ncdirect_putegc(), get true width in ncwidth #1899

NEWS.md

@@ -9,6 +9,9 @@ rearrangements of Notcurses.
   * Documented `ncplane_move_yx()` in `notcurses_plane.3`, and removed the
     false comment that "passing -1 as a coordinate will hold that axis
     constant" from `USGAE.md` and `notcurses.h`. This has never been true.
+  * Added `ncdirect_putegc()` to perform Unicode segmentation. It returns
+    the number of columns consumed, and makes available the number of bytes
+    used by the EGC.
 
 * 2.3.8 (2021-07-04)
   * Marked all capability functions `__attribute__ ((pure))`. If you were

USAGE.md

@@ -440,8 +440,19 @@ int ncdirect_cursor_pop(struct ncdirect* n);
 
 // Formatted printing (plus alignment relative to the terminal).
 int ncdirect_printf_aligned(struct ncdirect* n, int y, ncalign_e align,
-                            const char* fmt, ...)
-  __attribute__ ((format (printf, 4, 5)));
+                            const char* fmt, ...);
+
+// Output the string |utf8| according to the channels |channels|. Note that
+// ncdirect_putstr() does not explicitly flush output buffers, so it will not
+// necessarily be immediately visible.
+int ncdirect_putstr(struct ncdirect* nc, uint64_t channels, const char* utf8);
+
+// Output a single EGC (this might be several characters) from |utf8|,
+// according to the channels |channels|. On success, the number of columns
+// thought to have been used is returned, and if |sbytes| is not NULL,
+// the number of bytes consumed will be written there.
+int ncdirect_putegc(struct ncdirect* nc, uint64_t channels,
+                    const char* utf8, int* sbytes);
 
 // Draw horizontal/vertical lines using the specified channels, interpolating
 // between them as we go. The EGC may not use more than one column. For a

doc/man/man3/notcurses_direct.3.md

@@ -72,6 +72,8 @@ notcurses_direct - minimal notcurses instances for styling text
 
 **int ncdirect_putstr(struct ncdirect* ***nc***, uint64_t ***channels***, const char* ***utf8***);**
 
+**int ncdirect_putegc(struct ncdirect* ***nc***, uint64_t ***channels***, const char* ***utf8***, int* ***sbytes***);**
+
 **int ncdirect_printf_aligned(struct ncdirect* ***n***, int ***y***, ncalign_e ***align***, const char* ***fmt***, ***...***);**
 
 **const char* ncdirect_detected_terminal(const struct ncdirect* ***n***);**
@@ -225,6 +227,10 @@ to **ncdirect_stop**.
 **ncdirect_putstr** and **ncdirect_printf_aligned** return the number of bytes
 written on success. On failure, they return some negative number.
 
+**ncdirect_putegc** returns the number of columns consumed on success, or -1
+on failure. If ***sbytes*** is not **NULL**, the number of bytes consumed
+will be written to it.
+
 **ncdirect_check_pixel_support** returns -1 on error, 0 if there is no pixel
 support, and 1 if pixel support is successfully detected.
 

include/notcurses/direct.h

@@ -113,6 +113,14 @@ API unsigned ncdirect_palette_size(const struct ncdirect* nc)
 API int ncdirect_putstr(struct ncdirect* nc, uint64_t channels, const char* utf8)
   __attribute__ ((nonnull (1, 3)));
 
+// Output a single EGC (this might be several characters) from |utf8|,
+// according to the channels |channels|. On success, the number of columns
+// thought to have been used is returned, and if |sbytes| is not NULL,
+// the number of bytes consumed will be written there.
+API int ncdirect_putegc(struct ncdirect* nc, uint64_t channels,
+                        const char* utf8, int* sbytes)
+  __attribute__ ((nonnull (1, 3)));
+
 // Formatted printing (plus alignment relative to the terminal). Returns the
 // number of columns printed on success.
 API int ncdirect_printf_aligned(struct ncdirect* n, int y, ncalign_e align,

src/lib/direct.c

@@ -46,7 +46,26 @@ int ncdirect_putstr(ncdirect* nc, uint64_t channels, const char* utf8){
   if(activate_channels(nc, channels)){
     return -1;
   }
-  return fprintf(nc->ttyfp, "%s", utf8);
+  return ncfputs(utf8, nc->ttyfp);
+}
+
+int ncdirect_putegc(ncdirect* nc, uint64_t channels, const char* utf8,
+                    int* sbytes){
+  int cols;
+  int bytes = utf8_egc_len(utf8, &cols);
+  if(bytes < 0){
+    return -1;
+  }
+  if(sbytes){
+    *sbytes = bytes;
+  }
+  if(activate_channels(nc, channels)){
+    return -1;
+  }
+  if(fprintf(nc->ttyfp, "%.*s", bytes, utf8) < 0){
+    return -1;
+  }
+  return cols;
 }
 
 int ncdirect_cursor_up(ncdirect* nc, int num){

src/poc/ncwidth.c

@@ -4,6 +4,7 @@
 #include <stdlib.h>
 #include <string.h>
 #include <locale.h>
+#include <notcurses/direct.h>
 
 static int
 add_wchar(wchar_t** wbuf, size_t* bufsize, size_t* used, wchar_t wc){
@@ -21,11 +22,8 @@ add_wchar(wchar_t** wbuf, size_t* bufsize, size_t* used, wchar_t wc){
   return 0;
 }
 
-int main(int argc, char **argv){
-  if(!setlocale(LC_ALL, "")){
-    return EXIT_FAILURE;
-  }
-  if(argc <= 1){
+static int
+defaultout(void){
   for(int i = 0 ; i < 128 ; ++i){
     wchar_t w = i;
     int width = wcwidth(w);
@@ -35,7 +33,16 @@ int main(int argc, char **argv){
     }
   }
   printf("\n");
-    return EXIT_SUCCESS;
+  return 0;
+}
+
+int main(int argc, char **argv){
+  if(argc <= 1){
+    return defaultout() ? EXIT_FAILURE : EXIT_SUCCESS;
+  }
+  struct ncdirect* n;
+  if((n = ncdirect_core_init(NULL, NULL, 0)) == NULL){
+    return EXIT_FAILURE;
   }
   size_t bufsize = 0, used = 0;
   wchar_t* wbuf = NULL;
@@ -52,7 +59,7 @@ int main(int argc, char **argv){
       if(conv == (size_t)-1 || conv == (size_t)-2){
         fprintf(stderr, "Invalid UTF-8: %s\n", arg);
         free(wbuf);
-        return EXIT_FAILURE;
+        goto err;
       }
       int width = wcwidth(w);
       printf("0x%05lx: %d %lc\t", (long)w, width, w);
@@ -66,15 +73,63 @@ int main(int argc, char **argv){
       totalb += conv;
       add_wchar(&wbuf, &bufsize, &used, w);
     }
-    printf("\n total width: %d  total bytes: %zu wcswidth: %d\n\n", totalcols, totalb, wcswidth(wbuf, used));
-    // FIXME this will be broken if totalcols > screen width
-    printf("%s\n", *argv);
-    for(int z = 0 ; z < totalcols ; ++z){
+    int y, x, newy, newx;
+    putchar('\n');
+    ncdirect_cursor_yx(n, &y, &x);
+    printf("%s", *argv);
+    fflush(stdout);
+    ncdirect_cursor_yx(n, &newy, &newx);
+    int realcols = (newx - x) + ncdirect_dim_x(n) * (newy - y);
+    printf("\n iterated wcwidth: %d total bytes: %zu wcswidth: %d true width: %d\n\n",
+           totalcols, totalb, wcswidth(wbuf, used), realcols);
+    ncdirect_cursor_yx(n, &y, &x);
+    // throw up a background color for invisible glyphs
+    uint64_t chan = CHANNELS_RGB_INITIALIZER(0xff, 0xff, 0xff, 0, 0x80, 0);
+    int expy, expx;
+    int misses = 0;
+    int scrolls = 0;
+    while(**argv){
+      int sbytes;
+      int cols;
+      if((cols = ncdirect_putegc(n, chan, *argv, &sbytes)) < 0){
+        goto err;
+      }
+      fflush(stdout);
+      ncdirect_cursor_yx(n, &newy, &newx);
+      if(newy != y){
+        newx += ncdirect_dim_x(n) * (newy - y);
+      }
+      ncdirect_cursor_push(n);
+      if(x + cols != newx){
+        ++misses;
+        for(i = 0 ; i < misses ; ++i){
+          putchar('\v');
+        }
+        printf("True width: %d wcwidth: %d [%.*s]", newx - x, cols, sbytes, *argv);
+        ncdirect_cursor_yx(n, &expy, &expx);
+        scrolls = (newy + misses) - expy;
+        if(scrolls > 1){
+          ncdirect_cursor_up(n, scrolls - 1);
+        }
+      }
+      ncdirect_cursor_pop(n);
+      *argv += sbytes;
+      y = newy - (scrolls - 1);
+      x = newx;
+    }
+    for(i = 0 ; i < misses + 1 ; ++i){
+      putchar('\n');
+    }
+    ncdirect_set_fg_default(n);
+    ncdirect_set_bg_default(n);
+    for(int z = 0 ; z < realcols && z < ncdirect_dim_x(n) ; ++z){
       putchar('0' + z % 10);
     }
+    if(realcols < ncdirect_dim_x(n)){
       putchar('\n');
-    if(totalcols > 20){
-      for(int z = 0 ; z < totalcols ; ++z){
+    }
+    if(realcols > 20){
+      for(int z = 0 ; z < realcols && z < ncdirect_dim_x(n) ; ++z){
         if(z % 10){
           putchar(' ');
         }else{
@@ -85,5 +140,9 @@ int main(int argc, char **argv){
     }
   }
   free(wbuf);
-  return EXIT_SUCCESS;
+  return ncdirect_stop(n) ? EXIT_FAILURE : EXIT_SUCCESS;
+
+err:
+  ncdirect_stop(n);
+  return EXIT_FAILURE;
 }