LIST *
var_expand(
LIST *prefix,
const char *in,
const char *end,
LOL *lol,
int cancopyin )
{
BUFFER buff;
const char *inp = in;
int depth;
size_t save_buffer_pos, ov_save_buffer_pos;
int literal = 0;
if( DEBUG_VAREXP )
printf( "expand '%.*s'\n", end - in, in );
/* This gets alot of cases: $(<) and $(>) */
if( end - in == 4 && in[0] == '$' && in[1] == leftParen && in[3] == rightParen )
{
switch( in[2] )
{
case '1':
case '<':
return list_copy( prefix, lol_get( lol, 0 ) );
case '2':
case '>':
return list_copy( prefix, lol_get( lol, 1 ) );
}
}
buffer_init( &buff );
/* Just try simple copy of in to out. */
while( in < end ) {
char ch = *in++;
buffer_addchar( &buff, ch );
if( ch == '$' && *in == leftParen )
goto expand;
#ifdef OPT_EXPAND_LITERALS_EXT
if( ch == '@' && *in == leftParen ) {
literal = 1;
goto expand;
}
if( ch == '@' && in[0] == '$' && in[1] == leftParen ) {
++in;
literal = 1;
goto expand;
}
#endif
}
/* No variables expanded - just add copy of input string to list. */
/* Cancopyin is an optimization: if the input was already a list */
/* item, we can use the copystr() to put it on the new list. */
/* Otherwise, we use the slower newstr(). */
buffer_putchar( &buff, 0 );
if( cancopyin ) {
LIST *new_list = list_append( prefix, inp, 1 );
buffer_free( &buff );
return new_list;
}
else {
LIST *new_list = list_append( prefix, buffer_ptr( &buff ), 0 );
buffer_free( &buff );
return new_list;
}
expand:
/*
* Input so far (ignore blanks):
*
* stuff-in-outbuf $(variable) remainder
* ^ ^
* in end
* Output so far:
*
* stuff-in-outbuf $
* ^ ^
* out_buf out
*
*
* We just copied the $ of $(...), so back up one on the output.
* We now find the matching close paren, copying the variable and
* modifiers between the $( and ) temporarily into out_buf, so that
* we can replace :'s with MAGIC_COLON. This is necessary to avoid
* being confused by modifier values that are variables containing
* :'s. Ugly.
*/
depth = 1;
buffer_deltapos( &buff, -1 );
save_buffer_pos = buffer_pos( &buff );
in++;
while( in < end && depth )
{
char ch = *in++;
buffer_addchar( &buff, ch );
if ( ch == leftParen )
{
depth++;
}
else if ( ch == rightParen )
{
depth--;
}
else
{
switch( ch )
{
case ':': buffer_deltapos( &buff, -1 ); buffer_addchar( &buff, MAGIC_COLON ); break;
case '[': buffer_deltapos( &buff, -1 ); buffer_addchar( &buff, MAGIC_LEFT ); break;
case ']': buffer_deltapos( &buff, -1 ); buffer_addchar( &buff, MAGIC_RIGHT ); break;
}
}
}
/* Copied ) - back up. */
buffer_deltapos( &buff, -1 );
ov_save_buffer_pos = buffer_pos( &buff );
buffer_setpos( &buff, save_buffer_pos );
/*
* Input so far (ignore blanks):
*
* stuff-in-outbuf $(variable) remainder
* ^ ^
* in end
* Output so far:
*
* stuff-in-outbuf variable
* ^ ^ ^
* out_buf out ov
*
* Later we will overwrite 'variable' in out_buf, but we'll be
* done with it by then. 'variable' may be a multi-element list,
* so may each value for '$(variable element)', and so may 'remainder'.
* Thus we produce a product of three lists.
*/
{
LIST *variables = 0;
LIST *remainder = 0;
LISTITEM *vars;
/* Recursively expand variable name & rest of input */
if( save_buffer_pos < ov_save_buffer_pos )
variables = var_expand( L0, buffer_posptr( &buff ), buffer_ptr( &buff ) + ov_save_buffer_pos, lol, 0 );
if( in < end )
remainder = var_expand( L0, in, end, lol, 0 );
/* Now produce the result chain */
/* For each variable name */
for( vars = list_first(variables); vars; vars = list_next( vars ) )
{
LIST *value, *evalue = 0;
LISTITEM* valueSliceStart = NULL;
#ifdef OPT_EXPAND_LITERALS_EXT
LIST *origvalue = 0;
#endif
char *colon;
char *bracket;
BUFFER varnamebuff;
int sub1 = 0, sub2 = -1;
VAR_EDITS edits;
memset(&edits, 0, sizeof(VAR_EDITS));
if (leftParen == '{') {
edits.empty.ptr = "";
edits.empty.len = 0;
}
/* Look for a : modifier in the variable name */
/* Must copy into varname so we can modify it */
buffer_init( &varnamebuff );
buffer_addstring( &varnamebuff, list_value(vars), strlen( list_value(vars) ) );
buffer_addchar( &varnamebuff, 0 );
if( ( colon = strchr( buffer_ptr( &varnamebuff ), MAGIC_COLON ) ) )
{
*colon = '\0';
var_edit_parse( colon + 1, &edits );
}
/* Look for [x-y] and [x-] subscripting */
/* sub1 is x (0 default) */
/* sub2 is length (-1 means forever) */
if( ( bracket = strchr( buffer_ptr( &varnamebuff ), MAGIC_LEFT ) ) )
{
char *dash;
if( ( dash = strchr( bracket + 1, '-' ) ) )
*dash = '\0';
sub1 = atoi( bracket + 1 ) - 1;
if( !dash )
sub2 = 1;
else if( !dash[1] || dash[1] == MAGIC_RIGHT )
sub2 = -1;
else
sub2 = atoi( dash + 1 ) - sub1;
*bracket = '\0';
}
/* Get variable value, specially handling $(<), $(>), $(n) */
#ifdef OPT_EXPAND_LITERALS_EXT
if ( !literal )
#endif
{
const char* varname = buffer_ptr( &varnamebuff );
if( varname[0] == '<' && !varname[1] )
value = lol_get( lol, 0 );
else if( varname[0] == '>' && !varname[1] )
value = lol_get( lol, 1 );
else if( varname[0] >= '1' && varname[0] <= '9' && !varname[1] )
value = lol_get( lol, varname[0] - '1' );
else if ( edits.targetsetting ) {
TARGET* t = bindtarget(edits.targetname.ptr);
SETTINGS* settings = quicksettingslookup(t, varname);
if (settings)
value = list_copy(L0, settings->value);
else
value = L0;
} else
value = var_get( varname );
}
#ifdef OPT_EXPAND_LITERALS_EXT
else {
origvalue = value = list_append( L0, buffer_ptr( &varnamebuff ), 0 );
}
#endif
/* The fast path: $(x) - just copy the variable value. */
/* This is only an optimization */
if( buffer_isempty( &buff ) && !bracket && !colon && in == end )
{
prefix = list_copy( prefix, value );
buffer_free( &buff );
continue;
}
/* Handle start subscript */
valueSliceStart = list_first(value);
while(sub1 > 0 && valueSliceStart)
{
sub1 -= 1;
valueSliceStart = list_next(valueSliceStart);
}
/* Empty w/ :E=default? */
if( !valueSliceStart && (colon || leftParen == '{') && edits.empty.ptr ) {
evalue = value = list_append( L0, edits.empty.ptr, 0 );
valueSliceStart = list_first(value);
}
#ifdef OPT_EXPAND_LITERALS_EXT
if ( colon && edits.expandliteral ) {
LOL lol;
char const* string = list_value(list_first(value));
LIST *newvalue = var_expand( L0, string, string + strlen( string ), &lol, 0 );
if ( origvalue ) {
list_free( origvalue );
origvalue = 0;
}
value = newvalue;
valueSliceStart = list_first(value);
sub2 = -1;
}
#endif
#ifdef OPT_EXPAND_FILEGLOB_EXT
if ( edits.wildcard ) {
LIST *newl = L0;
for( ; valueSliceStart; valueSliceStart = list_next( valueSliceStart ) ) {
LIST *foundfiles = L0;
fileglob* glob;
/* Handle end subscript (length actually) */
if( sub2 >= 0 && --sub2 < 0 )
break;
glob = fileglob_Create( list_value(valueSliceStart) );
while ( fileglob_Next( glob ) ) {
foundfiles = list_append( foundfiles, fileglob_FileName( glob ) + edits.wildcard_remove_prepend.len, 0 );
}
fileglob_Destroy( glob );
/* TODO: Efficiency: Just append to newl above? */
newl = list_copy( newl, foundfiles );
list_free( foundfiles );
}
if ( origvalue ) {
list_free( origvalue );
origvalue = 0;
}
value = newl;
origvalue = value;
valueSliceStart = list_first(value);
}
#endif
/* For each variable value */
for( ; valueSliceStart; valueSliceStart = list_next( valueSliceStart ) )
{
LISTITEM *rem;
size_t save_buffer_pos;
size_t end_buffer_pos;
const char *valuestring;
/* Handle end subscript (length actually) */
if( sub2 >= 0 && --sub2 < 0 )
break;
/* Apply : mods, if present */
save_buffer_pos = buffer_pos( &buff );
valuestring = list_value(valueSliceStart);
#ifdef OPT_EXPAND_BINDING_EXT
if( colon && edits.expandbinding ) {
SETTINGS *expandText;
TARGET *t = bindtarget( valuestring );
expandText = quicksettingslookup( t, "EXPAND_TEXT" );
if ( expandText && list_first(expandText->value) ) {
valuestring = list_value(list_first(expandText->value));
} else {
if( t->binding == T_BIND_UNBOUND ) {
t->boundname = search_using_target_settings( t, t->name, &t->time );
t->binding = t->time ? T_BIND_EXISTS : T_BIND_MISSING;
}
valuestring = t->boundname;
}
}
#endif
if( colon && edits.filemods ) {
var_edit_file( valuestring, &buff, &edits );
} else {
buffer_addstring( &buff, valuestring, strlen( valuestring ) + 1 );
}
buffer_setpos( &buff, save_buffer_pos );
if( colon && ( edits.upshift || edits.downshift ) )
var_edit_shift( buffer_posptr( &buff ), &edits );
#ifdef OPT_SLASH_MODIFIERS_EXT
if( colon && ( edits.fslash || edits.bslash ) )
var_edit_slash( buffer_posptr( &buff ), &edits );
#endif
#ifdef OPT_EXPAND_ESCAPE_PATH_EXT
if ( colon && edits.escapepath )
{
const char* ptr = buffer_posptr( &buff );
const char* endptr = ptr + strlen( ptr );
BUFFER escapebuff;
buffer_init( &escapebuff );
save_buffer_pos = buffer_pos( &buff );
#ifdef NT
while ( ptr != endptr && *ptr != ' ' && *ptr != '/' )
++ptr;
if (*ptr == ' ' || *ptr == '/' ) {
buffer_addchar( &escapebuff, '"' );
buffer_addstring( &escapebuff, buffer_posptr( &buff ), endptr - buffer_posptr( &buff ) );
buffer_addchar( &escapebuff, '"' );
buffer_addchar( &escapebuff, 0 );
buffer_addstring( &buff, buffer_ptr( &escapebuff ), buffer_pos( &escapebuff ) );
}
#else
while ( ptr != endptr ) {
if ( *ptr == ' ' || *ptr == '\\' || *ptr == leftParen || *ptr == rightParen || *ptr == '"' ) {
buffer_addchar( &escapebuff, '\\' );
}
buffer_addchar( &escapebuff, *ptr );
++ptr;
}
buffer_addchar( &escapebuff, 0 );
buffer_addstring( &buff, buffer_ptr( &escapebuff ), buffer_pos( &escapebuff ) );
#endif
buffer_setpos( &buff, save_buffer_pos );
buffer_free( &escapebuff );
}
#endif
/* Handle :J=joinval */
/* If we have more values for this var, just */
/* keep appending them (with the join value) */
/* rather than creating separate LIST elements. */
if( colon && edits.join.ptr &&
( list_next( valueSliceStart ) || list_next( vars ) ) )
{
buffer_setpos( &buff, buffer_pos( &buff ) + strlen( buffer_posptr( &buff ) ) );
buffer_addstring( &buff, edits.join.ptr, strlen( edits.join.ptr ) + 1 );
buffer_deltapos( &buff, -1 );
continue;
}
/* If no remainder, append result to output chain. */
if( in == end )
{
prefix = list_append( prefix, buffer_ptr( &buff ), 0 );
continue;
}
/* For each remainder, append the complete string */
/* to the output chain. */
/* Remember the end of the variable expansion so */
/* we can just tack on each instance of 'remainder' */
save_buffer_pos = buffer_pos( &buff );
end_buffer_pos = strlen( buffer_ptr( &buff ) );
buffer_setpos( &buff, end_buffer_pos );
for( rem = list_first(remainder); rem; rem = list_next( rem ) )
{
buffer_addstring( &buff, list_value(rem), strlen( list_value(rem) ) + 1 );
buffer_setpos( &buff, end_buffer_pos );
prefix = list_append( prefix, buffer_ptr( &buff ), 0 );
}
buffer_setpos( &buff, save_buffer_pos );
}
/* Toss used empty */
if( evalue )
list_free( evalue );
#ifdef OPT_EXPAND_LITERALS_EXT
if ( origvalue )
list_free( origvalue );
#endif
#ifdef OPT_EXPAND_INCLUDES_EXCLUDES_EXT
if ( edits.includes_excludes ) {
LIST *newl = L0;
LISTITEM* l;
LIST *origprefix = prefix;
int hasInclude = 0;
if ( !regexhash )
regexhash = hashinit( sizeof(regexdata), "regex" );
{
LISTITEM *inex = list_first(edits.includes_excludes);
while ( inex ) {
char mod = list_value(inex)[0];
inex = list_next( inex );
if ( mod == 'I' ) {
hasInclude = 1;
}
}
}
for (l = list_first(prefix) ; l; l = list_next( l ) )
{
LISTITEM *inex = list_first(edits.includes_excludes);
int remove = hasInclude;
while ( inex ) {
char mod = list_value(inex)[0];
regexp *re;
regexdata data, *d = &data;
inex = list_next( inex );
data.name = list_value(inex);
if( !hashcheck( regexhash, (HASHDATA **)&d ) )
{
d->re = jam_regcomp( list_value(inex) );
(void)hashenter( regexhash, (HASHDATA **)&d );
}
re = d->re;
inex = list_next( inex );
if ( mod == 'X' ) {
if( jam_regexec( re, list_value(l) ) )
remove = 1;
} else if ( mod == 'I' ) {
if( jam_regexec( re, list_value(l) ) )
remove = 0;
}
}
if ( !remove )
newl = list_append( newl, list_value(l), 1 );
}
/* TODO: Efficiency: Just modify prefix? */
list_free( origprefix );
prefix = newl;
}
#endif
//#ifdef OPT_EXPAND_LITERALS_EXT
// buffer_free( &buff );
//#endif
#ifdef OPT_EXPAND_INCLUDES_EXCLUDES_EXT
list_free( edits.includes_excludes );
#endif
}
/* variables & remainder were gifts from var_expand */
/* and must be freed */
list_free( variables );
list_free( remainder );
if( DEBUG_VAREXP )
{
printf( "expanded to " );
list_print( prefix );
printf( "\n" );
}
buffer_free( &buff );
return prefix;
}
}
int _fileglob_GlobHelper(fileglob* self, const char* inPattern) {
fileglob_Context* context = self->context;
int hasWildcard;
int found;
Setup:
if (!context) {
context = (fileglob_Context*)self->allocFunction(self->userData, NULL, sizeof(fileglob_Context));
context->prev = self->context;
#if defined(WIN32)
context->handle = INVALID_HANDLE_VALUE;
#else
context->dirp = NULL;
context->hasattr = 0;
context->statted = 0;
#endif
context->pattern = NULL;
context->iterNode = NULL;
context->directoryListHead = context->directoryListTail = 0;
context->basePathLastSlashPos = 0;
buffer_initwithalloc(&context->patternBuf, self->allocFunction, self->userData);
buffer_addstring(&context->patternBuf, inPattern, strlen(inPattern) + 1);
buffer_initwithalloc(&context->basePath, self->allocFunction, self->userData);
buffer_initwithalloc(&context->matchPattern, self->allocFunction, self->userData);
self->context = context;
if (context->prev == NULL)
return 1;
}
DoRecursion:
found = 1;
if (!context->pattern) {
char* pattern;
context->basePathEndPos = context->basePathLastSlashPos = 0;
context->recurseAtPos = (size_t)-1;
// Split the path into base path and pattern to match against.
hasWildcard = 0;
for (pattern = buffer_ptr(&context->patternBuf); *pattern != '\0'; ++pattern) {
char ch = *pattern;
// Is it a '?' ?
if (ch == '?')
hasWildcard = 1;
// Is it a '*' ?
else if (ch == '*') {
hasWildcard = 1;
// Is there a '**'?
if (pattern[1] == '*') {
// If we're just starting the pattern or the characters immediately
// preceding the pattern are a drive letter ':' or a directory path
// '/', then set up the internals for later recursion.
if (pattern == buffer_ptr(&context->patternBuf) || pattern[-1] == '/' || pattern[-1] == ':') {
char ch2 = pattern[2];
if (ch2 == '/') {
context->recurseAtPos = pattern - buffer_ptr(&context->patternBuf);
memcpy(pattern, pattern + 3, strlen(pattern) - 2);
buffer_deltapos(&context->patternBuf, -3);
} else if (ch2 == '\0') {
context->recurseAtPos = pattern - buffer_ptr(&context->patternBuf);
*pattern = '\0';
}
}
}
}
// Is there a '/' or ':' in the pattern at this location?
if (ch == '/' || ch == ':') {
if (hasWildcard)
break;
else {
if (pattern[1])
context->basePathLastSlashPos = pattern - buffer_ptr(&context->patternBuf) + 1;
context->basePathEndPos = pattern - buffer_ptr(&context->patternBuf) + 1;
}
}
}
context->pattern = pattern;
// Copy the directory down.
context->basePathLen = context->basePathEndPos;
buffer_reset(&context->basePath);
buffer_addstring(&context->basePath, buffer_ptr(&context->patternBuf), context->basePathLen);
buffer_addchar(&context->basePath, 0);
if (context->iterNode) {
context->matchFiles = *context->pattern == 0;
goto NextDirectory;
}
}
#if defined(WIN32)
if (context->handle == INVALID_HANDLE_VALUE) {
#else
if (!context->dirp && !context->statted) {
#endif
size_t matchLen;
// Did we make it to the end of the pattern? If so, we should match files,
// since there were no slashes encountered.
context->matchFiles = *context->pattern == 0;
// Copy the wildcard matching string.
matchLen = (context->pattern - buffer_ptr(&context->patternBuf)) - context->basePathLen;
buffer_reset(&context->matchPattern);
buffer_addstring(&context->matchPattern, buffer_ptr(&context->patternBuf) + context->basePathLen, matchLen + 1);
buffer_deltapos(&context->matchPattern, -1);
if (*buffer_posptr(&context->matchPattern) == '/') {
buffer_deltapos(&context->matchPattern, 1);
buffer_addchar(&context->matchPattern, 0);
}
#if defined(WIN32)
// Do the file search with *.* in the directory specified in basePattern.
buffer_setpos(&context->basePath, context->basePathEndPos);
buffer_addstring(&context->basePath, "*.*", 4);
// Start the find.
context->handle = FindFirstFile(buffer_ptr(&context->basePath), &context->fd);
if (context->handle == INVALID_HANDLE_VALUE) {
found = 0;
}
#else
// Start the find.
buffer_setpos(&context->basePath, context->basePathEndPos);
buffer_addchar(&context->basePath, 0);
context->dirp = opendir(buffer_ptr(&context->basePath)[0] ? buffer_ptr(&context->basePath) : ".");
if (!context->dirp) {
found = 0;
} else {
context->dp = readdir(context->dirp);
found = context->dp != NULL;
}
#endif
// Clear out the *.* so we can use the original basePattern string.
buffer_setpos(&context->basePath, context->basePathEndPos);
buffer_putchar(&context->basePath, 0);
} else {
goto NextFile;
}
// Any files found?
#if defined(WIN32)
if (context->handle != INVALID_HANDLE_VALUE) {
#else
if (context->dirp) {
#endif
for (;;) {
#if defined(WIN32)
char* filename = context->fd.cFileName;
#else
char* filename = context->dp->d_name;
context->hasattr = 0;
#endif
// Is the file a directory?
#if defined(WIN32)
if (context->fd.dwFileAttributes & FILE_ATTRIBUTE_DIRECTORY) {
#else
if (context->dp->d_type == DT_DIR) {
#endif
// Knock out "." or ".."
int ignore = filename[0] == '.' &&
(filename[1] == 0 ||
(filename[1] == '.' && filename[2] == 0));
// Should this file be ignored?
int matches = 0;
if (!ignore) {
size_t len = strlen(filename);
filename[len] = '/';
filename[len + 1] = '\0';
matches = fileglob_WildMatch(buffer_ptr(&context->matchPattern), filename, 0);
}
// Do a wildcard match.
if (!ignore && matches) {
// It matched. Let's see if the file should be ignored.
// See if this is a directory to ignore.
ignore = _fileglob_MatchIgnoreDirectoryPattern(self, filename);
// Should this file be ignored?
if (!ignore) {
_fileglob_list_append(self, &context->directoryListHead, &context->directoryListTail, filename);
// Is this pattern exclusive?
if (self->exclusiveDirectoryPatternsHead) {
if (_fileglob_MatchExclusiveDirectoryPattern(self, filename))
break;
} else {
if ((!context->matchFiles && context->pattern[0] == '/' && context->pattern[1] == 0)
|| (self->filesAndFolders))
break;
}
}
}
} else {
// Do a wildcard match.
if (fileglob_WildMatch(buffer_ptr(&context->matchPattern), filename, 0)) {
// It matched. Let's see if the file should be ignored.
int ignore = _fileglob_MatchIgnoreFilePattern(self, filename);
// Is this pattern exclusive?
if (!ignore && self->exclusiveFilePatternsHead) {
ignore = !_fileglob_MatchExclusiveFilePattern(self, filename);
}
// Should this file be ignored?
if (!ignore) {
if (context->matchFiles)
break;
}
}
}
NextFile:
// Look up the next file.
#if defined(WIN32)
found = FindNextFile(context->handle, &context->fd) == TRUE;
#else
if (context->dirp) {
context->dp = readdir(context->dirp);
found = context->dp != NULL;
} else {
found = 0;
}
#endif
if (!found)
break;
}
if (!found) {
// Close down the file find handle.
#if defined(WIN32)
FindClose(context->handle);
context->handle = INVALID_HANDLE_VALUE;
#else
if (context->dirp) {
closedir(context->dirp);
context->dirp = NULL;
}
#endif
context->iterNode = context->directoryListHead;
}
}
// Iterate the file list and either recurse or add the file as a found
// file.
if (!context->matchFiles) {
if (found) {
return 1;
}
NextDirectory:
if (context->iterNode) {
// Need more directories.
SplicePath(&self->combinedName, buffer_ptr(&context->basePath), context->iterNode->buffer);
buffer_deltapos(&self->combinedName, -2);
buffer_addstring(&self->combinedName, context->pattern, strlen(context->pattern) + 1);
context->iterNode = context->iterNode->next;
context = NULL;
inPattern = buffer_ptr(&self->combinedName);
goto Setup;
}
} else {
if (found)
return 1;
}
// Do we need to recurse?
if (context->recurseAtPos == (size_t)-1) {
_fileglob_FreeContextLevel(self);
context = self->context;
if (!context)
return 0;
goto NextDirectory;
}
buffer_reset(&context->matchPattern);
buffer_setpos(&context->patternBuf, context->recurseAtPos);
buffer_addstring(&context->matchPattern, buffer_posptr(&context->patternBuf), strlen(buffer_posptr(&context->patternBuf)));
buffer_addstring(&context->patternBuf, "*/**/", 5);
buffer_addstring(&context->patternBuf, buffer_ptr(&context->matchPattern), buffer_pos(&context->matchPattern) + 1);
inPattern = buffer_ptr(&context->patternBuf);
context->pattern = NULL;
if (context->matchFiles) {
context->iterNode = context->directoryListHead;
} else {
_fileglob_list_clear(self, &context->directoryListHead, &context->directoryListTail);
}
goto DoRecursion;
}
/* Lexer FSM main loop */
TokenList *lexer_split(const char *src)
{
LexerState state = ST_WHITESPACE;
const size_t length = strlen(src);
Buffer *tokens = buffer_new();
size_t i;
/* terminating zero is handled like a normal character */
for (i=0; state!=ST_ERROR && i<length+1; i++)
{
const char c = src[i];
switch (state)
{
case ST_WHITESPACE:
if (c=='"') /* start a quoted string, new token */
state = ST_QUOTE;
else if (c=='\\') /* start escape-sequence, new token */
state = ST_ESCAPE;
else if (c!='\0' && !isspace(c)) /* start a new token */
{
state = ST_WORD;
buffer_putchar(tokens, c);
}
/* else: whitespace -> ignore */
break;
case ST_QUOTE:
if (c=='"') /* end a quoted string */
state = ST_WORD;
else if (c=='\\') /* start escape-sequence */
state = ST_QUOTE_ESCAPE;
else if (c!='\0') /* append a char from inside quotes to token */
buffer_putchar(tokens, c);
else /* '\0' -> set error flag */
state = ST_ERROR;
break;
case ST_WORD:
if (c=='\0' || isspace(c)) /* end a token */
{
state = ST_WHITESPACE;
buffer_putchar(tokens, '\0');
}
else if (c=='"') /* start a quoted string */
state = ST_QUOTE;
else if (c=='\\') /* start escape-sequence */
state = ST_ESCAPE;
else /* append a char to token */
buffer_putchar(tokens, c);
break;
case ST_ESCAPE:
if (c!='\0')
{
state = ST_WORD;
buffer_putchar(tokens, c);
}
else /* '\0' -> set error flag */
state = ST_ERROR;
break;
case ST_QUOTE_ESCAPE:
if (c!='\0')
{
state = ST_QUOTE;
buffer_putchar(tokens, c);
}
else /* '\0' -> set error flag */
state = ST_ERROR;
break;
default:
break;
}
}
/* Detect lexing errors */
if (state == ST_ERROR)
{
buffer_delete(tokens);
return NULL;
}
else
{
TokenList *tl = (TokenList *) malloc(sizeof(TokenList));
tl->buf = tokens;
tl->tokens = make_token_list(tokens);
return tl;
}
}
/* Lexer FSM main loop */
Buffer *lexer_split(const char *src, Diagnostic *diag)
{
LexerState state = ST_WHITESPACE;
LexerState prev_state;
const size_t length = strlen(src);
Buffer *tokens = buffer_alloc();
size_t i;
char op = 0;
/* terminating zero is handled like a normal character */
for (i=0; state!=ST_ERROR && i<length+1; i++)
{
const char c = src[i];
prev_state = state;
switch (state)
{
case ST_OPERATOR:
state = ST_WHITESPACE;
if (c==op)
{
/* double operator */
buffer_putchar(tokens, op);
buffer_putchar(tokens, op);
buffer_putchar(tokens, '\0');
break;
}
else
{
/* single operator */
buffer_putchar(tokens, op);
buffer_putchar(tokens, '\0');
}
/* no break (intended) -- fall back to whitespace case */
case ST_WHITESPACE:
if (c=='"') /* start a quoted string, new token */
state = ST_QUOTE;
else if (c=='\\') /* start escape-sequence, new token */
state = ST_ESCAPE;
else if (is_simple_operator(c))
{
/* a new token for operator */
buffer_putchar(tokens, c);
buffer_putchar(tokens, '\0');
}
else if (is_double_operator(c))
{
state = ST_OPERATOR;
op = c;
}
else if (c!='\0' && !isspace(c)) /* start a new token */
{
state = ST_WORD;
buffer_putchar(tokens, c);
}
/* else: whitespace -> ignore */
break;
case ST_QUOTE:
if (c=='"') /* end a quoted string */
state = ST_WORD;
else if (c=='\\') /* start escape-sequence */
state = ST_QUOTE_ESCAPE;
else if (c!='\0') /* append a char from inside quotes to token */
buffer_putchar(tokens, c);
else /* '\0' -> set error flag */
state = ST_ERROR;
break;
case ST_WORD:
if (c=='\0' || isspace(c)) /* end a token */
{
state = ST_WHITESPACE;
buffer_putchar(tokens, '\0');
}
else if (c=='"') /* start a quoted string */
state = ST_QUOTE;
else if (c=='\\') /* start escape-sequence */
state = ST_ESCAPE;
else if (is_simple_operator(c))
{
state = ST_WHITESPACE;
/* a new token for operator */
buffer_putchar(tokens, '\0');
buffer_putchar(tokens, c);
buffer_putchar(tokens, '\0');
}
else if (is_double_operator(c))
{
state = ST_OPERATOR;
buffer_putchar(tokens, '\0');
op = c;
}
else /* append a char to token */
buffer_putchar(tokens, c);
break;
case ST_ESCAPE:
if (c!='\0')
{
state = ST_WORD;
buffer_putchar(tokens, c);
}
else /* '\0' -> set error flag */
{
state = ST_ERROR;
}
break;
case ST_QUOTE_ESCAPE:
if (c!='\0')
{
state = ST_QUOTE;
buffer_putchar(tokens, c);
}
else /* '\0' -> set error flag */
state = ST_ERROR;
break;
default:
break;
}
}
/* Detect lexing errors */
if (state == ST_ERROR)
{
buffer_free(tokens);
diag->error = 1;
switch (prev_state)
{
case ST_QUOTE:
diag->error_message = "Unfinished quote string";
break;
case ST_ESCAPE:
case ST_QUOTE_ESCAPE:
diag->error_message = "Unfinished escape sequence";
break;
default:
break;
}
return NULL;
}
else
{
diag->error = 0;
return tokens;
}
}