Merge pull request #620 from sirosen/show-stringified-parse-errors

Refactor error formatting, show parse error cause

Author: sirosen

CHANGELOG.rst

@@ -8,13 +8,13 @@ CHANGELOG
 Unreleased
 ----------
 
-- Add official support for Python 3.14
-- Add Citation File Format schema and pre-commit hook. Thanks :user:`edgarrmondragon`! (:issue:`502`)
-
 .. vendor-insert-here
 
 - Update vendored schemas: bitbucket-pipelines, buildkite, circle-ci, compose-spec,
   dependabot, gitlab-ci, meltano, mergify, renovate, snapcraft (2025-11-11)
+- Add official support for Python 3.14
+- Add Citation File Format schema and pre-commit hook. Thanks :user:`edgarrmondragon`! (:issue:`502`)
+- Improved default text output when parsing errors are encountered. (:issue:`581`)
 
 0.34.1
 ------

src/check_jsonschema/checker.py

@@ -7,7 +7,7 @@
 import jsonschema
 import referencing.exceptions
 
-from . import utils
+from . import format_errors
 from .formats import FormatOptions
 from .instance_loader import InstanceLoader
 from .parsers import ParseError
@@ -31,7 +31,7 @@ def __init__(
         *,
         format_opts: FormatOptions,
         regex_impl: RegexImplementation,
-        traceback_mode: str = "short",
+        traceback_mode: t.Literal["minimal", "short", "full"] = "short",
         fill_defaults: bool = False,
     ) -> None:
         self._schema_loader = schema_loader
@@ -46,7 +46,7 @@ def __init__(
     def _fail(self, msg: str, err: Exception | None = None) -> t.NoReturn:
         click.echo(msg, err=True)
         if err is not None:
-            utils.print_error(err, mode=self._traceback_mode)
+            format_errors.print_error(err, mode=self._traceback_mode)
         raise _Exit(1)
 
     def get_validator(

src/check_jsonschema/cli/parse_result.py

@@ -41,7 +41,7 @@ def __init__(self) -> None:
         self.regex_variant: RegexVariantName = RegexVariantName.default
         # error and output controls
         self.verbosity: int = 1
-        self.traceback_mode: str = "short"
+        self.traceback_mode: t.Literal["short", "full"] = "short"
         self.output_format: str = "text"
 
     def set_regex_variant(

src/check_jsonschema/format_errors.py

@@ -0,0 +1,66 @@
+from __future__ import annotations
+
+import linecache
+import textwrap
+import traceback
+import typing as t
+
+import click
+
+
+def format_error_message(err: BaseException) -> str:
+    return f"{type(err).__name__}: {err}"
+
+
+def format_minimal_error(err: BaseException, *, indent: int = 0) -> str:
+    lines = [textwrap.indent(str(err), indent * " ")]
+    if err.__cause__ is not None:
+        lines.append(
+            textwrap.indent(format_error_message(err.__cause__), (indent + 2) * " ")
+        )
+
+    return "\n".join(lines)
+
+
+def format_shortened_error(err: BaseException, *, indent: int = 0) -> str:
+    lines = []
+    lines.append(textwrap.indent(format_error_message(err), indent * " "))
+    if err.__traceback__ is not None:
+        lineno = err.__traceback__.tb_lineno
+        tb_frame = err.__traceback__.tb_frame
+        filename = tb_frame.f_code.co_filename
+        line = linecache.getline(filename, lineno)
+        lines.append((indent + 2) * " " + f'in "{filename}", line {lineno}')
+        lines.append((indent + 2) * " " + ">>> " + line.strip())
+    return "\n".join(lines)
+
+
+def format_shortened_trace(caught_err: BaseException) -> str:
+    err_stack: list[BaseException] = [caught_err]
+    while err_stack[-1].__context__ is not None:
+        err_stack.append(err_stack[-1].__context__)  # type: ignore[arg-type]
+
+    parts = [format_shortened_error(caught_err)]
+    indent = 0
+    for err in err_stack[1:]:
+        indent += 2
+        parts.append("\n" + indent * " " + "caused by\n")
+        parts.append(format_shortened_error(err, indent=indent))
+    return "\n".join(parts)
+
+
+def format_error(
+    err: Exception, mode: t.Literal["minimal", "short", "full"] = "short"
+) -> str:
+    if mode == "minimal":
+        return format_minimal_error(err)
+    elif mode == "short":
+        return format_shortened_trace(err)
+    else:
+        return "".join(traceback.format_exception(type(err), err, err.__traceback__))
+
+
+def print_error(
+    err: Exception, mode: t.Literal["minimal", "short", "full"] = "short"
+) -> None:
+    click.echo(format_error(err, mode=mode), err=True)

src/check_jsonschema/reporter.py

@@ -13,9 +13,10 @@
 import click
 import jsonschema
 
+from . import format_errors
 from .parsers import ParseError
 from .result import CheckResult
-from .utils import format_error, iter_validation_error
+from .utils import iter_validation_error
 
 
 class Reporter(abc.ABC):
@@ -111,11 +112,12 @@ def _show_validation_error(
 
     def _show_parse_error(self, filename: str, err: ParseError) -> None:
         if self.verbosity < 2:
-            self._echo(click.style(str(err), fg="yellow"), indent=2)
+            mode: t.Literal["minimal", "short", "full"] = "minimal"
         elif self.verbosity < 3:
-            self._echo(textwrap.indent(format_error(err, mode="short"), "  "))
+            mode = "short"
         else:
-            self._echo(textwrap.indent(format_error(err, mode="full"), "  "))
+            mode = "full"
+        self._echo(textwrap.indent(format_errors.format_error(err, mode=mode), "  "))
 
     def report_errors(self, result: CheckResult) -> None:
         if self.verbosity < 1:

src/check_jsonschema/utils.py

@@ -1,16 +1,12 @@
 from __future__ import annotations
 
-import linecache
 import os
 import pathlib
 import re
-import textwrap
-import traceback
 import typing as t
 import urllib.parse
 import urllib.request
 
-import click
 import jsonschema
 
 WINDOWS = os.name == "nt"
@@ -95,44 +91,6 @@ def filename2path(filename: str) -> pathlib.Path:
     return p.resolve()
 
 
-def format_shortened_error(err: Exception, *, indent: int = 0) -> str:
-    lines = []
-    lines.append(textwrap.indent(f"{type(err).__name__}: {err}", indent * " "))
-    if err.__traceback__ is not None:
-        lineno = err.__traceback__.tb_lineno
-        tb_frame = err.__traceback__.tb_frame
-        filename = tb_frame.f_code.co_filename
-        line = linecache.getline(filename, lineno)
-        lines.append((indent + 2) * " " + f'in "{filename}", line {lineno}')
-        lines.append((indent + 2) * " " + ">>> " + line.strip())
-    return "\n".join(lines)
-
-
-def format_shortened_trace(caught_err: Exception) -> str:
-    err_stack: list[Exception] = [caught_err]
-    while err_stack[-1].__context__ is not None:
-        err_stack.append(err_stack[-1].__context__)  # type: ignore[arg-type]
-
-    parts = [format_shortened_error(caught_err)]
-    indent = 0
-    for err in err_stack[1:]:
-        indent += 2
-        parts.append("\n" + indent * " " + "caused by\n")
-        parts.append(format_shortened_error(err, indent=indent))
-    return "\n".join(parts)
-
-
-def format_error(err: Exception, mode: str = "short") -> str:
-    if mode == "short":
-        return format_shortened_trace(err)
-    else:
-        return "".join(traceback.format_exception(type(err), err, err.__traceback__))
-
-
-def print_error(err: Exception, mode: str = "short") -> None:
-    click.echo(format_error(err, mode=mode), err=True)
-
-
 def iter_validation_error(
     err: jsonschema.ValidationError,
 ) -> t.Iterator[jsonschema.ValidationError]:

tests/unit/test_reporters.py

@@ -4,6 +4,7 @@
 import pytest
 from jsonschema import Draft7Validator
 
+from check_jsonschema.parsers import ParseError
 from check_jsonschema.reporter import JsonReporter, TextReporter
 from check_jsonschema.result import CheckResult
 
@@ -240,3 +241,30 @@ def test_json_format_validation_error_nested(capsys, pretty_json, verbosity):
     assert "{'baz': 'buzz'} is not of type 'string'" in [
         item["message"] for item in bar_errors
     ]
+
+
+def test_text_print_parse_error_with_cause(capsys):
+    cause = json.JSONDecodeError("a bad thing happened", "{,}", 1)
+    error = ParseError("whoopsie during parsing")
+    error.__cause__ = cause
+
+    result = CheckResult()
+    result.record_parse_error("foo.json", error)
+
+    text_reporter = TextReporter(verbosity=1)
+    text_reporter.report_result(result)
+    captured = capsys.readouterr()
+
+    # nothing to stderr
+    assert captured.err == ""
+    # stdout contains a nicely formatted error
+    assert (
+        textwrap.dedent(
+            """\
+            Several files failed to parse.
+              whoopsie during parsing
+                JSONDecodeError: a bad thing happened: line 1 column 2 (char 1)
+            """
+        )
+        in captured.out
+    )