summaryrefslogtreecommitdiffabout
path: root/lib/consumer.cc
Unidiff
Diffstat (limited to 'lib/consumer.cc') (more/less context) (show whitespace changes)
-rw-r--r--lib/consumer.cc124
1 files changed, 65 insertions, 59 deletions
diff --git a/lib/consumer.cc b/lib/consumer.cc
index ff5da91..df95b64 100644
--- a/lib/consumer.cc
+++ b/lib/consumer.cc
@@ -12,17 +12,55 @@
12 12
13#include "config.h" 13#include "config.h"
14 14
15#if defined(USE_LIBPCRECPP) 15#include <pcre.h>
16# include <pcrecpp.h>
17#elif defined(USE_PCREPP)
18# include <pcre++.h>
19#else
20 /* internal implementation won't be built */
21#endif
22 16
23namespace opkele { 17namespace opkele {
24 using namespace std; 18 using namespace std;
25 19
20 class pcre_matches_t {
21 public:
22 int *_ov;
23 int _s;
24
25 pcre_matches_t() : _ov(0), _s(0) { }
26 pcre_matches_t(int s) : _ov(0), _s(s) {
27 if(_s&1) ++_s;
28 _s += _s>>1;
29 _ov = new int[_s];
30 }
31 ~pcre_matches_t() throw() { if(_ov) delete[] _ov; }
32
33 int begin(int i) const { return _ov[i<<1]; }
34 int end(int i) const { return _ov[(i<<1)+1]; }
35 int length(int i) const { int t=i<<1; return _ov[t+1]-_ov[t]; }
36 };
37
38 class pcre_t {
39 public:
40 pcre *_p;
41
42 pcre_t() : _p(0) { }
43 pcre_t(pcre *p) : _p(p) { }
44 pcre_t(const char *re,int opts) : _p(0) {
45 static const char *errptr; static int erroffset;
46 _p = pcre_compile(re,opts,&errptr,&erroffset,NULL);
47 if(!_p)
48 throw internal_error(OPKELE_CP_ string("Failed to compile regexp: ")+errptr);
49 }
50 ~pcre_t() throw() { if(_p) (*pcre_free)(_p); }
51
52 pcre_t& operator=(pcre *p) { if(_p) (*pcre_free)(_p); _p=p; return *this; }
53
54 operator const pcre*(void) const { return _p; }
55 operator pcre*(void) { return _p; }
56
57 int exec(const string& s,pcre_matches_t& m) {
58 if(!_p)
59 throw internal_error(OPKELE_CP_ "Trying to execute absent regexp");
60 return pcre_exec(_p,NULL,s.c_str(),s.length(),0,0,m._ov,m._s);
61 }
62 };
63
26 class curl_t { 64 class curl_t {
27 public: 65 public:
28 CURL *_c; 66 CURL *_c;
@@ -268,7 +306,6 @@ namespace opkele {
268 } 306 }
269 307
270 void consumer_t::retrieve_links(const string& url,string& server,string& delegate) { 308 void consumer_t::retrieve_links(const string& url,string& server,string& delegate) {
271#if defined(USE_LIBPCRECPP) || defined(USE_PCREPP)
272 server.erase(); 309 server.erase();
273 delegate.erase(); 310 delegate.erase();
274 curl_t curl = curl_easy_init(); 311 curl_t curl = curl_easy_init();
@@ -286,67 +323,36 @@ namespace opkele {
286 r = curl_easy_perform(curl); 323 r = curl_easy_perform(curl);
287 if(r && r!=CURLE_WRITE_ERROR) 324 if(r && r!=CURLE_WRITE_ERROR)
288 throw exception_curl(OPKELE_CP_ "failed to curl_easy_perform()",r); 325 throw exception_curl(OPKELE_CP_ "failed to curl_easy_perform()",r);
289 // strip out everything past body 326 static const char *re_bre = "<\\s*body\\b", *re_hdre = "<\\s*head[^>]*>",
290 static const char *re_hdre = "<\\s*head[^>]*>",
291 *re_lre = "<\\s*link\\b([^>]+)>", 327 *re_lre = "<\\s*link\\b([^>]+)>",
292 *re_rre = "\\brel\\s*=\\s*['\"]\\s*([^'\"\\s]+)\\s*['\"]", 328 *re_rre = "\\brel\\s*=\\s*['\"]\\s*([^'\"\\s]+)\\s*['\"]",
293 *re_hre = "\\bhref\\s*=\\s*['\"]\\s*([^'\"\\s]+)\\s*['\"]"; 329 *re_hre = "\\bhref\\s*=\\s*['\"]\\s*([^'\"\\s]+)\\s*['\"]";
294#if defined(USE_LIBPCRECPP) 330 pcre_matches_t m1(3), m2(3);
295 static pcrecpp::RE_Options ro(PCRE_CASELESS|PCRE_DOTALL); 331 pcre_t bre(re_bre,PCRE_CASELESS);
296 static pcrecpp::RE 332 if(bre.exec(html,m1)>0)
297 bre("<body\\b.*",ro), hdre(re_hdre,ro), 333 html.erase(m1.begin(0));
298 lre(re_lre,ro), rre(re_rre), hre(re_hre,ro); 334 pcre_t hdre(re_hdre,PCRE_CASELESS);
299 bre.Replace("",&html); 335 if(hdre.exec(html,m1)<=0)
300 pcrecpp::StringPiece hpiece(html); 336 throw bad_input(OPKELE_CP_ "failed to find <head>");
301 if(!hdre.FindAndConsume(&hpiece)) 337 html.erase(0,m1.end(0)+1);
302 throw bad_input(OPKELE_CP_ "failed to find head"); 338 pcre_t lre(re_lre,PCRE_CASELESS), rre(re_rre,PCRE_CASELESS), hre(re_hre,PCRE_CASELESS);
303 string attrs; 339 while(lre.exec(html,m1)>=2) {
304 while(lre.FindAndConsume(&hpiece,&attrs)) { 340 string attrs(html,m1.begin(1),m1.length(1));
305 pcrecpp::StringPiece rel, href; 341 html.erase(0,m1.end(0)+1);
306 if(!(rre.PartialMatch(attrs,&rel) && hre.PartialMatch(attrs,&href))) 342 if(!( rre.exec(attrs,m1)>=2 && hre.exec(attrs,m2)>=2 ))
307 continue; 343 continue;
344 string rel(attrs,m1.begin(1),m1.length(1));
345 string href(attrs,m2.begin(1),m2.length(1));
308 if(rel=="openid.server") { 346 if(rel=="openid.server") {
309 href.CopyToString(&server); 347 server = href;
310 if(!delegate.empty()) 348 if(!delegate.empty()) break;
311 break;
312 }else if(rel=="openid.delegate") { 349 }else if(rel=="openid.delegate") {
313 href.CopyToString(&delegate); 350 delegate = href;
314 if(!server.empty()) 351 if(!server.empty()) break;
315 break;
316 }
317 } 352 }
318#elif defined(USE_PCREPP)
319 pcrepp::Pcre bre("<body\\b",PCRE_CASELESS);
320 if(bre.search(html))
321 html.erase(bre.get_match_start());
322 pcrepp::Pcre hdre(re_hdre,PCRE_CASELESS);
323 if(!hdre.search(html))
324 throw bad_input(OPKELE_CP_ "failed to find head");
325 html.erase(0,hdre.get_match_end()+1);
326 pcrepp::Pcre lre(re_lre,PCRE_CASELESS), rre(re_rre,PCRE_CASELESS), hre(re_hre,PCRE_CASELESS);
327 while(lre.search(html)) {
328 string attrs = lre[0];
329 html.erase(0,lre.get_match_end()+1);
330 if(!(rre.search(attrs)&&hre.search(attrs)))
331 continue;
332 if(rre[0]=="openid.server") {
333 server = hre[0];
334 if(!delegate.empty())
335 break;
336 }else if(rre[0]=="openid.delegate") {
337 delegate = hre[0];
338 if(!server.empty())
339 break;
340 } 353 }
341 }
342#else
343 #error "I must have gone crazy"
344#endif
345 if(server.empty()) 354 if(server.empty())
346 throw failed_assertion(OPKELE_CP_ "The location has no openid.server declaration"); 355 throw failed_assertion(OPKELE_CP_ "The location has no openid.server declaration");
347#else /* none of the RE bindings enabled */
348 throw not_implemented(OPKELE_CP_ "No internal implementation of retrieve_links were provided at compile-time");
349#endif
350 } 356 }
351 357
352 assoc_t consumer_t::find_assoc(const string& server) { 358 assoc_t consumer_t::find_assoc(const string& server) {