Support validating multi-doc YAML files

src/command_validate.cc

@@ -288,7 +288,9 @@ auto sourcemeta::jsonschema::validate(const sourcemeta::core::Options &options)
                                "given a single instance"};
     }
     if (std::filesystem::is_directory(instance_path) ||
-        instance_path.extension() == ".jsonl") {
+        instance_path.extension() == ".jsonl" ||
+        instance_path.extension() == ".yaml" ||
+        instance_path.extension() == ".yml") {
       for (const auto &entry : for_each_json({instance_path_view}, options)) {
         std::ostringstream error;
         sourcemeta::blaze::SimpleOutput output{entry.second};

src/input.h

@@ -153,14 +153,59 @@ handle_json_entry(const std::filesystem::path &entry_path,
         if (index == 0) {
           LOG_WARNING() << "The JSONL file is empty\n";
         }
+      } else if (canonical.extension() == ".yaml" ||
+                 canonical.extension() == ".yml") {
+        if (std::filesystem::is_empty(canonical)) {
+          return;
+        }
+        auto stream{sourcemeta::core::read_file(canonical)};
+        std::vector<std::pair<sourcemeta::core::JSON,
+                              sourcemeta::core::PointerPositionTracker>>
+            documents;
+        std::uint64_t line_offset{0};
+        std::uint64_t max_line{0};
+        while (stream.peek() != std::char_traits<char>::eof()) {
+          sourcemeta::core::PointerPositionTracker positions;
+          const std::uint64_t current_offset{line_offset};
+          max_line = 0;
+          auto callback = [&positions, current_offset, &max_line](
+                              const sourcemeta::core::JSON::ParsePhase phase,
+                              const sourcemeta::core::JSON::Type type,
+                              const std::uint64_t line,
+                              const std::uint64_t column,
+                              const sourcemeta::core::JSON &value) {
+            max_line = std::max(max_line, line);
+            positions(phase, type, line + current_offset, column, value);
+          };
+          documents.emplace_back(sourcemeta::core::parse_yaml(stream, callback),
+                                 std::move(positions));
+          // The YAML parser reports the line of the next document separator,
+          // so we subtract 1 to get the actual lines consumed by this document
+          line_offset += max_line > 0 ? max_line - 1 : 0;
+        }
+
+        if (documents.size() > 1) {
+          LOG_VERBOSE(options) << "Interpreting input as YAML multi-document: "
+                               << canonical.string() << "\n";
+          std::size_t index{0};
+          for (auto &entry : documents) {
+            result.push_back({canonical, std::move(entry.first),
+                              std::move(entry.second), index, true});
+            index += 1;
+          }
+        } else if (documents.size() == 1) {
+          result.push_back({std::move(canonical),
+                            std::move(documents.front().first),
+                            std::move(documents.front().second)});
+        }
       } else {
         if (std::filesystem::is_empty(canonical)) {
           return;
         }
         sourcemeta::core::PointerPositionTracker positions;
         // TODO: Print a verbose message for what is getting parsed
-        auto contents{sourcemeta::core::read_yaml_or_json(canonical,
-                                                          std::ref(positions))};
+        auto contents{
+            sourcemeta::core::read_json(canonical, std::ref(positions))};
         result.push_back(
             {std::move(canonical), std::move(contents), std::move(positions)});
       }

test/CMakeLists.txt

@@ -141,6 +141,13 @@ add_jsonschema_test_unix(validate/pass_many_verbose)
 add_jsonschema_test_unix(validate/fail_many)
 add_jsonschema_test_unix(validate/fail_many_verbose)
 add_jsonschema_test_unix(validate/fail_yaml)
+add_jsonschema_test_unix(validate/pass_yaml_multi)
+add_jsonschema_test_unix(validate/pass_yaml_multi_verbose)
+add_jsonschema_test_unix(validate/pass_yaml_multi_json)
+add_jsonschema_test_unix(validate/fail_yaml_multi_one)
+add_jsonschema_test_unix(validate/fail_yaml_multi_one_verbose)
+add_jsonschema_test_unix(validate/fail_yaml_multi_one_json)
+add_jsonschema_test_unix(validate/fail_yaml_multi_blank_lines)
 add_jsonschema_test_unix(validate/pass_json_ref_yaml)
 add_jsonschema_test_unix(validate/pass_process_substitution)
 add_jsonschema_test_unix(validate/pass_2020_12_fast_with_template)

test/validate/fail_yaml_multi_blank_lines.sh

@@ -0,0 +1,51 @@
+#!/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#",
+  "type": "object"
+}
+EOF
+
+# YAML multi-document with blank lines around separators
+cat << 'EOF' > "$TMP/instance.yaml"
+---
+foo: first
+
+---
+
+- bar: second
+---
+baz: third
+EOF
+
+"$1" validate "$TMP/schema.json" "$TMP/instance.yaml" --verbose 2>"$TMP/stderr.txt" \
+  && EXIT_CODE="$?" || EXIT_CODE="$?"
+test "$EXIT_CODE" = "2" || exit 1
+
+cat << EOF > "$TMP/expected.txt"
+Interpreting input as YAML multi-document: $(realpath "$TMP")/instance.yaml
+ok: $(realpath "$TMP")/instance.yaml (entry #1)
+  matches $(realpath "$TMP")/schema.json
+fail: $(realpath "$TMP")/instance.yaml (entry #2)
+
+[
+  {
+    "bar": "second"
+  }
+]
+
+error: Schema validation failure
+  The value was expected to be of type object but it was of type array
+    at instance location "" (line 6, column 1)
+    at evaluate path "/type"
+EOF
+
+diff "$TMP/stderr.txt" "$TMP/expected.txt"

test/validate/fail_yaml_multi_one.sh

@@ -0,0 +1,45 @@
+#!/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#",
+  "type": "object"
+}
+EOF
+
+cat << 'EOF' > "$TMP/instance.yaml"
+---
+foo: 1
+---
+- foo: 2
+---
+foo: 3
+EOF
+
+"$1" validate "$TMP/schema.json" "$TMP/instance.yaml" 2>"$TMP/stderr.txt" \
+  && EXIT_CODE="$?" || EXIT_CODE="$?"
+test "$EXIT_CODE" = "2" || exit 1
+
+cat << EOF > "$TMP/expected.txt"
+fail: $(realpath "$TMP")/instance.yaml (entry #2)
+
+[
+  {
+    "foo": 2
+  }
+]
+
+error: Schema validation failure
+  The value was expected to be of type object but it was of type array
+    at instance location "" (line 4, column 1)
+    at evaluate path "/type"
+EOF
+
+diff "$TMP/stderr.txt" "$TMP/expected.txt"

test/validate/fail_yaml_multi_one_json.sh

@@ -0,0 +1,48 @@
+#!/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#",
+  "type": "object"
+}
+EOF
+
+cat << 'EOF' > "$TMP/instance.yaml"
+---
+foo: 1
+---
+- foo: 2
+---
+foo: 3
+EOF
+
+"$1" validate "$TMP/schema.json" "$TMP/instance.yaml" --json > "$TMP/stdout.txt" \
+  && EXIT_CODE="$?" || EXIT_CODE="$?"
+test "$EXIT_CODE" = "2" || exit 1
+
+cat << EOF > "$TMP/expected.txt"
+{
+  "valid": true
+}
+{
+  "valid": false,
+  "errors": [
+    {
+      "keywordLocation": "/type",
+      "absoluteKeywordLocation": "file://$(realpath "$TMP")/schema.json#/type",
+      "instanceLocation": "",
+      "instancePosition": [ 4, 1, 5, 0 ],
+      "error": "The value was expected to be of type object but it was of type array"
+    }
+  ]
+}
+EOF
+
+diff "$TMP/stdout.txt" "$TMP/expected.txt"

test/validate/fail_yaml_multi_one_verbose.sh

@@ -0,0 +1,48 @@
+#!/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#",
+  "type": "object"
+}
+EOF
+
+cat << 'EOF' > "$TMP/instance.yaml"
+---
+foo: 1
+---
+- foo: 2
+---
+foo: 3
+EOF
+
+"$1" validate "$TMP/schema.json" "$TMP/instance.yaml" --verbose 2>"$TMP/stderr.txt" \
+  && EXIT_CODE="$?" || EXIT_CODE="$?"
+test "$EXIT_CODE" = "2" || exit 1
+
+cat << EOF > "$TMP/expected.txt"
+Interpreting input as YAML multi-document: $(realpath "$TMP")/instance.yaml
+ok: $(realpath "$TMP")/instance.yaml (entry #1)
+  matches $(realpath "$TMP")/schema.json
+fail: $(realpath "$TMP")/instance.yaml (entry #2)
+
+[
+  {
+    "foo": 2
+  }
+]
+
+error: Schema validation failure
+  The value was expected to be of type object but it was of type array
+    at instance location "" (line 4, column 1)
+    at evaluate path "/type"
+EOF
+
+diff "$TMP/stderr.txt" "$TMP/expected.txt"

test/validate/pass_yaml_multi.sh

@@ -0,0 +1,35 @@
+#!/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
+
+cat << 'EOF' > "$TMP/instance.yaml"
+---
+foo: first
+---
+foo: second
+---
+foo: third
+EOF
+
+"$1" validate "$TMP/schema.json" "$TMP/instance.yaml" 2> "$TMP/output.txt" 1>&2
+
+cat << EOF > "$TMP/expected.txt"
+EOF
+
+diff "$TMP/output.txt" "$TMP/expected.txt"

test/validate/pass_yaml_multi_json.sh

@@ -0,0 +1,44 @@
+#!/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
+
+cat << 'EOF' > "$TMP/instance.yaml"
+---
+foo: first
+---
+foo: second
+---
+foo: third
+EOF
+
+"$1" validate "$TMP/schema.json" "$TMP/instance.yaml" --json > "$TMP/output.json" 2>&1
+
+cat << EOF > "$TMP/expected.json"
+{
+  "valid": true
+}
+{
+  "valid": true
+}
+{
+  "valid": true
+}
+EOF
+
+diff "$TMP/output.json" "$TMP/expected.json"