Merge branch 'master' of github.com:pks/lfpe

12 files changed, 167 insertions, 49 deletions

diff --git a/.htaccess b/.htaccess
index a2cb1b7..cf61c47 100644
--- a/.htaccess
+++ b/.htaccess
@@ -1,5 +1,5 @@
 AuthName "Post-Editing Interface"
 AuthType Basic
-AuthUserFile /fast_scratch/simianer/lfpe/.htpasswd
+AuthUserFile /srv/postedit/.htpasswd
 require valid-user
 
diff --git a/README.md b/README.md
index 403d745..a3d4e96 100644
--- a/README.md
+++ b/README.md
@@ -2,20 +2,35 @@
 Post-editing interface for learning from post-edited machine translations.
 
 # Setup
+
+`
+    export BASE_DIR=/srv/postedit
+`
+
 ## nanomsg lib
-    export LD_LIBRARY_PATH=/fast_scratch/simianer/lfpe/lib/nanomsg-0.5-beta/lib
+`
+    export LD_LIBRARY_PATH=$BASE_DIR/lib/nanomsg-0.5-beta/lib
+`
 
 ## ruby
-    [see $(pwd)/lib/ruby/gems/nanomsg-0.4.0/ext/extconf.rb]
-    gem install nanomsg -i $(pwd)/lib/ruby
-    export GEM_PATH=/fast_scratch/simianer/lfpe/lib/ruby/:$GEM_PATH
+`
+    [see $BASE_DIR/lib/ruby/gems/nanomsg-0.4.0/ext/extconf.rb]
+    gem install nanomsg -i $BSAE_DIR/lib/ruby
+    export GEM_PATH=$BASE_DIR/lib/ruby/:$GEM_PATH
+`
 
 ## iptables
+`
     iptables -A INPUT -i eth0 -p tcp -m multiport --dports 50000:50100 -j ACCEPT
+`
 
 ## apache
+`
     ln -s /etc/apache2/sites-available/lfpe /etc/apache2/sites-enabled/020-lfpe
+`
 
 ## python
-    export PYTHONPATH=/fast_scratch/simianer/lfpe/lib/python:$PYTHONPATH
+`
+    export PYTHONPATH=$BASE_DIR/lib/python:$PYTHONPATH
+`
 
diff --git a/external/lfpe-apache b/external/lfpe-apache
index e4de4be..eec5e5c 100644
--- a/external/lfpe-apache
+++ b/external/lfpe-apache
@@ -3,13 +3,13 @@
 
 	ServerAdmin simianer@cl.uni-heidelberg.de
 
-	DocumentRoot /fast_scratch/simianer/lfpe/lfpe
+	DocumentRoot /srv/postedit/lfpe
 
-	ErrorLog /fast_scratch/simianer/lfpe/lfpe/logs/apache2.error.log
+	ErrorLog /srv/postedit/lfpe/logs/apache2.error.log
 	LogLevel warn
-	CustomLog /fast_scratch/simianer/lfpe/lfpe/logs/apache2.access.log combined
+	CustomLog /srv/postedit/lfpe/logs/apache2.access.log combined
 
-	<Directory /fast_scratch/simianer/lfpe/lfpe>
+	<Directory /srv/postedit/lfpe>
                 Options +FollowSymLinks
                 AllowOverride All
                 order allow,deny
diff --git a/inc/db.inc.php b/inc/db.inc.php
index f08eda5..ed45f7b 100644
--- a/inc/db.inc.php
+++ b/inc/db.inc.php
@@ -1,6 +1,6 @@
 <?php
 
-$SESSION_DIR="/fast_scratch/simianer/lfpe/sessions";
+$SESSION_DIR="/srv/postedit/sessions";
 $json = file_get_contents($SESSION_DIR."/".$_GET["key"]."/data.json");
 $db = json_decode($json);
 
diff --git a/interface.php b/interface.php
index 8df7fd0..46b07f0 100644
--- a/interface.php
+++ b/interface.php
@@ -33,7 +33,7 @@
     <tr>
       <td align="right">Target:</td>
       <td>
-        <textarea id="target_textarea" name="target" cols="80" rows="1" onkeypress="catch_return(event);" disabled></textarea>
+        <textarea id="target_textarea" name="target" cols="80" rows="1" onkeypress="TEXT_handle_keypress(event);" disabled></textarea>
       </td>
     </tr>
   </table>
@@ -52,7 +52,7 @@ Note that the source word may be distorted.
 <div>
   <button id="help_button" class="button" onclick="$('#help').toggle('blind')">Help</button>
   <button id="pause_button" class='button' type="button" onclick="pause()">Pause</button>
-  <button id="reset_button" class='button' type="button" onclick="DE_init()">Reset</button>
+  <button id="reset_button" class='button' type="button" onclick="reset()">Reset</button>
   <button id="next" type="button" class='button' onclick="next();">Start/Continue</button>
   <span id="status"><strong>Working: <span id="status_detail">...</span></strong> <img src="static/ajax-loader-large.gif" width="20px" /></span>
 </div>
@@ -103,5 +103,6 @@ Note that the source word may be distorted.
 <textarea style="display:none" id="ui_type"           ><?php echo $_GET["ui_type"]; ?></textarea>
 <textarea style="display:none" id="data"              ></textarea>
 <textarea style="display:none" id="original_svg"      ></textarea>
+<textarea style="display:none" id="original_mt"       ></textarea>
 <!-- /Data -->
 
diff --git a/js/interface.js b/js/interface.js
index c946e0c..f009641 100644
--- a/js/interface.js
+++ b/js/interface.js
@@ -5,6 +5,9 @@
 var data,    // global data object
     ui_type; // 't' (text) or 'g' (graphical)
 
+var TEXT_count_click=0,
+    TEXT_count_kbd=0;
+
 /*
  * cross-site request
  *
@@ -108,6 +111,18 @@ var catch_return = function (e)
   return false;
 }
 
+var TEXT_handle_keypress = function (e)
+{
+   if (e.keyCode == 13) {
+    e.preventDefault();
+    next();
+  }
+
+  TEXT_count_kbd += 1;
+
+  return false;
+}
+
 /*
  * working/not working
  *
@@ -230,7 +245,7 @@ var next =  function ()
   working();
 
   // get metadata stored in DOM
-  var base_url = "http://coltrane.cl.uni-heidelberg.de";
+  var base_url = "http://lemmy.cl.uni-heidelberg.de";
   var port     = document.getElementById("port").value;
   var key      = document.getElementById("key").value;
 
@@ -259,6 +274,8 @@ var next =  function ()
     post_edit = $.trim(target_textarea.value);
     send_data["post_edit"] = encodeURIComponent(post_edit);
     send_data['type'] = 't';
+    send_data["count_click"] = TEXT_count_click;
+    send_data["count_kbd"] = TEXT_count_kbd;
   }
 
   send_data["key"] = key;
@@ -334,7 +351,7 @@ var request_and_process_next = function ()
   var last_post_edit      = document.getElementById("last_post_edit");
 
   // get metadata stored in DOM
-  var base_url = "http://coltrane.cl.uni-heidelberg.de";
+  var base_url = "http://lemmy.cl.uni-heidelberg.de";
   var port     = document.getElementById("port").value;
   var key      = document.getElementById("key").value;
 
@@ -456,8 +473,11 @@ var request_and_process_next = function ()
         target_textarea.rows     = Math.round(translation.length/80+0.5);
       //raw_source_textarea.rows = Math.round(raw_source.length/80+0.5);
       target_textarea.focus();
+      $("#original_mt").val(target_textarea.value);
       target_textarea.selectionStart = 0;
       target_textarea.selectionEnd   = 0;
+      TEXT_count_click = 0;
+      TEXT_count_kbd = 0;
 
       // remember aux data in DOM
       current_seg_id.value = id;
@@ -493,10 +513,35 @@ var init_text_editor = function ()
 {
   document.getElementById("target_textarea").value     = "";
   document.getElementById("target_textarea").setAttribute("disabled", "disabled");
+  
+  TEXT_count_click = 0;
+  TEXT_count_kbd = 0;
+
+  $("#target_textarea").click(function () {
+    TEXT_count_click += 1;
+  });
 
   return false;
 }
 
+var get_ui_type = function ()
+{
+  return document.getElementById("ui_type").value;
+}
+
+var reset = function ()
+{
+  var ui_type = get_ui_type();
+  if (ui_type == "t") {
+    if (!$("#init").val()) return;
+    TEXT_count_click = 0;
+    TEXT_count_kbd = 0;
+    $("#target_textarea").val($("#original_mt").val());
+  } else if (ui_type == "g") {
+    DE_init()
+  }
+}
+
 /*
  * init site
  *
@@ -514,7 +559,7 @@ $().ready(function()
 
   not_working();
 
-  ui_type = document.getElementById("ui_type").value;
+  ui_type = get_ui_type();
 
   // graphical derivation editor
   if (ui_type == "g") {
@@ -525,6 +570,7 @@ $().ready(function()
     init_text_editor();
     document.getElementById("textboxes").style.display = "block";
   }
+
 });
 
 
diff --git a/phrase2_extraction/phrase2_extraction.rb b/phrase2_extraction/phrase2_extraction.rb
index 48dfd73..547e0be 100755
--- a/phrase2_extraction/phrase2_extraction.rb
+++ b/phrase2_extraction/phrase2_extraction.rb
@@ -6,8 +6,8 @@ module PhrasePhraseExtraction
 
 DEBUG                      = false
 MAX_NT                     = 2    # Chiang: 2
-MAX_SEED_NUM_WORDS         = 3    # Chiang: 10 words
-MAX_SRC_SZ                 = 3    # Chiang: 5 words
+MAX_SEED_NUM_WORDS         = 4    # Chiang: 10 words, -> phrases!
+MAX_SRC_SZ                 = 10   # Chiang: 5 words, -> words!
 FORBID_SRC_ADJACENT_SRC_NT = true # Chiang:true
 
 class Rule
@@ -51,6 +51,21 @@ class Rule
     return src_len
   end
 
+  def len_src_w
+    src_len = 0
+    @source.each { |i|
+      if i.is_a? String
+        src_len += i.split.size #1
+      else
+        i.each { |j|
+          src_len += source_context[j].split.size
+        }
+      end
+    }
+
+    return src_len
+  end
+
   def len_tgt
     tgt_len = 0
     @target.each { |i|
@@ -64,6 +79,21 @@ class Rule
     return tgt_len
   end
 
+  def len_tgt_w
+    tgt_len = 0
+    @target.each { |i|
+      if i.is_a? String
+        tgt_len += i.split.size
+      else
+        i.each { |j|
+          tgt_len += target_context[j].split.size
+        }
+      end
+    }
+
+    return tgt_len
+  end
+
   def to_s
     source_string = ""
     @source.each { |i|
@@ -625,7 +655,7 @@ end
 
 def PhrasePhraseExtraction.remove_too_long_src_sides rules
   return rules.reject { |r|
-    r.len_src > PhrasePhraseExtraction::MAX_SRC_SZ
+    r.len_src_w > PhrasePhraseExtraction::MAX_SRC_SZ
   }
 end
 
diff --git a/server.rb b/server.rb
index 5a95131..599fdbd 100755
--- a/server.rb
+++ b/server.rb
@@ -34,7 +34,7 @@ $oov_corrected.default   = false
 # #############################################################################
 # Daemons
 # #############################################################################
-DIR="/fast_scratch/simianer/lfpe"
+DIR="/srv/postedit"
 $daemons = {
   :tokenizer        => "#{DIR}/lfpe/util/nanomsg_wrapper.rb -a tokenize   -S '__ADDR__' -e #{EXTERNAL} -l #{TARGET_LANG}",
   :tokenizer_src    => "#{DIR}/lfpe/util/nanomsg_wrapper.rb -a tokenize   -S '__ADDR__' -e #{EXTERNAL} -l #{SOURCE_LANG}",
@@ -123,6 +123,8 @@ def init
     $env[name] = { :socket => sock, :pid => pid }
     port += 1
   }
+
+  send_recv :truecaser, "lOaD iT"
                                                                    #  lock file
   `touch #{LOCK_FILE}`
   $status = "Initialized"                                              # status
@@ -293,6 +295,8 @@ def process_next reply
     $db['svg']                << data['svg']
     $db['original_svg']       << data['original_svg']
     $db['durations']          << data['duration'].to_f
+    $db['count_click']        << data['count_click'].to_i
+    $db['count_kbd']          << data['count_kbd'].to_i
     $db['post_edits_display'] << send_recv(:detokenizer, post_edit)
     $last_processed_postedit = $db['post_edits_display'].last
     # 1. tokenize
@@ -406,7 +410,7 @@ def process_next reply
     end
                                                                 # - known rules
     logmsg :server, "annotating known rules"
-    $status = "Adding rules to grammar"                                # status
+    $status = "Adding rules to the grammar"                            # status
     match = {}
     $known_rules.each { |r|
       _,src,tgt,_,_ = splitpipe r
@@ -421,13 +425,18 @@ def process_next reply
         all_rules[j] = ar
       end
     }
-    WriteFile.new(grammar).write all_rules.join("\n")+"\n"
                                                            # - additional rules
-    $new_rules.each { |rule|
-     logmsg :server, "adding rule '#{rule}' to grammar '#{grammar}'"
-      s = splitpipe(rule)[1..2].map{|i|i.strip.lstrip}.join(" ||| ")
-      `echo "#{rule}" >> #{grammar}`
-    }
+    #logmsg :server, $new_rules.to_s
+    if $new_rules.size > 0
+      all_rules += $new_rules
+      #`echo "#{s}" >> #{grammar}`
+    end
+    WriteFile.new(grammar).write all_rules.join("\n")+"\n"
+    #$new_rules.each { |rule|
+    # logmsg :server, "adding rule '#{rule}' to grammar '#{grammar}'"
+    #  s = splitpipe(rule)[1..2].map{|i|i.strip.lstrip}.join(" ||| ")
+    #  `echo "#{rule}" >> #{grammar}`
+    #}
                                                             # 2. check for OOVs
     if !$oov_corrected[$db['progress']]
     $status = "Checking for OOVs"                                      # status
@@ -664,11 +673,21 @@ get '/reset_extractor' do                             # reset grammar extractor
   return "reset extractor: done"
 end
 
+get '/reset_grammars' do                               # reset grammar extractor
+  logmsg :server, "reset grammars"
+  return "locked" if $lock
+  `cp #{SESSION_DIR}/g/original/* #{SESSION_DIR}/g/`
+  $last_reply = nil
+
+  return "reset grammars: done"
+end
+
 get '/reset_new_rules' do                               # removed learned rules
   $new_rules.clear
   $known_rules.clear
   `rm #{WORK_DIR}/*.*_rules`
   `rm #{WORK_DIR}/g/*`
+  $last_reply = nil
 
   return "reset new rules: done"
 end
diff --git a/static/pattr-abstracts.html b/static/pattr-abstracts.html
index c24598c..09f5779 100644
--- a/static/pattr-abstracts.html
+++ b/static/pattr-abstracts.html
@@ -20,7 +20,7 @@ div.ex:hover {
 
 <h2>Abstracts</h2>
 
-<h3><a href="http://coltrane.cl.uni-heidelberg.de:60666/load/0">EP-0005734-A1</a> (H01H)</h3>
+<h3><a href="http://lemmy.cl.uni-heidelberg.de:60666/load/0">EP-0005734-A1</a> (H01H)</h3>
   <div class="ex">
     <strong>Ein elektromagnetisch betätigtes Schaltgerät, z. B. ein elektrisches Schütz, hat einen magnetischen Eisenkern, der aus zwei gleichen E-förmigen Magnetteilen besteht, nämlich dem die Wicklung tragenden Magnetkern (8) und dem Anker (7).</strong><br/><hr/>
     <strong>An electromagnetically operated switchgear, e.g. an electrical contactor, has a magnetic iron core which consists of two identical E- shaped magnet parts, specifically the magnet core (8), carrying the winding, and the armature (7).</strong>
@@ -38,7 +38,7 @@ div.ex:hover {
 
 <hr />
 
-<h3><a href="http://coltrane.cl.uni-heidelberg.de:60666/load/1">EP-0003301-A1</a> (A01N,C07C)</h3>
+<h3><a href="http://lemmy.cl.uni-heidelberg.de:60666/load/1">EP-0003301-A1</a> (A01N,C07C)</h3>
 <div class="ex">
  <strong>Sie weisen starke insektizide, akarizide und nematizide Eigenschaften auf.</strong><br/><hr/>
  <strong>They have strong insecticidal, acaricidal and nematicidal properties.</strong>
@@ -50,7 +50,7 @@ div.ex:hover {
 
 <hr/>
 
-<h3><a href="http://coltrane.cl.uni-heidelberg.de:60666/load/2">EP-0003578-A2</a> (F25B)</h3>
+<h3><a href="http://lemmy.cl.uni-heidelberg.de:60666/load/2">EP-0003578-A2</a> (F25B)</h3>
 <div class="ex">
  <strong>Die Erfindung bezieht sich auf den Kältemittelkreislauf (1) einer Wärmepumpe.</strong><br/><hr/>
  <strong>The invention relates to the refrigerant circuit (1) of a heat pump.</strong>
@@ -78,7 +78,7 @@ div.ex:hover {
 
 <hr/>
 
-<h3><a href="http://coltrane.cl.uni-heidelberg.de:60666/load/3">EP-0002017-A1</a> (C25B)</h3>
+<h3><a href="http://lemmy.cl.uni-heidelberg.de:60666/load/3">EP-0002017-A1</a> (C25B)</h3>
 <div class="ex">
  <strong>Anoden für Elektrolysezwecke bestehen aus einem elektrisch leitenden Trägerkörper und einer darauf aufgebrachten Schicht aus metallischem Silicium und/oder Germanium.</strong><br/><hr/>
  <strong>Anodes for electrochemical purposes are composed of an electrically conducting support body and a layer of metallic silicon and/or germanium applied thereto.</strong>
@@ -94,7 +94,7 @@ div.ex:hover {
 
  <hr/>
 
- <h3><a href="http://coltrane.cl.uni-heidelberg.de:60666/load/4">EP-0018427-A1</a> (G05B)</h3>
+ <h3><a href="http://lemmy.cl.uni-heidelberg.de:60666/load/4">EP-0018427-A1</a> (G05B)</h3>
 <div class="ex">
  <strong>Elektrische Steuerschaltung mit einem Signalgenerator­ teil (20), der in Abhängigkeit von einem Führungssignal auf­ einanderfolgende gleichmäßige Impulse erzeugt.</strong><br/><hr/>
  <strong>Electrical control circuit comprising a signal generator section (20) which generates successive uniform pulses in dependence on a control signal.</strong>
diff --git a/util/run_beta_test b/util/run_beta_test
index a0fe20f..9180b52 100755
--- a/util/run_beta_test
+++ b/util/run_beta_test
@@ -1,8 +1,8 @@
 #!/bin/zsh -x
 
-cd /fast_scratch/simianer/lfpe/lfpe/util
+cd /srv/postedit/lfpe/util
 ./kill; ./kill; ./kill;
-for i in ../../sessions/product_de-en_beta_test_*; do
+for i in `ls -1 ../../sessions/ | grep -v "_1_" | grep -v data | grep -v toy`; do
   echo $i
   echo $(basename $i)
   ./run_server $(basename $i) &; sleep 600;
diff --git a/util/run_server b/util/run_server
index 7d45583..a4b7a6c 100755
--- a/util/run_server
+++ b/util/run_server
@@ -1,15 +1,17 @@
 #!/bin/bash -x
 
-export LD_LIBRARY_PATH=/fast_scratch/simianer/lfpe/lib/nanomsg-0.5-beta/lib:$LD_LIBRARY_PATH
-export PYTHONPATH=/fast_scratch/simianer/lfpe/lib/python:$PYTHONPATH
-export GEM_PATH=/fast_scratch/simianer/lfpe/lib/ruby/:$GEM_PATH
-UTIL=/fast_scratch/simianer/lfpe/lfpe/util
+BASE_DIR=/srv/postedit
+export LD_LIBRARY_PATH=$BASE_DIR/lib/nanomsg-0.5-beta/lib:$LD_LIBRARY_PATH
+export PYTHONPATH=$BASE_DIR/lib/python:$PYTHONPATH
+export GEM_PATH=$BASE_DIR/lib/ruby/:$GEM_PATH
+UTIL=$BASE_DIR/lfpe/util
 SESSION=$1
-DIR=/fast_scratch/simianer/lfpe/sessions/$SESSION
+SESSION_DIR=$BASE_DIR/sessions/$SESSION
 
-rm $DIR/work/lockfile
-rm -r $DIR/work/
-mkdir -p $DIR/work
-cp $DIR/data.json.original $DIR/data.json
-$UTIL/../server.rb $DIR/conf.rb &>$DIR/work/session.out
+rm $SESSION_DIR/work/lockfile
+rm -r $SESSION_DIR/work/
+mkdir -p $SESSION_DIR/work
+cp $SESSION_DIR/data.json.original $SESSION_DIR/data.json
+cp $SESSION_DIR/g/original/* $SESSION_DIR/g/
+$UTIL/../server.rb $SESSION_DIR/conf.rb &>$SESSION_DIR/work/session.out
 
diff --git a/views/debug.haml b/views/debug.haml
index 4ebb0a0..3c1e006 100644
--- a/views/debug.haml
+++ b/views/debug.haml
@@ -43,10 +43,13 @@
         %a.ajax{:tgt => "/reset_weights",        :href => "#controls"} Reset weights
       %li
         %a.ajax{:tgt => "/reset_learning_rates", :href => "#controls"} Reset learning rates
+      /
+        %li
+          %a.ajax{:tgt => "/reset_extractor",       :href => "#controls"} Reset extractor
       %li
-        %a.ajax{:tgt => "/reset_extractor",       :href => "#controls"} Reset extractor
+        %a.ajax{:tgt => "/reset_grammars",       :href => "#controls"} Reset grammars
       %li
-        %a.ajax{:tgt => "/reset_new_rules",       :href => "#controls"} Reset new rules
+        %a.ajax{:tgt => "/reset_new_rules",      :href => "#controls"} Reset new rules
       /
         %li
           %a.ajax{:tgt => "/shutdown",              :href => "#controls"} Initiate shutdown
@@ -141,6 +144,8 @@
     %p.updated <strong>Number of updates:</strong>  #{pairwise_ranking_data["num_up"]}
     %p.updated <strong>Updated features:</strong>   #{pairwise_ranking_data["updated_features"]}
     %p <strong>Duration:</strong>                   #{data["durations"][progress]}ms
+    %p <strong>Keypresses:</strong>                 #{data["count_kbd"][progress]}
+    %p <strong>Clicks:</strong>                     #{data["count_click"][progress]}
 
     %h3 Derivation
     %p
@@ -299,9 +304,9 @@
       %tr
         %td Shape_*
         %td.left Indicator features for rule shapes (39 in total)
-      %tr
-        %td IsSupportedOnline
-        %td.left Rules with support from local context (added by Denkowski's online suffix array extractor)
+      /=%tr
+      /=  %td IsSupportedOnline
+      /=  %td.left Rules with support from local context (added by Denkowski's online suffix array extractor)
 
     %p.up
       %a{ :href => "#" } ^ up