Make snapshot-lsif command more robust

These changes are needed to make the `snapshot-lsif` work the output of
an experimental Python indexer.

- Don't crash with `NoSuchElementException` when a range has no text
  document, silently ignore it instead.
- Use `#` instead of `//` for comments in Python snapshots.
- Print the absolute path of the generated output directory.
- Report a warning when skipping a file because it's not part of the
  sourceroot.

Author: olafurpg

lsif-java/src/main/scala/com/sourcegraph/lsif_java/SemanticdbPrinters.scala

@@ -2,20 +2,26 @@ package com.sourcegraph.lsif_java
 
 import scala.jdk.CollectionConverters._
 
+import com.sourcegraph.lsif_java.commands.CommentSyntax
 import com.sourcegraph.lsif_semanticdb.SignatureFormatter
 import com.sourcegraph.lsif_semanticdb.Symtab
 import com.sourcegraph.semanticdb_javac.Semanticdb.SymbolOccurrence
 import com.sourcegraph.semanticdb_javac.Semanticdb.SymbolOccurrence.Role
 import com.sourcegraph.semanticdb_javac.Semanticdb.TextDocument
 
 object SemanticdbPrinters {
-  def printTextDocument(doc: TextDocument): String = {
+  def printTextDocument(
+      doc: TextDocument,
+      comments: CommentSyntax = CommentSyntax.default
+  ): String = {
     val occurrencesByLine = doc
       .getOccurrencesList
       .asScala
       .groupBy(_.getRange.getStartLine)
     val out = new StringBuilder()
     val symtab = new Symtab(doc)
+    val extension = doc.getUri.split("\\.").lastOption.getOrElse("")
+    val commentSyntax = comments.extensionSyntax(extension)
     doc
       .getText
       .linesWithSeparators
@@ -32,7 +38,7 @@ object SemanticdbPrinters {
             )
           )
         occurrences.foreach { occ =>
-          formatOccurrence(out, occ, line, symtab)
+          formatOccurrence(out, occ, line, symtab, commentSyntax)
         }
       }
     out.toString()
@@ -42,7 +48,8 @@ object SemanticdbPrinters {
       out: StringBuilder,
       occ: SymbolOccurrence,
       line: String,
-      symtab: Symtab
+      symtab: Symtab,
+      commentSyntax: String
   ): Unit = {
     val r = occ.getRange
     val isMultiline = r.getStartLine != r.getEndLine
@@ -54,10 +61,12 @@ object SemanticdbPrinters {
       }
     out
       .append(
-        if (r.getStartCharacter > 2)
-          "// " + " " * (r.getStartCharacter - 3)
+        if (r.getStartCharacter > commentSyntax.length)
+          s"$commentSyntax " +
+            " " *
+            (r.getStartCharacter - commentSyntax.length - 1)
         else
-          "//"
+          commentSyntax
       )
       .append(
         if (r.getStartCharacter == 1)

lsif-java/src/main/scala/com/sourcegraph/lsif_java/commands/CommentSyntax.scala

@@ -0,0 +1,18 @@
+package com.sourcegraph.lsif_java.commands
+
+case class CommentSyntax(value: String) {
+  private val map =
+    value
+      .split("\\s+")
+      .map(_.split(","))
+      .collect { case Array(a, b) =>
+        a -> b
+      }
+      .toMap
+  def extensionSyntax(fileExtension: String): String =
+    map.getOrElse(fileExtension, "//")
+}
+object CommentSyntax {
+  val default = CommentSyntax("py,# sql,-- yaml,# yml,#")
+  implicit val codec = moped.macros.deriveCodec(default)
+}

lsif-java/src/main/scala/com/sourcegraph/lsif_java/commands/SnapshotCommand.scala

@@ -82,8 +82,12 @@ case class SnapshotCommand(
 }
 
 object SnapshotCommand {
-  def writeSnapshot(doc: TextDocument, outputDirectory: Path): Unit = {
-    val document = SemanticdbPrinters.printTextDocument(doc)
+  def writeSnapshot(
+      doc: TextDocument,
+      outputDirectory: Path,
+      commentSyntax: CommentSyntax = CommentSyntax.default
+  ): Unit = {
+    val document = SemanticdbPrinters.printTextDocument(doc, commentSyntax)
     val snapshotOutput = outputDirectory.resolve(doc.getUri)
     Files.createDirectories(snapshotOutput.getParent)
     Files.write(snapshotOutput, document.getBytes(StandardCharsets.UTF_8))

lsif-java/src/main/scala/com/sourcegraph/lsif_java/commands/SnapshotLsifCommand.scala

@@ -33,6 +33,8 @@ import com.sourcegraph.semanticdb_javac.Semanticdb.SymbolOccurrence
 import com.sourcegraph.semanticdb_javac.Semanticdb.SymbolOccurrence.Role
 import com.sourcegraph.semanticdb_javac.Semanticdb.TextDocument
 import moped.annotations.CommandName
+import moped.annotations.Description
+import moped.annotations.ExampleValue
 import moped.annotations.Inline
 import moped.annotations.PositionalArguments
 import moped.cli.Application
@@ -46,6 +48,12 @@ import org.scalameta.ascii.layout.prefs.LayoutPrefsImpl
 case class SnapshotLsifCommand(
     @Inline() app: Application = Application.default,
     output: Path = Paths.get("generated"),
+    @Description(
+      "Space-separated list of '$FILE_EXTENSION,$COMMENT_SYNTAX' that determines what syntax " +
+        "to use for comments depending on the file extension of the source file."
+    )
+    @ExampleValue("py,# scala,// ") commentSyntax: CommentSyntax =
+      CommentSyntax.default,
     @PositionalArguments() input: List[Path] = List(Paths.get("dump.lsif"))
 ) extends Command {
 
@@ -63,9 +71,14 @@ case class SnapshotLsifCommand(
         .of(Paths.get(doc.getUri), sourceroot)
         .normalize()
       if (docPath.toAbsolutePath.startsWith(sourceroot)) {
-        SnapshotCommand.writeSnapshot(doc, finalOutput)
+        SnapshotCommand.writeSnapshot(doc, finalOutput, commentSyntax)
+      } else {
+        app.warning(
+          s"skipping path '$docPath' because it is not part of the sourceroot '$sourceroot'"
+        )
       }
     }
+    app.info(finalOutput.toString())
     0
   }
 
@@ -85,6 +98,8 @@ object SnapshotLsifCommand {
     val lsif = new IndexedLsif(input, objects, sourceroot)
     lsif
       .ranges
+      .iterator
+      .filter(o => lsif.contains.contains(o.getId))
       .foreach { o =>
         val docId = lsif.contains(o.getId)
         val doc = lsif.textDocument(docId)