Add support for BEGINFILE and ENDFILE rules

[bertysentry]

Closes #457.

What changed

Jawk now parses and executes gawk's BEGINFILE / ENDFILE special patterns, together with the nextfile statement and the ERRNO / ARGIND special variables.

• Parser (AwkParser): new BEGINFILE / ENDFILE / nextfile keywords and rule ASTs. When any of these features is used, the main input loop is compiled with the per-file tuple layout from the issue (NEXT_FILE before begin_file, CONSUME_FILE_INPUT → end_file at each file's EOF, ENDFILE rules, then NEXT_FILE → no_more_input falling into the END blocks). Programs that don't use these features keep the exact previous flat layout, so there is zero overhead for existing scripts.
• Runtime (AVM, JRT, StreamInputSource): per-file stepping with gawk-compatible FILENAME, FNR = 0, cleared $0, ARGIND, and ERRNO on open failure (No such file or directory / Is a directory, synthesized deterministically cross-platform). A file that cannot be opened is non-fatal only if a BEGINFILE rule skips it with nextfile; otherwise the usual gawk-style fatal error fires. nextfile works in ordinary rules and BEGINFILE — including from user-defined functions (it unwinds the runtime/operand stacks the same way exit does) — and runs ENDFILE for successfully opened files while skipping it for failed ones, matching the vendored .ok fixtures. exit inside the hooks jumps to the END blocks.
• Restrictions, matching gawk: next is rejected inside BEGINFILE/ENDFILE and nextfile inside BEGIN/END/ENDFILE (compile time when direct, runtime when reached through a function); non-redirected getline is a fatal error inside both hooks, redirected forms are allowed.
• Opcodes: NEXT_FILE, CONSUME_FILE_INPUT, EXEC_NEXTFILE, SET_ENDFILE_ADDRESS, SET_NEXTFILE_ADDRESS, PUSH/ASSIGN_ERRNO, PUSH/ASSIGN_ARGIND — appended at the end of the Opcode enum because fromId resolves by ordinal, preserving precompiled tuple (-L) compatibility.
• POSIX mode: as in gawk, BEGINFILE and ENDFILE are not special under --posix / setPosix(true) and lex as ordinary identifiers. nextfile stays available in both modes (POSIX.1-2024 standardizes it).
• Lexer fix: a / inside a regexp bracket expression (/.*[/]/, used by beginfile1.awk) no longer terminates the literal. This also flipped the vendored gawk regexpbrack IT from error to pass.
• FNR fix (found while testing the hooks on stdin): FNR now counts records read from standard input (POSIX defines FNR as the record number in the current input file, which includes stdin) — printf 'a\nb\n' | jawk '{print FNR}' now prints 1 2 instead of 0 0. Custom InputSource implementations keep the documented isFromFilenameList() contract.

Tests

• New BeginFileEndFileTest (25 cases): execution order across files, empty files, stdin, ERRNO + nextfile skip, fatal open failures, nextfile from functions and without hooks, all restriction diagnostics, ARGIND, $0 clearing, exit in hooks, POSIX-mode non-specialness, bracket-slash regexp.
• Re-enabled the previously-skipped POSIX nextfile conformance test; added stdin-FNR conformance tests and an InputSourceTest guard for the custom-source FNR contract.
• GawkExtensionIT#test_beginfile1 now runs the vendored fixture for real and matches beginfile1.ok byte for byte (including the ERRNO lines for . and a missing file). Six beginfile2 sub-tests (2, 5, 9a, 9b, 12, 15) extract their programs and expected transcripts from the vendored beginfile2.in / .ok and pass. The remaining sub-tests rely on non-redirected getline in BEGIN/END driving the BEGINFILE/ENDFILE hooks mid-statement (or compare non-zero gawk CLI transcripts) — that reentrancy isn't reproducible in the tuple VM; the umbrella skip documents it. getfile / readfile2 need gawk's dynamic extension API (@load, get_file()), which Jawk doesn't provide; their skip reasons now say so.

Verification

• mvn verify site: BUILD SUCCESS; checkstyle / PMD / SpotBugs at 0 findings; 543 unit tests pass.
• Full compatibility-IT regression check: I ran the entire failsafe suite on a clean baseline worktree and diffed all per-test outcomes — the only changes are the 7 new beginfile tests passing and regexpbrack turning green; zero regressions across the 939 pre-existing cases, including after the FNR change.

Notes for reviewers

• Known deviation (documented in the umbrella test_beginfile2 skip): non-redirected getline in BEGIN/END reads ahead in the main input without firing BEGINFILE/ENDFILE; the loop then adopts the already-open file instead of reopening it.
• gawk 5.1.1+ semantics were chosen where versions differ: $0 and the fields are cleared in BEGINFILE.

🤖 Generated with Claude Code

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 697ebee809

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".