1 /*
2  * Copyright (C) 2016 The Android Open Source Project
3  *
4  * Licensed under the Apache License, Version 2.0 (the "License");
5  * you may not use this file except in compliance with the License.
6  * You may obtain a copy of the License at
7  *
8  *      http://www.apache.org/licenses/LICENSE-2.0
9  *
10  * Unless required by applicable law or agreed to in writing, software
11  * distributed under the License is distributed on an "AS IS" BASIS,
12  * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13  * See the License for the specific language governing permissions and
14  * limitations under the License.
15  */
16 
17 #include <array>
18 #include <cstdint>
19 #include <cstdlib>
20 #include <cstring>
21 #include <string>
22 #include <unordered_map>
23 #include <unordered_set>
24 
25 #include <androidfw/LocaleData.h>
26 
27 namespace android {
28 
29 #include "LocaleDataTables.cpp"
30 
31 inline uint32_t packLocale(const char* language, const char* region) {
32     return (((uint8_t) language[0]) << 24u) | (((uint8_t) language[1]) << 16u) |
33            (((uint8_t) region[0]) << 8u) | ((uint8_t) region[1]);
34 }
35 
36 inline uint32_t dropRegion(uint32_t packed_locale) {
37     return packed_locale & 0xFFFF0000LU;
38 }
39 
40 inline bool hasRegion(uint32_t packed_locale) {
41     return (packed_locale & 0x0000FFFFLU) != 0;
42 }
43 
44 const size_t SCRIPT_LENGTH = 4;
45 const size_t SCRIPT_PARENTS_COUNT = sizeof(SCRIPT_PARENTS)/sizeof(SCRIPT_PARENTS[0]);
46 const uint32_t PACKED_ROOT = 0; // to represent the root locale
47 
48 uint32_t findParent(uint32_t packed_locale, const char* script) {
49     if (hasRegion(packed_locale)) {
50         for (size_t i = 0; i < SCRIPT_PARENTS_COUNT; i++) {
51             if (memcmp(script, SCRIPT_PARENTS[i].script, SCRIPT_LENGTH) == 0) {
52                 auto map = SCRIPT_PARENTS[i].map;
53                 auto lookup_result = map->find(packed_locale);
54                 if (lookup_result != map->end()) {
55                     return lookup_result->second;
56                 }
57                 break;
58             }
59         }
60         return dropRegion(packed_locale);
61     }
62     return PACKED_ROOT;
63 }
64 
65 // Find the ancestors of a locale, and fill 'out' with it (assumes out has enough
66 // space). If any of the members of stop_list was seen, write it in the
67 // output but stop afterwards.
68 //
69 // This also outputs the index of the last written ancestor in the stop_list
70 // to stop_list_index, which will be -1 if it is not found in the stop_list.
71 //
72 // Returns the number of ancestors written in the output, which is always
73 // at least one.
74 //
75 // (If 'out' is nullptr, we do everything the same way but we simply don't write
76 // any results in 'out'.)
77 size_t findAncestors(uint32_t* out, ssize_t* stop_list_index,
78                      uint32_t packed_locale, const char* script,
79                      const uint32_t* stop_list, size_t stop_set_length) {
80     uint32_t ancestor = packed_locale;
81     size_t count = 0;
82     do {
83         if (out != nullptr) out[count] = ancestor;
84         count++;
85         for (size_t i = 0; i < stop_set_length; i++) {
86             if (stop_list[i] == ancestor) {
87                 *stop_list_index = (ssize_t) i;
88                 return count;
89             }
90         }
91         ancestor = findParent(ancestor, script);
92     } while (ancestor != PACKED_ROOT);
93     *stop_list_index = (ssize_t) -1;
94     return count;
95 }
96 
97 size_t findDistance(uint32_t supported,
98                     const char* script,
99                     const uint32_t* request_ancestors,
100                     size_t request_ancestors_count) {
101     ssize_t request_ancestors_index;
102     const size_t supported_ancestor_count = findAncestors(
103             nullptr, &request_ancestors_index,
104             supported, script,
105             request_ancestors, request_ancestors_count);
106     // Since both locales share the same root, there will always be a shared
107     // ancestor, so the distance in the parent tree is the sum of the distance
108     // of 'supported' to the lowest common ancestor (number of ancestors
109     // written for 'supported' minus 1) plus the distance of 'request' to the
110     // lowest common ancestor (the index of the ancestor in request_ancestors).
111     return supported_ancestor_count + request_ancestors_index - 1;
112 }
113 
114 inline bool isRepresentative(uint32_t language_and_region, const char* script) {
115     const uint64_t packed_locale = (
116             (((uint64_t) language_and_region) << 32u) |
117             (((uint64_t) script[0]) << 24u) |
118             (((uint64_t) script[1]) << 16u) |
119             (((uint64_t) script[2]) <<  8u) |
120             ((uint64_t) script[3]));
121 
122     return (REPRESENTATIVE_LOCALES.count(packed_locale) != 0);
123 }
124 
125 const uint32_t US_SPANISH = 0x65735553LU; // es-US
126 const uint32_t MEXICAN_SPANISH = 0x65734D58LU; // es-MX
127 const uint32_t LATIN_AMERICAN_SPANISH = 0x6573A424LU; // es-419
128 
129 // The two locales es-US and es-MX are treated as special fallbacks for es-419.
130 // If there is no es-419, they are considered its equivalent.
131 inline bool isSpecialSpanish(uint32_t language_and_region) {
132     return (language_and_region == US_SPANISH || language_and_region == MEXICAN_SPANISH);
133 }
134 
135 int localeDataCompareRegions(
136         const char* left_region, const char* right_region,
137         const char* requested_language, const char* requested_script,
138         const char* requested_region) {
139 
140     if (left_region[0] == right_region[0] && left_region[1] == right_region[1]) {
141         return 0;
142     }
143     uint32_t left = packLocale(requested_language, left_region);
144     uint32_t right = packLocale(requested_language, right_region);
145     const uint32_t request = packLocale(requested_language, requested_region);
146 
147     // If one and only one of the two locales is a special Spanish locale, we
148     // replace it with es-419. We don't do the replacement if the other locale
149     // is already es-419, or both locales are special Spanish locales (when
150     // es-US is being compared to es-MX).
151     const bool leftIsSpecialSpanish = isSpecialSpanish(left);
152     const bool rightIsSpecialSpanish = isSpecialSpanish(right);
153     if (leftIsSpecialSpanish && !rightIsSpecialSpanish && right != LATIN_AMERICAN_SPANISH) {
154         left = LATIN_AMERICAN_SPANISH;
155     } else if (rightIsSpecialSpanish && !leftIsSpecialSpanish && left != LATIN_AMERICAN_SPANISH) {
156         right = LATIN_AMERICAN_SPANISH;
157     }
158 
159     uint32_t request_ancestors[MAX_PARENT_DEPTH+1];
160     ssize_t left_right_index;
161     // Find the parents of the request, but stop as soon as we saw left or right
162     const std::array<uint32_t, 2> left_and_right = {{left, right}};
163     const size_t ancestor_count = findAncestors(
164             request_ancestors, &left_right_index,
165             request, requested_script,
166             left_and_right.data(), left_and_right.size());
167     if (left_right_index == 0) { // We saw left earlier
168         return 1;
169     }
170     if (left_right_index == 1) { // We saw right earlier
171         return -1;
172     }
173 
174     // If we are here, neither left nor right are an ancestor of the
175     // request. This means that all the ancestors have been computed and
176     // the last ancestor is just the language by itself. We will use the
177     // distance in the parent tree for determining the better match.
178     const size_t left_distance = findDistance(
179             left, requested_script, request_ancestors, ancestor_count);
180     const size_t right_distance = findDistance(
181             right, requested_script, request_ancestors, ancestor_count);
182     if (left_distance != right_distance) {
183         return (int) right_distance - (int) left_distance; // smaller distance is better
184     }
185 
186     // If we are here, left and right are equidistant from the request. We will
187     // try and see if any of them is a representative locale.
188     const bool left_is_representative = isRepresentative(left, requested_script);
189     const bool right_is_representative = isRepresentative(right, requested_script);
190     if (left_is_representative != right_is_representative) {
191         return (int) left_is_representative - (int) right_is_representative;
192     }
193 
194     // We have no way of figuring out which locale is a better match. For
195     // the sake of stability, we consider the locale with the lower region
196     // code (in dictionary order) better, with two-letter codes before
197     // three-digit codes (since two-letter codes are more specific).
198     return (int64_t) right - (int64_t) left;
199 }
200 
201 void localeDataComputeScript(char out[4], const char* language, const char* region) {
202     if (language[0] == '\0') {
203         memset(out, '\0', SCRIPT_LENGTH);
204         return;
205     }
206     uint32_t lookup_key = packLocale(language, region);
207     auto lookup_result = LIKELY_SCRIPTS.find(lookup_key);
208     if (lookup_result == LIKELY_SCRIPTS.end()) {
209         // We couldn't find the locale. Let's try without the region
210         if (region[0] != '\0') {
211             lookup_key = dropRegion(lookup_key);
212             lookup_result = LIKELY_SCRIPTS.find(lookup_key);
213             if (lookup_result != LIKELY_SCRIPTS.end()) {
214                 memcpy(out, SCRIPT_CODES[lookup_result->second], SCRIPT_LENGTH);
215                 return;
216             }
217         }
218         // We don't know anything about the locale
219         memset(out, '\0', SCRIPT_LENGTH);
220         return;
221     } else {
222         // We found the locale.
223         memcpy(out, SCRIPT_CODES[lookup_result->second], SCRIPT_LENGTH);
224     }
225 }
226 
227 const uint32_t ENGLISH_STOP_LIST[2] = {
228     0x656E0000LU, // en
229     0x656E8400LU, // en-001
230 };
231 const char ENGLISH_CHARS[2] = {'e', 'n'};
232 const char LATIN_CHARS[4] = {'L', 'a', 't', 'n'};
233 
234 bool localeDataIsCloseToUsEnglish(const char* region) {
235     const uint32_t locale = packLocale(ENGLISH_CHARS, region);
236     ssize_t stop_list_index;
237     findAncestors(nullptr, &stop_list_index, locale, LATIN_CHARS, ENGLISH_STOP_LIST, 2);
238     // A locale is like US English if we see "en" before "en-001" in its ancestor list.
239     return stop_list_index == 0; // 'en' is first in ENGLISH_STOP_LIST
240 }
241 
242 } // namespace android
243