35 #include <sys/types.h>
63 HTTPCache *HTTPCache::_instance = 0;
72 static pthread_mutex_t instance_mutex;
73 static pthread_once_t once_block = PTHREAD_ONCE_INIT;
76 #define NO_LM_EXPIRATION 24*3600 // 24 hours
78 #define DUMP_FREQUENCY 10 // Dump index every x loads
80 #define MEGA 0x100000L
81 #define CACHE_TOTAL_SIZE 20 // Default cache size is 20M
82 #define CACHE_FOLDER_PCT 10 // 10% of cache size for metainfo etc.
83 #define CACHE_GC_PCT 10 // 10% of cache size free after GC
84 #define MIN_CACHE_TOTAL_SIZE 5 // 5M Min cache size
85 #define MAX_CACHE_ENTRY_SIZE 3 // 3M Max size of single cached entry
91 status =
INIT(&instance_mutex);
94 throw InternalErr(__FILE__, __LINE__,
"Could not initialize the HTTP Cache mutex. Exiting.");
126 HTTPCache::instance(
const string &cache_root,
bool force)
128 int status = pthread_once(&once_block, once_init_routine);
130 throw InternalErr(__FILE__, __LINE__,
"Could not initialize the HTTP Cache mutex. Exiting.");
132 LOCK(&instance_mutex);
134 DBG(cerr <<
"Entering instance(); (" << hex << _instance << dec <<
")"
139 _instance =
new HTTPCache(cache_root, force);
141 DBG(cerr <<
"New instance: " << _instance <<
", cache root: "
142 << _instance->d_cache_root << endl);
144 atexit(delete_instance);
157 SignalHandler::instance()->register_handler(SIGINT, old_eh);
159 "Could not register event handler for SIGINT without superseding an existing one.");
164 SignalHandler::instance()->register_handler(SIGPIPE, old_eh);
166 "Could not register event handler for SIGPIPE without superseding an existing one.");
171 SignalHandler::instance()->register_handler(SIGTERM, old_eh);
173 "Could not register event handler for SIGTERM without superseding an existing one.");
179 DBG2(cerr <<
"The constructor threw an Error!" << endl);
185 DBGN(cerr <<
"returning " << hex << _instance << dec << endl);
194 HTTPCache::delete_instance()
196 DBG(cerr <<
"Entering delete_instance()..." << endl);
198 if (HTTPCache::_instance) {
199 DBG(cerr <<
"Deleting the cache: " << HTTPCache::_instance << endl);
200 delete HTTPCache::_instance;
201 HTTPCache::_instance = 0;
204 delete SignalHandler::instance()->remove_handler(SIGINT);
205 delete SignalHandler::instance()->remove_handler(SIGPIPE);
206 delete SignalHandler::instance()->remove_handler(SIGTERM);
209 DBG(cerr <<
"Exiting delete_instance()" << endl);
226 HTTPCache::HTTPCache(
string cache_root,
bool force) :
227 d_locked_open_file(0),
228 d_cache_enabled(false),
229 d_cache_protected(false),
230 d_expire_ignored(false),
231 d_always_validate(false),
240 d_http_cache_table(0)
242 DBG(cerr <<
"Entering the constructor for " <<
this <<
"... ");
244 int status = pthread_once(&once_block, once_init_routine);
246 throw InternalErr(__FILE__, __LINE__,
"Could not initialize the HTTP Cache mutex. Exiting.");
248 INIT(&d_cache_mutex);
260 set_cache_root(cache_root);
263 if (!get_single_user_lock(force))
264 throw Error(
"Could not get single user lock for the cache");
274 if (stat(cache_root.c_str(), &s) == 0)
275 block_size = s.st_blksize;
277 throw Error(
"Could not set file system block size.");
279 d_http_cache_table =
new HTTPCacheTable(d_cache_root, block_size);
280 d_cache_enabled =
true;
282 DBGN(cerr <<
"exiting" << endl);
299 DBG(cerr <<
"Entering the destructor for " <<
this <<
"... ");
303 perform_garbage_collection();
315 delete d_http_cache_table;
317 release_single_user_lock();
319 DBGN(cerr <<
"exiting destructor." << endl);
332 HTTPCache::stopGC()
const
334 return (d_http_cache_table->
get_current_size() + d_folder_size < d_total_size - d_gc_buffer);
344 HTTPCache::startGC()
const
347 return (d_http_cache_table->
get_current_size() + d_folder_size > d_total_size);
365 HTTPCache::perform_garbage_collection()
367 DBG(cerr <<
"Performing garbage collection" << endl);
386 HTTPCache::expired_gc()
388 if (!d_expire_ignored) {
426 void HTTPCache::too_big_gc() {
443 bool HTTPCache::get_single_user_lock(
bool force)
445 if (!d_locked_open_file) {
451 create_cache_root(d_cache_root);
457 DBG(cerr <<
"Failure to create the cache root" << endl);
463 if ((fp = fopen(lock.c_str(),
"r")) != NULL) {
464 int res = fclose(fp);
466 DBG(cerr <<
"Failed to close " << (
void *)fp << endl);
474 if ((fp = fopen(lock.c_str(),
"w")) == NULL) {
475 DBG(cerr <<
"Could not open for write access" << endl);
479 d_locked_open_file = fp;
483 DBG(cerr <<
"locked_open_file is true" << endl);
490 HTTPCache::release_single_user_lock()
492 if (d_locked_open_file) {
493 int res = fclose(d_locked_open_file);
495 DBG(cerr <<
"Failed to close " << (
void *)d_locked_open_file << endl) ;
497 d_locked_open_file = 0;
526 HTTPCache::create_cache_root(
const string &cache_root)
528 struct stat stat_info;
529 string::size_type cur = 0;
532 cur = cache_root[1] ==
':' ? 3 : 1;
538 string dir = cache_root.substr(0, cur);
539 if (stat(dir.c_str(), &stat_info) == -1) {
540 DBG2(cerr <<
"Cache....... Creating " << dir << endl);
541 mode_t mask =
UMASK(0);
542 if (
MKDIR(dir.c_str(), 0777) < 0) {
543 DBG2(cerr <<
"Error: can't create." << endl);
545 throw Error(
string(
"Could not create the directory for the cache. Failed when building path at ") + dir +
string(
"."));
550 DBG2(cerr <<
"Cache....... Found " << dir << endl);
571 HTTPCache::set_cache_root(
const string &root)
583 char * cr = (
char *) getenv(
"DODS_CACHE");
584 if (!cr) cr = (
char *) getenv(
"TMP");
585 if (!cr) cr = (
char *) getenv(
"TEMP");
601 if (d_http_cache_table)
621 d_cache_enabled = mode;
631 DBG2(cerr <<
"In HTTPCache::is_cache_enabled: (" << d_cache_enabled <<
")"
633 return d_cache_enabled;
651 d_cache_disconnected = mode;
661 return d_cache_disconnected;
677 d_expire_ignored = mode;
688 return d_expire_ignored;
714 (size > ULONG_MAX ? ULONG_MAX : size *
MEGA);
715 unsigned long old_size = d_total_size;
716 d_total_size = new_size;
720 if (new_size < old_size && startGC()) {
721 perform_garbage_collection();
727 DBGN(cerr <<
"Unlocking interface." << endl);
731 DBG2(cerr <<
"Cache....... Total cache size: " << d_total_size
732 <<
" with " << d_folder_size
733 <<
" bytes for meta information and folders and at least "
734 << d_gc_buffer <<
" bytes free after every gc" << endl);
744 return d_total_size /
MEGA;
761 unsigned long new_size = size *
MEGA;
762 if (new_size > 0 && new_size < d_total_size - d_folder_size) {
763 unsigned long old_size = d_max_entry_size;
764 d_max_entry_size = new_size;
765 if (new_size < old_size && startGC()) {
766 perform_garbage_collection();
776 DBG2(cerr <<
"Cache...... Max entry cache size is "
777 << d_max_entry_size << endl);
789 return d_max_entry_size /
MEGA;
807 d_default_expiration = exp_time;
817 return d_default_expiration;
827 d_always_validate = validate;
836 return d_always_validate;
861 d_cache_control = cc;
863 vector<string>::const_iterator i;
864 for (i = cc.begin(); i != cc.end(); ++i) {
865 string header = (*i).substr(0, (*i).find(
':'));
866 string value = (*i).substr((*i).find(
": ") + 2);
867 if (header !=
"Cache-Control") {
868 throw InternalErr(__FILE__, __LINE__,
"Expected cache control header not found.");
871 if (value ==
"no-cache" || value ==
"no-store")
872 d_cache_enabled =
false;
873 else if (value.find(
"max-age") != string::npos) {
874 string max_age = value.substr(value.find(
"=" + 1));
877 else if (value ==
"max-stale")
879 else if (value.find(
"max-stale") != string::npos) {
880 string max_stale = value.substr(value.find(
"=" + 1));
883 else if (value.find(
"min-fresh") != string::npos) {
884 string min_fresh = value.substr(value.find(
"=" + 1));
906 return d_cache_control;
920 HTTPCache::is_url_in_cache(
const string &url)
922 DBG(cerr <<
"Is this url in the cache? (" << url <<
")" << endl);
925 bool status = entry != 0;
940 return header.find(
"Connection") != string::npos
941 || header.find(
"Keep-Alive") != string::npos
942 || header.find(
"Proxy-Authenticate") != string::npos
943 || header.find(
"Proxy-Authorization") != string::npos
944 || header.find(
"Transfer-Encoding") != string::npos
945 || header.find(
"Upgrade") != string::npos;
960 HTTPCache::write_metadata(
const string &cachename,
const vector<string> &headers)
963 d_open_files.push_back(fname);
965 FILE *dest = fopen(fname.c_str(),
"w");
967 throw InternalErr(__FILE__, __LINE__,
968 "Could not open named cache entry file.");
971 vector<string>::const_iterator i;
972 for (i = headers.begin(); i != headers.end(); ++i) {
974 int s = fwrite((*i).c_str(), (*i).size(), 1, dest);
977 throw InternalErr(__FILE__, __LINE__,
"could not write header: '" + (*i) +
"' " +
long_to_string(s));
979 s = fwrite(
"\n", 1, 1, dest);
982 throw InternalErr(__FILE__, __LINE__,
"could not write header: " +
long_to_string(s));
987 int res = fclose(dest);
989 DBG(cerr <<
"HTTPCache::write_metadata - Failed to close "
993 d_open_files.pop_back();
1007 HTTPCache::read_metadata(
const string &cachename, vector<string> &headers)
1009 FILE *md = fopen(
string(cachename + CACHE_META).c_str(),
"r");
1011 throw InternalErr(__FILE__, __LINE__,
1012 "Could not open named cache entry meta data file.");
1016 while (!feof(md) && fgets(line, 1024, md)) {
1017 line[min(1024, static_cast<int>(strlen(line)))-1] =
'\0';
1018 headers.push_back(
string(line));
1021 int res = fclose(md);
1023 DBG(cerr <<
"HTTPCache::read_metadata - Failed to close "
1050 HTTPCache::write_body(
const string &cachename,
const FILE *src)
1052 d_open_files.push_back(cachename);
1054 FILE *dest = fopen(cachename.c_str(),
"wb");
1056 throw InternalErr(__FILE__, __LINE__,
1057 "Could not open named cache entry file.");
1065 while ((n = fread(line, 1, 1024, const_cast<FILE *>(src))) > 0) {
1066 total += fwrite(line, 1, n, dest);
1070 if (ferror(const_cast<FILE *>(src)) || ferror(dest)) {
1071 int res = fclose(dest);
1072 res = res & unlink(cachename.c_str());
1074 DBG(cerr <<
"HTTPCache::write_body - Failed to close/unlink "
1077 throw InternalErr(__FILE__, __LINE__,
1078 "I/O error transferring data to the cache.");
1081 rewind(const_cast<FILE *>(src));
1083 int res = fclose(dest);
1085 DBG(cerr <<
"HTTPCache::write_body - Failed to close "
1089 d_open_files.pop_back();
1103 HTTPCache::open_body(
const string &cachename)
1105 DBG(cerr <<
"cachename: " << cachename << endl);
1107 FILE *src = fopen(cachename.c_str(),
"rb");
1109 throw InternalErr(__FILE__, __LINE__,
"Could not open cache file.");
1141 const vector<string> &headers,
const FILE *body)
1145 DBG(cerr <<
"Caching url: " << url <<
"." << endl);
1149 if (url.find(
"http:") == string::npos &&
1150 url.find(
"https:") == string::npos) {
1164 d_http_cache_table->
parse_headers(entry, d_max_entry_size, headers);
1166 DBG(cerr <<
"Not cache-able; deleting HTTPCacheTable::CacheEntry: " << entry
1167 <<
"(" << url <<
")" << endl);
1169 delete entry; entry = 0;
1175 d_http_cache_table->
calculate_time(entry, d_default_expiration, request_time);
1189 DBG(cerr <<
"Too big; deleting HTTPCacheTable::CacheEntry: " << entry <<
"(" << url
1192 delete entry; entry = 0;
1199 perform_garbage_collection();
1238 vector<string> headers;
1240 DBG(cerr <<
"Getting conditional request headers for " << url << endl);
1243 entry = d_http_cache_table->get_locked_entry_from_cache_table(url);
1245 throw Error(
"There is no cache entry for the URL: " + url);
1248 headers.push_back(
string(
"If-None-Match: ") + entry->
get_etag());
1250 if (entry->
get_lm() > 0) {
1251 time_t lm = entry->
get_lm();
1252 headers.push_back(
string(
"If-Modified-Since: ")
1257 headers.push_back(
string(
"If-Modified-Since: ")
1262 headers.push_back(
string(
"If-Modified-Since: ")
1282 struct HeaderLess: binary_function<const string&, const string&, bool>
1284 bool operator()(
const string &s1,
const string &s2)
const {
1285 return s1.substr(0, s1.find(
':')) < s2.substr(0, s2.find(
':'));
1304 const vector<string> &headers)
1309 DBG(cerr <<
"Updating the response headers for: " << url << endl);
1314 throw Error(
"There is no cache entry for the URL: " + url);
1317 d_http_cache_table->
parse_headers(entry, d_max_entry_size, headers);
1320 d_http_cache_table->
calculate_time(entry, d_default_expiration, request_time);
1328 set<string, HeaderLess> merged_headers;
1331 copy(headers.begin(), headers.end(),
1332 inserter(merged_headers, merged_headers.begin()));
1335 vector<string> old_headers;
1337 copy(old_headers.begin(), old_headers.end(),
1338 inserter(merged_headers, merged_headers.begin()));
1343 vector<string> result;
1344 copy(merged_headers.rbegin(), merged_headers.rend(),
1345 back_inserter(result));
1379 DBG(cerr <<
"Is this URL valid? (" << url <<
")" << endl);
1382 if (d_always_validate) {
1387 entry = d_http_cache_table->get_locked_entry_from_cache_table(url);
1389 throw Error(
"There is no cache entry for the URL: " + url);
1407 if (d_max_age >= 0 && current_age > d_max_age) {
1408 DBG(cerr <<
"Cache....... Max-age validation" << endl);
1413 if (d_min_fresh >= 0
1415 DBG(cerr <<
"Cache....... Min-fresh validation" << endl);
1422 + (d_max_stale >= 0 ? d_max_stale : 0) > current_age);
1465 vector<string> &headers,
string &cacheName) {
1471 DBG(cerr <<
"Getting the cached response for " << url << endl);
1474 entry = d_http_cache_table->get_locked_entry_from_cache_table(url);
1483 DBG(cerr <<
"Headers just read from cache: " << endl);
1484 DBGN(copy(headers.begin(), headers.end(), ostream_iterator<string>(cerr,
"\n")));
1488 DBG(cerr <<
"Returning: " << url <<
" from the cache." << endl);
1520 string discard_name;
1537 string discard_name;
1538 vector<string> discard_headers;
1590 throw Error(
"Attempt to purge the cache with entries in use.");
vector< string > get_conditional_request_headers(const string &url)
bool is_cache_enabled() const
vector< string > get_cache_control()
void set_cache_enabled(bool mode)
time_t parse_time(const char *str, bool expand)
bool is_url_valid(const string &url)
string get_error_message() const
void create_location(CacheEntry *entry)
void unlock_read_response()
void set_max_size(unsigned long size)
void parse_headers(HTTPCacheTable::CacheEntry *entry, unsigned long max_entry_size, const vector< string > &headers)
void add_entry_to_cache_table(CacheEntry *entry)
FILE * get_cached_response(const string &url, vector< string > &headers, string &cacheName)
void calculate_time(HTTPCacheTable::CacheEntry *entry, int default_expiration, time_t request_time)
void delete_by_size(unsigned int size)
void lock_write_response()
int get_new_entries() const
int get_default_expiration() const
A class for software fault reporting.
void delete_all_entries()
bool is_hop_by_hop_header(const string &header)
unsigned long get_max_entry_size() const
bool is_expire_ignored() const
#define MAX_CACHE_ENTRY_SIZE
void set_size(unsigned long sz)
CacheDisconnectedMode get_cache_disconnected() const
#define DIR_SEPARATOR_CHAR
void update_response(const string &url, time_t request_time, const vector< string > &headers)
bool get_always_validate() const
void bind_entry_to_data(CacheEntry *entry, FILE *body)
string get_cache_root() const
void set_cache_control(const vector< string > &cc)
void delete_expired_entries(time_t time=0)
string date_time_str(time_t *calendar, bool local)
bool get_must_revalidate()
unsigned long get_current_size() const
bool cache_response(const string &url, time_t request_time, const vector< string > &headers, const FILE *body)
time_t get_freshness_lifetime()
#define MIN_CACHE_TOTAL_SIZE
string long_to_string(long val, int base)
void unlock_write_response()
void set_always_validate(bool validate)
void set_default_expiration(int exp_time)
void remove_entry_from_cache_table(const string &url)
void release_cached_response(FILE *response)
void unlock_cache_interface()
CacheEntry * get_write_locked_entry_from_cache_table(const string &url)
void lock_cache_interface()
time_t get_response_time()
unsigned long get_max_size() const
void delete_by_hits(int hits)
A class for error processing.
void set_expire_ignored(bool mode)
void set_max_entry_size(unsigned long size)
time_t get_corrected_initial_age()
void uncouple_entry_from_data(FILE *body)
bool is_locked_read_responses()
void set_cache_root(const string &cr)
void set_cache_disconnected(CacheDisconnectedMode mode)