chore(release): bump version to 0.0.6

Reviewed-on: #2

AGENTS.md

@@ -36,7 +36,7 @@ pkg/pgast/         — go-pgquery wrapper: SQL → real PG AST (lint, v2)
 pkg/lint/          — rule engine + rule packs (v2)
 pkg/lsp/           — LSP server (v3)
 editors/vscode/    — VSCode extension (v3)
-editors/datagrip/  — LSP4IJ integration (v4)
+editors/datagrip/  — native JetBrains plugin, shells out to the CLI (v4)
 testdata/corpus/   — real-world .pgsql procedures used as the safety/idempotence harness
 ```
 
@@ -111,8 +111,9 @@ go fmt ./...             # Format Go code
   Per-platform VSIX (`win32/linux/darwin × x64/arm64`) built in CI matrix.
 
 ### V4 — DataGrip
-- `editors/datagrip`: integrate via LSP4IJ (free, works across JetBrains editions incl.
+- `editors/datagrip`: native JetBrains plugin (no LSP4IJ dependency) that shells out to the
-  DataGrip). No core changes expected.
+  `pgtidy` binary directly. Formatting and version info run `pgtidy fmt`/`version`; lint
+  diagnostics and quick-fixes run via a native `ExternalAnnotator` calling `pgtidy lint --json`.
 
 ## Verification
 - **Formatter:** `go test ./...` runs golden-file tests + corpus harness asserting idempotence

cmd/pgtidy/lint.go

@@ -1,6 +1,7 @@
 package main
 
 import (
+	"encoding/json"
 	"fmt"
 	"io"
 	"os"
@@ -18,9 +19,10 @@ import (
 //	file:line:col: [RULEID] severity: message
 func cmdLint(args []string, stdin io.Reader, stdout, stderr io.Writer) int {
 	var (
-		only  []string // --only=ID,ID rule filter
+		only       []string // --only=ID,ID rule filter
-		fix   bool
+		fix        bool
-		files []string
+		jsonOutput bool
+		files      []string
 	)
 	for _, a := range args {
 		switch {
@@ -29,6 +31,8 @@ func cmdLint(args []string, stdin io.Reader, stdout, stderr io.Writer) int {
 			return 0
 		case a == "--fix":
 			fix = true
+		case a == "--json":
+			jsonOutput = true
 		case strings.HasPrefix(a, "--only="):
 			ids := strings.Split(strings.TrimPrefix(a, "--only="), ",")
 			for _, id := range ids {
@@ -66,7 +70,13 @@ func cmdLint(args []string, stdin io.Reader, stdout, stderr io.Writer) int {
 		return diags, nil
 	}
 
-	printDiags := func(diags []diagnostics.Diagnostic) {
+	var jsonDiags []diagnostics.Diagnostic
+
+	reportDiags := func(diags []diagnostics.Diagnostic) {
+		if jsonOutput {
+			jsonDiags = append(jsonDiags, diags...)
+			return
+		}
 		for _, d := range diags {
 			loc := d.File
 			if d.Line > 0 {
@@ -92,14 +102,14 @@ func cmdLint(args []string, stdin io.Reader, stdout, stderr io.Writer) int {
 			_, _ = fmt.Fprintf(stderr, "pgtidy: %v\n", err)
 			return 2
 		}
-		if fix {
+		if fix && !jsonOutput {
 			fixed := lint.ApplyFixes(string(src), diags)
 			if fixed != string(src) {
 				_, _ = io.WriteString(stdout, fixed)
 				return 0
 			}
 		}
-		printDiags(diags)
+		reportDiags(diags)
 		if len(diags) > 0 {
 			found = true
 		}
@@ -115,7 +125,7 @@ func cmdLint(args []string, stdin io.Reader, stdout, stderr io.Writer) int {
 				_, _ = fmt.Fprintf(stderr, "pgtidy: %v\n", err)
 				return 2
 			}
-			if fix {
+			if fix && !jsonOutput {
 				fixed := lint.ApplyFixes(string(src), diags)
 				if fixed != string(src) {
 					if err := os.WriteFile(path, []byte(fixed), 0o644); err != nil {
@@ -130,13 +140,24 @@ func cmdLint(args []string, stdin io.Reader, stdout, stderr io.Writer) int {
 					}
 				}
 			}
-			printDiags(diags)
+			reportDiags(diags)
 			if len(diags) > 0 {
 				found = true
 			}
 		}
 	}
 
+	if jsonOutput {
+		if jsonDiags == nil {
+			jsonDiags = []diagnostics.Diagnostic{}
+		}
+		enc := json.NewEncoder(stdout)
+		if err := enc.Encode(jsonDiags); err != nil {
+			_, _ = fmt.Fprintf(stderr, "pgtidy: encoding json: %v\n", err)
+			return 2
+		}
+	}
+
 	if found {
 		return 1
 	}
@@ -150,6 +171,7 @@ Read SQL from files (or stdin) and report lint findings.
 
 Flags:
   --fix           Apply autofixes for fixable rules (MIG001, MIG003) and rewrite files
+  --json          Emit findings as a JSON array instead of text (ignores --fix)
   --only=ID,...   comma-separated rule IDs to enable (default: all rules)
   -h, --help      show this help
 

editors/datagrip/README.md

@@ -1,37 +1,27 @@
 # PgTidy — DataGrip / JetBrains Plugin
 
-Formats and lints SQL files in DataGrip (and any JetBrains IDE) via the LSP4IJ plugin.
+Formats SQL files in DataGrip (and any JetBrains IDE) by shelling out to the `pgtidy` binary.
 
 ## Requirements
 
 - `pgtidy` binary on `PATH` — download from [releases](https://git.warky.dev/wdevs/PgTidy/releases) or build with `go install git.warky.dev/wdevs/pgtidy/cmd/pgtidy@latest`
-- [LSP4IJ](https://plugins.jetbrains.com/plugin/23257-lsp4ij) plugin installed (free, by Red Hat)
 - DataGrip 2024.3+ (or any JetBrains IDE 2024.3+)
 
 ## Install the plugin
 
-### Option A — Install from disk (`.zip`)
-
 1. Build: `./gradlew buildPlugin` (output in `build/distributions/`)
 2. In DataGrip: **Settings → Plugins → ⚙ → Install Plugin from Disk…** → select the `.zip`
 3. Restart the IDE
 
-### Option B — Install LSP4IJ and configure manually (no plugin build needed)
-
-1. Install **LSP4IJ** from the marketplace (**Settings → Plugins → Marketplace → search "LSP4IJ"**)
-2. Go to **Settings → Language Servers → + (Add)**
-3. Fill in:
-   - **Name:** `PgTidy`
-   - **Command:** `pgtidy lsp`
-4. Under **Mappings**, add file patterns: `*.sql`, `*.pgsql`
-5. Click **OK** and restart the IDE
-
 ## Features
 
-- **Formatting** — `Code → Reformat Code` (`Ctrl+Alt+L`) formats the current SQL file
+- **Formatting** — `Format with PgTidy` (`Ctrl+Alt+Shift+P` or right-click in the editor) formats the current SQL file via `pgtidy fmt`
-- **Diagnostics** — lint findings shown as inspections (MIG001–MIG003, COR001–003, NAM001–003)
+- **Version check** — `Tools → Show PgTidy Version` and the status bar widget show the installed `pgtidy` version
+- **Diagnostics** — lint findings (MIG001–MIG003, COR001–003, NAM001–003) are shown inline as editor annotations, via a native `ExternalAnnotator` that runs `pgtidy lint --json`
 - **Quick fixes** — intention actions for MIG001 (add `CONCURRENTLY`) and MIG003 (add `NOT VALID`)
 
 ## Troubleshooting
 
-**Server not starting:** Check **View → Tool Windows → LSP4IJ Consoles** for stderr output. Most common cause: `pgtidy` not found on `PATH` — update the command to the full path, e.g. `/usr/local/bin/pgtidy lsp`.
+**"Cannot start pgtidy" error:** the `pgtidy` binary isn't on `PATH` for the IDE process. Install it or restart the IDE after adding it to `PATH`.
+
+**No diagnostics showing:** diagnostics only run on files with a `.sql` or `.pgsql` extension; check `pgtidy lint --json` runs cleanly on the file from a terminal with the same `PATH` as the IDE.

editors/datagrip/build.gradle.kts

@@ -23,7 +23,6 @@ dependencies {
         val platformType    = providers.gradleProperty("platformType")
         create(platformType, platformVersion)
         bundledPlugin("com.intellij.database")
-        plugin("com.redhat.devtools.lsp4ij:${providers.gradleProperty("lsp4ijVersion").get()}")
     }
 }
 

editors/datagrip/gradle.properties

@@ -5,4 +5,3 @@ pluginSinceBuild=243
 # DataGrip 2024.3
 platformVersion=2024.3
 platformType=DB
-lsp4ijVersion=0.20.1

editors/datagrip/src/main/kotlin/com/pgtidy/datagrip/PgTidyDiagnostic.kt

@@ -0,0 +1,18 @@
+package com.pgtidy.datagrip
+
+data class PgTidyFix(
+    val offset: Int,
+    val end: Int,
+    val new: String,
+    val title: String,
+)
+
+data class PgTidyDiagnostic(
+    val ruleId: String,
+    val severity: String,
+    val message: String,
+    val file: String,
+    val line: Int,
+    val col: Int,
+    val fix: PgTidyFix?,
+)

editors/datagrip/src/main/kotlin/com/pgtidy/datagrip/PgTidyExternalAnnotator.kt

@@ -0,0 +1,50 @@
+package com.pgtidy.datagrip
+
+import com.intellij.lang.annotation.AnnotationHolder
+import com.intellij.lang.annotation.ExternalAnnotator
+import com.intellij.lang.annotation.HighlightSeverity
+import com.intellij.openapi.editor.Document
+import com.intellij.psi.PsiFile
+
+class PgTidyExternalAnnotator : ExternalAnnotator<String, List<PgTidyDiagnostic>>() {
+
+    override fun collectInformation(file: PsiFile): String? {
+        val ext = file.virtualFile?.extension?.lowercase()
+        if (ext != "sql" && ext != "pgsql") return null
+        return file.viewProvider.document?.text
+    }
+
+    override fun doAnnotate(source: String): List<PgTidyDiagnostic> = PgTidyLintRunner.run(source)
+
+    override fun apply(file: PsiFile, diagnostics: List<PgTidyDiagnostic>, holder: AnnotationHolder) {
+        val document = file.viewProvider.document ?: return
+        for (d in diagnostics) {
+            if (d.line <= 0 || d.line > document.lineCount) continue
+            val range = lineTailRange(document, d.line, d.col) ?: continue
+
+            val severity = when (d.severity) {
+                "error" -> HighlightSeverity.ERROR
+                "warning" -> HighlightSeverity.WARNING
+                else -> HighlightSeverity.WEAK_WARNING
+            }
+
+            val builder = holder.newAnnotation(severity, "[${d.ruleId}] ${d.message}").range(range)
+            if (d.fix != null) {
+                builder.withFix(PgTidyQuickFix(d.fix))
+            }
+            builder.create()
+        }
+    }
+
+    /** Byte column [col] (1-based) to end-of-line, converted to a document TextRange. */
+    private fun lineTailRange(document: Document, line1Based: Int, col1Based: Int): com.intellij.openapi.util.TextRange? {
+        val lineIdx = line1Based - 1
+        val lineStart = document.getLineStartOffset(lineIdx)
+        val lineEnd = document.getLineEndOffset(lineIdx)
+        val lineText = document.getText(com.intellij.openapi.util.TextRange(lineStart, lineEnd))
+        val charCol = PgTidyOffsets.byteOffsetToCharIndex(lineText, col1Based - 1)
+        val start = lineStart + charCol
+        if (start >= lineEnd) return null
+        return com.intellij.openapi.util.TextRange(start, lineEnd)
+    }
+}

editors/datagrip/src/main/kotlin/com/pgtidy/datagrip/PgTidyLintRunner.kt

@@ -0,0 +1,32 @@
+package com.pgtidy.datagrip
+
+import com.google.gson.Gson
+import com.google.gson.reflect.TypeToken
+
+object PgTidyLintRunner {
+    private val gson = Gson()
+    private val listType = TypeToken.getParameterized(List::class.java, PgTidyDiagnostic::class.java).type
+
+    /** Runs `pgtidy lint --json` over [source]. Returns an empty list if pgtidy is missing or output is unparseable. */
+    fun run(source: String): List<PgTidyDiagnostic> {
+        val proc = try {
+            ProcessBuilder("pgtidy", "lint", "--json")
+                .redirectErrorStream(false)
+                .start()
+        } catch (ex: Exception) {
+            return emptyList()
+        }
+
+        proc.outputStream.bufferedWriter().use { it.write(source) }
+        val output = proc.inputStream.bufferedReader().readText()
+        proc.errorStream.bufferedReader().readText()
+        proc.waitFor()
+
+        if (output.isBlank()) return emptyList()
+        return try {
+            gson.fromJson<List<PgTidyDiagnostic>>(output, listType) ?: emptyList()
+        } catch (ex: Exception) {
+            emptyList()
+        }
+    }
+}

editors/datagrip/src/main/kotlin/com/pgtidy/datagrip/PgTidyOffsets.kt

@@ -0,0 +1,28 @@
+package com.pgtidy.datagrip
+
+/**
+ * pgtidy reports positions as UTF-8 byte offsets (Go strings are byte slices);
+ * IntelliJ documents index text as UTF-16 chars. Converts one to the other.
+ */
+object PgTidyOffsets {
+    fun byteOffsetToCharIndex(text: String, byteOffset: Int): Int {
+        var bytes = 0
+        var i = 0
+        while (i < text.length) {
+            val cp = text.codePointAt(i)
+            val charCount = Character.charCount(cp)
+            val byteLen = utf8Length(cp)
+            if (bytes + byteLen > byteOffset) return i
+            bytes += byteLen
+            i += charCount
+        }
+        return text.length
+    }
+
+    private fun utf8Length(codePoint: Int): Int = when {
+        codePoint <= 0x7F -> 1
+        codePoint <= 0x7FF -> 2
+        codePoint <= 0xFFFF -> 3
+        else -> 4
+    }
+}

editors/datagrip/src/main/kotlin/com/pgtidy/datagrip/PgTidyQuickFix.kt

@@ -0,0 +1,25 @@
+package com.pgtidy.datagrip
+
+import com.intellij.codeInsight.intention.IntentionAction
+import com.intellij.openapi.editor.Editor
+import com.intellij.openapi.project.Project
+import com.intellij.psi.PsiFile
+import com.intellij.util.IncorrectOperationException
+
+class PgTidyQuickFix(private val fix: PgTidyFix) : IntentionAction {
+
+    override fun getText(): String = fix.title
+    override fun getFamilyName(): String = "PgTidy"
+    override fun startInWriteAction(): Boolean = true
+
+    override fun isAvailable(project: Project, editor: Editor?, file: PsiFile?): Boolean = editor != null
+
+    @Throws(IncorrectOperationException::class)
+    override fun invoke(project: Project, editor: Editor?, file: PsiFile?) {
+        val document = editor?.document ?: return
+        val text = document.text
+        val start = PgTidyOffsets.byteOffsetToCharIndex(text, fix.offset)
+        val end = PgTidyOffsets.byteOffsetToCharIndex(text, fix.end)
+        document.replaceString(start, end, fix.new)
+    }
+}

editors/datagrip/src/main/kotlin/com/pgtidy/datagrip/PgTidyServerConnection.kt

@@ -1,9 +0,0 @@
-package com.pgtidy.datagrip
-
-import com.intellij.openapi.project.Project
-import com.redhat.devtools.lsp4ij.server.ProcessStreamConnectionProvider
-
-class PgTidyServerConnection(project: Project) : ProcessStreamConnectionProvider(
-    listOf("pgtidy", "lsp"),
-    project.basePath ?: System.getProperty("user.home"),
-)

editors/datagrip/src/main/kotlin/com/pgtidy/datagrip/PgTidyServerFactory.kt

@@ -1,10 +0,0 @@
-package com.pgtidy.datagrip
-
-import com.intellij.openapi.project.Project
-import com.redhat.devtools.lsp4ij.LanguageServerFactory
-import com.redhat.devtools.lsp4ij.server.StreamConnectionProvider
-
-class PgTidyServerFactory : LanguageServerFactory {
-    override fun createConnectionProvider(project: Project): StreamConnectionProvider =
-        PgTidyServerConnection(project)
-}

editors/datagrip/src/main/resources/META-INF/plugin.xml

@@ -5,13 +5,11 @@
     <vendor url="https://git.warky.dev/wdevs/PgTidy">Warky Devs</vendor>
     <description><![CDATA[
         PostgreSQL formatter and linter powered by <a href="https://github.com/hein/pgtidy">pgtidy</a>.<br/>
-        Requires the <code>pgtidy</code> binary on PATH and the
+        Requires the <code>pgtidy</code> binary on PATH.
-        <a href="https://plugins.jetbrains.com/plugin/23257-lsp4ij">LSP4IJ</a> plugin.
     ]]></description>
 
     <depends>com.intellij.modules.platform</depends>
     <depends>com.intellij.database</depends>
-    <depends>com.redhat.devtools.lsp4ij</depends>
 
     <actions>
         <action id="com.pgtidy.FormatDocument"
@@ -33,17 +31,7 @@
         <statusBarWidgetFactory id="com.pgtidy.StatusBarWidget"
                                  implementation="com.pgtidy.datagrip.PgTidyStatusBarWidgetFactory"
                                  order="last"/>
+        <externalAnnotator language="" implementationClass="com.pgtidy.datagrip.PgTidyExternalAnnotator"/>
     </extensions>
 
-    <extensions defaultExtensionNs="com.redhat.devtools.lsp4ij">
-        <server id="com.pgtidy.lsp"
-                name="PgTidy"
-                factoryClass="com.pgtidy.datagrip.PgTidyServerFactory">
-            <description>PostgreSQL formatter and linter (pgtidy lsp)</description>
-        </server>
-
-        <fileNamePatternMapping patterns="*.sql;*.pgsql"
-                                serverId="com.pgtidy.lsp"
-                                languageId="SQL"/>
-    </extensions>
 </idea-plugin>

linux/arch/PKGBUILD

@@ -1,6 +1,6 @@
 # Maintainer: Hein (Warky Devs) <hein@warky.dev>
 pkgname=pgtidy-bin
-pkgver=0.0.4
+pkgver=0.0.6
 pkgrel=1
 pkgdesc="PostgreSQL SQL formatter and linter"
 arch=('x86_64' 'aarch64')

linux/centos/pgtidy.spec

@@ -1,5 +1,5 @@
 Name:           pgtidy
-Version:        0.0.4
+Version:        0.0.6
 Release:        1%{?dist}
 Summary:        PostgreSQL SQL formatter and linter
 

pkg/diagnostics/diagnostics.go

@@ -15,27 +15,27 @@ const (
 // TextFix is a byte-range replacement that can be applied to the source SQL.
 // Replace src[Offset:End] with New. An insertion has Offset == End.
 type TextFix struct {
-	Offset int    // byte offset in source (inclusive)
+	Offset int    `json:"offset"` // byte offset in source (inclusive)
-	End    int    // byte offset in source (exclusive)
+	End    int    `json:"end"`    // byte offset in source (exclusive)
-	New    string // replacement text
+	New    string `json:"new"`    // replacement text
-	Title  string // short description shown in editor UI
+	Title  string `json:"title"`  // short description shown in editor UI
 }
 
 // Diagnostic is a single lint finding.
 type Diagnostic struct {
 	// RuleID is the stable identifier for the rule that produced this finding
 	// (e.g. "MIG001").
-	RuleID string
+	RuleID string `json:"ruleId"`
 	// Severity is the urgency level.
-	Severity Severity
+	Severity Severity `json:"severity"`
 	// Message is a human-readable description of the finding.
-	Message string
+	Message string `json:"message"`
 	// File is the path to the source file, or "" for stdin.
-	File string
+	File string `json:"file"`
 	// Line is the 1-based line number of the finding.
-	Line int
+	Line int `json:"line"`
 	// Col is the 1-based column number of the finding.
-	Col int
+	Col int `json:"col"`
 	// Fix is non-nil when an autofix is available for this diagnostic.
-	Fix *TextFix
+	Fix *TextFix `json:"fix,omitempty"`
 }

pkg/format/body.go

@@ -441,13 +441,9 @@ func formatBodyStatements(text string, st config.Style) string {
 			stmtLines = joinThenToCondition(stmt)
 		}
 
-		for i, ll := range stmtLines {
+		formattedLines := formatBodyStmtLines(stmtLines, baseIndent, st)
-			if i == 0 || ll.indent == "" {
+		for _, line := range formattedLines {
-				result.WriteString(baseIndent)
+			result.WriteString(line)
-			} else {
-				result.WriteString(ll.indent)
-			}
-			result.WriteString(ll.text)
 			result.WriteString(nl)
 		}
 
@@ -574,6 +570,131 @@ func formatBodyStatements(text string, st config.Style) string {
 	return result.String()
 }
 
+// formatBodyStmtLines formats one flushed PL/pgSQL statement at its contextual
+// base indent. Multi-line UPDATE/DELETE statements inside PL/pgSQL get their
+// top-level SET/WHERE/AND/OR clauses realigned under the statement while nested
+// subqueries keep their original indentation. Non-DML statements keep
+// continuation indentation, except that standalone structural keywords such as
+// THEN are aligned with the block opener.
+func formatBodyStmtLines(lines []bline, baseIndent string, st config.Style) []string {
+	if len(lines) == 0 {
+		return nil
+	}
+
+	if looksLikeMultiLineBodyDML(lines) {
+		return reindentBodyDML(lines, baseIndent, st)
+	}
+
+	out := make([]string, 0, len(lines))
+	for i, ll := range lines {
+		text := ll.text
+		indent := baseIndent
+		if i > 0 && ll.indent != "" && !isStandaloneBodyKeyword(ll.text, "then", "else", "elsif", "elseif") {
+			indent = ll.indent
+		}
+		out = append(out, indent+text)
+	}
+	return out
+}
+
+func looksLikeMultiLineBodyDML(lines []bline) bool {
+	if len(lines) < 2 {
+		return false
+	}
+	kw := lowerASCII(firstBodyKeyword(lines[0].text))
+	return kw == "update" || kw == "delete"
+}
+
+func reindentBodyDML(lines []bline, baseIndent string, st config.Style) []string {
+	out := make([]string, 0, len(lines)+1)
+	afterWhere := false
+	parenDepth := 0
+	for i, ll := range lines {
+		text := strings.TrimRight(ll.text, " 	")
+		lineDepth := parenDepth
+		kw := lowerASCII(firstBodyKeyword(text))
+		if afterWhere && lineDepth == 0 && kw != "and" && kw != "or" {
+			out = append(out, baseIndent+st.Indent+st.Indent+strings.TrimSpace(text))
+			afterWhere = false
+			updateBodyParenDepth(text, &parenDepth)
+			continue
+		}
+		if lineDepth == 0 && (kw == "set" || kw == "where" || kw == "values" || kw == "returning") {
+			if kw == "where" {
+				whereText := strings.TrimSpace(text)
+				fields := strings.Fields(whereText)
+				nextKw := ""
+				if i+1 < len(lines) {
+					nextKw = lowerASCII(firstBodyKeyword(lines[i+1].text))
+				}
+				if len(fields) > 1 && (nextKw == "and" || nextKw == "or") {
+					out = append(out, baseIndent+fields[0])
+					out = append(out, baseIndent+st.Indent+st.Indent+strings.TrimSpace(whereText[len(fields[0]):]))
+					afterWhere = false
+					continue
+				}
+				afterWhere = len(fields) == 1
+			}
+			out = append(out, baseIndent+strings.TrimSpace(text))
+			continue
+		}
+		if lineDepth == 0 && (kw == "and" || kw == "or") {
+			out = append(out, baseIndent+st.Indent+strings.TrimSpace(text))
+			afterWhere = false
+			updateBodyParenDepth(text, &parenDepth)
+			continue
+		}
+		if i == 0 {
+			out = append(out, baseIndent+strings.TrimSpace(text))
+		} else if ll.indent != "" {
+			out = append(out, ll.indent+strings.TrimSpace(text))
+		} else {
+			out = append(out, baseIndent+strings.TrimSpace(text))
+		}
+		afterWhere = false
+		updateBodyParenDepth(text, &parenDepth)
+	}
+	return out
+}
+
+func updateBodyParenDepth(s string, depth *int) {
+	for _, tok := range lexer.Lex(s) {
+		switch tok.Kind {
+		case lexer.LParen, lexer.LBracket:
+			(*depth)++
+		case lexer.RParen, lexer.RBracket:
+			if *depth > 0 {
+				(*depth)--
+			}
+		}
+	}
+}
+
+func significantBodyTokens(s string) []cst.Tok {
+	var toks []cst.Tok
+	for _, tok := range lexer.Lex(s) {
+		if tok.IsTrivia() || tok.Kind == lexer.EOF {
+			continue
+		}
+		toks = append(toks, cst.Tok{Tok: tok})
+	}
+	return toks
+}
+
+func isStandaloneBodyKeyword(s string, kws ...string) bool {
+	toks := significantBodyTokens(s)
+	if len(toks) != 1 || toks[0].Tok.Kind != lexer.Ident {
+		return false
+	}
+	low := lowerASCII(toks[0].Tok.Text)
+	for _, kw := range kws {
+		if low == kw {
+			return true
+		}
+	}
+	return false
+}
+
 // joinThenToCondition merges a THEN line (on its own bline) into the preceding
 // condition line when plpgsql_if_then_newline is false.
 func joinThenToCondition(lines []bline) []bline {

pkg/format/format_test.go

@@ -107,6 +107,66 @@ func TestFormatMmProcBroken(t *testing.T) {
 	}
 }
 
+func TestFormatIssue1PLpgSQLIndenting(t *testing.T) {
+	src := "CREATE FUNCTION f() RETURNS void LANGUAGE plpgsql AS $$\n" +
+		"DECLARE\n" +
+		"  r_lp record;\n" +
+		"BEGIN\n" +
+		"   if r_lp.total > 0\n" +
+		"      and r_lp.totaldone >= r_lp.total\n" +
+		"    then\n" +
+		"    update core.process u\n" +
+		"      set status = 'done'\n" +
+		"      where u.rid_process = r_lp.rid_process\n" +
+		"        and nv(u.status) <> 'done';\n" +
+		"  elsif r_lp.total > 0\n" +
+		"    then\n" +
+		"    update core.process u\n" +
+		"      set status = 'open'\n" +
+		"      where u.rid_process = r_lp.rid_process\n" +
+		"        and nv(u.status) <> 'open';\n" +
+		"\n" +
+		"  end if;\n" +
+		"$$;\n"
+
+	want := "CREATE FUNCTION f(\n" +
+		")\n" +
+		"RETURNS void\n" +
+		"LANGUAGE plpgsql\n" +
+		"AS\n" +
+		"$$\n" +
+		"DECLARE\n" +
+		"  r_lp record;\n" +
+		"BEGIN\n" +
+		"  if r_lp.total > 0\n" +
+		"      and r_lp.totaldone >= r_lp.total\n" +
+		"  then\n" +
+		"    update core.process u\n" +
+		"    set status = 'done'\n" +
+		"    where\n" +
+		"        u.rid_process = r_lp.rid_process\n" +
+		"      and nv(u.status) <> 'done';\n" +
+		"  elsif r_lp.total > 0\n" +
+		"  then\n" +
+		"    update core.process u\n" +
+		"    set status = 'open'\n" +
+		"    where\n" +
+		"        u.rid_process = r_lp.rid_process\n" +
+		"      and nv(u.status) <> 'open';\n" +
+		"\n" +
+		"  end if;\n" +
+		"$$;\n"
+
+	got := format(src)
+	if got != want {
+		t.Errorf("issue #1 PL/pgSQL indenting\n--- got ---\n%s\n--- want ---\n%s", got, want)
+	}
+	checkDML(t, "issue #1 PL/pgSQL indenting", got)
+	if !semanticallyEqual(src, got) {
+		t.Errorf("issue #1 PL/pgSQL indenting changed semantics")
+	}
+}
+
 func TestCorpusIdempotentAndSafe(t *testing.T) {
 	dir := filepath.Join("..", "..", "testdata", "corpus")
 	entries, err := os.ReadDir(dir)

testdata/corpus/test_mm_proc.pgsql

@@ -632,7 +632,7 @@ BEGIN
     end if;*/
 
     if G_BENCHMARK = 1
-  then
+    then
       perform log_event(m_funcname,format('Perf Merge Replace SinceStart: %s Duration: %s', clock_timestamp() - m_start, clock_timestamp() - m_ltime),bt_enum('eventlog','local notice'));
       m_ltime = clock_timestamp();
     end if;
@@ -735,7 +735,7 @@ BEGIN
     , '\[(.*?)\]', 'ig') r(v) into m_errmsg,m_retval;
 
     if m_retval > 0
-      then
+    then
       raise exception E'The following data fields could not be found for prefix % \r\n%', ifblnk(m_data_prefix, 'Null'), m_errmsg using hint = 'in ID replace process';
     end if;
 
@@ -1136,16 +1136,16 @@ BEGIN
     --                 );
 
     if r_lp_t.table_name = any(a_inner_selected) and nv(r_lp_t.table_name ) <> ''
-      then
+    then
       raise notice 'Table used as inner table: %', r_lp_t.table_name;
       continue;
     end if;
     --raise notice 'Field:% Table:% Tag: % Type: %', r_lp_t.field_name, r_lp_t.table_name, r_lp_t.mergetag, r_lp_t.merge_type;
     if nv(r_lp_t.field_name) = '' and r_lp_t.merge_type not in (G_MTYPE_TBLROOT,G_MTYPE_SPECIAL)
-      then
+    then
 
       if G_DEBUG
-        then
+      then
         perform log_event(m_funcname,format('Blank field name on Complex merge for p_doctype=%s, p_commtype=%s, p_data_prefix=%s, p_data_rid=%s
           field_name=%s, merge_type=%s, table_name=%s
           ' ,p_doctype, p_commtype,p_data_prefix,p_data_rid
@@ -1159,7 +1159,7 @@ BEGIN
     end if;
 
     if nv(m_exec_orderstr) = '' and nv(r_lp_t.parent_order_string) <> ''
-      then
+    then
       m_exec_orderstr = r_lp_t.parent_order_string;
       --raise notice 'Applying order % by for % %.', r_lp_t.parent_order_string, r_lp_t.parent_table_name,r_lp_t.field_name;
         /*
@@ -1176,7 +1176,7 @@ BEGIN
     end if;
 
     if nv(r_lp_t.ops_string) = ''
-      then
+    then
       r_lp_t.ops_string = r_lp_t.field_name;
     end if;
 
@@ -1192,19 +1192,19 @@ BEGIN
     end loop;
 
     if r_lp_t.merge_type in (G_MTYPE_TBLFIELD, G_MTYPE_CONDFIELD)
-      then
+    then
       m_execstr = format($S$%s|| '%s"%s":' || json_build_object('value',json_agg(%s::text %s), 'type', '%s')::text %s$S$
         ,m_execstr,m_comma,r_lp_t.mergetag
         ,r_lp_t.ops_string
         , m_exec_orderstr, r_lp_t.merge_type, E'\r\n');
 
     elseif r_lp_t.merge_type = G_MTYPE_SPECIAL--special fields
-      then
+    then
       m_execstr = format($S$%s|| '%s"%s":' || json_build_object('value',%s, 'type', '%s')::text %s$S$
         ,m_execstr,m_comma,r_lp_t.mergetag,quote_literal(r_lp_t.tagvalue), r_lp_t.merge_type, E'\r\n');
       --raise notice 'Special Field: %s',r_lp_t;
     elseif r_lp.merge_type = G_MTYPE_PICTURE
-      then
+    then
       m_execstr = format($S$%s|| '%s"%s":'
           || json_build_object('value',%s::text, 'type', '%s'
           , 'w', mailmerge_specialfield('width', '%s', %s) ,'h', mailmerge_specialfield('height', '%s', %s))::text %s$S$
@@ -1212,7 +1212,7 @@ BEGIN
           ,r_lp.mergetag,quote_nullable(m_data_rid),r_lp.mergetag,quote_nullable(m_data_rid), E'\r\n');
 
     elseif nv(r_lp_t.field_name) <> ''
-      then
+    then
       m_execstr = format($S$%s|| '%s"%s":' || json_build_object('value',%s::text, 'type', '%s')::text %s$S$
           ,m_execstr,m_comma,r_lp_t.mergetag
           , r_lp_t.ops_string
@@ -1225,7 +1225,7 @@ BEGIN
     m_blankexec = format($S$%s|| '%s"%s":' || json_build_object('value','', 'type', '%s')::text %s$S$,m_blankexec,m_comma,r_lp_t.mergetag, r_lp_t.merge_type, E'\r\n');
 
     if r_lp_t.rn = 1
-      then
+    then
       --Inner level tables (2)
         --raise notice 'Begin: parent: %', r_lp_t;
       for r_lp_c in (
@@ -1345,7 +1345,7 @@ BEGIN
       end loop;
 
       if ifblnk(r_lp_t.parent_table_name,'') = ''
-        then
+      then
         m_execstr = format(E'select (''{'' %s \r\n || ''}'')::json  ;',m_execstr  );
       else
         select string_agg(s.filter_string, ' ')
@@ -1364,7 +1364,7 @@ BEGIN
       from exec_json(m_execstr, 'str json') r into m_retval,m_errmsg, m_json;
 
       if m_json is null
-        then
+      then
 
         select r.p_retval, r.p_errmsg, r.p_json - > 'str'
         from exec_json(m_execstr, 'str json') r into m_retval,m_errmsg, m_json;
@@ -1373,12 +1373,12 @@ BEGIN
       m_debug_exestr = nv(m_debug_exestr) || E'\r\n/*'|| nv(r_lp_t.parent_table_name) || ' len:' || nv(length(m_json::text)) ||E'*/ \r\n' || nv(m_execstr) || E'\r\n ';
 
       if m_json_full_complex is null
-        then
+      then
         m_json_full_complex = jsonb_build_object(r_lp_t.tblid::text,m_json);
       end if;
 
       if (m_json_full_complex->r_lp_t.tblid::text) is null
-        then
+      then
         m_json_full_complex = jsonb_set(m_json_full_complex, format('{%s}',r_lp_t.tblid)::text[], m_json::jsonb,true);
       else
         m_json_full_complex = jsonb_set(m_json_full_complex, format('{%s}',r_lp_t.tblid)::text[], _jsonb_object_cat(m_json_full_complex->r_lp_t.tblid,m_json::jsonb),true);
@@ -1399,13 +1399,13 @@ BEGIN
     end if;
 
     if nv(m_comma) = '' and length(m_execstr) > 2
-      then
+    then
       m_comma = ',';
     end if;
   end loop;
 
   if G_DEBUG
-    then
+  then
     perform pl_writefile(r_template.debugsql_filename, convert_to(m_debug_exestr,'utf8'));
   end if;