ox 2.9.4 → 2.10.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 9c220a3c8bc6e3606d936684b2cdde97146131cdee936c9a66bc8c4ce0473b90
-  data.tar.gz: 96e146a36eff4184918913f27cffb23a81de87835b14c4b30be612d13fb385b8
+  metadata.gz: e1967ed8377e3a3e7ccdb8b6e14a21e5dbd0c7f435b87474c422cf3572d639a9
+  data.tar.gz: cdb5712d31fd0641c9f444e8ea205fb18c3ccb246ba651d31b8dbd34db7ba579
 SHA512:
-  metadata.gz: 1b61e28ecced2336af416e61907c9f4b279c6f7b37df75a8d73a449fda4aac7ba62583255ba2b136f05286f69f840734d5bdd8573da5f11393106c3e6bd082c2
-  data.tar.gz: 703e1df77bb2da424be80c9f0491f57bf021ee312add658062c0c6fe736429fab3fe589a273cf32646ca331ec4e21c435fa51d4bfa2c8cedbc560f7e1b9f69ed
+  metadata.gz: 52f78021c4525f0c88548c2ce87a2a660d0ce60ea72ffef159611a4b39b883ce497adebf3a13c22f3c744325f4c463c4493ab66b3efbda7c975e99b0d46e7c71
+  data.tar.gz: b505855a76c09de2bb9c07b821b509e1d8d58ee3ce45816db6612086b10bb748dfe1c9ca07da8baab077a5ffdbc33cd3ed773fbd14f9cb7e8c2108b2e935797e

data/CHANGELOG.md

@@ -1,4 +1,9 @@
 
+## 2.10.0 - August 26, 2018
+
+  - `:element_key_mod` and `:attr_key_mod` options were added to allow keys to
+    be modified when loading.
+
 ## 2.9.4 - July 16, 2018
 
   - Fixed issue with malformed object mode input.

data/ext/ox/gen_load.c

@@ -84,10 +84,10 @@ create_doc(PInfo pi) {
 
 static void
 create_prolog_doc(PInfo pi, const char *target, Attr attrs) {
-    VALUE       doc;
-    VALUE       ah;
-    VALUE       nodes;
-    VALUE	sym;
+    volatile VALUE       doc;
+    volatile VALUE       ah;
+    volatile VALUE       nodes;
+    volatile VALUE	sym;
 
     if (!helper_stack_empty(&pi->helpers)) { /* top level object */
         ox_err_set(&pi->err, rb_eSyntaxError, "Prolog must be the first element in an XML document.\n");
@@ -96,7 +96,10 @@ create_prolog_doc(PInfo pi, const char *target, Attr attrs) {
     doc = rb_obj_alloc(ox_document_clas);
     ah = rb_hash_new();
     for (; 0 != attrs->name; attrs++) {
-	if (Yes == pi->options->sym_keys) {
+	if (Qnil != pi->options->attr_key_mod) {
+	    sym = rb_funcall(pi->options->attr_key_mod, ox_call_id, 1, rb_str_new2(attrs->name));
+	    rb_hash_aset(ah, sym, rb_str_new2(attrs->value));
+	} else if (Yes == pi->options->sym_keys) {
 #if HAS_ENCODING_SUPPORT
 	    if (0 != pi->options->rb_enc) {
 		VALUE	rstr = rb_str_new2(attrs->name);
@@ -120,7 +123,7 @@ create_prolog_doc(PInfo pi, const char *target, Attr attrs) {
 #endif
 	    rb_hash_aset(ah, sym, rb_str_new2(attrs->value));
 	} else {
-	    VALUE	rstr = rb_str_new2(attrs->name);
+	    volatile VALUE	rstr = rb_str_new2(attrs->name);
 
 #if HAS_ENCODING_SUPPORT
 	    if (0 != pi->options->rb_enc) {
@@ -291,6 +294,9 @@ add_element(PInfo pi, const char *ename, Attr attrs, int hasChildren) {
     VALUE       e;
     VALUE       s = rb_str_new2(ename);
 
+    if (Qnil != pi->options->element_key_mod) {
+	s = rb_funcall(pi->options->element_key_mod, ox_call_id, 1, s);
+    }
 #if HAS_ENCODING_SUPPORT
     if (0 != pi->options->rb_enc) {
         rb_enc_associate(s, pi->options->rb_enc);
@@ -308,7 +314,9 @@ add_element(PInfo pi, const char *ename, Attr attrs, int hasChildren) {
         for (; 0 != attrs->name; attrs++) {
             volatile VALUE	sym;
 
-	    if (Yes == pi->options->sym_keys) {
+	    if (Qnil != pi->options->attr_key_mod) {
+		sym = rb_funcall(pi->options->attr_key_mod, ox_call_id, 1, rb_str_new2(attrs->name));
+	    } else if (Yes == pi->options->sym_keys) {
 		VALUE	*slot;
 
 		if (Qundef == (sym = ox_cache_get(ox_symbol_cache, attrs->name, &slot, 0))) {
@@ -415,13 +423,15 @@ add_instruct(PInfo pi, const char *name, Attr attrs, const char *content) {
     if (0 != content) {
 	rb_ivar_set(inst, ox_at_content_id, c);
     } else if (0 != attrs->name) {
-        VALUE   ah = rb_hash_new();
+        volatile VALUE   ah = rb_hash_new();
         
         for (; 0 != attrs->name; attrs++) {
-            VALUE   sym;
-            VALUE   *slot;
+            volatile VALUE	sym;
+	    VALUE		*slot;
 
-	    if (Yes == pi->options->sym_keys) {
+	    if (Qnil != pi->options->attr_key_mod) {
+		sym = rb_funcall(pi->options->attr_key_mod, ox_call_id, 1, rb_str_new2(attrs->name));
+	    } else if (Yes == pi->options->sym_keys) {
 		if (Qundef == (sym = ox_cache_get(ox_symbol_cache, attrs->name, &slot, 0))) {
 #if HAS_ENCODING_SUPPORT
 		    if (0 != pi->options->rb_enc) {

data/ext/ox/hash_load.c

@@ -71,7 +71,9 @@ add_element(PInfo pi, const char *ename, Attr attrs, int hasChildren) {
 	volatile VALUE	a;
 	
         for (; 0 != attrs->name; attrs++) {
-	    if (Yes == pi->options->sym_keys) {
+	    if (Qnil != pi->options->attr_key_mod) {
+		key = rb_funcall(pi->options->attr_key_mod, ox_call_id, 1, rb_str_new2(attrs->name));
+	    } else if (Yes == pi->options->sym_keys) {
 		key = rb_id2sym(rb_intern(attrs->name));
 	    } else {
 		key = rb_str_new2(attrs->name);
@@ -125,7 +127,9 @@ end_element_core(PInfo pi, const char *ename, bool check_taint) {
     if (NoCode == e->type) {
 	e->obj = Qnil;
     }
-    if (Yes == pi->options->sym_keys) {
+    if (Qnil != pi->options->element_key_mod) {
+	key = rb_funcall(pi->options->element_key_mod, ox_call_id, 1, rb_id2str(e->var));
+    } else if (Yes == pi->options->sym_keys) {
 	key = rb_id2sym(e->var);
     } else {
 	key = rb_id2str(e->var);

data/ext/ox/ox.c

@@ -40,6 +40,7 @@ ID	ox_attributes_id;
 ID	ox_attrs_done_id;
 ID	ox_beg_id;
 ID	ox_bigdecimal_id;
+ID	ox_call_id;
 ID	ox_cdata_id;
 ID	ox_comment_id;
 ID	ox_den_id;
@@ -109,6 +110,7 @@ Cache	ox_attr_cache = 0;
 
 static VALUE	abort_sym;
 static VALUE	active_sym;
+static VALUE	attr_key_mod_sym;
 static VALUE	auto_define_sym;
 static VALUE	auto_sym;
 static VALUE	block_sym;
@@ -145,6 +147,7 @@ static VALUE	with_dtd_sym;
 static VALUE	with_instruct_sym;
 static VALUE	with_xml_sym;
 static VALUE	xsd_date_sym;
+static VALUE 	element_key_mod_sym;
 
 static ID	encoding_id;
 static ID	has_key_id;
@@ -158,30 +161,32 @@ void		*ox_utf8_encoding = 0;
 #endif
 
 struct _Options	 ox_default_options = {
-    { '\0' },		/* encoding */
-    { '\0' },		/* margin */
-    2,			/* indent */
-    0,			/* trace */
-    0,			/* margin_len */
-    No,			/* with_dtd */
-    No,			/* with_xml */
-    No,			/* with_instruct */
-    No,			/* circular */
-    No,			/* xsd_date */
-    NoMode,		/* mode */
-    StrictEffort,	/* effort */
-    Yes,		/* sym_keys */
-    SpcSkip,		/* skip */
-    No,			/* smart */
-    1,			/* convert_special */
-    No,			/* allow_invalid */
-    { '\0' },		/* inv_repl */
-    { '\0' },		/* strip_ns */
-    NULL,		/* html_hints */
+    { '\0' },		// encoding
+    { '\0' },		// margin
+    2,			// indent
+    0,			// trace
+    0,			// margin_len
+    No,			// with_dtd
+    No,			// with_xml
+    No,			// with_instruct
+    No,			// circular
+    No,			// xsd_date
+    NoMode,		// mode
+    StrictEffort,	// effort
+    Yes,		// sym_keys
+    SpcSkip,		// skip
+    No,			// smart
+    1,			// convert_special
+    No,			// allow_invalid
+    { '\0' },		// inv_repl
+    { '\0' },		// strip_ns
+    NULL,		// html_hints
+    Qnil,		// attr_key_mod;
+    Qnil,		// element_key_mod;
 #if HAS_PRIVATE_ENCODING
-    Qnil		/* rb_enc */
+    Qnil		// rb_enc
 #else
-    0			/* rb_enc */
+    0			// rb_enc
 #endif
 };
 
@@ -197,7 +202,7 @@ static void	parse_dump_options(VALUE ropts, Options copts);
 static char*
 defuse_bom(char *xml, Options options) {
     switch ((uint8_t)*xml) {
-    case 0xEF: /* UTF-8 */
+    case 0xEF: // UTF-8
 	if (0xBB == (uint8_t)xml[1] && 0xBF == (uint8_t)xml[2]) {
 	    options->rb_enc = ox_utf8_encoding;
 	    xml += 3;
@@ -206,7 +211,7 @@ defuse_bom(char *xml, Options options) {
 	}
 	break;
 #if 0
-    case 0xFE: /* UTF-16BE */
+    case 0xFE: // UTF-16BE
 	if (0xFF == (uint8_t)xml[1]) {
 	    options->rb_enc = ox_utf16be_encoding;
 	    xml += 2;
@@ -214,7 +219,7 @@ defuse_bom(char *xml, Options options) {
 	    rb_raise(ox_parse_error_class, "Invalid BOM in XML string.\n");
 	}
 	break;
-    case 0xFF: /* UTF-16LE or UTF-32LE */
+    case 0xFF: // UTF-16LE or UTF-32LE
 	if (0xFE == (uint8_t)xml[1]) {
 	    if (0x00 == (uint8_t)xml[2] && 0x00 == (uint8_t)xml[3]) {
 		options->rb_enc = ox_utf32le_encoding;
@@ -227,7 +232,7 @@ defuse_bom(char *xml, Options options) {
 	    rb_raise(ox_parse_error_class, "Invalid BOM in XML string.\n");
 	}
 	break;
-    case 0x00: /* UTF-32BE */
+    case 0x00: // UTF-32BE
 	if (0x00 == (uint8_t)xml[1] && 0xFE == (uint8_t)xml[2] && 0xFF == (uint8_t)xml[3]) {
 	    options->rb_enc = ox_utf32be_encoding;
 	    xml += 4;
@@ -237,7 +242,8 @@ defuse_bom(char *xml, Options options) {
 	break;
 #endif
     default:
-	/* Let it fail if there is a BOM that is not UTF-8. Other BOM options are not ASCII compatible. */
+	// Let it fail if there is a BOM that is not UTF-8. Other BOM options
+	// are not ASCII compatible.
 	break;
     }
     return xml;
@@ -280,6 +286,8 @@ hints_to_overlay(Hints hints) {
  * - _:mode_ [:object|:generic|:limited|:hash|:hash_no_attrs|nil] load method to use for XML
  * - _:effort_ [:strict|:tolerant|:auto_define] set the tolerance level for loading
  * - _:symbolize_keys_ [true|false|nil] symbolize element attribute keys or leave as Strings
+ * - _:element_key_mod_ [Proc|nil] converts element keys on parse if not nil
+ * - _:attr_key_mod_ [Proc|nil] converts attribute keys on parse if not nil
  * - _:skip_ [:skip_none|:skip_return|:skip_white|:skip_off] determines how to handle white space in text
  * - _:smart_ [true|false|nil] flag indicating the SAX parser uses hints if available (use with html)
  * - _:convert_special_ [true|false|nil] flag indicating special characters like < are converted with the SAX parser
@@ -313,6 +321,8 @@ get_def_opts(VALUE self) {
     rb_hash_aset(opts, circular_sym, (Yes == ox_default_options.circular) ? Qtrue : ((No == ox_default_options.circular) ? Qfalse : Qnil));
     rb_hash_aset(opts, xsd_date_sym, (Yes == ox_default_options.xsd_date) ? Qtrue : ((No == ox_default_options.xsd_date) ? Qfalse : Qnil));
     rb_hash_aset(opts, symbolize_keys_sym, (Yes == ox_default_options.sym_keys) ? Qtrue : ((No == ox_default_options.sym_keys) ? Qfalse : Qnil));
+    rb_hash_aset(opts, attr_key_mod_sym, ox_default_options.attr_key_mod);
+    rb_hash_aset(opts, element_key_mod_sym, ox_default_options.element_key_mod);
     rb_hash_aset(opts, smart_sym, (Yes == ox_default_options.smart) ? Qtrue : ((No == ox_default_options.smart) ? Qfalse : Qnil));
     rb_hash_aset(opts, convert_special_sym, (ox_default_options.convert_special) ? Qtrue : Qfalse);
     switch (ox_default_options.mode) {
@@ -416,6 +426,8 @@ sax_html_overlay(VALUE self) {
  *   - _:mode_ [:object|:generic|:limited|:hash|:hash_no_attrs|nil] load method to use for XML
  *   - _:effort_ [:strict|:tolerant|:auto_define] set the tolerance level for loading
  *   - _:symbolize_keys_ [true|false|nil] symbolize element attribute keys or leave as Strings
+ *   - _:element_key_mod_ [Proc|nil] converts element keys on parse if not nil
+ *   - _:attr_key_mod_ [Proc|nil] converts attribute keys on parse if not nil
  *   - _:skip_ [:skip_none|:skip_return|:skip_white|:skip_off] determines how to handle white space in text
  *   - _:smart_ [true|false|nil] flag indicating the SAX parser uses hints if available (use with html)
  *   - _:invalid_replace_ [nil|String] replacement string for invalid XML characters on dump. nil indicates include anyway as hex. A string, limited to 10 characters will replace the invalid character with the replace.
@@ -611,6 +623,9 @@ set_def_opts(VALUE self, VALUE opts) {
 	    rb_hash_foreach(v, set_overlay, (VALUE)ox_default_options.html_hints);
 	}
     }
+    ox_default_options.element_key_mod = rb_hash_lookup2(opts, element_key_mod_sym, ox_default_options.element_key_mod);
+    ox_default_options.attr_key_mod = rb_hash_lookup2(opts, attr_key_mod_sym, ox_default_options.attr_key_mod);
+    
     return Qnil;
 }
 
@@ -755,6 +770,9 @@ load(char *xml, size_t len, int argc, VALUE *argv, VALUE self, VALUE encoding, E
 	if (Qnil != (v = rb_hash_lookup(h, symbolize_keys_sym))) {
 	    options.sym_keys = (Qfalse == v) ? No : Yes;
 	}
+	options.element_key_mod = rb_hash_lookup2(h, element_key_mod_sym, options.element_key_mod);
+	options.attr_key_mod = rb_hash_lookup2(h, attr_key_mod_sym, options.attr_key_mod);
+
 	if (Qnil != (v = rb_hash_lookup(h, convert_special_sym))) {
 	    options.convert_special = (Qfalse != v);
 	}
@@ -1379,6 +1397,7 @@ void Init_ox() {
     ox_attrs_done_id = rb_intern("attrs_done");
     ox_beg_id = rb_intern("@beg");
     ox_bigdecimal_id = rb_intern("BigDecimal");
+    ox_call_id = rb_intern("call");
     ox_cdata_id = rb_intern("cdata");
     ox_comment_id = rb_intern("comment");
     ox_den_id = rb_intern("@den");
@@ -1442,12 +1461,14 @@ void Init_ox() {
 
     abort_sym = ID2SYM(rb_intern("abort"));			rb_gc_register_address(&abort_sym);
     active_sym = ID2SYM(rb_intern("active"));			rb_gc_register_address(&active_sym);
+    attr_key_mod_sym = ID2SYM(rb_intern("attr_key_mod"));	rb_gc_register_address(&attr_key_mod_sym);
     auto_define_sym = ID2SYM(rb_intern("auto_define"));		rb_gc_register_address(&auto_define_sym);
     auto_sym = ID2SYM(rb_intern("auto"));			rb_gc_register_address(&auto_sym);
     block_sym = ID2SYM(rb_intern("block"));			rb_gc_register_address(&block_sym);
     circular_sym = ID2SYM(rb_intern("circular"));		rb_gc_register_address(&circular_sym);
     convert_special_sym = ID2SYM(rb_intern("convert_special")); rb_gc_register_address(&convert_special_sym);
     effort_sym = ID2SYM(rb_intern("effort"));			rb_gc_register_address(&effort_sym);
+    element_key_mod_sym = ID2SYM(rb_intern("element_key_mod"));	rb_gc_register_address(&element_key_mod_sym);
     generic_sym = ID2SYM(rb_intern("generic"));			rb_gc_register_address(&generic_sym);
     hash_no_attrs_sym = ID2SYM(rb_intern("hash_no_attrs"));	rb_gc_register_address(&hash_no_attrs_sym);
     hash_sym = ID2SYM(rb_intern("hash"));			rb_gc_register_address(&hash_sym);

data/ext/ox/ox.h

@@ -143,6 +143,8 @@ typedef struct _Options {
     char		inv_repl[12];	/* max 10 valid characters, first character is the length */
     char		strip_ns[64];	/* namespace to strip, \0 is no-strip, \* is all, else only matches */
     struct _Hints	*html_hints;	/* html hints */
+    VALUE		attr_key_mod;
+    VALUE		element_key_mod;
 #if HAS_ENCODING_SUPPORT
     rb_encoding		*rb_enc;
 #elif HAS_PRIVATE_ENCODING
@@ -192,6 +194,7 @@ extern ID	ox_attrs_done_id;
 extern ID	ox_attributes_id;
 extern ID	ox_beg_id;
 extern ID	ox_bigdecimal_id;
+extern ID	ox_call_id;
 extern ID	ox_cdata_id;
 extern ID	ox_comment_id;
 extern ID	ox_den_id;

data/lib/ox/version.rb

@@ -1,5 +1,5 @@
 
 module Ox
   # Current version of the module.
-  VERSION = '2.9.4'
+  VERSION = '2.10.0'
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: ox
 version: !ruby/object:Gem::Version
-  version: 2.9.4
+  version: 2.10.0
 platform: ruby
 authors:
 - Peter Ohler
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2018-07-16 00:00:00.000000000 Z
+date: 2018-09-03 00:00:00.000000000 Z
 dependencies: []
 description: "A fast XML parser and object serializer that uses only standard C lib.\n
   \           \nOptimized XML (Ox), as the name implies was written to provide speed
@@ -35,7 +35,6 @@ files:
 - ext/ox/cache8.c
 - ext/ox/cache8.h
 - ext/ox/dump.c
-- ext/ox/encode.h
 - ext/ox/err.c
 - ext/ox/err.h
 - ext/ox/extconf.rb
@@ -100,7 +99,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
       version: '0'
 requirements: []
 rubyforge_project: ox
-rubygems_version: 2.7.3
+rubygems_version: 2.7.6
 signing_key: 
 specification_version: 4
 summary: A fast XML parser and object serializer.

data/ext/ox/encode.h

@@ -1,26 +0,0 @@
-/* encode.h
- * Copyright (c) 2011, Peter Ohler
- * All rights reserved.
- */
-
-#ifndef __OX_ENCODE_H__
-#define __OX_ENCODE_H__
-
-#include "ruby.h"
-#if HAS_ENCODING_SUPPORT
-#include "ruby/encoding.h"
-#endif
-
-static inline VALUE
-ox_encode(VALUE rstr) {
-#if HAS_ENCODING_SUPPORT
-    rb_enc_associate(rstr, ox_utf8_encoding);
-#else
-    if (Qnil != ox_utf8_encoding) {
-	rstr = rb_funcall(ox_utf8_encoding, ox_iconv_id, 1, rstr);
-    }
-#endif
-    return rstr;
-}
-
-#endif /* __OX_ENCODE_H__ */