1// Copyright (C) 2016 and later: Unicode, Inc. and others.
2// License & terms of use: http://www.unicode.org/copyright.html
3/*
4**********************************************************************
5*   Copyright (c) 2001-2016, International Business Machines
6*   Corporation and others.  All Rights Reserved.
7**********************************************************************
8*   Date        Name        Description
9*   11/19/2001  aliu        Creation.
10**********************************************************************
11*/
12
13#include "unicode/uchar.h"
14#include "unicode/utf16.h"
15#include "patternprops.h"
16#include "util.h"
17
18U_NAMESPACE_BEGIN
19
20/**
21 * Parse an integer at pos, either of the form \d+ or of the form
22 * 0x[0-9A-Fa-f]+ or 0[0-7]+, that is, in standard decimal, hex,
23 * or octal format.
24 * @param pos INPUT-OUTPUT parameter.  On input, the first
25 * character to parse.  On output, the character after the last
26 * parsed character.
27 */
28int32_t ICU_Utility::parseInteger(const UnicodeString& rule, int32_t& pos, int32_t limit) {
29    int32_t count = 0;
30    int32_t value = 0;
31    int32_t p = pos;
32    int8_t radix = 10;
33
34    if (p < limit && rule.charAt(p) == 48 /*0*/) {
35        if (p+1 < limit && (rule.charAt(p+1) == 0x78 /*x*/ || rule.charAt(p+1) == 0x58 /*X*/)) {
36            p += 2;
37            radix = 16;
38        }
39        else {
40            p++;
41            count = 1;
42            radix = 8;
43        }
44    }
45
46    while (p < limit) {
47        int32_t d = u_digit(rule.charAt(p++), radix);
48        if (d < 0) {
49            --p;
50            break;
51        }
52        ++count;
53        int32_t v = (value * radix) + d;
54        if (v <= value) {
55            // If there are too many input digits, at some point
56            // the value will go negative, e.g., if we have seen
57            // "0x8000000" already and there is another '0', when
58            // we parse the next 0 the value will go negative.
59            return 0;
60        }
61        value = v;
62    }
63    if (count > 0) {
64        pos = p;
65    }
66    return value;
67}
68
69/**
70 * Parse a pattern string starting at offset pos.  Keywords are
71 * matched case-insensitively.  Spaces may be skipped and may be
72 * optional or required.  Integer values may be parsed, and if
73 * they are, they will be returned in the given array.  If
74 * successful, the offset of the next non-space character is
75 * returned.  On failure, -1 is returned.
76 * @param pattern must only contain lowercase characters, which
77 * will match their uppercase equivalents as well.  A space
78 * character matches one or more required spaces.  A '~' character
79 * matches zero or more optional spaces.  A '#' character matches
80 * an integer and stores it in parsedInts, which the caller must
81 * ensure has enough capacity.
82 * @param parsedInts array to receive parsed integers.  Caller
83 * must ensure that parsedInts.length is >= the number of '#'
84 * signs in 'pattern'.
85 * @return the position after the last character parsed, or -1 if
86 * the parse failed
87 */
88int32_t ICU_Utility::parsePattern(const UnicodeString& rule, int32_t pos, int32_t limit,
89                              const UnicodeString& pattern, int32_t* parsedInts) {
90    // TODO Update this to handle surrogates
91    int32_t p;
92    int32_t intCount = 0; // number of integers parsed
93    for (int32_t i=0; i<pattern.length(); ++i) {
94        UChar cpat = pattern.charAt(i);
95        UChar c;
96        switch (cpat) {
97        case 32 /*' '*/:
98            if (pos >= limit) {
99                return -1;
100            }
101            c = rule.charAt(pos++);
102            if (!PatternProps::isWhiteSpace(c)) {
103                return -1;
104            }
105            // FALL THROUGH to skipWhitespace
106            U_FALLTHROUGH;
107        case 126 /*'~'*/:
108            pos = skipWhitespace(rule, pos);
109            break;
110        case 35 /*'#'*/:
111            p = pos;
112            parsedInts[intCount++] = parseInteger(rule, p, limit);
113            if (p == pos) {
114                // Syntax error; failed to parse integer
115                return -1;
116            }
117            pos = p;
118            break;
119        default:
120            if (pos >= limit) {
121                return -1;
122            }
123            c = (UChar) u_tolower(rule.charAt(pos++));
124            if (c != cpat) {
125                return -1;
126            }
127            break;
128        }
129    }
130    return pos;
131}
132
133/**
134 * Parse a Unicode identifier from the given string at the given
135 * position.  Return the identifier, or an empty string if there
136 * is no identifier.
137 * @param str the string to parse
138 * @param pos INPUT-OUPUT parameter.  On INPUT, pos is the
139 * first character to examine.  It must be less than str.length(),
140 * and it must not point to a whitespace character.  That is, must
141 * have pos < str.length().  On
142 * OUTPUT, the position after the last parsed character.
143 * @return the Unicode identifier, or an empty string if there is
144 * no valid identifier at pos.
145 */
146UnicodeString ICU_Utility::parseUnicodeIdentifier(const UnicodeString& str, int32_t& pos) {
147    // assert(pos < str.length());
148    UnicodeString buf;
149    int p = pos;
150    while (p < str.length()) {
151        UChar32 ch = str.char32At(p);
152        if (buf.length() == 0) {
153            if (u_isIDStart(ch)) {
154                buf.append(ch);
155            } else {
156                buf.truncate(0);
157                return buf;
158            }
159        } else {
160            if (u_isIDPart(ch)) {
161                buf.append(ch);
162            } else {
163                break;
164            }
165        }
166        p += U16_LENGTH(ch);
167    }
168    pos = p;
169    return buf;
170}
171
172/**
173 * Parse an unsigned 31-bit integer at the given offset.  Use
174 * UCharacter.digit() to parse individual characters into digits.
175 * @param text the text to be parsed
176 * @param pos INPUT-OUTPUT parameter.  On entry, pos[0] is the
177 * offset within text at which to start parsing; it should point
178 * to a valid digit.  On exit, pos[0] is the offset after the last
179 * parsed character.  If the parse failed, it will be unchanged on
180 * exit.  Must be >= 0 on entry.
181 * @param radix the radix in which to parse; must be >= 2 and <=
182 * 36.
183 * @return a non-negative parsed number, or -1 upon parse failure.
184 * Parse fails if there are no digits, that is, if pos[0] does not
185 * point to a valid digit on entry, or if the number to be parsed
186 * does not fit into a 31-bit unsigned integer.
187 */
188int32_t ICU_Utility::parseNumber(const UnicodeString& text,
189                                 int32_t& pos, int8_t radix) {
190    // assert(pos[0] >= 0);
191    // assert(radix >= 2);
192    // assert(radix <= 36);
193    int32_t n = 0;
194    int32_t p = pos;
195    while (p < text.length()) {
196        UChar32 ch = text.char32At(p);
197        int32_t d = u_digit(ch, radix);
198        if (d < 0) {
199            break;
200        }
201        n = radix*n + d;
202        // ASSUME that when a 32-bit integer overflows it becomes
203        // negative.  E.g., 214748364 * 10 + 8 => negative value.
204        if (n < 0) {
205            return -1;
206        }
207        ++p;
208    }
209    if (p == pos) {
210        return -1;
211    }
212    pos = p;
213    return n;
214}
215
216U_NAMESPACE_END
217
218