From 7e4708e456b17131f28ae7f1ff7b5c7f86e2f803 Mon Sep 17 00:00:00 2001
Message-Id: <7e4708e456b17131f28ae7f1ff7b5c7f86e2f803.1713895947.git.mdw@distorted.org.uk>
From: Mark Wooding <mdw@chiark.greenend.org.uk>
Date: Tue, 14 Mar 2006 16:41:41 +0000
Subject: [PATCH] url: Allow various `safe' characters unquoted in URL strings.
Organization: Straylight/Edgeware

From: Mark Wooding <mdw@distorted.org.uk>

Allow `.', `-', `_' and `/', because they're commonly used in filenames,
and it's nice for them to be displayed readably.  The `~' character
isn't actually safe but we allow it anyway.
---
 man/url.3 | 21 +++++++++++++++++++++
 url.c     | 36 +++++++++++++++++++++++++++---------
 url.h     |  2 ++
 3 files changed, 50 insertions(+), 9 deletions(-)

diff --git a/man/url.3 b/man/url.3
index 5a96403..919ce93 100644
--- a/man/url.3
+++ b/man/url.3
@@ -57,6 +57,27 @@ string (see
 .BR dstr (3)
 for details).
 .PP
+You can set flags in the encoding context's
+.B f
+member:
+.TP
+.B URLF_STRICT
+Be strict about escaping non-alphanumeric characters.  Without this,
+potentially unsafe characters such as 
+.RB ` / '
+and
+.RB ` ~ '
+will be left unescaped, which makes encoded filenames (for example) more
+readable.
+.TP
+.B URLF_LAX
+Be very lax about non-alphanumeric characters.  Everything except
+obviously-unsafe characters like
+.RB ` & '
+and 
+.RB ` = '
+are left unescaped.
+.PP
 Decoding a sequence of name/value pairs is performed using the
 .B url_dec
 function.  It requires as input a
diff --git a/url.c b/url.c
index d859bfa..f7e11b3 100644
--- a/url.c
+++ b/url.c
@@ -52,7 +52,8 @@ void url_initenc(url_ectx *ctx) { ctx->f = 0; }
 
 /* --- @encode@ --- *
  *
- * Arguments:	@dstr *d@ = pointer to output string
+ * Arguments:	@url_ectx *ctx@ = encoding context
+ *		@dstr *d@ = pointer to output string
  *		@const char *p@ = pointer to thing to encode
  *
  * Returns:	---
@@ -60,7 +61,7 @@ void url_initenc(url_ectx *ctx) { ctx->f = 0; }
  * Use:		Encodes the input string into the output string.
  */
 
-static void encode(dstr *d, const char *p)
+static void encode(url_ectx *ctx, dstr *d, const char *p)
 {
   while (*p) {
     switch (*p) {
@@ -68,11 +69,28 @@ static void encode(dstr *d, const char *p)
 	DPUTC(d, '+');
         break;
       default:
-        if (isalnum((unsigned char)*p))
-	  DPUTC(d, *p);
-        else
-	  dstr_putf(d, "%%%02x", *p);
-        break;
+        if ((ctx->f & URLF_LAX) || isalnum((unsigned char)*p))
+	  goto safe;
+	else
+	  goto unsafe;
+      case '/':
+      case '~':
+	if (ctx->f & URLF_STRICT)
+	  goto unsafe;
+      case '-':
+      case '.':
+      case '_':
+      safe:
+	DPUTC(d, *p);
+	break;
+      unsafe:
+      case '+':
+      case '%':
+      case '=':
+      case '&':
+      case ';':
+	dstr_putf(d, "%%%02x", *p);
+	break;
     }
     p++;
   }
@@ -95,9 +113,9 @@ void url_enc(url_ectx *ctx, dstr *d, const char *name, const char *value)
 {
   if (ctx->f & URLF_SEP)
     DPUTC(d, '&');
-  encode(d, name);
+  encode(ctx, d, name);
   DPUTC(d, '=');
-  encode(d, value);
+  encode(ctx, d, value);
   DPUTZ(d);
   ctx->f |= URLF_SEP;
 }
diff --git a/url.h b/url.h
index c351343..be38bdc 100644
--- a/url.h
+++ b/url.h
@@ -47,6 +47,8 @@ typedef struct url_ectx {
 } url_ectx;
 
 #define URLF_SEP 1u
+#define URLF_STRICT 2u
+#define URLF_LAX 4u
 
 typedef struct url_dctx {
   const char *p;
-- 
[mdw]