1 package simpledb.parse;
2
3 import java.util.*;
4 import java.io.*;
5
6 /**
7 * The lexical analyzer.
8 * @author Edward Sciore
9 */
10 public class Lexer {
11 private Collection<String> keywords;
12 private StreamTokenizer tok;
13
14 /**
15 * Creates a new lexical analyzer for SQL statement s.
16 * @param s the SQL statement
17 */
18 public Lexer(String s) {
19 initKeywords();
20 tok = new StreamTokenizer(new StringReader(s));
21 tok.ordinaryChar('.');
22 tok.lowerCaseMode(true); //ids and keywords are converted
23 nextToken();
24 }
25
26 //Methods to check the status of the current token
27
28 /**
29 * Returns true if the current token is
30 * the specified delimiter character.
31 * @param d a character denoting the delimiter
32 * @return true if the delimiter is the current token
33 */
34 public boolean matchDelim(char d) {
35 return d == (char)tok.ttype;
36 }
37
38 /**
39 * Returns true if the current token is an integer.
40 * @return true if the current token is an integer
41 */
42 public boolean matchIntConstant() {
43 return tok.ttype == StreamTokenizer.TT_NUMBER;
44 }
45
46 /**
47 * Returns true if the current token is a string.
48 * @return true if the current token is a string
49 */
50 public boolean matchStringConstant() {
51 return '\'' == (char)tok.ttype;
52 }
53
54 /**
55 * Returns true if the current token is the specified keyword.
56 * @param w the keyword string
57 * @return true if that keyword is the current token
58 */
59 public boolean matchKeyword(String w) {
60 return tok.ttype == StreamTokenizer.TT_WORD && tok.sval.equals(w);
61 }
62
63 /**
64 * Returns true if the current token is a legal identifier.
65 * @return true if the current token is an identifier
66 */
67 public boolean matchId() {
68 return tok.ttype==StreamTokenizer.TT_WORD && !keywords.contains(tok.sval);
69 }
70
71 //Methods to "eat" the current token
72
73 /**
74 * Throws an exception if the current token is not the
75 * specified delimiter.
76 * Otherwise, moves to the next token.
77 * @param d a character denoting the delimiter
78 */
79 public void eatDelim(char d) {
80 if (!matchDelim(d))
81 throw new BadSyntaxException();
82 nextToken();
83 }
84
85 /**
86 * Throws an exception if the current token is not
87 * an integer.
88 * Otherwise, returns that integer and moves to the next token.
89 * @return the integer value of the current token
90 */
91 public int eatIntConstant() {
92 if (!matchIntConstant())
93 throw new BadSyntaxException();
94 int i = (int) tok.nval;
95 nextToken();
96 return i;
97 }
98
99 /**
100 * Throws an exception if the current token is not
101 * a string.
102 * Otherwise, returns that string and moves to the next token.
103 * @return the string value of the current token
104 */
105 public String eatStringConstant() {
106 if (!matchStringConstant())
107 throw new BadSyntaxException();
108 String s = tok.sval; //constants are not converted to lower case
109 nextToken();
110 return s;
111 }
112
113 /**
114 * Throws an exception if the current token is not the
115 * specified keyword.
116 * Otherwise, moves to the next token.
117 * @param w the keyword string
118 */
119 public void eatKeyword(String w) {
120 if (!matchKeyword(w))
121 throw new BadSyntaxException();
122 nextToken();
123 }
124
125 /**
126 * Throws an exception if the current token is not
127 * an identifier.
128 * Otherwise, returns the identifier string
129 * and moves to the next token.
130 * @return the string value of the current token
131 */
132 public String eatId() {
133 if (!matchId())
134 throw new BadSyntaxException();
135 String s = tok.sval;
136 nextToken();
137 return s;
138 }
139
140 private void nextToken() {
141 try {
142 tok.nextToken();
143 }
144 catch(IOException e) {
145 throw new BadSyntaxException();
146 }
147 }
148
149 private void initKeywords() {
150 keywords = Arrays.asList("select", "from", "where", "and",
151 "insert", "into", "values", "delete", "update", "set",
152 "create", "table", "int", "varchar", "view", "as", "index", "on");
153 }
154 }