Annotation of parser3/src/main/pa_common.C, revision 1.183
1.15 paf 1: /** @file
1.16 paf 2: Parser: commonly functions.
3:
1.174 paf 4: Copyright(c) 2001-2004 ArtLebedev Group (http://www.artlebedev.com)
1.101 paf 5: Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru)
1.111 paf 6: */
1.16 paf 7:
1.183 ! paf 8: static const char * const IDENT_COMMON_C="$Date: 2004/04/05 09:23:31 $";
1.1 paf 9:
10: #include "pa_common.h"
1.4 paf 11: #include "pa_exception.h"
1.154 paf 12: #include "pa_hash.h"
1.14 paf 13: #include "pa_globals.h"
1.154 paf 14: #include "pa_request_charsets.h"
15: #include "pa_charsets.h"
16:
17: #define PA_HTTP
18:
19: #ifdef PA_HTTP
1.126 paf 20: #include "pa_vstring.h"
1.154 paf 21: #include "pa_vint.h"
1.155 paf 22: #include "pa_vhash.h"
23: #include "pa_vtable.h"
1.154 paf 24:
25: #ifdef CYGWIN
26: #define _GNU_H_WINDOWS32_SOCKETS
27: // for PASCAL
28: #include <windows.h>
29: // SOCKET
30: typedef u_int SOCKET;
31: int PASCAL closesocket(SOCKET);
32: #else
33: # if defined(WIN32)
34: # include <windows.h>
35: # else
36: # define closesocket close
37: # endif
38: #endif
1.1 paf 39:
1.126 paf 40: #else
1.154 paf 41:
42: # if defined(WIN32)
43: # include <windows.h>
44: # endif
45:
1.98 paf 46: #endif
47:
1.93 paf 48: // some maybe-undefined constants
49:
1.82 paf 50: #ifndef _O_TEXT
51: # define _O_TEXT 0
52: #endif
53: #ifndef _O_BINARY
54: # define _O_BINARY 0
1.47 paf 55: #endif
1.80 paf 56:
1.138 paf 57: #ifdef HAVE_FTRUNCATE
58: # define PA_O_TRUNC 0
59: #else
60: # ifdef _O_TRUNC
61: # define PA_O_TRUNC _O_TRUNC
62: # else
63: # error you must have either ftruncate function or _O_TRUNC bit declared
64: # endif
1.154 paf 65: #endif
1.176 paf 66:
67: # ifndef INADDR_NONE
68: # define INADDR_NONE ((ulong) -1)
69: # endif
1.154 paf 70:
71: // defines for globals
72:
73: #define FILE_STATUS_NAME "status"
74:
75: // globals
76:
77: const String file_status_name(FILE_STATUS_NAME);
78:
1.178 paf 79: // defines
1.154 paf 80:
81: #define HTTP_METHOD_NAME "method"
1.180 paf 82: #define HTTP_FORM_NAME "form"
83: #define HTTP_BODY_NAME "body"
1.154 paf 84: #define HTTP_TIMEOUT_NAME "timeout"
85: #define HTTP_HEADERS_NAME "headers"
86: #define HTTP_ANY_STATUS_NAME "any-status"
87: #define HTTP_CHARSET_NAME "charset"
1.155 paf 88: #define HTTP_TABLES_NAME "tables"
1.178 paf 89: #define HTTP_USER "user"
90: #define HTTP_PASSWORD "password"
1.154 paf 91:
92: // defines
93:
1.127 paf 94: #define DEFAULT_USER_AGENT "parser3"
95:
1.154 paf 96: // functions
1.127 paf 97:
1.154 paf 98: void fix_line_breaks(char *str, size_t& length) {
1.87 paf 99: //_asm int 3;
1.154 paf 100: const char* const eob=str+length;
101: char* dest=str;
1.72 parser 102: // fix DOS: \r\n -> \n
103: // fix Macintosh: \r -> \n
1.154 paf 104: char* bol=str;
1.137 paf 105: while(char* eol=(char*)memchr(bol, '\r', eob -bol)) {
1.72 parser 106: size_t len=eol-bol;
107: if(dest!=bol)
1.126 paf 108: memcpy(dest, bol, len);
1.72 parser 109: dest+=len;
1.126 paf 110: *dest++='\n';
1.72 parser 111:
1.126 paf 112: if(&eol[1]<eob && eol[1]=='\n') { // \r, \n = DOS
1.72 parser 113: bol=eol+2;
1.154 paf 114: length--;
1.126 paf 115: } else // \r, not \n = Macintosh
1.72 parser 116: bol=eol+1;
117: }
1.154 paf 118: // last piece without \r
1.72 parser 119: if(dest!=bol)
1.126 paf 120: memcpy(dest, bol, eob-bol);
1.154 paf 121: str[length]=0; // terminating
1.72 parser 122: }
1.18 paf 123:
1.154 paf 124: char* file_read_text(Request_charsets& charsets,
125: const String& file_spec,
126: bool fail_on_read_problem,
127: HashStringValue* params/*, HashStringValue* * out_fields*/) {
128: File_read_result file=
129: file_read(charsets, file_spec, true, params, fail_on_read_problem);
130: return file.success?file.str:0;
1.126 paf 131: }
132:
1.178 paf 133: //http request stuff
1.154 paf 134: #ifdef PA_HTTP
135:
136: #undef CRLF
137: #define CRLF "\r\n"
1.126 paf 138:
139: static bool set_addr(struct sockaddr_in *addr, const char* host, const short port){
140: memset(addr, 0, sizeof(*addr));
141: addr->sin_family=AF_INET;
142: addr->sin_port=htons(port);
143: if(host) {
1.175 paf 144: ulong packed_ip=inet_addr(host);
145: struct hostent *hostIP = packed_ip==INADDR_NONE?
146: gethostbyname(host)
147: : gethostbyaddr((char *)&packed_ip, sizeof(packed_ip), AF_INET);
148: if(hostIP)
1.126 paf 149: memcpy(&addr->sin_addr, hostIP->h_addr, hostIP->h_length);
150: else
151: return false;
152: } else
153: addr->sin_addr.s_addr=INADDR_ANY;
154: return true;
155: }
156:
1.171 paf 157: static int http_read_response(char*& response, size_t& response_size, int sock, bool fail_on_status_ne_200){
158: response=(char*)pa_malloc_atomic(1/*terminator*/); // setting memory block type
159: response[(response_size=0)]=0;
1.154 paf 160: int result=0;
1.171 paf 161: char* EOLat=0;
1.126 paf 162: while(true) {
1.171 paf 163: char buf[MAX_STRING*10];
164: ssize_t received_size=recv(sock, buf, sizeof(buf), 0);
165: if(received_size<=0)
1.126 paf 166: break;
1.171 paf 167: response=(char*)pa_realloc(response, response_size+received_size+1/*terminator*/);
168: memcpy(response+response_size, buf, received_size);
169: response_size+=received_size;
170: response[response_size]=0;
171:
172: if(!result && (EOLat=strstr(response, CRLF))) { // checking status in first response
173: const String status_line(pa_strdup(response, EOLat-response));
1.154 paf 174: ArrayString astatus;
175: size_t pos_after=0;
176: status_line.split(astatus, pos_after, " ");
1.182 paf 177: const String& status_code=*astatus.get(astatus.count()>1?1:0);
1.154 paf 178: result=status_code.as_int();
1.142 paf 179:
1.154 paf 180: if(fail_on_status_ne_200 && result!=200)
1.142 paf 181: throw Exception("http.status",
1.154 paf 182: &status_code,
1.142 paf 183: "invalid HTTP response status");
184: }
185: }
1.154 paf 186: if(result)
187: return result;
1.142 paf 188: else
189: throw Exception("http.response",
190: 0,
1.173 paf 191: "bad response from host - no status found (size=%u)", response_size);
1.126 paf 192: }
193:
194: /* ********************** request *************************** */
195:
196: #if defined(SIGALRM) && defined(HAVE_SIGSETJMP) && defined(HAVE_SIGLONGJMP)
1.145 paf 197: # define PA_USE_ALARM
1.126 paf 198: #endif
199:
1.145 paf 200: #ifdef PA_USE_ALARM
1.126 paf 201: static sigjmp_buf timeout_env;
202: static void timeout_handler(int sig){
203: siglongjmp(timeout_env, 1);
204: }
205: #endif
206:
1.171 paf 207: static int http_request(char*& response, size_t& response_size,
1.152 paf 208: const char* host, int port,
209: const char* request,
1.166 paf 210: int
211: #ifdef PA_USE_ALARM
212: timeout
213: #endif
214: ,
1.152 paf 215: bool fail_on_status_ne_200) {
1.126 paf 216: if(!host)
217: throw Exception("http.host",
1.154 paf 218: 0,
1.126 paf 219: "zero hostname"); //never
220:
1.145 paf 221: #ifdef PA_USE_ALARM
1.146 paf 222: signal(SIGALRM, timeout_handler);
1.126 paf 223: #endif
224: int sock=-1;
1.145 paf 225: #ifdef PA_USE_ALARM
226: if(sigsetjmp(timeout_env, 1)) {
227: // stupid gcc [2.95.4] generated bad code
228: // which failed to handle sigsetjmp+throw: crashed inside of pre-throw code.
229: // rewritten simplier [though duplicating closesocket code]
230: if(sock>=0)
231: closesocket(sock);
232: throw Exception("http.timeout",
233: origin_string,
234: "timeout occured while retrieving document");
1.146 paf 235: return 0; // never
1.145 paf 236: } else {
237: alarm(timeout);
238: #endif
1.146 paf 239: try {
240: int result;
1.126 paf 241: struct sockaddr_in dest;
1.154 paf 242:
243: if(!set_addr(&dest, host, port))
1.126 paf 244: throw Exception("http.host",
1.154 paf 245: 0,
1.127 paf 246: "can not resolve hostname \"%s\"", host);
1.126 paf 247:
248: if((sock=socket(AF_INET, SOCK_STREAM, IPPROTO_TCP/*0*/))<0)
249: throw Exception("http.connect",
1.154 paf 250: 0,
1.127 paf 251: "can not make socket: %s (%d)", strerror(errno), errno);
1.126 paf 252: if(connect(sock, (struct sockaddr *)&dest, sizeof(dest)))
253: throw Exception("http.connect",
1.154 paf 254: 0,
1.127 paf 255: "can not connect to host \"%s\": %s (%d)", host, strerror(errno), errno);
1.126 paf 256: size_t request_size=strlen(request);
257: if(send(sock, request, request_size, 0)!=(ssize_t)request_size)
258: throw Exception("http.connect",
1.154 paf 259: 0,
1.127 paf 260: "error sending request: %s (%d)", strerror(errno), errno);
1.126 paf 261:
1.171 paf 262: result=http_read_response(response, response_size, sock, fail_on_status_ne_200);
1.142 paf 263: closesocket(sock);
1.145 paf 264: #ifdef PA_USE_ALARM
1.142 paf 265: alarm(0);
1.126 paf 266: #endif
1.147 paf 267: return result;
1.146 paf 268: } catch(...) {
1.145 paf 269: #ifdef PA_USE_ALARM
1.146 paf 270: alarm(0);
1.126 paf 271: #endif
1.146 paf 272: if(sock>=0)
273: closesocket(sock);
1.154 paf 274: rethrow;
1.146 paf 275: }
1.148 paf 276: #ifdef PA_USE_ALARM
1.126 paf 277: }
1.148 paf 278: #endif
1.126 paf 279: }
280:
1.127 paf 281: #ifndef DOXYGEN
282: struct Http_pass_header_info {
1.154 paf 283: Request_charsets* charsets;
1.127 paf 284: String* request;
285: bool user_agent_specified;
286: };
287: #endif
1.154 paf 288: static void http_pass_header(HashStringValue::key_type key,
289: HashStringValue::value_type value,
290: Http_pass_header_info *info) {
291: *info->request <<key<<": "
292: << attributed_meaning_to_string(*value, String::L_HTTP_HEADER, false)
293: << CRLF;
1.135 paf 294:
1.154 paf 295: if(String(key, String::L_TAINTED).change_case(info->charsets->source(), String::CC_UPPER)=="USER-AGENT")
296: info->user_agent_specified=true;
1.126 paf 297: }
1.154 paf 298:
299:
300: static Charset* detect_charset(Charset& source_charset, const String& content_type_value) {
1.156 paf 301: const String::Body CONTENT_TYPE_VALUE=
1.154 paf 302: content_type_value.change_case(source_charset, String::CC_UPPER);
303: // content-type: xxx/xxx; source_charset=WE-NEED-THIS
304: // content-type: xxx/xxx; source_charset="WE-NEED-THIS"
305: // content-type: xxx/xxx; source_charset="WE-NEED-THIS";
306: size_t before_charseteq_pos=CONTENT_TYPE_VALUE.pos("CHARSET=");
307: if(before_charseteq_pos!=STRING_NOT_FOUND) {
308: size_t charset_begin=before_charseteq_pos+8/*CHARSET="*/;
309: size_t open_quote_pos=CONTENT_TYPE_VALUE.pos('"', charset_begin);
310: bool quoted=open_quote_pos==charset_begin;
311: if(quoted)
312: charset_begin++; // skip opening '"'
313: size_t charset_end=CONTENT_TYPE_VALUE.length();
314: if(quoted) {
315: size_t close_quote_pos=CONTENT_TYPE_VALUE.pos('"', charset_begin);
316: if(close_quote_pos!=STRING_NOT_FOUND)
317: charset_end=close_quote_pos;
318: } else {
319: size_t delim_pos=CONTENT_TYPE_VALUE.pos(';', charset_begin);
320: if(delim_pos!=STRING_NOT_FOUND)
321: charset_end=delim_pos;
322: }
1.156 paf 323: const String::Body CHARSET_NAME_BODY=
1.154 paf 324: CONTENT_TYPE_VALUE.mid(charset_begin, charset_end);
325:
326: return &charsets.get(CHARSET_NAME_BODY);
327: }
328:
329: return 0;
330: }
331:
1.178 paf 332: static const String* basic_authorization_field(const char* user, const char* pass) {
333: if(!user&& !pass)
334: return 0;
335:
336: String combined;
337: if(user)
338: combined<<user;
339: combined<<":";
340: if(pass)
341: combined<<pass;
342:
343: String* result=new String("Basic "); *result<<pa_base64(combined.cstr(), combined.length());
344: return result;
345: }
1.154 paf 346:
1.181 paf 347: static void form_string_value2string(
348: HashStringValue::key_type key,
349: const String& value,
350: String& result)
351: {
352: result << String(key, String::L_TAINTED) << "=";
353: result.append(value, String::L_URI, true);
354: result<< "&";
355: }
356: #ifndef DOXYGEN
357: struct Form_table_value2string_info {
358: HashStringValue::key_type key;
359: String& result;
360:
361: Form_table_value2string_info(HashStringValue::key_type akey, String& aresult):
362: key(akey), result(aresult) {}
363: };
364: #endif
365: static void form_table_value2string(Table::element_type row, Form_table_value2string_info* info) {
366: form_string_value2string(info->key, *row->get(0), info->result);
367: }
1.180 paf 368: static void form_value2string(
369: HashStringValue::key_type key,
370: HashStringValue::value_type value,
371: String* result)
372: {
1.181 paf 373: if(const String* svalue=value->get_string())
374: form_string_value2string(key, *svalue, *result);
375: else if(Table* tvalue=value->get_table()) {
376: Form_table_value2string_info info(key, *result);
377: tvalue->for_each(form_table_value2string, &info);
378: } else
379: throw Exception(0,
380: new String(key, String::L_TAINTED),
381: "is %s, "HTTP_FORM_NAME" option value must either string or table", value->type());
1.180 paf 382: }
383: static const char* form2string(HashStringValue& form) {
384: String string;
385: form.for_each(form_value2string, &string);
1.181 paf 386: return string.cstr(String::L_UNSPECIFIED);
1.180 paf 387: }
1.154 paf 388: #ifndef DOXYGEN
389: struct File_read_http_result {
390: char *str; size_t length;
391: HashStringValue* headers;
392: };
393: #endif
1.155 paf 394: /// @todo build .cookies field. use ^file.tables.SET-COOKIES.menu{ for now
1.154 paf 395: static File_read_http_result file_read_http(Request_charsets& charsets,
396: const String& file_spec,
1.169 paf 397: bool as_text,
398: HashStringValue *options=0) {
1.154 paf 399: File_read_http_result result;
1.126 paf 400: char host[MAX_STRING];
1.129 paf 401: const char* uri;
1.126 paf 402: int port;
1.180 paf 403: const char* method="GET"; bool method_is_get;
404: HashStringValue* form=0;
405: const char* body_cstr=0;
1.126 paf 406: int timeout=2;
1.142 paf 407: bool fail_on_status_ne_200=true;
1.154 paf 408: Value* vheaders=0;
409: Charset *asked_remote_charset=0;
1.178 paf 410: const char* user_cstr=0;
411: const char* password_cstr=0;
1.126 paf 412:
1.127 paf 413: if(options) {
414: int valid_options=0;
1.177 paf 415: if(Value* vmethod=options->get(HTTP_METHOD_NAME)) {
1.127 paf 416: valid_options++;
1.154 paf 417: method=vmethod->as_string().cstr();
1.181 paf 418: }
419: if(Value* vform=options->get(HTTP_FORM_NAME)) {
1.180 paf 420: valid_options++;
421: form=vform->get_hash();
1.181 paf 422: }
423: if(Value* vbody=options->get(HTTP_BODY_NAME)) {
1.180 paf 424: valid_options++;
425: body_cstr=vbody->as_string().cstr(String::L_UNSPECIFIED);
1.181 paf 426: }
427: if(Value* vtimeout=options->get(HTTP_TIMEOUT_NAME)) {
1.127 paf 428: valid_options++;
429: timeout=vtimeout->as_int();
1.181 paf 430: }
431: if((vheaders=options->get(HTTP_HEADERS_NAME))) {
1.127 paf 432: valid_options++;
1.181 paf 433: }
434: if(Value* vany_status=options->get(HTTP_ANY_STATUS_NAME)) {
1.142 paf 435: valid_options++;
436: fail_on_status_ne_200=!vany_status->as_bool();
1.181 paf 437: }
438: if(Value* vcharset_name=options->get(HTTP_CHARSET_NAME)) {
1.154 paf 439: valid_options++;
440: asked_remote_charset=&::charsets.get(vcharset_name->as_string().
441: change_case(charsets.source(), String::CC_UPPER));
1.181 paf 442: }
443: if(Value* vuser=options->get(HTTP_USER)) {
1.178 paf 444: valid_options++;
445: user_cstr=vuser->as_string().cstr();
1.181 paf 446: }
447: if(Value* vpassword=options->get(HTTP_PASSWORD)) {
1.178 paf 448: valid_options++;
449: password_cstr=vpassword->as_string().cstr();
450: }
1.142 paf 451:
1.154 paf 452: if(valid_options!=options->count())
1.127 paf 453: throw Exception("parser.runtime",
454: 0,
455: "invalid option passed");
1.154 paf 456: }
457: if(!asked_remote_charset) // defaulting to $request:charset
458: asked_remote_charset=&charsets.source();
459:
1.180 paf 460: method_is_get=strcmp(method, "GET")==0;
461: if(method_is_get && body_cstr)
462: throw Exception("parser.runtime",
463: 0,
464: "you can not use $."HTTP_BODY_NAME" option with method GET");
465:
1.154 paf 466: //preparing request
467: String& connect_string=*new String;
468: // not in ^sql{... L_SQL ...} spirit, but closer to ^file::load one
469: connect_string.append(file_spec, String::L_URI); // tainted pieces -> URI pieces
470:
1.180 paf 471: String request_head_and_body;
1.154 paf 472: {
473: // influence URLencoding of tainted pieces to String::L_URI lang
474: Temp_client_charset temp(charsets, *asked_remote_charset);
475:
476: const char* connect_string_cstr=connect_string.cstr(String::L_UNSPECIFIED);
1.126 paf 477:
1.154 paf 478: const char* current=connect_string_cstr;
479: if(strncmp(current, "http://", 7)!=0)
480: throw Exception(0,
481: &connect_string,
482: "does not start with http://"); //never
483: current+=7;
484:
485: strncpy(host, current, sizeof(host)-1); host[sizeof(host)-1]=0;
486: char* host_uri=lsplit(host, '/');
487: uri=host_uri?current+(host_uri-1-host):"/";
488: char* port_cstr=lsplit(host, ':');
489: char* error_pos=0;
490: port=port_cstr?strtol(port_cstr, &error_pos, 0):80;
491:
1.180 paf 492: if(strchr(uri, '?') && form)
493: throw Exception("parser.runtime",
494: 0,
495: "use either uri with ?params or $."HTTP_FORM_NAME" option");
496:
497: //making request head
498: String head;
499: head << method;
500: head << " " << uri;
501: if(form)
502: if(method_is_get)
1.181 paf 503: head << "?" << form2string(*form);
1.180 paf 504: head <<" HTTP/1.0" CRLF
1.154 paf 505: "host: "<< host << CRLF;
1.181 paf 506: if(form && !method_is_get) {
507: head << "content-type: application/x-www-form-urlencoded" CRLF;
508: body_cstr = form2string(*form);
509: }
1.178 paf 510:
1.179 paf 511: // http://www.ietf.org/rfc/rfc2617.txt
1.178 paf 512: if(const String* authorization_field_value=basic_authorization_field(user_cstr, password_cstr))
1.180 paf 513: head<<"authorization: "<<*authorization_field_value<<CRLF;
1.178 paf 514:
1.154 paf 515: bool user_agent_specified=false;
516: if(vheaders && !vheaders->is_string()) { // allow empty
517: if(HashStringValue *headers=vheaders->get_hash()) {
1.180 paf 518: Http_pass_header_info info={&charsets, &head, false};
1.154 paf 519: headers->for_each(http_pass_header, &info);
520: user_agent_specified=info.user_agent_specified;
521: } else
522: throw Exception("parser.runtime",
523: &connect_string,
524: "headers param must be hash");
525: };
526: if(!user_agent_specified) // defaulting
1.180 paf 527: head << "user-agent: " DEFAULT_USER_AGENT CRLF;
528:
529: if(body_cstr) {
530: // recode those pieces which are not in String::L_URI lang
531: // [those violating HTTP standard, but widly used]
532: body_cstr=Charset::transcode(
533: String::C(body_cstr, strlen(body_cstr)),
534: charsets.source(),
535: *asked_remote_charset);
1.181 paf 536:
537: head << "content-length: " << format(strlen(body_cstr), "%u") << CRLF;
1.180 paf 538: }
1.154 paf 539:
1.181 paf 540: const char* head_cstr=head.cstr(String::L_UNSPECIFIED);
541:
542: // recode those pieces which are not in String::L_URI lang
543: // [those violating HTTP standard, but widly used]
544: head_cstr=Charset::transcode(
545: String::C(head_cstr, strlen(head_cstr)),
546: charsets.source(),
547: *asked_remote_charset);
1.180 paf 548:
1.181 paf 549: // head + end of header
550: request_head_and_body << head_cstr << CRLF;
1.180 paf 551: // body
552: if(body_cstr)
553: request_head_and_body << body_cstr;
1.154 paf 554: }
1.126 paf 555:
556: //sending request
1.171 paf 557: char* response;
558: size_t response_size;
559: int status_code=http_request(response, response_size,
1.180 paf 560: host, port, request_head_and_body.cstr(),
1.142 paf 561: timeout, fail_on_status_ne_200);
1.126 paf 562:
563: //processing results
1.171 paf 564: char* raw_body; size_t raw_body_size;
565: char* headers_end_at=strstr(response, CRLF CRLF /*change '4' below along!*/);
566: if(headers_end_at) {
567: raw_body=headers_end_at+4;
568: raw_body_size=response_size-(raw_body-response);
569: } else
1.126 paf 570: throw Exception("http.response",
1.133 paf 571: &connect_string,
1.126 paf 572: "bad response from host - no headers found");
1.171 paf 573:
574: *headers_end_at=0;
575: const String header_block(response, headers_end_at-response, true);
1.126 paf 576:
1.154 paf 577: ArrayString aheaders;
578: result.headers=new HashStringValue;
1.155 paf 579: VHash* vtables=new VHash;
1.177 paf 580: result.headers->put(HTTP_TABLES_NAME, vtables);
1.155 paf 581: HashStringValue& tables=vtables->hash();
582:
1.154 paf 583: size_t pos_after=0;
584: header_block.split(aheaders, pos_after, CRLF);
1.126 paf 585:
586: //processing headers
1.154 paf 587: size_t aheaders_count=aheaders.count();
588: Charset* real_remote_charset=0; // undetected, yet
589: for(size_t i=1; i<aheaders_count; i++) {
590: const String& line=*aheaders.get(i);
1.171 paf 591: size_t pos=line.pos(": ", 2);
1.154 paf 592: if(pos==STRING_NOT_FOUND || pos<1)
1.126 paf 593: throw Exception("http.response",
1.154 paf 594: &connect_string,
595: "bad response from host - bad header \"%s\"", line.cstr());
1.156 paf 596: const String::Body HEADER_NAME=
1.154 paf 597: line.mid(0, pos).change_case(charsets.source(), String::CC_UPPER);
598: const String& header_value=line.mid(pos+2, line.length());
599: if(HEADER_NAME=="CONTENT-TYPE")
600: real_remote_charset=detect_charset(charsets.source(), header_value);
1.155 paf 601:
602: // tables
603: {
604: Value *valready=(Value *)tables.get(HEADER_NAME);
605: bool existed=valready!=0;
606: Table *table;
607: if(existed) {
608: // second+ appearence
609: table=valready->get_table();
610: } else {
611: // first appearence
612: Table::columns_type columns =new ArrayString(1);
613: *columns+=new String("value");
614: table=new Table(columns);
615: }
616: // this string becomes next row
617: ArrayString& row=*new ArrayString(1);
618: row+=&header_value;
619: *table+=&row;
620: // not existed before? add it
621: if(!existed)
622: tables.put(HEADER_NAME, new VTable(table));
623: }
624:
1.154 paf 625: result.headers->put(HEADER_NAME, new VString(header_value));
1.126 paf 626: }
1.154 paf 627: // defaulting to used-asked charset [it's never empty!]
628: if(!real_remote_charset)
629: real_remote_charset=asked_remote_charset;
1.126 paf 630:
631: // output response
1.171 paf 632: String::C real_body=String::C(raw_body, raw_body_size);
633: if(as_text && raw_body_size) // must be checked because transcode returns CONST string in case length==0, which contradicts hacking few lines below
1.169 paf 634: real_body=Charset::transcode(real_body, *real_remote_charset, charsets.source());
1.154 paf 635:
636: result.str=const_cast<char *>(real_body.str); // hacking a little
637: result.length=real_body.length;
638: result.headers->put(file_status_name, new VInt(status_code));
639: return result;
1.34 paf 640: }
1.123 paf 641:
1.154 paf 642: #endif
643:
1.123 paf 644: #ifndef DOXYGEN
645: struct File_read_action_info {
1.154 paf 646: char **data; size_t *data_size;
1.126 paf 647: };
1.123 paf 648: #endif
1.154 paf 649: static void file_read_action(
650: struct stat& finfo,
651: int f,
1.166 paf 652: const String& file_spec, const char* /*fname*/, bool as_text,
1.154 paf 653: void *context) {
1.126 paf 654: File_read_action_info& info=*static_cast<File_read_action_info *>(context);
1.123 paf 655: if(size_t to_read_size=(size_t)finfo.st_size) {
1.154 paf 656: *info.data=new(PointerFreeGC) char[to_read_size+(as_text?1:0)];
1.126 paf 657: *info.data_size=(size_t)read(f, *info.data, to_read_size);
1.123 paf 658:
659: if(ssize_t(*info.data_size)<0 || *info.data_size>to_read_size)
1.126 paf 660: throw Exception(0,
1.123 paf 661: &file_spec,
1.173 paf 662: "read failed: actually read %u bytes count not in [0..%u] valid range",
1.126 paf 663: *info.data_size, to_read_size);
1.123 paf 664: } else { // empty file
665: if(as_text) {
1.154 paf 666: *info.data=new(PointerFreeGC) char[1];
1.123 paf 667: *(char*)(*info.data)=0;
668: } else
669: *info.data=0;
670: *info.data_size=0;
671: return;
672: }
1.126 paf 673: }
1.154 paf 674: File_read_result file_read(Request_charsets& charsets, const String& file_spec,
675: bool as_text, HashStringValue *params,
1.126 paf 676: bool fail_on_read_problem) {
1.167 paf 677: File_read_result result={false, 0, 0, 0};
1.154 paf 678: #ifdef PA_HTTP
679: if(file_spec.starts_with("http://")) {
1.126 paf 680: // fail on read problem
1.169 paf 681: File_read_http_result http=file_read_http(charsets, file_spec, as_text, params);
1.154 paf 682: result.success=true;
683: result.str=http.str;
684: result.length=http.length;
685: result.headers=http.headers;
1.126 paf 686: } else {
1.154 paf 687: #endif
1.161 paf 688: if(params && params->count())
689: throw Exception("parser.runtime",
690: 0,
691: "invalid option passed");
692:
1.154 paf 693: File_read_action_info info={&result.str, &result.length};
694: result.success=file_read_action_under_lock(file_spec,
1.126 paf 695: "read", file_read_action, &info,
696: as_text, fail_on_read_problem);
1.154 paf 697: #ifdef PA_HTTP
1.126 paf 698: }
1.154 paf 699: #endif
1.123 paf 700:
1.154 paf 701: if(result.success && as_text) {
1.131 paf 702: // UTF-8 signature: EF BB BF
1.154 paf 703: if(result.length>=3) {
704: char *in=(char *)result.str;
1.159 paf 705: if(strncmp(in, "\xEF\xBB\xBF", 3)==0) {
1.154 paf 706: result.str=in+3; result.length-=3;// skip prefix
1.131 paf 707: }
708: }
709:
1.154 paf 710: fix_line_breaks((char *)(result.str), result.length);
1.123 paf 711: }
1.126 paf 712:
713: return result;
1.123 paf 714: }
715:
1.154 paf 716: #ifdef PA_SAFE_MODE
717: void check_safe_mode(struct stat finfo, const String& file_spec, const char* fname) {
718: if(finfo.st_uid/*foreign?*/!=geteuid()
719: && finfo.st_gid/*foreign?*/!=getegid())
720: throw Exception("parser.runtime",
721: &file_spec,
722: "parser is in safe mode: "
723: "reading files of foreign group and user disabled "
724: "[recompile parser with --disable-safe-mode configure option], "
725: "actual filename '%s', "
726: "fuid(%d)!=euid(%d) or fgid(%d)!=egid(%d)",
727: fname,
728: finfo.st_uid, geteuid(),
729: finfo.st_gid, getegid());
730: }
731: #endif
1.149 paf 732:
1.154 paf 733: bool file_read_action_under_lock(const String& file_spec,
1.126 paf 734: const char* action_name, File_read_action action, void *context,
735: bool as_text,
1.123 paf 736: bool fail_on_read_problem) {
1.154 paf 737: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.33 paf 738: int f;
739:
740: // first open, next stat:
1.45 paf 741: // directory update of NTFS hard links performed on open.
1.33 paf 742: // ex:
743: // a.html:^test[] and b.html hardlink to a.html
744: // user inserts ! before ^test in a.html
1.126 paf 745: // directory entry of b.html in NTFS not updated at once,
1.35 paf 746: // they delay update till open, so we would receive "!^test[" string
747: // if would do stat, next open.
1.123 paf 748: // later: it seems, even this does not help sometimes
1.98 paf 749: if((f=open(fname, O_RDONLY|(as_text?_O_TEXT:_O_BINARY)))>=0) {
1.123 paf 750: try {
1.162 paf 751: if(pa_lock_shared_blocking(f)!=0)
1.126 paf 752: throw Exception("file.lock",
1.123 paf 753: &file_spec,
754: "shared lock failed: %s (%d), actual filename '%s'",
1.154 paf 755: strerror(errno), errno, fname);
1.123 paf 756:
1.124 paf 757: struct stat finfo;
758: if(stat(fname, &finfo)!=0)
759: throw Exception("file.missing", // hardly possible: we just opened it OK
760: &file_spec,
761: "stat failed: %s (%d), actual filename '%s'",
1.154 paf 762: strerror(errno), errno, fname);
1.124 paf 763:
1.140 paf 764: #ifdef PA_SAFE_MODE
1.149 paf 765: check_safe_mode(finfo, file_spec, fname);
1.105 paf 766: #endif
1.32 paf 767:
1.154 paf 768: action(finfo, f, file_spec, fname, as_text, context);
1.123 paf 769: } catch(...) {
1.162 paf 770: pa_unlock(f);close(f);
1.123 paf 771: if(fail_on_read_problem)
1.154 paf 772: rethrow;
1.123 paf 773: return false;
774: }
1.87 paf 775:
1.162 paf 776: pa_unlock(f);close(f);
1.72 parser 777: return true;
1.118 paf 778: } else {
779: if(fail_on_read_problem)
1.126 paf 780: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.118 paf 781: &file_spec,
1.123 paf 782: "%s failed: %s (%d), actual filename '%s'",
1.154 paf 783: action_name, strerror(errno), errno, fname);
1.118 paf 784: return false;
785: }
1.8 paf 786: }
787:
1.63 parser 788: static void create_dir_for_file(const String& file_spec) {
789: size_t pos_after=1;
1.154 paf 790: size_t pos_before;
791: while((pos_before=file_spec.pos('/', pos_after))!=STRING_NOT_FOUND) {
792: mkdir(file_spec.mid(0, pos_before).cstr(String::L_FILE_SPEC), 0775);
1.63 parser 793: pos_after=pos_before+1;
794: }
795: }
796:
1.98 paf 797: bool file_write_action_under_lock(
1.28 paf 798: const String& file_spec,
1.126 paf 799: const char* action_name, File_write_action action, void *context,
800: bool as_text,
801: bool do_append,
802: bool do_block,
1.110 paf 803: bool fail_on_lock_problem) {
1.154 paf 804: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.28 paf 805: int f;
1.80 paf 806: if(access(fname, W_OK)!=0) // no
1.126 paf 807: create_dir_for_file(file_spec);
1.50 paf 808:
1.80 paf 809: if((f=open(fname,
810: O_CREAT|O_RDWR
811: |(as_text?_O_TEXT:_O_BINARY)
1.138 paf 812: |(do_append?O_APPEND:PA_O_TRUNC), 0664))>=0) {
1.162 paf 813: if((do_block?pa_lock_exclusive_blocking(f):pa_lock_exclusive_nonblocking(f))!=0) {
1.126 paf 814: Exception e("file.lock",
1.110 paf 815: &file_spec,
816: "shared lock failed: %s (%d), actual filename '%s'",
1.154 paf 817: strerror(errno), errno, fname);
1.126 paf 818: close(f);
1.110 paf 819: if(fail_on_lock_problem)
820: throw e;
1.98 paf 821: return false;
822: }
1.96 paf 823:
1.158 paf 824: try {
1.126 paf 825: action(f, context);
1.158 paf 826: } catch(...) {
1.138 paf 827: #ifdef HAVE_FTRUNCATE
1.104 paf 828: if(!do_append)
1.125 paf 829: ftruncate(f, lseek(f, 0, SEEK_CUR)); // one can not use O_TRUNC, read lower
1.138 paf 830: #endif
1.162 paf 831: pa_unlock(f);close(f);
1.154 paf 832: rethrow;
1.158 paf 833: }
1.80 paf 834:
1.138 paf 835: #ifdef HAVE_FTRUNCATE
1.104 paf 836: if(!do_append)
1.125 paf 837: ftruncate(f, lseek(f, 0, SEEK_CUR)); // O_TRUNC truncates even exclusevely write-locked file [thanks to Igor Milyakov <virtan@rotabanner.com> for discovering]
1.138 paf 838: #endif
1.162 paf 839: pa_unlock(f);close(f);
1.98 paf 840: return true;
1.80 paf 841: } else
1.126 paf 842: throw Exception(errno==EACCES?"file.access":0,
1.80 paf 843: &file_spec,
1.96 paf 844: "%s failed: %s (%d), actual filename '%s'",
1.154 paf 845: action_name, strerror(errno), errno, fname);
1.96 paf 846: // here should be nothing, see rethrow above
847: }
848:
849: #ifndef DOXYGEN
850: struct File_write_action_info {
1.154 paf 851: const char* str; size_t length;
1.126 paf 852: };
1.96 paf 853: #endif
854: static void file_write_action(int f, void *context) {
1.126 paf 855: File_write_action_info& info=*static_cast<File_write_action_info *>(context);
1.154 paf 856: if(info.length) {
857: int written=write(f, info.str, info.length);
1.116 paf 858: if(written<0)
1.126 paf 859: throw Exception(0,
860: 0,
861: "write failed: %s (%d)", strerror(errno), errno);
1.113 paf 862: }
1.96 paf 863: }
864: void file_write(
865: const String& file_spec,
1.154 paf 866: const char* data, size_t size,
1.126 paf 867: bool as_text,
1.96 paf 868: bool do_append) {
1.126 paf 869: File_write_action_info info={data, size};
1.98 paf 870: file_write_action_under_lock(
1.154 paf 871: file_spec,
872: "write", file_write_action, &info,
873: as_text,
874: do_append);
1.30 paf 875: }
876:
1.63 parser 877: // throws nothing! [this is required in file_move & file_delete]
1.50 paf 878: static void rmdir(const String& file_spec, size_t pos_after) {
1.154 paf 879: size_t pos_before;
880: if((pos_before=file_spec.pos('/', pos_after))!=STRING_NOT_FOUND)
1.126 paf 881: rmdir(file_spec, pos_before+1);
1.50 paf 882:
1.154 paf 883: rmdir(file_spec.mid(0, pos_after-1/* / */).cstr(String::L_FILE_SPEC));
1.50 paf 884: }
1.164 paf 885: bool file_delete(const String& file_spec, bool fail_on_problem) {
1.154 paf 886: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.54 parser 887: if(unlink(fname)!=0)
1.164 paf 888: if(fail_on_problem)
1.126 paf 889: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.93 paf 890: &file_spec,
891: "unlink failed: %s (%d), actual filename '%s'",
1.154 paf 892: strerror(errno), errno, fname);
1.93 paf 893: else
894: return false;
1.50 paf 895:
1.126 paf 896: rmdir(file_spec, 1);
1.93 paf 897: return true;
1.60 parser 898: }
1.95 paf 899: void file_move(const String& old_spec, const String& new_spec) {
1.154 paf 900: const char* old_spec_cstr=old_spec.cstr(String::L_FILE_SPEC);
901: const char* new_spec_cstr=new_spec.cstr(String::L_FILE_SPEC);
1.63 parser 902:
1.126 paf 903: create_dir_for_file(new_spec);
1.63 parser 904:
1.60 parser 905: if(rename(old_spec_cstr, new_spec_cstr)!=0)
1.126 paf 906: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.60 parser 907: &old_spec,
908: "rename failed: %s (%d), actual filename '%s' to '%s'",
1.154 paf 909: strerror(errno), errno, old_spec_cstr, new_spec_cstr);
1.63 parser 910:
1.126 paf 911: rmdir(old_spec, 1);
1.31 paf 912: }
913:
1.51 paf 914:
1.126 paf 915: bool entry_exists(const char* fname, struct stat *afinfo) {
1.118 paf 916: struct stat lfinfo;
917: bool result=stat(fname, &lfinfo)==0;
918: if(afinfo)
919: *afinfo=lfinfo;
920: return result;
1.119 paf 921: }
922:
923: bool entry_exists(const String& file_spec) {
1.154 paf 924: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.126 paf 925: return entry_exists(fname, 0);
1.118 paf 926: }
927:
1.51 paf 928: static bool entry_readable(const String& file_spec, bool need_dir) {
1.154 paf 929: char* fname=file_spec.cstrm(String::L_FILE_SPEC);
1.120 paf 930: if(need_dir) {
1.126 paf 931: size_t size=strlen(fname);
1.120 paf 932: while(size) {
1.126 paf 933: char c=fname[size-1];
1.120 paf 934: if(c=='/' || c=='\\')
935: fname[--size]=0;
936: else
937: break;
938: }
939: }
1.51 paf 940: struct stat finfo;
1.118 paf 941: if(access(fname, R_OK)==0 && entry_exists(fname, &finfo)) {
1.109 paf 942: bool is_dir=(finfo.st_mode&S_IFDIR) != 0;
1.51 paf 943: return is_dir==need_dir;
944: }
945: return false;
946: }
1.31 paf 947: bool file_readable(const String& file_spec) {
1.126 paf 948: return entry_readable(file_spec, false);
1.51 paf 949: }
950: bool dir_readable(const String& file_spec) {
1.126 paf 951: return entry_readable(file_spec, true);
1.65 parser 952: }
1.154 paf 953: const String* file_readable(const String& path, const String& name) {
954: String& result=*new String(path);
955: result << "/";
956: result << name;
957: return file_readable(result)?&result:0;
1.43 paf 958: }
959: bool file_executable(const String& file_spec) {
1.154 paf 960: return access(file_spec.cstr(String::L_FILE_SPEC), X_OK)==0;
1.44 paf 961: }
962:
1.64 parser 963: bool file_stat(const String& file_spec,
1.58 parser 964: size_t& rsize,
1.126 paf 965: time_t& ratime,
966: time_t& rmtime,
967: time_t& rctime,
1.64 parser 968: bool fail_on_read_problem) {
1.154 paf 969: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
970: struct stat finfo;
1.44 paf 971: if(stat(fname, &finfo)!=0)
1.64 parser 972: if(fail_on_read_problem)
1.126 paf 973: throw Exception("file.missing",
1.67 parser 974: &file_spec,
975: "getting file size failed: %s (%d), real filename '%s'",
1.154 paf 976: strerror(errno), errno, fname);
1.64 parser 977: else
978: return false;
1.58 parser 979: rsize=finfo.st_size;
980: ratime=finfo.st_atime;
981: rmtime=finfo.st_mtime;
982: rctime=finfo.st_ctime;
1.64 parser 983: return true;
1.18 paf 984: }
985:
1.126 paf 986: char* getrow(char* *row_ref, char delim) {
987: char* result=*row_ref;
1.8 paf 988: if(result) {
1.126 paf 989: *row_ref=strchr(result, delim);
1.8 paf 990: if(*row_ref)
991: *((*row_ref)++)=0;
992: else if(!*result)
993: return 0;
994: }
995: return result;
996: }
997:
1.126 paf 998: char* lsplit(char* string, char delim) {
1.23 paf 999: if(string) {
1.126 paf 1000: char* v=strchr(string, delim);
1.8 paf 1001: if(v) {
1002: *v=0;
1003: return v+1;
1004: }
1005: }
1006: return 0;
1007: }
1008:
1.126 paf 1009: char* lsplit(char* *string_ref, char delim) {
1010: char* result=*string_ref;
1011: char* next=lsplit(*string_ref, delim);
1.8 paf 1012: *string_ref=next;
1013: return result;
1.9 paf 1014: }
1015:
1.126 paf 1016: char* rsplit(char* string, char delim) {
1.18 paf 1017: if(string) {
1.126 paf 1018: char* v=strrchr(string, delim);
1.18 paf 1019: if(v) {
1.9 paf 1020: *v=0;
1021: return v+1;
1022: }
1023: }
1024: return NULL;
1.10 paf 1025: }
1026:
1.37 paf 1027: /// @todo less stupid type detection
1.154 paf 1028: const char* format(double value, char* fmt) {
1.126 paf 1029: char local_buf[MAX_NUMBER];
1.108 paf 1030: size_t size;
1031:
1.10 paf 1032: if(fmt)
1033: if(strpbrk(fmt, "diouxX"))
1034: if(strpbrk(fmt, "ouxX"))
1.126 paf 1035: size=snprintf(local_buf, sizeof(local_buf), fmt, (uint)value);
1.10 paf 1036: else
1.126 paf 1037: size=snprintf(local_buf, sizeof(local_buf), fmt, (int)value);
1.10 paf 1038: else
1.126 paf 1039: size=snprintf(local_buf, sizeof(local_buf), fmt, value);
1.10 paf 1040: else
1.126 paf 1041: size=snprintf(local_buf, sizeof(local_buf), "%d", (int)value);
1.10 paf 1042:
1.154 paf 1043: return pa_strdup(local_buf, size);
1.12 paf 1044: }
1045:
1.36 paf 1046: size_t stdout_write(const void *buf, size_t size) {
1.12 paf 1047: #ifdef WIN32
1048: do{
1.154 paf 1049: int chunk_written=fwrite(buf, 1, min((size_t)8*0x400, size), stdout);
1.12 paf 1050: if(chunk_written<=0)
1051: break;
1052: size-=chunk_written;
1.36 paf 1053: buf=((const char*)buf)+chunk_written;
1.126 paf 1054: } while(size>0);
1.12 paf 1055:
1056: return size;
1057: #else
1.126 paf 1058: return fwrite(buf, 1, size, stdout);
1.12 paf 1059: #endif
1.2 paf 1060: }
1.14 paf 1061:
1.154 paf 1062: char* unescape_chars(const char* cp, int len) {
1063: char* s=new(PointerFreeGC) char[len + 1];
1.14 paf 1064: enum EscapeState {
1.33 paf 1065: EscapeRest,
1066: EscapeFirst,
1.14 paf 1067: EscapeSecond
1068: } escapeState=EscapeRest;
1069: int escapedValue=0;
1070: int srcPos=0;
1071: int dstPos=0;
1072: while(srcPos < len) {
1.126 paf 1073: int ch=cp[srcPos];
1.14 paf 1074: switch(escapeState) {
1075: case EscapeRest:
1076: if(ch=='%') {
1077: escapeState=EscapeFirst;
1078: } else if(ch=='+') {
1.126 paf 1079: s[dstPos++]=' ';
1.14 paf 1080: } else {
1081: s[dstPos++]=ch;
1082: }
1083: break;
1084: case EscapeFirst:
1085: escapedValue=hex_value[ch] << 4;
1086: escapeState=EscapeSecond;
1087: break;
1088: case EscapeSecond:
1.126 paf 1089: escapedValue +=hex_value[ch];
1.14 paf 1090: s[dstPos++]=escapedValue;
1091: escapeState=EscapeRest;
1092: break;
1093: }
1.126 paf 1094: srcPos++;
1.14 paf 1095: }
1096: s[dstPos]=0;
1097: return s;
1.24 paf 1098: }
1099:
1100: #ifdef WIN32
1.126 paf 1101: void back_slashes_to_slashes(char* s) {
1.24 paf 1102: if(s)
1103: for(; *s; s++)
1104: if(*s=='\\')
1.126 paf 1105: *s='/';
1.24 paf 1106: }
1.42 paf 1107: /*
1.126 paf 1108: void slashes_to_back_slashes(char* s) {
1.42 paf 1109: if(s)
1110: for(; *s; s++)
1111: if(*s=='/')
1.126 paf 1112: *s='\\';
1.42 paf 1113: }
1114: */
1.24 paf 1115: #endif
1.41 paf 1116:
1.126 paf 1117: bool StrEqNc(const char* s1, const char* s2, bool strict) {
1.41 paf 1118: while(true) {
1119: if(!(*s1)) {
1120: if(!(*s2))
1121: return true;
1122: else
1123: return !strict;
1124: } else if(!(*s2))
1125: return !strict;
1126: if(isalpha(*s1)) {
1127: if(tolower(*s1) !=tolower(*s2))
1128: return false;
1129: } else if((*s1) !=(*s2))
1130: return false;
1.126 paf 1131: s1++;
1132: s2++;
1.41 paf 1133: }
1.57 parser 1134: }
1135:
1.84 paf 1136: static bool isLeap(int year) {
1.57 parser 1137: return !(
1138: (year % 4) || ((year % 400) && !(year % 100))
1.126 paf 1139: );
1.57 parser 1140: }
1141:
1142: int getMonthDays(int year, int month) {
1143: int monthDays[]={
1.126 paf 1144: 31,
1145: isLeap(year) ? 29 : 28,
1146: 31,
1147: 30,
1148: 31,
1149: 30,
1150: 31,
1151: 31,
1152: 30,
1153: 31,
1154: 30,
1.57 parser 1155: 31
1.126 paf 1156: };
1157: return monthDays[month];
1.41 paf 1158: }
1.69 parser 1159:
1.126 paf 1160: void remove_crlf(char* start, char* end) {
1161: for(char* p=start; p<end; p++)
1.69 parser 1162: switch(*p) {
1.126 paf 1163: case '\n': *p='|'; break;
1164: case '\r': *p=' '; break;
1.69 parser 1165: }
1.91 paf 1166: }
1167:
1168:
1169: /// must be last in this file
1170: #undef vsnprintf
1.126 paf 1171: int __vsnprintf(char* b, size_t s, const char* f, va_list l) {
1.91 paf 1172: if(!s)
1173: return 0;
1174:
1175: int r;
1176: // note: on win32& maybe somewhere else
1177: // vsnprintf do not writes terminating 0 in 'buffer full' case, reducing
1178: --s;
1.172 paf 1179:
1180: // clients do not check for negative 's', feature: ignore such prints
1181: if((ssize_t)s<0)
1182: return 0;
1183:
1.91 paf 1184: #if _MSC_VER
1185: /*
1186: win32:
1187: mk:@MSITStore:C:\Program%20Files\Microsoft%20Visual%20Studio\MSDN\2001APR\1033\vccore.chm::/html/_crt__vsnprintf.2c_._vsnwprintf.htm
1188:
1.154 paf 1189: if the number of bytes to write exceeds buffer, then count bytes are written and Ö1 is returned
1.91 paf 1190: */
1.126 paf 1191: r=_vsnprintf(b, s, f, l);
1.91 paf 1192: if(r<0)
1193: r=s;
1194: #else
1.126 paf 1195: r=vsnprintf(b, s, f, l);
1.91 paf 1196: /*
1197: solaris:
1198: man vsnprintf
1199:
1200: The snprintf() function returns the number of characters
1201: formatted, that is, the number of characters that would have
1202: been written to the buffer if it were large enough. If the
1203: value of n is 0 on a call to snprintf(), an unspecified
1204: value less than 1 is returned.
1205: */
1206:
1207: if(r<0)
1208: r=0;
1.167 paf 1209: else if((size_t)r>s)
1.91 paf 1210: r=s;
1211: #endif
1212: b[r]=0;
1213: return r;
1214: }
1215:
1.126 paf 1216: int __snprintf(char* b, size_t s, const char* f, ...) {
1.91 paf 1217: va_list l;
1.126 paf 1218: va_start(l, f);
1219: int r=__vsnprintf(b, s, f, l);
1220: va_end(l);
1.91 paf 1221: return r;
1.178 paf 1222: }
1223:
1224: /* mime64 functions are from libgmime[http://spruce.sourceforge.net/gmime/] lib */
1225: /*
1226: * Authors: Michael Zucchi <notzed@helixcode.com>
1227: * Jeffrey Stedfast <fejj@helixcode.com>
1228: *
1229: * Copyright 2000 Helix Code, Inc. (www.helixcode.com)
1230: *
1231: * This program is free software; you can redistribute it and/or modify
1232: * it under the terms of the GNU General Public License as published by
1233: * the Free Software Foundation; either version 2 of the License, or
1234: * (at your option) any later version.
1235: *
1236: * This program is distributed in the hope that it will be useful,
1237: * but WITHOUT ANY WARRANTY; without even the implied warranty of
1238: * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
1239: * GNU General Public License for more details.
1240: *
1241: * You should have received a copy of the GNU General Public License
1242: * along with this program; if not, write to the Free Software
1243: * Foundation, Inc., 59 Temple Street #330, Boston, MA 02111-1307, USA.
1244: *
1245: */
1246: static char *base64_alphabet =
1247: "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";
1248:
1249: /**
1250: * g_mime_utils_base64_encode_step:
1251: * @in: input stream
1252: * @inlen: length of the input
1253: * @out: output string
1254: * @state: holds the number of bits that are stored in @save
1255: * @save: leftover bits that have not yet been encoded
1256: *
1257: * Base64 encodes a chunk of data. Performs an 'encode step', only
1258: * encodes blocks of 3 characters to the output at a time, saves
1259: * left-over state in state and save (initialise to 0 on first
1260: * invocation).
1261: *
1262: * Returns the number of bytes encoded.
1263: **/
1264: static size_t
1265: g_mime_utils_base64_encode_step (const unsigned char *in, size_t inlen, unsigned char *out, int *state, int *save)
1266: {
1267: const register unsigned char *inptr;
1268: register unsigned char *outptr;
1269:
1270: if (inlen <= 0)
1271: return 0;
1272:
1273: inptr = in;
1274: outptr = out;
1275:
1276: if (inlen + ((unsigned char *)save)[0] > 2) {
1277: const unsigned char *inend = in + inlen - 2;
1278: register int c1 = 0, c2 = 0, c3 = 0;
1279: register int already;
1280:
1281: already = *state;
1282:
1283: switch (((char *)save)[0]) {
1284: case 1: c1 = ((unsigned char *)save)[1]; goto skip1;
1285: case 2: c1 = ((unsigned char *)save)[1];
1286: c2 = ((unsigned char *)save)[2]; goto skip2;
1287: }
1288:
1289: /* yes, we jump into the loop, no i'm not going to change it, its beautiful! */
1290: while (inptr < inend) {
1291: c1 = *inptr++;
1292: skip1:
1293: c2 = *inptr++;
1294: skip2:
1295: c3 = *inptr++;
1296: *outptr++ = base64_alphabet [c1 >> 2];
1297: *outptr++ = base64_alphabet [(c2 >> 4) | ((c1 & 0x3) << 4)];
1298: *outptr++ = base64_alphabet [((c2 & 0x0f) << 2) | (c3 >> 6)];
1299: *outptr++ = base64_alphabet [c3 & 0x3f];
1300: /* this is a bit ugly ... */
1301: if ((++already) >= 19) {
1302: *outptr++ = '\n';
1303: already = 0;
1304: }
1305: }
1306:
1307: ((unsigned char *)save)[0] = 0;
1308: inlen = 2 - (inptr - inend);
1309: *state = already;
1310: }
1311:
1312: //d(printf ("state = %d, inlen = %d\n", (int)((char *)save)[0], inlen));
1313:
1314: if (inlen > 0) {
1315: register char *saveout;
1316:
1317: /* points to the slot for the next char to save */
1318: saveout = & (((char *)save)[1]) + ((char *)save)[0];
1319:
1320: /* inlen can only be 0 1 or 2 */
1321: switch (inlen) {
1322: case 2: *saveout++ = *inptr++;
1323: case 1: *saveout++ = *inptr++;
1324: }
1325: ((char *)save)[0] += inlen;
1326: }
1327:
1328: /*d(printf ("mode = %d\nc1 = %c\nc2 = %c\n",
1329: (int)((char *)save)[0],
1330: (int)((char *)save)[1],
1331: (int)((char *)save)[2]));*/
1332:
1333: return (outptr - out);
1334: }
1335:
1336: /**
1337: * g_mime_utils_base64_encode_close:
1338: * @in: input stream
1339: * @inlen: length of the input
1340: * @out: output string
1341: * @state: holds the number of bits that are stored in @save
1342: * @save: leftover bits that have not yet been encoded
1343: *
1344: * Base64 encodes the input stream to the output stream. Call this
1345: * when finished encoding data with g_mime_utils_base64_encode_step to
1346: * flush off the last little bit.
1347: *
1348: * Returns the number of bytes encoded.
1349: **/
1350: static size_t
1351: g_mime_utils_base64_encode_close (const unsigned char *in, size_t inlen, unsigned char *out, int *state, int *save)
1352: {
1353: unsigned char *outptr = out;
1354: int c1, c2;
1355:
1356: if (inlen > 0)
1357: outptr += g_mime_utils_base64_encode_step (in, inlen, outptr, state, save);
1358:
1359: c1 = ((unsigned char *)save)[1];
1360: c2 = ((unsigned char *)save)[2];
1361:
1362: switch (((unsigned char *)save)[0]) {
1363: case 2:
1364: outptr[2] = base64_alphabet [(c2 & 0x0f) << 2];
1365: goto skip;
1366: case 1:
1367: outptr[2] = '=';
1368: skip:
1369: outptr[0] = base64_alphabet [c1 >> 2];
1370: outptr[1] = base64_alphabet [c2 >> 4 | ((c1 & 0x3) << 4)];
1371: outptr[3] = '=';
1372: outptr += 4;
1373: break;
1374: }
1375:
1376: *outptr++ = 0;
1377:
1378: *save = 0;
1379: *state = 0;
1380:
1381: return (outptr - out);
1382: }
1383:
1384: char* pa_base64(const char *in, size_t len)
1385: {
1386: /* wont go to more than 2x size (overly conservative) */
1387: char* result=new(PointerFreeGC) char[len * 2 + 6];
1388: int state=0;
1389: int save=0;
1.183 ! paf 1390: #ifndef NDEBUG
! 1391: size_t filled=
! 1392: #endif
! 1393: g_mime_utils_base64_encode_close ((const unsigned char*)in, len,
1.178 paf 1394: (unsigned char*)result, &state, &save);
1395: assert(filled <= len * 2 + 6);
1396:
1397: return result;
1.98 paf 1398: }
E-mail:
![[BACK]](/cvsweb/icons/back.gif){kind=icon}
Return to pa_common.C CVS log ![[TXT]](/cvsweb/icons/text.gif){kind=icon}
![[DIR]](/cvsweb/icons/dir.gif){kind=icon}
Up to [parser3project] / parser3 / src / main