From d631879609b4738494c433581b49bea4288db5c3 Mon Sep 17 00:00:00 2001
From: Johannes Heinecke <johannes.heinecke@orange.fr>
Date: Sat, 29 Oct 2022 12:37:04 +0200
Subject: [PATCH] check for absent features/misc-keys

---
 CHANGES.md                                     |  1 +
 doc/mass_editing.md                            |  7 ++++---
 gui/index.html                                 |  6 +++++-
 .../com/orange/labs/conllparser/Conditions.g4  |  8 ++++----
 .../orange/labs/conllparser/CEvalVisitor.java  | 18 +++++++++++++++---
 .../com/orange/labs/conllparser/ConllWord.java |  4 ++--
 6 files changed, 31 insertions(+), 13 deletions(-)

diff --git a/CHANGES.md b/CHANGES.md
index 0a689bf..df43947 100644
--- a/CHANGES.md
+++ b/CHANGES.md
@@ -4,6 +4,7 @@
 * more tests for value access
 * added compatibility test `~` in addition to strict comparison with `=`
 * validator shortcut changed from `=` to `!`
+* check for absent features/misc-keys
 
 ## Version 2.18.0
 * extension to mass-edit/complex search&replace: possibility to search heads/childs etc with same Feature value or same UPOS etc
diff --git a/doc/mass_editing.md b/doc/mass_editing.md
index 3687a93..2137640 100644
--- a/doc/mass_editing.md
+++ b/doc/mass_editing.md
@@ -30,17 +30,18 @@ Examples:
   * `Id:` (Values: integer)
   * `MWT:` (Values: length of the multi-word token `[2-9]`)
   * `IsEmpty` (no value, true if the current node is empty)
-  * `IsMWT` (no value, true if the current node is empty)
+  * `IsMWT` (no value, true if the current node is a MWT)
 
 `Form:`, `Lemma:` and `Xpos:` can contain simple regular expression (only the character ')' cannot be used
+In order to check for the absence of a given Featurename in the Feature or Misc column, use the following:
+  * `Feat:Gender:` true if the cyurrent word has no feature `Gender`
 
 `EUD` cannot deal (yet) with empty word ids (`n.m`)
 
- `Lemma` and `Form` can have either a regex as argument or a filename of a file which contains a list of forms or lemmas:
+`Lemma` and `Form` can have either a regex as argument or a filename of a file which contains a list of forms or lemmas:
   * `Lemma:sing.* > misc:"Value=Sing"`
   * `Lemma:#mylemmas.txt > misc:"Value=Sing"` (if the file `mylemmas.txt` does not exist, the condition is false)
 
-
 In addition to key keys listed above, four functions are available to take the context of the token into account:
 * `child()` child of current token
 * `head()` head of current token
diff --git a/gui/index.html b/gui/index.html
index d7741a3..b8eb911 100644
--- a/gui/index.html
+++ b/gui/index.html
@@ -489,7 +489,11 @@ <h3>Complex search</h3>
                             <li> <span class="userinputdoc">IsEmpty</span> (no value, true if the current node is empty)</li>
                             <li> <span class="userinputdoc">IsMWT</span> (no value, true if the current node is a multi token word)</li>
                         </ul>
-
+                        In order to check for the absence of a given Featurename in the Feature or Misc column, use the following:
+                        <ul>
+                            <li><span class="userinputdoc">Feat:Gender:</span> true if the cyurrent word has no feature <span class="userinputdoc">Gender</span>
+                        </ul>
+                        
                         In addition to key keys listed above, four functions are available to take the context of the token into account:
                         <ul>
                             <li>  <span class="userinputdoc">child()</span> child of current token</li>
diff --git a/src/main/antlr4/com/orange/labs/conllparser/Conditions.g4 b/src/main/antlr4/com/orange/labs/conllparser/Conditions.g4
index 3d4b9e2..71b6fda 100644
--- a/src/main/antlr4/com/orange/labs/conllparser/Conditions.g4
+++ b/src/main/antlr4/com/orange/labs/conllparser/Conditions.g4
@@ -85,14 +85,14 @@ field
 
 
 
-UPOS      : 'Upos:' [A-Z]+ ;
+UPOS      : 'Upos:' [A-Z_]+ ;
 LEMMA     : 'Lemma:' ~[ \n\t)&|]+ ;
 FORM      : 'Form:' ~[ \n\t)&|]+ ;
 XPOS      : 'Xpos:' ~[ \n\t)&|]+ ;
 //DEPREL : 'Deprel:' [a-z]+( ':' ~[ \n\t)&|]+)? ;
 DEPREL    : 'Deprel:' ~[ \n\t)&|]+( ':' ~[ \n\t)&|]+)? ;
-FEAT      : 'Feat:' [A-Za-z_[\]]+ [:=] [A-Za-z0-9]+ ;
-MISC      : 'Misc:' [A-Za-z_]+ [:=] ~[ \n\t)&|]+ ;
+FEAT      : 'Feat:' [A-Za-z_[\]]+ [:=] [A-Za-z0-9]* ;
+MISC      : 'Misc:' [A-Za-z_]+ [:=] ~[ \n\t)&|]* ;
 ID        : 'Id:' [1-9][0-9]* ; // no "n.m" nor "n-m" yet
 MWT       : 'MWT:' [2-9] ; // length of a MWT in tokens
 HEADID    : 'HeadId:' [+-]?[0-9]+ ;
@@ -100,7 +100,7 @@ RELEUD    : 'EUD:' ([+-][0-9]+) [:=]  [a-z]+( ':' ~[ \n\t)&|]+)? ;
 ABSEUD    : 'EUD:' ([0-9]+|'*') [:=]  [a-z]+( ':' ~[ \n\t)&|]+)? ;
 
 ISEMPTY   : 'IsEmpty' ; // emptyword
-ISMWT     : 'IsMWT' ; // multi word tokenemptyword
+ISMWT     : 'IsMWT' ; // multi word token
 
 AND    : 'and' | '&&' ;
 OR     : 'or' | '||' ;
diff --git a/src/main/java/com/orange/labs/conllparser/CEvalVisitor.java b/src/main/java/com/orange/labs/conllparser/CEvalVisitor.java
index c48ddd4..ae2d2b6 100644
--- a/src/main/java/com/orange/labs/conllparser/CEvalVisitor.java
+++ b/src/main/java/com/orange/labs/conllparser/CEvalVisitor.java
@@ -303,19 +303,31 @@ public Boolean visitCheckFeat(ConditionsParser.CheckFeatContext ctx) {
         if (use == null) {
             return false;
         }
-        boolean rtc = use.matchesFeatureValue(fv[0], fv[1]);
+        boolean rtc ;
+        if (fv.length == 2) {
+            rtc = use.matchesFeatureValue(fv[0], fv[1]);
+        } else {
+            // feature must not be in word
+            rtc = !use.getFeatures().containsKey(fv[0]);
+        }
         return rtc;
     }
 
     @Override
     public Boolean visitCheckMisc(ConditionsParser.CheckMiscContext ctx) {
         String text = ctx.MISC().getText();
-        String[] fv = text.substring(5).split("=");
+        String[] fv = text.substring(5).split("[:=]");
         ConllWord use = getCW();
         if (use == null) {
             return false;
         }
-        boolean rtc = use.matchesMiscValue(fv[0], fv[1]);
+        boolean rtc;
+        if (fv.length == 2) {
+            rtc = use.matchesMiscValue(fv[0], fv[1]);
+        } else {
+            // feature must not be in word
+            rtc = !use.getMisc().containsKey(fv[0]);
+        }
         return rtc;
     }
 
diff --git a/src/main/java/com/orange/labs/conllparser/ConllWord.java b/src/main/java/com/orange/labs/conllparser/ConllWord.java
index a00276f..3f568f1 100644
--- a/src/main/java/com/orange/labs/conllparser/ConllWord.java
+++ b/src/main/java/com/orange/labs/conllparser/ConllWord.java
@@ -1441,7 +1441,7 @@ public boolean matchCondition(String condition, Map<String, Set<String>> wordlis
             //System.err.println("\n\nEVAL " + this);
             return CheckConditions.evaluate(condition, wordlists, this, false); // debug: show tokenisation of condition
         } catch (Exception e) {
-            //e.printStackTrace();
+            e.printStackTrace();
             throw new ConllException(e.getMessage());
         }
     }
@@ -1539,7 +1539,7 @@ public boolean anyFeatures() {
         return !features.isEmpty();
     }
 
-    // check whether featyre with value is present
+    // check whether feature with value is present
     public boolean hasFeature(String name, String val) {
         if (features.isEmpty()) {
             return false;