fix: release RUN COMMAND expression temporaries

docs/CURRENT_STATUS.md

@@ -38,7 +38,7 @@ The `FIND ... IN ... AS ...` command now releases its evaluated search value and
 
 `PUT ... KEY ... VALUE ...` now releases the evaluated map, key, and value copies after runtime storage has copied them. `python3 tools/validate_put_expression_cleanup.py` passes with 2,000 repeated string writes under ASan/UBSan/LSan. Verified 2026-09-26: `make -s all`, `./claro test` (0 failures), `./claro doctor`, `./claro validate`, the focused sanitizer validator, and `git diff --check` pass. Separate malformed-argument sanitizer smoke tests for the existing builtin-arity and inverted-range diagnostics do not report out-of-bounds access, but LeakSanitizer reports the already documented `rt_error` diagnostic-string leaks (552 bytes/14 allocations for arity cases; 79 bytes/2 allocations for the inverted-range case). This slice is runtime-verified; those error-path leaks remain a separate blocker to claiming sanitizer-clean diagnostics.
 
-The `RUN COMMAND` path now decodes POSIX `pclose()` wait status before storing `LASTEXIT`, so a child that exits with code 3 exposes `3` rather than the encoded status 768. Focused coverage is `tests/42_last_exit_code.claro`. HTTP responses now have a 1,048,576-byte cap and marker-like response bodies are preserved while extracting the final HTTP status marker. Focused coverage is `tools/validate_http_hardening.py`. Remaining memory-growth areas include other expression temporaries. Claro remains a trusted-script interpreter, not a sandbox.
+The `RUN COMMAND` path now decodes POSIX `pclose()` wait status before storing `LASTEXIT`, so a child that exits with code 3 exposes `3` rather than the encoded status 768. Focused coverage is `tests/42_last_exit_code.claro`. HTTP responses now have a 1,048,576-byte cap and marker-like response bodies are preserved while extracting the final HTTP status marker. Focused coverage is `tools/validate_http_hardening.py`. Remaining memory-growth areas include other expression temporaries. Claro remains a trusted-script interpreter, not a sandbox. `RUN COMMAND` now releases its evaluated command expression and temporary captured-output value after copying output into runtime storage. The new `tools/validate_run_command_expression_cleanup.py` first detected 12,000 bytes leaked across 2,000 command executions, then passed under ASan/UBSan/LSan after the ownership fix.
 
 `RUN COMMAND` is documented and validated as a trusted-code capability: it executes shell commands with the user's permissions and is not a sandbox. Claro does not claim untrusted-script safety or use a fragile blacklist sanitizer. Focused documentation coverage is `tools/validate_trusted_command_docs.py`.
 

src/claro.c

@@ -456,7 +456,7 @@ static void exec_line(Runtime *rt,Program *p,int *pcp,const char *raw){ char buf
   else { char *ex=substr(pcur,as); Value v=eval_expr(rt,ex); char *txt=v_to_string(v); if(!strcmp(restup,"UPPER")){ char *r=text_upper_copy(txt); rt_set(rt,var,v_str(r)); free(r); } else if(!strcmp(restup,"LOWER")){ char *r=text_lower_copy(txt); rt_set(rt,var,v_str(r)); free(r); } else if(!strcmp(restup,"TRIM")){ char *r=text_trim_copy(txt); rt_set(rt,var,v_str(r)); free(r); } else if(!strcmp(restup,"LENGTH")){ rt_set(rt,var,v_num((double)strlen(txt))); } else if(!strcmp(restup,"CONTAINS")){ char *left=NULL,*right=NULL; const char *needle=find_word_ci(pcur,"NEEDLE"); if(!needle) needle=find_word_ci(pcur,"TEXT"); if(needle){ left=substr(pcur,needle); right=substr(needle+6,as); } else { char **args=NULL; int n=split_args(pcur,&args); if(n>=2){ left=xstrdup(args[0]); right=substr(args[1],as); } split_args_free(args,n); } if(left&&right){ Value lv=eval_expr(rt,left), rv=eval_expr(rt,right); char *ls=v_to_string(lv), *rs=v_to_string(rv); rt_set(rt,var,v_bool(strstr(ls,rs)!=NULL)); free(ls); free(rs); } free(left); free(right); } free(txt); free(ex); }
   free(var); } return; }
  if(strcmp(up,"RANDOM")==0){ const char *as=find_word_ci(t,"AS"); if(as&&starts_ci(t+6," NUMBER")){ char *range=substr(t+13,as); char *var=xstrdup(trim_inplace((char*)as+2)); char **args=NULL; int n=split_args(range,&args); double a=1,b=100; if(n>=2){ a=v_number(eval_expr(rt,args[0])); b=v_number(eval_expr(rt,args[1])); } else { char *tmp=range; char *one=unquote_token((const char**)&tmp); char *two=unquote_token((const char**)&tmp); if(*one) a=v_number(eval_expr(rt,one)); if(*two) b=v_number(eval_expr(rt,two)); free(one); free(two); } split_args_free(args,n); if(b<a){ double c=a; a=b; b=c; } rt_set(rt,var,v_num((int)a + (rand()%((int)(b-a+1))))); free(range); free(var); } return; }
- if(strcmp(up,"RUN")==0){ const char *as=find_word_ci(t,"AS"); if(as&&starts_ci(t+3," COMMAND")){ char *ex=substr(t+11,as); char *var=xstrdup(trim_inplace((char*)as+2)); Value cv=eval_expr(rt,ex); char *cmd=v_to_string(cv); int code=0; char *out=capture_command_output(cmd,&code,0); rt_set(rt,var,v_str(out)); rt_set(rt,"LASTEXIT",v_num(code)); free(out); free(cmd); free(ex); free(var); } return; }
+ if(strcmp(up,"RUN")==0){ const char *as=find_word_ci(t,"AS"); if(as&&starts_ci(t+3," COMMAND")){ char *ex=substr(t+11,as); char *var=xstrdup(trim_inplace((char*)as+2)); Value cv=eval_expr(rt,ex); char *cmd=v_to_string(cv); int code=0; char *out=capture_command_output(cmd,&code,0); Value output=v_str(out); rt_set(rt,var,output); value_free(output); rt_set(rt,"LASTEXIT",v_num(code)); value_free(cv); free(out); free(cmd); free(ex); free(var); } return; }
  if(strcmp(up,"SORT")==0){ char *name=xstrdup(trim_inplace(t+4)); Value v=rt_get(rt,name); value_list_sort(v); rt_set(rt,name,v); free(name); return; }
  if(strcmp(up,"REVERSE")==0){ char *name=xstrdup(trim_inplace(t+7)); Value v=rt_get(rt,name); value_list_reverse(v); rt_set(rt,name,v); free(name); return; }
  if(strcmp(up,"FIND")==0){ const char *in=find_word_ci(t,"IN"), *as=find_word_ci(t,"AS"); if(in&&as){ char *needle=substr(t+4,in); char *listname=substr(in+2,as); char *var=xstrdup(trim_inplace((char*)as+2)); Value nv=eval_expr(rt,needle), lv=eval_expr(rt,listname); int found=0,i; if(lv.type==V_LIST){ for(i=0;i<lv.list->count;i++) if(value_compare(nv,lv.list->items[i])==0){ found=i+1; break; } } rt_set(rt,var,v_num(found)); free(needle); free(listname); free(var); value_free(nv); value_free(lv); } return; }

tools/validate_run_command_expression_cleanup.py

@@ -0,0 +1,43 @@
+#!/usr/bin/env python3
+"""Regression check for RUN COMMAND expression temporary ownership."""
+from pathlib import Path
+import os
+import subprocess
+import tempfile
+
+ROOT = Path(__file__).resolve().parent.parent
+
+
+def main() -> int:
+    if os.name == "nt":
+        print("SKIP: LeakSanitizer check is POSIX-only")
+        return 0
+    with tempfile.TemporaryDirectory(prefix="claro-run-command-cleanup-") as tmp:
+        tmp_path = Path(tmp)
+        binary = tmp_path / "claro-lsan"
+        script = tmp_path / "run_command_expressions.claro"
+        script.write_text('RUN COMMAND "true" AS output\n' * 2000, encoding="utf-8")
+        build = subprocess.run(
+            ["gcc", "-std=c99", "-O0", "-g", "-fsanitize=address,undefined",
+             "src/claro.c", "-o", str(binary), "-lm"],
+            cwd=ROOT, text=True, capture_output=True,
+        )
+        if build.returncode:
+            print(build.stdout, end="")
+            print(build.stderr, end="")
+            return build.returncode
+        run = subprocess.run(
+            [str(binary), str(script)], cwd=ROOT, text=True, capture_output=True,
+            env={**os.environ, "ASAN_OPTIONS": "detect_leaks=1"},
+        )
+        if run.returncode or "LeakSanitizer" in run.stderr or "SUMMARY:" in run.stderr:
+            print("FAIL: RUN COMMAND expression values still leak")
+            print(run.stdout, end="")
+            print(run.stderr, end="")
+            return 1
+        print("PASS: RUN COMMAND expression values are released")
+        return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())