michael@0: /*
michael@0: **********************************************************************
michael@0: *   Copyright (c) 2001-2011, International Business Machines
michael@0: *   Corporation and others.  All Rights Reserved.
michael@0: **********************************************************************
michael@0: *   Date        Name        Description
michael@0: *   11/19/2001  aliu        Creation.
michael@0: **********************************************************************
michael@0: */
michael@0: 
michael@0: #include "unicode/unimatch.h"
michael@0: #include "unicode/utf16.h"
michael@0: #include "patternprops.h"
michael@0: #include "util.h"
michael@0: 
michael@0: // Define UChar constants using hex for EBCDIC compatibility
michael@0: 
michael@0: static const UChar BACKSLASH  = 0x005C; /*\*/
michael@0: static const UChar UPPER_U    = 0x0055; /*U*/
michael@0: static const UChar LOWER_U    = 0x0075; /*u*/
michael@0: static const UChar APOSTROPHE = 0x0027; // '\''
michael@0: static const UChar SPACE      = 0x0020; // ' '
michael@0: 
michael@0: // "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ"
michael@0: static const UChar DIGITS[] = {
michael@0:     48,49,50,51,52,53,54,55,56,57,
michael@0:     65,66,67,68,69,70,71,72,73,74,
michael@0:     75,76,77,78,79,80,81,82,83,84,
michael@0:     85,86,87,88,89,90
michael@0: };
michael@0: 
michael@0: U_NAMESPACE_BEGIN
michael@0: 
michael@0: UnicodeString& ICU_Utility::appendNumber(UnicodeString& result, int32_t n,
michael@0:                                      int32_t radix, int32_t minDigits) {
michael@0:     if (radix < 2 || radix > 36) {
michael@0:         // Bogus radix
michael@0:         return result.append((UChar)63/*?*/);
michael@0:     }
michael@0:     // Handle negatives
michael@0:     if (n < 0) {
michael@0:         n = -n;
michael@0:         result.append((UChar)45/*-*/);
michael@0:     }
michael@0:     // First determine the number of digits
michael@0:     int32_t nn = n;
michael@0:     int32_t r = 1;
michael@0:     while (nn >= radix) {
michael@0:         nn /= radix;
michael@0:         r *= radix;
michael@0:         --minDigits;
michael@0:     }
michael@0:     // Now generate the digits
michael@0:     while (--minDigits > 0) {
michael@0:         result.append(DIGITS[0]);
michael@0:     }
michael@0:     while (r > 0) {
michael@0:         int32_t digit = n / r;
michael@0:         result.append(DIGITS[digit]);
michael@0:         n -= digit * r;
michael@0:         r /= radix;
michael@0:     }
michael@0:     return result;
michael@0: }
michael@0: 
michael@0: /**
michael@0:  * Return true if the character is NOT printable ASCII.
michael@0:  */
michael@0: UBool ICU_Utility::isUnprintable(UChar32 c) {
michael@0:     return !(c >= 0x20 && c <= 0x7E);
michael@0: }
michael@0: 
michael@0: /**
michael@0:  * Escape unprintable characters using \uxxxx notation for U+0000 to
michael@0:  * U+FFFF and \Uxxxxxxxx for U+10000 and above.  If the character is
michael@0:  * printable ASCII, then do nothing and return FALSE.  Otherwise,
michael@0:  * append the escaped notation and return TRUE.
michael@0:  */
michael@0: UBool ICU_Utility::escapeUnprintable(UnicodeString& result, UChar32 c) {
michael@0:     if (isUnprintable(c)) {
michael@0:         result.append(BACKSLASH);
michael@0:         if (c & ~0xFFFF) {
michael@0:             result.append(UPPER_U);
michael@0:             result.append(DIGITS[0xF&(c>>28)]);
michael@0:             result.append(DIGITS[0xF&(c>>24)]);
michael@0:             result.append(DIGITS[0xF&(c>>20)]);
michael@0:             result.append(DIGITS[0xF&(c>>16)]);
michael@0:         } else {
michael@0:             result.append(LOWER_U);
michael@0:         }
michael@0:         result.append(DIGITS[0xF&(c>>12)]);
michael@0:         result.append(DIGITS[0xF&(c>>8)]);
michael@0:         result.append(DIGITS[0xF&(c>>4)]);
michael@0:         result.append(DIGITS[0xF&c]);
michael@0:         return TRUE;
michael@0:     }
michael@0:     return FALSE;
michael@0: }
michael@0: 
michael@0: /**
michael@0:  * Returns the index of a character, ignoring quoted text.
michael@0:  * For example, in the string "abc'hide'h", the 'h' in "hide" will not be
michael@0:  * found by a search for 'h'.
michael@0:  */
michael@0: // FOR FUTURE USE.  DISABLE FOR NOW for coverage reasons.
michael@0: /*
michael@0: int32_t ICU_Utility::quotedIndexOf(const UnicodeString& text,
michael@0:                                int32_t start, int32_t limit,
michael@0:                                UChar charToFind) {
michael@0:     for (int32_t i=start; i<limit; ++i) {
michael@0:         UChar c = text.charAt(i);
michael@0:         if (c == BACKSLASH) {
michael@0:             ++i;
michael@0:         } else if (c == APOSTROPHE) {
michael@0:             while (++i < limit
michael@0:                    && text.charAt(i) != APOSTROPHE) {}
michael@0:         } else if (c == charToFind) {
michael@0:             return i;
michael@0:         }
michael@0:     }
michael@0:     return -1;
michael@0: }
michael@0: */
michael@0: 
michael@0: /**
michael@0:  * Skip over a sequence of zero or more white space characters at pos.
michael@0:  * @param advance if true, advance pos to the first non-white-space
michael@0:  * character at or after pos, or str.length(), if there is none.
michael@0:  * Otherwise leave pos unchanged.
michael@0:  * @return the index of the first non-white-space character at or
michael@0:  * after pos, or str.length(), if there is none.
michael@0:  */
michael@0: int32_t ICU_Utility::skipWhitespace(const UnicodeString& str, int32_t& pos,
michael@0:                                     UBool advance) {
michael@0:     int32_t p = pos;
michael@0:     const UChar* s = str.getBuffer();
michael@0:     p = (int32_t)(PatternProps::skipWhiteSpace(s + p, str.length() - p) - s);
michael@0:     if (advance) {
michael@0:         pos = p;
michael@0:     }
michael@0:     return p;
michael@0: }
michael@0: 
michael@0: /**
michael@0:  * Skip over Pattern_White_Space in a Replaceable.
michael@0:  * Skipping may be done in the forward or
michael@0:  * reverse direction.  In either case, the leftmost index will be
michael@0:  * inclusive, and the rightmost index will be exclusive.  That is,
michael@0:  * given a range defined as [start, limit), the call
michael@0:  * skipWhitespace(text, start, limit) will advance start past leading
michael@0:  * whitespace, whereas the call skipWhitespace(text, limit, start),
michael@0:  * will back up limit past trailing whitespace.
michael@0:  * @param text the text to be analyzed
michael@0:  * @param pos either the start or limit of a range of 'text', to skip
michael@0:  * leading or trailing whitespace, respectively
michael@0:  * @param stop either the limit or start of a range of 'text', to skip
michael@0:  * leading or trailing whitespace, respectively
michael@0:  * @return the new start or limit, depending on what was passed in to
michael@0:  * 'pos'
michael@0:  */
michael@0: //?FOR FUTURE USE.  DISABLE FOR NOW for coverage reasons.
michael@0: //?int32_t ICU_Utility::skipWhitespace(const Replaceable& text,
michael@0: //?                                    int32_t pos, int32_t stop) {
michael@0: //?    UChar32 c;
michael@0: //?    UBool isForward = (stop >= pos);
michael@0: //?
michael@0: //?    if (!isForward) {
michael@0: //?        --pos; // pos is a limit, so back up by one
michael@0: //?    }
michael@0: //?    
michael@0: //?    while (pos != stop &&
michael@0: //?           PatternProps::isWhiteSpace(c = text.char32At(pos))) {
michael@0: //?        if (isForward) {
michael@0: //?            pos += U16_LENGTH(c);
michael@0: //?        } else {
michael@0: //?            pos -= U16_LENGTH(c);
michael@0: //?        }
michael@0: //?    }
michael@0: //?
michael@0: //?    if (!isForward) {
michael@0: //?        ++pos; // make pos back into a limit
michael@0: //?    }
michael@0: //?
michael@0: //?    return pos;
michael@0: //?}
michael@0: 
michael@0: /**
michael@0:  * Parse a single non-whitespace character 'ch', optionally
michael@0:  * preceded by whitespace.
michael@0:  * @param id the string to be parsed
michael@0:  * @param pos INPUT-OUTPUT parameter.  On input, pos[0] is the
michael@0:  * offset of the first character to be parsed.  On output, pos[0]
michael@0:  * is the index after the last parsed character.  If the parse
michael@0:  * fails, pos[0] will be unchanged.
michael@0:  * @param ch the non-whitespace character to be parsed.
michael@0:  * @return true if 'ch' is seen preceded by zero or more
michael@0:  * whitespace characters.
michael@0:  */
michael@0: UBool ICU_Utility::parseChar(const UnicodeString& id, int32_t& pos, UChar ch) {
michael@0:     int32_t start = pos;
michael@0:     skipWhitespace(id, pos, TRUE);
michael@0:     if (pos == id.length() ||
michael@0:         id.charAt(pos) != ch) {
michael@0:         pos = start;
michael@0:         return FALSE;
michael@0:     }
michael@0:     ++pos;
michael@0:     return TRUE;
michael@0: }
michael@0: 
michael@0: /**
michael@0:  * Parse a pattern string within the given Replaceable and a parsing
michael@0:  * pattern.  Characters are matched literally and case-sensitively
michael@0:  * except for the following special characters:
michael@0:  *
michael@0:  * ~  zero or more Pattern_White_Space chars
michael@0:  *
michael@0:  * If end of pattern is reached with all matches along the way,
michael@0:  * pos is advanced to the first unparsed index and returned.
michael@0:  * Otherwise -1 is returned.
michael@0:  * @param pat pattern that controls parsing
michael@0:  * @param text text to be parsed, starting at index
michael@0:  * @param index offset to first character to parse
michael@0:  * @param limit offset after last character to parse
michael@0:  * @return index after last parsed character, or -1 on parse failure.
michael@0:  */
michael@0: int32_t ICU_Utility::parsePattern(const UnicodeString& pat,
michael@0:                                   const Replaceable& text,
michael@0:                                   int32_t index,
michael@0:                                   int32_t limit) {
michael@0:     int32_t ipat = 0;
michael@0: 
michael@0:     // empty pattern matches immediately
michael@0:     if (ipat == pat.length()) {
michael@0:         return index;
michael@0:     }
michael@0: 
michael@0:     UChar32 cpat = pat.char32At(ipat);
michael@0: 
michael@0:     while (index < limit) {
michael@0:         UChar32 c = text.char32At(index);
michael@0: 
michael@0:         // parse \s*
michael@0:         if (cpat == 126 /*~*/) {
michael@0:             if (PatternProps::isWhiteSpace(c)) {
michael@0:                 index += U16_LENGTH(c);
michael@0:                 continue;
michael@0:             } else {
michael@0:                 if (++ipat == pat.length()) {
michael@0:                     return index; // success; c unparsed
michael@0:                 }
michael@0:                 // fall thru; process c again with next cpat
michael@0:             }
michael@0:         }
michael@0: 
michael@0:         // parse literal
michael@0:         else if (c == cpat) {
michael@0:             index += U16_LENGTH(c);
michael@0:             ipat += U16_LENGTH(cpat);
michael@0:             if (ipat == pat.length()) {
michael@0:                 return index; // success; c parsed
michael@0:             }
michael@0:             // fall thru; get next cpat
michael@0:         }
michael@0: 
michael@0:         // match failure of literal
michael@0:         else {
michael@0:             return -1;
michael@0:         }
michael@0: 
michael@0:         cpat = pat.char32At(ipat);
michael@0:     }
michael@0: 
michael@0:     return -1; // text ended before end of pat
michael@0: }
michael@0: 
michael@0: /**
michael@0:  * Append a character to a rule that is being built up.  To flush
michael@0:  * the quoteBuf to rule, make one final call with isLiteral == TRUE.
michael@0:  * If there is no final character, pass in (UChar32)-1 as c.
michael@0:  * @param rule the string to append the character to
michael@0:  * @param c the character to append, or (UChar32)-1 if none.
michael@0:  * @param isLiteral if true, then the given character should not be
michael@0:  * quoted or escaped.  Usually this means it is a syntactic element
michael@0:  * such as > or $
michael@0:  * @param escapeUnprintable if true, then unprintable characters
michael@0:  * should be escaped using \uxxxx or \Uxxxxxxxx.  These escapes will
michael@0:  * appear outside of quotes.
michael@0:  * @param quoteBuf a buffer which is used to build up quoted
michael@0:  * substrings.  The caller should initially supply an empty buffer,
michael@0:  * and thereafter should not modify the buffer.  The buffer should be
michael@0:  * cleared out by, at the end, calling this method with a literal
michael@0:  * character.
michael@0:  */
michael@0: void ICU_Utility::appendToRule(UnicodeString& rule,
michael@0:                                UChar32 c,
michael@0:                                UBool isLiteral,
michael@0:                                UBool escapeUnprintable,
michael@0:                                UnicodeString& quoteBuf) {
michael@0:     // If we are escaping unprintables, then escape them outside
michael@0:     // quotes.  \u and \U are not recognized within quotes.  The same
michael@0:     // logic applies to literals, but literals are never escaped.
michael@0:     if (isLiteral ||
michael@0:         (escapeUnprintable && ICU_Utility::isUnprintable(c))) {
michael@0:         if (quoteBuf.length() > 0) {
michael@0:             // We prefer backslash APOSTROPHE to double APOSTROPHE
michael@0:             // (more readable, less similar to ") so if there are
michael@0:             // double APOSTROPHEs at the ends, we pull them outside
michael@0:             // of the quote.
michael@0: 
michael@0:             // If the first thing in the quoteBuf is APOSTROPHE
michael@0:             // (doubled) then pull it out.
michael@0:             while (quoteBuf.length() >= 2 &&
michael@0:                    quoteBuf.charAt(0) == APOSTROPHE &&
michael@0:                    quoteBuf.charAt(1) == APOSTROPHE) {
michael@0:                 rule.append(BACKSLASH).append(APOSTROPHE);
michael@0:                 quoteBuf.remove(0, 2);
michael@0:             }
michael@0:             // If the last thing in the quoteBuf is APOSTROPHE
michael@0:             // (doubled) then remove and count it and add it after.
michael@0:             int32_t trailingCount = 0;
michael@0:             while (quoteBuf.length() >= 2 &&
michael@0:                    quoteBuf.charAt(quoteBuf.length()-2) == APOSTROPHE &&
michael@0:                    quoteBuf.charAt(quoteBuf.length()-1) == APOSTROPHE) {
michael@0:                 quoteBuf.truncate(quoteBuf.length()-2);
michael@0:                 ++trailingCount;
michael@0:             }
michael@0:             if (quoteBuf.length() > 0) {
michael@0:                 rule.append(APOSTROPHE);
michael@0:                 rule.append(quoteBuf);
michael@0:                 rule.append(APOSTROPHE);
michael@0:                 quoteBuf.truncate(0);
michael@0:             }
michael@0:             while (trailingCount-- > 0) {
michael@0:                 rule.append(BACKSLASH).append(APOSTROPHE);
michael@0:             }
michael@0:         }
michael@0:         if (c != (UChar32)-1) {
michael@0:             /* Since spaces are ignored during parsing, they are
michael@0:              * emitted only for readability.  We emit one here
michael@0:              * only if there isn't already one at the end of the
michael@0:              * rule.
michael@0:              */
michael@0:             if (c == SPACE) {
michael@0:                 int32_t len = rule.length();
michael@0:                 if (len > 0 && rule.charAt(len-1) != c) {
michael@0:                     rule.append(c);
michael@0:                 }
michael@0:             } else if (!escapeUnprintable || !ICU_Utility::escapeUnprintable(rule, c)) {
michael@0:                 rule.append(c);
michael@0:             }
michael@0:         }
michael@0:     }
michael@0: 
michael@0:     // Escape ' and '\' and don't begin a quote just for them
michael@0:     else if (quoteBuf.length() == 0 &&
michael@0:              (c == APOSTROPHE || c == BACKSLASH)) {
michael@0:         rule.append(BACKSLASH);
michael@0:         rule.append(c);
michael@0:     }
michael@0: 
michael@0:     // Specials (printable ascii that isn't [0-9a-zA-Z]) and
michael@0:     // whitespace need quoting.  Also append stuff to quotes if we are
michael@0:     // building up a quoted substring already.
michael@0:     else if (quoteBuf.length() > 0 ||
michael@0:              (c >= 0x0021 && c <= 0x007E &&
michael@0:               !((c >= 0x0030/*'0'*/ && c <= 0x0039/*'9'*/) ||
michael@0:                 (c >= 0x0041/*'A'*/ && c <= 0x005A/*'Z'*/) ||
michael@0:                 (c >= 0x0061/*'a'*/ && c <= 0x007A/*'z'*/))) ||
michael@0:              PatternProps::isWhiteSpace(c)) {
michael@0:         quoteBuf.append(c);
michael@0:         // Double ' within a quote
michael@0:         if (c == APOSTROPHE) {
michael@0:             quoteBuf.append(c);
michael@0:         }
michael@0:     }
michael@0:     
michael@0:     // Otherwise just append
michael@0:     else {
michael@0:         rule.append(c);
michael@0:     }
michael@0: }
michael@0: 
michael@0: void ICU_Utility::appendToRule(UnicodeString& rule,
michael@0:                                const UnicodeString& text,
michael@0:                                UBool isLiteral,
michael@0:                                UBool escapeUnprintable,
michael@0:                                UnicodeString& quoteBuf) {
michael@0:     for (int32_t i=0; i<text.length(); ++i) {
michael@0:         appendToRule(rule, text[i], isLiteral, escapeUnprintable, quoteBuf);
michael@0:     }
michael@0: }
michael@0: 
michael@0: /**
michael@0:  * Given a matcher reference, which may be null, append its
michael@0:  * pattern as a literal to the given rule.
michael@0:  */
michael@0: void ICU_Utility::appendToRule(UnicodeString& rule,
michael@0:                                const UnicodeMatcher* matcher,
michael@0:                                UBool escapeUnprintable,
michael@0:                                UnicodeString& quoteBuf) {
michael@0:     if (matcher != NULL) {
michael@0:         UnicodeString pat;
michael@0:         appendToRule(rule, matcher->toPattern(pat, escapeUnprintable),
michael@0:                      TRUE, escapeUnprintable, quoteBuf);
michael@0:     }
michael@0: }
michael@0: 
michael@0: U_NAMESPACE_END