Transliteration Table README ---------------------------- 1. Rationale: Identifiers within the database or programming languages only accept latin-1 characters, numbers and '_' character. Application developers can enter captions (titles) to give objects or variables a meaningful name using full unicode set. Transliteration is used to convert unicode captions to identifiers without loosing meaning of the names. More info: http://en.wikipedia.org/wiki/Transliteration http://en.wikipedia.org/wiki/Romanization 2. We use special kind of romanization as we only allow characters described in 1. 3. Implementation: transliteration table, was generated by generate_transliteration_table.sh shell script is used to transliterate any unicode character (having code < 65535) to an identifier, what gives constant time for converting single character. The resulting generated code is kept in transliteration_table.{h|cpp} files, included by identifier.cpp for use in public utility functions. For each item, the table (basically a table of c-strings) contains: - a NULL string it the resulting conversion have to be "_" string; - a c-string of size 1 or more containing a valid transliteration as described in 1; - an empty string "" if the transliteration should return empty string (can be useful e.g. for soft signs in Cyrillic) 4. Fixes: Because iconv/recode tools are not fully implemented in regards to transliteration to latin-1 (e.g. no good support for Greek and Cyrillic/Serbian characters), the transliteration_table.cpp file is patched with transliteration_table.cpp.patch which provides fixes written by hand. If you find invalid or missing transliterations: a) edit transliteration_table.cpp (using UTF-8-compliant text editor!) - if transliteration_table.cpp file does not exist, extract it from transliteration_table.bz2 archive b) run update_transliteration_table_patch.sh shell script, what will update the transliteration_table.cpp.patch file c) send the transliteration_table.cpp.patch file to the Kexi team 5. Credits Jaroslaw Staniek Michael Drueing Chusslove Illich Michal Svec