chore: more work done and planning with AI.

pkg/format/dml.go

@@ -179,9 +179,16 @@ func dmlSegText(seg dmlSeg, st config.Style) string {
 	}
 
 	switch kw {
-	case "select", "set", "returning":
+	case "select", "returning":
 		items := dmlSplitCommas(seg.body)
-		return dmlColList(kwText, items, st)
+		return dmlColListSelect(kwText, items, st)
+	case "set":
+		items := dmlSplitCommas(seg.body)
+		return dmlColListSet(kwText, items, st)
+	case "where":
+		return dmlWhereClause(kwText, seg.body, st)
+	case "join", "left", "right", "inner", "full", "cross", "natural":
+		return dmlJoinClause(kwText, seg.body, st)
 	case "with":
 		return formatWithBody(kwText, seg.body, st)
 	default:
@@ -193,6 +200,97 @@ func dmlSegText(seg dmlSeg, st config.Style) string {
 	}
 }
 
+// dmlJoinClause formats a JOIN clause, applying indent_join when configured.
+func dmlJoinClause(kwText string, body []cst.Tok, st config.Style) string {
+	text := dmlInline(body, st)
+	line := kwText
+	if text != "" {
+		line += " " + text
+	}
+	if !st.IndentJoin {
+		return line
+	}
+	indent := strings.Repeat(st.Indent, st.JoinIndentSize)
+	nl := st.Newline
+	var b strings.Builder
+	for i, part := range strings.Split(line, nl) {
+		if i > 0 {
+			b.WriteString(nl)
+		}
+		b.WriteString(indent)
+		b.WriteString(part)
+	}
+	return b.String()
+}
+
+// dmlWhereClause formats a WHERE clause, splitting AND/OR conditions per
+// the where_wrap and where_and_or_indent settings.
+func dmlWhereClause(kwText string, body []cst.Tok, st config.Style) string {
+	if st.WhereWrap == config.WrapNever {
+		text := dmlInline(body, st)
+		if text == "" {
+			return kwText
+		}
+		return kwText + " " + text
+	}
+
+	// Split at depth-0 AND/OR.
+	conditions := dmlSplitAndOr(body)
+	if len(conditions) <= 1 {
+		text := dmlInline(body, st)
+		if text == "" {
+			return kwText
+		}
+		return kwText + " " + text
+	}
+
+	nl := st.Newline
+	var b strings.Builder
+	b.WriteString(kwText)
+	for i, cond := range conditions {
+		b.WriteString(nl)
+		text := dmlInline(cond, st)
+		if st.WhereAndOrIndent {
+			b.WriteString(st.Indent)
+		}
+		if i == 0 {
+			// First condition: no leading AND/OR
+			b.WriteString("  ") // align with AND/OR token width
+			b.WriteString(text)
+		} else {
+			b.WriteString(text)
+		}
+	}
+	return b.String()
+}
+
+// dmlSplitAndOr splits toks at depth-0 AND/OR tokens, keeping the AND/OR with
+// the following condition.
+func dmlSplitAndOr(toks []cst.Tok) [][]cst.Tok {
+	var result [][]cst.Tok
+	depth := 0
+	start := 0
+	for i, t := range toks {
+		switch t.Tok.Kind {
+		case lexer.LParen, lexer.LBracket:
+			depth++
+		case lexer.RParen, lexer.RBracket:
+			if depth > 0 {
+				depth--
+			}
+		}
+		if depth == 0 && t.Tok.Kind == lexer.Ident {
+			low := lowerASCII(t.Tok.Text)
+			if (low == "and" || low == "or") && i > start {
+				result = append(result, toks[start:i])
+				start = i
+			}
+		}
+	}
+	result = append(result, toks[start:])
+	return result
+}
+
 // formatWithBody formats the body of a WITH clause by splitting CTE definitions
 // at depth-0 commas and formatting the subquery inside each AS (...) block.
 func formatWithBody(kwText string, body []cst.Tok, st config.Style) string {
@@ -370,7 +468,18 @@ func dmlInline(toks []cst.Tok, st config.Style) string {
 		if i > 0 && needSpace(toks[i-1].Tok, t.Tok) {
 			b.WriteByte(' ')
 		}
-		b.WriteString(caseText(t.Tok, st))
+		// Space after comma in calls: func(a, b) vs func(a,b).
+		if st.SpaceAfterCommaInCalls && i > 0 && toks[i-1].Tok.Kind == lexer.Comma {
+			// Only inside parens (caller manages this at depth > 0, but we add space
+			// when the comma is not a clause-level comma — heuristic: always add).
+			b.WriteByte(' ')
+		}
+		var prev lexer.Token
+		if i > 0 {
+			prev = toks[i-1].Tok
+		}
+		nextIsLParen := i+1 < len(toks) && toks[i+1].Tok.Kind == lexer.LParen
+		b.WriteString(caseTextCtx(t.Tok, prev, nextIsLParen, st))
 	}
 	return b.String()
 }
@@ -405,7 +514,12 @@ func dmlSplitCommas(toks []cst.Tok) [][]cst.Tok {
 // One item: kept on the same line as the keyword.
 // Multiple items: each on its own line with the configured comma style.
 func dmlColList(kwText string, items [][]cst.Tok, st config.Style) string {
-	// Filter out spurious empty items (e.g. trailing comma in source).
+	return dmlColListSelect(kwText, items, st)
+}
+
+// dmlColListSelect formats a SELECT / RETURNING column list with optional
+// align_columns and select_align_as settings.
+func dmlColListSelect(kwText string, items [][]cst.Tok, st config.Style) string {
 	var kept [][]cst.Tok
 	for _, item := range items {
 		if len(item) > 0 {
@@ -426,14 +540,23 @@ func dmlColList(kwText string, items [][]cst.Tok, st config.Style) string {
 		return kwText + " " + body
 	}
 
-	// Multiple items: one per line.
-	first := st.Indent + " " // aligns item text one column past the comma
+	// Render each item text.
+	texts := make([]string, len(items))
+	for i, item := range items {
+		texts[i] = dmlInline(item, st)
+	}
+
+	// align_columns / select_align_as: pad expressions so AS and aliases align.
+	if (st.AlignColumns || st.SelectAlignAs) && len(texts) > 1 {
+		texts = alignSelectItems(texts, st)
+	}
+
+	first := st.Indent + " "
 	cont := st.Indent + ","
 	var b strings.Builder
 	b.WriteString(kwText)
-	for i, item := range items {
+	for i, text := range texts {
 		b.WriteString(nl)
-		text := dmlInline(item, st)
 		if i == 0 || st.Commas != config.CommaLeading {
 			b.WriteString(first)
 			b.WriteString(text)
@@ -447,3 +570,136 @@ func dmlColList(kwText string, items [][]cst.Tok, st config.Style) string {
 	}
 	return b.String()
 }
+
+// dmlColListSet formats an UPDATE SET column list with optional set_align_equal.
+func dmlColListSet(kwText string, items [][]cst.Tok, st config.Style) string {
+	var kept [][]cst.Tok
+	for _, item := range items {
+		if len(item) > 0 {
+			kept = append(kept, item)
+		}
+	}
+	items = kept
+
+	nl := st.Newline
+	switch len(items) {
+	case 0:
+		return kwText
+	case 1:
+		body := dmlInline(items[0], st)
+		if body == "" {
+			return kwText
+		}
+		return kwText + " " + body
+	}
+
+	texts := make([]string, len(items))
+	for i, item := range items {
+		texts[i] = dmlInline(item, st)
+	}
+
+	// set_align_equal: pad lhs so = signs align.
+	if st.SetAlignEqual && len(texts) > 1 {
+		texts = alignSetItems(texts)
+	}
+
+	first := st.Indent + " "
+	cont := st.Indent + ","
+	var b strings.Builder
+	b.WriteString(kwText)
+	for i, text := range texts {
+		b.WriteString(nl)
+		if i == 0 || st.Commas != config.CommaLeading {
+			b.WriteString(first)
+			b.WriteString(text)
+			if st.Commas == config.CommaTrailing && i < len(items)-1 {
+				b.WriteString(",")
+			}
+		} else {
+			b.WriteString(cont)
+			b.WriteString(text)
+		}
+	}
+	return b.String()
+}
+
+// alignSelectItems pads SELECT list item expressions so that AS keywords and
+// alias names align vertically.
+func alignSelectItems(texts []string, st config.Style) []string {
+	// Split each text into (expr, " AS ", alias) or keep as-is.
+	type part struct{ expr, alias string; hasAs bool }
+	parts := make([]part, len(texts))
+	maxExpr := 0
+	for i, t := range texts {
+		// Find " AS " or " as " (case-insensitive).
+		if idx := findAsIndex(t); idx >= 0 {
+			parts[i] = part{expr: t[:idx], alias: t[idx:], hasAs: true}
+			if l := len(t[:idx]); l > maxExpr {
+				maxExpr = l
+			}
+		} else {
+			parts[i] = part{expr: t}
+			if st.AlignColumns {
+				if l := len(t); l > maxExpr {
+					maxExpr = l
+				}
+			}
+		}
+	}
+
+	out := make([]string, len(texts))
+	for i, p := range parts {
+		if !p.hasAs || maxExpr == 0 {
+			out[i] = texts[i]
+			continue
+		}
+		pad := strings.Repeat(" ", maxExpr-len(p.expr))
+		out[i] = p.expr + pad + p.alias
+	}
+	return out
+}
+
+// findAsIndex returns the byte index of " AS " (case-insensitive) in s,
+// or -1 if not present at depth 0.
+func findAsIndex(s string) int {
+	low := lowerASCII(s)
+	// Look for " as " boundary.
+	for i := 0; i < len(low)-3; i++ {
+		if low[i] == ' ' && low[i+1] == 'a' && low[i+2] == 's' && low[i+3] == ' ' {
+			return i + 1 // index of 'a'
+		}
+	}
+	return -1
+}
+
+// alignSetItems pads SET assignment lhs values so that = signs align.
+func alignSetItems(texts []string) []string {
+	maxLhs := 0
+	lhsWidths := make([]int, len(texts))
+	for i, t := range texts {
+		idx := strings.Index(t, " = ")
+		if idx < 0 {
+			idx = strings.Index(t, "=")
+		}
+		if idx >= 0 {
+			lhsWidths[i] = idx
+			if idx > maxLhs {
+				maxLhs = idx
+			}
+		}
+	}
+	if maxLhs == 0 {
+		return texts
+	}
+	out := make([]string, len(texts))
+	for i, t := range texts {
+		if lhsWidths[i] == 0 || lhsWidths[i] == maxLhs {
+			out[i] = t
+			continue
+		}
+		idx := lhsWidths[i]
+		pad := strings.Repeat(" ", maxLhs-idx)
+		out[i] = t[:idx] + pad + t[idx:]
+	}
+	return out
+}