From ef6f4c2fbf0d4adcb4698db6b54372a3499aeb7c Mon Sep 17 00:00:00 2001
From: Kovid Goyal <kovid@kovidgoyal.net>
Date: Thu, 7 Dec 2017 09:52:48 +0530
Subject: [PATCH] Improve fontconfig fallback font queries for emoji

---
 .gitattributes      |   1 +
 count-lines-of-code |   2 +-
 gen-emoji.py        |  75 ++++++++++++
 kitty/emoji.h       | 287 ++++++++++++++++++++++++++++++++++++++++++++
 kitty/fontconfig.c  |   8 +-
 5 files changed, 369 insertions(+), 4 deletions(-)
 create mode 100755 gen-emoji.py
 create mode 100644 kitty/emoji.h

diff --git a/.gitattributes b/.gitattributes
index 1b5a56bde..aca9cf858 100644
--- a/.gitattributes
+++ b/.gitattributes
@@ -1,4 +1,5 @@
 kitty/wcwidth9.h linguist-generated=true
+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
diff --git a/count-lines-of-code b/count-lines-of-code
index 0d5c4b1a3..2181d9694 100755
--- a/count-lines-of-code
+++ b/count-lines-of-code
@@ -1,2 +1,2 @@
 #!/bin/bash
-cloc --exclude-list-file <(echo -e 'kitty/wcwidth9.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') kitty
+cloc --exclude-list-file <(echo -e 'kitty/wcwidth9.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
diff --git a/gen-emoji.py b/gen-emoji.py
new file mode 100755
index 000000000..547cdc07b
--- /dev/null
+++ b/gen-emoji.py
@@ -0,0 +1,75 @@
+#!/usr/bin/env python3
+# vim:fileencoding=utf-8
+# License: GPL v3 Copyright: 2017, Kovid Goyal <kovid at kovidgoyal.net>
+
+import os
+from collections import defaultdict
+from functools import partial
+from itertools import groupby
+from operator import itemgetter
+from urllib.request import urlopen
+
+os.chdir(os.path.dirname(os.path.abspath(__file__)))
+
+raw = urlopen('http://unicode.org/Public/emoji/5.0/emoji-data.txt').read().decode('utf-8')
+seen = set()
+cmap = defaultdict(set)
+for line in raw.splitlines():
+    line = line.strip()
+    if not line or line.startswith('#'):
+        continue
+    spec, rest = line.partition(';')[::2]
+    spec, rest = spec.strip(), rest.strip().split(' ', 1)[0].strip()
+    if '.' in spec:
+        spec = tuple(map(lambda x: int(x, 16), filter(None, spec.split('.'))))
+        spec = set(range(spec[0], spec[1] + 1))
+    else:
+        spec = {int(spec, 16)}
+    cmap[rest] |= spec
+    seen |= spec
+items = list(seen)
+
+
+def get_ranges(items):
+    items.sort()
+    for k, g in groupby(enumerate(items), lambda m: m[0]-m[1]):
+        group = tuple(map(itemgetter(1), g))
+        a, b = group[0], group[-1]
+        if a == b:
+            yield a
+        else:
+            yield a, b
+
+
+def write_case(spec, p):
+    if isinstance(spec, tuple):
+        p('\t\tcase 0x{:x} ... 0x{:x}:'.format(*spec))
+    else:
+        p('\t\tcase 0x{:x}:'.format(spec))
+
+
+with open('kitty/emoji.h', 'w') as f:
+    p = partial(print, file=f)
+    p('#pragma once')
+    p('#include "data-types.h"\n')
+    p('START_ALLOW_CASE_RANGE')
+    p('static inline bool is_emoji(uint32_t code) {')
+    p('\tswitch(code) {')
+    for spec in get_ranges(items):
+        last = spec[1] if isinstance(spec, tuple) else spec
+        if last < 0x231a:
+            continue
+        write_case(spec, p)
+        p('\t\t\treturn true;')
+    p('\t\tdefault: return false;')
+    p('\t}')
+    p('\treturn false; \n}')
+    p('static inline bool is_emoji_modifier(uint32_t code) {')
+    p('\tswitch(code) {')
+    for spec in get_ranges(list(cmap['Emoji_Modifier'])):
+        write_case(spec, p)
+        p('\t\t\treturn true;')
+    p('\t\tdefault: return false;')
+    p('\t}')
+    p('\treturn false; \n}')
+    p('END_ALLOW_CASE_RANGE')
diff --git a/kitty/emoji.h b/kitty/emoji.h
new file mode 100644
index 000000000..fe76e1164
--- /dev/null
+++ b/kitty/emoji.h
@@ -0,0 +1,287 @@
+#pragma once
+#include "data-types.h"
+
+START_ALLOW_CASE_RANGE
+static inline bool is_emoji(uint32_t code) {
+	switch(code) {
+		case 0x231a ... 0x231b:
+			return true;
+		case 0x2328:
+			return true;
+		case 0x23cf:
+			return true;
+		case 0x23e9 ... 0x23f3:
+			return true;
+		case 0x23f8 ... 0x23fa:
+			return true;
+		case 0x24c2:
+			return true;
+		case 0x25aa ... 0x25ab:
+			return true;
+		case 0x25b6:
+			return true;
+		case 0x25c0:
+			return true;
+		case 0x25fb ... 0x25fe:
+			return true;
+		case 0x2600 ... 0x2604:
+			return true;
+		case 0x260e:
+			return true;
+		case 0x2611:
+			return true;
+		case 0x2614 ... 0x2615:
+			return true;
+		case 0x2618:
+			return true;
+		case 0x261d:
+			return true;
+		case 0x2620:
+			return true;
+		case 0x2622 ... 0x2623:
+			return true;
+		case 0x2626:
+			return true;
+		case 0x262a:
+			return true;
+		case 0x262e ... 0x262f:
+			return true;
+		case 0x2638 ... 0x263a:
+			return true;
+		case 0x2640:
+			return true;
+		case 0x2642:
+			return true;
+		case 0x2648 ... 0x2653:
+			return true;
+		case 0x2660:
+			return true;
+		case 0x2663:
+			return true;
+		case 0x2665 ... 0x2666:
+			return true;
+		case 0x2668:
+			return true;
+		case 0x267b:
+			return true;
+		case 0x267f:
+			return true;
+		case 0x2692 ... 0x2697:
+			return true;
+		case 0x2699:
+			return true;
+		case 0x269b ... 0x269c:
+			return true;
+		case 0x26a0 ... 0x26a1:
+			return true;
+		case 0x26aa ... 0x26ab:
+			return true;
+		case 0x26b0 ... 0x26b1:
+			return true;
+		case 0x26bd ... 0x26be:
+			return true;
+		case 0x26c4 ... 0x26c5:
+			return true;
+		case 0x26c8:
+			return true;
+		case 0x26ce ... 0x26cf:
+			return true;
+		case 0x26d1:
+			return true;
+		case 0x26d3 ... 0x26d4:
+			return true;
+		case 0x26e9 ... 0x26ea:
+			return true;
+		case 0x26f0 ... 0x26f5:
+			return true;
+		case 0x26f7 ... 0x26fa:
+			return true;
+		case 0x26fd:
+			return true;
+		case 0x2702:
+			return true;
+		case 0x2705:
+			return true;
+		case 0x2708 ... 0x270d:
+			return true;
+		case 0x270f:
+			return true;
+		case 0x2712:
+			return true;
+		case 0x2714:
+			return true;
+		case 0x2716:
+			return true;
+		case 0x271d:
+			return true;
+		case 0x2721:
+			return true;
+		case 0x2728:
+			return true;
+		case 0x2733 ... 0x2734:
+			return true;
+		case 0x2744:
+			return true;
+		case 0x2747:
+			return true;
+		case 0x274c:
+			return true;
+		case 0x274e:
+			return true;
+		case 0x2753 ... 0x2755:
+			return true;
+		case 0x2757:
+			return true;
+		case 0x2763 ... 0x2764:
+			return true;
+		case 0x2795 ... 0x2797:
+			return true;
+		case 0x27a1:
+			return true;
+		case 0x27b0:
+			return true;
+		case 0x27bf:
+			return true;
+		case 0x2934 ... 0x2935:
+			return true;
+		case 0x2b05 ... 0x2b07:
+			return true;
+		case 0x2b1b ... 0x2b1c:
+			return true;
+		case 0x2b50:
+			return true;
+		case 0x2b55:
+			return true;
+		case 0x3030:
+			return true;
+		case 0x303d:
+			return true;
+		case 0x3297:
+			return true;
+		case 0x3299:
+			return true;
+		case 0x1f004:
+			return true;
+		case 0x1f0cf:
+			return true;
+		case 0x1f170 ... 0x1f171:
+			return true;
+		case 0x1f17e ... 0x1f17f:
+			return true;
+		case 0x1f18e:
+			return true;
+		case 0x1f191 ... 0x1f19a:
+			return true;
+		case 0x1f1e6 ... 0x1f1ff:
+			return true;
+		case 0x1f201 ... 0x1f202:
+			return true;
+		case 0x1f21a:
+			return true;
+		case 0x1f22f:
+			return true;
+		case 0x1f232 ... 0x1f23a:
+			return true;
+		case 0x1f250 ... 0x1f251:
+			return true;
+		case 0x1f300 ... 0x1f321:
+			return true;
+		case 0x1f324 ... 0x1f393:
+			return true;
+		case 0x1f396 ... 0x1f397:
+			return true;
+		case 0x1f399 ... 0x1f39b:
+			return true;
+		case 0x1f39e ... 0x1f3f0:
+			return true;
+		case 0x1f3f3 ... 0x1f3f5:
+			return true;
+		case 0x1f3f7 ... 0x1f4fd:
+			return true;
+		case 0x1f4ff ... 0x1f53d:
+			return true;
+		case 0x1f549 ... 0x1f54e:
+			return true;
+		case 0x1f550 ... 0x1f567:
+			return true;
+		case 0x1f56f ... 0x1f570:
+			return true;
+		case 0x1f573 ... 0x1f57a:
+			return true;
+		case 0x1f587:
+			return true;
+		case 0x1f58a ... 0x1f58d:
+			return true;
+		case 0x1f590:
+			return true;
+		case 0x1f595 ... 0x1f596:
+			return true;
+		case 0x1f5a4 ... 0x1f5a5:
+			return true;
+		case 0x1f5a8:
+			return true;
+		case 0x1f5b1 ... 0x1f5b2:
+			return true;
+		case 0x1f5bc:
+			return true;
+		case 0x1f5c2 ... 0x1f5c4:
+			return true;
+		case 0x1f5d1 ... 0x1f5d3:
+			return true;
+		case 0x1f5dc ... 0x1f5de:
+			return true;
+		case 0x1f5e1:
+			return true;
+		case 0x1f5e3:
+			return true;
+		case 0x1f5e8:
+			return true;
+		case 0x1f5ef:
+			return true;
+		case 0x1f5f3:
+			return true;
+		case 0x1f5fa ... 0x1f64f:
+			return true;
+		case 0x1f680 ... 0x1f6c5:
+			return true;
+		case 0x1f6cb ... 0x1f6d2:
+			return true;
+		case 0x1f6e0 ... 0x1f6e5:
+			return true;
+		case 0x1f6e9:
+			return true;
+		case 0x1f6eb ... 0x1f6ec:
+			return true;
+		case 0x1f6f0:
+			return true;
+		case 0x1f6f3 ... 0x1f6f8:
+			return true;
+		case 0x1f910 ... 0x1f93a:
+			return true;
+		case 0x1f93c ... 0x1f93e:
+			return true;
+		case 0x1f940 ... 0x1f945:
+			return true;
+		case 0x1f947 ... 0x1f94c:
+			return true;
+		case 0x1f950 ... 0x1f96b:
+			return true;
+		case 0x1f980 ... 0x1f997:
+			return true;
+		case 0x1f9c0:
+			return true;
+		case 0x1f9d0 ... 0x1f9e6:
+			return true;
+		default: return false;
+	}
+	return false; 
+}
+static inline bool is_emoji_modifier(uint32_t code) {
+	switch(code) {
+		case 0x1f3fb ... 0x1f3ff:
+			return true;
+		default: return false;
+	}
+	return false; 
+}
+END_ALLOW_CASE_RANGE
diff --git a/kitty/fontconfig.c b/kitty/fontconfig.c
index ac1906a15..5ae94bdcb 100644
--- a/kitty/fontconfig.c
+++ b/kitty/fontconfig.c
@@ -9,6 +9,7 @@
 #include "lineops.h"
 #include "fonts.h"
 #include <fontconfig/fontconfig.h>
+#include "emoji.h"
 #ifndef FC_COLOR
 #define FC_COLOR "color"
 #endif
@@ -190,9 +191,10 @@ create_fallback_face(PyObject UNUSED *base_face, Cell* cell, bool bold, bool ita
     PyObject *ans = NULL;
     FcPattern *pat = FcPatternCreate();
     if (pat == NULL) return PyErr_NoMemory();
-    AP(FcPatternAddString, FC_FAMILY, (const FcChar8*)"monospace", "family");
-    if (bold) { AP(FcPatternAddInteger, FC_WEIGHT, FC_WEIGHT_BOLD, "weight"); }
-    if (italic) { AP(FcPatternAddInteger, FC_SLANT, FC_SLANT_ITALIC, "slant"); }
+    bool emoji = is_emoji(cell->ch);
+    AP(FcPatternAddString, FC_FAMILY, (const FcChar8*)(emoji ? "emoji" : "monospace"), "family");
+    if (!emoji && bold) { AP(FcPatternAddInteger, FC_WEIGHT, FC_WEIGHT_BOLD, "weight"); }
+    if (!emoji && italic) { AP(FcPatternAddInteger, FC_SLANT, FC_SLANT_ITALIC, "slant"); }
     size_t num = cell_as_unicode(cell, true, char_buf, ' ');
     add_charset(pat, num);
     PyObject *d = _fc_match(pat);