clean up xml parsing, fix numbers xlsx decryption, unskip some tests

source/detail/constants.cpp

@@ -24,6 +24,7 @@
 
 #include <detail/constants.hpp>
 #include <xlnt/xlnt_config.hpp>
+#include <xlnt/utils/exceptions.hpp>
 
 namespace xlnt {
 
@@ -69,7 +70,7 @@ const std::unordered_map<std::string, std::string> &constants::get_namespaces()
     static const std::unordered_map<std::string, std::string> *namespaces =
         new std::unordered_map<std::string, std::string>
         {
-            { "worksheet", "http://schemas.openxmlformats.org/spreadsheetml/2006/main" },
+            { "spreadsheetml", "http://schemas.openxmlformats.org/spreadsheetml/2006/main" },
             { "content-types", "http://schemas.openxmlformats.org/package/2006/content-types" },
             { "relationships", "http://schemas.openxmlformats.org/package/2006/relationships" },
             { "drawingml", "http://schemas.openxmlformats.org/drawingml/2006/main" },
@@ -102,7 +103,14 @@ const std::unordered_map<std::string, std::string> &constants::get_namespaces()
 
 const std::string &constants::get_namespace(const std::string &id)
 {
-    return get_namespaces().find(id)->second;
+    auto match = get_namespaces().find(id);
+
+    if (match == get_namespaces().end())
+    {
+        throw xlnt::exception("bad namespace");
+    }
+
+    return match->second;
 }
 
 }

source/detail/xlsx_consumer.hpp

@@ -35,6 +35,7 @@
 
 namespace xlnt {
 
+class formatted_text;
 class path;
 class relationship;
 class workbook;
@@ -220,13 +221,31 @@ private:
 	/// </summary>
 	void read_unknown_relationships();
     
+    std::vector<relationship> read_relationships(const path &part);
+    
     std::string read_text();
     
-    void read_block(const std::unordered_map<xml::qname, std::function<void(xlsx_consumer &)>> &handlers);
+    formatted_text read_formatted_text(const std::string &xmlns);
     
     void read_part(const std::vector<relationship> &rel_chain);
 
-    void skip_attribute(const std::string &name);
+    void skip_attributes();
+
+    void skip_attributes(const std::vector<xml::qname> &names);
+
+    void skip_attributes(const std::vector<std::string> &names);
+
+    void skip_remaining_content(const xml::qname &name);
+
+    xml::qname expect_start_element(xml::content content);
+
+    void expect_start_element(const xml::qname &name, xml::content content);
+    
+    void expect_end_element(const xml::qname &name);
+    
+    bool in_element(const xml::qname &name);
+
+    std::vector<std::string> read_namespaces();
 
 	/// <summary>
 	/// The ZIP file containing the files that make up the OOXML package.
@@ -255,6 +274,8 @@ private:
 	/// to access the object.
 	/// </summary>
 	xml::parser *parser_;
+    
+    std::vector<xml::qname> stack_;
 };
 
 } // namespace detail

source/detail/xlsx_crypto.cpp

@@ -586,6 +586,7 @@ struct crypto_helper
             = { {0xd7, 0xaa, 0x0f, 0x6d, 0x30, 0x61, 0x34, 0x4e} };
         auto expected_verifier = calculate_block(h_n, verifier_block_key,
             result.key_encryptor.verifier_hash_value);
+        expected_verifier.resize(calculated_verifier.size());
 
         if (calculated_verifier.size() != expected_verifier.size()
             || std::mismatch(calculated_verifier.begin(), calculated_verifier.end(),

source/detail/xlsx_producer.cpp

@@ -352,7 +352,7 @@ void xlsx_producer::write_workbook(const relationship &rel)
     static const auto &xmlns_mc = constants::get_namespace("mc");
     static const auto &xmlns_mx = constants::get_namespace("mx");
     static const auto &xmlns_r = constants::get_namespace("r");
-    static const auto &xmlns_s = constants::get_namespace("worksheet");
+    static const auto &xmlns_s = constants::get_namespace("spreadsheetml");
     static const auto &xmlns_x15 = constants::get_namespace("x15");
     static const auto &xmlns_x15ac = constants::get_namespace("x15ac");
 
@@ -671,7 +671,7 @@ void xlsx_producer::write_pivot_table(const relationship &/*rel*/)
 
 void xlsx_producer::write_shared_string_table(const relationship &/*rel*/)
 {
-    static const auto &xmlns = constants::get_namespace("worksheet");
+    static const auto &xmlns = constants::get_namespace("spreadsheetml");
 
 	serializer().start_element(xmlns, "sst");
     serializer().namespace_decl(xmlns, "");
@@ -796,7 +796,7 @@ void xlsx_producer::write_shared_workbook_user_data(const relationship &/*rel*/)
 
 void xlsx_producer::write_styles(const relationship &/*rel*/)
 {
-    static const auto &xmlns = constants::get_namespace("worksheet");
+    static const auto &xmlns = constants::get_namespace("spreadsheetml");
     static const auto &xmlns_mc = constants::get_namespace("mc");
     static const auto &xmlns_x14 = constants::get_namespace("x14");
     static const auto &xmlns_x14ac = constants::get_namespace("x14ac");
@@ -1830,7 +1830,7 @@ void xlsx_producer::write_volatile_dependencies(const relationship &/*rel*/)
 
 void xlsx_producer::write_worksheet(const relationship &rel)
 {
-	static const auto &xmlns = constants::get_namespace("worksheet");
+	static const auto &xmlns = constants::get_namespace("spreadsheetml");
 	static const auto &xmlns_r = constants::get_namespace("r");
 	static const auto &xmlns_mc = constants::get_namespace("mc");
 	static const auto &xmlns_x14ac = constants::get_namespace("x14ac");
@@ -2400,7 +2400,7 @@ void xlsx_producer::write_worksheet(const relationship &rel)
 void xlsx_producer::write_comments(const relationship &/*rel*/, worksheet ws, 
 	const std::vector<cell_reference> &cells)
 {
-	static const auto &xmlns = constants::get_namespace("worksheet");
+	static const auto &xmlns = constants::get_namespace("spreadsheetml");
 
 	serializer().start_element(xmlns, "comments");
 	serializer().namespace_decl(xmlns, "");

source/workbook/tests/test_consume_xlsx.hpp

@@ -7,6 +7,8 @@
 #include <helpers/path_helper.hpp>
 #include <xlnt/workbook/workbook.hpp>
 
+// Cryptographic key generation can take a few seconds, particularly in unoptomized builds.
+// Set this to false to skip those tests that use cryptography.
 #define TEST_CRYPTO true
 
 #ifndef TEST_CRYPTO
@@ -42,7 +44,6 @@ public:
 
     void test_decrypt_numbers()
     {
-        TS_SKIP("");
         xlnt::workbook wb;
 #if TEST_CRYPTO
         wb.load(path_helper::get_data_directory("17_encrypted_numbers.xlsx"), "secret");

source/workbook/tests/test_produce_xlsx.hpp

@@ -43,7 +43,6 @@ public:
 
 	void test_produce_default_libre_office()
 	{
-        TS_SKIP("");
 		xlnt::workbook wb = xlnt::workbook::empty_libre_office();
 		TS_ASSERT(workbook_matches_file(wb, path_helper::get_data_directory("10_default-libre-office.xlsx")));
 	}
@@ -185,6 +184,6 @@ public:
 
 		wb.save("debug.xlsx");
 
-		//TS_ASSERT(workbook_matches_file(wb, xlnt::path("data/18_basic_comments.xlsx")));
+		TS_ASSERT(workbook_matches_file(wb, xlnt::path("data/18_basic_comments.xlsx")));
 	}
 };

source/workbook/tests/test_round_trip.hpp

@@ -70,14 +70,13 @@ public:
 
 	void test_round_trip_empty_libre_office_wrw()
 	{
-		TS_SKIP("");
 		xlnt::workbook wb = xlnt::workbook::empty_libre_office();
 		TS_ASSERT(round_trip_matches_wrw(wb));
 	}
 
-	void test_round_trip_empty_pages_wrw()
+	void test_round_trip_empty_numbers_wrw()
 	{
-		TS_SKIP("");
+        TS_SKIP("");
 		xlnt::workbook wb = xlnt::workbook::empty_numbers();
 		TS_ASSERT(round_trip_matches_wrw(wb));
 	}
@@ -96,11 +95,17 @@ public:
 
 	void test_round_trip_empty_libre_rw()
 	{
-		TS_SKIP("");
 		auto path = path_helper::get_data_directory("10_default-libre-office.xlsx");
 		TS_ASSERT(round_trip_matches_rw(path));
 	}
 
+	void test_round_trip_empty_numbers_rw()
+	{
+		TS_SKIP("");
+		auto path = path_helper::get_data_directory("10_default-numbers.xlsx");
+		TS_ASSERT(round_trip_matches_rw(path));
+	}
+
 	void test_round_trip_all_styles_rw()
 	{
 		auto path = path_helper::get_data_directory("13_all_styles.xlsx");
@@ -109,6 +114,6 @@ public:
         
         std::vector<std::uint8_t> buffer;
         original_workbook.save(buffer);
-		//TS_ASSERT(round_trip_matches_rw(path));
+		TS_ASSERT(round_trip_matches_rw(path));
 	}
 };

source/workbook/workbook.cpp

@@ -483,15 +483,16 @@ void workbook::register_comments_in_manifest(worksheet ws)
 	{
 		std::size_t file_number = 1;
 		path filename("vmlDrawing1.vml");
+        bool filename_exists = true;
 
-		while (true)
+		while (filename_exists)
 		{
-			bool filename_exists = false;
+            filename_exists = false;
 
 			for (auto current_ws_rel : get_manifest().get_relationships(wb_rel.get_target().get_path(), xlnt::relationship_type::worksheet))
 			{
 				path current_ws_path(current_ws_rel.get_source().get_path().parent().append(current_ws_rel.get_target().get_path()));
-				if (!get_manifest().has_relationship(current_ws_path, xlnt::relationship_type::vml_drawing)) break;
+				if (!get_manifest().has_relationship(current_ws_path, xlnt::relationship_type::vml_drawing)) continue;
 
 				for (auto current_ws_child_rel : get_manifest().get_relationships(current_ws_path, xlnt::relationship_type::vml_drawing))
 				{
@@ -501,17 +502,13 @@ void workbook::register_comments_in_manifest(worksheet ws)
 						break;
 					}
 				}
-
-				if (filename_exists)
-				{
-					break;
-				}
 			}
 
-			if (!filename_exists) break;
+            if (filename_exists)
-
+            {
-			file_number++;
+                file_number++;
-			filename = path("vmlDrawing" + std::to_string(file_number) + ".vml");
+                filename = path("vmlDrawing" + std::to_string(file_number) + ".vml");
+            }
 		}
 
 		get_manifest().register_default_type("vml",

tests/helpers/xml_helper.hpp

@@ -199,7 +199,8 @@ public:
         auto is_xml = (content_type.substr(0, 12) == "application/"
             && content_type.substr(content_type.size() - 4) == "+xml")
             || content_type == "application/xml"
-            || content_type == "[Content_Types].xml";
+            || content_type == "[Content_Types].xml"
+            || content_type == "application/vnd.openxmlformats-officedocument.vmlDrawing";
         
         if (is_xml)
         {

third-party/botan

@@ -1 +1 @@
-Subproject commit 523b2a4ca48fa5cf04ea371aabe7167ce2e5cd13
+Subproject commit 923a95d546df5b6d31f39b0af900d0361fb2e6a6