re-enabled select distinct aggregations

pjshumphreys · Jun 3, 2019 · ecd255b · ecd255b
1 parent 2e9d2c9
commit ecd255b
Show file tree

Hide file tree

Showing 8 changed files with 43 additions and 48 deletions.
diff --git a/env/dos/dos.c b/env/dos/dos.c
@@ -67,7 +67,7 @@ int fputs_dos(const char *str, FILE *stream) {
     output = d_charsetEncode((char *)str, consoleEncoding, &len);
 
     /* eat last trailing newline (if we print something else we'll display it then) */
-    if(output[len-1] == '\n') {
+    if(output && output[len-1] == '\n') {
       newline = TRUE;
       lastWasErr = stream == stderr;
       output[len-1] = '\0';

diff --git a/env/win32/win32.c b/env/win32/win32.c
@@ -64,7 +64,7 @@ int fputs_w32(const char *str, FILE *stream) {
 
       output = d_charsetEncode((char *)str, ENC_UTF16LE, &len);
 
-      if(output[len-2] == '\n') {
+      if(output && output[len-2] == '\n') {
         newline = TRUE;
         len -= 2;
       }
@@ -118,7 +118,7 @@ int fputs_w32(const char *str, FILE *stream) {
     output = d_charsetEncode((char *)str, consoleEncoding, &len);
 
     /* eat last trailing newline. If we print something else we'll display it at that time */
-    if(output[len-1] == '\n') {
+    if(output && output[len-1] == '\n') {
       newline = TRUE;
       lastWasErr = stream == stderr;
       output[len-1] = '\0';

diff --git a/querycsv.h b/querycsv.h
@@ -80,6 +80,7 @@
 #define TRE_BLACK 1
 #define TRE_RED 2
 #define TRE_CONVERTED 3
+#define TRE_SKIP 4
 #define isRed(x) (x != NULL && x->type == TRE_RED)
 
 /* short codes for the character encodings we want to support */

diff --git a/split/groupResultsInner.c b/split/groupResultsInner.c
@@ -2,50 +2,39 @@
 void groupResultsInner(
     struct qryData *query,
     struct resultColumnValue *columns,
-    int i
+    int i,
+    struct resultTree *item
 ) {
-  struct resultColumnValue *previousMatch;
+  struct resultTree *tempItem;
 
   MAC_YIELD
 
-  if(i == 0) {
-    updateRunningCounts(query, columns, TRUE);
-
-    if(query->recordCount > 1) {
-      query->match = columns;
-    }
-    else {
-      addGroupedResult(query, columns);
-    }
-
-    return;
+  if(item->type == TRE_SKIP) {
+    cleanup_matchValues(query, &columns);
   }
+  else {
+    query->groupCount = 0;
 
-  previousMatch = query->match;
+    updateRunningCounts(query, item);
 
-  /* if the current record to look at is identical to the previous one */
-  /* if no group by clause then every record is part of one group */
-  if(
-      query->groupByClause == NULL ||
+    /* look at the next item */
+    tempItem = item->link[1];
+
+    /* while the next nth result is part of the same group
+    as this result, add to the number of items to look ahead */
+    while(
+      tempItem &&
       recordCompare(
-        (void *)previousMatch,
         (void *)columns,
+        (void *)tempItem->columns,
         (void *)query
-      ) != 0
+      ) == 0
     ) {
-    addGroupedResult(query, previousMatch);
-
-    updateRunningCounts(query, columns, TRUE);
-
-    previousMatch = query->match = columns;
-  }
-  else {
-    updateRunningCounts(query, columns, FALSE);
-
-    cleanup_matchValues(query, &columns);
-  }
+      updateRunningCounts(query, tempItem);
+      tempItem->type = TRE_SKIP;
+      tempItem = tempItem->link[1];
+    }
 
-  if(i == query->recordCount - 1) {
-    addGroupedResult(query, previousMatch);
+    addGroupedResult(query, columns);
   }
 }
diff --git a/split/outputResult.c b/split/outputResult.c
@@ -2,7 +2,8 @@
 void outputResult(
     struct qryData *query,
     struct resultColumnValue *columns,
-    int currentIndex
+    int currentIndex,
+    struct resultTree *item
 ) {
   struct resultColumn *currentResultColumn;
   struct resultColumnValue *field;
@@ -82,5 +83,4 @@ void outputResult(
   if(currentIndex > -1) {
     cleanup_matchValues(query, &columns);
   }
-
 }
diff --git a/split/runQuery.c b/split/runQuery.c
@@ -33,7 +33,7 @@ int runQuery(char *queryFileName) {
     while(getMatchingRecord(&query, match)) {
       /* print record to stdout */
       query.recordCount++;
-      outputResult(&query, match, -1);
+      outputResult(&query, match, -1, NULL);
     }
 
     cleanup_matchValues(&query, &match);

diff --git a/split/tree_walkAndCleanup.c b/split/tree_walkAndCleanup.c
@@ -4,7 +4,7 @@
 void tree_walkAndCleanup(
     struct qryData *query,
     struct resultTree **root,
-    void (*callback)(struct qryData *, struct resultColumnValue *, int)
+    void (*callback)(struct qryData *, struct resultColumnValue *, int, struct resultTree *)
 ) {
   struct resultTree *currentResult;
   struct resultTree *tempResult;
@@ -85,7 +85,8 @@ void tree_walkAndCleanup(
     callback(
       query,
       currentResult->columns,
-      i++
+      i++,
+      currentResult
     );
 
     tempResult = currentResult;

diff --git a/split/updateRunningCounts.c b/split/updateRunningCounts.c
@@ -1,8 +1,10 @@
 void updateRunningCounts(
     struct qryData *query,
-    struct resultColumnValue *match,
-    int startNewGroup
+    struct resultTree *item
 ) {
+  struct resultTree *tempItem;
+  struct resultColumnValue *match = item->columns;
+
   struct columnRefHashEntry *currentHashEntry;
   struct columnReference *currentReference;
   struct resultColumn *currentResultColumn;
@@ -34,7 +36,7 @@ void updateRunningCounts(
           ) {
 
           /* if query->groupCount = 1 then we're starting a new group */
-          if(startNewGroup) {
+          if(query->groupCount == 1) {
             freeAndZero(currentResultColumn->groupText);
             currentResultColumn->groupNum = ctof(0);
             currentResultColumn->groupCount = 0;
@@ -46,11 +48,12 @@ void updateRunningCounts(
             stringGet((unsigned char **)(&tempString), field, query->params);
 
             /* distinct groupings. only add to the count if the column value hasn't aready been seen */
-            /* TODO: fix this code. we'll need to keep all results until the grouping is finished because of this */
-            /*if(currentResultColumn->groupType > GRP_STAR) {
+            if(currentResultColumn->groupType > GRP_STAR) { /* distinct variants */
               if(query->groupCount > 1) {
+                tempItem = item->link[0];
+
                 for(j = 1; j < query->groupCount; j++) {
-                  stringGet((unsigned char **)(&tempString2), &(match[(currentResultColumn->resultColumnIndex) - (query->columnCount)]), query->params);
+                  stringGet((unsigned char **)(&tempString2), &(tempItem->columns[currentResultColumn->resultColumnIndex]), query->params);
 
                   if(strCompare(
                     (unsigned char **)(&tempString),
@@ -64,12 +67,13 @@ void updateRunningCounts(
                   }
 
                   freeAndZero(tempString2);
+                  tempItem = tempItem->link[0];
                 }
               }
               else {
                 j = query->groupCount;
               }
-            }*/
+            }
 
             switch(currentResultColumn->groupType) {
               case GRP_DIS_COUNT: {