ariadne
/
smol
2
1
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

overhaul the justify code to make it leave the right number of spaces at 

the ends of the lines of a paragraph, and also to make it simpler


git-svn-id: svn://svn.savannah.gnu.org/nano/trunk/nano@2354 35c25a1d-7b9e-4130-9fde-d3aeb78583b8
master
David Lawrence Ramsey 2005-03-13 03:28:37 +00:00
parent 59fe758a10
commit e9ac1d7930
3 changed files with 254 additions and 265 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
ChangeLog
View File

@ -171,6 +171,10 @@ CVS code -
parse_syntax(), parse_colors(), parse_rcfile(), do_rcfile(), parse_syntax(), parse_colors(), parse_rcfile(), do_rcfile(),
etc. (David Benbennick) DLR: Rename colortoint() to etc. (David Benbennick) DLR: Rename colortoint() to
color_to_int(), and add a few miscellaneous tweaks. color_to_int(), and add a few miscellaneous tweaks.
- Overhaul the justify code to make it leave the right number of
spaces at the ends of the lines of a paragraph, and also to
make it simpler. Changes to justify_format() and
do_justify(); removal of breakable(). (DLR)
- Still more steps toward full wide/multibyte character support. - Still more steps toward full wide/multibyte character support.
Make whitespace display mode work with multibyte characters, Make whitespace display mode work with multibyte characters,
and add a few related documentation updates. New function and add a few related documentation updates. New function

512
src/nano.c
View File

@ -2354,8 +2354,8 @@ size_t indent_length(const char *line)
#ifndef DISABLE_JUSTIFY #ifndef DISABLE_JUSTIFY
/* justify_format() replaces tabs with spaces and multiple spaces by 1 /* justify_format() replaces tabs with spaces and multiple spaces by 1
* (except it maintains 2 after a non-repeated character in punct * (except it maintains 2 after a non-repeated character in punct
* followed by a character in brackets). Note that the terminating \0 * followed by a character in brackets, and removes all at the end of
* counts as a space. * the line).
* *
* justify_format() might make line->data shorter, and change the actual * justify_format() might make line->data shorter, and change the actual
* pointer with null_at(). * pointer with null_at().
@ -2363,74 +2363,146 @@ size_t indent_length(const char *line)
* justify_format() will not look at the first skip characters of line. * justify_format() will not look at the first skip characters of line.
* skip should be at most strlen(line->data). The character at * skip should be at most strlen(line->data). The character at
* line[skip + 1] must not be whitespace. */ * line[skip + 1] must not be whitespace. */
void justify_format(filestruct *line, size_t skip) void justify_format(filestruct *paragraph, size_t skip)
{ {
char *back, *front; char *end, *new_end, *new_paragraph_data;
size_t shift = 0;
#ifndef NANO_SMALL
size_t mark_shift = 0;
#endif
/* These four asserts are assumptions about the input data. */ /* These four asserts are assumptions about the input data. */
assert(line != NULL); assert(paragraph != NULL);
assert(line->data != NULL); assert(paragraph->data != NULL);
assert(skip < strlen(line->data)); assert(skip < strlen(paragraph->data));
assert(!is_blank_char(line->data[skip])); assert(!is_blank_char(paragraph->data[skip]));
back = line->data + skip; end = paragraph->data + skip;
for (front = back; ; front++) { new_paragraph_data = charalloc(strlen(paragraph->data) + 1);
bool remove_space = FALSE; charcpy(new_paragraph_data, paragraph->data, skip);
/* Do we want to remove this space? */ new_end = new_paragraph_data + skip;
if (*front == '\t') while (*end != '\0') {
*front = ' '; /* If this character is blank, make sure that it's a space with
* no blanks after it. */
if (is_blank_char(*end)) {
*new_end = ' ';
new_end++;
end++;
/* These tests are safe since line->data + skip is not a while (*end != '\0' && is_blank_char(*end)) {
* space. */ end++;
if ((*front == '\0' || *front == ' ') && *(front - 1) == ' ') { shift++;
const char *bob = back - 2; #ifndef NANO_SMALL
if (mark_beginbuf == paragraph &&
remove_space = TRUE; mark_beginx >= end - paragraph->data)
for (; bob >= line->data + skip; bob--) { mark_shift++;
if (strchr(punct, *bob) != NULL) { #endif
/* If this character is in punct, don't remove the
* space unless this character and the character
* before it are the same. */
remove_space = (bob > line->data + skip &&
*bob == *(bob - 1));
break;
}
if (strchr(brackets, *bob) == NULL)
break;
} }
} /* If this character is punctuation, there are two ways we can
* handle it. */
} else if (strchr(punct, *end) != NULL) {
*new_end = *end;
new_end++;
end++;
if (remove_space) { /* If this character is punctuation followed by itself and
/* Now *front is a space we want to remove. We do that by * optionally followed by a bracket, make sure there is no
* simply failing to assign it to *back. */ * more than one blank after it, and make sure that the
* blank is a space. */
if (*end != '\0' && *end == *(end - 1)) {
*new_end = *end;
new_end++;
end++;
if (*end != '\0' && strchr(brackets, *end) != NULL) {
*new_end = *end;
new_end++;
end++;
}
if (*end != '\0' && is_blank_char(*end)) {
*new_end = ' ';
new_end++;
end++;
}
while (*end != '\0' && is_blank_char(*end)) {
end++;
shift++;
#ifndef NANO_SMALL #ifndef NANO_SMALL
if (mark_beginbuf == line && back - line->data < mark_beginx) if (mark_beginbuf == paragraph &&
mark_beginx--; mark_beginx >= end - paragraph->data)
mark_shift++;
#endif #endif
if (*front == '\0') }
*(back - 1) = '\0'; /* If this character is punctuation optionally followed by a
* bracket and then followed by spaces, make sure there are
* no more than two blanks after it, and make sure that the
* blanks are spaces. */
} else {
if (*end != '\0' && strchr(brackets, *end) != NULL) {
*new_end = *end;
new_end++;
end++;
}
if (*end != '\0' && is_blank_char(*end)) {
*new_end = ' ';
new_end++;
end++;
}
if (*end != '\0' && is_blank_char(*end)) {
*new_end = ' ';
new_end++;
end++;
}
while (*end != '\0' && is_blank_char(*end)) {
end++;
shift++;
#ifndef NANO_SMALL
if (mark_beginbuf == paragraph &&
mark_beginx >= end - paragraph->data)
mark_shift++;
#endif
}
}
} else { } else {
*back = *front; *new_end = *end;
back++; new_end++;
end++;
} }
if (*front == '\0')
break;
} }
back--; assert(*end == '\0');
assert(*back == '\0' && *front == '\0'); *new_end = *end;
while (new_end > new_paragraph_data + skip &&
*(new_end - 1) == ' ') {
new_end--;
shift++;
}
if (shift > 0) {
totsize -= shift;
null_at(&new_paragraph_data, new_end - new_paragraph_data);
free(paragraph->data);
paragraph->data = new_paragraph_data;
/* Now back is the new end of line->data. */
if (back != front) {
totsize -= front - back;
null_at(&line->data, back - line->data);
#ifndef NANO_SMALL #ifndef NANO_SMALL
if (mark_beginbuf == line && back - line->data < mark_beginx) /* Adjust the mark coordinates to compensate for the change in
mark_beginx = back - line->data; * the current line. */
if (mark_beginbuf == paragraph) {
mark_beginx -= mark_shift;
if (mark_beginx > new_end - new_paragraph_data)
mark_beginx = new_end - new_paragraph_data;
}
#endif #endif
} } else
free(new_paragraph_data);
} }
/* The "quote part" of a line is the largest initial substring matching /* The "quote part" of a line is the largest initial substring matching
@ -2668,25 +2740,6 @@ filestruct *backup_lines(filestruct *first_line, size_t par_len, size_t
return first_line; return first_line;
} }
/* Is it possible to break line at or before goal? */
bool breakable(const char *line, ssize_t goal)
{
while (*line != '\0' && goal >= 0) {
size_t pos = 0;
if (is_blank_char(*line))
return TRUE;
line += parse_mbchar(line, NULL, NULL, &pos);
goal -= pos;
}
/* If goal is not negative, the whole line (one word) was short
* enough. */
return (goal >= 0);
}
/* We are trying to break a chunk off line. We find the last space such /* We are trying to break a chunk off line. We find the last space such
* that the display length to there is at most goal + 1. If there is no * that the display length to there is at most goal + 1. If there is no
* such space, and force is TRUE, then we find the first space. Anyway, * such space, and force is TRUE, then we find the first space. Anyway,
@ -2871,10 +2924,6 @@ void do_justify(bool full_justify)
filestruct *last_par_line; filestruct *last_par_line;
/* Will be the line containing the newline after the last line /* Will be the line containing the newline after the last line
* of the result. Also for restoring after unjustify. */ * of the result. Also for restoring after unjustify. */
bool allow_respacing;
/* Whether we should change the spacing at the end of a line
* after justifying it. This should be TRUE whenever we move
* to the next line after justifying the current line. */
/* We save these global variables to be restored if the user /* We save these global variables to be restored if the user
* unjustifies. Note that we don't need to save totlines. */ * unjustifies. Note that we don't need to save totlines. */
@ -2897,11 +2946,18 @@ void do_justify(bool full_justify)
last_par_line = current; last_par_line = current;
while (TRUE) { while (TRUE) {
size_t i;
/* Generic loop variable. */
size_t quote_len; size_t quote_len;
/* Length of the initial quotation of the paragraph we /* Length of the initial quotation of the paragraph we
* justify. */ * justify. */
size_t indent_len;
/* Length of the initial indentation of the paragraph we
* justify. */
size_t par_len; size_t par_len;
/* Number of lines in that paragraph. */ /* Number of lines in the paragraph we justify. */
ssize_t break_pos;
/* Where we will break lines. */
/* Find the first line of the paragraph to be justified. That /* Find the first line of the paragraph to be justified. That
* is the start of this paragraph if we're in one, or the start * is the start of this paragraph if we're in one, or the start
@ -2923,213 +2979,143 @@ void do_justify(bool full_justify)
} }
} }
/* Next step, we loop through the lines of this paragraph, /* If we haven't already done it, copy the original paragraph(s)
* justifying each one individually. */ * to the justify buffer. */
for (; par_len > 0; current_y++, par_len--) { if (first_par_line == NULL)
size_t indent_len; first_par_line = backup_lines(current, full_justify ?
/* Generic indentation length. */ filebot->lineno - current->lineno : par_len, quote_len);
size_t line_len;
size_t display_len;
/* The width of current in screen columns. */
ssize_t break_pos;
/* Where we will break the line. */
/* We'll be moving to the next line after justifying the /* Next step, we tack all the lines of the paragraph together,
* current line in almost all cases, so allow changing the * skipping the quoting and indentation on all lines after the
* spacing at the ends of justified lines by default. */ * first. */
allow_respacing = TRUE; for (i = 0; i < par_len - 1; i++) {
filestruct *next_line = current->next;
size_t line_len = strlen(current->data);
size_t next_line_len = strlen(current->next->data);
indent_len = quote_len + indent_length(current->next->data +
quote_len);
next_line_len -= indent_len;
totsize -= indent_len;
if (line_len > 0 && current->data[line_len - 1] != ' ') {
line_len++;
current->data = charealloc(current->data, line_len + 1);
current->data[line_len - 1] = ' ';
current->data[line_len] = '\0';
totsize++;
}
current->data = charealloc(current->data, line_len +
next_line_len + 1);
strcat(current->data, next_line->data + indent_len);
/* Don't destroy edittop! */
if (edittop == next_line)
edittop = current;
#ifndef NANO_SMALL
/* Adjust the mark coordinates to compensate for the change
* in the next line. */
if (mark_beginbuf == next_line) {
mark_beginbuf = current;
mark_beginx += line_len;
}
#endif
unlink_node(next_line);
delete_node(next_line);
/* If we've removed the next line, we need to go through
* this line again. */
i--;
par_len--;
totlines--;
totsize--;
}
/* Now we call justify_format() on the paragraph, which will
* remove excess spaces from it and change tabs to spaces. */
justify_format(current, quote_len +
indent_length(current->data + quote_len));
while (par_len > 0 && strlenpt(current->data) > fill) {
size_t line_len = strlen(current->data);
indent_len = quote_len + indent_length(current->data + indent_len = quote_len + indent_length(current->data +
quote_len); quote_len);
/* If we haven't already done it, copy the original /* If this line is too long, try to wrap it to the next line
* paragraph to the justify buffer. */ * to make it short enough. */
if (first_par_line == NULL) break_pos = break_line(current->data + indent_len,
first_par_line = backup_lines(current, full_justify ? fill - strnlenpt(current->data, indent_len), TRUE);
filebot->lineno - current->lineno : par_len,
quote_len);
/* Now we call justify_format() on the current line of the /* We can't break the line, or don't need to, so get out. */
* paragraph, which will remove excess spaces from it and if (break_pos == -1 || break_pos + indent_len == line_len)
* change tabs to spaces. */ break;
justify_format(current, quote_len +
indent_length(current->data + quote_len));
line_len = strlen(current->data); break_pos += indent_len;
display_len = strlenpt(current->data);
if (display_len > fill) { assert(break_pos < line_len);
/* The line is too long. Try to wrap it to the next. */
break_pos = break_line(current->data + indent_len,
fill - strnlenpt(current->data, indent_len),
TRUE);
if (break_pos == -1 ||
break_pos + indent_len == line_len)
/* We can't break the line, or don't need to, so
* just go on to the next. */
goto continue_loc;
break_pos += indent_len;
assert(break_pos < line_len); /* Make a new line and copy the text after where we broke
* this line to the beginning of the new line. */
splice_node(current, make_new_node(current), current->next);
if (par_len == 1) { /* If this paragraph is non-quoted, and autoindent is turned
/* There is no next line in this paragraph. We make * on, set the indentation length to zero so that
* a new line and copy text after break_pos into * indentation is treated as part of the line. */
* it. */ if (quote_len == 0
splice_node(current, make_new_node(current),
current->next);
/* In a non-quoted paragraph, we copy the indent
* only if AUTOINDENT is turned on. */
if (quote_len == 0
#ifndef NANO_SMALL #ifndef NANO_SMALL
&& !ISSET(AUTOINDENT) && !ISSET(AUTOINDENT)
#endif #endif
) )
indent_len = 0; indent_len = 0;
current->next->data = charalloc(indent_len +
line_len - break_pos);
strncpy(current->next->data, current->data,
indent_len);
strcpy(current->next->data + indent_len,
current->data + break_pos + 1);
assert(strlen(current->next->data) == current->next->data = charalloc(indent_len + line_len -
indent_len + line_len - break_pos - 1); break_pos);
charcpy(current->next->data, current->data, indent_len);
strcpy(current->next->data + indent_len, current->data +
break_pos + 1);
totlines++; assert(strlen(current->next->data) == indent_len + line_len - break_pos - 1);
totsize += indent_len;
par_len++;
} else {
size_t next_line_len = strlen(current->next->data);
indent_len = quote_len + par_len++;
indent_length(current->next->data + quote_len); totlines++;
current->next->data = totsize += indent_len;
charealloc(current->next->data, next_line_len +
line_len - break_pos + 1);
charmove(current->next->data + indent_len +
line_len - break_pos, current->next->data +
indent_len, next_line_len - indent_len + 1);
strcpy(current->next->data + indent_len,
current->data + break_pos + 1);
current->next->data[indent_len + line_len -
break_pos - 1] = ' ';
#ifndef NANO_SMALL #ifndef NANO_SMALL
if (mark_beginbuf == current->next) { /* Adjust the mark coordinates to compensate for the change
if (mark_beginx < indent_len) * in the current line. */
mark_beginx = indent_len; if (mark_beginbuf == current && mark_beginx > break_pos) {
mark_beginx += line_len - break_pos; mark_beginbuf = current->next;
} mark_beginx -= break_pos + 1 - indent_len;
#endif
}
#ifndef NANO_SMALL
if (mark_beginbuf == current &&
mark_beginx > break_pos) {
mark_beginbuf = current->next;
mark_beginx -= break_pos + 1 - indent_len;
}
#endif
null_at(&current->data, break_pos);
/* Go to the next line. */
current = current->next;
} else if (display_len < fill && par_len > 1) {
size_t next_line_len;
indent_len = quote_len +
indent_length(current->next->data + quote_len);
/* If we can't pull a word from the next line up to this
* one, just go on. */
if (!breakable(current->next->data + indent_len,
fill - display_len - 1))
goto continue_loc;
break_pos = break_line(current->next->data + indent_len,
fill - display_len - 1, FALSE);
assert(break_pos != -1);
current->data = charealloc(current->data,
line_len + break_pos + 2);
current->data[line_len] = ' ';
strncpy(current->data + line_len + 1,
current->next->data + indent_len, break_pos);
current->data[line_len + break_pos + 1] = '\0';
#ifndef NANO_SMALL
if (mark_beginbuf == current->next) {
if (mark_beginx < indent_len + break_pos) {
mark_beginbuf = current;
if (mark_beginx <= indent_len)
mark_beginx = line_len + 1;
else
mark_beginx = line_len + 1 + mark_beginx -
indent_len;
} else
mark_beginx -= break_pos + 1;
}
#endif
next_line_len = strlen(current->next->data);
if (indent_len + break_pos == next_line_len) {
filestruct *line = current->next;
/* Don't destroy edittop! */
if (line == edittop)
edittop = current;
unlink_node(line);
delete_node(line);
totlines--;
totsize -= indent_len;
current_y--;
/* Don't go to the next line. Accordingly, don't
* allow changing the spacing at the end of the
* previous justified line, so that we don't end up
* doing it more than once on the same line. */
allow_respacing = FALSE;
} else {
charmove(current->next->data + indent_len,
current->next->data + indent_len + break_pos + 1,
next_line_len - break_pos - indent_len);
null_at(&current->next->data,
next_line_len - break_pos);
/* Go to the next line. */
current = current->next;
}
} else
continue_loc:
/* Go to the next line. */
current = current->next;
/* We've moved to the next line after justifying the
* current line. If the justified line was not the last
* line of the paragraph, add a space to the end of it to
* replace the one removed or left out by justify_format().
* If it was the last line of the paragraph, and
* justify_format() left a space on the end of it, remove
* the space. */
if (allow_respacing) {
size_t prev_line_len = strlen(current->prev->data);
if (par_len > 1) {
current->prev->data =
charealloc(current->prev->data,
prev_line_len + 2);
current->prev->data[prev_line_len] = ' ';
current->prev->data[prev_line_len + 1] = '\0';
totsize++;
} else if (par_len == 1 &&
current->prev->data[prev_line_len - 1] == ' ') {
current->prev->data =
charealloc(current->prev->data, prev_line_len);
current->prev->data[prev_line_len - 1] = '\0';
totsize--;
}
} }
#endif
/* Break the line. If this isn't the last line of the
* paragraph, add a space after where we break it. */
null_at(&current->data, break_pos);
if (par_len > 1) {
current->data = charealloc(current->data,
break_pos + 2);
current->data[break_pos] = ' ';
current->data[break_pos + 1] = '\0';
totsize++;
}
/* Go to the next line. */
par_len--;
current_y++;
current = current->next;
} }
/* Go to the next line, the line after the last line of the
* paragraph. */
current_y++;
current = current->next;
/* We've just justified a paragraph. If we're not justifying the /* We've just justified a paragraph. If we're not justifying the
* entire file, break out of the loop. Otherwise, continue the * entire file, break out of the loop. Otherwise, continue the
* loop so that we justify all the paragraphs in the file. */ * loop so that we justify all the paragraphs in the file. */

3
src/proto.h
View File

@ -414,7 +414,7 @@ void do_spell(void);
size_t indent_length(const char *line); size_t indent_length(const char *line);
#endif #endif
#ifndef DISABLE_JUSTIFY #ifndef DISABLE_JUSTIFY
void justify_format(filestruct *line, size_t skip); void justify_format(filestruct *paragraph, size_t skip);
size_t quote_length(const char *line); size_t quote_length(const char *line);
bool quotes_match(const char *a_line, size_t a_quote, const char bool quotes_match(const char *a_line, size_t a_quote, const char
*b_line); *b_line);
@ -426,7 +426,6 @@ bool inpar(const char *str);
void do_para_end(void); void do_para_end(void);
filestruct *backup_lines(filestruct *first_line, size_t par_len, size_t filestruct *backup_lines(filestruct *first_line, size_t par_len, size_t
quote_len); quote_len);
bool breakable(const char *line, ssize_t goal);
ssize_t break_line(const char *line, ssize_t goal, bool force); ssize_t break_line(const char *line, ssize_t goal, bool force);
bool do_para_search(size_t *const quote, size_t *const par); bool do_para_search(size_t *const quote, size_t *const par);
void do_justify(bool full_justify); void do_justify(bool full_justify);