From: Kirill Yukhin <kyukhin@tarantool.org>
To: Vladislav Shpilevoy <v.shpilevoy@tarantool.org>
Cc: tarantool-patches@freelists.org
Subject: [tarantool-patches] Re: [PATCH] sql: allow any space symbols to be a white space
Date: Wed, 23 May 2018 08:15:00 +0300 [thread overview]
Message-ID: <20180523051500.3fx5uoetyzsjkwp5@tarantool.org> (raw)
In-Reply-To: <e69e8e43-c483-8800-60b2-9f599bc2a397@tarantool.org>
Hello Vlad,
Thanks a lot for your exctensive review.
Updated patch (still faiuling on Travis) is in the bottom.
On 22 мая 21:06, Vladislav Shpilevoy wrote:
> Hello. Thanks for the patch! At first, no one test is passed on my
> machine. I see infinite errors like this:
>
> [012] Worker "012_box" cannot start tarantool server; the tasks will be ignored...
> [012] The raised exception is '' of type '<class 'lib.tarantool_server.TarantoolStartError'>'.
> [012] Worker "012_box" received the following error:
> [012] Traceback (most recent call last):
> [012] File "/Users/v.shpilevoy/Work/Repositories/tarantool/test-run/lib/worker.py", line 211, in __init__
> [012] self.inspector = suite.start_server(self.server)
> [012] File "/Users/v.shpilevoy/Work/Repositories/tarantool/test-run/lib/test_suite.py", line 158, in start_server
> [012] server.deploy(silent=False)
> [012] File "/Users/v.shpilevoy/Work/Repositories/tarantool/test-run/lib/tarantool_server.py", line 510, in deploy
> [012] self.start(silent=silent, **kwargs)
> [012] File "/Users/v.shpilevoy/Work/Repositories/tarantool/test-run/lib/tarantool_server.py", line 578, in start
> [012] self.wait_until_started(wait_load)
> [012] File "/Users/v.shpilevoy/Work/Repositories/tarantool/test-run/lib/tarantool_server.py", line 745, in wait_until_started
> [012] msg, self.process if not self.gdb and not self.lldb else None)
> [012] File "/Users/v.shpilevoy/Work/Repositories/tarantool/test-run/lib/tarantool_server.py", line 211, in seek_wait
> [012] raise TarantoolStartError
> [012] TarantoolStartError
> [012]
>
> On Travis I see the same.
I have no idea what the hell is that. I suspect that there's an issue w/ linking of ICU.
Will investigate.
> At second, see 17 comments below. Most of them are about style, but a pair
> of functional ones too.
>
> On 22/05/2018 18:51, Kirill Yukhin wrote:
> > Branch: https://github.com/tarantool/tarantool/tree/kyukhin/gh-2371-utf8-spaces
> > Issue: https://github.com/tarantool/tarantool/issues/2371
> >
> > ANSI SQL allows any of Unicode classes ZI, Zp or Zs to
> > act as white space symbol. Allow this in lexical analyzer.
> > Refactor lexical analyzer routine to follow Tarantool's
> > coding style.
> > Also, remove dead encoding: ABCDIC.
> 1. Maybe EBCDIC?
Sure.
> > diff --git a/src/box/sql/alter.c b/src/box/sql/alter.c
> > index c9c8f9b..f509d4e 100644
> > --- a/src/box/sql/alter.c
> > +++ b/src/box/sql/alter.c
> > @@ -351,7 +351,7 @@ rename_table(sqlite3 *db, const char *sql_stmt, const char *table_name,
> > int token;
> > Token old_name;
> > - unsigned char const *csr = (unsigned const char *)sql_stmt;
> > + char const *csr = sql_stmt;
> 2. const char *
Fixed.
> > diff --git a/src/box/sql/complete.c b/src/box/sql/complete.c
> > index 092d4fb..047e09e 100644
> > --- a/src/box/sql/complete.c
> > +++ b/src/box/sql/complete.c
> > @@ -40,19 +40,6 @@
> > #include "sqliteInt.h"
> > #ifndef SQLITE_OMIT_COMPLETE
> > -/*
> > - * This is defined in tokenize.c. We just have to import the definition.
> > - */
> > -#ifndef SQLITE_AMALGAMATION
> > -#ifdef SQLITE_ASCII
> > -#define IdChar(C) ((sqlite3CtypeMap[(unsigned char)C]&0x46)!=0)
> > -#endif
> > -#ifdef SQLITE_EBCDIC
> 3. I still can grep SQLITE_EBCDIC in tarantool/extra/mkkeywordhash.c. It is not
> generated file. Same about SQLITE_ASCII.
Purged.
> > diff --git a/src/box/sql/sqliteInt.h b/src/box/sql/sqliteInt.h
> > index b3db468..f55c734 100644
> > --- a/src/box/sql/sqliteInt.h
> > +++ b/src/box/sql/sqliteInt.h
> > @@ -36,6 +36,8 @@
> > #ifndef SQLITEINT_H
> > #define SQLITEINT_H
> > +#define IdChar(C) ((sqlite3CtypeMap[(unsigned char)C]&0x46)!=0)
> 4. Why can not you leave it in sql/complete.c?
Since it is used in tokenize.c as well. I'd avoid such a duplicates.
> > @@ -4164,7 +4136,18 @@ extern int sqlite3PendingByte;
> > #endif
> > void sqlite3Reindex(Parse *, Token *, Token *);
> > void sqlite3AlterRenameTable(Parse *, SrcList *, Token *);
> > -int sqlite3GetToken(const unsigned char *, int *, bool *);
> > +
> > +/**
> > + * Return the length (in bytes) of the token that begins at z[0].
> > + * Store the token type in *tokenType before returning.
> 5. No tokenType.
Fixed.
> > --- a/src/box/sql/tokenize.c
> > +++ b/src/box/sql/tokenize.c
> > @@ -77,10 +81,9 @@
> > #define CC_DOT 26 /* '.' */
> > #define CC_ILLEGAL 27 /* Illegal character */
> > -static const unsigned char aiClass[] = {
> > -#ifdef SQLITE_ASCII
> > +static const char sql_ascii_class[] = {
> > /* x0 x1 x2 x3 x4 x5 x6 x7 x8 x9 xa xb xc xd xe xf */
> > -/* 0x */ 27, 27, 27, 27, 27, 27, 27, 27, 27, 7, 7, 27, 7, 7, 27, 27,
> > +/* 0x */ 27, 27, 27, 27, 27, 27, 27, 27, 27, 7, 7, 7, 7, 7, 27, 27,
> 6. Why did you change 27 to 7 in xb column?
Since according to Issue this symbol should be classified as space as well.
> > };
> > -/*
> > +/**
> > * The charMap() macro maps alphabetic characters (only) into their
> 7. Out of 66.
Fixed.
> > - * For ASCII, any character with the high-order bit set is
> > + * Any character with the high-order bit set is
> > * allowed in an identifier. For 7-bit characters,
> > * sqlite3IsIdChar[X] must be 1.
> 8. No sqlite3IsIdChar.
Fixed.
>
> 9. Now I see two identical implementations of this macro:
> sql_id_char here and IdChar in sqliteInt.h. And sql_id_char looks
> to be unused.
Removed.
> > *
> > - * For EBCDIC, the rules are more complex but have the same
> > - * end result.
> > - *
> > - * Ticket #1066. the SQL standard does not allow '$' in the
> > + * The SQL standard does not allow '$' in the
> > * middle of identifiers. But many SQL implementations do.
> > * SQLite will allow '$' in identifiers for compatibility.
> > * But the feature is undocumented.
> > + *
> 10. Trailing white space.
Removed.
> > + * @param z Input stream.
> > + * @retval True if current symbl space.
> 11. symbol.
Fixed.
> > + */
> > +static inline bool
> > +sql_is_space_char(const char *z)
> > +{
> > + if (sqlite3Isspace(z[0]))
> > + return true;
> > + if (sqlite3CtypeMap[*(unsigned char*)z] & 0x40) {
> 12. Why 0x40?
0x40 stands for possible urf-8 sequence, please consult comment on
sqlite3CtypeMap definition in sql/global.c
> > +static inline int
> > +sql_skip_spaces(const char *z)
> > +{
> > + int idx = 0;
> > + while (true) {
> > + if (sqlite3Isspace(z[idx])) {
> > + idx += 1;
> > + } else if ((sqlite3CtypeMap[*(unsigned char *)(z + idx)] & 0x40)
> 13. Already second place with 0x40. What is it? Maybe you should create a separate
> macro like it is done here?
> > + != 0) {
> > + UChar32 c;
> > + int prev_offset = idx;
> > + U8_NEXT(z, idx, INT32_MAX, c);
>
> 14. If you do not check for errors and don't know a length,
> you can use U8_NEXT_UNSAFE. It is much shorter. Same about U8_NEXT
> in sql_is_space_char I think.
Done.
> > + if (!u_isspace(c)) {
> > + idx = prev_offset;
>
> 15. In most cases you will have one white space, so maybe it is faster
> to make this:
> int new_offset = idx;
> U8_NEXT(z, new_offset, INT32_MAX, c);
> if (! u_isspace(c))
> break;
> idx = new_offset;
> Here you have not to reset the offset in most cases.
Good idea. Done.
> > int
> > -sqlite3GetToken(const unsigned char *z, int *tokenType, bool *is_reserved)
> > +sql_token(const char *z, int *type, bool *is_reserved)
>
> 16. Please, consider my light refactoring.
>
> diff --git a/src/box/sql/tokenize.c b/src/box/sql/tokenize.c
> index 8df58e949..4cdd375ee 100644
> --- a/src/box/sql/tokenize.c
> +++ b/src/box/sql/tokenize.c
> @@ -209,8 +209,7 @@ sql_token(const char *z, int *type, bool *is_reserved)
> return i;
> case CC_MINUS:
> if (z[1] == '-') {
> - for (i = 2; (c = z[i]) != 0 && c != '\n'; i++) {
> - }
> + for (i = 2; (c = z[i]) != 0 && c != '\n'; i++);
I believe that is evil.
> @@ -237,10 +236,8 @@ sql_token(const char *z, int *type, bool *is_reserved)
> return 1;
> }
> for (i = 3, c = z[2];
> - (c != '*' || z[i] != '/') && (c = z[i]) != 0;
> - i++) {
> - }
> - if (c)
> + (c != '*' || z[i] != '/') && (c = z[i]) != 0; i++);
> + if (c != 0)
Ditto.
> @@ -323,8 +320,7 @@ sql_token(const char *z, int *type, bool *is_reserved)
> }
> FALLTHROUGH;
> case CC_DOT:
> - if (!sqlite3Isdigit(z[1]))
> - {
> + if (! sqlite3Isdigit(z[1])) {
Picked.
> @@ -335,38 +331,32 @@ sql_token(const char *z, int *type, bool *is_reserved)
> FALLTHROUGH;
> case CC_DIGIT:
> *type = TK_INTEGER;
> - if (z[0] == '0' && (z[1] == 'x' || z[1] == 'X')
> - && sqlite3Isxdigit(z[2])) {
> - for (i = 3; sqlite3Isxdigit(z[i]); i++) {
> - }
> + if (z[0] == '0' && (z[1] == 'x' || z[1] == 'X') &&
> + sqlite3Isxdigit(z[2])) {
> + for (i = 3; sqlite3Isxdigit(z[i]); i++);
Partially picked.
> return i;
> }
> - for (i = 0; sqlite3Isdigit(z[i]); i++) {
> - }
> + for (i = 0; sqlite3Isdigit(z[i]); i++);
Evil.
> if (z[i] == '.') {
> - i++;
> - while (sqlite3Isdigit(z[i]))
> - i++;
> + while (sqlite3Isdigit(z[++i]));
Picked.
> *type = TK_FLOAT;
> }
> if ((z[i] == 'e' || z[i] == 'E') &&
> (sqlite3Isdigit(z[i + 1])
> - || ((z[i + 1] == '+' || z[i + 1] == '-')
> - && sqlite3Isdigit(z[i + 2])))) {
> - i += 2;
> - while (sqlite3Isdigit(z[i]))
> - i++;
> + || ((z[i + 1] == '+' || z[i + 1] == '-') &&
> + sqlite3Isdigit(z[i + 2])))) {
> + ++i;
> + while (sqlite3Isdigit(z[++i]));
IMHO, refatoring a loop in such a way makes him less readable and
might reduce performance, since you are doing addtion twice at least.
> - while (IdChar(z[i])) {
> + if (IdChar(z[i])) {
> *type = TK_ILLEGAL;
> - i++;
> + while(IdChar(z[++i]));
> }
Picked.
> return i;
> case CC_VARNUM:
> *type = TK_VARIABLE;
> - for (i = 1; sqlite3Isdigit(z[i]); i++) {
> - }
> + for (i = 1; sqlite3Isdigit(z[i]); i++);
Evil.
> @@ -383,9 +373,8 @@ sql_token(const char *z, int *type, bool *is_reserved)
> return i;
> case CC_KYWD:
> for (i = 1; sql_ascii_class[*(unsigned char*)(z+i)] <= CC_KYWD;
> - i++) {
> - }
> - if (!sql_is_space_char((const char *)z + i) && IdChar(z[i])) {
> + i++);
> + if (!sql_is_space_char(z + i) && IdChar(z[i])) {
Partially picked.
> @@ -395,18 +384,17 @@ sql_token(const char *z, int *type, bool *is_reserved)
> break;
> }
> *type = TK_ID;
> - return keywordCode((char *)z, i, type, is_reserved);
> + return keywordCode(z, i, type, is_reserved);
Picked.
> case CC_X:
> if (z[1] == '\'') {
> *type = TK_BLOB;
> - for (i = 2; sqlite3Isxdigit(z[i]); i++) {
> - }
> + for (i = 2; sqlite3Isxdigit(z[i]); i++);
Evil.
> if (z[i] != '\'' || i % 2) {
> *type = TK_ILLEGAL;
> - while (z[i] && z[i] != '\'')
> + while (z[i] != 0 && z[i] != '\'')
Picked.
> }
> - if (z[i])
> + if (z[i] != 0)
Picked.
> 17. And the last comment. I see, that you use sql_is_space_char to
> detect if a char is a white space. But the caller code propagates the
> offset to 1 byte always. Regardless of the fact the white space symbol
> can occupy more than 1 byte.
>
> > if (!sql_is_space_char(z + i) && IdChar(z[i])) {
> > /* This token started out using characters
> > * that can appear in keywords, but z[i] is
> > * a character not allowed within keywords,
> > * so this must be an identifier instead.
> > */
> > i++;
> ^^^^^^^^^
> \|/
> here --------+
>
> > break;
> > }
> I propose to inline it and propagate 'i' on the value got from
> U8_NEXT_UNSAFE as offset.
I think here's a misunderstanding. `sql_is_space_char` is a predicate. It does
not promote anything. It's responsibility to detect if current character might
be a white space. If *it is not* a white space and it is an ID char then it must
be identifier which first letters coincide w/ some SQL keyword. Scan further in
that case, beyond switch() stmt.
Else, if it is a white space then we should stop lexer, fetch a keyword from
hash and return current token.
--
Regards, Kirill Yukhin
commit 80a1056f7d939a90223dfd65772954746b485640
Author: Kirill Yukhin <kyukhin@tarantool.org>
Date: Tue May 22 18:45:35 2018 +0300
sql: allow any space symbols to be a white space
ANSI SQL allows any of Unicode classes ZI, Zp or Zs to
act as white space symbol. Allow this in lexical analyzer.
Refactor lexical analyzer routine to follow Tarantool's
coding style.
Also, remove dead encoding: EBCDIC.
Closes #2371
diff --git a/extra/mkkeywordhash.c b/extra/mkkeywordhash.c
index cf34831..9e0e24b 100644
--- a/extra/mkkeywordhash.c
+++ b/extra/mkkeywordhash.c
@@ -611,12 +611,7 @@ int main(int argc, char **argv){
printf(" if( aLen[i]!=n ) continue;\n");
printf(" j = 0;\n");
printf(" zKW = &zText[aOffset[i]];\n");
- printf("#ifdef SQLITE_ASCII\n");
printf(" while( j<n && (z[j]&~0x20)==zKW[j] ){ j++; }\n");
- printf("#endif\n");
- printf("#ifdef SQLITE_EBCDIC\n");
- printf(" while( j<n && toupper(z[j])==zKW[j] ){ j++; }\n");
- printf("#endif\n");
printf(" if( j<n ) continue;\n");
for(i=0; i<nKeyword; i++){
printf(" testcase( i==%d ); /* %s */\n",
diff --git a/src/box/sql/alter.c b/src/box/sql/alter.c
index c9c8f9b..c9325c5 100644
--- a/src/box/sql/alter.c
+++ b/src/box/sql/alter.c
@@ -351,7 +351,7 @@ rename_table(sqlite3 *db, const char *sql_stmt, const char *table_name,
int token;
Token old_name;
- unsigned char const *csr = (unsigned const char *)sql_stmt;
+ const char *csr = sql_stmt;
int len = 0;
char *new_sql_stmt;
bool unused;
@@ -374,7 +374,7 @@ rename_table(sqlite3 *db, const char *sql_stmt, const char *table_name,
*/
do {
csr += len;
- len = sqlite3GetToken(csr, &token, &unused);
+ len = sql_token(csr, &token, &unused);
} while (token == TK_SPACE);
assert(len > 0);
} while (token != TK_LP && token != TK_USING);
@@ -430,13 +430,12 @@ rename_parent_table(sqlite3 *db, const char *sql_stmt, const char *old_name,
bool is_quoted;
for (csr = sql_stmt; *csr; csr = csr + n) {
- n = sqlite3GetToken((const unsigned char *)csr, &token, &unused);
+ n = sql_token(csr, &token, &unused);
if (token == TK_REFERENCES) {
char *zParent;
do {
csr += n;
- n = sqlite3GetToken((const unsigned char *)csr,
- &token, &unused);
+ n = sql_token(csr, &token, &unused);
} while (token == TK_SPACE);
if (token == TK_ILLEGAL)
break;
@@ -482,7 +481,7 @@ rename_trigger(sqlite3 *db, char const *sql_stmt, char const *table_name,
int token;
Token tname;
int dist = 3;
- unsigned char const *csr = (unsigned char const*)sql_stmt;
+ char const *csr = (char const*)sql_stmt;
int len = 0;
char *new_sql_stmt;
bool unused;
@@ -505,7 +504,7 @@ rename_trigger(sqlite3 *db, char const *sql_stmt, char const *table_name,
*/
do {
csr += len;
- len = sqlite3GetToken(csr, &token, &unused);
+ len = sql_token(csr, &token, &unused);
} while (token == TK_SPACE);
assert(len > 0);
/* Variable 'dist' stores the number of tokens read since the most
diff --git a/src/box/sql/complete.c b/src/box/sql/complete.c
index 092d4fb..74b057b 100644
--- a/src/box/sql/complete.c
+++ b/src/box/sql/complete.c
@@ -40,18 +40,7 @@
#include "sqliteInt.h"
#ifndef SQLITE_OMIT_COMPLETE
-/*
- * This is defined in tokenize.c. We just have to import the definition.
- */
-#ifndef SQLITE_AMALGAMATION
-#ifdef SQLITE_ASCII
#define IdChar(C) ((sqlite3CtypeMap[(unsigned char)C]&0x46)!=0)
-#endif
-#ifdef SQLITE_EBCDIC
-extern const char sqlite3IsEbcdicIdChar[];
-#define IdChar(C) (((c=C)>=0x42 && sqlite3IsEbcdicIdChar[c-0x40]))
-#endif
-#endif /* SQLITE_AMALGAMATION */
/*
* Token types used by the sqlite3_complete() routine. See the header
@@ -230,9 +219,6 @@ sqlite3_complete(const char *zSql)
break;
}
default:{
-#ifdef SQLITE_EBCDIC
- unsigned char c;
-#endif
if (IdChar((u8) * zSql)) {
/* Keywords and unquoted identifiers */
int nId;
diff --git a/src/box/sql/func.c b/src/box/sql/func.c
index dcac22c..c06e3bd 100644
--- a/src/box/sql/func.c
+++ b/src/box/sql/func.c
@@ -623,12 +623,7 @@ struct compareInfo {
* macro for fast reading of the next character in the common case where
* the next character is ASCII.
*/
-#if defined(SQLITE_EBCDIC)
-#define sqlite3Utf8Read(A) (*((*A)++))
-#define Utf8Read(A) (*(A++))
-#else
#define Utf8Read(s, e) ucnv_getNextUChar(pUtf8conv, &s, e, &status)
-#endif
static const struct compareInfo globInfo = { '*', '?', '[', 0 };
diff --git a/src/box/sql/global.c b/src/box/sql/global.c
index cd6f9c4..8e53bcc 100644
--- a/src/box/sql/global.c
+++ b/src/box/sql/global.c
@@ -43,7 +43,6 @@
* involved are nearly as big or bigger than SQLite itself.
*/
const unsigned char sqlite3UpperToLower[] = {
-#ifdef SQLITE_ASCII
0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17,
18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35,
36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53,
@@ -70,25 +69,6 @@ const unsigned char sqlite3UpperToLower[] = {
234, 235, 236, 237, 238, 239, 240, 241, 242, 243, 244, 245, 246, 247,
248, 249, 250, 251,
252, 253, 254, 255
-#endif
-#ifdef SQLITE_EBCDIC
- 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, /* 0x */
- 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, /* 1x */
- 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, /* 2x */
- 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, /* 3x */
- 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, /* 4x */
- 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, /* 5x */
- 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, /* 6x */
- 112, 113, 114, 115, 116, 117, 118, 119, 120, 121, 122, 123, 124, 125, 126, 127, /* 7x */
- 128, 129, 130, 131, 132, 133, 134, 135, 136, 137, 138, 139, 140, 141, 142, 143, /* 8x */
- 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, /* 9x */
- 160, 161, 162, 163, 164, 165, 166, 167, 168, 169, 170, 171, 140, 141, 142, 175, /* Ax */
- 176, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191, /* Bx */
- 192, 129, 130, 131, 132, 133, 134, 135, 136, 137, 202, 203, 204, 205, 206, 207, /* Cx */
- 208, 145, 146, 147, 148, 149, 150, 151, 152, 153, 218, 219, 220, 221, 222, 223, /* Dx */
- 224, 225, 162, 163, 164, 165, 166, 167, 168, 169, 234, 235, 236, 237, 238, 239, /* Ex */
- 240, 241, 242, 243, 244, 245, 246, 247, 248, 249, 250, 251, 252, 253, 254, 255, /* Fx */
-#endif
};
/*
@@ -119,7 +99,6 @@ const unsigned char sqlite3UpperToLower[] = {
* non-ASCII UTF character. Hence the test for whether or not a character is
* part of an identifier is 0x46.
*/
-#ifdef SQLITE_ASCII
const unsigned char sqlite3CtypeMap[256] = {
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 00..07 ........ */
0x00, 0x01, 0x01, 0x01, 0x01, 0x01, 0x00, 0x00, /* 08..0f ........ */
@@ -157,7 +136,6 @@ const unsigned char sqlite3CtypeMap[256] = {
0x40, 0x40, 0x40, 0x40, 0x40, 0x40, 0x40, 0x40, /* f0..f7 ........ */
0x40, 0x40, 0x40, 0x40, 0x40, 0x40, 0x40, 0x40 /* f8..ff ........ */
};
-#endif
/* EVIDENCE-OF: R-02982-34736 In order to maintain full backwards
* compatibility for legacy applications, the URI filename capability is
diff --git a/src/box/sql/sqliteInt.h b/src/box/sql/sqliteInt.h
index b3db468..e6ffda4 100644
--- a/src/box/sql/sqliteInt.h
+++ b/src/box/sql/sqliteInt.h
@@ -36,6 +36,8 @@
#ifndef SQLITEINT_H
#define SQLITEINT_H
+#define IdChar(C) ((sqlite3CtypeMap[(unsigned char)C]&0x46)!=0)
+
/* Special Comments:
*
* Some comments have special meaning to the tools that measure test
@@ -1129,16 +1131,6 @@ sqlite3_bind_parameter_lindex(sqlite3_stmt * pStmt, const char *zName,
#define MAX(A,B) ((A)>(B)?(A):(B))
#endif
-/*
- * Check to see if this machine uses EBCDIC. (Yes, believe it or
- * not, there are still machines out there that use EBCDIC.)
- */
-#if 'A' == '\301'
-#define SQLITE_EBCDIC 1
-#else
-#define SQLITE_ASCII 1
-#endif
-
/*
* Integers of known sizes. These typedefs might change for architectures
* where the sizes very. Preprocessor macros are available so that the
@@ -3368,21 +3360,11 @@ int sqlite3IoerrnomemError(int);
#define SQLITE_ENABLE_FTS3 1
#endif
-/*
- * The ctype.h header is needed for non-ASCII systems. It is also
- * needed by FTS3 when FTS3 is included in the amalgamation.
- */
-#if !defined(SQLITE_ASCII) || \
- (defined(SQLITE_ENABLE_FTS3) && defined(SQLITE_AMALGAMATION))
-#include <ctype.h>
-#endif
-
/*
* The following macros mimic the standard library functions toupper(),
* isspace(), isalnum(), isdigit() and isxdigit(), respectively. The
* sqlite versions only work for ASCII characters, regardless of locale.
*/
-#ifdef SQLITE_ASCII
#define sqlite3Toupper(x) ((x)&~(sqlite3CtypeMap[(unsigned char)(x)]&0x20))
#define sqlite3Isspace(x) (sqlite3CtypeMap[(unsigned char)(x)]&0x01)
#define sqlite3Isalnum(x) (sqlite3CtypeMap[(unsigned char)(x)]&0x06)
@@ -3391,16 +3373,6 @@ int sqlite3IoerrnomemError(int);
#define sqlite3Isxdigit(x) (sqlite3CtypeMap[(unsigned char)(x)]&0x08)
#define sqlite3Tolower(x) (sqlite3UpperToLower[(unsigned char)(x)])
#define sqlite3Isquote(x) (sqlite3CtypeMap[(unsigned char)(x)]&0x80)
-#else
-#define sqlite3Toupper(x) toupper((unsigned char)(x))
-#define sqlite3Isspace(x) isspace((unsigned char)(x))
-#define sqlite3Isalnum(x) isalnum((unsigned char)(x))
-#define sqlite3Isalpha(x) isalpha((unsigned char)(x))
-#define sqlite3Isdigit(x) isdigit((unsigned char)(x))
-#define sqlite3Isxdigit(x) isxdigit((unsigned char)(x))
-#define sqlite3Tolower(x) tolower((unsigned char)(x))
-#define sqlite3Isquote(x) ((x)=='"'||(x)=='\''||(x)=='['||(x)=='`')
-#endif
/*
* Internal function prototypes
@@ -4164,7 +4136,18 @@ extern int sqlite3PendingByte;
#endif
void sqlite3Reindex(Parse *, Token *, Token *);
void sqlite3AlterRenameTable(Parse *, SrcList *, Token *);
-int sqlite3GetToken(const unsigned char *, int *, bool *);
+
+/**
+ * Return the length (in bytes) of the token that begins at z[0].
+ * Store the token type in *type before returning.
+ *
+ * @param z Input stream.
+ * @param[out] type Detected type of token.
+ * @param[out] is_reserved True if reserved word.
+ */
+int
+sql_token(const char *z, int *type, bool *is_reserved);
+
void sqlite3NestedParse(Parse *, const char *, ...);
void sqlite3ExpirePreparedStatements(sqlite3 *);
int sqlite3CodeSubselect(Parse *, Expr *, int);
diff --git a/src/box/sql/tokenize.c b/src/box/sql/tokenize.c
index c77aa9b..4c01066 100644
--- a/src/box/sql/tokenize.c
+++ b/src/box/sql/tokenize.c
@@ -36,17 +36,21 @@
* individual tokens and sends those tokens one-by-one over to the
* parser for analysis.
*/
-#include "sqliteInt.h"
#include <stdlib.h>
+#include <unicode/utf8.h>
+#include <unicode/uchar.h>
+
#include "say.h"
+#include "sqliteInt.h"
/* Character classes for tokenizing
*
- * In the sqlite3GetToken() function, a switch() on aiClass[c] is implemented
- * using a lookup table, whereas a switch() directly on c uses a binary search.
- * The lookup table is much faster. To maximize speed, and to ensure that
- * a lookup table is used, all of the classes need to be small integers and
- * all of them need to be used within the switch.
+ * In the sql_token() function, a switch() on sql_ascii[c] is
+ * implemented using a lookup table, whereas a switch() directly
+ * on c uses a binary search. The lookup table is much faster.
+ * To maximize speed, and to ensure that a lookup table is used,
+ * all of the classes need to be small integers and all of them
+ * need to be used within the switch.
*/
#define CC_X 0 /* The letter 'x', or start of BLOB literal */
#define CC_KYWD 1 /* Alphabetics or '_'. Usable in a keyword */
@@ -77,10 +81,9 @@
#define CC_DOT 26 /* '.' */
#define CC_ILLEGAL 27 /* Illegal character */
-static const unsigned char aiClass[] = {
-#ifdef SQLITE_ASCII
+static const char sql_ascii_class[] = {
/* x0 x1 x2 x3 x4 x5 x6 x7 x8 x9 xa xb xc xd xe xf */
-/* 0x */ 27, 27, 27, 27, 27, 27, 27, 27, 27, 7, 7, 27, 7, 7, 27, 27,
+/* 0x */ 27, 27, 27, 27, 27, 27, 27, 27, 27, 7, 7, 7, 7, 7, 27, 27,
/* 1x */ 27, 27, 27, 27, 27, 27, 27, 27, 27, 27, 27, 27, 27, 27, 27, 27,
/* 2x */ 7, 15, 9, 5, 4, 22, 24, 8, 17, 18, 21, 20, 23, 11, 26, 16,
/* 3x */ 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 5, 19, 12, 14, 13, 6,
@@ -96,63 +99,16 @@ static const unsigned char aiClass[] = {
/* Dx */ 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
/* Ex */ 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
/* Fx */ 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2
-#endif
-#ifdef SQLITE_EBCDIC
-/* x0 x1 x2 x3 x4 x5 x6 x7 x8 x9 xa xb xc xd xe xf */
-/* 0x */ 27, 27, 27, 27, 27, 7, 27, 27, 27, 27, 27, 27, 7, 7, 27,
- 27,
-/* 1x */ 27, 27, 27, 27, 27, 27, 27, 27, 27, 27, 27, 27, 27, 27, 27, 27,
-/* 2x */ 27, 27, 27, 27, 27, 7, 27, 27, 27, 27, 27, 27, 27, 27, 27, 27,
-/* 3x */ 27, 27, 27, 27, 27, 27, 27, 27, 27, 27, 27, 27, 27, 27, 27, 27,
-/* 4x */ 7, 27, 27, 27, 27, 27, 27, 27, 27, 27, 27, 26, 12, 17, 20, 10,
-/* 5x */ 24, 27, 27, 27, 27, 27, 27, 27, 27, 27, 15, 4, 21, 18, 19, 27,
-/* 6x */ 11, 16, 27, 27, 27, 27, 27, 27, 27, 27, 27, 23, 22, 1, 13, 6,
-/* 7x */ 27, 27, 27, 27, 27, 27, 27, 27, 27, 8, 5, 5, 5, 8, 14, 8,
-/* 8x */ 27, 1, 1, 1, 1, 1, 1, 1, 1, 1, 27, 27, 27, 27, 27, 27,
-/* 9x */ 27, 1, 1, 1, 1, 1, 1, 1, 1, 1, 27, 27, 27, 27, 27, 27,
-/* Ax */ 27, 25, 1, 1, 1, 1, 1, 0, 1, 1, 27, 27, 27, 27, 27, 27,
-/* Bx */ 27, 27, 27, 27, 27, 27, 27, 27, 27, 27, 9, 27, 27, 27, 27, 27,
-/* Cx */ 27, 1, 1, 1, 1, 1, 1, 1, 1, 1, 27, 27, 27, 27, 27, 27,
-/* Dx */ 27, 1, 1, 1, 1, 1, 1, 1, 1, 1, 27, 27, 27, 27, 27, 27,
-/* Ex */ 27, 27, 1, 1, 1, 1, 1, 0, 1, 1, 27, 27, 27, 27, 27, 27,
-/* Fx */ 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 27, 27, 27, 27, 27, 27,
-#endif
};
-/*
- * The charMap() macro maps alphabetic characters (only) into their
- * lower-case ASCII equivalent. On ASCII machines, this is just
- * an upper-to-lower case map. On EBCDIC machines we also need
- * to adjust the encoding. The mapping is only valid for alphabetics
- * which are the only characters for which this feature is used.
+/**
+ * The charMap() macro maps alphabetic characters (only) into
+ * their lower-case ASCII equivalent. On ASCII machines, this
+ * is just an upper-to-lower case map.
*
* Used by keywordhash.h
*/
-#ifdef SQLITE_ASCII
#define charMap(X) sqlite3UpperToLower[(unsigned char)X]
-#endif
-#ifdef SQLITE_EBCDIC
-#define charMap(X) ebcdicToAscii[(unsigned char)X]
-const unsigned char ebcdicToAscii[] = {
-/* 0 1 2 3 4 5 6 7 8 9 A B C D E F */
- 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 0x */
- 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 1x */
- 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 2x */
- 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 3x */
- 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 4x */
- 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 5x */
- 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 95, 0, 0, /* 6x */
- 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 7x */
- 0, 97, 98, 99, 100, 101, 102, 103, 104, 105, 0, 0, 0, 0, 0, 0, /* 8x */
- 0, 106, 107, 108, 109, 110, 111, 112, 113, 114, 0, 0, 0, 0, 0, 0, /* 9x */
- 0, 0, 115, 116, 117, 118, 119, 120, 121, 122, 0, 0, 0, 0, 0, 0, /* Ax */
- 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* Bx */
- 0, 97, 98, 99, 100, 101, 102, 103, 104, 105, 0, 0, 0, 0, 0, 0, /* Cx */
- 0, 106, 107, 108, 109, 110, 111, 112, 113, 114, 0, 0, 0, 0, 0, 0, /* Dx */
- 0, 0, 115, 116, 117, 118, 119, 120, 121, 122, 0, 0, 0, 0, 0, 0, /* Ex */
- 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* Fx */
-};
-#endif
/*
* The sqlite3KeywordCode function looks up an identifier to determine if
@@ -167,360 +123,295 @@ const unsigned char ebcdicToAscii[] = {
*/
#include "keywordhash.h"
-/*
- * If X is a character that can be used in an identifier then
- * IdChar(X) will be true. Otherwise it is false.
- *
- * For ASCII, any character with the high-order bit set is
- * allowed in an identifier. For 7-bit characters,
- * sqlite3IsIdChar[X] must be 1.
- *
- * For EBCDIC, the rules are more complex but have the same
- * end result.
+#define maybe_utf8(c) ((sqlite3CtypeMap[c] & 0x40) != 0)
+
+/**
+ * Return true if current symbol is space.
*
- * Ticket #1066. the SQL standard does not allow '$' in the
- * middle of identifiers. But many SQL implementations do.
- * SQLite will allow '$' in identifiers for compatibility.
- * But the feature is undocumented.
+ * @param z Input stream.
+ * @retval True if current symbo1l space.
*/
-#ifdef SQLITE_ASCII
-#define IdChar(C) ((sqlite3CtypeMap[(unsigned char)C]&0x46)!=0)
-#endif
-#ifdef SQLITE_EBCDIC
-const char sqlite3IsEbcdicIdChar[] = {
-/* x0 x1 x2 x3 x4 x5 x6 x7 x8 x9 xA xB xC xD xE xF */
- 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, /* 4x */
- 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 0, 0, 0, 0, /* 5x */
- 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 1, 0, 0, /* 6x */
- 0, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, /* 7x */
- 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 0, /* 8x */
- 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 1, 0, 1, 0, /* 9x */
- 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 0, /* Ax */
- 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* Bx */
- 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, /* Cx */
- 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, /* Dx */
- 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, /* Ex */
- 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 0, /* Fx */
-};
-
-#define IdChar(C) (((c=C)>=0x42 && sqlite3IsEbcdicIdChar[c-0x40]))
-#endif
+static inline bool
+sql_is_space_char(const char *z)
+{
+ if (sqlite3Isspace(z[0]))
+ return true;
+ if (maybe_utf8(*(unsigned char*)z)) {
+ UChar32 c;
+ int unused = 0;
+ U8_NEXT_UNSAFE(z, unused, c);
+ if (u_isspace(c))
+ return true;
+ }
+ return false;
+}
-/*
- * Return the length (in bytes) of the token that begins at z[0].
- * Store the token type in *tokenType before returning.
+/**
+ * Calculate length of continuous sequence of
+ * space symbols.
+ *
+ * @param z Input stream.
+ * @retval Number of bytes which constitute sequence of spaces.
+ * Can be 0 if first symbol in stram is not space.
*/
+static inline int
+sql_skip_spaces(const char *z)
+{
+ int idx = 0;
+ while (true) {
+ if (sqlite3Isspace(z[idx])) {
+ idx += 1;
+ } else if (maybe_utf8(*(unsigned char *)(z + idx))) {
+ UChar32 c;
+ int new_offset = idx;
+ U8_NEXT_UNSAFE(z, new_offset, c);
+ if (!u_isspace(c))
+ break;
+ idx = new_offset;
+ } else {
+ break;
+ }
+ }
+ return idx;
+}
+
int
-sqlite3GetToken(const unsigned char *z, int *tokenType, bool *is_reserved)
+sql_token(const char *z, int *type, bool *is_reserved)
{
*is_reserved = false;
- int i, c;
- switch (aiClass[*z]) { /* Switch on the character-class of the first byte
- * of the token. See the comment on the CC_ defines
- * above.
- */
- case CC_SPACE:{
- testcase(z[0] == ' ');
- testcase(z[0] == '\t');
- testcase(z[0] == '\n');
- testcase(z[0] == '\f');
- testcase(z[0] == '\r');
- for (i = 1; sqlite3Isspace(z[i]); i++) {
+ int i, n;
+ char c, delim;
+ /* Switch on the character-class of the first byte
+ * of the token. See the comment on the CC_ defines
+ * above.
+ */
+ switch (sql_ascii_class[*(unsigned char*)z]) {
+ case CC_SPACE:
+ i = 1 + sql_skip_spaces(z+1);
+ *type = TK_SPACE;
+ return i;
+ case CC_MINUS:
+ if (z[1] == '-') {
+ for (i = 2; (c = z[i]) != 0 && c != '\n'; i++) {
}
- *tokenType = TK_SPACE;
+ *type = TK_SPACE;
return i;
}
- case CC_MINUS:{
- if (z[1] == '-') {
- for (i = 2; (c = z[i]) != 0 && c != '\n'; i++) {
- }
- *tokenType = TK_SPACE; /* IMP: R-22934-25134
- */
- return i;
- }
- *tokenType = TK_MINUS;
+ *type = TK_MINUS;
+ return 1;
+ case CC_LP:
+ *type = TK_LP;
+ return 1;
+ case CC_RP:
+ *type = TK_RP;
+ return 1;
+ case CC_SEMI:
+ *type = TK_SEMI;
+ return 1;
+ case CC_PLUS:
+ *type = TK_PLUS;
+ return 1;
+ case CC_STAR:
+ *type = TK_STAR;
+ return 1;
+ case CC_SLASH:
+ if (z[1] != '*' || z[2] == 0) {
+ *type = TK_SLASH;
return 1;
}
- case CC_LP:{
- *tokenType = TK_LP;
- return 1;
+ for (i = 3, c = z[2];
+ (c != '*' || z[i] != '/') && (c = z[i]) != 0;
+ i++) {
}
- case CC_RP:{
- *tokenType = TK_RP;
+ if (c)
+ i++;
+ *type = TK_SPACE;
+ return i;
+ case CC_PERCENT:
+ *type = TK_REM;
+ return 1;
+ case CC_EQ:
+ *type = TK_EQ;
+ return 1 + (z[1] == '=');
+ case CC_LT:
+ if ((c = z[1]) == '=') {
+ *type = TK_LE;
+ return 2;
+ } else if (c == '>') {
+ *type = TK_NE;
+ return 2;
+ } else if (c == '<') {
+ *type = TK_LSHIFT;
+ return 2;
+ } else {
+ *type = TK_LT;
return 1;
}
- case CC_SEMI:{
- *tokenType = TK_SEMI;
+ case CC_GT:
+ if ((c = z[1]) == '=') {
+ *type = TK_GE;
+ return 2;
+ } else if (c == '>') {
+ *type = TK_RSHIFT;
+ return 2;
+ } else {
+ *type = TK_GT;
return 1;
}
- case CC_PLUS:{
- *tokenType = TK_PLUS;
+ case CC_BANG:
+ if (z[1] != '=') {
+ *type = TK_ILLEGAL;
return 1;
+ } else {
+ *type = TK_NE;
+ return 2;
}
- case CC_STAR:{
- *tokenType = TK_STAR;
+ case CC_PIPE:
+ if (z[1] != '|') {
+ *type = TK_BITOR;
return 1;
+ } else {
+ *type = TK_CONCAT;
+ return 2;
}
- case CC_SLASH:{
- if (z[1] != '*' || z[2] == 0) {
- *tokenType = TK_SLASH;
- return 1;
- }
- for (i = 3, c = z[2];
- (c != '*' || z[i] != '/') && (c = z[i]) != 0;
- i++) {
+ case CC_COMMA:
+ *type = TK_COMMA;
+ return 1;
+ case CC_AND:
+ *type = TK_BITAND;
+ return 1;
+ case CC_TILDA:
+ *type = TK_BITNOT;
+ return 1;
+ case CC_QUOTE:
+ case CC_DQUOTE:
+ delim = z[0];
+ for (i = 1; (c = z[i]) != 0; i++) {
+ if (c == delim) {
+ if (z[i + 1] == delim)
+ i++;
+ else
+ break;
}
- if (c)
- i++;
- *tokenType = TK_SPACE; /* IMP: R-22934-25134
- */
+ }
+ if (c == '\'') {
+ *type = TK_STRING;
+ return i + 1;
+ } else if (c != 0) {
+ *type = TK_ID;
+ return i + 1;
+ } else {
+ *type = TK_ILLEGAL;
return i;
}
- case CC_PERCENT:{
- *tokenType = TK_REM;
+ FALLTHROUGH;
+ case CC_DOT:
+ if (!sqlite3Isdigit(z[1])) {
+ *type = TK_DOT;
return 1;
}
- case CC_EQ:{
- *tokenType = TK_EQ;
- return 1 + (z[1] == '=');
- }
- case CC_LT:{
- if ((c = z[1]) == '=') {
- *tokenType = TK_LE;
- return 2;
- } else if (c == '>') {
- *tokenType = TK_NE;
- return 2;
- } else if (c == '<') {
- *tokenType = TK_LSHIFT;
- return 2;
- } else {
- *tokenType = TK_LT;
- return 1;
+ /* If the next character is a digit, this is a
+ * floating point number that begins with ".".
+ * Fall thru into the next case.
+ */
+ FALLTHROUGH;
+ case CC_DIGIT:
+ *type = TK_INTEGER;
+ if (z[0] == '0' && (z[1] == 'x' || z[1] == 'X') &&
+ sqlite3Isxdigit(z[2])) {
+ for (i = 3; sqlite3Isxdigit(z[i]); i++) {
}
+ return i;
}
- case CC_GT:{
- if ((c = z[1]) == '=') {
- *tokenType = TK_GE;
- return 2;
- } else if (c == '>') {
- *tokenType = TK_RSHIFT;
- return 2;
- } else {
- *tokenType = TK_GT;
- return 1;
- }
+ for (i = 0; sqlite3Isdigit(z[i]); i++) {
}
- case CC_BANG:{
- if (z[1] != '=') {
- *tokenType = TK_ILLEGAL;
- return 1;
- } else {
- *tokenType = TK_NE;
- return 2;
+ if (z[i] == '.') {
+ while (sqlite3Isdigit(z[++i])) {
}
+ *type = TK_FLOAT;
}
- case CC_PIPE:{
- if (z[1] != '|') {
- *tokenType = TK_BITOR;
- return 1;
- } else {
- *tokenType = TK_CONCAT;
- return 2;
- }
+ if ((z[i] == 'e' || z[i] == 'E') &&
+ (sqlite3Isdigit(z[i + 1])
+ || ((z[i + 1] == '+' || z[i + 1] == '-') &&
+ sqlite3Isdigit(z[i + 2])))) {
+ i += 2;
+ while (sqlite3Isdigit(z[i]))
+ i++;
+ *type = TK_FLOAT;
}
- case CC_COMMA:{
- *tokenType = TK_COMMA;
- return 1;
+ if (IdChar(z[i])) {
+ *type = TK_ILLEGAL;
+ while (IdChar(z[++i])) {
+ }
}
- case CC_AND:{
- *tokenType = TK_BITAND;
- return 1;
+ return i;
+ case CC_VARNUM:
+ *type = TK_VARIABLE;
+ for (i = 1; sqlite3Isdigit(z[i]); i++) {
}
- case CC_TILDA:{
- *tokenType = TK_BITNOT;
- return 1;
+ return i;
+ case CC_DOLLAR:
+ case CC_VARALPHA:
+ n = 0;
+ *type = TK_VARIABLE;
+ for (i = 1; (c = z[i]) != 0; i++) {
+ if (IdChar(c))
+ n++;
+ else
+ break;
}
- case CC_QUOTE:
- case CC_DQUOTE:{
- int delim = z[0];
- testcase(delim == '\'');
- testcase(delim == '"');
- for (i = 1; (c = z[i]) != 0; i++) {
- if (c == delim) {
- if (z[i + 1] == delim) {
- i++;
- } else {
- break;
- }
- }
- }
- if (c == '\'') {
- *tokenType = TK_STRING;
- return i + 1;
- } else if (c != 0) {
- *tokenType = TK_ID;
- return i + 1;
- } else {
- *tokenType = TK_ILLEGAL;
- return i;
- }
- FALLTHROUGH;
+ if (n == 0)
+ *type = TK_ILLEGAL;
+ return i;
+ case CC_KYWD:
+ for (i = 1; sql_ascii_class[*(unsigned char*)(z+i)] <= CC_KYWD;
+ i++) {
}
- case CC_DOT:{
-#ifndef SQLITE_OMIT_FLOATING_POINT
- if (!sqlite3Isdigit(z[1]))
-#endif
- {
- *tokenType = TK_DOT;
- return 1;
- }
- /* If the next character is a digit, this is a floating point
- * number that begins with ".". Fall thru into the next case
+ if (!sql_is_space_char(z + i) && IdChar(z[i])) {
+ /* This token started out using characters
+ * that can appear in keywords, but z[i] is
+ * a character not allowed within keywords,
+ * so this must be an identifier instead.
*/
- FALLTHROUGH;
+ i++;
+ break;
}
- case CC_DIGIT:{
- testcase(z[0] == '0');
- testcase(z[0] == '1');
- testcase(z[0] == '2');
- testcase(z[0] == '3');
- testcase(z[0] == '4');
- testcase(z[0] == '5');
- testcase(z[0] == '6');
- testcase(z[0] == '7');
- testcase(z[0] == '8');
- testcase(z[0] == '9');
- *tokenType = TK_INTEGER;
-#ifndef SQLITE_OMIT_HEX_INTEGER
- if (z[0] == '0' && (z[1] == 'x' || z[1] == 'X')
- && sqlite3Isxdigit(z[2])) {
- for (i = 3; sqlite3Isxdigit(z[i]); i++) {
- }
- return i;
+ *type = TK_ID;
+ return keywordCode(z, i, type, is_reserved);
+ case CC_X:
+ if (z[1] == '\'') {
+ *type = TK_BLOB;
+ for (i = 2; sqlite3Isxdigit(z[i]); i++) {
}
-#endif
- for (i = 0; sqlite3Isdigit(z[i]); i++) {
- }
-#ifndef SQLITE_OMIT_FLOATING_POINT
- if (z[i] == '.') {
- i++;
- while (sqlite3Isdigit(z[i])) {
+ if (z[i] != '\'' || i % 2) {
+ *type = TK_ILLEGAL;
+ while (z[i] != 0 && z[i] != '\'')
i++;
- }
- *tokenType = TK_FLOAT;
}
- if ((z[i] == 'e' || z[i] == 'E') &&
- (sqlite3Isdigit(z[i + 1])
- || ((z[i + 1] == '+' || z[i + 1] == '-')
- && sqlite3Isdigit(z[i + 2]))
- )
- ) {
- i += 2;
- while (sqlite3Isdigit(z[i])) {
- i++;
- }
- *tokenType = TK_FLOAT;
- }
-#endif
- while (IdChar(z[i])) {
- *tokenType = TK_ILLEGAL;
+ if (z[i] != 0)
i++;
- }
- return i;
- }
- case CC_VARNUM:{
- *tokenType = TK_VARIABLE;
- for (i = 1; sqlite3Isdigit(z[i]); i++) {
- }
return i;
}
- case CC_DOLLAR:
- case CC_VARALPHA:{
- int n = 0;
- testcase(z[0] == '$');
- testcase(z[0] == '@');
- testcase(z[0] == ':');
- testcase(z[0] == '#');
- *tokenType = TK_VARIABLE;
- for (i = 1; (c = z[i]) != 0; i++) {
- if (IdChar(c)) {
- n++;
-#ifndef SQLITE_OMIT_TCL_VARIABLE
- } else if (c == '(' && n > 0) {
- do {
- i++;
- } while ((c = z[i]) != 0
- && !sqlite3Isspace(c)
- && c != ')');
- if (c == ')') {
- i++;
- } else {
- *tokenType = TK_ILLEGAL;
- }
- break;
- } else if (c == ':' && z[i + 1] == ':') {
- i++;
-#endif
- } else {
- break;
- }
- }
- if (n == 0)
- *tokenType = TK_ILLEGAL;
- return i;
- }
- case CC_KYWD:{
- for (i = 1; aiClass[z[i]] <= CC_KYWD; i++) {
- }
- if (IdChar(z[i])) {
- /* This token started out using characters that can appear in keywords,
- * but z[i] is a character not allowed within keywords, so this must
- * be an identifier instead
- */
- i++;
- break;
- }
- *tokenType = TK_ID;
- return keywordCode((char *)z, i, tokenType, is_reserved);
- }
- case CC_X:{
-#ifndef SQLITE_OMIT_BLOB_LITERAL
- testcase(z[0] == 'x');
- testcase(z[0] == 'X');
- if (z[1] == '\'') {
- *tokenType = TK_BLOB;
- for (i = 2; sqlite3Isxdigit(z[i]); i++) {
- }
- if (z[i] != '\'' || i % 2) {
- *tokenType = TK_ILLEGAL;
- while (z[i] && z[i] != '\'') {
- i++;
- }
- }
- if (z[i])
- i++;
- return i;
- }
-#endif
- /* If it is not a BLOB literal, then it must be an ID, since no
- * SQL keywords start with the letter 'x'. Fall through
- */
- FALLTHROUGH;
- }
- case CC_ID:{
- i = 1;
- break;
- }
- default:{
- *tokenType = TK_ILLEGAL;
- return 1;
- }
+ /* If it is not a BLOB literal, then it must be an
+ * ID, since no SQL keywords start with the letter
+ * 'x'. Fall through.
+ */
+ FALLTHROUGH;
+ case CC_ID:
+ i = 1;
+ break;
+ default:
+ *type = TK_ILLEGAL;
+ return 1;
}
- while (IdChar(z[i])) {
- i++;
+ int spaces_len = sql_skip_spaces(z);
+ if (spaces_len > 0) {
+ *type = TK_SPACE;
+ return spaces_len;
}
- *tokenType = TK_ID;
+ while (IdChar(z[i]))
+ i++;
+ *type = TK_ID;
return i;
}
@@ -566,8 +457,8 @@ sqlite3RunParser(Parse * pParse, const char *zSql, char **pzErrMsg)
if (zSql[i] != 0) {
pParse->sLastToken.z = &zSql[i];
pParse->sLastToken.n =
- sqlite3GetToken((u8 *) & zSql[i], &tokenType,
- &pParse->sLastToken.isReserved);
+ sql_token(&zSql[i], &tokenType,
+ &pParse->sLastToken.isReserved);
i += pParse->sLastToken.n;
if (i > mxSqlLen) {
pParse->rc = SQLITE_TOOBIG;
diff --git a/src/box/sql/util.c b/src/box/sql/util.c
index 8c4e7b9..0c2a050 100644
--- a/src/box/sql/util.c
+++ b/src/box/sql/util.c
@@ -1228,12 +1228,7 @@ sqlite3HexToInt(int h)
{
assert((h >= '0' && h <= '9') || (h >= 'a' && h <= 'f')
|| (h >= 'A' && h <= 'F'));
-#ifdef SQLITE_ASCII
h += 9 * (1 & (h >> 6));
-#endif
-#ifdef SQLITE_EBCDIC
- h += 9 * (1 & ~(h >> 4));
-#endif
return (u8) (h & 0xf);
}
diff --git a/src/box/sql/vdbetrace.c b/src/box/sql/vdbetrace.c
index 8623e68..63e2311 100644
--- a/src/box/sql/vdbetrace.c
+++ b/src/box/sql/vdbetrace.c
@@ -57,7 +57,7 @@ findNextHostParameter(const char *zSql, int *pnToken)
*pnToken = 0;
while (zSql[0]) {
- n = sqlite3GetToken((u8 *) zSql, &tokenType, &unused);
+ n = sql_token(zSql, &tokenType, &unused);
assert(n > 0 && tokenType != TK_ILLEGAL);
if (tokenType == TK_VARIABLE) {
*pnToken = n;
diff --git a/src/box/sql/whereexpr.c b/src/box/sql/whereexpr.c
index 34a1f13..c3a8634 100644
--- a/src/box/sql/whereexpr.c
+++ b/src/box/sql/whereexpr.c
@@ -256,10 +256,6 @@ isLikeOrGlob(Parse * pParse, /* Parsing and code generating context */
if (!sqlite3IsLikeFunction(db, pExpr, pnoCase, wc)) {
return 0;
}
-#ifdef SQLITE_EBCDIC
- if (*pnoCase)
- return 0;
-#endif
pList = pExpr->x.pList;
pLeft = pList->a[1].pExpr;
if (pLeft->op != TK_COLUMN || sqlite3ExprAffinity(pLeft) != SQLITE_AFF_TEXT /* Value might be numeric */
diff --git a/test/sql-tap/e_expr.test.lua b/test/sql-tap/e_expr.test.lua
index d0f6895..f7f3b15 100755
--- a/test/sql-tap/e_expr.test.lua
+++ b/test/sql-tap/e_expr.test.lua
@@ -1,6 +1,6 @@
#!/usr/bin/env tarantool
test = require("sqltester")
-test:plan(14750)
+test:plan(14748)
--!./tcltestrunner.lua
-- 2010 July 16
@@ -1506,8 +1506,6 @@ local test_cases12 ={
{10, "?123"},
{11, "@hello"},
{12, ":world"},
- {13, "$tcl"},
- {14, "$tcl(array)"},
{15, "cname"},
{16, "tblname.cname"},
diff --git a/test/sql-tap/gh-2371-utf8-spaces.test.lua b/test/sql-tap/gh-2371-utf8-spaces.test.lua
new file mode 100755
index 0000000..191cc1c
--- /dev/null
+++ b/test/sql-tap/gh-2371-utf8-spaces.test.lua
@@ -0,0 +1,37 @@
+#!/usr/bin/env tarantool
+test = require("sqltester")
+test:plan(23 * 3)
+
+-- 23 entities
+local utf8_spaces = {"\u{0009}", "\u{000A}", "\u{000B}", "\u{000C}", "\u{000D}",
+ "\u{0085}", "\u{1680}", "\u{2000}", "\u{2001}", "\u{2002}",
+ "\u{2003}", "\u{2004}", "\u{2005}", "\u{2006}", "\u{2007}",
+ "\u{2008}", "\u{2009}", "\u{200A}", "\u{2028}", "\u{2029}",
+ "\u{202F}", "\u{205F}", "\u{3000}"}
+local spaces_cnt = 23
+
+-- 1. Check UTF-8 single space
+for i, v in pairs(utf8_spaces) do
+ test:do_execsql_test(
+ "utf8-spaces-1."..i,
+ "select" .. v .. "1",
+ { 1 })
+end
+
+-- 2. Check pair simple + UTF-8 space
+for i, v in pairs(utf8_spaces) do
+ test:do_execsql_test(
+ "utf8-spaces-2."..i,
+ "select" .. v .. "1",
+ { 1 })
+end
+
+-- 3. Sequence of spaces
+for i, v in pairs(utf8_spaces) do
+ test:do_execsql_test(
+ "utf8-spaces-3."..i,
+ "select" .. v .. " " .. utf8_spaces[spaces_cnt - i + 1] .. " 1",
+ { 1 })
+end
+
+test:finish_test()
next prev parent reply other threads:[~2018-05-23 5:15 UTC|newest]
Thread overview: 8+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-05-22 15:51 [tarantool-patches] " Kirill Yukhin
2018-05-22 18:06 ` [tarantool-patches] " Vladislav Shpilevoy
2018-05-23 5:15 ` Kirill Yukhin [this message]
2018-05-23 5:54 ` Kirill Yukhin
2018-05-23 10:29 ` Vladislav Shpilevoy
2018-05-23 14:05 ` Kirill Yukhin
2018-05-24 11:09 ` Vladislav Shpilevoy
2018-05-24 14:23 ` Kirill Yukhin
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20180523051500.3fx5uoetyzsjkwp5@tarantool.org \
--to=kyukhin@tarantool.org \
--cc=tarantool-patches@freelists.org \
--cc=v.shpilevoy@tarantool.org \
--subject='[tarantool-patches] Re: [PATCH] sql: allow any space symbols to be a white space' \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox