Annotation of parser3/src/main/pa_common.C, revision 1.202
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:
1.202 ! paf 8: static const char * const IDENT_COMMON_C="$Date: 2004/12/10 08:37:53 $";
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"
1.195 paf 24: #include "pa_socks.h"
1.154 paf 25:
26: #ifdef CYGWIN
27: #define _GNU_H_WINDOWS32_SOCKETS
28: // for PASCAL
29: #include <windows.h>
30: // SOCKET
31: typedef u_int SOCKET;
32: int PASCAL closesocket(SOCKET);
33: #else
34: # if defined(WIN32)
35: # include <windows.h>
36: # else
37: # define closesocket close
38: # endif
39: #endif
1.1 paf 40:
1.126 paf 41: #else
1.154 paf 42:
43: # if defined(WIN32)
44: # include <windows.h>
45: # endif
46:
1.98 paf 47: #endif
48:
1.93 paf 49: // some maybe-undefined constants
50:
1.82 paf 51: #ifndef _O_TEXT
52: # define _O_TEXT 0
53: #endif
54: #ifndef _O_BINARY
55: # define _O_BINARY 0
1.47 paf 56: #endif
1.80 paf 57:
1.138 paf 58: #ifdef HAVE_FTRUNCATE
59: # define PA_O_TRUNC 0
60: #else
61: # ifdef _O_TRUNC
62: # define PA_O_TRUNC _O_TRUNC
63: # else
64: # error you must have either ftruncate function or _O_TRUNC bit declared
65: # endif
1.154 paf 66: #endif
1.176 paf 67:
68: # ifndef INADDR_NONE
69: # define INADDR_NONE ((ulong) -1)
70: # endif
1.154 paf 71:
72: // defines for globals
73:
74: #define FILE_STATUS_NAME "status"
75:
76: // globals
77:
78: const String file_status_name(FILE_STATUS_NAME);
79:
1.178 paf 80: // defines
1.154 paf 81:
82: #define HTTP_METHOD_NAME "method"
1.180 paf 83: #define HTTP_FORM_NAME "form"
84: #define HTTP_BODY_NAME "body"
1.154 paf 85: #define HTTP_TIMEOUT_NAME "timeout"
86: #define HTTP_HEADERS_NAME "headers"
87: #define HTTP_ANY_STATUS_NAME "any-status"
88: #define HTTP_CHARSET_NAME "charset"
1.155 paf 89: #define HTTP_TABLES_NAME "tables"
1.178 paf 90: #define HTTP_USER "user"
91: #define HTTP_PASSWORD "password"
1.154 paf 92:
93: // defines
94:
1.127 paf 95: #define DEFAULT_USER_AGENT "parser3"
96:
1.154 paf 97: // functions
1.127 paf 98:
1.154 paf 99: void fix_line_breaks(char *str, size_t& length) {
1.87 paf 100: //_asm int 3;
1.154 paf 101: const char* const eob=str+length;
102: char* dest=str;
1.72 parser 103: // fix DOS: \r\n -> \n
104: // fix Macintosh: \r -> \n
1.154 paf 105: char* bol=str;
1.137 paf 106: while(char* eol=(char*)memchr(bol, '\r', eob -bol)) {
1.72 parser 107: size_t len=eol-bol;
108: if(dest!=bol)
1.126 paf 109: memcpy(dest, bol, len);
1.72 parser 110: dest+=len;
1.126 paf 111: *dest++='\n';
1.72 parser 112:
1.126 paf 113: if(&eol[1]<eob && eol[1]=='\n') { // \r, \n = DOS
1.72 parser 114: bol=eol+2;
1.154 paf 115: length--;
1.126 paf 116: } else // \r, not \n = Macintosh
1.72 parser 117: bol=eol+1;
118: }
1.154 paf 119: // last piece without \r
1.72 parser 120: if(dest!=bol)
1.126 paf 121: memcpy(dest, bol, eob-bol);
1.154 paf 122: str[length]=0; // terminating
1.72 parser 123: }
1.18 paf 124:
1.154 paf 125: char* file_read_text(Request_charsets& charsets,
126: const String& file_spec,
127: bool fail_on_read_problem,
128: HashStringValue* params/*, HashStringValue* * out_fields*/) {
129: File_read_result file=
130: file_read(charsets, file_spec, true, params, fail_on_read_problem);
131: return file.success?file.str:0;
1.126 paf 132: }
133:
1.178 paf 134: //http request stuff
1.154 paf 135: #ifdef PA_HTTP
136:
137: #undef CRLF
138: #define CRLF "\r\n"
1.126 paf 139:
140: static bool set_addr(struct sockaddr_in *addr, const char* host, const short port){
141: memset(addr, 0, sizeof(*addr));
142: addr->sin_family=AF_INET;
143: addr->sin_port=htons(port);
144: if(host) {
1.175 paf 145: ulong packed_ip=inet_addr(host);
1.184 paf 146: if(packed_ip!=INADDR_NONE)
1.185 paf 147: memcpy(&addr->sin_addr, &packed_ip, sizeof(packed_ip));
1.184 paf 148: else {
149: struct hostent *hostIP=gethostbyname(host);
150: if(hostIP)
151: memcpy(&addr->sin_addr, hostIP->h_addr, hostIP->h_length);
152: else
153: return false;
154: }
1.126 paf 155: } else
156: addr->sin_addr.s_addr=INADDR_ANY;
157: return true;
158: }
159:
1.171 paf 160: static int http_read_response(char*& response, size_t& response_size, int sock, bool fail_on_status_ne_200){
161: response=(char*)pa_malloc_atomic(1/*terminator*/); // setting memory block type
162: response[(response_size=0)]=0;
1.154 paf 163: int result=0;
1.171 paf 164: char* EOLat=0;
1.126 paf 165: while(true) {
1.171 paf 166: char buf[MAX_STRING*10];
167: ssize_t received_size=recv(sock, buf, sizeof(buf), 0);
1.196 paf 168: if(received_size<=0) {
169: if(int no=pa_socks_errno())
1.197 paf 170: throw Exception("http.timeout",
1.196 paf 171: 0,
172: "error receiving response: %s (%d)", pa_socks_strerr(no), no);
1.126 paf 173: break;
1.196 paf 174: }
1.171 paf 175: response=(char*)pa_realloc(response, response_size+received_size+1/*terminator*/);
176: memcpy(response+response_size, buf, received_size);
177: response_size+=received_size;
178: response[response_size]=0;
179:
1.192 paf 180: if(!result && (EOLat=strstr(response, "\n"))) { // checking status in first response
1.171 paf 181: const String status_line(pa_strdup(response, EOLat-response));
1.154 paf 182: ArrayString astatus;
183: size_t pos_after=0;
184: status_line.split(astatus, pos_after, " ");
1.182 paf 185: const String& status_code=*astatus.get(astatus.count()>1?1:0);
1.154 paf 186: result=status_code.as_int();
1.142 paf 187:
1.154 paf 188: if(fail_on_status_ne_200 && result!=200)
1.142 paf 189: throw Exception("http.status",
1.154 paf 190: &status_code,
1.142 paf 191: "invalid HTTP response status");
192: }
193: }
1.154 paf 194: if(result)
195: return result;
1.142 paf 196: else
197: throw Exception("http.response",
198: 0,
1.173 paf 199: "bad response from host - no status found (size=%u)", response_size);
1.126 paf 200: }
201:
202: /* ********************** request *************************** */
203:
204: #if defined(SIGALRM) && defined(HAVE_SIGSETJMP) && defined(HAVE_SIGLONGJMP)
1.145 paf 205: # define PA_USE_ALARM
1.126 paf 206: #endif
207:
1.145 paf 208: #ifdef PA_USE_ALARM
1.126 paf 209: static sigjmp_buf timeout_env;
1.199 paf 210: static void timeout_handler(int /*sig*/){
1.126 paf 211: siglongjmp(timeout_env, 1);
212: }
213: #endif
214:
1.171 paf 215: static int http_request(char*& response, size_t& response_size,
1.193 paf 216: const char* host, short port,
1.152 paf 217: const char* request,
1.196 paf 218: int timeout_secs,
1.152 paf 219: bool fail_on_status_ne_200) {
1.126 paf 220: if(!host)
221: throw Exception("http.host",
1.154 paf 222: 0,
1.126 paf 223: "zero hostname"); //never
224:
1.200 paf 225: volatile // to prevent makeing it register variable, because it will be clobbered by longjmp [thanks gcc warning]
226: int sock=-1;
1.145 paf 227: #ifdef PA_USE_ALARM
1.146 paf 228: signal(SIGALRM, timeout_handler);
1.126 paf 229: #endif
1.145 paf 230: #ifdef PA_USE_ALARM
231: if(sigsetjmp(timeout_env, 1)) {
232: // stupid gcc [2.95.4] generated bad code
233: // which failed to handle sigsetjmp+throw: crashed inside of pre-throw code.
1.199 paf 234: // rewritten simplier [athough duplicating closesocket code]
1.145 paf 235: if(sock>=0)
236: closesocket(sock);
237: throw Exception("http.timeout",
1.199 paf 238: 0,
1.145 paf 239: "timeout occured while retrieving document");
1.146 paf 240: return 0; // never
1.145 paf 241: } else {
1.196 paf 242: alarm(timeout_secs);
1.145 paf 243: #endif
1.146 paf 244: try {
245: int result;
1.126 paf 246: struct sockaddr_in dest;
1.154 paf 247:
248: if(!set_addr(&dest, host, port))
1.126 paf 249: throw Exception("http.host",
1.154 paf 250: 0,
1.127 paf 251: "can not resolve hostname \"%s\"", host);
1.126 paf 252:
1.195 paf 253: if((sock=socket(AF_INET, SOCK_STREAM, IPPROTO_TCP/*0*/))<0) {
254: int no=pa_socks_errno();
1.126 paf 255: throw Exception("http.connect",
1.154 paf 256: 0,
1.195 paf 257: "can not make socket: %s (%d)", pa_socks_strerr(no), no);
258: }
1.196 paf 259:
260: // To enable SO_DONTLINGER (that is, disable SO_LINGER)
261: // l_onoff should be set to zero and setsockopt should be called
262: linger dont_linger={0,0};
263: setsockopt(sock, SOL_SOCKET, SO_LINGER, (const char *)&dont_linger, sizeof(dont_linger));
264:
1.201 paf 265: #ifdef WIN32
266: // SO_*TIMEO can be defined in .h but not implemlemented in protocol,
267: // failing subsequently with Option not supported by protocol (99) message
268: // could not suppress that, so leaving this only for win32
1.196 paf 269: int timeout_ms=timeout_secs*1000;
270: setsockopt(sock, SOL_SOCKET, SO_SNDTIMEO, (const char*)&timeout_ms, sizeof(timeout_ms));
271: setsockopt(sock, SOL_SOCKET, SO_RCVTIMEO, (const char*)&timeout_ms, sizeof(timeout_ms));
272: #endif
273:
1.195 paf 274: if(connect(sock, (struct sockaddr *)&dest, sizeof(dest))) {
275: int no=pa_socks_errno();
1.126 paf 276: throw Exception("http.connect",
1.154 paf 277: 0,
1.195 paf 278: "can not connect to host \"%s\": %s (%d)", host, pa_socks_strerr(no), no);
279: }
1.126 paf 280: size_t request_size=strlen(request);
1.195 paf 281: if(send(sock, request, request_size, 0)!=(ssize_t)request_size) {
282: int no=pa_socks_errno();
1.197 paf 283: throw Exception("http.timeout",
1.154 paf 284: 0,
1.195 paf 285: "error sending request: %s (%d)", pa_socks_strerr(no), no);
286: }
1.126 paf 287:
1.171 paf 288: result=http_read_response(response, response_size, sock, fail_on_status_ne_200);
1.142 paf 289: closesocket(sock);
1.145 paf 290: #ifdef PA_USE_ALARM
1.142 paf 291: alarm(0);
1.126 paf 292: #endif
1.147 paf 293: return result;
1.146 paf 294: } catch(...) {
1.145 paf 295: #ifdef PA_USE_ALARM
1.146 paf 296: alarm(0);
1.126 paf 297: #endif
1.146 paf 298: if(sock>=0)
299: closesocket(sock);
1.154 paf 300: rethrow;
1.146 paf 301: }
1.148 paf 302: #ifdef PA_USE_ALARM
1.126 paf 303: }
1.148 paf 304: #endif
1.126 paf 305: }
306:
1.127 paf 307: #ifndef DOXYGEN
308: struct Http_pass_header_info {
1.154 paf 309: Request_charsets* charsets;
1.127 paf 310: String* request;
311: bool user_agent_specified;
312: };
313: #endif
1.154 paf 314: static void http_pass_header(HashStringValue::key_type key,
315: HashStringValue::value_type value,
316: Http_pass_header_info *info) {
317: *info->request <<key<<": "
318: << attributed_meaning_to_string(*value, String::L_HTTP_HEADER, false)
319: << CRLF;
1.135 paf 320:
1.154 paf 321: if(String(key, String::L_TAINTED).change_case(info->charsets->source(), String::CC_UPPER)=="USER-AGENT")
322: info->user_agent_specified=true;
1.126 paf 323: }
1.154 paf 324:
325:
326: static Charset* detect_charset(Charset& source_charset, const String& content_type_value) {
1.156 paf 327: const String::Body CONTENT_TYPE_VALUE=
1.154 paf 328: content_type_value.change_case(source_charset, String::CC_UPPER);
329: // content-type: xxx/xxx; source_charset=WE-NEED-THIS
330: // content-type: xxx/xxx; source_charset="WE-NEED-THIS"
331: // content-type: xxx/xxx; source_charset="WE-NEED-THIS";
332: size_t before_charseteq_pos=CONTENT_TYPE_VALUE.pos("CHARSET=");
333: if(before_charseteq_pos!=STRING_NOT_FOUND) {
334: size_t charset_begin=before_charseteq_pos+8/*CHARSET="*/;
335: size_t open_quote_pos=CONTENT_TYPE_VALUE.pos('"', charset_begin);
336: bool quoted=open_quote_pos==charset_begin;
337: if(quoted)
338: charset_begin++; // skip opening '"'
339: size_t charset_end=CONTENT_TYPE_VALUE.length();
340: if(quoted) {
341: size_t close_quote_pos=CONTENT_TYPE_VALUE.pos('"', charset_begin);
342: if(close_quote_pos!=STRING_NOT_FOUND)
343: charset_end=close_quote_pos;
344: } else {
345: size_t delim_pos=CONTENT_TYPE_VALUE.pos(';', charset_begin);
346: if(delim_pos!=STRING_NOT_FOUND)
347: charset_end=delim_pos;
348: }
1.156 paf 349: const String::Body CHARSET_NAME_BODY=
1.154 paf 350: CONTENT_TYPE_VALUE.mid(charset_begin, charset_end);
351:
352: return &charsets.get(CHARSET_NAME_BODY);
353: }
354:
355: return 0;
356: }
357:
1.178 paf 358: static const String* basic_authorization_field(const char* user, const char* pass) {
359: if(!user&& !pass)
360: return 0;
361:
362: String combined;
363: if(user)
364: combined<<user;
365: combined<<":";
366: if(pass)
367: combined<<pass;
368:
369: String* result=new String("Basic "); *result<<pa_base64(combined.cstr(), combined.length());
370: return result;
371: }
1.154 paf 372:
1.181 paf 373: static void form_string_value2string(
374: HashStringValue::key_type key,
375: const String& value,
376: String& result)
377: {
378: result << String(key, String::L_TAINTED) << "=";
379: result.append(value, String::L_URI, true);
380: result<< "&";
381: }
382: #ifndef DOXYGEN
383: struct Form_table_value2string_info {
384: HashStringValue::key_type key;
385: String& result;
386:
387: Form_table_value2string_info(HashStringValue::key_type akey, String& aresult):
388: key(akey), result(aresult) {}
389: };
390: #endif
391: static void form_table_value2string(Table::element_type row, Form_table_value2string_info* info) {
392: form_string_value2string(info->key, *row->get(0), info->result);
393: }
1.180 paf 394: static void form_value2string(
395: HashStringValue::key_type key,
396: HashStringValue::value_type value,
397: String* result)
398: {
1.181 paf 399: if(const String* svalue=value->get_string())
400: form_string_value2string(key, *svalue, *result);
401: else if(Table* tvalue=value->get_table()) {
402: Form_table_value2string_info info(key, *result);
403: tvalue->for_each(form_table_value2string, &info);
404: } else
405: throw Exception(0,
406: new String(key, String::L_TAINTED),
407: "is %s, "HTTP_FORM_NAME" option value must either string or table", value->type());
1.180 paf 408: }
409: static const char* form2string(HashStringValue& form) {
410: String string;
411: form.for_each(form_value2string, &string);
1.181 paf 412: return string.cstr(String::L_UNSPECIFIED);
1.180 paf 413: }
1.154 paf 414: #ifndef DOXYGEN
415: struct File_read_http_result {
416: char *str; size_t length;
417: HashStringValue* headers;
418: };
419: #endif
1.194 paf 420: static void find_headers_end(char* p,
421: char*& headers_end_at,
422: char*& raw_body)
423: {
424: raw_body=p;
425: // \n\n
426: // \r\n\r\n
427: while((p=strchr(p, '\n'))) {
428: headers_end_at=++p; // \n>.<
429: if(*p=='\r') // \r\n>\r?<\n
430: p++;
431: if(*p=='\n') { // \r\n\r>\n?<
432: raw_body=p+1;
433: return;
434: }
435: }
436: headers_end_at=0;
437: }
438:
1.155 paf 439: /// @todo build .cookies field. use ^file.tables.SET-COOKIES.menu{ for now
1.154 paf 440: static File_read_http_result file_read_http(Request_charsets& charsets,
441: const String& file_spec,
1.169 paf 442: bool as_text,
443: HashStringValue *options=0) {
1.154 paf 444: File_read_http_result result;
1.126 paf 445: char host[MAX_STRING];
1.129 paf 446: const char* uri;
1.193 paf 447: short port;
1.180 paf 448: const char* method="GET"; bool method_is_get;
449: HashStringValue* form=0;
450: const char* body_cstr=0;
1.196 paf 451: int timeout_secs=2;
1.142 paf 452: bool fail_on_status_ne_200=true;
1.154 paf 453: Value* vheaders=0;
454: Charset *asked_remote_charset=0;
1.178 paf 455: const char* user_cstr=0;
456: const char* password_cstr=0;
1.126 paf 457:
1.127 paf 458: if(options) {
459: int valid_options=0;
1.177 paf 460: if(Value* vmethod=options->get(HTTP_METHOD_NAME)) {
1.127 paf 461: valid_options++;
1.154 paf 462: method=vmethod->as_string().cstr();
1.181 paf 463: }
464: if(Value* vform=options->get(HTTP_FORM_NAME)) {
1.180 paf 465: valid_options++;
466: form=vform->get_hash();
1.181 paf 467: }
468: if(Value* vbody=options->get(HTTP_BODY_NAME)) {
1.180 paf 469: valid_options++;
470: body_cstr=vbody->as_string().cstr(String::L_UNSPECIFIED);
1.181 paf 471: }
472: if(Value* vtimeout=options->get(HTTP_TIMEOUT_NAME)) {
1.127 paf 473: valid_options++;
1.196 paf 474: timeout_secs=vtimeout->as_int();
1.181 paf 475: }
476: if((vheaders=options->get(HTTP_HEADERS_NAME))) {
1.127 paf 477: valid_options++;
1.181 paf 478: }
479: if(Value* vany_status=options->get(HTTP_ANY_STATUS_NAME)) {
1.142 paf 480: valid_options++;
481: fail_on_status_ne_200=!vany_status->as_bool();
1.181 paf 482: }
483: if(Value* vcharset_name=options->get(HTTP_CHARSET_NAME)) {
1.154 paf 484: valid_options++;
485: asked_remote_charset=&::charsets.get(vcharset_name->as_string().
486: change_case(charsets.source(), String::CC_UPPER));
1.181 paf 487: }
488: if(Value* vuser=options->get(HTTP_USER)) {
1.178 paf 489: valid_options++;
490: user_cstr=vuser->as_string().cstr();
1.181 paf 491: }
492: if(Value* vpassword=options->get(HTTP_PASSWORD)) {
1.178 paf 493: valid_options++;
494: password_cstr=vpassword->as_string().cstr();
495: }
1.142 paf 496:
1.154 paf 497: if(valid_options!=options->count())
1.127 paf 498: throw Exception("parser.runtime",
499: 0,
500: "invalid option passed");
1.154 paf 501: }
502: if(!asked_remote_charset) // defaulting to $request:charset
503: asked_remote_charset=&charsets.source();
504:
1.180 paf 505: method_is_get=strcmp(method, "GET")==0;
506: if(method_is_get && body_cstr)
507: throw Exception("parser.runtime",
508: 0,
509: "you can not use $."HTTP_BODY_NAME" option with method GET");
510:
1.154 paf 511: //preparing request
512: String& connect_string=*new String;
513: // not in ^sql{... L_SQL ...} spirit, but closer to ^file::load one
514: connect_string.append(file_spec, String::L_URI); // tainted pieces -> URI pieces
515:
1.180 paf 516: String request_head_and_body;
1.154 paf 517: {
518: // influence URLencoding of tainted pieces to String::L_URI lang
519: Temp_client_charset temp(charsets, *asked_remote_charset);
520:
521: const char* connect_string_cstr=connect_string.cstr(String::L_UNSPECIFIED);
1.126 paf 522:
1.154 paf 523: const char* current=connect_string_cstr;
524: if(strncmp(current, "http://", 7)!=0)
525: throw Exception(0,
526: &connect_string,
527: "does not start with http://"); //never
528: current+=7;
529:
530: strncpy(host, current, sizeof(host)-1); host[sizeof(host)-1]=0;
531: char* host_uri=lsplit(host, '/');
532: uri=host_uri?current+(host_uri-1-host):"/";
533: char* port_cstr=lsplit(host, ':');
534: char* error_pos=0;
1.193 paf 535: port=port_cstr?(short)strtol(port_cstr, &error_pos, 0):80;
1.154 paf 536:
1.180 paf 537: if(strchr(uri, '?') && form)
538: throw Exception("parser.runtime",
539: 0,
540: "use either uri with ?params or $."HTTP_FORM_NAME" option");
541:
542: //making request head
543: String head;
544: head << method;
545: head << " " << uri;
546: if(form)
547: if(method_is_get)
1.181 paf 548: head << "?" << form2string(*form);
1.180 paf 549: head <<" HTTP/1.0" CRLF
1.154 paf 550: "host: "<< host << CRLF;
1.181 paf 551: if(form && !method_is_get) {
552: head << "content-type: application/x-www-form-urlencoded" CRLF;
553: body_cstr = form2string(*form);
554: }
1.178 paf 555:
1.179 paf 556: // http://www.ietf.org/rfc/rfc2617.txt
1.178 paf 557: if(const String* authorization_field_value=basic_authorization_field(user_cstr, password_cstr))
1.180 paf 558: head<<"authorization: "<<*authorization_field_value<<CRLF;
1.178 paf 559:
1.154 paf 560: bool user_agent_specified=false;
561: if(vheaders && !vheaders->is_string()) { // allow empty
562: if(HashStringValue *headers=vheaders->get_hash()) {
1.180 paf 563: Http_pass_header_info info={&charsets, &head, false};
1.154 paf 564: headers->for_each(http_pass_header, &info);
565: user_agent_specified=info.user_agent_specified;
566: } else
567: throw Exception("parser.runtime",
568: &connect_string,
569: "headers param must be hash");
570: };
571: if(!user_agent_specified) // defaulting
1.180 paf 572: head << "user-agent: " DEFAULT_USER_AGENT CRLF;
573:
574: if(body_cstr) {
575: // recode those pieces which are not in String::L_URI lang
576: // [those violating HTTP standard, but widly used]
577: body_cstr=Charset::transcode(
578: String::C(body_cstr, strlen(body_cstr)),
579: charsets.source(),
580: *asked_remote_charset);
1.181 paf 581:
582: head << "content-length: " << format(strlen(body_cstr), "%u") << CRLF;
1.180 paf 583: }
1.154 paf 584:
1.181 paf 585: const char* head_cstr=head.cstr(String::L_UNSPECIFIED);
586:
587: // recode those pieces which are not in String::L_URI lang
588: // [those violating HTTP standard, but widly used]
589: head_cstr=Charset::transcode(
590: String::C(head_cstr, strlen(head_cstr)),
591: charsets.source(),
592: *asked_remote_charset);
1.180 paf 593:
1.181 paf 594: // head + end of header
595: request_head_and_body << head_cstr << CRLF;
1.180 paf 596: // body
597: if(body_cstr)
598: request_head_and_body << body_cstr;
1.154 paf 599: }
1.126 paf 600:
601: //sending request
1.171 paf 602: char* response;
603: size_t response_size;
604: int status_code=http_request(response, response_size,
1.180 paf 605: host, port, request_head_and_body.cstr(),
1.196 paf 606: timeout_secs, fail_on_status_ne_200);
1.126 paf 607:
608: //processing results
1.171 paf 609: char* raw_body; size_t raw_body_size;
1.194 paf 610: char* headers_end_at;
611: find_headers_end(response,
612: headers_end_at,
613: raw_body);
1.191 paf 614: raw_body_size=response_size-(raw_body-response);
1.171 paf 615:
1.154 paf 616: result.headers=new HashStringValue;
1.155 paf 617: VHash* vtables=new VHash;
1.177 paf 618: result.headers->put(HTTP_TABLES_NAME, vtables);
1.194 paf 619: Charset* real_remote_charset=0; // undetected, yet
620:
621: if(headers_end_at) {
622: *headers_end_at=0;
623: const String header_block(String::C(response, headers_end_at-response), true);
624:
625: ArrayString aheaders;
626: HashStringValue& tables=vtables->hash();
1.155 paf 627:
1.194 paf 628: size_t pos_after=0;
629: header_block.split(aheaders, pos_after, "\n");
630:
631: //processing headers
632: size_t aheaders_count=aheaders.count();
633: for(size_t i=1; i<aheaders_count; i++) {
634: const String& line=*aheaders.get(i);
635: size_t pos=line.pos(':');
636: if(pos==STRING_NOT_FOUND || pos<1)
637: throw Exception("http.response",
638: &connect_string,
639: "bad response from host - bad header \"%s\"", line.cstr());
640: const String::Body HEADER_NAME=
641: line.mid(0, pos).change_case(charsets.source(), String::CC_UPPER);
642: const String& header_value=line.mid(pos+1, line.length()).trim(String::TRIM_BOTH, " \t\r");
643: if(as_text && HEADER_NAME=="CONTENT-TYPE")
644: real_remote_charset=detect_charset(charsets.source(), header_value);
645:
646: // tables
647: {
648: Value *valready=(Value *)tables.get(HEADER_NAME);
649: bool existed=valready!=0;
650: Table *table;
651: if(existed) {
652: // second+ appearence
653: table=valready->get_table();
654: } else {
655: // first appearence
656: Table::columns_type columns =new ArrayString(1);
657: *columns+=new String("value");
658: table=new Table(columns);
659: }
660: // this string becomes next row
661: ArrayString& row=*new ArrayString(1);
662: row+=&header_value;
663: *table+=&row;
664: // not existed before? add it
665: if(!existed)
666: tables.put(HEADER_NAME, new VTable(table));
1.155 paf 667: }
1.194 paf 668:
669: result.headers->put(HEADER_NAME, new VString(header_value));
1.155 paf 670: }
1.126 paf 671: }
672:
673: // output response
1.171 paf 674: String::C real_body=String::C(raw_body, raw_body_size);
1.192 paf 675: if(as_text && raw_body_size) { // must be checked because transcode returns CONST string in case length==0, which contradicts hacking few lines below
676: // defaulting to used-asked charset [it's never empty!]
677: if(!real_remote_charset)
678: real_remote_charset=asked_remote_charset;
1.169 paf 679: real_body=Charset::transcode(real_body, *real_remote_charset, charsets.source());
1.192 paf 680: }
1.154 paf 681:
682: result.str=const_cast<char *>(real_body.str); // hacking a little
683: result.length=real_body.length;
684: result.headers->put(file_status_name, new VInt(status_code));
685: return result;
1.34 paf 686: }
1.123 paf 687:
1.154 paf 688: #endif
689:
1.123 paf 690: #ifndef DOXYGEN
691: struct File_read_action_info {
1.154 paf 692: char **data; size_t *data_size;
1.188 paf 693: char* buf; size_t offset; size_t count;
1.126 paf 694: };
1.123 paf 695: #endif
1.154 paf 696: static void file_read_action(
697: struct stat& finfo,
698: int f,
1.166 paf 699: const String& file_spec, const char* /*fname*/, bool as_text,
1.154 paf 700: void *context) {
1.126 paf 701: File_read_action_info& info=*static_cast<File_read_action_info *>(context);
1.188 paf 702: size_t to_read_size=info.count;
703: if(!to_read_size)
704: to_read_size=(size_t)finfo.st_size;
705: assert( !(info.buf && as_text) );
706: if(to_read_size) {
707: if(info.offset)
708: lseek(f, info.offset, SEEK_SET);
709: *info.data=info.buf
710: ? info.buf
711: : new(PointerFreeGC) char[to_read_size+(as_text?1:0)];
1.126 paf 712: *info.data_size=(size_t)read(f, *info.data, to_read_size);
1.123 paf 713:
714: if(ssize_t(*info.data_size)<0 || *info.data_size>to_read_size)
1.126 paf 715: throw Exception(0,
1.123 paf 716: &file_spec,
1.173 paf 717: "read failed: actually read %u bytes count not in [0..%u] valid range",
1.126 paf 718: *info.data_size, to_read_size);
1.123 paf 719: } else { // empty file
720: if(as_text) {
1.154 paf 721: *info.data=new(PointerFreeGC) char[1];
1.123 paf 722: *(char*)(*info.data)=0;
723: } else
724: *info.data=0;
725: *info.data_size=0;
726: return;
727: }
1.126 paf 728: }
1.154 paf 729: File_read_result file_read(Request_charsets& charsets, const String& file_spec,
730: bool as_text, HashStringValue *params,
1.188 paf 731: bool fail_on_read_problem,
732: char* buf, size_t offset, size_t count) {
1.167 paf 733: File_read_result result={false, 0, 0, 0};
1.154 paf 734: #ifdef PA_HTTP
735: if(file_spec.starts_with("http://")) {
1.126 paf 736: // fail on read problem
1.169 paf 737: File_read_http_result http=file_read_http(charsets, file_spec, as_text, params);
1.154 paf 738: result.success=true;
739: result.str=http.str;
740: result.length=http.length;
741: result.headers=http.headers;
1.126 paf 742: } else {
1.154 paf 743: #endif
1.161 paf 744: if(params && params->count())
745: throw Exception("parser.runtime",
746: 0,
747: "invalid option passed");
748:
1.188 paf 749: File_read_action_info info={&result.str, &result.length,
750: buf, offset, count};
1.154 paf 751: result.success=file_read_action_under_lock(file_spec,
1.126 paf 752: "read", file_read_action, &info,
753: as_text, fail_on_read_problem);
1.154 paf 754: #ifdef PA_HTTP
1.126 paf 755: }
1.154 paf 756: #endif
1.123 paf 757:
1.154 paf 758: if(result.success && as_text) {
1.131 paf 759: // UTF-8 signature: EF BB BF
1.154 paf 760: if(result.length>=3) {
761: char *in=(char *)result.str;
1.159 paf 762: if(strncmp(in, "\xEF\xBB\xBF", 3)==0) {
1.154 paf 763: result.str=in+3; result.length-=3;// skip prefix
1.131 paf 764: }
765: }
766:
1.154 paf 767: fix_line_breaks((char *)(result.str), result.length);
1.123 paf 768: }
1.126 paf 769:
770: return result;
1.123 paf 771: }
772:
1.154 paf 773: #ifdef PA_SAFE_MODE
774: void check_safe_mode(struct stat finfo, const String& file_spec, const char* fname) {
775: if(finfo.st_uid/*foreign?*/!=geteuid()
776: && finfo.st_gid/*foreign?*/!=getegid())
777: throw Exception("parser.runtime",
778: &file_spec,
779: "parser is in safe mode: "
780: "reading files of foreign group and user disabled "
781: "[recompile parser with --disable-safe-mode configure option], "
782: "actual filename '%s', "
783: "fuid(%d)!=euid(%d) or fgid(%d)!=egid(%d)",
784: fname,
785: finfo.st_uid, geteuid(),
786: finfo.st_gid, getegid());
787: }
788: #endif
1.149 paf 789:
1.154 paf 790: bool file_read_action_under_lock(const String& file_spec,
1.126 paf 791: const char* action_name, File_read_action action, void *context,
792: bool as_text,
1.123 paf 793: bool fail_on_read_problem) {
1.154 paf 794: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.33 paf 795: int f;
796:
797: // first open, next stat:
1.45 paf 798: // directory update of NTFS hard links performed on open.
1.33 paf 799: // ex:
800: // a.html:^test[] and b.html hardlink to a.html
801: // user inserts ! before ^test in a.html
1.126 paf 802: // directory entry of b.html in NTFS not updated at once,
1.35 paf 803: // they delay update till open, so we would receive "!^test[" string
804: // if would do stat, next open.
1.123 paf 805: // later: it seems, even this does not help sometimes
1.98 paf 806: if((f=open(fname, O_RDONLY|(as_text?_O_TEXT:_O_BINARY)))>=0) {
1.123 paf 807: try {
1.162 paf 808: if(pa_lock_shared_blocking(f)!=0)
1.126 paf 809: throw Exception("file.lock",
1.123 paf 810: &file_spec,
811: "shared lock failed: %s (%d), actual filename '%s'",
1.154 paf 812: strerror(errno), errno, fname);
1.123 paf 813:
1.124 paf 814: struct stat finfo;
815: if(stat(fname, &finfo)!=0)
816: throw Exception("file.missing", // hardly possible: we just opened it OK
817: &file_spec,
818: "stat failed: %s (%d), actual filename '%s'",
1.154 paf 819: strerror(errno), errno, fname);
1.124 paf 820:
1.140 paf 821: #ifdef PA_SAFE_MODE
1.149 paf 822: check_safe_mode(finfo, file_spec, fname);
1.105 paf 823: #endif
1.32 paf 824:
1.154 paf 825: action(finfo, f, file_spec, fname, as_text, context);
1.123 paf 826: } catch(...) {
1.162 paf 827: pa_unlock(f);close(f);
1.123 paf 828: if(fail_on_read_problem)
1.154 paf 829: rethrow;
1.123 paf 830: return false;
831: }
1.87 paf 832:
1.162 paf 833: pa_unlock(f);close(f);
1.72 parser 834: return true;
1.118 paf 835: } else {
836: if(fail_on_read_problem)
1.126 paf 837: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.118 paf 838: &file_spec,
1.123 paf 839: "%s failed: %s (%d), actual filename '%s'",
1.154 paf 840: action_name, strerror(errno), errno, fname);
1.118 paf 841: return false;
842: }
1.8 paf 843: }
844:
1.202 ! paf 845: void create_dir_for_file(const String& file_spec) {
1.63 parser 846: size_t pos_after=1;
1.154 paf 847: size_t pos_before;
848: while((pos_before=file_spec.pos('/', pos_after))!=STRING_NOT_FOUND) {
849: mkdir(file_spec.mid(0, pos_before).cstr(String::L_FILE_SPEC), 0775);
1.63 parser 850: pos_after=pos_before+1;
851: }
852: }
853:
1.98 paf 854: bool file_write_action_under_lock(
1.28 paf 855: const String& file_spec,
1.126 paf 856: const char* action_name, File_write_action action, void *context,
857: bool as_text,
858: bool do_append,
859: bool do_block,
1.110 paf 860: bool fail_on_lock_problem) {
1.154 paf 861: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.28 paf 862: int f;
1.80 paf 863: if(access(fname, W_OK)!=0) // no
1.126 paf 864: create_dir_for_file(file_spec);
1.50 paf 865:
1.80 paf 866: if((f=open(fname,
867: O_CREAT|O_RDWR
868: |(as_text?_O_TEXT:_O_BINARY)
1.138 paf 869: |(do_append?O_APPEND:PA_O_TRUNC), 0664))>=0) {
1.162 paf 870: if((do_block?pa_lock_exclusive_blocking(f):pa_lock_exclusive_nonblocking(f))!=0) {
1.126 paf 871: Exception e("file.lock",
1.110 paf 872: &file_spec,
873: "shared lock failed: %s (%d), actual filename '%s'",
1.154 paf 874: strerror(errno), errno, fname);
1.126 paf 875: close(f);
1.110 paf 876: if(fail_on_lock_problem)
877: throw e;
1.98 paf 878: return false;
879: }
1.96 paf 880:
1.158 paf 881: try {
1.126 paf 882: action(f, context);
1.158 paf 883: } catch(...) {
1.138 paf 884: #ifdef HAVE_FTRUNCATE
1.104 paf 885: if(!do_append)
1.125 paf 886: ftruncate(f, lseek(f, 0, SEEK_CUR)); // one can not use O_TRUNC, read lower
1.138 paf 887: #endif
1.162 paf 888: pa_unlock(f);close(f);
1.154 paf 889: rethrow;
1.158 paf 890: }
1.80 paf 891:
1.138 paf 892: #ifdef HAVE_FTRUNCATE
1.104 paf 893: if(!do_append)
1.125 paf 894: 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 895: #endif
1.162 paf 896: pa_unlock(f);close(f);
1.98 paf 897: return true;
1.80 paf 898: } else
1.126 paf 899: throw Exception(errno==EACCES?"file.access":0,
1.80 paf 900: &file_spec,
1.96 paf 901: "%s failed: %s (%d), actual filename '%s'",
1.154 paf 902: action_name, strerror(errno), errno, fname);
1.96 paf 903: // here should be nothing, see rethrow above
904: }
905:
906: #ifndef DOXYGEN
907: struct File_write_action_info {
1.154 paf 908: const char* str; size_t length;
1.126 paf 909: };
1.96 paf 910: #endif
911: static void file_write_action(int f, void *context) {
1.126 paf 912: File_write_action_info& info=*static_cast<File_write_action_info *>(context);
1.154 paf 913: if(info.length) {
914: int written=write(f, info.str, info.length);
1.116 paf 915: if(written<0)
1.126 paf 916: throw Exception(0,
917: 0,
918: "write failed: %s (%d)", strerror(errno), errno);
1.113 paf 919: }
1.96 paf 920: }
921: void file_write(
922: const String& file_spec,
1.154 paf 923: const char* data, size_t size,
1.126 paf 924: bool as_text,
1.96 paf 925: bool do_append) {
1.126 paf 926: File_write_action_info info={data, size};
1.98 paf 927: file_write_action_under_lock(
1.154 paf 928: file_spec,
929: "write", file_write_action, &info,
930: as_text,
931: do_append);
1.30 paf 932: }
933:
1.63 parser 934: // throws nothing! [this is required in file_move & file_delete]
1.50 paf 935: static void rmdir(const String& file_spec, size_t pos_after) {
1.154 paf 936: size_t pos_before;
937: if((pos_before=file_spec.pos('/', pos_after))!=STRING_NOT_FOUND)
1.126 paf 938: rmdir(file_spec, pos_before+1);
1.50 paf 939:
1.154 paf 940: rmdir(file_spec.mid(0, pos_after-1/* / */).cstr(String::L_FILE_SPEC));
1.50 paf 941: }
1.164 paf 942: bool file_delete(const String& file_spec, bool fail_on_problem) {
1.154 paf 943: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.54 parser 944: if(unlink(fname)!=0)
1.164 paf 945: if(fail_on_problem)
1.126 paf 946: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.93 paf 947: &file_spec,
948: "unlink failed: %s (%d), actual filename '%s'",
1.154 paf 949: strerror(errno), errno, fname);
1.93 paf 950: else
951: return false;
1.50 paf 952:
1.126 paf 953: rmdir(file_spec, 1);
1.93 paf 954: return true;
1.60 parser 955: }
1.95 paf 956: void file_move(const String& old_spec, const String& new_spec) {
1.154 paf 957: const char* old_spec_cstr=old_spec.cstr(String::L_FILE_SPEC);
958: const char* new_spec_cstr=new_spec.cstr(String::L_FILE_SPEC);
1.63 parser 959:
1.126 paf 960: create_dir_for_file(new_spec);
1.63 parser 961:
1.60 parser 962: if(rename(old_spec_cstr, new_spec_cstr)!=0)
1.126 paf 963: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.60 parser 964: &old_spec,
965: "rename failed: %s (%d), actual filename '%s' to '%s'",
1.154 paf 966: strerror(errno), errno, old_spec_cstr, new_spec_cstr);
1.63 parser 967:
1.126 paf 968: rmdir(old_spec, 1);
1.31 paf 969: }
970:
1.51 paf 971:
1.126 paf 972: bool entry_exists(const char* fname, struct stat *afinfo) {
1.118 paf 973: struct stat lfinfo;
974: bool result=stat(fname, &lfinfo)==0;
975: if(afinfo)
976: *afinfo=lfinfo;
977: return result;
1.119 paf 978: }
979:
980: bool entry_exists(const String& file_spec) {
1.154 paf 981: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.126 paf 982: return entry_exists(fname, 0);
1.118 paf 983: }
984:
1.51 paf 985: static bool entry_readable(const String& file_spec, bool need_dir) {
1.154 paf 986: char* fname=file_spec.cstrm(String::L_FILE_SPEC);
1.120 paf 987: if(need_dir) {
1.126 paf 988: size_t size=strlen(fname);
1.120 paf 989: while(size) {
1.126 paf 990: char c=fname[size-1];
1.120 paf 991: if(c=='/' || c=='\\')
992: fname[--size]=0;
993: else
994: break;
995: }
996: }
1.51 paf 997: struct stat finfo;
1.118 paf 998: if(access(fname, R_OK)==0 && entry_exists(fname, &finfo)) {
1.109 paf 999: bool is_dir=(finfo.st_mode&S_IFDIR) != 0;
1.51 paf 1000: return is_dir==need_dir;
1001: }
1002: return false;
1003: }
1.31 paf 1004: bool file_readable(const String& file_spec) {
1.126 paf 1005: return entry_readable(file_spec, false);
1.51 paf 1006: }
1007: bool dir_readable(const String& file_spec) {
1.126 paf 1008: return entry_readable(file_spec, true);
1.65 parser 1009: }
1.154 paf 1010: const String* file_readable(const String& path, const String& name) {
1011: String& result=*new String(path);
1012: result << "/";
1013: result << name;
1014: return file_readable(result)?&result:0;
1.43 paf 1015: }
1016: bool file_executable(const String& file_spec) {
1.154 paf 1017: return access(file_spec.cstr(String::L_FILE_SPEC), X_OK)==0;
1.44 paf 1018: }
1019:
1.64 parser 1020: bool file_stat(const String& file_spec,
1.58 parser 1021: size_t& rsize,
1.126 paf 1022: time_t& ratime,
1023: time_t& rmtime,
1024: time_t& rctime,
1.64 parser 1025: bool fail_on_read_problem) {
1.154 paf 1026: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1027: struct stat finfo;
1.44 paf 1028: if(stat(fname, &finfo)!=0)
1.64 parser 1029: if(fail_on_read_problem)
1.126 paf 1030: throw Exception("file.missing",
1.67 parser 1031: &file_spec,
1032: "getting file size failed: %s (%d), real filename '%s'",
1.154 paf 1033: strerror(errno), errno, fname);
1.64 parser 1034: else
1035: return false;
1.58 parser 1036: rsize=finfo.st_size;
1037: ratime=finfo.st_atime;
1038: rmtime=finfo.st_mtime;
1039: rctime=finfo.st_ctime;
1.64 parser 1040: return true;
1.18 paf 1041: }
1042:
1.126 paf 1043: char* getrow(char* *row_ref, char delim) {
1044: char* result=*row_ref;
1.8 paf 1045: if(result) {
1.126 paf 1046: *row_ref=strchr(result, delim);
1.8 paf 1047: if(*row_ref)
1048: *((*row_ref)++)=0;
1049: else if(!*result)
1050: return 0;
1051: }
1052: return result;
1053: }
1054:
1.126 paf 1055: char* lsplit(char* string, char delim) {
1.23 paf 1056: if(string) {
1.126 paf 1057: char* v=strchr(string, delim);
1.8 paf 1058: if(v) {
1059: *v=0;
1060: return v+1;
1061: }
1062: }
1063: return 0;
1064: }
1065:
1.126 paf 1066: char* lsplit(char* *string_ref, char delim) {
1067: char* result=*string_ref;
1068: char* next=lsplit(*string_ref, delim);
1.8 paf 1069: *string_ref=next;
1070: return result;
1.9 paf 1071: }
1072:
1.126 paf 1073: char* rsplit(char* string, char delim) {
1.18 paf 1074: if(string) {
1.126 paf 1075: char* v=strrchr(string, delim);
1.18 paf 1076: if(v) {
1.9 paf 1077: *v=0;
1078: return v+1;
1079: }
1080: }
1081: return NULL;
1.10 paf 1082: }
1083:
1.37 paf 1084: /// @todo less stupid type detection
1.154 paf 1085: const char* format(double value, char* fmt) {
1.126 paf 1086: char local_buf[MAX_NUMBER];
1.108 paf 1087: size_t size;
1088:
1.10 paf 1089: if(fmt)
1090: if(strpbrk(fmt, "diouxX"))
1091: if(strpbrk(fmt, "ouxX"))
1.126 paf 1092: size=snprintf(local_buf, sizeof(local_buf), fmt, (uint)value);
1.10 paf 1093: else
1.126 paf 1094: size=snprintf(local_buf, sizeof(local_buf), fmt, (int)value);
1.10 paf 1095: else
1.126 paf 1096: size=snprintf(local_buf, sizeof(local_buf), fmt, value);
1.10 paf 1097: else
1.126 paf 1098: size=snprintf(local_buf, sizeof(local_buf), "%d", (int)value);
1.10 paf 1099:
1.154 paf 1100: return pa_strdup(local_buf, size);
1.12 paf 1101: }
1102:
1.36 paf 1103: size_t stdout_write(const void *buf, size_t size) {
1.12 paf 1104: #ifdef WIN32
1.187 paf 1105: size_t to_write = size;
1.12 paf 1106: do{
1.154 paf 1107: int chunk_written=fwrite(buf, 1, min((size_t)8*0x400, size), stdout);
1.12 paf 1108: if(chunk_written<=0)
1109: break;
1110: size-=chunk_written;
1.36 paf 1111: buf=((const char*)buf)+chunk_written;
1.126 paf 1112: } while(size>0);
1.12 paf 1113:
1.187 paf 1114: return to_write-size;
1.12 paf 1115: #else
1.126 paf 1116: return fwrite(buf, 1, size, stdout);
1.12 paf 1117: #endif
1.2 paf 1118: }
1.14 paf 1119:
1.154 paf 1120: char* unescape_chars(const char* cp, int len) {
1121: char* s=new(PointerFreeGC) char[len + 1];
1.14 paf 1122: enum EscapeState {
1.33 paf 1123: EscapeRest,
1124: EscapeFirst,
1.14 paf 1125: EscapeSecond
1126: } escapeState=EscapeRest;
1.193 paf 1127: uchar escapedValue=0;
1.14 paf 1128: int srcPos=0;
1129: int dstPos=0;
1130: while(srcPos < len) {
1.193 paf 1131: uchar ch=(uchar)cp[srcPos];
1.14 paf 1132: switch(escapeState) {
1133: case EscapeRest:
1134: if(ch=='%') {
1135: escapeState=EscapeFirst;
1136: } else if(ch=='+') {
1.126 paf 1137: s[dstPos++]=' ';
1.14 paf 1138: } else {
1139: s[dstPos++]=ch;
1140: }
1141: break;
1142: case EscapeFirst:
1.193 paf 1143: escapedValue=(uchar)(hex_value[ch] << 4);
1.14 paf 1144: escapeState=EscapeSecond;
1145: break;
1146: case EscapeSecond:
1.126 paf 1147: escapedValue +=hex_value[ch];
1.14 paf 1148: s[dstPos++]=escapedValue;
1149: escapeState=EscapeRest;
1150: break;
1151: }
1.126 paf 1152: srcPos++;
1.14 paf 1153: }
1154: s[dstPos]=0;
1155: return s;
1.24 paf 1156: }
1157:
1158: #ifdef WIN32
1.126 paf 1159: void back_slashes_to_slashes(char* s) {
1.24 paf 1160: if(s)
1161: for(; *s; s++)
1162: if(*s=='\\')
1.126 paf 1163: *s='/';
1.24 paf 1164: }
1.42 paf 1165: /*
1.126 paf 1166: void slashes_to_back_slashes(char* s) {
1.42 paf 1167: if(s)
1168: for(; *s; s++)
1169: if(*s=='/')
1.126 paf 1170: *s='\\';
1.42 paf 1171: }
1172: */
1.24 paf 1173: #endif
1.41 paf 1174:
1.126 paf 1175: bool StrEqNc(const char* s1, const char* s2, bool strict) {
1.41 paf 1176: while(true) {
1177: if(!(*s1)) {
1178: if(!(*s2))
1179: return true;
1180: else
1181: return !strict;
1182: } else if(!(*s2))
1183: return !strict;
1.189 paf 1184: if(isalpha((unsigned char)*s1)) {
1.190 paf 1185: if(tolower((unsigned char)*s1) !=tolower((unsigned char)*s2))
1.41 paf 1186: return false;
1187: } else if((*s1) !=(*s2))
1188: return false;
1.126 paf 1189: s1++;
1190: s2++;
1.41 paf 1191: }
1.57 parser 1192: }
1193:
1.84 paf 1194: static bool isLeap(int year) {
1.57 parser 1195: return !(
1196: (year % 4) || ((year % 400) && !(year % 100))
1.126 paf 1197: );
1.57 parser 1198: }
1199:
1200: int getMonthDays(int year, int month) {
1201: int monthDays[]={
1.126 paf 1202: 31,
1203: isLeap(year) ? 29 : 28,
1204: 31,
1205: 30,
1206: 31,
1207: 30,
1208: 31,
1209: 31,
1210: 30,
1211: 31,
1212: 30,
1.57 parser 1213: 31
1.126 paf 1214: };
1215: return monthDays[month];
1.41 paf 1216: }
1.69 parser 1217:
1.126 paf 1218: void remove_crlf(char* start, char* end) {
1219: for(char* p=start; p<end; p++)
1.69 parser 1220: switch(*p) {
1.126 paf 1221: case '\n': *p='|'; break;
1222: case '\r': *p=' '; break;
1.69 parser 1223: }
1.91 paf 1224: }
1225:
1226:
1227: /// must be last in this file
1228: #undef vsnprintf
1.126 paf 1229: int __vsnprintf(char* b, size_t s, const char* f, va_list l) {
1.91 paf 1230: if(!s)
1231: return 0;
1232:
1233: int r;
1234: // note: on win32& maybe somewhere else
1235: // vsnprintf do not writes terminating 0 in 'buffer full' case, reducing
1236: --s;
1.172 paf 1237:
1238: // clients do not check for negative 's', feature: ignore such prints
1239: if((ssize_t)s<0)
1240: return 0;
1241:
1.91 paf 1242: #if _MSC_VER
1243: /*
1244: win32:
1245: mk:@MSITStore:C:\Program%20Files\Microsoft%20Visual%20Studio\MSDN\2001APR\1033\vccore.chm::/html/_crt__vsnprintf.2c_._vsnwprintf.htm
1246:
1.154 paf 1247: if the number of bytes to write exceeds buffer, then count bytes are written and Ö1 is returned
1.91 paf 1248: */
1.126 paf 1249: r=_vsnprintf(b, s, f, l);
1.91 paf 1250: if(r<0)
1251: r=s;
1252: #else
1.126 paf 1253: r=vsnprintf(b, s, f, l);
1.91 paf 1254: /*
1255: solaris:
1256: man vsnprintf
1257:
1258: The snprintf() function returns the number of characters
1259: formatted, that is, the number of characters that would have
1260: been written to the buffer if it were large enough. If the
1261: value of n is 0 on a call to snprintf(), an unspecified
1262: value less than 1 is returned.
1263: */
1264:
1265: if(r<0)
1266: r=0;
1.167 paf 1267: else if((size_t)r>s)
1.91 paf 1268: r=s;
1269: #endif
1270: b[r]=0;
1271: return r;
1272: }
1273:
1.126 paf 1274: int __snprintf(char* b, size_t s, const char* f, ...) {
1.91 paf 1275: va_list l;
1.126 paf 1276: va_start(l, f);
1277: int r=__vsnprintf(b, s, f, l);
1278: va_end(l);
1.91 paf 1279: return r;
1.178 paf 1280: }
1281:
1282: /* mime64 functions are from libgmime[http://spruce.sourceforge.net/gmime/] lib */
1283: /*
1284: * Authors: Michael Zucchi <notzed@helixcode.com>
1285: * Jeffrey Stedfast <fejj@helixcode.com>
1286: *
1287: * Copyright 2000 Helix Code, Inc. (www.helixcode.com)
1288: *
1289: * This program is free software; you can redistribute it and/or modify
1290: * it under the terms of the GNU General Public License as published by
1291: * the Free Software Foundation; either version 2 of the License, or
1292: * (at your option) any later version.
1293: *
1294: * This program is distributed in the hope that it will be useful,
1295: * but WITHOUT ANY WARRANTY; without even the implied warranty of
1296: * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
1297: * GNU General Public License for more details.
1298: *
1299: * You should have received a copy of the GNU General Public License
1300: * along with this program; if not, write to the Free Software
1301: * Foundation, Inc., 59 Temple Street #330, Boston, MA 02111-1307, USA.
1302: *
1303: */
1304: static char *base64_alphabet =
1305: "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";
1306:
1307: /**
1308: * g_mime_utils_base64_encode_step:
1309: * @in: input stream
1310: * @inlen: length of the input
1311: * @out: output string
1312: * @state: holds the number of bits that are stored in @save
1313: * @save: leftover bits that have not yet been encoded
1314: *
1315: * Base64 encodes a chunk of data. Performs an 'encode step', only
1316: * encodes blocks of 3 characters to the output at a time, saves
1317: * left-over state in state and save (initialise to 0 on first
1318: * invocation).
1319: *
1320: * Returns the number of bytes encoded.
1321: **/
1322: static size_t
1323: g_mime_utils_base64_encode_step (const unsigned char *in, size_t inlen, unsigned char *out, int *state, int *save)
1324: {
1.186 paf 1325: register const unsigned char *inptr;
1.178 paf 1326: register unsigned char *outptr;
1327:
1328: if (inlen <= 0)
1329: return 0;
1330:
1331: inptr = in;
1332: outptr = out;
1333:
1334: if (inlen + ((unsigned char *)save)[0] > 2) {
1335: const unsigned char *inend = in + inlen - 2;
1336: register int c1 = 0, c2 = 0, c3 = 0;
1337: register int already;
1338:
1339: already = *state;
1340:
1341: switch (((char *)save)[0]) {
1342: case 1: c1 = ((unsigned char *)save)[1]; goto skip1;
1343: case 2: c1 = ((unsigned char *)save)[1];
1344: c2 = ((unsigned char *)save)[2]; goto skip2;
1345: }
1346:
1347: /* yes, we jump into the loop, no i'm not going to change it, its beautiful! */
1348: while (inptr < inend) {
1349: c1 = *inptr++;
1350: skip1:
1351: c2 = *inptr++;
1352: skip2:
1353: c3 = *inptr++;
1354: *outptr++ = base64_alphabet [c1 >> 2];
1355: *outptr++ = base64_alphabet [(c2 >> 4) | ((c1 & 0x3) << 4)];
1356: *outptr++ = base64_alphabet [((c2 & 0x0f) << 2) | (c3 >> 6)];
1357: *outptr++ = base64_alphabet [c3 & 0x3f];
1358: /* this is a bit ugly ... */
1359: if ((++already) >= 19) {
1360: *outptr++ = '\n';
1361: already = 0;
1362: }
1363: }
1364:
1365: ((unsigned char *)save)[0] = 0;
1366: inlen = 2 - (inptr - inend);
1367: *state = already;
1368: }
1369:
1370: //d(printf ("state = %d, inlen = %d\n", (int)((char *)save)[0], inlen));
1371:
1372: if (inlen > 0) {
1373: register char *saveout;
1374:
1375: /* points to the slot for the next char to save */
1376: saveout = & (((char *)save)[1]) + ((char *)save)[0];
1377:
1378: /* inlen can only be 0 1 or 2 */
1379: switch (inlen) {
1380: case 2: *saveout++ = *inptr++;
1381: case 1: *saveout++ = *inptr++;
1382: }
1383: ((char *)save)[0] += inlen;
1384: }
1385:
1386: /*d(printf ("mode = %d\nc1 = %c\nc2 = %c\n",
1387: (int)((char *)save)[0],
1388: (int)((char *)save)[1],
1389: (int)((char *)save)[2]));*/
1390:
1391: return (outptr - out);
1392: }
1393:
1394: /**
1395: * g_mime_utils_base64_encode_close:
1396: * @in: input stream
1397: * @inlen: length of the input
1398: * @out: output string
1399: * @state: holds the number of bits that are stored in @save
1400: * @save: leftover bits that have not yet been encoded
1401: *
1402: * Base64 encodes the input stream to the output stream. Call this
1403: * when finished encoding data with g_mime_utils_base64_encode_step to
1404: * flush off the last little bit.
1405: *
1406: * Returns the number of bytes encoded.
1407: **/
1408: static size_t
1409: g_mime_utils_base64_encode_close (const unsigned char *in, size_t inlen, unsigned char *out, int *state, int *save)
1410: {
1411: unsigned char *outptr = out;
1412: int c1, c2;
1413:
1414: if (inlen > 0)
1415: outptr += g_mime_utils_base64_encode_step (in, inlen, outptr, state, save);
1416:
1417: c1 = ((unsigned char *)save)[1];
1418: c2 = ((unsigned char *)save)[2];
1419:
1420: switch (((unsigned char *)save)[0]) {
1421: case 2:
1422: outptr[2] = base64_alphabet [(c2 & 0x0f) << 2];
1423: goto skip;
1424: case 1:
1425: outptr[2] = '=';
1426: skip:
1427: outptr[0] = base64_alphabet [c1 >> 2];
1428: outptr[1] = base64_alphabet [c2 >> 4 | ((c1 & 0x3) << 4)];
1429: outptr[3] = '=';
1430: outptr += 4;
1431: break;
1432: }
1433:
1434: *outptr++ = 0;
1435:
1436: *save = 0;
1437: *state = 0;
1438:
1439: return (outptr - out);
1440: }
1441:
1442: char* pa_base64(const char *in, size_t len)
1443: {
1444: /* wont go to more than 2x size (overly conservative) */
1445: char* result=new(PointerFreeGC) char[len * 2 + 6];
1446: int state=0;
1447: int save=0;
1.183 paf 1448: #ifndef NDEBUG
1449: size_t filled=
1450: #endif
1451: g_mime_utils_base64_encode_close ((const unsigned char*)in, len,
1.178 paf 1452: (unsigned char*)result, &state, &save);
1453: assert(filled <= len * 2 + 6);
1454:
1455: return result;
1.98 paf 1456: }
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