Port diagnostics format to Prism from CRuby

Author: kddnewton

ext/prism/extension.c

@@ -14,6 +14,7 @@ extern "C" {
 #include "prism/ast.h"
 #include "prism/buffer.h"
 #include "prism/diagnostic.h"
+#include "prism/errors_format.h"
 #include "prism/json.h"
 #include "prism/node.h"
 #include "prism/options.h"

include/prism.h

@@ -0,0 +1,53 @@
+/**
+ * @file errors_format.h
+ *
+ * A function for formatting the errors in a parser into a buffer.
+ */
+#ifndef PRISM_ERRORS_FORMAT_H
+#define PRISM_ERRORS_FORMAT_H
+
+#include "prism/compiler/exported.h"
+#include "prism/compiler/nodiscard.h"
+#include "prism/compiler/nonnull.h"
+
+#include "prism/buffer.h"
+#include "prism/diagnostic.h"
+#include "prism/parser.h"
+
+/** The type of formatting to use when formatting errors. */
+typedef enum {
+    /** Format errors in a plain format with no colors or styles. Typically
+     * used when outputting to a file or when the output is not a terminal.
+     */
+    PM_ERRORS_FORMAT_PLAIN = 1,
+
+    /** Format errors in a style format with bold only. Typically used when
+     * outputting to a terminal where you do not want colors.
+     */
+    PM_ERRORS_FORMAT_STYLE = 2,
+
+    /** Format errors in a color format with bold and colors. The default case
+     * when outputting on the terminal.
+     */
+    PM_ERRORS_FORMAT_COLOR = 3
+} pm_errors_format_type_t;
+
+/**
+ * Format the errors in a parser into a buffer. After return, the buffer will
+ * contain the formatted errors and any relevant source snippets if available
+ * and possible.
+ *
+ * @param parser The parser containing the errors to format. Note that the
+ *   parser does not need to be fully initialized, but it will need to have
+ *   start, end, start_line, encoding, line_offsets, filepath, and error_list
+ *   initialized.
+ * @param buffer The buffer to format the errors into.
+ * @param format_type The type of formatting to use when formatting the errors.
+ * @returns The error level of the error with the highest precedence that was
+ *   formatted. In practice this means it will be the argument or load error
+ *   level if any were found, otherwise it will be the syntax error level.
+ *   Callers should raise appropriate error types based on the returned level.
+ */
+PRISM_EXPORTED_FUNCTION pm_error_level_t pm_errors_format(const pm_parser_t *parser, pm_buffer_t *buffer, pm_errors_format_type_t format_type) PRISM_NONNULL(1, 2);
+
+#endif

include/prism/errors_format.h

@@ -105,6 +105,18 @@ struct pm_options_t {
     */
     int8_t frozen_string_literal;
 
+    /*
+     * Whether or not the parser should raise an error when it encounters an
+     * error. By default this is off, but can be turned on in the case that you
+     * want nicely formatted syntax errors. The values that this can be set to
+     * are:
+     *  - 0 - do not raise an error, just return the errors in the result
+     *  - PM_ERRORS_FORMAT_PLAIN - raise a plain error with no formatting
+     *  - PM_ERRORS_FORMAT_STYLE - raise an error with bold formatting
+     *  - PM_ERRORS_FORMAT_COLOR - raise an error with color formatting
+     */
+    uint8_t raise_error;
+
     /*
      * Whether or not the encoding magic comments should be respected. This is a
      * niche use-case where you want to parse a file with a specific encoding

include/prism/internal/options.h

@@ -249,6 +249,22 @@ PRISM_EXPORTED_FUNCTION bool pm_options_freeze(const pm_options_t *options) PRIS
  */
 PRISM_EXPORTED_FUNCTION void pm_options_freeze_set(pm_options_t *options, bool freeze) PRISM_NONNULL(1);
 
+/**
+ * Get the raise_error option on the given options struct.
+ *
+ * @param options The options struct to get the raise_error value from.
+ * @returns The raise_error value.
+ */
+PRISM_EXPORTED_FUNCTION uint8_t pm_options_raise_error(const pm_options_t *options) PRISM_NONNULL(1);
+
+/**
+ * Set the raise_error option on the given options struct.
+ *
+ * @param options The options struct to set the raise_error value on.
+ * @param raise_error The raise_error value to set.
+ */
+PRISM_EXPORTED_FUNCTION void pm_options_raise_error_set(pm_options_t *options, uint8_t raise_error) PRISM_NONNULL(1);
+
 /**
  * Allocate and zero out the scopes array on the given options struct.
  *

include/prism/options.h

@@ -17,6 +17,7 @@
 #include "prism/compiler/nonnull.h"
 
 #include "prism/buffer.h"
+#include "prism/errors_format.h"
 #include "prism/parser.h"
 #include "prism/source.h"
 #include "prism/stream.h"
@@ -91,6 +92,24 @@ PRISM_EXPORTED_FUNCTION void pm_serialize_parse_lex(pm_buffer_t *buffer, const u
  */
 PRISM_EXPORTED_FUNCTION bool pm_serialize_parse_success_p(const uint8_t *source, size_t size, const char *data) PRISM_NONNULL(1);
 
+/**
+ * Parse the given source and format any errors that are encountered into the
+ * given buffer using the given format type. If the source parses without any
+ * errors, then -1 is returned and the buffer is left empty. Otherwise, the
+ * name of the encoding of the source is written to the buffer, followed by a
+ * null byte, followed by the formatted errors, and the error level of the
+ * error with the highest precedence is returned.
+ *
+ * @param buffer The buffer to write the encoding name and formatted errors to.
+ * @param source The source to parse.
+ * @param size The size of the source.
+ * @param data The optional data to pass to the parser.
+ * @param format_type The type of formatting to use when formatting the errors.
+ * @returns The error level of the error with the highest precedence, or -1 if
+ *   the source parsed without errors.
+ */
+PRISM_EXPORTED_FUNCTION int8_t pm_serialize_parse_errors_format(pm_buffer_t *buffer, const uint8_t *source, size_t size, const char *data, pm_errors_format_type_t format_type) PRISM_NONNULL(1, 2);
+
 #endif
 
 #endif

include/prism/serialize.h

@@ -103,23 +103,23 @@ def self.find(callable, rubyvm: !!defined?(RubyVM))
   #      def gets: (?Integer integer) -> (String | nil)
   #    end
   #
-  #    def self.parse:               (String source,  ?filepath: String, ?command_line: String, ?encoding: Encoding | false, ?freeze: bool, ?frozen_string_literal: bool, ?line: Integer, ?main_script: bool, ?partial_script: bool, ?scopes: Array[Array[Symbol]], ?version: String) -> ParseResult
-  #    def self.profile:             (String source,  ?filepath: String, ?command_line: String, ?encoding: Encoding | false, ?freeze: bool, ?frozen_string_literal: bool, ?line: Integer, ?main_script: bool, ?partial_script: bool, ?scopes: Array[Array[Symbol]], ?version: String) -> void
-  #    def self.lex:                 (String source,  ?filepath: String, ?command_line: String, ?encoding: Encoding | false, ?freeze: bool, ?frozen_string_literal: bool, ?line: Integer, ?main_script: bool, ?partial_script: bool, ?scopes: Array[Array[Symbol]], ?version: String) -> LexResult
-  #    def self.parse_lex:           (String source,  ?filepath: String, ?command_line: String, ?encoding: Encoding | false, ?freeze: bool, ?frozen_string_literal: bool, ?line: Integer, ?main_script: bool, ?partial_script: bool, ?scopes: Array[Array[Symbol]], ?version: String) -> ParseLexResult
-  #    def self.dump:                (String source,  ?filepath: String, ?command_line: String, ?encoding: Encoding | false, ?freeze: bool, ?frozen_string_literal: bool, ?line: Integer, ?main_script: bool, ?partial_script: bool, ?scopes: Array[Array[Symbol]], ?version: String) -> String
-  #    def self.parse_comments:      (String source,  ?filepath: String, ?command_line: String, ?encoding: Encoding | false, ?freeze: bool, ?frozen_string_literal: bool, ?line: Integer, ?main_script: bool, ?partial_script: bool, ?scopes: Array[Array[Symbol]], ?version: String) -> Array[Comment]
-  #    def self.parse_success?:      (String source,  ?filepath: String, ?command_line: String, ?encoding: Encoding | false, ?freeze: bool, ?frozen_string_literal: bool, ?line: Integer, ?main_script: bool, ?partial_script: bool, ?scopes: Array[Array[Symbol]], ?version: String) -> bool
-  #    def self.parse_failure?:      (String source,  ?filepath: String, ?command_line: String, ?encoding: Encoding | false, ?freeze: bool, ?frozen_string_literal: bool, ?line: Integer, ?main_script: bool, ?partial_script: bool, ?scopes: Array[Array[Symbol]], ?version: String) -> bool
-  #    def self.parse_stream:        (_Stream stream, ?filepath: String, ?command_line: String, ?encoding: Encoding | false, ?freeze: bool, ?frozen_string_literal: bool, ?line: Integer, ?main_script: bool, ?partial_script: bool, ?scopes: Array[Array[Symbol]], ?version: String) -> ParseResult
-  #    def self.parse_file:          (String filepath,                   ?command_line: String, ?encoding: Encoding | false, ?freeze: bool, ?frozen_string_literal: bool, ?line: Integer, ?main_script: bool, ?partial_script: bool, ?scopes: Array[Array[Symbol]], ?version: String) -> ParseResult
-  #    def self.profile_file:        (String filepath,                   ?command_line: String, ?encoding: Encoding | false, ?freeze: bool, ?frozen_string_literal: bool, ?line: Integer, ?main_script: bool, ?partial_script: bool, ?scopes: Array[Array[Symbol]], ?version: String) -> void
-  #    def self.lex_file:            (String filepath,                   ?command_line: String, ?encoding: Encoding | false, ?freeze: bool, ?frozen_string_literal: bool, ?line: Integer, ?main_script: bool, ?partial_script: bool, ?scopes: Array[Array[Symbol]], ?version: String) -> LexResult
-  #    def self.parse_lex_file:      (String filepath,                   ?command_line: String, ?encoding: Encoding | false, ?freeze: bool, ?frozen_string_literal: bool, ?line: Integer, ?main_script: bool, ?partial_script: bool, ?scopes: Array[Array[Symbol]], ?version: String) -> ParseLexResult
-  #    def self.dump_file:           (String filepath,                   ?command_line: String, ?encoding: Encoding | false, ?freeze: bool, ?frozen_string_literal: bool, ?line: Integer, ?main_script: bool, ?partial_script: bool, ?scopes: Array[Array[Symbol]], ?version: String) -> String
-  #    def self.parse_file_comments: (String filepath,                   ?command_line: String, ?encoding: Encoding | false, ?freeze: bool, ?frozen_string_literal: bool, ?line: Integer, ?main_script: bool, ?partial_script: bool, ?scopes: Array[Array[Symbol]], ?version: String) -> Array[Comment]
-  #    def self.parse_file_success?: (String filepath,                   ?command_line: String, ?encoding: Encoding | false, ?freeze: bool, ?frozen_string_literal: bool, ?line: Integer, ?main_script: bool, ?partial_script: bool, ?scopes: Array[Array[Symbol]], ?version: String) -> bool
-  #    def self.parse_file_failure?: (String filepath,                   ?command_line: String, ?encoding: Encoding | false, ?freeze: bool, ?frozen_string_literal: bool, ?line: Integer, ?main_script: bool, ?partial_script: bool, ?scopes: Array[Array[Symbol]], ?version: String) -> bool
+  #    def self.parse:               (String source,  ?filepath: String, ?command_line: String, ?encoding: Encoding | false, ?freeze: bool, ?frozen_string_literal: bool, ?line: Integer, ?main_script: bool, ?partial_script: bool, ?raise_error: Symbol | true, ?scopes: Array[Array[Symbol]], ?version: String) -> ParseResult
+  #    def self.profile:             (String source,  ?filepath: String, ?command_line: String, ?encoding: Encoding | false, ?freeze: bool, ?frozen_string_literal: bool, ?line: Integer, ?main_script: bool, ?partial_script: bool, ?raise_error: Symbol | true, ?scopes: Array[Array[Symbol]], ?version: String) -> void
+  #    def self.lex:                 (String source,  ?filepath: String, ?command_line: String, ?encoding: Encoding | false, ?freeze: bool, ?frozen_string_literal: bool, ?line: Integer, ?main_script: bool, ?partial_script: bool, ?raise_error: Symbol | true, ?scopes: Array[Array[Symbol]], ?version: String) -> LexResult
+  #    def self.parse_lex:           (String source,  ?filepath: String, ?command_line: String, ?encoding: Encoding | false, ?freeze: bool, ?frozen_string_literal: bool, ?line: Integer, ?main_script: bool, ?partial_script: bool, ?raise_error: Symbol | true, ?scopes: Array[Array[Symbol]], ?version: String) -> ParseLexResult
+  #    def self.dump:                (String source,  ?filepath: String, ?command_line: String, ?encoding: Encoding | false, ?freeze: bool, ?frozen_string_literal: bool, ?line: Integer, ?main_script: bool, ?partial_script: bool, ?raise_error: Symbol | true, ?scopes: Array[Array[Symbol]], ?version: String) -> String
+  #    def self.parse_comments:      (String source,  ?filepath: String, ?command_line: String, ?encoding: Encoding | false, ?freeze: bool, ?frozen_string_literal: bool, ?line: Integer, ?main_script: bool, ?partial_script: bool, ?raise_error: Symbol | true, ?scopes: Array[Array[Symbol]], ?version: String) -> Array[Comment]
+  #    def self.parse_success?:      (String source,  ?filepath: String, ?command_line: String, ?encoding: Encoding | false, ?freeze: bool, ?frozen_string_literal: bool, ?line: Integer, ?main_script: bool, ?partial_script: bool, ?raise_error: Symbol | true, ?scopes: Array[Array[Symbol]], ?version: String) -> bool
+  #    def self.parse_failure?:      (String source,  ?filepath: String, ?command_line: String, ?encoding: Encoding | false, ?freeze: bool, ?frozen_string_literal: bool, ?line: Integer, ?main_script: bool, ?partial_script: bool, ?raise_error: Symbol | true, ?scopes: Array[Array[Symbol]], ?version: String) -> bool
+  #    def self.parse_stream:        (_Stream stream, ?filepath: String, ?command_line: String, ?encoding: Encoding | false, ?freeze: bool, ?frozen_string_literal: bool, ?line: Integer, ?main_script: bool, ?partial_script: bool, ?raise_error: Symbol | true, ?scopes: Array[Array[Symbol]], ?version: String) -> ParseResult
+  #    def self.parse_file:          (String filepath,                   ?command_line: String, ?encoding: Encoding | false, ?freeze: bool, ?frozen_string_literal: bool, ?line: Integer, ?main_script: bool, ?partial_script: bool, ?raise_error: Symbol | true, ?scopes: Array[Array[Symbol]], ?version: String) -> ParseResult
+  #    def self.profile_file:        (String filepath,                   ?command_line: String, ?encoding: Encoding | false, ?freeze: bool, ?frozen_string_literal: bool, ?line: Integer, ?main_script: bool, ?partial_script: bool, ?raise_error: Symbol | true, ?scopes: Array[Array[Symbol]], ?version: String) -> void
+  #    def self.lex_file:            (String filepath,                   ?command_line: String, ?encoding: Encoding | false, ?freeze: bool, ?frozen_string_literal: bool, ?line: Integer, ?main_script: bool, ?partial_script: bool, ?raise_error: Symbol | true, ?scopes: Array[Array[Symbol]], ?version: String) -> LexResult
+  #    def self.parse_lex_file:      (String filepath,                   ?command_line: String, ?encoding: Encoding | false, ?freeze: bool, ?frozen_string_literal: bool, ?line: Integer, ?main_script: bool, ?partial_script: bool, ?raise_error: Symbol | true, ?scopes: Array[Array[Symbol]], ?version: String) -> ParseLexResult
+  #    def self.dump_file:           (String filepath,                   ?command_line: String, ?encoding: Encoding | false, ?freeze: bool, ?frozen_string_literal: bool, ?line: Integer, ?main_script: bool, ?partial_script: bool, ?raise_error: Symbol | true, ?scopes: Array[Array[Symbol]], ?version: String) -> String
+  #    def self.parse_file_comments: (String filepath,                   ?command_line: String, ?encoding: Encoding | false, ?freeze: bool, ?frozen_string_literal: bool, ?line: Integer, ?main_script: bool, ?partial_script: bool, ?raise_error: Symbol | true, ?scopes: Array[Array[Symbol]], ?version: String) -> Array[Comment]
+  #    def self.parse_file_success?: (String filepath,                   ?command_line: String, ?encoding: Encoding | false, ?freeze: bool, ?frozen_string_literal: bool, ?line: Integer, ?main_script: bool, ?partial_script: bool, ?raise_error: Symbol | true, ?scopes: Array[Array[Symbol]], ?version: String) -> bool
+  #    def self.parse_file_failure?: (String filepath,                   ?command_line: String, ?encoding: Encoding | false, ?freeze: bool, ?frozen_string_literal: bool, ?line: Integer, ?main_script: bool, ?partial_script: bool, ?raise_error: Symbol | true, ?scopes: Array[Array[Symbol]], ?version: String) -> bool
 end
 
 require_relative "prism/polyfill/byteindex"