From 3ae52c8cfdb1f8d6e887e9c8c980bbbec7fcb246 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Juan=20Manuel=20Tom=C3=A1s?= <jtomas1815@gmail.com>
Date: Tue, 26 May 2020 03:25:05 -0300
Subject: Add partial support for utf-8

---
 point.cpp | 112 ++++++++++++++++++++++++++++++++++++++++++++++----------------
 1 file changed, 83 insertions(+), 29 deletions(-)

(limited to 'point.cpp')

diff --git a/point.cpp b/point.cpp
index e31f5de..e03b43e 100644
--- a/point.cpp
+++ b/point.cpp
@@ -1,14 +1,11 @@
 #include <stdint.h>
+#include <assert.h>
 #include <stdbool.h>
 
-struct Point {
+class Point {
 	Page *page;
 	uint16_t index;
 
-	Point() : page(nullptr), index(0) {}
-	Point(Page *page) : page(page), index(0) {}
-	Point(const Point& p) : page(p.page), index(p.index) {}
-
 	uint16_t index_to_offset() {
 		if (index < page->gap_start) {
 			return index;
@@ -17,28 +14,20 @@ struct Point {
 		}
 	}
 
-	bool operator==(Point p) {
-		return page == p.page && index == p.index;
+	uint8_t byte() {
+		return index == page->element_count ? (page->next ? page->next->elements[0] : 0) : page->elements[index_to_offset()];
 	}
 
-	uint8_t element() {
-		if (index == page->element_count) {
-			return !page->next ? 0 : page->next->elements[0];
-		} else {
-			return page->elements[index_to_offset()];
-		}
-	}
-
-	void operator++(int) {
+	void next_byte() {
 		if (index < page->element_count) {
 			index++;
 		} else if (page->next) {
-			index = 1;
 			page = page->next;
+			index = 1;
 		}
 	}
 
-	void operator--(int) {
+	void prev_byte() {
 		if (index > 1) {
 			index--;
 		} else if (page->prev) {
@@ -49,10 +38,75 @@ struct Point {
 		}
 	}
 
+	size_t rune_type() {
+		if (byte() & 1 << 7) {
+			if (byte() & 1 << 6) {
+				if (byte() & 1 << 5) {
+					if (byte() & 1 << 4) {
+						return 4;
+					} else {
+						return 3;
+					}
+				} else {
+					return 2;
+				}
+			} else {
+				return 0;
+			}
+		} else {
+			return 1;
+		}
+	}
+
+	public:
+
+	Point() : page(new Page()), index(0) {}
+	Point(const Point& p) : page(p.page), index(p.index) {}
+
+	bool operator==(Point p) {
+		return page == p.page && index == p.index;
+	}
+
+	bool operator!=(Point p) {
+		return page != p.page || index != p.index;
+	}
+
+	bool at_start() {
+		return index == 0;
+	}
+
+	bool at_end() {
+		return index == page->element_count && !page->next;
+	}
+
+	void operator++(int) {
+		do {
+			next_byte();
+		} while (!rune_type());
+	}
+
+	void operator--(int) {
+		do {
+			prev_byte();
+		} while (!rune_type());
+	}
+
+	wchar_t element() {
+		size_t type = rune_type();
+		wchar_t rune = byte() & (0xff >> type);
+		Point iter(*this);
+		for (size_t i = 1; i < type; i++) {
+			rune <<= 6;
+			iter.next_byte();
+			rune |= (iter.byte() & 0x3f);
+		}
+		return rune;
+	}
+
 	uint64_t seek(uint8_t c, uint64_t limit) {
 		uint64_t travel_distance = 0;
-		while (element() && element() != c && travel_distance < limit) {
-			(*this)++;
+		while (!at_end() && byte() != c && travel_distance < limit) {
+			next_byte();
 			travel_distance++;
 		}
 		return travel_distance;
@@ -60,8 +114,8 @@ struct Point {
 
 	uint64_t rseek(uint8_t c, uint64_t limit) {
 		uint64_t travel_distance = 0;
-		while (index != 0 && element() != c && travel_distance < limit) {
-			(*this)--;
+		while (!at_start() && byte() != c && travel_distance < limit) {
+			prev_byte();
 			travel_distance++;
 		}
 		return travel_distance;
@@ -69,15 +123,15 @@ struct Point {
 
 	void align_gap() {
 		while (page->gap_end < index_to_offset()) {
-			(*page)++;
+			page->move_gap_forward();
 		}
 		while (page->gap_end > index_to_offset()) {
-			(*page)--;
+			page->move_gap_backward();
 		}
 	}
 
 	void push(uint8_t c) {
-		if (page->gap_start == page->gap_end) {
+		if (page->is_full()) {
 			page->split();
 			if (index >= PAGE_SIZE / 2) {
 				page = page->next;
@@ -86,13 +140,13 @@ struct Point {
 		}
 		align_gap();
 		page->push(c);
-		(*this)++;
+		next_byte();
 	}
 
 	void pop() {
 		if (page->element_count == 1 && index == 1) {
 			if (page->prev) {
-				(*this)--;
+				prev_byte();
 				delete page->next;
 			} else if (page->next) {
 				page->next->copy_to(page);
@@ -100,12 +154,12 @@ struct Point {
 				index = 0;
 			} else {
 				page->pop();
-				index = 0;
+				prev_byte();
 			}
 		} else if (index > 0) {
 			align_gap();
 			page->pop();
-			(*this)--;
+			prev_byte();
 		}
 	}
 
-- 
cgit v1.2.3