Add rewritten awk scenario harness

.claude/skills/implement-awk/SKILL.md

@@ -74,6 +74,7 @@ Run this sequence after every coherent implementation step:
 ```bash
 make fmt
 go test ./...
+RSHELL_BIN=./rshell AWK_UNDER_TEST=tools/awk-harness/rshell-awk tools/awk-harness/run.sh rewritten
 RSHELL_BIN=./rshell AWK_UNDER_TEST=tools/awk-harness/rshell-awk tools/awk-harness/run.sh gawk
 RSHELL_BIN=./rshell AWK_UNDER_TEST=tools/awk-harness/rshell-awk tools/awk-harness/run.sh onetrueawk
 ```
@@ -104,6 +105,8 @@ make build
    - safety rejection behavior
    - runtime or resource limit
 7. Add or update original rshell tests for the intended behavior.
+   Prefer `tests/awk_scenarios` for GNU awk behavior that came from upstream
+   AWK coverage, and include upstream metadata for traceability.
 8. Implement the smallest code change that addresses the cluster.
 9. Run `make fmt`.
 10. Run focused tests.

.github/workflows/awk-harness.yml

@@ -18,6 +18,8 @@ jobs:
       - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
       - name: Install pinned GNU awk oracle
         run: tools/awk-harness/run.sh install-gawk
+      - name: Run rewritten AWK scenarios against pinned GNU awk
+        run: tools/awk-harness/run.sh rewritten
       - name: Fetch One True Awk tests
         env:
           AWK_HARNESS_BOOTSTRAP: "1"

Makefile

@@ -1,4 +1,4 @@
-.PHONY: build fmt test test_all test_against_bash compliance
+.PHONY: build fmt test test_all test_against_bash test_awk_rewritten compliance
 
 build:
 	go build -o rshell ./cmd/rshell
@@ -15,5 +15,8 @@ test_all:
 test_against_bash:
 	RSHELL_BASH_TEST=1 go test -v ./tests/ -run TestShellScenariosAgainstBash -count=1
 
+test_awk_rewritten:
+	tools/awk-harness/run.sh rewritten
+
 compliance:
 	RSHELL_COMPLIANCE_TEST=1 go test -v ./tests/ -run TestCompliance -count=1

tests/awk_scenarios/README.md

@@ -0,0 +1,39 @@
+# AWK Scenario Rewrites
+
+This directory contains rshell-owned AWK tests rewritten from upstream behavior
+coverage. Do not copy upstream test bodies, helper scripts, comments, fixtures,
+or expected output into this directory.
+
+Each scenario is a small GNU awk behavior case with metadata that identifies
+which upstream suite or coverage area it belongs to and what behavior it covers.
+The tests run through the AWK-specific Go runner in
+`tests/awk_scenarios_test.go`.
+
+`enabled.txt` is the only run list. Each non-comment line is a path relative to
+this directory:
+
+```text
+gawk/basic/begin_end_records.yaml
+onetrueawk/basic/pattern_action.yaml
+```
+
+`upstream-map.yaml` is an audit ledger for rewrite progress. It does not decide
+which tests run.
+
+Run the rewritten scenarios against the pinned GNU awk oracle:
+
+```bash
+tools/awk-harness/run.sh install-gawk
+tools/awk-harness/run.sh rewritten
+```
+
+Run the same scenarios against rshell's `awk` adapter once the builtin exists:
+
+```bash
+make build
+RSHELL_BIN=./rshell AWK_UNDER_TEST=tools/awk-harness/rshell-awk tools/awk-harness/run.sh rewritten
+```
+
+The runner still compares rshell output to the pinned GNU awk oracle when
+`GAWK_ORACLE` is set, so expected output in these files and live GNU awk
+behavior must stay aligned.

tests/awk_scenarios/enabled.txt

@@ -0,0 +1,10 @@
+# Each non-comment line is a path relative to tests/awk_scenarios.
+gawk/basic/begin_end_records.yaml
+gawk/basic/field_separator.yaml
+gawk/cli/variable_assignment.yaml
+gawk/expressions/arithmetic_comparison.yaml
+gawk/fields/assign_rebuilds_record.yaml
+gawk/functions/split.yaml
+gawk/output/print_separators.yaml
+gawk/regex/pattern_match.yaml
+onetrueawk/basic/pattern_action.yaml

tests/awk_scenarios/gawk/basic/begin_end_records.yaml

@@ -0,0 +1,25 @@
+description: BEGIN and END blocks wrap per-record actions
+upstream:
+  suite: gawk
+  id: test/beginfile1.awk
+  ref: gawk-5.4.0
+covers:
+  - BEGIN actions run before input records are processed
+  - END actions run after all input records are processed
+  - NR is incremented for each input record
+  - $1 and $NF read the first and last fields of the current record
+input:
+  program: |
+    BEGIN { print "start" }
+    { print NR ":" $1 "-" $NF }
+    END { print "count=" NR }
+  stdin: |
+    alpha beta
+    gamma delta epsilon
+expect:
+  stdout: |
+    start
+    1:alpha-beta
+    2:gamma-epsilon
+    count=2
+  exit_code: 0

tests/awk_scenarios/gawk/basic/field_separator.yaml

@@ -0,0 +1,20 @@
+description: FS controls field splitting and NF for comma-separated records
+upstream:
+  suite: gawk
+  id: test/fieldwdth.awk
+  ref: gawk-5.4.0
+covers:
+  - FS controls field splitting for subsequent records
+  - NF reflects the number of fields in the current record
+input:
+  program: |
+    BEGIN { FS = "," }
+    { print $2 ":" NF }
+  stdin: |
+    a,b,c
+    x,y
+expect:
+  stdout: |
+    b:3
+    y:2
+  exit_code: 0

tests/awk_scenarios/gawk/cli/variable_assignment.yaml

@@ -0,0 +1,30 @@
+description: -v assignments are visible before BEGIN and file arguments set FILENAME and FNR
+upstream:
+  suite: gawk
+  id: test/argtest.awk
+  ref: gawk-5.4.0
+covers:
+  - -v assignments are visible before BEGIN runs
+  - FILENAME is set to the current input file path
+  - FNR counts records within the current input file
+setup:
+  files:
+    - path: records.txt
+      content: |
+        red
+        blue
+input:
+  awk_args:
+    - -v
+    - label=color
+  program: |
+    BEGIN { print label }
+    { print FILENAME ":" FNR ":" $0 }
+  args:
+    - records.txt
+expect:
+  stdout: |
+    color
+    records.txt:1:red
+    records.txt:2:blue
+  exit_code: 0

tests/awk_scenarios/gawk/expressions/arithmetic_comparison.yaml

@@ -0,0 +1,30 @@
+description: Arithmetic and comparison expressions drive conditional actions
+upstream:
+  suite: gawk
+  id: test/compare.awk
+  ref: gawk-5.4.0
+covers:
+  - numeric addition updates an accumulator
+  - modulo participates in equality comparisons
+  - if and else choose actions from numeric comparisons
+input:
+  program: |
+    {
+      total += $1
+      if ($1 % 2 == 0)
+        print $1, "even"
+      else
+        print $1, "odd"
+    }
+    END { print "total", total }
+  stdin: |
+    3
+    4
+    9
+expect:
+  stdout: |
+    3 odd
+    4 even
+    9 odd
+    total 16
+  exit_code: 0

tests/awk_scenarios/gawk/fields/assign_rebuilds_record.yaml

@@ -0,0 +1,27 @@
+description: Assigning a numbered field rebuilds $0 using OFS
+upstream:
+  suite: gawk
+  id: test/assignnumfield.awk
+  ref: gawk-5.4.0
+covers:
+  - assigning to a numbered field changes that field
+  - rebuilding $0 after a field assignment uses OFS
+  - NF remains the number of fields when assigning an existing field
+input:
+  program: |
+    BEGIN { OFS = "|" }
+    {
+      $2 = "patched"
+      print $0
+      print NF
+    }
+  stdin: |
+    alpha beta gamma
+    solo pair
+expect:
+  stdout: |
+    alpha|patched|gamma
+    3
+    solo|patched
+    2
+  exit_code: 0

tests/awk_scenarios/gawk/functions/split.yaml

@@ -0,0 +1,23 @@
+description: split populates array elements and returns the element count
+upstream:
+  suite: gawk
+  id: test/splitargv.awk
+  ref: gawk-5.4.0
+covers:
+  - split returns the number of elements it created
+  - split stores array elements using one-based numeric indexes
+  - split accepts a string field separator argument
+input:
+  program: |
+    BEGIN {
+      n = split("north:south:east", parts, ":")
+      print n
+      for (i = 1; i <= n; i++) print i "=" parts[i]
+    }
+expect:
+  stdout: |
+    3
+    1=north
+    2=south
+    3=east
+  exit_code: 0

tests/awk_scenarios/gawk/output/print_separators.yaml

@@ -0,0 +1,20 @@
+description: print uses OFS between arguments and ORS after each record
+upstream:
+  suite: gawk
+  id: test/ofs1.awk
+  ref: gawk-5.4.0
+covers:
+  - print inserts OFS between arguments
+  - print appends ORS after the output record
+  - numeric arguments are formatted for print output
+input:
+  program: |
+    BEGIN {
+      OFS = "::"
+      ORS = "<end>\n"
+      print "left", "right", 7
+    }
+expect:
+  stdout: |
+    left::right::7<end>
+  exit_code: 0

tests/awk_scenarios/gawk/regex/pattern_match.yaml

@@ -0,0 +1,26 @@
+description: Regular expression patterns and !~ select records
+upstream:
+  suite: gawk
+  id: test/re_test.awk
+  ref: gawk-5.4.0
+covers:
+  - regex patterns select matching input records
+  - anchors constrain regex matches to the whole record
+  - !~ negates a regex match expression
+input:
+  program: |
+    /^[[:alpha:]]+[0-9]$/ { print NR ":" $0 }
+    $0 !~ /[aeiou]/ { print "no-vowel:" $0 }
+  stdin: |
+    abc1
+    sky
+    lake2
+    B7
+expect:
+  stdout: |
+    1:abc1
+    no-vowel:sky
+    3:lake2
+    4:B7
+    no-vowel:B7
+  exit_code: 0

tests/awk_scenarios/onetrueawk/basic/pattern_action.yaml

@@ -0,0 +1,25 @@
+description: Pattern-only and action-only rules compose across records
+upstream:
+  suite: onetrueawk
+  id: testdir/t.3
+  ref: 3c2e168a8f794ed61c93131b05fb998d79d155df
+  notes: Also covers the default action behavior represented by testdir/t.0.
+covers:
+  - pattern-only rules print the current record
+  - action-only rules run for every record
+  - END actions can observe accumulated values
+input:
+  program: |
+    /keep/
+    { total += $2 }
+    END { print "total=" total }
+  stdin: |
+    keep 4
+    drop 7
+    keep 2
+expect:
+  stdout: |
+    keep 4
+    keep 2
+    total=13
+  exit_code: 0