Add: [OSX] Native text caret handling.

src/os/macosx/string_osx.cpp

@@ -11,6 +11,7 @@
 
 #include "../../stdafx.h"
 #include "string_osx.h"
+#include "../../string_func.h"
 #include "macos.h"
 
 #include <CoreFoundation/CoreFoundation.h>
@@ -56,6 +57,122 @@ int MacOSStringCompare(const char *s1, const char *s2)
 	return (int)res + 2;
 }
 
+
+/* virtual */ void OSXStringIterator::SetString(const char *s)
+{
+	const char *string_base = s;
+
+	this->utf16_to_utf8.clear();
+	this->str_info.clear();
+	this->cur_pos = 0;
+
+	/* CoreText operates on UTF-16, thus we have to convert the input string.
+	 * To be able to return proper offsets, we have to create a mapping at the same time. */
+	std::vector<UniChar> utf16_str;     ///< UTF-16 copy of the string.
+	while (*s != '\0') {
+		size_t idx = s - string_base;
+
+		WChar c = Utf8Consume(&s);
+		if (c < 0x10000) {
+			utf16_str.push_back((UniChar)c);
+		} else {
+			/* Make a surrogate pair. */
+			utf16_str.push_back((UniChar)(0xD800 + ((c - 0x10000) >> 10)));
+			utf16_str.push_back((UniChar)(0xDC00 + ((c - 0x10000) & 0x3FF)));
+			this->utf16_to_utf8.push_back(idx);
+		}
+		this->utf16_to_utf8.push_back(idx);
+	}
+	this->utf16_to_utf8.push_back(s - string_base);
+
+	/* Query CoreText for word and cluster break information. */
+	this->str_info.resize(utf16_to_utf8.size());
+
+	if (utf16_str.size() > 0) {
+		CFStringRef str = CFStringCreateWithCharactersNoCopy(kCFAllocatorDefault, &utf16_str[0], utf16_str.size(), kCFAllocatorNull);
+
+		/* Get cluster breaks. */
+		for (CFIndex i = 0; i < CFStringGetLength(str); ) {
+			CFRange r = CFStringGetRangeOfComposedCharactersAtIndex(str, i);
+			this->str_info[r.location].char_stop = true;
+
+			i += r.length;
+		}
+
+		/* Get word breaks. */
+		CFStringTokenizerRef tokenizer = CFStringTokenizerCreate(kCFAllocatorDefault, str, CFRangeMake(0, CFStringGetLength(str)), kCFStringTokenizerUnitWordBoundary, _osx_locale);
+
+		CFStringTokenizerTokenType tokenType = kCFStringTokenizerTokenNone;
+		while ((tokenType = CFStringTokenizerAdvanceToNextToken(tokenizer)) != kCFStringTokenizerTokenNone) {
+			/* Skip tokens that are white-space or punctuation tokens. */
+			if ((tokenType & kCFStringTokenizerTokenHasNonLettersMask) != kCFStringTokenizerTokenHasNonLettersMask) {
+				CFRange r = CFStringTokenizerGetCurrentTokenRange(tokenizer);
+				this->str_info[r.location].word_stop = true;
+			}
+		}
+
+		CFRelease(tokenizer);
+		CFRelease(str);
+	}
+
+	/* End-of-string is always a valid stopping point. */
+	this->str_info.back().char_stop = true;
+	this->str_info.back().word_stop = true;
+}
+
+/* virtual */ size_t OSXStringIterator::SetCurPosition(size_t pos)
+{
+	/* Convert incoming position to an UTF-16 string index. */
+	size_t utf16_pos = 0;
+	for (size_t i = 0; i < this->utf16_to_utf8.size(); i++) {
+		if (this->utf16_to_utf8[i] == pos) {
+			utf16_pos = i;
+			break;
+		}
+	}
+
+	/* Sanitize in case we get a position inside a grapheme cluster. */
+	while (utf16_pos > 0 && !this->str_info[utf16_pos].char_stop) utf16_pos--;
+	this->cur_pos = utf16_pos;
+
+	return this->utf16_to_utf8[this->cur_pos];
+}
+
+/* virtual */ size_t OSXStringIterator::Next(IterType what)
+{
+	assert(this->cur_pos <= this->utf16_to_utf8.size());
+	assert(what == StringIterator::ITER_CHARACTER || what == StringIterator::ITER_WORD);
+
+	if (this->cur_pos == this->utf16_to_utf8.size()) return END;
+
+	do {
+		this->cur_pos++;
+	} while (this->cur_pos < this->utf16_to_utf8.size() && (what  == ITER_WORD ? !this->str_info[this->cur_pos].word_stop : !this->str_info[this->cur_pos].char_stop));
+
+	return this->cur_pos == this->utf16_to_utf8.size() ? END : this->utf16_to_utf8[this->cur_pos];
+}
+
+/* virtual */ size_t OSXStringIterator::Prev(IterType what)
+{
+	assert(this->cur_pos <= this->utf16_to_utf8.size());
+	assert(what == StringIterator::ITER_CHARACTER || what == StringIterator::ITER_WORD);
+
+	if (this->cur_pos == 0) return END;
+
+	do {
+		this->cur_pos--;
+	} while (this->cur_pos > 0 && (what == ITER_WORD ? !this->str_info[this->cur_pos].word_stop : !this->str_info[this->cur_pos].char_stop));
+
+	return this->utf16_to_utf8[this->cur_pos];
+}
+
+/* static */ StringIterator *OSXStringIterator::Create()
+{
+	if (!MacOSVersionIsAtLeast(10, 5, 0)) return NULL;
+
+	return new OSXStringIterator();
+}
+
 #else
 void MacOSSetCurrentLocaleName(const char *iso_code) {}
 
@@ -63,4 +180,9 @@ int MacOSStringCompare(const char *s1, const char *s2)
 {
 	return 0;
 }
+
+/* static */ StringIterator *OSXStringIterator::Create()
+{
+	return NULL;
+}
 #endif /* (MAC_OS_X_VERSION_MAX_ALLOWED >= MAC_OS_X_VERSION_10_5) */

src/os/macosx/string_osx.h

@@ -12,6 +12,32 @@
 #ifndef STRING_OSX_H
 #define STRING_OSX_H
 
+#include "../../gfx_layout.h"
+#include "../../string_base.h"
+#include <vector>
+
+/** String iterator using CoreText as a backend. */
+class OSXStringIterator : public StringIterator {
+	/** Break info for a character. */
+	struct CharInfo {
+		bool word_stop : 1; ///< Code point is suitable as a word break.
+		bool char_stop : 1; ///< Code point is the start of a grapheme cluster, i.e. a "character".
+	};
+
+	std::vector<CharInfo> str_info;      ///< Break information for each code point.
+	std::vector<size_t>   utf16_to_utf8; ///< Mapping from UTF-16 code point position to index in the UTF-8 source string.
+
+	size_t cur_pos; ///< Current iteration position.
+
+public:
+	virtual void SetString(const char *s);
+	virtual size_t SetCurPosition(size_t pos);
+	virtual size_t Next(IterType what);
+	virtual size_t Prev(IterType what);
+
+	static StringIterator *Create();
+};
+
 
 void MacOSSetCurrentLocaleName(const char *iso_code);
 int MacOSStringCompare(const char *s1, const char *s2);

src/string.cpp

@@ -874,9 +874,19 @@ public:
 	}
 };
 
+#if defined(WITH_COCOA) && !defined(STRGEN) && !defined(SETTINGSGEN)
+/* static */ StringIterator *StringIterator::Create()
+{
+	StringIterator *i = OSXStringIterator::Create();
+	if (i != NULL) return i;
+
+	return new DefaultStringIterator();
+}
+#else
 /* static */ StringIterator *StringIterator::Create()
 {
 	return new DefaultStringIterator();
 }
+#endif /* defined(WITH_COCOA) && !defined(STRGEN) && !defined(SETTINGSGEN) */
 
 #endif