further along with popcon

src/shellpop.zig

@@ -17,11 +17,10 @@ const ShellIndex = struct {
 
 // MaxShells is the maximum number of "popular" shells.
 const MaxShells = 63;
+const MaxShellLen = 64;
 
 // ShellPopcon is a shell popularity contest: collect shells and return the
 // popular ones, sorted by score. score := len(shell) * number_of_shells.
-// String values are copied, the returned slice of shells is allocated
-// using an allocator.
 const ShellPopcon = struct {
     counts: std.StringHashMap(u32),
     allocator: Allocator,
@@ -32,20 +31,50 @@ const ShellPopcon = struct {
     };
 
     const ShellSections = struct {
-        index: []ShellIndex,
-        blob: []const u8,
+        index: BoundedArray(ShellIndex, MaxShells),
+        blob: BoundedArray(u8, MaxShells * MaxShellLen),
 
         offsets: StringHashMap(u10),
 
-        pub fn getOffset(self: *ShellSections, shell: []const u8) ?u10 {
-            return self.offsets.get(shell);
+        // initializes and populates shell sections. All strings are copied, nothing is owned.
+        pub fn init(allocator: Allocator, shells: BoundedArray([]const u8, MaxShells)) !ShellSections {
+            var self = ShellSections{
+                .index = try BoundedArray(ShellIndex, MaxShells).init(shells.len),
+                .blob = try BoundedArray(u8, MaxShells * MaxShellLen).init(0),
+                .offsets = StringHashMap(u10).init(allocator),
+            };
+            var offset: u10 = 0;
+            var idx: u10 = 0;
+            while (idx < shells.len) {
+                //const stderr = std.io.getStdErr().writer();
+                //try stderr.print("\n", .{});
+
+                const len = @intCast(u6, shells.get(idx).len);
+                try self.blob.appendSlice(shells.get(idx));
+                const ourShell = self.blob.constSlice()[offset .. offset + len];
+                try self.offsets.put(ourShell, idx);
+                self.index.set(idx, ShellIndex{
+                    .offset = offset >> 2, // all shells are padded by 4
+                    .len = len,
+                });
+                offset += len;
+
+                // if offset is not multiple by 4, make it so, and append the offset.
+                const padding = (offset + 3) & ~@intCast(u10, 3);
+                offset += padding;
+                try self.blob.appendNTimes(0, padding);
+                idx += 1;
+            }
+            return self;
         }
 
-        // initializes ShellSections. All strings are copied, nothing is owned.
-        pub fn init(allocator: Allocator, shells: BoundedArray([]const u8, MaxShells)) ShellSections {
-            self.offsets = StringHashMap(u10).init(allocator);
-            _ = allocator;
-            _ = shells;
+        pub fn deinit(self: *ShellSections) void {
+            self.offsets.deinit();
+            self.* = undefined;
+        }
+
+        pub fn getOffset(self: *ShellSections, shell: []const u8) ?u10 {
+            return self.offsets.get(shell);
         }
     };
 
@@ -82,10 +111,10 @@ const ShellPopcon = struct {
         return std.math.order(a.score, b.score);
     }
 
-    pub fn getSections(self: *Self, limit: u32) ShellSections {
-        const stderr = std.io.getStdErr().writer();
-        _ = stderr;
-
+    // toOwnedSections returns the analyzed ShellSections. Resets the shell
+    // popularity contest. ShellSections memory is allocated by the ShellPopcon
+    // allocator, and must be deInit'ed by the caller.
+    pub fn toOwnedSections(self: *Self, limit: u10) !ShellSections {
         var deque = PriorityDequeue(KV, void, cmpShells).init(self.allocator, {});
         defer deque.deinit();
 
@@ -99,55 +128,45 @@ const ShellPopcon = struct {
         }
 
         const total = std.math.min(deque.count(), limit);
-        var strSlice = self.allocator.alloc([]u8, total);
-        defer strSlice.deinit();
+        var topShells = try BoundedArray([]const u8, MaxShells).init(total);
 
         var i: u32 = 0;
         while (i < total) {
-            strSlice[i] = deque.removeMax();
+            const elem: []const u8 = deque.removeMax().shell;
+            topShells.set(i, elem);
             i += 1;
         }
 
-        return ShellSections.init(self.allocator, strSlice);
+        const result = ShellSections.init(self.allocator, topShells);
+        const allocator = self.allocator;
+        self.* = init(allocator);
+        return result;
     }
 };
 
-test "[]u8 comparison" {
-    var s1: []const u8 = "/bin/bash";
-    var s2: []const u8 = "/bin/bash";
-    try testing.expectEqual(s1, s2);
-}
-
-test "basic shellpop" {
+test "basic shellpopcon" {
     var popcon = ShellPopcon.init(testing.allocator);
     defer popcon.deinit();
 
-    try popcon.put("/bin/bash");
-    try popcon.put("/bin/bash");
-    try popcon.put("/bin/bash");
-    try popcon.put("/bin/zsh");
-    try popcon.put("/bin/zsh");
-    try popcon.put("/bin/zsh");
-    try popcon.put("/bin/zsh");
-    try popcon.put("/bin/nobody");
-    try popcon.put("/bin/very-long-shell-name-ought-to-be-first");
-    try popcon.put("/bin/very-long-shell-name-ought-to-be-first");
+    const bash = "/bin/bash"; // 9 chars
+    const zsh = "/bin/zsh"; // 8 chars
+    const nobody = "/bin/nobody"; // only 1 instance, ought to ignore
+    const long = "/bin/very-long-shell-name-ought-to-be-first";
+    const input = [_][]const u8{
+        zsh, zsh, zsh, zsh, // zsh score 8*4=32
+        bash, bash, bash, nobody, // bash score 3*9=27
+        long, long, // long score 2*42=84
+    };
 
-    const stderr = std.io.getStdErr().writer();
+    for (input) |shell| {
+        try popcon.put(shell);
+    }
 
-    var topshells = try popcon.top(2);
-    defer topshells.deinit();
-    var shellStrings = topshells.keys();
-    try testing.expectEqual(shellStrings.len, 2);
+    var sections = try popcon.toOwnedSections(MaxShells);
+    defer sections.deinit();
+    try testing.expectEqual(sections.index.len, 3); // all but "nobody" qualify
 
-    try stderr.print("\n", .{});
-    try stderr.print("0th type: {s}\n", .{@typeName(@TypeOf(shellStrings[0]))});
-    try stderr.print("1st type: {s}\n", .{@typeName(@TypeOf(shellStrings[1]))});
-    try stderr.print("0th: {s}, len: {d}\n", .{ shellStrings[0], shellStrings[0].len });
-    try stderr.print("0ww: /bin/very-long-shell-name-ought-to-be-first\n", .{});
-    try stderr.print("1st: {s}, len: {d}\n", .{ shellStrings[1], shellStrings[1].len });
-    try stderr.print("1ww: /bin/zsh\n", .{});
-
-    try testing.expectEqual(shellStrings[0], "/bin/very-long-shell-name-ought-to-be-first");
-    try testing.expectEqual(shellStrings[1], "/bin/zsh");
+    try testing.expectEqual(sections.getOffset(long).?, 0);
+    try testing.expectEqual(sections.getOffset(zsh).?, 1);
+    try testing.expectEqual(sections.getOffset(bash).?, 2);
 }