PostgreSQL Source Code  git master
parser.c
Go to the documentation of this file.
1 /*-------------------------------------------------------------------------
2  *
3  * parser.c
4  * Main entry point/driver for PostgreSQL grammar
5  *
6  * This should match src/backend/parser/parser.c, except that we do not
7  * need to bother with re-entrant interfaces.
8  *
9  * Note: ECPG doesn't report error location like the backend does.
10  * This file will need work if we ever want it to.
11  *
12  *
13  * Portions Copyright (c) 1996-2022, PostgreSQL Global Development Group
14  * Portions Copyright (c) 1994, Regents of the University of California
15  *
16  * IDENTIFICATION
17  * src/interfaces/ecpg/preproc/parser.c
18  *
19  *-------------------------------------------------------------------------
20  */
21 
22 #include "postgres_fe.h"
23 
24 #include "preproc_extern.h"
25 #include "preproc.h"
26 
27 
28 static bool have_lookahead; /* is lookahead info valid? */
29 static int lookahead_token; /* one-token lookahead */
30 static YYSTYPE lookahead_yylval; /* yylval for lookahead token */
31 static YYLTYPE lookahead_yylloc; /* yylloc for lookahead token */
32 static char *lookahead_yytext; /* start current token */
33 
34 static bool check_uescapechar(unsigned char escape);
35 static bool ecpg_isspace(char ch);
36 
37 
38 /*
39  * Intermediate filter between parser and base lexer (base_yylex in scan.l).
40  *
41  * This filter is needed because in some cases the standard SQL grammar
42  * requires more than one token lookahead. We reduce these cases to one-token
43  * lookahead by replacing tokens here, in order to keep the grammar LALR(1).
44  *
45  * Using a filter is simpler than trying to recognize multiword tokens
46  * directly in scan.l, because we'd have to allow for comments between the
47  * words. Furthermore it's not clear how to do that without re-introducing
48  * scanner backtrack, which would cost more performance than this filter
49  * layer does.
50  *
51  * We also use this filter to convert UIDENT and USCONST sequences into
52  * plain IDENT and SCONST tokens. While that could be handled by additional
53  * productions in the main grammar, it's more efficient to do it like this.
54  */
55 int
57 {
58  int cur_token;
59  int next_token;
60  YYSTYPE cur_yylval;
61  YYLTYPE cur_yylloc;
62  char *cur_yytext;
63 
64  /* Get next token --- we might already have it */
65  if (have_lookahead)
66  {
67  cur_token = lookahead_token;
68  base_yylval = lookahead_yylval;
69  base_yylloc = lookahead_yylloc;
71  have_lookahead = false;
72  }
73  else
74  cur_token = base_yylex();
75 
76  /*
77  * If this token isn't one that requires lookahead, just return it.
78  */
79  switch (cur_token)
80  {
81  case NOT:
82  case NULLS_P:
83  case WITH:
84  case UIDENT:
85  case USCONST:
86  case WITHOUT:
87  break;
88  default:
89  return cur_token;
90  }
91 
92  /* Save and restore lexer output variables around the call */
93  cur_yylval = base_yylval;
94  cur_yylloc = base_yylloc;
95  cur_yytext = base_yytext;
96 
97  /* Get next token, saving outputs into lookahead variables */
99 
101  lookahead_yylval = base_yylval;
102  lookahead_yylloc = base_yylloc;
104 
105  base_yylval = cur_yylval;
106  base_yylloc = cur_yylloc;
107  base_yytext = cur_yytext;
108 
109  have_lookahead = true;
110 
111  /* Replace cur_token if needed, based on lookahead */
112  switch (cur_token)
113  {
114  case NOT:
115  /* Replace NOT by NOT_LA if it's followed by BETWEEN, IN, etc */
116  switch (next_token)
117  {
118  case BETWEEN:
119  case IN_P:
120  case LIKE:
121  case ILIKE:
122  case SIMILAR:
123  cur_token = NOT_LA;
124  break;
125  }
126  break;
127 
128  case NULLS_P:
129  /* Replace NULLS_P by NULLS_LA if it's followed by FIRST or LAST */
130  switch (next_token)
131  {
132  case FIRST_P:
133  case LAST_P:
134  cur_token = NULLS_LA;
135  break;
136  }
137  break;
138 
139  case WITH:
140  /* Replace WITH by WITH_LA if it's followed by TIME or ORDINALITY */
141  switch (next_token)
142  {
143  case TIME:
144  case ORDINALITY:
145  cur_token = WITH_LA;
146  break;
147  case UNIQUE:
148  cur_token = WITH_LA_UNIQUE;
149  break;
150  }
151  break;
152 
153  case WITHOUT:
154  /* Replace WITHOUT by WITHOUT_LA if it's followed by TIME */
155  switch (next_token)
156  {
157  case TIME:
158  cur_token = WITHOUT_LA;
159  break;
160  }
161  break;
162  case UIDENT:
163  case USCONST:
164  /* Look ahead for UESCAPE */
165  if (next_token == UESCAPE)
166  {
167  /* Yup, so get third token, which had better be SCONST */
168  const char *escstr;
169 
170  /*
171  * Again save and restore lexer output variables around the
172  * call
173  */
174  cur_yylval = base_yylval;
175  cur_yylloc = base_yylloc;
176  cur_yytext = base_yytext;
177 
178  /* Get third token */
180 
181  if (next_token != SCONST)
182  mmerror(PARSE_ERROR, ET_ERROR, "UESCAPE must be followed by a simple string literal");
183 
184  /*
185  * Save and check escape string, which the scanner returns
186  * with quotes
187  */
188  escstr = base_yylval.str;
189  if (strlen(escstr) != 3 || !check_uescapechar(escstr[1]))
190  mmerror(PARSE_ERROR, ET_ERROR, "invalid Unicode escape character");
191 
192  base_yylval = cur_yylval;
193  base_yylloc = cur_yylloc;
194  base_yytext = cur_yytext;
195 
196  /* Combine 3 tokens into 1 */
197  base_yylval.str = psprintf("%s UESCAPE %s", base_yylval.str, escstr);
198 
199  /* Clear have_lookahead, thereby consuming all three tokens */
200  have_lookahead = false;
201  }
202 
203  if (cur_token == UIDENT)
204  cur_token = IDENT;
205  else if (cur_token == USCONST)
206  cur_token = SCONST;
207  break;
208  }
209 
210  return cur_token;
211 }
212 
213 /*
214  * check_uescapechar() and ecpg_isspace() should match their equivalents
215  * in pgc.l.
216  */
217 
218 /* is 'escape' acceptable as Unicode escape character (UESCAPE syntax) ? */
219 static bool
220 check_uescapechar(unsigned char escape)
221 {
222  if (isxdigit(escape)
223  || escape == '+'
224  || escape == '\''
225  || escape == '"'
226  || ecpg_isspace(escape))
227  return false;
228  else
229  return true;
230 }
231 
232 /*
233  * ecpg_isspace() --- return true if flex scanner considers char whitespace
234  */
235 static bool
236 ecpg_isspace(char ch)
237 {
238  if (ch == ' ' ||
239  ch == '\t' ||
240  ch == '\n' ||
241  ch == '\r' ||
242  ch == '\f')
243  return true;
244  return false;
245 }
int base_yylex(YYSTYPE *lvalp, YYLTYPE *llocp, core_yyscan_t yyscanner)
Definition: parser.c:111
static bool next_token(char **lineptr, char *buf, int bufsz, bool *initial_quote, bool *terminating_comma, int elevel, char **err_msg)
Definition: hba.c:169
static YYLTYPE lookahead_yylloc
Definition: parser.c:31
static bool have_lookahead
Definition: parser.c:28
static int lookahead_token
Definition: parser.c:29
static YYSTYPE lookahead_yylval
Definition: parser.c:30
static bool ecpg_isspace(char ch)
Definition: parser.c:236
static char * lookahead_yytext
Definition: parser.c:32
int filtered_base_yylex(void)
Definition: parser.c:56
static bool check_uescapechar(unsigned char escape)
Definition: parser.c:220
#define PARSE_ERROR
char * base_yytext
void mmerror(int errorcode, enum errortype type, const char *error,...) pg_attribute_printf(3
char * psprintf(const char *fmt,...)
Definition: psprintf.c:46
#define YYLTYPE
Definition: scanner.h:44
@ ET_ERROR
Definition: type.h:201