forgedock 1.16.0 → 1.16.2

package/scripts/diff-size.sh

@@ -19,8 +19,14 @@
 #   lockfiles (package-lock.json yarn.lock pnpm-lock.yaml Cargo.lock poetry.lock uv.lock go.sum Gemfile.lock
 #   composer.lock), *.min.*, *.snap, root-anchored dist/ build/ generated/ fixtures/ vendor/, any-depth
 #   __generated__/ __snapshots__/
-# Output (stdout, key=value lines): diff_lines=N excluded_lines=M threshold=T over=true|false, then up to 10
+# Output (stdout, key=value lines): diff_lines=N excluded_lines=M threshold=T over=true|false, then the cohesion
+#   measure new_lines=N test_lines=N cohesive_lines=N cohesive=true|false, then up to 10
 #   "top=<lines> <path>" lines, largest first. threshold 0 disables the gate (over=false always).
+#   Cohesion (forge#3790) counts only non-excluded files. new_lines = lines in files ADDED by the diff;
+#   test_lines = lines in test files (a test/ tests/ __tests__/ directory segment, *.test.* *.spec.* test_* *_test.*
+#   by basename); cohesive_lines = the union (a new test file counts once). cohesive=true when diff_lines>0 and
+#   cohesive_lines*100 >= 70*diff_lines and diff_lines <= threshold*3/2 (integer math; threshold 0 -> false).
+#   Test lines are classified, never discounted from diff_lines.
 # Exit: 0 measured (even when over); 2 usage or git failure with EMPTY stdout (fail closed: unknown is not zero).
 
 set -euo pipefail
@@ -62,7 +68,21 @@ MB="$(git -C "$REPO_PATH" merge-base HEAD "origin/$BASE" 2>/dev/null)" && [ -n "
 RAW="$(git -C "$REPO_PATH" diff --cached --numstat --no-renames -z "$MB" -- | tr '\0' '\001')" \
   || die "git diff against merge base $MB failed"
 
-TOTAL=0; EXCL=0; LIST=""
+ADDED="$(git -C "$REPO_PATH" diff --cached --diff-filter=A --no-renames --name-only -z "$MB" -- | tr '\0' '\001')" \
+  || die "git diff --diff-filter=A against merge base $MB failed"
+is_added() { # path -> 0 when the diff adds the file
+  case "$ADDED_SET" in *$'\001'"$1"$'\001'*) return 0 ;; esac
+  return 1
+}
+is_test() { # path -> 0 when it is a test file (directory segments exact, name patterns on the basename)
+  local p="$1" b="${1##*/}"
+  case "/$p" in */test/*|*/tests/*|*/__tests__/*) return 0 ;; esac
+  case "$b" in *.test.*|*.spec.*|test_*|*_test.*) return 0 ;; esac
+  return 1
+}
+ADDED_SET=$'\001'"$ADDED"
+
+TOTAL=0; EXCL=0; LIST=""; NEWL=0; TESTL=0; COHL=0
 # -z numstat records are "added<TAB>deleted<TAB>path" separated by NUL (translated to \001 above).
 while IFS= read -r -d $'\001' rec; do
   [ -n "$rec" ] || continue
@@ -70,7 +90,13 @@ while IFS= read -r -d $'\001' rec; do
   case "$a" in ''|*[!0-9]*) a=0 ;; esac
   case "$d" in ''|*[!0-9]*) d=0 ;; esac
   n=$((a + d))
-  if excluded "$path"; then EXCL=$((EXCL + n)); else TOTAL=$((TOTAL + n)); LIST="${LIST}${n} ${path}"$'\n'; fi
+  if excluded "$path"; then EXCL=$((EXCL + n)); else
+    TOTAL=$((TOTAL + n)); LIST="${LIST}${n} ${path}"$'\n'
+    isn=0; ist=0
+    if is_added "$path"; then isn=1; NEWL=$((NEWL + n)); fi
+    if is_test "$path"; then ist=1; TESTL=$((TESTL + n)); fi
+    if [ "$isn" = 1 ] || [ "$ist" = 1 ]; then COHL=$((COHL + n)); fi
+  fi
 done <<< "$RAW"
 
 OVER=false
@@ -79,6 +105,13 @@ echo "diff_lines=$TOTAL"
 echo "excluded_lines=$EXCL"
 echo "threshold=$THRESHOLD"
 echo "over=$OVER"
+COHESIVE=false
+if [ "$THRESHOLD" -gt 0 ] && [ "$TOTAL" -gt 0 ] && [ $((COHL * 100)) -ge $((70 * TOTAL)) ] \
+  && [ $((TOTAL * 2)) -le $((THRESHOLD * 3)) ]; then COHESIVE=true; fi
+echo "new_lines=$NEWL"
+echo "test_lines=$TESTL"
+echo "cohesive_lines=$COHL"
+echo "cohesive=$COHESIVE"
 if [ -n "$LIST" ]; then
   printf '%s' "$LIST" | sort -t' ' -k1,1nr | head -n 10 | sed 's/^/top=/' || true
 fi

package/scripts/diff-size.test.sh

@@ -95,6 +95,53 @@ reset; gen 3 > s.txt; gen 9 > l.txt; git add -A
 OUT="$(run)"; FIRST="$(printf '%s\n' "$OUT" | sed -n 's/^top=//p' | head -n1)"
 [ "$FIRST" = "9 l.txt" ] && ok || bad "top ordering (got '$FIRST')"
 
+# cohesion (forge#3790): new files + test files >= 70% of diff_lines and diff_lines <= threshold*3/2
+shape() { # all-new fixture: $1=new module lines, $2=new test file lines
+  reset; mkdir -p src tests; gen "$1" > src/newmod.py; gen "$2" > tests/test_newmod.py; git add -A
+}
+shape 685 458; OUT="$(run --threshold 1000)"
+check "34844 shape cohesive" "$OUT" cohesive true; check "34844 shape new (test file is also new)" "$OUT" new_lines 1143
+check "34844 shape test" "$OUT" test_lines 458; check "34844 shape cohesive_lines" "$OUT" cohesive_lines 1143
+shape 471 593; OUT="$(run --threshold 1000)"; check "35124 shape cohesive" "$OUT" cohesive true
+shape 516 697; OUT="$(run --threshold 1000)"; check "35143 shape cohesive" "$OUT" cohesive true
+shape 1478 0; OUT="$(run --threshold 1000)"; check "35100 all-new at 1.5x cohesive" "$OUT" cohesive true
+shape 1500 1; OUT="$(run --threshold 1000)"; check "over 1.5x not cohesive" "$OUT" cohesive false
+shape 1000 0; OUT="$(run --threshold 0)"; check "threshold 0 not cohesive" "$OUT" cohesive false
+
+# new test file counts once (union), under the threshold still reports cohesive
+reset; mkdir -p tests; gen 50 > tests/test_x.py; git add -A
+OUT="$(run --threshold 1000)"; check "new test file new" "$OUT" new_lines 50; check "new test file test" "$OUT" test_lines 50
+check "new test file union once" "$OUT" cohesive_lines 50
+
+# a modified (not added) test file counts as test, not new
+reset; mkdir -p tests; gen 40 > tests/test_old.py
+git add -A; git commit -qm t; git push -q origin feat:staging 2>/dev/null
+gen 10 | sed 's/line/chg/' > tests/test_old.py; gen 30 > src_new.py; git add -A
+OUT="$(run)"; check "modified test new_lines" "$OUT" new_lines 30; check "modified test test_lines" "$OUT" test_lines 50
+check "modified test union" "$OUT" cohesive_lines 80
+git reset -q --hard origin/staging
+
+# sprawl: 1300 lines across 15 modified files is not cohesive
+reset; for i in $(seq 1 15); do gen 50 > "m$i.txt"; done; git add -A; git commit -qm m; git push -q origin feat:staging 2>/dev/null
+for i in $(seq 1 15); do gen 43 | sed 's/line/edit/' > "m$i.txt"; done; git add -A
+OUT="$(run --threshold 1000)"
+check "sprawl lines" "$OUT" diff_lines 1395; check "sprawl cohesive_lines" "$OUT" cohesive_lines 0; check "sprawl not cohesive" "$OUT" cohesive false
+git reset -q --hard origin/staging
+
+# modified (not added) test file is test-classified; modified non-test stays out; basename patterns, no substring match
+reset; mkdir -p t; gen 10 > latest.sh; gen 10 > contest.txt; gen 10 > a.spec.ts; gen 10 > b_test.go; gen 10 > test_c.py; git add -A
+OUT="$(run)"; check "latest.sh and contest.txt are new, all counted" "$OUT" new_lines 50
+check "test name patterns" "$OUT" test_lines 30
+
+# excluded files are not counted by cohesion
+reset; gen 20 > keep.py; gen 900 > package-lock.json; git add -A
+OUT="$(run --threshold 1000)"; check "excluded not in new_lines" "$OUT" new_lines 20
+
+# existing key order is unchanged: cohesion keys sit between over= and top=
+reset; gen 3 > s.txt; git add -A
+KEYS="$(run | sed -n 's/^\([a-z_]*\)=.*/\1/p' | tr '\n' ' ')"
+[ "$KEYS" = "diff_lines excluded_lines threshold over new_lines test_lines cohesive_lines cohesive top " ] && ok || bad "key order ($KEYS)"
+
 # fail closed: bad repo path, bad base, bad threshold, unknown arg, missing base
 OUT="$(bash "$S" --repo-path "$TMP/nope" --base staging 2>/dev/null)"; rc=$?
 { [ "$rc" = 2 ] && [ -z "$OUT" ]; } && ok || bad "bad repo path (rc=$rc out='$OUT')"