From e1742d53d529eaafa06b06ee96fcfd3c68ae67a7 Mon Sep 17 00:00:00 2001 From: drh <> Date: Fri, 8 Apr 2022 12:50:31 +0000 Subject: [PATCH] Grammar rules for a hypothetical FOR EACH ROW statement. No new keywords required. However, the sqlite3_complete() machine had to be reworked because of the fact that FOR EACH ROW can contain internal semicolon tokens. FossilOrigin-Name: 32ae9ee7129a9f0779accef8cad499ebf044eebf08a0ecfe4cec41e5dcdd15a9 --- manifest | 17 +++--- manifest.uuid | 2 +- src/complete.c | 153 +++++++++++++++++++++++++++---------------------- src/parse.y | 22 +++++++ 4 files changed, 116 insertions(+), 78 deletions(-) diff --git a/manifest b/manifest index 9e29deaa06..50f64304dc 100644 --- a/manifest +++ b/manifest @@ -1,5 +1,5 @@ -C Fix\stwo\sunreachable\sbranches\sintroduced\sby\sthe\srecent\nsqlite3TriggersExist()\soptimization. -D 2022-04-07T20:45:38.697 +C Grammar\srules\sfor\sa\shypothetical\sFOR\sEACH\sROW\sstatement.\s\sNo\snew\skeywords\nrequired.\s\sHowever,\sthe\ssqlite3_complete()\smachine\shad\sto\sbe\sreworked\sbecause\nof\sthe\sfact\sthat\sFOR\sEACH\sROW\scan\scontain\sinternal\ssemicolon\stokens. +D 2022-04-08T12:50:31.318 F .fossil-settings/empty-dirs dbb81e8fc0401ac46a1491ab34a7f2c7c0452f2f06b54ebb845d024ca8283ef1 F .fossil-settings/ignore-glob 35175cdfcf539b2318cb04a9901442804be81cd677d8b889fcc9149c21f239ea F LICENSE.md df5091916dbb40e6e9686186587125e1b2ff51f022cc334e886c19a0e9982724 @@ -497,7 +497,7 @@ F src/btree.h 74d64b8f28cfa4a894d14d4ed64fa432cd697b98b61708d4351482ae15913e22 F src/btreeInt.h 8ce1332edd89dfd2461d561ac10a0ab5601c8e06200cb5230596c3caaf54482e F src/build.c ff119be98394a65bc8be7afc39d4a791a66f03a778d396de3ec456f5dfaf39e8 F src/callback.c 4c19af69835787bfe790ac560f3071a824eb629f34e41f97b52ce5235c77de1c -F src/complete.c a3634ab1e687055cd002e11b8f43eb75c17da23e +F src/complete.c 5d1146ae4ac8ced756be3e1e93e63eec3e6e3fa3ac92f6c52ffb2b694e153347 F src/ctime.c 026dbdcdbd8c3cde98a88483ee88310ff43150ab164ad768f12cc700a11495ad F src/date.c 15082566229d4b1e5f24fdb490bf9bcc68824b911d70e3573ef075a1b9e2d26f F src/dbpage.c a70be9a4879ac5392673a1050d526a72b8b2f9938df7049f65348566a2637db3 @@ -541,7 +541,7 @@ F src/os_win.c a8ea80037e81127ca01959daa87387cc135f325c88dc745376c4f760de852a10 F src/os_win.h 7b073010f1451abe501be30d12f6bc599824944a F src/pager.c 42120492784fc9bcd9082b5c9b5e329b7318c357f9f3574a1bbfcf7418910356 F src/pager.h f82e9844166e1585f5786837ddc7709966138ced17f568c16af7ccf946c2baa3 -F src/parse.y 9130a936927658e7c574e97e5d4f0a747a3f6b9a42bc5c63e455c8e40cc74216 +F src/parse.y bbb1709a4c5245c224f8451cab2a27604833f9c75eac237991d5da1e76b24ecc F src/pcache.c 084e638432c610f95aea72b8509f0845d2791293f39d1b82f0c0a7e089c3bb6b F src/pcache.h 4f87acd914cef5016fae3030343540d75f5b85a1877eed1a2a19b9f284248586 F src/pcache1.c 54881292a9a5db202b2c0ac541c5e3ef9a5e8c4f1c1383adb2601d5499a60e65 @@ -1945,8 +1945,11 @@ F vsixtest/vsixtest.tcl 6a9a6ab600c25a91a7acc6293828957a386a8a93 F vsixtest/vsixtest.vcxproj.data 2ed517e100c66dc455b492e1a33350c1b20fbcdc F vsixtest/vsixtest.vcxproj.filters 37e51ffedcdb064aad6ff33b6148725226cd608e F vsixtest/vsixtest_TemporaryKey.pfx e5b1b036facdb453873e7084e1cae9102ccc67a0 -P c2965fea9df7076b235d3eadaf84f0a36242476d0329030b0e57557b66e2540c -R 07d0b0c26e9cd2798bbb00ddbafc6a3b +P 1b5475d212cf9de0bff69eee8c607b4fcd8e04bf4df72171429e7609c4153951 +R f561994920f732a9d1a57dcea0b115a2 +T *branch * foreachrow-statement +T *sym-foreachrow-statement * +T -sym-trunk * U drh -Z 93237070ff101cf9292ceb1ee884c28a +Z 4231091c94d4a7d97bb7ff93181450a2 # Remove this line to create a well-formed Fossil manifest. diff --git a/manifest.uuid b/manifest.uuid index 55edbcf9bb..4c28fa2822 100644 --- a/manifest.uuid +++ b/manifest.uuid @@ -1 +1 @@ -1b5475d212cf9de0bff69eee8c607b4fcd8e04bf4df72171429e7609c4153951 \ No newline at end of file +32ae9ee7129a9f0779accef8cad499ebf044eebf08a0ecfe4cec41e5dcdd15a9 \ No newline at end of file diff --git a/src/complete.c b/src/complete.c index bb2c03098e..2474921364 100644 --- a/src/complete.c +++ b/src/complete.c @@ -37,16 +37,17 @@ extern const char sqlite3IsEbcdicIdChar[]; ** Token types used by the sqlite3_complete() routine. See the header ** comments on that procedure for additional information. */ -#define tkSEMI 0 -#define tkWS 1 -#define tkOTHER 2 -#ifndef SQLITE_OMIT_TRIGGER -#define tkEXPLAIN 3 -#define tkCREATE 4 -#define tkTEMP 5 -#define tkTRIGGER 6 -#define tkEND 7 -#endif +#define tkSEMI 0 /* ; */ +#define tkWS 1 /* whitespace */ +#define tkOTHER 2 /* other puntuation, literal-value, or quoted ID */ +#define tkEXPLAIN 3 /* EXPLAIN keyword */ +#define tkCREATE 4 /* CREATE keyword */ +#define tkTRIGGER 5 /* TRIGGER keyword */ +#define tkEND 6 /* END keyword */ +#define tkFOR 7 /* FOR keyword */ +#define tkEACH 8 /* EACH keyword */ +#define tkROW 9 /* ROW keyword */ +#define tkKYWD 10 /* Any other keyword */ /* ** Return TRUE if the given SQL string ends in a semicolon. @@ -57,8 +58,6 @@ extern const char sqlite3IsEbcdicIdChar[]; ** ** This implementation uses a state machine with 8 states: ** -** (0) INVALID We have not yet seen a non-whitespace character. -** ** (1) START At the beginning or end of an SQL statement. This routine ** returns 1 if it ends in the START state and 0 if it ends ** in any other state. @@ -66,33 +65,45 @@ extern const char sqlite3IsEbcdicIdChar[]; ** (2) NORMAL We are in the middle of statement which ends with a single ** semicolon. ** -** (3) EXPLAIN The keyword EXPLAIN has been seen at the beginning of +** (2) EXPLAIN The keyword EXPLAIN has been seen at the beginning of ** a statement. ** -** (4) CREATE The keyword CREATE has been seen at the beginning of a +** (3) CREATE The keyword CREATE has been seen at the beginning of a ** statement, possibly preceded by EXPLAIN and/or followed by ** TEMP or TEMPORARY ** -** (5) TRIGGER We are in the middle of a trigger definition that must be -** ended by a semicolon, the keyword END, and another semicolon. +** (4) TRIGGER We are in the middle of a trigger definition or FOR EACH ROW +** statement that must be ended by ";END:" ** -** (6) SEMI We've seen the first semicolon in the ";END;" that occurs at +** (5) SEMI We've seen the first semicolon in the ";END;" that occurs at ** the end of a trigger definition. ** -** (7) END We've seen the ";END" of the ";END;" that occurs at the end +** (6) END We've seen the ";END" of the ";END;" that occurs at the end ** of a trigger definition. ** +** (7) FOR We've seen the "FOR" keyword near the beginning of a +** statement. Awaiting a following "EACH". +** +** (8) EACH We've seen keywrods "FOR" and "EACH" near the beginning of +** of a statement and are awaiting a "ROW" token. +** +** (9) INIT No non-whitespace input seen so far. +** +** ** Transitions between states above are determined by tokens extracted ** from the input. The following tokens are significant: ** ** (0) tkSEMI A semicolon. ** (1) tkWS Whitespace. -** (2) tkOTHER Any other SQL token. +** (2) tkOTHER Any other token. ** (3) tkEXPLAIN The "explain" keyword. ** (4) tkCREATE The "create" keyword. -** (5) tkTEMP The "temp" or "temporary" keyword. -** (6) tkTRIGGER The "trigger" keyword. -** (7) tkEND The "end" keyword. +** (5) tkTRIGGER The "trigger" keyword. +** (6) tkEND The "end" keyword. +** (7) tkFOR The "for" keyword. +** (8) tkEACH The "each" keyword. +** (9) tkROW The "row" keyword. +** (10) tkKYWD Any other keyword-like token. ** ** Whitespace never causes a state transition and is always ignored. ** This means that a SQL string of all whitespace is invalid. @@ -102,37 +113,28 @@ extern const char sqlite3IsEbcdicIdChar[]; ** is look for a semicolon that is not part of an string or comment. */ int sqlite3_complete(const char *zSql){ - u8 state = 0; /* Current state, using numbers defined in header comment */ + u8 state = 9; /* Current state, using numbers defined in header comment */ u8 token; /* Value of the next token */ -#ifndef SQLITE_OMIT_TRIGGER /* A complex statement machine used to detect the end of a CREATE TRIGGER ** statement. This is the normal case. */ - static const u8 trans[8][8] = { - /* Token: */ - /* State: ** SEMI WS OTHER EXPLAIN CREATE TEMP TRIGGER END */ - /* 0 INVALID: */ { 1, 0, 2, 3, 4, 2, 2, 2, }, - /* 1 START: */ { 1, 1, 2, 3, 4, 2, 2, 2, }, - /* 2 NORMAL: */ { 1, 2, 2, 2, 2, 2, 2, 2, }, - /* 3 EXPLAIN: */ { 1, 3, 3, 2, 4, 2, 2, 2, }, - /* 4 CREATE: */ { 1, 4, 2, 2, 2, 4, 5, 2, }, - /* 5 TRIGGER: */ { 6, 5, 5, 5, 5, 5, 5, 5, }, - /* 6 SEMI: */ { 6, 6, 5, 5, 5, 5, 5, 7, }, - /* 7 END: */ { 1, 7, 5, 5, 5, 5, 5, 5, }, - }; -#else - /* If triggers are not supported by this compile then the statement machine - ** used to detect the end of a statement is much simpler - */ - static const u8 trans[3][3] = { - /* Token: */ - /* State: ** SEMI WS OTHER */ - /* 0 INVALID: */ { 1, 0, 2, }, - /* 1 START: */ { 1, 1, 2, }, - /* 2 NORMAL: */ { 1, 2, 2, }, + static const u8 trans[10][11] = { + /* Token: */ + /* 0 1 2 3 4 5 6 7 8 9 10 */ +/* State: ** SEMI WS OTHER EXPLN CREATE TRIGR END FOR EACH ROW KYWD */ +/* 0 START: */{ 0, 0, 1, 2, 3, 1, 1, 7, 1, 1, 1 }, +/* 1 NORMAL: */{ 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1 }, +/* 2 EXPLAIN: */{ 0, 2, 1, 1, 3, 1, 1, 7, 1, 1, 2 }, +/* 3 CREATE: */{ 0, 3, 1, 1, 1, 4, 1, 1, 1, 1, 3 }, +/* 4 TRIGGER: */{ 5, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4 }, +/* 5 SEMI: */{ 5, 5, 4, 4, 4, 4, 6, 4, 4, 4, 4 }, +/* 6 END: */{ 0, 6, 4, 4, 4, 4, 4, 4, 4, 4, 4 }, +/* 7 FOR: */{ 0, 7, 1, 1, 1, 1, 1, 1, 1, 1, 1 }, +/* 8 EACH: */{ 0, 8, 1, 1, 1, 1, 1, 1, 1, 4, 1 }, +/* 9 INIT: */{ 0, 9, 1, 2, 3, 1, 1, 7, 1, 1, 1 } }; -#endif /* SQLITE_OMIT_TRIGGER */ + #ifdef SQLITE_ENABLE_API_ARMOR if( zSql==0 ){ @@ -202,27 +204,12 @@ int sqlite3_complete(const char *zSql){ /* Keywords and unquoted identifiers */ int nId; for(nId=1; IdChar(zSql[nId]); nId++){} -#ifdef SQLITE_OMIT_TRIGGER - token = tkOTHER; -#else switch( *zSql ){ case 'c': case 'C': { if( nId==6 && sqlite3StrNICmp(zSql, "create", 6)==0 ){ token = tkCREATE; }else{ - token = tkOTHER; - } - break; - } - case 't': case 'T': { - if( nId==7 && sqlite3StrNICmp(zSql, "trigger", 7)==0 ){ - token = tkTRIGGER; - }else if( nId==4 && sqlite3StrNICmp(zSql, "temp", 4)==0 ){ - token = tkTEMP; - }else if( nId==9 && sqlite3StrNICmp(zSql, "temporary", 9)==0 ){ - token = tkTEMP; - }else{ - token = tkOTHER; + token = tkKYWD; } break; } @@ -230,22 +217,45 @@ int sqlite3_complete(const char *zSql){ if( nId==3 && sqlite3StrNICmp(zSql, "end", 3)==0 ){ token = tkEND; }else -#ifndef SQLITE_OMIT_EXPLAIN if( nId==7 && sqlite3StrNICmp(zSql, "explain", 7)==0 ){ token = tkEXPLAIN; }else -#endif - { - token = tkOTHER; + if( nId==4 && sqlite3StrNICmp(zSql, "each", 4)==0 ){ + token = tkEACH; + }else{ + token = tkKYWD; + } + break; + } + case 'f': case 'F': { + if( nId==3 && sqlite3StrNICmp(zSql, "for", 3)==0 ){ + token = tkFOR; + }else{ + token = tkKYWD; + } + break; + } + case 'r': case 'R': { + if( nId==3 && sqlite3StrNICmp(zSql, "row", 3)==0 ){ + token = tkROW; + }else{ + token = tkKYWD; + } + break; + } + case 't': case 'T': { + if( nId==7 && sqlite3StrNICmp(zSql, "trigger", 7)==0 ){ + token = tkTRIGGER; + }else{ + token = tkKYWD; } break; } default: { - token = tkOTHER; + token = tkKYWD; break; } } -#endif /* SQLITE_OMIT_TRIGGER */ zSql += nId-1; }else{ /* Operators and special symbols */ @@ -254,10 +264,13 @@ int sqlite3_complete(const char *zSql){ break; } } +#if 0 + printf("COMPLETER: %d.%d -> %d\n", state, token, trans[state][token]); +#endif state = trans[state][token]; zSql++; } - return state==1; + return state==0; } #ifndef SQLITE_OMIT_UTF16 diff --git a/src/parse.y b/src/parse.y index 4deae14b4c..42d1e126de 100644 --- a/src/parse.y +++ b/src/parse.y @@ -1852,6 +1852,28 @@ over_clause(A) ::= OVER nm(Z). { filter_clause(A) ::= FILTER LP WHERE expr(X) RP. { A = X; } %endif /* SQLITE_OMIT_WINDOWFUNC */ +//////////////////////// The FOR EACH ROW Command /////////////////////////////////// +// +%ifndef SQLITE_OMIT_FOREACHROW +cmd ::= foreachcmd. +foreachcmd ::= FOR EACH ROW IN select fewhenlist END. +foreachcmd ::= FOR EACH ROW IN select DO feactionlist END. +fewhenlist ::= fewhen. +fewhenlist ::= fewhenlist fewhen. +fewhen ::= WHEN expr THEN feactionlist. +fewhen ::= WHEN expr THEN DO NOTHING. +fewhen ::= ELSE feactionlist. +feactionlist ::= feaction SEMI. +feactionlist ::= feactionlist feaction SEMI. +feaction ::= insert_cmd INTO xfullname idlist_opt select upsert. +feaction ::= insert_cmd INTO xfullname idlist_opt DEFAULT VALUES. +feaction ::= UPDATE orconf xfullname indexed_opt SET setlist from where_opt upsert. +feaction ::= DELETE FROM xfullname indexed_opt where_opt. +feaction ::= select. +feaction ::= . +%endif + + /* ** The code generator needs some extra TK_ token values for tokens that ** are synthesized and do not actually appear in the grammar: -- 2.39.5