From bd2ea84028e172d245b49952fd6dc3ea9eb20c6f Mon Sep 17 00:00:00 2001
From: Juan Cruz Viotti <jv@jviotti.com>
Date: Tue, 3 Mar 2026 15:09:40 -0400
Subject: [PATCH] [WIP] Fix subpar caching usage during ingestion

Signed-off-by: Juan Cruz Viotti <jv@jviotti.com>
---
 test/cli/CMakeLists.txt                       |   2 +
 test/cli/index/common/rebuild-modify-cache.sh | 104 +++++++++
 test/cli/index/common/rebuild-two-to-three.sh | 204 ++++++++++++++++++
 3 files changed, 310 insertions(+)
 create mode 100755 test/cli/index/common/rebuild-modify-cache.sh
 create mode 100755 test/cli/index/common/rebuild-two-to-three.sh

diff --git a/test/cli/CMakeLists.txt b/test/cli/CMakeLists.txt
index 0cf210ae..ec9c729b 100644
--- a/test/cli/CMakeLists.txt
+++ b/test/cli/CMakeLists.txt
@@ -33,6 +33,8 @@ if(ONE_INDEX)
   sourcemeta_one_test_cli(common index extra-files-on-rebuild)
   sourcemeta_one_test_cli(common index extra-directories-on-rebuild)
   sourcemeta_one_test_cli(common index directory-schema-same-name)
+  sourcemeta_one_test_cli(common index rebuild-two-to-three)
+  sourcemeta_one_test_cli(common index rebuild-modify-cache)
   sourcemeta_one_test_cli(common index rebuild-cache)
   sourcemeta_one_test_cli(common index rebuild-to-empty)
   sourcemeta_one_test_cli(common index verbose-long)
diff --git a/test/cli/index/common/rebuild-modify-cache.sh b/test/cli/index/common/rebuild-modify-cache.sh
new file mode 100755
index 00000000..41cf74bf
--- /dev/null
+++ b/test/cli/index/common/rebuild-modify-cache.sh
@@ -0,0 +1,104 @@
+#!/bin/sh
+
+set -o errexit
+set -o nounset
+
+TMP="$(mktemp -d)"
+clean() { rm -rf "$TMP"; }
+trap clean EXIT
+
+cat << EOF > "$TMP/one.json"
+{
+  "url": "https://sourcemeta.com/",
+  "contents": {
+    "example": {
+      "contents": {
+        "schemas": {
+          "baseUri": "https://example.com/",
+          "path": "./schemas"
+        }
+      }
+    }
+  }
+}
+EOF
+
+mkdir "$TMP/schemas"
+
+cat << 'EOF' > "$TMP/schemas/a.json"
+{
+  "$schema": "http://json-schema.org/draft-07/schema#",
+  "$id": "https://example.com/a"
+}
+EOF
+
+remove_threads_information() {
+  expr='s/ \[[^]]*[^a-z-][^]]*\]//g'
+  if [ "$(uname -s)" = "Darwin" ]; then
+    sed -i '' "$expr" "$1"
+  else
+    sed -i "$expr" "$1"
+  fi
+}
+
+# Run 1: index one schema from scratch
+"$1" --skip-banner "$TMP/one.json" "$TMP/output" --concurrency 1 > /dev/null 2>&1
+
+# Run 2: modify the schema and re-index
+cat << 'EOF' > "$TMP/schemas/a.json"
+{
+  "$schema": "http://json-schema.org/draft-07/schema#",
+  "$id": "https://example.com/a",
+  "type": "string"
+}
+EOF
+"$1" --skip-banner "$TMP/one.json" "$TMP/output" --concurrency 1 > /dev/null 2>&1
+
+# Run 3: re-index with no changes. Everything should be fully cached.
+"$1" --skip-banner "$TMP/one.json" "$TMP/output" --concurrency 1 2> "$TMP/output.txt"
+remove_threads_information "$TMP/output.txt"
+
+cat << EOF > "$TMP/expected.txt"
+Writing output to: $(realpath "$TMP")/output
+Using configuration: $(realpath "$TMP")/one.json
+Detecting: $(realpath "$TMP")/schemas/a.json (#1)
+(100%) Resolving: a.json
+(100%) Ingesting: https://sourcemeta.com/example/schemas/a
+(skip) Ingesting: https://sourcemeta.com/example/schemas/a [materialise]
+(100%) Analysing: https://sourcemeta.com/example/schemas/a
+(skip) Analysing: https://sourcemeta.com/example/schemas/a [positions]
+(skip) Analysing: https://sourcemeta.com/example/schemas/a [locations]
+(skip) Analysing: https://sourcemeta.com/example/schemas/a [dependencies]
+(skip) Analysing: https://sourcemeta.com/example/schemas/a [stats]
+(skip) Analysing: https://sourcemeta.com/example/schemas/a [health]
+(skip) Analysing: https://sourcemeta.com/example/schemas/a [bundle]
+(skip) Analysing: https://sourcemeta.com/example/schemas/a [editor]
+(skip) Analysing: https://sourcemeta.com/example/schemas/a [blaze-exhaustive]
+(skip) Analysing: https://sourcemeta.com/example/schemas/a [blaze-fast]
+(skip) Analysing: https://sourcemeta.com/example/schemas/a [metadata]
+( 50%) Reviewing: schemas
+(100%) Reviewing: schemas
+(skip) Reviewing: schemas [dependencies]
+(100%) Reworking: https://sourcemeta.com/example/schemas/a
+(skip) Reworking: https://sourcemeta.com/example/schemas/a [dependents]
+(  0%) Producing: explorer
+(skip) Producing: explorer [search]
+( 33%) Producing: example/schemas
+(skip) Producing: example/schemas [directory]
+( 66%) Producing: example
+(skip) Producing: example [directory]
+(100%) Producing: .
+(skip) Producing: . [directory]
+( 25%) Rendering: example/schemas
+(skip) Rendering: example/schemas [directory]
+( 50%) Rendering: example
+(skip) Rendering: example [directory]
+( 75%) Rendering: .
+(skip) Rendering: . [index]
+(skip) Rendering: . [not-found]
+(100%) Rendering: example/schemas/a
+(skip) Rendering: example/schemas/a [schema]
+(skip) Producing: routes.bin [routes]
+EOF
+
+diff "$TMP/output.txt" "$TMP/expected.txt"
diff --git a/test/cli/index/common/rebuild-two-to-three.sh b/test/cli/index/common/rebuild-two-to-three.sh
new file mode 100755
index 00000000..208a1a74
--- /dev/null
+++ b/test/cli/index/common/rebuild-two-to-three.sh
@@ -0,0 +1,204 @@
+#!/bin/sh
+
+set -o errexit
+set -o nounset
+
+TMP="$(mktemp -d)"
+clean() { rm -rf "$TMP"; }
+trap clean EXIT
+
+cat << EOF > "$TMP/one.json"
+{
+  "url": "https://sourcemeta.com/",
+  "contents": {
+    "example": {
+      "contents": {
+        "schemas": {
+          "baseUri": "https://example.com/",
+          "path": "./schemas"
+        }
+      }
+    }
+  }
+}
+EOF
+
+mkdir "$TMP/schemas"
+
+cat << 'EOF' > "$TMP/schemas/a.json"
+{
+  "$schema": "http://json-schema.org/draft-07/schema#",
+  "$id": "https://example.com/a"
+}
+EOF
+
+cat << 'EOF' > "$TMP/schemas/b.json"
+{
+  "$schema": "http://json-schema.org/draft-07/schema#",
+  "$id": "https://example.com/b"
+}
+EOF
+
+remove_threads_information() {
+  expr='s/ \[[^]]*[^a-z-][^]]*\]//g'
+  if [ "$(uname -s)" = "Darwin" ]; then
+    sed -i '' "$expr" "$1"
+  else
+    sed -i "$expr" "$1"
+  fi
+}
+
+# Run 1: index two schemas from scratch
+"$1" --skip-banner "$TMP/one.json" "$TMP/output" --concurrency 1 > /dev/null 2>&1
+
+# Run 2: add a third schema and re-index
+cat << 'EOF' > "$TMP/schemas/c.json"
+{
+  "$schema": "http://json-schema.org/draft-07/schema#",
+  "$id": "https://example.com/c"
+}
+EOF
+"$1" --skip-banner "$TMP/one.json" "$TMP/output" --concurrency 1 > /dev/null 2>&1
+
+# Run 3: re-index with no changes. All three schemas should be fully cached.
+"$1" --skip-banner "$TMP/one.json" "$TMP/output" --concurrency 1 2> "$TMP/output.txt"
+remove_threads_information "$TMP/output.txt"
+grep '(skip) Ingesting:' "$TMP/output.txt" | sort > "$TMP/ingest_actual.txt"
+
+cat << 'EOF' | sort > "$TMP/ingest_expected.txt"
+(skip) Ingesting: https://sourcemeta.com/example/schemas/a [materialise]
+(skip) Ingesting: https://sourcemeta.com/example/schemas/b [materialise]
+(skip) Ingesting: https://sourcemeta.com/example/schemas/c [materialise]
+EOF
+
+diff "$TMP/ingest_actual.txt" "$TMP/ingest_expected.txt"
+
+cd "$TMP/output"
+find . -mindepth 1 | LC_ALL=C sort > "$TMP/manifest.txt"
+cd - > /dev/null
+
+cat << 'EOF' > "$TMP/expected_manifest.txt"
+./configuration.json
+./dependency-tree.metapack
+./dependency-tree.metapack.deps
+./explorer
+./explorer/%
+./explorer/%/404.metapack
+./explorer/%/404.metapack.deps
+./explorer/%/directory-html.metapack
+./explorer/%/directory-html.metapack.deps
+./explorer/%/directory.metapack
+./explorer/%/directory.metapack.deps
+./explorer/%/search.metapack
+./explorer/%/search.metapack.deps
+./explorer/example
+./explorer/example/%
+./explorer/example/%/directory-html.metapack
+./explorer/example/%/directory-html.metapack.deps
+./explorer/example/%/directory.metapack
+./explorer/example/%/directory.metapack.deps
+./explorer/example/schemas
+./explorer/example/schemas/%
+./explorer/example/schemas/%/directory-html.metapack
+./explorer/example/schemas/%/directory-html.metapack.deps
+./explorer/example/schemas/%/directory.metapack
+./explorer/example/schemas/%/directory.metapack.deps
+./explorer/example/schemas/a
+./explorer/example/schemas/a/%
+./explorer/example/schemas/a/%/schema-html.metapack
+./explorer/example/schemas/a/%/schema-html.metapack.deps
+./explorer/example/schemas/a/%/schema.metapack
+./explorer/example/schemas/a/%/schema.metapack.deps
+./explorer/example/schemas/b
+./explorer/example/schemas/b/%
+./explorer/example/schemas/b/%/schema-html.metapack
+./explorer/example/schemas/b/%/schema-html.metapack.deps
+./explorer/example/schemas/b/%/schema.metapack
+./explorer/example/schemas/b/%/schema.metapack.deps
+./explorer/example/schemas/c
+./explorer/example/schemas/c/%
+./explorer/example/schemas/c/%/schema-html.metapack
+./explorer/example/schemas/c/%/schema-html.metapack.deps
+./explorer/example/schemas/c/%/schema.metapack
+./explorer/example/schemas/c/%/schema.metapack.deps
+./routes.bin
+./routes.bin.deps
+./schemas
+./schemas/example
+./schemas/example/schemas
+./schemas/example/schemas/a
+./schemas/example/schemas/a/%
+./schemas/example/schemas/a/%/blaze-exhaustive.metapack
+./schemas/example/schemas/a/%/blaze-exhaustive.metapack.deps
+./schemas/example/schemas/a/%/blaze-fast.metapack
+./schemas/example/schemas/a/%/blaze-fast.metapack.deps
+./schemas/example/schemas/a/%/bundle.metapack
+./schemas/example/schemas/a/%/bundle.metapack.deps
+./schemas/example/schemas/a/%/dependencies.metapack
+./schemas/example/schemas/a/%/dependencies.metapack.deps
+./schemas/example/schemas/a/%/dependents.metapack
+./schemas/example/schemas/a/%/dependents.metapack.deps
+./schemas/example/schemas/a/%/editor.metapack
+./schemas/example/schemas/a/%/editor.metapack.deps
+./schemas/example/schemas/a/%/health.metapack
+./schemas/example/schemas/a/%/health.metapack.deps
+./schemas/example/schemas/a/%/locations.metapack
+./schemas/example/schemas/a/%/locations.metapack.deps
+./schemas/example/schemas/a/%/positions.metapack
+./schemas/example/schemas/a/%/positions.metapack.deps
+./schemas/example/schemas/a/%/schema.metapack
+./schemas/example/schemas/a/%/schema.metapack.deps
+./schemas/example/schemas/a/%/stats.metapack
+./schemas/example/schemas/a/%/stats.metapack.deps
+./schemas/example/schemas/b
+./schemas/example/schemas/b/%
+./schemas/example/schemas/b/%/blaze-exhaustive.metapack
+./schemas/example/schemas/b/%/blaze-exhaustive.metapack.deps
+./schemas/example/schemas/b/%/blaze-fast.metapack
+./schemas/example/schemas/b/%/blaze-fast.metapack.deps
+./schemas/example/schemas/b/%/bundle.metapack
+./schemas/example/schemas/b/%/bundle.metapack.deps
+./schemas/example/schemas/b/%/dependencies.metapack
+./schemas/example/schemas/b/%/dependencies.metapack.deps
+./schemas/example/schemas/b/%/dependents.metapack
+./schemas/example/schemas/b/%/dependents.metapack.deps
+./schemas/example/schemas/b/%/editor.metapack
+./schemas/example/schemas/b/%/editor.metapack.deps
+./schemas/example/schemas/b/%/health.metapack
+./schemas/example/schemas/b/%/health.metapack.deps
+./schemas/example/schemas/b/%/locations.metapack
+./schemas/example/schemas/b/%/locations.metapack.deps
+./schemas/example/schemas/b/%/positions.metapack
+./schemas/example/schemas/b/%/positions.metapack.deps
+./schemas/example/schemas/b/%/schema.metapack
+./schemas/example/schemas/b/%/schema.metapack.deps
+./schemas/example/schemas/b/%/stats.metapack
+./schemas/example/schemas/b/%/stats.metapack.deps
+./schemas/example/schemas/c
+./schemas/example/schemas/c/%
+./schemas/example/schemas/c/%/blaze-exhaustive.metapack
+./schemas/example/schemas/c/%/blaze-exhaustive.metapack.deps
+./schemas/example/schemas/c/%/blaze-fast.metapack
+./schemas/example/schemas/c/%/blaze-fast.metapack.deps
+./schemas/example/schemas/c/%/bundle.metapack
+./schemas/example/schemas/c/%/bundle.metapack.deps
+./schemas/example/schemas/c/%/dependencies.metapack
+./schemas/example/schemas/c/%/dependencies.metapack.deps
+./schemas/example/schemas/c/%/dependents.metapack
+./schemas/example/schemas/c/%/dependents.metapack.deps
+./schemas/example/schemas/c/%/editor.metapack
+./schemas/example/schemas/c/%/editor.metapack.deps
+./schemas/example/schemas/c/%/health.metapack
+./schemas/example/schemas/c/%/health.metapack.deps
+./schemas/example/schemas/c/%/locations.metapack
+./schemas/example/schemas/c/%/locations.metapack.deps
+./schemas/example/schemas/c/%/positions.metapack
+./schemas/example/schemas/c/%/positions.metapack.deps
+./schemas/example/schemas/c/%/schema.metapack
+./schemas/example/schemas/c/%/schema.metapack.deps
+./schemas/example/schemas/c/%/stats.metapack
+./schemas/example/schemas/c/%/stats.metapack.deps
+./version.json
+EOF
+
+diff "$TMP/manifest.txt" "$TMP/expected_manifest.txt"