- Make original buffer use a disk backed mapping to save memory.
- rearrange api function headers for search etc.
- fix bugs with iterators.
- support uint64_t for all byte and line offsets to allow for much larger files to be opened.
- and more minor bug fixes.
This commit is contained in:
2026-08-04 07:48:26 +01:00
parent 72d0ff85a3
commit 0bfa9300db
19 changed files with 612 additions and 396 deletions
+3
View File
@@ -40,9 +40,12 @@ extern "C" {
#include <shared_mutex>
#include <signal.h>
#include <stack>
#include <stdexcept>
#include <string.h>
#include <string>
#include <sys/ioctl.h>
#include <sys/mman.h>
#include <sys/stat.h>
#include <sys/types.h>
#include <sys/wait.h>
#include <termios.h>
+6 -6
View File
@@ -8,16 +8,16 @@ struct AppendBuffer : Buffer {
using TChunk = char[APPEND_CHUNK_SIZE];
std::vector<TChunk *> buf;
TChunk *t_current;
uint32_t current_offset{0};
uint32_t t_offset{0};
uint64_t current_offset{0};
uint64_t t_offset{0};
AppendBuffer();
~AppendBuffer();
uint32_t key(char c);
uint32_t append(const char *text, uint32_t length);
uint64_t append(char c);
uint64_t append(const char *text, uint64_t length);
const char *read(uint32_t pos, uint32_t *out_len);
inline uint32_t length();
const char *read(uint64_t pos, uint64_t *out_len);
inline uint64_t length();
};
+2 -2
View File
@@ -3,7 +3,7 @@
#include "pch.h"
struct Buffer {
virtual const char *read(uint32_t pos, uint32_t *out_len) = 0;
virtual inline uint32_t length() = 0;
virtual const char *read(uint64_t pos, uint64_t *out_len) = 0;
virtual inline uint64_t length() = 0;
virtual ~Buffer() = default;
};
+7 -7
View File
@@ -4,14 +4,14 @@
#include "pch.h"
struct OriginalBuffer : Buffer {
char *buf;
uint32_t len;
// Add 2 modes here, normal and lazy, lazy copies the file to a safe location and uses mmap.
OriginalBuffer(std::string path);
const char *buf;
uint64_t len;
int fd = -1;
OriginalBuffer();
~OriginalBuffer();
const char *read(uint32_t pos, uint32_t *out_len);
uint32_t length();
void initialize();
const char *read(uint64_t pos, uint64_t *out_len);
uint64_t length();
};
+2 -2
View File
@@ -2,8 +2,8 @@
#include "pch.h"
constexpr uint32_t APPEND_CHUNK_SIZE = 64 * 1024;
constexpr uint32_t PETAL_SIZE_MAX = 32 * 1024;
constexpr uint64_t APPEND_CHUNK_SIZE = 64 * 1024;
constexpr uint64_t PETAL_SIZE_MAX = 32 * 1024;
static_assert(
APPEND_CHUNK_SIZE > PETAL_SIZE_MAX,
+9 -6
View File
@@ -13,15 +13,18 @@ struct ChunkIterator {
Shard *root = nullptr;
std::vector<Shard *> stack;
Petal *petal = nullptr;
uint32_t petal_offset = 0;
uint32_t global_offset = 0;
uint64_t petal_offset = 0;
uint64_t global_offset = 0;
uint64_t global_line = 0;
bool at_end = false;
ChunkIterator(Shard *r, Direction dir);
~ChunkIterator();
void seek_offset(uint32_t offset);
void seek_line(uint32_t offset);
bool next(const char **data, uint32_t *out_len);
uint32_t byte_offset();
void seek_offset(uint64_t offset);
void seek_line(uint64_t offset);
bool next(const char **data, uint64_t *out_len);
uint64_t byte_offset();
uint64_t line_offset();
};
+6 -3
View File
@@ -9,11 +9,14 @@ struct LineIterator {
Direction dir;
const char *chunk;
uint32_t len;
uint64_t len;
uint64_t offset = 0;
uint64_t chunk_offset = 0;
bool at_end = false;
LineIterator(Shard *r, uint32_t start_line, Direction dir);
LineIterator(Shard *r, uint64_t start_line, Direction dir);
~LineIterator() = default;
bool next(std::string &line);
uint32_t byte_offset();
uint64_t byte_offset();
};
-21
View File
@@ -1,21 +0,0 @@
#pragma once
#include "pch.h"
#include "vase/shard.h"
struct RegexGroup {
uint32_t start{UINT32_MAX};
uint32_t end{UINT32_MAX};
};
struct RegexMatch {
uint32_t start;
uint32_t end;
RegexGroup groups[9]{};
};
std::vector<RegexMatch> regex_search(
Shard *root, std::string_view pattern_str,
uint32_t start_offset, uint32_t end_offset,
std::string_view options
);
+10 -9
View File
@@ -14,12 +14,12 @@ struct Shard {
uint8_t height;
uint32_t length;
uint32_t lines;
uint64_t length;
uint64_t lines;
std::atomic_uint32_t refs;
std::atomic_uint64_t refs;
Shard(ShardKind kind, uint32_t length, uint32_t lines, uint8_t height)
Shard(ShardKind kind, uint64_t length, uint64_t lines, uint8_t height)
: kind(kind), height(height), length(length), lines(lines), refs(1) {};
static void retain(Shard *n);
@@ -46,18 +46,19 @@ struct Branch : Shard {
struct Petal : Shard {
Buffer *source;
uint32_t pos;
uint64_t pos;
Petal(uint32_t length, uint32_t lines, Buffer *source, uint32_t pos)
Petal(uint64_t length, uint64_t lines, Buffer *source, uint64_t pos)
: Shard(ShardKind::Petal, length, lines, 1),
source(source), pos(pos) {};
};
Shard *create_shards(Buffer *o);
Shard *create_file_shards(std::string &path, OriginalBuffer *b);
Shard *create_swap_shards(std::string &path, OriginalBuffer *b);
std::pair<Shard *, Shard *> split_shard(Shard *n, uint32_t offset);
std::pair<Shard *, Shard *> split_shard(Shard *n, uint64_t offset);
Shard *concat_shard(Shard *left, Shard *right);
Shard *merge(Shard *a, Shard *b);
Shard *merge_leaves(Shard *a, Shard *b);
Shard *append_leaf(Shard *root, Shard *leaf);
Shard *build_balanced(Shard **pieces, uint32_t lo, uint32_t hi);
Shard *build_balanced(Shard **pieces, uint64_t lo, uint64_t hi);
+52 -17
View File
@@ -10,8 +10,8 @@
#include "utils/utils.h"
struct Point {
uint32_t row;
uint32_t col;
uint64_t row;
uint64_t col;
};
struct Range {
@@ -19,30 +19,52 @@ struct Range {
Point end;
};
struct RegexGroup {
uint64_t start{UINT32_MAX};
uint64_t end{UINT32_MAX};
};
struct RegexMatch {
uint64_t start;
uint64_t end;
RegexGroup groups[9]{};
};
struct ReplacePart {
enum struct PartType {
FullMatch,
CaptureGroup,
Constant
} type;
std::variant<uint8_t, Shard *> value;
};
struct Vase {
OriginalBuffer original;
AppendBuffer append;
OriginalBuffer *original;
AppendBuffer *append;
Shard *root;
Vase(std::string path);
~Vase();
uint32_t length();
uint64_t length();
std::string to_string();
std::string to_string(Range range);
Point resolve_column(uint32_t line, uint32_t column);
Point resolve_column(uint64_t line, uint64_t column);
LineIterator iterate(uint32_t line);
LineIterator iterate(uint64_t line);
void insert(Point *point, char key);
void insert(Point *point, const char *data, uint32_t len);
void erase(Point *point, int64_t amount);
void insert(Point *point, const char *data, uint64_t len);
void erase(Point *point, uint64_t amount, Direction dir);
void erase(Range range);
void replace(Range range, const char *data, uint32_t len);
void replace(Range range, const char *data, uint64_t len);
void move_clusters(Point *point, int64_t amount);
void move_lines(Point *point, int64_t amount);
void move_clusters(Point *point, uint64_t amount, Direction dir);
void move_lines(Point *point, uint64_t amount, Direction dir);
void clamp(Point *point);
void regex_search_replace(
@@ -50,17 +72,30 @@ struct Vase {
std::string_view replace, std::string_view options
);
std::vector<Range> regex_search(
std::string_view pattern, Range range, std::string_view options
);
bool undo();
bool redo();
void snapshot();
void prune_history(uint32_t n);
void prune_history(uint64_t n);
bool save(std::string path);
bool save_swap(std::string path);
uint64_t offset_of(Point point);
Point point_of(uint64_t offset);
private:
std::vector<Shard *> history;
uint32_t history_top;
uint64_t history_top;
uint32_t offset_of(Point point);
static void flatten(Shard *s, std::string &out);
void _insert(Point *point, const char *data, uint32_t len);
void _insert(Point *point, const char *data, uint64_t len);
std::vector<ReplacePart> parse_replace(std::string_view s);
std::vector<RegexMatch> _regex_search(
std::string_view pattern, Range range, std::string_view options
);
};