ruby_rnv 0.2.3 → 0.4.1

data/ext/rnv/src/xsd.c

@@ -1,5 +1,3 @@
-#include "type.h"
-
 /* $Id: xsd.c,v 1.47 2005/01/06 21:04:06 dvd Exp $ */
 
 #include <limits.h> /*INT_MAX*/
@@ -7,6 +5,9 @@
 #include <string.h> /*strlen*/
 #include <math.h> /*HUGE_VAL*/
 #include <assert.h>
+
+#include "type.h"
+
 #include "u.h"
 #include "xmlc.h"
 #include "s.h"
@@ -16,11 +17,10 @@
 #include "er.h"
 #include "xsd.h"
 
-#define err(msg) (*rnv->verror_handler)(rnv,erno|ERBIT_XSD,msg"\n",ap)
-void xsd_default_verror_handler(rnv_t *rnv, int erno,va_list ap) {
-  (*er_printf)("XML Schema datatypes: ");
+#define err(msg) (*handler)(data,erno|ERBIT_XSD,"XML Schema datatypes: "msg"\n",ap)
+void xsd_default_verror_handler(void *data, int erno, int (*handler)(void *data, int erno,char *format, va_list ap), va_list ap) {
   if(erno&ERBIT_RX) {
-    rx_default_verror_handler(rnv, erno&~ERBIT_RX,ap);
+    rx_default_verror_handler(data, erno&~ERBIT_RX,handler,ap);
   } else {
     switch(erno) {
     case XSD_ER_TYP: err("unknown type %s"); break;
@@ -36,24 +36,11 @@ void xsd_default_verror_handler(rnv_t *rnv, int erno,va_list ap) {
 }
 
 static void error_handler(rx_st_t *rx_st, int erno,...) {
-  va_list ap; va_start(ap,erno); (*rx_st->rnv->xsd_verror_handler)(rx_st->rnv,erno,ap); va_end(ap);
+  va_list ap; va_start(ap,erno); xsd_default_verror_handler(rx_st->user_data,erno,rx_st->verror_handler,ap); va_end(ap);
 }
 
-static void verror_handler_rx(rnv_t *rnv, int erno,va_list ap) {xsd_default_verror_handler(rnv,erno|ERBIT_RX,ap);}
-
-static void windup(void);
 void xsd_init(rx_st_t *rx_st) {
-    rx_st->rnv->xsd_verror_handler = &xsd_default_verror_handler;
     rx_init(rx_st);
-    rx_st->rnv->rx_verror_handler=&verror_handler_rx;
-    windup();
-}
-
-void xsd_clear(void) {
-  windup();
-}
-
-static void windup(void) {
 }
 
 #define FCT_ENUMERATION 0
@@ -81,7 +68,7 @@ static char *fcttab[NFCT]={
 #define WS_REPLACE 1
 #define WS_COLLAPSE 2
 
-static int (*match[])(rnv_t *rnv, rx_st_t *rx_st, char *r,char *s,int n)={&rx_match, &rx_rmatch,&rx_cmatch};
+static int (*match[])(rx_st_t *rx_st, char *r,char *s,int n)={&rx_match, &rx_rmatch,&rx_cmatch};
 
 #define TYP_ENTITIES 0
 #define TYP_ENTITY 1
@@ -327,7 +314,7 @@ struct facets {
 #define PAT_MONTH0 "(0[1-9]|1[0-2])"
 #define PAT_DAY0 "([0-2][0-9]|3[01])"
 #define PAT_YEAR "-?"PAT_YEAR0 PAT_ZONE"?"
-#define PAT_MONTH "--"PAT_MONTH0"--"PAT_ZONE"?"
+#define PAT_MONTH "--"PAT_MONTH0 PAT_ZONE"?"
 #define PAT_DAY "---"PAT_DAY0 PAT_ZONE"?"
 #define PAT_YEAR_MONTH "-?"PAT_YEAR0"-"PAT_MONTH0 PAT_ZONE"?"
 #define PAT_MONTH_DAY "--"PAT_MONTH0"-"PAT_DAY0 PAT_ZONE"?"
@@ -661,7 +648,7 @@ int xsd_allows(rx_st_t *rx_st, char *typ,char *ps,char *s,int n) {
   default: assert(0);
   }
 
-  while(fct.npat--) ok=ok&&match[fct.whiteSpace](rx_st->rnv, rx_st, fct.pattern[fct.npat],s,n);
+  while(fct.npat--) ok=ok&&match[fct.whiteSpace](rx_st, fct.pattern[fct.npat],s,n);
 
   if(fct.set&(1<<FCT_LENGTH)) ok=ok&&length==fct.length;
   if(fct.set&(1<<FCT_MAX_LENGTH)) ok=ok&&length<=fct.maxLength;

data/ext/rnv/src/xsd.h

@@ -14,10 +14,9 @@
 #define XSD_ER_WS 5
 #define XSD_ER_ENUM 6
 
-extern void xsd_default_verror_handler(rnv_t *rnv, int erno,va_list ap);
+extern void xsd_default_verror_handler(void *data, int erno, int (*handler)(void *data, int erno,char *format, va_list ap), va_list ap);
 
 extern void xsd_init(rx_st_t *rx_st);
-extern void xsd_clear(void);
 
 extern int xsd_allows(rx_st_t *rx_st, char *typ,char *ps,char *s,int n);
 extern int xsd_equal(rx_st_t *rx_st, char *typ,char *val,char *s,int n);

data/lib/rnv.rb

@@ -1,2 +1,5 @@
 require 'rnv/rnv'
+require 'rnv/error'
 require 'rnv/validator'
+require 'rnv/data_type_library'
+require 'rnv/pre_processor'

data/lib/rnv/data_type_library.rb

@@ -0,0 +1,21 @@
+module RNV
+  # Datatype library callback object
+  # @see https://www.oasis-open.org/committees/relax-ng/spec-20010811.html#IDA1I1R
+  class DataTypeLibrary
+    # @param [String] typ the data type
+    # @param [String] val
+    # @param [String] s
+    # @return [Boolean]
+    def equal(typ, val, s)
+      true
+    end
+
+    # @param [String] typ the data type
+    # @param [String] ps
+    # @param [String] s the value
+    # @return [Boolean]
+    def allows(typ, ps, s)
+      true
+    end
+  end
+end

data/lib/rnv/error.rb

@@ -0,0 +1,102 @@
+module RNV
+  class Error
+    attr_reader :original_message, :original_expected
+    attr_reader :ns_uri, :element, :attr, :value, :required, :allowed
+
+    # @return [String]
+    def message
+      @original_message
+    end
+
+    # @return [String]
+    def expected
+      @original_expected
+    end
+
+    # @return [String]
+    def compact_message
+      case @code
+      when :rnv_er_elem
+        "element <#{@element}> is not allowed"
+      when :rnv_er_ufin
+        "element <#{@element}> is incomplete"
+      when :rnv_er_akey
+        "attribute #{@attr} is not allowed"
+      when :rnv_er_aval
+        "attribute #{@attr} has invalid value '#{@value}'"
+      when :rnv_er_amis
+        "element <#{@element}> attribute missing"
+      else
+        @original_message
+      end
+    end
+
+    # @return [String]
+    def compact_expected
+      [(compact_required_set.length > 0 ? compact_expected_phrase(compact_required_set) +" required" : nil),
+      (compact_allowed_set.length > 0 ? compact_expected_phrase(compact_allowed_set) + " allowed" : nil)].join("; ")
+    end
+
+    # @param [String] data original content
+    # @return [String] context
+    def contextualize(data, max_len = 60)
+      out = ""
+      if data and @line
+        err_data = data.split("\n")[@line - 1]
+        if err_data
+          err_data.insert(@col - 1, "🢑")
+
+          start = 0
+          if @col > max_len
+            start = @col - max_len
+          end
+          out += (start > 0 ? "…" : "") + err_data[start..(@col + max_len)].strip + (@col + max_len < err_data.length ? "…" : "")
+        end
+      end
+      out
+    end
+
+    # @return [String]
+    def to_s
+      "#{@line}:#{@col}:error: #{self.message}\n#{self.expected}"
+    end
+
+    # @return [String]
+    def inspect
+      "#<RNV::Error code: :#{@code}, message: '#{self.message}', expected: '#{self.expected}, line: #{@line}, column: #{@col}>"
+    end
+
+    private
+
+    def compact_expected_to_s(e, with_ns = false)
+      case e.first
+      when :rn_p_attribute
+        compact_expected_to_s(e.last, with_ns)
+      when :rn_p_element
+        "<"+compact_expected_to_s(e.last, with_ns)+">"
+      when :rn_p_data
+        compact_expected_to_s(e.last, with_ns)
+      when :rn_nc_datatype
+        "datatype #{e[-2]}:#{e.last}"
+      when :rn_nc_qname
+        with_ns ? "#{e.first}:#{e.last}" : e.last
+      end
+    end
+
+    def compact_required_set
+      @compact_required_set ||= @required.map{|r| compact_expected_to_s(r)}.compact.uniq
+    end
+
+    def compact_allowed_set
+      @compact_allowed_set ||= @allowed.map{|r| compact_expected_to_s(r)}.compact.uniq
+    end
+
+    def compact_expected_phrase(set)
+      if set.length > 1
+        set[0..-2].join(", ") + " or "+set.last
+      else
+        set.last
+      end
+    end
+  end
+end

data/lib/rnv/pre_processor.rb

@@ -0,0 +1,25 @@
+module RNV
+  # Object to modify data before validation (eg: HTML5 data-* attributes)
+  class PreProcessor
+    # replace attributes before validate
+    # @param [Array<Array<String>>] attrs
+    # @return [Array<Array<String>>]
+    def attributes(attrs)
+      attrs
+    end
+
+    # replace tag name before validate
+    # @param [String] tag
+    # @return [String]
+    def tag(tag)
+      tag
+    end
+
+    # replace content text before validate
+    # @param [String] txt
+    # @return [String]
+    def text(txt)
+      txt
+    end
+  end
+end

data/lib/rnv/validator.rb

@@ -1,37 +1,40 @@
 require 'nokogiri'
 require 'rnv/rnv'
+require 'rnv/pre_processor'
 
 module RNV
+  class DocumentEmpty < StandardError; end
+
   # @!visibility private
   class NokogiriSaxDocument < Nokogiri::XML::SAX::Document
     # @return [Nokogiri::XML::SAX::ParserContext]
-    attr_accessor :ctx
+    attr_accessor :ctx, :pre_processor
     # @param [RNV::Document] document
     def initialize(document)
       @document = document
     end
 
     def start_element_namespace(name, attrs = [], prefix = nil, uri = nil, ns = nil)
-      update_line_col
       tag_attrs = attrs.map { |attr| [attr.uri ? "#{attr.uri}:#{attr.localname}" : attr.localname, attr.value] }
       tag_name = uri ? "#{uri}:#{name}" : name
-      @document.start_tag(tag_name, tag_attrs.flatten)
+      @document.start_tag(@pre_processor.tag(tag_name), @pre_processor.attributes(tag_attrs))
+      update_line_col
     end
 
     def end_element_namespace(name, prefix = nil, uri = nil)
-      update_line_col
       tag_name = uri ? "#{uri}:#{name}" : name
-      @document.end_tag(tag_name)
+      @document.end_tag(@pre_processor.tag(tag_name))
+      update_line_col
     end
 
     def characters str
+      @document.characters(@pre_processor.text(str))
       update_line_col
-      @document.characters(str)
     end
 
     def cdata_block str
+      @document.characters(@pre_processor.text(str))
       update_line_col
-      @document.characters(str)
     end
 
     private
@@ -57,25 +60,31 @@ module RNV
     end
 
     # @param [String] str RNC schema buffer
+    # @return [Boolean] true if schema loaded successfuly
     def load_schema_from_string(str)
       @document.load_string(str)
     end
 
     # @param [String] file RNC schema filename
+    # @return [Boolean] true if schema loaded successfuly
     def load_schema_from_file(file)
       @document.load_file(file)
     end
 
     # parse and validate buffer
     # @param [String] str XML buffer to parse
+    # @param [RNV::PreProcessor] pre_processor an optional pre-processor for tag and attributes data
     # @return [Boolean] true if valid
-    def parse_string(str)
-      @document.errors = [] # reset errors
+    def parse_string(str, pre_processor = PreProcessor.new)
       @document.start_document
       rnv_doc = NokogiriSaxDocument.new(@document)
+      rnv_doc.pre_processor = pre_processor
+
+      raise RNV::DocumentEmpty if str.nil? or str.empty?
 
       parser = Nokogiri::XML::SAX::Parser.new(rnv_doc)
       parser.parse_memory(str) do |ctx|
+        ctx.replace_entities = true
         rnv_doc.ctx = ctx
       end
 
@@ -84,16 +93,20 @@ module RNV
 
     # parse and validate file
     # @param [String, File] xml XML file to parse
+    # @param [RNV::PreProcessor] pre_processor an optional pre-processor for tag and attributes data
     # @return [Boolean] true if valid
-    def parse_file(xml)
-      @document.errors = [] # reset errors
+    def parse_file(xml, pre_processor = PreProcessor.new)
       @document.start_document
       rnv_doc = NokogiriSaxDocument.new(@document)
+      rnv_doc.pre_processor = pre_processor
 
       file = xml.is_a?(File) ? xml : File.open(xml)
 
+      raise RNV::DocumentEmpty if file.size == 0
+
       parser = Nokogiri::XML::SAX::Parser.new(rnv_doc)
       parser.parse(file) do |ctx|
+        ctx.replace_entities = true
         rnv_doc.ctx = ctx
       end
 

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: ruby_rnv
 version: !ruby/object:Gem::Version
-  version: 0.2.3
+  version: 0.4.1
 platform: ruby
 authors:
 - Julien Boulnois
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2021-04-08 00:00:00.000000000 Z
+date: 2021-04-12 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bundler
@@ -104,6 +104,8 @@ files:
 - Gemfile
 - ext/rnv/extconf.rb
 - ext/rnv/ruby_rnv.c
+- ext/rnv/ruby_rnv.h
+- ext/rnv/ruby_rnv_err.c
 - ext/rnv/src/ary.c
 - ext/rnv/src/ary.h
 - ext/rnv/src/drv.c
@@ -147,6 +149,9 @@ files:
 - ext/rnv/src/xsd_tm.c
 - ext/rnv/src/xsd_tm.h
 - lib/rnv.rb
+- lib/rnv/data_type_library.rb
+- lib/rnv/error.rb
+- lib/rnv/pre_processor.rb
 - lib/rnv/validator.rb
 homepage: https://github.com/immateriel/ruby_rnv
 licenses: