Skip to content

Commit

Permalink
Initial commit: ape-6.6-110816
Browse files Browse the repository at this point in the history
  • Loading branch information
@tkuhn
tkuhn committed Nov 21, 2011
0 parents commit 72c6ab2
Show file tree
Hide file tree
Showing 96 changed files with 175,083 additions and 0 deletions.
2 changes: 2 additions & 0 deletions .gitignore
Original file line number Diff line number Diff line change
@@ -0,0 +1,2 @@
/ape.exe
/.project
845 changes: 845 additions & 0 deletions LICENSE.txt

Large diffs are not rendered by default.

323 changes: 323 additions & 0 deletions README.html
Original file line number Diff line number Diff line change
@@ -0,0 +1,323 @@
<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
<html><head>
<meta http-equiv="content-type" content="text/html; charset=UTF-8" />
<title>ape</title>

</head>
<body>

<h1 class="wiki">APE - ACE Parsing Engine</h1>

<p>
<i>Authors: Kaarel Kaljurand, Norbert E. Fuchs, Tobias Kuhn</i>

</p><p>
<i>Last update: 2011-08-16</i>

</p><h2 class="wiki">Introduction</h2>

<p>
This document explains how APE (ACE Parsing Engine) is compiled and used.

</p><p>
In order to compile and run APE, you first need to install a recent version of SWI Prolog.
SWI Prolog is free software and can be downloaded from <a href="http://www.swi-prolog.org/">http://www.swi-prolog.org</a>.
Note that you minimally need to install the following SWI Prolog packages:
<code>clib</code>, <code>sgml</code>, and <code>http</code>.
To view the documentation embedded in the source files you also need <code>pldoc</code>.

</p><h2 class="wiki">Compilation</h2>

<p>
Before you can run APE, you have to compile the APE source code.
Just execute the file <code>make_exe.bat</code> in the case of Windows
or <code>make_exe.sh</code> in the case of Mac OS X, Linux, or any other Unix system. Both files are located in the root directory
of the APE distribution.
As a result (and given that there were no compilation errors),
a new file <code>ape.exe</code> is created into the current directory.

</p><h2 class="wiki">Execution</h2>

<p>
APE has to be executed from the command-line. In the command-line terminal, go to the
root directory of APE (where <code>ape.exe</code> is located).
Then type <code>ape.exe</code> in the case of Windows or <code>./ape.exe</code> otherwise.
As a result, you should see the following output:

</p><pre class="code">Attempto Parsing Engine for ACE 6.6, version 6.6-110816
Copyright 2008-2011 Attempto Group, University of Zurich
This program comes with ABSOLUTELY NO WARRANTY.
This is free software, and you are welcome to redistribute it under certain conditions.
Please visit http://attempto.ifi.uzh.ch for details.

Command-line arguments:
-text "TEXT" The input ACE text. If neither -text nor -file is present then the ACE text is read from stdin.
-file FILENAME The name or URL of the input file containing the ACE text.
-ulextext "TEXT" The user lexicon (taken from a string).
-ulexfile FILENAME The user lexicon (taken from a file or URL).
-solo OUTPUT Output just one output component. OUTPUT has to be one of {paraphrase,paraphrase1,paraphrase2,owlfss,owlfsspp,owlrdf,owlxml,fol,pnf,tptp,tokens,syntax,syntaxpp,syntaxd,syntaxdpp,drs,drsxml,drspp,drshtml}.
-cdrs Output the DRS as a Prolog term.
-cdrsxml Output the DRS in XML.
-cdrspp Output the DRS in pretty-printed form in plain text.
-cdrshtml Output the DRS in pretty-printed form in HTML.
-cparaphrase Output a paraphrase which is a "best-effort" combination of paraphrase1 and paraphrase2.
-cparaphrase1 Output a paraphrase which uses full sentences instead of relative clauses.
-cparaphrase2 Output a paraphrase which uses relative clauses instead of full sentences.
-ctokens Output tokens as a Prolog list of lists.
-csentences Output sentences as a Prolog list.
-csyntax Output simplified syntax trees as a Prolog list.
-csyntaxpp Output simplified syntax trees in pretty-printed form.
-csyntaxd Output plain syntax trees as a Prolog list (for debugging).
-csyntaxdpp Output plain syntax trees in pretty-printed form (for debugging).
-cowlfss Output OWL/SWRL in the Functional-Style Syntax representation (as Prolog term).
-cowlfsspp Output OWL/SWRL in the Functional-Style Syntax representation (pretty-printed).
-cowlxml Output OWL/SWRL in the XML representation.
-cowlrdf Output OWL/SWRL in the RDF/XML representation. DEPRECATED
-cfol Output standard first-order logic representations (default form) of the DRS as a Prolog term.
-cpnf Output standard first-order logic representations (prenex normal form) of the DRS as a Prolog term.
-ctptp Output TPTP representation of the DRS.
-uri URI URI for the OWL outputs.
-noclex Ignore the lexicon entries that are compiled into the executable.
-guess Guess the word-class of unknown words.
-server Launch a socket interface to APE at port 2766 (0xACE).
-httpserver Launch an HTTP interface to APE at port 8000.
-port NUMBER Override the default port of either the socket or the HTTP interface.
-version Shows version information.
-help Shows this help page.
</pre>

<p>
APE can be used via four different interfaces:
</p><ul>
<li>command-line interface</li>
<li>socket interface</li>
<li>HTTP interface (webservice)</li>
<li>directly from Prolog or Java</li>
</ul>

<p>
All these possibilities are described in the following sections.

</p><h3 class="wiki">Command-line interface to APE</h3>

<p>
The following command parses the text
"John waits." and outputs the DRS in XML representation and the syntax tree:

</p><pre class="code"> ./ape.exe -text 'John waits.' -cdrsxml -csyntax
</pre>

<p>
In the case of Windows, you have to omit the first two characters "<code>./</code>". The next example parses the text that
is inside of the file <code>ace.txt</code> and outputs the OWL FSS representation:

</p><pre class="code"> ./ape.exe -file ace.txt -solo owlfss
</pre>

<p>
If you omit both arguments, <code>text</code> and <code>file</code>, the ACE text is read from the standard input:

</p><pre class="code"> echo 'Every mammal is an animal.' | ./ape.exe -solo drspp
</pre>

<p>
Note that this does not work under Windows.

</p><p>
If you just execute the line

</p><pre class="code"> ./ape.exe -solo drspp
</pre>

<p>
then the terminal waits for an input. In this case, you can type your ACE text into the terminal window.
Once you have done so, press <code>Enter</code> and <code>Ctrl-D</code> to tell the terminal that you are finished.
The output (the pretty printed DRS in our case) is then shown below the ACE text you just entered.
Again, this does not work under Windows.

</p><h3 class="wiki">Socket interface to APE</h3>

<p>
A socket interface to APE is started by giving the argument <code>-server</code> to APE.
The default port, 2766 (0xACE), can be overridden by specifying a different
port number as a value to the argument <code>-port</code>.

</p><p>
A good way to start the server on a Unix command-line (e.g. <i>bash</i>) is:

</p><pre class="code">nohup swipl -x ape.exe -- -server -port 3453 &gt; stdout.txt 2&gt; stderr.txt &amp;
</pre>

<p>
On Mac OS X, one could use <i>launchctl</i> instead.

</p><p>
The socket interface knows only one command:

</p><pre class="code">get(Parameters).
</pre>

<p>
where <i>Parameters</i> is a list of parameters accepted by <code>get_ape_results/2</code>.
Note the dot at the end of the command! (There should also be a newline after the dot.)
Given the input, the server runs <code>get_ape_results/2</code> (with a 20 second timelimit)
on it and sends back the results, followed by <code>APESERVERSTREAMEND</code> on a separate line.
If the input command is syntactically incorrect then an error message is logged into STDERR.
In any case, the connection to the client is closed.

</p><p>
Examples of input commands:
</p><ul>
<li><code>get([text='Every man is a human.', cparaphrase1=on]).</code></li>
<li><code>get([text='Every man is a human.', solo=paraphrase1]).</code></li>
<li><code>get([text='Every man is a a human.', cparaphrase1=on]).</code></li>
<li><code>get([text='Every man is a human.', cinput=on, cdrs=on, cparaphrase=on, ctokens=on, csyntax=on]).</code></li>
</ul>

<p>
Session example, assuming that the APE socket server listens at port 3453.

</p><pre class="code">$ telnet 127.0.0.1 3453
Trying 127.0.0.1...
Connected to localhost.
Escape character is '^]'.
get([text='Every man is a human.', solo=paraphrase1]).
If there is a man X1 then the man X1 is a human.
APESERVERSTREAMEND
Connection closed by foreign host.
</pre>

<h3 class="wiki">HTTP interface to APE</h3>

<p>
An HTTP interface to APE is started by giving the argument <code>-httpserver</code> to APE.
The default port, 8000, can be overridden by specifying a different
port number as a value to the argument <code>-port</code>.

</p><p>
A good way to start the server on a Unix command-line (e.g. <i>bash</i>) is:

</p><pre class="code">nohup swipl -x ape.exe -- -httpserver -port 8001 &gt; stdout.txt 2&gt; stderr.txt &amp;
</pre>

<p>
On Mac OS X, one could use <i>launchctl</i> instead.

</p><p>
In case there is an error message in the error output, saying that

</p><pre class="code">ERROR: socket_error(Address already in use)
</pre>

<p>
then try another port number. For example, one would get this error when running
the above command twice: during the first time a server is started which starts
to listen requests at port 8001, during the second time a second server is started
but since the port is already being used by the first server, there is a conflict
and the second server crashes.

</p><p>
You can test the server by loading the following URL in your browser.

</p><pre class="code">http://localhost:8001/?text=Every+man+is+a+human.&amp;solo=drshtml
</pre>

<p>
The result should be an HTML-rendering of the DRS of the sentence "Every man is a human.".

</p><p>
The complete description of this webservice interface as well as some example clients
are available at <a href="http://attempto.ifi.uzh.ch/site/docs/ape_webservice.html">http://attempto.ifi.uzh.ch/site/docs/ape_webservice.html</a>.
(Note though that the parameter 'ulexreload' is not supported by this HTTP interface.)

</p><p>
Note that the parameter names and values, and their meaning is the same as for the
command-line client. Also the results' format is the same. The only difference
is that for security reasons the webservice cannot access local files, i.e. the input ACE text
can be passed as a string or via a pointer an HTTP resource (i.e. URL) but not
via pointing to a local file (such as <code>/var/acetexts/sometext.ace.txt</code>).

</p><h3 class="wiki">Using APE from Java programs</h3>

<p>
The <code>ACEParser</code> implementation <code>APELocal</code> (in <code>attempto_ape.jar</code>,
see <a href="http://attempto.ifi.uzh.ch/site/downloads/">http://attempto.ifi.uzh.ch/site/downloads/</a>)
provides a high level Java interface to <code>ape.exe</code>.
It is effectively a Java-style front-end to querying the predicate <code>get_ape_results/2</code>.

</p><p>
Even if you need to write your own Java interface to <code>ape.exe</code>, it might be useful
to take a look at the source code of <code>APELocal.java</code> for instructions on how to
access JPL, the bidirectional Prolog/Java interface, which is included in the SWI Prolog
installation.

</p><h3 class="wiki">Some examples</h3>

<p>
ACE text from URL

</p><pre class="code">./ape.exe -file http://attempto.ifi.uzh.ch/site/acetexts/example1.ace.txt -solo owlxml
</pre>

<p>
Lexicon from string

</p><pre class="code">./ape.exe -text 'Every mman is a hhuman.' -ulextext 'noun_sg(mman, mman, masc). noun_sg(hhuman, hhuman, neutr).' -cparaphrase1
</pre>

<p>
Reading from STDIN and writing to STDOUT can be used in order to chain several executions
of APE together. The following example paraphrases the paraphrase of "Every man is a human.".

</p><pre class="code">./ape.exe -text 'Every man is a human.' -solo paraphrase1 | ./ape.exe -solo paraphrase2
</pre>

<p>
The following commands have the same meaning.

</p><pre class="code">./ape.exe -httpserver
swipl -x ape.exe -- -httpserver
swipl -x ape.exe -g http_server
</pre>

<p>
The only difference is in the the way SWI Prolog is called, either via the full path name
embedded in <code>ape.exe</code> or via the name <code>swipl</code> which the command-line environment
must resolve to the full path name.

</p><p>
By overriding the goal (with <code>-g</code>), it is also possible to execute SWI-Prolog commands
in the context of <code>ape.exe</code>.
For example, the following command displays the source code of <code>ape.exe</code>.

</p><pre class="code">swipl -x ./ape.exe -g listing
</pre>

<h2 class="wiki">Code</h2>

<p>
The distribution includes the following packages:
</p><ul>
<li>logger/ contains the error logger module</li>
<li>lexicon/ contains various lexicon files, notably a content words lexicon with ~2,000 entries</li>
<li>utils/ contains various modules, mostly for translating the
Discourse Representation Structure (DRS) generated by APE into other
logical forms</li>
<li>parser/ contains the Attempto Parsing Engine (APE) (tokenizer, grammar files, anaphoric reference resolver)</li>
</ul>

<p>
This listing reflects the dependencies: <code>logger</code> depends on no other package. <code>lexicon</code> depends
only on <code>logger</code>. The package <code>utils</code> depends only on <code>logger</code> and <code>utils</code>. The package <code>parser</code>,
finally, depends on all three other packages.
The files in the root directory depend on those packages, but not vice versa.

</p><p>
For more information consult the commented source files.
Provided that you have PlDoc installed (SWI-Prolog package <code>pldoc</code>),
you can view the documentation by:

</p><pre class="code">?- doc_server(1234), [get_ape_results], doc_browser.
</pre>

</body></html>
Loading

0 comments on commit 72c6ab2

Please sign in to comment.