NOTE: All 0.x releases are kind of "alpha-versions" without expectations of API backward-compatibility.

Table of Contents

mdz_xml Overview
mdz_xml Advantages
Performance Comparison
mdz_xml Usage
Licensing info

mdz_xml Overview

Wiki: [mdz_ansi Wiki]
file: "mdz_xml.h"

Please take a look at "mdz_xml.h" file or mdz_xml Wiki site for detailed functions descriptions.

mdz_xml - is a very lightweight, versatile and speedy C library for parsing XML-documents, building DOM-structure and navigating it. The library is developed by maxdz Software GmbH. Source code of library is highly-portable, conforms to ANSI C 89/90 Standard.

Only shared/dynamically loaded libraries (.so and .dll files with import libraries) are available for evaluation testing purposes. Static libraries are covered by our commercial licenses.

Linux binaries are built against Linux Kernel 2.6.18 - and thus should be compatible with Debian (from ver. 4.0), Ubuntu (from ver. 8.04), Fedora (from ver. 9)

FreeBSD binaries - may be used from FreeBSD ver. 7.0

Win32 binaries are built using Visual Studio Platform Toolset "v90", thus are compatible with Windows versions from Windows 2000.
Win64 binaries are built using Visual Studio Platform Toolset "v100", thus are compatible with Windows versions from Windows XP.

Android x86/armeabi-v7a binaries - may be used from Android API level 16 ("Jelly Bean" ver. 4.1.x)
Android x86_64/arm64-v8a binaries - may be used from Android API level 21 ("Lollipop" ver. 5.0)

macOS binaries - x86_64, from MacOS X v10.6.0

mdz_xml Advantages

1. High portability: the whole code conforms to ANSI C 89/90 Standard.

2. Little dependencies: basically mdz_xml functions are only dependend on standard C-library memory-management/access functions. It means you can use library in your code without any further dependencies except standard platform libraries/APIs.

3. Very fast: the parser is ca. 50% faster and needs ca. 40% less memory than pugiXml (which is considered as one of the fastest parsers overall). And is ca. 20 times faster than of "MSXML" from Microsoft. Please take a look at comparison tables below Performance Comparison.

4. Parsing of very large XML-documents: basically, for 64-bit version the only limitation is your available RAM. Still the parser uses memory very effectively, thus parsing of very large files (we mean 2 GB large and bigger files here) is still possible using mdz_xml with a 50% less RAM-consumption than "pugiXML" or "MSXML" (which are very memory-effective too).

5. Parse error position: you are getting position of invalid (non-parced) character in XML-document if there is parse error.

6. Flexibility: you can parse read-only XML-documents. You can parse writable XML-documents even faster and with less memory-consumption than read-only parse. You can parse files - our file reading is very fast too.

7. Navigation: you can navigate through XML DOM-structure to collect necessary information from elements/children, attributes or texts.

8. Control: you can allow/disallow characters to be parsed in element names, attributes, texts, etc. Parser itself handles instructions, comments, and mixed content

Performance Comparison

Performance comparison tables give an idea about mdz_xml parsing overall performance compared to some well-known C/C++ alternatives.

Following tests are executed:

- parsing of ~800 MB large XML file
http://dumps.wikimedia.your.org/wikidatawiki/20210320/wikidatawiki-20210320-stub-articles6.xml.gz

- parsing of ~1.5 GB large XML file
http://dumps.wikimedia.your.org/wikidatawiki/20210320/wikidatawiki-20210320-stub-meta-current16.xml.gz

- parsing of ~2.0 GB large XML file
http://dumps.wikimedia.your.org/enwiki/20210320/enwiki-20210320-stub-meta-current20.xml.gz

- parsing of ~2.7 GB large XML file
http://dumps.wikimedia.your.org/wikidatawiki/20210320/wikidatawiki-20210320-stub-meta-current18.xml.gz

Parsing has been tested on Intel i5-6600 @ 3.30GHz (4 cores/4 threads), 16 GB RAM, Windows 10 (64-bit) OS.
Build using VC++, toolset v100/64bit.

- mdz_xml (ver. 0.1) is used as shared library (dll)
- Rapid-XML (latest version 1.13), used from sources
- TinyXML-2 (latest version on 23.03.2021), used from sources
(only actual parsing time in Parse() function is measured for TinyXML-2)
- MSXML 6.0 (6.30.18362.418), used via IXMLDOMDocument interface
- pugiXml (latest version on 23.03.2021), used from sources

- times are taken in microseconds using QueryPerformanceCounter() in main execution thread
- memory usage is taken after parsing finished, from Task-Manager/Processes tab for concrete process

Raw parse speed:

Test	mdz_xml_ parse()	mdz_xml_ parseWritable()	Rapid-XML	TinyXML-2	MSXML 6.0	pugiXml
800 MB
speed, μs	965,468	830,137	1,686,864	7,424,695	12,940,236	1,243,289
memory, MB	2028.4	1810.4	4530.5	5477.5	3428.5	3298.5
1.5 GB
speed, μs	1,686,615	1,451,198	2,965,762	12,958,705	22,529,768	2,181,464
memory, MB	3594.8	3214.9	7951.3	9601.6	6095.3	5806.0
2.0 GB
speed, μs	2,463,885	2,091,761	crash	crash	32,466,422	3,428,855
memory, MB	4892.3	4367.6			8279.7	7911.0
2.7 GB
speed, μs	3,148,216	2,682,474	crash	crash	43,521,423	4,124,275
memory, MB	6403.0	5738.1			10809.6	10274.6

"Load file - parse - unload data" test scenario ("warm" cache):

Test speed, μs	mdz_xml_ parseFile() writable	Rapid-XML	TinyXML-2	MSXML 6.0	pugiXml
800 MB	1,373,711	6,586,134	9,278,020	16,457,514	1,899,272
1.5 GB	2,439,384	16,474,472	16,493,514	29,422,628	3,515,159
2.0 GB	3,388,064	crash	crash	43,360,017	5,411,959
2.7 GB	4,406,550	crash	crash	58,679,884	9,551,191

"Load file - parse - unload data" test scenario ("cold" cache):

Test speed, μs	mdz_xml_ parseFile() writable	Rapid-XML	TinyXML-2	MSXML 6.0	pugiXml
1.5 GB	4,924,833	19,006,339	20,971,263	31,736,102	6,729,579

mdz_xml Usage

Test license generation: - in order to get free test-license, please proceed to our Shop page maxdz Shop and register an account. After registration you will be able to obtain free 14-days test-licenses for our products using "Obtain for free" button. Test license data should be used in mdz_xml_init() call for library initialization.

NOTE: All 0.x releases are kind of "beta-versions" and can be used 1) only with test-license (during test period of 14 days, with necessity to re-generate license for the next 14 days test period) and 2) without expectations of interface backward-compatibility.

Several usage-scenarios are possible:

• low-level - raw C interface, using mdz_xml.h header file
• higher-level - using MdzXml C++ "wrapper" around mdz_xml.h functions

Code Example (low-level use)

mdz_xml_init() with license information should be called for library initialization before any subsequent calls:

#include <mdz_xml.h>

int main(int argc, char* argv[])
{
  /* mdz_xml library initialization using license info retrieved after license generation (see "Test license generation" above) */
  
  mdz_bool bRet = mdz_xml_init("<first-name-hash>", "<last-name-hash>", "<email-hash>", "<license-hash>");
  ...
  
  mdz_xml_uninit(); /* call for un-initialization of library */
  
  return 0;
}

After library initialization call mdz_xml_create() for xml-parser creation. There should be also symmetric mdz_xml_destroy() call for every create, otherwise allocated for xml-parser memory remains occupied:

#include <mdz_xml.h>

int main(int argc, char* argv[])
{
  /* initialize mdz_xml library */
  mdz_bool bRet = mdz_xml_init("<first-name-hash>", "<last-name-hash>", "<email-hash>", "<license-hash>");
  
  struct mdz_Xml* pXml = mdz_xml_create(); /* create xml-parser */
  ...
  ...
  /* use pXml */
  ...
  ...
  /* destroy pXml */
  
  mdz_xml_destroy(&pXml); /* after this pXml should be NULL */
  
  mdz_xml_uninit(); /* un-initialize mdz_xml library */

  ...
}

Use mdz_Xml* pointer for parsing text from file:

#include <mdz_xml.h>

int main(int argc, char* argv[])
{
  mdz_bool bRet = mdz_xml_init("<first-name-hash>", "<last-name-hash>", "<email-hash>", "<license-hash>");
  
  struct mdz_Xml* pAnsi = mdz_xml_create();

  /* parse file "test.xml */
  
  bRet = mdz_xml_parseFile(pXml, "test.xml", mdz_true);
 
  ...
  
  mdz_xml_destroy(&pXml);
  
  mdz_xml_uninit();
  ...
}

Licensing info

Use of mdz_xml library is regulated by license agreement in LICENSE.txt

Basically private non-commercial "test" usage is unrestricted. Commercial usage of library (incl. its source code) will be regulated by according license agreement.