diff options
author | Dan Gudmundsson <[email protected]> | 2017-06-02 14:25:02 +0200 |
---|---|---|
committer | Dan Gudmundsson <[email protected]> | 2017-08-16 10:03:28 +0200 |
commit | 5f56b49c752a16ee28244981ca9b197ffd5fa691 (patch) | |
tree | 17a9563e03ecb40ba4cfe96cea2b89209ed1875b /erts | |
parent | b182febe36aa63eb8290f24ba4b7932673a9a9bc (diff) | |
download | otp-5f56b49c752a16ee28244981ca9b197ffd5fa691.tar.gz otp-5f56b49c752a16ee28244981ca9b197ffd5fa691.tar.bz2 otp-5f56b49c752a16ee28244981ca9b197ffd5fa691.zip |
stdlib: Improve edlin handling of unicode chars
Let edlin handle grapheme clusters instead of codepoints to
improve the handling multi-codepoints characters.
The ttsl driver (and protocol) still expects all lengths as
codepoints.
Previously it was expected that each codepoint used (at least) one
terminal column for each codepoint, and a hack was made for wide
characters (multicolumn) by patching in TAGGED characters to occupy
the extra space so that codepoint index was equal column index.
This didn't work at all for combining codepoints that do not occupy any
more space than the previous character.
Improved this handling by calculating column positions in move_cursor.
This is based on wcwidth() and is not perfect, wcwidth() is wrong for
some codepoints and wcwidth() can not know with Hangul graphemes for
example. But it works better than before without making a major change
in the protocol.
Diffstat (limited to 'erts')
-rw-r--r-- | erts/emulator/drivers/unix/ttsl_drv.c | 93 |
1 files changed, 50 insertions, 43 deletions
diff --git a/erts/emulator/drivers/unix/ttsl_drv.c b/erts/emulator/drivers/unix/ttsl_drv.c index e425b99f16..bce097d944 100644 --- a/erts/emulator/drivers/unix/ttsl_drv.c +++ b/erts/emulator/drivers/unix/ttsl_drv.c @@ -108,16 +108,15 @@ static int lbuf_size = BUFSIZ; static Uint32 *lbuf; /* The current line buffer */ static int llen; /* The current line length */ static int lpos; /* The current "cursor position" in the line buffer */ - + /* NOTE: not the same as column position a char may not take a" + * column to display or it might take many columns + */ /* * Tags used in line buffer to show that these bytes represent special characters, * Max unicode is 0x0010ffff, so we have lots of place for meta tags... */ #define CONTROL_TAG 0x10000000U /* Control character, value in first position */ #define ESCAPED_TAG 0x01000000U /* Escaped character, value in first position */ -#ifdef HAVE_WCWIDTH -#define WIDE_TAG 0x02000000U /* Wide character, value in first position */ -#endif #define TAG_MASK 0xFF000000U #define MAXSIZE (1 << 16) @@ -156,6 +155,8 @@ static int insert_buf(byte*,int); static int write_buf(Uint32 *,int); static int outc(int c); static int move_cursor(int,int); +static int cp_pos_to_col(int cp_pos); + /* Termcap functions. */ static int start_termcap(void); @@ -991,24 +992,26 @@ static int del_chars(int n) { int i, l, r; int pos; + int gcs; /* deleted grapheme characters */ update_cols(); /* Step forward or backwards over n logical characters. */ pos = step_over_chars(n); - + DEBUGLOG(("del_chars: %d from %d %d %d\n", n, lpos, pos, llen)); if (pos > lpos) { l = pos - lpos; /* Buffer characters to delete */ r = llen - lpos - l; /* Characters after deleted */ + gcs = cp_pos_to_col(pos) - cp_pos_to_col(lpos); /* Fix up buffer and buffer pointers. */ if (r > 0) memmove(lbuf + lpos, lbuf + pos, r * sizeof(Uint32)); llen -= l; /* Write out characters after, blank the tail and jump back to lpos. */ write_buf(lbuf + lpos, r); - for (i = l ; i > 0; --i) + for (i = gcs ; i > 0; --i) outc(' '); - if (COL(llen+l) == 0 && xn) + if (xn && COL(cp_pos_to_col(llen)+gcs) == 0) { outc(' '); move_left(1); @@ -1018,7 +1021,7 @@ static int del_chars(int n) else if (pos < lpos) { l = lpos - pos; /* Buffer characters */ r = llen - lpos; /* Characters after deleted */ - move_cursor(lpos, lpos-l); /* Move back */ + gcs = -move_cursor(lpos, lpos-l); /* Move back */ /* Fix up buffer and buffer pointers. */ if (r > 0) memmove(lbuf + pos, lbuf + lpos, r * sizeof(Uint32)); @@ -1026,14 +1029,14 @@ static int del_chars(int n) llen -= l; /* Write out characters after, blank the tail and jump back to lpos. */ write_buf(lbuf + lpos, r); - for (i = l ; i > 0; --i) - outc(' '); - if (COL(llen+l) == 0 && xn) + for (i = gcs ; i > 0; --i) + outc(' '); + if (xn && COL(cp_pos_to_col(llen)+gcs) == 0) { - outc(' '); - move_left(1); + outc(' '); + move_left(1); } - move_cursor(llen + l, lpos); + move_cursor(llen + l, lpos); } return TRUE; } @@ -1047,22 +1050,12 @@ static int step_over_chars(int n) end = lbuf + llen; c = lbuf + lpos; for ( ; n > 0 && c < end; --n) { -#ifdef HAVE_WCWIDTH - while (*c & WIDE_TAG) { - c++; - } -#endif c++; while (c < end && (*c & TAG_MASK) && ((*c & ~TAG_MASK) == 0)) c++; } for ( ; n < 0 && c > beg; n++) { --c; -#ifdef HAVE_WCWIDTH - while (c > beg + 1 && (c[-1] & WIDE_TAG)) { - --c; - } -#endif while (c > beg && (*c & TAG_MASK) && ((*c & ~TAG_MASK) == 0)) --c; } @@ -1088,15 +1081,6 @@ static int insert_buf(byte *s, int n) ++pos; } if ((utf8_mode && (ch >= 128 || isprint(ch))) || (ch <= 255 && isprint(ch))) { -#ifdef HAVE_WCWIDTH - int width; - if ((width = wcwidth(ch)) > 1) { - while (--width) { - DEBUGLOG(("insert_buf: Wide(UTF-8):%d,%d",width,ch)); - lbuf[lpos++] = (WIDE_TAG | ((Uint32) ch)); - } - } -#endif DEBUGLOG(("insert_buf: Printable(UTF-8):%d",ch)); lbuf[lpos++] = (Uint32) ch; } else if (ch >= 128) { /* not utf8 mode */ @@ -1204,10 +1188,6 @@ static int write_buf(Uint32 *s, int n) if (octbuff != octtmp) { driver_free(octbuff); } -#ifdef HAVE_WCWIDTH - } else if (*s & WIDE_TAG) { - --n; s++; -#endif } else { DEBUGLOG(("write_buf: Very unexpected character %d",(int) *s)); ++n; @@ -1216,7 +1196,7 @@ static int write_buf(Uint32 *s, int n) } /* Check landed in first column of new line and have 'xn' bug. */ n = s - lbuf; - if (COL(n) == 0 && xn && n != 0) { + if (xn && n != 0 && COL(cp_pos_to_col(n)) == 0) { if (n >= llen) { outc(' '); } else if (lastput == 0) { /* A multibyte UTF8 character */ @@ -1246,14 +1226,19 @@ static int outc(int c) return 1; } -static int move_cursor(int from, int to) +static int move_cursor(int from_pos, int to_pos) { + int from_col, to_col; int dc, dl; - update_cols(); - dc = COL(to) - COL(from); - dl = LINE(to) - LINE(from); + from_col = cp_pos_to_col(from_pos); + to_col = cp_pos_to_col(to_pos); + + dc = COL(to_col) - COL(from_col); + dl = LINE(to_col) - LINE(from_col); + DEBUGLOG(("move_cursor: from %d %d to %d %d => %d %d\n", + from_pos, from_col, to_pos, to_col, dl, dc)); if (dl > 0) move_down(dl); else if (dl < 0) @@ -1262,7 +1247,29 @@ static int move_cursor(int from, int to) move_right(dc); else if (dc < 0) move_left(-dc); - return TRUE; + return to_col-from_col; +} + +static int cp_pos_to_col(int cp_pos) +{ +#ifdef HAVE_WCWIDTH + int i; + int col = 0; + + for (i = 0; i < cp_pos; i++) { + int w = wcwidth(lbuf[i]); + if (w > 0) { + col += w; + } + } + return col; +#else + /* + * We dont' have any character width information. Assume that + * code points are one column wide. + */ + return cp_pos; +#endif } static int start_termcap(void) |