From: Claes Jakobsson Date: Wed, 6 Jan 2010 16:01:48 +0000 (+0000) Subject: Julien Chaffraix fixed so that the fragment part in an URL is not sent to the server... X-Git-Tag: curl-7_20_0~179 X-Git-Url: https://granicus.if.org/sourcecode?a=commitdiff_plain;h=9bd03483ce6983852d41f5f69cb74827c9defc26;p=curl Julien Chaffraix fixed so that the fragment part in an URL is not sent to the server anymore --- diff --git a/CHANGES b/CHANGES index ad3b8c478..506054914 100644 --- a/CHANGES +++ b/CHANGES @@ -6,6 +6,10 @@ Changelog +Claes Jakobsson (6 Jan 2010) +- Julien Chaffraix fixed so that the fragment part in an URL is not sent + to the server anymore. + Kamil Dudka (3 Jan 2010) - Julien Chaffraix eliminated a duplicated initialization in singlesocket(). diff --git a/RELEASE-NOTES b/RELEASE-NOTES index 46b93e8e6..19d8752e4 100644 --- a/RELEASE-NOTES +++ b/RELEASE-NOTES @@ -40,6 +40,7 @@ This release includes the following bugfixes: o some *_proxy environment variables didn't function o libcurl-OpenSSL engine cleanup o header include fix for FreeBSD versions before v8 + o fragment part of URLs are no longer sent to the server This release includes the following known bugs: diff --git a/docs/libcurl/curl_easy_setopt.3 b/docs/libcurl/curl_easy_setopt.3 index 97baa359a..dd1a38ae4 100644 --- a/docs/libcurl/curl_easy_setopt.3 +++ b/docs/libcurl/curl_easy_setopt.3 @@ -449,6 +449,9 @@ on which protocols are supported. The string given to CURLOPT_URL must be url-encoded and follow RFC 2396 (http://curl.haxx.se/rfc/rfc2396.txt). +Please note that starting with version 7.20.0, the fragment part of the URI will +not be send as part of the path, which was the case previously. + \fICURLOPT_URL\fP is the only option that \fBmust\fP be set before \fIcurl_easy_perform(3)\fP is called. diff --git a/lib/url.c b/lib/url.c index 13f80987a..5c7699d88 100644 --- a/lib/url.c +++ b/lib/url.c @@ -3311,8 +3311,9 @@ static CURLcode parseurlandfillconn(struct SessionHandle *data, bool *prot_missing) { char *at; - char *tmp; + char *fragment; char *path = data->state.path; + char *query; int rc; char protobuf[16]; const char *protop; @@ -3438,11 +3439,11 @@ static CURLcode parseurlandfillconn(struct SessionHandle *data, */ at = strchr(conn->host.name, '@'); if(at) - tmp = strchr(at+1, '?'); + query = strchr(at+1, '?'); else - tmp = strchr(conn->host.name, '?'); + query = strchr(conn->host.name, '?'); - if(tmp) { + if(query) { /* We must insert a slash before the '?'-letter in the URL. If the URL had a slash after the '?', that is where the path currently begins and the '?string' is still part of the host name. @@ -3451,7 +3452,7 @@ static CURLcode parseurlandfillconn(struct SessionHandle *data, the path. And have it all prefixed with a slash. */ - size_t hostlen = strlen(tmp); + size_t hostlen = strlen(query); size_t pathlen = strlen(path); /* move the existing path plus the zero byte forward, to make room for @@ -3459,11 +3460,11 @@ static CURLcode parseurlandfillconn(struct SessionHandle *data, memmove(path+hostlen+1, path, pathlen+1); /* now copy the trailing host part in front of the existing path */ - memcpy(path+1, tmp, hostlen); + memcpy(path+1, query, hostlen); path[0]='/'; /* prepend the missing slash */ - *tmp=0; /* now cut off the hostname at the ? */ + *query=0; /* now cut off the hostname at the ? */ } else if(!path[0]) { /* if there's no path set, use a single slash */ @@ -3500,12 +3501,19 @@ static CURLcode parseurlandfillconn(struct SessionHandle *data, } } - if (data->set.scope) + if(data->set.scope) /* Override any scope that was set above. */ conn->scope = data->set.scope; + /* Remove the fragment part of the path. Per RFC 2396, this is always the + last part of the URI. We are looking for the first '#' so that we deal gracefully + with non conformant URI such as http://example.com#foo#bar. */ + fragment = strchr(path, '#'); + if(fragment) + *fragment = 0; + /* - * So if the URL was A://B/C, + * So if the URL was A://B/C#D, * protop is A * conn->host.name is B * data->state.path is /C diff --git a/tests/data/test1109 b/tests/data/test1109 new file mode 100644 index 000000000..fc7fc225f --- /dev/null +++ b/tests/data/test1109 @@ -0,0 +1,46 @@ + +# Test that the fragment is not send as part of the path. + + +HTTP +CURLOPT_URL + + + +# Server-side + + +HTTP/1.1 200 OK +Content-Length: 6 + +hello + + + +# Client-side + + +http + + +HTTP GET + + +http://%HOSTIP:%HTTPPORT/1109#test + + + + +# Verify data after the test has been "shot" + + +^User-Agent:.* + + +GET /1109 HTTP/1.1 +Host: %HOSTIP:%HTTPPORT +Accept: */* + + + + diff --git a/tests/data/test1110 b/tests/data/test1110 new file mode 100644 index 000000000..0597d3ccf --- /dev/null +++ b/tests/data/test1110 @@ -0,0 +1,46 @@ + +# Test that the fragment is not send as part of the path when the path contains a query. + + +HTTP +CURLOPT_URL + + + +# Server-side + + +HTTP/1.1 200 OK +Content-Length: 6 + +hello + + + +# Client-side + + +http + + +HTTP GET + + +http://%HOSTIP:%HTTPPORT/1110?q=foobar#fragment + + + + +# Verify data after the test has been "shot" + + +^User-Agent:.* + + +GET /1110?q=foobar HTTP/1.1 +Host: %HOSTIP:%HTTPPORT +Accept: */* + + + + diff --git a/tests/data/test1111 b/tests/data/test1111 new file mode 100644 index 000000000..0f8e0563c --- /dev/null +++ b/tests/data/test1111 @@ -0,0 +1,46 @@ + +# Test that no fragment is not send as part of the path when the URI contains 2 '#' (does not follow RFC 2396) + + +HTTP +CURLOPT_URL + + + +# Server-side + + +HTTP/1.1 200 OK +Content-Length: 6 + +hello + + + +# Client-side + + +http + + +HTTP GET + + +http://%HOSTIP:%HTTPPORT/1111?q=foobar#fragment#fragment2 + + + + +# Verify data after the test has been "shot" + + +^User-Agent:.* + + +GET /1111?q=foobar HTTP/1.1 +Host: %HOSTIP:%HTTPPORT +Accept: */* + + + +