int get()
{
if (more())
{
int ch = data[c.idx++];
switch (ch)
{
case '\t':
c.col = calc_next_tab_column(c.col, cpd.settings[UO_input_tab_size].n);
break;
case '\n':
if (c.last_ch != '\r')
{
c.row++;
c.col = 1;
}
break;
case '\r':
c.row++;
c.col = 1;
break;
default:
c.col++;
break;
}
c.last_ch = ch;
return(ch);
}
return(-1);
}
/**
* Count the number of whitespace characters.
*
* @param pc The structure to update, str is an input.
* @return Whether whitespace was parsed
*/
static bool parse_whitespace(tok_ctx& ctx, chunk_t& pc)
{
int nl_count = 0;
int ch = -2;
/* REVISIT: use a better whitespace detector? */
while (ctx.more() && unc_isspace(ctx.peek()))
{
ch = ctx.get(); /* throw away the whitespace char */
switch (ch)
{
case '\r':
if (ctx.expect('\n'))
{
/* CRLF ending */
cpd.le_counts[LE_CRLF]++;
}
else
{
/* CR ending */
cpd.le_counts[LE_CR]++;
}
nl_count++;
pc.orig_prev_sp = 0;
break;
case '\n':
/* LF ending */
cpd.le_counts[LE_LF]++;
nl_count++;
pc.orig_prev_sp = 0;
break;
case '\t':
pc.orig_prev_sp += calc_next_tab_column(cpd.column, cpd.settings[UO_input_tab_size].n) - cpd.column;
break;
case ' ':
pc.orig_prev_sp++;
break;
default:
break;
}
}
if (ch != -2)
{
pc.str.clear();
pc.nl_count = nl_count;
pc.type = nl_count ? CT_NEWLINE : CT_WHITESPACE;
pc.after_tab = (ctx.c.last_ch == '\t');
return(true);
}
return(false);
} // parse_whitespace
/**
* Output a multiline comment without any reformatting other than shifting
* it left or right to get the column right.
* Oh, and trim trailing whitespace.
*/
static void output_comment_multi_simple(chunk_t *pc)
{
int cmt_idx;
char ch;
int line_count = 0;
int ccol;
int col_diff = 0;
bool nl_end = false;
cmt_reflow cmt;
unc_text line;
output_cmt_start(cmt, pc);
if (chunk_is_newline(chunk_get_prev(pc)))
{
/* The comment should be indented correctly */
col_diff = pc->orig_col - pc->column;
}
else
{
/* The comment starts after something else */
col_diff = 0;
}
ccol = pc->column;
cmt_idx = 0;
line.clear();
while (cmt_idx < pc->len())
{
ch = pc->str[cmt_idx++];
/* handle the CRLF and CR endings. convert both to LF */
if (ch == '\r')
{
ch = '\n';
if ((cmt_idx < pc->len()) && (pc->str[cmt_idx] == '\n'))
{
cmt_idx++;
}
}
/* Find the start column */
if (line.size() == 0)
{
nl_end = false;
if (ch == ' ')
{
ccol++;
continue;
}
else if (ch == '\t')
{
ccol = calc_next_tab_column(ccol, cpd.settings[UO_input_tab_size].n);
continue;
}
else
{
//LOG_FMT(LSYS, "%d] Text starts in col %d, col_diff=%d, real=%d\n",
// line_count, ccol, col_diff, ccol - col_diff);
}
}
line.append(ch);
/* If we just hit an end of line OR we just hit end-of-comment... */
if ((ch == '\n') || (cmt_idx == pc->len()))
{
line_count++;
/* strip trailing tabs and spaces before the newline */
if (ch == '\n')
{
line.pop_back();
nl_end = true;
/* Say we aren't in a preproc to prevent changing any bs-nl */
cmt_trim_whitespace(line, false);
}
if (line_count > 1)
{
ccol -= col_diff;
}
if (line.size() > 0)
{
cmt.column = ccol;
cmt_output_indent(cmt.brace_col, cmt.base_col, cmt.column);
add_text(line);
}
if (nl_end)
{
add_char('\n');
}
line.clear();
ccol = 1;
}
}
}
void output_comment_multi(chunk_t *pc)
{
int cmt_col = pc->column;
const char *cmt_str;
int remaining;
char ch;
chunk_t *prev;
char line[1024];
int line_len;
int line_count = 0;
int ccol;
int col_diff = 0;
int xtra = 1;
prev = chunk_get_prev(pc);
if ((prev != NULL) && (prev->type != CT_NEWLINE))
{
cmt_col = pc->orig_col;
}
else
{
col_diff = pc->orig_col - pc->column;
}
// fprintf(stderr, "Indenting1 line %d to col %d (orig=%d) col_diff=%d\n",
// pc->orig_line, cmt_col, pc->orig_col, col_diff);
xtra = calculate_comment_body_indent(pc->str, pc->len, pc->column);
ccol = 1;
remaining = pc->len;
cmt_str = pc->str;
line_len = 0;
while (remaining > 0)
{
ch = *cmt_str;
cmt_str++;
remaining--;
/* handle the CRLF and CR endings. convert both to LF */
if (ch == '\r')
{
ch = '\n';
if (*cmt_str == '\n')
{
cmt_str++;
remaining--;
}
}
/* Find the start column */
if (line_len == 0)
{
if (ch == ' ')
{
ccol++;
continue;
}
else if (ch == '\t')
{
ccol = calc_next_tab_column(ccol, cpd.settings[UO_input_tab_size].n);
continue;
}
else
{
//fprintf(stderr, "%d] Text starts in col %d\n", line_count, ccol);
}
}
line[line_len++] = ch;
/* If we just hit an end of line OR we just hit end-of-comment... */
if ((ch == '\n') || (remaining == 0))
{
line_count++;
/* strip trailing tabs and spaces before the newline */
if (ch == '\n')
{
line_len--;
while ((line_len > 0) &&
((line[line_len - 1] == ' ') ||
(line[line_len - 1] == '\t')))
{
line_len--;
}
line[line_len++] = ch;
}
line[line_len] = 0;
if (line_count == 1)
{
/* this is the first line - add unchanged */
/*TODO: need to support indent_with_tabs mode 1 */
output_to_column(cmt_col, cpd.settings[UO_indent_with_tabs].b);
add_text_len(line, line_len);
}
else
{
/* This is not the first line, so we need to indent to the
* correct column.
*/
ccol -= col_diff;
if (ccol < cmt_col)
{
ccol = cmt_col;
}
if (line[0] == '\n')
{
/* Emtpy line - just a '\n' */
if (cpd.settings[UO_cmt_star_cont].b)
{
output_to_column(cmt_col, cpd.settings[UO_indent_with_tabs].b);
add_text((xtra == 1) ? " *" : "*");
}
add_char('\n');
}
else
{
/* If this doesn't start with a '*' or '|' */
if ((line[0] != '*') && (line[0] != '|') && (line[0] != '#') &&
(line[0] != '\\') && (line[0] != '+'))
{
output_to_column(cmt_col, cpd.settings[UO_indent_with_tabs].b);
if (cpd.settings[UO_cmt_star_cont].b)
{
add_text((xtra == 1) ? " * " : "* ");
}
else
{
add_text(" ");
}
output_to_column(ccol, cpd.settings[UO_indent_with_tabs].b);
}
else
{
output_to_column(cmt_col + xtra, cpd.settings[UO_indent_with_tabs].b);
}
add_text_len(line, line_len);
}
}
line_len = 0;
ccol = 1;
}
}
}
/**
* A multiline comment -- woopeee!
* The only trick here is that we have to trim out whitespace characters
* to get the comment to line up.
*/
static void output_comment_multi(chunk_t *pc)
{
int cmt_col;
int cmt_idx;
int ch;
unc_text line;
int line_count = 0;
int ccol; /* the col of subsequent comment lines */
int col_diff = 0;
bool nl_end = false;
cmt_reflow cmt;
//LOG_FMT(LSYS, "%s: line %d\n", __func__, pc->orig_line);
output_cmt_start(cmt, pc);
cmt.reflow = (cpd.settings[UO_cmt_reflow_mode].n != 1);
cmt_col = cmt.base_col;
col_diff = pc->orig_col - cmt.base_col;
calculate_comment_body_indent(cmt, pc->str);
cmt.cont_text = !cpd.settings[UO_cmt_indent_multi].b ? "" :
(cpd.settings[UO_cmt_star_cont].b ? "* " : " ");
LOG_CONTTEXT();
//LOG_FMT(LSYS, "Indenting1 line %d to col %d (orig=%d) col_diff=%d xtra=%d cont='%s'\n",
// pc->orig_line, cmt_col, pc->orig_col, col_diff, cmt.xtra_indent, cmt.cont_text.c_str());
ccol = pc->column;
cmt_idx = 0;
line.clear();
while (cmt_idx < pc->len())
{
ch = pc->str[cmt_idx++];
/* handle the CRLF and CR endings. convert both to LF */
if (ch == '\r')
{
ch = '\n';
if ((cmt_idx < pc->len()) && (pc->str[cmt_idx] == '\n'))
{
cmt_idx++;
}
}
/* Find the start column */
if (line.size() == 0)
{
nl_end = false;
if (ch == ' ')
{
ccol++;
continue;
}
else if (ch == '\t')
{
ccol = calc_next_tab_column(ccol, cpd.settings[UO_input_tab_size].n);
continue;
}
else
{
//LOG_FMT(LSYS, "%d] Text starts in col %d\n", line_count, ccol);
}
}
/*
* Now see if we need/must fold the next line with the current to enable
* full reflow
*/
if ((cpd.settings[UO_cmt_reflow_mode].n == 2) &&
(ch == '\n') &&
(cmt_idx < pc->len()))
{
int nxt_len = 0;
int next_nonempty_line = -1;
int prev_nonempty_line = -1;
int nwidx = line.size();
bool star_is_bullet = false;
/* strip trailing whitespace from the line collected so far */
while (nwidx > 0)
{
nwidx--;
if ((prev_nonempty_line < 0) &&
!unc_isspace(line[nwidx]) &&
(line[nwidx] != '*') && // block comment: skip '*' at end of line
((pc->flags & PCF_IN_PREPROC)
? (line[nwidx] != '\\') ||
((line[nwidx + 1] != 'r') &&
(line[nwidx + 1] != '\n'))
: true))
{
prev_nonempty_line = nwidx; // last nonwhitespace char in the previous line
}
}
int remaining = pc->len() - cmt_idx;
for (nxt_len = 0;
(nxt_len <= remaining) &&
(pc->str[nxt_len] != 'r') &&
(pc->str[nxt_len] != '\n');
nxt_len++)
{
if ((next_nonempty_line < 0) &&
!unc_isspace(pc->str[nxt_len]) &&
(pc->str[nxt_len] != '*') &&
((nxt_len == remaining) ||
((pc->flags & PCF_IN_PREPROC)
? (pc->str[nxt_len] != '\\') ||
((pc->str[nxt_len + 1] != 'r') &&
(pc->str[nxt_len + 1] != '\n'))
: true)))
{
next_nonempty_line = nxt_len; // first nonwhitespace char in the next line
}
}
/*
* see if we should fold up; usually that'd be a YES, but there are a few
* situations where folding/reflowing by merging lines is frowned upon:
*
* - ASCII art in the comments (most often, these are drawings done in +-\/|.,*)
*
* - Doxygen/JavaDoc/etc. parameters: these often start with \ or @, at least
* something clearly non-alphanumeric (you see where we're going with this?)
*
* - bullet lists that are closely spaced: bullets are always non-alphanumeric
* characters, such as '-' or '+' (or, oh horor, '*' - that's bloody ambiguous
* to parse :-( ... with or without '*' comment start prefix, that's the
* question, then.)
*
* - semi-HTML formatted code, e.g. <pre>...</pre> comment sections (NDoc, etc.)
*
* - New lines which form a new paragraph without there having been added an
* extra empty line between the last sentence and the new one.
* A bit like this, really; so it is opportune to check if the last line ended
* in a terminal (that would be the set '.:;!?') and the new line starts with
* a capital.
* Though new lines starting with comment delimiters, such as '(', should be
* pulled up.
*
* So it bores down to this: the only folding (& reflowing) that's going to happen
* is when the next line starts with an alphanumeric character AND the last
* line didn't end with an non-alphanumeric character, except: ',' AND the next
* line didn't start with a '*' all of a sudden while the previous one didn't
* (the ambiguous '*'-for-bullet case!)
*/
if ((prev_nonempty_line >= 0) && (next_nonempty_line >= 0) &&
(((unc_isalnum(line[prev_nonempty_line]) ||
strchr(",)]", line[prev_nonempty_line])) &&
(unc_isalnum(pc->str[next_nonempty_line]) ||
strchr("([", pc->str[next_nonempty_line]))) ||
(('.' == line[prev_nonempty_line]) && // dot followed by non-capital is NOT a new sentence start
unc_isupper(pc->str[next_nonempty_line]))) &&
!star_is_bullet)
{
// rewind the line to the last non-alpha:
line.resize(prev_nonempty_line + 1);
// roll the current line forward to the first non-alpha:
cmt_idx += next_nonempty_line;
// override the NL and make it a single whitespace:
ch = ' ';
}
}
line.append(ch);
/* If we just hit an end of line OR we just hit end-of-comment... */
if ((ch == '\n') || (cmt_idx == pc->len()))
{
line_count++;
/* strip trailing tabs and spaces before the newline */
if (ch == '\n')
{
nl_end = true;
line.pop_back();
cmt_trim_whitespace(line, pc->flags & PCF_IN_PREPROC);
}
//LOG_FMT(LSYS, "[%3d]%s\n", ccol, line);
if (line_count == 1)
{
/* this is the first line - add unchanged */
add_comment_text(line, cmt, false);
if (nl_end)
{
add_char('\n');
}
}
else
{
/* This is not the first line, so we need to indent to the
* correct column. Each line is indented 0 or more spaces.
*/
ccol -= col_diff;
if (ccol < (cmt_col + 3))
{
ccol = cmt_col + 3;
}
if (line.size() == 0)
{
/* Empty line - just a '\n' */
if (cpd.settings[UO_cmt_star_cont].b)
{
cmt.column = cmt_col + cpd.settings[UO_cmt_sp_before_star_cont].n;
cmt_output_indent(cmt.brace_col, cmt.base_col, cmt.column);
if (cmt.xtra_indent)
{
add_char(' ');
}
add_text(cmt.cont_text);
}
add_char('\n');
}
else
{
/* If this doesn't start with a '*' or '|'.
* '\name' is a common parameter documentation thing.
*/
if (cpd.settings[UO_cmt_indent_multi].b &&
(line[0] != '*') && (line[0] != '|') && (line[0] != '#') &&
((line[0] != '\\') || unc_isalpha(line[1])) && (line[0] != '+'))
{
int start_col = cmt_col + cpd.settings[UO_cmt_sp_before_star_cont].n;
if (cpd.settings[UO_cmt_star_cont].b)
{
cmt.column = start_col;
cmt_output_indent(cmt.brace_col, cmt.base_col, cmt.column);
if (cmt.xtra_indent)
{
add_char(' ');
}
add_text(cmt.cont_text);
output_to_column(ccol + cpd.settings[UO_cmt_sp_after_star_cont].n,
false);
}
else
{
cmt.column = ccol;
cmt_output_indent(cmt.brace_col, cmt.base_col, cmt.column);
}
}
else
{
cmt.column = cmt_col + cpd.settings[UO_cmt_sp_before_star_cont].n;
cmt_output_indent(cmt.brace_col, cmt.base_col, cmt.column);
if (cmt.xtra_indent)
{
add_char(' ');
}
int idx;
idx = cmt_parse_lead(line, (cmt_idx == pc->len()));
if (idx > 0)
{
cmt.cont_text.set(line, 0, idx);
LOG_CONTTEXT();
if ((line.size() >= 2) && (line[0] == '*') && unc_isalnum(line[1]))
{
line.insert(1, ' ');
}
}
else
{
add_text(cmt.cont_text);
}
}
add_comment_text(line, cmt, false);
if (nl_end)
{
add_text("\n");
}
}
}
line.clear();
ccol = 1;
}
}
}
