Annotation of parser3/src/main/pa_common.C, revision 1.207
1.15 paf 1: /** @file
1.16 paf 2: Parser: commonly functions.
3:
1.205 paf 4: Copyright(c) 2001-2005 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:
1.207 ! paf 8: static const char * const IDENT_COMMON_C="$Date: 2005/11/18 09:11:21 $";
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:
1.206 paf 134: /// these options were handled but not checked elsewhere, now check them
135: static int get_valid_file_options_count(HashStringValue& options)
136: {
137: int result=0;
138: if(options.get(PA_SQL_LIMIT_NAME))
139: result++;
140: if(options.get(PA_SQL_OFFSET_NAME))
141: result++;
142: if(options.get(PA_COLUMN_SEPARATOR_NAME))
143: result++;
144: if(options.get(PA_COLUMN_ENCLOSER_NAME))
145: result++;
146: return result;
147: }
148:
1.178 paf 149: //http request stuff
1.154 paf 150: #ifdef PA_HTTP
151:
152: #undef CRLF
153: #define CRLF "\r\n"
1.126 paf 154:
155: static bool set_addr(struct sockaddr_in *addr, const char* host, const short port){
156: memset(addr, 0, sizeof(*addr));
157: addr->sin_family=AF_INET;
158: addr->sin_port=htons(port);
159: if(host) {
1.175 paf 160: ulong packed_ip=inet_addr(host);
1.184 paf 161: if(packed_ip!=INADDR_NONE)
1.185 paf 162: memcpy(&addr->sin_addr, &packed_ip, sizeof(packed_ip));
1.184 paf 163: else {
164: struct hostent *hostIP=gethostbyname(host);
165: if(hostIP)
166: memcpy(&addr->sin_addr, hostIP->h_addr, hostIP->h_length);
167: else
168: return false;
169: }
1.126 paf 170: } else
171: addr->sin_addr.s_addr=INADDR_ANY;
172: return true;
173: }
174:
1.207 ! paf 175: size_t guess_content_length(char* buf) {
! 176: char* ptr;
! 177: if((ptr=strstr(buf, "Content-Length:"))) // Apache
! 178: goto found;
! 179: if((ptr=strstr(buf, "content-length:"))) // Parser 3
! 180: goto found;
! 181: if((ptr=strstr(buf, "Content-length:"))) // maybe 1
! 182: goto found;
! 183: if((ptr=strstr(buf, "CONTENT-LENGTH:"))) // maybe 2
! 184: goto found;
! 185: return 0;
! 186: found:
! 187: char *error_pos;
! 188: size_t result=(size_t)strtol(ptr+15/*strlen("CONTENT-LENGTH:")*/, &error_pos, 0);
! 189:
! 190: const size_t reasonable_initial_max=0x400*0x400*10 /*10M*/;
! 191: if(result>reasonable_initial_max) // sanity check
! 192: return reasonable_initial_max;
! 193: return 0;//result;
! 194: }
! 195:
! 196: static int http_read_response(char*& response, size_t& response_size, int sock, bool fail_on_status_ne_200) {
1.154 paf 197: int result=0;
1.207 ! paf 198: // fetching some to local buffer, guessing on possible content-length
! 199: response_size=0x400*20; // initial size if content-length could not be determined
! 200: const size_t preview_size=0x400*20;
! 201: char preview_buf[preview_size+1/*terminator*/]; // 20K buffer to preview headers
! 202: ssize_t received_size=recv(sock, preview_buf, preview_size, 0);
! 203: if(received_size==0)
! 204: goto done;
! 205: if(received_size<0) {
! 206: if(int no=pa_socks_errno())
! 207: throw Exception("http.timeout",
! 208: 0,
! 209: "error receiving response header: %s (%d)", pa_socks_strerr(no), no);
! 210: goto done;
! 211: }
! 212: // detecting response_size
! 213: {
! 214: preview_buf[received_size]=0; // terminator
! 215: if(size_t content_length=guess_content_length(preview_buf))
! 216: response_size=preview_size+content_length; // a little more than needed, will adjust response_size by actual received size later
! 217: }
! 218:
! 219: // allocating initial buf
! 220: response=(char*)pa_malloc_atomic(response_size+1/*terminator*/); // just setting memory block type
! 221: char* ptr=response;
! 222: size_t todo_size=response_size;
! 223: // coping part of already received body
! 224: memcpy(ptr, preview_buf, received_size);
! 225: ptr+=received_size;
! 226: todo_size-=received_size;
! 227:
! 228: // we use terminator byte for two purposes here:
! 229: // 1. we return there zero always, not knowing: maybe they would want to create String form $file.body?
! 230: // invariant: all Strings should have zero-terminated buffers
! 231: // 2. we use that out-of-size byte to detect if our content-length guess was wrong
! 232: // when recv gets more than we expected
! 233: // a) we know that the content-length guess was wrong
! 234: // b) we have space to put the first byte of extra data
! 235: // c) we use less code to detect normal situation: on last while-cycle recv expected to just return 0
1.126 paf 236: while(true) {
1.207 ! paf 237: received_size=recv(sock, ptr, todo_size+1/*there is always a place for terminator*/, 0);
! 238: if(received_size==0) {
! 239: response_size-=todo_size; // in case we received less than expected, cut down the reported size
1.204 paf 240: break;
1.207 ! paf 241: }
1.204 paf 242: if(received_size<0) {
1.196 paf 243: if(int no=pa_socks_errno())
1.197 paf 244: throw Exception("http.timeout",
1.196 paf 245: 0,
1.207 ! paf 246: "error receiving response body: %s (%d)", pa_socks_strerr(no), no);
1.126 paf 247: break;
1.196 paf 248: }
1.207 ! paf 249: // they've touched the terminator?
! 250: if((size_t)received_size>todo_size)
! 251: {
! 252: // that means that our guessed response_size was not big enough
! 253: const size_t grow_chunk_size=0x400*0x400; // 1M
! 254: response_size+=grow_chunk_size;
! 255: size_t ptr_offset=ptr-response;
! 256: response=(char*)pa_realloc(response, response_size+1/*terminator*/);
! 257: ptr=response+ptr_offset;
! 258: todo_size+=grow_chunk_size;
! 259: }
! 260: // can't do this before realloc: we need <todo_size check
! 261: ptr+=received_size;
! 262: todo_size-=received_size;
1.171 paf 263:
1.207 ! paf 264: char* EOLat=0;
1.192 paf 265: if(!result && (EOLat=strstr(response, "\n"))) { // checking status in first response
1.171 paf 266: const String status_line(pa_strdup(response, EOLat-response));
1.154 paf 267: ArrayString astatus;
268: size_t pos_after=0;
269: status_line.split(astatus, pos_after, " ");
1.182 paf 270: const String& status_code=*astatus.get(astatus.count()>1?1:0);
1.154 paf 271: result=status_code.as_int();
1.142 paf 272:
1.154 paf 273: if(fail_on_status_ne_200 && result!=200)
1.142 paf 274: throw Exception("http.status",
1.154 paf 275: &status_code,
1.142 paf 276: "invalid HTTP response status");
277: }
278: }
1.207 ! paf 279: done:
1.154 paf 280: if(result)
1.207 ! paf 281: {
! 282: response[response_size]=0;
1.154 paf 283: return result;
1.207 ! paf 284: }
1.142 paf 285: else
286: throw Exception("http.response",
287: 0,
1.173 paf 288: "bad response from host - no status found (size=%u)", response_size);
1.126 paf 289: }
290:
291: /* ********************** request *************************** */
292:
293: #if defined(SIGALRM) && defined(HAVE_SIGSETJMP) && defined(HAVE_SIGLONGJMP)
1.145 paf 294: # define PA_USE_ALARM
1.126 paf 295: #endif
296:
1.145 paf 297: #ifdef PA_USE_ALARM
1.126 paf 298: static sigjmp_buf timeout_env;
1.199 paf 299: static void timeout_handler(int /*sig*/){
1.126 paf 300: siglongjmp(timeout_env, 1);
301: }
302: #endif
303:
1.171 paf 304: static int http_request(char*& response, size_t& response_size,
1.193 paf 305: const char* host, short port,
1.152 paf 306: const char* request,
1.196 paf 307: int timeout_secs,
1.152 paf 308: bool fail_on_status_ne_200) {
1.126 paf 309: if(!host)
310: throw Exception("http.host",
1.154 paf 311: 0,
1.126 paf 312: "zero hostname"); //never
313:
1.200 paf 314: volatile // to prevent makeing it register variable, because it will be clobbered by longjmp [thanks gcc warning]
315: int sock=-1;
1.145 paf 316: #ifdef PA_USE_ALARM
1.146 paf 317: signal(SIGALRM, timeout_handler);
1.126 paf 318: #endif
1.145 paf 319: #ifdef PA_USE_ALARM
320: if(sigsetjmp(timeout_env, 1)) {
321: // stupid gcc [2.95.4] generated bad code
322: // which failed to handle sigsetjmp+throw: crashed inside of pre-throw code.
1.199 paf 323: // rewritten simplier [athough duplicating closesocket code]
1.145 paf 324: if(sock>=0)
325: closesocket(sock);
326: throw Exception("http.timeout",
1.199 paf 327: 0,
1.145 paf 328: "timeout occured while retrieving document");
1.146 paf 329: return 0; // never
1.145 paf 330: } else {
1.196 paf 331: alarm(timeout_secs);
1.145 paf 332: #endif
1.146 paf 333: try {
334: int result;
1.126 paf 335: struct sockaddr_in dest;
1.154 paf 336:
337: if(!set_addr(&dest, host, port))
1.126 paf 338: throw Exception("http.host",
1.154 paf 339: 0,
1.127 paf 340: "can not resolve hostname \"%s\"", host);
1.126 paf 341:
1.195 paf 342: if((sock=socket(AF_INET, SOCK_STREAM, IPPROTO_TCP/*0*/))<0) {
343: int no=pa_socks_errno();
1.126 paf 344: throw Exception("http.connect",
1.154 paf 345: 0,
1.195 paf 346: "can not make socket: %s (%d)", pa_socks_strerr(no), no);
347: }
1.196 paf 348:
349: // To enable SO_DONTLINGER (that is, disable SO_LINGER)
350: // l_onoff should be set to zero and setsockopt should be called
351: linger dont_linger={0,0};
352: setsockopt(sock, SOL_SOCKET, SO_LINGER, (const char *)&dont_linger, sizeof(dont_linger));
353:
1.201 paf 354: #ifdef WIN32
355: // SO_*TIMEO can be defined in .h but not implemlemented in protocol,
356: // failing subsequently with Option not supported by protocol (99) message
357: // could not suppress that, so leaving this only for win32
1.196 paf 358: int timeout_ms=timeout_secs*1000;
359: setsockopt(sock, SOL_SOCKET, SO_SNDTIMEO, (const char*)&timeout_ms, sizeof(timeout_ms));
360: setsockopt(sock, SOL_SOCKET, SO_RCVTIMEO, (const char*)&timeout_ms, sizeof(timeout_ms));
361: #endif
362:
1.195 paf 363: if(connect(sock, (struct sockaddr *)&dest, sizeof(dest))) {
364: int no=pa_socks_errno();
1.126 paf 365: throw Exception("http.connect",
1.154 paf 366: 0,
1.195 paf 367: "can not connect to host \"%s\": %s (%d)", host, pa_socks_strerr(no), no);
368: }
1.126 paf 369: size_t request_size=strlen(request);
1.195 paf 370: if(send(sock, request, request_size, 0)!=(ssize_t)request_size) {
371: int no=pa_socks_errno();
1.197 paf 372: throw Exception("http.timeout",
1.154 paf 373: 0,
1.195 paf 374: "error sending request: %s (%d)", pa_socks_strerr(no), no);
375: }
1.126 paf 376:
1.171 paf 377: result=http_read_response(response, response_size, sock, fail_on_status_ne_200);
1.142 paf 378: closesocket(sock);
1.145 paf 379: #ifdef PA_USE_ALARM
1.142 paf 380: alarm(0);
1.126 paf 381: #endif
1.147 paf 382: return result;
1.146 paf 383: } catch(...) {
1.145 paf 384: #ifdef PA_USE_ALARM
1.146 paf 385: alarm(0);
1.126 paf 386: #endif
1.146 paf 387: if(sock>=0)
388: closesocket(sock);
1.154 paf 389: rethrow;
1.146 paf 390: }
1.148 paf 391: #ifdef PA_USE_ALARM
1.126 paf 392: }
1.148 paf 393: #endif
1.126 paf 394: }
395:
1.127 paf 396: #ifndef DOXYGEN
397: struct Http_pass_header_info {
1.154 paf 398: Request_charsets* charsets;
1.127 paf 399: String* request;
400: bool user_agent_specified;
401: };
402: #endif
1.154 paf 403: static void http_pass_header(HashStringValue::key_type key,
404: HashStringValue::value_type value,
405: Http_pass_header_info *info) {
406: *info->request <<key<<": "
407: << attributed_meaning_to_string(*value, String::L_HTTP_HEADER, false)
408: << CRLF;
1.135 paf 409:
1.154 paf 410: if(String(key, String::L_TAINTED).change_case(info->charsets->source(), String::CC_UPPER)=="USER-AGENT")
411: info->user_agent_specified=true;
1.126 paf 412: }
1.154 paf 413:
414:
415: static Charset* detect_charset(Charset& source_charset, const String& content_type_value) {
1.156 paf 416: const String::Body CONTENT_TYPE_VALUE=
1.154 paf 417: content_type_value.change_case(source_charset, String::CC_UPPER);
418: // content-type: xxx/xxx; source_charset=WE-NEED-THIS
419: // content-type: xxx/xxx; source_charset="WE-NEED-THIS"
420: // content-type: xxx/xxx; source_charset="WE-NEED-THIS";
421: size_t before_charseteq_pos=CONTENT_TYPE_VALUE.pos("CHARSET=");
422: if(before_charseteq_pos!=STRING_NOT_FOUND) {
423: size_t charset_begin=before_charseteq_pos+8/*CHARSET="*/;
424: size_t open_quote_pos=CONTENT_TYPE_VALUE.pos('"', charset_begin);
425: bool quoted=open_quote_pos==charset_begin;
426: if(quoted)
427: charset_begin++; // skip opening '"'
428: size_t charset_end=CONTENT_TYPE_VALUE.length();
429: if(quoted) {
430: size_t close_quote_pos=CONTENT_TYPE_VALUE.pos('"', charset_begin);
431: if(close_quote_pos!=STRING_NOT_FOUND)
432: charset_end=close_quote_pos;
433: } else {
434: size_t delim_pos=CONTENT_TYPE_VALUE.pos(';', charset_begin);
435: if(delim_pos!=STRING_NOT_FOUND)
436: charset_end=delim_pos;
437: }
1.156 paf 438: const String::Body CHARSET_NAME_BODY=
1.154 paf 439: CONTENT_TYPE_VALUE.mid(charset_begin, charset_end);
440:
441: return &charsets.get(CHARSET_NAME_BODY);
442: }
443:
444: return 0;
445: }
446:
1.178 paf 447: static const String* basic_authorization_field(const char* user, const char* pass) {
448: if(!user&& !pass)
449: return 0;
450:
451: String combined;
452: if(user)
453: combined<<user;
454: combined<<":";
455: if(pass)
456: combined<<pass;
457:
458: String* result=new String("Basic "); *result<<pa_base64(combined.cstr(), combined.length());
459: return result;
460: }
1.154 paf 461:
1.181 paf 462: static void form_string_value2string(
463: HashStringValue::key_type key,
464: const String& value,
465: String& result)
466: {
467: result << String(key, String::L_TAINTED) << "=";
468: result.append(value, String::L_URI, true);
469: result<< "&";
470: }
471: #ifndef DOXYGEN
472: struct Form_table_value2string_info {
473: HashStringValue::key_type key;
474: String& result;
475:
476: Form_table_value2string_info(HashStringValue::key_type akey, String& aresult):
477: key(akey), result(aresult) {}
478: };
479: #endif
480: static void form_table_value2string(Table::element_type row, Form_table_value2string_info* info) {
481: form_string_value2string(info->key, *row->get(0), info->result);
482: }
1.180 paf 483: static void form_value2string(
484: HashStringValue::key_type key,
485: HashStringValue::value_type value,
486: String* result)
487: {
1.181 paf 488: if(const String* svalue=value->get_string())
489: form_string_value2string(key, *svalue, *result);
490: else if(Table* tvalue=value->get_table()) {
491: Form_table_value2string_info info(key, *result);
492: tvalue->for_each(form_table_value2string, &info);
493: } else
494: throw Exception(0,
495: new String(key, String::L_TAINTED),
496: "is %s, "HTTP_FORM_NAME" option value must either string or table", value->type());
1.180 paf 497: }
498: static const char* form2string(HashStringValue& form) {
499: String string;
500: form.for_each(form_value2string, &string);
1.181 paf 501: return string.cstr(String::L_UNSPECIFIED);
1.180 paf 502: }
1.154 paf 503: #ifndef DOXYGEN
504: struct File_read_http_result {
505: char *str; size_t length;
506: HashStringValue* headers;
507: };
508: #endif
1.194 paf 509: static void find_headers_end(char* p,
510: char*& headers_end_at,
511: char*& raw_body)
512: {
513: raw_body=p;
514: // \n\n
515: // \r\n\r\n
516: while((p=strchr(p, '\n'))) {
517: headers_end_at=++p; // \n>.<
518: if(*p=='\r') // \r\n>\r?<\n
519: p++;
520: if(*p=='\n') { // \r\n\r>\n?<
521: raw_body=p+1;
522: return;
523: }
524: }
525: headers_end_at=0;
526: }
527:
1.155 paf 528: /// @todo build .cookies field. use ^file.tables.SET-COOKIES.menu{ for now
1.154 paf 529: static File_read_http_result file_read_http(Request_charsets& charsets,
530: const String& file_spec,
1.169 paf 531: bool as_text,
532: HashStringValue *options=0) {
1.154 paf 533: File_read_http_result result;
1.126 paf 534: char host[MAX_STRING];
1.129 paf 535: const char* uri;
1.193 paf 536: short port;
1.180 paf 537: const char* method="GET"; bool method_is_get;
538: HashStringValue* form=0;
539: const char* body_cstr=0;
1.196 paf 540: int timeout_secs=2;
1.142 paf 541: bool fail_on_status_ne_200=true;
1.154 paf 542: Value* vheaders=0;
543: Charset *asked_remote_charset=0;
1.178 paf 544: const char* user_cstr=0;
545: const char* password_cstr=0;
1.126 paf 546:
1.127 paf 547: if(options) {
1.206 paf 548: int valid_options=get_valid_file_options_count(*options);
549:
1.177 paf 550: if(Value* vmethod=options->get(HTTP_METHOD_NAME)) {
1.127 paf 551: valid_options++;
1.154 paf 552: method=vmethod->as_string().cstr();
1.181 paf 553: }
554: if(Value* vform=options->get(HTTP_FORM_NAME)) {
1.180 paf 555: valid_options++;
556: form=vform->get_hash();
1.181 paf 557: }
558: if(Value* vbody=options->get(HTTP_BODY_NAME)) {
1.180 paf 559: valid_options++;
560: body_cstr=vbody->as_string().cstr(String::L_UNSPECIFIED);
1.181 paf 561: }
562: if(Value* vtimeout=options->get(HTTP_TIMEOUT_NAME)) {
1.127 paf 563: valid_options++;
1.196 paf 564: timeout_secs=vtimeout->as_int();
1.181 paf 565: }
566: if((vheaders=options->get(HTTP_HEADERS_NAME))) {
1.127 paf 567: valid_options++;
1.181 paf 568: }
569: if(Value* vany_status=options->get(HTTP_ANY_STATUS_NAME)) {
1.142 paf 570: valid_options++;
571: fail_on_status_ne_200=!vany_status->as_bool();
1.181 paf 572: }
573: if(Value* vcharset_name=options->get(HTTP_CHARSET_NAME)) {
1.154 paf 574: valid_options++;
575: asked_remote_charset=&::charsets.get(vcharset_name->as_string().
576: change_case(charsets.source(), String::CC_UPPER));
1.181 paf 577: }
578: if(Value* vuser=options->get(HTTP_USER)) {
1.178 paf 579: valid_options++;
580: user_cstr=vuser->as_string().cstr();
1.181 paf 581: }
582: if(Value* vpassword=options->get(HTTP_PASSWORD)) {
1.178 paf 583: valid_options++;
584: password_cstr=vpassword->as_string().cstr();
585: }
1.142 paf 586:
1.154 paf 587: if(valid_options!=options->count())
1.127 paf 588: throw Exception("parser.runtime",
589: 0,
590: "invalid option passed");
1.154 paf 591: }
592: if(!asked_remote_charset) // defaulting to $request:charset
593: asked_remote_charset=&charsets.source();
594:
1.180 paf 595: method_is_get=strcmp(method, "GET")==0;
596: if(method_is_get && body_cstr)
597: throw Exception("parser.runtime",
598: 0,
599: "you can not use $."HTTP_BODY_NAME" option with method GET");
600:
1.154 paf 601: //preparing request
602: String& connect_string=*new String;
603: // not in ^sql{... L_SQL ...} spirit, but closer to ^file::load one
604: connect_string.append(file_spec, String::L_URI); // tainted pieces -> URI pieces
605:
1.180 paf 606: String request_head_and_body;
1.154 paf 607: {
608: // influence URLencoding of tainted pieces to String::L_URI lang
609: Temp_client_charset temp(charsets, *asked_remote_charset);
610:
611: const char* connect_string_cstr=connect_string.cstr(String::L_UNSPECIFIED);
1.126 paf 612:
1.154 paf 613: const char* current=connect_string_cstr;
614: if(strncmp(current, "http://", 7)!=0)
615: throw Exception(0,
616: &connect_string,
617: "does not start with http://"); //never
618: current+=7;
619:
620: strncpy(host, current, sizeof(host)-1); host[sizeof(host)-1]=0;
621: char* host_uri=lsplit(host, '/');
622: uri=host_uri?current+(host_uri-1-host):"/";
623: char* port_cstr=lsplit(host, ':');
624: char* error_pos=0;
1.193 paf 625: port=port_cstr?(short)strtol(port_cstr, &error_pos, 0):80;
1.154 paf 626:
1.180 paf 627: if(strchr(uri, '?') && form)
628: throw Exception("parser.runtime",
629: 0,
630: "use either uri with ?params or $."HTTP_FORM_NAME" option");
631:
632: //making request head
633: String head;
634: head << method;
635: head << " " << uri;
636: if(form)
637: if(method_is_get)
1.181 paf 638: head << "?" << form2string(*form);
1.180 paf 639: head <<" HTTP/1.0" CRLF
1.154 paf 640: "host: "<< host << CRLF;
1.181 paf 641: if(form && !method_is_get) {
642: head << "content-type: application/x-www-form-urlencoded" CRLF;
643: body_cstr = form2string(*form);
644: }
1.178 paf 645:
1.179 paf 646: // http://www.ietf.org/rfc/rfc2617.txt
1.178 paf 647: if(const String* authorization_field_value=basic_authorization_field(user_cstr, password_cstr))
1.180 paf 648: head<<"authorization: "<<*authorization_field_value<<CRLF;
1.178 paf 649:
1.154 paf 650: bool user_agent_specified=false;
651: if(vheaders && !vheaders->is_string()) { // allow empty
652: if(HashStringValue *headers=vheaders->get_hash()) {
1.180 paf 653: Http_pass_header_info info={&charsets, &head, false};
1.154 paf 654: headers->for_each(http_pass_header, &info);
655: user_agent_specified=info.user_agent_specified;
656: } else
657: throw Exception("parser.runtime",
658: &connect_string,
659: "headers param must be hash");
660: };
661: if(!user_agent_specified) // defaulting
1.180 paf 662: head << "user-agent: " DEFAULT_USER_AGENT CRLF;
663:
664: if(body_cstr) {
665: // recode those pieces which are not in String::L_URI lang
666: // [those violating HTTP standard, but widly used]
667: body_cstr=Charset::transcode(
668: String::C(body_cstr, strlen(body_cstr)),
669: charsets.source(),
670: *asked_remote_charset);
1.181 paf 671:
672: head << "content-length: " << format(strlen(body_cstr), "%u") << CRLF;
1.180 paf 673: }
1.154 paf 674:
1.181 paf 675: const char* head_cstr=head.cstr(String::L_UNSPECIFIED);
676:
677: // recode those pieces which are not in String::L_URI lang
678: // [those violating HTTP standard, but widly used]
679: head_cstr=Charset::transcode(
680: String::C(head_cstr, strlen(head_cstr)),
681: charsets.source(),
682: *asked_remote_charset);
1.180 paf 683:
1.181 paf 684: // head + end of header
685: request_head_and_body << head_cstr << CRLF;
1.180 paf 686: // body
687: if(body_cstr)
688: request_head_and_body << body_cstr;
1.154 paf 689: }
1.126 paf 690:
691: //sending request
1.171 paf 692: char* response;
693: size_t response_size;
694: int status_code=http_request(response, response_size,
1.180 paf 695: host, port, request_head_and_body.cstr(),
1.196 paf 696: timeout_secs, fail_on_status_ne_200);
1.126 paf 697:
698: //processing results
1.171 paf 699: char* raw_body; size_t raw_body_size;
1.194 paf 700: char* headers_end_at;
701: find_headers_end(response,
702: headers_end_at,
703: raw_body);
1.191 paf 704: raw_body_size=response_size-(raw_body-response);
1.171 paf 705:
1.154 paf 706: result.headers=new HashStringValue;
1.155 paf 707: VHash* vtables=new VHash;
1.177 paf 708: result.headers->put(HTTP_TABLES_NAME, vtables);
1.194 paf 709: Charset* real_remote_charset=0; // undetected, yet
710:
711: if(headers_end_at) {
712: *headers_end_at=0;
713: const String header_block(String::C(response, headers_end_at-response), true);
714:
715: ArrayString aheaders;
716: HashStringValue& tables=vtables->hash();
1.155 paf 717:
1.194 paf 718: size_t pos_after=0;
719: header_block.split(aheaders, pos_after, "\n");
720:
721: //processing headers
722: size_t aheaders_count=aheaders.count();
723: for(size_t i=1; i<aheaders_count; i++) {
724: const String& line=*aheaders.get(i);
725: size_t pos=line.pos(':');
726: if(pos==STRING_NOT_FOUND || pos<1)
727: throw Exception("http.response",
728: &connect_string,
729: "bad response from host - bad header \"%s\"", line.cstr());
730: const String::Body HEADER_NAME=
731: line.mid(0, pos).change_case(charsets.source(), String::CC_UPPER);
732: const String& header_value=line.mid(pos+1, line.length()).trim(String::TRIM_BOTH, " \t\r");
733: if(as_text && HEADER_NAME=="CONTENT-TYPE")
734: real_remote_charset=detect_charset(charsets.source(), header_value);
735:
736: // tables
737: {
738: Value *valready=(Value *)tables.get(HEADER_NAME);
739: bool existed=valready!=0;
740: Table *table;
741: if(existed) {
742: // second+ appearence
743: table=valready->get_table();
744: } else {
745: // first appearence
746: Table::columns_type columns =new ArrayString(1);
747: *columns+=new String("value");
748: table=new Table(columns);
749: }
750: // this string becomes next row
751: ArrayString& row=*new ArrayString(1);
752: row+=&header_value;
753: *table+=&row;
754: // not existed before? add it
755: if(!existed)
756: tables.put(HEADER_NAME, new VTable(table));
1.155 paf 757: }
1.194 paf 758:
759: result.headers->put(HEADER_NAME, new VString(header_value));
1.155 paf 760: }
1.126 paf 761: }
762:
763: // output response
1.171 paf 764: String::C real_body=String::C(raw_body, raw_body_size);
1.192 paf 765: if(as_text && raw_body_size) { // must be checked because transcode returns CONST string in case length==0, which contradicts hacking few lines below
766: // defaulting to used-asked charset [it's never empty!]
767: if(!real_remote_charset)
768: real_remote_charset=asked_remote_charset;
1.169 paf 769: real_body=Charset::transcode(real_body, *real_remote_charset, charsets.source());
1.192 paf 770: }
1.154 paf 771:
772: result.str=const_cast<char *>(real_body.str); // hacking a little
773: result.length=real_body.length;
774: result.headers->put(file_status_name, new VInt(status_code));
775: return result;
1.34 paf 776: }
1.123 paf 777:
1.154 paf 778: #endif
779:
1.123 paf 780: #ifndef DOXYGEN
781: struct File_read_action_info {
1.154 paf 782: char **data; size_t *data_size;
1.188 paf 783: char* buf; size_t offset; size_t count;
1.126 paf 784: };
1.123 paf 785: #endif
1.154 paf 786: static void file_read_action(
787: struct stat& finfo,
788: int f,
1.166 paf 789: const String& file_spec, const char* /*fname*/, bool as_text,
1.154 paf 790: void *context) {
1.126 paf 791: File_read_action_info& info=*static_cast<File_read_action_info *>(context);
1.188 paf 792: size_t to_read_size=info.count;
793: if(!to_read_size)
794: to_read_size=(size_t)finfo.st_size;
795: assert( !(info.buf && as_text) );
796: if(to_read_size) {
797: if(info.offset)
798: lseek(f, info.offset, SEEK_SET);
799: *info.data=info.buf
800: ? info.buf
801: : new(PointerFreeGC) char[to_read_size+(as_text?1:0)];
1.126 paf 802: *info.data_size=(size_t)read(f, *info.data, to_read_size);
1.123 paf 803:
804: if(ssize_t(*info.data_size)<0 || *info.data_size>to_read_size)
1.126 paf 805: throw Exception(0,
1.123 paf 806: &file_spec,
1.173 paf 807: "read failed: actually read %u bytes count not in [0..%u] valid range",
1.126 paf 808: *info.data_size, to_read_size);
1.123 paf 809: } else { // empty file
810: if(as_text) {
1.154 paf 811: *info.data=new(PointerFreeGC) char[1];
1.123 paf 812: *(char*)(*info.data)=0;
813: } else
814: *info.data=0;
815: *info.data_size=0;
816: return;
817: }
1.126 paf 818: }
1.154 paf 819: File_read_result file_read(Request_charsets& charsets, const String& file_spec,
820: bool as_text, HashStringValue *params,
1.188 paf 821: bool fail_on_read_problem,
822: char* buf, size_t offset, size_t count) {
1.167 paf 823: File_read_result result={false, 0, 0, 0};
1.154 paf 824: #ifdef PA_HTTP
825: if(file_spec.starts_with("http://")) {
1.203 paf 826: if(offset || count)
827: throw Exception("parser.runtime",
828: 0,
829: "offset and load options are not supported for HTTP:// file load");
830:
1.126 paf 831: // fail on read problem
1.169 paf 832: File_read_http_result http=file_read_http(charsets, file_spec, as_text, params);
1.154 paf 833: result.success=true;
834: result.str=http.str;
835: result.length=http.length;
836: result.headers=http.headers;
1.126 paf 837: } else {
1.154 paf 838: #endif
1.206 paf 839: if(params) {
840: int valid_options=get_valid_file_options_count(*params);
841: if(valid_options!=params->count())
842: throw Exception("parser.runtime",
843: 0,
844: "invalid option passed");
845: }
1.161 paf 846:
1.188 paf 847: File_read_action_info info={&result.str, &result.length,
848: buf, offset, count};
1.154 paf 849: result.success=file_read_action_under_lock(file_spec,
1.126 paf 850: "read", file_read_action, &info,
851: as_text, fail_on_read_problem);
1.154 paf 852: #ifdef PA_HTTP
1.126 paf 853: }
1.154 paf 854: #endif
1.123 paf 855:
1.154 paf 856: if(result.success && as_text) {
1.131 paf 857: // UTF-8 signature: EF BB BF
1.154 paf 858: if(result.length>=3) {
859: char *in=(char *)result.str;
1.159 paf 860: if(strncmp(in, "\xEF\xBB\xBF", 3)==0) {
1.154 paf 861: result.str=in+3; result.length-=3;// skip prefix
1.131 paf 862: }
863: }
864:
1.154 paf 865: fix_line_breaks((char *)(result.str), result.length);
1.123 paf 866: }
1.126 paf 867:
868: return result;
1.123 paf 869: }
870:
1.154 paf 871: #ifdef PA_SAFE_MODE
872: void check_safe_mode(struct stat finfo, const String& file_spec, const char* fname) {
873: if(finfo.st_uid/*foreign?*/!=geteuid()
874: && finfo.st_gid/*foreign?*/!=getegid())
875: throw Exception("parser.runtime",
876: &file_spec,
877: "parser is in safe mode: "
878: "reading files of foreign group and user disabled "
879: "[recompile parser with --disable-safe-mode configure option], "
880: "actual filename '%s', "
881: "fuid(%d)!=euid(%d) or fgid(%d)!=egid(%d)",
882: fname,
883: finfo.st_uid, geteuid(),
884: finfo.st_gid, getegid());
885: }
886: #endif
1.149 paf 887:
1.154 paf 888: bool file_read_action_under_lock(const String& file_spec,
1.126 paf 889: const char* action_name, File_read_action action, void *context,
890: bool as_text,
1.123 paf 891: bool fail_on_read_problem) {
1.154 paf 892: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.33 paf 893: int f;
894:
895: // first open, next stat:
1.45 paf 896: // directory update of NTFS hard links performed on open.
1.33 paf 897: // ex:
898: // a.html:^test[] and b.html hardlink to a.html
899: // user inserts ! before ^test in a.html
1.126 paf 900: // directory entry of b.html in NTFS not updated at once,
1.35 paf 901: // they delay update till open, so we would receive "!^test[" string
902: // if would do stat, next open.
1.123 paf 903: // later: it seems, even this does not help sometimes
1.98 paf 904: if((f=open(fname, O_RDONLY|(as_text?_O_TEXT:_O_BINARY)))>=0) {
1.123 paf 905: try {
1.162 paf 906: if(pa_lock_shared_blocking(f)!=0)
1.126 paf 907: throw Exception("file.lock",
1.123 paf 908: &file_spec,
909: "shared lock failed: %s (%d), actual filename '%s'",
1.154 paf 910: strerror(errno), errno, fname);
1.123 paf 911:
1.124 paf 912: struct stat finfo;
913: if(stat(fname, &finfo)!=0)
914: throw Exception("file.missing", // hardly possible: we just opened it OK
915: &file_spec,
916: "stat failed: %s (%d), actual filename '%s'",
1.154 paf 917: strerror(errno), errno, fname);
1.124 paf 918:
1.140 paf 919: #ifdef PA_SAFE_MODE
1.149 paf 920: check_safe_mode(finfo, file_spec, fname);
1.105 paf 921: #endif
1.32 paf 922:
1.154 paf 923: action(finfo, f, file_spec, fname, as_text, context);
1.123 paf 924: } catch(...) {
1.162 paf 925: pa_unlock(f);close(f);
1.123 paf 926: if(fail_on_read_problem)
1.154 paf 927: rethrow;
1.123 paf 928: return false;
929: }
1.87 paf 930:
1.162 paf 931: pa_unlock(f);close(f);
1.72 parser 932: return true;
1.118 paf 933: } else {
934: if(fail_on_read_problem)
1.126 paf 935: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.118 paf 936: &file_spec,
1.123 paf 937: "%s failed: %s (%d), actual filename '%s'",
1.154 paf 938: action_name, strerror(errno), errno, fname);
1.118 paf 939: return false;
940: }
1.8 paf 941: }
942:
1.202 paf 943: void create_dir_for_file(const String& file_spec) {
1.63 parser 944: size_t pos_after=1;
1.154 paf 945: size_t pos_before;
946: while((pos_before=file_spec.pos('/', pos_after))!=STRING_NOT_FOUND) {
947: mkdir(file_spec.mid(0, pos_before).cstr(String::L_FILE_SPEC), 0775);
1.63 parser 948: pos_after=pos_before+1;
949: }
950: }
951:
1.98 paf 952: bool file_write_action_under_lock(
1.28 paf 953: const String& file_spec,
1.126 paf 954: const char* action_name, File_write_action action, void *context,
955: bool as_text,
956: bool do_append,
957: bool do_block,
1.110 paf 958: bool fail_on_lock_problem) {
1.154 paf 959: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.28 paf 960: int f;
1.80 paf 961: if(access(fname, W_OK)!=0) // no
1.126 paf 962: create_dir_for_file(file_spec);
1.50 paf 963:
1.80 paf 964: if((f=open(fname,
965: O_CREAT|O_RDWR
966: |(as_text?_O_TEXT:_O_BINARY)
1.138 paf 967: |(do_append?O_APPEND:PA_O_TRUNC), 0664))>=0) {
1.162 paf 968: if((do_block?pa_lock_exclusive_blocking(f):pa_lock_exclusive_nonblocking(f))!=0) {
1.126 paf 969: Exception e("file.lock",
1.110 paf 970: &file_spec,
971: "shared lock failed: %s (%d), actual filename '%s'",
1.154 paf 972: strerror(errno), errno, fname);
1.126 paf 973: close(f);
1.110 paf 974: if(fail_on_lock_problem)
975: throw e;
1.98 paf 976: return false;
977: }
1.96 paf 978:
1.158 paf 979: try {
1.126 paf 980: action(f, context);
1.158 paf 981: } catch(...) {
1.138 paf 982: #ifdef HAVE_FTRUNCATE
1.104 paf 983: if(!do_append)
1.125 paf 984: ftruncate(f, lseek(f, 0, SEEK_CUR)); // one can not use O_TRUNC, read lower
1.138 paf 985: #endif
1.162 paf 986: pa_unlock(f);close(f);
1.154 paf 987: rethrow;
1.158 paf 988: }
1.80 paf 989:
1.138 paf 990: #ifdef HAVE_FTRUNCATE
1.104 paf 991: if(!do_append)
1.125 paf 992: 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 993: #endif
1.162 paf 994: pa_unlock(f);close(f);
1.98 paf 995: return true;
1.80 paf 996: } else
1.126 paf 997: throw Exception(errno==EACCES?"file.access":0,
1.80 paf 998: &file_spec,
1.96 paf 999: "%s failed: %s (%d), actual filename '%s'",
1.154 paf 1000: action_name, strerror(errno), errno, fname);
1.96 paf 1001: // here should be nothing, see rethrow above
1002: }
1003:
1004: #ifndef DOXYGEN
1005: struct File_write_action_info {
1.154 paf 1006: const char* str; size_t length;
1.126 paf 1007: };
1.96 paf 1008: #endif
1009: static void file_write_action(int f, void *context) {
1.126 paf 1010: File_write_action_info& info=*static_cast<File_write_action_info *>(context);
1.154 paf 1011: if(info.length) {
1012: int written=write(f, info.str, info.length);
1.116 paf 1013: if(written<0)
1.126 paf 1014: throw Exception(0,
1015: 0,
1016: "write failed: %s (%d)", strerror(errno), errno);
1.113 paf 1017: }
1.96 paf 1018: }
1019: void file_write(
1020: const String& file_spec,
1.154 paf 1021: const char* data, size_t size,
1.126 paf 1022: bool as_text,
1.96 paf 1023: bool do_append) {
1.126 paf 1024: File_write_action_info info={data, size};
1.98 paf 1025: file_write_action_under_lock(
1.154 paf 1026: file_spec,
1027: "write", file_write_action, &info,
1028: as_text,
1029: do_append);
1.30 paf 1030: }
1031:
1.63 parser 1032: // throws nothing! [this is required in file_move & file_delete]
1.50 paf 1033: static void rmdir(const String& file_spec, size_t pos_after) {
1.154 paf 1034: size_t pos_before;
1035: if((pos_before=file_spec.pos('/', pos_after))!=STRING_NOT_FOUND)
1.126 paf 1036: rmdir(file_spec, pos_before+1);
1.50 paf 1037:
1.154 paf 1038: rmdir(file_spec.mid(0, pos_after-1/* / */).cstr(String::L_FILE_SPEC));
1.50 paf 1039: }
1.164 paf 1040: bool file_delete(const String& file_spec, bool fail_on_problem) {
1.154 paf 1041: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.54 parser 1042: if(unlink(fname)!=0)
1.164 paf 1043: if(fail_on_problem)
1.126 paf 1044: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.93 paf 1045: &file_spec,
1046: "unlink failed: %s (%d), actual filename '%s'",
1.154 paf 1047: strerror(errno), errno, fname);
1.93 paf 1048: else
1049: return false;
1.50 paf 1050:
1.126 paf 1051: rmdir(file_spec, 1);
1.93 paf 1052: return true;
1.60 parser 1053: }
1.95 paf 1054: void file_move(const String& old_spec, const String& new_spec) {
1.154 paf 1055: const char* old_spec_cstr=old_spec.cstr(String::L_FILE_SPEC);
1056: const char* new_spec_cstr=new_spec.cstr(String::L_FILE_SPEC);
1.63 parser 1057:
1.126 paf 1058: create_dir_for_file(new_spec);
1.63 parser 1059:
1.60 parser 1060: if(rename(old_spec_cstr, new_spec_cstr)!=0)
1.126 paf 1061: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.60 parser 1062: &old_spec,
1063: "rename failed: %s (%d), actual filename '%s' to '%s'",
1.154 paf 1064: strerror(errno), errno, old_spec_cstr, new_spec_cstr);
1.63 parser 1065:
1.126 paf 1066: rmdir(old_spec, 1);
1.31 paf 1067: }
1068:
1.51 paf 1069:
1.126 paf 1070: bool entry_exists(const char* fname, struct stat *afinfo) {
1.118 paf 1071: struct stat lfinfo;
1072: bool result=stat(fname, &lfinfo)==0;
1073: if(afinfo)
1074: *afinfo=lfinfo;
1075: return result;
1.119 paf 1076: }
1077:
1078: bool entry_exists(const String& file_spec) {
1.154 paf 1079: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.126 paf 1080: return entry_exists(fname, 0);
1.118 paf 1081: }
1082:
1.51 paf 1083: static bool entry_readable(const String& file_spec, bool need_dir) {
1.154 paf 1084: char* fname=file_spec.cstrm(String::L_FILE_SPEC);
1.120 paf 1085: if(need_dir) {
1.126 paf 1086: size_t size=strlen(fname);
1.120 paf 1087: while(size) {
1.126 paf 1088: char c=fname[size-1];
1.120 paf 1089: if(c=='/' || c=='\\')
1090: fname[--size]=0;
1091: else
1092: break;
1093: }
1094: }
1.51 paf 1095: struct stat finfo;
1.118 paf 1096: if(access(fname, R_OK)==0 && entry_exists(fname, &finfo)) {
1.109 paf 1097: bool is_dir=(finfo.st_mode&S_IFDIR) != 0;
1.51 paf 1098: return is_dir==need_dir;
1099: }
1100: return false;
1101: }
1.31 paf 1102: bool file_readable(const String& file_spec) {
1.126 paf 1103: return entry_readable(file_spec, false);
1.51 paf 1104: }
1105: bool dir_readable(const String& file_spec) {
1.126 paf 1106: return entry_readable(file_spec, true);
1.65 parser 1107: }
1.154 paf 1108: const String* file_readable(const String& path, const String& name) {
1109: String& result=*new String(path);
1110: result << "/";
1111: result << name;
1112: return file_readable(result)?&result:0;
1.43 paf 1113: }
1114: bool file_executable(const String& file_spec) {
1.154 paf 1115: return access(file_spec.cstr(String::L_FILE_SPEC), X_OK)==0;
1.44 paf 1116: }
1117:
1.64 parser 1118: bool file_stat(const String& file_spec,
1.58 parser 1119: size_t& rsize,
1.126 paf 1120: time_t& ratime,
1121: time_t& rmtime,
1122: time_t& rctime,
1.64 parser 1123: bool fail_on_read_problem) {
1.154 paf 1124: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1125: struct stat finfo;
1.44 paf 1126: if(stat(fname, &finfo)!=0)
1.64 parser 1127: if(fail_on_read_problem)
1.126 paf 1128: throw Exception("file.missing",
1.67 parser 1129: &file_spec,
1130: "getting file size failed: %s (%d), real filename '%s'",
1.154 paf 1131: strerror(errno), errno, fname);
1.64 parser 1132: else
1133: return false;
1.58 parser 1134: rsize=finfo.st_size;
1135: ratime=finfo.st_atime;
1136: rmtime=finfo.st_mtime;
1137: rctime=finfo.st_ctime;
1.64 parser 1138: return true;
1.18 paf 1139: }
1140:
1.126 paf 1141: char* getrow(char* *row_ref, char delim) {
1142: char* result=*row_ref;
1.8 paf 1143: if(result) {
1.126 paf 1144: *row_ref=strchr(result, delim);
1.8 paf 1145: if(*row_ref)
1146: *((*row_ref)++)=0;
1147: else if(!*result)
1148: return 0;
1149: }
1150: return result;
1151: }
1152:
1.126 paf 1153: char* lsplit(char* string, char delim) {
1.23 paf 1154: if(string) {
1.126 paf 1155: char* v=strchr(string, delim);
1.8 paf 1156: if(v) {
1157: *v=0;
1158: return v+1;
1159: }
1160: }
1161: return 0;
1162: }
1163:
1.126 paf 1164: char* lsplit(char* *string_ref, char delim) {
1165: char* result=*string_ref;
1166: char* next=lsplit(*string_ref, delim);
1.8 paf 1167: *string_ref=next;
1168: return result;
1.9 paf 1169: }
1170:
1.126 paf 1171: char* rsplit(char* string, char delim) {
1.18 paf 1172: if(string) {
1.126 paf 1173: char* v=strrchr(string, delim);
1.18 paf 1174: if(v) {
1.9 paf 1175: *v=0;
1176: return v+1;
1177: }
1178: }
1179: return NULL;
1.10 paf 1180: }
1181:
1.37 paf 1182: /// @todo less stupid type detection
1.154 paf 1183: const char* format(double value, char* fmt) {
1.126 paf 1184: char local_buf[MAX_NUMBER];
1.108 paf 1185: size_t size;
1186:
1.10 paf 1187: if(fmt)
1188: if(strpbrk(fmt, "diouxX"))
1189: if(strpbrk(fmt, "ouxX"))
1.126 paf 1190: size=snprintf(local_buf, sizeof(local_buf), fmt, (uint)value);
1.10 paf 1191: else
1.126 paf 1192: size=snprintf(local_buf, sizeof(local_buf), fmt, (int)value);
1.10 paf 1193: else
1.126 paf 1194: size=snprintf(local_buf, sizeof(local_buf), fmt, value);
1.10 paf 1195: else
1.126 paf 1196: size=snprintf(local_buf, sizeof(local_buf), "%d", (int)value);
1.10 paf 1197:
1.154 paf 1198: return pa_strdup(local_buf, size);
1.12 paf 1199: }
1200:
1.36 paf 1201: size_t stdout_write(const void *buf, size_t size) {
1.12 paf 1202: #ifdef WIN32
1.187 paf 1203: size_t to_write = size;
1.12 paf 1204: do{
1.154 paf 1205: int chunk_written=fwrite(buf, 1, min((size_t)8*0x400, size), stdout);
1.12 paf 1206: if(chunk_written<=0)
1207: break;
1208: size-=chunk_written;
1.36 paf 1209: buf=((const char*)buf)+chunk_written;
1.126 paf 1210: } while(size>0);
1.12 paf 1211:
1.187 paf 1212: return to_write-size;
1.12 paf 1213: #else
1.126 paf 1214: return fwrite(buf, 1, size, stdout);
1.12 paf 1215: #endif
1.2 paf 1216: }
1.14 paf 1217:
1.154 paf 1218: char* unescape_chars(const char* cp, int len) {
1219: char* s=new(PointerFreeGC) char[len + 1];
1.14 paf 1220: enum EscapeState {
1.33 paf 1221: EscapeRest,
1222: EscapeFirst,
1.14 paf 1223: EscapeSecond
1224: } escapeState=EscapeRest;
1.193 paf 1225: uchar escapedValue=0;
1.14 paf 1226: int srcPos=0;
1227: int dstPos=0;
1228: while(srcPos < len) {
1.193 paf 1229: uchar ch=(uchar)cp[srcPos];
1.14 paf 1230: switch(escapeState) {
1231: case EscapeRest:
1232: if(ch=='%') {
1233: escapeState=EscapeFirst;
1234: } else if(ch=='+') {
1.126 paf 1235: s[dstPos++]=' ';
1.14 paf 1236: } else {
1237: s[dstPos++]=ch;
1238: }
1239: break;
1240: case EscapeFirst:
1.193 paf 1241: escapedValue=(uchar)(hex_value[ch] << 4);
1.14 paf 1242: escapeState=EscapeSecond;
1243: break;
1244: case EscapeSecond:
1.126 paf 1245: escapedValue +=hex_value[ch];
1.14 paf 1246: s[dstPos++]=escapedValue;
1247: escapeState=EscapeRest;
1248: break;
1249: }
1.126 paf 1250: srcPos++;
1.14 paf 1251: }
1252: s[dstPos]=0;
1253: return s;
1.24 paf 1254: }
1255:
1256: #ifdef WIN32
1.126 paf 1257: void back_slashes_to_slashes(char* s) {
1.24 paf 1258: if(s)
1259: for(; *s; s++)
1260: if(*s=='\\')
1.126 paf 1261: *s='/';
1.24 paf 1262: }
1.42 paf 1263: /*
1.126 paf 1264: void slashes_to_back_slashes(char* s) {
1.42 paf 1265: if(s)
1266: for(; *s; s++)
1267: if(*s=='/')
1.126 paf 1268: *s='\\';
1.42 paf 1269: }
1270: */
1.24 paf 1271: #endif
1.41 paf 1272:
1.126 paf 1273: bool StrEqNc(const char* s1, const char* s2, bool strict) {
1.41 paf 1274: while(true) {
1275: if(!(*s1)) {
1276: if(!(*s2))
1277: return true;
1278: else
1279: return !strict;
1280: } else if(!(*s2))
1281: return !strict;
1.189 paf 1282: if(isalpha((unsigned char)*s1)) {
1.190 paf 1283: if(tolower((unsigned char)*s1) !=tolower((unsigned char)*s2))
1.41 paf 1284: return false;
1285: } else if((*s1) !=(*s2))
1286: return false;
1.126 paf 1287: s1++;
1288: s2++;
1.41 paf 1289: }
1.57 parser 1290: }
1291:
1.84 paf 1292: static bool isLeap(int year) {
1.57 parser 1293: return !(
1294: (year % 4) || ((year % 400) && !(year % 100))
1.126 paf 1295: );
1.57 parser 1296: }
1297:
1298: int getMonthDays(int year, int month) {
1299: int monthDays[]={
1.126 paf 1300: 31,
1301: isLeap(year) ? 29 : 28,
1302: 31,
1303: 30,
1304: 31,
1305: 30,
1306: 31,
1307: 31,
1308: 30,
1309: 31,
1310: 30,
1.57 parser 1311: 31
1.126 paf 1312: };
1313: return monthDays[month];
1.41 paf 1314: }
1.69 parser 1315:
1.126 paf 1316: void remove_crlf(char* start, char* end) {
1317: for(char* p=start; p<end; p++)
1.69 parser 1318: switch(*p) {
1.126 paf 1319: case '\n': *p='|'; break;
1320: case '\r': *p=' '; break;
1.69 parser 1321: }
1.91 paf 1322: }
1323:
1324:
1325: /// must be last in this file
1326: #undef vsnprintf
1.126 paf 1327: int __vsnprintf(char* b, size_t s, const char* f, va_list l) {
1.91 paf 1328: if(!s)
1329: return 0;
1330:
1331: int r;
1332: // note: on win32& maybe somewhere else
1333: // vsnprintf do not writes terminating 0 in 'buffer full' case, reducing
1334: --s;
1.172 paf 1335:
1336: // clients do not check for negative 's', feature: ignore such prints
1337: if((ssize_t)s<0)
1338: return 0;
1339:
1.91 paf 1340: #if _MSC_VER
1341: /*
1342: win32:
1343: mk:@MSITStore:C:\Program%20Files\Microsoft%20Visual%20Studio\MSDN\2001APR\1033\vccore.chm::/html/_crt__vsnprintf.2c_._vsnwprintf.htm
1344:
1.154 paf 1345: if the number of bytes to write exceeds buffer, then count bytes are written and Ö1 is returned
1.91 paf 1346: */
1.126 paf 1347: r=_vsnprintf(b, s, f, l);
1.91 paf 1348: if(r<0)
1349: r=s;
1350: #else
1.126 paf 1351: r=vsnprintf(b, s, f, l);
1.91 paf 1352: /*
1353: solaris:
1354: man vsnprintf
1355:
1356: The snprintf() function returns the number of characters
1357: formatted, that is, the number of characters that would have
1358: been written to the buffer if it were large enough. If the
1359: value of n is 0 on a call to snprintf(), an unspecified
1360: value less than 1 is returned.
1361: */
1362:
1363: if(r<0)
1364: r=0;
1.167 paf 1365: else if((size_t)r>s)
1.91 paf 1366: r=s;
1367: #endif
1368: b[r]=0;
1369: return r;
1370: }
1371:
1.126 paf 1372: int __snprintf(char* b, size_t s, const char* f, ...) {
1.91 paf 1373: va_list l;
1.126 paf 1374: va_start(l, f);
1375: int r=__vsnprintf(b, s, f, l);
1376: va_end(l);
1.91 paf 1377: return r;
1.178 paf 1378: }
1379:
1380: /* mime64 functions are from libgmime[http://spruce.sourceforge.net/gmime/] lib */
1381: /*
1382: * Authors: Michael Zucchi <notzed@helixcode.com>
1383: * Jeffrey Stedfast <fejj@helixcode.com>
1384: *
1385: * Copyright 2000 Helix Code, Inc. (www.helixcode.com)
1386: *
1387: * This program is free software; you can redistribute it and/or modify
1388: * it under the terms of the GNU General Public License as published by
1389: * the Free Software Foundation; either version 2 of the License, or
1390: * (at your option) any later version.
1391: *
1392: * This program is distributed in the hope that it will be useful,
1393: * but WITHOUT ANY WARRANTY; without even the implied warranty of
1394: * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
1395: * GNU General Public License for more details.
1396: *
1397: * You should have received a copy of the GNU General Public License
1398: * along with this program; if not, write to the Free Software
1399: * Foundation, Inc., 59 Temple Street #330, Boston, MA 02111-1307, USA.
1400: *
1401: */
1402: static char *base64_alphabet =
1403: "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";
1404:
1405: /**
1406: * g_mime_utils_base64_encode_step:
1407: * @in: input stream
1408: * @inlen: length of the input
1409: * @out: output string
1410: * @state: holds the number of bits that are stored in @save
1411: * @save: leftover bits that have not yet been encoded
1412: *
1413: * Base64 encodes a chunk of data. Performs an 'encode step', only
1414: * encodes blocks of 3 characters to the output at a time, saves
1415: * left-over state in state and save (initialise to 0 on first
1416: * invocation).
1417: *
1418: * Returns the number of bytes encoded.
1419: **/
1420: static size_t
1421: g_mime_utils_base64_encode_step (const unsigned char *in, size_t inlen, unsigned char *out, int *state, int *save)
1422: {
1.186 paf 1423: register const unsigned char *inptr;
1.178 paf 1424: register unsigned char *outptr;
1425:
1426: if (inlen <= 0)
1427: return 0;
1428:
1429: inptr = in;
1430: outptr = out;
1431:
1432: if (inlen + ((unsigned char *)save)[0] > 2) {
1433: const unsigned char *inend = in + inlen - 2;
1434: register int c1 = 0, c2 = 0, c3 = 0;
1435: register int already;
1436:
1437: already = *state;
1438:
1439: switch (((char *)save)[0]) {
1440: case 1: c1 = ((unsigned char *)save)[1]; goto skip1;
1441: case 2: c1 = ((unsigned char *)save)[1];
1442: c2 = ((unsigned char *)save)[2]; goto skip2;
1443: }
1444:
1445: /* yes, we jump into the loop, no i'm not going to change it, its beautiful! */
1446: while (inptr < inend) {
1447: c1 = *inptr++;
1448: skip1:
1449: c2 = *inptr++;
1450: skip2:
1451: c3 = *inptr++;
1452: *outptr++ = base64_alphabet [c1 >> 2];
1453: *outptr++ = base64_alphabet [(c2 >> 4) | ((c1 & 0x3) << 4)];
1454: *outptr++ = base64_alphabet [((c2 & 0x0f) << 2) | (c3 >> 6)];
1455: *outptr++ = base64_alphabet [c3 & 0x3f];
1456: /* this is a bit ugly ... */
1457: if ((++already) >= 19) {
1458: *outptr++ = '\n';
1459: already = 0;
1460: }
1461: }
1462:
1463: ((unsigned char *)save)[0] = 0;
1464: inlen = 2 - (inptr - inend);
1465: *state = already;
1466: }
1467:
1468: //d(printf ("state = %d, inlen = %d\n", (int)((char *)save)[0], inlen));
1469:
1470: if (inlen > 0) {
1471: register char *saveout;
1472:
1473: /* points to the slot for the next char to save */
1474: saveout = & (((char *)save)[1]) + ((char *)save)[0];
1475:
1476: /* inlen can only be 0 1 or 2 */
1477: switch (inlen) {
1478: case 2: *saveout++ = *inptr++;
1479: case 1: *saveout++ = *inptr++;
1480: }
1481: ((char *)save)[0] += inlen;
1482: }
1483:
1484: /*d(printf ("mode = %d\nc1 = %c\nc2 = %c\n",
1485: (int)((char *)save)[0],
1486: (int)((char *)save)[1],
1487: (int)((char *)save)[2]));*/
1488:
1489: return (outptr - out);
1490: }
1491:
1492: /**
1493: * g_mime_utils_base64_encode_close:
1494: * @in: input stream
1495: * @inlen: length of the input
1496: * @out: output string
1497: * @state: holds the number of bits that are stored in @save
1498: * @save: leftover bits that have not yet been encoded
1499: *
1500: * Base64 encodes the input stream to the output stream. Call this
1501: * when finished encoding data with g_mime_utils_base64_encode_step to
1502: * flush off the last little bit.
1503: *
1504: * Returns the number of bytes encoded.
1505: **/
1506: static size_t
1507: g_mime_utils_base64_encode_close (const unsigned char *in, size_t inlen, unsigned char *out, int *state, int *save)
1508: {
1509: unsigned char *outptr = out;
1510: int c1, c2;
1511:
1512: if (inlen > 0)
1513: outptr += g_mime_utils_base64_encode_step (in, inlen, outptr, state, save);
1514:
1515: c1 = ((unsigned char *)save)[1];
1516: c2 = ((unsigned char *)save)[2];
1517:
1518: switch (((unsigned char *)save)[0]) {
1519: case 2:
1520: outptr[2] = base64_alphabet [(c2 & 0x0f) << 2];
1521: goto skip;
1522: case 1:
1523: outptr[2] = '=';
1524: skip:
1525: outptr[0] = base64_alphabet [c1 >> 2];
1526: outptr[1] = base64_alphabet [c2 >> 4 | ((c1 & 0x3) << 4)];
1527: outptr[3] = '=';
1528: outptr += 4;
1529: break;
1530: }
1531:
1532: *outptr++ = 0;
1533:
1534: *save = 0;
1535: *state = 0;
1536:
1537: return (outptr - out);
1538: }
1539:
1540: char* pa_base64(const char *in, size_t len)
1541: {
1542: /* wont go to more than 2x size (overly conservative) */
1543: char* result=new(PointerFreeGC) char[len * 2 + 6];
1544: int state=0;
1545: int save=0;
1.183 paf 1546: #ifndef NDEBUG
1547: size_t filled=
1548: #endif
1549: g_mime_utils_base64_encode_close ((const unsigned char*)in, len,
1.178 paf 1550: (unsigned char*)result, &state, &save);
1551: assert(filled <= len * 2 + 6);
1552:
1553: return result;
1.98 paf 1554: }
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