MarcoBuster/Gadgetbridge
1
0
Fork 0
mirror of https://codeberg.org/Freeyourgadget/Gadgetbridge synced 2024-07-23 23:32:23 +02:00
Code Releases Wiki Activity

Bangle.js: when converting text to bitmaps, try and split up emoji so we can render just those in color.

Browse Source 
Rendering text as well uses more memory but means text is rendered white, which doesn't work on some backgrounds
This commit is contained in:
Gordon Williams 2024-06-03 15:58:05 +01:00
parent 4459b30ee6
commit 0c19010ca2
1 changed files with 43 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

43
app/src/main/java/nodomain/freeyourgadget/gadgetbridge/service/devices/banglejs/BangleJSDeviceSupport.java
View File

@ -1236,6 +1236,21 @@ public class BangleJSDeviceSupport extends AbstractBTLEDeviceSupport {
return renderUnicodeWordPartAsImage(word);
}
/* is the given character code an emoji? Note that `char` is only 0..65535, so many emoji will
actually be made of 2 surrogate chars. To work around this, we just assume that anything with a surrogate
is an emoji, which will be true in most cases */
public boolean isCharCodeEmoji(char ch) {
if (ch>=0x2190 && ch<=0x21FF) return true;
if (ch>=0x2600 && ch<=0x26FF) return true;
if (ch>=0x2700 && ch<=0x27BF) return true;
if (ch>=0x3000 && ch<=0x303F) return true;
if (ch>=0xD800 && ch<=0xDFFF) return true; // high/low surrogate
if (ch>=0xFE00 && ch<=0xFE0F) return true; // variation selector 1 (2 is U+E0100 so is in a surrogate)
//if (ch>=0x1F300 && ch<=0x1F64F) return true; // needs a surrogate
//if (ch>=0x1F680 && ch<=0x1F6FF) return true; // needs a surrogate
return false;
}
public String renderUnicodeAsImage(String txt) {
// FIXME: it looks like we could implement this as customStringFilter now so it happens automatically
if (txt==null) return null;
@ -1250,6 +1265,7 @@ public class BangleJSDeviceSupport extends AbstractBTLEDeviceSupport {
String word = "";
StringBuilder result = new StringBuilder();
boolean needsTranslate = false;
boolean wordIsAllEmoji = true;
for (int i=0;i<txt.length();i++) {
char ch = txt.charAt(i);
// Special cases where we can just use a built-in character...
@ -1261,6 +1277,32 @@ public class BangleJSDeviceSupport extends AbstractBTLEDeviceSupport {
else if (ch =='】') ch=']';
else if (ch=='‘' || ch=='’' || ch=='‛' || ch=='′' || ch=='ʹ') ch='\'';
else if (ch=='“' || ch=='”' || ch =='„' || ch=='‟' || ch=='″') ch='"';
else if (ch == 0xFEFF) continue; // nonbreaking space - ignore
boolean isCharEmoji = isCharCodeEmoji(ch);
if (isCharEmoji) {
if (!wordIsAllEmoji) {
// if the word is all emoji we are fine, just fall through.
// but here we have characters already and we want to translate those
// separately, so we can have a color emoji on its own.
if (needsTranslate) { // convert word
LOG.info("renderUnicodeAsImage converting " + word);
result.append(renderUnicodeWordAsImage(word));
} else { // or just copy across
result.append(word);
}
word = "";
wordIsAllEmoji = true;
}
needsTranslate = true;
} else {
if (!word.isEmpty() && wordIsAllEmoji) {
// this isn't am emoji, but it follows one - render that separately!
result.append(renderUnicodeWordAsImage(word));
word = "";
needsTranslate = false;
}
wordIsAllEmoji = ch>=0xFE00;
}
// chars which break words up
if (" -_/:.,?!'\"&*()[]".indexOf(ch)>=0) {
// word split
@ -1272,6 +1314,7 @@ public class BangleJSDeviceSupport extends AbstractBTLEDeviceSupport {
}
word = "";
needsTranslate = false;
wordIsAllEmoji = true;
} else {
// TODO: better check?
if (ch>255) needsTranslate = true;