void align_struct_initializers(void)
{
LOG_FUNC_ENTRY();
chunk_t *pc = chunk_get_head();
while (pc != nullptr)
{
chunk_t *prev = chunk_get_prev_ncnl(pc);
if ( chunk_is_token(prev, CT_ASSIGN)
&& ( chunk_is_token(pc, CT_BRACE_OPEN)
|| (language_is_set(LANG_D) && chunk_is_token(pc, CT_SQUARE_OPEN))))
{
align_init_brace(pc);
}
pc = chunk_get_next_type(pc, CT_BRACE_OPEN, -1);
}
} // align_struct_initializers
void align_add(ChunkStack &cs, chunk_t *pc, size_t &max_col)
{
LOG_FUNC_ENTRY();
size_t min_col;
chunk_t *prev = chunk_get_prev(pc);
if (prev == nullptr || chunk_is_newline(prev))
{
min_col = 1;
LOG_FMT(LALADD, "%s(%d): pc->orig_line=%zu, pc->col=%zu max_col=%zu min_col=%zu\n",
__func__, __LINE__, pc->orig_line, pc->column, max_col, min_col);
}
else
{
if (chunk_is_token(prev, CT_COMMENT_MULTI))
{
min_col = prev->orig_col_end + 1;
}
else
{
min_col = prev->column + prev->len() + 1;
}
LOG_FMT(LALADD, "%s(%d): pc->orig_line=%zu, pc->col=%zu max_col=%zu min_col=%zu multi:%s prev->col=%zu prev->len()=%zu %s\n",
__func__, __LINE__, pc->orig_line, pc->column, max_col, min_col, (chunk_is_token(prev, CT_COMMENT_MULTI)) ? "Y" : "N",
(chunk_is_token(prev, CT_COMMENT_MULTI)) ? prev->orig_col_end : (UINT32)prev->column, prev->len(), get_token_name(prev->type));
}
if (cs.Empty())
{
max_col = 0;
}
cs.Push_Back(pc);
if (min_col > max_col)
{
max_col = min_col;
}
} // align_add
void tokenize_cleanup(void)
{
LOG_FUNC_ENTRY();
chunk_t *pc = chunk_get_head();
chunk_t *prev = NULL;
chunk_t *next;
chunk_t *tmp;
chunk_t *tmp2;
bool in_type_cast = false;
cpd.unc_stage = US_TOKENIZE_CLEANUP;
/* Since [] is expected to be TSQUARE for the 'operator', we need to make
* this change in the first pass.
*/
for (pc = chunk_get_head(); pc != NULL; pc = chunk_get_next_ncnl(pc))
{
if (pc->type == CT_SQUARE_OPEN)
{
next = chunk_get_next_ncnl(pc);
if (chunk_is_token(next, CT_SQUARE_CLOSE))
{
/* Change '[' + ']' into '[]' */
set_chunk_type(pc, CT_TSQUARE);
pc->str = "[]";
// bug # 664
// The original orig_col_end of CT_SQUARE_CLOSE is stored at orig_col_end of CT_TSQUARE.
// pc->orig_col_end += 1;
pc->orig_col_end = next->orig_col_end;
chunk_del(next);
}
}
if ((pc->type == CT_SEMICOLON) &&
(pc->flags & PCF_IN_PREPROC) &&
!chunk_get_next_ncnl(pc, CNAV_PREPROC))
{
LOG_FMT(LNOTE, "%s:%d Detected a macro that ends with a semicolon. Possible failures if used.\n",
cpd.filename, pc->orig_line);
}
}
/* We can handle everything else in the second pass */
pc = chunk_get_head();
next = chunk_get_next_ncnl(pc);
while ((pc != NULL) && (next != NULL))
{
if ((pc->type == CT_DOT) && (cpd.lang_flags & LANG_ALLC))
{
set_chunk_type(pc, CT_MEMBER);
}
if ((pc->type == CT_NULLCOND) && (cpd.lang_flags & LANG_CS))
{
set_chunk_type(pc, CT_MEMBER);
}
/* Determine the version stuff (D only) */
if (pc->type == CT_D_VERSION)
{
if (next->type == CT_PAREN_OPEN)
{
set_chunk_type(pc, CT_D_VERSION_IF);
}
else
{
if (next->type != CT_ASSIGN)
{
LOG_FMT(LERR, "%s:%d %s: version: Unexpected token %s\n",
cpd.filename, pc->orig_line, __func__, get_token_name(next->type));
cpd.error_count++;
}
set_chunk_type(pc, CT_WORD);
}
}
/* Determine the scope stuff (D only) */
if (pc->type == CT_D_SCOPE)
{
if (next->type == CT_PAREN_OPEN)
{
set_chunk_type(pc, CT_D_SCOPE_IF);
}
else
{
set_chunk_type(pc, CT_TYPE);
}
}
/**
* Change CT_BASE before CT_PAREN_OPEN to CT_WORD.
* public myclass() : base() {
* }
*/
if ((pc->type == CT_BASE) && (next->type == CT_PAREN_OPEN))
{
set_chunk_type(pc, CT_WORD);
}
if ((pc->type == CT_ENUM) && (next->type == CT_CLASS))
{
set_chunk_type(next, CT_ENUM_CLASS);
}
/**
* Change CT_WORD after CT_ENUM, CT_UNION, or CT_STRUCT to CT_TYPE
* Change CT_WORD before CT_WORD to CT_TYPE
*/
if (next->type == CT_WORD)
{
if ((pc->type == CT_ENUM) ||
(pc->type == CT_ENUM_CLASS) ||
(pc->type == CT_UNION) ||
(pc->type == CT_STRUCT))
{
set_chunk_type(next, CT_TYPE);
}
if (pc->type == CT_WORD)
{
set_chunk_type(pc, CT_TYPE);
}
}
/* change extern to qualifier if extern isn't followed by a string or
* an open paren
*/
if (pc->type == CT_EXTERN)
{
if (next->type == CT_STRING)
{
/* Probably 'extern "C"' */
}
else if (next->type == CT_PAREN_OPEN)
{
/* Probably 'extern (C)' */
}
else
{
/* Something else followed by a open brace */
tmp = chunk_get_next_ncnl(next);
if ((tmp == NULL) || (tmp->type != CT_BRACE_OPEN))
{
set_chunk_type(pc, CT_QUALIFIER);
}
}
}
/**
* Change CT_STAR to CT_PTR_TYPE if preceded by CT_TYPE,
* CT_QUALIFIER, or CT_PTR_TYPE.
*/
if ((next->type == CT_STAR) &&
((pc->type == CT_TYPE) ||
(pc->type == CT_QUALIFIER) ||
(pc->type == CT_PTR_TYPE)))
{
set_chunk_type(next, CT_PTR_TYPE);
}
if ((pc->type == CT_TYPE_CAST) &&
(next->type == CT_ANGLE_OPEN))
{
set_chunk_parent(next, CT_TYPE_CAST);
in_type_cast = true;
}
/**
* Change angle open/close to CT_COMPARE, if not a template thingy
*/
if ((pc->type == CT_ANGLE_OPEN) && (pc->parent_type != CT_TYPE_CAST))
{
/* pretty much all languages except C use <> for something other than
* comparisons. "#include<xxx>" is handled elsewhere.
*/
if (cpd.lang_flags & (LANG_CPP | LANG_CS | LANG_JAVA | LANG_VALA | LANG_OC))
{
// bug #663
check_template(pc);
}
else
{
/* convert CT_ANGLE_OPEN to CT_COMPARE */
set_chunk_type(pc, CT_COMPARE);
}
}
if ((pc->type == CT_ANGLE_CLOSE) && (pc->parent_type != CT_TEMPLATE))
{
if (in_type_cast)
{
in_type_cast = false;
set_chunk_parent(pc, CT_TYPE_CAST);
}
else
{
next = handle_double_angle_close(pc);
}
}
if (cpd.lang_flags & LANG_D)
{
/* Check for the D string concat symbol '~' */
if ((pc->type == CT_INV) &&
((prev->type == CT_STRING) ||
(prev->type == CT_WORD) ||
(next->type == CT_STRING)))
{
set_chunk_type(pc, CT_CONCAT);
}
/* Check for the D template symbol '!' (word + '!' + word or '(') */
if ((pc->type == CT_NOT) &&
(prev->type == CT_WORD) &&
((next->type == CT_PAREN_OPEN) ||
(next->type == CT_WORD) ||
(next->type == CT_TYPE)))
{
set_chunk_type(pc, CT_D_TEMPLATE);
}
/* handle "version(unittest) { }" vs "unittest { }" */
if (prev && (pc->type == CT_UNITTEST) && (prev->type == CT_PAREN_OPEN))
{
set_chunk_type(pc, CT_WORD);
}
/* handle 'static if' and merge the tokens */
if (prev && (pc->type == CT_IF) && chunk_is_str(prev, "static", 6))
{
/* delete PREV and merge with IF */
pc->str.insert(0, ' ');
pc->str.insert(0, prev->str);
pc->orig_col = prev->orig_col;
pc->orig_line = prev->orig_line;
chunk_t *to_be_deleted = prev;
prev = chunk_get_prev_ncnl(prev);
chunk_del(to_be_deleted);
}
}
if (cpd.lang_flags & LANG_CPP)
{
/* Change Word before '::' into a type */
if ((pc->type == CT_WORD) && (next->type == CT_DC_MEMBER))
{
set_chunk_type(pc, CT_TYPE);
}
}
/* Change get/set to CT_WORD if not followed by a brace open */
if ((pc->type == CT_GETSET) && (next->type != CT_BRACE_OPEN))
{
if ((next->type == CT_SEMICOLON) &&
((prev->type == CT_BRACE_CLOSE) ||
(prev->type == CT_BRACE_OPEN) ||
(prev->type == CT_SEMICOLON)))
{
set_chunk_type(pc, CT_GETSET_EMPTY);
set_chunk_parent(next, CT_GETSET);
}
else
{
set_chunk_type(pc, CT_WORD);
}
}
/* Interface is only a keyword in MS land if followed by 'class' or 'struct'
* likewise, 'class' may be a member name in Java.
*/
if ((pc->type == CT_CLASS) && !CharTable::IsKw1(next->str[0]))
{
set_chunk_type(pc, CT_WORD);
}
/* Change item after operator (>=, ==, etc) to a CT_OPERATOR_VAL
* Usually the next item is part of the operator.
* In a few cases the next few tokens are part of it:
* operator + - common case
* operator >> - need to combine '>' and '>'
* operator ()
* operator [] - already converted to TSQUARE
* operator new []
* operator delete []
* operator const char *
* operator const B&
* operator std::allocator<U>
*
* In all cases except the last, this will put the entire operator value
* in one chunk.
*/
if (pc->type == CT_OPERATOR)
{
tmp2 = chunk_get_next(next);
/* Handle special case of () operator -- [] already handled */
if (next->type == CT_PAREN_OPEN)
{
tmp = chunk_get_next(next);
if ((tmp != NULL) && (tmp->type == CT_PAREN_CLOSE))
{
next->str = "()";
set_chunk_type(next, CT_OPERATOR_VAL);
chunk_del(tmp);
next->orig_col_end += 1;
}
}
else if ((next->type == CT_ANGLE_CLOSE) &&
tmp2 && (tmp2->type == CT_ANGLE_CLOSE) &&
(tmp2->orig_col == next->orig_col_end))
{
next->str.append('>');
next->orig_col_end++;
set_chunk_type(next, CT_OPERATOR_VAL);
chunk_del(tmp2);
}
else if (next->flags & PCF_PUNCTUATOR)
{
set_chunk_type(next, CT_OPERATOR_VAL);
}
else
{
set_chunk_type(next, CT_TYPE);
/* Replace next with a collection of all tokens that are part of
* the type.
*/
tmp2 = next;
while ((tmp = chunk_get_next(tmp2)) != NULL)
{
if ((tmp->type != CT_WORD) &&
(tmp->type != CT_TYPE) &&
(tmp->type != CT_QUALIFIER) &&
(tmp->type != CT_STAR) &&
(tmp->type != CT_CARET) &&
(tmp->type != CT_AMP) &&
(tmp->type != CT_TSQUARE))
{
break;
}
/* Change tmp into a type so that space_needed() works right */
make_type(tmp);
int num_sp = space_needed(tmp2, tmp);
while (num_sp-- > 0)
{
next->str.append(" ");
}
next->str.append(tmp->str);
tmp2 = tmp;
}
while ((tmp2 = chunk_get_next(next)) != tmp)
{
chunk_del(tmp2);
}
set_chunk_type(next, CT_OPERATOR_VAL);
next->orig_col_end = next->orig_col + next->len();
}
set_chunk_parent(next, CT_OPERATOR);
LOG_FMT(LOPERATOR, "%s: %d:%d operator '%s'\n",
__func__, pc->orig_line, pc->orig_col, next->text());
}
/* Change private, public, protected into either a qualifier or label */
if (pc->type == CT_PRIVATE)
{
/* Handle Qt slots - maybe should just check for a CT_WORD? */
if (chunk_is_str(next, "slots", 5) || chunk_is_str(next, "Q_SLOTS", 7))
{
tmp = chunk_get_next(next);
if ((tmp != NULL) && (tmp->type == CT_COLON))
{
next = tmp;
}
}
if (next->type == CT_COLON)
{
set_chunk_type(next, CT_PRIVATE_COLON);
if ((tmp = chunk_get_next_ncnl(next)) != NULL)
{
chunk_flags_set(tmp, PCF_STMT_START | PCF_EXPR_START);
}
}
else
{
set_chunk_type(pc, (chunk_is_str(pc, "signals", 7) || chunk_is_str(pc, "Q_SIGNALS", 9)) ? CT_WORD : CT_QUALIFIER);
}
}
/* Look for <newline> 'EXEC' 'SQL' */
if ((chunk_is_str(pc, "EXEC", 4) && chunk_is_str(next, "SQL", 3)) ||
((*pc->str == '$') && (pc->type != CT_SQL_WORD)))
{
tmp = chunk_get_prev(pc);
if (chunk_is_newline(tmp))
{
if (*pc->str == '$')
{
set_chunk_type(pc, CT_SQL_EXEC);
if (pc->len() > 1)
{
/* SPLIT OFF '$' */
chunk_t nc;
nc = *pc;
pc->str.resize(1);
pc->orig_col_end = pc->orig_col + 1;
nc.type = CT_SQL_WORD;
nc.str.pop_front();
nc.orig_col++;
nc.column++;
chunk_add_after(&nc, pc);
next = chunk_get_next(pc);
}
}
tmp = chunk_get_next(next);
if (chunk_is_str_case(tmp, "BEGIN", 5))
{
set_chunk_type(pc, CT_SQL_BEGIN);
}
else if (chunk_is_str_case(tmp, "END", 3))
{
set_chunk_type(pc, CT_SQL_END);
}
else
{
set_chunk_type(pc, CT_SQL_EXEC);
}
/* Change words into CT_SQL_WORD until CT_SEMICOLON */
while (tmp != NULL)
{
if (tmp->type == CT_SEMICOLON)
{
break;
}
if ((tmp->len() > 0) && (unc_isalpha(*tmp->str) || (*tmp->str == '$')))
{
set_chunk_type(tmp, CT_SQL_WORD);
}
tmp = chunk_get_next_ncnl(tmp);
}
}
}
/* handle MS abomination 'for each' */
if ((pc->type == CT_FOR) && chunk_is_str(next, "each", 4) &&
(next == chunk_get_next(pc)))
{
/* merge the two with a space between */
pc->str.append(' ');
pc->str += next->str;
pc->orig_col_end = next->orig_col_end;
chunk_del(next);
next = chunk_get_next_ncnl(pc);
/* label the 'in' */
if (next && (next->type == CT_PAREN_OPEN))
{
tmp = chunk_get_next_ncnl(next);
while (tmp && (tmp->type != CT_PAREN_CLOSE))
{
if (chunk_is_str(tmp, "in", 2))
{
set_chunk_type(tmp, CT_IN);
break;
}
tmp = chunk_get_next_ncnl(tmp);
}
}
}
/* ObjectiveC allows keywords to be used as identifiers in some situations
* This is a dirty hack to allow some of the more common situations.
*/
if (cpd.lang_flags & LANG_OC)
{
if (((pc->type == CT_IF) ||
(pc->type == CT_FOR) ||
(pc->type == CT_WHILE)) &&
!chunk_is_token(next, CT_PAREN_OPEN))
{
set_chunk_type(pc, CT_WORD);
}
if ((pc->type == CT_DO) &&
(chunk_is_token(prev, CT_MINUS) ||
chunk_is_token(next, CT_SQUARE_CLOSE)))
{
set_chunk_type(pc, CT_WORD);
}
}
/* Another hack to clean up more keyword abuse */
if ((pc->type == CT_CLASS) &&
(chunk_is_token(prev, CT_DOT) ||
chunk_is_token(next, CT_DOT)))
{
set_chunk_type(pc, CT_WORD);
}
/* Detect Objective C class name */
if ((pc->type == CT_OC_IMPL) ||
(pc->type == CT_OC_INTF) ||
(pc->type == CT_OC_PROTOCOL))
{
if (next->type != CT_PAREN_OPEN)
{
set_chunk_type(next, CT_OC_CLASS);
}
set_chunk_parent(next, pc->type);
tmp = chunk_get_next_ncnl(next);
if (tmp != NULL)
{
chunk_flags_set(tmp, PCF_STMT_START | PCF_EXPR_START);
}
tmp = chunk_get_next_type(pc, CT_OC_END, pc->level);
if (tmp != NULL)
{
set_chunk_parent(tmp, pc->type);
}
}
if (pc->type == CT_OC_INTF)
{
tmp = chunk_get_next_ncnl(pc, CNAV_PREPROC);
while ((tmp != NULL) && (tmp->type != CT_OC_END))
{
if (get_token_pattern_class(tmp->type) != PATCLS_NONE)
{
LOG_FMT(LOBJCWORD, "@interface %d:%d change '%s' (%s) to CT_WORD\n",
pc->orig_line, pc->orig_col, tmp->text(),
get_token_name(tmp->type));
set_chunk_type(tmp, CT_WORD);
}
tmp = chunk_get_next_ncnl(tmp, CNAV_PREPROC);
}
}
/* Detect Objective-C categories and class extensions */
/* @interface ClassName (CategoryName) */
/* @implementation ClassName (CategoryName) */
/* @interface ClassName () */
/* @implementation ClassName () */
if (((pc->parent_type == CT_OC_IMPL) ||
(pc->parent_type == CT_OC_INTF) ||
(pc->type == CT_OC_CLASS)) &&
(next->type == CT_PAREN_OPEN))
{
set_chunk_parent(next, pc->parent_type);
tmp = chunk_get_next(next);
if ((tmp != NULL) && (tmp->next != NULL))
{
if (tmp->type == CT_PAREN_CLOSE)
{
//set_chunk_type(tmp, CT_OC_CLASS_EXT);
set_chunk_parent(tmp, pc->parent_type);
}
else
{
set_chunk_type(tmp, CT_OC_CATEGORY);
set_chunk_parent(tmp, pc->parent_type);
}
}
tmp = chunk_get_next_type(pc, CT_PAREN_CLOSE, pc->level);
if (tmp != NULL)
{
set_chunk_parent(tmp, pc->parent_type);
}
}
/* Detect Objective C @property
* @property NSString *stringProperty;
* @property(nonatomic, retain) NSMutableDictionary *shareWith;
*/
if (pc->type == CT_OC_PROPERTY)
{
if (next->type != CT_PAREN_OPEN)
{
chunk_flags_set(next, PCF_STMT_START | PCF_EXPR_START);
}
else
{
set_chunk_parent(next, pc->type);
tmp = chunk_get_next_type(pc, CT_PAREN_CLOSE, pc->level);
if (tmp != NULL)
{
set_chunk_parent(tmp, pc->type);
tmp = chunk_get_next_ncnl(tmp);
if (tmp != NULL)
{
chunk_flags_set(tmp, PCF_STMT_START | PCF_EXPR_START);
tmp = chunk_get_next_type(tmp, CT_SEMICOLON, pc->level);
if (tmp != NULL)
{
set_chunk_parent(tmp, pc->type);
}
}
}
}
}
/* Detect Objective C @selector
* @selector(msgNameWithNoArg)
* @selector(msgNameWith1Arg:)
* @selector(msgNameWith2Args:arg2Name:)
*/
if ((pc->type == CT_OC_SEL) && (next->type == CT_PAREN_OPEN))
{
set_chunk_parent(next, pc->type);
tmp = chunk_get_next(next);
if (tmp != NULL)
{
set_chunk_type(tmp, CT_OC_SEL_NAME);
set_chunk_parent(tmp, pc->type);
while ((tmp = chunk_get_next_ncnl(tmp)) != NULL)
{
if (tmp->type == CT_PAREN_CLOSE)
{
set_chunk_parent(tmp, CT_OC_SEL);
break;
}
set_chunk_type(tmp, CT_OC_SEL_NAME);
set_chunk_parent(tmp, pc->type);
}
}
}
/* Handle special preprocessor junk */
if (pc->type == CT_PREPROC)
{
set_chunk_parent(pc, next->type);
}
/* Detect "pragma region" and "pragma endregion" */
if ((pc->type == CT_PP_PRAGMA) && (next->type == CT_PREPROC_BODY))
{
if ((memcmp(next->str, "region", 6) == 0) ||
(memcmp(next->str, "endregion", 9) == 0))
{
set_chunk_type(pc, (*next->str == 'r') ? CT_PP_REGION : CT_PP_ENDREGION);
set_chunk_parent(prev, pc->type);
}
}
/* Check for C# nullable types '?' is in next */
if ((cpd.lang_flags & LANG_CS) &&
(next->type == CT_QUESTION) &&
(next->orig_col == (pc->orig_col + pc->len())))
{
tmp = chunk_get_next_ncnl(next);
if (tmp != NULL)
{
bool doit = ((tmp->type == CT_PAREN_CLOSE) ||
(tmp->type == CT_ANGLE_CLOSE));
if (tmp->type == CT_WORD)
{
tmp2 = chunk_get_next_ncnl(tmp);
if ((tmp2 != NULL) &&
((tmp2->type == CT_SEMICOLON) ||
(tmp2->type == CT_ASSIGN) ||
(tmp2->type == CT_COMMA) ||
(tmp2->type == CT_BRACE_OPEN)))
{
doit = true;
}
}
if (doit)
{
pc->str += next->str;
pc->orig_col_end = next->orig_col_end;
chunk_del(next);
next = tmp;
}
}
}
/* Change 'default(' into a sizeof-like statement */
if ((cpd.lang_flags & LANG_CS) &&
(pc->type == CT_DEFAULT) &&
(next->type == CT_PAREN_OPEN))
{
set_chunk_type(pc, CT_SIZEOF);
}
if ((pc->type == CT_UNSAFE) && (next->type != CT_BRACE_OPEN))
{
set_chunk_type(pc, CT_QUALIFIER);
}
if (((pc->type == CT_USING) ||
((pc->type == CT_TRY) && (cpd.lang_flags & LANG_JAVA))) &&
(next->type == CT_PAREN_OPEN))
{
set_chunk_type(pc, CT_USING_STMT);
}
/* Add minimal support for C++0x rvalue references */
if ((pc->type == CT_BOOL) && chunk_is_str(pc, "&&", 2))
{
if (prev->type == CT_TYPE)
{
set_chunk_type(pc, CT_BYREF);
}
}
/* HACK: treat try followed by a colon as a qualifier to handle this:
* A::A(int) try : B() { } catch (...) { }
*/
if ((pc->type == CT_TRY) && chunk_is_str(pc, "try", 3) &&
(next != NULL) && (next->type == CT_COLON))
{
set_chunk_type(pc, CT_QUALIFIER);
}
/* If Java's 'synchronized' is in a method declaration, it should be
* a qualifier. */
if ((cpd.lang_flags & LANG_JAVA) &&
(pc->type == CT_SYNCHRONIZED) &&
(next->type != CT_PAREN_OPEN))
{
set_chunk_type(pc, CT_QUALIFIER);
}
// guy 2015-11-05
// change CT_DC_MEMBER + CT_FOR into CT_DC_MEMBER + CT_FUNC_CALL
if ((pc->type == CT_FOR) &&
(pc->prev->type == CT_DC_MEMBER))
{
set_chunk_type(pc, CT_FUNC_CALL);
}
/* TODO: determine other stuff here */
prev = pc;
pc = next;
next = chunk_get_next_ncnl(pc);
}
} // tokenize_cleanup
/**
* Scan backwards to find the most appropriate spot to split the line
* and insert a newline.
*
* See if this needs special function handling.
* Scan backwards and find the best token for the split.
*
* @param start The first chunk that exceeded the limit
*/
static void split_line(chunk_t *start)
{
LOG_FMT(LSPLIT, "%s: line %d, col %d token:%s[%s] (IN_FUNC=%d) ",
__func__, start->orig_line, start->column, start->str.c_str(),
get_token_name(start->type),
(start->flags & (PCF_IN_FCN_DEF | PCF_IN_FCN_CALL)) != 0);
/**
* break at maximum line length if indent_continue is absolute
*/
if (cpd.settings[UO_indent_continue].n < 0)
{
}
/* Check to see if we are in a for statement */
else if ((start->flags & PCF_IN_FOR) != 0)
{
LOG_FMT(LSPLIT, " ** FOR SPLIT **\n");
split_for_stmt(start);
if (!is_past_width(start))
{
return;
}
LOG_FMT(LSPLIT, "%s: for split didn't work\n", __func__);
}
/* If this is in a function call or prototype, split on commas or right
* after the open paren
*/
else if (((start->flags & PCF_IN_FCN_DEF) != 0) ||
((start->level == (start->brace_level + 1)) &&
((start->flags & PCF_IN_FCN_CALL) != 0)))
{
LOG_FMT(LSPLIT, " ** FUNC SPLIT **\n");
if (cpd.settings[UO_ls_func_split_full].b)
{
split_fcn_params_full(start);
if (!is_past_width(start))
{
return;
}
}
split_fcn_params(start);
if (!is_past_width(start))
{
return;
}
LOG_FMT(LSPLIT, "%s: func split didn't work\n", __func__);
}
/**
* Try to find the best spot to split the line
*/
cw_entry ent;
memset(&ent, 0, sizeof(ent));
chunk_t *pc = start;
chunk_t *prev;
while (((pc = chunk_get_prev(pc)) != NULL) && !chunk_is_newline(pc))
{
if (pc->type != CT_SPACE)
{
try_split_here(ent, pc);
// break at maximum line length if indent_continue is absolute
if ((ent.pc != NULL) && (cpd.settings[UO_indent_continue].n < 0))
break;
}
}
if (ent.pc == NULL)
{
LOG_FMT(LSPLIT, "%s: TRY_SPLIT yielded NO SOLUTION for line %d at %s [%s]\n",
__func__, start->orig_line, start->str.c_str(), get_token_name(start->type));
}
else
{
LOG_FMT(LSPLIT, "%s: TRY_SPLIT yielded '%s' [%s] on line %d\n", __func__,
ent.pc->str.c_str(), get_token_name(ent.pc->type), ent.pc->orig_line);
}
/* Break before the token instead of after it according to the pos_xxx rules */
if ((chunk_is_token(ent.pc, CT_ARITH) &&
(cpd.settings[UO_pos_arith].tp & TP_LEAD)) ||
(chunk_is_token(ent.pc, CT_ASSIGN) &&
(cpd.settings[UO_pos_assign].tp & TP_LEAD)) ||
(chunk_is_token(ent.pc, CT_COMPARE) &&
(cpd.settings[UO_pos_compare].tp & TP_LEAD)) ||
((chunk_is_token(ent.pc, CT_COND_COLON) ||
chunk_is_token(ent.pc, CT_QUESTION)) &&
(cpd.settings[UO_pos_conditional].tp & TP_LEAD)) ||
(chunk_is_token(ent.pc, CT_BOOL) &&
(cpd.settings[UO_pos_bool].tp & TP_LEAD)))
{
pc = ent.pc;
}
else
{
pc = chunk_get_next(ent.pc);
}
if (pc == NULL)
{
pc = start;
/* Don't break before a close, comma, or colon */
if ((start->type == CT_PAREN_CLOSE) ||
(start->type == CT_PAREN_OPEN) ||
(start->type == CT_FPAREN_CLOSE) ||
(start->type == CT_FPAREN_OPEN) ||
(start->type == CT_SPAREN_CLOSE) ||
(start->type == CT_SPAREN_OPEN) ||
(start->type == CT_ANGLE_CLOSE) ||
(start->type == CT_BRACE_CLOSE) ||
(start->type == CT_COMMA) ||
(start->type == CT_SEMICOLON) ||
(start->type == CT_VSEMICOLON) ||
(start->len() == 0))
{
LOG_FMT(LSPLIT, " ** NO GO **\n");
/*TODO: Add in logic to handle 'hard' limits by backing up a token */
return;
}
}
/* add a newline before pc */
prev = chunk_get_prev(pc);
if ((prev != NULL) && !chunk_is_newline(pc) && !chunk_is_newline(prev))
{
int plen = (pc->len() < 5) ? pc->len() : 5;
int slen = (start->len() < 5) ? start->len() : 5;
LOG_FMT(LSPLIT, " '%.*s' [%s], started on token '%.*s' [%s]\n",
plen, pc->str.c_str(), get_token_name(pc->type),
slen, start->str.c_str(), get_token_name(start->type));
split_before_chunk(pc);
}
}
static bool split_line(chunk_t *start)
{
LOG_FUNC_ENTRY();
LOG_FMT(LSPLIT, "%s(%d): start->text() '%s', orig_line is %zu, orig_col is %zu, type is %s\n",
__func__, __LINE__, start->text(), start->orig_line, start->orig_col, get_token_name(start->type));
LOG_FMT(LSPLIT, " start->flags ");
log_pcf_flags(LSPLIT, start->flags);
LOG_FMT(LSPLIT, " start->parent_type %s, (PCF_IN_FCN_DEF is %s), (PCF_IN_FCN_CALL is %s)\n",
get_token_name(start->parent_type),
((start->flags & (PCF_IN_FCN_DEF)) != 0) ? "TRUE" : "FALSE",
((start->flags & (PCF_IN_FCN_CALL)) != 0) ? "TRUE" : "FALSE");
// break at maximum line length if ls_code_width is true
if (start->flags & PCF_ONE_LINER)
{
LOG_FMT(LSPLIT, "%s(%d): ** ONCE LINER SPLIT **\n", __func__, __LINE__);
undo_one_liner(start);
newlines_cleanup_braces(false);
// Issue #1352
cpd.changes++;
return(false);
}
LOG_FMT(LSPLIT, "%s(%d):\n", __func__, __LINE__);
if (cpd.settings[UO_ls_code_width].b)
{
}
// Check to see if we are in a for statement
else if (start->flags & PCF_IN_FOR)
{
LOG_FMT(LSPLIT, " ** FOR SPLIT **\n");
split_for_stmt(start);
if (!is_past_width(start))
{
return(true);
}
LOG_FMT(LSPLIT, "%s(%d): for split didn't work\n", __func__, __LINE__);
}
/*
* If this is in a function call or prototype, split on commas or right
* after the open parenthesis
*/
else if ( (start->flags & PCF_IN_FCN_DEF)
|| start->parent_type == CT_FUNC_PROTO // Issue #1169
|| ( (start->level == (start->brace_level + 1))
&& (start->flags & PCF_IN_FCN_CALL)))
{
LOG_FMT(LSPLIT, " ** FUNC SPLIT **\n");
if (cpd.settings[UO_ls_func_split_full].b)
{
split_fcn_params_full(start);
if (!is_past_width(start))
{
return(true);
}
}
split_fcn_params(start);
return(true);
}
/*
* If this is in a template, split on commas, Issue #1170
*/
else if (start->flags & PCF_IN_TEMPLATE)
{
LOG_FMT(LSPLIT, " ** TEMPLATE SPLIT **\n");
split_template(start);
return(true);
}
LOG_FMT(LSPLIT, "%s(%d):\n", __func__, __LINE__);
// Try to find the best spot to split the line
cw_entry ent;
memset(&ent, 0, sizeof(ent));
chunk_t *pc = start;
chunk_t *prev;
while (((pc = chunk_get_prev(pc)) != nullptr) && !chunk_is_newline(pc))
{
LOG_FMT(LSPLIT, "%s(%d): at %s, orig_line=%zu, orig_col=%zu\n",
__func__, __LINE__, pc->text(), pc->orig_line, pc->orig_col);
if (pc->type != CT_SPACE)
{
try_split_here(ent, pc);
// break at maximum line length
if (ent.pc != nullptr && (cpd.settings[UO_ls_code_width].b))
{
break;
}
}
}
if (ent.pc == nullptr)
{
LOG_FMT(LSPLIT, "\n%s(%d): TRY_SPLIT yielded NO SOLUTION for orig_line %zu at '%s' [%s]\n",
__func__, __LINE__, start->orig_line, start->text(), get_token_name(start->type));
}
else
{
LOG_FMT(LSPLIT, "\n%s(%d): TRY_SPLIT yielded '%s' [%s] on orig_line %zu\n",
__func__, __LINE__, ent.pc->text(), get_token_name(ent.pc->type), ent.pc->orig_line);
LOG_FMT(LSPLIT, "%s(%d): ent at '%s', orig_col is %zu\n",
__func__, __LINE__, ent.pc->text(), ent.pc->orig_col);
}
// Break before the token instead of after it according to the pos_xxx rules
if (ent.pc == nullptr)
{
pc = nullptr;
}
else
{
if ( ( ( chunk_is_token(ent.pc, CT_ARITH)
|| chunk_is_token(ent.pc, CT_CARET))
&& (cpd.settings[UO_pos_arith].tp & TP_LEAD))
|| ( chunk_is_token(ent.pc, CT_ASSIGN)
&& (cpd.settings[UO_pos_assign].tp & TP_LEAD))
|| ( chunk_is_token(ent.pc, CT_COMPARE)
&& (cpd.settings[UO_pos_compare].tp & TP_LEAD))
|| ( ( chunk_is_token(ent.pc, CT_COND_COLON)
|| chunk_is_token(ent.pc, CT_QUESTION))
&& (cpd.settings[UO_pos_conditional].tp & TP_LEAD))
|| ( chunk_is_token(ent.pc, CT_BOOL)
&& (cpd.settings[UO_pos_bool].tp & TP_LEAD)))
{
pc = ent.pc;
}
else
{
pc = chunk_get_next(ent.pc);
}
LOG_FMT(LSPLIT, "%s(%d): at '%s', orig_col is %zu\n",
__func__, __LINE__, pc->text(), pc->orig_col);
}
if (pc == nullptr)
{
pc = start;
// Don't break before a close, comma, or colon
if ( start->type == CT_PAREN_CLOSE
|| start->type == CT_PAREN_OPEN
|| start->type == CT_FPAREN_CLOSE
|| start->type == CT_FPAREN_OPEN
|| start->type == CT_SPAREN_CLOSE
|| start->type == CT_SPAREN_OPEN
|| start->type == CT_ANGLE_CLOSE
|| start->type == CT_BRACE_CLOSE
|| start->type == CT_COMMA
|| start->type == CT_SEMICOLON
|| start->type == CT_VSEMICOLON
|| start->len() == 0)
{
LOG_FMT(LSPLIT, " ** NO GO **\n");
// TODO: Add in logic to handle 'hard' limits by backing up a token
return(true);
}
}
// add a newline before pc
prev = chunk_get_prev(pc);
if ( prev != nullptr
&& !chunk_is_newline(pc)
&& !chunk_is_newline(prev))
{
//int plen = (pc->len() < 5) ? pc->len() : 5;
//int slen = (start->len() < 5) ? start->len() : 5;
//LOG_FMT(LSPLIT, " '%.*s' [%s], started on token '%.*s' [%s]\n",
// plen, pc->text(), get_token_name(pc->type),
// slen, start->text(), get_token_name(start->type));
LOG_FMT(LSPLIT, "%s(%d): text() '%s', type [%s], started on token '%s', type [%s]\n",
__func__, __LINE__, pc->text(), get_token_name(pc->type),
start->text(), get_token_name(start->type));
split_before_chunk(pc);
}
return(true);
} // split_line
/**
* Adds an entry to the appropriate stack.
*
* @param pc The chunk
* @param seqnum Optional seqnum (0=assign one)
*/
void AlignStack::Add(chunk_t *start, int seqnum)
{
/* Assign a seqnum if needed */
if (seqnum == 0)
{
seqnum = m_seqnum;
}
chunk_t *ali;
chunk_t *ref;
chunk_t *tmp;
chunk_t *prev;
chunk_t *next;
int col_adj = 0; /* Amount the column is shifted for 'dangle' mode */
int tmp_col;
int endcol;
int gap;
m_last_added = 0;
/* Check threshold limits */
if ((m_max_col == 0) || (m_thresh == 0) ||
(((start->column + m_gap) <= (m_max_col + m_thresh)) &&
(((start->column + m_gap) >= (m_max_col - m_thresh)) ||
(start->column >= m_min_col))))
{
/* we are adding it, so update the newline seqnum */
if (seqnum > m_nl_seqnum)
{
m_nl_seqnum = seqnum;
}
/**
* SS_IGNORE: no special handling of '*' or '&', only 'foo' is aligned
* void foo; // gap=5, 'foo' is aligned
* char * foo; // gap=3, 'foo' is aligned
* foomatic foo; // gap=1, 'foo' is aligned
* The gap is the columns between 'foo' and the previous token.
* [void - foo], ['*' - foo], etc
*
* SS_INCLUDE: - space between variable and '*' or '&' is eaten
* void foo; // gap=5, 'foo' is aligned
* char *foo; // gap=5, '*' is aligned
* foomatic foo; // gap=1, 'foo' is aligned
* The gap is the columns between the first '*' or '&' before foo
* and the previous token. [void - foo], [char - '*'], etc
*
* SS_DANGLE: - space between variable and '*' or '&' is eaten
* void foo; // gap=5
* char *bar; // gap=5, as the '*' doesn't count
* foomatic foo; // gap=1
* The gap is the columns between 'foo' and the chunk before the first
* '*' or '&'. [void - foo], [char - bar], etc
*
* If the gap < m_gap, then the column is bumped out by the difference.
* So, if m_gap is 2, then the above would be:
* SS_IGNORE:
* void foo; // gap=6
* char * foo; // gap=4
* foomatic foo; // gap=2
* SS_INCLUDE:
* void foo; // gap=6
* char *foo; // gap=6
* foomatic foo; // gap=2
* SS_DANGLE:
* void foo; // gap=6
* char *bar; // gap=6, as the '*' doesn't count
* foomatic foo; // gap=2
* Right aligned numbers:
* #define A -1
* #define B 631
* #define C 3
* Left aligned numbers:
* #define A -1
* #define B 631
* #define C 3
*
* In the code below, pc is set to the item that is aligned.
* In the above examples, that is 'foo', '*', '-', or 63.
*
* Ref is set to the last part of the type.
* In the above examples, that is 'void', 'char', 'foomatic', 'A', or 'B'.
*
* The '*' and '&' can float between the two.
*
* If align_on_tabstop=true, then SS_DANGLE is changed to SS_INCLUDE.
*/
if (cpd.settings[UO_align_on_tabstop].b && (m_star_style == SS_DANGLE))
{
m_star_style = SS_INCLUDE;
}
/* Find ref. Back up to the real item that is aligned. */
prev = start;
while (((prev = chunk_get_prev(prev)) != NULL) &&
(chunk_is_star(prev) ||
chunk_is_addr(prev) ||
(prev->type == CT_TPAREN_OPEN)))
{
/* do nothing - we want prev when this exits */
}
ref = prev;
if (chunk_is_newline(ref))
{
ref = chunk_get_next(ref);
}
/* Find the item that we are going to align. */
ali = start;
if (m_star_style != SS_IGNORE)
{
/* back up to the first '*' preceding the token */
prev = chunk_get_prev(ali);
while (chunk_is_star(prev))
{
ali = prev;
prev = chunk_get_prev(ali);
}
if (chunk_is_token(prev, CT_TPAREN_OPEN))
{
ali = prev;
prev = chunk_get_prev(ali);
}
}
if (m_amp_style != SS_IGNORE)
{
/* back up to the first '&' preceding the token */
prev = chunk_get_prev(ali);
while (chunk_is_addr(prev))
{
ali = prev;
prev = chunk_get_prev(ali);
}
}
/* Tighten down the spacing between ref and start */
if (!cpd.settings[UO_align_keep_extra_space].b)
{
tmp_col = ref->column;
tmp = ref;
while (tmp != start)
{
next = chunk_get_next(tmp);
tmp_col += space_col_align(tmp, next);
if (next->column != tmp_col)
{
align_to_column(next, tmp_col);
}
tmp = next;
}
}
/* Set the column adjust and gap */
col_adj = 0;
gap = 0;
if (ref != ali)
{
gap = ali->column - (ref->column + ref->len());
}
tmp = ali;
if (chunk_is_token(tmp, CT_TPAREN_OPEN))
{
tmp = chunk_get_next(tmp);
}
if ((chunk_is_star(tmp) && (m_star_style == SS_DANGLE)) ||
(chunk_is_addr(tmp) && (m_amp_style == SS_DANGLE)))
{
col_adj = start->column - ali->column;
gap = start->column - (ref->column + ref->len());
}
/* See if this pushes out the max_col */
endcol = ali->column + col_adj;
if (gap < m_gap)
{
endcol += m_gap - gap;
}
// LOG_FMT(LSYS, "[%p] line %d pc='%s' [%s] col:%d ali='%s' [%s] col:%d ref='%s' [%s] col:%d col_adj=%d endcol=%d, ss=%d as=%d, gap=%d\n",
// this,
// start->orig_line,
// start->str.c_str(), get_token_name(start->type), start->column,
// ali->str.c_str(), get_token_name(ali->type), ali->column,
// ref->str.c_str(), get_token_name(ref->type), ref->column,
// col_adj, endcol, m_star_style, m_amp_style, gap);
ali->align.col_adj = col_adj;
ali->align.ref = ref;
ali->align.start = start;
m_aligned.Push_Back(ali, seqnum);
m_last_added = 1;
LOG_FMT(LAS, "Add-[%s]: line %d, col %d, adj %d : ref=[%s] endcol=%d\n",
ali->str.c_str(), ali->orig_line, ali->column, ali->align.col_adj,
ref->str.c_str(), endcol);
if (m_min_col > endcol)
{
m_min_col = endcol;
}
if (endcol > m_max_col)
{
LOG_FMT(LAS, "Add-aligned [%d/%d/%d]: line %d, col %d : max_col old %d, new %d - min_col %d\n",
seqnum, m_nl_seqnum, m_seqnum,
ali->orig_line, ali->column, m_max_col, endcol, m_min_col);
m_max_col = endcol;
/**
* If there were any entries that were skipped, re-add them as they
* may now be within the threshold
*/
if (!m_skipped.Empty())
{
ReAddSkipped();
}
}
else
{
LOG_FMT(LAS, "Add-aligned [%d/%d/%d]: line %d, col %d : col %d <= %d - min_col %d\n",
seqnum, m_nl_seqnum, m_seqnum,
ali->orig_line, ali->column, endcol, m_max_col, m_min_col);
}
}
else
{
/* The threshold check failed, so add it to the skipped list */
m_skipped.Push_Back(start, seqnum);
m_last_added = 2;
LOG_FMT(LAS, "Add-skipped [%d/%d/%d]: line %d, col %d <= %d + %d\n",
seqnum, m_nl_seqnum, m_seqnum,
start->orig_line, start->column, m_max_col, m_thresh);
}
}
static void check_bool_parens(chunk_t *popen, chunk_t *pclose, int nest)
{
LOG_FUNC_ENTRY();
chunk_t *ref = popen;
bool hit_compare = false;
LOG_FMT(LPARADD, "%s(%d): popen on %zu, col %zu, pclose on %zu, col %zu, level=%zu\n",
__func__, nest,
popen->orig_line, popen->orig_col,
pclose->orig_line, pclose->orig_col,
popen->level);
chunk_t *pc = popen;
while ((pc = chunk_get_next_ncnl(pc)) != nullptr && pc != pclose)
{
if (pc->flags & PCF_IN_PREPROC)
{
LOG_FMT(LPARADD2, " -- bail on PP %s [%s] at line %zu col %zu, level %zu\n",
get_token_name(pc->type),
pc->text(), pc->orig_line, pc->orig_col, pc->level);
return;
}
if ( chunk_is_token(pc, CT_BOOL)
|| chunk_is_token(pc, CT_QUESTION)
|| chunk_is_token(pc, CT_COND_COLON)
|| chunk_is_token(pc, CT_COMMA))
{
LOG_FMT(LPARADD2, " -- %s [%s] at line %zu col %zu, level %zu\n",
get_token_name(pc->type),
pc->text(), pc->orig_line, pc->orig_col, pc->level);
if (hit_compare)
{
hit_compare = false;
add_parens_between(ref, pc);
}
ref = pc;
}
else if (chunk_is_token(pc, CT_COMPARE))
{
LOG_FMT(LPARADD2, " -- compare [%s] at line %zu col %zu, level %zu\n",
pc->text(), pc->orig_line, pc->orig_col, pc->level);
hit_compare = true;
}
else if (chunk_is_paren_open(pc))
{
chunk_t *next = chunk_skip_to_match(pc);
if (next != nullptr)
{
check_bool_parens(pc, next, nest + 1);
pc = next;
}
}
else if ( chunk_is_token(pc, CT_BRACE_OPEN)
|| chunk_is_token(pc, CT_SQUARE_OPEN)
|| chunk_is_token(pc, CT_ANGLE_OPEN))
{
// Skip [], {}, and <>
pc = chunk_skip_to_match(pc);
}
}
if (hit_compare && ref != popen)
{
add_parens_between(ref, pclose);
}
} // check_bool_parens
void tokenize_cleanup(void)
{
chunk_t *pc = chunk_get_head();
chunk_t *prev = NULL;
chunk_t *next;
chunk_t *tmp;
chunk_t *tmp2;
bool in_type_cast = false;
pc = chunk_get_head();
next = chunk_get_next_ncnl(pc);
while ((pc != NULL) && (next != NULL))
{
/* Change '[' + ']' into '[]' */
if ((pc->type == CT_SQUARE_OPEN) && (next->type == CT_SQUARE_CLOSE))
{
pc->type = CT_TSQUARE;
pc->str = "[]";
pc->len = 2;
chunk_del(next);
pc->orig_col_end += 1;
next = chunk_get_next_ncnl(pc);
}
if ((pc->type == CT_DOT) && ((cpd.lang_flags & LANG_ALLC) != 0))
{
pc->type = CT_MEMBER;
}
/* Determine the version stuff (D only) */
if (pc->type == CT_D_VERSION)
{
if (next->type == CT_PAREN_OPEN)
{
pc->type = CT_D_VERSION_IF;
}
else
{
if (next->type != CT_ASSIGN)
{
LOG_FMT(LERR, "%s:%d %s: version: Unexpected token %s\n",
cpd.filename, pc->orig_line, __func__, get_token_name(next->type));
cpd.error_count++;
}
pc->type = CT_WORD;
}
}
/* Determine the scope stuff (D only) */
if (pc->type == CT_D_SCOPE)
{
if (next->type == CT_PAREN_OPEN)
{
pc->type = CT_D_SCOPE_IF;
}
else
{
pc->type = CT_TYPE;
}
}
/**
* Change CT_BASE before CT_PAREN_OPEN to CT_WORD.
* public myclass() : base() {
* }
*/
if ((pc->type == CT_BASE) && (next->type == CT_PAREN_OPEN))
{
pc->type = CT_WORD;
}
/**
* Change CT_WORD after CT_ENUM, CT_UNION, or CT_STRUCT to CT_TYPE
* Change CT_WORD before CT_WORD to CT_TYPE
*/
if (next->type == CT_WORD)
{
if ((pc->type == CT_ENUM) ||
(pc->type == CT_UNION) ||
(pc->type == CT_STRUCT))
{
next->type = CT_TYPE;
}
if (pc->type == CT_WORD)
{
pc->type = CT_TYPE;
}
}
/* change extern to qualifier if extern isn't followed by a string or
* an open paren
*/
if (pc->type == CT_EXTERN)
{
if (next->type == CT_STRING)
{
/* Probably 'extern "C"' */
}
else if (next->type == CT_PAREN_OPEN)
{
/* Probably 'extern (C)' */
}
else
{
/* Something else followed by a open brace */
tmp = chunk_get_next_ncnl(next);
if ((tmp != NULL) || (tmp->type != CT_BRACE_OPEN))
{
pc->type = CT_QUALIFIER;
}
}
}
/**
* Change CT_STAR to CT_PTR_TYPE if preceded by CT_TYPE,
* CT_QUALIFIER, or CT_PTR_TYPE.
*/
if ((next->type == CT_STAR) &&
((pc->type == CT_TYPE) ||
(pc->type == CT_QUALIFIER) ||
(pc->type == CT_PTR_TYPE)))
{
next->type = CT_PTR_TYPE;
}
if ((pc->type == CT_TYPE_CAST) &&
(next->type == CT_ANGLE_OPEN))
{
next->parent_type = CT_TYPE_CAST;
in_type_cast = true;
}
/**
* Change angle open/close to CT_COMPARE, if not a template thingy
*/
if ((pc->type == CT_ANGLE_OPEN) && (pc->parent_type != CT_TYPE_CAST))
{
check_template(pc);
}
if ((pc->type == CT_ANGLE_CLOSE) && (pc->parent_type != CT_TEMPLATE))
{
if (in_type_cast)
{
in_type_cast = false;
pc->parent_type = CT_TYPE_CAST;
}
else
{
next = handle_double_angle_close(pc);
}
}
if ((cpd.lang_flags & LANG_D) != 0)
{
/* Check for the D string concat symbol '~' */
if ((pc->type == CT_INV) &&
((prev->type == CT_STRING) ||
(prev->type == CT_WORD) ||
(next->type == CT_STRING)))
{
pc->type = CT_CONCAT;
}
/* Check for the D template symbol '!' (word + '!' + word or '(') */
if ((pc->type == CT_NOT) &&
(prev->type == CT_WORD) &&
((next->type == CT_PAREN_OPEN) ||
(next->type == CT_WORD) ||
(next->type == CT_TYPE)))
{
pc->type = CT_D_TEMPLATE;
}
/* handle "version(unittest) { }" vs "unittest { }" */
if ((pc->type == CT_UNITTEST) && (prev->type == CT_PAREN_OPEN))
{
pc->type = CT_WORD;
}
}
if ((cpd.lang_flags & LANG_CPP) != 0)
{
/* Change Word before '::' into a type */
if ((pc->type == CT_WORD) && (next->type == CT_DC_MEMBER))
{
pc->type = CT_TYPE;
}
}
/* Change get/set to CT_WORD if not followed by a brace open */
if ((pc->type == CT_GETSET) && (next->type != CT_BRACE_OPEN))
{
if ((next->type == CT_SEMICOLON) &&
((prev->type == CT_BRACE_CLOSE) ||
(prev->type == CT_BRACE_OPEN) ||
(prev->type == CT_SEMICOLON)))
{
pc->type = CT_GETSET_EMPTY;
next->parent_type = CT_GETSET;
}
else
{
pc->type = CT_WORD;
}
}
/* Change item after operator (>=, ==, etc) to a CT_OPERATOR_VAL
* Usually the next item is part of the operator.
* In a few cases the next few tokens are part of it:
* operator + - common case
* operator >> - need to combine '>' and '>'
* operator ()
* operator [] - already converted to TSQUARE
* operator new []
* operator delete []
* operator const char *
* operator const B&
* operator std::allocator<U>
*
* In all cases except the last, this will put the entire operator value
* in one chunk.
*/
if (pc->type == CT_OPERATOR)
{
/* Handle special case of () operator -- [] already handled */
if (next->type == CT_PAREN_OPEN)
{
tmp = chunk_get_next(next);
if ((tmp != NULL) && (tmp->type == CT_PAREN_CLOSE))
{
next->str = "()";
next->len = 2;
next->type = CT_OPERATOR_VAL;
chunk_del(tmp);
next->orig_col_end += 1;
}
}
else if ((next->type == CT_ANGLE_CLOSE) &&
(memcmp(next->str, ">>", 2) == 0))
{
next->len++;
next->orig_col_end++;
next->type = CT_OPERATOR_VAL;
chunk_del(chunk_get_next(next));
}
else if (next->flags & PCF_PUNCTUATOR)
{
next->type = CT_OPERATOR_VAL;
}
else
{
next->type = CT_TYPE;
/* Replace next with a collection of all tokens that are part of
* the type.
*/
char opbuf[256];
int len;
len = snprintf(opbuf, sizeof(opbuf), "%.*s",
next->len, next->str);
tmp2 = next;
while ((tmp = chunk_get_next(tmp2)) != NULL)
{
if ((tmp->type != CT_WORD) &&
(tmp->type != CT_TYPE) &&
(tmp->type != CT_QUALIFIER) &&
(tmp->type != CT_STAR) &&
(tmp->type != CT_AMP) &&
(tmp->type != CT_TSQUARE))
{
break;
}
/* Change tmp into a type so that space_needed() works right */
make_type(tmp);
len += snprintf(opbuf + len, sizeof(opbuf) - len, "%s%.*s",
space_needed(tmp2, tmp) ? " " : "",
tmp->len, tmp->str);
tmp2 = tmp;
}
while ((tmp2 = chunk_get_next(next)) != tmp)
{
chunk_del(tmp2);
}
next->str = strdup(opbuf);
next->len = len;
next->flags |= PCF_OWN_STR;
next->type = CT_OPERATOR_VAL;
next->orig_col_end = next->orig_col + next->len;
}
next->parent_type = CT_OPERATOR;
LOG_FMT(LOPERATOR, "%s: %d:%d operator '%.*s'\n",
__func__, pc->orig_line, pc->orig_col,
next->len, next->str);
}
/* Change private, public, protected into either a qualifier or label */
if (pc->type == CT_PRIVATE)
{
/* Handle Qt slots - maybe should just check for a CT_WORD? */
if (chunk_is_str(next, "slots", 5) || chunk_is_str(next, "Q_SLOTS", 7))
{
tmp = chunk_get_next(next);
if ((tmp != NULL) && (tmp->type == CT_COLON))
{
next = tmp;
}
}
if (next->type == CT_COLON)
{
next->type = CT_PRIVATE_COLON;
if ((tmp = chunk_get_next_ncnl(next)) != NULL)
{
tmp->flags |= PCF_STMT_START | PCF_EXPR_START;
}
}
else
{
pc->type = (chunk_is_str(pc, "signals", 7) || chunk_is_str(pc, "Q_SIGNALS", 9)) ? CT_WORD : CT_QUALIFIER;
}
}
/* Look for <newline> 'EXEC' 'SQL' */
if (chunk_is_str(pc, "EXEC", 4) && chunk_is_str(next, "SQL", 3))
{
tmp = chunk_get_prev(pc);
if (chunk_is_newline(tmp))
{
tmp = chunk_get_next(next);
if (chunk_is_str_case(tmp, "BEGIN", 5))
{
pc->type = CT_SQL_BEGIN;
}
else if (chunk_is_str_case(tmp, "END", 3))
{
pc->type = CT_SQL_END;
}
else
{
pc->type = CT_SQL_EXEC;
}
/* Change words into CT_SQL_WORD until CT_SEMICOLON */
while (tmp != NULL)
{
if (tmp->type == CT_SEMICOLON)
{
break;
}
if ((tmp->len > 0) && isalpha(*tmp->str))
{
tmp->type = CT_SQL_WORD;
}
tmp = chunk_get_next_ncnl(tmp);
}
}
}
/* handle MS abomination 'for each' */
if ((pc->type == CT_FOR) && chunk_is_str(next, "each", 4) &&
(next == chunk_get_next(pc)))
{
/* merge the two */
pc->len = next->orig_col_end - pc->orig_col;
pc->orig_col_end = next->orig_col_end;
chunk_del(next);
next = chunk_get_next_ncnl(pc);
/* label the 'in' */
if (next && (next->type == CT_PAREN_OPEN))
{
tmp = chunk_get_next_ncnl(next);
while (tmp && (tmp->type != CT_PAREN_CLOSE))
{
if (chunk_is_str(tmp, "in", 2))
{
tmp->type = CT_IN;
break;
}
tmp = chunk_get_next_ncnl(tmp);
}
}
}
/* ObjectiveC allows keywords to be used as identifiers in some situations
* This is a dirty hack to allow some of the more common situations.
*/
if (cpd.lang_flags & LANG_OC)
{
if (((pc->type == CT_IF) ||
(pc->type == CT_FOR) ||
(pc->type == CT_WHILE)) &&
!chunk_is_token(next, CT_PAREN_OPEN))
{
pc->type = CT_WORD;
}
if ((pc->type == CT_DO) &&
(chunk_is_token(prev, CT_MINUS) ||
chunk_is_token(next, CT_SQUARE_CLOSE)))
{
pc->type = CT_WORD;
}
}
/* Another hack to clean up more keyword abuse */
if ((pc->type == CT_CLASS) &&
(chunk_is_token(prev, CT_DOT) ||
chunk_is_token(next, CT_DOT)))
{
pc->type = CT_WORD;
}
/* Detect Objective C class name */
if ((pc->type == CT_OC_IMPL) ||
(pc->type == CT_OC_INTF) ||
(pc->type == CT_OC_PROTOCOL))
{
if (next->type != CT_PAREN_OPEN)
{
next->type = CT_OC_CLASS;
}
next->parent_type = pc->type;
tmp = chunk_get_next_ncnl(next);
if (tmp != NULL)
{
tmp->flags |= PCF_STMT_START | PCF_EXPR_START;
}
tmp = chunk_get_next_type(pc, CT_OC_END, pc->level);
if (tmp != NULL)
{
tmp->parent_type = pc->type;
}
}
if (pc->type == CT_OC_INTF)
{
tmp = chunk_get_next_ncnl(pc, CNAV_PREPROC);
while ((tmp != NULL) && (tmp->type != CT_OC_END))
{
if (get_token_pattern_class(tmp->type) != PATCLS_NONE)
{
LOG_FMT(LOBJCWORD, "@interface %d:%d change '%.*s' (%s) to CT_WORD\n",
pc->orig_line, pc->orig_col, tmp->len, tmp->str,
get_token_name(tmp->type));
tmp->type = CT_WORD;
}
tmp = chunk_get_next_ncnl(tmp, CNAV_PREPROC);
}
}
/* Detect Objective-C categories and class extensions */
/* @interface ClassName (CategoryName) */
/* @implementation ClassName (CategoryName) */
/* @interface ClassName () */
/* @implementation ClassName () */
if (((pc->parent_type == CT_OC_IMPL) ||
(pc->parent_type == CT_OC_INTF) ||
(pc->type == CT_OC_CLASS)) &&
(next->type == CT_PAREN_OPEN))
{
next->parent_type = pc->parent_type;
tmp = chunk_get_next(next);
if ((tmp != NULL) && (tmp->next != NULL))
{
if (tmp->type == CT_PAREN_CLOSE)
{
//tmp->type = CT_OC_CLASS_EXT;
tmp->parent_type = pc->parent_type;
}
else
{
tmp->type = CT_OC_CATEGORY;
tmp->parent_type = pc->parent_type;
}
}
tmp = chunk_get_next_type(pc, CT_PAREN_CLOSE, pc->level);
if (tmp != NULL)
{
tmp->parent_type = pc->parent_type;
}
}
/**
* Objective C @dynamic and @synthesize
* @dynamic xxx, yyy;
* @synthesize xxx, yyy;
* Just find the semicolon and mark it.
*/
if (pc->type == CT_OC_DYNAMIC)
{
tmp = chunk_get_next_type(pc, CT_SEMICOLON, pc->level);
if (tmp != NULL)
{
tmp->parent_type = pc->type;
}
}
/* Detect Objective C @property
* @property NSString *stringProperty;
* @property(nonatomic, retain) NSMutableDictionary *shareWith;
*/
if (pc->type == CT_OC_PROPERTY)
{
if (next->type != CT_PAREN_OPEN)
{
next->flags |= PCF_STMT_START | PCF_EXPR_START;
}
else
{
next->parent_type = pc->type;
tmp = chunk_get_next_type(pc, CT_PAREN_CLOSE, pc->level);
if (tmp != NULL)
{
tmp->parent_type = pc->type;
tmp = chunk_get_next_ncnl(tmp);
if (tmp != NULL)
{
tmp->flags |= PCF_STMT_START | PCF_EXPR_START;
tmp = chunk_get_next_type(tmp, CT_SEMICOLON, pc->level);
if (tmp != NULL)
{
tmp->parent_type = pc->type;
}
}
}
}
}
/* Detect Objective C @selector
* @selector(msgNameWithNoArg)
* @selector(msgNameWith1Arg:)
* @selector(msgNameWith2Args:arg2Name:)
*/
if ((pc->type == CT_OC_SEL) && (next->type == CT_PAREN_OPEN))
{
next->parent_type = pc->type;
tmp = chunk_get_next(next);
if (tmp != NULL)
{
tmp->type = CT_OC_SEL_NAME;
tmp->parent_type = pc->type;
while ((tmp = chunk_get_next_ncnl(tmp)) != NULL)
{
if (tmp->type == CT_PAREN_CLOSE)
{
tmp->parent_type = CT_OC_SEL;
break;
}
tmp->type = CT_OC_SEL_NAME;
tmp->parent_type = pc->type;
}
}
}
/* Mark Objective-C blocks (aka lambdas or closures)
* The syntax and usage is exactly like C function pointers with two exceptions:
* Instead of an asterisk they have a caret as pointer symbol.
* In method declarations which take a block as parameter, there can be anonymous blocks, e.g.: (^)
* 1. block literal: ^{ ... };
* 2. block declaration: return_t (^name) (int arg1, int arg2, ...) NB: return_t is optional and name can be optional if found as param in a method declaration.
* 3. block expression: ^ return_t (int arg) { ... }; NB: return_t is optional
*
* See http://developer.apple.com/mac/library/documentation/Cocoa/Conceptual/Blocks for more info...
*/
if ((cpd.lang_flags & LANG_OC) &&
(pc->type == CT_ARITH) &&
(chunk_is_str(pc, "^", 1)) &&
(prev->type != CT_NUMBER) &&
(prev->type != CT_NUMBER_FP))
{
/* mark objc blocks caret so that we can process it later*/
pc->type = CT_OC_BLOCK_CARET;
if (prev->type == CT_PAREN_OPEN)
{
/* block declaration */
pc->parent_type = CT_OC_BLOCK_TYPE;
}
else if ((next->type == CT_PAREN_OPEN) ||
(next->type == CT_BRACE_OPEN))
{
/* block expression without return type */
/* block literal */
pc->parent_type = CT_OC_BLOCK_EXPR;
}
else
{
/* block expression with return type (seldomly used) */
if (prev->type == CT_ASSIGN)
{
/* shortcut to spare the peeking below
* the XOR operator wouldn't be used directly
* after an assign all by itself */
pc->parent_type = CT_OC_BLOCK_EXPR;
}
else
{
/* this ones tricky because we don't know how many
* stars the return type has - if there even is one */
tmp = pc;
while ((tmp = chunk_get_next(tmp)) != NULL)
{
/* we just peek ahead and see if the line contains
* an open brace somewhere.
* FIXME: this check needs to be more thorough. */
if (tmp->type == CT_BRACE_OPEN)
{
pc->parent_type = CT_OC_BLOCK_EXPR;
break;
}
}
}
}
}
/* Handle special preprocessor junk */
if (pc->type == CT_PREPROC)
{
pc->parent_type = next->type;
}
/* Detect "pragma region" and "pragma endregion" */
if ((pc->type == CT_PP_PRAGMA) && (next->type == CT_PREPROC_BODY))
{
if ((memcmp(next->str, "region", 6) == 0) ||
(memcmp(next->str, "endregion", 9) == 0))
{
pc->type = (*next->str == 'r') ? CT_PP_REGION : CT_PP_ENDREGION;
prev->parent_type = pc->type;
}
}
/* Check for C# nullable types '?' is in next */
if ((cpd.lang_flags & LANG_CS) &&
(next->type == CT_QUESTION) &&
(next->orig_col == (pc->orig_col + pc->len)))
{
tmp = chunk_get_next_ncnl(next);
if (tmp != NULL)
{
bool doit = ((tmp->type == CT_PAREN_CLOSE) ||
(tmp->type == CT_ANGLE_CLOSE));
if (tmp->type == CT_WORD)
{
tmp2 = chunk_get_next_ncnl(tmp);
if ((tmp2 != NULL) &&
((tmp2->type == CT_SEMICOLON) ||
(tmp2->type == CT_ASSIGN) ||
(tmp2->type == CT_COMMA)))
{
doit = true;
}
}
if (doit)
{
pc->len++;
chunk_del(next);
next = tmp;
}
}
}
/* Change 'default(' into a sizeof-like statement */
if ((cpd.lang_flags & LANG_CS) &&
(pc->type == CT_DEFAULT) &&
(next->type == CT_PAREN_OPEN))
{
pc->type = CT_SIZEOF;
}
if ((pc->type == CT_UNSAFE) && (next->type != CT_BRACE_OPEN))
{
pc->type = CT_QUALIFIER;
}
if ((pc->type == CT_USING) && (next->type == CT_PAREN_OPEN))
{
pc->type = CT_USING_STMT;
}
/* Add minimal support for C++0x rvalue references */
if ((pc->type == CT_BOOL) && chunk_is_str(pc, "&&", 2))
{
if (prev->type == CT_TYPE)
{
pc->type = CT_BYREF;
}
}
/* HACK: treat try followed by a colon as a qualifier to handle this:
* A::A(int) try : B() { } catch (...) { }
*/
if ((pc->type == CT_TRY) && chunk_is_str(pc, "try", 3) &&
(next != NULL) && (next->type == CT_COLON))
{
pc->type = CT_QUALIFIER;
}
/* TODO: determine other stuff here */
prev = pc;
pc = next;
next = chunk_get_next_ncnl(pc);
}
}