00001
00002
00003
00004
00005
00006
00007
00008
00009
00010
00011
00012
00013
00014
00015
00016
00017
00018
00019
00020
00021
00022
00023
00024
00025
00026 #include "config.h"
00027
00028
00029
00030
00031 #include <pthread.h>
00032 #include <limits.h>
00033 #include <unistd.h>
00034 #include <sys/types.h>
00035 #include <sys/stat.h>
00036
00037 #include <cstring>
00038 #include <iostream>
00039 #include <sstream>
00040 #include <algorithm>
00041 #include <iterator>
00042 #include <set>
00043
00044 #include "Error.h"
00045 #include "InternalErr.h"
00046 #include "ResponseTooBigErr.h"
00047 #ifndef WIN32
00048 #include "SignalHandler.h"
00049 #endif
00050 #include "HTTPCacheInterruptHandler.h"
00051 #include "HTTPCacheTable.h"
00052 #include "HTTPCache.h"
00053
00054 #include "util_mit.h"
00055 #include "debug.h"
00056
00057 using namespace std;
00058
00059 namespace libdap {
00060
00061 HTTPCache *HTTPCache::_instance = 0;
00062
00063
00064
00065
00066
00067
00068
00069
00070 static pthread_mutex_t instance_mutex;
00071 static pthread_once_t once_block = PTHREAD_ONCE_INIT;
00072
00073 #ifdef WIN32
00074 #include <direct.h>
00075 #include <time.h>
00076 #include <fcntl.h>
00077 #define MKDIR(a,b) _mkdir((a))
00078 #define UMASK(a) _umask((a))
00079 #define REMOVE(a) remove((a))
00080 #define MKSTEMP(a) _open(_mktemp((a)),_O_CREAT,_S_IREAD|_S_IWRITE)
00081 #define DIR_SEPARATOR_CHAR '\\'
00082 #define DIR_SEPARATOR_STR "\\"
00083 #else
00084 #define MKDIR(a,b) mkdir((a), (b))
00085 #define UMASK(a) umask((a))
00086 #define REMOVE(a) remove((a))
00087 #define MKSTEMP(a) mkstemp((a))
00088 #define DIR_SEPARATOR_CHAR '/'
00089 #define DIR_SEPARATOR_STR "/"
00090 #endif
00091
00092 #ifdef WIN32
00093 #define CACHE_LOC "\\tmp\\"
00094 #define CACHE_ROOT "dods-cache\\"
00095 #else
00096 #define CACHE_LOC "/tmp/"
00097 #define CACHE_ROOT "dods-cache/"
00098 #endif
00099 #define CACHE_INDEX ".index"
00100 #define CACHE_LOCK ".lock"
00101 #define CACHE_META ".meta"
00102
00103
00104 #define NO_LM_EXPIRATION 24*3600 // 24 hours
00105
00106 #define DUMP_FREQUENCY 10 // Dump index every x loads
00107
00108 #define MEGA 0x100000L
00109 #define CACHE_TOTAL_SIZE 20 // Default cache size is 20M
00110 #define CACHE_FOLDER_PCT 10 // 10% of cache size for metainfo etc.
00111 #define CACHE_GC_PCT 10 // 10% of cache size free after GC
00112 #define MIN_CACHE_TOTAL_SIZE 5 // 5M Min cache size
00113 #define MAX_CACHE_ENTRY_SIZE 3 // 3M Max size of single cached entry
00114
00115 static void
00116 once_init_routine()
00117 {
00118 int status;
00119 status = INIT(&instance_mutex);
00120
00121 if (status != 0)
00122 throw InternalErr(__FILE__, __LINE__, "Could not initialize the HTTP Cache mutex. Exiting.");
00123 }
00124
00153 HTTPCache *
00154 HTTPCache::instance(const string &cache_root, bool force)
00155 {
00156 LOCK(&instance_mutex);
00157 DBG(cerr << "Entering instance(); (" << hex << _instance << dec << ")"
00158 << "... ");
00159
00160 try {
00161 if (!_instance) {
00162 _instance = new HTTPCache(cache_root, force);
00163
00164 DBG(cerr << "New instance: " << _instance << ", cache root: "
00165 << _instance->d_cache_root << endl);
00166
00167 atexit(delete_instance);
00168
00169 #ifndef WIN32
00170
00171
00172
00173
00174
00175
00176
00177
00178 EventHandler *old_eh = SignalHandler::instance()->register_handler
00179 (SIGINT, new HTTPCacheInterruptHandler);
00180 if (old_eh) {
00181 SignalHandler::instance()->register_handler(SIGINT, old_eh);
00182 throw SignalHandlerRegisteredErr(
00183 "Could not register event handler for SIGINT without superseding an existing one.");
00184 }
00185
00186 old_eh = SignalHandler::instance()->register_handler
00187 (SIGPIPE, new HTTPCacheInterruptHandler);
00188 if (old_eh) {
00189 SignalHandler::instance()->register_handler(SIGPIPE, old_eh);
00190 throw SignalHandlerRegisteredErr(
00191 "Could not register event handler for SIGPIPE without superseding an existing one.");
00192 }
00193
00194 old_eh = SignalHandler::instance()->register_handler
00195 (SIGTERM, new HTTPCacheInterruptHandler);
00196 if (old_eh) {
00197 SignalHandler::instance()->register_handler(SIGTERM, old_eh);
00198 throw SignalHandlerRegisteredErr(
00199 "Could not register event handler for SIGTERM without superseding an existing one.");
00200 }
00201 #endif
00202 }
00203 }
00204 catch (...) {
00205 DBG2(cerr << "The constructor threw an Error!" << endl);
00206 UNLOCK(&instance_mutex);
00207 throw;
00208 }
00209
00210 UNLOCK(&instance_mutex);
00211 DBGN(cerr << "returning " << hex << _instance << dec << endl);
00212
00213 return _instance;
00214 }
00215
00219 void
00220 HTTPCache::delete_instance()
00221 {
00222 DBG(cerr << "Entering delete_instance()..." << endl);
00223 if (HTTPCache::_instance) {
00224 DBG(cerr << "Deleting the cache: " << HTTPCache::_instance << endl);
00225 delete HTTPCache::_instance;
00226 HTTPCache::_instance = 0;
00227 }
00228
00229 DBG(cerr << "Exiting delete_instance()" << endl);
00230 }
00231
00246 HTTPCache::HTTPCache(string cache_root, bool force) :
00247 d_locked_open_file(0),
00248 d_cache_enabled(false),
00249 d_cache_protected(false),
00250 d_expire_ignored(false),
00251 d_always_validate(false),
00252 d_total_size(CACHE_TOTAL_SIZE * MEGA),
00253 d_folder_size(CACHE_TOTAL_SIZE / CACHE_FOLDER_PCT),
00254 d_gc_buffer(CACHE_TOTAL_SIZE / CACHE_GC_PCT),
00255 d_max_entry_size(MAX_CACHE_ENTRY_SIZE * MEGA),
00256 d_default_expiration(NO_LM_EXPIRATION),
00257 d_max_age(-1),
00258 d_max_stale(-1),
00259 d_min_fresh(-1),
00260 d_http_cache_table(0)
00261 {
00262 DBG(cerr << "Entering the constructor for " << this << "... ");
00263
00264 int status = pthread_once(&once_block, once_init_routine);
00265 if (status != 0)
00266 throw InternalErr(__FILE__, __LINE__, "Could not initialize the HTTP Cache mutex. Exiting.");
00267
00268 INIT(&d_cache_mutex);
00269
00270
00271
00272
00273
00274
00275
00276
00277
00278
00279
00280 set_cache_root(cache_root);
00281 int block_size;
00282
00283 if (!get_single_user_lock(force))
00284 throw Error("Could not get single user lock for the cache");
00285
00286 #ifdef WIN32
00287
00288
00289
00290
00291 block_size = 4096;
00292 #else
00293 struct stat s;
00294 if (stat(cache_root.c_str(), &s) == 0)
00295 block_size = s.st_blksize;
00296 else
00297 throw Error("Could not set file system block size.");
00298 #endif
00299 d_http_cache_table = new HTTPCacheTable(d_cache_root, block_size);
00300 d_cache_enabled = true;
00301
00302 DBGN(cerr << "exiting" << endl);
00303 }
00304
00317 HTTPCache::~HTTPCache()
00318 {
00319 DBG(cerr << "Entering the destructor for " << this << "... ");
00320
00321 try {
00322 if (startGC())
00323 perform_garbage_collection();
00324
00325 d_http_cache_table->cache_index_write();
00326 }
00327 catch (Error &e) {
00328
00329
00330
00331
00332 DBG(cerr << e.get_error_message() << endl);
00333 }
00334
00335 delete d_http_cache_table;
00336
00337 release_single_user_lock();
00338
00339 DBGN(cerr << "exiting destructor." << endl);
00340 DESTROY(&d_cache_mutex);
00341 }
00342
00343
00347
00351 bool
00352 HTTPCache::stopGC() const
00353 {
00354 return (d_http_cache_table->get_current_size() + d_folder_size < d_total_size - d_gc_buffer);
00355 }
00356
00363 bool
00364 HTTPCache::startGC() const
00365 {
00366 DBG(cerr << "startGC, current_size: " << d_http_cache_table->get_current_size() << endl);
00367 return (d_http_cache_table->get_current_size() + d_folder_size > d_total_size);
00368 }
00369
00384 void
00385 HTTPCache::perform_garbage_collection()
00386 {
00387 DBG(cerr << "Performing garbage collection" << endl);
00388
00389
00390 expired_gc();
00391
00392
00393 too_big_gc();
00394
00395
00396
00397 hits_gc();
00398 }
00399
00405 void
00406 HTTPCache::expired_gc()
00407 {
00408 if (!d_expire_ignored) {
00409 d_http_cache_table->delete_expired_entries();
00410 }
00411 }
00412
00429 void
00430 HTTPCache::hits_gc()
00431 {
00432 int hits = 0;
00433
00434 if (startGC()) {
00435 while (!stopGC()) {
00436 d_http_cache_table->delete_by_hits(hits);
00437 hits++;
00438 }
00439 }
00440 }
00441
00446 void HTTPCache::too_big_gc() {
00447 if (startGC())
00448 d_http_cache_table->delete_by_size(d_max_entry_size);
00449 }
00450
00452
00463 bool HTTPCache::get_single_user_lock(bool force) {
00464 if (!d_locked_open_file) {
00465 FILE * fp = NULL;
00466
00467 try {
00468
00469
00470 create_cache_root(d_cache_root);
00471 }
00472 catch (Error &e) {
00473
00474
00475
00476 return false;
00477 }
00478
00479
00480 string lock = d_cache_root + CACHE_LOCK;
00481 if ((fp = fopen(lock.c_str(), "r")) != NULL) {
00482 int res = fclose(fp);
00483 if (res) {
00484 DBG(cerr << "Failed to close " << (void *)fp << endl);
00485 }
00486 if (force)
00487 REMOVE(lock.c_str());
00488 else
00489 return false;
00490 }
00491
00492 if ((fp = fopen(lock.c_str(), "w")) == NULL)
00493 return false;
00494
00495 d_locked_open_file = fp;
00496 return true;
00497 }
00498
00499 return false;
00500 }
00501
00504 void
00505 HTTPCache::release_single_user_lock()
00506 {
00507 if (d_locked_open_file) {
00508 int res = fclose(d_locked_open_file);
00509 if (res) {
00510 DBG(cerr << "Failed to close " << (void *)d_locked_open_file << endl) ;
00511 }
00512 d_locked_open_file = 0;
00513 }
00514
00515 string lock = d_cache_root + CACHE_LOCK;
00516 REMOVE(lock.c_str());
00517 }
00518
00521
00525 string
00526 HTTPCache::get_cache_root() const
00527 {
00528 return d_cache_root;
00529 }
00530
00531
00540 void
00541 HTTPCache::create_cache_root(const string &cache_root)
00542 {
00543 struct stat stat_info;
00544 string::size_type cur = 0;
00545
00546 #ifdef WIN32
00547 cur = cache_root[1] == ':' ? 3 : 1;
00548 typedef int mode_t;
00549 #else
00550 cur = 1;
00551 #endif
00552 while ((cur = cache_root.find(DIR_SEPARATOR_CHAR, cur)) != string::npos) {
00553 string dir = cache_root.substr(0, cur);
00554 if (stat(dir.c_str(), &stat_info) == -1) {
00555 DBG2(cerr << "Cache....... Creating " << dir << endl);
00556 mode_t mask = UMASK(0);
00557 if (MKDIR(dir.c_str(), 0777) < 0) {
00558 DBG2(cerr << "Error: can't create." << endl);
00559 UMASK(mask);
00560 throw Error(string("Could not create the directory for the cache. Failed when building path at ") + dir + string("."));
00561 }
00562 UMASK(mask);
00563 }
00564 else {
00565 DBG2(cerr << "Cache....... Found " << dir << endl);
00566 }
00567 cur++;
00568 }
00569 }
00570
00585 void
00586 HTTPCache::set_cache_root(const string &root)
00587 {
00588 if (root != "") {
00589 d_cache_root = root;
00590
00591 if (d_cache_root[d_cache_root.size()-1] != DIR_SEPARATOR_CHAR)
00592 d_cache_root += DIR_SEPARATOR_CHAR;
00593 }
00594 else {
00595
00596
00597 char * cr = (char *) getenv("DODS_CACHE");
00598 if (!cr) cr = (char *) getenv("TMP");
00599 if (!cr) cr = (char *) getenv("TEMP");
00600 if (!cr) cr = CACHE_LOC;
00601
00602 d_cache_root = cr;
00603 if (d_cache_root[d_cache_root.size()-1] != DIR_SEPARATOR_CHAR)
00604 d_cache_root += DIR_SEPARATOR_CHAR;
00605
00606 d_cache_root += CACHE_ROOT;
00607 }
00608
00609
00610
00611
00612 if (d_http_cache_table)
00613 d_http_cache_table->set_cache_root(d_cache_root);
00614 }
00615
00627 void
00628 HTTPCache::set_cache_enabled(bool mode)
00629 {
00630 lock_cache_interface();
00631
00632 d_cache_enabled = mode;
00633
00634 unlock_cache_interface();
00635 }
00636
00639 bool
00640 HTTPCache::is_cache_enabled() const
00641 {
00642 DBG2(cerr << "In HTTPCache::is_cache_enabled: (" << d_cache_enabled << ")"
00643 << endl);
00644 return d_cache_enabled;
00645 }
00646
00657 void
00658 HTTPCache::set_cache_disconnected(CacheDisconnectedMode mode)
00659 {
00660 lock_cache_interface();
00661
00662 d_cache_disconnected = mode;
00663
00664 unlock_cache_interface();
00665 }
00666
00669 CacheDisconnectedMode
00670 HTTPCache::get_cache_disconnected() const
00671 {
00672 return d_cache_disconnected;
00673 }
00674
00683 void
00684 HTTPCache::set_expire_ignored(bool mode)
00685 {
00686 lock_cache_interface();
00687
00688 d_expire_ignored = mode;
00689
00690 unlock_cache_interface();
00691 }
00692
00693
00694
00695
00696 bool
00697 HTTPCache::is_expire_ignored() const
00698 {
00699 return d_expire_ignored;
00700 }
00701
00717 void
00718 HTTPCache::set_max_size(unsigned long size)
00719 {
00720 lock_cache_interface();
00721
00722 try {
00723 unsigned long new_size = size < MIN_CACHE_TOTAL_SIZE ?
00724 MIN_CACHE_TOTAL_SIZE * MEGA :
00725 (size > ULONG_MAX ? ULONG_MAX : size * MEGA);
00726 unsigned long old_size = d_total_size;
00727 d_total_size = new_size;
00728 d_folder_size = d_total_size / CACHE_FOLDER_PCT;
00729 d_gc_buffer = d_total_size / CACHE_GC_PCT;
00730
00731 if (new_size < old_size && startGC()) {
00732 perform_garbage_collection();
00733 d_http_cache_table->cache_index_write();
00734 }
00735 }
00736 catch (...) {
00737 unlock_cache_interface();
00738 DBGN(cerr << "Unlocking interface." << endl);
00739 throw;
00740 }
00741
00742 DBG2(cerr << "Cache....... Total cache size: " << d_total_size
00743 << " with " << d_folder_size
00744 << " bytes for meta information and folders and at least "
00745 << d_gc_buffer << " bytes free after every gc" << endl);
00746
00747 unlock_cache_interface();
00748 }
00749
00752 unsigned long
00753 HTTPCache::get_max_size() const
00754 {
00755 return d_total_size / MEGA;
00756 }
00757
00766 void
00767 HTTPCache::set_max_entry_size(unsigned long size)
00768 {
00769 lock_cache_interface();
00770
00771 try {
00772 unsigned long new_size = size * MEGA;
00773 if (new_size > 0 && new_size < d_total_size - d_folder_size) {
00774 unsigned long old_size = d_max_entry_size;
00775 d_max_entry_size = new_size;
00776 if (new_size < old_size && startGC()) {
00777 perform_garbage_collection();
00778 d_http_cache_table->cache_index_write();
00779 }
00780 }
00781 }
00782 catch (...) {
00783 unlock_cache_interface();
00784 throw;
00785 }
00786
00787 DBG2(cerr << "Cache...... Max entry cache size is "
00788 << d_max_entry_size << endl);
00789
00790 unlock_cache_interface();
00791 }
00792
00797 unsigned long
00798 HTTPCache::get_max_entry_size() const
00799 {
00800 return d_max_entry_size / MEGA;
00801 }
00802
00813 void
00814 HTTPCache::set_default_expiration(const int exp_time)
00815 {
00816 lock_cache_interface();
00817
00818 d_default_expiration = exp_time;
00819
00820 unlock_cache_interface();
00821 }
00822
00825 int
00826 HTTPCache::get_default_expiration() const
00827 {
00828 return d_default_expiration;
00829 }
00830
00835 void
00836 HTTPCache::set_always_validate(bool validate)
00837 {
00838 d_always_validate = validate;
00839 }
00840
00844 bool
00845 HTTPCache::get_always_validate() const
00846 {
00847 return d_always_validate;
00848 }
00849
00866 void
00867 HTTPCache::set_cache_control(const vector<string> &cc)
00868 {
00869 lock_cache_interface();
00870
00871 try {
00872 d_cache_control = cc;
00873
00874 vector<string>::const_iterator i;
00875 for (i = cc.begin(); i != cc.end(); ++i) {
00876 string header = (*i).substr(0, (*i).find(':'));
00877 string value = (*i).substr((*i).find(": ") + 2);
00878 if (header != "Cache-Control") {
00879 throw InternalErr(__FILE__, __LINE__, "Expected cache control header not found.");
00880 }
00881 else {
00882 if (value == "no-cache" || value == "no-store")
00883 d_cache_enabled = false;
00884 else if (value.find("max-age") != string::npos) {
00885 string max_age = value.substr(value.find("=" + 1));
00886 d_max_age = parse_time(max_age.c_str());
00887 }
00888 else if (value == "max-stale")
00889 d_max_stale = 0;
00890 else if (value.find("max-stale") != string::npos) {
00891 string max_stale = value.substr(value.find("=" + 1));
00892 d_max_stale = parse_time(max_stale.c_str());
00893 }
00894 else if (value.find("min-fresh") != string::npos) {
00895 string min_fresh = value.substr(value.find("=" + 1));
00896 d_min_fresh = parse_time(min_fresh.c_str());
00897 }
00898 }
00899 }
00900 }
00901 catch (...) {
00902 unlock_cache_interface();
00903 throw;
00904 }
00905
00906 unlock_cache_interface();
00907 }
00908
00909
00914 vector<string>
00915 HTTPCache::get_cache_control()
00916 {
00917 return d_cache_control;
00918 }
00919
00921
00930 bool
00931 HTTPCache::is_url_in_cache(const string &url)
00932 {
00933 DBG(cerr << "Is this url in the cache? (" << url << ")" << endl);
00934
00935 HTTPCacheTable::CacheEntry *entry = d_http_cache_table->get_locked_entry_from_cache_table(url);
00936 bool status = entry != 0;
00937 if (entry) {
00938 entry->unlock_read_response();
00939 }
00940 return status;
00941 }
00942
00948 bool
00949 is_hop_by_hop_header(const string &header)
00950 {
00951 return header.find("Connection") != string::npos
00952 || header.find("Keep-Alive") != string::npos
00953 || header.find("Proxy-Authenticate") != string::npos
00954 || header.find("Proxy-Authorization") != string::npos
00955 || header.find("Transfer-Encoding") != string::npos
00956 || header.find("Upgrade") != string::npos;
00957 }
00958
00970 void
00971 HTTPCache::write_metadata(const string &cachename, const vector<string> &headers)
00972 {
00973 string fname = cachename + CACHE_META;
00974 d_open_files.push_back(fname);
00975
00976 FILE *dest = fopen(fname.c_str(), "w");
00977 if (!dest) {
00978 throw InternalErr(__FILE__, __LINE__,
00979 "Could not open named cache entry file.");
00980 }
00981
00982 vector<string>::const_iterator i;
00983 for (i = headers.begin(); i != headers.end(); ++i) {
00984 if (!is_hop_by_hop_header(*i)) {
00985 fwrite((*i).c_str(), (*i).size(), 1, dest);
00986 fwrite("\n", 1, 1, dest);
00987 }
00988 }
00989
00990 int res = fclose(dest);
00991 if (res) {
00992 DBG(cerr << "HTTPCache::write_metadata - Failed to close "
00993 << dest << endl);
00994 }
00995
00996 d_open_files.pop_back();
00997 }
00998
01009 void
01010 HTTPCache::read_metadata(const string &cachename, vector<string> &headers)
01011 {
01012 FILE *md = fopen(string(cachename + CACHE_META).c_str(), "r");
01013 if (!md) {
01014 throw InternalErr(__FILE__, __LINE__,
01015 "Could not open named cache entry meta data file.");
01016 }
01017
01018 char line[1024];
01019 while (!feof(md) && fgets(line, 1024, md)) {
01020 line[min(1024, static_cast<int>(strlen(line)))-1] = '\0';
01021 headers.push_back(string(line));
01022 }
01023
01024 int res = fclose(md);
01025 if (res) {
01026 DBG(cerr << "HTTPCache::read_metadata - Failed to close "
01027 << md << endl);
01028 }
01029 }
01030
01052 int
01053 HTTPCache::write_body(const string &cachename, const FILE *src)
01054 {
01055 d_open_files.push_back(cachename);
01056
01057 FILE *dest = fopen(cachename.c_str(), "wb");
01058 if (!dest) {
01059 throw InternalErr(__FILE__, __LINE__,
01060 "Could not open named cache entry file.");
01061 }
01062
01063
01064
01065 char line[1024];
01066 size_t n;
01067 int total = 0;
01068 while ((n = fread(line, 1, 1024, const_cast<FILE *>(src))) > 0) {
01069 total += fwrite(line, 1, n, dest);
01070 DBG2(sleep(3));
01071 }
01072
01073 if (ferror(const_cast<FILE *>(src)) || ferror(dest)) {
01074 int res = fclose(dest);
01075 res = res & unlink(cachename.c_str());
01076 if (res) {
01077 DBG(cerr << "HTTPCache::write_body - Failed to close/unlink "
01078 << dest << endl);
01079 }
01080 throw InternalErr(__FILE__, __LINE__,
01081 "I/O error transferring data to the cache.");
01082 }
01083
01084 rewind(const_cast<FILE *>(src));
01085
01086 int res = fclose(dest);
01087 if (res) {
01088 DBG(cerr << "HTTPCache::write_body - Failed to close "
01089 << dest << endl);
01090 }
01091
01092 d_open_files.pop_back();
01093
01094 return total;
01095 }
01096
01105 FILE *
01106 HTTPCache::open_body(const string &cachename)
01107 {
01108 FILE *src = fopen(cachename.c_str(), "rb");
01109 if (!src)
01110 throw InternalErr(__FILE__, __LINE__, "Could not open cache file.");
01111
01112 return src;
01113 }
01114
01140 bool
01141 HTTPCache::cache_response(const string &url, time_t request_time,
01142 const vector<string> &headers, const FILE *body)
01143 {
01144 lock_cache_interface();
01145
01146 DBG(cerr << "Caching url: " << url << "." << endl);
01147
01148 try {
01149
01150 if (url.find("http:") == string::npos &&
01151 url.find("https:") == string::npos) {
01152 unlock_cache_interface();
01153 return false;
01154 }
01155
01156
01157
01158
01159 d_http_cache_table->remove_entry_from_cache_table(url);
01160
01161 HTTPCacheTable::CacheEntry *entry = new HTTPCacheTable::CacheEntry(url);
01162 entry->lock_write_response();
01163
01164 try {
01165 d_http_cache_table->parse_headers(entry, d_max_entry_size, headers);
01166 if (entry->is_no_cache()) {
01167 DBG(cerr << "Not cache-able; deleting HTTPCacheTable::CacheEntry: " << entry
01168 << "(" << url << ")" << endl);
01169 entry->unlock_write_response();
01170 delete entry; entry = 0;
01171 unlock_cache_interface();
01172 return false;
01173 }
01174
01175
01176 d_http_cache_table->calculate_time(entry, d_default_expiration, request_time);
01177
01178 d_http_cache_table->create_location(entry);
01179
01180 entry->set_size(write_body(entry->get_cachename(), body));
01181 write_metadata(entry->get_cachename(), headers);
01182 d_http_cache_table->add_entry_to_cache_table(entry);
01183 entry->unlock_write_response();
01184 }
01185 catch (ResponseTooBigErr &e) {
01186
01187 DBG(cerr << e.get_error_message() << endl);
01188 REMOVE(entry->get_cachename().c_str());
01189 REMOVE(string(entry->get_cachename() + CACHE_META).c_str());
01190 DBG(cerr << "Too big; deleting HTTPCacheTable::CacheEntry: " << entry << "(" << url
01191 << ")" << endl);
01192 entry->unlock_write_response();
01193 delete entry; entry = 0;
01194 unlock_cache_interface();
01195 return false;
01196 }
01197
01198 if (d_http_cache_table->get_new_entries() > DUMP_FREQUENCY) {
01199 if (startGC())
01200 perform_garbage_collection();
01201
01202 d_http_cache_table->cache_index_write();
01203 }
01204 }
01205 catch (...) {
01206 unlock_cache_interface();
01207 throw;
01208 }
01209
01210 unlock_cache_interface();
01211
01212 return true;
01213 }
01214
01233 vector<string>
01234 HTTPCache::get_conditional_request_headers(const string &url)
01235 {
01236 lock_cache_interface();
01237
01238 HTTPCacheTable::CacheEntry *entry = 0;
01239 vector<string> headers;
01240
01241 DBG(cerr << "Getting conditional request headers for " << url << endl);
01242
01243 try {
01244 entry = d_http_cache_table->get_locked_entry_from_cache_table(url);
01245 if (!entry)
01246 throw Error("There is no cache entry for the URL: " + url);
01247
01248 if (entry->get_etag() != "")
01249 headers.push_back(string("If-None-Match: ") + entry->get_etag());
01250
01251 if (entry->get_lm() > 0) {
01252 time_t lm = entry->get_lm();
01253 headers.push_back(string("If-Modified-Since: ")
01254 + date_time_str(&lm));
01255 }
01256 else if (entry->get_max_age() > 0) {
01257 time_t max_age = entry->get_max_age();
01258 headers.push_back(string("If-Modified-Since: ")
01259 + date_time_str(&max_age));
01260 }
01261 else if (entry->get_expires() > 0) {
01262 time_t expires = entry->get_expires();
01263 headers.push_back(string("If-Modified-Since: ")
01264 + date_time_str(&expires));
01265 }
01266 entry->unlock_read_response();
01267 unlock_cache_interface();
01268 }
01269 catch (...) {
01270 unlock_cache_interface();
01271 if (entry) {
01272 entry->unlock_read_response();
01273 }
01274 throw;
01275 }
01276
01277 return headers;
01278 }
01279
01283 struct HeaderLess: binary_function<const string&, const string&, bool>
01284 {
01285 bool operator()(const string &s1, const string &s2) const {
01286 return s1.substr(0, s1.find(':')) < s2.substr(0, s2.find(':'));
01287 }
01288 };
01289
01303 void
01304 HTTPCache::update_response(const string &url, time_t request_time,
01305 const vector<string> &headers)
01306 {
01307 lock_cache_interface();
01308
01309 HTTPCacheTable::CacheEntry *entry = 0;
01310 DBG(cerr << "Updating the response headers for: " << url << endl);
01311
01312 try {
01313 entry = d_http_cache_table->get_write_locked_entry_from_cache_table(url);
01314 if (!entry)
01315 throw Error("There is no cache entry for the URL: " + url);
01316
01317
01318 d_http_cache_table->parse_headers(entry, d_max_entry_size, headers);
01319
01320
01321 d_http_cache_table->calculate_time(entry, d_default_expiration, request_time);
01322
01323
01324
01325
01326
01327
01328
01329 set<string, HeaderLess> merged_headers;
01330
01331
01332 copy(headers.begin(), headers.end(),
01333 inserter(merged_headers, merged_headers.begin()));
01334
01335
01336 vector<string> old_headers;
01337 read_metadata(entry->get_cachename(), old_headers);
01338 copy(old_headers.begin(), old_headers.end(),
01339 inserter(merged_headers, merged_headers.begin()));
01340
01341
01342
01343
01344 vector<string> result;
01345 copy(merged_headers.rbegin(), merged_headers.rend(),
01346 back_inserter(result));
01347
01348 write_metadata(entry->get_cachename(), result);
01349 entry->unlock_write_response();
01350 unlock_cache_interface();
01351 }
01352 catch (...) {
01353 if (entry) {
01354 entry->unlock_read_response();
01355 }
01356 unlock_cache_interface();
01357 throw;
01358 }
01359 }
01360
01372 bool
01373 HTTPCache::is_url_valid(const string &url)
01374 {
01375 lock_cache_interface();
01376
01377 bool freshness;
01378 HTTPCacheTable::CacheEntry *entry = 0;
01379
01380 DBG(cerr << "Is this URL valid? (" << url << ")" << endl);
01381
01382 try {
01383 if (d_always_validate) {
01384 unlock_cache_interface();
01385 return false;
01386 }
01387
01388 entry = d_http_cache_table->get_locked_entry_from_cache_table(url);
01389 if (!entry)
01390 throw Error("There is no cache entry for the URL: " + url);
01391
01392
01393
01394
01395
01396
01397 if (entry->get_must_revalidate()) {
01398 entry->unlock_read_response();
01399 unlock_cache_interface();
01400 return false;
01401 }
01402
01403 time_t resident_time = time(NULL) - entry->get_response_time();
01404 time_t current_age = entry->get_corrected_initial_age() + resident_time;
01405
01406
01407
01408 if (d_max_age >= 0 && current_age > d_max_age) {
01409 DBG(cerr << "Cache....... Max-age validation" << endl);
01410 entry->unlock_read_response();
01411 unlock_cache_interface();
01412 return false;
01413 }
01414 if (d_min_fresh >= 0
01415 && entry->get_freshness_lifetime() < current_age + d_min_fresh) {
01416 DBG(cerr << "Cache....... Min-fresh validation" << endl);
01417 entry->unlock_read_response();
01418 unlock_cache_interface();
01419 return false;
01420 }
01421
01422 freshness = (entry->get_freshness_lifetime()
01423 + (d_max_stale >= 0 ? d_max_stale : 0) > current_age);
01424 entry->unlock_read_response();
01425 unlock_cache_interface();
01426 }
01427 catch (...) {
01428 if (entry) {
01429 entry->unlock_read_response();
01430 }
01431 unlock_cache_interface();
01432 throw;
01433 }
01434
01435 return freshness;
01436 }
01437
01465 FILE * HTTPCache::get_cached_response(const string &url,
01466 vector<string> &headers, string &cacheName) {
01467 lock_cache_interface();
01468
01469 FILE *body;
01470 HTTPCacheTable::CacheEntry *entry = 0;
01471
01472 DBG(cerr << "Getting the cached response for " << url << endl);
01473
01474 try {
01475 entry = d_http_cache_table->get_locked_entry_from_cache_table(url);
01476 if (!entry) {
01477 unlock_cache_interface();
01478 return 0;
01479 }
01480
01481 cacheName = entry->get_cachename();
01482 read_metadata(entry->get_cachename(), headers);
01483
01484 DBG(cerr << "Headers just read from cache: " << endl);
01485 DBGN(copy(headers.begin(), headers.end(), ostream_iterator<string>(cerr, "\n")));
01486
01487 body = open_body(entry->get_cachename());
01488
01489 DBG(cerr << "Returning: " << url << " from the cache." << endl);
01490
01491 d_http_cache_table->bind_entry_to_data(entry, body);
01492 }
01493 catch (...) {
01494 if (entry)
01495 unlock_cache_interface();
01496 throw;
01497 }
01498
01499 unlock_cache_interface();
01500
01501 return body;
01502 }
01514 FILE *
01515 HTTPCache::get_cached_response(const string &url, vector<string> &headers)
01516 {
01517 string discard_name;
01518 return get_cached_response(url, headers, discard_name);
01519 }
01520
01531 FILE *
01532 HTTPCache::get_cached_response(const string &url)
01533 {
01534 string discard_name;
01535 vector<string> discard_headers;
01536 return get_cached_response(url, discard_headers, discard_name);
01537 }
01538
01551 void
01552 HTTPCache::release_cached_response(FILE *body)
01553 {
01554 lock_cache_interface();
01555
01556 try {
01557 d_http_cache_table->uncouple_entry_from_data(body);
01558 }
01559 catch (...) {
01560 unlock_cache_interface();
01561 throw;
01562 }
01563
01564 unlock_cache_interface();
01565 }
01566
01579 void
01580 HTTPCache::purge_cache()
01581 {
01582 lock_cache_interface();
01583
01584 try {
01585 if (d_http_cache_table->is_locked_read_responses())
01586 throw Error("Attempt to purge the cache with entries in use.");
01587
01588 d_http_cache_table->delete_all_entries();
01589 }
01590 catch (...) {
01591 unlock_cache_interface();
01592 throw;
01593 }
01594
01595 unlock_cache_interface();
01596 }
01597
01598 }