motiejus/turbonss
1
Fork 0
Code Releases Activity

packing shell sections

Browse Source
This commit is contained in:
Motiejus Jakštys
2022-03-03 18:05:46 +02:00
committed by Motiejus Jakštys
parent e1bdb6c529
commit a4e3e08f5f
4 changed files with 104 additions and 37 deletions
Download Patch File Download Diff File Expand all files Collapse all files

48
src/shell.zig
View File

@@ -7,11 +7,22 @@ const StringHashMap = std.StringHashMap;
const BoundedArray = std.BoundedArray;
const StringContext = std.hash_map.StringContext;
// MaxShells is the maximum number of "popular" shells.
pub const MaxShells = 63;
pub const MaxShellLen = 64;
// maxShells is the maximum number of "popular" shells.
pub const max_shells = 63;
pub const max_shell_len = 64;
const ShellAlignment = 2; // bits
// ShellIndex is an index to the shell strings. As shell can be up to 64 bytes
// (1<<6), maximum number of shells is 63 (1<<6-1), the maximum location offset
// is 1<<12. To make location resolvable in 10 bits, all shells will be padded
// to 4 bytes.
// The actual shell length is len+1: we don't allow empty shells, and the real
// length of the shell is 1-64 bytes.
const ShellIndex = packed struct {
offset: u10,
len: u6,
};
// ShellReader interprets "Shell Index" and "Shell Blob" sections.
pub const ShellReader = struct {
sectionIndex: []const ShellIndex,
@@ -44,19 +55,20 @@ pub const ShellWriter = struct {
};
const ShellSections = struct {
index: BoundedArray(ShellIndex, MaxShells),
blob: BoundedArray(u8, MaxShells * MaxShellLen),
index: BoundedArray(ShellIndex, max_shells),
blob: BoundedArray(u8, max_shells * max_shell_len),
indices: StringHashMap(u6),
// initializes and populates shell sections. All strings are copied,
// nothing is owned.
pub const initErr = Allocator.Error || error{Overflow};
pub fn init(
allocator: Allocator,
shells: BoundedArray([]const u8, MaxShells),
) !ShellSections {
shells: BoundedArray([]const u8, max_shells),
) initErr!ShellSections {
var self = ShellSections{
.index = try BoundedArray(ShellIndex, MaxShells).init(shells.len),
.blob = try BoundedArray(u8, MaxShells * MaxShellLen).init(0),
.index = try BoundedArray(ShellIndex, max_shells).init(shells.len),
.blob = try BoundedArray(u8, max_shells * max_shell_len).init(0),
.indices = StringHashMap(u6).init(allocator),
};
var fullOffset: u12 = 0;
@@ -132,7 +144,8 @@ pub const ShellWriter = struct {
// toOwnedSections returns the analyzed ShellSections. Resets the shell
// popularity contest. ShellSections memory is allocated by the ShellWriter
// allocator, and must be deInit'ed by the caller.
pub fn toOwnedSections(self: *ShellWriter, limit: u10) !ShellSections {
const toOwnedSectionsErr = Allocator.Error || error{Overflow};
pub fn toOwnedSections(self: *ShellWriter, limit: u10) toOwnedSectionsErr!ShellSections {
var deque = PriorityDequeue(KV, void, cmpShells).init(self.allocator, {});
defer deque.deinit();
@@ -145,7 +158,7 @@ pub const ShellWriter = struct {
}
const total = std.math.min(deque.count(), limit);
var topShells = try BoundedArray([]const u8, MaxShells).init(total);
var topShells = try BoundedArray([]const u8, max_shells).init(total);
var i: u32 = 0;
while (i < total) : (i += 1) {
@@ -161,17 +174,6 @@ pub const ShellWriter = struct {
}
};
// ShellIndex is an index to the shell strings. As shell can be up to 64 bytes
// (1<<6), maximum number of shells is 63 (1<<6-1), the maximum location offset
// is 1<<12. To make location resolvable in 10 bits, all shells will be padded
// to 4 bytes.
// The actual shell length is len+1: we don't allow empty shells, and the real
// length of the shell is 1-64 bytes.
const ShellIndex = packed struct {
offset: u10,
len: u6,
};
const testing = std.testing;
test "basic shellpopcon" {
@@ -192,7 +194,7 @@ test "basic shellpopcon" {
try popcon.put(shell);
}
var sections = try popcon.toOwnedSections(MaxShells);
var sections = try popcon.toOwnedSections(max_shells);
defer sections.deinit();
try testing.expectEqual(sections.index.len, 3); // all but "nobody" qualify