/// <summary>
/// Default implementation of UnicodeMatcher::matches() for Unicode filters.
/// Matches a single 16-bit code unit at offset.
/// </summary>
///
/// @stable ICU 2.0
public virtual int Matches(Replaceable text, int[] offset, int limit,
bool incremental)
{
int c;
if (offset[0] < limit && Contains(c = text.Char32At(offset[0])))
{
offset[0] += IBM.ICU.Text.UTF16.GetCharCount(c);
return(IBM.ICU.Text.UnicodeMatcher_Constants.U_MATCH);
}
if (offset[0] > limit && Contains(c = text.Char32At(offset[0])))
{
// Backup offset by 1, unless the preceding character is a
// surrogate pair -- then backup by 2 (keep offset pointing at
// the lead surrogate).
--offset[0];
if (offset[0] >= 0)
{
offset[0] -= IBM.ICU.Text.UTF16.GetCharCount(text.Char32At(offset[0])) - 1;
}
return(IBM.ICU.Text.UnicodeMatcher_Constants.U_MATCH);
}
if (incremental && offset[0] == limit)
{
return(IBM.ICU.Text.UnicodeMatcher_Constants.U_PARTIAL_MATCH);
}
return(IBM.ICU.Text.UnicodeMatcher_Constants.U_MISMATCH);
}
/// <summary>
/// Returns TRUE if there are any more runs. TRUE is always returned at
/// least once. Upon return, the caller should examine scriptCode, start,
/// and limit.
/// </summary>
///
public bool Next()
{
int ch;
int s;
scriptCode = IBM.ICU.Lang.UScript.INVALID_CODE; // don't know script yet
start = limit;
// Are we done?
if (start == textLimit)
{
return(false);
}
// Move start back to include adjacent COMMON or INHERITED
// characters
while (start > textStart)
{
ch = text.Char32At(start - 1); // look back
s = IBM.ICU.Lang.UScript.GetScript(ch);
if (s == IBM.ICU.Lang.UScript.COMMON || s == IBM.ICU.Lang.UScript.INHERITED)
{
--start;
}
else
{
break;
}
}
// Move limit ahead to include COMMON, INHERITED, and characters
// of the current script.
while (limit < textLimit)
{
ch = text.Char32At(limit); // look ahead
s = IBM.ICU.Lang.UScript.GetScript(ch);
if (s != IBM.ICU.Lang.UScript.COMMON && s != IBM.ICU.Lang.UScript.INHERITED)
{
if (scriptCode == IBM.ICU.Lang.UScript.INVALID_CODE)
{
scriptCode = s;
}
else if (s != scriptCode)
{
break;
}
}
++limit;
}
// Return TRUE even if the entire text is COMMON / INHERITED, in
// which case scriptCode will be UScript.INVALID_CODE.
return(true);
}
/// <summary>
/// Transliterate the given text with the given UTransPosition indices.
/// Return TRUE if the transliteration should continue or FALSE if it should
/// halt (because of a U_PARTIAL_MATCH match). Note that FALSE is only ever
/// returned if isIncremental is TRUE.
/// </summary>
///
/// <param name="text">the text to be transliterated</param>
/// <param name="pos">the position indices, which will be updated</param>
/// <param name="incremental">if TRUE, assume new text may be inserted at index.limit, andreturn FALSE if thre is a partial match.</param>
/// <returns>TRUE unless a U_PARTIAL_MATCH has been obtained, indicating that
/// transliteration should stop until more text arrives.</returns>
public bool Transliterate(Replaceable text, Transliterator.Position pos,
bool incremental)
{
int indexByte = text.Char32At(pos.start) & 0xFF;
for (int i = index[indexByte]; i < index[indexByte + 1]; ++i)
{
int m = rules[i].MatchAndReplace(text, pos, incremental);
switch (m)
{
case IBM.ICU.Text.UnicodeMatcher_Constants.U_MATCH:
if (IBM.ICU.Text.Transliterator.DEBUG)
{
System.Console.Out.WriteLine(((incremental) ? "Rule.i: match "
: "Rule: match ")
+ rules[i].ToRule(true)
+ " => "
+ IBM.ICU.Impl.UtilityExtensions.FormatInput(text, pos));
}
return(true);
case IBM.ICU.Text.UnicodeMatcher_Constants.U_PARTIAL_MATCH:
if (IBM.ICU.Text.Transliterator.DEBUG)
{
System.Console.Out
.WriteLine(((incremental) ? "Rule.i: partial match "
: "Rule: partial match ")
+ rules[i].ToRule(true)
+ " => "
+ IBM.ICU.Impl.UtilityExtensions.FormatInput(text, pos));
}
return(false);
default:
if (IBM.ICU.Text.Transliterator.DEBUG)
{
System.Console.Out.WriteLine("Rule: no match " + rules[i]);
}
break;
}
}
// No match or partial match from any rule
pos.start += IBM.ICU.Text.UTF16.GetCharCount(text.Char32At(pos.start));
if (IBM.ICU.Text.Transliterator.DEBUG)
{
System.Console.Out.WriteLine(((incremental) ? "Rule.i: no match => "
: "Rule: no match => ")
+ IBM.ICU.Impl.UtilityExtensions.FormatInput(text, pos));
}
return(true);
}
/// <summary>
/// Implements <see cref="M:IBM.ICU.Text.Transliterator.HandleTransliterate(IBM.ICU.Text.Replaceable, null, System.Boolean)"/>.
/// </summary>
///
protected internal override void HandleTransliterate(Replaceable text, Transliterator.Position offsets,
bool isIncremental)
{
int start = offsets.start;
int limit = offsets.limit;
if (start >= limit)
{
return;
}
int overallDelta = 0;
// Walk through the string looking for safe characters.
// Whenever you hit one normalize from the start of the last
// safe character up to just before the next safe character
// Also, if you hit the end and we are not in incremental mode,
// do to end.
// TODO: fix for surrogates
// TODO: add QuickCheck, so we rarely convert OK stuff
int lastSafe = start; // go back to start in any event
int cp;
for (int i = start + 1; i < limit; i += IBM.ICU.Text.UTF16.GetCharCount(cp))
{
cp = text.Char32At(i);
if (IBM.ICU.Lang.UCharacter.GetCombiningClass(cp) == 0 &&
!unsafeStart.Contains(cp))
{
int delta = Convert(text, lastSafe, i, null);
i += delta;
limit += delta;
overallDelta += delta;
lastSafe = i;
}
}
if (!isIncremental)
{
int delta_0 = Convert(text, lastSafe, limit, null);
overallDelta += delta_0;
lastSafe = limit + delta_0;
}
else
{
// We are incremental, so accept the last characters IF they turn
// into skippables
int delta_1 = Convert(text, lastSafe, limit, skippable);
if (delta_1 != Int32.MinValue)
{
overallDelta += delta_1;
lastSafe = limit + delta_1;
}
}
offsets.contextLimit += overallDelta;
offsets.limit += overallDelta;
offsets.start = lastSafe;
}
/// <summary>
/// Implements <see cref="M:IBM.ICU.Text.Transliterator.HandleTransliterate(IBM.ICU.Text.Replaceable, null, System.Boolean)"/>.
/// </summary>
///
protected internal override void HandleTransliterate(Replaceable text, Transliterator.Position pos,
bool incremental)
{
int start = pos.start;
int limit = pos.limit;
StringBuilder buf = new StringBuilder(prefix);
int prefixLen = prefix.Length;
bool redoPrefix = false;
while (start < limit)
{
int c = (grokSupplementals) ? (int)(text.Char32At(start)) : (int)(text.CharAt(start));
int charLen = (grokSupplementals) ? IBM.ICU.Text.UTF16.GetCharCount(c) : 1;
if ((c & -65536) != 0 && supplementalHandler != null)
{
buf.Length = 0;
buf.Append(supplementalHandler.prefix);
IBM.ICU.Impl.Utility.AppendNumber(buf, c, supplementalHandler.radix,
supplementalHandler.minDigits);
buf.Append(supplementalHandler.suffix);
redoPrefix = true;
}
else
{
if (redoPrefix)
{
buf.Length = 0;
buf.Append(prefix);
redoPrefix = false;
}
else
{
buf.Length = prefixLen;
}
IBM.ICU.Impl.Utility.AppendNumber(buf, c, radix, minDigits);
buf.Append(suffix);
}
text.Replace(start, start + charLen, buf.ToString());
start += buf.Length;
limit += buf.Length - charLen;
}
pos.contextLimit += limit - pos.limit;
pos.limit = limit;
pos.start = start;
}
/// <summary>
/// Implements <see cref="M:IBM.ICU.Text.Transliterator.HandleTransliterate(IBM.ICU.Text.Replaceable, null, System.Boolean)"/>.
/// </summary>
///
protected internal override void HandleTransliterate(Replaceable text, Transliterator.Position offsets,
bool isIncremental)
{
int cursor = offsets.start;
int limit = offsets.limit;
StringBuilder str = new StringBuilder();
str.Append(OPEN_DELIM);
int len;
String name;
while (cursor < limit)
{
int c = text.Char32At(cursor);
if ((name = IBM.ICU.Lang.UCharacter.GetExtendedName(c)) != null)
{
str.Length = OPEN_DELIM_LEN;
str.Append(name).Append(CLOSE_DELIM);
int clen = IBM.ICU.Text.UTF16.GetCharCount(c);
text.Replace(cursor, cursor + clen, str.ToString());
len = str.Length;
cursor += len; // advance cursor by 1 and adjust for new text
limit += len - clen; // change in length
}
else
{
++cursor;
}
}
offsets.contextLimit += limit - offsets.limit;
offsets.limit = limit;
offsets.start = cursor;
}
/// <summary>
/// Implements <see cref="M:IBM.ICU.Text.Transliterator.HandleTransliterate(IBM.ICU.Text.Replaceable, null, System.Boolean)"/>.
/// </summary>
///
protected internal override void HandleTransliterate(Replaceable text, Transliterator.Position offsets,
bool isIncremental)
{
int maxLen = IBM.ICU.Impl.UCharacterName.GetInstance().GetMaxCharNameLength() + 1; // allow
// for
// temporary
// trailing
// space
StringBuilder name = new StringBuilder(maxLen);
// Get the legal character set
UnicodeSet legal = new UnicodeSet();
IBM.ICU.Impl.UCharacterName.GetInstance().GetCharNameCharacters(legal);
int cursor = offsets.start;
int limit = offsets.limit;
// Modes:
// 0 - looking for open delimiter
// 1 - after open delimiter
int mode = 0;
int openPos = -1; // open delim candidate pos
int c;
while (cursor < limit)
{
c = text.Char32At(cursor);
switch (mode)
{
case 0: // looking for open delimiter
if (c == OPEN_DELIM) // quick check first
{
openPos = cursor;
int i = IBM.ICU.Impl.Utility.ParsePattern(OPEN_PAT, text, cursor, limit);
if (i >= 0 && i < limit)
{
mode = 1;
name.Length = 0;
cursor = i;
continue; // *** reprocess char32At(cursor)
}
}
break;
case 1: // after open delimiter
// Look for legal chars. If \s+ is found, convert it
// to a single space. If closeDelimiter is found, exit
// the loop. If any other character is found, exit the
// loop. If the limit is reached, exit the loop.
// Convert \s+ => SPACE. This assumes there are no
// runs of >1 space characters in names.
if (IBM.ICU.Impl.UCharacterProperty.IsRuleWhiteSpace(c))
{
// Ignore leading whitespace
if (name.Length > 0 &&
name[name.Length - 1] != SPACE)
{
name.Append(SPACE);
// If we are too long then abort. maxLen includes
// temporary trailing space, so use '>'.
if (name.Length > maxLen)
{
mode = 0;
}
}
break;
}
if (c == CLOSE_DELIM)
{
int len = name.Length;
// Delete trailing space, if any
if (len > 0 && name[len - 1] == SPACE)
{
name.Length = --len;
}
c = IBM.ICU.Lang.UCharacter.GetCharFromExtendedName(name.ToString());
if (c != -1)
{
// Lookup succeeded
// assert(UTF16.getCharCount(CLOSE_DELIM) == 1);
cursor++; // advance over CLOSE_DELIM
String str = IBM.ICU.Text.UTF16.ValueOf(c);
text.Replace(openPos, cursor, str);
// Adjust indices for the change in the length of
// the string. Do not assume that str.length() ==
// 1, in case of surrogates.
int delta = cursor - openPos - str.Length;
cursor -= delta;
limit -= delta;
// assert(cursor == openPos + str.length());
}
// If the lookup failed, we leave things as-is and
// still switch to mode 0 and continue.
mode = 0;
openPos = -1; // close off candidate
continue; // *** reprocess char32At(cursor)
}
if (legal.Contains(c))
{
IBM.ICU.Text.UTF16.Append(name, c);
// If we go past the longest possible name then abort.
// maxLen includes temporary trailing space, so use '>='.
if (name.Length >= maxLen)
{
mode = 0;
}
}
// Invalid character
else
{
--cursor; // Backup and reprocess this character
mode = 0;
}
break;
}
cursor += IBM.ICU.Text.UTF16.GetCharCount(c);
}
offsets.contextLimit += limit - offsets.limit;
offsets.limit = limit;
// In incremental mode, only advance the cursor up to the last
// open delimiter candidate.
offsets.start = (isIncremental && openPos >= 0) ? openPos : cursor;
}
// = public static UnicodeReplacer valueOf(String output,
// = int cursorPos,
// = RuleBasedTransliterator.Data data) {
// = if (output.length() == 1) {
// = char c = output.charAt(0);
// = UnicodeReplacer r = data.lookupReplacer(c);
// = if (r != null) {
// = return r;
// = }
// = }
// = return new StringReplacer(output, cursorPos, data);
// = }
/// <summary>
/// UnicodeReplacer API
/// </summary>
///
public virtual int Replace(Replaceable text, int start, int limit, int[] cursor)
{
int outLen;
int newStart = 0;
// NOTE: It should be possible to _always_ run the complex
// processing code; just slower. If not, then there is a bug
// in the complex processing code.
// Simple (no nested replacers) Processing Code :
if (!isComplex)
{
text.Replace(start, limit, output);
outLen = output.Length;
// Setup default cursor position (for cursorPos within output)
newStart = cursorPos;
}
// Complex (nested replacers) Processing Code :
else
{
/*
* When there are segments to be copied, use the Replaceable.copy()
* API in order to retain out-of-band data. Copy everything to the
* end of the string, then copy them back over the key. This
* preserves the integrity of indices into the key and surrounding
* context while generating the output text.
*/
StringBuilder buf = new StringBuilder();
int oOutput; // offset into 'output'
isComplex = false;
// The temporary buffer starts at tempStart, and :
// to destLimit + tempExtra. The start of the buffer has a single
// character from before the key. This provides style
// data when addition characters are filled into the
// temporary buffer. If there is nothing to the left, use
// the non-character U+FFFF, which Replaceable subclasses
// should treat specially as a "no-style character."
// destStart points to the point after the style context
// character, so it is tempStart+1 or tempStart+2.
int tempStart = text.Length(); // start of temp buffer
int destStart = tempStart; // copy new text to here
if (start > 0)
{
int len = IBM.ICU.Text.UTF16.GetCharCount(text.Char32At(start - 1));
text.Copy(start - len, start, tempStart);
destStart += len;
}
else
{
text.Replace(tempStart, tempStart, "\uFFFF");
destStart++;
}
int destLimit = destStart;
int tempExtra = 0; // temp chars after destLimit
for (oOutput = 0; oOutput < output.Length;)
{
if (oOutput == cursorPos)
{
// Record the position of the cursor
newStart = buf.Length + destLimit - destStart; // relative
// to start
// the buf.length() was inserted for bug 5789
// the problem is that if we are accumulating into a buffer
// (when r == null below)
// then the actual length of the text at that point needs to
// add the buf length.
// there was an alternative suggested in #5789, but that
// looks like it won't work
// if we have accumulated some stuff in the dest part AND
// have a non-zero buffer.
}
int c = IBM.ICU.Text.UTF16.CharAt(output, oOutput);
// When we are at the last position copy the right style
// context character into the temporary buffer. We don't
// do this before because it will provide an incorrect
// right context for previous replace() operations.
int nextIndex = oOutput + IBM.ICU.Text.UTF16.GetCharCount(c);
if (nextIndex == output.Length)
{
tempExtra = IBM.ICU.Text.UTF16.GetCharCount(text.Char32At(limit));
text.Copy(limit, limit + tempExtra, destLimit);
}
UnicodeReplacer r = data.LookupReplacer(c);
if (r == null)
{
// Accumulate straight (non-segment) text.
IBM.ICU.Text.UTF16.Append(buf, c);
}
else
{
isComplex = true;
// Insert any accumulated straight text.
if (buf.Length > 0)
{
text.Replace(destLimit, destLimit, buf.ToString());
destLimit += buf.Length;
buf.Length = 0;
}
// Delegate output generation to replacer object
int len_0 = r.Replace(text, destLimit, destLimit, cursor);
destLimit += len_0;
}
oOutput = nextIndex;
}
// Insert any accumulated straight text.
if (buf.Length > 0)
{
text.Replace(destLimit, destLimit, buf.ToString());
destLimit += buf.Length;
}
if (oOutput == cursorPos)
{
// Record the position of the cursor
newStart = destLimit - destStart; // relative to start
}
outLen = destLimit - destStart;
// Copy new text to start, and delete it
text.Copy(destStart, destLimit, start);
text.Replace(tempStart + outLen, destLimit + tempExtra + outLen, "");
// Delete the old text (the key)
text.Replace(start + outLen, limit + outLen, "");
}
if (hasCursor)
{
// Adjust the cursor for positions outside the key. These
// refer to code points rather than code units. If cursorPos
// is within the output string, then use newStart, which has
// already been set above.
if (cursorPos < 0)
{
newStart = start;
int n = cursorPos;
// Outside the output string, cursorPos counts code points
while (n < 0 && newStart > 0)
{
newStart -= IBM.ICU.Text.UTF16.GetCharCount(text.Char32At(newStart - 1));
++n;
}
newStart += n;
}
else if (cursorPos > output.Length)
{
newStart = start + outLen;
int n_1 = cursorPos - output.Length;
// Outside the output string, cursorPos counts code points
while (n_1 > 0 && newStart < text.Length())
{
newStart += IBM.ICU.Text.UTF16.GetCharCount(text.Char32At(newStart));
--n_1;
}
newStart += n_1;
}
else
{
// Cursor is within output string. It has been set up above
// to be relative to start.
newStart += start;
}
cursor[0] = newStart;
}
return(outLen);
}
/// <summary>
/// Implements <see cref="M:IBM.ICU.Text.Transliterator.HandleTransliterate(IBM.ICU.Text.Replaceable, null, System.Boolean)"/>.
/// </summary>
///
protected internal override void HandleTransliterate(Replaceable text, Transliterator.Position pos,
bool isIncremental)
{
int start = pos.start;
int limit = pos.limit;
int i, j, ipat;
loop : {
while (start < limit)
{
// Loop over the forms in spec[]. Exit this loop when we
// match one of the specs. Exit the outer loop if a
// partial match is detected and isIncremental is true.
for (j = 0, ipat = 0; spec[ipat] != END; ++j)
{
// Read the header
int prefixLen = spec[ipat++];
int suffixLen = spec[ipat++];
int radix = spec[ipat++];
int minDigits = spec[ipat++];
int maxDigits = spec[ipat++];
// s is a copy of start that is advanced over the
// characters as we parse them.
int s = start;
bool match = true;
for (i = 0; i < prefixLen; ++i)
{
if (s >= limit)
{
if (i > 0)
{
// We've already matched a character. This is
// a partial match, so we return if in
// incremental mode. In non-incremental mode,
// go to the next spec.
if (isIncremental)
{
goto gotoloop;
}
match = false;
break;
}
}
char c = text.CharAt(s++);
if (c != spec[ipat + i])
{
match = false;
break;
}
}
if (match)
{
int u = 0;
int digitCount = 0;
for (;;)
{
if (s >= limit)
{
// Check for partial match in incremental mode.
if (s > start && isIncremental)
{
goto gotoloop;
}
break;
}
int ch = text.Char32At(s);
int digit = IBM.ICU.Lang.UCharacter.Digit(ch, radix);
if (digit < 0)
{
break;
}
s += IBM.ICU.Text.UTF16.GetCharCount(ch);
u = (u * radix) + digit;
if (++digitCount == maxDigits)
{
break;
}
}
match = (digitCount >= minDigits);
if (match)
{
for (i = 0; i < suffixLen; ++i)
{
if (s >= limit)
{
// Check for partial match in incremental mode.
if (s > start && isIncremental)
{
goto gotoloop;
}
match = false;
break;
}
char c_0 = text.CharAt(s++);
if (c_0 != spec[ipat + prefixLen + i])
{
match = false;
break;
}
}
if (match)
{
// At this point, we have a match
String str = IBM.ICU.Text.UTF16.ValueOf(u);
text.Replace(start, s, str);
limit -= s - start - str.Length;
// The following break statement leaves the
// loop that is traversing the forms in
// spec[]. We then parse the next input
// character.
break;
}
}
}
ipat += prefixLen + suffixLen;
}
if (start < limit)
{
start += IBM.ICU.Text.UTF16.GetCharCount(text.Char32At(start));
}
}
}
gotoloop:
;
pos.contextLimit += limit - pos.limit;
pos.limit = limit;
pos.start = start;
}
static internal int PosAfter(Replaceable str, int pos)
{
return((pos >= 0 && pos < str.Length()) ? pos
+ IBM.ICU.Text.UTF16.GetCharCount(str.Char32At(pos)) : pos + 1);
}
static internal int PosBefore(Replaceable str, int pos)
{
return((pos > 0) ? pos - IBM.ICU.Text.UTF16.GetCharCount(str.Char32At(pos - 1))
: pos - 1);
}
/// <summary>
/// Implements <see cref="M:IBM.ICU.Text.Transliterator.HandleTransliterate(IBM.ICU.Text.Replaceable, null, System.Boolean)"/>.
/// </summary>
///
protected internal override void HandleTransliterate(Replaceable text, Transliterator.Position offsets,
bool isIncremental)
{
// TODO reimplement, see ustrcase.c
// using a real word break iterator
// instead of just looking for a transition between cased and uncased
// characters
// call CaseMapTransliterator::handleTransliterate() for lowercasing?
// (set fMap)
// needs to take isIncremental into account because case mappings are
// context-sensitive
// also detect when lowercasing function did not finish because of
// context
if (offsets.start >= offsets.limit)
{
return;
}
// case type: >0 cased (UCaseProps.LOWER etc.) ==0 uncased <0
// case-ignorable
int type;
// Our mode; we are either converting letter toTitle or
// toLower.
bool doTitle = true;
// Determine if there is a preceding context of cased case-ignorable*,
// in which case we want to start in toLower mode. If the
// prior context is anything else (including empty) then start
// in toTitle mode.
int c, start;
for (start = offsets.start - 1; start >= offsets.contextStart; start -= IBM.ICU.Text.UTF16
.GetCharCount(c))
{
c = text.Char32At(start);
type = csp.GetTypeOrIgnorable(c);
if (type > 0) // cased
{
doTitle = false;
break;
}
else if (type == 0) // uncased but not ignorable
{
break;
}
// else (type<0) case-ignorable: continue
}
// Convert things after a cased character toLower; things
// after a uncased, non-case-ignorable character toTitle. Case-ignorable
// characters are copied directly and do not change the mode.
iter.SetText(text);
iter.SetIndex(offsets.start);
iter.SetLimit(offsets.limit);
iter.SetContextLimits(offsets.contextStart, offsets.contextLimit);
result.Length = 0;
// Walk through original string
// If there is a case change, modify corresponding position in
// replaceable
int delta;
while ((c = iter.NextCaseMapCP()) >= 0)
{
type = csp.GetTypeOrIgnorable(c);
if (type >= 0) // not case-ignorable
{
if (doTitle)
{
c = csp.ToFullTitle(c, iter, result, locale, locCache);
}
else
{
c = csp.ToFullLower(c, iter, result, locale, locCache);
}
doTitle = type == 0; // doTitle=isUncased
if (iter.DidReachLimit() && isIncremental)
{
// the case mapping function tried to look beyond the
// context limit
// wait for more input
offsets.start = iter.GetCaseMapCPStart();
return;
}
/* decode the result */
if (c < 0)
{
/* c mapped to itself, no change */
continue;
}
else if (c <= IBM.ICU.Impl.UCaseProps.MAX_STRING_LENGTH)
{
/* replace by the mapping string */
delta = iter.Replace(result.ToString());
result.Length = 0;
}
else
{
/* replace by single-code point mapping */
delta = iter.Replace(IBM.ICU.Text.UTF16.ValueOf(c));
}
if (delta != 0)
{
offsets.limit += delta;
offsets.contextLimit += delta;
}
}
}
offsets.start = offsets.limit;
}
