1/*
2**********************************************************************
3*   Copyright (c) 2001-2006, International Business Machines
4*   Corporation and others.  All Rights Reserved.
5**********************************************************************
6*   Date        Name        Description
7*   11/19/2001  aliu        Creation.
8**********************************************************************
9*/
10
11#include "util.h"
12#include "unicode/uchar.h"
13
14U_NAMESPACE_BEGIN
15
16/**
17 * Parse an integer at pos, either of the form \d+ or of the form
18 * 0x[0-9A-Fa-f]+ or 0[0-7]+, that is, in standard decimal, hex,
19 * or octal format.
20 * @param pos INPUT-OUTPUT parameter.  On input, the first
21 * character to parse.  On output, the character after the last
22 * parsed character.
23 */
24int32_t ICU_Utility::parseInteger(const UnicodeString& rule, int32_t& pos, int32_t limit) {
25    int32_t count = 0;
26    int32_t value = 0;
27    int32_t p = pos;
28    int8_t radix = 10;
29
30    if (p < limit && rule.charAt(p) == 48 /*0*/) {
31        if (p+1 < limit && (rule.charAt(p+1) == 0x78 /*x*/ || rule.charAt(p+1) == 0x58 /*X*/)) {
32            p += 2;
33            radix = 16;
34        }
35        else {
36            p++;
37            count = 1;
38            radix = 8;
39        }
40    }
41
42    while (p < limit) {
43        int32_t d = u_digit(rule.charAt(p++), radix);
44        if (d < 0) {
45            --p;
46            break;
47        }
48        ++count;
49        int32_t v = (value * radix) + d;
50        if (v <= value) {
51            // If there are too many input digits, at some point
52            // the value will go negative, e.g., if we have seen
53            // "0x8000000" already and there is another '0', when
54            // we parse the next 0 the value will go negative.
55            return 0;
56        }
57        value = v;
58    }
59    if (count > 0) {
60        pos = p;
61    }
62    return value;
63}
64
65/**
66 * Parse a pattern string starting at offset pos.  Keywords are
67 * matched case-insensitively.  Spaces may be skipped and may be
68 * optional or required.  Integer values may be parsed, and if
69 * they are, they will be returned in the given array.  If
70 * successful, the offset of the next non-space character is
71 * returned.  On failure, -1 is returned.
72 * @param pattern must only contain lowercase characters, which
73 * will match their uppercase equivalents as well.  A space
74 * character matches one or more required spaces.  A '~' character
75 * matches zero or more optional spaces.  A '#' character matches
76 * an integer and stores it in parsedInts, which the caller must
77 * ensure has enough capacity.
78 * @param parsedInts array to receive parsed integers.  Caller
79 * must ensure that parsedInts.length is >= the number of '#'
80 * signs in 'pattern'.
81 * @return the position after the last character parsed, or -1 if
82 * the parse failed
83 */
84int32_t ICU_Utility::parsePattern(const UnicodeString& rule, int32_t pos, int32_t limit,
85                              const UnicodeString& pattern, int32_t* parsedInts) {
86    // TODO Update this to handle surrogates
87    int32_t p;
88    int32_t intCount = 0; // number of integers parsed
89    for (int32_t i=0; i<pattern.length(); ++i) {
90        UChar cpat = pattern.charAt(i);
91        UChar c;
92        switch (cpat) {
93        case 32 /*' '*/:
94            if (pos >= limit) {
95                return -1;
96            }
97            c = rule.charAt(pos++);
98            if (!uprv_isRuleWhiteSpace(c)) {
99                return -1;
100            }
101            // FALL THROUGH to skipWhitespace
102        case 126 /*'~'*/:
103            pos = skipWhitespace(rule, pos);
104            break;
105        case 35 /*'#'*/:
106            p = pos;
107            parsedInts[intCount++] = parseInteger(rule, p, limit);
108            if (p == pos) {
109                // Syntax error; failed to parse integer
110                return -1;
111            }
112            pos = p;
113            break;
114        default:
115            if (pos >= limit) {
116                return -1;
117            }
118            c = (UChar) u_tolower(rule.charAt(pos++));
119            if (c != cpat) {
120                return -1;
121            }
122            break;
123        }
124    }
125    return pos;
126}
127
128/**
129 * Parse a Unicode identifier from the given string at the given
130 * position.  Return the identifier, or an empty string if there
131 * is no identifier.
132 * @param str the string to parse
133 * @param pos INPUT-OUPUT parameter.  On INPUT, pos is the
134 * first character to examine.  It must be less than str.length(),
135 * and it must not point to a whitespace character.  That is, must
136 * have pos < str.length() and
137 * !uprv_isRuleWhiteSpace(str.char32At(pos)).  On
138 * OUTPUT, the position after the last parsed character.
139 * @return the Unicode identifier, or an empty string if there is
140 * no valid identifier at pos.
141 */
142UnicodeString ICU_Utility::parseUnicodeIdentifier(const UnicodeString& str, int32_t& pos) {
143    // assert(pos < str.length());
144    // assert(!uprv_isRuleWhiteSpace(str.char32At(pos)));
145    UnicodeString buf;
146    int p = pos;
147    while (p < str.length()) {
148        UChar32 ch = str.char32At(p);
149        if (buf.length() == 0) {
150            if (u_isIDStart(ch)) {
151                buf.append(ch);
152            } else {
153                buf.truncate(0);
154                return buf;
155            }
156        } else {
157            if (u_isIDPart(ch)) {
158                buf.append(ch);
159            } else {
160                break;
161            }
162        }
163        p += UTF_CHAR_LENGTH(ch);
164    }
165    pos = p;
166    return buf;
167}
168
169/**
170 * Parse an unsigned 31-bit integer at the given offset.  Use
171 * UCharacter.digit() to parse individual characters into digits.
172 * @param text the text to be parsed
173 * @param pos INPUT-OUTPUT parameter.  On entry, pos[0] is the
174 * offset within text at which to start parsing; it should point
175 * to a valid digit.  On exit, pos[0] is the offset after the last
176 * parsed character.  If the parse failed, it will be unchanged on
177 * exit.  Must be >= 0 on entry.
178 * @param radix the radix in which to parse; must be >= 2 and <=
179 * 36.
180 * @return a non-negative parsed number, or -1 upon parse failure.
181 * Parse fails if there are no digits, that is, if pos[0] does not
182 * point to a valid digit on entry, or if the number to be parsed
183 * does not fit into a 31-bit unsigned integer.
184 */
185int32_t ICU_Utility::parseNumber(const UnicodeString& text,
186                                 int32_t& pos, int8_t radix) {
187    // assert(pos[0] >= 0);
188    // assert(radix >= 2);
189    // assert(radix <= 36);
190    int32_t n = 0;
191    int32_t p = pos;
192    while (p < text.length()) {
193        UChar32 ch = text.char32At(p);
194        int32_t d = u_digit(ch, radix);
195        if (d < 0) {
196            break;
197        }
198        n = radix*n + d;
199        // ASSUME that when a 32-bit integer overflows it becomes
200        // negative.  E.g., 214748364 * 10 + 8 => negative value.
201        if (n < 0) {
202            return -1;
203        }
204        ++p;
205    }
206    if (p == pos) {
207        return -1;
208    }
209    pos = p;
210    return n;
211}
212
213U_NAMESPACE_END
214
215