|
|
An automatic generation of logging database triggers
Matásek, Luboš ; Matouš, Václav (advisor) ; Mlýnková, Irena (referee)
In the thesis we study automatic generation of logging database triggers. As the result an application called Database Logger for logging triggers management was developed. The application allows DML events logging (insert, update, delete) including column update triggers. Current version supports two database platforms: Oracle and PostgreSQL. The log format preserves the maximum of information value and brings the possibility of easy querying. Users can also create historical tables to produce historical data views. Application supports log and historical table querying and their export into XML.
|
|
|
Effective storage for html documents
Matouš, Václav ; Žemlička, Michal (advisor) ; Lánský, Jan (referee)
In the presented work a prototype of a storage for html documents is designed and implemented. The storing of complete documents is based on both current compression techniques (gzip, bzip2) and new methods. An input document is modi ed by the new methods and compressed by the current techniques. If there are stored many versions of the same document, some of them can be stored in the form of di erential les { in the case that it is more useful. These di erential les can be compressed as well. The compression methods with preprocessed input are a bit more e ffective than the original methods. The diff erence of efficiency of lossy and lossless compression methods is in this case relatively small. The performed tests showed that a combination of mentioned techniques could be a suitable substitution of compact archives for versioned documents. The implementation guarantees quick access to the actual version what is very important as we suppose that such versions will be the most requested ones.
|
|
|
Heuristics for compression of non-well-formed XML
Szabó, Mária ; Lánský, Jan (advisor) ; Matouš, Václav (referee)
XBW [9] is a modular application for lossless text compression, which enables to use several compression algorithms. The best results were reached with the combination of XML parser and Burrows-Wheeler transformation. Thus XBW stands for merged shortcuts XML and BWT. Therefore we try to improve the results in combination with BWT in the thesis. On les with size about 20MB, generated from hundreds of concatenated webpages, we achieve 37 % faster compression time at the cost of 5% worse compression ratio. However, this compression ratio is by 38% better when it comes to confrontation with Rar software. This acceleration was reached by a new type of parser based on dictionaries of tags and elements. Thesis contains also a new, completely rewritten, implementation of original parser, based on the same principle of tag and attribute dictionaries. With this reimplemetation we improved the average compression speed by 4% and average compression ratio by 2%.
|
|
|
Heuristics for compression of non-well-formed XML
Szabó, Mária ; Matouš, Václav (referee) ; Lánský, Jan (advisor)
XBW [9] is a modular application for lossless text compression, which enables to use several compression algorithms. The best results were reached with the combination of XML parser and Burrows-Wheeler transformation. Thus XBW stands for merged shortcuts XML and BWT. Therefore we try to improve the results in combination with BWT in the thesis. On les with size about 20MB, generated from hundreds of concatenated webpages, we achieve 37 % faster compression time at the cost of 5% worse compression ratio. However, this compression ratio is by 38% better when it comes to confrontation with Rar software. This acceleration was reached by a new type of parser based on dictionaries of tags and elements. Thesis contains also a new, completely rewritten, implementation of original parser, based on the same principle of tag and attribute dictionaries. With this reimplemetation we improved the average compression speed by 4% and average compression ratio by 2%.
|
|
|
Effective storage for html documents
Matouš, Václav ; Lánský, Jan (referee) ; Žemlička, Michal (advisor)
In the presented work a prototype of a storage for html documents is designed and implemented. The storing of complete documents is based on both current compression techniques (gzip, bzip2) and new methods. An input document is modi ed by the new methods and compressed by the current techniques. If there are stored many versions of the same document, some of them can be stored in the form of di erential les { in the case that it is more useful. These di erential les can be compressed as well. The compression methods with preprocessed input are a bit more e ffective than the original methods. The diff erence of efficiency of lossy and lossless compression methods is in this case relatively small. The performed tests showed that a combination of mentioned techniques could be a suitable substitution of compact archives for versioned documents. The implementation guarantees quick access to the actual version what is very important as we suppose that such versions will be the most requested ones.
|
|
|
An automatic generation of logging database triggers
Matásek, Luboš ; Matouš, Václav (advisor) ; Mlýnková, Irena (referee)
In the thesis we study automatic generation of logging database triggers. As the result an application called Database Logger for logging triggers management was developed. The application allows DML events logging (insert, update, delete) including column update triggers. Current version supports two database platforms: Oracle and PostgreSQL. The log format preserves the maximum of information value and brings the possibility of easy querying. Users can also create historical tables to produce historical data views. Application supports log and historical table querying and their export into XML.
|
|
|
CTL-to-Java compiler for clover.ETL
Tomčányi, Michal ; Bednárek, David (referee) ; Matouš, Václav (advisor)
CloverETL is a set of data integration tools aimed for rapid development of data transformations. CTL is a proprietary language of CloverETL for de nition of data manipulations. Our goal is to design and implement translator of CTL programs into Java source code which can be compiled and executed as part of CloverETL data transformation. The work discusses key features of CTL (and CloverETL) and proposes necessary changes to translate CTL into Java. Design and implementation of compiler and runtime environment for amended CTL is presented afterwards. The result is a working implementation of compiler for the CTL language allowing future extensions, fully integrated with CloverETL environment.
|
guest ::
