sql_cli/data/
datatable.rs

1use crate::api_client::QueryResponse;
2use crate::data::data_provider::DataProvider;
3use crate::data::type_inference::{InferredType, TypeInference};
4use serde::{Deserialize, Serialize};
5use serde_json::Value as JsonValue;
6use std::collections::HashMap;
7use std::fmt;
8use std::sync::Arc;
9use tracing::debug;
10
11/// Represents the data type of a column
12#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
13pub enum DataType {
14    String,
15    Integer,
16    Float,
17    Boolean,
18    DateTime,
19    Null,
20    Mixed, // For columns with mixed types
21}
22
23impl DataType {
24    /// Infer type from a string value
25    #[must_use]
26    pub fn infer_from_string(value: &str) -> Self {
27        // Handle explicit null string
28        if value.eq_ignore_ascii_case("null") {
29            return DataType::Null;
30        }
31
32        // Use the shared type inference logic
33        match TypeInference::infer_from_string(value) {
34            InferredType::Null => DataType::Null,
35            InferredType::Boolean => DataType::Boolean,
36            InferredType::Integer => DataType::Integer,
37            InferredType::Float => DataType::Float,
38            InferredType::DateTime => DataType::DateTime,
39            InferredType::String => DataType::String,
40        }
41    }
42
43    /// Check if a string looks like a datetime value
44    /// Delegates to shared type inference logic
45    fn looks_like_datetime(value: &str) -> bool {
46        TypeInference::looks_like_datetime(value)
47    }
48
49    /// Merge two types (for columns with mixed types)
50    #[must_use]
51    pub fn merge(&self, other: &DataType) -> DataType {
52        if self == other {
53            return self.clone();
54        }
55
56        match (self, other) {
57            (DataType::Null, t) | (t, DataType::Null) => t.clone(),
58            (DataType::Integer, DataType::Float) | (DataType::Float, DataType::Integer) => {
59                DataType::Float
60            }
61            _ => DataType::Mixed,
62        }
63    }
64}
65
66/// Column metadata and definition
67#[derive(Debug, Clone, Serialize, Deserialize)]
68pub struct DataColumn {
69    pub name: String,
70    pub data_type: DataType,
71    pub nullable: bool,
72    pub unique_values: Option<usize>,
73    pub null_count: usize,
74    pub metadata: HashMap<String, String>,
75}
76
77impl DataColumn {
78    pub fn new(name: impl Into<String>) -> Self {
79        Self {
80            name: name.into(),
81            data_type: DataType::String,
82            nullable: true,
83            unique_values: None,
84            null_count: 0,
85            metadata: HashMap::new(),
86        }
87    }
88
89    #[must_use]
90    pub fn with_type(mut self, data_type: DataType) -> Self {
91        self.data_type = data_type;
92        self
93    }
94
95    #[must_use]
96    pub fn with_nullable(mut self, nullable: bool) -> Self {
97        self.nullable = nullable;
98        self
99    }
100}
101
102/// A single cell value in the table
103#[derive(Debug, Clone, PartialEq, PartialOrd)]
104pub enum DataValue {
105    String(String),
106    InternedString(Arc<String>), // For repeated strings (e.g., status, trader names)
107    Integer(i64),
108    Float(f64),
109    Boolean(bool),
110    DateTime(String), // Store as ISO 8601 string for now
111    Null,
112}
113
114// Custom Hash implementation for DataValue to handle f64
115impl std::hash::Hash for DataValue {
116    fn hash<H: std::hash::Hasher>(&self, state: &mut H) {
117        match self {
118            DataValue::String(s) => {
119                0u8.hash(state);
120                s.hash(state);
121            }
122            DataValue::InternedString(s) => {
123                1u8.hash(state);
124                s.hash(state);
125            }
126            DataValue::Integer(i) => {
127                2u8.hash(state);
128                i.hash(state);
129            }
130            DataValue::Float(f) => {
131                3u8.hash(state);
132                // Hash the bits of the float for consistency
133                f.to_bits().hash(state);
134            }
135            DataValue::Boolean(b) => {
136                4u8.hash(state);
137                b.hash(state);
138            }
139            DataValue::DateTime(dt) => {
140                5u8.hash(state);
141                dt.hash(state);
142            }
143            DataValue::Null => {
144                6u8.hash(state);
145            }
146        }
147    }
148}
149
150// Custom Eq implementation for DataValue
151impl Eq for DataValue {}
152
153impl DataValue {
154    pub fn from_string(s: &str, data_type: &DataType) -> Self {
155        if s.is_empty() || s.eq_ignore_ascii_case("null") {
156            return DataValue::Null;
157        }
158
159        match data_type {
160            DataType::String => DataValue::String(s.to_string()),
161            DataType::Integer => s
162                .parse::<i64>()
163                .map_or_else(|_| DataValue::String(s.to_string()), DataValue::Integer),
164            DataType::Float => s
165                .parse::<f64>()
166                .map_or_else(|_| DataValue::String(s.to_string()), DataValue::Float),
167            DataType::Boolean => {
168                let lower = s.to_lowercase();
169                DataValue::Boolean(lower == "true" || lower == "1" || lower == "yes")
170            }
171            DataType::DateTime => DataValue::DateTime(s.to_string()),
172            DataType::Null => DataValue::Null,
173            DataType::Mixed => {
174                // Try to infer for mixed columns
175                let inferred = DataType::infer_from_string(s);
176                Self::from_string(s, &inferred)
177            }
178        }
179    }
180
181    #[must_use]
182    pub fn is_null(&self) -> bool {
183        matches!(self, DataValue::Null)
184    }
185
186    #[must_use]
187    pub fn data_type(&self) -> DataType {
188        match self {
189            DataValue::String(_) | DataValue::InternedString(_) => DataType::String,
190            DataValue::Integer(_) => DataType::Integer,
191            DataValue::Float(_) => DataType::Float,
192            DataValue::Boolean(_) => DataType::Boolean,
193            DataValue::DateTime(_) => DataType::DateTime,
194            DataValue::Null => DataType::Null,
195        }
196    }
197
198    /// Get string representation without allocation when possible
199    /// Returns owned String for compatibility but tries to reuse existing strings
200    #[must_use]
201    pub fn to_string_optimized(&self) -> String {
202        match self {
203            DataValue::String(s) => s.clone(), // Clone existing string
204            DataValue::InternedString(s) => s.as_ref().clone(), // Clone from Rc
205            DataValue::DateTime(s) => s.clone(), // Clone existing string
206            DataValue::Integer(i) => i.to_string(),
207            DataValue::Float(f) => f.to_string(),
208            DataValue::Boolean(b) => {
209                if *b {
210                    "true".to_string()
211                } else {
212                    "false".to_string()
213                }
214            }
215            DataValue::Null => String::new(), // Empty string, minimal allocation
216        }
217    }
218}
219
220impl fmt::Display for DataValue {
221    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
222        match self {
223            DataValue::String(s) => write!(f, "{s}"),
224            DataValue::InternedString(s) => write!(f, "{s}"),
225            DataValue::Integer(i) => write!(f, "{i}"),
226            DataValue::Float(fl) => write!(f, "{fl}"),
227            DataValue::Boolean(b) => write!(f, "{b}"),
228            DataValue::DateTime(dt) => write!(f, "{dt}"),
229            DataValue::Null => write!(f, ""),
230        }
231    }
232}
233
234/// A row of data in the table
235#[derive(Debug, Clone)]
236pub struct DataRow {
237    pub values: Vec<DataValue>,
238}
239
240impl DataRow {
241    #[must_use]
242    pub fn new(values: Vec<DataValue>) -> Self {
243        Self { values }
244    }
245
246    #[must_use]
247    pub fn get(&self, index: usize) -> Option<&DataValue> {
248        self.values.get(index)
249    }
250
251    pub fn get_mut(&mut self, index: usize) -> Option<&mut DataValue> {
252        self.values.get_mut(index)
253    }
254
255    #[must_use]
256    pub fn len(&self) -> usize {
257        self.values.len()
258    }
259
260    #[must_use]
261    pub fn is_empty(&self) -> bool {
262        self.values.is_empty()
263    }
264}
265
266/// The main `DataTable` structure
267#[derive(Debug, Clone)]
268pub struct DataTable {
269    pub name: String,
270    pub columns: Vec<DataColumn>,
271    pub rows: Vec<DataRow>,
272    pub metadata: HashMap<String, String>,
273}
274
275impl DataTable {
276    pub fn new(name: impl Into<String>) -> Self {
277        Self {
278            name: name.into(),
279            columns: Vec::new(),
280            rows: Vec::new(),
281            metadata: HashMap::new(),
282        }
283    }
284
285    /// Create a DUAL table (similar to Oracle's DUAL) with one row and one column
286    /// Used for evaluating expressions without a data source
287    #[must_use]
288    pub fn dual() -> Self {
289        let mut table = DataTable::new("DUAL");
290        table.add_column(DataColumn::new("DUMMY").with_type(DataType::String));
291        table
292            .add_row(DataRow::new(vec![DataValue::String("X".to_string())]))
293            .unwrap();
294        table
295    }
296
297    pub fn add_column(&mut self, column: DataColumn) -> &mut Self {
298        self.columns.push(column);
299        self
300    }
301
302    pub fn add_row(&mut self, row: DataRow) -> Result<(), String> {
303        if row.len() != self.columns.len() {
304            return Err(format!(
305                "Row has {} values but table has {} columns",
306                row.len(),
307                self.columns.len()
308            ));
309        }
310        self.rows.push(row);
311        Ok(())
312    }
313
314    #[must_use]
315    pub fn get_column(&self, name: &str) -> Option<&DataColumn> {
316        self.columns.iter().find(|c| c.name == name)
317    }
318
319    #[must_use]
320    pub fn get_column_index(&self, name: &str) -> Option<usize> {
321        self.columns.iter().position(|c| c.name == name)
322    }
323
324    #[must_use]
325    pub fn column_count(&self) -> usize {
326        self.columns.len()
327    }
328
329    #[must_use]
330    pub fn row_count(&self) -> usize {
331        self.rows.len()
332    }
333
334    #[must_use]
335    pub fn is_empty(&self) -> bool {
336        self.rows.is_empty()
337    }
338
339    /// Get column names as a vector
340    #[must_use]
341    pub fn column_names(&self) -> Vec<String> {
342        self.columns.iter().map(|c| c.name.clone()).collect()
343    }
344
345    /// Infer and update column types based on data
346    pub fn infer_column_types(&mut self) {
347        for (col_idx, column) in self.columns.iter_mut().enumerate() {
348            let mut inferred_type = DataType::Null;
349            let mut null_count = 0;
350            let mut unique_values = std::collections::HashSet::new();
351
352            for row in &self.rows {
353                if let Some(value) = row.get(col_idx) {
354                    if value.is_null() {
355                        null_count += 1;
356                    } else {
357                        let value_type = value.data_type();
358                        inferred_type = inferred_type.merge(&value_type);
359                        unique_values.insert(value.to_string());
360                    }
361                }
362            }
363
364            column.data_type = inferred_type;
365            column.null_count = null_count;
366            column.nullable = null_count > 0;
367            column.unique_values = Some(unique_values.len());
368        }
369    }
370
371    /// Get a value at specific row and column
372    #[must_use]
373    pub fn get_value(&self, row: usize, col: usize) -> Option<&DataValue> {
374        self.rows.get(row)?.get(col)
375    }
376
377    /// Get a value by row index and column name
378    #[must_use]
379    pub fn get_value_by_name(&self, row: usize, col_name: &str) -> Option<&DataValue> {
380        let col_idx = self.get_column_index(col_name)?;
381        self.get_value(row, col_idx)
382    }
383
384    /// Convert to a vector of string vectors (for display/compatibility)
385    #[must_use]
386    pub fn to_string_table(&self) -> Vec<Vec<String>> {
387        self.rows
388            .iter()
389            .map(|row| {
390                row.values
391                    .iter()
392                    .map(DataValue::to_string_optimized)
393                    .collect()
394            })
395            .collect()
396    }
397
398    /// Get table statistics
399    #[must_use]
400    pub fn get_stats(&self) -> DataTableStats {
401        DataTableStats {
402            row_count: self.row_count(),
403            column_count: self.column_count(),
404            memory_size: self.estimate_memory_size(),
405            null_count: self.columns.iter().map(|c| c.null_count).sum(),
406        }
407    }
408
409    /// Generate a debug dump string for display
410    #[must_use]
411    pub fn debug_dump(&self) -> String {
412        let mut output = String::new();
413
414        output.push_str(&format!("DataTable: {}\n", self.name));
415        output.push_str(&format!(
416            "Rows: {} | Columns: {}\n",
417            self.row_count(),
418            self.column_count()
419        ));
420
421        if !self.metadata.is_empty() {
422            output.push_str("Metadata:\n");
423            for (key, value) in &self.metadata {
424                output.push_str(&format!("  {key}: {value}\n"));
425            }
426        }
427
428        output.push_str("\nColumns:\n");
429        for column in &self.columns {
430            output.push_str(&format!("  {} ({:?})", column.name, column.data_type));
431            if column.nullable {
432                output.push_str(&format!(" - nullable, {} nulls", column.null_count));
433            }
434            if let Some(unique) = column.unique_values {
435                output.push_str(&format!(", {unique} unique"));
436            }
437            output.push('\n');
438        }
439
440        // Show first few rows
441        if self.row_count() > 0 {
442            let sample_size = 5.min(self.row_count());
443            output.push_str(&format!("\nFirst {sample_size} rows:\n"));
444
445            for row_idx in 0..sample_size {
446                output.push_str(&format!("  [{row_idx}]: "));
447                for (col_idx, value) in self.rows[row_idx].values.iter().enumerate() {
448                    if col_idx > 0 {
449                        output.push_str(", ");
450                    }
451                    output.push_str(&value.to_string());
452                }
453                output.push('\n');
454            }
455        }
456
457        output
458    }
459
460    #[must_use]
461    pub fn estimate_memory_size(&self) -> usize {
462        // Base structure size
463        let mut size = std::mem::size_of::<Self>();
464
465        // Column metadata
466        size += self.columns.len() * std::mem::size_of::<DataColumn>();
467        for col in &self.columns {
468            size += col.name.len();
469        }
470
471        // Row structure overhead
472        size += self.rows.len() * std::mem::size_of::<DataRow>();
473
474        // Actual data values
475        for row in &self.rows {
476            for value in &row.values {
477                // Base enum size
478                size += std::mem::size_of::<DataValue>();
479                // Add string content size
480                match value {
481                    DataValue::String(s) | DataValue::DateTime(s) => size += s.len(),
482                    _ => {} // Numbers and booleans are inline
483                }
484            }
485        }
486
487        size
488    }
489
490    /// V46: Create `DataTable` from `QueryResponse`
491    /// This is the key conversion function that bridges old and new systems
492    pub fn from_query_response(response: &QueryResponse, table_name: &str) -> Result<Self, String> {
493        debug!(
494            "V46: Converting QueryResponse to DataTable for table '{}'",
495            table_name
496        );
497
498        // Track memory before conversion
499        crate::utils::memory_tracker::track_memory("start_from_query_response");
500
501        let mut table = DataTable::new(table_name);
502
503        // Extract column names and types from first row
504        if let Some(first_row) = response.data.first() {
505            if let Some(obj) = first_row.as_object() {
506                // Create columns based on the keys in the JSON object
507                for key in obj.keys() {
508                    let column = DataColumn::new(key.clone());
509                    table.add_column(column);
510                }
511
512                // Now convert all rows
513                for json_row in &response.data {
514                    if let Some(row_obj) = json_row.as_object() {
515                        let mut values = Vec::new();
516
517                        // Ensure we get values in the same order as columns
518                        for column in &table.columns {
519                            let value = row_obj
520                                .get(&column.name)
521                                .map_or(DataValue::Null, json_value_to_data_value);
522                            values.push(value);
523                        }
524
525                        table.add_row(DataRow::new(values))?;
526                    }
527                }
528
529                // Infer column types from the data
530                table.infer_column_types();
531
532                // Add metadata
533                if let Some(source) = &response.source {
534                    table.metadata.insert("source".to_string(), source.clone());
535                }
536                if let Some(cached) = response.cached {
537                    table
538                        .metadata
539                        .insert("cached".to_string(), cached.to_string());
540                }
541                table
542                    .metadata
543                    .insert("original_count".to_string(), response.count.to_string());
544
545                debug!(
546                    "V46: Created DataTable with {} columns and {} rows",
547                    table.column_count(),
548                    table.row_count()
549                );
550            } else {
551                // Handle non-object JSON (single values)
552                table.add_column(DataColumn::new("value"));
553                for json_value in &response.data {
554                    let value = json_value_to_data_value(json_value);
555                    table.add_row(DataRow::new(vec![value]))?;
556                }
557            }
558        }
559
560        Ok(table)
561    }
562
563    /// Get a single row by index
564    #[must_use]
565    pub fn get_row(&self, index: usize) -> Option<&DataRow> {
566        self.rows.get(index)
567    }
568
569    /// V50: Get a single row as strings
570    #[must_use]
571    pub fn get_row_as_strings(&self, index: usize) -> Option<Vec<String>> {
572        self.rows.get(index).map(|row| {
573            row.values
574                .iter()
575                .map(DataValue::to_string_optimized)
576                .collect()
577        })
578    }
579
580    /// Pretty print the `DataTable` with a nice box drawing
581    #[must_use]
582    pub fn pretty_print(&self) -> String {
583        let mut output = String::new();
584
585        // Header
586        output.push_str("╔═══════════════════════════════════════════════════════╗\n");
587        output.push_str(&format!("║ DataTable: {:^41} ║\n", self.name));
588        output.push_str("╠═══════════════════════════════════════════════════════╣\n");
589
590        // Summary stats
591        output.push_str(&format!(
592            "║ Rows: {:6} | Columns: {:3} | Memory: ~{:6} bytes ║\n",
593            self.row_count(),
594            self.column_count(),
595            self.get_stats().memory_size
596        ));
597
598        // Metadata if any
599        if !self.metadata.is_empty() {
600            output.push_str("╠═══════════════════════════════════════════════════════╣\n");
601            output.push_str("║ Metadata:                                             ║\n");
602            for (key, value) in &self.metadata {
603                let truncated_value = if value.len() > 35 {
604                    format!("{}...", &value[..32])
605                } else {
606                    value.clone()
607                };
608                output.push_str(&format!(
609                    "║   {:15} : {:35} ║\n",
610                    Self::truncate_string(key, 15),
611                    truncated_value
612                ));
613            }
614        }
615
616        // Column details
617        output.push_str("╠═══════════════════════════════════════════════════════╣\n");
618        output.push_str("║ Columns:                                              ║\n");
619        output.push_str("╟───────────────────┬──────────┬─────────┬──────┬──────╢\n");
620        output.push_str("║ Name              │ Type     │ Nullable│ Nulls│Unique║\n");
621        output.push_str("╟───────────────────┼──────────┼─────────┼──────┼──────╢\n");
622
623        for column in &self.columns {
624            let type_str = match &column.data_type {
625                DataType::String => "String",
626                DataType::Integer => "Integer",
627                DataType::Float => "Float",
628                DataType::Boolean => "Boolean",
629                DataType::DateTime => "DateTime",
630                DataType::Null => "Null",
631                DataType::Mixed => "Mixed",
632            };
633
634            output.push_str(&format!(
635                "║ {:17} │ {:8} │ {:7} │ {:4} │ {:4} ║\n",
636                Self::truncate_string(&column.name, 17),
637                type_str,
638                if column.nullable { "Yes" } else { "No" },
639                column.null_count,
640                column.unique_values.unwrap_or(0)
641            ));
642        }
643
644        output.push_str("╚═══════════════════════════════════════════════════════╝\n");
645
646        // Sample data (first 5 rows)
647        output.push_str("\nSample Data (first 5 rows):\n");
648        let sample_count = self.rows.len().min(5);
649
650        if sample_count > 0 {
651            // Column headers
652            output.push('┌');
653            for (i, _col) in self.columns.iter().enumerate() {
654                if i > 0 {
655                    output.push('┬');
656                }
657                output.push_str(&"─".repeat(20));
658            }
659            output.push_str("┐\n");
660
661            output.push('│');
662            for col in &self.columns {
663                output.push_str(&format!(" {:^18} │", Self::truncate_string(&col.name, 18)));
664            }
665            output.push('\n');
666
667            output.push('├');
668            for (i, _) in self.columns.iter().enumerate() {
669                if i > 0 {
670                    output.push('┼');
671                }
672                output.push_str(&"─".repeat(20));
673            }
674            output.push_str("┤\n");
675
676            // Data rows
677            for row_idx in 0..sample_count {
678                if let Some(row) = self.rows.get(row_idx) {
679                    output.push('│');
680                    for value in &row.values {
681                        let value_str = value.to_string();
682                        output
683                            .push_str(&format!(" {:18} │", Self::truncate_string(&value_str, 18)));
684                    }
685                    output.push('\n');
686                }
687            }
688
689            output.push('└');
690            for (i, _) in self.columns.iter().enumerate() {
691                if i > 0 {
692                    output.push('┴');
693                }
694                output.push_str(&"─".repeat(20));
695            }
696            output.push_str("┘\n");
697        }
698
699        output
700    }
701
702    fn truncate_string(s: &str, max_len: usize) -> String {
703        if s.len() > max_len {
704            format!("{}...", &s[..max_len - 3])
705        } else {
706            s.to_string()
707        }
708    }
709
710    /// Get a schema summary of the `DataTable`
711    #[must_use]
712    pub fn get_schema_summary(&self) -> String {
713        let mut summary = String::new();
714        summary.push_str(&format!(
715            "DataTable Schema ({} columns, {} rows):\n",
716            self.columns.len(),
717            self.rows.len()
718        ));
719
720        for (idx, column) in self.columns.iter().enumerate() {
721            let type_str = match &column.data_type {
722                DataType::String => "String",
723                DataType::Integer => "Integer",
724                DataType::Float => "Float",
725                DataType::Boolean => "Boolean",
726                DataType::DateTime => "DateTime",
727                DataType::Null => "Null",
728                DataType::Mixed => "Mixed",
729            };
730
731            let nullable_str = if column.nullable {
732                "nullable"
733            } else {
734                "not null"
735            };
736            let null_info = if column.null_count > 0 {
737                format!(", {} nulls", column.null_count)
738            } else {
739                String::new()
740            };
741
742            summary.push_str(&format!(
743                "  [{:3}] {} : {} ({}{})\n",
744                idx, column.name, type_str, nullable_str, null_info
745            ));
746        }
747
748        summary
749    }
750
751    /// Get detailed schema information as a structured format
752    #[must_use]
753    pub fn get_schema_info(&self) -> Vec<(String, String, bool, usize)> {
754        self.columns
755            .iter()
756            .map(|col| {
757                let type_name = format!("{:?}", col.data_type);
758                (col.name.clone(), type_name, col.nullable, col.null_count)
759            })
760            .collect()
761    }
762
763    /// Reserve capacity for rows to avoid reallocations
764    pub fn reserve_rows(&mut self, additional: usize) {
765        self.rows.reserve(additional);
766    }
767
768    /// Shrink vectors to fit actual data (removes excess capacity)
769    pub fn shrink_to_fit(&mut self) {
770        self.rows.shrink_to_fit();
771        for _column in &mut self.columns {
772            // Shrink any column-specific data if needed
773        }
774    }
775
776    /// Get actual memory usage estimate (more accurate than `estimate_memory_size`)
777    #[must_use]
778    pub fn get_memory_usage(&self) -> usize {
779        let mut size = std::mem::size_of::<Self>();
780
781        // Account for string allocations
782        size += self.name.capacity();
783
784        // Account for columns
785        size += self.columns.capacity() * std::mem::size_of::<DataColumn>();
786        for col in &self.columns {
787            size += col.name.capacity();
788        }
789
790        // Account for rows and their capacity
791        size += self.rows.capacity() * std::mem::size_of::<DataRow>();
792
793        // Account for actual data values
794        for row in &self.rows {
795            size += row.values.capacity() * std::mem::size_of::<DataValue>();
796            for value in &row.values {
797                match value {
798                    DataValue::String(s) => size += s.capacity(),
799                    DataValue::InternedString(_) => size += std::mem::size_of::<Arc<String>>(),
800                    DataValue::DateTime(s) => size += s.capacity(),
801                    _ => {} // Other types are inline
802                }
803            }
804        }
805
806        // Account for metadata
807        size += self.metadata.capacity() * std::mem::size_of::<(String, String)>();
808        for (k, v) in &self.metadata {
809            size += k.capacity() + v.capacity();
810        }
811
812        size
813    }
814}
815
816/// V46: Helper function to convert JSON value to `DataValue`
817fn json_value_to_data_value(json: &JsonValue) -> DataValue {
818    match json {
819        JsonValue::Null => DataValue::Null,
820        JsonValue::Bool(b) => DataValue::Boolean(*b),
821        JsonValue::Number(n) => {
822            if let Some(i) = n.as_i64() {
823                DataValue::Integer(i)
824            } else if let Some(f) = n.as_f64() {
825                DataValue::Float(f)
826            } else {
827                DataValue::String(n.to_string())
828            }
829        }
830        JsonValue::String(s) => {
831            // Try to detect if it's a date/time
832            if s.contains('-') && s.len() >= 8 && s.len() <= 30 {
833                // Simple heuristic for dates
834                DataValue::DateTime(s.clone())
835            } else {
836                DataValue::String(s.clone())
837            }
838        }
839        JsonValue::Array(_) | JsonValue::Object(_) => {
840            // Store complex types as JSON string
841            DataValue::String(json.to_string())
842        }
843    }
844}
845
846/// Statistics about a `DataTable`
847#[derive(Debug, Clone)]
848pub struct DataTableStats {
849    pub row_count: usize,
850    pub column_count: usize,
851    pub memory_size: usize,
852    pub null_count: usize,
853}
854
855/// Implementation of `DataProvider` for `DataTable`
856/// This allows `DataTable` to be used wherever `DataProvider` trait is expected
857impl DataProvider for DataTable {
858    fn get_row(&self, index: usize) -> Option<Vec<String>> {
859        self.rows.get(index).map(|row| {
860            row.values
861                .iter()
862                .map(DataValue::to_string_optimized)
863                .collect()
864        })
865    }
866
867    fn get_column_names(&self) -> Vec<String> {
868        self.column_names()
869    }
870
871    fn get_row_count(&self) -> usize {
872        self.row_count()
873    }
874
875    fn get_column_count(&self) -> usize {
876        self.column_count()
877    }
878}
879
880#[cfg(test)]
881mod tests {
882    use super::*;
883
884    #[test]
885    fn test_data_type_inference() {
886        assert_eq!(DataType::infer_from_string("123"), DataType::Integer);
887        assert_eq!(DataType::infer_from_string("123.45"), DataType::Float);
888        assert_eq!(DataType::infer_from_string("true"), DataType::Boolean);
889        assert_eq!(DataType::infer_from_string("hello"), DataType::String);
890        assert_eq!(DataType::infer_from_string(""), DataType::Null);
891        assert_eq!(
892            DataType::infer_from_string("2024-01-01"),
893            DataType::DateTime
894        );
895    }
896
897    #[test]
898    fn test_datatable_creation() {
899        let mut table = DataTable::new("test");
900
901        table.add_column(DataColumn::new("id").with_type(DataType::Integer));
902        table.add_column(DataColumn::new("name").with_type(DataType::String));
903        table.add_column(DataColumn::new("active").with_type(DataType::Boolean));
904
905        assert_eq!(table.column_count(), 3);
906        assert_eq!(table.row_count(), 0);
907
908        let row = DataRow::new(vec![
909            DataValue::Integer(1),
910            DataValue::String("Alice".to_string()),
911            DataValue::Boolean(true),
912        ]);
913
914        table.add_row(row).unwrap();
915        assert_eq!(table.row_count(), 1);
916
917        let value = table.get_value_by_name(0, "name").unwrap();
918        assert_eq!(value.to_string(), "Alice");
919    }
920
921    #[test]
922    fn test_type_inference() {
923        let mut table = DataTable::new("test");
924
925        // Add columns without types
926        table.add_column(DataColumn::new("mixed"));
927
928        // Add rows with different types
929        table
930            .add_row(DataRow::new(vec![DataValue::Integer(1)]))
931            .unwrap();
932        table
933            .add_row(DataRow::new(vec![DataValue::Float(2.5)]))
934            .unwrap();
935        table.add_row(DataRow::new(vec![DataValue::Null])).unwrap();
936
937        table.infer_column_types();
938
939        // Should infer Float since we have both Integer and Float
940        assert_eq!(table.columns[0].data_type, DataType::Float);
941        assert_eq!(table.columns[0].null_count, 1);
942        assert!(table.columns[0].nullable);
943    }
944
945    #[test]
946    fn test_from_query_response() {
947        use crate::api_client::{QueryInfo, QueryResponse};
948        use serde_json::json;
949
950        let response = QueryResponse {
951            query: QueryInfo {
952                select: vec!["id".to_string(), "name".to_string(), "age".to_string()],
953                where_clause: None,
954                order_by: None,
955            },
956            data: vec![
957                json!({
958                    "id": 1,
959                    "name": "Alice",
960                    "age": 30
961                }),
962                json!({
963                    "id": 2,
964                    "name": "Bob",
965                    "age": 25
966                }),
967                json!({
968                    "id": 3,
969                    "name": "Carol",
970                    "age": null
971                }),
972            ],
973            count: 3,
974            source: Some("test.csv".to_string()),
975            table: Some("test".to_string()),
976            cached: Some(false),
977        };
978
979        let table = DataTable::from_query_response(&response, "test").unwrap();
980
981        assert_eq!(table.name, "test");
982        assert_eq!(table.row_count(), 3);
983        assert_eq!(table.column_count(), 3);
984
985        // Check column names
986        let col_names = table.column_names();
987        assert!(col_names.contains(&"id".to_string()));
988        assert!(col_names.contains(&"name".to_string()));
989        assert!(col_names.contains(&"age".to_string()));
990
991        // Check metadata
992        assert_eq!(table.metadata.get("source"), Some(&"test.csv".to_string()));
993        assert_eq!(table.metadata.get("cached"), Some(&"false".to_string()));
994
995        // Check first row values
996        assert_eq!(
997            table.get_value_by_name(0, "id"),
998            Some(&DataValue::Integer(1))
999        );
1000        assert_eq!(
1001            table.get_value_by_name(0, "name"),
1002            Some(&DataValue::String("Alice".to_string()))
1003        );
1004        assert_eq!(
1005            table.get_value_by_name(0, "age"),
1006            Some(&DataValue::Integer(30))
1007        );
1008
1009        // Check null handling
1010        assert_eq!(table.get_value_by_name(2, "age"), Some(&DataValue::Null));
1011    }
1012}
sql_cli/data/datatable.rs

sql_cli/data/
datatable.rs
