You can not select more than 25 topics
Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
261 lines
7.6 KiB
261 lines
7.6 KiB
// Scintilla source code edit control
|
|
/** @file LexAPDL.cxx
|
|
** Lexer for APDL. Based on the lexer for Assembler by The Black Horus.
|
|
** By Hadar Raz.
|
|
**/
|
|
// Copyright 1998-2003 by Neil Hodgson <neilh@scintilla.org>
|
|
// The License.txt file describes the conditions under which this software may be distributed.
|
|
|
|
#include <stdlib.h>
|
|
#include <string.h>
|
|
#include <stdio.h>
|
|
#include <stdarg.h>
|
|
#include <assert.h>
|
|
#include <ctype.h>
|
|
|
|
#include <string>
|
|
#include <string_view>
|
|
|
|
#include "ILexer.h"
|
|
#include "Scintilla.h"
|
|
#include "SciLexer.h"
|
|
|
|
#include "WordList.h"
|
|
#include "LexAccessor.h"
|
|
#include "Accessor.h"
|
|
#include "StyleContext.h"
|
|
#include "CharacterSet.h"
|
|
#include "LexerModule.h"
|
|
|
|
using namespace Lexilla;
|
|
|
|
static inline bool IsAWordChar(const int ch) {
|
|
return (ch < 0x80 && (isalnum(ch) || ch == '_'));
|
|
}
|
|
|
|
static inline bool IsAnOperator(char ch) {
|
|
// '.' left out as it is used to make up numbers
|
|
if (ch == '*' || ch == '/' || ch == '-' || ch == '+' ||
|
|
ch == '(' || ch == ')' || ch == '=' || ch == '^' ||
|
|
ch == '[' || ch == ']' || ch == '<' || ch == '&' ||
|
|
ch == '>' || ch == ',' || ch == '|' || ch == '~' ||
|
|
ch == '$' || ch == ':' || ch == '%')
|
|
return true;
|
|
return false;
|
|
}
|
|
|
|
static void ColouriseAPDLDoc(Sci_PositionU startPos, Sci_Position length, int initStyle, WordList *keywordlists[],
|
|
Accessor &styler) {
|
|
|
|
int stringStart = ' ';
|
|
|
|
WordList &processors = *keywordlists[0];
|
|
WordList &commands = *keywordlists[1];
|
|
WordList &slashcommands = *keywordlists[2];
|
|
WordList &starcommands = *keywordlists[3];
|
|
WordList &arguments = *keywordlists[4];
|
|
WordList &functions = *keywordlists[5];
|
|
|
|
// Do not leak onto next line
|
|
initStyle = SCE_APDL_DEFAULT;
|
|
StyleContext sc(startPos, length, initStyle, styler);
|
|
|
|
for (; sc.More(); sc.Forward()) {
|
|
// Determine if the current state should terminate.
|
|
if (sc.state == SCE_APDL_NUMBER) {
|
|
if (!(IsADigit(sc.ch) || sc.ch == '.' || (sc.ch == 'e' || sc.ch == 'E') ||
|
|
((sc.ch == '+' || sc.ch == '-') && (sc.chPrev == 'e' || sc.chPrev == 'E')))) {
|
|
sc.SetState(SCE_APDL_DEFAULT);
|
|
}
|
|
} else if (sc.state == SCE_APDL_COMMENT) {
|
|
if (sc.atLineEnd) {
|
|
sc.SetState(SCE_APDL_DEFAULT);
|
|
}
|
|
} else if (sc.state == SCE_APDL_COMMENTBLOCK) {
|
|
if (sc.atLineEnd) {
|
|
if (sc.ch == '\r') {
|
|
sc.Forward();
|
|
}
|
|
sc.ForwardSetState(SCE_APDL_DEFAULT);
|
|
}
|
|
} else if (sc.state == SCE_APDL_STRING) {
|
|
if (sc.atLineEnd) {
|
|
sc.SetState(SCE_APDL_DEFAULT);
|
|
} else if ((sc.ch == '\'' && stringStart == '\'') || (sc.ch == '\"' && stringStart == '\"')) {
|
|
sc.ForwardSetState(SCE_APDL_DEFAULT);
|
|
}
|
|
} else if (sc.state == SCE_APDL_WORD) {
|
|
if (!IsAWordChar(sc.ch)) {
|
|
char s[100];
|
|
sc.GetCurrentLowered(s, sizeof(s));
|
|
if (processors.InList(s)) {
|
|
sc.ChangeState(SCE_APDL_PROCESSOR);
|
|
} else if (slashcommands.InList(s)) {
|
|
sc.ChangeState(SCE_APDL_SLASHCOMMAND);
|
|
} else if (starcommands.InList(s)) {
|
|
sc.ChangeState(SCE_APDL_STARCOMMAND);
|
|
} else if (commands.InList(s)) {
|
|
sc.ChangeState(SCE_APDL_COMMAND);
|
|
} else if (arguments.InList(s)) {
|
|
sc.ChangeState(SCE_APDL_ARGUMENT);
|
|
} else if (functions.InList(s)) {
|
|
sc.ChangeState(SCE_APDL_FUNCTION);
|
|
}
|
|
sc.SetState(SCE_APDL_DEFAULT);
|
|
}
|
|
} else if (sc.state == SCE_APDL_OPERATOR) {
|
|
if (!IsAnOperator(static_cast<char>(sc.ch))) {
|
|
sc.SetState(SCE_APDL_DEFAULT);
|
|
}
|
|
}
|
|
|
|
// Determine if a new state should be entered.
|
|
if (sc.state == SCE_APDL_DEFAULT) {
|
|
if (sc.ch == '!' && sc.chNext == '!') {
|
|
sc.SetState(SCE_APDL_COMMENTBLOCK);
|
|
} else if (sc.ch == '!') {
|
|
sc.SetState(SCE_APDL_COMMENT);
|
|
} else if (IsADigit(sc.ch) || (sc.ch == '.' && IsADigit(sc.chNext))) {
|
|
sc.SetState(SCE_APDL_NUMBER);
|
|
} else if (sc.ch == '\'' || sc.ch == '\"') {
|
|
sc.SetState(SCE_APDL_STRING);
|
|
stringStart = sc.ch;
|
|
} else if (IsAWordChar(sc.ch) || ((sc.ch == '*' || sc.ch == '/') && !isgraph(sc.chPrev))) {
|
|
sc.SetState(SCE_APDL_WORD);
|
|
} else if (IsAnOperator(static_cast<char>(sc.ch))) {
|
|
sc.SetState(SCE_APDL_OPERATOR);
|
|
}
|
|
}
|
|
}
|
|
sc.Complete();
|
|
}
|
|
|
|
//------------------------------------------------------------------------------
|
|
// 06-27-07 Sergio Lucato
|
|
// - Included code folding for Ansys APDL lexer
|
|
// - Copyied from LexBasic.cxx and modified for APDL
|
|
//------------------------------------------------------------------------------
|
|
|
|
/* Bits:
|
|
* 1 - whitespace
|
|
* 2 - operator
|
|
* 4 - identifier
|
|
* 8 - decimal digit
|
|
* 16 - hex digit
|
|
* 32 - bin digit
|
|
*/
|
|
static int character_classification[128] =
|
|
{
|
|
0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 0, 0, 1, 0, 0,
|
|
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
|
|
1, 2, 0, 2, 2, 2, 2, 2, 2, 2, 6, 2, 2, 2, 10, 6,
|
|
60, 60, 28, 28, 28, 28, 28, 28, 28, 28, 2, 2, 2, 2, 2, 2,
|
|
2, 20, 20, 20, 20, 20, 20, 4, 4, 4, 4, 4, 4, 4, 4, 4,
|
|
4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 2, 2, 2, 2, 4,
|
|
2, 20, 20, 20, 20, 20, 20, 4, 4, 4, 4, 4, 4, 4, 4, 4,
|
|
4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 2, 2, 2, 2, 0
|
|
};
|
|
|
|
static bool IsSpace(int c) {
|
|
return c < 128 && (character_classification[c] & 1);
|
|
}
|
|
|
|
static bool IsIdentifier(int c) {
|
|
return c < 128 && (character_classification[c] & 4);
|
|
}
|
|
|
|
static int LowerCase(int c)
|
|
{
|
|
if (c >= 'A' && c <= 'Z')
|
|
return 'a' + c - 'A';
|
|
return c;
|
|
}
|
|
|
|
static int CheckAPDLFoldPoint(char const *token, int &level) {
|
|
if (!strcmp(token, "*if") ||
|
|
!strcmp(token, "*do") ||
|
|
!strcmp(token, "*dowhile") ) {
|
|
level |= SC_FOLDLEVELHEADERFLAG;
|
|
return 1;
|
|
}
|
|
if (!strcmp(token, "*endif") ||
|
|
!strcmp(token, "*enddo") ) {
|
|
return -1;
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
static void FoldAPDLDoc(Sci_PositionU startPos, Sci_Position length, int,
|
|
WordList *[], Accessor &styler) {
|
|
|
|
Sci_Position line = styler.GetLine(startPos);
|
|
int level = styler.LevelAt(line);
|
|
int go = 0, done = 0;
|
|
Sci_Position endPos = startPos + length;
|
|
char word[256];
|
|
int wordlen = 0;
|
|
Sci_Position i;
|
|
bool foldCompact = styler.GetPropertyInt("fold.compact", 1) != 0;
|
|
// Scan for tokens at the start of the line (they may include
|
|
// whitespace, for tokens like "End Function"
|
|
for (i = startPos; i < endPos; i++) {
|
|
int c = styler.SafeGetCharAt(i);
|
|
if (!done && !go) {
|
|
if (wordlen) { // are we scanning a token already?
|
|
word[wordlen] = static_cast<char>(LowerCase(c));
|
|
if (!IsIdentifier(c)) { // done with token
|
|
word[wordlen] = '\0';
|
|
go = CheckAPDLFoldPoint(word, level);
|
|
if (!go) {
|
|
// Treat any whitespace as single blank, for
|
|
// things like "End Function".
|
|
if (IsSpace(c) && IsIdentifier(word[wordlen - 1])) {
|
|
word[wordlen] = ' ';
|
|
if (wordlen < 255)
|
|
wordlen++;
|
|
}
|
|
else // done with this line
|
|
done = 1;
|
|
}
|
|
} else if (wordlen < 255) {
|
|
wordlen++;
|
|
}
|
|
} else { // start scanning at first non-whitespace character
|
|
if (!IsSpace(c)) {
|
|
if (IsIdentifier(c)) {
|
|
word[0] = static_cast<char>(LowerCase(c));
|
|
wordlen = 1;
|
|
} else // done with this line
|
|
done = 1;
|
|
}
|
|
}
|
|
}
|
|
if (c == '\n') { // line end
|
|
if (!done && wordlen == 0 && foldCompact) // line was only space
|
|
level |= SC_FOLDLEVELWHITEFLAG;
|
|
if (level != styler.LevelAt(line))
|
|
styler.SetLevel(line, level);
|
|
level += go;
|
|
line++;
|
|
// reset state
|
|
wordlen = 0;
|
|
level &= ~SC_FOLDLEVELHEADERFLAG;
|
|
level &= ~SC_FOLDLEVELWHITEFLAG;
|
|
go = 0;
|
|
done = 0;
|
|
}
|
|
}
|
|
}
|
|
|
|
static const char * const apdlWordListDesc[] = {
|
|
"processors",
|
|
"commands",
|
|
"slashommands",
|
|
"starcommands",
|
|
"arguments",
|
|
"functions",
|
|
0
|
|
};
|
|
|
|
extern const LexerModule lmAPDL(SCLEX_APDL, ColouriseAPDLDoc, "apdl", FoldAPDLDoc, apdlWordListDesc);
|