add c3

.gitignore

@@ -17,6 +17,7 @@ crystal/.shards/
 crystal/crystal
 ols.json
 odin/related
+c3/related
 .dart_tool/
 v/related
 

Dockerfile

@@ -45,6 +45,12 @@ RUN wget 'https://github.com/odin-lang/Odin/releases/download/dev-2023-12/odin-u
 
 RUN unzip /home/builduser/odin.zip -d /home/builduser/odin
 
+# install c3
+RUN wget 'https://github.com/c3lang/c3c/releases/download/v0.6.6/c3-linux.tar.gz' -O /home/builduser/c3.tar.gz
+
+RUN tar zxvf /home/builduser/c3.tar.gz -C /home/builduser
+ENV PATH="$PATH:/home/builduser/c3"
+
 # install vlang
 RUN wget 'https://github.com/vlang/v/releases/download/weekly.2023.40.1/v_linux.zip' -O /home/builduser/v.zip
 

c3/json.c3

@@ -0,0 +1,120 @@
+// Using std::collections::object to serialize json would require
+// converting into objects; instead, I just directly wrote the json.
+// It's ugly copy-paste one-time-use code.
+module related::json;
+
+import std::collections;
+import std::encoding::json;
+import std::io;
+
+// Nothing is freed, and assumes that input outlives output
+fn Post[]! parse_posts(String input, Allocator alloc = allocator::heap()) {
+	List(<Post>) posts;
+	Object* posts_arr = json::parse_string(input, alloc)!;
+	foreach (i, post_obj : posts_arr.array) {
+		List(<String>) tags;
+		foreach (tag_obj : post_obj.get("tags")!.array) {
+			tags.push(tag_obj.s);
+		}
+		posts.push(Post {
+			._id = post_obj.get_string("_id")!,
+			.title = post_obj.get_string("title")!,
+			.tags = tags.array_view(),
+		});
+	}
+	return posts.array_view();
+}
+
+fn String Post.to_json(post) {
+	DString out;
+	out.append('{');
+
+	// _id
+	out.append("\"_id\":\"");
+	out.append(post._id);
+	out.append("\",");
+
+	// title
+	out.append("\"title\":\"");
+	out.append(post.title);
+	out.append("\",");
+
+	// tags
+	out.append("\"tags\":[");
+	if (post.tags.len > 0) {
+		out.append("\"");
+		out.append(post.tags[0]);
+		out.append("\"");
+	}
+	if (post.tags.len > 1) {
+		foreach(tag : post.tags[1..]) {
+			out.append(",");
+			out.append("\"");
+			out.append(tag);
+			out.append("\"");
+		}
+	}
+	out.append("]");
+	out.append('}');
+	return out.str_view();
+}
+
+fn String TopPosts.to_json(topposts) {
+	DString out;
+	out.append('{');
+
+	// _id
+	out.append("\"_id\":\"");
+	out.append(*topposts._id);
+	out.append("\",");
+
+	// tags
+	out.append("\"tags\":[");
+	if (topposts.tags.len > 0) {
+		out.append("\"");
+		out.append((*topposts.tags)[0]);
+		out.append("\"");
+	}
+	if (topposts.tags.len > 1) {
+		foreach(tag : (*topposts.tags)[1..]) {
+			out.append(",");
+			out.append("\"");
+			out.append(tag);
+			out.append("\"");
+		}
+	}
+	out.append("],");
+
+	// related
+	out.append("\"related\":[");
+	if (topposts.related.len > 0) {
+		out.append(topposts.related[0].to_json());
+	}
+	if (topposts.related.len > 1) {
+		foreach(related_post : topposts.related[1..]) {
+			out.append(',');
+			out.append(related_post.to_json());
+		}
+	}
+	out.append("]");
+
+	// finish
+	out.append('}');
+	return out.str_view();
+}
+
+fn String TopPosts[].to_json(topposts_list) {
+	DString out;
+	out.append("[\n");
+	if (topposts_list.len > 0) {
+		out.append(topposts_list[0].to_json());
+	}
+	if (topposts_list.len > 1) {
+		foreach(topposts : topposts_list[1..]) {
+			out.append(",\n");
+			out.append(topposts.to_json());
+		}
+	}
+	out.append("\n]");
+	return out.str_view();
+}

c3/related.c3

@@ -0,0 +1,122 @@
+module related;
+
+import std::collections;
+import std::core::mem;
+import std::io;
+import std::time::clock;
+
+struct Post {
+	String _id;
+	String title;
+	Tag[] tags;
+}
+def Posts = Post[];
+
+def PostIdx = uint;
+def PostIdxList = List(<PostIdx>);
+def Tag = String;
+def Tag2PostIdxList = HashMap(<Tag, PostIdxList>);
+
+struct TopPosts {
+	String* _id;
+	Tag[]* tags;
+	Post*[] related;
+}
+struct Score {
+	char s;
+	uint pos;
+}
+
+fn char is_top(char m, char[] score) @inline {
+	char x;
+	foreach (s : score) {
+		// TODO is there a compiler builtin for bool_to_int?
+		x |= (s > m) ? 1 : 0;
+	}
+	return x;
+}
+
+fn void get_top(uint b, char[] score, char* min, Score[] t5) @inline {
+	uint i = b;
+	int score_idx;
+	while(score_idx < score.len) {
+		char s = score[score_idx];
+		if (s > *min) {
+			ichar u = 3;
+			while (u >= 0 && s > t5[u].s) {
+				t5[u + 1] = t5[u];
+				u -= 1;
+			}
+			t5[u + 1] = Score{ .s = s, .pos = i };
+			*min = t5[4].s;
+		}
+		i += 1;
+		score_idx += 1;
+	}
+}
+
+fn void top5(Post*[] related, char[] score, Post[] ps) @inline {
+	Score s = { .s = 0, .pos = 0 };
+	Score[5] t5 = { s, s, s, s, s };
+	char min_tags;
+
+	uint b;
+	uint cache_line = 64;
+	while (b < score.len) {
+		uint e = min(b + cache_line, (uint)score.len);
+		char[] chunk = score[b..e - 1];
+		if (is_top(min_tags, chunk) > 0) {
+			get_top(b, chunk, &min_tags, t5[0..]);
+		}
+		b += cache_line;
+	}
+	foreach (i, t : t5) {
+		related[i] = &ps[t.pos];
+	}
+}
+
+fn void main() {
+	String input = (String)file::load_new("../posts.json")!!;
+	Post[] posts = json::parse_posts(input)!!;
+
+	Clock start = clock::now();
+
+	Tag2PostIdxList tag2postidxs;
+	foreach (post_idx, post : posts) {
+		foreach (tag : post.tags) {
+			// TODO hashmap doesn't have the API
+			// that would remove extra hash comparisons
+			if (!tag2postidxs.has_key(tag)) {
+				tag2postidxs.set(tag, {});
+			}
+			tag2postidxs.get_ref(tag)!!.push((uint)post_idx);
+		}
+	}
+
+	TopPosts[] op = mem::new_array(TopPosts, posts.len);
+	Post*[] rl = mem::new_array(Post*, posts.len * 5);
+
+	char[] tagged_post_count = mem::new_array(char, posts.len);
+
+	for (int post_idx = 0; post_idx < posts.len; post_idx += 1) {
+		// reset tagged_post_count
+		mem::zero_volatile(tagged_post_count);
+
+		foreach (tag : posts[post_idx].tags) {
+			foreach (tagged_post_idx : tag2postidxs.get(tag)!!) {
+				tagged_post_count[tagged_post_idx] += 1;
+			}
+		}
+
+		tagged_post_count[post_idx] = 0; // Don't count self
+
+		Post*[] related = rl[post_idx * 5 : 5];
+		top5(related, tagged_post_count, posts);
+		op[post_idx] = { ._id = &posts[post_idx]._id, .tags = &posts[post_idx].tags, .related = related };
+	}
+	NanoDuration end = start.to_now();
+	io::printf("Processing time (w/o IO): %sms\n", end.to_ms());
+
+	File op_file = file::open("../related_posts_c3.json", "wb")!!;
+	io::fprint(&op_file, op.to_json())!!;
+}

run.sh

@@ -253,6 +253,14 @@ run_odin() {
         check_output "related_posts_odin.json"
 }
 
+run_c3() {
+    echo "Running c3" &&
+        cd ./c3 &&
+        c3c compile related.c3 json.c3 -O5 &&
+        run_command "c3" $runs ./related &&
+        check_output "related_posts_c3.json"
+}
+
 run_vlang() {
     echo "Running Vlang" &&
         cd ./v &&
@@ -763,6 +771,10 @@ elif [ "$first_arg" = "odin" ]; then
 
     run_odin
 
+elif [ "$first_arg" = "c3" ]; then
+
+    run_c3
+
 elif [ "$first_arg" = "jq" ]; then
 
     run_jq
@@ -972,6 +984,7 @@ elif [ "$first_arg" = "all" ]; then
         run_julia_highly_optimized || echo -e "\n" &&
         run_julia_con || echo -e "\n" &&
         run_odin || echo -e "\n" &&
+        run_c3 || echo -e "\n" &&
         run_vlang || echo -e "\n" &&
         run_dart || echo -e "\n" &&
         run_dart_aot || echo -e "\n" &&
@@ -1024,6 +1037,8 @@ elif [ "$first_arg" = "clean" ]; then
         cd .. &&
         cd rust_con && cargo clean &&
         cd .. &&
+        cd c3 && rm -f related &&
+        cd .. &&
         cd d && rm -f related &&
         cd .. &&
         cd d_con && rm -f related_concurrent &&
@@ -1050,6 +1065,6 @@ elif [ "$first_arg" = "clean" ]; then
 
 else
 
-    echo "Valid args: go | go_con | rust | rust_con | d | d_con | r | py | numpy | erlang | cl | numba | numba_con | cr | zig | zig_con | odin | jq | julia | julia_highly_optimized | julia_con | v | dart | swift | swift_con | node | bun | deno | java | java_graal | java_graal_con | nim | luajit | lua | fsharp | fsharp_aot | fsharp_con | csharp | csharp_aot | dascript | all | clean. Unknown argument: $first_arg"
+    echo "Valid args: go | go_con | rust | rust_con | d | d_con | r | py | numpy | erlang | cl | numba | numba_con | cr | zig | zig_con | odin | c3 | jq | julia | julia_highly_optimized | julia_con | v | dart | swift | swift_con | node | bun | deno | java | java_graal | java_graal_con | nim | luajit | lua | fsharp | fsharp_aot | fsharp_con | csharp | csharp_aot | dascript | all | clean. Unknown argument: $first_arg"
 
 fi