Improve invalid pattern regex error messages

Fixes: #678
Signed-off-by: Juan Cruz Viotti <jv@jviotti.com>

Author: jviotti

.github/workflows/test.yml

@@ -72,7 +72,7 @@ jobs:
       - name: Install dependencies (Alpine)
         if: matrix.platform.type == 'container'
         run: |
-          apk add --no-cache cmake make g++ zsh bash jq shellcheck pipx git perl-utils ca-certificates
+          apk add --no-cache cmake make g++ linux-headers zsh bash jq shellcheck pipx git perl-utils ca-certificates
           apk add --no-cache --repository https://dl-cdn.alpinelinux.org/alpine/edge/main --repository https://dl-cdn.alpinelinux.org/alpine/edge/community --repository https://dl-cdn.alpinelinux.org/alpine/edge/testing fuse3 bindfs
 
       - name: Install pre-commit

DEPENDENCIES

@@ -1,7 +1,7 @@
 vendorpull https://github.com/sourcemeta/vendorpull 1dcbac42809cf87cb5b045106b863e17ad84ba02
-core https://github.com/sourcemeta/core 6cdc6b362fef03e3efb5b674095287d983bafffa
+core https://github.com/sourcemeta/core ac96b5fba550a5097d174c1681ba3155cb3e3914
 jsonbinpack https://github.com/sourcemeta/jsonbinpack 3898774a945ebf7392bad8a9015ead97a0518a19
-blaze https://github.com/sourcemeta/blaze a7d43ba8a4de39918f44b97602950e260809cd0d
+blaze https://github.com/sourcemeta/blaze 6ea546a2698c636693e2cdc8a30fcc933b602429
 hydra https://github.com/sourcemeta/hydra 6e0ad118846ce57275bc7d6434f8440baae2c27e
 codegen https://github.com/sourcemeta/codegen 7733b3071cb140bc0d09406ebb9c3f306d71f7bf
 ctrf https://github.com/ctrf-io/ctrf 93ea827d951390190171d37443bff169cf47c808

src/command_compile.cc

@@ -79,6 +79,9 @@ auto sourcemeta::jsonschema::compile(const sourcemeta::core::Options &options)
   } catch (const sourcemeta::blaze::CompilerInvalidEntryPoint &error) {
     throw FileError<sourcemeta::blaze::CompilerInvalidEntryPoint>(schema_path,
                                                                   error);
+  } catch (const sourcemeta::blaze::CompilerInvalidRegexError &error) {
+    throw FileError<sourcemeta::blaze::CompilerInvalidRegexError>(schema_path,
+                                                                  error);
   } catch (
       const sourcemeta::blaze::CompilerReferenceTargetNotSchemaError &error) {
     throw FileError<sourcemeta::blaze::CompilerReferenceTargetNotSchemaError>(

src/command_lint.cc

@@ -166,6 +166,9 @@ static auto load_rule(sourcemeta::core::SchemaTransformer &bundle,
   } catch (const sourcemeta::blaze::LinterInvalidNameError &error) {
     throw sourcemeta::jsonschema::FileError<
         sourcemeta::blaze::LinterInvalidNameError>(rule_path, error);
+  } catch (const sourcemeta::blaze::LinterInvalidNamePatternError &error) {
+    throw sourcemeta::jsonschema::FileError<
+        sourcemeta::blaze::LinterInvalidNamePatternError>(rule_path, error);
   } catch (
       const sourcemeta::blaze::CompilerReferenceTargetNotSchemaError &error) {
     throw sourcemeta::jsonschema::FileError<
@@ -379,6 +382,14 @@ auto sourcemeta::jsonschema::lint(const sourcemeta::core::Options &options)
                   "Could not autofix the schema without breaking its internal "
                   "references",
                   entry.resolution_base, error.location()};
+            } catch (
+                const sourcemeta::blaze::CompilerInvalidRegexError &error) {
+              if (printed_progress) {
+                std::cerr << "\n";
+              }
+
+              throw FileError<sourcemeta::blaze::CompilerInvalidRegexError>(
+                  entry.resolution_base, error);
             } catch (
                 const sourcemeta::blaze::CompilerReferenceTargetNotSchemaError
                     &error) {
@@ -494,6 +505,10 @@ auto sourcemeta::jsonschema::lint(const sourcemeta::core::Options &options)
                 // Return 2 for logical lint failures
                 return EXIT_EXPECTED_FAILURE;
               }
+            } catch (
+                const sourcemeta::blaze::CompilerInvalidRegexError &error) {
+              throw FileError<sourcemeta::blaze::CompilerInvalidRegexError>(
+                  entry.resolution_base, error);
             } catch (
                 const sourcemeta::blaze::CompilerReferenceTargetNotSchemaError
                     &error) {

src/command_metaschema.cc

@@ -105,6 +105,9 @@ auto sourcemeta::jsonschema::metaschema(
           result = false;
         }
       }
+    } catch (const sourcemeta::blaze::CompilerInvalidRegexError &error) {
+      throw FileError<sourcemeta::blaze::CompilerInvalidRegexError>(
+          entry.resolution_base, error);
     } catch (
         const sourcemeta::blaze::CompilerReferenceTargetNotSchemaError &error) {
       throw FileError<sourcemeta::blaze::CompilerReferenceTargetNotSchemaError>(

src/command_validate.cc

@@ -258,6 +258,9 @@ auto sourcemeta::jsonschema::validate(const sourcemeta::core::Options &options)
     } catch (const sourcemeta::blaze::CompilerInvalidEntryPoint &error) {
       throw FileError<sourcemeta::blaze::CompilerInvalidEntryPoint>(schema_path,
                                                                     error);
+    } catch (const sourcemeta::blaze::CompilerInvalidRegexError &error) {
+      throw FileError<sourcemeta::blaze::CompilerInvalidRegexError>(schema_path,
+                                                                    error);
     } catch (
         const sourcemeta::blaze::CompilerReferenceTargetNotSchemaError &error) {
       throw FileError<sourcemeta::blaze::CompilerReferenceTargetNotSchemaError>(

src/error.h

@@ -340,6 +340,14 @@ inline auto print_exception(const bool is_json, const Exception &exception)
     }
   }
 
+  if constexpr (requires(const Exception &current) { current.regex(); }) {
+    if (is_json) {
+      error_json.assign("regex", sourcemeta::core::JSON{exception.regex()});
+    } else {
+      std::cerr << "  at regex " << exception.regex() << "\n";
+    }
+  }
+
   if constexpr (requires(const Exception &current) {
                   {
                     current.path()
@@ -496,6 +504,11 @@ inline auto try_catch(const sourcemeta::core::Options &options,
     const auto is_json{options.contains("json")};
     print_exception(is_json, error);
     return EXIT_OTHER_INPUT_ERROR;
+  } catch (const FileError<sourcemeta::blaze::LinterInvalidNamePatternError>
+               &error) {
+    const auto is_json{options.contains("json")};
+    print_exception(is_json, error);
+    return EXIT_OTHER_INPUT_ERROR;
   } catch (const FileError<sourcemeta::blaze::LinterMissingNameError> &error) {
     const auto is_json{options.contains("json")};
     print_exception(is_json, error);
@@ -562,6 +575,16 @@ inline auto try_catch(const sourcemeta::core::Options &options,
           << "\nUse the `inspect` command to find valid schema locations\n";
     }
 
+    return EXIT_SCHEMA_INPUT_ERROR;
+  } catch (
+      const FileError<sourcemeta::blaze::CompilerInvalidRegexError> &error) {
+    const auto is_json{options.contains("json")};
+    print_exception(is_json, error);
+    if (!is_json) {
+      std::cerr << "\nDetailed regex error messages are not yet supported\n"
+                   "Try tools like https://regex101.com to debug further\n";
+    }
+
     return EXIT_SCHEMA_INPUT_ERROR;
   } catch (const FileError<sourcemeta::core::SchemaReferenceError> &error) {
     const auto is_json{options.contains("json")};

test/CMakeLists.txt

@@ -77,6 +77,7 @@ add_jsonschema_test_unix(format/pass_config_ignore)
 add_jsonschema_test_unix(validate/fail_instance_enoent)
 add_jsonschema_test_unix(validate/fail_instance_invalid_json)
 add_jsonschema_test_unix(validate/fail_invalid_ref)
+add_jsonschema_test_unix(validate/fail_invalid_regex)
 add_jsonschema_test_unix(validate/fail_no_schema)
 add_jsonschema_test_unix(validate/fail_relative_external_ref_missing)
 add_jsonschema_test_unix(validate/fail_resolve_enoent)

test/lint/fail_draft7_defs_ref_target_fix.sh

@@ -25,38 +25,32 @@ cat << 'EOF' > "$TMP/schema.json"
 EOF
 
 "$1" lint "$TMP/schema.json" --fix 2>"$TMP/stderr.txt" && EXIT_CODE="$?" || EXIT_CODE="$?"
-# Unexpected error
-test "$EXIT_CODE" = "1" || exit 1
+# Schema input error
+test "$EXIT_CODE" = "4" || exit 1
 
 cat << EOF > "$TMP/expected.txt"
 .
-error: Could not autofix the schema without breaking its internal references
+error: The referenced schema is not considered to be a valid subschema given the dialect and vocabularies in use
+  at identifier file://$(realpath "$TMP")/schema.json#/x-\$defs/not-a-schema
   at file path $(realpath "$TMP")/schema.json
-  at location "/allOf/0/\$ref"
+  at location "/x-\$defs"
 
-This is an unexpected error, as making the auto-fix functionality work in all
-cases is tricky. We are working hard to improve the auto-fixing functionality
-to handle all possible edge cases, but for now, try again without \`--fix/-f\`
-and apply the suggestions by hand.
-
-Also consider consider reporting this problematic case to the issue tracker,
-so we can add it to the test suite and fix it:
-
-https://github.com/sourcemeta/jsonschema/issues
+Are you sure the reported location is a valid JSON Schema keyword in this dialect?
 EOF
 
 diff "$TMP/stderr.txt" "$TMP/expected.txt"
 
 # JSON error
 "$1" lint "$TMP/schema.json" --fix --json >"$TMP/stdout.txt" && EXIT_CODE="$?" || EXIT_CODE="$?"
-# Unexpected error
-test "$EXIT_CODE" = "1" || exit 1
+# Schema input error
+test "$EXIT_CODE" = "4" || exit 1
 
 cat << EOF > "$TMP/expected.txt"
 {
-  "error": "Could not autofix the schema without breaking its internal references",
+  "error": "The referenced schema is not considered to be a valid subschema given the dialect and vocabularies in use",
+  "identifier": "file://$(realpath "$TMP")/schema.json#/x-\$defs/not-a-schema",
   "filePath": "$(realpath "$TMP")/schema.json",
-  "location": "/allOf/0/\$ref"
+  "location": "/x-\$defs"
 }
 EOF
 

test/lint/fail_lint_config_rule_invalid_title.sh

@@ -40,8 +40,9 @@ cd "$TMP"
 test "$EXIT_CODE" = "6" || exit 1
 
 cat << EOF > "$TMP/expected.txt"
-error: The schema rule name must match ^[a-z0-9_/]+\$
+error: The schema rule name does not match the required pattern
   at identifier Invalid Title!
+  at regex ^[a-z0-9_/]+\$
   at file path $(realpath "$TMP")/rule.json
 EOF
 
@@ -55,8 +56,9 @@ test "$EXIT_CODE" = "6" || exit 1
 
 cat << EOF > "$TMP/expected_json.txt"
 {
-  "error": "The schema rule name must match ^[a-z0-9_/]+\$",
+  "error": "The schema rule name does not match the required pattern",
   "identifier": "Invalid Title!",
+  "regex": "^[a-z0-9_/]+\$",
   "filePath": "$(realpath "$TMP")/rule.json"
 }
 EOF