Implemented fallback for backward compatibility in offset validation (p…

…roycon/folia#88)
LanguageMachines · Mar 19, 2021 · 56afe68 · 56afe68
1 parent 35a6cf5
commit 56afe68
Show file tree

Hide file tree

Showing 3 changed files with 45 additions and 11 deletions.
diff --git a/include/libfolia/folia_impl.h b/include/libfolia/folia_impl.h
@@ -1350,7 +1350,7 @@ namespace folia {
 	return res;
       }
       FoliaElement *postappend();
-      FoliaElement *get_reference() const;
+      FoliaElement *get_reference(bool trim_spaces=true) const;
       std::string ref() const { return _ref; };
     private:
       void init();
@@ -1376,7 +1376,7 @@ namespace folia {
 			      TEXT_FLAGS = TEXT_FLAGS::NONE ) const;
     int offset() const { return _offset; };
     FoliaElement *postappend();
-    FoliaElement *get_reference() const;
+    FoliaElement *get_reference(bool trim_spaces=true) const;
     std::string ref() const { return _ref; };
   private:
     void init();

diff --git a/src/folia_document.cxx b/src/folia_document.cxx
@@ -1714,13 +1714,28 @@ namespace folia {
 	  string msg = "Text for " + txt->parent()->xmltag() + "(ID="
 	    + txt->parent()->id() + ", textclass='" + txt->cls()
 	    + "'), has incorrect offset " + TiCC::toString(offset);
+
+
 	  string ref = txt->ref();
 	  if ( !ref.empty() ){
 	    msg += " or invalid reference:" + ref;
 	  }
 	  msg += "\n\toriginal msg=";
 	  msg += e.what();
-	  throw UnresolvableTextContent( msg );
+
+          bool warn = false;
+          try {
+             txt->get_reference(false); //trim_spaces = false
+             msg += "\nHowever, according to the older rules (<v2.4.1) the offsets are accepted. So we are treating this as a warning rather than an error. We do recommend fixing this if this is a document you intend to publish.";
+             warn = true;
+          } catch (UnresolvableTextContent& e2) {
+             msg += "\n(also checked against older rules prior to FoLiA v2.4.1)";
+          }
+
+          if (warn)
+              cerr << "WARNING: " << msg << endl;
+          else
+              throw UnresolvableTextContent( msg );
 	}
       }
     }
@@ -1739,13 +1754,28 @@ namespace folia {
 	  string msg = "Phoneme for " + phon->parent()->xmltag() + ", ID="
 	    + phon->parent()->id() + ", textclass='" + phon->cls()
 	    + "', has incorrect offset " + TiCC::toString(offset);
+
+
 	  string ref = phon->ref();
 	  if ( !ref.empty() ){
 	    msg += " or invalid reference:" + ref;
 	  }
 	  msg += "\n\toriginal msg=";
 	  msg += e.what();
-	  throw UnresolvableTextContent( msg );
+
+          bool warn = false;
+          try {
+             phon->get_reference(false); //trim_spaces = false
+             msg += "\nHowever, according to the older rules (<v2.4.1) the offsets are accepted. So we are treating this as a warning rather than an error. We do recommend fixing this if this is a document you intend to publish.";
+             warn = true;
+          } catch (UnresolvableTextContent& e2) {
+             msg += "\n(also checked against older rules prior to FoLiA v2.4.1)";
+          }
+
+          if (warn)
+              cerr << "WARNING: " << msg << endl;
+          else
+              throw UnresolvableTextContent( msg );
 	}
       }
     }

diff --git a/src/folia_impl.cxx b/src/folia_impl.cxx
@@ -3895,8 +3895,8 @@ namespace folia {
     return 0;
   }
 
-  FoliaElement *TextContent::get_reference() const {
-    /// get the FoliaElement _ref is refering to
+  FoliaElement *TextContent::get_reference(bool trim_spaces) const {
+    /// get the FoliaElement _ref is refering to and does offset validation
     /*!
      * \return the refered element OR the default parent when _ref is 0
      */
@@ -3921,8 +3921,10 @@ namespace folia {
       throw UnresolvableTextContent( "Reference (ID " + _ref + ") has no such text (class=" + cls() + ")" );
     }
     else if ( doc()->checktext() || doc()->fixtext() ){
-      UnicodeString mt = this->text( this->cls(), TEXT_FLAGS::STRICT );
-      UnicodeString pt = ref->text( this->cls(), TEXT_FLAGS::STRICT );
+      TEXT_FLAGS flags = TEXT_FLAGS::STRICT;
+      if ( !trim_spaces ) flags |= TEXT_FLAGS::NO_TRIM_SPACES;
+      UnicodeString mt = this->text( this->cls(), flags );
+      UnicodeString pt = ref->text( this->cls(), flags );
       UnicodeString sub( pt, this->offset(), mt.length() );
       if ( mt != sub ){
 	if ( doc()->fixtext() ){
@@ -3992,7 +3994,7 @@ namespace folia {
     return 0;
   }
 
-  FoliaElement *PhonContent::get_reference() const {
+  FoliaElement *PhonContent::get_reference(bool trim_spaces) const {
     /// get the FoliaElement _ref is refering to
     /*!
      * \return the refered element OR the default parent when _ref is 0
@@ -4018,8 +4020,10 @@ namespace folia {
       throw UnresolvableTextContent( "Reference (ID " + _ref + ") has no such phonetic content (class=" + cls() + ")" );
     }
     else if ( doc()->checktext() || doc()->fixtext() ){
-      UnicodeString mt = this->phon( this->cls() );
-      UnicodeString pt = ref->phon( this->cls() );
+      TEXT_FLAGS flags = TEXT_FLAGS::STRICT;
+      if ( !trim_spaces ) flags |= TEXT_FLAGS::NO_TRIM_SPACES;
+      UnicodeString mt = this->phon( this->cls(), flags );
+      UnicodeString pt = ref->phon( this->cls(), flags );
       UnicodeString sub( pt, this->offset(), mt.length() );
       if ( mt != sub ){
 	if ( doc()->fixtext() ){