Fix tabular view

Change-Id: I86f35ccc811833f896b68daccd1c68d0a6b2fa9c

Author: kupietz

src/main/java/de/ids_mannheim/korap/response/Match.java

@@ -1635,6 +1635,10 @@ public ObjectNode getSnippetTokens () {
         PositionsToOffset pto = this.positionsToOffset;
         int ldid = this.localDocID;
 
+        // Work on local copies of match bounds to avoid mutating global state
+        int matchStartTok = this.startPos;
+        int matchEndTok = this.endPos; // exclusive bound
+
         int startContext = -1;
         int endContext = -1;
         int startContextChar = -1;
@@ -1707,20 +1711,20 @@ public ObjectNode getSnippetTokens () {
         int kwicMax = KrillProperties.getMaxTokenKwicSize();
         if (kwicMax > 0) {
             // Convert endContext to exclusive bound for iteration ease
-            int leftLen = (startContext < this.startPos) ? (this.startPos - startContext) : 0;
-            int matchLen = (this.endPos > this.startPos) ? (this.endPos - this.startPos) : 0;
-            int rightLen = (endContext > this.endPos) ? (endContext - this.endPos) : 0;
+            int leftLen = (startContext < matchStartTok) ? (matchStartTok - startContext) : 0;
+            int matchLen = (matchEndTok > matchStartTok) ? (matchEndTok - matchStartTok) : 0;
+            int rightLen = (endContext > matchEndTok) ? (endContext - matchEndTok) : 0;
             int total = leftLen + matchLen + rightLen;
             if (DEBUG)
                 log.info("KWIC tokens pre-cap: total={} (L/M/R={}/{}/{}) cap={} id={} uid={}",
                          total, leftLen, matchLen, rightLen, kwicMax, this.getID(), this.getUID());
 
             if (matchLen >= kwicMax) {
                 // Cut match to kwicMax, drop all context
-                this.endPos = this.startPos + kwicMax;
+                matchEndTok = matchStartTok + kwicMax;
                 this.endCutted = true;
-                startContext = this.startPos;
-                endContext = this.endPos; // exclusive bound
+                startContext = matchStartTok;
+                endContext = matchEndTok; // exclusive bound
             }
             else if (total > kwicMax) {
                 int toReduce = total - kwicMax;
@@ -1749,10 +1753,10 @@ else if (total > kwicMax) {
                     // Trim remaining from match end
                     int newMatchLen = matchLen - rest;
                     if (newMatchLen < 0) newMatchLen = 0;
-                    this.endPos = this.startPos + newMatchLen;
+                    matchEndTok = matchStartTok + newMatchLen;
                     this.endCutted = true;
-                    if (endContext < this.endPos)
-                        endContext = this.endPos;
+                    if (endContext < matchEndTok)
+                        endContext = matchEndTok;
                 }
             }
             else {
@@ -1806,7 +1810,7 @@ else if (total > kwicMax) {
         };
 
         tokens = json.putArray("match");
-        for (i = this.startPos; i < this.endPos; i++) {
+        for (i = matchStartTok; i < matchEndTok; i++) {
             offsets = pto.span(ldid,i);
             if (offsets == null) {
                 continue;
@@ -1820,7 +1824,7 @@ else if (total > kwicMax) {
         // Create right context token list
         if (endContext > this.endPos) {
             tokens = null;
-            for (i = this.endPos; i < endContext; i++) {
+            for (i = matchEndTok; i < endContext; i++) {
                 offsets = pto.span(ldid,i);
                 if (offsets == null) {
                     break;
@@ -1849,14 +1853,28 @@ else if (total > kwicMax) {
                 if (highlight.end == PB_MARKER || highlight.end == ALL_MARKER)
                     continue;
 
-                if (classes == null)
-                    classes = json.putArray("classes");
-                
-                ArrayNode cls = mapper.createArrayNode();                
-                cls.add(highlight.number);
-                cls.add(highlight.start - this.startPos);
-                cls.add(highlight.end - this.startPos);
-                classes.add(cls);
+                // Clamp class span to visible match window
+                int clsStart = highlight.start - matchStartTok;
+                int clsEnd = highlight.end - matchStartTok; // inclusive
+
+                int visibleMatchLen = Math.max(0, matchEndTok - matchStartTok);
+                int maxIdx = visibleMatchLen > 0 ? (visibleMatchLen - 1) : -1;
+
+                if (maxIdx >= 0) {
+                    if (clsStart < 0) clsStart = 0;
+                    if (clsEnd > maxIdx) clsEnd = maxIdx;
+                }
+
+                if (clsStart <= clsEnd && maxIdx >= 0) {
+                    if (classes == null)
+                        classes = json.putArray("classes");
+
+                    ArrayNode cls = mapper.createArrayNode();
+                    cls.add(highlight.number);
+                    cls.add(clsStart);
+                    cls.add(clsEnd);
+                    classes.add(cls);
+                }
             };
         };
 

src/main/resources/krill.properties.info

@@ -22,6 +22,21 @@ krill.index.textSize.max = 20000000
 # 2*krill.context.max.token + krill.match.max.token and log a deprecation warning.
 # krill.kwic.max.token = 100
 
+# For tokenized KWIC (tabular) views, optionally include a class entry
+# that covers the whole match span. Frontends can use this to highlight
+# the complete match in the token window in addition to annotation classes.
+# Default: false
+# krill.kwic.tokens.match.class = false
+
+# If true, align the HTML KWIC window strictly to the token window produced
+# by the tokenized KWIC. This can help ensure both representations show the
+# exact same left/match/right boundaries. Default: false
+# krill.kwic.enforce.html = false
+
+# If true, clamp HTML snippet character ranges to safe boundaries when
+# offsets are incomplete. Default: false
+# krill.snippet.safeCharBounds = false
+
 # Deprecated: These are ignored when 'krill.kwic.max.token' is set and will be removed.
 # They were previously used to cap match length and per-side context lengths, but
 # licensing limits apply to the total snippet size, not to the match alone.