1 /*
2  * Copyright (C) 2019 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 package android.database.sqlite;
18 
19 import android.annotation.NonNull;
20 import android.annotation.Nullable;
21 
22 import java.util.ArrayList;
23 import java.util.List;
24 import java.util.Locale;
25 import java.util.function.Consumer;
26 
27 /**
28  * SQL Tokenizer specialized to extract tokens from SQL (snippets).
29  * <p>
30  * Based on sqlite3GetToken() in tokenzie.c in SQLite.
31  * <p>
32  * Source for v3.8.6 (which android uses): http://www.sqlite.org/src/artifact/ae45399d6252b4d7
33  * (Latest source as of now: http://www.sqlite.org/src/artifact/78c8085bc7af1922)
34  * <p>
35  * Also draft spec: http://www.sqlite.org/draft/tokenreq.html
36  *
37  * @hide
38  */
39 public class SQLiteTokenizer {
isAlpha(char ch)40     private static boolean isAlpha(char ch) {
41         return ('a' <= ch && ch <= 'z') || ('A' <= ch && ch <= 'Z') || (ch == '_');
42     }
43 
isNum(char ch)44     private static boolean isNum(char ch) {
45         return ('0' <= ch && ch <= '9');
46     }
47 
isAlNum(char ch)48     private static boolean isAlNum(char ch) {
49         return isAlpha(ch) || isNum(ch);
50     }
51 
isAnyOf(char ch, String set)52     private static boolean isAnyOf(char ch, String set) {
53         return set.indexOf(ch) >= 0;
54     }
55 
genException(String message, String sql)56     private static IllegalArgumentException genException(String message, String sql) {
57         throw new IllegalArgumentException(message + " in '" + sql + "'");
58     }
59 
peek(String s, int index)60     private static char peek(String s, int index) {
61         return index < s.length() ? s.charAt(index) : '\0';
62     }
63 
64     public static final int OPTION_NONE = 0;
65 
66     /**
67      * Require that SQL contains only tokens; any comments or values will result
68      * in an exception.
69      */
70     public static final int OPTION_TOKEN_ONLY = 1 << 0;
71 
72     /**
73      * Tokenize the given SQL, returning the list of each encountered token.
74      *
75      * @throws IllegalArgumentException if invalid SQL is encountered.
76      */
tokenize(@ullable String sql, int options)77     public static List<String> tokenize(@Nullable String sql, int options) {
78         final ArrayList<String> res = new ArrayList<>();
79         tokenize(sql, options, res::add);
80         return res;
81     }
82 
83     /**
84      * Tokenize the given SQL, sending each encountered token to the given
85      * {@link Consumer}.
86      *
87      * @throws IllegalArgumentException if invalid SQL is encountered.
88      */
tokenize(@ullable String sql, int options, Consumer<String> checker)89     public static void tokenize(@Nullable String sql, int options, Consumer<String> checker) {
90         if (sql == null) {
91             return;
92         }
93         int pos = 0;
94         final int len = sql.length();
95         while (pos < len) {
96             final char ch = peek(sql, pos);
97 
98             // Regular token.
99             if (isAlpha(ch)) {
100                 final int start = pos;
101                 pos++;
102                 while (isAlNum(peek(sql, pos))) {
103                     pos++;
104                 }
105                 final int end = pos;
106 
107                 final String token = sql.substring(start, end);
108                 checker.accept(token);
109 
110                 continue;
111             }
112 
113             // Handle quoted tokens
114             if (isAnyOf(ch, "'\"`")) {
115                 final int quoteStart = pos;
116                 pos++;
117 
118                 for (;;) {
119                     pos = sql.indexOf(ch, pos);
120                     if (pos < 0) {
121                         throw genException("Unterminated quote", sql);
122                     }
123                     if (peek(sql, pos + 1) != ch) {
124                         break;
125                     }
126                     // Quoted quote char -- e.g. "abc""def" is a single string.
127                     pos += 2;
128                 }
129                 final int quoteEnd = pos;
130                 pos++;
131 
132                 if (ch != '\'') {
133                     // Extract the token
134                     final String tokenUnquoted = sql.substring(quoteStart + 1, quoteEnd);
135 
136                     final String token;
137 
138                     // Unquote if needed. i.e. "aa""bb" -> aa"bb
139                     if (tokenUnquoted.indexOf(ch) >= 0) {
140                         token = tokenUnquoted.replaceAll(
141                                 String.valueOf(ch) + ch, String.valueOf(ch));
142                     } else {
143                         token = tokenUnquoted;
144                     }
145                     checker.accept(token);
146                 } else {
147                     if ((options &= OPTION_TOKEN_ONLY) != 0) {
148                         throw genException("Non-token detected", sql);
149                     }
150                 }
151                 continue;
152             }
153             // Handle tokens enclosed in [...]
154             if (ch == '[') {
155                 final int quoteStart = pos;
156                 pos++;
157 
158                 pos = sql.indexOf(']', pos);
159                 if (pos < 0) {
160                     throw genException("Unterminated quote", sql);
161                 }
162                 final int quoteEnd = pos;
163                 pos++;
164 
165                 final String token = sql.substring(quoteStart + 1, quoteEnd);
166 
167                 checker.accept(token);
168                 continue;
169             }
170             if ((options &= OPTION_TOKEN_ONLY) != 0) {
171                 throw genException("Non-token detected", sql);
172             }
173 
174             // Detect comments.
175             if (ch == '-' && peek(sql, pos + 1) == '-') {
176                 pos += 2;
177                 pos = sql.indexOf('\n', pos);
178                 if (pos < 0) {
179                     // We disallow strings ending in an inline comment.
180                     throw genException("Unterminated comment", sql);
181                 }
182                 pos++;
183 
184                 continue;
185             }
186             if (ch == '/' && peek(sql, pos + 1) == '*') {
187                 pos += 2;
188                 pos = sql.indexOf("*/", pos);
189                 if (pos < 0) {
190                     throw genException("Unterminated comment", sql);
191                 }
192                 pos += 2;
193 
194                 continue;
195             }
196 
197             // Semicolon is never allowed.
198             if (ch == ';') {
199                 throw genException("Semicolon is not allowed", sql);
200             }
201 
202             // For this purpose, we can simply ignore other characters.
203             // (Note it doesn't handle the X'' literal properly and reports this X as a token,
204             // but that should be fine...)
205             pos++;
206         }
207     }
208 
209     /**
210      * Test if given token is a
211      * <a href="https://www.sqlite.org/lang_keywords.html">SQLite reserved
212      * keyword</a>.
213      */
isKeyword(@onNull String token)214     public static boolean isKeyword(@NonNull String token) {
215         switch (token.toUpperCase(Locale.US)) {
216             case "ABORT": case "ACTION": case "ADD": case "AFTER":
217             case "ALL": case "ALTER": case "ANALYZE": case "AND":
218             case "AS": case "ASC": case "ATTACH": case "AUTOINCREMENT":
219             case "BEFORE": case "BEGIN": case "BETWEEN": case "BINARY":
220             case "BY": case "CASCADE": case "CASE": case "CAST":
221             case "CHECK": case "COLLATE": case "COLUMN": case "COMMIT":
222             case "CONFLICT": case "CONSTRAINT": case "CREATE": case "CROSS":
223             case "CURRENT": case "CURRENT_DATE": case "CURRENT_TIME": case "CURRENT_TIMESTAMP":
224             case "DATABASE": case "DEFAULT": case "DEFERRABLE": case "DEFERRED":
225             case "DELETE": case "DESC": case "DETACH": case "DISTINCT":
226             case "DO": case "DROP": case "EACH": case "ELSE":
227             case "END": case "ESCAPE": case "EXCEPT": case "EXCLUDE":
228             case "EXCLUSIVE": case "EXISTS": case "EXPLAIN": case "FAIL":
229             case "FILTER": case "FOLLOWING": case "FOR": case "FOREIGN":
230             case "FROM": case "FULL": case "GLOB": case "GROUP":
231             case "GROUPS": case "HAVING": case "IF": case "IGNORE":
232             case "IMMEDIATE": case "IN": case "INDEX": case "INDEXED":
233             case "INITIALLY": case "INNER": case "INSERT": case "INSTEAD":
234             case "INTERSECT": case "INTO": case "IS": case "ISNULL":
235             case "JOIN": case "KEY": case "LEFT": case "LIKE":
236             case "LIMIT": case "MATCH": case "NATURAL": case "NO":
237             case "NOCASE": case "NOT": case "NOTHING": case "NOTNULL":
238             case "NULL": case "OF": case "OFFSET": case "ON":
239             case "OR": case "ORDER": case "OTHERS": case "OUTER":
240             case "OVER": case "PARTITION": case "PLAN": case "PRAGMA":
241             case "PRECEDING": case "PRIMARY": case "QUERY": case "RAISE":
242             case "RANGE": case "RECURSIVE": case "REFERENCES": case "REGEXP":
243             case "REINDEX": case "RELEASE": case "RENAME": case "REPLACE":
244             case "RESTRICT": case "RIGHT": case "ROLLBACK": case "ROW":
245             case "ROWS": case "RTRIM": case "SAVEPOINT": case "SELECT":
246             case "SET": case "TABLE": case "TEMP": case "TEMPORARY":
247             case "THEN": case "TIES": case "TO": case "TRANSACTION":
248             case "TRIGGER": case "UNBOUNDED": case "UNION": case "UNIQUE":
249             case "UPDATE": case "USING": case "VACUUM": case "VALUES":
250             case "VIEW": case "VIRTUAL": case "WHEN": case "WHERE":
251             case "WINDOW": case "WITH": case "WITHOUT":
252                 return true;
253             default:
254                 return false;
255         }
256     }
257 
258     /**
259      * Test if given token is a
260      * <a href="https://www.sqlite.org/lang_corefunc.html">SQLite reserved
261      * function</a>.
262      */
isFunction(@onNull String token)263     public static boolean isFunction(@NonNull String token) {
264         switch (token.toLowerCase(Locale.US)) {
265             case "abs": case "avg": case "char": case "coalesce":
266             case "count": case "glob": case "group_concat": case "hex":
267             case "ifnull": case "instr": case "length": case "like":
268             case "likelihood": case "likely": case "lower": case "ltrim":
269             case "max": case "min": case "nullif": case "random":
270             case "randomblob": case "replace": case "round": case "rtrim":
271             case "substr": case "sum": case "total": case "trim":
272             case "typeof": case "unicode": case "unlikely": case "upper":
273             case "zeroblob":
274                 return true;
275             default:
276                 return false;
277         }
278     }
279 
280     /**
281      * Test if given token is a
282      * <a href="https://www.sqlite.org/datatype3.html">SQLite reserved type</a>.
283      */
isType(@onNull String token)284     public static boolean isType(@NonNull String token) {
285         switch (token.toUpperCase(Locale.US)) {
286             case "INT": case "INTEGER": case "TINYINT": case "SMALLINT":
287             case "MEDIUMINT": case "BIGINT": case "INT2": case "INT8":
288             case "CHARACTER": case "VARCHAR": case "NCHAR": case "NVARCHAR":
289             case "TEXT": case "CLOB": case "BLOB": case "REAL":
290             case "DOUBLE": case "FLOAT": case "NUMERIC": case "DECIMAL":
291             case "BOOLEAN": case "DATE": case "DATETIME":
292                 return true;
293             default:
294                 return false;
295         }
296     }
297 }
298