From 8601ffd0567b3b85dd2df85d90bcf58258ced1d9 Mon Sep 17 00:00:00 2001
From: Juan Cruz Viotti <jv@jviotti.com>
Date: Tue, 20 Aug 2024 11:43:34 -0400
Subject: [PATCH] [WIP] Gracefully handle validating empty JSONL datasets

Signed-off-by: Juan Cruz Viotti <jv@jviotti.com>
---
 src/command_validate.cc                   |  7 ++++--
 test/CMakeLists.txt                       |  2 ++
 test/validate/fail_jsonl_all.sh           |  2 +-
 test/validate/fail_jsonl_all_verbose.sh   |  2 +-
 test/validate/fail_jsonl_one.sh           |  2 +-
 test/validate/fail_jsonl_one_verbose.sh   |  4 +--
 test/validate/pass_jsonl_empty.sh         | 28 +++++++++++++++++++++
 test/validate/pass_jsonl_empty_verbose.sh | 30 +++++++++++++++++++++++
 test/validate/pass_jsonl_verbose.sh       |  4 +--
 9 files changed, 72 insertions(+), 9 deletions(-)
 create mode 100755 test/validate/pass_jsonl_empty.sh
 create mode 100755 test/validate/pass_jsonl_empty_verbose.sh

diff --git a/src/command_validate.cc b/src/command_validate.cc
index 0a4bb203..3405c153 100644
--- a/src/command_validate.cc
+++ b/src/command_validate.cc
@@ -67,6 +67,7 @@ auto intelligence::jsonschema::cli::validate(
       auto stream{sourcemeta::jsontoolkit::read_file(instance_path)};
       try {
         for (const auto &instance : sourcemeta::jsontoolkit::JSONL{stream}) {
+          index += 1;
           std::ostringstream error;
           bool subresult = true;
           if (benchmark) {
@@ -110,13 +111,15 @@ auto intelligence::jsonschema::cli::validate(
             result = false;
             break;
           }
-
-          index += 1;
         }
       } catch (const sourcemeta::jsontoolkit::ParseError &error) {
         // For producing better error messages
         throw sourcemeta::jsontoolkit::FileParseError(instance_path, error);
       }
+
+      if (index == 0) {
+        log_verbose(options) << "warning: The JSONL file is empty\n";
+      }
     } else {
       const auto instance{sourcemeta::jsontoolkit::from_file(instance_path)};
       std::ostringstream error;
diff --git a/test/CMakeLists.txt b/test/CMakeLists.txt
index 829295ea..4da6ec16 100644
--- a/test/CMakeLists.txt
+++ b/test/CMakeLists.txt
@@ -66,6 +66,8 @@ add_jsonschema_test_unix(validate/fail_draft7)
 add_jsonschema_test_unix(validate/fail_2019_09)
 add_jsonschema_test_unix(validate/fail_2020_12)
 add_jsonschema_test_unix(validate/pass_jsonl)
+add_jsonschema_test_unix(validate/pass_jsonl_empty)
+add_jsonschema_test_unix(validate/pass_jsonl_empty_verbose)
 add_jsonschema_test_unix(validate/pass_jsonl_verbose)
 add_jsonschema_test_unix(validate/fail_jsonl_invalid_entry)
 add_jsonschema_test_unix(validate/fail_jsonl_one)
diff --git a/test/validate/fail_jsonl_all.sh b/test/validate/fail_jsonl_all.sh
index aa5771b1..2567f660 100755
--- a/test/validate/fail_jsonl_all.sh
+++ b/test/validate/fail_jsonl_all.sh
@@ -25,7 +25,7 @@ EOF
 test "$CODE" = "1" || exit 1
 
 cat << EOF > "$TMP/expected.txt"
-fail: $(realpath "$TMP")/instance.jsonl (entry #0)
+fail: $(realpath "$TMP")/instance.jsonl (entry #1)
 
 {
   "foo": 1
diff --git a/test/validate/fail_jsonl_all_verbose.sh b/test/validate/fail_jsonl_all_verbose.sh
index 2a96bc0a..5a7b570a 100755
--- a/test/validate/fail_jsonl_all_verbose.sh
+++ b/test/validate/fail_jsonl_all_verbose.sh
@@ -26,7 +26,7 @@ test "$CODE" = "1" || exit 1
 
 cat << EOF > "$TMP/expected.txt"
 Interpreting input as JSONL: $(realpath "$TMP")/instance.jsonl
-fail: $(realpath "$TMP")/instance.jsonl (entry #0)
+fail: $(realpath "$TMP")/instance.jsonl (entry #1)
 
 {
   "foo": 1
diff --git a/test/validate/fail_jsonl_one.sh b/test/validate/fail_jsonl_one.sh
index 9c2f407b..bc219c5e 100755
--- a/test/validate/fail_jsonl_one.sh
+++ b/test/validate/fail_jsonl_one.sh
@@ -25,7 +25,7 @@ EOF
 test "$CODE" = "1" || exit 1
 
 cat << EOF > "$TMP/expected.txt"
-fail: $(realpath "$TMP")/instance.jsonl (entry #1)
+fail: $(realpath "$TMP")/instance.jsonl (entry #2)
 
 [
   {
diff --git a/test/validate/fail_jsonl_one_verbose.sh b/test/validate/fail_jsonl_one_verbose.sh
index 132114e7..7c3fdb31 100755
--- a/test/validate/fail_jsonl_one_verbose.sh
+++ b/test/validate/fail_jsonl_one_verbose.sh
@@ -26,9 +26,9 @@ test "$CODE" = "1" || exit 1
 
 cat << EOF > "$TMP/expected.txt"
 Interpreting input as JSONL: $(realpath "$TMP")/instance.jsonl
-ok: $(realpath "$TMP")/instance.jsonl (entry #0)
+ok: $(realpath "$TMP")/instance.jsonl (entry #1)
   matches $(realpath "$TMP")/schema.json
-fail: $(realpath "$TMP")/instance.jsonl (entry #1)
+fail: $(realpath "$TMP")/instance.jsonl (entry #2)
 
 [
   {
diff --git a/test/validate/pass_jsonl_empty.sh b/test/validate/pass_jsonl_empty.sh
new file mode 100755
index 00000000..15526a96
--- /dev/null
+++ b/test/validate/pass_jsonl_empty.sh
@@ -0,0 +1,28 @@
+#!/bin/sh
+
+set -o errexit
+set -o nounset
+
+TMP="$(mktemp -d)"
+clean() { rm -rf "$TMP"; }
+trap clean EXIT
+
+cat << 'EOF' > "$TMP/schema.json"
+{
+  "$schema": "http://json-schema.org/draft-04/schema#",
+  "properties": {
+    "foo": {
+      "type": "string"
+    }
+  }
+}
+EOF
+
+touch "$TMP/instance.jsonl"
+
+"$1" validate "$TMP/schema.json" "$TMP/instance.jsonl" 2> "$TMP/output.txt" 1>&2
+
+cat << EOF > "$TMP/expected.txt"
+EOF
+
+diff "$TMP/output.txt" "$TMP/expected.txt"
diff --git a/test/validate/pass_jsonl_empty_verbose.sh b/test/validate/pass_jsonl_empty_verbose.sh
new file mode 100755
index 00000000..2fd4ecd9
--- /dev/null
+++ b/test/validate/pass_jsonl_empty_verbose.sh
@@ -0,0 +1,30 @@
+#!/bin/sh
+
+set -o errexit
+set -o nounset
+
+TMP="$(mktemp -d)"
+clean() { rm -rf "$TMP"; }
+trap clean EXIT
+
+cat << 'EOF' > "$TMP/schema.json"
+{
+  "$schema": "http://json-schema.org/draft-04/schema#",
+  "properties": {
+    "foo": {
+      "type": "string"
+    }
+  }
+}
+EOF
+
+touch "$TMP/instance.jsonl"
+
+"$1" validate "$TMP/schema.json" "$TMP/instance.jsonl" --verbose 2> "$TMP/output.txt" 1>&2
+
+cat << EOF > "$TMP/expected.txt"
+Interpreting input as JSONL: $(realpath "$TMP")/instance.jsonl
+warning: The JSONL file is empty
+EOF
+
+diff "$TMP/output.txt" "$TMP/expected.txt"
diff --git a/test/validate/pass_jsonl_verbose.sh b/test/validate/pass_jsonl_verbose.sh
index 0b20dcb9..e94e0735 100755
--- a/test/validate/pass_jsonl_verbose.sh
+++ b/test/validate/pass_jsonl_verbose.sh
@@ -28,12 +28,12 @@ EOF
 
 cat << EOF > "$TMP/expected.txt"
 Interpreting input as JSONL: $(realpath "$TMP")/instance.jsonl
-ok: $(realpath "$TMP")/instance.jsonl (entry #0)
-  matches $(realpath "$TMP")/schema.json
 ok: $(realpath "$TMP")/instance.jsonl (entry #1)
   matches $(realpath "$TMP")/schema.json
 ok: $(realpath "$TMP")/instance.jsonl (entry #2)
   matches $(realpath "$TMP")/schema.json
+ok: $(realpath "$TMP")/instance.jsonl (entry #3)
+  matches $(realpath "$TMP")/schema.json
 EOF
 
 diff "$TMP/output.txt" "$TMP/expected.txt"