Glen Sanft <[EMAIL PROTECTED]> writes:
> This is not a problem which has just arisen in the recentest
> release, but "#" in a path/file name, when properly escaped in a
> URI, becomes itself in the filesystem, thus breaking the subsequent
> local fetch.
It sounds like Wget should encode "#" as "%23" when generating the
link.
Does this patch work for you?
2004-03-04 Hrvoje Niksic <[EMAIL PROTECTED]>
* convert.c (local_quote_string): Also quote "#" as "%23".
Index: src/convert.c
===================================================================
RCS file: /pack/anoncvs/wget/src/convert.c,v
retrieving revision 1.7
diff -u -r1.7 convert.c
--- src/convert.c 2003/11/08 04:55:44 1.7
+++ src/convert.c 2004/03/04 00:24:28
@@ -576,49 +576,49 @@
"index.html%3Ffoo=bar" would break local browsing, as the latter
isn't even recognized as an HTML file! However, converting
"index.html?foo=bar.html" to "index.html%3Ffoo=bar.html" should be
- safe for both local and HTTP-served browsing. */
+ safe for both local and HTTP-served browsing.
+ We always quote "#" as "%23" because # in a URL is a fragment
+ delimiter. */
+
static char *
local_quote_string (const char *file)
{
- const char *file_sans_qmark;
- int qm;
+ const char *from;
+ char *newname, *to;
+ int newlen;
- if (!opt.html_extension)
- return html_quote_string (file);
+ int qmark = count_char (file, '?');
+ int pound = count_char (file, '#');
- qm = count_char (file, '?');
+ if (pound == 0 && (qmark == 0 || !opt.html_extension))
+ return html_quote_string (file);
- if (qm)
- {
- const char *from = file;
- char *to, *newname;
-
- /* qm * 2 because we replace each question mark with "%3F",
- i.e. replace one char with three, hence two more. */
- int fsqlen = strlen (file) + qm * 2;
-
- to = newname = (char *)alloca (fsqlen + 1);
- for (; *from; from++)
- {
- if (*from != '?')
- *to++ = *from;
- else
- {
- *to++ = '%';
- *to++ = '3';
- *to++ = 'F';
- }
- }
- assert (to - newname == fsqlen);
- *to = '\0';
-
- file_sans_qmark = newname;
- }
- else
- file_sans_qmark = file;
+ /* Add (qmark + pound) * 2 because we replace each such character
+ with "%3F", i.e. replace one char with three, hence two more. */
+ newlen = strlen (file) + (qmark + pound) * 2;
+
+ to = newname = (char *)alloca (newlen + 1);
+ for (from = file; *from; from++)
+ switch (*from)
+ {
+ case '?':
+ *to++ = '%';
+ *to++ = '3';
+ *to++ = 'F';
+ break;
+ case '#':
+ *to++ = '%';
+ *to++ = '2';
+ *to++ = '3';
+ break;
+ default:
+ *to++ = *from;
+ }
+ assert (to - newname == newlen);
+ *to = '\0';
- return html_quote_string (file_sans_qmark);
+ return html_quote_string (newname);
}
/* Book-keeping code for dl_file_url_map, dl_url_file_map,
