[Fix] pass-2 session glob skips legacy .html artifacts; issue #483

[Fix] Three glob-filter sites (_lmd_render_json_list in lmd_alert.sh;
      _resolve_latest_session_id and _view_session_list pass 2 in
      lmd_session.sh) used session.[0-9]* to enumerate legacy plaintext
      sessions and only excluded .tsv./.hits. variants — picking up
      stale session.N.html artifacts left by pre-on-demand-HTML code
      paths. _parse_session_metadata then slurped multi-MB HTML files
      line-by-line searching for FILE HIT LIST markers that never
      appear in HTML. Local reproducer with 3 HTML artifacts (one 23 MB,
      491k lines): maldet -e list --format json took 12.4s; after fix
      0.5s (24x). Filter now excludes *.html in all three sites.
[New] tests/31-json-report.bats: regression test 33 fabricates a
      session.NNN.html with embedded "SCAN ID:" / "STARTED:" lines
      (pre-fix _parse_session_metadata would parse these and emit as
      a report); asserts absence from output. Guards the class — same
      filter bug repeated across text + json + latest-resolution paths.
[Change] CHANGELOG + CHANGELOG.RELEASE: v2.0.1 Bug Fixes entry.

Author: rfxn

CHANGELOG

@@ -339,6 +339,13 @@ v2.0.1 | Mar 25 2026:
 [New] --json-report list: active[] and stopped[] entries now include
       started and started_epoch for consistency with reports[]; stopped[]
       also gains stopped_epoch alongside existing stopped_hr; issue #483
+[Fix] --report list / --json-report list / -e newest: pass-2 glob now
+      skips legacy session.*.html artifacts (caught pre-fix by
+      session.[0-9]* glob). _parse_session_metadata was slurping
+      multi-MB HTML files line-by-line searching for break markers that
+      never appear in HTML, producing 12s+ hangs on installs carrying
+      pre-on-demand-HTML legacy artifacts. Fix at 3 sites in lmd_alert.sh
+      + lmd_session.sh
 
 v1.6.6.1 | Feb 25 2025:
 [Fix] find_recentopts incorrectly escaping find options to the right of ( -mtime .. -ctime ); previously normalized by eval; issue #440, pr#442

CHANGELOG.RELEASE

@@ -339,3 +339,10 @@ v2.0.1 | Mar 25 2026:
 [New] --json-report list: active[] and stopped[] entries now include
       started and started_epoch for consistency with reports[]; stopped[]
       also gains stopped_epoch alongside existing stopped_hr; issue #483
+[Fix] --report list / --json-report list / -e newest: pass-2 glob now
+      skips legacy session.*.html artifacts (caught pre-fix by
+      session.[0-9]* glob). _parse_session_metadata was slurping
+      multi-MB HTML files line-by-line searching for break markers that
+      never appear in HTML, producing 12s+ hangs on installs carrying
+      pre-on-demand-HTML legacy artifacts. Fix at 3 sites in lmd_alert.sh
+      + lmd_session.sh

files/internals/lmd_alert.sh

@@ -855,7 +855,7 @@ _lmd_render_json_list() {
 	local _file
 	for _file in "$sessdir"/session.[0-9]*; do
 		[ -f "$_file" ] || continue
-		case "$_file" in *.tsv.*|*.hits.*) continue ;; esac
+		case "$_file" in *.tsv.*|*.hits.*|*.html) continue ;; esac
 		local _sid="${_file##*session.}"
 		[ -n "${_seen_ids[$_sid]:-}" ] && continue  # skip if already in index
 		# Clear vars before parsing (prevent stale data from prior iteration)

files/internals/lmd_session.sh

@@ -143,9 +143,8 @@ _session_legacy_check() {
 	local _latest _latest_file=""
 	for _latest in "$sessdir"/session.[0-9]*; do
 		[ -f "$_latest" ] || continue
-		# Skip .tsv. and .hits. variants
 		case "$_latest" in
-			*.tsv.*|*.hits.*) continue ;;
+			*.tsv.*|*.hits.*|*.html) continue ;;
 		esac
 		_latest_file="$_latest"
 	done
@@ -382,7 +381,7 @@ view_report() {
 		# Pass 2: Legacy plaintext session files (skip if TSV exists)
 		for file in "$sessdir"/session.[0-9]*; do
 			[ -f "$file" ] || continue
-			case "$file" in *.tsv.*|*.hits.*) continue ;; esac
+			case "$file" in *.tsv.*|*.hits.*|*.html) continue ;; esac
 			local _sid="${file##*session.}"
 			case "$_seen_ids" in *" $_sid"*) continue ;; esac
 			_meta=$(command grep -E "^SCAN ID|^(TOTAL )?FILES|^(TOTAL )?HITS|^(TOTAL )?CLEANED|^TIME:|^STARTED:|^ELAPSED|^PATH" "$file")

tests/31-json-report.bats

@@ -674,3 +674,32 @@ EOF
     [[ "$output" =~ \"started_epoch\":[[:space:]]+$_started_epoch ]]
     [[ "$output" =~ \"stopped_epoch\":[[:space:]]+$_stopped_epoch ]]
 }
+
+# --- Test 33: pass-2 glob skips legacy .html artifacts ---
+# Regression guard: pre-fix the session.[0-9]* glob caught stale
+# session.N.html files left behind by pre-on-demand-HTML code paths.
+# _parse_session_metadata would slurp multi-MB HTML files line-by-line
+# looking for break markers that don't exist in HTML, producing a 12s+
+# hang on installs with such artifacts. Fix: exclude *.html from pass 2
+# in lmd_alert.sh, lmd_session.sh:148 (_resolve_latest_session_id), and
+# lmd_session.sh:385 (text list pass 2).
+@test "--json-report list skips legacy session.*.html artifacts" {
+    local sessdir="$LMD_INSTALL/sess"
+    local sid="200102-0100.88882"
+    # Fabricate an HTML artifact that LOOKS like a session (has SCAN ID
+    # and STARTED lines). Pre-fix: _parse_session_metadata parses these,
+    # emits a report[] entry. Post-fix: *.html filter skips it entirely.
+    cat > "$sessdir/session.$sid.html" <<EOF
+<!DOCTYPE html><html><body>
+<p>SCAN ID: $sid</p>
+<p>STARTED: Jan 02 2020 01:00:00 +0000</p>
+<p>TOTAL FILES: 1</p>
+<p>TOTAL HITS: 0</p>
+</body></html>
+EOF
+    run timeout 10 maldet --json-report list
+    rm -f "$sessdir/session.$sid.html"
+    assert_success
+    # HTML scanid must NOT appear in reports[] — proves the glob filter caught it.
+    refute_output --partial "\"$sid\""
+}