there is no more data, it simply returns
\texttt{nil}. In the event of an error, the source can inform the
caller by returning \texttt{nil} followed by an error message.
Below are two simple source factories. The \texttt{empty} source
returns no data, possibly returning an associated error
message. The \texttt{file} source yields the contents of a file
in a chunk by chunk fashion:
message. The \texttt{file} source works harder, and
yields the contents of a file in a chunk by chunk fashion:
\begin{quote}
\begin{lua}
@stick#
@ -407,7 +398,7 @@ A filtered source passes its data through the
associated filter before returning it to the caller.
Filtered sources are useful when working with
functions that get their input data from a source (such as
the pumps in our examples). By chaining a source with one or
the pump in our first example). By chaining a source with one or
more filters, the function can be transparently provided
with filtered data, with no need to change its interface.
Here is a factory that does the job:
@ -415,18 +406,14 @@ Here is a factory that does the job:
\begin{lua}
@stick#
function source.chain(src, f)
return function()
if not src then
return nil
end
return source.simplify(function()
if not src then return nil end
local chunk, err = src()
if not chunk then
src = nil
return f(nil)
else
return f(chunk)
end
end
else return f(chunk) end
end)
end
%
\end{lua}
@ -434,20 +421,20 @@ end
\subsection{Sinks}
Just as we defined an interface for source of data,
Just as we defined an interface a data source,
we can also define an interface for a data destination.
We call any function respecting this
interface a \emph{sink}. In our first example, we used a
file sink connected to the standard output.
Sinks receive consecutive chunks of data, until the end of
data is signaled by a \nil\ input chunk. A sink can be
data is signaled by a \texttt{nil} chunk. A sink can be
notified of an error with an optional extra argument that
contains the error message, following a \nil\ chunk.
contains the error message, following a \texttt{nil} chunk.
If a sink detects an error itself, and
wishes not to be called again, it can return \nil,
wishes not to be called again, it can return \texttt{nil},
followed by an error message. A return value that
is not \nil\ means the sink will accept more data.
is not \texttt{nil} means the source will accept more data.
Below are two useful sink factories.
The table factory creates a sink that stores
@ -482,7 +469,7 @@ end
Naturally, filtered sinks are just as useful as filtered
sources. A filtered sink passes each chunk it receives
through the associated filter before handing it down to the
through the associated filter before handing it to the
original sink. In the following example, we use a source
that reads from the standard input. The input chunks are
sent to a table sink, which has been coupled with a
@ -492,10 +479,10 @@ standard out:
\begin{quote}
\begin{lua}
@stick#
local input = source.file(io.stdin)
local output, t = sink.table()
output = sink.chain(normalize(CRLF), output)
pump.all(input, output)
local in = source.file(io.stdin)
local out, t = sink.table()
out = sink.chain(normalize("\r\n"), out)
pump.all(in, out)
io.write(table.concat(t))
%
\end{lua}
@ -503,11 +490,11 @@ io.write(table.concat(t))
\subsection{Pumps}
Although not on purpose, our interface for sources is
compatible with Lua iterators. That is, a source can be
neatly used in conjunction with \texttt{for} loops. Using
our file source as an iterator, we can write the following
code:
Adrian Sietsma noticed that, although not on purpose, our
interface for sources is compatible with Lua iterators.
That is, a source can be neatly used in conjunction
with \texttt{for} loops. Using our file
source as an iterator, we can write the following code:
\begin{quote}
\begin{lua}
@stick#
@ -552,22 +539,20 @@ end
The \texttt{pump.step} function moves one chunk of data from
the source to the sink. The \texttt{pump.all} function takes
an optional \texttt{step} function and uses it to pump all the
data from the source to the sink.
Here is an example that uses the Base64 and the
line wrapping filters from the \texttt{LuaSocket}
distribution. The program reads a binary file from
data from the source to the sink. We can now use everything
we have to write a program that reads a binary file from
disk and stores it in another file, after encoding it to the
Base64 transfer content encoding:
\begin{quote}
\begin{lua}
@stick#
local input = source.chain(
local in = source.chain(
source.file(io.open("input.bin", "rb")),
encode("base64"))
local output = sink.chain(
local out = sink.chain(
wrap(76),
sink.file(io.open("output.b64", "w")))
pump.all(input, output)
pump.all(in, out)
%
\end{lua}
\end{quote}
@ -576,17 +561,19 @@ The way we split the filters here is not intuitive, on
purpose. Alternatively, we could have chained the Base64
encode filter and the line-wrap filter together, and then
chain the resulting filter with either the file source or
the file sink. It doesn't really matter.
the file sink. It doesn't really matter. The Base64 and the
line wrapping filters are part of the \texttt{LuaSocket}
distribution.
\section{Exploding filters}
Our current filter interface has one serious shortcoming.
Consider for example a \texttt{gzip} decompression filter.
During decompression, a small input chunk can be exploded
into a huge amount of data. To address this problem, we
decided to change the filter interface and allow exploding
filters to return large quantities of output data in a chunk
by chunk manner.
Our current filter interface has one flagrant shortcoming.
When David Burgess was writing his \texttt{gzip} filter, he
noticed that a decompression filter can explode a small
input chunk into a huge amount of data. To address this
problem, we decided to change the filter interface and allow
exploding filters to return large quantities of output data
in a chunk by chunk manner.
More specifically, after passing each chunk of input to
a filter, and collecting the first chunk of output, the
@ -595,11 +582,11 @@ filtered data is left. Within these secondary calls, the
caller passes an empty string to the filter. The filter
responds with an empty string when it is ready for the next
input chunk. In the end, after the user passes a
\nil\ chunk notifying the filter that there is no
\texttt{nil} chunk notifying the filter that there is no
more input data, the filter might still have to produce too
much output data to return in a single chunk. The user has
to loop again, now passing \nil\ to the filter each time,
until the filter itself returns \nil\ to notify the
to loop again, now passing \texttt{nil} to the filter each time,
until the filter itself returns \texttt{nil} to notify the
user it is finally done.
Fortunately, it is very easy to modify a filter to respect
@ -612,13 +599,13 @@ Interestingly, the modifications do not have a measurable
negative impact in the performance of filters that do
not need the added flexibility. On the other hand, for a
small price in complexity, the changes make exploding
filters practical.
filters practical.
\section{A complex example}
The LTN12 module in the \texttt{LuaSocket} distribution
implements all the ideas we have described. The MIME
and SMTP modules are tightly integrated with LTN12,
implements the ideas we have described. The MIME
and SMTP modules are especially integrated with LTN12,
and can be used to showcase the expressive power of filters,
sources, sinks, and pumps. Below is an example
of how a user would proceed to define and send a
@ -635,9 +622,9 @@ local message = smtp.message{
to = "Fulano <fulano@example.com>",
subject = "A message with an attachment"},
body = {
preamble = "Hope you can see the attachment" .. CRLF,
preamble = "Hope you can see the attachment\r\n",
[1] = {
body = "Here is our logo" .. CRLF},
body = "Here is our logo\r\n"},
[2] = {
headers = {
["content-type"] = 'image/png; name="luasocket.png"',
@ -678,18 +665,6 @@ abstraction for final data destinations. Filters define an
interface for data transformations. The chaining of
filters, sources and sinks provides an elegant way to create
arbitrarily complex data transformations from simpler
components. Pumps simply push the data through.
\section{Acknowledgements}
The concepts described in this text are the result of long
discussions with David Burgess. A version of this text has
been released on-line as the Lua Technical Note 012, hence
the name of the corresponding LuaSocket module,
\texttt{ltn12}. Wim Couwenberg contributed to the
implementation of the module, and Adrian Sietsma was the
first to notice the correspondence between sources and Lua
iterators.
components. Pumps simply move the data through.
\end{document}

View File

@ -12,12 +12,3 @@ clean:
pdf: ltn012.pdf
open ltn012.pdf
test: gem.so
gem.o: gem.c
gcc -c -o gem.o -Wall -ansi -W -O2 gem.c
gem.so: gem.o
export MACOSX_DEPLOYMENT_TARGET="10.3"; gcc -bundle -undefined dynamic_lookup -o gem.so gem.o