> However, I'm in doubt about the part that mangles path#fragment?query to > path?query. As specified by the URI RFC the fragment is always the last part > and anything after it can be disregarded. This would also make the patch a > lot simpler.
You are right. I misunderstood the fragment's position. I have attached the updated patch (which also add a mention about this change in the curl_easy_setopt manpage) and test case to match the URI RFC. Thanks for the review! Regards, Julien
diff --git a/docs/libcurl/curl_easy_setopt.3 b/docs/libcurl/curl_easy_setopt.3 index 9229d09..a852028 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 aeb0f91..8a4525b 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 +3501,18 @@ 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. */ + fragment = strrchr(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 0000000..6068380 --- /dev/null +++ b/tests/data/test1109 @@ -0,0 +1,45 @@ +<testcase> +<info> +<keywords> +HTTP +CURLOPT_URL +</keywords> +</info> + +# Server-side +<reply name="1"> +<data> +HTTP/1.1 200 OK +Content-Length: 6 + +hello +</data> +</reply> + +# Client-side +<client> +<server> +http +</server> + <name> +HTTP GET + </name> + <command> +http://%HOSTIP:%HTTPPORT/1109#test +</command> +</client> + + +# Verify data after the test has been "shot" +<verify> +<strip> +^User-Agent:.* +</strip> +<protocol> +GET /1109 HTTP/1.1 +Host: %HOSTIP:%HTTPPORT +Accept: */* + +</protocol> +</verify> +</testcase> diff --git a/tests/data/test1110 b/tests/data/test1110 new file mode 100644 index 0000000..70a9486 --- /dev/null +++ b/tests/data/test1110 @@ -0,0 +1,45 @@ +<testcase> +<info> +<keywords> +HTTP +CURLOPT_URL +</keywords> +</info> + +# Server-side +<reply name="1"> +<data> +HTTP/1.1 200 OK +Content-Length: 6 + +hello +</data> +</reply> + +# Client-side +<client> +<server> +http +</server> + <name> +HTTP GET + </name> + <command> +http://%HOSTIP:%HTTPPORT/1110?q=foobar#fragment +</command> +</client> + + +# Verify data after the test has been "shot" +<verify> +<strip> +^User-Agent:.* +</strip> +<protocol> +GET /1110?q=foobar HTTP/1.1 +Host: %HOSTIP:%HTTPPORT +Accept: */* + +</protocol> +</verify> +</testcase>
test1110
Description: Binary data
------------------------------------------------------------------- List admin: http://cool.haxx.se/list/listinfo/curl-library Etiquette: http://curl.haxx.se/mail/etiquette.html
