todo

SCHEMA_TABLE_HANDLING.md

@@ -0,0 +1,138 @@
+# Schema and Table Name Handling
+
+This document explains how the handlers properly separate and handle schema and table names.
+
+## Implementation
+
+Both `resolvespec` and `restheadspec` handlers now properly handle schema and table name separation through the following functions:
+
+- `parseTableName(fullTableName)` - Splits "schema.table" into separate components
+- `getSchemaAndTable(defaultSchema, entity, model)` - Returns schema and table separately
+- `getTableName(schema, entity, model)` - Returns the full "schema.table" format
+
+## Priority Order
+
+When determining the schema and table name, the following priority is used:
+
+1. **If `TableName()` contains a schema** (e.g., "myschema.mytable"), that schema takes precedence
+2. **If model implements `SchemaProvider`**, use that schema
+3. **Otherwise**, use the `defaultSchema` parameter from the URL/request
+
+## Scenarios
+
+### Scenario 1: Simple table name, default schema
+```go
+type User struct {
+    ID   string
+    Name string
+}
+
+func (User) TableName() string {
+    return "users"
+}
+```
+- Request URL: `/api/public/users`
+- Result: `schema="public"`, `table="users"`, `fullName="public.users"`
+
+### Scenario 2: Table name includes schema
+```go
+type User struct {
+    ID   string
+    Name string
+}
+
+func (User) TableName() string {
+    return "auth.users"  // Schema included!
+}
+```
+- Request URL: `/api/public/users` (public is ignored)
+- Result: `schema="auth"`, `table="users"`, `fullName="auth.users"`
+- **Note**: The schema from `TableName()` takes precedence over the URL schema
+
+### Scenario 3: Using SchemaProvider
+```go
+type User struct {
+    ID   string
+    Name string
+}
+
+func (User) TableName() string {
+    return "users"
+}
+
+func (User) SchemaName() string {
+    return "auth"
+}
+```
+- Request URL: `/api/public/users` (public is ignored)
+- Result: `schema="auth"`, `table="users"`, `fullName="auth.users"`
+
+### Scenario 4: Table name includes schema AND SchemaProvider
+```go
+type User struct {
+    ID   string
+    Name string
+}
+
+func (User) TableName() string {
+    return "core.users"  // This wins!
+}
+
+func (User) SchemaName() string {
+    return "auth"  // This is ignored
+}
+```
+- Request URL: `/api/public/users`
+- Result: `schema="core"`, `table="users"`, `fullName="core.users"`
+- **Note**: Schema from `TableName()` takes highest precedence
+
+### Scenario 5: No providers at all
+```go
+type User struct {
+    ID   string
+    Name string
+}
+// No TableName() or SchemaName()
+```
+- Request URL: `/api/public/users`
+- Result: `schema="public"`, `table="users"`, `fullName="public.users"`
+- Uses URL schema and entity name
+
+## Key Features
+
+1. **Automatic detection**: The code automatically detects if `TableName()` includes a schema by checking for "."
+2. **Backward compatible**: Existing code continues to work
+3. **Flexible**: Supports multiple ways to specify schema and table
+4. **Debug logging**: Logs when schema is detected in `TableName()` for debugging
+
+## Code Locations
+
+### Handlers
+- `/pkg/resolvespec/handler.go:472-531`
+- `/pkg/restheadspec/handler.go:534-593`
+
+### Database Adapters
+- `/pkg/common/adapters/database/utils.go` - Shared `parseTableName()` function
+- `/pkg/common/adapters/database/bun.go` - Bun adapter with separated schema/table
+- `/pkg/common/adapters/database/gorm.go` - GORM adapter with separated schema/table
+
+## Adapter Implementation
+
+Both Bun and GORM adapters now properly separate schema and table name:
+
+```go
+// BunSelectQuery/GormSelectQuery now have separated fields:
+type BunSelectQuery struct {
+    query      *bun.SelectQuery
+    schema     string // Separated schema name
+    tableName  string // Just the table name, without schema
+    tableAlias string
+}
+```
+
+When `Model()` or `Table()` is called:
+1. The full table name (which may include schema) is parsed
+2. Schema and table name are stored separately
+3. When building joins, the already-separated table name is used directly
+
+This ensures consistent handling of schema-qualified table names throughout the codebase.

make_release.sh

@@ -15,12 +15,12 @@ if [[ $make_release =~ ^[Yy]$ ]]; then
     fi
 
     # Create an annotated tag
-    git tag -a "$version" -m "Released Core $version"
+    git tag -a "$version" -m "Released $version"
 
     # Push the tag to the remote repository
     git push origin "$version"
 
-    echo "Tag $version created for Core and pushed to the remote repository."
+    echo "Tag $version created and pushed to the remote repository."
 else
     echo "No release version created."
 fi

pkg/common/adapters/database/bun.go

@@ -4,6 +4,7 @@ import (
 	"context"
 	"database/sql"
 	"fmt"
+	"strings"
 
 	"github.com/Warky-Devs/ResolveSpec/pkg/common"
 	"github.com/uptrace/bun"
@@ -21,7 +22,10 @@ func NewBunAdapter(db *bun.DB) *BunAdapter {
 }
 
 func (b *BunAdapter) NewSelect() common.SelectQuery {
-	return &BunSelectQuery{query: b.db.NewSelect()}
+	return &BunSelectQuery{
+		query: b.db.NewSelect(),
+		db:    b.db,
+	}
 }
 
 func (b *BunAdapter) NewInsert() common.InsertQuery {
@@ -76,16 +80,32 @@ func (b *BunAdapter) RunInTransaction(ctx context.Context, fn func(common.Databa
 
 // BunSelectQuery implements SelectQuery for Bun
 type BunSelectQuery struct {
-	query *bun.SelectQuery
+	query      *bun.SelectQuery
+	db         bun.IDB // Store DB connection for count queries
+	hasModel   bool    // Track if Model() was called
+	schema     string  // Separated schema name
+	tableName  string  // Just the table name, without schema
+	tableAlias string
 }
 
 func (b *BunSelectQuery) Model(model interface{}) common.SelectQuery {
 	b.query = b.query.Model(model)
+	b.hasModel = true // Mark that we have a model
+
+	// Try to get table name from model if it implements TableNameProvider
+	if provider, ok := model.(common.TableNameProvider); ok {
+		fullTableName := provider.TableName()
+		// Check if the table name contains schema (e.g., "schema.table")
+		b.schema, b.tableName = parseTableName(fullTableName)
+	}
+
 	return b
 }
 
 func (b *BunSelectQuery) Table(table string) common.SelectQuery {
 	b.query = b.query.Table(table)
+	// Check if the table name contains schema (e.g., "schema.table")
+	b.schema, b.tableName = parseTableName(table)
 	return b
 }
 
@@ -105,12 +125,74 @@ func (b *BunSelectQuery) WhereOr(query string, args ...interface{}) common.Selec
 }
 
 func (b *BunSelectQuery) Join(query string, args ...interface{}) common.SelectQuery {
-	b.query = b.query.Join(query, args...)
+	// Extract optional prefix from args
+	// If the last arg is a string that looks like a table prefix, use it
+	var prefix string
+	sqlArgs := args
+
+	if len(args) > 0 {
+		if lastArg, ok := args[len(args)-1].(string); ok && len(lastArg) < 50 && !strings.Contains(lastArg, " ") {
+			// Likely a prefix, not a SQL parameter
+			prefix = lastArg
+			sqlArgs = args[:len(args)-1]
+		}
+	}
+
+	// If no prefix provided, use the table name as prefix (already separated from schema)
+	if prefix == "" && b.tableName != "" {
+		prefix = b.tableName
+	}
+
+	// If prefix is provided, add it as an alias in the join
+	// Bun expects: "JOIN table AS alias ON condition"
+	joinClause := query
+	if prefix != "" && !strings.Contains(strings.ToUpper(query), " AS ") {
+		// If query doesn't already have AS, check if it's a simple table name
+		parts := strings.Fields(query)
+		if len(parts) > 0 && !strings.HasPrefix(strings.ToUpper(parts[0]), "JOIN") {
+			// Simple table name, add prefix: "table AS prefix"
+			joinClause = fmt.Sprintf("%s AS %s", parts[0], prefix)
+			if len(parts) > 1 {
+				// Has ON clause: "table ON ..." becomes "table AS prefix ON ..."
+				joinClause += " " + strings.Join(parts[1:], " ")
+			}
+		}
+	}
+
+	b.query = b.query.Join(joinClause, sqlArgs...)
 	return b
 }
 
 func (b *BunSelectQuery) LeftJoin(query string, args ...interface{}) common.SelectQuery {
-	b.query = b.query.Join("LEFT JOIN " + query, args...)
+	// Extract optional prefix from args
+	var prefix string
+	sqlArgs := args
+
+	if len(args) > 0 {
+		if lastArg, ok := args[len(args)-1].(string); ok && len(lastArg) < 50 && !strings.Contains(lastArg, " ") {
+			prefix = lastArg
+			sqlArgs = args[:len(args)-1]
+		}
+	}
+
+	// If no prefix provided, use the table name as prefix (already separated from schema)
+	if prefix == "" && b.tableName != "" {
+		prefix = b.tableName
+	}
+
+	// Construct LEFT JOIN with prefix
+	joinClause := query
+	if prefix != "" && !strings.Contains(strings.ToUpper(query), " AS ") {
+		parts := strings.Fields(query)
+		if len(parts) > 0 && !strings.HasPrefix(strings.ToUpper(parts[0]), "LEFT") && !strings.HasPrefix(strings.ToUpper(parts[0]), "JOIN") {
+			joinClause = fmt.Sprintf("%s AS %s", parts[0], prefix)
+			if len(parts) > 1 {
+				joinClause += " " + strings.Join(parts[1:], " ")
+			}
+		}
+	}
+
+	b.query = b.query.Join("LEFT JOIN " + joinClause, sqlArgs...)
 	return b
 }
 
@@ -152,7 +234,19 @@ func (b *BunSelectQuery) Scan(ctx context.Context, dest interface{}) error {
 }
 
 func (b *BunSelectQuery) Count(ctx context.Context) (int, error) {
-	count, err := b.query.Count(ctx)
+	// If Model() was set, use bun's native Count() which works properly
+	if b.hasModel {
+		count, err := b.query.Count(ctx)
+		return count, err
+	}
+
+	// Otherwise, wrap as subquery to avoid "Model(nil)" error
+	// This is needed when only Table() is set without a model
+	var count int
+	err := b.db.NewSelect().
+		TableExpr("(?) AS subquery", b.query).
+		ColumnExpr("COUNT(*)").
+		Scan(ctx, &count)
 	return count, err
 }
 
@@ -302,7 +396,10 @@ type BunTxAdapter struct {
 }
 
 func (b *BunTxAdapter) NewSelect() common.SelectQuery {
-	return &BunSelectQuery{query: b.tx.NewSelect()}
+	return &BunSelectQuery{
+		query: b.tx.NewSelect(),
+		db:    b.tx,
+	}
 }
 
 func (b *BunTxAdapter) NewInsert() common.InsertQuery {

pkg/common/adapters/database/gorm.go

@@ -2,6 +2,8 @@ package database
 
 import (
 	"context"
+	"fmt"
+	"strings"
 
 	"github.com/Warky-Devs/ResolveSpec/pkg/common"
 	"gorm.io/gorm"
@@ -67,16 +69,29 @@ func (g *GormAdapter) RunInTransaction(ctx context.Context, fn func(common.Datab
 
 // GormSelectQuery implements SelectQuery for GORM
 type GormSelectQuery struct {
-	db *gorm.DB
+	db         *gorm.DB
+	schema     string // Separated schema name
+	tableName  string // Just the table name, without schema
+	tableAlias string
 }
 
 func (g *GormSelectQuery) Model(model interface{}) common.SelectQuery {
 	g.db = g.db.Model(model)
+
+	// Try to get table name from model if it implements TableNameProvider
+	if provider, ok := model.(common.TableNameProvider); ok {
+		fullTableName := provider.TableName()
+		// Check if the table name contains schema (e.g., "schema.table")
+		g.schema, g.tableName = parseTableName(fullTableName)
+	}
+
 	return g
 }
 
 func (g *GormSelectQuery) Table(table string) common.SelectQuery {
 	g.db = g.db.Table(table)
+	// Check if the table name contains schema (e.g., "schema.table")
+	g.schema, g.tableName = parseTableName(table)
 	return g
 }
 
@@ -96,12 +111,74 @@ func (g *GormSelectQuery) WhereOr(query string, args ...interface{}) common.Sele
 }
 
 func (g *GormSelectQuery) Join(query string, args ...interface{}) common.SelectQuery {
-	g.db = g.db.Joins(query, args...)
+	// Extract optional prefix from args
+	// If the last arg is a string that looks like a table prefix, use it
+	var prefix string
+	sqlArgs := args
+
+	if len(args) > 0 {
+		if lastArg, ok := args[len(args)-1].(string); ok && len(lastArg) < 50 && !strings.Contains(lastArg, " ") {
+			// Likely a prefix, not a SQL parameter
+			prefix = lastArg
+			sqlArgs = args[:len(args)-1]
+		}
+	}
+
+	// If no prefix provided, use the table name as prefix (already separated from schema)
+	if prefix == "" && g.tableName != "" {
+		prefix = g.tableName
+	}
+
+	// If prefix is provided, add it as an alias in the join
+	// GORM expects: "JOIN table AS alias ON condition"
+	joinClause := query
+	if prefix != "" && !strings.Contains(strings.ToUpper(query), " AS ") {
+		// If query doesn't already have AS, check if it's a simple table name
+		parts := strings.Fields(query)
+		if len(parts) > 0 && !strings.HasPrefix(strings.ToUpper(parts[0]), "JOIN") {
+			// Simple table name, add prefix: "table AS prefix"
+			joinClause = fmt.Sprintf("%s AS %s", parts[0], prefix)
+			if len(parts) > 1 {
+				// Has ON clause: "table ON ..." becomes "table AS prefix ON ..."
+				joinClause += " " + strings.Join(parts[1:], " ")
+			}
+		}
+	}
+
+	g.db = g.db.Joins(joinClause, sqlArgs...)
 	return g
 }
 
 func (g *GormSelectQuery) LeftJoin(query string, args ...interface{}) common.SelectQuery {
-	g.db = g.db.Joins("LEFT JOIN "+query, args...)
+	// Extract optional prefix from args
+	var prefix string
+	sqlArgs := args
+
+	if len(args) > 0 {
+		if lastArg, ok := args[len(args)-1].(string); ok && len(lastArg) < 50 && !strings.Contains(lastArg, " ") {
+			prefix = lastArg
+			sqlArgs = args[:len(args)-1]
+		}
+	}
+
+	// If no prefix provided, use the table name as prefix (already separated from schema)
+	if prefix == "" && g.tableName != "" {
+		prefix = g.tableName
+	}
+
+	// Construct LEFT JOIN with prefix
+	joinClause := query
+	if prefix != "" && !strings.Contains(strings.ToUpper(query), " AS ") {
+		parts := strings.Fields(query)
+		if len(parts) > 0 && !strings.HasPrefix(strings.ToUpper(parts[0]), "LEFT") && !strings.HasPrefix(strings.ToUpper(parts[0]), "JOIN") {
+			joinClause = fmt.Sprintf("%s AS %s", parts[0], prefix)
+			if len(parts) > 1 {
+				joinClause += " " + strings.Join(parts[1:], " ")
+			}
+		}
+	}
+
+	g.db = g.db.Joins("LEFT JOIN "+joinClause, sqlArgs...)
 	return g
 }
 

pkg/common/adapters/database/utils.go

@@ -0,0 +1,13 @@
+package database
+
+import "strings"
+
+// parseTableName splits a table name that may contain schema into separate schema and table
+// For example: "public.users" -> ("public", "users")
+//              "users" -> ("", "users")
+func parseTableName(fullTableName string) (schema, table string) {
+	if idx := strings.LastIndex(fullTableName, "."); idx != -1 {
+		return fullTableName[:idx], fullTableName[idx+1:]
+	}
+	return "", fullTableName
+}

pkg/common/types.go

@@ -37,9 +37,10 @@ type PreloadOption struct {
 }
 
 type FilterOption struct {
-	Column   string      `json:"column"`
-	Operator string      `json:"operator"`
-	Value    interface{} `json:"value"`
+	Column        string      `json:"column"`
+	Operator      string      `json:"operator"`
+	Value         interface{} `json:"value"`
+	LogicOperator string      `json:"logic_operator"` // "AND" or "OR" - how this filter combines with previous filters
 }
 
 type SortOption struct {

pkg/common/validation.go

@@ -0,0 +1,272 @@
+package common
+
+import (
+	"fmt"
+	"reflect"
+	"strings"
+
+	"github.com/Warky-Devs/ResolveSpec/pkg/logger"
+)
+
+// ColumnValidator validates column names against a model's fields
+type ColumnValidator struct {
+	validColumns map[string]bool
+	model        interface{}
+}
+
+// NewColumnValidator creates a new column validator for a given model
+func NewColumnValidator(model interface{}) *ColumnValidator {
+	validator := &ColumnValidator{
+		validColumns: make(map[string]bool),
+		model:        model,
+	}
+	validator.buildValidColumns()
+	return validator
+}
+
+// buildValidColumns extracts all valid column names from the model using reflection
+func (v *ColumnValidator) buildValidColumns() {
+	modelType := reflect.TypeOf(v.model)
+
+	// Unwrap pointers, slices, and arrays to get to the base struct type
+	for modelType != nil && (modelType.Kind() == reflect.Ptr || modelType.Kind() == reflect.Slice || modelType.Kind() == reflect.Array) {
+		modelType = modelType.Elem()
+	}
+
+	// Validate that we have a struct type
+	if modelType == nil || modelType.Kind() != reflect.Struct {
+		return
+	}
+
+	// Extract column names from struct fields
+	for i := 0; i < modelType.NumField(); i++ {
+		field := modelType.Field(i)
+
+		if !field.IsExported() {
+			continue
+		}
+
+		// Get column name from bun, gorm, or json tag
+		columnName := v.getColumnName(field)
+		if columnName != "" && columnName != "-" {
+			v.validColumns[strings.ToLower(columnName)] = true
+		}
+	}
+}
+
+// getColumnName extracts the column name from a struct field's tags
+// Supports both Bun and GORM tags
+func (v *ColumnValidator) getColumnName(field reflect.StructField) string {
+	// First check Bun tag for column name
+	bunTag := field.Tag.Get("bun")
+	if bunTag != "" && bunTag != "-" {
+		parts := strings.Split(bunTag, ",")
+		// The first part is usually the column name
+		columnName := strings.TrimSpace(parts[0])
+		if columnName != "" && columnName != "-" {
+			return columnName
+		}
+	}
+
+	// Check GORM tag for column name
+	gormTag := field.Tag.Get("gorm")
+	if strings.Contains(gormTag, "column:") {
+		parts := strings.Split(gormTag, ";")
+		for _, part := range parts {
+			part = strings.TrimSpace(part)
+			if strings.HasPrefix(part, "column:") {
+				return strings.TrimPrefix(part, "column:")
+			}
+		}
+	}
+
+	// Fall back to JSON tag
+	jsonTag := field.Tag.Get("json")
+	if jsonTag != "" && jsonTag != "-" {
+		// Extract just the name part (before any comma)
+		jsonName := strings.Split(jsonTag, ",")[0]
+		return jsonName
+	}
+
+	// Fall back to field name in lowercase (snake_case conversion would be better)
+	return strings.ToLower(field.Name)
+}
+
+// ValidateColumn validates a single column name
+// Returns nil if valid, error if invalid
+// Columns prefixed with "cql" (case insensitive) are always valid
+func (v *ColumnValidator) ValidateColumn(column string) error {
+	// Allow empty columns
+	if column == "" {
+		return nil
+	}
+
+	// Allow columns prefixed with "cql" (case insensitive) for computed columns
+	if strings.HasPrefix(strings.ToLower(column), "cql") {
+		return nil
+	}
+
+	// Check if column exists in model
+	if _, exists := v.validColumns[strings.ToLower(column)]; !exists {
+		return fmt.Errorf("invalid column '%s': column does not exist in model", column)
+	}
+
+	return nil
+}
+
+// IsValidColumn checks if a column is valid
+// Returns true if valid, false if invalid
+func (v *ColumnValidator) IsValidColumn(column string) bool {
+	return v.ValidateColumn(column) == nil
+}
+
+// FilterValidColumns filters a list of columns, returning only valid ones
+// Logs warnings for any invalid columns
+func (v *ColumnValidator) FilterValidColumns(columns []string) []string {
+	if len(columns) == 0 {
+		return columns
+	}
+
+	validColumns := make([]string, 0, len(columns))
+	for _, col := range columns {
+		if v.IsValidColumn(col) {
+			validColumns = append(validColumns, col)
+		} else {
+			logger.Warn("Invalid column '%s' filtered out: column does not exist in model", col)
+		}
+	}
+	return validColumns
+}
+
+// ValidateColumns validates multiple column names
+// Returns error with details about all invalid columns
+func (v *ColumnValidator) ValidateColumns(columns []string) error {
+	var invalidColumns []string
+
+	for _, column := range columns {
+		if err := v.ValidateColumn(column); err != nil {
+			invalidColumns = append(invalidColumns, column)
+		}
+	}
+
+	if len(invalidColumns) > 0 {
+		return fmt.Errorf("invalid columns: %s", strings.Join(invalidColumns, ", "))
+	}
+
+	return nil
+}
+
+// ValidateRequestOptions validates all column references in RequestOptions
+func (v *ColumnValidator) ValidateRequestOptions(options RequestOptions) error {
+	// Validate Columns
+	if err := v.ValidateColumns(options.Columns); err != nil {
+		return fmt.Errorf("in select columns: %w", err)
+	}
+
+	// Validate OmitColumns
+	if err := v.ValidateColumns(options.OmitColumns); err != nil {
+		return fmt.Errorf("in omit columns: %w", err)
+	}
+
+	// Validate Filter columns
+	for _, filter := range options.Filters {
+		if err := v.ValidateColumn(filter.Column); err != nil {
+			return fmt.Errorf("in filter: %w", err)
+		}
+	}
+
+	// Validate Sort columns
+	for _, sort := range options.Sort {
+		if err := v.ValidateColumn(sort.Column); err != nil {
+			return fmt.Errorf("in sort: %w", err)
+		}
+	}
+
+	// Validate Preload columns (if specified)
+	for _, preload := range options.Preload {
+		// Note: We don't validate the relation name itself, as it's a relationship
+		// Only validate columns if specified for the preload
+		if err := v.ValidateColumns(preload.Columns); err != nil {
+			return fmt.Errorf("in preload '%s' columns: %w", preload.Relation, err)
+		}
+		if err := v.ValidateColumns(preload.OmitColumns); err != nil {
+			return fmt.Errorf("in preload '%s' omit columns: %w", preload.Relation, err)
+		}
+
+		// Validate filter columns in preload
+		for _, filter := range preload.Filters {
+			if err := v.ValidateColumn(filter.Column); err != nil {
+				return fmt.Errorf("in preload '%s' filter: %w", preload.Relation, err)
+			}
+		}
+	}
+
+	return nil
+}
+
+// FilterRequestOptions filters all column references in RequestOptions
+// Returns a new RequestOptions with only valid columns, logging warnings for invalid ones
+func (v *ColumnValidator) FilterRequestOptions(options RequestOptions) RequestOptions {
+	filtered := options
+
+	// Filter Columns
+	filtered.Columns = v.FilterValidColumns(options.Columns)
+
+	// Filter OmitColumns
+	filtered.OmitColumns = v.FilterValidColumns(options.OmitColumns)
+
+	// Filter Filter columns
+	validFilters := make([]FilterOption, 0, len(options.Filters))
+	for _, filter := range options.Filters {
+		if v.IsValidColumn(filter.Column) {
+			validFilters = append(validFilters, filter)
+		} else {
+			logger.Warn("Invalid column in filter '%s' removed", filter.Column)
+		}
+	}
+	filtered.Filters = validFilters
+
+	// Filter Sort columns
+	validSorts := make([]SortOption, 0, len(options.Sort))
+	for _, sort := range options.Sort {
+		if v.IsValidColumn(sort.Column) {
+			validSorts = append(validSorts, sort)
+		} else {
+			logger.Warn("Invalid column in sort '%s' removed", sort.Column)
+		}
+	}
+	filtered.Sort = validSorts
+
+	// Filter Preload columns
+	validPreloads := make([]PreloadOption, 0, len(options.Preload))
+	for _, preload := range options.Preload {
+		filteredPreload := preload
+		filteredPreload.Columns = v.FilterValidColumns(preload.Columns)
+		filteredPreload.OmitColumns = v.FilterValidColumns(preload.OmitColumns)
+
+		// Filter preload filters
+		validPreloadFilters := make([]FilterOption, 0, len(preload.Filters))
+		for _, filter := range preload.Filters {
+			if v.IsValidColumn(filter.Column) {
+				validPreloadFilters = append(validPreloadFilters, filter)
+			} else {
+				logger.Warn("Invalid column in preload '%s' filter '%s' removed", preload.Relation, filter.Column)
+			}
+		}
+		filteredPreload.Filters = validPreloadFilters
+
+		validPreloads = append(validPreloads, filteredPreload)
+	}
+	filtered.Preload = validPreloads
+
+	return filtered
+}
+
+// GetValidColumns returns a list of all valid column names for debugging purposes
+func (v *ColumnValidator) GetValidColumns() []string {
+	columns := make([]string, 0, len(v.validColumns))
+	for col := range v.validColumns {
+		columns = append(columns, col)
+	}
+	return columns
+}

pkg/common/validation_test.go

@@ -0,0 +1,363 @@
+package common
+
+import (
+	"strings"
+	"testing"
+)
+
+// TestModel represents a sample model for testing
+type TestModel struct {
+	ID        int64  `json:"id" gorm:"primaryKey"`
+	Name      string `json:"name" gorm:"column:name"`
+	Email     string `json:"email" bun:"email"`
+	Age       int    `json:"age"`
+	IsActive  bool   `json:"is_active"`
+	CreatedAt string `json:"created_at"`
+}
+
+func TestNewColumnValidator(t *testing.T) {
+	model := TestModel{}
+	validator := NewColumnValidator(model)
+
+	if validator == nil {
+		t.Fatal("Expected validator to be created")
+	}
+
+	if len(validator.validColumns) == 0 {
+		t.Fatal("Expected validator to have valid columns")
+	}
+
+	// Check that expected columns are present
+	expectedColumns := []string{"id", "name", "email", "age", "is_active", "created_at"}
+	for _, col := range expectedColumns {
+		if !validator.validColumns[col] {
+			t.Errorf("Expected column '%s' to be valid", col)
+		}
+	}
+}
+
+func TestValidateColumn(t *testing.T) {
+	model := TestModel{}
+	validator := NewColumnValidator(model)
+
+	tests := []struct {
+		name        string
+		column      string
+		shouldError bool
+	}{
+		{"Valid column - id", "id", false},
+		{"Valid column - name", "name", false},
+		{"Valid column - email", "email", false},
+		{"Valid column - uppercase", "ID", false}, // Case insensitive
+		{"Invalid column", "invalid_column", true},
+		{"CQL prefixed - should be valid", "cqlComputedField", false},
+		{"CQL prefixed uppercase - should be valid", "CQLComputedField", false},
+		{"Empty column", "", false}, // Empty columns are allowed
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			err := validator.ValidateColumn(tt.column)
+			if tt.shouldError && err == nil {
+				t.Errorf("Expected error for column '%s', got nil", tt.column)
+			}
+			if !tt.shouldError && err != nil {
+				t.Errorf("Expected no error for column '%s', got: %v", tt.column, err)
+			}
+		})
+	}
+}
+
+func TestValidateColumns(t *testing.T) {
+	model := TestModel{}
+	validator := NewColumnValidator(model)
+
+	tests := []struct {
+		name        string
+		columns     []string
+		shouldError bool
+	}{
+		{"All valid columns", []string{"id", "name", "email"}, false},
+		{"One invalid column", []string{"id", "invalid_col", "name"}, true},
+		{"All invalid columns", []string{"bad1", "bad2"}, true},
+		{"With CQL prefix", []string{"id", "cqlComputed", "name"}, false},
+		{"Empty list", []string{}, false},
+		{"Nil list", nil, false},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			err := validator.ValidateColumns(tt.columns)
+			if tt.shouldError && err == nil {
+				t.Errorf("Expected error for columns %v, got nil", tt.columns)
+			}
+			if !tt.shouldError && err != nil {
+				t.Errorf("Expected no error for columns %v, got: %v", tt.columns, err)
+			}
+		})
+	}
+}
+
+func TestValidateRequestOptions(t *testing.T) {
+	model := TestModel{}
+	validator := NewColumnValidator(model)
+
+	tests := []struct {
+		name        string
+		options     RequestOptions
+		shouldError bool
+		errorMsg    string
+	}{
+		{
+			name: "Valid options with columns",
+			options: RequestOptions{
+				Columns: []string{"id", "name"},
+				Filters: []FilterOption{
+					{Column: "name", Operator: "eq", Value: "test"},
+				},
+				Sort: []SortOption{
+					{Column: "id", Direction: "ASC"},
+				},
+			},
+			shouldError: false,
+		},
+		{
+			name: "Invalid column in Columns",
+			options: RequestOptions{
+				Columns: []string{"id", "invalid_column"},
+			},
+			shouldError: true,
+			errorMsg:    "select columns",
+		},
+		{
+			name: "Invalid column in Filters",
+			options: RequestOptions{
+				Filters: []FilterOption{
+					{Column: "invalid_col", Operator: "eq", Value: "test"},
+				},
+			},
+			shouldError: true,
+			errorMsg:    "filter",
+		},
+		{
+			name: "Invalid column in Sort",
+			options: RequestOptions{
+				Sort: []SortOption{
+					{Column: "invalid_col", Direction: "ASC"},
+				},
+			},
+			shouldError: true,
+			errorMsg:    "sort",
+		},
+		{
+			name: "Valid CQL prefixed columns",
+			options: RequestOptions{
+				Columns: []string{"id", "cqlComputedField"},
+				Filters: []FilterOption{
+					{Column: "cqlCustomFilter", Operator: "eq", Value: "test"},
+				},
+			},
+			shouldError: false,
+		},
+		{
+			name: "Invalid column in Preload",
+			options: RequestOptions{
+				Preload: []PreloadOption{
+					{
+						Relation: "SomeRelation",
+						Columns:  []string{"id", "invalid_col"},
+					},
+				},
+			},
+			shouldError: true,
+			errorMsg:    "preload",
+		},
+		{
+			name: "Valid preload with valid columns",
+			options: RequestOptions{
+				Preload: []PreloadOption{
+					{
+						Relation: "SomeRelation",
+						Columns:  []string{"id", "name"},
+					},
+				},
+			},
+			shouldError: false,
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			err := validator.ValidateRequestOptions(tt.options)
+			if tt.shouldError {
+				if err == nil {
+					t.Errorf("Expected error, got nil")
+				} else if tt.errorMsg != "" && !strings.Contains(err.Error(), tt.errorMsg) {
+					t.Errorf("Expected error to contain '%s', got: %v", tt.errorMsg, err)
+				}
+			} else {
+				if err != nil {
+					t.Errorf("Expected no error, got: %v", err)
+				}
+			}
+		})
+	}
+}
+
+func TestGetValidColumns(t *testing.T) {
+	model := TestModel{}
+	validator := NewColumnValidator(model)
+
+	columns := validator.GetValidColumns()
+	if len(columns) == 0 {
+		t.Error("Expected to get valid columns, got empty list")
+	}
+
+	// Should have at least the columns from TestModel
+	if len(columns) < 6 {
+		t.Errorf("Expected at least 6 columns, got %d", len(columns))
+	}
+}
+
+// Test with Bun tags specifically
+type BunModel struct {
+	ID    int64  `bun:"id,pk"`
+	Name  string `bun:"name"`
+	Email string `bun:"user_email"`
+}
+
+func TestBunTagSupport(t *testing.T) {
+	model := BunModel{}
+	validator := NewColumnValidator(model)
+
+	// Test that bun tags are properly recognized
+	tests := []struct {
+		column      string
+		shouldError bool
+	}{
+		{"id", false},
+		{"name", false},
+		{"user_email", false}, // Bun tag specifies this name
+		{"email", true},       // JSON tag would be "email", but bun tag says "user_email"
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.column, func(t *testing.T) {
+			err := validator.ValidateColumn(tt.column)
+			if tt.shouldError && err == nil {
+				t.Errorf("Expected error for column '%s'", tt.column)
+			}
+			if !tt.shouldError && err != nil {
+				t.Errorf("Expected no error for column '%s', got: %v", tt.column, err)
+			}
+		})
+	}
+}
+
+func TestFilterValidColumns(t *testing.T) {
+	model := TestModel{}
+	validator := NewColumnValidator(model)
+
+	tests := []struct {
+		name           string
+		input          []string
+		expectedOutput []string
+	}{
+		{
+			name:           "All valid columns",
+			input:          []string{"id", "name", "email"},
+			expectedOutput: []string{"id", "name", "email"},
+		},
+		{
+			name:           "Mix of valid and invalid",
+			input:          []string{"id", "invalid_col", "name", "bad_col", "email"},
+			expectedOutput: []string{"id", "name", "email"},
+		},
+		{
+			name:           "All invalid columns",
+			input:          []string{"bad1", "bad2"},
+			expectedOutput: []string{},
+		},
+		{
+			name:           "With CQL prefix (should pass)",
+			input:          []string{"id", "cqlComputed", "name"},
+			expectedOutput: []string{"id", "cqlComputed", "name"},
+		},
+		{
+			name:           "Empty input",
+			input:          []string{},
+			expectedOutput: []string{},
+		},
+		{
+			name:           "Nil input",
+			input:          nil,
+			expectedOutput: nil,
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			result := validator.FilterValidColumns(tt.input)
+			if len(result) != len(tt.expectedOutput) {
+				t.Errorf("Expected %d columns, got %d", len(tt.expectedOutput), len(result))
+			}
+			for i, col := range result {
+				if col != tt.expectedOutput[i] {
+					t.Errorf("At index %d: expected %s, got %s", i, tt.expectedOutput[i], col)
+				}
+			}
+		})
+	}
+}
+
+func TestFilterRequestOptions(t *testing.T) {
+	model := TestModel{}
+	validator := NewColumnValidator(model)
+
+	options := RequestOptions{
+		Columns:     []string{"id", "name", "invalid_col"},
+		OmitColumns: []string{"email", "bad_col"},
+		Filters: []FilterOption{
+			{Column: "name", Operator: "eq", Value: "test"},
+			{Column: "invalid_col", Operator: "eq", Value: "test"},
+		},
+		Sort: []SortOption{
+			{Column: "id", Direction: "ASC"},
+			{Column: "bad_col", Direction: "DESC"},
+		},
+	}
+
+	filtered := validator.FilterRequestOptions(options)
+
+	// Check Columns
+	if len(filtered.Columns) != 2 {
+		t.Errorf("Expected 2 columns, got %d", len(filtered.Columns))
+	}
+	if filtered.Columns[0] != "id" || filtered.Columns[1] != "name" {
+		t.Errorf("Expected columns [id, name], got %v", filtered.Columns)
+	}
+
+	// Check OmitColumns
+	if len(filtered.OmitColumns) != 1 {
+		t.Errorf("Expected 1 omit column, got %d", len(filtered.OmitColumns))
+	}
+	if filtered.OmitColumns[0] != "email" {
+		t.Errorf("Expected omit column [email], got %v", filtered.OmitColumns)
+	}
+
+	// Check Filters
+	if len(filtered.Filters) != 1 {
+		t.Errorf("Expected 1 filter, got %d", len(filtered.Filters))
+	}
+	if filtered.Filters[0].Column != "name" {
+		t.Errorf("Expected filter column 'name', got %s", filtered.Filters[0].Column)
+	}
+
+	// Check Sort
+	if len(filtered.Sort) != 1 {
+		t.Errorf("Expected 1 sort, got %d", len(filtered.Sort))
+	}
+	if filtered.Sort[0].Column != "id" {
+		t.Errorf("Expected sort column 'id', got %s", filtered.Sort[0].Column)
+	}
+}

pkg/modelregistry/model_registry.go

@@ -2,6 +2,7 @@ package modelregistry
 
 import (
 	"fmt"
+	"reflect"
 	"sync"
 )
 
@@ -26,11 +27,41 @@ func NewModelRegistry() *DefaultModelRegistry {
 func (r *DefaultModelRegistry) RegisterModel(name string, model interface{}) error {
 	r.mutex.Lock()
 	defer r.mutex.Unlock()
-	
+
 	if _, exists := r.models[name]; exists {
 		return fmt.Errorf("model %s already registered", name)
 	}
-	
+
+	// Validate that model is a non-pointer struct
+	modelType := reflect.TypeOf(model)
+	if modelType == nil {
+		return fmt.Errorf("model cannot be nil")
+	}
+
+	originalType := modelType
+
+	// Unwrap pointers, slices, and arrays to check the underlying type
+	for modelType.Kind() == reflect.Ptr || modelType.Kind() == reflect.Slice || modelType.Kind() == reflect.Array {
+		modelType = modelType.Elem()
+	}
+
+	// Validate that the underlying type is a struct
+	if modelType.Kind() != reflect.Struct {
+		return fmt.Errorf("model must be a struct or pointer to struct, got %s", originalType.String())
+	}
+
+	// If a pointer/slice/array was passed, unwrap to the base struct
+	if originalType != modelType {
+		// Create a zero value of the struct type
+		model = reflect.New(modelType).Elem().Interface()
+	}
+
+	// Additional check: ensure model is not a pointer
+	finalType := reflect.TypeOf(model)
+	if finalType.Kind() == reflect.Ptr {
+		return fmt.Errorf("model must be a non-pointer struct, got pointer to %s. Use MyModel{} instead of &MyModel{}", finalType.Elem().Name())
+	}
+
 	r.models[name] = model
 	return nil
 }

pkg/resolvespec/context.go

@@ -0,0 +1,85 @@
+package resolvespec
+
+import (
+	"context"
+)
+
+// Context keys for request-scoped data
+type contextKey string
+
+const (
+	contextKeySchema    contextKey = "schema"
+	contextKeyEntity    contextKey = "entity"
+	contextKeyTableName contextKey = "tableName"
+	contextKeyModel     contextKey = "model"
+	contextKeyModelPtr  contextKey = "modelPtr"
+)
+
+// WithSchema adds schema to context
+func WithSchema(ctx context.Context, schema string) context.Context {
+	return context.WithValue(ctx, contextKeySchema, schema)
+}
+
+// GetSchema retrieves schema from context
+func GetSchema(ctx context.Context) string {
+	if v := ctx.Value(contextKeySchema); v != nil {
+		return v.(string)
+	}
+	return ""
+}
+
+// WithEntity adds entity to context
+func WithEntity(ctx context.Context, entity string) context.Context {
+	return context.WithValue(ctx, contextKeyEntity, entity)
+}
+
+// GetEntity retrieves entity from context
+func GetEntity(ctx context.Context) string {
+	if v := ctx.Value(contextKeyEntity); v != nil {
+		return v.(string)
+	}
+	return ""
+}
+
+// WithTableName adds table name to context
+func WithTableName(ctx context.Context, tableName string) context.Context {
+	return context.WithValue(ctx, contextKeyTableName, tableName)
+}
+
+// GetTableName retrieves table name from context
+func GetTableName(ctx context.Context) string {
+	if v := ctx.Value(contextKeyTableName); v != nil {
+		return v.(string)
+	}
+	return ""
+}
+
+// WithModel adds model to context
+func WithModel(ctx context.Context, model interface{}) context.Context {
+	return context.WithValue(ctx, contextKeyModel, model)
+}
+
+// GetModel retrieves model from context
+func GetModel(ctx context.Context) interface{} {
+	return ctx.Value(contextKeyModel)
+}
+
+// WithModelPtr adds model pointer to context
+func WithModelPtr(ctx context.Context, modelPtr interface{}) context.Context {
+	return context.WithValue(ctx, contextKeyModelPtr, modelPtr)
+}
+
+// GetModelPtr retrieves model pointer from context
+func GetModelPtr(ctx context.Context) interface{} {
+	return ctx.Value(contextKeyModelPtr)
+}
+
+// WithRequestData adds all request-scoped data to context at once
+func WithRequestData(ctx context.Context, schema, entity, tableName string, model, modelPtr interface{}) context.Context {
+	ctx = WithSchema(ctx, schema)
+	ctx = WithEntity(ctx, entity)
+	ctx = WithTableName(ctx, tableName)
+	ctx = WithModel(ctx, model)
+	ctx = WithModelPtr(ctx, modelPtr)
+	return ctx
+}

pkg/resolvespec/handler.go

@@ -6,6 +6,7 @@ import (
 	"fmt"
 	"net/http"
 	"reflect"
+	"runtime/debug"
 	"strings"
 
 	"github.com/Warky-Devs/ResolveSpec/pkg/common"
@@ -26,8 +27,22 @@ func NewHandler(db common.Database, registry common.ModelRegistry) *Handler {
 	}
 }
 
+// handlePanic is a helper function to handle panics with stack traces
+func (h *Handler) handlePanic(w common.ResponseWriter, method string, err interface{}) {
+	stack := debug.Stack()
+	logger.Error("Panic in %s: %v\nStack trace:\n%s", method, err, string(stack))
+	h.sendError(w, http.StatusInternalServerError, "internal_error", fmt.Sprintf("Internal server error in %s", method), fmt.Errorf("%v", err))
+}
+
 // Handle processes API requests through router-agnostic interface
 func (h *Handler) Handle(w common.ResponseWriter, r common.Request, params map[string]string) {
+	// Capture panics and return error response
+	defer func() {
+		if err := recover(); err != nil {
+			h.handlePanic(w, "Handle", err)
+		}
+	}()
+
 	ctx := context.Background()
 
 	body, err := r.Body()
@@ -50,15 +65,54 @@ func (h *Handler) Handle(w common.ResponseWriter, r common.Request, params map[s
 
 	logger.Info("Handling %s operation for %s.%s", req.Operation, schema, entity)
 
+	// Get model and populate context with request-scoped data
+	model, err := h.registry.GetModelByEntity(schema, entity)
+	if err != nil {
+		logger.Error("Invalid entity: %v", err)
+		h.sendError(w, http.StatusBadRequest, "invalid_entity", "Invalid entity", err)
+		return
+	}
+
+	// Validate that the model is a struct type (not a slice or pointer to slice)
+	modelType := reflect.TypeOf(model)
+	originalType := modelType
+	for modelType != nil && (modelType.Kind() == reflect.Ptr || modelType.Kind() == reflect.Slice || modelType.Kind() == reflect.Array) {
+		modelType = modelType.Elem()
+	}
+
+	if modelType == nil || modelType.Kind() != reflect.Struct {
+		logger.Error("Model for %s.%s must be a struct type, got %v. Please register models as struct types, not slices or pointers to slices.", schema, entity, originalType)
+		h.sendError(w, http.StatusInternalServerError, "invalid_model_type",
+			fmt.Sprintf("Model must be a struct type, got %v. Ensure you register the struct (e.g., ModelCoreAccount{}) not a slice (e.g., []*ModelCoreAccount)", originalType),
+			fmt.Errorf("invalid model type: %v", originalType))
+		return
+	}
+
+	// If the registered model was a pointer or slice, use the unwrapped struct type
+	if originalType != modelType {
+		model = reflect.New(modelType).Elem().Interface()
+	}
+
+	// Create a pointer to the model type for database operations
+	modelPtr := reflect.New(reflect.TypeOf(model)).Interface()
+	tableName := h.getTableName(schema, entity, model)
+
+	// Add request-scoped data to context
+	ctx = WithRequestData(ctx, schema, entity, tableName, model, modelPtr)
+
+	// Validate and filter columns in options (log warnings for invalid columns)
+	validator := common.NewColumnValidator(model)
+	req.Options = validator.FilterRequestOptions(req.Options)
+
 	switch req.Operation {
 	case "read":
-		h.handleRead(ctx, w, schema, entity, id, req.Options)
+		h.handleRead(ctx, w, id, req.Options)
 	case "create":
-		h.handleCreate(ctx, w, schema, entity, req.Data, req.Options)
+		h.handleCreate(ctx, w, req.Data, req.Options)
 	case "update":
-		h.handleUpdate(ctx, w, schema, entity, id, req.ID, req.Data, req.Options)
+		h.handleUpdate(ctx, w, id, req.ID, req.Data, req.Options)
 	case "delete":
-		h.handleDelete(ctx, w, schema, entity, id)
+		h.handleDelete(ctx, w, id)
 	default:
 		logger.Error("Invalid operation: %s", req.Operation)
 		h.sendError(w, http.StatusBadRequest, "invalid_operation", "Invalid operation", nil)
@@ -67,6 +121,13 @@ func (h *Handler) Handle(w common.ResponseWriter, r common.Request, params map[s
 
 // HandleGet processes GET requests for metadata
 func (h *Handler) HandleGet(w common.ResponseWriter, r common.Request, params map[string]string) {
+	// Capture panics and return error response
+	defer func() {
+		if err := recover(); err != nil {
+			h.handlePanic(w, "HandleGet", err)
+		}
+	}()
+
 	schema := params["schema"]
 	entity := params["entity"]
 
@@ -83,21 +144,47 @@ func (h *Handler) HandleGet(w common.ResponseWriter, r common.Request, params ma
 	h.sendResponse(w, metadata, nil)
 }
 
-func (h *Handler) handleRead(ctx context.Context, w common.ResponseWriter, schema, entity, id string, options common.RequestOptions) {
-	logger.Info("Reading records from %s.%s", schema, entity)
+func (h *Handler) handleRead(ctx context.Context, w common.ResponseWriter, id string, options common.RequestOptions) {
+	// Capture panics and return error response
+	defer func() {
+		if err := recover(); err != nil {
+			h.handlePanic(w, "handleRead", err)
+		}
+	}()
 
-	model, err := h.registry.GetModelByEntity(schema, entity)
-	if err != nil {
-		logger.Error("Invalid entity: %v", err)
-		h.sendError(w, http.StatusBadRequest, "invalid_entity", "Invalid entity", err)
+	schema := GetSchema(ctx)
+	entity := GetEntity(ctx)
+	tableName := GetTableName(ctx)
+	model := GetModel(ctx)
+
+	// Validate and unwrap model type to get base struct
+	modelType := reflect.TypeOf(model)
+	for modelType != nil && (modelType.Kind() == reflect.Ptr || modelType.Kind() == reflect.Slice || modelType.Kind() == reflect.Array) {
+		modelType = modelType.Elem()
+	}
+
+	if modelType == nil || modelType.Kind() != reflect.Struct {
+		logger.Error("Model must be a struct type, got %v for %s.%s", modelType, schema, entity)
+		h.sendError(w, http.StatusInternalServerError, "invalid_model", "Model must be a struct type", fmt.Errorf("invalid model type: %v", modelType))
 		return
 	}
 
-	query := h.db.NewSelect().Model(model)
+	logger.Info("Reading records from %s.%s", schema, entity)
 
-	// Get table name
-	tableName := h.getTableName(schema, entity, model)
-	query = query.Table(tableName)
+	// Create the model pointer for Scan() operations
+	sliceType := reflect.SliceOf(reflect.PointerTo(modelType))
+	modelPtr := reflect.New(sliceType).Interface()
+
+	// Start with Model() using the slice pointer to avoid "Model(nil)" errors in Count()
+	// Bun's Model() accepts both single pointers and slice pointers
+	query := h.db.NewSelect().Model(modelPtr)
+
+	// Only set Table() if the model doesn't provide a table name via the underlying type
+	// Create a temporary instance to check for TableNameProvider
+	tempInstance := reflect.New(modelType).Interface()
+	if provider, ok := tempInstance.(common.TableNameProvider); !ok || provider.TableName() == "" {
+		query = query.Table(tableName)
+	}
 
 	// Apply column selection
 	if len(options.Columns) > 0 {
@@ -149,7 +236,8 @@ func (h *Handler) handleRead(ctx context.Context, w common.ResponseWriter, schem
 	var result interface{}
 	if id != "" {
 		logger.Debug("Querying single record with ID: %s", id)
-		singleResult := model
+		// For single record, create a new pointer to the struct type
+		singleResult := reflect.New(modelType).Interface()
 		query = query.Where("id = ?", id)
 		if err := query.Scan(ctx, singleResult); err != nil {
 			logger.Error("Error querying record: %v", err)
@@ -159,15 +247,13 @@ func (h *Handler) handleRead(ctx context.Context, w common.ResponseWriter, schem
 		result = singleResult
 	} else {
 		logger.Debug("Querying multiple records")
-		sliceType := reflect.SliceOf(reflect.TypeOf(model))
-		results := reflect.New(sliceType).Interface()
-
-		if err := query.Scan(ctx, results); err != nil {
+		// Use the modelPtr already created and set on the query
+		if err := query.Scan(ctx, modelPtr); err != nil {
 			logger.Error("Error querying records: %v", err)
 			h.sendError(w, http.StatusInternalServerError, "query_error", "Error executing query", err)
 			return
 		}
-		result = reflect.ValueOf(results).Elem().Interface()
+		result = reflect.ValueOf(modelPtr).Elem().Interface()
 	}
 
 	logger.Info("Successfully retrieved records")
@@ -189,17 +275,20 @@ func (h *Handler) handleRead(ctx context.Context, w common.ResponseWriter, schem
 	})
 }
 
-func (h *Handler) handleCreate(ctx context.Context, w common.ResponseWriter, schema, entity string, data interface{}, options common.RequestOptions) {
+func (h *Handler) handleCreate(ctx context.Context, w common.ResponseWriter, data interface{}, options common.RequestOptions) {
+	// Capture panics and return error response
+	defer func() {
+		if err := recover(); err != nil {
+			h.handlePanic(w, "handleCreate", err)
+		}
+	}()
+
+	schema := GetSchema(ctx)
+	entity := GetEntity(ctx)
+	tableName := GetTableName(ctx)
+
 	logger.Info("Creating records for %s.%s", schema, entity)
 
-	// Get the model to determine the actual table name
-	model, err := h.registry.GetModelByEntity(schema, entity)
-	if err != nil {
-		logger.Warn("Model not found, using default table name")
-		model = nil
-	}
-
-	tableName := h.getTableName(schema, entity, model)
 	query := h.db.NewInsert().Table(tableName)
 
 	switch v := data.(type) {
@@ -269,18 +358,20 @@ func (h *Handler) handleCreate(ctx context.Context, w common.ResponseWriter, sch
 	}
 }
 
-func (h *Handler) handleUpdate(ctx context.Context, w common.ResponseWriter, schema, entity, urlID string, reqID interface{}, data interface{}, options common.RequestOptions) {
+func (h *Handler) handleUpdate(ctx context.Context, w common.ResponseWriter, urlID string, reqID interface{}, data interface{}, options common.RequestOptions) {
+	// Capture panics and return error response
+	defer func() {
+		if err := recover(); err != nil {
+			h.handlePanic(w, "handleUpdate", err)
+		}
+	}()
+
+	schema := GetSchema(ctx)
+	entity := GetEntity(ctx)
+	tableName := GetTableName(ctx)
+
 	logger.Info("Updating records for %s.%s", schema, entity)
 
-	// Get the model to determine the actual table name
-	model, err := h.registry.GetModelByEntity(schema, entity)
-	if err != nil {
-		logger.Warn("Model not found, using default table name")
-		// Fallback to entity name (without schema for SQLite compatibility)
-		model = nil
-	}
-
-	tableName := h.getTableName(schema, entity, model)
 	query := h.db.NewUpdate().Table(tableName)
 
 	switch updates := data.(type) {
@@ -324,7 +415,18 @@ func (h *Handler) handleUpdate(ctx context.Context, w common.ResponseWriter, sch
 	h.sendResponse(w, data, nil)
 }
 
-func (h *Handler) handleDelete(ctx context.Context, w common.ResponseWriter, schema, entity, id string) {
+func (h *Handler) handleDelete(ctx context.Context, w common.ResponseWriter, id string) {
+	// Capture panics and return error response
+	defer func() {
+		if err := recover(); err != nil {
+			h.handlePanic(w, "handleDelete", err)
+		}
+	}()
+
+	schema := GetSchema(ctx)
+	entity := GetEntity(ctx)
+	tableName := GetTableName(ctx)
+
 	logger.Info("Deleting records from %s.%s", schema, entity)
 
 	if id == "" {
@@ -333,14 +435,6 @@ func (h *Handler) handleDelete(ctx context.Context, w common.ResponseWriter, sch
 		return
 	}
 
-	// Get the model to determine the actual table name
-	model, err := h.registry.GetModelByEntity(schema, entity)
-	if err != nil {
-		logger.Warn("Model not found, using default table name")
-		model = nil
-	}
-
-	tableName := h.getTableName(schema, entity, model)
 	query := h.db.NewDelete().Table(tableName).Where("id = ?", id)
 
 	result, err := query.Exec(ctx)
@@ -385,19 +479,86 @@ func (h *Handler) applyFilter(query common.SelectQuery, filter common.FilterOpti
 	}
 }
 
-func (h *Handler) getTableName(schema, entity string, model interface{}) string {
-	if provider, ok := model.(common.TableNameProvider); ok {
-		return provider.TableName()
+// parseTableName splits a table name that may contain schema into separate schema and table
+func (h *Handler) parseTableName(fullTableName string) (schema, table string) {
+	if idx := strings.LastIndex(fullTableName, "."); idx != -1 {
+		return fullTableName[:idx], fullTableName[idx+1:]
 	}
-	return fmt.Sprintf("%s.%s", schema, entity)
+	return "", fullTableName
+}
+
+// getSchemaAndTable returns the schema and table name separately
+// It checks SchemaProvider and TableNameProvider interfaces and handles cases where
+// the table name may already include the schema (e.g., "public.users")
+//
+// Priority order:
+// 1. If TableName() contains a schema (e.g., "myschema.mytable"), that schema takes precedence
+// 2. If model implements SchemaProvider, use that schema
+// 3. Otherwise, use the defaultSchema parameter
+func (h *Handler) getSchemaAndTable(defaultSchema, entity string, model interface{}) (schema, table string) {
+	// First check if model provides a table name
+	// We check this FIRST because the table name might already contain the schema
+	if tableProvider, ok := model.(common.TableNameProvider); ok {
+		tableName := tableProvider.TableName()
+
+		// IMPORTANT: Check if the table name already contains a schema (e.g., "schema.table")
+		// This is common when models need to specify a different schema than the default
+		if tableSchema, tableOnly := h.parseTableName(tableName); tableSchema != "" {
+			// Table name includes schema - use it and ignore any other schema providers
+			logger.Debug("TableName() includes schema: %s.%s", tableSchema, tableOnly)
+			return tableSchema, tableOnly
+		}
+
+		// Table name is just the table name without schema
+		// Now determine which schema to use
+		if schemaProvider, ok := model.(common.SchemaProvider); ok {
+			schema = schemaProvider.SchemaName()
+		} else {
+			schema = defaultSchema
+		}
+
+		return schema, tableName
+	}
+
+	// No TableNameProvider, so check for schema and use entity as table name
+	if schemaProvider, ok := model.(common.SchemaProvider); ok {
+		schema = schemaProvider.SchemaName()
+	} else {
+		schema = defaultSchema
+	}
+
+	// Default to entity name as table
+	return schema, entity
+}
+
+// getTableName returns the full table name including schema (schema.table)
+func (h *Handler) getTableName(schema, entity string, model interface{}) string {
+	schemaName, tableName := h.getSchemaAndTable(schema, entity, model)
+	if schemaName != "" {
+		return fmt.Sprintf("%s.%s", schemaName, tableName)
+	}
+	return tableName
 }
 
 func (h *Handler) generateMetadata(schema, entity string, model interface{}) *common.TableMetadata {
 	modelType := reflect.TypeOf(model)
-	if modelType.Kind() == reflect.Ptr {
+
+	// Unwrap pointers, slices, and arrays to get to the base struct type
+	for modelType != nil && (modelType.Kind() == reflect.Ptr || modelType.Kind() == reflect.Slice || modelType.Kind() == reflect.Array) {
 		modelType = modelType.Elem()
 	}
 
+	// Validate that we have a struct type
+	if modelType == nil || modelType.Kind() != reflect.Struct {
+		logger.Error("Model type must be a struct, got %v for %s.%s", modelType, schema, entity)
+		return &common.TableMetadata{
+			Schema:    schema,
+			Table:     entity,
+			Columns:   make([]common.Column, 0),
+			Relations: make([]string, 0),
+		}
+	}
+
 	metadata := &common.TableMetadata{
 		Schema:    schema,
 		Table:     entity,
@@ -541,10 +702,18 @@ type relationshipInfo struct {
 
 func (h *Handler) applyPreloads(model interface{}, query common.SelectQuery, preloads []common.PreloadOption) common.SelectQuery {
 	modelType := reflect.TypeOf(model)
-	if modelType.Kind() == reflect.Ptr {
+
+	// Unwrap pointers, slices, and arrays to get to the base struct type
+	for modelType != nil && (modelType.Kind() == reflect.Ptr || modelType.Kind() == reflect.Slice || modelType.Kind() == reflect.Array) {
 		modelType = modelType.Elem()
 	}
 
+	// Validate that we have a struct type
+	if modelType == nil || modelType.Kind() != reflect.Struct {
+		logger.Warn("Cannot apply preloads to non-struct type: %v", modelType)
+		return query
+	}
+
 	for _, preload := range preloads {
 		logger.Debug("Processing preload for relation: %s", preload.Relation)
 		relInfo := h.getRelationshipInfo(modelType, preload.Relation)
@@ -568,6 +737,12 @@ func (h *Handler) applyPreloads(model interface{}, query common.SelectQuery, pre
 }
 
 func (h *Handler) getRelationshipInfo(modelType reflect.Type, relationName string) *relationshipInfo {
+	// Ensure we have a struct type
+	if modelType == nil || modelType.Kind() != reflect.Struct {
+		logger.Warn("Cannot get relationship info from non-struct type: %v", modelType)
+		return nil
+	}
+
 	for i := 0; i < modelType.NumField(); i++ {
 		field := modelType.Field(i)
 		jsonTag := field.Tag.Get("json")

pkg/restheadspec/context.go

@@ -0,0 +1,85 @@
+package restheadspec
+
+import (
+	"context"
+)
+
+// Context keys for request-scoped data
+type contextKey string
+
+const (
+	contextKeySchema    contextKey = "schema"
+	contextKeyEntity    contextKey = "entity"
+	contextKeyTableName contextKey = "tableName"
+	contextKeyModel     contextKey = "model"
+	contextKeyModelPtr  contextKey = "modelPtr"
+)
+
+// WithSchema adds schema to context
+func WithSchema(ctx context.Context, schema string) context.Context {
+	return context.WithValue(ctx, contextKeySchema, schema)
+}
+
+// GetSchema retrieves schema from context
+func GetSchema(ctx context.Context) string {
+	if v := ctx.Value(contextKeySchema); v != nil {
+		return v.(string)
+	}
+	return ""
+}
+
+// WithEntity adds entity to context
+func WithEntity(ctx context.Context, entity string) context.Context {
+	return context.WithValue(ctx, contextKeyEntity, entity)
+}
+
+// GetEntity retrieves entity from context
+func GetEntity(ctx context.Context) string {
+	if v := ctx.Value(contextKeyEntity); v != nil {
+		return v.(string)
+	}
+	return ""
+}
+
+// WithTableName adds table name to context
+func WithTableName(ctx context.Context, tableName string) context.Context {
+	return context.WithValue(ctx, contextKeyTableName, tableName)
+}
+
+// GetTableName retrieves table name from context
+func GetTableName(ctx context.Context) string {
+	if v := ctx.Value(contextKeyTableName); v != nil {
+		return v.(string)
+	}
+	return ""
+}
+
+// WithModel adds model to context
+func WithModel(ctx context.Context, model interface{}) context.Context {
+	return context.WithValue(ctx, contextKeyModel, model)
+}
+
+// GetModel retrieves model from context
+func GetModel(ctx context.Context) interface{} {
+	return ctx.Value(contextKeyModel)
+}
+
+// WithModelPtr adds model pointer to context
+func WithModelPtr(ctx context.Context, modelPtr interface{}) context.Context {
+	return context.WithValue(ctx, contextKeyModelPtr, modelPtr)
+}
+
+// GetModelPtr retrieves model pointer from context
+func GetModelPtr(ctx context.Context) interface{} {
+	return ctx.Value(contextKeyModelPtr)
+}
+
+// WithRequestData adds all request-scoped data to context at once
+func WithRequestData(ctx context.Context, schema, entity, tableName string, model, modelPtr interface{}) context.Context {
+	ctx = WithSchema(ctx, schema)
+	ctx = WithEntity(ctx, entity)
+	ctx = WithTableName(ctx, tableName)
+	ctx = WithModel(ctx, model)
+	ctx = WithModelPtr(ctx, modelPtr)
+	return ctx
+}

pkg/restheadspec/handler.go

@@ -6,6 +6,7 @@ import (
 	"fmt"
 	"net/http"
 	"reflect"
+	"runtime/debug"
 	"strings"
 
 	"github.com/Warky-Devs/ResolveSpec/pkg/common"
@@ -27,9 +28,23 @@ func NewHandler(db common.Database, registry common.ModelRegistry) *Handler {
 	}
 }
 
+// handlePanic is a helper function to handle panics with stack traces
+func (h *Handler) handlePanic(w common.ResponseWriter, method string, err interface{}) {
+	stack := debug.Stack()
+	logger.Error("Panic in %s: %v\nStack trace:\n%s", method, err, string(stack))
+	h.sendError(w, http.StatusInternalServerError, "internal_error", fmt.Sprintf("Internal server error in %s", method), fmt.Errorf("%v", err))
+}
+
 // Handle processes API requests through router-agnostic interface
 // Options are read from HTTP headers instead of request body
 func (h *Handler) Handle(w common.ResponseWriter, r common.Request, params map[string]string) {
+	// Capture panics and return error response
+	defer func() {
+		if err := recover(); err != nil {
+			h.handlePanic(w, "Handle", err)
+		}
+	}()
+
 	ctx := context.Background()
 
 	schema := params["schema"]
@@ -44,14 +59,52 @@ func (h *Handler) Handle(w common.ResponseWriter, r common.Request, params map[s
 
 	logger.Info("Handling %s request for %s.%s", method, schema, entity)
 
+	// Get model and populate context with request-scoped data
+	model, err := h.registry.GetModelByEntity(schema, entity)
+	if err != nil {
+		logger.Error("Invalid entity: %v", err)
+		h.sendError(w, http.StatusBadRequest, "invalid_entity", "Invalid entity", err)
+		return
+	}
+
+	// Validate that the model is a struct type (not a slice or pointer to slice)
+	modelType := reflect.TypeOf(model)
+	originalType := modelType
+	for modelType != nil && (modelType.Kind() == reflect.Ptr || modelType.Kind() == reflect.Slice || modelType.Kind() == reflect.Array) {
+		modelType = modelType.Elem()
+	}
+
+	if modelType == nil || modelType.Kind() != reflect.Struct {
+		logger.Error("Model for %s.%s must be a struct type, got %v. Please register models as struct types, not slices or pointers to slices.", schema, entity, originalType)
+		h.sendError(w, http.StatusInternalServerError, "invalid_model_type",
+			fmt.Sprintf("Model must be a struct type, got %v. Ensure you register the struct (e.g., ModelCoreAccount{}) not a slice (e.g., []*ModelCoreAccount)", originalType),
+			fmt.Errorf("invalid model type: %v", originalType))
+		return
+	}
+
+	// If the registered model was a pointer or slice, use the unwrapped struct type
+	if originalType != modelType {
+		model = reflect.New(modelType).Elem().Interface()
+	}
+
+	modelPtr := reflect.New(reflect.TypeOf(model)).Interface()
+	tableName := h.getTableName(schema, entity, model)
+
+	// Add request-scoped data to context
+	ctx = WithRequestData(ctx, schema, entity, tableName, model, modelPtr)
+
+	// Validate and filter columns in options (log warnings for invalid columns)
+	validator := common.NewColumnValidator(model)
+	options = filterExtendedOptions(validator, options)
+
 	switch method {
 	case "GET":
 		if id != "" {
 			// GET with ID - read single record
-			h.handleRead(ctx, w, schema, entity, id, options)
+			h.handleRead(ctx, w, id, options)
 		} else {
 			// GET without ID - read multiple records
-			h.handleRead(ctx, w, schema, entity, "", options)
+			h.handleRead(ctx, w, "", options)
 		}
 	case "POST":
 		// Create operation
@@ -67,7 +120,7 @@ func (h *Handler) Handle(w common.ResponseWriter, r common.Request, params map[s
 			h.sendError(w, http.StatusBadRequest, "invalid_request", "Invalid request body", err)
 			return
 		}
-		h.handleCreate(ctx, w, schema, entity, data, options)
+		h.handleCreate(ctx, w, data, options)
 	case "PUT", "PATCH":
 		// Update operation
 		body, err := r.Body()
@@ -82,9 +135,9 @@ func (h *Handler) Handle(w common.ResponseWriter, r common.Request, params map[s
 			h.sendError(w, http.StatusBadRequest, "invalid_request", "Invalid request body", err)
 			return
 		}
-		h.handleUpdate(ctx, w, schema, entity, id, nil, data, options)
+		h.handleUpdate(ctx, w, id, nil, data, options)
 	case "DELETE":
-		h.handleDelete(ctx, w, schema, entity, id)
+		h.handleDelete(ctx, w, id)
 	default:
 		logger.Error("Invalid HTTP method: %s", method)
 		h.sendError(w, http.StatusMethodNotAllowed, "invalid_method", "Invalid HTTP method", nil)
@@ -93,6 +146,13 @@ func (h *Handler) Handle(w common.ResponseWriter, r common.Request, params map[s
 
 // HandleGet processes GET requests for metadata
 func (h *Handler) HandleGet(w common.ResponseWriter, r common.Request, params map[string]string) {
+	// Capture panics and return error response
+	defer func() {
+		if err := recover(); err != nil {
+			h.handlePanic(w, "HandleGet", err)
+		}
+	}()
+
 	schema := params["schema"]
 	entity := params["entity"]
 
@@ -111,21 +171,46 @@ func (h *Handler) HandleGet(w common.ResponseWriter, r common.Request, params ma
 
 // parseOptionsFromHeaders is now implemented in headers.go
 
-func (h *Handler) handleRead(ctx context.Context, w common.ResponseWriter, schema, entity, id string, options ExtendedRequestOptions) {
-	logger.Info("Reading records from %s.%s", schema, entity)
+func (h *Handler) handleRead(ctx context.Context, w common.ResponseWriter, id string, options ExtendedRequestOptions) {
+	// Capture panics and return error response
+	defer func() {
+		if err := recover(); err != nil {
+			h.handlePanic(w, "handleRead", err)
+		}
+	}()
 
-	model, err := h.registry.GetModelByEntity(schema, entity)
-	if err != nil {
-		logger.Error("Invalid entity: %v", err)
-		h.sendError(w, http.StatusBadRequest, "invalid_entity", "Invalid entity", err)
+	schema := GetSchema(ctx)
+	entity := GetEntity(ctx)
+	tableName := GetTableName(ctx)
+	model := GetModel(ctx)
+
+	// Validate and unwrap model type to get base struct
+	modelType := reflect.TypeOf(model)
+	for modelType != nil && (modelType.Kind() == reflect.Ptr || modelType.Kind() == reflect.Slice || modelType.Kind() == reflect.Array) {
+		modelType = modelType.Elem()
+	}
+
+	if modelType == nil || modelType.Kind() != reflect.Struct {
+		logger.Error("Model must be a struct type, got %v for %s.%s", modelType, schema, entity)
+		h.sendError(w, http.StatusInternalServerError, "invalid_model", "Model must be a struct type", fmt.Errorf("invalid model type: %v", modelType))
 		return
 	}
 
-	query := h.db.NewSelect().Model(model)
+	// Create a pointer to a slice of pointers to the model type for query results
+	modelPtr := reflect.New(reflect.SliceOf(reflect.PointerTo(modelType))).Interface()
 
-	// Get table name
-	tableName := h.getTableName(schema, entity, model)
-	query = query.Table(tableName)
+	logger.Info("Reading records from %s.%s", schema, entity)
+
+	// Start with Model() using the slice pointer to avoid "Model(nil)" errors in Count()
+	// Bun's Model() accepts both single pointers and slice pointers
+	query := h.db.NewSelect().Model(modelPtr)
+
+	// Only set Table() if the model doesn't provide a table name via the underlying type
+	// Create a temporary instance to check for TableNameProvider
+	tempInstance := reflect.New(modelType).Interface()
+	if provider, ok := tempInstance.(common.TableNameProvider); !ok || provider.TableName() == "" {
+		query = query.Table(tableName)
+	}
 
 	// Apply column selection
 	if len(options.Columns) > 0 {
@@ -154,10 +239,21 @@ func (h *Handler) handleRead(ctx context.Context, w common.ResponseWriter, schem
 		// This may need to be handled differently per database adapter
 	}
 
-	// Apply filters
-	for _, filter := range options.Filters {
-		logger.Debug("Applying filter: %s %s %v", filter.Column, filter.Operator, filter.Value)
-		query = h.applyFilter(query, filter)
+	// Apply filters - validate and adjust for column types first
+	for i := range options.Filters {
+		filter := &options.Filters[i]
+
+		// Validate and adjust filter based on column type
+		castInfo := h.ValidateAndAdjustFilterForColumnType(filter, model)
+
+		// Default to AND if LogicOperator is not set
+		logicOp := filter.LogicOperator
+		if logicOp == "" {
+			logicOp = "AND"
+		}
+
+		logger.Debug("Applying filter: %s %s %v (needsCast=%v, logic=%s)", filter.Column, filter.Operator, filter.Value, castInfo.NeedsCast, logicOp)
+		query = h.applyFilter(query, *filter, tableName, castInfo.NeedsCast, logicOp)
 	}
 
 	// Apply custom SQL WHERE clause (AND condition)
@@ -214,9 +310,8 @@ func (h *Handler) handleRead(ctx context.Context, w common.ResponseWriter, schem
 		query = query.Offset(*options.Offset)
 	}
 
-	// Execute query
-	resultSlice := reflect.New(reflect.SliceOf(reflect.TypeOf(model))).Interface()
-	if err := query.Scan(ctx, resultSlice); err != nil {
+	// Execute query - modelPtr was already created earlier
+	if err := query.Scan(ctx, modelPtr); err != nil {
 		logger.Error("Error executing query: %v", err)
 		h.sendError(w, http.StatusInternalServerError, "query_error", "Error executing query", err)
 		return
@@ -238,21 +333,24 @@ func (h *Handler) handleRead(ctx context.Context, w common.ResponseWriter, schem
 		Offset:   offset,
 	}
 
-	h.sendFormattedResponse(w, resultSlice, metadata, options)
+	h.sendFormattedResponse(w, modelPtr, metadata, options)
 }
 
-func (h *Handler) handleCreate(ctx context.Context, w common.ResponseWriter, schema, entity string, data interface{}, options ExtendedRequestOptions) {
+func (h *Handler) handleCreate(ctx context.Context, w common.ResponseWriter, data interface{}, options ExtendedRequestOptions) {
+	// Capture panics and return error response
+	defer func() {
+		if err := recover(); err != nil {
+			h.handlePanic(w, "handleCreate", err)
+		}
+	}()
+
+	schema := GetSchema(ctx)
+	entity := GetEntity(ctx)
+	tableName := GetTableName(ctx)
+	model := GetModel(ctx)
+
 	logger.Info("Creating record in %s.%s", schema, entity)
 
-	model, err := h.registry.GetModelByEntity(schema, entity)
-	if err != nil {
-		logger.Error("Invalid entity: %v", err)
-		h.sendError(w, http.StatusBadRequest, "invalid_entity", "Invalid entity", err)
-		return
-	}
-
-	tableName := h.getTableName(schema, entity, model)
-
 	// Handle batch creation
 	dataValue := reflect.ValueOf(data)
 	if dataValue.Kind() == reflect.Slice || dataValue.Kind() == reflect.Array {
@@ -263,8 +361,8 @@ func (h *Handler) handleCreate(ctx context.Context, w common.ResponseWriter, sch
 			for i := 0; i < dataValue.Len(); i++ {
 				item := dataValue.Index(i).Interface()
 
-				// Convert item to model type
-				modelValue := reflect.New(reflect.TypeOf(model).Elem()).Interface()
+				// Convert item to model type - create a pointer to the model
+				modelValue := reflect.New(reflect.TypeOf(model)).Interface()
 				jsonData, err := json.Marshal(item)
 				if err != nil {
 					return fmt.Errorf("failed to marshal item: %w", err)
@@ -291,8 +389,8 @@ func (h *Handler) handleCreate(ctx context.Context, w common.ResponseWriter, sch
 		return
 	}
 
-	// Single record creation
-	modelValue := reflect.New(reflect.TypeOf(model).Elem()).Interface()
+	// Single record creation - create a pointer to the model
+	modelValue := reflect.New(reflect.TypeOf(model)).Interface()
 	jsonData, err := json.Marshal(data)
 	if err != nil {
 		logger.Error("Error marshaling data: %v", err)
@@ -315,18 +413,20 @@ func (h *Handler) handleCreate(ctx context.Context, w common.ResponseWriter, sch
 	h.sendResponse(w, modelValue, nil)
 }
 
-func (h *Handler) handleUpdate(ctx context.Context, w common.ResponseWriter, schema, entity, id string, idPtr *int64, data interface{}, options ExtendedRequestOptions) {
+func (h *Handler) handleUpdate(ctx context.Context, w common.ResponseWriter, id string, idPtr *int64, data interface{}, options ExtendedRequestOptions) {
+	// Capture panics and return error response
+	defer func() {
+		if err := recover(); err != nil {
+			h.handlePanic(w, "handleUpdate", err)
+		}
+	}()
+
+	schema := GetSchema(ctx)
+	entity := GetEntity(ctx)
+	tableName := GetTableName(ctx)
+
 	logger.Info("Updating record in %s.%s", schema, entity)
 
-	model, err := h.registry.GetModelByEntity(schema, entity)
-	if err != nil {
-		logger.Error("Invalid entity: %v", err)
-		h.sendError(w, http.StatusBadRequest, "invalid_entity", "Invalid entity", err)
-		return
-	}
-
-	tableName := h.getTableName(schema, entity, model)
-
 	// Convert data to map
 	dataMap, ok := data.(map[string]interface{})
 	if !ok {
@@ -367,18 +467,20 @@ func (h *Handler) handleUpdate(ctx context.Context, w common.ResponseWriter, sch
 	}, nil)
 }
 
-func (h *Handler) handleDelete(ctx context.Context, w common.ResponseWriter, schema, entity, id string) {
+func (h *Handler) handleDelete(ctx context.Context, w common.ResponseWriter, id string) {
+	// Capture panics and return error response
+	defer func() {
+		if err := recover(); err != nil {
+			h.handlePanic(w, "handleDelete", err)
+		}
+	}()
+
+	schema := GetSchema(ctx)
+	entity := GetEntity(ctx)
+	tableName := GetTableName(ctx)
+
 	logger.Info("Deleting record from %s.%s", schema, entity)
 
-	model, err := h.registry.GetModelByEntity(schema, entity)
-	if err != nil {
-		logger.Error("Invalid entity: %v", err)
-		h.sendError(w, http.StatusBadRequest, "invalid_entity", "Invalid entity", err)
-		return
-	}
-
-	tableName := h.getTableName(schema, entity, model)
-
 	query := h.db.NewDelete().Table(tableName)
 
 	if id == "" {
@@ -400,80 +502,178 @@ func (h *Handler) handleDelete(ctx context.Context, w common.ResponseWriter, sch
 	}, nil)
 }
 
-func (h *Handler) applyFilter(query common.SelectQuery, filter common.FilterOption) common.SelectQuery {
+// qualifyColumnName ensures column name is fully qualified with table name if not already
+func (h *Handler) qualifyColumnName(columnName, fullTableName string) string {
+	// Check if column already has a table/schema prefix (contains a dot)
+	if strings.Contains(columnName, ".") {
+		return columnName
+	}
+
+	// If no table name provided, return column as-is
+	if fullTableName == "" {
+		return columnName
+	}
+
+	// Extract just the table name from "schema.table" format
+	// Only use the table name part, not the schema
+	tableOnly := fullTableName
+	if idx := strings.LastIndex(fullTableName, "."); idx != -1 {
+		tableOnly = fullTableName[idx+1:]
+	}
+
+	// Return column qualified with just the table name
+	return fmt.Sprintf("%s.%s", tableOnly, columnName)
+}
+
+func (h *Handler) applyFilter(query common.SelectQuery, filter common.FilterOption, tableName string, needsCast bool, logicOp string) common.SelectQuery {
+	// Qualify the column name with table name if not already qualified
+	qualifiedColumn := h.qualifyColumnName(filter.Column, tableName)
+
+	// Apply casting to text if needed for non-numeric columns or non-numeric values
+	if needsCast {
+		qualifiedColumn = fmt.Sprintf("CAST(%s AS TEXT)", qualifiedColumn)
+	}
+
+	// Helper function to apply the correct Where method based on logic operator
+	applyWhere := func(condition string, args ...interface{}) common.SelectQuery {
+		if logicOp == "OR" {
+			return query.WhereOr(condition, args...)
+		}
+		return query.Where(condition, args...)
+	}
+
 	switch strings.ToLower(filter.Operator) {
 	case "eq", "equals":
-		return query.Where(fmt.Sprintf("%s = ?", filter.Column), filter.Value)
+		return applyWhere(fmt.Sprintf("%s = ?", qualifiedColumn), filter.Value)
 	case "neq", "not_equals", "ne":
-		return query.Where(fmt.Sprintf("%s != ?", filter.Column), filter.Value)
+		return applyWhere(fmt.Sprintf("%s != ?", qualifiedColumn), filter.Value)
 	case "gt", "greater_than":
-		return query.Where(fmt.Sprintf("%s > ?", filter.Column), filter.Value)
+		return applyWhere(fmt.Sprintf("%s > ?", qualifiedColumn), filter.Value)
 	case "gte", "greater_than_equals", "ge":
-		return query.Where(fmt.Sprintf("%s >= ?", filter.Column), filter.Value)
+		return applyWhere(fmt.Sprintf("%s >= ?", qualifiedColumn), filter.Value)
 	case "lt", "less_than":
-		return query.Where(fmt.Sprintf("%s < ?", filter.Column), filter.Value)
+		return applyWhere(fmt.Sprintf("%s < ?", qualifiedColumn), filter.Value)
 	case "lte", "less_than_equals", "le":
-		return query.Where(fmt.Sprintf("%s <= ?", filter.Column), filter.Value)
+		return applyWhere(fmt.Sprintf("%s <= ?", qualifiedColumn), filter.Value)
 	case "like":
-		return query.Where(fmt.Sprintf("%s LIKE ?", filter.Column), filter.Value)
+		return applyWhere(fmt.Sprintf("%s LIKE ?", qualifiedColumn), filter.Value)
 	case "ilike":
 		// Use ILIKE for case-insensitive search (PostgreSQL)
-		// For other databases, cast to citext or use LOWER()
-		return query.Where(fmt.Sprintf("CAST(%s AS TEXT) ILIKE ?", filter.Column), filter.Value)
+		// Column is already cast to TEXT if needed
+		return applyWhere(fmt.Sprintf("%s ILIKE ?", qualifiedColumn), filter.Value)
 	case "in":
-		return query.Where(fmt.Sprintf("%s IN (?)", filter.Column), filter.Value)
+		return applyWhere(fmt.Sprintf("%s IN (?)", qualifiedColumn), filter.Value)
 	case "between":
 		// Handle between operator - exclusive (> val1 AND < val2)
 		if values, ok := filter.Value.([]interface{}); ok && len(values) == 2 {
-			return query.Where(fmt.Sprintf("%s > ? AND %s < ?", filter.Column, filter.Column), values[0], values[1])
+			return applyWhere(fmt.Sprintf("%s > ? AND %s < ?", qualifiedColumn, qualifiedColumn), values[0], values[1])
 		} else if values, ok := filter.Value.([]string); ok && len(values) == 2 {
-			return query.Where(fmt.Sprintf("%s > ? AND %s < ?", filter.Column, filter.Column), values[0], values[1])
+			return applyWhere(fmt.Sprintf("%s > ? AND %s < ?", qualifiedColumn, qualifiedColumn), values[0], values[1])
 		}
 		logger.Warn("Invalid BETWEEN filter value format")
 		return query
 	case "between_inclusive":
 		// Handle between inclusive operator - inclusive (>= val1 AND <= val2)
 		if values, ok := filter.Value.([]interface{}); ok && len(values) == 2 {
-			return query.Where(fmt.Sprintf("%s >= ? AND %s <= ?", filter.Column, filter.Column), values[0], values[1])
+			return applyWhere(fmt.Sprintf("%s >= ? AND %s <= ?", qualifiedColumn, qualifiedColumn), values[0], values[1])
 		} else if values, ok := filter.Value.([]string); ok && len(values) == 2 {
-			return query.Where(fmt.Sprintf("%s >= ? AND %s <= ?", filter.Column, filter.Column), values[0], values[1])
+			return applyWhere(fmt.Sprintf("%s >= ? AND %s <= ?", qualifiedColumn, qualifiedColumn), values[0], values[1])
 		}
 		logger.Warn("Invalid BETWEEN INCLUSIVE filter value format")
 		return query
 	case "is_null", "isnull":
-		// Check for NULL values
-		return query.Where(fmt.Sprintf("(%s IS NULL OR %s = '')", filter.Column, filter.Column))
+		// Check for NULL values - don't use cast for NULL checks
+		colName := h.qualifyColumnName(filter.Column, tableName)
+		return applyWhere(fmt.Sprintf("(%s IS NULL OR %s = '')", colName, colName))
 	case "is_not_null", "isnotnull":
-		// Check for NOT NULL values
-		return query.Where(fmt.Sprintf("(%s IS NOT NULL AND %s != '')", filter.Column, filter.Column))
+		// Check for NOT NULL values - don't use cast for NULL checks
+		colName := h.qualifyColumnName(filter.Column, tableName)
+		return applyWhere(fmt.Sprintf("(%s IS NOT NULL AND %s != '')", colName, colName))
 	default:
 		logger.Warn("Unknown filter operator: %s, defaulting to equals", filter.Operator)
-		return query.Where(fmt.Sprintf("%s = ?", filter.Column), filter.Value)
+		return applyWhere(fmt.Sprintf("%s = ?", qualifiedColumn), filter.Value)
 	}
 }
 
-func (h *Handler) getTableName(schema, entity string, model interface{}) string {
-	// Check if model implements TableNameProvider
-	if provider, ok := model.(common.TableNameProvider); ok {
-		tableName := provider.TableName()
-		if tableName != "" {
-			return tableName
+// parseTableName splits a table name that may contain schema into separate schema and table
+func (h *Handler) parseTableName(fullTableName string) (schema, table string) {
+	if idx := strings.LastIndex(fullTableName, "."); idx != -1 {
+		return fullTableName[:idx], fullTableName[idx+1:]
+	}
+	return "", fullTableName
+}
+
+// getSchemaAndTable returns the schema and table name separately
+// It checks SchemaProvider and TableNameProvider interfaces and handles cases where
+// the table name may already include the schema (e.g., "public.users")
+//
+// Priority order:
+// 1. If TableName() contains a schema (e.g., "myschema.mytable"), that schema takes precedence
+// 2. If model implements SchemaProvider, use that schema
+// 3. Otherwise, use the defaultSchema parameter
+func (h *Handler) getSchemaAndTable(defaultSchema, entity string, model interface{}) (schema, table string) {
+	// First check if model provides a table name
+	// We check this FIRST because the table name might already contain the schema
+	if tableProvider, ok := model.(common.TableNameProvider); ok {
+		tableName := tableProvider.TableName()
+
+		// IMPORTANT: Check if the table name already contains a schema (e.g., "schema.table")
+		// This is common when models need to specify a different schema than the default
+		if tableSchema, tableOnly := h.parseTableName(tableName); tableSchema != "" {
+			// Table name includes schema - use it and ignore any other schema providers
+			logger.Debug("TableName() includes schema: %s.%s", tableSchema, tableOnly)
+			return tableSchema, tableOnly
 		}
+
+		// Table name is just the table name without schema
+		// Now determine which schema to use
+		if schemaProvider, ok := model.(common.SchemaProvider); ok {
+			schema = schemaProvider.SchemaName()
+		} else {
+			schema = defaultSchema
+		}
+
+		return schema, tableName
 	}
 
-	// Default to schema.entity
-	if schema != "" {
-		return fmt.Sprintf("%s.%s", schema, entity)
+	// No TableNameProvider, so check for schema and use entity as table name
+	if schemaProvider, ok := model.(common.SchemaProvider); ok {
+		schema = schemaProvider.SchemaName()
+	} else {
+		schema = defaultSchema
 	}
-	return entity
+
+	// Default to entity name as table
+	return schema, entity
+}
+
+// getTableName returns the full table name including schema (schema.table)
+func (h *Handler) getTableName(schema, entity string, model interface{}) string {
+	schemaName, tableName := h.getSchemaAndTable(schema, entity, model)
+	if schemaName != "" {
+		return fmt.Sprintf("%s.%s", schemaName, tableName)
+	}
+	return tableName
 }
 
 func (h *Handler) generateMetadata(schema, entity string, model interface{}) *common.TableMetadata {
 	modelType := reflect.TypeOf(model)
-	if modelType.Kind() == reflect.Ptr {
+
+	// Unwrap pointers, slices, and arrays to get to the base struct type
+	for modelType.Kind() == reflect.Ptr || modelType.Kind() == reflect.Slice || modelType.Kind() == reflect.Array {
 		modelType = modelType.Elem()
 	}
 
+	// Validate that we have a struct type
+	if modelType.Kind() != reflect.Struct {
+		logger.Error("Model type must be a struct, got %s for %s.%s", modelType.Kind(), schema, entity)
+		return &common.TableMetadata{
+			Schema:  schema,
+			Table:   h.getTableName(schema, entity, model),
+			Columns: []common.Column{},
+		}
+	}
+
 	tableName := h.getTableName(schema, entity, model)
 
 	metadata := &common.TableMetadata{
@@ -559,7 +759,7 @@ func (h *Handler) sendFormattedResponse(w common.ResponseWriter, data interface{
 	if options.CleanJSON {
 		data = h.cleanJSON(data)
 	}
-
+	w.SetHeader("Content-Type", "application/json")
 	// Format response based on response format option
 	switch options.ResponseFormat {
 	case "simple":
@@ -614,3 +814,41 @@ func (h *Handler) sendError(w common.ResponseWriter, statusCode int, code, messa
 	w.WriteHeader(statusCode)
 	w.WriteJSON(response)
 }
+
+// filterExtendedOptions filters all column references, removing invalid ones and logging warnings
+func filterExtendedOptions(validator *common.ColumnValidator, options ExtendedRequestOptions) ExtendedRequestOptions {
+	filtered := options
+
+	// Filter base RequestOptions
+	filtered.RequestOptions = validator.FilterRequestOptions(options.RequestOptions)
+
+	// Filter SearchColumns
+	filtered.SearchColumns = validator.FilterValidColumns(options.SearchColumns)
+
+	// Filter AdvancedSQL column keys
+	filteredAdvSQL := make(map[string]string)
+	for colName, sqlExpr := range options.AdvancedSQL {
+		if validator.IsValidColumn(colName) {
+			filteredAdvSQL[colName] = sqlExpr
+		} else {
+			logger.Warn("Invalid column in advanced SQL removed: %s", colName)
+		}
+	}
+	filtered.AdvancedSQL = filteredAdvSQL
+
+	// ComputedQL columns are allowed to be any name since they're computed
+	// No filtering needed for ComputedQL keys
+	filtered.ComputedQL = options.ComputedQL
+
+	// Filter Expand columns
+	filteredExpands := make([]ExpandOption, 0, len(options.Expand))
+	for _, expand := range options.Expand {
+		filteredExpand := expand
+		// Don't validate relation name, only columns
+		filteredExpand.Columns = validator.FilterValidColumns(expand.Columns)
+		filteredExpands = append(filteredExpands, filteredExpand)
+	}
+	filtered.Expand = filteredExpands
+
+	return filtered
+}

pkg/restheadspec/headers.go

@@ -4,6 +4,7 @@ import (
 	"encoding/base64"
 	"encoding/json"
 	"fmt"
+	"reflect"
 	"strconv"
 	"strings"
 
@@ -19,21 +20,21 @@ type ExtendedRequestOptions struct {
 	CleanJSON bool
 
 	// Advanced filtering
-	SearchColumns []string
+	SearchColumns  []string
 	CustomSQLWhere string
-	CustomSQLOr string
+	CustomSQLOr    string
 
 	// Joins
 	Expand []ExpandOption
 
 	// Advanced features
-	AdvancedSQL map[string]string // Column -> SQL expression
-	ComputedQL map[string]string  // Column -> CQL expression
-	Distinct bool
-	SkipCount bool
-	SkipCache bool
+	AdvancedSQL    map[string]string // Column -> SQL expression
+	ComputedQL     map[string]string // Column -> CQL expression
+	Distinct       bool
+	SkipCount      bool
+	SkipCache      bool
 	FetchRowNumber *string
-	PKRow *string
+	PKRow          *string
 
 	// Response format
 	ResponseFormat string // "simple", "detail", "syncfusion"
@@ -42,42 +43,58 @@ type ExtendedRequestOptions struct {
 	AtomicTransaction bool
 
 	// Cursor pagination
-	CursorForward string
+	CursorForward  string
 	CursorBackward string
 }
 
 // ExpandOption represents a relation expansion configuration
 type ExpandOption struct {
 	Relation string
-	Columns []string
-	Where string
-	Sort string
+	Columns  []string
+	Where    string
+	Sort     string
 }
 
 // decodeHeaderValue decodes base64 encoded header values
 // Supports ZIP_ and __ prefixes for base64 encoding
 func decodeHeaderValue(value string) string {
-	// Check for ZIP_ prefix
-	if strings.HasPrefix(value, "ZIP_") {
-		decoded, err := base64.StdEncoding.DecodeString(value[4:])
-		if err == nil {
-			return string(decoded)
+	str, _ := DecodeParam(value)
+	return str
+}
+
+// DecodeParam - Decodes parameter string and returns unencoded string
+func DecodeParam(pStr string) (string, error) {
+	var code string = pStr
+	if strings.HasPrefix(pStr, "ZIP_") {
+		code = strings.ReplaceAll(pStr, "ZIP_", "")
+		code = strings.ReplaceAll(code, "\n", "")
+		code = strings.ReplaceAll(code, "\r", "")
+		code = strings.ReplaceAll(code, " ", "")
+		strDat, err := base64.StdEncoding.DecodeString(code)
+		if err != nil {
+			return code, fmt.Errorf("failed to read parameter base64: %v", err)
+		} else {
+			code = string(strDat)
+		}
+	} else if strings.HasPrefix(pStr, "__") {
+		code = strings.ReplaceAll(pStr, "__", "")
+		code = strings.ReplaceAll(code, "\n", "")
+		code = strings.ReplaceAll(code, "\r", "")
+		code = strings.ReplaceAll(code, " ", "")
+
+		strDat, err := base64.StdEncoding.DecodeString(code)
+		if err != nil {
+			return code, fmt.Errorf("failed to read parameter base64: %v", err)
+		} else {
+			code = string(strDat)
 		}
-		logger.Warn("Failed to decode ZIP_ prefixed value: %v", err)
-		return value
 	}
 
-	// Check for __ prefix
-	if strings.HasPrefix(value, "__") {
-		decoded, err := base64.StdEncoding.DecodeString(value[2:])
-		if err == nil {
-			return string(decoded)
-		}
-		logger.Warn("Failed to decode __ prefixed value: %v", err)
-		return value
+	if strings.HasPrefix(code, "ZIP_") || strings.HasPrefix(code, "__") {
+		code, _ = DecodeParam(code)
 	}
 
-	return value
+	return code, nil
 }
 
 // parseOptionsFromHeaders parses all request options from HTTP headers
@@ -85,12 +102,13 @@ func (h *Handler) parseOptionsFromHeaders(r common.Request) ExtendedRequestOptio
 	options := ExtendedRequestOptions{
 		RequestOptions: common.RequestOptions{
 			Filters: make([]common.FilterOption, 0),
-			Sort: make([]common.SortOption, 0),
+			Sort:    make([]common.SortOption, 0),
 			Preload: make([]common.PreloadOption, 0),
 		},
-		AdvancedSQL: make(map[string]string),
-		ComputedQL: make(map[string]string),
-		Expand: make([]ExpandOption, 0),
+		AdvancedSQL:    make(map[string]string),
+		ComputedQL:     make(map[string]string),
+		Expand:         make([]ExpandOption, 0),
+		ResponseFormat: "simple", // Default response format
 	}
 
 	// Get all headers
@@ -198,6 +216,9 @@ func (h *Handler) parseSelectFields(options *ExtendedRequestOptions, value strin
 		return
 	}
 	options.Columns = h.parseCommaSeparated(value)
+	if len(options.Columns) > 1 {
+		options.CleanJSON = true
+	}
 }
 
 // parseNotSelectFields parses x-not-select-fields header
@@ -206,15 +227,19 @@ func (h *Handler) parseNotSelectFields(options *ExtendedRequestOptions, value st
 		return
 	}
 	options.OmitColumns = h.parseCommaSeparated(value)
+	if len(options.OmitColumns) > 1 {
+		options.CleanJSON = true
+	}
 }
 
 // parseFieldFilter parses x-fieldfilter-{colname} header (exact match)
 func (h *Handler) parseFieldFilter(options *ExtendedRequestOptions, headerKey, value string) {
 	colName := strings.TrimPrefix(headerKey, "x-fieldfilter-")
 	options.Filters = append(options.Filters, common.FilterOption{
-		Column: colName,
-		Operator: "eq",
-		Value: value,
+		Column:        colName,
+		Operator:      "eq",
+		Value:         value,
+		LogicOperator: "AND", // Default to AND
 	})
 }
 
@@ -223,9 +248,10 @@ func (h *Handler) parseSearchFilter(options *ExtendedRequestOptions, headerKey,
 	colName := strings.TrimPrefix(headerKey, "x-searchfilter-")
 	// Use ILIKE for fuzzy search
 	options.Filters = append(options.Filters, common.FilterOption{
-		Column: colName,
-		Operator: "ilike",
-		Value: "%" + value + "%",
+		Column:        colName,
+		Operator:      "ilike",
+		Value:         "%" + value + "%",
+		LogicOperator: "AND", // Default to AND
 	})
 }
 
@@ -254,70 +280,68 @@ func (h *Handler) parseSearchOp(options *ExtendedRequestOptions, headerKey, valu
 	colName := parts[1]
 
 	// Map operator names to filter operators
-	filterOp := h.mapSearchOperator(operator, value)
+	filterOp := h.mapSearchOperator(colName, operator, value)
+
+	// Set the logic operator (AND or OR)
+	filterOp.LogicOperator = logicOp
 
 	options.Filters = append(options.Filters, filterOp)
 
-	// Note: OR logic would need special handling in query builder
-	// For now, we'll add a comment to indicate OR logic
-	if logicOp == "OR" {
-		// TODO: Implement OR logic in query builder
-		logger.Debug("OR logic filter: %s %s %v", colName, filterOp.Operator, filterOp.Value)
-	}
+	logger.Debug("%s logic filter: %s %s %v", logicOp, colName, filterOp.Operator, filterOp.Value)
 }
 
 // mapSearchOperator maps search operator names to filter operators
-func (h *Handler) mapSearchOperator(operator, value string) common.FilterOption {
+func (h *Handler) mapSearchOperator(colName, operator, value string) common.FilterOption {
 	operator = strings.ToLower(operator)
 
 	switch operator {
-	case "contains":
-		return common.FilterOption{Operator: "ilike", Value: "%" + value + "%"}
+	case "contains", "contain", "like":
+		return common.FilterOption{Column: colName, Operator: "ilike", Value: "%" + value + "%"}
 	case "beginswith", "startswith":
-		return common.FilterOption{Operator: "ilike", Value: value + "%"}
+		return common.FilterOption{Column: colName, Operator: "ilike", Value: value + "%"}
 	case "endswith":
-		return common.FilterOption{Operator: "ilike", Value: "%" + value}
-	case "equals", "eq":
-		return common.FilterOption{Operator: "eq", Value: value}
-	case "notequals", "neq", "ne":
-		return common.FilterOption{Operator: "neq", Value: value}
-	case "greaterthan", "gt":
-		return common.FilterOption{Operator: "gt", Value: value}
-	case "lessthan", "lt":
-		return common.FilterOption{Operator: "lt", Value: value}
-	case "greaterthanorequal", "gte", "ge":
-		return common.FilterOption{Operator: "gte", Value: value}
-	case "lessthanorequal", "lte", "le":
-		return common.FilterOption{Operator: "lte", Value: value}
+		return common.FilterOption{Column: colName, Operator: "ilike", Value: "%" + value}
+	case "equals", "eq", "=":
+		return common.FilterOption{Column: colName, Operator: "eq", Value: value}
+	case "notequals", "neq", "ne", "!=", "<>":
+		return common.FilterOption{Column: colName, Operator: "neq", Value: value}
+	case "greaterthan", "gt", ">":
+		return common.FilterOption{Column: colName, Operator: "gt", Value: value}
+	case "lessthan", "lt", "<":
+		return common.FilterOption{Column: colName, Operator: "lt", Value: value}
+	case "greaterthanorequal", "gte", "ge", ">=":
+		return common.FilterOption{Column: colName, Operator: "gte", Value: value}
+	case "lessthanorequal", "lte", "le", "<=":
+		return common.FilterOption{Column: colName, Operator: "lte", Value: value}
 	case "between":
 		// Parse between values (format: "value1,value2")
 		// Between is exclusive (> value1 AND < value2)
 		parts := strings.Split(value, ",")
 		if len(parts) == 2 {
-			return common.FilterOption{Operator: "between", Value: parts}
+			return common.FilterOption{Column: colName, Operator: "between", Value: parts}
 		}
-		return common.FilterOption{Operator: "eq", Value: value}
+		return common.FilterOption{Column: colName, Operator: "eq", Value: value}
 	case "betweeninclusive":
 		// Parse between values (format: "value1,value2")
 		// Between inclusive is >= value1 AND <= value2
 		parts := strings.Split(value, ",")
 		if len(parts) == 2 {
-			return common.FilterOption{Operator: "between_inclusive", Value: parts}
+			return common.FilterOption{Column: colName, Operator: "between_inclusive", Value: parts}
 		}
-		return common.FilterOption{Operator: "eq", Value: value}
+		return common.FilterOption{Column: colName, Operator: "eq", Value: value}
 	case "in":
 		// Parse IN values (format: "value1,value2,value3")
 		values := strings.Split(value, ",")
-		return common.FilterOption{Operator: "in", Value: values}
+		return common.FilterOption{Column: colName, Operator: "in", Value: values}
 	case "empty", "isnull", "null":
 		// Check for NULL or empty string
-		return common.FilterOption{Operator: "is_null", Value: nil}
+		return common.FilterOption{Column: colName, Operator: "is_null", Value: nil}
 	case "notempty", "isnotnull", "notnull":
 		// Check for NOT NULL
-		return common.FilterOption{Operator: "is_not_null", Value: nil}
+		return common.FilterOption{Column: colName, Operator: "is_not_null", Value: nil}
 	default:
 		logger.Warn("Unknown search operator: %s, defaulting to equals", operator)
-		return common.FilterOption{Operator: "eq", Value: value}
+		return common.FilterOption{Column: colName, Operator: "eq", Value: value}
 	}
 }
 
@@ -404,10 +428,16 @@ func (h *Handler) parseSorting(options *ExtendedRequestOptions, value string) {
 		} else if strings.HasPrefix(field, "+") {
 			direction = "ASC"
 			colName = strings.TrimPrefix(field, "+")
+		} else if strings.HasSuffix(field, " desc") {
+			direction = "DESC"
+			colName = strings.TrimSuffix(field, "desc")
+		} else if strings.HasSuffix(field, " asc") {
+			direction = "ASC"
+			colName = strings.TrimSuffix(field, "asc")
 		}
 
 		options.Sort = append(options.Sort, common.SortOption{
-			Column: colName,
+			Column:    strings.Trim(colName, " "),
 			Direction: direction,
 		})
 	}
@@ -439,3 +469,235 @@ func (h *Handler) parseJSONHeader(value string) (map[string]interface{}, error)
 	}
 	return result, nil
 }
+
+// getColumnTypeFromModel uses reflection to determine the Go type of a column in a model
+func (h *Handler) getColumnTypeFromModel(model interface{}, colName string) reflect.Kind {
+	if model == nil {
+		return reflect.Invalid
+	}
+
+	modelType := reflect.TypeOf(model)
+	// Dereference pointer if needed
+	if modelType.Kind() == reflect.Ptr {
+		modelType = modelType.Elem()
+	}
+
+	// Ensure it's a struct
+	if modelType.Kind() != reflect.Struct {
+		return reflect.Invalid
+	}
+
+	// Find the field by JSON tag or field name
+	for i := 0; i < modelType.NumField(); i++ {
+		field := modelType.Field(i)
+
+		// Check JSON tag
+		jsonTag := field.Tag.Get("json")
+		if jsonTag != "" {
+			// Parse JSON tag (format: "name,omitempty")
+			parts := strings.Split(jsonTag, ",")
+			if parts[0] == colName {
+				return field.Type.Kind()
+			}
+		}
+
+		// Check field name (case-insensitive)
+		if strings.EqualFold(field.Name, colName) {
+			return field.Type.Kind()
+		}
+
+		// Check snake_case conversion
+		snakeCaseName := toSnakeCase(field.Name)
+		if snakeCaseName == colName {
+			return field.Type.Kind()
+		}
+	}
+
+	return reflect.Invalid
+}
+
+// toSnakeCase converts a string from CamelCase to snake_case
+func toSnakeCase(s string) string {
+	var result strings.Builder
+	for i, r := range s {
+		if i > 0 && r >= 'A' && r <= 'Z' {
+			result.WriteRune('_')
+		}
+		result.WriteRune(r)
+	}
+	return strings.ToLower(result.String())
+}
+
+// isNumericType checks if a reflect.Kind is a numeric type
+func isNumericType(kind reflect.Kind) bool {
+	return kind == reflect.Int || kind == reflect.Int8 || kind == reflect.Int16 ||
+		kind == reflect.Int32 || kind == reflect.Int64 || kind == reflect.Uint ||
+		kind == reflect.Uint8 || kind == reflect.Uint16 || kind == reflect.Uint32 ||
+		kind == reflect.Uint64 || kind == reflect.Float32 || kind == reflect.Float64
+}
+
+// isStringType checks if a reflect.Kind is a string type
+func isStringType(kind reflect.Kind) bool {
+	return kind == reflect.String
+}
+
+// isBoolType checks if a reflect.Kind is a boolean type
+func isBoolType(kind reflect.Kind) bool {
+	return kind == reflect.Bool
+}
+
+// convertToNumericType converts a string value to the appropriate numeric type
+func convertToNumericType(value string, kind reflect.Kind) (interface{}, error) {
+	value = strings.TrimSpace(value)
+
+	switch kind {
+	case reflect.Int, reflect.Int8, reflect.Int16, reflect.Int32, reflect.Int64:
+		// Parse as integer
+		bitSize := 64
+		switch kind {
+		case reflect.Int8:
+			bitSize = 8
+		case reflect.Int16:
+			bitSize = 16
+		case reflect.Int32:
+			bitSize = 32
+		}
+
+		intVal, err := strconv.ParseInt(value, 10, bitSize)
+		if err != nil {
+			return nil, fmt.Errorf("invalid integer value: %w", err)
+		}
+
+		// Return the appropriate type
+		switch kind {
+		case reflect.Int:
+			return int(intVal), nil
+		case reflect.Int8:
+			return int8(intVal), nil
+		case reflect.Int16:
+			return int16(intVal), nil
+		case reflect.Int32:
+			return int32(intVal), nil
+		case reflect.Int64:
+			return intVal, nil
+		}
+
+	case reflect.Uint, reflect.Uint8, reflect.Uint16, reflect.Uint32, reflect.Uint64:
+		// Parse as unsigned integer
+		bitSize := 64
+		switch kind {
+		case reflect.Uint8:
+			bitSize = 8
+		case reflect.Uint16:
+			bitSize = 16
+		case reflect.Uint32:
+			bitSize = 32
+		}
+
+		uintVal, err := strconv.ParseUint(value, 10, bitSize)
+		if err != nil {
+			return nil, fmt.Errorf("invalid unsigned integer value: %w", err)
+		}
+
+		// Return the appropriate type
+		switch kind {
+		case reflect.Uint:
+			return uint(uintVal), nil
+		case reflect.Uint8:
+			return uint8(uintVal), nil
+		case reflect.Uint16:
+			return uint16(uintVal), nil
+		case reflect.Uint32:
+			return uint32(uintVal), nil
+		case reflect.Uint64:
+			return uintVal, nil
+		}
+
+	case reflect.Float32, reflect.Float64:
+		// Parse as float
+		bitSize := 64
+		if kind == reflect.Float32 {
+			bitSize = 32
+		}
+
+		floatVal, err := strconv.ParseFloat(value, bitSize)
+		if err != nil {
+			return nil, fmt.Errorf("invalid float value: %w", err)
+		}
+
+		if kind == reflect.Float32 {
+			return float32(floatVal), nil
+		}
+		return floatVal, nil
+	}
+
+	return nil, fmt.Errorf("unsupported numeric type: %v", kind)
+}
+
+// isNumericValue checks if a string value can be parsed as a number
+func isNumericValue(value string) bool {
+	value = strings.TrimSpace(value)
+	_, err := strconv.ParseFloat(value, 64)
+	return err == nil
+}
+
+// ColumnCastInfo holds information about whether a column needs casting
+type ColumnCastInfo struct {
+	NeedsCast     bool
+	IsNumericType bool
+}
+
+// ValidateAndAdjustFilterForColumnType validates and adjusts a filter based on column type
+// Returns ColumnCastInfo indicating whether the column should be cast to text in SQL
+func (h *Handler) ValidateAndAdjustFilterForColumnType(filter *common.FilterOption, model interface{}) ColumnCastInfo {
+	if filter == nil || model == nil {
+		return ColumnCastInfo{NeedsCast: false, IsNumericType: false}
+	}
+
+	colType := h.getColumnTypeFromModel(model, filter.Column)
+	if colType == reflect.Invalid {
+		// Column not found in model, no casting needed
+		logger.Debug("Column %s not found in model, skipping type validation", filter.Column)
+		return ColumnCastInfo{NeedsCast: false, IsNumericType: false}
+	}
+
+	// Check if the input value is numeric
+	valueIsNumeric := false
+	if strVal, ok := filter.Value.(string); ok {
+		strVal = strings.Trim(strVal, "%")
+		valueIsNumeric = isNumericValue(strVal)
+	}
+
+	// Adjust based on column type
+	switch {
+	case isNumericType(colType):
+		// Column is numeric
+		if valueIsNumeric {
+			// Value is numeric - try to convert it
+			if strVal, ok := filter.Value.(string); ok {
+				strVal = strings.Trim(strVal, "%")
+				numericVal, err := convertToNumericType(strVal, colType)
+				if err != nil {
+					logger.Debug("Failed to convert value '%s' to numeric type for column %s, will use text cast", strVal, filter.Column)
+					return ColumnCastInfo{NeedsCast: true, IsNumericType: true}
+				}
+				filter.Value = numericVal
+			}
+			// No cast needed - numeric column with numeric value
+			return ColumnCastInfo{NeedsCast: false, IsNumericType: true}
+		} else {
+			// Value is not numeric - cast column to text for comparison
+			logger.Debug("Non-numeric value for numeric column %s, will cast to text", filter.Column)
+			return ColumnCastInfo{NeedsCast: true, IsNumericType: true}
+		}
+
+	case isStringType(colType):
+		// String columns don't need casting
+		return ColumnCastInfo{NeedsCast: false, IsNumericType: false}
+
+	default:
+		// For bool, time.Time, and other complex types - cast to text
+		logger.Debug("Complex type column %s, will cast to text", filter.Column)
+		return ColumnCastInfo{NeedsCast: true, IsNumericType: false}
+	}
+}

pkg/restheadspec/restheadspec.go

@@ -1,3 +1,55 @@
+// Package restheadspec provides the Rest Header Spec API framework.
+//
+// Rest Header Spec (restheadspec) is a RESTful API framework that reads query options,
+// filters, sorting, pagination, and other parameters from HTTP headers instead of
+// request bodies or query parameters. This approach provides a clean separation between
+// data and metadata in API requests.
+//
+// # Key Features
+//
+//   - Header-based API configuration: All query options are passed via HTTP headers
+//   - Database-agnostic: Works with both GORM and Bun ORM through adapters
+//   - Router-agnostic: Supports multiple HTTP routers (Mux, BunRouter, etc.)
+//   - Advanced filtering: Supports complex filter operations (eq, gt, lt, like, between, etc.)
+//   - Pagination and sorting: Built-in support for limit, offset, and multi-column sorting
+//   - Preloading and expansion: Support for eager loading relationships
+//   - Multiple response formats: Default, simple, and Syncfusion formats
+//
+// # HTTP Headers
+//
+// The following headers are supported for configuring API requests:
+//
+//   - X-Filters: JSON array of filter conditions
+//   - X-Columns: Comma-separated list of columns to select
+//   - X-Sort: JSON array of sort specifications
+//   - X-Limit: Maximum number of records to return
+//   - X-Offset: Number of records to skip
+//   - X-Preload: Comma-separated list of relations to preload
+//   - X-Expand: Comma-separated list of relations to expand (LEFT JOIN)
+//   - X-Distinct: Boolean to enable DISTINCT queries
+//   - X-Skip-Count: Boolean to skip total count query
+//   - X-Response-Format: Response format (detail, simple, syncfusion)
+//   - X-Clean-JSON: Boolean to remove null/empty fields
+//   - X-Custom-SQL-Where: Custom SQL WHERE clause (AND)
+//   - X-Custom-SQL-Or: Custom SQL WHERE clause (OR)
+//
+// # Usage Example
+//
+//	// Create a handler with GORM
+//	handler := restheadspec.NewHandlerWithGORM(db)
+//
+//	// Register models
+//	handler.Registry.RegisterModel("users", User{})
+//
+//	// Setup routes with Mux
+//	muxRouter := mux.NewRouter()
+//	restheadspec.SetupMuxRoutes(muxRouter, handler)
+//
+//	// Make a request with headers
+//	// GET /public/users
+//	// X-Filters: [{"column":"age","operator":"gt","value":18}]
+//	// X-Sort: [{"column":"name","direction":"asc"}]
+//	// X-Limit: 10
 package restheadspec
 
 import (

pkg/testmodels/business.go

@@ -140,22 +140,22 @@ func (Comment) TableName() string {
 
 // RegisterTestModels registers all test models with the provided registry
 func RegisterTestModels(registry *modelregistry.DefaultModelRegistry) {
-	registry.RegisterModel("departments", &Department{})
-	registry.RegisterModel("employees", &Employee{})
-	registry.RegisterModel("projects", &Project{})
-	registry.RegisterModel("project_tasks", &ProjectTask{})
-	registry.RegisterModel("documents", &Document{})
-	registry.RegisterModel("comments", &Comment{})
+	registry.RegisterModel("departments", Department{})
+	registry.RegisterModel("employees", Employee{})
+	registry.RegisterModel("projects", Project{})
+	registry.RegisterModel("project_tasks", ProjectTask{})
+	registry.RegisterModel("documents", Document{})
+	registry.RegisterModel("comments", Comment{})
 }
 
 // GetTestModels returns a list of all test model instances
 func GetTestModels() []interface{} {
 	return []interface{}{
-		&Department{},
-		&Employee{},
-		&Project{},
-		&ProjectTask{},
-		&Document{},
-		&Comment{},
+		Department{},
+		Employee{},
+		Project{},
+		ProjectTask{},
+		Document{},
+		Comment{},
 	}
 }

todo.md

@@ -0,0 +1,152 @@
+# ResolveSpec - TODO List
+
+This document tracks incomplete features and improvements for the ResolveSpec project.
+
+## Core Features to Implement
+
+### 1. Column Selection and Filtering for Preloads
+**Location:** `pkg/resolvespec/handler.go:730`
+**Status:** Not Implemented
+**Description:** Currently, preloads are applied without any column selection or filtering. This feature would allow clients to:
+- Select specific columns for preloaded relationships
+- Apply filters to preloaded data
+- Reduce payload size and improve performance
+
+**Current Limitation:**
+```go
+// For now, we'll preload without conditions
+// TODO: Implement column selection and filtering for preloads
+// This requires a more sophisticated approach with callbacks or query builders
+query = query.Preload(relationFieldName)
+```
+
+**Required Implementation:**
+- Add support for column selection in preloaded relationships
+- Implement filtering conditions for preloaded data
+- Design a callback or query builder approach that works across different ORMs
+
+---
+
+### 2. Recursive JSON Cleaning
+**Location:** `pkg/restheadspec/handler.go:796`
+**Status:** Partially Implemented (Simplified)
+**Description:** The current `cleanJSON` function returns data as-is without recursively removing null and empty fields from nested structures.
+
+**Current Limitation:**
+```go
+// This is a simplified implementation
+// A full implementation would recursively clean nested structures
+// For now, we'll return the data as-is
+// TODO: Implement recursive cleaning
+return data
+```
+
+**Required Implementation:**
+- Recursively traverse nested structures (maps, slices, structs)
+- Remove null values
+- Remove empty objects and arrays
+- Handle edge cases (circular references, pointers, etc.)
+
+---
+
+### 3. Custom SQL Join Support
+**Location:** `pkg/restheadspec/headers.go:159`
+**Status:** Not Implemented
+**Description:** Support for custom SQL joins via the `X-Custom-SQL-Join` header is currently logged but not executed.
+
+**Current Limitation:**
+```go
+case strings.HasPrefix(normalizedKey, "x-custom-sql-join"):
+    // TODO: Implement custom SQL join
+    logger.Debug("Custom SQL join not yet implemented: %s", decodedValue)
+```
+
+**Required Implementation:**
+- Parse custom SQL join expressions from headers
+- Apply joins to the query builder
+- Ensure security (SQL injection prevention)
+- Support for different join types (INNER, LEFT, RIGHT, FULL)
+- Works across different database adapters (GORM, Bun)
+
+---
+
+### 4. Proper Condition Handling for Bun Preloads
+**Location:** `pkg/common/adapters/database/bun.go:202`
+**Status:** Partially Implemented
+**Description:** The Bun adapter's `Preload` method currently ignores conditions passed to it.
+
+**Current Limitation:**
+```go
+func (b *BunSelectQuery) Preload(relation string, conditions ...interface{}) common.SelectQuery {
+    // Bun uses Relation() method for preloading
+    // For now, we'll just pass the relation name without conditions
+    // TODO: Implement proper condition handling for Bun
+    b.query = b.query.Relation(relation)
+    return b
+}
+```
+
+**Required Implementation:**
+- Properly handle condition parameters in Bun's Relation() method
+- Support filtering on preloaded relationships
+- Ensure compatibility with GORM's condition syntax where possible
+- Test with various condition types
+
+---
+
+## Code Quality Improvements
+
+### 5. Modernize Go Type Declarations
+**Location:** `pkg/common/types.go:5, 42, 64, 79`
+**Status:** Pending
+**Priority:** Low
+**Description:** Replace legacy `interface{}` with modern `any` type alias (Go 1.18+).
+
+**Affected Lines:**
+- Line 5: Function parameter or return type
+- Line 42: Function parameter or return type
+- Line 64: Function parameter or return type
+- Line 79: Function parameter or return type
+
+**Benefits:**
+- More modern and idiomatic Go code
+- Better readability
+- Aligns with current Go best practices
+
+---
+
+## Additional Considerations
+
+### Documentation
+- Ensure all new features are documented in README.md
+- Update examples to showcase new functionality
+- Add migration notes if any breaking changes are introduced
+
+### Testing
+- Add unit tests for each new feature
+- Add integration tests for database adapter compatibility
+- Ensure backward compatibility is maintained
+
+### Performance
+- Profile preload performance with column selection and filtering
+- Optimize recursive JSON cleaning for large payloads
+- Benchmark custom SQL join performance
+
+---
+
+## Priority Ranking
+
+1. **High Priority**
+   - Column Selection and Filtering for Preloads (#1)
+   - Proper Condition Handling for Bun Preloads (#4)
+
+2. **Medium Priority**
+   - Custom SQL Join Support (#3)
+   - Recursive JSON Cleaning (#2)
+
+3. **Low Priority**
+   - Modernize Go Type Declarations (#5)
+
+---
+
+**Last Updated:** 2025-11-07