>From d8f4127d54dd09a0834a50c414be02594d828d03 Mon Sep 17 00:00:00 2001
From: Gianluca Cannata <[email protected]>
Date: Wed, 11 Nov 2026 11:42:15 +0100
Subject: [PATCH] open_connection: use libcurl for HTTP transfer
---
Makefile.am | 2 +-
configure.in | 3 +-
http.c | 174 +++++++++++++++++---------------------------
httpfs.c | 7 ++
httpfs.h | 4 +-
netfs.c | 11 +--
parsehtml.c | 11 +--
tests/Makefile.am | 2 +-
tests/test_httpfs.c | 43 +++--------
9 files changed, 98 insertions(+), 159 deletions(-)
diff --git a/Makefile.am b/Makefile.am
index 27d2588..473f60b 100644
--- a/Makefile.am
+++ b/Makefile.am
@@ -2,7 +2,7 @@
bin_PROGRAMS=httpfs
-HTTPFS_SHARED_LIBS = -lxml2 -lpthread -lnetfs -lfshelp -lports
-liohelp -lshouldbeinlibc -lhurdbugaddr
+HTTPFS_SHARED_LIBS = -lcurl -lxml2 -lpthread -lnetfs -lfshelp -lports
-liohelp -lshouldbeinlibc -lhurdbugaddr
httpfs_LDADD = $(HTTPFS_SHARED_LIBS) $(LIBS)
httpfs_SOURCES=httpfs.c \
diff --git a/configure.in b/configure.in
index 0643c19..5098e68 100644
--- a/configure.in
+++ b/configure.in
@@ -4,9 +4,10 @@ AM_INIT_AUTOMAKE
AC_PROG_CC
+PKG_CHECK_MODULES([CURL], [libcurl >= 7.0.0])
PKG_CHECK_MODULES(HTTPFS, libxml-2.0)
-HTTPFS_SHARED_LIBS="-lxml2 -lpthread -lnetfs -lfshelp -lports
-liohelp -lshouldbeinlibc -lhurdbugaddr"
+HTTPFS_SHARED_LIBS="-lcurl -lxml2 -lpthread -lnetfs -lfshelp -lports
-liohelp -lshouldbeinlibc -lhurdbugaddr"
AC_SUBST([HTTPFS_SHARED_LIBS])
# Optional test suite integration (auto-enabled if dependencies are found)
diff --git a/http.c b/http.c
index 50184c0..3b50fb5 100644
--- a/http.c
+++ b/http.c
@@ -34,140 +34,96 @@
#include <hurd/hurd_types.h>
#include <hurd/netfs.h>
-/* do a DNS lookup for NAME and store result in *ENT */
-error_t lookup_host (char *url, struct hostent **ent)
+struct curl_read_request {
+ char *buffer;
+ loff_t offset;
+};
+
+static size_t curl_write_response_to_buffer(char *data, size_t size,
size_t nmemb, void *userdata)
{
- if ( (*ent = gethostbyname(url)) == NULL )
- {
- fprintf(stderr,"wrong host name\n");
- return EINVAL;
+ struct curl_read_request *request = (struct curl_read_request *) userdata;
+ if (!request) {
+ return 0;
}
- return 0;
+
+ size_t total_size = size * nmemb - request->offset;
+ if (total_size == 0) {
+ return 0;
+ }
+
+ memcpy(request->buffer, data + request->offset, total_size);
+
+ return total_size;
}
/* open a connection with the remote web server */
-error_t open_connection(struct netnode *node, int *fd,off_t *head_len)
+error_t open_connection(struct netnode *node, char *buffer, loff_t
offset, off_t *head_len)
{
/* HTTP GET command returns head and body so we have to prune the
* head. *HEAD_LEN indicates the header length, for pruning upto that */
-
- error_t err;
- struct hostent *hptr;
- struct sockaddr_in dest;
- ssize_t written;
- size_t towrite;
- char buffer[4096];
- ssize_t bytes_read;
- char *token,*mesg;
- int code;
- char delimiters0[] = " ";
- char delimiters1[] = "\n";
+ error_t err = 0;
+ long http_status_code = 200;
+ double http_content_length = 0.0f;
+ CURLcode curl_err = CURLE_OK;
+ struct curl_read_request request;
- bzero(&dest,sizeof(dest));
- dest.sin_family = AF_INET;
- dest.sin_port = htons (port);
+ request.buffer = buffer;
+ request.offset = offset;
- if ( !strcmp(ip_addr,"0.0.0.0") )
- {
- /* connection is not through a proxy server
- * find IP addr. of remote server */
- err = lookup_host (node->url, &hptr);
- if (err)
- {
- fprintf(stderr,"Could not find IP addr %s\n",node->url);
- return err;
- }
- dest.sin_addr = *(struct in_addr *)hptr->h_addr;
+ CURL *curl = curl_easy_init();
+ if (!curl) {
+ return (err = ENOMEM);
}
- else
- {
- /* connection is through the proxy server
- * need not find IP of remote server
- * find IP of the proxy server */
- if ( inet_aton(ip_addr,&dest.sin_addr) == 0 )
- {
- fprintf(stderr,"Invalid IP for proxy\n");
- return -1;
- }
+
+ if (!strcmp(ip_addr, "0.0.0.0")) {
+ /* Connection is not through a proxy server. Do nothing. */
+ } else {
+ /* Connection is through a proxy server. */
+ curl_easy_setopt(curl, CURLOPT_PROXY, ip_addr);
}
+ curl_easy_setopt(curl, CURLOPT_URL, node->url);
+ curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, curl_write_response_to_buffer);
+ curl_easy_setopt(curl, CURLOPT_WRITEDATA, buffer);
+ curl_easy_setopt(curl, CURLOPT_HTTPGET, 1L);
+
if (debug_flag)
fprintf (stderr, "trying to open %s:%d/%s\n", node->url,
port, node->conn_req);
- *fd = socket (AF_INET, SOCK_STREAM, 0);
- if (*fd == -1)
- {
- fprintf(stderr,"Socket creation error\n");
- return errno;
+ curl_err = curl_easy_perform(curl);
+ if (curl_err != CURLE_OK) {
+ fprintf (stderr, "curl_easy_perform(): %s\n", curl_easy_strerror(curl_err));
+ err = EAGAIN;
+ goto cleanup;
}
- err = connect (*fd, (struct sockaddr *)&dest, sizeof (dest));
- if (err == -1)
- {
- fprintf(stderr,"Cannot connect to remote host\n");
- return errno;
+ curl_err = curl_easy_getinfo(curl, CURLINFO_RESPONSE_CODE, &http_status_code);
+ if (curl_err != CURLE_OK) {
+ fprintf(stderr, "Cannot get HTTP status code: %s\n",
curl_easy_strerror(curl_err));
+ err = EAGAIN;
+ goto cleanup;
+ } else if (http_status_code != 200) {
+ fprintf(stderr, "HTTP status code is: %ld\n", http_status_code);
+ err = EIO;
+ goto cleanup;
}
- /* Send a HEAD request find header length */
- sprintf(buffer,"HEAD %s HTTP/1.0\r\nHost:
%s\r\n\r\n",node->conn_req,node->url);
- towrite = strlen (buffer);
- written = TEMP_FAILURE_RETRY (write (*fd, buffer, towrite));
- if ( written == -1 || written < towrite )
- {
- fprintf(stderr,"Could not send an HTTP request to host\n");
- return errno;
- }
-
- bytes_read = read(*fd,buffer,sizeof(buffer));
- if ( bytes_read < 0 )
- {
- fprintf(stderr,"Error with HEAD read\n");
- return errno;
+ curl_err = curl_easy_getinfo(curl,
CURLINFO_CONTENT_LENGTH_DOWNLOAD_T, &http_content_length);
+ if (curl_err != CURLE_OK) {
+ fprintf(stderr, "Could not determine HTTP content length: %s\n",
curl_easy_strerror(curl_err));
+ err = EIO;
+ goto cleanup;
}
- *head_len = bytes_read;
- token = strtok(buffer,delimiters0);
- token = strtok(NULL,delimiters0);
- sscanf(token,"%d",&code);
- token = strtok(NULL,delimiters1);
- mesg = strdup(token);
- if ( code != 200 )
- {
- /* page does not exist */
- fprintf(stderr,"Error Page not Accesible\n");
- fprintf(stderr,"%d %s\n",code,mesg);
- return EBADF;
- }
-
- close(*fd);
-
- /* Send the GET request for the url */
- *fd = socket (AF_INET, SOCK_STREAM, 0);
- if (*fd == -1)
- {
- fprintf(stderr,"Socket creation error\n");
- return errno;
- }
-
- err = connect (*fd, (struct sockaddr *)&dest, sizeof (dest));
- if (err == -1)
- {
- fprintf(stderr,"Cannot connect to remote host\n");
- return errno;
- }
+ *head_len = (off_t) http_content_length;
- towrite = strlen (node->comm_buf);
+ err = 0;
+ goto cleanup;
- /* guard against EINTR failures */
- written = TEMP_FAILURE_RETRY (write (*fd, node->comm_buf, towrite));
- written += TEMP_FAILURE_RETRY (write (*fd, "\n\n", 2));
- if (written == -1 || written < (towrite+2))
- {
- fprintf(stderr,"Could not send GET request to remote host\n");
- return errno;
- }
- return 0;
+ cleanup:
+ curl_easy_cleanup(curl);
+ return err;
}
/* fetch a directory node from the web server
diff --git a/httpfs.c b/httpfs.c
index 2da7f0d..67a2b6d 100644
--- a/httpfs.c
+++ b/httpfs.c
@@ -42,6 +42,12 @@ main (int argc, char **argv)
mode = 1; /* means directory */
ip_addr = strdup ("0.0.0.0");
+ CURLcode curl_err = CURLE_OK;
+ curl_err = curl_global_init(CURL_GLOBAL_ALL);
+ if (curl_err != CURLE_OK) {
+ error (1, 0, "libcurl cannot be initialized.");
+ }
+
if (debug_flag)
fprintf (stderr, "pid %d\n", getpid ());
httpfs_parse_args (argc, argv);
@@ -165,5 +171,6 @@ main (int argc, char **argv)
/* NOT REACHED */
free (httpfs);
+ curl_global_cleanup();
return 0;
}
diff --git a/httpfs.h b/httpfs.h
index 3b54e48..1ab0669 100644
--- a/httpfs.h
+++ b/httpfs.h
@@ -21,6 +21,8 @@
#include <sys/stat.h>
#include <netdb.h>
+#include <curl/curl.h>
+
#include <hurd/hurd_types.h>
#include "config.h"
@@ -92,7 +94,7 @@ error_t lookup_host (char *url, struct hostent **ent);
/* store the remote socket in *FD and establish the connection and send the
* http GET request */
-error_t open_connection(struct netnode *node, int *fd,off_t *head_len);
+error_t open_connection(struct netnode *node, char *buffer, loff_t
offset, off_t *head_len);
/* make an instance of `struct netnode' with the specified parameters,
* return NULL on error */
diff --git a/netfs.c b/netfs.c
index 0b49d93..e0e6b4b 100644
--- a/netfs.c
+++ b/netfs.c
@@ -464,18 +464,16 @@ error_t netfs_attempt_readlink (struct iouser
*user, struct node *node,
error_t netfs_attempt_read (struct iouser *cred, struct node *node,
loff_t offset, size_t *len, void *data)
{
- error_t err;
- static int remote_fd;
+ error_t err = 0;
char buffer[4096];
int bytes_read;
off_t head_len;
if ( offset==0) {
- err = open_connection (node->nn, &remote_fd,&head_len);
+ err = open_connection (node->nn, buffer, offset, &head_len);
if (err)
return err;
- bytes_read = pread(remote_fd, buffer, sizeof(buffer), offset);
- memcpy(data,(void *)(buffer+head_len),sizeof(buffer)-(sizeof(char)*head_len));
+ bytes_read = strlen(buffer) + head_len;
if (bytes_read < 0)
err = errno;
else {
@@ -486,8 +484,7 @@ error_t netfs_attempt_read (struct iouser *cred,
struct node *node,
}
else {
- bzero(buffer,sizeof(buffer));
- bytes_read = pread(remote_fd, data, sizeof(buffer), offset);
+ bytes_read = strlen(buffer) + head_len;
if (bytes_read < 0)
err = errno;
else {
diff --git a/parsehtml.c b/parsehtml.c
index 3bd202f..5f3ede0 100644
--- a/parsehtml.c
+++ b/parsehtml.c
@@ -98,7 +98,6 @@ error_t parse(struct netnode *node)
/* SAX HTML Parsing Variables */
htmlSAXHandlerPtr emptysax = ∅
error_t err;
- static int remote_fd;
off_t head_len=0,offset=0;
char buffer[MAXBUF];
int bytes_read=0;
@@ -125,20 +124,16 @@ error_t parse(struct netnode *node)
/* get the HTML stream corresponding to the directory
* send a GET request */
- err = open_connection(node,&remote_fd,&head_len);
+ err = open_connection(node, buffer, offset, &head_len);
if ( err )
return err;
- bytes_read = pread(remote_fd,buffer,sizeof(char)*6,offset);
+ bytes_read = strlen(buffer) + head_len;
if ( bytes_read > 0 )
{
/* Parser operates in Push mode */
ctxt =
htmlCreatePushParserCtxt(emptysax,NULL,buffer,bytes_read,"abracabadra",0);
//file name is irrelevant abracabadra
-
- while ( (bytes_read=pread(remote_fd,buffer,sizeof(buffer),offset)) > 0) {
- htmlParseChunk(ctxt,buffer,bytes_read,0);
- }
- htmlParseChunk(ctxt,buffer,0,1);
+ htmlParseChunk(ctxt, buffer, bytes_read, 0);
doc = ctxt->myDoc;
htmlFreeParserCtxt(ctxt);
if ( doc != NULL )
diff --git a/tests/Makefile.am b/tests/Makefile.am
index 2823d0a..dfb741b 100644
--- a/tests/Makefile.am
+++ b/tests/Makefile.am
@@ -2,7 +2,7 @@ check_PROGRAMS = test_httpfs
test_httpfs_SOURCES = test_httpfs.c ../http.c ../parsehtml.c
../node.c ../extract.c ../netfs.c ../args.c ../globals.c
test_httpfs_CFLAGS = -I$(top_srcdir) $(CMOCKA_CFLAGS)
-test_httpfs_LDFLAGS = -Wl,--wrap=gethostbyname -Wl,--export-dynamic
+test_httpfs_LDFLAGS = -Wl,--wrap=curl_easy_init -Wl,--export-dynamic
test_httpfs_LDADD = $(CMOCKA_LIBS) $(MHD_LIBS) $(HTTPFS_SHARED_LIBS)
TESTS = $(check_PROGRAMS)
diff --git a/tests/test_httpfs.c b/tests/test_httpfs.c
index bd2d65b..f46c8f5 100644
--- a/tests/test_httpfs.c
+++ b/tests/test_httpfs.c
@@ -37,50 +37,31 @@
#include "globals.h"
-struct hostent *__wrap_gethostbyname(const char *name) {
- assert_non_null(name);
-
- if (strcmp(name, "www.example.com") == 0) {
- return (struct hostent *)mock();
+/* http.c unit tests */
+CURL *__wrap_curl_easy_init(int is_curl_null) {
+ if (is_curl_null == 1) {
+ return NULL;
}
- return (struct hostent *)mock();
+ return (CURL *)mock();
}
-/* http.c unit tests */
-static void test_lookup_host_1(void **state) {
+static void test_open_connection_curl_easy_init_fails(void **state)
+{
(void) state;
- struct hostent mock_hostent;
- memset(&mock_hostent, 0, sizeof(mock_hostent));
- mock_hostent.h_name = "www.example.com";
-
- will_return(__wrap_gethostbyname, &mock_hostent);
-
- struct hostent *result = NULL;
- int ret = lookup_host("www.example.com", &result);
-
- assert_int_equal(ret, 0);
- assert_non_null(result);
- assert_string_equal(result->h_name, "www.example.com");
-}
-
-static void test_lookup_host_2(void **state) {
- (void) state;
+ CURL *mock_curl;
- will_return(__wrap_gethostbyname, NULL);
+ will_return(__wrap_curl_easy_init, NULL);
- struct hostent *result = NULL;
- int ret = lookup_host("invalid.url", &result);
+ error_t err = open_connection(NULL, NULL, 0, NULL);
- assert_int_equal(ret, EINVAL);
- assert_null(result);
+ assert_int_equal(err, ENOMEM);
}
int main(void) {
const struct CMUnitTest tests[] = {
- cmocka_unit_test(test_lookup_host_1),
- cmocka_unit_test(test_lookup_host_2),
+ cmocka_unit_test(test_open_connection_curl_easy_init_fails),
};
return cmocka_run_group_tests(tests, NULL, NULL);
--
2.56.0
