[Xml] More bounds-checking during xml parse to avoid certain common e…

…rrors.

Fix .to_xml to use indents and newlines to be slightly prettier.

src/unstable/xml/Document.winxed

@@ -40,12 +40,13 @@ class Rosella.Xml.Document
 
     function read_from_file(string filename)
     {
-        string xml = (new Rosella.FileSystem.File(filename)).read_all_text();
-        Rosella.Xml.__parse(xml, self);
+        var f = new Rosella.FileSystem.File(filename);
+        string xml = f.read_all_text();
+        Rosella.Xml.parse(xml, self);
     }
 
     function read_from_string(string xml)
     {
-        Rosella.Xml._parse(xml, self);
+        Rosella.Xml.parse(xml, self);
     }
 }

src/unstable/xml/Includes.winxed

@@ -1,3 +1,5 @@
+const string INDENT_STR = "  ";
+
 inline is_whitespace(int c) return int
 {
     return (c <= 0x20 || c >= 127); // Any non-printable characters and all whitespace
@@ -12,13 +14,31 @@ inline is_alphanumeric(int c) return int
             c == 0x5F || c == 0x2D);         // c == '_' || c == '-'
 }
 
+inline error_on_out_of_bounds(int c, var buffer, int j)
+{
+    if (c == 0) {
+        int buflen = elements(buffer);
+        if (j >= buflen)
+            Rosella.Error.error("Attempt to read position %d, passed the end of the buffer %d", j, buflen);
+
+    }
+}
+
+inline read_char(var buffer, int j) return int
+{
+    int c = buffer[j];
+    error_on_out_of_bounds(c, buffer, j);
+    return c;
+}
+
 inline eat_whitespace(var buffer, int j) return int
 {
     int c = buffer[j];
     while (is_whitespace(c)) {
         j++;
         c = buffer[j];
     }
+    error_on_out_of_bounds(c, buffer, j);
     return j;
 }
 

src/unstable/xml/Tag.winxed

@@ -37,23 +37,26 @@ class Rosella.Xml.Tag
     function to_xml()
     {
         var sb = new 'StringBuilder';
-        self.to_xml_sb(sb);
+        self.to_xml_sb(sb, 0);
         return sb;
     }
 
-    function to_xml_sb(var sb)
+    function to_xml_sb(var sb, int indent)
     {
+        string indent_str = "\n" + repeat_string(INDENT_STR, indent);
+        push(sb, indent_str);
         push(sb, "<");
         push(sb, self.name);
-        self.__to_xml_sb_attrs(sb);
+        self.__to_xml_sb_attrs(sb, indent);
         push(sb, ">");
-        self.__to_xml_sb_children(sb);
+        self.__to_xml_sb_children(sb, indent + 1);
+        push(sb, indent_str);
         push(sb, "</");
         push(sb, self.name);
         push(sb, ">");
     }
 
-    function __to_xml_sb_attrs(var sb)
+    function __to_xml_sb_attrs(var sb, int indent)
     {
         for (string attr_name in self.attributes) {
             string attr = self.attributes[attr_name];
@@ -65,18 +68,18 @@ class Rosella.Xml.Tag
         }
     }
 
-    function __to_xml_sb_children(var sb)
+    function __to_xml_sb_children(var sb, int indent)
     {
         for (var child in self.children)
-            child.to_xml_sb(sb);
+            child.to_xml_sb(sb, indent);
     }
 }
 
 class Rosella.Xml.Tag.XmlHeader : Rosella.Xml.Tag
 {
     function XmlHeader() { self.Tag(""); }
 
-    function to_xml_sb(var sb)
+    function to_xml_sb(var sb, int indent)
     {
         push(sb, "<?");
         push(sb, self.name);
@@ -97,12 +100,14 @@ class Rosella.Xml.Tag.Comment
     function to_xml()
     {
         var sb = new 'StringBuilder';
-        self.to_xml_sb(sb);
+        self.to_xml_sb(sb, 0);
         return sb;
     }
 
-    function to_xml_sb(var sb)
+    function to_xml_sb(var sb, int indent)
     {
+        string indent_str = "\n" + repeat_string(INDENT_STR, indent);
+        push(sb, indent_str);
         push(sb, "<!--");
         push(sb, self.text);
         push(sb, "-->");

src/unstable/xml/Text.winxed

@@ -11,12 +11,14 @@ class Rosella.Xml.Text
     function to_xml()
     {
         var sb = new 'StringBuilder';
-        self.to_xml_sb(sb);
+        self.to_xml_sb(sb, 0);
         return sb;
     }
 
-    function to_xml_sb(var sb)
+    function to_xml_sb(var sb, int indent)
     {
+        string indent_str = "\n" + repeat_string(INDENT_STR, indent);
+        push(sb, indent_str);
         push(sb, self.text);
     }
 }

src/unstable/xml/Xml.winxed

@@ -13,7 +13,7 @@ namespace Rosella.Xml
     }
 }
 
-class Rosella.Xml
+namespace Rosella.Xml
 {
     function read_file(string filename)
     {
@@ -29,7 +29,7 @@ class Rosella.Xml
         return document;
     }
 
-    function __parse(string xml, var document)
+    function parse(string xml, var document)
     {
         int open_bracket = get_codepoint("<", 0);
         int question_mark = get_codepoint("?", 0);
@@ -43,12 +43,13 @@ class Rosella.Xml
         int i = eat_whitespace(buffer, 0);
         int c = buffer[i];
         if (buf_length > 5 && c == open_bracket && int(buffer[i + 1]) == question_mark)
-            i = self.__parse_xml_header(buffer, i + 2, document);
+            i = Rosella.Xml.__parse_xml_header(buffer, i + 2, document);
 
+        error_on_out_of_bounds(c, buffer, i);
         for ( ; i < elements(buffer); ) {
             int c = buffer[i];
             if (c == open_bracket) {
-                :(var tag, i, int is_close, int contained, int is_comment) = self.__parse_tag(buffer, i + 1, current_tag);
+                :(var tag, i, int is_close, int contained, int is_comment) = Rosella.Xml.__parse_tag(buffer, i + 1, current_tag);
 
                 // A closing tag: </foo>
                 if (!is_comment && !contained && is_close) {
@@ -98,7 +99,7 @@ class Rosella.Xml
         int close_bracket = get_codepoint(">", 0);
         var header_tag = new Rosella.Xml.Tag.XmlHeader();
         i = eat_whitespace(buffer, i);
-        i = self.__parse_attributes(buffer, i, header_tag);
+        i = Rosella.Xml.__parse_attributes(buffer, i, header_tag);
         i = eat_whitespace(buffer, i);
         int c = buffer[i];
         if (c == question_mark && int(buffer[i + 1]) == close_bracket) {
@@ -119,7 +120,7 @@ class Rosella.Xml
         int j = i;
         int c = buffer[j];
         if (c == exclamation_point)
-            return self.__parse_comment(buffer, i + 1);
+            return Rosella.Xml.__parse_comment(buffer, i + 1);
         int is_close_tag = false;
 
         if (c == close_slash) {
@@ -135,7 +136,7 @@ class Rosella.Xml
 
         j = eat_whitespace(buffer, j);
         if (!is_close_tag)
-            j = self.__parse_attributes(buffer, j, tag);
+            j = Rosella.Xml.__parse_attributes(buffer, j, tag);
 
         c = buffer[j];
         if (c == close_bracket)
@@ -178,6 +179,7 @@ class Rosella.Xml
                 return comment, j + 3, true, true, true;
             }
         }
+        Rosella.Error.error("Unterminated comment starting at position %d", i - 2);
     }
 
     function __parse_attributes(var buffer, int i, var tag)
@@ -228,6 +230,7 @@ class Rosella.Xml
             }
             c = buffer[j];
         }
+        error_on_out_of_bounds(c, buffer, i);
         return j;
     }
 }