Annotation of parser3/src/main/pa_common.C, revision 1.180
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:
1.180 ! paf 8: static const char * const IDENT_COMMON_C="$Date: 2004/03/05 13:17:59 $";
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, " ");
177: const String& status_code=*astatus.get(1);
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:
1.180 ! paf 347: /// @todo table value
! 348: static void form_value2string(
! 349: HashStringValue::key_type key,
! 350: HashStringValue::value_type value,
! 351: String* result)
! 352: {
! 353: *result << String(key, String::L_TAINTED) << "=";
! 354: result->append(value->as_string(), String::L_URI, true);
! 355: }
! 356: static const char* form2string(HashStringValue& form) {
! 357: String string;
! 358: form.for_each(form_value2string, &string);
! 359: return 0; // todo
! 360: }
1.154 paf 361: #ifndef DOXYGEN
362: struct File_read_http_result {
363: char *str; size_t length;
364: HashStringValue* headers;
365: };
366: #endif
1.155 paf 367: /// @todo build .cookies field. use ^file.tables.SET-COOKIES.menu{ for now
1.154 paf 368: static File_read_http_result file_read_http(Request_charsets& charsets,
369: const String& file_spec,
1.169 paf 370: bool as_text,
371: HashStringValue *options=0) {
1.154 paf 372: File_read_http_result result;
1.126 paf 373: char host[MAX_STRING];
1.129 paf 374: const char* uri;
1.126 paf 375: int port;
1.180 ! paf 376: const char* method="GET"; bool method_is_get;
! 377: HashStringValue* form=0;
! 378: const char* body_cstr=0;
1.126 paf 379: int timeout=2;
1.142 paf 380: bool fail_on_status_ne_200=true;
1.154 paf 381: Value* vheaders=0;
382: Charset *asked_remote_charset=0;
1.178 paf 383: const char* user_cstr=0;
384: const char* password_cstr=0;
1.126 paf 385:
1.127 paf 386: if(options) {
387: int valid_options=0;
1.177 paf 388: if(Value* vmethod=options->get(HTTP_METHOD_NAME)) {
1.127 paf 389: valid_options++;
1.154 paf 390: method=vmethod->as_string().cstr();
1.180 ! paf 391: } else if(Value* vform=options->get(HTTP_FORM_NAME)) {
! 392: valid_options++;
! 393: form=vform->get_hash();
! 394: } else if(Value* vbody=options->get(HTTP_BODY_NAME)) {
! 395: valid_options++;
! 396: body_cstr=vbody->as_string().cstr(String::L_UNSPECIFIED);
! 397: } else if(Value* vtimeout=options->get(HTTP_TIMEOUT_NAME)) {
1.127 paf 398: valid_options++;
399: timeout=vtimeout->as_int();
1.180 ! paf 400: } else if((vheaders=options->get(HTTP_HEADERS_NAME))) {
1.127 paf 401: valid_options++;
1.180 ! paf 402: } else if(Value* vany_status=options->get(HTTP_ANY_STATUS_NAME)) {
1.142 paf 403: valid_options++;
404: fail_on_status_ne_200=!vany_status->as_bool();
1.180 ! paf 405: } else if(Value* vcharset_name=options->get(HTTP_CHARSET_NAME)) {
1.154 paf 406: valid_options++;
407: asked_remote_charset=&::charsets.get(vcharset_name->as_string().
408: change_case(charsets.source(), String::CC_UPPER));
1.180 ! paf 409: } else if(Value* vuser=options->get(HTTP_USER)) {
1.178 paf 410: valid_options++;
411: user_cstr=vuser->as_string().cstr();
1.180 ! paf 412: } else if(Value* vpassword=options->get(HTTP_PASSWORD)) {
1.178 paf 413: valid_options++;
414: password_cstr=vpassword->as_string().cstr();
415: }
1.142 paf 416:
1.154 paf 417: if(valid_options!=options->count())
1.127 paf 418: throw Exception("parser.runtime",
419: 0,
420: "invalid option passed");
1.154 paf 421: }
422: if(!asked_remote_charset) // defaulting to $request:charset
423: asked_remote_charset=&charsets.source();
424:
1.180 ! paf 425: method_is_get=strcmp(method, "GET")==0;
! 426: if(method_is_get && body_cstr)
! 427: throw Exception("parser.runtime",
! 428: 0,
! 429: "you can not use $."HTTP_BODY_NAME" option with method GET");
! 430:
1.154 paf 431: //preparing request
432: String& connect_string=*new String;
433: // not in ^sql{... L_SQL ...} spirit, but closer to ^file::load one
434: connect_string.append(file_spec, String::L_URI); // tainted pieces -> URI pieces
435:
1.180 ! paf 436: String request_head_and_body;
1.154 paf 437: {
438: // influence URLencoding of tainted pieces to String::L_URI lang
439: Temp_client_charset temp(charsets, *asked_remote_charset);
440:
441: const char* connect_string_cstr=connect_string.cstr(String::L_UNSPECIFIED);
1.126 paf 442:
1.154 paf 443: const char* current=connect_string_cstr;
444: if(strncmp(current, "http://", 7)!=0)
445: throw Exception(0,
446: &connect_string,
447: "does not start with http://"); //never
448: current+=7;
449:
450: strncpy(host, current, sizeof(host)-1); host[sizeof(host)-1]=0;
451: char* host_uri=lsplit(host, '/');
452: uri=host_uri?current+(host_uri-1-host):"/";
453: char* port_cstr=lsplit(host, ':');
454: char* error_pos=0;
455: port=port_cstr?strtol(port_cstr, &error_pos, 0):80;
456:
1.180 ! paf 457: if(strchr(uri, '?') && form)
! 458: throw Exception("parser.runtime",
! 459: 0,
! 460: "use either uri with ?params or $."HTTP_FORM_NAME" option");
! 461:
! 462: //making request head
! 463: String head;
! 464: head << method;
! 465: head << " " << uri;
! 466: if(form)
! 467: if(method_is_get)
! 468: head << form2string(*form);
! 469: else
! 470: body_cstr = form2string(*form);
! 471: head <<" HTTP/1.0" CRLF
1.154 paf 472: "host: "<< host << CRLF;
1.178 paf 473:
1.179 paf 474: // http://www.ietf.org/rfc/rfc2617.txt
1.178 paf 475: if(const String* authorization_field_value=basic_authorization_field(user_cstr, password_cstr))
1.180 ! paf 476: head<<"authorization: "<<*authorization_field_value<<CRLF;
1.178 paf 477:
1.154 paf 478: bool user_agent_specified=false;
479: if(vheaders && !vheaders->is_string()) { // allow empty
480: if(HashStringValue *headers=vheaders->get_hash()) {
1.180 ! paf 481: Http_pass_header_info info={&charsets, &head, false};
1.154 paf 482: headers->for_each(http_pass_header, &info);
483: user_agent_specified=info.user_agent_specified;
484: } else
485: throw Exception("parser.runtime",
486: &connect_string,
487: "headers param must be hash");
488: };
489: if(!user_agent_specified) // defaulting
1.180 ! paf 490: head << "user-agent: " DEFAULT_USER_AGENT CRLF;
! 491:
! 492: const char* head_cstr=head.cstr(String::L_UNSPECIFIED);
! 493:
! 494: // recode those pieces which are not in String::L_URI lang
! 495: // [those violating HTTP standard, but widly used]
! 496: head_cstr=Charset::transcode(
! 497: String::C(head_cstr, strlen(head_cstr)),
! 498: charsets.source(),
! 499: *asked_remote_charset);
! 500:
! 501: if(body_cstr) {
! 502: // recode those pieces which are not in String::L_URI lang
! 503: // [those violating HTTP standard, but widly used]
! 504: body_cstr=Charset::transcode(
! 505: String::C(body_cstr, strlen(body_cstr)),
! 506: charsets.source(),
! 507: *asked_remote_charset);
! 508: }
1.154 paf 509:
1.180 ! paf 510: request_head_and_body << head_cstr;
! 511: if(body_cstr)
! 512: head << "content-length: " << format(strlen(body_cstr), "%u") << CRLF;
! 513:
! 514: // end of header
! 515: request_head_and_body << CRLF;
! 516: // body
! 517: if(body_cstr)
! 518: request_head_and_body << body_cstr;
1.154 paf 519: }
1.126 paf 520:
521: //sending request
1.171 paf 522: char* response;
523: size_t response_size;
524: int status_code=http_request(response, response_size,
1.180 ! paf 525: host, port, request_head_and_body.cstr(),
1.142 paf 526: timeout, fail_on_status_ne_200);
1.126 paf 527:
528: //processing results
1.171 paf 529: char* raw_body; size_t raw_body_size;
530: char* headers_end_at=strstr(response, CRLF CRLF /*change '4' below along!*/);
531: if(headers_end_at) {
532: raw_body=headers_end_at+4;
533: raw_body_size=response_size-(raw_body-response);
534: } else
1.126 paf 535: throw Exception("http.response",
1.133 paf 536: &connect_string,
1.126 paf 537: "bad response from host - no headers found");
1.171 paf 538:
539: *headers_end_at=0;
540: const String header_block(response, headers_end_at-response, true);
1.126 paf 541:
1.154 paf 542: ArrayString aheaders;
543: result.headers=new HashStringValue;
1.155 paf 544: VHash* vtables=new VHash;
1.177 paf 545: result.headers->put(HTTP_TABLES_NAME, vtables);
1.155 paf 546: HashStringValue& tables=vtables->hash();
547:
1.154 paf 548: size_t pos_after=0;
549: header_block.split(aheaders, pos_after, CRLF);
1.126 paf 550:
551: //processing headers
1.154 paf 552: size_t aheaders_count=aheaders.count();
553: Charset* real_remote_charset=0; // undetected, yet
554: for(size_t i=1; i<aheaders_count; i++) {
555: const String& line=*aheaders.get(i);
1.171 paf 556: size_t pos=line.pos(": ", 2);
1.154 paf 557: if(pos==STRING_NOT_FOUND || pos<1)
1.126 paf 558: throw Exception("http.response",
1.154 paf 559: &connect_string,
560: "bad response from host - bad header \"%s\"", line.cstr());
1.156 paf 561: const String::Body HEADER_NAME=
1.154 paf 562: line.mid(0, pos).change_case(charsets.source(), String::CC_UPPER);
563: const String& header_value=line.mid(pos+2, line.length());
564: if(HEADER_NAME=="CONTENT-TYPE")
565: real_remote_charset=detect_charset(charsets.source(), header_value);
1.155 paf 566:
567: // tables
568: {
569: Value *valready=(Value *)tables.get(HEADER_NAME);
570: bool existed=valready!=0;
571: Table *table;
572: if(existed) {
573: // second+ appearence
574: table=valready->get_table();
575: } else {
576: // first appearence
577: Table::columns_type columns =new ArrayString(1);
578: *columns+=new String("value");
579: table=new Table(columns);
580: }
581: // this string becomes next row
582: ArrayString& row=*new ArrayString(1);
583: row+=&header_value;
584: *table+=&row;
585: // not existed before? add it
586: if(!existed)
587: tables.put(HEADER_NAME, new VTable(table));
588: }
589:
1.154 paf 590: result.headers->put(HEADER_NAME, new VString(header_value));
1.126 paf 591: }
1.154 paf 592: // defaulting to used-asked charset [it's never empty!]
593: if(!real_remote_charset)
594: real_remote_charset=asked_remote_charset;
1.126 paf 595:
596: // output response
1.171 paf 597: String::C real_body=String::C(raw_body, raw_body_size);
598: 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 599: real_body=Charset::transcode(real_body, *real_remote_charset, charsets.source());
1.154 paf 600:
601: result.str=const_cast<char *>(real_body.str); // hacking a little
602: result.length=real_body.length;
603: result.headers->put(file_status_name, new VInt(status_code));
604: return result;
1.34 paf 605: }
1.123 paf 606:
1.154 paf 607: #endif
608:
1.123 paf 609: #ifndef DOXYGEN
610: struct File_read_action_info {
1.154 paf 611: char **data; size_t *data_size;
1.126 paf 612: };
1.123 paf 613: #endif
1.154 paf 614: static void file_read_action(
615: struct stat& finfo,
616: int f,
1.166 paf 617: const String& file_spec, const char* /*fname*/, bool as_text,
1.154 paf 618: void *context) {
1.126 paf 619: File_read_action_info& info=*static_cast<File_read_action_info *>(context);
1.123 paf 620: if(size_t to_read_size=(size_t)finfo.st_size) {
1.154 paf 621: *info.data=new(PointerFreeGC) char[to_read_size+(as_text?1:0)];
1.126 paf 622: *info.data_size=(size_t)read(f, *info.data, to_read_size);
1.123 paf 623:
624: if(ssize_t(*info.data_size)<0 || *info.data_size>to_read_size)
1.126 paf 625: throw Exception(0,
1.123 paf 626: &file_spec,
1.173 paf 627: "read failed: actually read %u bytes count not in [0..%u] valid range",
1.126 paf 628: *info.data_size, to_read_size);
1.123 paf 629: } else { // empty file
630: if(as_text) {
1.154 paf 631: *info.data=new(PointerFreeGC) char[1];
1.123 paf 632: *(char*)(*info.data)=0;
633: } else
634: *info.data=0;
635: *info.data_size=0;
636: return;
637: }
1.126 paf 638: }
1.154 paf 639: File_read_result file_read(Request_charsets& charsets, const String& file_spec,
640: bool as_text, HashStringValue *params,
1.126 paf 641: bool fail_on_read_problem) {
1.167 paf 642: File_read_result result={false, 0, 0, 0};
1.154 paf 643: #ifdef PA_HTTP
644: if(file_spec.starts_with("http://")) {
1.126 paf 645: // fail on read problem
1.169 paf 646: File_read_http_result http=file_read_http(charsets, file_spec, as_text, params);
1.154 paf 647: result.success=true;
648: result.str=http.str;
649: result.length=http.length;
650: result.headers=http.headers;
1.126 paf 651: } else {
1.154 paf 652: #endif
1.161 paf 653: if(params && params->count())
654: throw Exception("parser.runtime",
655: 0,
656: "invalid option passed");
657:
1.154 paf 658: File_read_action_info info={&result.str, &result.length};
659: result.success=file_read_action_under_lock(file_spec,
1.126 paf 660: "read", file_read_action, &info,
661: as_text, fail_on_read_problem);
1.154 paf 662: #ifdef PA_HTTP
1.126 paf 663: }
1.154 paf 664: #endif
1.123 paf 665:
1.154 paf 666: if(result.success && as_text) {
1.131 paf 667: // UTF-8 signature: EF BB BF
1.154 paf 668: if(result.length>=3) {
669: char *in=(char *)result.str;
1.159 paf 670: if(strncmp(in, "\xEF\xBB\xBF", 3)==0) {
1.154 paf 671: result.str=in+3; result.length-=3;// skip prefix
1.131 paf 672: }
673: }
674:
1.154 paf 675: fix_line_breaks((char *)(result.str), result.length);
1.123 paf 676: }
1.126 paf 677:
678: return result;
1.123 paf 679: }
680:
1.154 paf 681: #ifdef PA_SAFE_MODE
682: void check_safe_mode(struct stat finfo, const String& file_spec, const char* fname) {
683: if(finfo.st_uid/*foreign?*/!=geteuid()
684: && finfo.st_gid/*foreign?*/!=getegid())
685: throw Exception("parser.runtime",
686: &file_spec,
687: "parser is in safe mode: "
688: "reading files of foreign group and user disabled "
689: "[recompile parser with --disable-safe-mode configure option], "
690: "actual filename '%s', "
691: "fuid(%d)!=euid(%d) or fgid(%d)!=egid(%d)",
692: fname,
693: finfo.st_uid, geteuid(),
694: finfo.st_gid, getegid());
695: }
696: #endif
1.149 paf 697:
1.154 paf 698: bool file_read_action_under_lock(const String& file_spec,
1.126 paf 699: const char* action_name, File_read_action action, void *context,
700: bool as_text,
1.123 paf 701: bool fail_on_read_problem) {
1.154 paf 702: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.33 paf 703: int f;
704:
705: // first open, next stat:
1.45 paf 706: // directory update of NTFS hard links performed on open.
1.33 paf 707: // ex:
708: // a.html:^test[] and b.html hardlink to a.html
709: // user inserts ! before ^test in a.html
1.126 paf 710: // directory entry of b.html in NTFS not updated at once,
1.35 paf 711: // they delay update till open, so we would receive "!^test[" string
712: // if would do stat, next open.
1.123 paf 713: // later: it seems, even this does not help sometimes
1.98 paf 714: if((f=open(fname, O_RDONLY|(as_text?_O_TEXT:_O_BINARY)))>=0) {
1.123 paf 715: try {
1.162 paf 716: if(pa_lock_shared_blocking(f)!=0)
1.126 paf 717: throw Exception("file.lock",
1.123 paf 718: &file_spec,
719: "shared lock failed: %s (%d), actual filename '%s'",
1.154 paf 720: strerror(errno), errno, fname);
1.123 paf 721:
1.124 paf 722: struct stat finfo;
723: if(stat(fname, &finfo)!=0)
724: throw Exception("file.missing", // hardly possible: we just opened it OK
725: &file_spec,
726: "stat failed: %s (%d), actual filename '%s'",
1.154 paf 727: strerror(errno), errno, fname);
1.124 paf 728:
1.140 paf 729: #ifdef PA_SAFE_MODE
1.149 paf 730: check_safe_mode(finfo, file_spec, fname);
1.105 paf 731: #endif
1.32 paf 732:
1.154 paf 733: action(finfo, f, file_spec, fname, as_text, context);
1.123 paf 734: } catch(...) {
1.162 paf 735: pa_unlock(f);close(f);
1.123 paf 736: if(fail_on_read_problem)
1.154 paf 737: rethrow;
1.123 paf 738: return false;
739: }
1.87 paf 740:
1.162 paf 741: pa_unlock(f);close(f);
1.72 parser 742: return true;
1.118 paf 743: } else {
744: if(fail_on_read_problem)
1.126 paf 745: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.118 paf 746: &file_spec,
1.123 paf 747: "%s failed: %s (%d), actual filename '%s'",
1.154 paf 748: action_name, strerror(errno), errno, fname);
1.118 paf 749: return false;
750: }
1.8 paf 751: }
752:
1.63 parser 753: static void create_dir_for_file(const String& file_spec) {
754: size_t pos_after=1;
1.154 paf 755: size_t pos_before;
756: while((pos_before=file_spec.pos('/', pos_after))!=STRING_NOT_FOUND) {
757: mkdir(file_spec.mid(0, pos_before).cstr(String::L_FILE_SPEC), 0775);
1.63 parser 758: pos_after=pos_before+1;
759: }
760: }
761:
1.98 paf 762: bool file_write_action_under_lock(
1.28 paf 763: const String& file_spec,
1.126 paf 764: const char* action_name, File_write_action action, void *context,
765: bool as_text,
766: bool do_append,
767: bool do_block,
1.110 paf 768: bool fail_on_lock_problem) {
1.154 paf 769: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.28 paf 770: int f;
1.80 paf 771: if(access(fname, W_OK)!=0) // no
1.126 paf 772: create_dir_for_file(file_spec);
1.50 paf 773:
1.80 paf 774: if((f=open(fname,
775: O_CREAT|O_RDWR
776: |(as_text?_O_TEXT:_O_BINARY)
1.138 paf 777: |(do_append?O_APPEND:PA_O_TRUNC), 0664))>=0) {
1.162 paf 778: if((do_block?pa_lock_exclusive_blocking(f):pa_lock_exclusive_nonblocking(f))!=0) {
1.126 paf 779: Exception e("file.lock",
1.110 paf 780: &file_spec,
781: "shared lock failed: %s (%d), actual filename '%s'",
1.154 paf 782: strerror(errno), errno, fname);
1.126 paf 783: close(f);
1.110 paf 784: if(fail_on_lock_problem)
785: throw e;
1.98 paf 786: return false;
787: }
1.96 paf 788:
1.158 paf 789: try {
1.126 paf 790: action(f, context);
1.158 paf 791: } catch(...) {
1.138 paf 792: #ifdef HAVE_FTRUNCATE
1.104 paf 793: if(!do_append)
1.125 paf 794: ftruncate(f, lseek(f, 0, SEEK_CUR)); // one can not use O_TRUNC, read lower
1.138 paf 795: #endif
1.162 paf 796: pa_unlock(f);close(f);
1.154 paf 797: rethrow;
1.158 paf 798: }
1.80 paf 799:
1.138 paf 800: #ifdef HAVE_FTRUNCATE
1.104 paf 801: if(!do_append)
1.125 paf 802: 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 803: #endif
1.162 paf 804: pa_unlock(f);close(f);
1.98 paf 805: return true;
1.80 paf 806: } else
1.126 paf 807: throw Exception(errno==EACCES?"file.access":0,
1.80 paf 808: &file_spec,
1.96 paf 809: "%s failed: %s (%d), actual filename '%s'",
1.154 paf 810: action_name, strerror(errno), errno, fname);
1.96 paf 811: // here should be nothing, see rethrow above
812: }
813:
814: #ifndef DOXYGEN
815: struct File_write_action_info {
1.154 paf 816: const char* str; size_t length;
1.126 paf 817: };
1.96 paf 818: #endif
819: static void file_write_action(int f, void *context) {
1.126 paf 820: File_write_action_info& info=*static_cast<File_write_action_info *>(context);
1.154 paf 821: if(info.length) {
822: int written=write(f, info.str, info.length);
1.116 paf 823: if(written<0)
1.126 paf 824: throw Exception(0,
825: 0,
826: "write failed: %s (%d)", strerror(errno), errno);
1.113 paf 827: }
1.96 paf 828: }
829: void file_write(
830: const String& file_spec,
1.154 paf 831: const char* data, size_t size,
1.126 paf 832: bool as_text,
1.96 paf 833: bool do_append) {
1.126 paf 834: File_write_action_info info={data, size};
1.98 paf 835: file_write_action_under_lock(
1.154 paf 836: file_spec,
837: "write", file_write_action, &info,
838: as_text,
839: do_append);
1.30 paf 840: }
841:
1.63 parser 842: // throws nothing! [this is required in file_move & file_delete]
1.50 paf 843: static void rmdir(const String& file_spec, size_t pos_after) {
1.154 paf 844: size_t pos_before;
845: if((pos_before=file_spec.pos('/', pos_after))!=STRING_NOT_FOUND)
1.126 paf 846: rmdir(file_spec, pos_before+1);
1.50 paf 847:
1.154 paf 848: rmdir(file_spec.mid(0, pos_after-1/* / */).cstr(String::L_FILE_SPEC));
1.50 paf 849: }
1.164 paf 850: bool file_delete(const String& file_spec, bool fail_on_problem) {
1.154 paf 851: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.54 parser 852: if(unlink(fname)!=0)
1.164 paf 853: if(fail_on_problem)
1.126 paf 854: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.93 paf 855: &file_spec,
856: "unlink failed: %s (%d), actual filename '%s'",
1.154 paf 857: strerror(errno), errno, fname);
1.93 paf 858: else
859: return false;
1.50 paf 860:
1.126 paf 861: rmdir(file_spec, 1);
1.93 paf 862: return true;
1.60 parser 863: }
1.95 paf 864: void file_move(const String& old_spec, const String& new_spec) {
1.154 paf 865: const char* old_spec_cstr=old_spec.cstr(String::L_FILE_SPEC);
866: const char* new_spec_cstr=new_spec.cstr(String::L_FILE_SPEC);
1.63 parser 867:
1.126 paf 868: create_dir_for_file(new_spec);
1.63 parser 869:
1.60 parser 870: if(rename(old_spec_cstr, new_spec_cstr)!=0)
1.126 paf 871: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.60 parser 872: &old_spec,
873: "rename failed: %s (%d), actual filename '%s' to '%s'",
1.154 paf 874: strerror(errno), errno, old_spec_cstr, new_spec_cstr);
1.63 parser 875:
1.126 paf 876: rmdir(old_spec, 1);
1.31 paf 877: }
878:
1.51 paf 879:
1.126 paf 880: bool entry_exists(const char* fname, struct stat *afinfo) {
1.118 paf 881: struct stat lfinfo;
882: bool result=stat(fname, &lfinfo)==0;
883: if(afinfo)
884: *afinfo=lfinfo;
885: return result;
1.119 paf 886: }
887:
888: bool entry_exists(const String& file_spec) {
1.154 paf 889: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.126 paf 890: return entry_exists(fname, 0);
1.118 paf 891: }
892:
1.51 paf 893: static bool entry_readable(const String& file_spec, bool need_dir) {
1.154 paf 894: char* fname=file_spec.cstrm(String::L_FILE_SPEC);
1.120 paf 895: if(need_dir) {
1.126 paf 896: size_t size=strlen(fname);
1.120 paf 897: while(size) {
1.126 paf 898: char c=fname[size-1];
1.120 paf 899: if(c=='/' || c=='\\')
900: fname[--size]=0;
901: else
902: break;
903: }
904: }
1.51 paf 905: struct stat finfo;
1.118 paf 906: if(access(fname, R_OK)==0 && entry_exists(fname, &finfo)) {
1.109 paf 907: bool is_dir=(finfo.st_mode&S_IFDIR) != 0;
1.51 paf 908: return is_dir==need_dir;
909: }
910: return false;
911: }
1.31 paf 912: bool file_readable(const String& file_spec) {
1.126 paf 913: return entry_readable(file_spec, false);
1.51 paf 914: }
915: bool dir_readable(const String& file_spec) {
1.126 paf 916: return entry_readable(file_spec, true);
1.65 parser 917: }
1.154 paf 918: const String* file_readable(const String& path, const String& name) {
919: String& result=*new String(path);
920: result << "/";
921: result << name;
922: return file_readable(result)?&result:0;
1.43 paf 923: }
924: bool file_executable(const String& file_spec) {
1.154 paf 925: return access(file_spec.cstr(String::L_FILE_SPEC), X_OK)==0;
1.44 paf 926: }
927:
1.64 parser 928: bool file_stat(const String& file_spec,
1.58 parser 929: size_t& rsize,
1.126 paf 930: time_t& ratime,
931: time_t& rmtime,
932: time_t& rctime,
1.64 parser 933: bool fail_on_read_problem) {
1.154 paf 934: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
935: struct stat finfo;
1.44 paf 936: if(stat(fname, &finfo)!=0)
1.64 parser 937: if(fail_on_read_problem)
1.126 paf 938: throw Exception("file.missing",
1.67 parser 939: &file_spec,
940: "getting file size failed: %s (%d), real filename '%s'",
1.154 paf 941: strerror(errno), errno, fname);
1.64 parser 942: else
943: return false;
1.58 parser 944: rsize=finfo.st_size;
945: ratime=finfo.st_atime;
946: rmtime=finfo.st_mtime;
947: rctime=finfo.st_ctime;
1.64 parser 948: return true;
1.18 paf 949: }
950:
1.126 paf 951: char* getrow(char* *row_ref, char delim) {
952: char* result=*row_ref;
1.8 paf 953: if(result) {
1.126 paf 954: *row_ref=strchr(result, delim);
1.8 paf 955: if(*row_ref)
956: *((*row_ref)++)=0;
957: else if(!*result)
958: return 0;
959: }
960: return result;
961: }
962:
1.126 paf 963: char* lsplit(char* string, char delim) {
1.23 paf 964: if(string) {
1.126 paf 965: char* v=strchr(string, delim);
1.8 paf 966: if(v) {
967: *v=0;
968: return v+1;
969: }
970: }
971: return 0;
972: }
973:
1.126 paf 974: char* lsplit(char* *string_ref, char delim) {
975: char* result=*string_ref;
976: char* next=lsplit(*string_ref, delim);
1.8 paf 977: *string_ref=next;
978: return result;
1.9 paf 979: }
980:
1.126 paf 981: char* rsplit(char* string, char delim) {
1.18 paf 982: if(string) {
1.126 paf 983: char* v=strrchr(string, delim);
1.18 paf 984: if(v) {
1.9 paf 985: *v=0;
986: return v+1;
987: }
988: }
989: return NULL;
1.10 paf 990: }
991:
1.37 paf 992: /// @todo less stupid type detection
1.154 paf 993: const char* format(double value, char* fmt) {
1.126 paf 994: char local_buf[MAX_NUMBER];
1.108 paf 995: size_t size;
996:
1.10 paf 997: if(fmt)
998: if(strpbrk(fmt, "diouxX"))
999: if(strpbrk(fmt, "ouxX"))
1.126 paf 1000: size=snprintf(local_buf, sizeof(local_buf), fmt, (uint)value);
1.10 paf 1001: else
1.126 paf 1002: size=snprintf(local_buf, sizeof(local_buf), fmt, (int)value);
1.10 paf 1003: else
1.126 paf 1004: size=snprintf(local_buf, sizeof(local_buf), fmt, value);
1.10 paf 1005: else
1.126 paf 1006: size=snprintf(local_buf, sizeof(local_buf), "%d", (int)value);
1.10 paf 1007:
1.154 paf 1008: return pa_strdup(local_buf, size);
1.12 paf 1009: }
1010:
1.36 paf 1011: size_t stdout_write(const void *buf, size_t size) {
1.12 paf 1012: #ifdef WIN32
1013: do{
1.154 paf 1014: int chunk_written=fwrite(buf, 1, min((size_t)8*0x400, size), stdout);
1.12 paf 1015: if(chunk_written<=0)
1016: break;
1017: size-=chunk_written;
1.36 paf 1018: buf=((const char*)buf)+chunk_written;
1.126 paf 1019: } while(size>0);
1.12 paf 1020:
1021: return size;
1022: #else
1.126 paf 1023: return fwrite(buf, 1, size, stdout);
1.12 paf 1024: #endif
1.2 paf 1025: }
1.14 paf 1026:
1.154 paf 1027: char* unescape_chars(const char* cp, int len) {
1028: char* s=new(PointerFreeGC) char[len + 1];
1.14 paf 1029: enum EscapeState {
1.33 paf 1030: EscapeRest,
1031: EscapeFirst,
1.14 paf 1032: EscapeSecond
1033: } escapeState=EscapeRest;
1034: int escapedValue=0;
1035: int srcPos=0;
1036: int dstPos=0;
1037: while(srcPos < len) {
1.126 paf 1038: int ch=cp[srcPos];
1.14 paf 1039: switch(escapeState) {
1040: case EscapeRest:
1041: if(ch=='%') {
1042: escapeState=EscapeFirst;
1043: } else if(ch=='+') {
1.126 paf 1044: s[dstPos++]=' ';
1.14 paf 1045: } else {
1046: s[dstPos++]=ch;
1047: }
1048: break;
1049: case EscapeFirst:
1050: escapedValue=hex_value[ch] << 4;
1051: escapeState=EscapeSecond;
1052: break;
1053: case EscapeSecond:
1.126 paf 1054: escapedValue +=hex_value[ch];
1.14 paf 1055: s[dstPos++]=escapedValue;
1056: escapeState=EscapeRest;
1057: break;
1058: }
1.126 paf 1059: srcPos++;
1.14 paf 1060: }
1061: s[dstPos]=0;
1062: return s;
1.24 paf 1063: }
1064:
1065: #ifdef WIN32
1.126 paf 1066: void back_slashes_to_slashes(char* s) {
1.24 paf 1067: if(s)
1068: for(; *s; s++)
1069: if(*s=='\\')
1.126 paf 1070: *s='/';
1.24 paf 1071: }
1.42 paf 1072: /*
1.126 paf 1073: void slashes_to_back_slashes(char* s) {
1.42 paf 1074: if(s)
1075: for(; *s; s++)
1076: if(*s=='/')
1.126 paf 1077: *s='\\';
1.42 paf 1078: }
1079: */
1.24 paf 1080: #endif
1.41 paf 1081:
1.126 paf 1082: bool StrEqNc(const char* s1, const char* s2, bool strict) {
1.41 paf 1083: while(true) {
1084: if(!(*s1)) {
1085: if(!(*s2))
1086: return true;
1087: else
1088: return !strict;
1089: } else if(!(*s2))
1090: return !strict;
1091: if(isalpha(*s1)) {
1092: if(tolower(*s1) !=tolower(*s2))
1093: return false;
1094: } else if((*s1) !=(*s2))
1095: return false;
1.126 paf 1096: s1++;
1097: s2++;
1.41 paf 1098: }
1.57 parser 1099: }
1100:
1.84 paf 1101: static bool isLeap(int year) {
1.57 parser 1102: return !(
1103: (year % 4) || ((year % 400) && !(year % 100))
1.126 paf 1104: );
1.57 parser 1105: }
1106:
1107: int getMonthDays(int year, int month) {
1108: int monthDays[]={
1.126 paf 1109: 31,
1110: isLeap(year) ? 29 : 28,
1111: 31,
1112: 30,
1113: 31,
1114: 30,
1115: 31,
1116: 31,
1117: 30,
1118: 31,
1119: 30,
1.57 parser 1120: 31
1.126 paf 1121: };
1122: return monthDays[month];
1.41 paf 1123: }
1.69 parser 1124:
1.126 paf 1125: void remove_crlf(char* start, char* end) {
1126: for(char* p=start; p<end; p++)
1.69 parser 1127: switch(*p) {
1.126 paf 1128: case '\n': *p='|'; break;
1129: case '\r': *p=' '; break;
1.69 parser 1130: }
1.91 paf 1131: }
1132:
1133:
1134: /// must be last in this file
1135: #undef vsnprintf
1.126 paf 1136: int __vsnprintf(char* b, size_t s, const char* f, va_list l) {
1.91 paf 1137: if(!s)
1138: return 0;
1139:
1140: int r;
1141: // note: on win32& maybe somewhere else
1142: // vsnprintf do not writes terminating 0 in 'buffer full' case, reducing
1143: --s;
1.172 paf 1144:
1145: // clients do not check for negative 's', feature: ignore such prints
1146: if((ssize_t)s<0)
1147: return 0;
1148:
1.91 paf 1149: #if _MSC_VER
1150: /*
1151: win32:
1152: mk:@MSITStore:C:\Program%20Files\Microsoft%20Visual%20Studio\MSDN\2001APR\1033\vccore.chm::/html/_crt__vsnprintf.2c_._vsnwprintf.htm
1153:
1.154 paf 1154: if the number of bytes to write exceeds buffer, then count bytes are written and Ö1 is returned
1.91 paf 1155: */
1.126 paf 1156: r=_vsnprintf(b, s, f, l);
1.91 paf 1157: if(r<0)
1158: r=s;
1159: #else
1.126 paf 1160: r=vsnprintf(b, s, f, l);
1.91 paf 1161: /*
1162: solaris:
1163: man vsnprintf
1164:
1165: The snprintf() function returns the number of characters
1166: formatted, that is, the number of characters that would have
1167: been written to the buffer if it were large enough. If the
1168: value of n is 0 on a call to snprintf(), an unspecified
1169: value less than 1 is returned.
1170: */
1171:
1172: if(r<0)
1173: r=0;
1.167 paf 1174: else if((size_t)r>s)
1.91 paf 1175: r=s;
1176: #endif
1177: b[r]=0;
1178: return r;
1179: }
1180:
1.126 paf 1181: int __snprintf(char* b, size_t s, const char* f, ...) {
1.91 paf 1182: va_list l;
1.126 paf 1183: va_start(l, f);
1184: int r=__vsnprintf(b, s, f, l);
1185: va_end(l);
1.91 paf 1186: return r;
1.178 paf 1187: }
1188:
1189: /* mime64 functions are from libgmime[http://spruce.sourceforge.net/gmime/] lib */
1190: /*
1191: * Authors: Michael Zucchi <notzed@helixcode.com>
1192: * Jeffrey Stedfast <fejj@helixcode.com>
1193: *
1194: * Copyright 2000 Helix Code, Inc. (www.helixcode.com)
1195: *
1196: * This program is free software; you can redistribute it and/or modify
1197: * it under the terms of the GNU General Public License as published by
1198: * the Free Software Foundation; either version 2 of the License, or
1199: * (at your option) any later version.
1200: *
1201: * This program is distributed in the hope that it will be useful,
1202: * but WITHOUT ANY WARRANTY; without even the implied warranty of
1203: * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
1204: * GNU General Public License for more details.
1205: *
1206: * You should have received a copy of the GNU General Public License
1207: * along with this program; if not, write to the Free Software
1208: * Foundation, Inc., 59 Temple Street #330, Boston, MA 02111-1307, USA.
1209: *
1210: */
1211: static char *base64_alphabet =
1212: "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";
1213:
1214: /**
1215: * g_mime_utils_base64_encode_step:
1216: * @in: input stream
1217: * @inlen: length of the input
1218: * @out: output string
1219: * @state: holds the number of bits that are stored in @save
1220: * @save: leftover bits that have not yet been encoded
1221: *
1222: * Base64 encodes a chunk of data. Performs an 'encode step', only
1223: * encodes blocks of 3 characters to the output at a time, saves
1224: * left-over state in state and save (initialise to 0 on first
1225: * invocation).
1226: *
1227: * Returns the number of bytes encoded.
1228: **/
1229: static size_t
1230: g_mime_utils_base64_encode_step (const unsigned char *in, size_t inlen, unsigned char *out, int *state, int *save)
1231: {
1232: const register unsigned char *inptr;
1233: register unsigned char *outptr;
1234:
1235: if (inlen <= 0)
1236: return 0;
1237:
1238: inptr = in;
1239: outptr = out;
1240:
1241: if (inlen + ((unsigned char *)save)[0] > 2) {
1242: const unsigned char *inend = in + inlen - 2;
1243: register int c1 = 0, c2 = 0, c3 = 0;
1244: register int already;
1245:
1246: already = *state;
1247:
1248: switch (((char *)save)[0]) {
1249: case 1: c1 = ((unsigned char *)save)[1]; goto skip1;
1250: case 2: c1 = ((unsigned char *)save)[1];
1251: c2 = ((unsigned char *)save)[2]; goto skip2;
1252: }
1253:
1254: /* yes, we jump into the loop, no i'm not going to change it, its beautiful! */
1255: while (inptr < inend) {
1256: c1 = *inptr++;
1257: skip1:
1258: c2 = *inptr++;
1259: skip2:
1260: c3 = *inptr++;
1261: *outptr++ = base64_alphabet [c1 >> 2];
1262: *outptr++ = base64_alphabet [(c2 >> 4) | ((c1 & 0x3) << 4)];
1263: *outptr++ = base64_alphabet [((c2 & 0x0f) << 2) | (c3 >> 6)];
1264: *outptr++ = base64_alphabet [c3 & 0x3f];
1265: /* this is a bit ugly ... */
1266: if ((++already) >= 19) {
1267: *outptr++ = '\n';
1268: already = 0;
1269: }
1270: }
1271:
1272: ((unsigned char *)save)[0] = 0;
1273: inlen = 2 - (inptr - inend);
1274: *state = already;
1275: }
1276:
1277: //d(printf ("state = %d, inlen = %d\n", (int)((char *)save)[0], inlen));
1278:
1279: if (inlen > 0) {
1280: register char *saveout;
1281:
1282: /* points to the slot for the next char to save */
1283: saveout = & (((char *)save)[1]) + ((char *)save)[0];
1284:
1285: /* inlen can only be 0 1 or 2 */
1286: switch (inlen) {
1287: case 2: *saveout++ = *inptr++;
1288: case 1: *saveout++ = *inptr++;
1289: }
1290: ((char *)save)[0] += inlen;
1291: }
1292:
1293: /*d(printf ("mode = %d\nc1 = %c\nc2 = %c\n",
1294: (int)((char *)save)[0],
1295: (int)((char *)save)[1],
1296: (int)((char *)save)[2]));*/
1297:
1298: return (outptr - out);
1299: }
1300:
1301: /**
1302: * g_mime_utils_base64_encode_close:
1303: * @in: input stream
1304: * @inlen: length of the input
1305: * @out: output string
1306: * @state: holds the number of bits that are stored in @save
1307: * @save: leftover bits that have not yet been encoded
1308: *
1309: * Base64 encodes the input stream to the output stream. Call this
1310: * when finished encoding data with g_mime_utils_base64_encode_step to
1311: * flush off the last little bit.
1312: *
1313: * Returns the number of bytes encoded.
1314: **/
1315: static size_t
1316: g_mime_utils_base64_encode_close (const unsigned char *in, size_t inlen, unsigned char *out, int *state, int *save)
1317: {
1318: unsigned char *outptr = out;
1319: int c1, c2;
1320:
1321: if (inlen > 0)
1322: outptr += g_mime_utils_base64_encode_step (in, inlen, outptr, state, save);
1323:
1324: c1 = ((unsigned char *)save)[1];
1325: c2 = ((unsigned char *)save)[2];
1326:
1327: switch (((unsigned char *)save)[0]) {
1328: case 2:
1329: outptr[2] = base64_alphabet [(c2 & 0x0f) << 2];
1330: goto skip;
1331: case 1:
1332: outptr[2] = '=';
1333: skip:
1334: outptr[0] = base64_alphabet [c1 >> 2];
1335: outptr[1] = base64_alphabet [c2 >> 4 | ((c1 & 0x3) << 4)];
1336: outptr[3] = '=';
1337: outptr += 4;
1338: break;
1339: }
1340:
1341: *outptr++ = 0;
1342:
1343: *save = 0;
1344: *state = 0;
1345:
1346: return (outptr - out);
1347: }
1348:
1349: char* pa_base64(const char *in, size_t len)
1350: {
1351: /* wont go to more than 2x size (overly conservative) */
1352: char* result=new(PointerFreeGC) char[len * 2 + 6];
1353: int state=0;
1354: int save=0;
1355: size_t filled=g_mime_utils_base64_encode_close ((const unsigned char*)in, len,
1356: (unsigned char*)result, &state, &save);
1357: assert(filled <= len * 2 + 6);
1358:
1359: return result;
1.98 paf 1360: }
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