Annotation of parser3/src/main/pa_common.C, revision 1.193
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:
1.193 ! paf 8: static const char * const IDENT_COMMON_C="$Date: 2004/08/30 09:33:13 $";
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);
1.184 paf 145: if(packed_ip!=INADDR_NONE)
1.185 paf 146: memcpy(&addr->sin_addr, &packed_ip, sizeof(packed_ip));
1.184 paf 147: else {
148: struct hostent *hostIP=gethostbyname(host);
149: if(hostIP)
150: memcpy(&addr->sin_addr, hostIP->h_addr, hostIP->h_length);
151: else
152: return false;
153: }
1.126 paf 154: } else
155: addr->sin_addr.s_addr=INADDR_ANY;
156: return true;
157: }
158:
1.171 paf 159: static int http_read_response(char*& response, size_t& response_size, int sock, bool fail_on_status_ne_200){
160: response=(char*)pa_malloc_atomic(1/*terminator*/); // setting memory block type
161: response[(response_size=0)]=0;
1.154 paf 162: int result=0;
1.171 paf 163: char* EOLat=0;
1.126 paf 164: while(true) {
1.171 paf 165: char buf[MAX_STRING*10];
166: ssize_t received_size=recv(sock, buf, sizeof(buf), 0);
167: if(received_size<=0)
1.126 paf 168: break;
1.171 paf 169: response=(char*)pa_realloc(response, response_size+received_size+1/*terminator*/);
170: memcpy(response+response_size, buf, received_size);
171: response_size+=received_size;
172: response[response_size]=0;
173:
1.192 paf 174: if(!result && (EOLat=strstr(response, "\n"))) { // checking status in first response
1.171 paf 175: const String status_line(pa_strdup(response, EOLat-response));
1.154 paf 176: ArrayString astatus;
177: size_t pos_after=0;
178: status_line.split(astatus, pos_after, " ");
1.182 paf 179: const String& status_code=*astatus.get(astatus.count()>1?1:0);
1.154 paf 180: result=status_code.as_int();
1.142 paf 181:
1.154 paf 182: if(fail_on_status_ne_200 && result!=200)
1.142 paf 183: throw Exception("http.status",
1.154 paf 184: &status_code,
1.142 paf 185: "invalid HTTP response status");
186: }
187: }
1.154 paf 188: if(result)
189: return result;
1.142 paf 190: else
191: throw Exception("http.response",
192: 0,
1.173 paf 193: "bad response from host - no status found (size=%u)", response_size);
1.126 paf 194: }
195:
196: /* ********************** request *************************** */
197:
198: #if defined(SIGALRM) && defined(HAVE_SIGSETJMP) && defined(HAVE_SIGLONGJMP)
1.145 paf 199: # define PA_USE_ALARM
1.126 paf 200: #endif
201:
1.145 paf 202: #ifdef PA_USE_ALARM
1.126 paf 203: static sigjmp_buf timeout_env;
204: static void timeout_handler(int sig){
205: siglongjmp(timeout_env, 1);
206: }
207: #endif
208:
1.171 paf 209: static int http_request(char*& response, size_t& response_size,
1.193 ! paf 210: const char* host, short port,
1.152 paf 211: const char* request,
1.166 paf 212: int
213: #ifdef PA_USE_ALARM
214: timeout
215: #endif
216: ,
1.152 paf 217: bool fail_on_status_ne_200) {
1.126 paf 218: if(!host)
219: throw Exception("http.host",
1.154 paf 220: 0,
1.126 paf 221: "zero hostname"); //never
222:
1.145 paf 223: #ifdef PA_USE_ALARM
1.146 paf 224: signal(SIGALRM, timeout_handler);
1.126 paf 225: #endif
226: int sock=-1;
1.145 paf 227: #ifdef PA_USE_ALARM
228: if(sigsetjmp(timeout_env, 1)) {
229: // stupid gcc [2.95.4] generated bad code
230: // which failed to handle sigsetjmp+throw: crashed inside of pre-throw code.
231: // rewritten simplier [though duplicating closesocket code]
232: if(sock>=0)
233: closesocket(sock);
234: throw Exception("http.timeout",
235: origin_string,
236: "timeout occured while retrieving document");
1.146 paf 237: return 0; // never
1.145 paf 238: } else {
239: alarm(timeout);
240: #endif
1.146 paf 241: try {
242: int result;
1.126 paf 243: struct sockaddr_in dest;
1.154 paf 244:
245: if(!set_addr(&dest, host, port))
1.126 paf 246: throw Exception("http.host",
1.154 paf 247: 0,
1.127 paf 248: "can not resolve hostname \"%s\"", host);
1.126 paf 249:
250: if((sock=socket(AF_INET, SOCK_STREAM, IPPROTO_TCP/*0*/))<0)
251: throw Exception("http.connect",
1.154 paf 252: 0,
1.127 paf 253: "can not make socket: %s (%d)", strerror(errno), errno);
1.126 paf 254: if(connect(sock, (struct sockaddr *)&dest, sizeof(dest)))
255: throw Exception("http.connect",
1.154 paf 256: 0,
1.127 paf 257: "can not connect to host \"%s\": %s (%d)", host, strerror(errno), errno);
1.126 paf 258: size_t request_size=strlen(request);
259: if(send(sock, request, request_size, 0)!=(ssize_t)request_size)
260: throw Exception("http.connect",
1.154 paf 261: 0,
1.127 paf 262: "error sending request: %s (%d)", strerror(errno), errno);
1.126 paf 263:
1.171 paf 264: result=http_read_response(response, response_size, sock, fail_on_status_ne_200);
1.142 paf 265: closesocket(sock);
1.145 paf 266: #ifdef PA_USE_ALARM
1.142 paf 267: alarm(0);
1.126 paf 268: #endif
1.147 paf 269: return result;
1.146 paf 270: } catch(...) {
1.145 paf 271: #ifdef PA_USE_ALARM
1.146 paf 272: alarm(0);
1.126 paf 273: #endif
1.146 paf 274: if(sock>=0)
275: closesocket(sock);
1.154 paf 276: rethrow;
1.146 paf 277: }
1.148 paf 278: #ifdef PA_USE_ALARM
1.126 paf 279: }
1.148 paf 280: #endif
1.126 paf 281: }
282:
1.127 paf 283: #ifndef DOXYGEN
284: struct Http_pass_header_info {
1.154 paf 285: Request_charsets* charsets;
1.127 paf 286: String* request;
287: bool user_agent_specified;
288: };
289: #endif
1.154 paf 290: static void http_pass_header(HashStringValue::key_type key,
291: HashStringValue::value_type value,
292: Http_pass_header_info *info) {
293: *info->request <<key<<": "
294: << attributed_meaning_to_string(*value, String::L_HTTP_HEADER, false)
295: << CRLF;
1.135 paf 296:
1.154 paf 297: if(String(key, String::L_TAINTED).change_case(info->charsets->source(), String::CC_UPPER)=="USER-AGENT")
298: info->user_agent_specified=true;
1.126 paf 299: }
1.154 paf 300:
301:
302: static Charset* detect_charset(Charset& source_charset, const String& content_type_value) {
1.156 paf 303: const String::Body CONTENT_TYPE_VALUE=
1.154 paf 304: content_type_value.change_case(source_charset, String::CC_UPPER);
305: // content-type: xxx/xxx; source_charset=WE-NEED-THIS
306: // content-type: xxx/xxx; source_charset="WE-NEED-THIS"
307: // content-type: xxx/xxx; source_charset="WE-NEED-THIS";
308: size_t before_charseteq_pos=CONTENT_TYPE_VALUE.pos("CHARSET=");
309: if(before_charseteq_pos!=STRING_NOT_FOUND) {
310: size_t charset_begin=before_charseteq_pos+8/*CHARSET="*/;
311: size_t open_quote_pos=CONTENT_TYPE_VALUE.pos('"', charset_begin);
312: bool quoted=open_quote_pos==charset_begin;
313: if(quoted)
314: charset_begin++; // skip opening '"'
315: size_t charset_end=CONTENT_TYPE_VALUE.length();
316: if(quoted) {
317: size_t close_quote_pos=CONTENT_TYPE_VALUE.pos('"', charset_begin);
318: if(close_quote_pos!=STRING_NOT_FOUND)
319: charset_end=close_quote_pos;
320: } else {
321: size_t delim_pos=CONTENT_TYPE_VALUE.pos(';', charset_begin);
322: if(delim_pos!=STRING_NOT_FOUND)
323: charset_end=delim_pos;
324: }
1.156 paf 325: const String::Body CHARSET_NAME_BODY=
1.154 paf 326: CONTENT_TYPE_VALUE.mid(charset_begin, charset_end);
327:
328: return &charsets.get(CHARSET_NAME_BODY);
329: }
330:
331: return 0;
332: }
333:
1.178 paf 334: static const String* basic_authorization_field(const char* user, const char* pass) {
335: if(!user&& !pass)
336: return 0;
337:
338: String combined;
339: if(user)
340: combined<<user;
341: combined<<":";
342: if(pass)
343: combined<<pass;
344:
345: String* result=new String("Basic "); *result<<pa_base64(combined.cstr(), combined.length());
346: return result;
347: }
1.154 paf 348:
1.181 paf 349: static void form_string_value2string(
350: HashStringValue::key_type key,
351: const String& value,
352: String& result)
353: {
354: result << String(key, String::L_TAINTED) << "=";
355: result.append(value, String::L_URI, true);
356: result<< "&";
357: }
358: #ifndef DOXYGEN
359: struct Form_table_value2string_info {
360: HashStringValue::key_type key;
361: String& result;
362:
363: Form_table_value2string_info(HashStringValue::key_type akey, String& aresult):
364: key(akey), result(aresult) {}
365: };
366: #endif
367: static void form_table_value2string(Table::element_type row, Form_table_value2string_info* info) {
368: form_string_value2string(info->key, *row->get(0), info->result);
369: }
1.180 paf 370: static void form_value2string(
371: HashStringValue::key_type key,
372: HashStringValue::value_type value,
373: String* result)
374: {
1.181 paf 375: if(const String* svalue=value->get_string())
376: form_string_value2string(key, *svalue, *result);
377: else if(Table* tvalue=value->get_table()) {
378: Form_table_value2string_info info(key, *result);
379: tvalue->for_each(form_table_value2string, &info);
380: } else
381: throw Exception(0,
382: new String(key, String::L_TAINTED),
383: "is %s, "HTTP_FORM_NAME" option value must either string or table", value->type());
1.180 paf 384: }
385: static const char* form2string(HashStringValue& form) {
386: String string;
387: form.for_each(form_value2string, &string);
1.181 paf 388: return string.cstr(String::L_UNSPECIFIED);
1.180 paf 389: }
1.154 paf 390: #ifndef DOXYGEN
391: struct File_read_http_result {
392: char *str; size_t length;
393: HashStringValue* headers;
394: };
395: #endif
1.155 paf 396: /// @todo build .cookies field. use ^file.tables.SET-COOKIES.menu{ for now
1.154 paf 397: static File_read_http_result file_read_http(Request_charsets& charsets,
398: const String& file_spec,
1.169 paf 399: bool as_text,
400: HashStringValue *options=0) {
1.154 paf 401: File_read_http_result result;
1.126 paf 402: char host[MAX_STRING];
1.129 paf 403: const char* uri;
1.193 ! paf 404: short port;
1.180 paf 405: const char* method="GET"; bool method_is_get;
406: HashStringValue* form=0;
407: const char* body_cstr=0;
1.126 paf 408: int timeout=2;
1.142 paf 409: bool fail_on_status_ne_200=true;
1.154 paf 410: Value* vheaders=0;
411: Charset *asked_remote_charset=0;
1.178 paf 412: const char* user_cstr=0;
413: const char* password_cstr=0;
1.126 paf 414:
1.127 paf 415: if(options) {
416: int valid_options=0;
1.177 paf 417: if(Value* vmethod=options->get(HTTP_METHOD_NAME)) {
1.127 paf 418: valid_options++;
1.154 paf 419: method=vmethod->as_string().cstr();
1.181 paf 420: }
421: if(Value* vform=options->get(HTTP_FORM_NAME)) {
1.180 paf 422: valid_options++;
423: form=vform->get_hash();
1.181 paf 424: }
425: if(Value* vbody=options->get(HTTP_BODY_NAME)) {
1.180 paf 426: valid_options++;
427: body_cstr=vbody->as_string().cstr(String::L_UNSPECIFIED);
1.181 paf 428: }
429: if(Value* vtimeout=options->get(HTTP_TIMEOUT_NAME)) {
1.127 paf 430: valid_options++;
431: timeout=vtimeout->as_int();
1.181 paf 432: }
433: if((vheaders=options->get(HTTP_HEADERS_NAME))) {
1.127 paf 434: valid_options++;
1.181 paf 435: }
436: if(Value* vany_status=options->get(HTTP_ANY_STATUS_NAME)) {
1.142 paf 437: valid_options++;
438: fail_on_status_ne_200=!vany_status->as_bool();
1.181 paf 439: }
440: if(Value* vcharset_name=options->get(HTTP_CHARSET_NAME)) {
1.154 paf 441: valid_options++;
442: asked_remote_charset=&::charsets.get(vcharset_name->as_string().
443: change_case(charsets.source(), String::CC_UPPER));
1.181 paf 444: }
445: if(Value* vuser=options->get(HTTP_USER)) {
1.178 paf 446: valid_options++;
447: user_cstr=vuser->as_string().cstr();
1.181 paf 448: }
449: if(Value* vpassword=options->get(HTTP_PASSWORD)) {
1.178 paf 450: valid_options++;
451: password_cstr=vpassword->as_string().cstr();
452: }
1.142 paf 453:
1.154 paf 454: if(valid_options!=options->count())
1.127 paf 455: throw Exception("parser.runtime",
456: 0,
457: "invalid option passed");
1.154 paf 458: }
459: if(!asked_remote_charset) // defaulting to $request:charset
460: asked_remote_charset=&charsets.source();
461:
1.180 paf 462: method_is_get=strcmp(method, "GET")==0;
463: if(method_is_get && body_cstr)
464: throw Exception("parser.runtime",
465: 0,
466: "you can not use $."HTTP_BODY_NAME" option with method GET");
467:
1.154 paf 468: //preparing request
469: String& connect_string=*new String;
470: // not in ^sql{... L_SQL ...} spirit, but closer to ^file::load one
471: connect_string.append(file_spec, String::L_URI); // tainted pieces -> URI pieces
472:
1.180 paf 473: String request_head_and_body;
1.154 paf 474: {
475: // influence URLencoding of tainted pieces to String::L_URI lang
476: Temp_client_charset temp(charsets, *asked_remote_charset);
477:
478: const char* connect_string_cstr=connect_string.cstr(String::L_UNSPECIFIED);
1.126 paf 479:
1.154 paf 480: const char* current=connect_string_cstr;
481: if(strncmp(current, "http://", 7)!=0)
482: throw Exception(0,
483: &connect_string,
484: "does not start with http://"); //never
485: current+=7;
486:
487: strncpy(host, current, sizeof(host)-1); host[sizeof(host)-1]=0;
488: char* host_uri=lsplit(host, '/');
489: uri=host_uri?current+(host_uri-1-host):"/";
490: char* port_cstr=lsplit(host, ':');
491: char* error_pos=0;
1.193 ! paf 492: port=port_cstr?(short)strtol(port_cstr, &error_pos, 0):80;
1.154 paf 493:
1.180 paf 494: if(strchr(uri, '?') && form)
495: throw Exception("parser.runtime",
496: 0,
497: "use either uri with ?params or $."HTTP_FORM_NAME" option");
498:
499: //making request head
500: String head;
501: head << method;
502: head << " " << uri;
503: if(form)
504: if(method_is_get)
1.181 paf 505: head << "?" << form2string(*form);
1.180 paf 506: head <<" HTTP/1.0" CRLF
1.154 paf 507: "host: "<< host << CRLF;
1.181 paf 508: if(form && !method_is_get) {
509: head << "content-type: application/x-www-form-urlencoded" CRLF;
510: body_cstr = form2string(*form);
511: }
1.178 paf 512:
1.179 paf 513: // http://www.ietf.org/rfc/rfc2617.txt
1.178 paf 514: if(const String* authorization_field_value=basic_authorization_field(user_cstr, password_cstr))
1.180 paf 515: head<<"authorization: "<<*authorization_field_value<<CRLF;
1.178 paf 516:
1.154 paf 517: bool user_agent_specified=false;
518: if(vheaders && !vheaders->is_string()) { // allow empty
519: if(HashStringValue *headers=vheaders->get_hash()) {
1.180 paf 520: Http_pass_header_info info={&charsets, &head, false};
1.154 paf 521: headers->for_each(http_pass_header, &info);
522: user_agent_specified=info.user_agent_specified;
523: } else
524: throw Exception("parser.runtime",
525: &connect_string,
526: "headers param must be hash");
527: };
528: if(!user_agent_specified) // defaulting
1.180 paf 529: head << "user-agent: " DEFAULT_USER_AGENT CRLF;
530:
531: if(body_cstr) {
532: // recode those pieces which are not in String::L_URI lang
533: // [those violating HTTP standard, but widly used]
534: body_cstr=Charset::transcode(
535: String::C(body_cstr, strlen(body_cstr)),
536: charsets.source(),
537: *asked_remote_charset);
1.181 paf 538:
539: head << "content-length: " << format(strlen(body_cstr), "%u") << CRLF;
1.180 paf 540: }
1.154 paf 541:
1.181 paf 542: const char* head_cstr=head.cstr(String::L_UNSPECIFIED);
543:
544: // recode those pieces which are not in String::L_URI lang
545: // [those violating HTTP standard, but widly used]
546: head_cstr=Charset::transcode(
547: String::C(head_cstr, strlen(head_cstr)),
548: charsets.source(),
549: *asked_remote_charset);
1.180 paf 550:
1.181 paf 551: // head + end of header
552: request_head_and_body << head_cstr << CRLF;
1.180 paf 553: // body
554: if(body_cstr)
555: request_head_and_body << body_cstr;
1.154 paf 556: }
1.126 paf 557:
558: //sending request
1.171 paf 559: char* response;
560: size_t response_size;
561: int status_code=http_request(response, response_size,
1.180 paf 562: host, port, request_head_and_body.cstr(),
1.142 paf 563: timeout, fail_on_status_ne_200);
1.126 paf 564:
565: //processing results
1.171 paf 566: char* raw_body; size_t raw_body_size;
1.191 paf 567: char* headers_end_at=strstr(response, "\n\n" /*change '2' below along!*/);
568: if(headers_end_at)
569: raw_body=headers_end_at+2;
570: else {
571: headers_end_at=strstr(response, CRLF CRLF /*change '4' below along!*/);
572: if(headers_end_at)
573: raw_body=headers_end_at+4;
574: else {
575: // yandex web server (http://localhost:17000)
576: // returns "\n\r\n"
577: headers_end_at=strstr(response, "\n\r\n" /*change '3' below along!*/);
578: if(!headers_end_at)
579: throw Exception("http.response",
580: &connect_string,
581: "bad response from host - no headers found");
582:
583: raw_body=headers_end_at+3;
584: }
585: }
586: raw_body_size=response_size-(raw_body-response);
1.171 paf 587:
588: *headers_end_at=0;
589: const String header_block(response, headers_end_at-response, true);
1.126 paf 590:
1.154 paf 591: ArrayString aheaders;
592: result.headers=new HashStringValue;
1.155 paf 593: VHash* vtables=new VHash;
1.177 paf 594: result.headers->put(HTTP_TABLES_NAME, vtables);
1.155 paf 595: HashStringValue& tables=vtables->hash();
596:
1.154 paf 597: size_t pos_after=0;
598: header_block.split(aheaders, pos_after, CRLF);
1.126 paf 599:
600: //processing headers
1.154 paf 601: size_t aheaders_count=aheaders.count();
602: Charset* real_remote_charset=0; // undetected, yet
603: for(size_t i=1; i<aheaders_count; i++) {
604: const String& line=*aheaders.get(i);
1.171 paf 605: size_t pos=line.pos(": ", 2);
1.154 paf 606: if(pos==STRING_NOT_FOUND || pos<1)
1.126 paf 607: throw Exception("http.response",
1.154 paf 608: &connect_string,
609: "bad response from host - bad header \"%s\"", line.cstr());
1.156 paf 610: const String::Body HEADER_NAME=
1.154 paf 611: line.mid(0, pos).change_case(charsets.source(), String::CC_UPPER);
612: const String& header_value=line.mid(pos+2, line.length());
1.192 paf 613: if(as_text && HEADER_NAME=="CONTENT-TYPE")
1.154 paf 614: real_remote_charset=detect_charset(charsets.source(), header_value);
1.155 paf 615:
616: // tables
617: {
618: Value *valready=(Value *)tables.get(HEADER_NAME);
619: bool existed=valready!=0;
620: Table *table;
621: if(existed) {
622: // second+ appearence
623: table=valready->get_table();
624: } else {
625: // first appearence
626: Table::columns_type columns =new ArrayString(1);
627: *columns+=new String("value");
628: table=new Table(columns);
629: }
630: // this string becomes next row
631: ArrayString& row=*new ArrayString(1);
632: row+=&header_value;
633: *table+=&row;
634: // not existed before? add it
635: if(!existed)
636: tables.put(HEADER_NAME, new VTable(table));
637: }
638:
1.154 paf 639: result.headers->put(HEADER_NAME, new VString(header_value));
1.126 paf 640: }
641:
642: // output response
1.171 paf 643: String::C real_body=String::C(raw_body, raw_body_size);
1.192 paf 644: if(as_text && raw_body_size) { // must be checked because transcode returns CONST string in case length==0, which contradicts hacking few lines below
645: // defaulting to used-asked charset [it's never empty!]
646: if(!real_remote_charset)
647: real_remote_charset=asked_remote_charset;
1.169 paf 648: real_body=Charset::transcode(real_body, *real_remote_charset, charsets.source());
1.192 paf 649: }
1.154 paf 650:
651: result.str=const_cast<char *>(real_body.str); // hacking a little
652: result.length=real_body.length;
653: result.headers->put(file_status_name, new VInt(status_code));
654: return result;
1.34 paf 655: }
1.123 paf 656:
1.154 paf 657: #endif
658:
1.123 paf 659: #ifndef DOXYGEN
660: struct File_read_action_info {
1.154 paf 661: char **data; size_t *data_size;
1.188 paf 662: char* buf; size_t offset; size_t count;
1.126 paf 663: };
1.123 paf 664: #endif
1.154 paf 665: static void file_read_action(
666: struct stat& finfo,
667: int f,
1.166 paf 668: const String& file_spec, const char* /*fname*/, bool as_text,
1.154 paf 669: void *context) {
1.126 paf 670: File_read_action_info& info=*static_cast<File_read_action_info *>(context);
1.188 paf 671: size_t to_read_size=info.count;
672: if(!to_read_size)
673: to_read_size=(size_t)finfo.st_size;
674: assert( !(info.buf && as_text) );
675: if(to_read_size) {
676: if(info.offset)
677: lseek(f, info.offset, SEEK_SET);
678: *info.data=info.buf
679: ? info.buf
680: : new(PointerFreeGC) char[to_read_size+(as_text?1:0)];
1.126 paf 681: *info.data_size=(size_t)read(f, *info.data, to_read_size);
1.123 paf 682:
683: if(ssize_t(*info.data_size)<0 || *info.data_size>to_read_size)
1.126 paf 684: throw Exception(0,
1.123 paf 685: &file_spec,
1.173 paf 686: "read failed: actually read %u bytes count not in [0..%u] valid range",
1.126 paf 687: *info.data_size, to_read_size);
1.123 paf 688: } else { // empty file
689: if(as_text) {
1.154 paf 690: *info.data=new(PointerFreeGC) char[1];
1.123 paf 691: *(char*)(*info.data)=0;
692: } else
693: *info.data=0;
694: *info.data_size=0;
695: return;
696: }
1.126 paf 697: }
1.154 paf 698: File_read_result file_read(Request_charsets& charsets, const String& file_spec,
699: bool as_text, HashStringValue *params,
1.188 paf 700: bool fail_on_read_problem,
701: char* buf, size_t offset, size_t count) {
1.167 paf 702: File_read_result result={false, 0, 0, 0};
1.154 paf 703: #ifdef PA_HTTP
704: if(file_spec.starts_with("http://")) {
1.126 paf 705: // fail on read problem
1.169 paf 706: File_read_http_result http=file_read_http(charsets, file_spec, as_text, params);
1.154 paf 707: result.success=true;
708: result.str=http.str;
709: result.length=http.length;
710: result.headers=http.headers;
1.126 paf 711: } else {
1.154 paf 712: #endif
1.161 paf 713: if(params && params->count())
714: throw Exception("parser.runtime",
715: 0,
716: "invalid option passed");
717:
1.188 paf 718: File_read_action_info info={&result.str, &result.length,
719: buf, offset, count};
1.154 paf 720: result.success=file_read_action_under_lock(file_spec,
1.126 paf 721: "read", file_read_action, &info,
722: as_text, fail_on_read_problem);
1.154 paf 723: #ifdef PA_HTTP
1.126 paf 724: }
1.154 paf 725: #endif
1.123 paf 726:
1.154 paf 727: if(result.success && as_text) {
1.131 paf 728: // UTF-8 signature: EF BB BF
1.154 paf 729: if(result.length>=3) {
730: char *in=(char *)result.str;
1.159 paf 731: if(strncmp(in, "\xEF\xBB\xBF", 3)==0) {
1.154 paf 732: result.str=in+3; result.length-=3;// skip prefix
1.131 paf 733: }
734: }
735:
1.154 paf 736: fix_line_breaks((char *)(result.str), result.length);
1.123 paf 737: }
1.126 paf 738:
739: return result;
1.123 paf 740: }
741:
1.154 paf 742: #ifdef PA_SAFE_MODE
743: void check_safe_mode(struct stat finfo, const String& file_spec, const char* fname) {
744: if(finfo.st_uid/*foreign?*/!=geteuid()
745: && finfo.st_gid/*foreign?*/!=getegid())
746: throw Exception("parser.runtime",
747: &file_spec,
748: "parser is in safe mode: "
749: "reading files of foreign group and user disabled "
750: "[recompile parser with --disable-safe-mode configure option], "
751: "actual filename '%s', "
752: "fuid(%d)!=euid(%d) or fgid(%d)!=egid(%d)",
753: fname,
754: finfo.st_uid, geteuid(),
755: finfo.st_gid, getegid());
756: }
757: #endif
1.149 paf 758:
1.154 paf 759: bool file_read_action_under_lock(const String& file_spec,
1.126 paf 760: const char* action_name, File_read_action action, void *context,
761: bool as_text,
1.123 paf 762: bool fail_on_read_problem) {
1.154 paf 763: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.33 paf 764: int f;
765:
766: // first open, next stat:
1.45 paf 767: // directory update of NTFS hard links performed on open.
1.33 paf 768: // ex:
769: // a.html:^test[] and b.html hardlink to a.html
770: // user inserts ! before ^test in a.html
1.126 paf 771: // directory entry of b.html in NTFS not updated at once,
1.35 paf 772: // they delay update till open, so we would receive "!^test[" string
773: // if would do stat, next open.
1.123 paf 774: // later: it seems, even this does not help sometimes
1.98 paf 775: if((f=open(fname, O_RDONLY|(as_text?_O_TEXT:_O_BINARY)))>=0) {
1.123 paf 776: try {
1.162 paf 777: if(pa_lock_shared_blocking(f)!=0)
1.126 paf 778: throw Exception("file.lock",
1.123 paf 779: &file_spec,
780: "shared lock failed: %s (%d), actual filename '%s'",
1.154 paf 781: strerror(errno), errno, fname);
1.123 paf 782:
1.124 paf 783: struct stat finfo;
784: if(stat(fname, &finfo)!=0)
785: throw Exception("file.missing", // hardly possible: we just opened it OK
786: &file_spec,
787: "stat failed: %s (%d), actual filename '%s'",
1.154 paf 788: strerror(errno), errno, fname);
1.124 paf 789:
1.140 paf 790: #ifdef PA_SAFE_MODE
1.149 paf 791: check_safe_mode(finfo, file_spec, fname);
1.105 paf 792: #endif
1.32 paf 793:
1.154 paf 794: action(finfo, f, file_spec, fname, as_text, context);
1.123 paf 795: } catch(...) {
1.162 paf 796: pa_unlock(f);close(f);
1.123 paf 797: if(fail_on_read_problem)
1.154 paf 798: rethrow;
1.123 paf 799: return false;
800: }
1.87 paf 801:
1.162 paf 802: pa_unlock(f);close(f);
1.72 parser 803: return true;
1.118 paf 804: } else {
805: if(fail_on_read_problem)
1.126 paf 806: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.118 paf 807: &file_spec,
1.123 paf 808: "%s failed: %s (%d), actual filename '%s'",
1.154 paf 809: action_name, strerror(errno), errno, fname);
1.118 paf 810: return false;
811: }
1.8 paf 812: }
813:
1.63 parser 814: static void create_dir_for_file(const String& file_spec) {
815: size_t pos_after=1;
1.154 paf 816: size_t pos_before;
817: while((pos_before=file_spec.pos('/', pos_after))!=STRING_NOT_FOUND) {
818: mkdir(file_spec.mid(0, pos_before).cstr(String::L_FILE_SPEC), 0775);
1.63 parser 819: pos_after=pos_before+1;
820: }
821: }
822:
1.98 paf 823: bool file_write_action_under_lock(
1.28 paf 824: const String& file_spec,
1.126 paf 825: const char* action_name, File_write_action action, void *context,
826: bool as_text,
827: bool do_append,
828: bool do_block,
1.110 paf 829: bool fail_on_lock_problem) {
1.154 paf 830: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.28 paf 831: int f;
1.80 paf 832: if(access(fname, W_OK)!=0) // no
1.126 paf 833: create_dir_for_file(file_spec);
1.50 paf 834:
1.80 paf 835: if((f=open(fname,
836: O_CREAT|O_RDWR
837: |(as_text?_O_TEXT:_O_BINARY)
1.138 paf 838: |(do_append?O_APPEND:PA_O_TRUNC), 0664))>=0) {
1.162 paf 839: if((do_block?pa_lock_exclusive_blocking(f):pa_lock_exclusive_nonblocking(f))!=0) {
1.126 paf 840: Exception e("file.lock",
1.110 paf 841: &file_spec,
842: "shared lock failed: %s (%d), actual filename '%s'",
1.154 paf 843: strerror(errno), errno, fname);
1.126 paf 844: close(f);
1.110 paf 845: if(fail_on_lock_problem)
846: throw e;
1.98 paf 847: return false;
848: }
1.96 paf 849:
1.158 paf 850: try {
1.126 paf 851: action(f, context);
1.158 paf 852: } catch(...) {
1.138 paf 853: #ifdef HAVE_FTRUNCATE
1.104 paf 854: if(!do_append)
1.125 paf 855: ftruncate(f, lseek(f, 0, SEEK_CUR)); // one can not use O_TRUNC, read lower
1.138 paf 856: #endif
1.162 paf 857: pa_unlock(f);close(f);
1.154 paf 858: rethrow;
1.158 paf 859: }
1.80 paf 860:
1.138 paf 861: #ifdef HAVE_FTRUNCATE
1.104 paf 862: if(!do_append)
1.125 paf 863: 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 864: #endif
1.162 paf 865: pa_unlock(f);close(f);
1.98 paf 866: return true;
1.80 paf 867: } else
1.126 paf 868: throw Exception(errno==EACCES?"file.access":0,
1.80 paf 869: &file_spec,
1.96 paf 870: "%s failed: %s (%d), actual filename '%s'",
1.154 paf 871: action_name, strerror(errno), errno, fname);
1.96 paf 872: // here should be nothing, see rethrow above
873: }
874:
875: #ifndef DOXYGEN
876: struct File_write_action_info {
1.154 paf 877: const char* str; size_t length;
1.126 paf 878: };
1.96 paf 879: #endif
880: static void file_write_action(int f, void *context) {
1.126 paf 881: File_write_action_info& info=*static_cast<File_write_action_info *>(context);
1.154 paf 882: if(info.length) {
883: int written=write(f, info.str, info.length);
1.116 paf 884: if(written<0)
1.126 paf 885: throw Exception(0,
886: 0,
887: "write failed: %s (%d)", strerror(errno), errno);
1.113 paf 888: }
1.96 paf 889: }
890: void file_write(
891: const String& file_spec,
1.154 paf 892: const char* data, size_t size,
1.126 paf 893: bool as_text,
1.96 paf 894: bool do_append) {
1.126 paf 895: File_write_action_info info={data, size};
1.98 paf 896: file_write_action_under_lock(
1.154 paf 897: file_spec,
898: "write", file_write_action, &info,
899: as_text,
900: do_append);
1.30 paf 901: }
902:
1.63 parser 903: // throws nothing! [this is required in file_move & file_delete]
1.50 paf 904: static void rmdir(const String& file_spec, size_t pos_after) {
1.154 paf 905: size_t pos_before;
906: if((pos_before=file_spec.pos('/', pos_after))!=STRING_NOT_FOUND)
1.126 paf 907: rmdir(file_spec, pos_before+1);
1.50 paf 908:
1.154 paf 909: rmdir(file_spec.mid(0, pos_after-1/* / */).cstr(String::L_FILE_SPEC));
1.50 paf 910: }
1.164 paf 911: bool file_delete(const String& file_spec, bool fail_on_problem) {
1.154 paf 912: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.54 parser 913: if(unlink(fname)!=0)
1.164 paf 914: if(fail_on_problem)
1.126 paf 915: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.93 paf 916: &file_spec,
917: "unlink failed: %s (%d), actual filename '%s'",
1.154 paf 918: strerror(errno), errno, fname);
1.93 paf 919: else
920: return false;
1.50 paf 921:
1.126 paf 922: rmdir(file_spec, 1);
1.93 paf 923: return true;
1.60 parser 924: }
1.95 paf 925: void file_move(const String& old_spec, const String& new_spec) {
1.154 paf 926: const char* old_spec_cstr=old_spec.cstr(String::L_FILE_SPEC);
927: const char* new_spec_cstr=new_spec.cstr(String::L_FILE_SPEC);
1.63 parser 928:
1.126 paf 929: create_dir_for_file(new_spec);
1.63 parser 930:
1.60 parser 931: if(rename(old_spec_cstr, new_spec_cstr)!=0)
1.126 paf 932: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.60 parser 933: &old_spec,
934: "rename failed: %s (%d), actual filename '%s' to '%s'",
1.154 paf 935: strerror(errno), errno, old_spec_cstr, new_spec_cstr);
1.63 parser 936:
1.126 paf 937: rmdir(old_spec, 1);
1.31 paf 938: }
939:
1.51 paf 940:
1.126 paf 941: bool entry_exists(const char* fname, struct stat *afinfo) {
1.118 paf 942: struct stat lfinfo;
943: bool result=stat(fname, &lfinfo)==0;
944: if(afinfo)
945: *afinfo=lfinfo;
946: return result;
1.119 paf 947: }
948:
949: bool entry_exists(const String& file_spec) {
1.154 paf 950: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.126 paf 951: return entry_exists(fname, 0);
1.118 paf 952: }
953:
1.51 paf 954: static bool entry_readable(const String& file_spec, bool need_dir) {
1.154 paf 955: char* fname=file_spec.cstrm(String::L_FILE_SPEC);
1.120 paf 956: if(need_dir) {
1.126 paf 957: size_t size=strlen(fname);
1.120 paf 958: while(size) {
1.126 paf 959: char c=fname[size-1];
1.120 paf 960: if(c=='/' || c=='\\')
961: fname[--size]=0;
962: else
963: break;
964: }
965: }
1.51 paf 966: struct stat finfo;
1.118 paf 967: if(access(fname, R_OK)==0 && entry_exists(fname, &finfo)) {
1.109 paf 968: bool is_dir=(finfo.st_mode&S_IFDIR) != 0;
1.51 paf 969: return is_dir==need_dir;
970: }
971: return false;
972: }
1.31 paf 973: bool file_readable(const String& file_spec) {
1.126 paf 974: return entry_readable(file_spec, false);
1.51 paf 975: }
976: bool dir_readable(const String& file_spec) {
1.126 paf 977: return entry_readable(file_spec, true);
1.65 parser 978: }
1.154 paf 979: const String* file_readable(const String& path, const String& name) {
980: String& result=*new String(path);
981: result << "/";
982: result << name;
983: return file_readable(result)?&result:0;
1.43 paf 984: }
985: bool file_executable(const String& file_spec) {
1.154 paf 986: return access(file_spec.cstr(String::L_FILE_SPEC), X_OK)==0;
1.44 paf 987: }
988:
1.64 parser 989: bool file_stat(const String& file_spec,
1.58 parser 990: size_t& rsize,
1.126 paf 991: time_t& ratime,
992: time_t& rmtime,
993: time_t& rctime,
1.64 parser 994: bool fail_on_read_problem) {
1.154 paf 995: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
996: struct stat finfo;
1.44 paf 997: if(stat(fname, &finfo)!=0)
1.64 parser 998: if(fail_on_read_problem)
1.126 paf 999: throw Exception("file.missing",
1.67 parser 1000: &file_spec,
1001: "getting file size failed: %s (%d), real filename '%s'",
1.154 paf 1002: strerror(errno), errno, fname);
1.64 parser 1003: else
1004: return false;
1.58 parser 1005: rsize=finfo.st_size;
1006: ratime=finfo.st_atime;
1007: rmtime=finfo.st_mtime;
1008: rctime=finfo.st_ctime;
1.64 parser 1009: return true;
1.18 paf 1010: }
1011:
1.126 paf 1012: char* getrow(char* *row_ref, char delim) {
1013: char* result=*row_ref;
1.8 paf 1014: if(result) {
1.126 paf 1015: *row_ref=strchr(result, delim);
1.8 paf 1016: if(*row_ref)
1017: *((*row_ref)++)=0;
1018: else if(!*result)
1019: return 0;
1020: }
1021: return result;
1022: }
1023:
1.126 paf 1024: char* lsplit(char* string, char delim) {
1.23 paf 1025: if(string) {
1.126 paf 1026: char* v=strchr(string, delim);
1.8 paf 1027: if(v) {
1028: *v=0;
1029: return v+1;
1030: }
1031: }
1032: return 0;
1033: }
1034:
1.126 paf 1035: char* lsplit(char* *string_ref, char delim) {
1036: char* result=*string_ref;
1037: char* next=lsplit(*string_ref, delim);
1.8 paf 1038: *string_ref=next;
1039: return result;
1.9 paf 1040: }
1041:
1.126 paf 1042: char* rsplit(char* string, char delim) {
1.18 paf 1043: if(string) {
1.126 paf 1044: char* v=strrchr(string, delim);
1.18 paf 1045: if(v) {
1.9 paf 1046: *v=0;
1047: return v+1;
1048: }
1049: }
1050: return NULL;
1.10 paf 1051: }
1052:
1.37 paf 1053: /// @todo less stupid type detection
1.154 paf 1054: const char* format(double value, char* fmt) {
1.126 paf 1055: char local_buf[MAX_NUMBER];
1.108 paf 1056: size_t size;
1057:
1.10 paf 1058: if(fmt)
1059: if(strpbrk(fmt, "diouxX"))
1060: if(strpbrk(fmt, "ouxX"))
1.126 paf 1061: size=snprintf(local_buf, sizeof(local_buf), fmt, (uint)value);
1.10 paf 1062: else
1.126 paf 1063: size=snprintf(local_buf, sizeof(local_buf), fmt, (int)value);
1.10 paf 1064: else
1.126 paf 1065: size=snprintf(local_buf, sizeof(local_buf), fmt, value);
1.10 paf 1066: else
1.126 paf 1067: size=snprintf(local_buf, sizeof(local_buf), "%d", (int)value);
1.10 paf 1068:
1.154 paf 1069: return pa_strdup(local_buf, size);
1.12 paf 1070: }
1071:
1.36 paf 1072: size_t stdout_write(const void *buf, size_t size) {
1.12 paf 1073: #ifdef WIN32
1.187 paf 1074: size_t to_write = size;
1.12 paf 1075: do{
1.154 paf 1076: int chunk_written=fwrite(buf, 1, min((size_t)8*0x400, size), stdout);
1.12 paf 1077: if(chunk_written<=0)
1078: break;
1079: size-=chunk_written;
1.36 paf 1080: buf=((const char*)buf)+chunk_written;
1.126 paf 1081: } while(size>0);
1.12 paf 1082:
1.187 paf 1083: return to_write-size;
1.12 paf 1084: #else
1.126 paf 1085: return fwrite(buf, 1, size, stdout);
1.12 paf 1086: #endif
1.2 paf 1087: }
1.14 paf 1088:
1.154 paf 1089: char* unescape_chars(const char* cp, int len) {
1090: char* s=new(PointerFreeGC) char[len + 1];
1.14 paf 1091: enum EscapeState {
1.33 paf 1092: EscapeRest,
1093: EscapeFirst,
1.14 paf 1094: EscapeSecond
1095: } escapeState=EscapeRest;
1.193 ! paf 1096: uchar escapedValue=0;
1.14 paf 1097: int srcPos=0;
1098: int dstPos=0;
1099: while(srcPos < len) {
1.193 ! paf 1100: uchar ch=(uchar)cp[srcPos];
1.14 paf 1101: switch(escapeState) {
1102: case EscapeRest:
1103: if(ch=='%') {
1104: escapeState=EscapeFirst;
1105: } else if(ch=='+') {
1.126 paf 1106: s[dstPos++]=' ';
1.14 paf 1107: } else {
1108: s[dstPos++]=ch;
1109: }
1110: break;
1111: case EscapeFirst:
1.193 ! paf 1112: escapedValue=(uchar)(hex_value[ch] << 4);
1.14 paf 1113: escapeState=EscapeSecond;
1114: break;
1115: case EscapeSecond:
1.126 paf 1116: escapedValue +=hex_value[ch];
1.14 paf 1117: s[dstPos++]=escapedValue;
1118: escapeState=EscapeRest;
1119: break;
1120: }
1.126 paf 1121: srcPos++;
1.14 paf 1122: }
1123: s[dstPos]=0;
1124: return s;
1.24 paf 1125: }
1126:
1127: #ifdef WIN32
1.126 paf 1128: void back_slashes_to_slashes(char* s) {
1.24 paf 1129: if(s)
1130: for(; *s; s++)
1131: if(*s=='\\')
1.126 paf 1132: *s='/';
1.24 paf 1133: }
1.42 paf 1134: /*
1.126 paf 1135: void slashes_to_back_slashes(char* s) {
1.42 paf 1136: if(s)
1137: for(; *s; s++)
1138: if(*s=='/')
1.126 paf 1139: *s='\\';
1.42 paf 1140: }
1141: */
1.24 paf 1142: #endif
1.41 paf 1143:
1.126 paf 1144: bool StrEqNc(const char* s1, const char* s2, bool strict) {
1.41 paf 1145: while(true) {
1146: if(!(*s1)) {
1147: if(!(*s2))
1148: return true;
1149: else
1150: return !strict;
1151: } else if(!(*s2))
1152: return !strict;
1.189 paf 1153: if(isalpha((unsigned char)*s1)) {
1.190 paf 1154: if(tolower((unsigned char)*s1) !=tolower((unsigned char)*s2))
1.41 paf 1155: return false;
1156: } else if((*s1) !=(*s2))
1157: return false;
1.126 paf 1158: s1++;
1159: s2++;
1.41 paf 1160: }
1.57 parser 1161: }
1162:
1.84 paf 1163: static bool isLeap(int year) {
1.57 parser 1164: return !(
1165: (year % 4) || ((year % 400) && !(year % 100))
1.126 paf 1166: );
1.57 parser 1167: }
1168:
1169: int getMonthDays(int year, int month) {
1170: int monthDays[]={
1.126 paf 1171: 31,
1172: isLeap(year) ? 29 : 28,
1173: 31,
1174: 30,
1175: 31,
1176: 30,
1177: 31,
1178: 31,
1179: 30,
1180: 31,
1181: 30,
1.57 parser 1182: 31
1.126 paf 1183: };
1184: return monthDays[month];
1.41 paf 1185: }
1.69 parser 1186:
1.126 paf 1187: void remove_crlf(char* start, char* end) {
1188: for(char* p=start; p<end; p++)
1.69 parser 1189: switch(*p) {
1.126 paf 1190: case '\n': *p='|'; break;
1191: case '\r': *p=' '; break;
1.69 parser 1192: }
1.91 paf 1193: }
1194:
1195:
1196: /// must be last in this file
1197: #undef vsnprintf
1.126 paf 1198: int __vsnprintf(char* b, size_t s, const char* f, va_list l) {
1.91 paf 1199: if(!s)
1200: return 0;
1201:
1202: int r;
1203: // note: on win32& maybe somewhere else
1204: // vsnprintf do not writes terminating 0 in 'buffer full' case, reducing
1205: --s;
1.172 paf 1206:
1207: // clients do not check for negative 's', feature: ignore such prints
1208: if((ssize_t)s<0)
1209: return 0;
1210:
1.91 paf 1211: #if _MSC_VER
1212: /*
1213: win32:
1214: mk:@MSITStore:C:\Program%20Files\Microsoft%20Visual%20Studio\MSDN\2001APR\1033\vccore.chm::/html/_crt__vsnprintf.2c_._vsnwprintf.htm
1215:
1.154 paf 1216: if the number of bytes to write exceeds buffer, then count bytes are written and Ö1 is returned
1.91 paf 1217: */
1.126 paf 1218: r=_vsnprintf(b, s, f, l);
1.91 paf 1219: if(r<0)
1220: r=s;
1221: #else
1.126 paf 1222: r=vsnprintf(b, s, f, l);
1.91 paf 1223: /*
1224: solaris:
1225: man vsnprintf
1226:
1227: The snprintf() function returns the number of characters
1228: formatted, that is, the number of characters that would have
1229: been written to the buffer if it were large enough. If the
1230: value of n is 0 on a call to snprintf(), an unspecified
1231: value less than 1 is returned.
1232: */
1233:
1234: if(r<0)
1235: r=0;
1.167 paf 1236: else if((size_t)r>s)
1.91 paf 1237: r=s;
1238: #endif
1239: b[r]=0;
1240: return r;
1241: }
1242:
1.126 paf 1243: int __snprintf(char* b, size_t s, const char* f, ...) {
1.91 paf 1244: va_list l;
1.126 paf 1245: va_start(l, f);
1246: int r=__vsnprintf(b, s, f, l);
1247: va_end(l);
1.91 paf 1248: return r;
1.178 paf 1249: }
1250:
1251: /* mime64 functions are from libgmime[http://spruce.sourceforge.net/gmime/] lib */
1252: /*
1253: * Authors: Michael Zucchi <notzed@helixcode.com>
1254: * Jeffrey Stedfast <fejj@helixcode.com>
1255: *
1256: * Copyright 2000 Helix Code, Inc. (www.helixcode.com)
1257: *
1258: * This program is free software; you can redistribute it and/or modify
1259: * it under the terms of the GNU General Public License as published by
1260: * the Free Software Foundation; either version 2 of the License, or
1261: * (at your option) any later version.
1262: *
1263: * This program is distributed in the hope that it will be useful,
1264: * but WITHOUT ANY WARRANTY; without even the implied warranty of
1265: * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
1266: * GNU General Public License for more details.
1267: *
1268: * You should have received a copy of the GNU General Public License
1269: * along with this program; if not, write to the Free Software
1270: * Foundation, Inc., 59 Temple Street #330, Boston, MA 02111-1307, USA.
1271: *
1272: */
1273: static char *base64_alphabet =
1274: "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";
1275:
1276: /**
1277: * g_mime_utils_base64_encode_step:
1278: * @in: input stream
1279: * @inlen: length of the input
1280: * @out: output string
1281: * @state: holds the number of bits that are stored in @save
1282: * @save: leftover bits that have not yet been encoded
1283: *
1284: * Base64 encodes a chunk of data. Performs an 'encode step', only
1285: * encodes blocks of 3 characters to the output at a time, saves
1286: * left-over state in state and save (initialise to 0 on first
1287: * invocation).
1288: *
1289: * Returns the number of bytes encoded.
1290: **/
1291: static size_t
1292: g_mime_utils_base64_encode_step (const unsigned char *in, size_t inlen, unsigned char *out, int *state, int *save)
1293: {
1.186 paf 1294: register const unsigned char *inptr;
1.178 paf 1295: register unsigned char *outptr;
1296:
1297: if (inlen <= 0)
1298: return 0;
1299:
1300: inptr = in;
1301: outptr = out;
1302:
1303: if (inlen + ((unsigned char *)save)[0] > 2) {
1304: const unsigned char *inend = in + inlen - 2;
1305: register int c1 = 0, c2 = 0, c3 = 0;
1306: register int already;
1307:
1308: already = *state;
1309:
1310: switch (((char *)save)[0]) {
1311: case 1: c1 = ((unsigned char *)save)[1]; goto skip1;
1312: case 2: c1 = ((unsigned char *)save)[1];
1313: c2 = ((unsigned char *)save)[2]; goto skip2;
1314: }
1315:
1316: /* yes, we jump into the loop, no i'm not going to change it, its beautiful! */
1317: while (inptr < inend) {
1318: c1 = *inptr++;
1319: skip1:
1320: c2 = *inptr++;
1321: skip2:
1322: c3 = *inptr++;
1323: *outptr++ = base64_alphabet [c1 >> 2];
1324: *outptr++ = base64_alphabet [(c2 >> 4) | ((c1 & 0x3) << 4)];
1325: *outptr++ = base64_alphabet [((c2 & 0x0f) << 2) | (c3 >> 6)];
1326: *outptr++ = base64_alphabet [c3 & 0x3f];
1327: /* this is a bit ugly ... */
1328: if ((++already) >= 19) {
1329: *outptr++ = '\n';
1330: already = 0;
1331: }
1332: }
1333:
1334: ((unsigned char *)save)[0] = 0;
1335: inlen = 2 - (inptr - inend);
1336: *state = already;
1337: }
1338:
1339: //d(printf ("state = %d, inlen = %d\n", (int)((char *)save)[0], inlen));
1340:
1341: if (inlen > 0) {
1342: register char *saveout;
1343:
1344: /* points to the slot for the next char to save */
1345: saveout = & (((char *)save)[1]) + ((char *)save)[0];
1346:
1347: /* inlen can only be 0 1 or 2 */
1348: switch (inlen) {
1349: case 2: *saveout++ = *inptr++;
1350: case 1: *saveout++ = *inptr++;
1351: }
1352: ((char *)save)[0] += inlen;
1353: }
1354:
1355: /*d(printf ("mode = %d\nc1 = %c\nc2 = %c\n",
1356: (int)((char *)save)[0],
1357: (int)((char *)save)[1],
1358: (int)((char *)save)[2]));*/
1359:
1360: return (outptr - out);
1361: }
1362:
1363: /**
1364: * g_mime_utils_base64_encode_close:
1365: * @in: input stream
1366: * @inlen: length of the input
1367: * @out: output string
1368: * @state: holds the number of bits that are stored in @save
1369: * @save: leftover bits that have not yet been encoded
1370: *
1371: * Base64 encodes the input stream to the output stream. Call this
1372: * when finished encoding data with g_mime_utils_base64_encode_step to
1373: * flush off the last little bit.
1374: *
1375: * Returns the number of bytes encoded.
1376: **/
1377: static size_t
1378: g_mime_utils_base64_encode_close (const unsigned char *in, size_t inlen, unsigned char *out, int *state, int *save)
1379: {
1380: unsigned char *outptr = out;
1381: int c1, c2;
1382:
1383: if (inlen > 0)
1384: outptr += g_mime_utils_base64_encode_step (in, inlen, outptr, state, save);
1385:
1386: c1 = ((unsigned char *)save)[1];
1387: c2 = ((unsigned char *)save)[2];
1388:
1389: switch (((unsigned char *)save)[0]) {
1390: case 2:
1391: outptr[2] = base64_alphabet [(c2 & 0x0f) << 2];
1392: goto skip;
1393: case 1:
1394: outptr[2] = '=';
1395: skip:
1396: outptr[0] = base64_alphabet [c1 >> 2];
1397: outptr[1] = base64_alphabet [c2 >> 4 | ((c1 & 0x3) << 4)];
1398: outptr[3] = '=';
1399: outptr += 4;
1400: break;
1401: }
1402:
1403: *outptr++ = 0;
1404:
1405: *save = 0;
1406: *state = 0;
1407:
1408: return (outptr - out);
1409: }
1410:
1411: char* pa_base64(const char *in, size_t len)
1412: {
1413: /* wont go to more than 2x size (overly conservative) */
1414: char* result=new(PointerFreeGC) char[len * 2 + 6];
1415: int state=0;
1416: int save=0;
1.183 paf 1417: #ifndef NDEBUG
1418: size_t filled=
1419: #endif
1420: g_mime_utils_base64_encode_close ((const unsigned char*)in, len,
1.178 paf 1421: (unsigned char*)result, &state, &save);
1422: assert(filled <= len * 2 + 6);
1423:
1424: return result;
1.98 paf 1425: }
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