From 44f98163d91b33c48c9d1f57b141a0c0bd6fd22a Mon Sep 17 00:00:00 2001
From: Emery Hemingway <ehmry@posteo.net>
Date: Sun, 24 Dec 2023 21:07:05 +0200
Subject: [PATCH] Unify string and symbol escaping

---
 src/preserves/private/texts.nim | 107 +++++++++++++++-----------------
 1 file changed, 50 insertions(+), 57 deletions(-)

diff --git a/src/preserves/private/texts.nim b/src/preserves/private/texts.nim
index 247d659..de62564 100644
--- a/src/preserves/private/texts.nim
+++ b/src/preserves/private/texts.nim
@@ -1,42 +1,53 @@
 # SPDX-FileCopyrightText: ☭ Emery Hemingway
 # SPDX-License-Identifier: Unlicense
 
-import std/[base64, endians, json, math, options, sets, sequtils, streams, strutils, tables, typetraits]
+import std/[base64, bitops, endians, math, sequtils, streams, strutils, unicode]
+import bigints
 import ./values
 
-proc `$`*(s: Symbol): string =
-  let sym = string s
-  if sym.len > 0 and sym[0] in {'A'..'z'} and not sym.anyIt(char(it) in { '\x00'..'\x19', '"', '\\', '|' }):
-    result = sym
-  else:
-    result = newStringOfCap(sym.len shl 1)
-    result.add('|')
-    for c in sym:
-      case c
-      of '\\':
-        result.add("\\\\")
-      of '/':
-        result.add("\\/")
-      of '\x08':
-        result.add("\\b")
-      of '\x0c':
-        result.add("\\f")
-      of '\x0a':
-        result.add("\\n")
-      of '\x0d':
-        result.add("\\r")
-      of '\x09':
-        result.add("\\t")
-      of '|':
-        result.add("\\|")
-      else:
-        result.add(c)
-    result.add('|')
-
 const hexAlphabet = "0123456789abcdef"
 
 type TextMode* = enum textPreserves, textJson
 
+template writeEscaped(stream: Stream; text: string; delim: char) =
+  const escaped = { '"', '\\', '\b', '\f', '\n', '\r', '\t' }
+  var
+    i: int
+    r: Rune
+    c: char
+  while i < text.len:
+    c = text[i]
+    if (c.ord and 0x80) == 0x00:
+      case c
+      of delim:
+        write(stream, '\\')
+        write(stream, delim)
+      of '\\': write(stream, "\\\\")
+      of '\b': write(stream, "\\b")
+      of '\f': write(stream, "\\f")
+      of '\n': write(stream, "\\n")
+      of '\r': write(stream, "\\r")
+      of '\t': write(stream, "\\t")
+      of { '\x00'..'\x1f', '\x7f' } - escaped:
+        # do not use \x__ notation because
+        # it is a subset of \u____.
+        write(stream, "\\u00")
+        write(stream, c.uint8.toHex(2))
+      else: write(stream, c)
+      inc i
+    else:
+      fastRuneAt(text, i, r)
+      write(stream, "\\u")
+      write(stream, r.uint16.toHex(4))
+
+proc writeSymbol(stream: Stream; sym: string) =
+  if sym.len > 0 and sym[0] in {'A'..'z'} and not sym.anyIt(char(it) in { '\x00'..'\x19', '"', '\\', '|' }):
+    write(stream, sym)
+  else:
+    write(stream, '|')
+    writeEscaped(stream, sym, '|')
+    write(stream, '|')
+
 proc writeText*[E](stream: Stream; pr: Preserve[E]; mode = textPreserves) =
   ## Encode Preserves to a `Stream` as text.
   if pr.embedded: write(stream, "#!")
@@ -75,7 +86,9 @@ proc writeText*[E](stream: Stream; pr: Preserve[E]; mode = textPreserves) =
   of pkBigInt:
     write(stream, $pr.bigint)
   of pkString:
-    write(stream, escapeJson(pr.string))
+    write(stream, '"')
+    writeEscaped(stream, pr.string, '"')
+    write(stream, '"')
   of pkByteString:
     if pr.bytes.allIt(char(it) in {' '..'!', '#'..'~'}):
       write(stream, "#\"")
@@ -93,32 +106,7 @@ proc writeText*[E](stream: Stream; pr: Preserve[E]; mode = textPreserves) =
           write(stream, hexAlphabet[b.int and 0xf])
         write(stream, '"')
   of pkSymbol:
-    let sym = pr.symbol.string
-    if sym.len > 0 and sym[0] in {'A'..'z'} and not sym.anyIt(char(it) in { '\x00'..'\x19', '"', '\\', '|' }):
-      write(stream, sym)
-    else:
-      write(stream, '|')
-      for c in sym:
-        case c
-        of '\\':
-          write(stream, "\\\\")
-        of '/':
-          write(stream, "\\/")
-        of '\x08':
-          write(stream, "\\b")
-        of '\x0c':
-          write(stream, "\\f")
-        of '\x0a':
-          write(stream, "\\n")
-        of '\x0d':
-          write(stream, "\\r")
-        of '\x09':
-          write(stream, "\\t")
-        of '|':
-          write(stream, "\\|")
-        else:
-          write(stream, c)
-      write(stream, '|')
+    writeSymbol(stream, pr.symbol.string)
   of pkRecord:
     assert(pr.record.len > 0)
     write(stream, '<')
@@ -176,6 +164,11 @@ proc writeText*[E](stream: Stream; pr: Preserve[E]; mode = textPreserves) =
     else:
       write(stream, "…")
 
+proc `$`*(sym: Symbol): string =
+  var stream = newStringStream()
+  writeSymbol(stream, sym.string)
+  result = move stream.data
+
 proc `$`*[E](pr: Preserve[E]): string =
   ## Generate the textual representation of ``pr``.
   var stream = newStringStream()