From 7135ea28654d0761967f57f01c782be9a7337c2f Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Remko=20Tron=C3=A7on?= <remko@el-tramo.be>
Date: Sun, 21 Jan 2024 17:45:46 +0100
Subject: [PATCH] thurtle: Add URL compression

---
 src/web/thurtle/thurtle.tsx |  31 ++++++++---
 src/web/thurtle/ulz.ts      | 108 ++++++++++++++++++++++++++++++++++++
 2 files changed, 130 insertions(+), 9 deletions(-)
 create mode 100644 src/web/thurtle/ulz.ts

diff --git a/src/web/thurtle/thurtle.tsx b/src/web/thurtle/thurtle.tsx
index 33c9e96..7bf6a34 100644
--- a/src/web/thurtle/thurtle.tsx
+++ b/src/web/thurtle/thurtle.tsx
@@ -1,6 +1,7 @@
 /////////////////////////////////////////////////////////////////////////
 // Query parameters:
 // `p`: Base64-encoded program
+// `u`: Base64-encoded ULZ-compressed program
 // `pn`: Program name. If `p` is not provided, looks up builtin example
 // `ar`: Auto-run program
 // `sn`: Show navbar (default: 1)
@@ -18,9 +19,12 @@ import {
 import Editor from "./Editor";
 import { saveAs } from "file-saver";
 import draw from "./draw";
+import { ulzEncode, ulzDecode } from "./ulz";
 
 declare let bootstrap: any;
 
+////////////////////////////////////////////////////////////////////////////////
+
 async function b64encode(bs: Uint8Array, forURL?: boolean) {
   const url = await new Promise<string>((resolve) => {
     const reader = new FileReader();
@@ -46,6 +50,8 @@ function b64decode(s: string) {
   );
 }
 
+////////////////////////////////////////////////////////////////////////////////
+
 function parseQS(sqs = window.location.search) {
   const qs: Record<string, string> = {};
   const sqss = sqs
@@ -65,6 +71,8 @@ function parseQS(sqs = window.location.search) {
 
 const qs = parseQS();
 
+////////////////////////////////////////////////////////////////////////////////
+
 function About() {
   return (
     <>
@@ -604,13 +612,17 @@ async function downloadPNG(ev: MouseEvent) {
 async function share(ev: MouseEvent) {
   ev.preventDefault();
   shareModalTitleEl.innerText = programsEl.value;
-  shareModalURLEl.value = `${window.location.protocol}//${
-    window.location.host
-  }${window.location.pathname}?pn=${encodeURIComponent(
-    programsEl.value
-  )}&p=${encodeURIComponent(
-    await b64encode(new TextEncoder().encode(editor.getValue()), true)
-  )}&ar=1`;
+  let url = `${window.location.protocol}//${window.location.host}${
+    window.location.pathname
+  }?pn=${encodeURIComponent(programsEl.value)}&ar=1&`;
+  const program = new TextEncoder().encode(editor.getValue());
+  const compressed = ulzEncode(program);
+  if (compressed.length < program.length) {
+    url += `u=${encodeURIComponent(await b64encode(compressed, true))}`;
+  } else {
+    url += `p=${encodeURIComponent(await b64encode(program, true))}`;
+  }
+  shareModalURLEl.value = url;
   shareModal.show();
 }
 
@@ -676,8 +688,9 @@ async function reset() {
 
 /////////////////////////////////////////////////////////////////////////
 
-if (qs.p) {
-  saveProgram(qs.pn ?? "", new TextDecoder().decode(b64decode(qs.p)), true);
+if (qs.p || qs.u) {
+  const program = qs.p ? b64decode(qs.p) : ulzDecode(b64decode(qs.u));
+  saveProgram(qs.pn ?? "", new TextDecoder().decode(program), true);
   loadPrograms();
   loadProgram(qs.pn);
 } else {
diff --git a/src/web/thurtle/ulz.ts b/src/web/thurtle/ulz.ts
new file mode 100644
index 0000000..86de837
--- /dev/null
+++ b/src/web/thurtle/ulz.ts
@@ -0,0 +1,108 @@
+/*
+ * ULZ ( http://wiki.xxiivv.com/site/ulz_format ) encoder & decoder
+ */
+
+export function ulzDecode(src: Uint8Array) {
+  const dst: Array<number> = [];
+  let sp = 0;
+  while (sp < src.length) {
+    const c = src[sp++];
+    if (c & 0x80) {
+      // CPY
+      let length;
+      if (c & 0x40) {
+        if (sp >= src.length) {
+          throw new Error(`incomplete CPY2`);
+        }
+        length = ((c & 0x3f) << 8) | src[sp++];
+      } else {
+        length = c & 0x3f;
+      }
+      if (sp >= src.length) {
+        throw new Error(`incomplete CPY`);
+      }
+      let cp = dst.length - (src[sp++] + 1);
+      if (cp < 0) {
+        throw new Error(`CPY underflow`);
+      }
+      for (let i = 0; i < length + 4; i++) {
+        dst.push(dst[cp++]);
+      }
+    } else {
+      // LIT
+      if (sp + c >= src.length) {
+        throw new Error(`LIT out of bounds: ${sp} + ${c} >= ${src.length}`);
+      }
+      for (let i = 0; i < c + 1; i++) {
+        dst.push(src[sp++]);
+      }
+    }
+  }
+  return new Uint8Array(dst);
+}
+
+const MIN_MAX_LENGTH = 4;
+
+function findBestMatch(
+  src: Uint8Array,
+  sp: number,
+  dlen: number,
+  slen: number
+) {
+  let bmlen = 0;
+  let bmp = 0;
+  let dp = sp - dlen;
+  for (; dlen; dp++, dlen--) {
+    let i = 0;
+    for (; ; i++) {
+      if (i == slen) {
+        return [dp, i];
+      }
+      if (src[sp + i] != src[dp + (i % dlen)]) {
+        break;
+      }
+    }
+    if (i > bmlen) {
+      bmlen = i;
+      bmp = dp;
+    }
+  }
+  return [bmp, bmlen];
+}
+
+export function ulzEncode(src: Uint8Array) {
+  const dst: Array<number> = [];
+  let sp = 0;
+  let litp = -1;
+  while (sp < src.length) {
+    const dlen = Math.min(sp, 256);
+    const slen = Math.min(src.length - sp, 0x3fff + MIN_MAX_LENGTH);
+    const [bmp, bmlen] = findBestMatch(src, sp, dlen, slen);
+    if (bmlen >= MIN_MAX_LENGTH) {
+      // CPY
+      const bmctl = bmlen - MIN_MAX_LENGTH;
+      if (bmctl > 0x3f) {
+        // 	CPY2
+        dst.push((bmctl >> 8) | 0xc0);
+        dst.push(bmctl & 0xff);
+      } else {
+        dst.push(bmctl | 0x80);
+      }
+      dst.push(sp - bmp - 1);
+      sp += bmlen;
+      litp = -1;
+    } else {
+      // LIT
+      if (litp >= 0) {
+        if ((dst[litp] += 1) == 127) {
+          litp = -1;
+        }
+      } else {
+        dst.push(0);
+        litp = dst.length - 1;
+      }
+      dst.push(src[sp++]);
+    }
+  }
+  return new Uint8Array(dst);
+}