From 08d58fe13e84a929c72c5d4aa98001279463a79f Mon Sep 17 00:00:00 2001
From: Simon Tatham <anakin@pobox.com>
Date: Sun, 19 Jun 2022 10:36:50 +0100
Subject: [PATCH] Routines for %-encoding and %-decoding.

These make a good storage format for mostly-textual data in
configuration, if it can't afford to reserve any character as a
delimiter. Assuming very few characters need to be escaped, the space
cost is lower than base64, and also you can read it by eye.
---
 misc.h                 |  7 +++++++
 utils/CMakeLists.txt   |  2 ++
 utils/percent_decode.c | 41 +++++++++++++++++++++++++++++++++++++++++
 utils/percent_encode.c | 34 ++++++++++++++++++++++++++++++++++
 4 files changed, 84 insertions(+)
 create mode 100644 utils/percent_decode.c
 create mode 100644 utils/percent_encode.c

diff --git a/misc.h b/misc.h
index afde1502..9cf656fd 100644
--- a/misc.h
+++ b/misc.h
@@ -116,6 +116,13 @@ void base64_encode_fp(FILE *fp, ptrlen data, int cpl);
 strbuf *base64_encode_sb(ptrlen data, int cpl);
 bool base64_valid(ptrlen data);
 
+void percent_encode_bs(BinarySink *bs, ptrlen data, const char *badchars);
+void percent_encode_fp(FILE *fp, ptrlen data, const char *badchars);
+strbuf *percent_encode_sb(ptrlen data, const char *badchars);
+void percent_decode_bs(BinarySink *bs, ptrlen data);
+void percent_decode_fp(FILE *fp, ptrlen data);
+strbuf *percent_decode_sb(ptrlen data);
+
 struct bufchain_granule;
 struct bufchain_tag {
     struct bufchain_granule *head, *tail;
diff --git a/utils/CMakeLists.txt b/utils/CMakeLists.txt
index 957b861c..56cc3c1b 100644
--- a/utils/CMakeLists.txt
+++ b/utils/CMakeLists.txt
@@ -46,6 +46,8 @@ add_sources_from_current_dir(utils
   nullstrcmp.c
   out_of_memory.c
   parse_blocksize.c
+  percent_decode.c
+  percent_encode.c
   prompts.c
   ptrlen.c
   read_file_into.c
diff --git a/utils/percent_decode.c b/utils/percent_decode.c
new file mode 100644
index 00000000..dff2c233
--- /dev/null
+++ b/utils/percent_decode.c
@@ -0,0 +1,41 @@
+/*
+ * Decode %-encoding in URL style.
+ */
+
+#include <ctype.h>
+
+#include "misc.h"
+
+void percent_decode_bs(BinarySink *bs, ptrlen data)
+{
+    for (const char *p = data.ptr, *e = ptrlen_end(data); p < e; p++) {
+        char c = *p;
+        if (c == '%' && e-p >= 3 &&
+            isxdigit((unsigned char)p[1]) &&
+            isxdigit((unsigned char)p[2])) {
+            char hex[3];
+            hex[0] = p[1];
+            hex[1] = p[2];
+            hex[2] = '\0';
+            put_byte(bs, strtoul(hex, NULL, 16));
+            p += 2;
+        } else {
+            put_byte(bs, c);
+        }
+    }
+
+}
+
+void percent_decode_fp(FILE *fp, ptrlen data)
+{
+    stdio_sink ss;
+    stdio_sink_init(&ss, fp);
+    percent_decode_bs(BinarySink_UPCAST(&ss), data);
+}
+
+strbuf *percent_decode_sb(ptrlen data)
+{
+    strbuf *sb = strbuf_new();
+    percent_decode_bs(BinarySink_UPCAST(sb), data);
+    return sb;
+}
diff --git a/utils/percent_encode.c b/utils/percent_encode.c
new file mode 100644
index 00000000..ea04b0ac
--- /dev/null
+++ b/utils/percent_encode.c
@@ -0,0 +1,34 @@
+/*
+ * %-encoding in URL style.
+ *
+ * Defaults to escaping % itself (necessary for decoding to even
+ * work), and any C0 escape character. Further bad characters can be
+ * provided in 'badchars'.
+ */
+
+#include "misc.h"
+
+void percent_encode_bs(BinarySink *bs, ptrlen data, const char *badchars)
+{
+    for (const char *p = data.ptr, *e = ptrlen_end(data); p < e; p++) {
+        char c = *p;
+        if (c == '%' || c < ' ' || (badchars && strchr(badchars, c)))
+            put_fmt(bs, "%%%02X", (unsigned char)c);
+        else
+            put_byte(bs, c);
+    }
+}
+
+void percent_encode_fp(FILE *fp, ptrlen data, const char *badchars)
+{
+    stdio_sink ss;
+    stdio_sink_init(&ss, fp);
+    percent_encode_bs(BinarySink_UPCAST(&ss), data, badchars);
+}
+
+strbuf *percent_encode_sb(ptrlen data, const char *badchars)
+{
+    strbuf *sb = strbuf_new();
+    percent_encode_bs(BinarySink_UPCAST(sb), data, badchars);
+    return sb;
+}