Add preliminary support for UTF-8 handling in the main view

It is very simple but fixes the issue with unaligned drawing of the line
between the author and log title columns.

diff --git a/tig.c b/tig.c
index ca57f75ca25b8b9408adf91a5049bce21e0ef213..95da3db69d04463ee299a05d3b6f2bbb4d4eec0f 100644 (file)
--- a/tig.c
+++ b/tig.c
@@ -55,6 +55,7 @@
 static void die(const char *err, ...);
 static void report(const char *msg, ...);
 static void set_nonblocking_input(bool loading);
+static size_t utf8_length(const char *string, size_t max_width, int *coloffset, int *trimmed);
 
 #define ABS(x)         ((x) >= 0  ? (x) : -(x))
 #define MIN(x, y)      ((x) < (y) ? (x) :  (y))
@@ -74,6 +75,8 @@ static void set_nonblocking_input(bool loading);
 #define DATE_FORMAT    "%Y-%m-%d %H:%M"
 #define DATE_COLS      STRING_SIZE("2006-04-29 14:21 ")
 
+#define AUTHOR_COLS    20
+
 /* The default interval between line numbers. */
 #define NUMBER_INTERVAL        1
 
@@ -1566,6 +1569,8 @@ main_draw(struct view *view, unsigned int lineno)
        enum line_type type;
        int col = 0;
        size_t timelen;
+       size_t authorlen;
+       int trimmed;
 
        if (view->offset + lineno >= view->lines)
                return FALSE;
@@ -1597,8 +1602,11 @@ main_draw(struct view *view, unsigned int lineno)
        if (type != LINE_CURSOR)
                wattrset(view->win, get_line_attr(LINE_MAIN_AUTHOR));
 
-       if (strlen(commit->author) > 19) {
-               waddnstr(view->win, commit->author, 18);
+       /* FIXME: Make this optional, and add i18n.commitEncoding support. */
+       authorlen = utf8_length(commit->author, AUTHOR_COLS - 2, &col, &trimmed);
+
+       if (trimmed) {
+               waddnstr(view->win, commit->author, authorlen);
                if (type != LINE_CURSOR)
                        wattrset(view->win, get_line_attr(LINE_MAIN_DELIM));
                waddch(view->win, '~');
@@ -1606,7 +1614,7 @@ main_draw(struct view *view, unsigned int lineno)
                waddstr(view->win, commit->author);
        }
 
-       col += 20;
+       col += AUTHOR_COLS;
        if (type != LINE_CURSOR)
                wattrset(view->win, A_NORMAL);
 
@@ -1897,6 +1905,166 @@ get_request(int key)
 }
 
 
+/*
+ * Unicode / UTF-8 handling
+ *
+ * NOTE: Much of the following code for dealing with unicode is derived from
+ * ELinks' UTF-8 code developed by Scrool <scroolik@gmail.com>. Origin file is
+ * src/intl/charset.c from the utf8 branch commit elinks-0.11.0-g31f2c28.
+ */
+
+/* I've (over)annotated a lot of code snippets because I am not entirely
+ * confident that the approach taken by this small UTF-8 interface is correct.
+ * --jonas */
+
+static inline int
+unicode_width(unsigned long c)
+{
+       if (c >= 0x1100 &&
+          (c <= 0x115f                         /* Hangul Jamo */
+           || c == 0x2329
+           || c == 0x232a
+           || (c >= 0x2e80  && c <= 0xa4cf && c != 0x303f)
+                                               /* CJK ... Yi */
+           || (c >= 0xac00  && c <= 0xd7a3)    /* Hangul Syllables */
+           || (c >= 0xf900  && c <= 0xfaff)    /* CJK Compatibility Ideographs */
+           || (c >= 0xfe30  && c <= 0xfe6f)    /* CJK Compatibility Forms */
+           || (c >= 0xff00  && c <= 0xff60)    /* Fullwidth Forms */
+           || (c >= 0xffe0  && c <= 0xffe6)
+           || (c >= 0x20000 && c <= 0x2fffd)
+           || (c >= 0x30000 && c <= 0x3fffd)))
+               return 2;
+
+       return 1;
+}
+
+/* Number of bytes used for encoding a UTF-8 character indexed by first byte.
+ * Illegal bytes are set one. */
+static const unsigned char utf8_bytes[256] = {
+       1,1,1,1,1,1,1,1, 1,1,1,1,1,1,1,1, 1,1,1,1,1,1,1,1, 1,1,1,1,1,1,1,1,
+       1,1,1,1,1,1,1,1, 1,1,1,1,1,1,1,1, 1,1,1,1,1,1,1,1, 1,1,1,1,1,1,1,1,
+       1,1,1,1,1,1,1,1, 1,1,1,1,1,1,1,1, 1,1,1,1,1,1,1,1, 1,1,1,1,1,1,1,1,
+       1,1,1,1,1,1,1,1, 1,1,1,1,1,1,1,1, 1,1,1,1,1,1,1,1, 1,1,1,1,1,1,1,1,
+       1,1,1,1,1,1,1,1, 1,1,1,1,1,1,1,1, 1,1,1,1,1,1,1,1, 1,1,1,1,1,1,1,1,
+       1,1,1,1,1,1,1,1, 1,1,1,1,1,1,1,1, 1,1,1,1,1,1,1,1, 1,1,1,1,1,1,1,1,
+       2,2,2,2,2,2,2,2, 2,2,2,2,2,2,2,2, 2,2,2,2,2,2,2,2, 2,2,2,2,2,2,2,2,
+       3,3,3,3,3,3,3,3, 3,3,3,3,3,3,3,3, 4,4,4,4,4,4,4,4, 5,5,5,5,6,6,1,1,
+};
+
+/* Decode UTF-8 multi-byte representation into a unicode character. */
+static inline unsigned long
+utf8_to_unicode(const char *string, size_t length)
+{
+       unsigned long unicode;
+
+       switch (length) {
+       case 1:
+               unicode  =   string[0];
+               break;
+       case 2:
+               unicode  =  (string[0] & 0x1f) << 6;
+               unicode +=  (string[1] & 0x3f);
+               break;
+       case 3:
+               unicode  =  (string[0] & 0x0f) << 12;
+               unicode += ((string[1] & 0x3f) << 6);
+               unicode +=  (string[2] & 0x3f);
+               break;
+       case 4:
+               unicode  =  (string[0] & 0x0f) << 18;
+               unicode += ((string[1] & 0x3f) << 12);
+               unicode += ((string[2] & 0x3f) << 6);
+               unicode +=  (string[3] & 0x3f);
+               break;
+       case 5:
+               unicode  =  (string[0] & 0x0f) << 24;
+               unicode += ((string[1] & 0x3f) << 18);
+               unicode += ((string[2] & 0x3f) << 12);
+               unicode += ((string[3] & 0x3f) << 6);
+               unicode +=  (string[4] & 0x3f);
+               break;
+       case 6: 
+               unicode  =  (string[0] & 0x01) << 30;
+               unicode += ((string[1] & 0x3f) << 24);
+               unicode += ((string[2] & 0x3f) << 18);
+               unicode += ((string[3] & 0x3f) << 12);
+               unicode += ((string[4] & 0x3f) << 6);
+               unicode +=  (string[5] & 0x3f);
+               break;
+       default:
+               die("Invalid unicode length");
+       }
+
+       /* Invalid characters could return the special 0xfffd value but NUL
+        * should be just as good. */
+       return unicode > 0xffff ? 0 : unicode;
+}
+
+/* Calculates how much of string can be shown within the given maximum width
+ * and sets trimmed parameter to non-zero value if all of string could not be
+ * shown.
+ *
+ * Additionally, adds to coloffset how many many columns to move to align with
+ * the expected position. Takes into account how multi-byte and double-width
+ * characters will effect the cursor position.
+ *
+ * Returns the number of bytes to output from string to satisfy max_width. */
+static size_t
+utf8_length(const char *string, size_t max_width, int *coloffset, int *trimmed)
+{
+       const char *start = string;
+       const char *end = strchr(string, '\0');
+       size_t mbwidth = 0;
+       size_t width = 0;
+
+       *trimmed = 0;
+
+       while (string < end) {
+               int c = *(unsigned char *) string;
+               unsigned char bytes = utf8_bytes[c];
+               size_t ucwidth;
+               unsigned long unicode;
+
+               if (string + bytes > end)
+                       break;
+
+               /* Change representation to figure out whether
+                * it is a single- or double-width character. */
+
+               unicode = utf8_to_unicode(string, bytes);
+               /* FIXME: Graceful handling of invalid unicode character. */
+               if (!unicode)
+                       break;
+
+               ucwidth = unicode_width(unicode);
+               width  += ucwidth;
+               if (width > max_width) {
+                       *trimmed = 1;
+                       break;
+               }
+
+               /* The column offset collects the differences between the
+                * number of bytes encoding a character and the number of
+                * columns will be used for rendering said character.
+                *
+                * So if some character A is encoded in 2 bytes, but will be
+                * represented on the screen using only 1 byte this will and up
+                * adding 1 to the multi-byte column offset.
+                *
+                * Assumes that no double-width character can be encoding in
+                * less than two bytes. */
+               if (bytes > ucwidth)
+                       mbwidth += bytes - ucwidth;
+
+               string  += bytes;
+       }
+
+       *coloffset += mbwidth;
+
+       return string - start;
+}
+
+
 /*
  * Status management
  */
@@ -2316,6 +2484,8 @@ main(int argc, char *argv[])
  * ----
  * Known bugs and problems:
  *
+ * - In it's current state tig is pretty much UTF-8 only.
+ *
  * - If the screen width is very small the main view can draw
  *   outside the current view causing bad wrapping. Same goes
  *   for title and status windows.