Drop the dependency on libunistring

.gitattributes

@@ -3,6 +3,7 @@ kitty/emoji.h linguist-generated=true
 kitty/keys.h linguist-generated=true
 kitty/charsets.c linguist-generated=true
 kitty/key_encoding.py linguist-generated=true
+kitty/unicode-data.c
 kitty/rgb.py linguist-generated=true
 kitty/gl-wrapper.* linguist-generated=true
 kitty/khrplatform.h linguist-generated=true

CHANGELOG.rst

@@ -36,6 +36,8 @@ version 0.7.0 [future]
 
 - Fix incorrect handling of some unicode combining marks that are not re-ordered
 
+- Drop the dependency on libunistring
+
 
 version 0.6.1 [2017-12-28]
 ---------------------------

README.asciidoc

@@ -90,7 +90,6 @@ the following dependencies are installed first.
 
 * python >= 3.5
 * harfbuzz >= 1.5.0
-* libunistring
 * zlib
 * libpng
 * freetype (not needed on macOS)

count-lines-of-code

@@ -1,2 +1,2 @@
 #!/bin/bash
-cloc --exclude-list-file <(echo -e 'kitty/wcwidth-std.h\nkitty/glfw.c\nkitty/keys.h\nkitty/charsets.c\nkitty/key_encoding.py\nkitty/rgb.py\nkitty/gl.h\nkitty/gl-wrapper.h\nkitty/gl-wrapper.c\nkitty/khrplatform.h\nkitty/glfw-wrapper.h\nkitty/glfw-wrapper.c\nkitty/emoji.h') kitty
+cloc --exclude-list-file <(echo -e 'kitty/wcwidth-std.h\nkitty/glfw.c\nkitty/keys.h\nkitty/charsets.c\nkitty/unicode-data.c\nkitty/key_encoding.py\nkitty/rgb.py\nkitty/gl.h\nkitty/gl-wrapper.h\nkitty/gl-wrapper.c\nkitty/khrplatform.h\nkitty/glfw-wrapper.h\nkitty/glfw-wrapper.c\nkitty/emoji.h') kitty

gen-wcwidth.py

@@ -130,6 +130,7 @@ def create_header(path):
     p = partial(print, file=f)
     p('// unicode data, built from the unicode standard on:', date.today())
     p('// see gen-wcwidth.py')
+    if path.endswith('.h'):
         p('#pragma once')
     p('#include "data-types.h"\n')
     p('START_ALLOW_CASE_RANGE')
@@ -160,13 +161,38 @@ def gen_emoji():
         p('\treturn false;\n}')
 
 
+def category_test(name, p, classes, comment, static=False):
+    static = 'static inline ' if static else ''
+    chars = set()
+    for c in classes:
+        chars |= class_maps[c]
+    p(f'{static}bool\n{name}(char_type code) {{')
+    p(f'\t// {comment} ({len(chars)} codepoints)' + ' {{' '{')
+    p('\tswitch(code) {')
+    for spec in get_ranges(list(chars)):
+        write_case(spec, p)
+        p(f'\t\t\treturn true;')
+    p('\t} // }}}\n')
+    p('\treturn false;\n}\n')
+
+
+def gen_ucd():
+    with create_header('kitty/unicode-data.c') as p:
+        p('#include "unicode-data.h"')
+        category_test('is_combining_char', p, {c for c in class_maps if c.startswith('M')}, 'M category (marks)')
+        category_test('is_ignored_char', p, 'Cc Cf Cs'.split(), 'Control characters (Cc Cf Cs)')
+        category_test('is_word_char', p, {c for c in class_maps if c[0] in 'LN'}, 'L and N categories')
+        category_test('is_CZ_category', p, {c for c in class_maps if c[0] in 'CZ'}, 'C and Z categories')
+        category_test('is_P_category', p, {c for c in class_maps if c[0] == 'P'}, 'P category (punctuation)')
+
+
 def gen_wcwidth():
     seen = set()
 
     def add(p, comment, chars_, ret):
         chars = chars_ - seen
         seen.update(chars)
-        p(f'\t\t// {comment} ({len(chars)} codepoints)' + ' {{{')
+        p(f'\t\t// {comment} ({len(chars)} codepoints)' + ' {{' '{')
         for spec in get_ranges(list(chars)):
             write_case(spec, p)
             p(f'\t\t\treturn {ret};')
@@ -194,5 +220,6 @@ def gen_wcwidth():
 parse_ucd()
 parse_emoji()
 parse_eaw()
+gen_ucd()
 gen_wcwidth()
 gen_emoji()

kitty/emoji.h

@@ -1,4 +1,4 @@
-// unicode data, built from the unicode standard on: 2018-01-17
+// unicode data, built from the unicode standard on: 2018-01-18
 // see gen-wcwidth.py
 #pragma once
 #include "data-types.h"

kitty/line.c

@@ -42,9 +42,7 @@ line_text_at(char_type ch, combining_type cc) {
     } else {
         Py_UCS4 buf[3];
         buf[0] = ch; buf[1] = cc & CC_MASK; buf[2] = cc >> 16;
-        Py_UCS4 normalized = normalize(ch, buf[1], buf[2]);
+        ans = PyUnicode_FromKindAndData(PyUnicode_4BYTE_KIND, buf, buf[2] ? 3 : 2);
-        if (normalized) ans = PyUnicode_FromKindAndData(PyUnicode_4BYTE_KIND, &normalized, 1);
-        else ans = PyUnicode_FromKindAndData(PyUnicode_4BYTE_KIND, buf, buf[2] ? 3 : 2);
     }
     return ans;
 }

kitty/unicode-data.h

@@ -1,41 +1,22 @@
 #pragma once
+#include "data-types.h"
 
-#include <unictype.h>
+bool is_combining_char(char_type ch);
-#include <uninorm.h>
+bool is_ignored_char(char_type ch);
-
+bool is_word_char(char_type ch);
-static inline bool
+bool is_CZ_category(char_type);
-is_combining_char(uint32_t ch) {
+bool is_P_category(char_type);
-    return uc_is_general_category_withtable(ch, UC_CATEGORY_MASK_M);
-}
-
-
-static inline bool
-is_ignored_char(uint32_t ch) {
-    return uc_is_general_category_withtable(ch, UC_CATEGORY_MASK_Cc | UC_CATEGORY_MASK_Cf | UC_CATEGORY_MASK_Cs);
-}
-
-static inline bool
-is_word_char(uint32_t ch) {
-    return uc_is_general_category_withtable(ch, UC_CATEGORY_MASK_L | UC_CATEGORY_MASK_N);
-}
 
 static inline bool
 is_url_char(uint32_t ch) {
-    return ch && !uc_is_general_category_withtable(ch, UC_CATEGORY_MASK_C | UC_CATEGORY_MASK_Z);
+    return ch && !is_CZ_category(ch);
-}
-
-static inline uint32_t
-normalize(uint32_t ch, uint32_t cc1, uint32_t cc2) {
-    uint32_t ans = uc_composition(ch, cc1);
-    if (ans && cc2) ans = uc_composition(ans, cc2);
-    return ans;
 }
 
 static inline bool
 can_strip_from_end_of_url(uint32_t ch) {
     // remove trailing punctuation
     return (
-        (uc_is_general_category_withtable(ch, UC_CATEGORY_MASK_P) && ch != '/') ||
+        (is_P_category(ch) && ch != '/') ||
         ch == '>'
     ) ? true : false;
 }

kitty/wcwidth-std.h

@@ -1,4 +1,4 @@
-// unicode data, built from the unicode standard on: 2018-01-17
+// unicode data, built from the unicode standard on: 2018-01-18
 // see gen-wcwidth.py
 #pragma once
 #include "data-types.h"

setup.py

@@ -220,9 +220,7 @@ def kitty_env():
     pylib = get_python_flags(cflags)
     gl_libs = ['-framework', 'OpenGL'] if is_macos else pkg_config('gl', '--libs')
     libpng = pkg_config('libpng', '--libs')
-    ans.ldpaths += pylib + font_libs + gl_libs + libpng + [
+    ans.ldpaths += pylib + font_libs + gl_libs + libpng
-        '-lunistring'
-    ]
     if is_macos:
         ans.ldpaths.extend('-framework Cocoa'.split())
         if is_travis and 'SW' in os.environ: