Introduced encoding_latin1 support (conversion on loading, conversion on saving, encoding name in declaration in document::save)

git-svn-id: http://pugixml.googlecode.com/svn/trunk@829 99668b35-9821-0410-8761-19e4c4f06640

6 files changed, 189 insertions, 12 deletions

diff --git a/src/pugixml.cpp b/src/pugixml.cpp
index a6196dc..eb924a2 100644
--- a/src/pugixml.cpp
+++ b/src/pugixml.cpp
@@ -754,6 +754,27 @@ namespace
 		}
 	};
 
+	struct latin1_writer
+	{
+		typedef uint8_t* value_type;
+
+		static value_type low(value_type result, uint32_t ch)
+		{
+			*result = static_cast<uint8_t>(ch > 255 ? '?' : ch);
+
+			return result + 1;
+		}
+
+		static value_type high(value_type result, uint32_t ch)
+		{
+            (void)ch;
+
+			*result = '?';
+
+			return result + 1;
+		}
+	};
+
 	template <size_t size> struct wchar_selector;
 
 	template <> struct wchar_selector<2>
@@ -904,6 +925,16 @@ namespace
 
 			return result;
 		}
+
+		static inline typename Traits::value_type decode_latin1_block(const uint8_t* data, size_t size, typename Traits::value_type result)
+		{
+			for (size_t i = 0; i < size; ++i)
+            {
+                result = Traits::low(result, data[i]);
+            }
+
+			return result;
+		}
 	};
 
 	template <typename T> inline void convert_utf_endian_swap(T* result, const T* data, size_t length)
@@ -1172,6 +1203,27 @@ namespace
 		return true;
 	}
 
+	bool convert_buffer_latin1(char_t*& out_buffer, size_t& out_length, const void* contents, size_t size)
+	{
+		const uint8_t* data = static_cast<const uint8_t*>(contents);
+
+		// get length in wchar_t units
+		out_length = size;
+
+		// allocate buffer of suitable length
+		out_buffer = static_cast<char_t*>(global_allocate((out_length > 0 ? out_length : 1) * sizeof(char_t)));
+		if (!out_buffer) return false;
+
+		// convert latin1 input to wchar_t
+		wchar_writer::value_type out_begin = reinterpret_cast<wchar_writer::value_type>(out_buffer);
+		wchar_writer::value_type out_end = utf_decoder<wchar_writer>::decode_latin1_block(data, size, out_begin);
+
+		assert(out_end == out_begin + out_length);
+		(void)!out_end;
+
+		return true;
+	}
+
 	bool convert_buffer(char_t*& out_buffer, size_t& out_length, xml_encoding encoding, const void* contents, size_t size, bool is_mutable)
 	{
 		// get native encoding
@@ -1206,6 +1258,9 @@ namespace
 				convert_buffer_utf32(out_buffer, out_length, contents, size, opt_true());
 		}
 
+        // source encoding is latin1
+		if (encoding == encoding_latin1) return convert_buffer_latin1(out_buffer, out_length, contents, size);
+
 		assert(!"Invalid encoding");
 		return false;
 	}
@@ -1254,6 +1309,48 @@ namespace
 		return true;
 	}
 
+    size_t get_latin1_7bit_prefix_length(const uint8_t* data, size_t size)
+    {
+        for (size_t i = 0; i < size; ++i)
+            if (data[i] > 127)
+                return i;
+
+        return size;
+    }
+
+	bool convert_buffer_latin1(char_t*& out_buffer, size_t& out_length, const void* contents, size_t size, bool is_mutable)
+	{
+		const uint8_t* data = static_cast<const uint8_t*>(contents);
+
+        // get size of prefix that does not need utf8 conversion
+        size_t prefix_length = get_latin1_7bit_prefix_length(data, size);
+        assert(prefix_length <= size);
+
+        const uint8_t* postfix = data + prefix_length;
+        size_t postfix_length = size - prefix_length;
+
+        // if no conversion is needed, just return the original buffer
+        if (postfix_length == 0) return get_mutable_buffer(out_buffer, out_length, contents, size, is_mutable);
+
+		// first pass: get length in utf8 units
+		out_length = prefix_length + utf_decoder<utf8_counter>::decode_latin1_block(postfix, postfix_length, 0);
+
+		// allocate buffer of suitable length
+		out_buffer = static_cast<char_t*>(global_allocate((out_length > 0 ? out_length : 1) * sizeof(char_t)));
+		if (!out_buffer) return false;
+
+		// second pass: convert latin1 input to utf8
+        memcpy(out_buffer, data, prefix_length);
+
+		uint8_t* out_begin = reinterpret_cast<uint8_t*>(out_buffer);
+		uint8_t* out_end = utf_decoder<utf8_writer>::decode_latin1_block(postfix, postfix_length, out_begin + prefix_length);
+
+		assert(out_end == out_begin + out_length);
+		(void)!out_end;
+
+		return true;
+	}
+
 	bool convert_buffer(char_t*& out_buffer, size_t& out_length, xml_encoding encoding, const void* contents, size_t size, bool is_mutable)
 	{
 		// fast path: no conversion required
@@ -1279,6 +1376,9 @@ namespace
 				convert_buffer_utf32(out_buffer, out_length, contents, size, opt_true());
 		}
 
+        // source encoding is latin1
+		if (encoding == encoding_latin1) return convert_buffer_latin1(out_buffer, out_length, contents, size, is_mutable);
+
 		assert(!"Invalid encoding");
 		return false;
 	}
@@ -2580,6 +2680,18 @@ namespace
 			return static_cast<size_t>(end - dest) * sizeof(uint32_t);
 		}
 
+		// convert to latin1
+		if (encoding == encoding_latin1)
+		{
+			uint8_t* dest = reinterpret_cast<uint8_t*>(result);
+
+			uint8_t* end = sizeof(wchar_t) == 2 ?
+				utf_decoder<latin1_writer>::decode_utf16_block(reinterpret_cast<const uint16_t*>(data), length, dest) :
+				utf_decoder<latin1_writer>::decode_utf32_block(reinterpret_cast<const uint32_t*>(data), length, dest);
+
+			return static_cast<size_t>(end - dest);
+		}
+
 		assert(!"Invalid encoding");
 		return 0;
 	}
@@ -2632,6 +2744,14 @@ namespace
 			return static_cast<size_t>(end - dest) * sizeof(uint32_t);
 		}
 
+		if (encoding == encoding_latin1)
+		{
+			uint8_t* dest = reinterpret_cast<uint8_t*>(result);
+			uint8_t* end = utf_decoder<latin1_writer>::decode_utf8_block(reinterpret_cast<const uint8_t*>(data), length, dest);
+
+			return static_cast<size_t>(end - dest);
+		}
+
 		assert(!"Invalid encoding");
 		return 0;
 	}
@@ -2822,6 +2942,9 @@ namespace
 			writer.write("\xff\xfe\x00\x00", 4);
 			break;
 
+        case encoding_latin1:
+            break;
+
 		default:
 			assert(!"Invalid encoding");
 		}
@@ -4806,7 +4929,9 @@ namespace pugi
 
 		if (!(flags & format_no_declaration) && !has_declaration(*this))
 		{
-			buffered_writer.write(PUGIXML_TEXT("<?xml version=\"1.0\"?>"));
+			buffered_writer.write(PUGIXML_TEXT("<?xml version=\"1.0\""));
+            if (encoding == encoding_latin1) buffered_writer.write(PUGIXML_TEXT(" encoding=\"ISO-8859-1\""));
+			buffered_writer.write('?', '>');
 			if (!(flags & format_raw)) buffered_writer.write('\n');
 		}
 
diff --git a/src/pugixml.hpp b/src/pugixml.hpp
index 1946bfb..d0a8623 100644
--- a/src/pugixml.hpp
+++ b/src/pugixml.hpp
@@ -194,7 +194,8 @@ namespace pugi
 		encoding_utf32_le,  // Little-endian UTF32
 		encoding_utf32_be,  // Big-endian UTF32
 		encoding_utf32,     // UTF32 with native endianness
-		encoding_wchar      // The same encoding wchar_t has (either UTF16 or UTF32)
+		encoding_wchar,     // The same encoding wchar_t has (either UTF16 or UTF32)
+        encoding_latin1
 	};
 
 	// Formatting flags
diff --git a/tests/data/latintest_latin1.xml b/tests/data/latintest_latin1.xml
new file mode 100644
index 0000000..3336f0c
--- /dev/null
+++ b/tests/data/latintest_latin1.xml
@@ -0,0 +1 @@
+<?xml version="1.0" encoding="ISO-8859-1"?><EXAMPLE><!--This is a comment with special chars: <äöü>--><ORDER version="1.0" xml:lang="de"><!--This is another comment with special chars: <äöü>--><HEADER><X_ORDER_ID>0000053535</X_ORDER_ID><CUSTOMER_ID>1010</CUSTOMER_ID><NAME_1>Müller</NAME_1><NAME_2>Jörg</NAME_2></HEADER><ENTRIES><ENTRY><ARTICLE>&lt;Test&gt;</ARTICLE><ENTRY_NO>10</ENTRY_NO></ENTRY><ENTRY><ARTICLE>&lt;Test 2&gt;</ARTICLE><ENTRY_NO>20</ENTRY_NO></ENTRY></ENTRIES><FOOTER><TEXT>This is a text.</TEXT></FOOTER></ORDER></EXAMPLE>
\ No newline at end of file
diff --git a/tests/data/latintest_utf8.xml b/tests/data/latintest_utf8.xml
new file mode 100644
index 0000000..3efad30
--- /dev/null
+++ b/tests/data/latintest_utf8.xml
@@ -0,0 +1 @@
+<?xml version="1.0" encoding="ISO-8859-1"?><EXAMPLE><!--This is a comment with special chars: <äöü>--><ORDER version="1.0" xml:lang="de"><!--This is another comment with special chars: <äöü>--><HEADER><X_ORDER_ID>0000053535</X_ORDER_ID><CUSTOMER_ID>1010</CUSTOMER_ID><NAME_1>Müller</NAME_1><NAME_2>Jörg</NAME_2></HEADER><ENTRIES><ENTRY><ARTICLE>&lt;Test&gt;</ARTICLE><ENTRY_NO>10</ENTRY_NO></ENTRY><ENTRY><ARTICLE>&lt;Test 2&gt;</ARTICLE><ENTRY_NO>20</ENTRY_NO></ENTRY></ENTRIES><FOOTER><TEXT>This is a text.</TEXT></FOOTER></ORDER></EXAMPLE>
\ No newline at end of file
diff --git a/tests/test_document.cpp b/tests/test_document.cpp
index a799fbf..a49efcd 100644
--- a/tests/test_document.cpp
+++ b/tests/test_document.cpp
@@ -307,6 +307,7 @@ TEST_XML(document_save_bom, "<n/>")
 	CHECK(test_save_narrow(doc, flags, encoding_utf16_le, "\xff\xfe<\x00n\x00 \x00/\x00>\x00", 12));
 	CHECK(test_save_narrow(doc, flags, encoding_utf32_be, "\x00\x00\xfe\xff\x00\x00\x00<\x00\x00\x00n\x00\x00\x00 \x00\x00\x00/\x00\x00\x00>", 24));
 	CHECK(test_save_narrow(doc, flags, encoding_utf32_le, "\xff\xfe\x00\x00<\x00\x00\x00n\x00\x00\x00 \x00\x00\x00/\x00\x00\x00>\x00\x00\x00", 24));
+	CHECK(test_save_narrow(doc, flags, encoding_latin1, "<n />", 5));
 
 	// encodings synonyms
 	CHECK(save_narrow(doc, flags, encoding_utf16) == save_narrow(doc, flags, (is_little_endian() ? encoding_utf16_le : encoding_utf16_be)));
@@ -371,6 +372,15 @@ TEST_XML(document_save_declaration_present_last, "<node/>")
 	CHECK(writer.as_string() == STR("<?xml version=\"1.0\"?>\n<node />\n<?xml encoding=\"utf8\"?>\n"));
 }
 
+TEST_XML(document_save_declaration_latin1, "<node/>")
+{
+	xml_writer_string writer;
+
+	doc.save(writer, STR(""), pugi::format_default, encoding_latin1);
+
+	CHECK(writer.as_narrow() == "<?xml version=\"1.0\" encoding=\"ISO-8859-1\"?>\n<node />\n");
+}
+
 struct temp_file
 {
 	char path[512];
@@ -704,18 +714,19 @@ static bool load_file_in_memory(const char* path, char*& data, size_t& size)
 	return true;
 }
 
-TEST(document_contents_preserve)
+struct file_data_t
 {
-	struct file_t
-	{
-		const char* path;
-		xml_encoding encoding;
+    const char* path;
+    xml_encoding encoding;
+
+    char* data;
+    size_t size;
+};
 
-		char* data;
-		size_t size;
-	};
 
-	file_t files[] =
+TEST(document_contents_preserve)
+{
+	file_data_t files[] =
 	{
 		{"tests/data/utftest_utf16_be_clean.xml", encoding_utf16_be, 0, 0},
 		{"tests/data/utftest_utf16_le_clean.xml", encoding_utf16_le, 0, 0},
@@ -751,6 +762,41 @@ TEST(document_contents_preserve)
 	}
 }
 
+TEST(document_contents_preserve_latin1)
+{
+	file_data_t files[] =
+	{
+		{"tests/data/latintest_utf8.xml", encoding_utf8, 0, 0},
+		{"tests/data/latintest_latin1.xml", encoding_latin1, 0, 0}
+	};
+
+	// load files in memory
+	for (unsigned int i = 0; i < sizeof(files) / sizeof(files[0]); ++i)
+	{
+		CHECK(load_file_in_memory(files[i].path, files[i].data, files[i].size));
+	}
+
+	// convert each file to each format and compare bitwise
+	for (unsigned int src = 0; src < sizeof(files) / sizeof(files[0]); ++src)
+	{
+		for (unsigned int dst = 0; dst < sizeof(files) / sizeof(files[0]); ++dst)
+		{
+			// parse into document (preserve comments, declaration and whitespace pcdata)
+			xml_document doc;
+			CHECK(doc.load_buffer(files[src].data, files[src].size, parse_default | parse_ws_pcdata | parse_declaration | parse_comments, files[src].encoding));
+
+			// compare saved document with the original (raw formatting, without extra declaration, write bom if it was in original file)
+			CHECK(test_save_narrow(doc, format_raw | format_no_declaration | format_write_bom, files[dst].encoding, files[dst].data, files[dst].size));
+		}
+	}
+
+	// cleanup
+	for (unsigned int j = 0; j < sizeof(files) / sizeof(files[0]); ++j)
+	{
+		delete[] files[j].data;
+	}
+}
+
 static bool test_parse_fail(const void* buffer, size_t size, xml_encoding encoding = encoding_utf8)
 {
 	// copy buffer to heap (to enable out-of-bounds checks)
@@ -811,7 +857,8 @@ TEST(document_load_buffer_empty)
 		encoding_utf32_le,
 		encoding_utf32_be,
 		encoding_utf32,
-		encoding_wchar
+		encoding_wchar,
+        encoding_latin1
 	};
 
 	char buffer[1];
diff --git a/tests/test_write.cpp b/tests/test_write.cpp
index 094bf59..93f5bd9 100644
--- a/tests/test_write.cpp
+++ b/tests/test_write.cpp
@@ -189,6 +189,8 @@ TEST(write_encodings)
 	{
 		CHECK(v.size() == 10 && v[0] == '<' && v[1] == 0x54 && v[2] == 0xA2 && v[3] == 0x20AC && v[4] == 0xd852 && v[5] == 0xdf62 && v[6] == ' ' && v[7] == '/' && v[8] == '>' && v[9] == '\n');
 	}
+
+    CHECK(test_write_narrow(doc, format_default, encoding_latin1, "<\x54\xA2?? />\n", 9));
 }
 
 #ifdef PUGIXML_WCHAR_MODE