pg_query 0.5.0 → 0.6.0
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/Rakefile +13 -10
- data/ext/pg_query/extconf.rb +16 -14
- data/ext/pg_query/pg_query.c +3 -466
- data/ext/pg_query/pg_query.h +18 -0
- data/ext/pg_query/pg_query_normalize.c +363 -0
- data/ext/pg_query/pg_query_parse.c +102 -0
- data/lib/pg_query.rb +4 -0
- data/lib/pg_query/filter_columns.rb +52 -39
- data/lib/pg_query/fingerprint.rb +25 -50
- data/lib/pg_query/param_refs.rb +37 -32
- data/lib/pg_query/parse.rb +65 -45
- data/lib/pg_query/parse_error.rb +1 -1
- data/lib/pg_query/version.rb +1 -1
- metadata +35 -4
checksums.yaml
CHANGED
@@ -1,7 +1,7 @@
|
|
1
1
|
---
|
2
2
|
SHA1:
|
3
|
-
metadata.gz:
|
4
|
-
data.tar.gz:
|
3
|
+
metadata.gz: e4626094955c34766049a41a21623847e80cb0d8
|
4
|
+
data.tar.gz: ea9c73a5e6e410f14dd3baa3971dbb87046efab3
|
5
5
|
SHA512:
|
6
|
-
metadata.gz:
|
7
|
-
data.tar.gz:
|
6
|
+
metadata.gz: eb545f475347e8add8714fd95b9c5022d79c24296c16c73a7990e1742704de0778d713faa47ec0cb34de609ffa7e05a8a60e0fe3357910034e9d3ec633bc50b2
|
7
|
+
data.tar.gz: c4981b0e1107bf5ce207376207c8d7b90244536cc480b2d0a7a3c2f311938cb7bf95bfaf7eae171e4448993da335fbd65984dadc37af0543c2970b5178d98f9c
|
data/Rakefile
CHANGED
@@ -1,20 +1,23 @@
|
|
1
|
-
require
|
2
|
-
require
|
3
|
-
require
|
1
|
+
require 'bundler/gem_tasks'
|
2
|
+
require 'rake/extensiontask'
|
3
|
+
require 'rspec/core/rake_task'
|
4
|
+
require 'rubocop/rake_task'
|
4
5
|
|
5
|
-
Rake::ExtensionTask.new
|
6
|
-
ext.lib_dir =
|
6
|
+
Rake::ExtensionTask.new 'pg_query' do |ext|
|
7
|
+
ext.lib_dir = 'lib/pg_query'
|
7
8
|
end
|
8
9
|
|
9
10
|
RSpec::Core::RakeTask.new
|
11
|
+
RuboCop::RakeTask.new
|
10
12
|
|
11
13
|
task spec: :compile
|
12
14
|
|
13
|
-
task default: :spec
|
15
|
+
task default: [:lint, :spec]
|
14
16
|
task test: :spec
|
17
|
+
task lint: :rubocop
|
15
18
|
|
16
19
|
task :clean do
|
17
|
-
FileUtils.rm_rf File.join(File.dirname(__FILE__),
|
18
|
-
FileUtils.rm_f Dir.glob(File.join(File.dirname(__FILE__),
|
19
|
-
FileUtils.rm_f File.join(File.dirname(__FILE__),
|
20
|
-
end
|
20
|
+
FileUtils.rm_rf File.join(File.dirname(__FILE__), 'tmp/')
|
21
|
+
FileUtils.rm_f Dir.glob(File.join(File.dirname(__FILE__), 'ext/pg_query/*.o'))
|
22
|
+
FileUtils.rm_f File.join(File.dirname(__FILE__), 'lib/pg_query/pg_query.bundle')
|
23
|
+
end
|
data/ext/pg_query/extconf.rb
CHANGED
@@ -1,8 +1,10 @@
|
|
1
|
+
# rubocop:disable Style/GlobalVars
|
2
|
+
|
1
3
|
require 'mkmf'
|
2
4
|
require 'open-uri'
|
3
5
|
|
4
6
|
workdir = Dir.pwd
|
5
|
-
pgdir = File.join(workdir,
|
7
|
+
pgdir = File.join(workdir, 'postgres')
|
6
8
|
|
7
9
|
# Limit the objects we build to speed up compilation times
|
8
10
|
PG_OBJS = {
|
@@ -16,43 +18,43 @@ PG_OBJS = {
|
|
16
18
|
'gram.o', 'parser.o', 'keywords.o', 'kwlookup.o', 'scansup.o'
|
17
19
|
],
|
18
20
|
'backend/nodes' => [
|
19
|
-
'nodeFuncs.o', 'makefuncs.o', 'value.o', 'list.o', 'outfuncs_json.o'
|
21
|
+
'copyfuncs.o', 'nodeFuncs.o', 'makefuncs.o', 'value.o', 'list.o', 'outfuncs_json.o'
|
20
22
|
],
|
21
23
|
'backend/lib' => ['stringinfo.o'],
|
22
24
|
'port' => ['qsort.o'],
|
23
25
|
'common' => ['psprintf.o'],
|
24
|
-
'timezone' => ['pgtz.o']
|
26
|
+
'timezone' => ['pgtz.o']
|
25
27
|
}
|
26
28
|
|
27
29
|
# Download & compile PostgreSQL if we don't have it yet
|
28
30
|
#
|
29
31
|
# Note: We intentionally use a patched version that fixes bugs in outfuncs.c
|
30
|
-
|
31
|
-
unless File.
|
32
|
-
File.open("#{workdir}/postgres.tar.gz",
|
33
|
-
open(
|
32
|
+
unless Dir.exist?(pgdir)
|
33
|
+
unless File.exist?("#{workdir}/postgres.tar.gz")
|
34
|
+
File.open("#{workdir}/postgres.tar.gz", 'wb') do |target_file|
|
35
|
+
open('https://codeload.github.com/pganalyze/postgres/tar.gz/pg_query', 'rb') do |read_file|
|
34
36
|
target_file.write(read_file.read)
|
35
37
|
end
|
36
38
|
end
|
37
39
|
end
|
38
|
-
system("tar -xf #{workdir}/postgres.tar.gz") ||
|
39
|
-
system("mv #{workdir}/postgres-pg_query #{pgdir}") ||
|
40
|
-
system("cd #{pgdir}; CFLAGS=-fPIC ./configure -q") ||
|
40
|
+
system("tar -xf #{workdir}/postgres.tar.gz") || fail('ERROR')
|
41
|
+
system("mv #{workdir}/postgres-pg_query #{pgdir}") || fail('ERROR')
|
42
|
+
system("cd #{pgdir}; CFLAGS=-fPIC ./configure -q") || fail('ERROR')
|
41
43
|
system("cd #{pgdir}; make -C src/backend lib-recursive") # Ensures headers are generated
|
42
44
|
PG_OBJS.each do |directory, objs|
|
43
|
-
system("cd #{pgdir}; make -C src/#{directory} #{objs.join(' ')}")
|
45
|
+
system("cd #{pgdir}; make -C src/#{directory} #{objs.join(' ')}") || fail('ERROR')
|
44
46
|
end
|
45
47
|
end
|
46
48
|
|
47
49
|
$objs = PG_OBJS.map { |directory, objs| objs.map { |obj| "#{pgdir}/src/#{directory}/#{obj}" } }.flatten
|
48
|
-
$objs +=
|
50
|
+
$objs += %w(pg_query.o pg_query_parse.o pg_query_normalize.o pg_polyfills.o)
|
49
51
|
|
50
52
|
$CFLAGS << " -I #{pgdir}/src/include"
|
51
53
|
|
52
54
|
# Similar to those used by PostgreSQL
|
53
|
-
$CFLAGS <<
|
55
|
+
$CFLAGS << ' -O2 -Wall -Wmissing-prototypes -Wpointer-arith -Wdeclaration-after-statement -Wendif-labels -Wmissing-format-attribute -Wformat-security -fno-strict-aliasing -fwrapv'
|
54
56
|
|
55
|
-
SYMFILE = File.join(File.dirname(__FILE__),
|
57
|
+
SYMFILE = File.join(File.dirname(__FILE__), 'pg_query.sym')
|
56
58
|
if RUBY_PLATFORM =~ /darwin/
|
57
59
|
$DLDFLAGS << " -Wl,-exported_symbols_list #{SYMFILE}" unless defined?(::Rubinius)
|
58
60
|
else
|
data/ext/pg_query/pg_query.c
CHANGED
@@ -1,478 +1,15 @@
|
|
1
|
-
#include "
|
2
|
-
#include "utils/memutils.h"
|
3
|
-
#include "parser/parser.h"
|
4
|
-
#include "parser/scanner.h"
|
5
|
-
#include "parser/scansup.h"
|
6
|
-
#include "nodes/print.h"
|
7
|
-
#include "nodes/nodeFuncs.h"
|
8
|
-
#include "mb/pg_wchar.h"
|
9
|
-
|
10
|
-
#include <unistd.h>
|
11
|
-
#include <fcntl.h>
|
12
|
-
|
13
|
-
#include <ruby.h>
|
1
|
+
#include "pg_query.h"
|
14
2
|
|
15
3
|
const char* progname = "pg_query";
|
16
4
|
|
17
|
-
void Init_pg_query(void);
|
18
|
-
|
19
|
-
static VALUE new_parse_error(ErrorData* error)
|
20
|
-
{
|
21
|
-
VALUE cPgQuery, cParseError;
|
22
|
-
VALUE args[2];
|
23
|
-
|
24
|
-
cPgQuery = rb_const_get(rb_cObject, rb_intern("PgQuery"));
|
25
|
-
cParseError = rb_const_get_at(cPgQuery, rb_intern("ParseError"));
|
26
|
-
|
27
|
-
args[0] = rb_str_new2(error->message);
|
28
|
-
args[1] = INT2NUM(error->cursorpos);
|
29
|
-
|
30
|
-
return rb_class_new_instance(2, args, cParseError);
|
31
|
-
}
|
32
|
-
|
33
|
-
#define STDERR_BUFFER_LEN 4096
|
34
|
-
//#define DEBUG
|
35
|
-
|
36
|
-
static VALUE pg_query_raw_parse(VALUE self, VALUE input)
|
37
|
-
{
|
38
|
-
Check_Type(input, T_STRING);
|
39
|
-
|
40
|
-
MemoryContext ctx = NULL;
|
41
|
-
VALUE result = Qnil;
|
42
|
-
VALUE error = Qnil;
|
43
|
-
char stderr_buffer[STDERR_BUFFER_LEN + 1] = {0};
|
44
|
-
#ifndef DEBUG
|
45
|
-
int stderr_global;
|
46
|
-
int stderr_pipe[2];
|
47
|
-
#endif
|
48
|
-
|
49
|
-
ctx = AllocSetContextCreate(TopMemoryContext,
|
50
|
-
"pg_query_raw_parse",
|
51
|
-
ALLOCSET_DEFAULT_MINSIZE,
|
52
|
-
ALLOCSET_DEFAULT_INITSIZE,
|
53
|
-
ALLOCSET_DEFAULT_MAXSIZE);
|
54
|
-
MemoryContextSwitchTo(ctx);
|
55
|
-
|
56
|
-
#ifndef DEBUG
|
57
|
-
// Setup pipe for stderr redirection
|
58
|
-
if (pipe(stderr_pipe) != 0)
|
59
|
-
rb_raise(rb_eIOError, "Failed to open pipe, too many open file descriptors");
|
60
|
-
|
61
|
-
fcntl(stderr_pipe[0], F_SETFL, fcntl(stderr_pipe[0], F_GETFL) | O_NONBLOCK);
|
62
|
-
|
63
|
-
// Redirect stderr to the pipe
|
64
|
-
stderr_global = dup(STDERR_FILENO);
|
65
|
-
dup2(stderr_pipe[1], STDERR_FILENO);
|
66
|
-
close(stderr_pipe[1]);
|
67
|
-
#endif
|
68
|
-
|
69
|
-
// Parse it!
|
70
|
-
PG_TRY();
|
71
|
-
{
|
72
|
-
List *tree;
|
73
|
-
char *str;
|
74
|
-
|
75
|
-
str = StringValueCStr(input);
|
76
|
-
tree = raw_parser(str);
|
77
|
-
|
78
|
-
str = nodeToJSONString(tree);
|
79
|
-
|
80
|
-
#ifndef DEBUG
|
81
|
-
// Save stderr for result
|
82
|
-
read(stderr_pipe[0], stderr_buffer, STDERR_BUFFER_LEN);
|
83
|
-
#endif
|
84
|
-
|
85
|
-
result = rb_ary_new();
|
86
|
-
rb_ary_push(result, rb_str_new2(str));
|
87
|
-
rb_ary_push(result, rb_str_new2(stderr_buffer));
|
88
|
-
|
89
|
-
pfree(str);
|
90
|
-
}
|
91
|
-
PG_CATCH();
|
92
|
-
{
|
93
|
-
ErrorData* error_data = CopyErrorData();
|
94
|
-
error = new_parse_error(error_data);
|
95
|
-
FlushErrorState();
|
96
|
-
}
|
97
|
-
PG_END_TRY();
|
98
|
-
|
99
|
-
#ifndef DEBUG
|
100
|
-
// Restore stderr, close pipe
|
101
|
-
dup2(stderr_global, STDERR_FILENO);
|
102
|
-
close(stderr_pipe[0]);
|
103
|
-
close(stderr_global);
|
104
|
-
#endif
|
105
|
-
|
106
|
-
// Return to previous PostgreSQL memory context
|
107
|
-
MemoryContextSwitchTo(TopMemoryContext);
|
108
|
-
MemoryContextDelete(ctx);
|
109
|
-
|
110
|
-
// If we got an error, throw it
|
111
|
-
if (!NIL_P(error)) rb_exc_raise(error);
|
112
|
-
|
113
|
-
return result;
|
114
|
-
}
|
115
|
-
|
116
|
-
/*
|
117
|
-
* Struct for tracking locations/lengths of constants during normalization
|
118
|
-
*/
|
119
|
-
typedef struct pgssLocationLen
|
120
|
-
{
|
121
|
-
int location; /* start offset in query text */
|
122
|
-
int length; /* length in bytes, or -1 to ignore */
|
123
|
-
} pgssLocationLen;
|
124
|
-
|
125
|
-
/*
|
126
|
-
* Working state for constant tree walker
|
127
|
-
*/
|
128
|
-
typedef struct pgssConstLocations
|
129
|
-
{
|
130
|
-
/* Array of locations of constants that should be removed */
|
131
|
-
pgssLocationLen *clocations;
|
132
|
-
|
133
|
-
/* Allocated length of clocations array */
|
134
|
-
int clocations_buf_size;
|
135
|
-
|
136
|
-
/* Current number of valid entries in clocations array */
|
137
|
-
int clocations_count;
|
138
|
-
} pgssConstLocations;
|
139
|
-
|
140
|
-
/*
|
141
|
-
* comp_location: comparator for qsorting pgssLocationLen structs by location
|
142
|
-
*/
|
143
|
-
static int
|
144
|
-
comp_location(const void *a, const void *b)
|
145
|
-
{
|
146
|
-
int l = ((const pgssLocationLen *) a)->location;
|
147
|
-
int r = ((const pgssLocationLen *) b)->location;
|
148
|
-
|
149
|
-
if (l < r)
|
150
|
-
return -1;
|
151
|
-
else if (l > r)
|
152
|
-
return +1;
|
153
|
-
else
|
154
|
-
return 0;
|
155
|
-
}
|
156
|
-
|
157
|
-
/*
|
158
|
-
* Given a valid SQL string and an array of constant-location records,
|
159
|
-
* fill in the textual lengths of those constants.
|
160
|
-
*
|
161
|
-
* The constants may use any allowed constant syntax, such as float literals,
|
162
|
-
* bit-strings, single-quoted strings and dollar-quoted strings. This is
|
163
|
-
* accomplished by using the public API for the core scanner.
|
164
|
-
*
|
165
|
-
* It is the caller's job to ensure that the string is a valid SQL statement
|
166
|
-
* with constants at the indicated locations. Since in practice the string
|
167
|
-
* has already been parsed, and the locations that the caller provides will
|
168
|
-
* have originated from within the authoritative parser, this should not be
|
169
|
-
* a problem.
|
170
|
-
*
|
171
|
-
* Duplicate constant pointers are possible, and will have their lengths
|
172
|
-
* marked as '-1', so that they are later ignored. (Actually, we assume the
|
173
|
-
* lengths were initialized as -1 to start with, and don't change them here.)
|
174
|
-
*
|
175
|
-
* N.B. There is an assumption that a '-' character at a Const location begins
|
176
|
-
* a negative numeric constant. This precludes there ever being another
|
177
|
-
* reason for a constant to start with a '-'.
|
178
|
-
*/
|
179
|
-
static void
|
180
|
-
fill_in_constant_lengths(pgssConstLocations *jstate, const char *query)
|
181
|
-
{
|
182
|
-
pgssLocationLen *locs;
|
183
|
-
core_yyscan_t yyscanner;
|
184
|
-
core_yy_extra_type yyextra;
|
185
|
-
core_YYSTYPE yylval;
|
186
|
-
YYLTYPE yylloc;
|
187
|
-
int last_loc = -1;
|
188
|
-
int i;
|
189
|
-
|
190
|
-
/*
|
191
|
-
* Sort the records by location so that we can process them in order while
|
192
|
-
* scanning the query text.
|
193
|
-
*/
|
194
|
-
if (jstate->clocations_count > 1)
|
195
|
-
qsort(jstate->clocations, jstate->clocations_count,
|
196
|
-
sizeof(pgssLocationLen), comp_location);
|
197
|
-
locs = jstate->clocations;
|
198
|
-
|
199
|
-
/* initialize the flex scanner --- should match raw_parser() */
|
200
|
-
yyscanner = scanner_init(query,
|
201
|
-
&yyextra,
|
202
|
-
ScanKeywords,
|
203
|
-
NumScanKeywords);
|
204
|
-
|
205
|
-
/* Search for each constant, in sequence */
|
206
|
-
for (i = 0; i < jstate->clocations_count; i++)
|
207
|
-
{
|
208
|
-
int loc = locs[i].location;
|
209
|
-
int tok;
|
210
|
-
|
211
|
-
Assert(loc >= 0);
|
212
|
-
|
213
|
-
if (loc <= last_loc)
|
214
|
-
continue; /* Duplicate constant, ignore */
|
215
|
-
|
216
|
-
/* Lex tokens until we find the desired constant */
|
217
|
-
for (;;)
|
218
|
-
{
|
219
|
-
tok = core_yylex(&yylval, &yylloc, yyscanner);
|
220
|
-
|
221
|
-
/* We should not hit end-of-string, but if we do, behave sanely */
|
222
|
-
if (tok == 0)
|
223
|
-
break; /* out of inner for-loop */
|
224
|
-
|
225
|
-
/*
|
226
|
-
* We should find the token position exactly, but if we somehow
|
227
|
-
* run past it, work with that.
|
228
|
-
*/
|
229
|
-
if (yylloc >= loc)
|
230
|
-
{
|
231
|
-
if (query[loc] == '-')
|
232
|
-
{
|
233
|
-
/*
|
234
|
-
* It's a negative value - this is the one and only case
|
235
|
-
* where we replace more than a single token.
|
236
|
-
*
|
237
|
-
* Do not compensate for the core system's special-case
|
238
|
-
* adjustment of location to that of the leading '-'
|
239
|
-
* operator in the event of a negative constant. It is
|
240
|
-
* also useful for our purposes to start from the minus
|
241
|
-
* symbol. In this way, queries like "select * from foo
|
242
|
-
* where bar = 1" and "select * from foo where bar = -2"
|
243
|
-
* will have identical normalized query strings.
|
244
|
-
*/
|
245
|
-
tok = core_yylex(&yylval, &yylloc, yyscanner);
|
246
|
-
if (tok == 0)
|
247
|
-
break; /* out of inner for-loop */
|
248
|
-
}
|
249
|
-
|
250
|
-
/*
|
251
|
-
* We now rely on the assumption that flex has placed a zero
|
252
|
-
* byte after the text of the current token in scanbuf.
|
253
|
-
*/
|
254
|
-
locs[i].length = (int) strlen(yyextra.scanbuf + loc);
|
255
|
-
|
256
|
-
/* Quoted string with Unicode escapes
|
257
|
-
*
|
258
|
-
* The lexer consumes trailing whitespace in order to find UESCAPE, but if there
|
259
|
-
* is no UESCAPE it has still consumed it - don't include it in constant length.
|
260
|
-
*/
|
261
|
-
if (locs[i].length > 4 && /* U&'' */
|
262
|
-
(yyextra.scanbuf[loc] == 'u' || yyextra.scanbuf[loc] == 'U') &&
|
263
|
-
yyextra.scanbuf[loc + 1] == '&' && yyextra.scanbuf[loc + 2] == '\'')
|
264
|
-
{
|
265
|
-
int j = locs[i].length - 1; /* Skip the \0 */
|
266
|
-
for (; j >= 0 && scanner_isspace(yyextra.scanbuf[loc + j]); j--) {}
|
267
|
-
locs[i].length = j + 1; /* Count the \0 */
|
268
|
-
}
|
269
|
-
|
270
|
-
break; /* out of inner for-loop */
|
271
|
-
}
|
272
|
-
}
|
273
|
-
|
274
|
-
/* If we hit end-of-string, give up, leaving remaining lengths -1 */
|
275
|
-
if (tok == 0)
|
276
|
-
break;
|
277
|
-
|
278
|
-
last_loc = loc;
|
279
|
-
}
|
280
|
-
|
281
|
-
scanner_finish(yyscanner);
|
282
|
-
}
|
283
|
-
|
284
|
-
/*
|
285
|
-
* Generate a normalized version of the query string that will be used to
|
286
|
-
* represent all similar queries.
|
287
|
-
*
|
288
|
-
* Note that the normalized representation may well vary depending on
|
289
|
-
* just which "equivalent" query is used to create the hashtable entry.
|
290
|
-
* We assume this is OK.
|
291
|
-
*
|
292
|
-
* *query_len_p contains the input string length, and is updated with
|
293
|
-
* the result string length (which cannot be longer) on exit.
|
294
|
-
*
|
295
|
-
* Returns a palloc'd string.
|
296
|
-
*/
|
297
|
-
static char *
|
298
|
-
generate_normalized_query(pgssConstLocations *jstate, const char *query,
|
299
|
-
int *query_len_p, int encoding)
|
300
|
-
{
|
301
|
-
char *norm_query;
|
302
|
-
int query_len = *query_len_p;
|
303
|
-
int i,
|
304
|
-
len_to_wrt, /* Length (in bytes) to write */
|
305
|
-
quer_loc = 0, /* Source query byte location */
|
306
|
-
n_quer_loc = 0, /* Normalized query byte location */
|
307
|
-
last_off = 0, /* Offset from start for previous tok */
|
308
|
-
last_tok_len = 0; /* Length (in bytes) of that tok */
|
309
|
-
|
310
|
-
/*
|
311
|
-
* Get constants' lengths (core system only gives us locations). Note
|
312
|
-
* this also ensures the items are sorted by location.
|
313
|
-
*/
|
314
|
-
fill_in_constant_lengths(jstate, query);
|
315
|
-
|
316
|
-
/* Allocate result buffer */
|
317
|
-
norm_query = palloc(query_len + 1);
|
318
|
-
|
319
|
-
for (i = 0; i < jstate->clocations_count; i++)
|
320
|
-
{
|
321
|
-
int off, /* Offset from start for cur tok */
|
322
|
-
tok_len; /* Length (in bytes) of that tok */
|
323
|
-
|
324
|
-
off = jstate->clocations[i].location;
|
325
|
-
tok_len = jstate->clocations[i].length;
|
326
|
-
|
327
|
-
if (tok_len < 0)
|
328
|
-
continue; /* ignore any duplicates */
|
329
|
-
|
330
|
-
/* Copy next chunk (what precedes the next constant) */
|
331
|
-
len_to_wrt = off - last_off;
|
332
|
-
len_to_wrt -= last_tok_len;
|
333
|
-
|
334
|
-
Assert(len_to_wrt >= 0);
|
335
|
-
memcpy(norm_query + n_quer_loc, query + quer_loc, len_to_wrt);
|
336
|
-
n_quer_loc += len_to_wrt;
|
337
|
-
|
338
|
-
/* And insert a '?' in place of the constant token */
|
339
|
-
norm_query[n_quer_loc++] = '?';
|
340
|
-
|
341
|
-
quer_loc = off + tok_len;
|
342
|
-
last_off = off;
|
343
|
-
last_tok_len = tok_len;
|
344
|
-
}
|
345
|
-
|
346
|
-
/*
|
347
|
-
* We've copied up until the last ignorable constant. Copy over the
|
348
|
-
* remaining bytes of the original query string.
|
349
|
-
*/
|
350
|
-
len_to_wrt = query_len - quer_loc;
|
351
|
-
|
352
|
-
Assert(len_to_wrt >= 0);
|
353
|
-
memcpy(norm_query + n_quer_loc, query + quer_loc, len_to_wrt);
|
354
|
-
n_quer_loc += len_to_wrt;
|
355
|
-
|
356
|
-
Assert(n_quer_loc <= query_len);
|
357
|
-
norm_query[n_quer_loc] = '\0';
|
358
|
-
|
359
|
-
*query_len_p = n_quer_loc;
|
360
|
-
return norm_query;
|
361
|
-
}
|
362
|
-
|
363
|
-
static bool const_record_walker(Node *node, pgssConstLocations *jstate)
|
364
|
-
{
|
365
|
-
bool result;
|
366
|
-
|
367
|
-
if (node == NULL) return false;
|
368
|
-
|
369
|
-
if (IsA(node, A_Const) && ((A_Const *) node)->location >= 0)
|
370
|
-
{
|
371
|
-
/* enlarge array if needed */
|
372
|
-
if (jstate->clocations_count >= jstate->clocations_buf_size)
|
373
|
-
{
|
374
|
-
jstate->clocations_buf_size *= 2;
|
375
|
-
jstate->clocations = (pgssLocationLen *)
|
376
|
-
repalloc(jstate->clocations,
|
377
|
-
jstate->clocations_buf_size *
|
378
|
-
sizeof(pgssLocationLen));
|
379
|
-
}
|
380
|
-
jstate->clocations[jstate->clocations_count].location = ((A_Const *) node)->location;
|
381
|
-
/* initialize lengths to -1 to simplify fill_in_constant_lengths */
|
382
|
-
jstate->clocations[jstate->clocations_count].length = -1;
|
383
|
-
jstate->clocations_count++;
|
384
|
-
}
|
385
|
-
else if (IsA(node, VariableSetStmt))
|
386
|
-
{
|
387
|
-
return const_record_walker((Node *) ((VariableSetStmt *) node)->args, jstate);
|
388
|
-
}
|
389
|
-
else if (IsA(node, CopyStmt))
|
390
|
-
{
|
391
|
-
return const_record_walker((Node *) ((CopyStmt *) node)->query, jstate);
|
392
|
-
}
|
393
|
-
|
394
|
-
PG_TRY();
|
395
|
-
{
|
396
|
-
result = raw_expression_tree_walker(node, const_record_walker, (void*) jstate);
|
397
|
-
}
|
398
|
-
PG_CATCH();
|
399
|
-
{
|
400
|
-
FlushErrorState();
|
401
|
-
result = false;
|
402
|
-
}
|
403
|
-
PG_END_TRY();
|
404
|
-
|
405
|
-
return result;
|
406
|
-
}
|
407
|
-
|
408
|
-
static VALUE pg_query_normalize(VALUE self, VALUE input)
|
409
|
-
{
|
410
|
-
Check_Type(input, T_STRING);
|
411
|
-
|
412
|
-
MemoryContext ctx = NULL;
|
413
|
-
VALUE result = Qnil;
|
414
|
-
VALUE error = Qnil;
|
415
|
-
|
416
|
-
ctx = AllocSetContextCreate(TopMemoryContext,
|
417
|
-
"pg_query_normalize",
|
418
|
-
ALLOCSET_DEFAULT_MINSIZE,
|
419
|
-
ALLOCSET_DEFAULT_INITSIZE,
|
420
|
-
ALLOCSET_DEFAULT_MAXSIZE);
|
421
|
-
MemoryContextSwitchTo(ctx);
|
422
|
-
|
423
|
-
PG_TRY();
|
424
|
-
{
|
425
|
-
List *tree;
|
426
|
-
char *str;
|
427
|
-
pgssConstLocations jstate;
|
428
|
-
int query_len;
|
429
|
-
|
430
|
-
/* Parse query */
|
431
|
-
str = StringValueCStr(input);
|
432
|
-
tree = raw_parser(str);
|
433
|
-
|
434
|
-
/* Set up workspace for constant recording */
|
435
|
-
jstate.clocations_buf_size = 32;
|
436
|
-
jstate.clocations = (pgssLocationLen *)
|
437
|
-
palloc(jstate.clocations_buf_size * sizeof(pgssLocationLen));
|
438
|
-
jstate.clocations_count = 0;
|
439
|
-
|
440
|
-
/* Walk tree and record const locations */
|
441
|
-
const_record_walker((Node *) tree, &jstate);
|
442
|
-
|
443
|
-
/* Normalize query */
|
444
|
-
query_len = (int) strlen(str);
|
445
|
-
str = generate_normalized_query(&jstate, str, &query_len, PG_UTF8);
|
446
|
-
|
447
|
-
result = rb_str_new2(str);
|
448
|
-
|
449
|
-
pfree(str);
|
450
|
-
}
|
451
|
-
PG_CATCH();
|
452
|
-
{
|
453
|
-
ErrorData* error_data = CopyErrorData();
|
454
|
-
error = new_parse_error(error_data);
|
455
|
-
FlushErrorState();
|
456
|
-
}
|
457
|
-
PG_END_TRY();
|
458
|
-
|
459
|
-
MemoryContextSwitchTo(TopMemoryContext);
|
460
|
-
MemoryContextDelete(ctx);
|
461
|
-
|
462
|
-
// If we got an error, throw it
|
463
|
-
if (!NIL_P(error)) rb_exc_raise(error);
|
464
|
-
|
465
|
-
return result;
|
466
|
-
}
|
467
|
-
|
468
5
|
void Init_pg_query(void)
|
469
6
|
{
|
470
7
|
VALUE cPgQuery;
|
471
|
-
|
8
|
+
|
472
9
|
MemoryContextInit();
|
473
10
|
|
474
11
|
cPgQuery = rb_const_get(rb_cObject, rb_intern("PgQuery"));
|
475
12
|
|
476
13
|
rb_define_singleton_method(cPgQuery, "_raw_parse", pg_query_raw_parse, 1);
|
477
14
|
rb_define_singleton_method(cPgQuery, "normalize", pg_query_normalize, 1);
|
478
|
-
}
|
15
|
+
}
|