Class JavapSectionExtractor

java.lang.Object
org.ek9lang.compiler.backend.jvm.JavapSectionExtractor
public final class JavapSectionExtractor extends Object
THE indentation-aware slicer for javap -v -p -l output — the shared helper that both BytecodeNormalizer (strip a section) and the @DEBUG directive (keep a section) build on. It exists because the lookahead strip regexes in BytecodeNormalizer ((?s)LocalVariableTable:.*?(?=\n\n|\n +|\n})}) mis-slice: javap prints LineNumberTable / LocalVariableTable / StackMapTable contiguously with no blank line between them, so a lookahead that stops at the next blank line swallows the following sibling section. Slicing by INDENTATION is robust to that (Wave-1 A.0; see docs/strategic/SPEC-agentic-debugger.md §9.2).

A "section" is a header line whose trimmed text is <Name>: (e.g. LocalVariableTable:, LineNumberTable:, SourceDebugExtension:) plus every following line indented MORE deeply than the header. The section ends at the first line whose indentation is less-than-or-equal to the header's (a sibling section, the enclosing Code:'s next item, or a method/class boundary), or at a blank line — javap never puts a blank inside a section body. A class can contain many occurrences of the same section (one LocalVariableTable per method); all are returned in source order.

  • Method Details

    • extract

      public static String extract(String javapOutput, String sectionName)
      Every occurrence of the named section (header + its indented body), in order, each terminated by a newline. Empty string when the section does not occur (e.g. no LocalVariableTable was emitted). sectionName is the header word without the colon, e.g. "LocalVariableTable".

    • occurrences

      public static List<String> occurrences(String javapOutput, String sectionName)
      The header+body text block for each occurrence of the section (no trailing newline per block).