chore(execplan): MCOL: 6144 prerry print filters and havings as a tree (#3712)

* MCOL-6144: print Filters and Having indented for better CSEP visualization

dbcon/execplan/aggregatecolumn.cpp

@@ -152,13 +152,49 @@ const string AggregateColumn::toString() const
   return output.str();
 }
 
+const string AggregateColumn::toString(bool compact) const
+{
+  if (!compact)
+  {
+    return toString();
+  }
+
+  ostringstream output;
+
+  // Compact format for tree display - let tree printer handle indentation
+  output << "AggregateColumn:";
+  output << endl << "Function: " << (int)fAggOp << ", Distinct: " << fDistinct;
+
+  if (fAlias.length() > 0)
+    output << endl << "Alias: " << fAlias;
+
+  if (fAggParms.size() > 0)
+  {
+    for (uint32_t i = 0; i < fAggParms.size(); ++i)
+    {
+      SimpleColumn* sc = dynamic_cast<SimpleColumn*>(fAggParms[i].get());
+      if (sc)
+      {
+        output << endl << sc->toString(true);
+      }
+      else
+      {
+        output << endl << "Param: " << fAggParms[i]->data();
+      }
+    }
+  }
+
+  return output.str();
+}
+
 string AggregateColumn::toCppCode(IncludeSet& includes) const
 {
   includes.insert("aggregatecolumn.h");
   stringstream ss;
   auto fContent = fData.substr(fFunctionName.size() + 1, fData.size() - fFunctionName.size() - 2);
 
-  ss << "AggregateColumn(" << std::quoted(fFunctionName) << ", " << std::quoted(fContent) << ", " << sessionID() << ")";
+  ss << "AggregateColumn(" << std::quoted(fFunctionName) << ", " << std::quoted(fContent) << ", "
+     << sessionID() << ")";
 
   return ss.str();
 }
@@ -588,15 +624,15 @@ void AggregateColumn::evaluate(Row& row, bool& isNull)
 
     case CalpontSystemCatalog::VARBINARY:
     case CalpontSystemCatalog::BLOB:
-      {
-        auto const str = row.getConstString(fInputIndex);
-        fResult.strVal.dropString();
-        if (!str.isNull())
-          fResult.strVal.assign((const uint8_t*)str.str(), str.length());
+    {
+      auto const str = row.getConstString(fInputIndex);
+      fResult.strVal.dropString();
+      if (!str.isNull())
+        fResult.strVal.assign((const uint8_t*)str.str(), str.length());
 
-        isNull = isNull || fResult.strVal.isNull();
-      }
-      break;
+      isNull = isNull || fResult.strVal.isNull();
+    }
+    break;
 
     default:  // treat as int64
       if (row.equals<8>(BIGINTNULL, fInputIndex))

dbcon/execplan/aggregatecolumn.h

@@ -220,6 +220,7 @@ class AggregateColumn : public ReturnedColumn
    * Overloaded stream operator
    */
   const std::string toString() const override;
+  const std::string toString(bool compact) const;
   std::string toCppCode(IncludeSet& includes) const override;
 
   /**

dbcon/execplan/calpontselectexecutionplan.cpp

@@ -22,6 +22,7 @@
  ***********************************************************************/
 #include <iostream>
 #include <algorithm>
+#include <sstream>
 using namespace std;
 
 #include <boost/uuid/uuid_io.hpp>
@@ -35,6 +36,9 @@ using namespace messageqcpp;
 #include "returnedcolumn.h"
 #include "simplecolumn.h"
 #include "querystats.h"
+#include "simplefilter.h"
+#include "operator.h"
+#include <boost/core/demangle.hpp>
 
 #include "querytele.h"
 #include "utils/pron/pron.h"
@@ -197,12 +201,140 @@ std::string endlWithIndent(const size_t ident)
 {
   ostringstream output;
   output << endl;
-  output << std::string(ident, ' ');
+
+  for (size_t i = 0; i < ident; i++)
+    output << " ";
+
   return output.str();
 }
 
+// Iterative tree printer that preserves vertical branches for multi-line nodes
+static void printIndentedFilterTreeImpl(const ParseTree* root, ostringstream& output, size_t indent,
+                                        const std::vector<bool>& rootAncestors, bool rootIsLast)
+{
+  // Stack holds frames: (node, ancestorHasNext, isLastAtLevel)
+  struct Frame
+  {
+    const ParseTree* node;
+    std::vector<bool> ancestors;
+    bool isLast;
+  };
+
+  std::vector<Frame> stack;
+  stack.push_back(Frame{root, rootAncestors, rootIsLast});
+
+  while (!stack.empty())
+  {
+    Frame frame = std::move(stack.back());
+    stack.pop_back();
+
+    const ParseTree* node = frame.node;
+    const std::vector<bool>& ancestorHasNext = frame.ancestors;
+    const bool isLastAtLevel = frame.isLast;
+
+    if (!node)
+    {
+      // Build prefix for a null placeholder
+      std::string base;
+      for (bool hasNext : ancestorHasNext)
+        base += hasNext ? "│   " : "    ";
+      std::string nodePrefix = base + (isLastAtLevel ? "└── " : "├── ");
+      output << endlWithIndent(indent) << nodePrefix << "(null)";
+      continue;
+    }
+
+    // Gather children in left-to-right order
+    std::vector<const ParseTree*> children;
+    if (node->left())
+      children.push_back(node->left());
+    if (node->right())
+      children.push_back(node->right());
+
+    // Helper to build prefixes
+    auto makePrefixes = [&](bool isLast)
+    {
+      std::string base;
+      for (bool hasNext : ancestorHasNext)
+        base += hasNext ? "│   " : "    ";
+      std::string first = base + (isLast ? "└── " : "├── ");
+      std::string cont = base + (isLast ? "    " : "│   ");
+      return std::pair<std::string, std::string>(first, cont);
+    };
+
+    // Build node content string
+    TreeNode* data = node->data();
+    std::string nodeContent;
+    if (data)
+    {
+      if (auto sf = dynamic_cast<SimpleFilter*>(data))
+      {
+        nodeContent = sf->toString(true);
+      }
+      else if (auto op = dynamic_cast<Operator*>(data))
+      {
+        nodeContent = op->toString();
+      }
+      else
+      {
+        nodeContent = boost::core::demangle(typeid(*data).name()) + ": " + data->toString();
+      }
+    }
+    else
+    {
+      nodeContent = "(null data)";
+    }
+
+    // Print current node content
+    if (ancestorHasNext.empty())
+    {
+      // Root: print without branch glyphs
+      std::istringstream contentStream(nodeContent);
+      std::string line;
+      while (std::getline(contentStream, line))
+        output << endlWithIndent(indent) << line;
+    }
+    else
+    {
+      auto prefixes = makePrefixes(isLastAtLevel);
+      std::istringstream contentStream(nodeContent);
+      std::string line;
+      bool firstLine = true;
+      while (std::getline(contentStream, line))
+      {
+        if (firstLine)
+        {
+          output << endlWithIndent(indent) << prefixes.first << line;
+          firstLine = false;
+        }
+        else
+        {
+          output << endlWithIndent(indent) << prefixes.second << line;
+        }
+      }
+    }
+
+    // Push children onto stack in reverse order to process left child first
+    for (size_t i = children.size(); i-- > 0;)
+    {
+      bool childIsLast = (i == children.size() - 1);
+      std::vector<bool> nextAncestors = ancestorHasNext;
+      // For children, push whether THIS node has a next sibling; this keeps the vertical bar under this node
+      // if we are not the last at our level.
+      nextAncestors.push_back(!isLastAtLevel);
+      stack.push_back(Frame{children[i], std::move(nextAncestors), childIsLast});
+    }
+  }
+}
+
+void printIndentedFilterTree(const ParseTree* tree, ostringstream& output, size_t indent)
+{
+  // Start with empty ancestor vector and indicate root is last at its (virtual) level
+  std::vector<bool> ancestors;  // empty => root
+  printIndentedFilterTreeImpl(tree, output, indent, ancestors, true);
+}
+
 void CalpontSelectExecutionPlan::printSubCSEP(const size_t& ident, ostringstream& output,
-                                           CalpontSelectExecutionPlan*& plan) const
+                                              CalpontSelectExecutionPlan*& plan) const
 {
   if (plan)
   {
@@ -243,7 +375,7 @@ string CalpontSelectExecutionPlan::toString(const size_t ident) const
 
   for (unsigned int i = 0; i < retCols.size(); i++)
   {
-    output << endlWithIndent(ident+2) << *retCols[i]; // WIP replace with constant
+    output << endlWithIndent(ident + 2) << *retCols[i];  // WIP replace with constant
 
     if (retCols[i]->colSource() & SELECT_SUB)
     {
@@ -257,7 +389,7 @@ string CalpontSelectExecutionPlan::toString(const size_t ident) const
 
   // From Clause
   CalpontSelectExecutionPlan::TableList tables = tableList();
-  output << endlWithIndent(ident) <<">>From Tables";
+  output << endlWithIndent(ident) << ">>From Tables";
   seq = 0;
 
   for (unsigned int i = 0; i < tables.size(); i++)
@@ -265,7 +397,7 @@ string CalpontSelectExecutionPlan::toString(const size_t ident) const
     // derived table
     if (tables[i].schema.length() == 0 && tables[i].table.length() == 0)
     {
-      output << endlWithIndent(ident+2) << "derived table - " << tables[i].alias;
+      output << endlWithIndent(ident + 2) << "derived table - " << tables[i].alias;
       CalpontSelectExecutionPlan* plan =
           dynamic_cast<CalpontSelectExecutionPlan*>(fDerivedTableList[seq++].get());
 
@@ -273,7 +405,7 @@ string CalpontSelectExecutionPlan::toString(const size_t ident) const
     }
     else
     {
-      output << endlWithIndent(ident+2) << tables[i];
+      output << endlWithIndent(ident + 2) << tables[i];
     }
   }
 
@@ -282,8 +414,8 @@ string CalpontSelectExecutionPlan::toString(const size_t ident) const
 
   if (filters() != nullptr)
   {
-    output << endlWithIndent(ident + 2);
-    filters()->walk(ParseTree::print, output);
+    output << endlWithIndent(ident + 2) << "Filter Tree:";
+    printIndentedFilterTree(filters(), output, ident + 4);
   }
   else
   {
@@ -308,8 +440,9 @@ string CalpontSelectExecutionPlan::toString(const size_t ident) const
   // Having
   if (having() != nullptr)
   {
-    output << endlWithIndent(ident) << ">>Having" << endlWithIndent(ident + 2);
-    having()->walk(ParseTree::print, output);
+    output << endlWithIndent(ident) << ">>Having";
+    output << endlWithIndent(ident + 2) << "Having Tree:";
+    printIndentedFilterTree(having(), output, ident + 4);
   }
 
   // Order by columns
@@ -863,7 +996,8 @@ void CalpontSelectExecutionPlan::pron(std::string&& pron)
   fPron = pron;
 }
 
-// This routine doesn't copy derived table list, union vector, select subqueries, subquery list, and subselects.
+// This routine doesn't copy derived table list, union vector, select subqueries, subquery list, and
+// subselects.
 execplan::SCSEP CalpontSelectExecutionPlan::cloneWORecursiveSelects()
 {
   execplan::SCSEP newPlan(new CalpontSelectExecutionPlan(fLocation));

dbcon/execplan/simplecolumn.cpp

@@ -292,6 +292,26 @@ const string SimpleColumn::toString() const
   return output.str();
 }
 
+const string SimpleColumn::toString(bool compact) const
+{
+  if (!compact)
+  {
+    // Use the original detailed format
+    return toString();
+  }
+
+  ostringstream output;
+
+  // Compact format for tree display - let tree printer handle indentation
+  output << "Column: " << data();
+  datatypes::Charset cs(fResultType.charsetNumber);
+  output << endl
+         << "Info: " << schemaName() << "." << tableName() << "." << columnName()
+         << " (Type: " << colDataTypeToString(fResultType.colDataType) << ", OID: " << oid() << ")";
+
+  return output.str();
+}
+
 string SimpleColumn::toCppCode(IncludeSet& includes) const
 {
   includes.insert("simplecolumn.h");

dbcon/execplan/simplecolumn.h

@@ -205,6 +205,7 @@ class SimpleColumn : public ReturnedColumn
   void unserialize(messageqcpp::ByteStream&) override;
 
   const std::string toString() const override;
+  const std::string toString(bool compact) const;
   std::string toCppCode(IncludeSet& includes) const override;
   /** @brief Do a deep, strict (as opposed to semantic) equivalence test
    *

dbcon/execplan/simplefilter.cpp

@@ -29,16 +29,22 @@ using namespace std;
 
 #include "returnedcolumn.h"
 #include "constantcolumn.h"
-#include "simplecolumn.h"
-#include "operator.h"
-#include "constantfilter.h"
+#include "simplefilter.h"
 #include "bytestream.h"
 #include "objectreader.h"
-#include "functioncolumn.h"
-#include "arithmeticcolumn.h"
-#include "simplefilter.h"
+#include "simplecolumn.h"
 #include "aggregatecolumn.h"
+#include "arithmeticcolumn.h"
+#include "functioncolumn.h"
+#include "constantcolumn.h"
+#include "selectfilter.h"
+#include "operator.h"
 #include "windowfunctioncolumn.h"
+#include "rowgroup.h"
+#include "joblisttypes.h"
+using namespace messageqcpp;
+using namespace std;
+using namespace execplan;
 
 namespace execplan
 {
@@ -250,6 +256,53 @@ const string SimpleFilter::toString() const
   return output.str();
 }
 
+const string SimpleFilter::toString(bool compact) const
+{
+  if (!compact)
+  {
+    // Use the original detailed format
+    return toString();
+  }
+
+  ostringstream output;
+
+  // Compact format for tree display
+  output << "SimpleFilter(indexflag=" << fIndexFlag;
+  output << " joinFlag= " << fJoinFlag;
+  output << " card= " << fCardinality << ")";
+
+  // Handle different types of left operands with compact formatting
+  SimpleColumn* sc = dynamic_cast<SimpleColumn*>(fLhs);
+  AggregateColumn* ac = dynamic_cast<AggregateColumn*>(fLhs);
+
+  if (ac)
+  {
+    string acCompact = ac->toString(true);
+    if (!acCompact.empty())
+    {
+      output << endl << acCompact;
+    }
+  }
+  else if (sc)
+  {
+    string scCompact = sc->toString(true);
+    if (!scCompact.empty())
+    {
+      output << endl << scCompact;
+    }
+  }
+  else
+  {
+    // For other operand types, show them compactly
+    output << endl << "LHS: " << fLhs->data();
+  }
+
+  output << endl << "Op: " << fOp->data();
+  output << endl << "RHS: " << fRhs->data();
+
+  return output.str();
+}
+
 string SimpleFilter::toCppCode(IncludeSet& includes) const
 {
   includes.insert("simplefilter.h");

dbcon/execplan/simplefilter.h

@@ -132,6 +132,7 @@ class SimpleFilter : public Filter
   void rhs(ReturnedColumn* rhs);
 
   const std::string toString() const override;
+  const std::string toString(bool compact) const;
 
   /**
    * The serialization interface