/// <summary> /// UnicodeReplacer API /// </summary> /// public virtual int Replace(Replaceable text, int start, int limit, int[] cursor) { int outLen = 0; // Copy segment with out-of-band data int dest = limit; // If there was no match, that means that a quantifier // matched zero-length. E.g., x (a)* y matched "xy". if (matchStart >= 0) { if (matchStart != matchLimit) { text.Copy(matchStart, matchLimit, dest); outLen = matchLimit - matchStart; } } text.Replace(start, limit, ""); // delete original text return(outLen); }
// = 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); }