[ruby/prism] Provide options for reducing size

ruby/prism@592128de4d

prism/encoding.h

@@ -135,7 +135,14 @@ extern const uint8_t pm_encoding_unicode_table[256];
  */
 typedef enum {
     PM_ENCODING_UTF_8 = 0,
+    PM_ENCODING_US_ASCII,
     PM_ENCODING_ASCII_8BIT,
+    PM_ENCODING_EUC_JP,
+    PM_ENCODING_WINDOWS_31J,
+
+// We optionally support excluding the full set of encodings to only support the
+// minimum necessary to process Ruby code without encoding comments.
+#ifndef PRISM_ENCODING_EXCLUDE_FULL
     PM_ENCODING_BIG5,
     PM_ENCODING_BIG5_HKSCS,
     PM_ENCODING_BIG5_UAO,
@@ -148,7 +155,6 @@ typedef enum {
     PM_ENCODING_CP950,
     PM_ENCODING_CP951,
     PM_ENCODING_EMACS_MULE,
-    PM_ENCODING_EUC_JP,
     PM_ENCODING_EUC_JP_MS,
     PM_ENCODING_EUC_JIS_2004,
     PM_ENCODING_EUC_KR,
@@ -208,7 +214,6 @@ typedef enum {
     PM_ENCODING_STATELESS_ISO_2022_JP,
     PM_ENCODING_STATELESS_ISO_2022_JP_KDDI,
     PM_ENCODING_TIS_620,
-    PM_ENCODING_US_ASCII,
     PM_ENCODING_UTF8_MAC,
     PM_ENCODING_UTF8_DOCOMO,
     PM_ENCODING_UTF8_KDDI,
@@ -222,8 +227,9 @@ typedef enum {
     PM_ENCODING_WINDOWS_1256,
     PM_ENCODING_WINDOWS_1257,
     PM_ENCODING_WINDOWS_1258,
-    PM_ENCODING_WINDOWS_31J,
     PM_ENCODING_WINDOWS_874,
+#endif
+
     PM_ENCODING_MAXIMUM
 } pm_encoding_type_t;
 

prism/extension.c

@@ -311,7 +311,7 @@ dump(int argc, VALUE *argv, VALUE self) {
     pm_options_t options = { 0 };
     string_options(argc, argv, &input, &options);
 
-#ifdef PRISM_DEBUG_MODE_BUILD
+#ifdef PRISM_BUILD_DEBUG
     size_t length = pm_string_length(&input);
     char* dup = xmalloc(length);
     memcpy(dup, pm_string_source(&input), length);
@@ -320,7 +320,7 @@ dump(int argc, VALUE *argv, VALUE self) {
 
     VALUE value = dump_input(&input, &options);
 
-#ifdef PRISM_DEBUG_MODE_BUILD
+#ifdef PRISM_BUILD_DEBUG
     xfree(dup);
 #endif
 
@@ -737,7 +737,7 @@ parse(int argc, VALUE *argv, VALUE self) {
     pm_options_t options = { 0 };
     string_options(argc, argv, &input, &options);
 
-#ifdef PRISM_DEBUG_MODE_BUILD
+#ifdef PRISM_BUILD_DEBUG
     size_t length = pm_string_length(&input);
     char* dup = xmalloc(length);
     memcpy(dup, pm_string_source(&input), length);
@@ -746,7 +746,7 @@ parse(int argc, VALUE *argv, VALUE self) {
 
     VALUE value = parse_input(&input, &options);
 
-#ifdef PRISM_DEBUG_MODE_BUILD
+#ifdef PRISM_BUILD_DEBUG
     xfree(dup);
 #endif
 

prism/pack.c

@@ -1,16 +1,43 @@
 #include "prism/pack.h"
 
+// We optionally support parsing String#pack templates. For systems that don't
+// want or need this functionality, it can be turned off with the
+// PRISM_EXCLUDE_PACK define.
+#ifdef PRISM_EXCLUDE_PACK
+
+void pm_pack_parse(void) {}
+
+#else
+
 #include <stdbool.h>
 #include <errno.h>
 
 static uintmax_t
-strtoumaxc(const char **format);
+strtoumaxc(const char **format) {
+    uintmax_t value = 0;
+    while (**format >= '0' && **format <= '9') {
+        if (value > UINTMAX_MAX / 10) {
+            errno = ERANGE;
+        }
+        value = value * 10 + ((uintmax_t) (**format - '0'));
+        (*format)++;
+    }
+    return value;
+}
 
 PRISM_EXPORTED_FUNCTION pm_pack_result
-pm_pack_parse(pm_pack_variant variant, const char **format, const char *format_end,
-                            pm_pack_type *type, pm_pack_signed *signed_type, pm_pack_endian *endian, pm_pack_size *size,
-                            pm_pack_length_type *length_type, uint64_t *length, pm_pack_encoding *encoding) {
-
+pm_pack_parse(
+    pm_pack_variant variant,
+    const char **format,
+    const char *format_end,
+    pm_pack_type *type,
+    pm_pack_signed *signed_type,
+    pm_pack_endian *endian,
+    pm_pack_size *size,
+    pm_pack_length_type *length_type,
+    uint64_t *length,
+    pm_pack_encoding *encoding
+) {
     if (*encoding == PM_PACK_ENCODING_START) {
         *encoding = PM_PACK_ENCODING_US_ASCII;
     }
@@ -479,15 +506,4 @@ pm_size_to_native(pm_pack_size size) {
     }
 }
 
-static uintmax_t
-strtoumaxc(const char **format) {
-    uintmax_t value = 0;
-    while (**format >= '0' && **format <= '9') {
-        if (value > UINTMAX_MAX / 10) {
-            errno = ERANGE;
-        }
-        value = value * 10 + ((uintmax_t) (**format - '0'));
-        (*format)++;
-    }
-    return value;
-}
+#endif

prism/pack.h

@@ -6,6 +6,15 @@
 #ifndef PRISM_PACK_H
 #define PRISM_PACK_H
 
+// We optionally support parsing String#pack templates. For systems that don't
+// want or need this functionality, it can be turned off with the
+// PRISM_EXCLUDE_PACK define.
+#ifdef PRISM_EXCLUDE_PACK
+
+void pm_pack_parse(void);
+
+#else
+
 #include "prism/defines.h"
 
 #include <stdint.h>
@@ -150,3 +159,5 @@ pm_pack_parse(
 PRISM_EXPORTED_FUNCTION size_t pm_size_to_native(pm_pack_size size);
 
 #endif
+
+#endif

prism/prettyprint.h

@@ -6,6 +6,12 @@
 #ifndef PRISM_PRETTYPRINT_H
 #define PRISM_PRETTYPRINT_H
 
+#ifdef PRISM_EXCLUDE_PRETTYPRINT
+
+void pm_prettyprint(void);
+
+#else
+
 #include "prism/defines.h"
 
 #include <stdio.h>
@@ -24,3 +30,5 @@
 PRISM_EXPORTED_FUNCTION void pm_prettyprint(pm_buffer_t *output_buffer, const pm_parser_t *parser, const pm_node_t *node);
 
 #endif
+
+#endif

prism/prism.c

@@ -19316,6 +19316,41 @@ pm_parse_stream(pm_parser_t *parser, pm_buffer_t *buffer, void *stream, pm_parse
     return node;
 }
 
+/**
+ * Parse the source and return true if it parses without errors or warnings.
+ */
+PRISM_EXPORTED_FUNCTION bool
+pm_parse_success_p(const uint8_t *source, size_t size, const char *data) {
+    pm_options_t options = { 0 };
+    pm_options_read(&options, data);
+
+    pm_parser_t parser;
+    pm_parser_init(&parser, source, size, &options);
+
+    pm_node_t *node = pm_parse(&parser);
+    pm_node_destroy(&parser, node);
+
+    bool result = parser.error_list.size == 0 && parser.warning_list.size == 0;
+    pm_parser_free(&parser);
+    pm_options_free(&options);
+
+    return result;
+}
+
+#undef PM_CASE_KEYWORD
+#undef PM_CASE_OPERATOR
+#undef PM_CASE_WRITABLE
+#undef PM_STRING_EMPTY
+#undef PM_LOCATION_NODE_BASE_VALUE
+#undef PM_LOCATION_NODE_VALUE
+#undef PM_LOCATION_NULL_VALUE
+#undef PM_LOCATION_TOKEN_VALUE
+
+// We optionally support serializing to a binary string. For systems that don't
+// want or need this functionality, it can be turned off with the
+// PRISM_EXCLUDE_SERIALIZATION define.
+#ifndef PRISM_EXCLUDE_SERIALIZATION
+
 static inline void
 pm_serialize_header(pm_buffer_t *buffer) {
     pm_buffer_append_string(buffer, "PRISM", 5);
@@ -19402,14 +19437,7 @@ pm_serialize_parse_comments(pm_buffer_t *buffer, const uint8_t *source, size_t s
     pm_options_free(&options);
 }
 
-#undef PM_CASE_KEYWORD
-#undef PM_CASE_OPERATOR
-#undef PM_CASE_WRITABLE
-#undef PM_STRING_EMPTY
-#undef PM_LOCATION_NODE_BASE_VALUE
-#undef PM_LOCATION_NODE_VALUE
-#undef PM_LOCATION_NULL_VALUE
-#undef PM_LOCATION_TOKEN_VALUE
+#endif
 
 /** An error that is going to be formatted into the output. */
 typedef struct {

prism/prism.h

@@ -98,6 +98,11 @@ typedef char * (pm_parse_stream_fgets_t)(char *string, int size, void *stream);
  */
 PRISM_EXPORTED_FUNCTION pm_node_t * pm_parse_stream(pm_parser_t *parser, pm_buffer_t *buffer, void *stream, pm_parse_stream_fgets_t *fgets, const pm_options_t *options);
 
+// We optionally support serializing to a binary string. For systems that don't
+// want or need this functionality, it can be turned off with the
+// PRISM_EXCLUDE_SERIALIZATION define.
+#ifndef PRISM_EXCLUDE_SERIALIZATION
+
 /**
  * Parse and serialize the AST represented by the source that is read out of the
  * given stream into to the given buffer.
@@ -185,6 +190,8 @@ PRISM_EXPORTED_FUNCTION void pm_serialize_lex(pm_buffer_t *buffer, const uint8_t
  */
 PRISM_EXPORTED_FUNCTION void pm_serialize_parse_lex(pm_buffer_t *buffer, const uint8_t *source, size_t size, const char *data);
 
+#endif
+
 /**
  * Parse the source and return true if it parses without errors or warnings.
  *
@@ -220,6 +227,10 @@ const char * pm_token_type_human(pm_token_type_t token_type);
  */
 PRISM_EXPORTED_FUNCTION void pm_parser_errors_format(const pm_parser_t *parser, pm_buffer_t *buffer, bool colorize);
 
+// We optionally support dumping to JSON. For systems that don't want or need
+// this functionality, it can be turned off with the PRISM_EXCLUDE_JSON define.
+#ifndef PRISM_EXCLUDE_JSON
+
 /**
  * Dump JSON to the given buffer.
  *
@@ -229,6 +240,8 @@ PRISM_EXPORTED_FUNCTION void pm_parser_errors_format(const pm_parser_t *parser,
  */
 PRISM_EXPORTED_FUNCTION void pm_dump_json(pm_buffer_t *buffer, const pm_parser_t *parser, const pm_node_t *node);
 
+#endif
+
 /**
  * @mainpage
  *

prism/templates/src/node.c.erb

@@ -247,6 +247,10 @@ pm_visit_child_nodes(const pm_node_t *node, bool (*visitor)(const pm_node_t *nod
     }
 }
 
+// We optionally support dumping to JSON. For systems that don't want or need
+// this functionality, it can be turned off with the PRISM_EXCLUDE_JSON define.
+#ifndef PRISM_EXCLUDE_JSON
+
 static void
 pm_dump_json_constant(pm_buffer_t *buffer, const pm_parser_t *parser, pm_constant_id_t constant_id) {
     const pm_constant_t *constant = pm_constant_pool_id_to_constant(&parser->constant_pool, constant_id);
@@ -360,3 +364,5 @@ pm_dump_json(pm_buffer_t *buffer, const pm_parser_t *parser, const pm_node_t *no
             break;
     }
 }
+
+#endif

prism/templates/src/prettyprint.c.erb

@@ -1,6 +1,15 @@
 <%# encoding: ASCII -%>
 #include "prism/prettyprint.h"
 
+// We optionally support pretty printing nodes. For systems that don't want or
+// need this functionality, it can be turned off with the
+// PRISM_EXCLUDE_PRETTYPRINT define.
+#ifdef PRISM_EXCLUDE_PRETTYPRINT
+
+void pm_prettyprint(void) {}
+
+#else
+
 static inline void
 prettyprint_location(pm_buffer_t *output_buffer, const pm_parser_t *parser, const pm_location_t *location) {
     pm_line_column_t start = pm_newline_list_line_column(&parser->newline_list, location->start, parser->start_line);
@@ -154,3 +163,5 @@ pm_prettyprint(pm_buffer_t *output_buffer, const pm_parser_t *parser, const pm_n
     prettyprint_node(output_buffer, parser, node, &prefix_buffer);
     pm_buffer_free(&prefix_buffer);
 }
+
+#endif

prism/templates/src/serialize.c.erb

@@ -1,5 +1,10 @@
 #include "prism.h"
 
+// We optionally support serializing to a binary string. For systems that don't
+// want or need this functionality, it can be turned off with the
+// PRISM_EXCLUDE_SERIALIZATION define.
+#ifndef PRISM_EXCLUDE_SERIALIZATION
+
 #include <stdio.h>
 
 static inline uint32_t
@@ -394,23 +399,4 @@ pm_serialize_parse_lex(pm_buffer_t *buffer, const uint8_t *source, size_t size,
     pm_options_free(&options);
 }
 
-/**
- * Parse the source and return true if it parses without errors or warnings.
- */
-PRISM_EXPORTED_FUNCTION bool
-pm_parse_success_p(const uint8_t *source, size_t size, const char *data) {
-    pm_options_t options = { 0 };
-    pm_options_read(&options, data);
-
-    pm_parser_t parser;
-    pm_parser_init(&parser, source, size, &options);
-
-    pm_node_t *node = pm_parse(&parser);
-    pm_node_destroy(&parser, node);
-
-    bool result = parser.error_list.size == 0 && parser.warning_list.size == 0;
-    pm_parser_free(&parser);
-    pm_options_free(&options);
-
-    return result;
-}
+#endif