diff options
Diffstat (limited to 'offapi')
| -rw-r--r-- | offapi/com/sun/star/i18n/TransliterationModules.idl | 8 | ||||
| -rw-r--r-- | offapi/com/sun/star/util/XTextSearch.idl | 150 |
2 files changed, 108 insertions, 50 deletions
diff --git a/offapi/com/sun/star/i18n/TransliterationModules.idl b/offapi/com/sun/star/i18n/TransliterationModules.idl index 1ac752d52fd8..69618dee0c9f 100644 --- a/offapi/com/sun/star/i18n/TransliterationModules.idl +++ b/offapi/com/sun/star/i18n/TransliterationModules.idl @@ -33,8 +33,12 @@ module com { module sun { module star { module i18n { //============================================================================= -/** Old transliteration module enumeration to use with - <member>XTransliteration::loadModule()</member> +/** Old transliteration module enumeration. + + <p> Use with <member>XTransliteration::loadModule()</member> and + <member scope="com::sun::star::util">SearchOptions::transliterateFlags</member> </p> + + <p> Note that values >=0x100 are logically or'ed with other values! </p> */ published enum TransliterationModules diff --git a/offapi/com/sun/star/util/XTextSearch.idl b/offapi/com/sun/star/util/XTextSearch.idl index cc9e45058c47..3b0170c2ec26 100644 --- a/offapi/com/sun/star/util/XTextSearch.idl +++ b/offapi/com/sun/star/util/XTextSearch.idl @@ -40,97 +40,151 @@ module com { module sun { module star { module util { //============================================================================= -published enum SearchAlgorithms { - ABSOLUTE, // "normal" a kind of Boyer-Moore - REGEXP, // regular expression - APPROXIMATE // Leveinstein distance +published enum SearchAlgorithms +{ + /// Literal + ABSOLUTE, // implemented as a kind of Boyer-Moore + /// Regular expression + REGEXP, + /// Weighted Levenshtein Distance + APPROXIMATE }; +/// Flags for search methods published constants SearchFlags { - //Flag for all search methods - /** - @deprecated : the constant ALL_IGNORE_CASE is never supported - - it must use the transliteration flags of - the SearchOptions. - <p>@see TransliterationModulesNew + @deprecated The constant ALL_IGNORE_CASE is never supported - use + <const scope="com::sun::star::i18n">TransliterationModules::IGNORE_CASE</const> + with + <member>SearchOptions::transliterateFlags</member> + instead. + + @see <type scope="com::sun::star::i18n">TransliterationModules</type> */ const long ALL_IGNORE_CASE = 0x00000001; - /// Flag for normal (Boyer-Moore) search + /** Flag for normal (Boyer-Moore) search / Search for word only. */ const long NORM_WORD_ONLY = 0x00000010; - /// Flag for "regular expression" search / interpret as extended regular expression + /** Flag for "regular expression" search / Interpret as extended + regular expression. + + @deprecated The flag is currently not supported by OOo. + */ const long REG_EXTENDED = 0x00000100; - /** Flag for "regular expression" search / No replace, i.e., avoid sub regular - expresions, return true/false to match -<!-- JRH: Check this for sense of the expression. --> + /** Flag for "regular expression" search / No register information + or backreferences, i.e., avoid sub expressions. Return only + true/false if matched or not. + + @deprecated The flag is currently not supported by OOo. + */ + const long REG_NOSUB = 0x00000200; + + /** Flag for "regular expression" search / Special new line + treatment. + + @deprecated The flag is currently not supported by OOo. + + <p> A NEWLINE character in string will not be matched by a + period outside bracket expression or by any form of a non + matching list. </p> + + <p> A circumflex (^) in pattern when used to specify expression + anchoring will match the zero length string immediately after a + newline in string, regardless of the setting of + REG_NOT_BEGINOFLINE. </p> + + <p> A dollar-sign ($) in pattern when used to specify expression + anchoring, will match zero-length string immediately before a + new line in string, regardless of the setting of + REG_NOT_ENDOFLINE. </p> + */ + const long REG_NEWLINE = 0x00000400; + + /** The first character in the string is not the beginning of the + line therefore ^ will not match with first character of the + string. */ - const long REG_NOSUB = 0x00000200; - - /// Flag for "regular expression" search / Special new line treatment - const long REG_NEWLINE = 0x00000400; - - /** A NEWLINE character in string will not be matched by a period outside bracket - expression or by any form of a non matching list. - A circumflex (^) in pattern when used to specify expression anchoring -<!-- JRH: anhoring to anchoring. --> - will match the zero length string immediately after a newline in string, - regardless of the setting of REG_NOTBOL - A dollar-sign ($) in pattern when used to specify expression anchoring, - will match zero-length string immediately before a new line in string, - regardless of the setting of REG_NOTEOL -*/ const long REG_NOT_BEGINOFLINE = 0x00000800; - /** The first character in the string is not the beginning of the line therefore ^ will not - match with first character of the string + /** The last character in the string is not the end of the line + therefore $ will not match with last character of the string. */ const long REG_NOT_ENDOFLINE = 0x00001000; - - /// Flags for "Weight Levenshtein-Distance" search + /** Flag for "Weighted Levenshtein Distance" search / Relaxed + checking of limit, split weigh pools. + + <p> If not specified (<b>strict</b>), the search is sucessful if + the WLD is within a calculated limit where each insertion, + deletion and replacement adds a weight to a common pool of + weights. This is the mathematically correct WLD. </p> + + <p> From a user's point of view the strict WLD is an + exclusive-OR of the arguments given, for example if allowed + insertions=2 and allowed replacements=2, the search fails if 2 + characters had been inserted and an additional operation would + be needed to match. Depending on the weights it may also fail if + 1 character was inserted and 1 character replaced and an + additional operation would be needed to match. The strict + algorithm may match less than expected from a first glance of + the specified arguments, but does not return false positives. </p> + + <p> If specified (<b>relaxed</b>), the search is also successful + if the combined pool for insertions and deletions is below a + doubled calculated limit and replacements are treated + differently. Additionally, swapped characters are counted as one + replacement. </p> + + <p> From a user's point of view the relaxed WLD is an + inclusive-OR of the arguments given, for example if allowed + insertions=2 and allowed replacements=2, the search succeeds if + 2 characters had been inserted and an additional replacement is + needed to match. The relaxed algorithm may return false + positives, but meets user expectation better. </p> + */ const long LEV_RELAXED = 0x00010000; }; published struct SearchOptions { //------------------------------------------------------------------------- - /// search type, can be: ABSOLUTE, REGEXP, APPROXIMATE + /** search type */ SearchAlgorithms algorithmType; /** some flags - can be mixed - @see SearchFlags + @see <type>SearchFlags</type> */ long searchFlag; - /// the search text + /** The text or pattern to be searched. */ string searchString; - /** the replace text - (is for optional replacing - SearchOption is only the data container for it)*/ + /** The replacement text + (is for optional replacing - SearchOption is only the data container for it) */ string replaceString; - /// this is the language for case insensitive search + /** The locale for case insensitive search. */ ::com::sun::star::lang::Locale Locale; - /** this many characters can be different between the found and search word - in a "Weight Levenshtein-Distance"*/ + /** This many characters can be different (as a replacement) between + the found word and the search pattern in a "Weighted Levenshtein + Distance" search. */ long changedChars; - /** this many characters can be missed in the found word - in a "Weight Levenshtein-Distance"*/ + /** This many characters can be missing in the found word in a + "Weighted Levenshtein Distance" search. */ long deletedChars; - /// this many characters can be additional in the found word in a "Weight Levenshtein-Distance" + /** This many characters can be additional in the found word in a + "Weighted Levenshtein Distance" search. */ long insertedChars; - /** asian flags for the transliteration. Same meaning as the enum of TransliteratorModule - - @see com/sun/star/i18n/XTransliteration.idl + /** Flags for the transliteration. Same meaning as the enum of + <type scope="com::sun::star::i18n">TransliterationModules</type> */ long transliterateFlags; }; |