notcurses_puttext: use libunistring for wordbreaking #772

CMakeLists.txt

@@ -79,18 +79,24 @@ elseif(${USE_OIIO})
 pkg_check_modules(OIIO REQUIRED OpenImageIO>=2.1)
 endif()
 find_library(MATH_LIBRARIES m)
-# don't cache this, or installing it requires clearing the cache to be found
+if(${USE_DOCTEST})
+find_package(doctest 2.3.5 REQUIRED)
+endif()
+
+# don't cache these, or installing them requires clearing the cache to be found.
+# this is going to be true for anything lacking pkg-config/CMake support.
 unset(HAVE_QRCODEGEN_H CACHE)
-check_include_file("qrcodegen/qrcodegen.h" HAVE_QRCODEGEN_H)
+check_include_file("uniwbrk.h" HAVE_UNISTRING_H)
+if(NOT "${HAVE_UNISTRING_H}")
+  message(FATAL_ERROR "Couldn't find uniwbrk.h from GNU libunistring")
+endif()
 if("${USE_QRCODEGEN}")
+check_include_file("qrcodegen/qrcodegen.h" HAVE_QRCODEGEN_H)
 if(NOT "${HAVE_QRCODEGEN_H}")
   message(FATAL_ERROR "USE_QRCODEGEN is active, but couldn't find qrcodegen.h")
 endif()
 endif()
 find_library(LIBRT rt)
-if(${USE_DOCTEST})
-find_package(doctest 2.3.5 REQUIRED)
-endif()
 
 # libnotcurses (core shared library and static library)
 file(GLOB NCSRCS CONFIGURE_DEPENDS src/lib/*.c src/lib/*.cpp)
@@ -129,6 +135,7 @@ target_link_libraries(notcurses
   PRIVATE
     "${TERMINFO_LIBRARIES}"
     "${LIBRT}"
+    unistring
   PUBLIC
     Threads::Threads
 )
@@ -136,6 +143,7 @@ target_link_libraries(notcurses-static
   PRIVATE
     "${TERMINFO_STATIC_LIBRARIES}"
     "${LIBRT}"
+    unistring
   PUBLIC
     Threads::Threads
 )

NEWS.md

@@ -1,6 +1,9 @@
 This document attempts to list user-visible changes and any major internal
 rearrangements of Notcurses.
 
+* 1.6.7 (not yet released)
+  * GNU libunistring is now required to build/load Notcurses.
+
 * 1.6.6 (2020-07-19)
   * `notcurses-pydemo` is now only installed alongside the Python module,
     using setuptools. CMake no longer installs it.

README.md

@@ -114,6 +114,7 @@ that fine library.
 * (build) A C11 and a C++17 compiler
 * (build) CMake 3.14.0+
 * (build+runtime) From NCURSES: terminfo 6.1+
+* (build+runtime) GNU libunistring 0.9.10+
 * (OPTIONAL) (build+runtime) From QR-Code-generator: [libqrcodegen](https://github.com/nayuki/QR-Code-generator) 1.5.0+
 * (OPTIONAL) (build+runtime) From [FFmpeg](https://www.ffmpeg.org/): libswscale 5.0+, libavformat 57.0+, libavutil 56.0+
 * (OPTIONAL) (build+runtime) [OpenImageIO](https://github.com/OpenImageIO/oiio) 2.15.0+

src/lib/direct.cpp

@@ -341,7 +341,7 @@ nc_err_e ncdirect_render_image(ncdirect* n, const char* file, ncalign_e align,
   struct ncplane* faken = ncplane_create(nullptr, nullptr,
                                          disprows / encoding_y_scale(bset),
                                          dispcols / encoding_x_scale(bset),
-                                         0, 0, nullptr);
+                                         0, 0, nullptr, nullptr);
   if(faken == nullptr){
     return NCERR_NOMEM;
   }

src/lib/internal.h

@@ -80,6 +80,7 @@ typedef struct ncplane {
   cell basecell;         // cell written anywhere that fb[i].gcluster == 0
   struct notcurses* nc;  // notcurses object of which we are a part
   bool scrolling;        // is scrolling enabled? always disabled by default
+  char* name;            // used only for debugging
 } ncplane;
 
 #include "blitset.h"
@@ -782,7 +783,7 @@ calc_gradient_channels(uint64_t* channels, uint64_t ul, uint64_t ur,
 // ncvisual_render(), and thus calls these low-level internal functions.
 // they are not for general use -- check ncplane_new() and ncplane_destroy().
 ncplane* ncplane_create(notcurses* nc, ncplane* n, int rows, int cols,
-                        int yoff, int xoff, void* opaque);
+                        int yoff, int xoff, void* opaque, const char* name);
 void free_plane(ncplane* p);
 
 // heap-allocated formatted output

src/lib/notcurses.c

@@ -14,6 +14,7 @@
 #include <stdlib.h>
 #include <signal.h>
 #include <locale.h>
+#include <uniwbrk.h>
 #include <langinfo.h>
 #include <stdatomic.h>
 #include <sys/ioctl.h>
@@ -265,6 +266,7 @@ void free_plane(ncplane* p){
       p->nc->stats.fbbytes -= sizeof(*p->fb) * p->leny * p->lenx;
     }
     egcpool_dump(&p->pool);
+    free(p->name);
     free(p->fb);
     free(p);
   }
@@ -281,7 +283,7 @@ void free_plane(ncplane* p){
 // ncplane created by ncdirect for rendering visuals. in that case (and only in
 // that case), nc is NULL.
 ncplane* ncplane_create(notcurses* nc, ncplane* n, int rows, int cols,
-                        int yoff, int xoff, void* opaque){
+                        int yoff, int xoff, void* opaque, const char* name){
   if(rows <= 0 || cols <= 0){
     return NULL;
   }
@@ -299,6 +301,7 @@ ncplane* ncplane_create(notcurses* nc, ncplane* n, int rows, int cols,
   p->x = p->y = 0;
   p->logrow = 0;
   p->blist = NULL;
+  p->name = name ? strdup(name) : NULL;
   if( (p->boundto = n) ){
     p->absx = xoff + n->absx;
     p->absy = yoff + n->absy;
@@ -339,7 +342,8 @@ ncplane* ncplane_create(notcurses* nc, ncplane* n, int rows, int cols,
 static ncplane*
 create_initial_ncplane(notcurses* nc, int dimy, int dimx){
   nc->stdplane = ncplane_create(nc, NULL, dimy - (nc->margin_t + nc->margin_b),
-                              dimx - (nc->margin_l + nc->margin_r), 0, 0, NULL);
+                                dimx - (nc->margin_l + nc->margin_r), 0, 0, NULL,
+                                "std");
   return nc->stdplane;
 }
 
@@ -352,16 +356,17 @@ const ncplane* notcurses_stdplane_const(const notcurses* nc){
 }
 
 ncplane* ncplane_new(notcurses* nc, int rows, int cols, int yoff, int xoff, void* opaque){
-  return ncplane_create(nc, NULL, rows, cols, yoff, xoff, opaque);
+  return ncplane_create(nc, NULL, rows, cols, yoff, xoff, opaque, NULL);
 }
 
 ncplane* ncplane_bound(ncplane* n, int rows, int cols, int yoff, int xoff, void* opaque){
-  return ncplane_create(n->nc, n, rows, cols, yoff, xoff, opaque);
+  return ncplane_create(n->nc, n, rows, cols, yoff, xoff, opaque, NULL);
 }
 
 ncplane* ncplane_aligned(ncplane* n, int rows, int cols, int yoff,
                          ncalign_e align, void* opaque){
-  return ncplane_create(n->nc, n, rows, cols, yoff, ncplane_align(n, align, cols), opaque);
+  return ncplane_create(n->nc, n, rows, cols, yoff,
+                        ncplane_align(n, align, cols), opaque, NULL);
 }
 
 void ncplane_home(ncplane* n){
@@ -408,7 +413,8 @@ ncplane* ncplane_dup(const ncplane* n, void* opaque){
   const struct notcurses* nc = ncplane_notcurses_const(n);
   const int placey = n->absy - nc->margin_t;
   const int placex = n->absx - nc->margin_l;
-  ncplane* newn = ncplane_create(n->nc, n->boundto, dimy, dimx, placey, placex, opaque);
+  ncplane* newn = ncplane_create(n->nc, n->boundto, dimy, dimx,
+                                 placey, placex, opaque, n->name);
   if(newn){
     if(egcpool_dup(&newn->pool, &n->pool)){
       ncplane_destroy(newn);
@@ -1595,10 +1601,10 @@ int ncplane_hline_interp(ncplane* n, const cell* c, int len,
   return ret;
 }
 
-// FIXME there are more advanced means of wordbreaking within unicode; use them
 static bool
-iswordbreak(wchar_t w){
-  return iswspace(w);
+iswordbreak(wchar_t wchar){
+  int w = uc_wordbreak_property(wchar);
+  return (w == WBP_OTHER || w == WBP_NEWLINE || w == WBP_CR || w == WBP_LF);
 }
 
 static bool
@@ -1623,6 +1629,7 @@ overlong_word(const char* text, int dimx){
   return false;
 }
 
+// FIXME probably best to use u8_wordbreaks() and get all wordbreaks at once...
 int ncplane_puttext(ncplane* n, int y, ncalign_e align, const char* text, size_t* bytes){
   int totalcols = 0;
   // save the beginning for diagnostic

tests/ncplane.cpp

@@ -498,6 +498,7 @@ TEST_CASE("NCPlane") {
     ncplane_cursor_yx(n_, &y, &x);
     REQUIRE(1 == y);
     REQUIRE(dimx == x);
+    // this ought not print anything, since we're at the end of the row
     REQUIRE(0 == ncplane_putstr(n_, STR3));
     REQUIRE(0 == ncplane_cursor_move_yx(n_, 0, 0));
     REQUIRE(0 < ncplane_at_cursor_cell(n_, &testcell)); // want first char of STR1
@@ -534,6 +535,7 @@ TEST_CASE("NCPlane") {
     ncplane_cursor_yx(n_, &y, &x);
     REQUIRE(1 == y);
     REQUIRE(dimx == x);
+    // this ought not print anything, since we're at the end of the row
     REQUIRE(0 == ncplane_putstr(n_, STR3));
     REQUIRE(0 == ncplane_cursor_move_yx(n_, 0, 0));
     REQUIRE(0 < ncplane_at_cursor_cell(n_, &testcell)); // want first char of STR1