The talk will consist of two parts: Collection and Search. Collection: Projects DML-CZ and EuDML will be described and demoed, and the main lessons from them enumerated. Attention will be given to the process of creation and delivery of new metadata to global DML as EuDML via local repositories like DML-CZ. Search: The recent move to the semantic search and MathML has brought renewed attention to the need of unambiguous canonical math representation in texts. As part of the project of building the European Digital Mathematics Library (http://www.eudml.eu) we have designed and implemented a math search engine, MIaS (http://nlp.fi.muni.cz/projekty/eudml/mias). It currently indexes and searches more than 160,000,000 formulae originally written by authors in TeX in their scientific papers. We will present the system and will discuss the ways towards a global math search engine based on the TeX math notation.