prism 1.8.1 → 2.0.0.rc1

data/src/diagnostic.c

@@ -1,5 +1,3 @@
-/* :markup: markdown */
-
 /*----------------------------------------------------------------------------*/
 /* This file is generated by the templates/template.rb script and should not  */
 /* be modified manually. See                                                  */
@@ -8,9 +6,22 @@
 /* template                                                                   */
 /*----------------------------------------------------------------------------*/
 
-#include "prism/diagnostic.h"
+#include "prism/internal/diagnostic.h"
+
+#include "prism/compiler/assume.h"
+#include "prism/compiler/inline.h"
+
+#include "prism/internal/allocator.h"
+#include "prism/internal/arena.h"
+#include "prism/internal/list.h"
 
-#define PM_DIAGNOSTIC_ID_MAX 326
+#include <assert.h>
+#include <inttypes.h>
+#include <stdarg.h>
+#include <stdio.h>
+#include <stdlib.h>
+
+#define PM_DIAGNOSTIC_ID_MAX 327
 
 /** This struct holds the data for each diagnostic. */
 typedef struct {
@@ -85,16 +96,16 @@ typedef struct {
  * * `PM_WARNING_LEVEL_VERBOSE` - Warnings that appear with `-w`, as in `ruby -w -c -e 'code'`.
  */
 static const pm_diagnostic_data_t diagnostic_messages[PM_DIAGNOSTIC_ID_MAX] = {
-    // Special error that can be replaced
+    /* Special error that can be replaced */
     [PM_ERR_CANNOT_PARSE_EXPRESSION]            = { "cannot parse the expression", PM_ERROR_LEVEL_SYNTAX },
 
-    // Errors that should raise argument errors
+    /* Errors that should raise argument errors */
     [PM_ERR_INVALID_ENCODING_MAGIC_COMMENT]     = { "unknown or invalid encoding in the magic comment", PM_ERROR_LEVEL_ARGUMENT },
 
-    // Errors that should raise load errors
+    /* Errors that should raise load errors */
     [PM_ERR_SCRIPT_NOT_FOUND]                   = { "no Ruby script found in input", PM_ERROR_LEVEL_LOAD },
 
-    // Errors that should raise syntax errors
+    /* Errors that should raise syntax errors */
     [PM_ERR_ALIAS_ARGUMENT]                     = { "invalid argument being passed to `alias`; expected a bare word, symbol, constant, or global variable", PM_ERROR_LEVEL_SYNTAX },
     [PM_ERR_ALIAS_ARGUMENT_NUMBERED_REFERENCE]  = { "invalid argument being passed to `alias`; can't make alias for the number variables", PM_ERROR_LEVEL_SYNTAX },
     [PM_ERR_AMPAMPEQ_MULTI_ASSIGN]              = { "unexpected `&&=` in a multiple assignment", PM_ERROR_LEVEL_SYNTAX },
@@ -158,6 +169,7 @@ static const pm_diagnostic_data_t diagnostic_messages[PM_DIAGNOSTIC_ID_MAX] = {
     [PM_ERR_DEF_ENDLESS]                        = { "could not parse the endless method body", PM_ERROR_LEVEL_SYNTAX },
     [PM_ERR_DEF_ENDLESS_PARAMETERS]             = { "could not parse the endless method parameters", PM_ERROR_LEVEL_SYNTAX },
     [PM_ERR_DEF_ENDLESS_SETTER]                 = { "invalid method name; a setter method cannot be defined in an endless method definition", PM_ERROR_LEVEL_SYNTAX },
+    [PM_ERR_DEF_ENDLESS_DO_BLOCK]               = { "unexpected `do` for block in an endless method definition", PM_ERROR_LEVEL_SYNTAX },
     [PM_ERR_DEF_NAME]                           = { "unexpected %s; expected a method name", PM_ERROR_LEVEL_SYNTAX },
     [PM_ERR_DEF_PARAMS_TERM]                    = { "expected a delimiter to close the parameters", PM_ERROR_LEVEL_SYNTAX },
     [PM_ERR_DEF_PARAMS_TERM_PAREN]              = { "unexpected %s; expected a `)` to close the parameters", PM_ERROR_LEVEL_SYNTAX },
@@ -339,13 +351,15 @@ static const pm_diagnostic_data_t diagnostic_messages[PM_DIAGNOSTIC_ID_MAX] = {
     [PM_ERR_PATTERN_TERM_PAREN]                 = { "expected a `)` to close the pattern expression", PM_ERROR_LEVEL_SYNTAX },
     [PM_ERR_PIPEPIPEEQ_MULTI_ASSIGN]            = { "unexpected `||=` in a multiple assignment", PM_ERROR_LEVEL_SYNTAX },
     [PM_ERR_REGEXP_ENCODING_OPTION_MISMATCH]    = { "regexp encoding option '%c' differs from source encoding '%s'", PM_ERROR_LEVEL_SYNTAX },
+    [PM_ERR_REGEXP_ESCAPED_NON_ASCII_IN_UTF8]   = { "escaped non ASCII character in UTF-8 regexp: /%.*s/", PM_ERROR_LEVEL_SYNTAX },
     [PM_ERR_REGEXP_INCOMPAT_CHAR_ENCODING]      = { "incompatible character encoding: /%.*s/", PM_ERROR_LEVEL_SYNTAX },
-    [PM_ERR_REGEXP_NON_ESCAPED_MBC]             = { "/.../n has a non escaped non ASCII character in non ASCII-8BIT script: /%.*s/", PM_ERROR_LEVEL_SYNTAX },
+    [PM_ERR_REGEXP_INVALID_CHAR_PROPERTY]       = { "invalid character property name {%.*s}: /%.*s/", PM_ERROR_LEVEL_SYNTAX },
     [PM_ERR_REGEXP_INVALID_UNICODE_RANGE]       = { "invalid Unicode range: /%.*s/", PM_ERROR_LEVEL_SYNTAX },
+    [PM_ERR_REGEXP_NON_ESCAPED_MBC]             = { "/.../n has a non escaped non ASCII character in non ASCII-8BIT script: /%.*s/", PM_ERROR_LEVEL_SYNTAX },
     [PM_ERR_REGEXP_PARSE_ERROR]                 = { "%s", PM_ERROR_LEVEL_SYNTAX },
     [PM_ERR_REGEXP_UNKNOWN_OPTIONS]             = { "unknown regexp %s - %.*s", PM_ERROR_LEVEL_SYNTAX },
     [PM_ERR_REGEXP_TERM]                        = { "unterminated regexp meets end of file; expected a closing delimiter", PM_ERROR_LEVEL_SYNTAX },
-    [PM_ERR_REGEXP_UTF8_CHAR_NON_UTF8_REGEXP]   = { "UTF-8 character in non UTF-8 regexp: /%s/", PM_ERROR_LEVEL_SYNTAX },
+    [PM_ERR_REGEXP_UTF8_CHAR_NON_UTF8_REGEXP]   = { "UTF-8 character in non UTF-8 regexp: /%.*s/", PM_ERROR_LEVEL_SYNTAX },
     [PM_ERR_RESCUE_EXPRESSION]                  = { "expected a rescued expression", PM_ERROR_LEVEL_SYNTAX },
     [PM_ERR_RESCUE_MODIFIER_VALUE]              = { "expected a value after the `rescue` modifier", PM_ERROR_LEVEL_SYNTAX },
     [PM_ERR_RESCUE_TERM]                        = { "expected a closing delimiter for the `rescue` clause", PM_ERROR_LEVEL_SYNTAX },
@@ -360,7 +374,7 @@ static const pm_diagnostic_data_t diagnostic_messages[PM_DIAGNOSTIC_ID_MAX] = {
     [PM_ERR_STRING_INTERPOLATED_TERM]           = { "unterminated string; expected a closing delimiter for the interpolated string", PM_ERROR_LEVEL_SYNTAX },
     [PM_ERR_STRING_LITERAL_EOF]                 = { "unterminated string meets end of file", PM_ERROR_LEVEL_SYNTAX },
     [PM_ERR_STRING_LITERAL_TERM]                = { "unexpected %s, expected a string literal terminator", PM_ERROR_LEVEL_SYNTAX },
-    [PM_ERR_SYMBOL_INVALID]                     = { "invalid symbol", PM_ERROR_LEVEL_SYNTAX }, // TODO expected symbol? prism.c ~9719
+    [PM_ERR_SYMBOL_INVALID]                     = { "invalid symbol", PM_ERROR_LEVEL_SYNTAX }, /* TODO expected symbol? prism.c ~9719 */
     [PM_ERR_SYMBOL_TERM_DYNAMIC]                = { "unterminated quoted string; expected a closing delimiter for the dynamic symbol", PM_ERROR_LEVEL_SYNTAX },
     [PM_ERR_SYMBOL_TERM_INTERPOLATED]           = { "unterminated symbol; expected a closing delimiter for the interpolated symbol", PM_ERROR_LEVEL_SYNTAX },
     [PM_ERR_TERNARY_COLON]                      = { "expected a `:` after the true expression of a ternary operator", PM_ERROR_LEVEL_SYNTAX },
@@ -387,7 +401,7 @@ static const pm_diagnostic_data_t diagnostic_messages[PM_DIAGNOSTIC_ID_MAX] = {
     [PM_ERR_WRITE_TARGET_UNEXPECTED]            = { "unexpected write target", PM_ERROR_LEVEL_SYNTAX },
     [PM_ERR_XSTRING_TERM]                       = { "expected a closing delimiter for the `%x` or backtick string", PM_ERROR_LEVEL_SYNTAX },
 
-    // Warnings
+    /* Warnings */
     [PM_WARN_AMBIGUOUS_BINARY_OPERATOR]         = { "'%s' after local variable or literal is interpreted as binary operator even though it seems like %s", PM_WARNING_LEVEL_VERBOSE },
     [PM_WARN_AMBIGUOUS_FIRST_ARGUMENT_MINUS]    = { "ambiguous first argument; put parentheses or a space even after `-` operator", PM_WARNING_LEVEL_VERBOSE },
     [PM_WARN_AMBIGUOUS_FIRST_ARGUMENT_PLUS]     = { "ambiguous first argument; put parentheses or a space even after `+` operator", PM_WARNING_LEVEL_VERBOSE },
@@ -423,8 +437,8 @@ static const pm_diagnostic_data_t diagnostic_messages[PM_DIAGNOSTIC_ID_MAX] = {
 /**
  * Get the human-readable name of the given diagnostic ID.
  */
-const char *
-pm_diagnostic_id_human(pm_diagnostic_id_t diag_id) {
+static const char *
+pm_diagnostic_id_name(pm_diagnostic_id_t diag_id) {
     switch (diag_id) {
         case PM_ERR_ALIAS_ARGUMENT: return "alias_argument";
         case PM_ERR_ALIAS_ARGUMENT_NUMBERED_REFERENCE: return "alias_argument_numbered_reference";
@@ -432,7 +446,6 @@ pm_diagnostic_id_human(pm_diagnostic_id_t diag_id) {
         case PM_ERR_ARGUMENT_AFTER_BLOCK: return "argument_after_block";
         case PM_ERR_ARGUMENT_AFTER_FORWARDING_ELLIPSES: return "argument_after_forwarding_ellipses";
         case PM_ERR_ARGUMENT_BARE_HASH: return "argument_bare_hash";
-        case PM_ERR_ARGUMENT_BLOCK_FORWARDING: return "argument_block_forwarding";
         case PM_ERR_ARGUMENT_BLOCK_MULTI: return "argument_block_multi";
         case PM_ERR_ARGUMENT_CONFLICT_AMPERSAND: return "argument_conflict_ampersand";
         case PM_ERR_ARGUMENT_CONFLICT_STAR: return "argument_conflict_star";
@@ -491,6 +504,7 @@ pm_diagnostic_id_human(pm_diagnostic_id_t diag_id) {
         case PM_ERR_DEF_ENDLESS: return "def_endless";
         case PM_ERR_DEF_ENDLESS_PARAMETERS: return "def_endless_parameters";
         case PM_ERR_DEF_ENDLESS_SETTER: return "def_endless_setter";
+        case PM_ERR_DEF_ENDLESS_DO_BLOCK: return "def_endless_do_block";
         case PM_ERR_DEF_NAME: return "def_name";
         case PM_ERR_DEF_PARAMS_TERM: return "def_params_term";
         case PM_ERR_DEF_PARAMS_TERM_PAREN: return "def_params_term_paren";
@@ -524,7 +538,6 @@ pm_diagnostic_id_human(pm_diagnostic_id_t diag_id) {
         case PM_ERR_EXPECT_EXPRESSION_AFTER_LPAREN: return "expect_expression_after_lparen";
         case PM_ERR_EXPECT_EXPRESSION_AFTER_OPERATOR: return "expect_expression_after_operator";
         case PM_ERR_EXPECT_EXPRESSION_AFTER_PIPEPIPEEQ: return "expect_expression_after_pipepipeeq";
-        case PM_ERR_EXPECT_EXPRESSION_AFTER_QUESTION: return "expect_expression_after_question";
         case PM_ERR_EXPECT_EXPRESSION_AFTER_SPLAT: return "expect_expression_after_splat";
         case PM_ERR_EXPECT_EXPRESSION_AFTER_SPLAT_HASH: return "expect_expression_after_splat_hash";
         case PM_ERR_EXPECT_EXPRESSION_AFTER_STAR: return "expect_expression_after_star";
@@ -674,7 +687,9 @@ pm_diagnostic_id_human(pm_diagnostic_id_t diag_id) {
         case PM_ERR_PATTERN_TERM_PAREN: return "pattern_term_paren";
         case PM_ERR_PIPEPIPEEQ_MULTI_ASSIGN: return "pipepipeeq_multi_assign";
         case PM_ERR_REGEXP_ENCODING_OPTION_MISMATCH: return "regexp_encoding_option_mismatch";
+        case PM_ERR_REGEXP_ESCAPED_NON_ASCII_IN_UTF8: return "regexp_escaped_non_ascii_in_utf8";
         case PM_ERR_REGEXP_INCOMPAT_CHAR_ENCODING: return "regexp_incompat_char_encoding";
+        case PM_ERR_REGEXP_INVALID_CHAR_PROPERTY: return "regexp_invalid_char_property";
         case PM_ERR_REGEXP_INVALID_UNICODE_RANGE: return "regexp_invalid_unicode_range";
         case PM_ERR_REGEXP_NON_ESCAPED_MBC: return "regexp_non_escaped_mbc";
         case PM_ERR_REGEXP_PARSE_ERROR: return "regexp_parse_error";
@@ -758,9 +773,10 @@ pm_diagnostic_id_human(pm_diagnostic_id_t diag_id) {
     return "";
 }
 
-static inline const char *
-pm_diagnostic_message(pm_diagnostic_id_t diag_id) {
+static PRISM_INLINE const char *
+pm_diagnostic_id_message(pm_diagnostic_id_t diag_id) {
     assert(diag_id < PM_DIAGNOSTIC_ID_MAX);
+    PRISM_ASSUME(diag_id < PM_DIAGNOSTIC_ID_MAX);
 
     const char *message = diagnostic_messages[diag_id].message;
     assert(message);
@@ -768,91 +784,103 @@ pm_diagnostic_message(pm_diagnostic_id_t diag_id) {
     return message;
 }
 
-static inline uint8_t
-pm_diagnostic_level(pm_diagnostic_id_t diag_id) {
+static PRISM_INLINE uint8_t
+pm_diagnostic_id_level(pm_diagnostic_id_t diag_id) {
     assert(diag_id < PM_DIAGNOSTIC_ID_MAX);
+    PRISM_ASSUME(diag_id < PM_DIAGNOSTIC_ID_MAX);
 
     return (uint8_t) diagnostic_messages[diag_id].level;
 }
 
+/**
+ * Get the type of the given diagnostic.
+ */
+const char *
+pm_diagnostic_type(const pm_diagnostic_t *diagnostic) {
+    return pm_diagnostic_id_name(diagnostic->diag_id);
+}
+
+/**
+ * Get the location of the given diagnostic.
+ */
+pm_location_t
+pm_diagnostic_location(const pm_diagnostic_t *diagnostic) {
+    return diagnostic->location;
+}
+
+/**
+ * Get the message of the given diagnostic.
+ */
+const char *
+pm_diagnostic_message(const pm_diagnostic_t *diagnostic) {
+    return diagnostic->message;
+}
+
+/**
+ * Get the error level associated with the given diagnostic.
+ */
+pm_error_level_t
+pm_diagnostic_error_level(const pm_diagnostic_t *diagnostic) {
+    return (pm_error_level_t) pm_diagnostic_id_level(diagnostic->diag_id);
+}
+
+/**
+ * Get the warning level associated with the given diagnostic.
+ */
+pm_warning_level_t
+pm_diagnostic_warning_level(const pm_diagnostic_t *diagnostic) {
+    return (pm_warning_level_t) pm_diagnostic_id_level(diagnostic->diag_id);
+}
+
 /**
  * Append an error to the given list of diagnostic.
  */
-bool
-pm_diagnostic_list_append(pm_list_t *list, const uint8_t *start, const uint8_t *end, pm_diagnostic_id_t diag_id) {
-    pm_diagnostic_t *diagnostic = (pm_diagnostic_t *) xcalloc(1, sizeof(pm_diagnostic_t));
-    if (diagnostic == NULL) return false;
+void
+pm_diagnostic_list_append(pm_arena_t *arena, pm_list_t *list, uint32_t start, uint32_t length, pm_diagnostic_id_t diag_id) {
+    pm_diagnostic_t *diagnostic = (pm_diagnostic_t *) pm_arena_zalloc(arena, sizeof(pm_diagnostic_t), PRISM_ALIGNOF(pm_diagnostic_t));
 
     *diagnostic = (pm_diagnostic_t) {
-        .location = { start, end },
+        .location = { .start = start, .length = length },
         .diag_id = diag_id,
-        .message = pm_diagnostic_message(diag_id),
-        .owned = false,
-        .level = pm_diagnostic_level(diag_id)
+        .message = pm_diagnostic_id_message(diag_id),
+        .level = pm_diagnostic_id_level(diag_id)
     };
 
     pm_list_append(list, (pm_list_node_t *) diagnostic);
-    return true;
 }
 
 /**
  * Append a diagnostic to the given list of diagnostics that is using a format
  * string for its message.
  */
-bool
-pm_diagnostic_list_append_format(pm_list_t *list, const uint8_t *start, const uint8_t *end, pm_diagnostic_id_t diag_id, ...) {
+void
+pm_diagnostic_list_append_format(pm_arena_t *arena, pm_list_t *list, uint32_t start, uint32_t length, pm_diagnostic_id_t diag_id, ...) {
     va_list arguments;
     va_start(arguments, diag_id);
 
-    const char *format = pm_diagnostic_message(diag_id);
+    const char *format = pm_diagnostic_id_message(diag_id);
     int result = vsnprintf(NULL, 0, format, arguments);
     va_end(arguments);
 
     if (result < 0) {
-        return false;
+        return;
     }
 
-    pm_diagnostic_t *diagnostic = (pm_diagnostic_t *) xcalloc(1, sizeof(pm_diagnostic_t));
-    if (diagnostic == NULL) {
-        return false;
-    }
+    pm_diagnostic_t *diagnostic = (pm_diagnostic_t *) pm_arena_zalloc(arena, sizeof(pm_diagnostic_t), PRISM_ALIGNOF(pm_diagnostic_t));
 
-    size_t length = (size_t) (result + 1);
-    char *message = (char *) xmalloc(length);
-    if (message == NULL) {
-        xfree(diagnostic);
-        return false;
-    }
+    size_t message_length = (size_t) (result + 1);
+    char *message = (char *) pm_arena_alloc(arena, message_length, 1);
 
     va_start(arguments, diag_id);
-    vsnprintf(message, length, format, arguments);
+    vsnprintf(message, message_length, format, arguments);
     va_end(arguments);
 
     *diagnostic = (pm_diagnostic_t) {
-        .location = { start, end },
+        .location = { .start = start, .length = length },
         .diag_id = diag_id,
         .message = message,
-        .owned = true,
-        .level = pm_diagnostic_level(diag_id)
+        .level = pm_diagnostic_id_level(diag_id)
     };
 
     pm_list_append(list, (pm_list_node_t *) diagnostic);
-    return true;
-}
-
-/**
- * Deallocate the internal state of the given diagnostic list.
- */
-void
-pm_diagnostic_list_free(pm_list_t *list) {
-    pm_diagnostic_t *node = (pm_diagnostic_t *) list->head;
-
-    while (node != NULL) {
-        pm_diagnostic_t *next = (pm_diagnostic_t *) node->node.next;
-
-        if (node->owned) xfree((void *) node->message);
-        xfree(node);
-
-        node = next;
-    }
 }

data/src/encoding.c

@@ -1,4 +1,9 @@
-#include "prism/encoding.h"
+#include "prism/internal/encoding.h"
+
+#include "prism/compiler/unused.h"
+#include "prism/internal/strncasecmp.h"
+
+#include <assert.h>
 
 typedef uint32_t pm_unicode_codepoint_t;
 
@@ -4089,7 +4094,7 @@ pm_encoding_ascii_isupper_char(const uint8_t *b, ptrdiff_t n) {
  * matter what the codepoint, so this function is shared between them.
  */
 static size_t
-pm_encoding_single_char_width(PRISM_ATTRIBUTE_UNUSED const uint8_t *b, PRISM_ATTRIBUTE_UNUSED ptrdiff_t n) {
+pm_encoding_single_char_width(PRISM_UNUSED const uint8_t *b, PRISM_UNUSED ptrdiff_t n) {
     return 1;
 }
 
@@ -4110,7 +4115,7 @@ pm_encoding_euc_jp_char_width(const uint8_t *b, ptrdiff_t n) {
     }
 
     // These are the triple byte characters.
-    if ((n > 2) && (b[0] == 0x8F) && (b[1] >= 0xA1 && b[2] <= 0xFE) && (b[2] >= 0xA1 && b[2] <= 0xFE)) {
+    if ((n > 2) && (b[0] == 0x8F) && (b[1] >= 0xA1 && b[1] <= 0xFE) && (b[2] >= 0xA1 && b[2] <= 0xFE)) {
         return 3;
     }