sql_cli/sql/parser/
ast_formatter.rs

1//! AST-based SQL Formatter
2//!
3//! This module provides proper SQL formatting by traversing the parsed AST,
4//! which is more reliable than regex-based formatting and handles complex
5//! features like CTEs, subqueries, and expressions correctly.
6
7use crate::sql::parser::ast::*;
8use std::fmt::Write;
9
10/// Configuration for SQL formatting
11pub struct FormatConfig {
12    /// Indentation string (e.g., "  " for 2 spaces, "\t" for tab)
13    pub indent: String,
14    /// Maximum number of items per line for lists (SELECT columns, etc.)
15    pub items_per_line: usize,
16    /// Whether to uppercase keywords
17    pub uppercase_keywords: bool,
18    /// Whether to add newlines between major clauses
19    pub compact: bool,
20}
21
22impl Default for FormatConfig {
23    fn default() -> Self {
24        Self {
25            indent: "    ".to_string(),
26            items_per_line: 5,
27            uppercase_keywords: true,
28            compact: false,
29        }
30    }
31}
32
33/// Format a SELECT statement into pretty SQL
34pub fn format_select_statement(stmt: &SelectStatement) -> String {
35    format_select_with_config(stmt, &FormatConfig::default())
36}
37
38/// Format a SELECT statement with custom configuration
39pub fn format_select_with_config(stmt: &SelectStatement, config: &FormatConfig) -> String {
40    let formatter = AstFormatter::new(config);
41    formatter.format_select(stmt, 0)
42}
43
44/// Format a single SQL expression into a string
45/// This is useful for extracting and displaying parts of a query
46pub fn format_expression(expr: &SqlExpression) -> String {
47    let config = FormatConfig::default();
48    let formatter = AstFormatter::new(&config);
49    formatter.format_expression(expr)
50}
51
52struct AstFormatter<'a> {
53    config: &'a FormatConfig,
54}
55
56impl<'a> AstFormatter<'a> {
57    fn new(config: &'a FormatConfig) -> Self {
58        Self { config }
59    }
60
61    fn keyword(&self, word: &str) -> String {
62        if self.config.uppercase_keywords {
63            word.to_uppercase()
64        } else {
65            word.to_lowercase()
66        }
67    }
68
69    fn indent(&self, level: usize) -> String {
70        self.config.indent.repeat(level)
71    }
72
73    fn format_select(&self, stmt: &SelectStatement, indent_level: usize) -> String {
74        let mut result = String::new();
75        let indent = self.indent(indent_level);
76
77        // Emit leading comments if present
78        for comment in &stmt.leading_comments {
79            self.format_comment(&mut result, comment, &indent);
80        }
81
82        // CTEs (WITH clause)
83        if !stmt.ctes.is_empty() {
84            writeln!(&mut result, "{}{}", indent, self.keyword("WITH")).unwrap();
85            for (i, cte) in stmt.ctes.iter().enumerate() {
86                let is_last = i == stmt.ctes.len() - 1;
87                self.format_cte(&mut result, cte, indent_level + 1, is_last);
88            }
89        }
90
91        // SELECT clause (with newline if we had leading comments OR CTEs)
92        if !stmt.leading_comments.is_empty() || !stmt.ctes.is_empty() {
93            writeln!(&mut result, "{}{}", indent, self.keyword("SELECT")).unwrap();
94        } else {
95            write!(&mut result, "{}{}", indent, self.keyword("SELECT")).unwrap();
96        }
97        if stmt.distinct {
98            write!(&mut result, " {}", self.keyword("DISTINCT")).unwrap();
99        }
100
101        // Format select items
102        if stmt.select_items.is_empty() && !stmt.columns.is_empty() {
103            // Legacy columns field
104            self.format_column_list(&mut result, &stmt.columns, indent_level);
105        } else {
106            self.format_select_items(&mut result, &stmt.select_items, indent_level);
107        }
108
109        // INTO clause (for SELECT INTO #temp)
110        if let Some(ref into_table) = stmt.into_table {
111            writeln!(&mut result).unwrap();
112            write!(
113                &mut result,
114                "{}{} {}",
115                indent,
116                self.keyword("INTO"),
117                into_table.name
118            )
119            .unwrap();
120        }
121
122        // FROM clause
123        if let Some(ref table) = stmt.from_table {
124            writeln!(&mut result).unwrap();
125            write!(&mut result, "{}{} {}", indent, self.keyword("FROM"), table).unwrap();
126        } else if let Some(ref subquery) = stmt.from_subquery {
127            writeln!(&mut result).unwrap();
128            write!(&mut result, "{}{} (", indent, self.keyword("FROM")).unwrap();
129            writeln!(&mut result).unwrap();
130            let subquery_sql = self.format_select(subquery, indent_level + 1);
131            write!(&mut result, "{}", subquery_sql).unwrap();
132            write!(&mut result, "\n{}", indent).unwrap();
133            write!(&mut result, ")").unwrap();
134            if let Some(ref alias) = stmt.from_alias {
135                write!(&mut result, " {} {}", self.keyword("AS"), alias).unwrap();
136            }
137        } else if let Some(ref func) = stmt.from_function {
138            writeln!(&mut result).unwrap();
139            write!(&mut result, "{}{} ", indent, self.keyword("FROM")).unwrap();
140            self.format_table_function(&mut result, func);
141            if let Some(ref alias) = stmt.from_alias {
142                write!(&mut result, " {} {}", self.keyword("AS"), alias).unwrap();
143            }
144        }
145
146        // JOIN clauses
147        for join in &stmt.joins {
148            writeln!(&mut result).unwrap();
149            self.format_join(&mut result, join, indent_level);
150        }
151
152        // WHERE clause
153        if let Some(ref where_clause) = stmt.where_clause {
154            writeln!(&mut result).unwrap();
155            write!(&mut result, "{}{}", indent, self.keyword("WHERE")).unwrap();
156            self.format_where_clause(&mut result, where_clause, indent_level);
157        }
158
159        // GROUP BY clause
160        if let Some(ref group_by) = stmt.group_by {
161            writeln!(&mut result).unwrap();
162            write!(&mut result, "{}{} ", indent, self.keyword("GROUP BY")).unwrap();
163            for (i, expr) in group_by.iter().enumerate() {
164                if i > 0 {
165                    write!(&mut result, ", ").unwrap();
166                }
167                write!(&mut result, "{}", self.format_expression(expr)).unwrap();
168            }
169        }
170
171        // HAVING clause
172        if let Some(ref having) = stmt.having {
173            writeln!(&mut result).unwrap();
174            write!(
175                &mut result,
176                "{}{} {}",
177                indent,
178                self.keyword("HAVING"),
179                self.format_expression(having)
180            )
181            .unwrap();
182        }
183
184        // ORDER BY clause
185        if let Some(ref order_by) = stmt.order_by {
186            writeln!(&mut result).unwrap();
187            write!(&mut result, "{}{} ", indent, self.keyword("ORDER BY")).unwrap();
188            for (i, col) in order_by.iter().enumerate() {
189                if i > 0 {
190                    write!(&mut result, ", ").unwrap();
191                }
192                write!(&mut result, "{}", self.format_expression(&col.expr)).unwrap();
193                match col.direction {
194                    SortDirection::Asc => write!(&mut result, " {}", self.keyword("ASC")).unwrap(),
195                    SortDirection::Desc => {
196                        write!(&mut result, " {}", self.keyword("DESC")).unwrap()
197                    }
198                }
199            }
200        }
201
202        // LIMIT clause
203        if let Some(limit) = stmt.limit {
204            writeln!(&mut result).unwrap();
205            write!(&mut result, "{}{} {}", indent, self.keyword("LIMIT"), limit).unwrap();
206        }
207
208        // OFFSET clause
209        if let Some(offset) = stmt.offset {
210            writeln!(&mut result).unwrap();
211            write!(
212                &mut result,
213                "{}{} {}",
214                indent,
215                self.keyword("OFFSET"),
216                offset
217            )
218            .unwrap();
219        }
220
221        // Emit trailing comment if present
222        if let Some(ref comment) = stmt.trailing_comment {
223            write!(&mut result, "  ").unwrap();
224            self.format_inline_comment(&mut result, comment);
225        }
226
227        result
228    }
229
230    /// Format a comment on its own line with proper indentation
231    fn format_comment(&self, result: &mut String, comment: &Comment, indent: &str) {
232        if comment.is_line_comment {
233            writeln!(result, "{}-- {}", indent, comment.text.trim()).unwrap();
234        } else {
235            // Block comments
236            writeln!(result, "{}/* {} */", indent, comment.text.trim()).unwrap();
237        }
238    }
239
240    /// Format an inline trailing comment (on the same line as SQL)
241    fn format_inline_comment(&self, result: &mut String, comment: &Comment) {
242        if comment.is_line_comment {
243            write!(result, "-- {}", comment.text.trim()).unwrap();
244        } else {
245            write!(result, "/* {} */", comment.text.trim()).unwrap();
246        }
247    }
248
249    fn format_cte(&self, result: &mut String, cte: &CTE, indent_level: usize, is_last: bool) {
250        let indent = self.indent(indent_level);
251
252        // Add WEB keyword for Web CTEs
253        let is_web = matches!(&cte.cte_type, crate::sql::parser::ast::CTEType::Web(_));
254        if is_web {
255            write!(result, "{}{} {}", indent, self.keyword("WEB"), cte.name).unwrap();
256        } else {
257            write!(result, "{}{}", indent, cte.name).unwrap();
258        }
259
260        if let Some(ref columns) = cte.column_list {
261            write!(result, "(").unwrap();
262            for (i, col) in columns.iter().enumerate() {
263                if i > 0 {
264                    write!(result, ", ").unwrap();
265                }
266                write!(result, "{}", col).unwrap();
267            }
268            write!(result, ")").unwrap();
269        }
270
271        writeln!(result, " {} (", self.keyword("AS")).unwrap();
272        let cte_sql = match &cte.cte_type {
273            crate::sql::parser::ast::CTEType::Standard(query) => {
274                self.format_select(query, indent_level + 1)
275            }
276            crate::sql::parser::ast::CTEType::Web(web_spec) => {
277                // Format WEB CTE
278                let mut web_str = format!(
279                    "{}{} '{}'",
280                    "    ".repeat(indent_level + 1),
281                    self.keyword("URL"),
282                    web_spec.url
283                );
284
285                // Add METHOD if specified
286                if let Some(method) = &web_spec.method {
287                    web_str.push_str(&format!(
288                        " {} {}",
289                        self.keyword("METHOD"),
290                        match method {
291                            crate::sql::parser::ast::HttpMethod::GET => "GET",
292                            crate::sql::parser::ast::HttpMethod::POST => "POST",
293                            crate::sql::parser::ast::HttpMethod::PUT => "PUT",
294                            crate::sql::parser::ast::HttpMethod::DELETE => "DELETE",
295                            crate::sql::parser::ast::HttpMethod::PATCH => "PATCH",
296                        }
297                    ));
298                }
299
300                // Add BODY if specified
301                if let Some(body) = &web_spec.body {
302                    // Check if the body looks like JSON (starts with { or [)
303                    let trimmed_body = body.trim();
304                    if (trimmed_body.starts_with('{') && trimmed_body.ends_with('}'))
305                        || (trimmed_body.starts_with('[') && trimmed_body.ends_with(']'))
306                    {
307                        // Try to prettify JSON
308                        match serde_json::from_str::<serde_json::Value>(trimmed_body) {
309                            Ok(json_val) => {
310                                // Pretty print JSON with 2-space indentation
311                                match serde_json::to_string_pretty(&json_val) {
312                                    Ok(pretty_json) => {
313                                        // Check if JSON is complex (multiline or has special chars)
314                                        let is_complex = pretty_json.lines().count() > 1
315                                            || pretty_json.contains('"')
316                                            || pretty_json.contains('\\');
317
318                                        if is_complex {
319                                            // Use $JSON$ delimiters for complex JSON
320                                            let base_indent = "    ".repeat(indent_level + 1);
321                                            let json_lines: Vec<String> = pretty_json
322                                                .lines()
323                                                .enumerate()
324                                                .map(|(i, line)| {
325                                                    if i == 0 {
326                                                        line.to_string()
327                                                    } else {
328                                                        format!("{}{}", base_indent, line)
329                                                    }
330                                                })
331                                                .collect();
332                                            let formatted_json = json_lines.join("\n");
333
334                                            web_str.push_str(&format!(
335                                                " {} $JSON${}\n{}$JSON$\n{}",
336                                                self.keyword("BODY"),
337                                                formatted_json,
338                                                base_indent,
339                                                base_indent
340                                            ));
341                                        } else {
342                                            // Simple JSON, use regular single quotes
343                                            web_str.push_str(&format!(
344                                                " {} '{}'",
345                                                self.keyword("BODY"),
346                                                pretty_json
347                                            ));
348                                        }
349                                    }
350                                    Err(_) => {
351                                        // Fall back to original if pretty print fails
352                                        web_str.push_str(&format!(
353                                            " {} '{}'",
354                                            self.keyword("BODY"),
355                                            body
356                                        ));
357                                    }
358                                }
359                            }
360                            Err(_) => {
361                                // Not valid JSON, use as-is
362                                web_str.push_str(&format!(" {} '{}'", self.keyword("BODY"), body));
363                            }
364                        }
365                    } else {
366                        // Not JSON, use as-is
367                        web_str.push_str(&format!(" {} '{}'", self.keyword("BODY"), body));
368                    }
369                }
370
371                // Add FORMAT if specified
372                if let Some(format) = &web_spec.format {
373                    web_str.push_str(&format!(
374                        " {} {}",
375                        self.keyword("FORMAT"),
376                        match format {
377                            crate::sql::parser::ast::DataFormat::CSV => "CSV",
378                            crate::sql::parser::ast::DataFormat::JSON => "JSON",
379                            crate::sql::parser::ast::DataFormat::Auto => "AUTO",
380                        }
381                    ));
382                }
383
384                // Add JSON_PATH if specified
385                if let Some(json_path) = &web_spec.json_path {
386                    web_str.push_str(&format!(" {} '{}'", self.keyword("JSON_PATH"), json_path));
387                }
388
389                // Add CACHE if specified
390                if let Some(cache) = web_spec.cache_seconds {
391                    web_str.push_str(&format!(" {} {}", self.keyword("CACHE"), cache));
392                }
393
394                // Add FORM_FILE entries if specified
395                for (field_name, file_path) in &web_spec.form_files {
396                    web_str.push_str(&format!(
397                        "\n{}{} '{}' '{}'",
398                        "    ".repeat(indent_level + 1),
399                        self.keyword("FORM_FILE"),
400                        field_name,
401                        file_path
402                    ));
403                }
404
405                // Add FORM_FIELD entries if specified
406                for (field_name, value) in &web_spec.form_fields {
407                    // Check if the value looks like JSON (starts with { or [)
408                    let trimmed_value = value.trim();
409                    if (trimmed_value.starts_with('{') && trimmed_value.ends_with('}'))
410                        || (trimmed_value.starts_with('[') && trimmed_value.ends_with(']'))
411                    {
412                        // Try to prettify JSON
413                        match serde_json::from_str::<serde_json::Value>(trimmed_value) {
414                            Ok(json_val) => {
415                                // Pretty print JSON with 2-space indentation
416                                match serde_json::to_string_pretty(&json_val) {
417                                    Ok(pretty_json) => {
418                                        // Check if JSON is complex (multiline or has special chars)
419                                        let is_complex = pretty_json.lines().count() > 1
420                                            || pretty_json.contains('"')
421                                            || pretty_json.contains('\\');
422
423                                        if is_complex {
424                                            // Use $JSON$ delimiters for complex JSON
425                                            let base_indent = "    ".repeat(indent_level + 1);
426                                            let json_lines: Vec<String> = pretty_json
427                                                .lines()
428                                                .enumerate()
429                                                .map(|(i, line)| {
430                                                    if i == 0 {
431                                                        line.to_string()
432                                                    } else {
433                                                        format!("{}{}", base_indent, line)
434                                                    }
435                                                })
436                                                .collect();
437                                            let formatted_json = json_lines.join("\n");
438
439                                            web_str.push_str(&format!(
440                                                "\n{}{} '{}' $JSON${}\n{}$JSON$",
441                                                base_indent,
442                                                self.keyword("FORM_FIELD"),
443                                                field_name,
444                                                formatted_json,
445                                                base_indent
446                                            ));
447                                        } else {
448                                            // Simple JSON, use regular single quotes
449                                            web_str.push_str(&format!(
450                                                "\n{}{} '{}' '{}'",
451                                                "    ".repeat(indent_level + 1),
452                                                self.keyword("FORM_FIELD"),
453                                                field_name,
454                                                pretty_json
455                                            ));
456                                        }
457                                    }
458                                    Err(_) => {
459                                        // Fall back to original if pretty print fails
460                                        web_str.push_str(&format!(
461                                            "\n{}{} '{}' '{}'",
462                                            "    ".repeat(indent_level + 1),
463                                            self.keyword("FORM_FIELD"),
464                                            field_name,
465                                            value
466                                        ));
467                                    }
468                                }
469                            }
470                            Err(_) => {
471                                // Not valid JSON, use as-is
472                                web_str.push_str(&format!(
473                                    "\n{}{} '{}' '{}'",
474                                    "    ".repeat(indent_level + 1),
475                                    self.keyword("FORM_FIELD"),
476                                    field_name,
477                                    value
478                                ));
479                            }
480                        }
481                    } else {
482                        // Not JSON, use as-is
483                        web_str.push_str(&format!(
484                            "\n{}{} '{}' '{}'",
485                            "    ".repeat(indent_level + 1),
486                            self.keyword("FORM_FIELD"),
487                            field_name,
488                            value
489                        ));
490                    }
491                }
492
493                // Add HEADERS if specified
494                if !web_spec.headers.is_empty() {
495                    web_str.push_str(&format!(" {} (", self.keyword("HEADERS")));
496                    for (i, (key, value)) in web_spec.headers.iter().enumerate() {
497                        if i > 0 {
498                            web_str.push_str(", ");
499                        }
500                        web_str.push_str(&format!("'{}': '{}'", key, value));
501                    }
502                    web_str.push(')');
503                }
504
505                web_str
506            }
507        };
508        write!(result, "{}", cte_sql).unwrap();
509        writeln!(result).unwrap();
510        write!(result, "{}", indent).unwrap();
511        if is_last {
512            writeln!(result, ")").unwrap();
513        } else {
514            writeln!(result, "),").unwrap();
515        }
516    }
517
518    fn format_column_list(&self, result: &mut String, columns: &[String], indent_level: usize) {
519        if columns.len() <= self.config.items_per_line {
520            // Single line
521            write!(result, " ").unwrap();
522            for (i, col) in columns.iter().enumerate() {
523                if i > 0 {
524                    write!(result, ", ").unwrap();
525                }
526                write!(result, "{}", col).unwrap();
527            }
528        } else {
529            // Multi-line
530            writeln!(result).unwrap();
531            let indent = self.indent(indent_level + 1);
532            for (i, col) in columns.iter().enumerate() {
533                write!(result, "{}{}", indent, col).unwrap();
534                if i < columns.len() - 1 {
535                    writeln!(result, ",").unwrap();
536                }
537            }
538        }
539    }
540
541    fn format_select_items(&self, result: &mut String, items: &[SelectItem], indent_level: usize) {
542        if items.is_empty() {
543            write!(result, " *").unwrap();
544            return;
545        }
546
547        // Count non-star items for formatting decision
548        let _non_star_count = items
549            .iter()
550            .filter(|i| !matches!(i, SelectItem::Star { .. }))
551            .count();
552
553        // Check if any item is complex (function calls, CASE expressions, etc.)
554        let has_complex_items = items.iter().any(|item| match item {
555            SelectItem::Expression { expr, .. } => self.is_complex_expression(expr),
556            _ => false,
557        });
558
559        // Calculate total approximate length if on single line
560        let single_line_length: usize = items
561            .iter()
562            .map(|item| {
563                match item {
564                    SelectItem::Star { .. } => 1,
565                    SelectItem::StarExclude {
566                        excluded_columns, ..
567                    } => {
568                        // "* EXCLUDE (" + column names + ")"
569                        11 + excluded_columns.iter().map(|c| c.len()).sum::<usize>()
570                            + (excluded_columns.len().saturating_sub(1) * 2) // ", " separators
571                    }
572                    SelectItem::Column { column: col, .. } => col.name.len(),
573                    SelectItem::Expression { expr, alias, .. } => {
574                        self.format_expression(expr).len() + 4 + alias.len() // " AS " = 4
575                    }
576                }
577            })
578            .sum::<usize>()
579            + (items.len() - 1) * 2; // ", " between items
580
581        // Use multi-line formatting by default unless:
582        // - It's a single simple column or star
583        // - It's 2-3 simple columns with total length < 40 chars
584        let use_single_line = match items.len() {
585            1 => !has_complex_items, // Single item: only if simple
586            2..=3 => !has_complex_items && single_line_length < 40, // 2-3 items: only if very short
587            _ => false,              // 4+ items: always multi-line
588        };
589
590        if !use_single_line {
591            // Multi-line
592            writeln!(result).unwrap();
593            let indent = self.indent(indent_level + 1);
594            for (i, item) in items.iter().enumerate() {
595                write!(result, "{}", indent).unwrap();
596                self.format_select_item(result, item);
597                if i < items.len() - 1 {
598                    writeln!(result, ",").unwrap();
599                }
600            }
601        } else {
602            // Single line
603            write!(result, " ").unwrap();
604            for (i, item) in items.iter().enumerate() {
605                if i > 0 {
606                    write!(result, ", ").unwrap();
607                }
608                self.format_select_item(result, item);
609            }
610        }
611    }
612
613    fn is_complex_expression(&self, expr: &SqlExpression) -> bool {
614        match expr {
615            SqlExpression::CaseExpression { .. } => true,
616            SqlExpression::FunctionCall { .. } => true,
617            SqlExpression::WindowFunction { .. } => true,
618            SqlExpression::ScalarSubquery { .. } => true,
619            SqlExpression::InSubquery { .. } => true,
620            SqlExpression::NotInSubquery { .. } => true,
621            SqlExpression::BinaryOp { left, right, .. } => {
622                self.is_complex_expression(left) || self.is_complex_expression(right)
623            }
624            _ => false,
625        }
626    }
627
628    fn format_select_item(&self, result: &mut String, item: &SelectItem) {
629        match item {
630            SelectItem::Star { .. } => write!(result, "*").unwrap(),
631            SelectItem::StarExclude {
632                excluded_columns, ..
633            } => {
634                write!(
635                    result,
636                    "* {} ({})",
637                    self.keyword("EXCLUDE"),
638                    excluded_columns.join(", ")
639                )
640                .unwrap();
641            }
642            SelectItem::Column { column: col, .. } => write!(result, "{}", col.to_sql()).unwrap(),
643            SelectItem::Expression { expr, alias, .. } => {
644                write!(
645                    result,
646                    "{} {} {}",
647                    self.format_expression(expr),
648                    self.keyword("AS"),
649                    alias
650                )
651                .unwrap();
652            }
653        }
654    }
655
656    fn format_expression(&self, expr: &SqlExpression) -> String {
657        match expr {
658            SqlExpression::Column(column_ref) => column_ref.to_sql(),
659            SqlExpression::StringLiteral(s) => format!("'{}'", s),
660            SqlExpression::NumberLiteral(n) => n.clone(),
661            SqlExpression::BooleanLiteral(b) => b.to_string().to_uppercase(),
662            SqlExpression::Null => self.keyword("NULL"),
663            SqlExpression::BinaryOp { left, op, right } => {
664                // Special handling for IS NULL / IS NOT NULL operators
665                if op == "IS NULL" || op == "IS NOT NULL" {
666                    format!("{} {}", self.format_expression(left), op)
667                } else {
668                    format!(
669                        "{} {} {}",
670                        self.format_expression(left),
671                        op,
672                        self.format_expression(right)
673                    )
674                }
675            }
676            SqlExpression::FunctionCall {
677                name,
678                args,
679                distinct,
680            } => {
681                let mut result = name.clone();
682                result.push('(');
683                if *distinct {
684                    result.push_str(&self.keyword("DISTINCT"));
685                    result.push(' ');
686                }
687                for (i, arg) in args.iter().enumerate() {
688                    if i > 0 {
689                        result.push_str(", ");
690                    }
691                    result.push_str(&self.format_expression(arg));
692                }
693                result.push(')');
694                result
695            }
696            SqlExpression::CaseExpression {
697                when_branches,
698                else_branch,
699            } => {
700                // Format CASE expressions on multiple lines for readability
701                let mut result = String::new();
702                result.push_str(&self.keyword("CASE"));
703                result.push('\n');
704
705                // Format each WHEN branch on its own line with indentation
706                for branch in when_branches {
707                    result.push_str("        "); // 8 spaces for WHEN indent
708                    result.push_str(&format!(
709                        "{} {} {} {}",
710                        self.keyword("WHEN"),
711                        self.format_expression(&branch.condition),
712                        self.keyword("THEN"),
713                        self.format_expression(&branch.result)
714                    ));
715                    result.push('\n');
716                }
717
718                // Format ELSE clause if present
719                if let Some(else_expr) = else_branch {
720                    result.push_str("        "); // 8 spaces for ELSE indent
721                    result.push_str(&format!(
722                        "{} {}",
723                        self.keyword("ELSE"),
724                        self.format_expression(else_expr)
725                    ));
726                    result.push('\n');
727                }
728
729                result.push_str("    "); // 4 spaces for END
730                result.push_str(&self.keyword("END"));
731                result
732            }
733            SqlExpression::SimpleCaseExpression {
734                expr,
735                when_branches,
736                else_branch,
737            } => {
738                // Format simple CASE expressions on multiple lines for readability
739                let mut result = String::new();
740                result.push_str(&format!(
741                    "{} {}",
742                    self.keyword("CASE"),
743                    self.format_expression(expr)
744                ));
745                result.push('\n');
746
747                // Format each WHEN branch on its own line with indentation
748                for branch in when_branches {
749                    result.push_str("        "); // 8 spaces for WHEN indent
750                    result.push_str(&format!(
751                        "{} {} {} {}",
752                        self.keyword("WHEN"),
753                        self.format_expression(&branch.value),
754                        self.keyword("THEN"),
755                        self.format_expression(&branch.result)
756                    ));
757                    result.push('\n');
758                }
759
760                // Format ELSE clause if present
761                if let Some(else_expr) = else_branch {
762                    result.push_str("        "); // 8 spaces for ELSE indent
763                    result.push_str(&format!(
764                        "{} {}",
765                        self.keyword("ELSE"),
766                        self.format_expression(else_expr)
767                    ));
768                    result.push('\n');
769                }
770
771                result.push_str("    "); // 4 spaces for END
772                result.push_str(&self.keyword("END"));
773                result
774            }
775            SqlExpression::Between { expr, lower, upper } => {
776                format!(
777                    "{} {} {} {} {}",
778                    self.format_expression(expr),
779                    self.keyword("BETWEEN"),
780                    self.format_expression(lower),
781                    self.keyword("AND"),
782                    self.format_expression(upper)
783                )
784            }
785            SqlExpression::InList { expr, values } => {
786                let mut result =
787                    format!("{} {} (", self.format_expression(expr), self.keyword("IN"));
788                for (i, val) in values.iter().enumerate() {
789                    if i > 0 {
790                        result.push_str(", ");
791                    }
792                    result.push_str(&self.format_expression(val));
793                }
794                result.push(')');
795                result
796            }
797            SqlExpression::NotInList { expr, values } => {
798                let mut result = format!(
799                    "{} {} {} (",
800                    self.format_expression(expr),
801                    self.keyword("NOT"),
802                    self.keyword("IN")
803                );
804                for (i, val) in values.iter().enumerate() {
805                    if i > 0 {
806                        result.push_str(", ");
807                    }
808                    result.push_str(&self.format_expression(val));
809                }
810                result.push(')');
811                result
812            }
813            SqlExpression::Not { expr } => {
814                format!("{} {}", self.keyword("NOT"), self.format_expression(expr))
815            }
816            SqlExpression::ScalarSubquery { query } => {
817                // Check if subquery is complex enough to warrant multi-line formatting
818                let subquery_str = self.format_select(query, 0);
819                if subquery_str.contains('\n') || subquery_str.len() > 60 {
820                    // Multi-line formatting
821                    format!("(\n{}\n)", self.format_select(query, 1))
822                } else {
823                    // Inline formatting
824                    format!("({})", subquery_str)
825                }
826            }
827            SqlExpression::InSubquery { expr, subquery } => {
828                let subquery_str = self.format_select(subquery, 0);
829                if subquery_str.contains('\n') || subquery_str.len() > 60 {
830                    // Multi-line formatting
831                    format!(
832                        "{} {} (\n{}\n)",
833                        self.format_expression(expr),
834                        self.keyword("IN"),
835                        self.format_select(subquery, 1)
836                    )
837                } else {
838                    // Inline formatting
839                    format!(
840                        "{} {} ({})",
841                        self.format_expression(expr),
842                        self.keyword("IN"),
843                        subquery_str
844                    )
845                }
846            }
847            SqlExpression::NotInSubquery { expr, subquery } => {
848                let subquery_str = self.format_select(subquery, 0);
849                if subquery_str.contains('\n') || subquery_str.len() > 60 {
850                    // Multi-line formatting
851                    format!(
852                        "{} {} {} (\n{}\n)",
853                        self.format_expression(expr),
854                        self.keyword("NOT"),
855                        self.keyword("IN"),
856                        self.format_select(subquery, 1)
857                    )
858                } else {
859                    // Inline formatting
860                    format!(
861                        "{} {} {} ({})",
862                        self.format_expression(expr),
863                        self.keyword("NOT"),
864                        self.keyword("IN"),
865                        subquery_str
866                    )
867                }
868            }
869            SqlExpression::MethodCall {
870                object,
871                method,
872                args,
873            } => {
874                let mut result = format!("{}.{}", object, method);
875                result.push('(');
876                for (i, arg) in args.iter().enumerate() {
877                    if i > 0 {
878                        result.push_str(", ");
879                    }
880                    result.push_str(&self.format_expression(arg));
881                }
882                result.push(')');
883                result
884            }
885            SqlExpression::ChainedMethodCall { base, method, args } => {
886                let mut result = format!("{}.{}", self.format_expression(base), method);
887                result.push('(');
888                for (i, arg) in args.iter().enumerate() {
889                    if i > 0 {
890                        result.push_str(", ");
891                    }
892                    result.push_str(&self.format_expression(arg));
893                }
894                result.push(')');
895                result
896            }
897            SqlExpression::WindowFunction {
898                name,
899                args,
900                window_spec,
901            } => {
902                let mut result = format!("{}(", name);
903
904                // Add function arguments
905                for (i, arg) in args.iter().enumerate() {
906                    if i > 0 {
907                        result.push_str(", ");
908                    }
909                    result.push_str(&self.format_expression(arg));
910                }
911                result.push_str(") ");
912                result.push_str(&self.keyword("OVER"));
913                result.push_str(" (");
914
915                // Add PARTITION BY clause if present
916                if !window_spec.partition_by.is_empty() {
917                    result.push_str(&self.keyword("PARTITION BY"));
918                    result.push(' ');
919                    for (i, col) in window_spec.partition_by.iter().enumerate() {
920                        if i > 0 {
921                            result.push_str(", ");
922                        }
923                        result.push_str(col);
924                    }
925                }
926
927                // Add ORDER BY clause if present
928                if !window_spec.order_by.is_empty() {
929                    if !window_spec.partition_by.is_empty() {
930                        result.push(' ');
931                    }
932                    result.push_str(&self.keyword("ORDER BY"));
933                    result.push(' ');
934                    for (i, col) in window_spec.order_by.iter().enumerate() {
935                        if i > 0 {
936                            result.push_str(", ");
937                        }
938                        result.push_str(&self.format_expression(&col.expr));
939                        match col.direction {
940                            SortDirection::Asc => {
941                                result.push(' ');
942                                result.push_str(&self.keyword("ASC"));
943                            }
944                            SortDirection::Desc => {
945                                result.push(' ');
946                                result.push_str(&self.keyword("DESC"));
947                            }
948                        }
949                    }
950                }
951
952                // Add window frame specification if present
953                if let Some(frame) = &window_spec.frame {
954                    // Add space before frame specification
955                    if !window_spec.partition_by.is_empty() || !window_spec.order_by.is_empty() {
956                        result.push(' ');
957                    }
958
959                    // Format frame unit (ROWS or RANGE)
960                    match frame.unit {
961                        FrameUnit::Rows => result.push_str(&self.keyword("ROWS")),
962                        FrameUnit::Range => result.push_str(&self.keyword("RANGE")),
963                    }
964
965                    result.push(' ');
966
967                    // Format frame bounds
968                    if let Some(end) = &frame.end {
969                        // BETWEEN start AND end
970                        result.push_str(&self.keyword("BETWEEN"));
971                        result.push(' ');
972                        result.push_str(&self.format_frame_bound(&frame.start));
973                        result.push(' ');
974                        result.push_str(&self.keyword("AND"));
975                        result.push(' ');
976                        result.push_str(&self.format_frame_bound(end));
977                    } else {
978                        // Just a single bound (uncommon but valid)
979                        result.push_str(&self.format_frame_bound(&frame.start));
980                    }
981                }
982
983                result.push(')');
984                result
985            }
986            SqlExpression::DateTimeConstructor {
987                year,
988                month,
989                day,
990                hour,
991                minute,
992                second,
993            } => {
994                if let (Some(h), Some(m), Some(s)) = (hour, minute, second) {
995                    format!(
996                        "DateTime({}, {}, {}, {}, {}, {})",
997                        year, month, day, h, m, s
998                    )
999                } else {
1000                    format!("DateTime({}, {}, {})", year, month, day)
1001                }
1002            }
1003            SqlExpression::DateTimeToday {
1004                hour,
1005                minute,
1006                second,
1007            } => {
1008                if let (Some(h), Some(m), Some(s)) = (hour, minute, second) {
1009                    format!("Today({}, {}, {})", h, m, s)
1010                } else {
1011                    "Today()".to_string()
1012                }
1013            }
1014            _ => format!("{:?}", expr), // Fallback for unhandled expression types
1015        }
1016    }
1017
1018    fn format_where_clause(
1019        &self,
1020        result: &mut String,
1021        where_clause: &WhereClause,
1022        indent_level: usize,
1023    ) {
1024        let needs_multiline = where_clause.conditions.len() > 1;
1025
1026        if needs_multiline {
1027            writeln!(result).unwrap();
1028            let indent = self.indent(indent_level + 1);
1029            for (i, condition) in where_clause.conditions.iter().enumerate() {
1030                if i > 0 {
1031                    if let Some(ref connector) = where_clause.conditions[i - 1].connector {
1032                        let connector_str = match connector {
1033                            LogicalOp::And => self.keyword("AND"),
1034                            LogicalOp::Or => self.keyword("OR"),
1035                        };
1036                        writeln!(result).unwrap();
1037                        write!(result, "{}{} ", indent, connector_str).unwrap();
1038                    }
1039                } else {
1040                    write!(result, "{}", indent).unwrap();
1041                }
1042                write!(result, "{}", self.format_expression(&condition.expr)).unwrap();
1043            }
1044        } else if let Some(condition) = where_clause.conditions.first() {
1045            write!(result, " {}", self.format_expression(&condition.expr)).unwrap();
1046        }
1047    }
1048
1049    fn format_frame_bound(&self, bound: &FrameBound) -> String {
1050        match bound {
1051            FrameBound::UnboundedPreceding => self.keyword("UNBOUNDED PRECEDING"),
1052            FrameBound::CurrentRow => self.keyword("CURRENT ROW"),
1053            FrameBound::UnboundedFollowing => self.keyword("UNBOUNDED FOLLOWING"),
1054            FrameBound::Preceding(n) => format!("{} {}", n, self.keyword("PRECEDING")),
1055            FrameBound::Following(n) => format!("{} {}", n, self.keyword("FOLLOWING")),
1056        }
1057    }
1058
1059    fn format_join(&self, result: &mut String, join: &JoinClause, indent_level: usize) {
1060        let indent = self.indent(indent_level);
1061        let join_type = match join.join_type {
1062            JoinType::Inner => self.keyword("INNER JOIN"),
1063            JoinType::Left => self.keyword("LEFT JOIN"),
1064            JoinType::Right => self.keyword("RIGHT JOIN"),
1065            JoinType::Full => self.keyword("FULL JOIN"),
1066            JoinType::Cross => self.keyword("CROSS JOIN"),
1067        };
1068
1069        write!(result, "{}{} ", indent, join_type).unwrap();
1070
1071        match &join.table {
1072            TableSource::Table(name) => write!(result, "{}", name).unwrap(),
1073            TableSource::DerivedTable { query, alias } => {
1074                writeln!(result, "(").unwrap();
1075                let subquery_sql = self.format_select(query, indent_level + 1);
1076                write!(result, "{}", subquery_sql).unwrap();
1077                writeln!(result).unwrap();
1078                write!(result, "{}) {} {}", indent, self.keyword("AS"), alias).unwrap();
1079            }
1080        }
1081
1082        if let Some(ref alias) = join.alias {
1083            write!(result, " {} {}", self.keyword("AS"), alias).unwrap();
1084        }
1085
1086        if !join.condition.conditions.is_empty() {
1087            write!(result, " {}", self.keyword("ON")).unwrap();
1088            for (i, condition) in join.condition.conditions.iter().enumerate() {
1089                if i > 0 {
1090                    write!(result, " {}", self.keyword("AND")).unwrap();
1091                }
1092                write!(
1093                    result,
1094                    " {} {} {}",
1095                    self.format_expression(&condition.left_expr),
1096                    self.format_join_operator(&condition.operator),
1097                    self.format_expression(&condition.right_expr)
1098                )
1099                .unwrap();
1100            }
1101        }
1102    }
1103
1104    fn format_join_operator(&self, op: &JoinOperator) -> String {
1105        match op {
1106            JoinOperator::Equal => "=",
1107            JoinOperator::NotEqual => "!=",
1108            JoinOperator::LessThan => "<",
1109            JoinOperator::GreaterThan => ">",
1110            JoinOperator::LessThanOrEqual => "<=",
1111            JoinOperator::GreaterThanOrEqual => ">=",
1112        }
1113        .to_string()
1114    }
1115
1116    fn format_table_function(&self, result: &mut String, func: &TableFunction) {
1117        match func {
1118            TableFunction::Generator { name, args } => {
1119                write!(result, "{}(", self.keyword(&name.to_uppercase())).unwrap();
1120                for (i, arg) in args.iter().enumerate() {
1121                    if i > 0 {
1122                        write!(result, ", ").unwrap();
1123                    }
1124                    write!(result, "{}", self.format_expression(arg)).unwrap();
1125                }
1126                write!(result, ")").unwrap();
1127            }
1128        }
1129    }
1130}
1131
1132/// Parse and format SQL query using the AST
1133pub fn format_sql_ast(query: &str) -> Result<String, String> {
1134    use crate::sql::recursive_parser::Parser;
1135
1136    let mut parser = Parser::new(query);
1137    match parser.parse() {
1138        Ok(stmt) => Ok(format_select_statement(&stmt)),
1139        Err(e) => Err(format!("Parse error: {}", e)),
1140    }
1141}
1142
1143/// Parse and format SQL with custom configuration
1144pub fn format_sql_ast_with_config(query: &str, config: &FormatConfig) -> Result<String, String> {
1145    use crate::sql::recursive_parser::Parser;
1146
1147    let mut parser = Parser::new(query);
1148    match parser.parse() {
1149        Ok(stmt) => Ok(format_select_with_config(&stmt, &config)),
1150        Err(e) => Err(format!("Parse error: {}", e)),
1151    }
1152}
sql_cli/sql/parser/ast_formatter.rs

sql_cli/sql/parser/
ast_formatter.rs
