scintilla/LexSQL.cxx

   1 // Scintilla source code edit control
   2 /** @file LexSQL.cxx
   3  ** Lexer for SQL, including PL/SQL and SQL*Plus.
   4  **/
   5 // Copyright 1998-2005 by Neil Hodgson <neilh@scintilla.org>
   6 // The License.txt file describes the conditions under which this software may be distributed.
   7
   8 #include <stdlib.h>
   9 #include <string.h>
  10 #include <ctype.h>
  11 #include <stdio.h>
  12 #include <stdarg.h>
  13
  14 #include "Platform.h"
  15
  16 #include "PropSet.h"
  17 #include "Accessor.h"
  18 #include "StyleContext.h"
  19 #include "KeyWords.h"
  20 #include "Scintilla.h"
  21 #include "SciLexer.h"
  22
  23 #ifdef SCI_NAMESPACE
  24 using namespace Scintilla;
  25 #endif
  26
  27 static inline bool IsAWordChar(int ch) {
  28         return (ch < 0x80) && (isalnum(ch) || ch == '_');
  29 }
  30
  31 static inline bool IsAWordStart(int ch) {
  32         return (ch < 0x80) && (isalpha(ch) || ch == '_');
  33 }
  34
  35 static inline bool IsADoxygenChar(int ch) {
  36         return (islower(ch) || ch == '$' || ch == '@' ||
  37                 ch == '\\' || ch == '&' || ch == '<' ||
  38                 ch == '>' || ch == '#' || ch == '{' ||
  39                 ch == '}' || ch == '[' || ch == ']');
  40 }
  41
  42 static inline bool IsANumberChar(int ch) {
  43         // Not exactly following number definition (several dots are seen as OK, etc.)
  44         // but probably enough in most cases.
  45         return (ch < 0x80) &&
  46                 (isdigit(ch) || toupper(ch) == 'E' ||
  47              ch == '.' || ch == '-' || ch == '+');
  48 }
  49
  50 static void ColouriseSQLDoc(unsigned int startPos, int length, int initStyle, WordList *keywordlists[],
  51                             Accessor &styler) {
  52
  53         WordList &keywords1 = *keywordlists[0];
  54         WordList &keywords2 = *keywordlists[1];
  55         WordList &kw_pldoc = *keywordlists[2];
  56         WordList &kw_sqlplus = *keywordlists[3];
  57         WordList &kw_user1 = *keywordlists[4];
  58         WordList &kw_user2 = *keywordlists[5];
  59         WordList &kw_user3 = *keywordlists[6];
  60         WordList &kw_user4 = *keywordlists[7];
  61
  62         StyleContext sc(startPos, length, initStyle, styler);
  63
  64         // property sql.backslash.escapes
  65         //      Enables backslash as an escape character in SQL.
  66         bool sqlBackslashEscapes = styler.GetPropertyInt("sql.backslash.escapes", 0) != 0;
  67
  68         bool sqlBackticksIdentifier = styler.GetPropertyInt("lexer.sql.backticks.identifier", 0) != 0;
  69         int styleBeforeDCKeyword = SCE_SQL_DEFAULT;
  70         for (; sc.More(); sc.Forward()) {
  71                 // Determine if the current state should terminate.
  72                 switch (sc.state) {
  73                 case SCE_SQL_OPERATOR:
  74                         sc.SetState(SCE_SQL_DEFAULT);
  75                         break;
  76                 case SCE_SQL_NUMBER:
  77                         // We stop the number definition on non-numerical non-dot non-eE non-sign char
  78                         if (!IsANumberChar(sc.ch)) {
  79                                 sc.SetState(SCE_SQL_DEFAULT);
  80                         }
  81                         break;
  82                 case SCE_SQL_IDENTIFIER:
  83                         if (!IsAWordChar(sc.ch)) {
  84                                 int nextState = SCE_SQL_DEFAULT;
  85                                 char s[1000];
  86                                 sc.GetCurrentLowered(s, sizeof(s));
  87                                 if (keywords1.InList(s)) {
  88                                         sc.ChangeState(SCE_SQL_WORD);
  89                                 } else if (keywords2.InList(s)) {
  90                                         sc.ChangeState(SCE_SQL_WORD2);
  91                                 } else if (kw_sqlplus.InListAbbreviated(s, '~')) {
  92                                         sc.ChangeState(SCE_SQL_SQLPLUS);
  93                                         if (strncmp(s, "rem", 3) == 0) {
  94                                                 nextState = SCE_SQL_SQLPLUS_COMMENT;
  95                                         } else if (strncmp(s, "pro", 3) == 0) {
  96                                                 nextState = SCE_SQL_SQLPLUS_PROMPT;
  97                                         }
  98                                 } else if (kw_user1.InList(s)) {
  99                                         sc.ChangeState(SCE_SQL_USER1);
 100                                 } else if (kw_user2.InList(s)) {
 101                                         sc.ChangeState(SCE_SQL_USER2);
 102                                 } else if (kw_user3.InList(s)) {
 103                                         sc.ChangeState(SCE_SQL_USER3);
 104                                 } else if (kw_user4.InList(s)) {
 105                                         sc.ChangeState(SCE_SQL_USER4);
 106                                 }
 107                                 sc.SetState(nextState);
 108                         }
 109                         break;
 110                 case SCE_SQL_QUOTEDIDENTIFIER:
 111                         if (sc.ch == 0x60) {
 112                                 if (sc.chNext == 0x60) {
 113                                         sc.Forward();   // Ignore it
 114                                 } else {
 115                                         sc.ForwardSetState(SCE_SQL_DEFAULT);
 116                                 }
 117                         }
 118                         break;
 119                 case SCE_SQL_COMMENT:
 120                         if (sc.Match('*', '/')) {
 121                                 sc.Forward();
 122                                 sc.ForwardSetState(SCE_SQL_DEFAULT);
 123                         }
 124                         break;
 125                 case SCE_SQL_COMMENTDOC:
 126                         if (sc.Match('*', '/')) {
 127                                 sc.Forward();
 128                                 sc.ForwardSetState(SCE_SQL_DEFAULT);
 129                         } else if (sc.ch == '@' || sc.ch == '\\') { // Doxygen support
 130                                 // Verify that we have the conditions to mark a comment-doc-keyword
 131                                 if ((IsASpace(sc.chPrev) || sc.chPrev == '*') && (!IsASpace(sc.chNext))) {
 132                                         styleBeforeDCKeyword = SCE_SQL_COMMENTDOC;
 133                                         sc.SetState(SCE_SQL_COMMENTDOCKEYWORD);
 134                                 }
 135                         }
 136                         break;
 137                 case SCE_SQL_COMMENTLINE:
 138                 case SCE_SQL_COMMENTLINEDOC:
 139                 case SCE_SQL_SQLPLUS_COMMENT:
 140                 case SCE_SQL_SQLPLUS_PROMPT:
 141                         if (sc.atLineStart) {
 142                                 sc.SetState(SCE_SQL_DEFAULT);
 143                         }
 144                         break;
 145                 case SCE_SQL_COMMENTDOCKEYWORD:
 146                         if ((styleBeforeDCKeyword == SCE_SQL_COMMENTDOC) && sc.Match('*', '/')) {
 147                                 sc.ChangeState(SCE_SQL_COMMENTDOCKEYWORDERROR);
 148                                 sc.Forward();
 149                                 sc.ForwardSetState(SCE_SQL_DEFAULT);
 150                         } else if (!IsADoxygenChar(sc.ch)) {
 151                                 char s[100];
 152                                 sc.GetCurrentLowered(s, sizeof(s));
 153                                 if (!isspace(sc.ch) || !kw_pldoc.InList(s + 1)) {
 154                                         sc.ChangeState(SCE_SQL_COMMENTDOCKEYWORDERROR);
 155                                 }
 156                                 sc.SetState(styleBeforeDCKeyword);
 157                         }
 158                         break;
 159                 case SCE_SQL_CHARACTER:
 160                         if (sqlBackslashEscapes && sc.ch == '\\') {
 161                                 sc.Forward();
 162                         } else if (sc.ch == '\'') {
 163                                 if (sc.chNext == '\"') {
 164                                         sc.Forward();
 165                                 } else {
 166                                         sc.ForwardSetState(SCE_SQL_DEFAULT);
 167                                 }
 168                         }
 169                         break;
 170                 case SCE_SQL_STRING:
 171                         if (sc.ch == '\\') {
 172                                 // Escape sequence
 173                                 sc.Forward();
 174                         } else if (sc.ch == '\"') {
 175                                 if (sc.chNext == '\"') {
 176                                         sc.Forward();
 177                                 } else {
 178                                         sc.ForwardSetState(SCE_SQL_DEFAULT);
 179                                 }
 180                         }
 181                         break;
 182                 }
 183
 184                 // Determine if a new state should be entered.
 185                 if (sc.state == SCE_SQL_DEFAULT) {
 186                         if (IsADigit(sc.ch) || (sc.ch == '.' && IsADigit(sc.chNext))) {
 187                                 sc.SetState(SCE_SQL_NUMBER);
 188                         } else if (IsAWordStart(sc.ch)) {
 189                                 sc.SetState(SCE_SQL_IDENTIFIER);
 190                         } else if (sc.ch == 0x60 && sqlBackticksIdentifier) {
 191                                 sc.SetState(SCE_SQL_QUOTEDIDENTIFIER);
 192                         } else if (sc.Match('/', '*')) {
 193                                 if (sc.Match("/**") || sc.Match("/*!")) {       // Support of Doxygen doc. style
 194                                         sc.SetState(SCE_SQL_COMMENTDOC);
 195                                 } else {
 196                                         sc.SetState(SCE_SQL_COMMENT);
 197                                 }
 198                                 sc.Forward();   // Eat the * so it isn't used for the end of the comment
 199                         } else if (sc.Match('-', '-')) {
 200                                 // MySQL requires a space or control char after --
 201                                 // http://dev.mysql.com/doc/mysql/en/ansi-diff-comments.html
 202                                 // Perhaps we should enforce that with proper property:
 203 //~                     } else if (sc.Match("-- ")) {
 204                                 sc.SetState(SCE_SQL_COMMENTLINE);
 205                         } else if (sc.ch == '#') {
 206                                 sc.SetState(SCE_SQL_COMMENTLINEDOC);
 207                         } else if (sc.ch == '\'') {
 208                                 sc.SetState(SCE_SQL_CHARACTER);
 209                         } else if (sc.ch == '\"') {
 210                                 sc.SetState(SCE_SQL_STRING);
 211                         } else if (isoperator(static_cast<char>(sc.ch))) {
 212                                 sc.SetState(SCE_SQL_OPERATOR);
 213                         }
 214                 }
 215         }
 216         sc.Complete();
 217 }
 218
 219 static bool IsStreamCommentStyle(int style) {
 220         return style == SCE_SQL_COMMENT ||
 221                style == SCE_SQL_COMMENTDOC ||
 222                style == SCE_SQL_COMMENTDOCKEYWORD ||
 223                style == SCE_SQL_COMMENTDOCKEYWORDERROR;
 224 }
 225
 226 // Store both the current line's fold level and the next lines in the
 227 // level store to make it easy to pick up with each increment.
 228 static void FoldSQLDoc(unsigned int startPos, int length, int initStyle,
 229                             WordList *[], Accessor &styler) {
 230         bool foldComment = styler.GetPropertyInt("fold.comment") != 0;
 231         bool foldCompact = styler.GetPropertyInt("fold.compact", 1) != 0;
 232         bool foldOnlyBegin = styler.GetPropertyInt("fold.sql.only.begin", 0) != 0;
 233
 234         // property fold.sql.exists
 235         //      Enables "EXISTS" to end a fold as is started by "IF" in "DROP TABLE IF EXISTS".
 236         bool foldSqlExists = styler.GetPropertyInt("fold.sql.exists", 1) != 0;
 237
 238         unsigned int endPos = startPos + length;
 239         int visibleChars = 0;
 240         int lineCurrent = styler.GetLine(startPos);
 241         int levelCurrent = SC_FOLDLEVELBASE;
 242         if (lineCurrent > 0) {
 243                 levelCurrent = styler.LevelAt(lineCurrent - 1) >> 16;
 244         }
 245         int levelNext = levelCurrent;
 246         char chNext = styler[startPos];
 247         int styleNext = styler.StyleAt(startPos);
 248         int style = initStyle;
 249         bool endFound = false;
 250         for (unsigned int i = startPos; i < endPos; i++) {
 251                 char ch = chNext;
 252                 chNext = styler.SafeGetCharAt(i + 1);
 253                 int stylePrev = style;
 254                 style = styleNext;
 255                 styleNext = styler.StyleAt(i + 1);
 256                 bool atEOL = (ch == '\r' && chNext != '\n') || (ch == '\n');
 257                 if (foldComment && IsStreamCommentStyle(style)) {
 258                         if (!IsStreamCommentStyle(stylePrev)) {
 259                                 levelNext++;
 260                         } else if (!IsStreamCommentStyle(styleNext) && !atEOL) {
 261                                 // Comments don't end at end of line and the next character may be unstyled.
 262                                 levelNext--;
 263                         }
 264                 }
 265                 if (foldComment && (style == SCE_SQL_COMMENTLINE)) {
 266                         // MySQL needs -- comments to be followed by space or control char
 267                         if ((ch == '-') && (chNext == '-')) {
 268                                 char chNext2 = styler.SafeGetCharAt(i + 2);
 269                                 char chNext3 = styler.SafeGetCharAt(i + 3);
 270                                 if (chNext2 == '{' || chNext3 == '{') {
 271                                         levelNext++;
 272                                 } else if (chNext2 == '}' || chNext3 == '}') {
 273                                         levelNext--;
 274                                 }
 275                         }
 276                 }
 277                 if (style == SCE_SQL_OPERATOR) {
 278                         if (ch == '(') {
 279                                 levelNext++;
 280                         } else if (ch == ')') {
 281                                 levelNext--;
 282                         }
 283                 }
 284                 // If new keyword (cannot trigger on elseif or nullif, does less tests)
 285                 if (style == SCE_SQL_WORD && stylePrev != SCE_SQL_WORD) {
 286                         const int MAX_KW_LEN = 6;       // Maximum length of folding keywords
 287                         char s[MAX_KW_LEN + 2];
 288                         unsigned int j = 0;
 289                         for (; j < MAX_KW_LEN + 1; j++) {
 290                                 if (!iswordchar(styler[i + j])) {
 291                                         break;
 292                                 }
 293                                 s[j] = static_cast<char>(tolower(styler[i + j]));
 294                         }
 295                         if (j == MAX_KW_LEN + 1) {
 296                                 // Keyword too long, don't test it
 297                                 s[0] = '\0';
 298                         } else {
 299                                 s[j] = '\0';
 300                         }
 301                         if ((!foldOnlyBegin) && (strcmp(s, "if") == 0 || strcmp(s, "loop") == 0)) {
 302                                 if (endFound) {
 303                                         // ignore
 304                                         endFound = false;
 305                                 } else {
 306                                         levelNext++;
 307                                 }
 308                         } else if (strcmp(s, "begin") == 0) {
 309                                 levelNext++;
 310                         } else if ((strcmp(s, "end") == 0) ||
 311 //                                              // DROP TABLE IF EXISTS or CREATE TABLE IF NOT EXISTS
 312                                                 (foldSqlExists && (strcmp(s, "exists") == 0)) ||
 313 //                                              //  SQL Anywhere permits IF ... ELSE ... ENDIF
 314 //                                              //      will only be active if "endif" appears in the
 315 //                                              //              keyword list.
 316                                                 (strcmp(s, "endif") == 0)) {
 317                                 endFound = true;
 318                                 levelNext--;
 319                                 if (levelNext < SC_FOLDLEVELBASE) {
 320                                         levelNext = SC_FOLDLEVELBASE;
 321                                 }
 322                         }
 323                 }
 324                 if (atEOL) {
 325                         int levelUse = levelCurrent;
 326                         int lev = levelUse | levelNext << 16;
 327                         if (visibleChars == 0 && foldCompact)
 328                                 lev |= SC_FOLDLEVELWHITEFLAG;
 329                         if (levelUse < levelNext)
 330                                 lev |= SC_FOLDLEVELHEADERFLAG;
 331                         if (lev != styler.LevelAt(lineCurrent)) {
 332                                 styler.SetLevel(lineCurrent, lev);
 333                         }
 334                         lineCurrent++;
 335                         levelCurrent = levelNext;
 336                         visibleChars = 0;
 337                         endFound = false;
 338                 }
 339                 if (!isspacechar(ch)) {
 340                         visibleChars++;
 341                 }
 342         }
 343 }
 344
 345 static const char * const sqlWordListDesc[] = {
 346         "Keywords",
 347         "Database Objects",
 348         "PLDoc",
 349         "SQL*Plus",
 350         "User Keywords 1",
 351         "User Keywords 2",
 352         "User Keywords 3",
 353         "User Keywords 4",
 354         0
 355 };
 356
 357 LexerModule lmSQL(SCLEX_SQL, ColouriseSQLDoc, "sql", FoldSQLDoc, sqlWordListDesc);