Annotation of parser3/src/main/pa_common.C, revision 1.209
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:
1.209 ! paf 8: static const char * const IDENT_COMMON_C="$Date: 2005/11/18 09:20:24 $";
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:
1.208 paf 219: // [gcc is happier this way, see goto above]
220: {
221: // allocating initial buf
222: response=(char*)pa_malloc_atomic(response_size+1/*terminator*/); // just setting memory block type
223: char* ptr=response;
224: size_t todo_size=response_size;
225: // coping part of already received body
226: memcpy(ptr, preview_buf, received_size);
1.207 paf 227: ptr+=received_size;
1.208 paf 228: todo_size-=received_size;
1.171 paf 229:
1.208 paf 230: // we use terminator byte for two purposes here:
231: // 1. we return there zero always, not knowing: maybe they would want to create String form $file.body?
232: // invariant: all Strings should have zero-terminated buffers
233: // 2. we use that out-of-size byte to detect if our content-length guess was wrong
234: // when recv gets more than we expected
235: // a) we know that the content-length guess was wrong
236: // b) we have space to put the first byte of extra data
237: // c) we use less code to detect normal situation: on last while-cycle recv expected to just return 0
238: while(true) {
239: received_size=recv(sock, ptr, todo_size+1/*there is always a place for terminator*/, 0);
240: if(received_size==0) {
241: response_size-=todo_size; // in case we received less than expected, cut down the reported size
242: break;
243: }
244: if(received_size<0) {
245: if(int no=pa_socks_errno())
246: throw Exception("http.timeout",
247: 0,
248: "error receiving response body: %s (%d)", pa_socks_strerr(no), no);
249: break;
250: }
251: // they've touched the terminator?
252: if((size_t)received_size>todo_size)
253: {
254: // that means that our guessed response_size was not big enough
255: const size_t grow_chunk_size=0x400*0x400; // 1M
256: response_size+=grow_chunk_size;
257: size_t ptr_offset=ptr-response;
258: response=(char*)pa_realloc(response, response_size+1/*terminator*/);
259: ptr=response+ptr_offset;
260: todo_size+=grow_chunk_size;
261: }
262: // can't do this before realloc: we need <todo_size check
263: ptr+=received_size;
264: todo_size-=received_size;
265:
266: char* EOLat=0;
267: if(!result && (EOLat=strstr(response, "\n"))) { // checking status in first response
268: const String status_line(pa_strdup(response, EOLat-response));
269: ArrayString astatus;
270: size_t pos_after=0;
271: status_line.split(astatus, pos_after, " ");
272: const String& status_code=*astatus.get(astatus.count()>1?1:0);
273: result=status_code.as_int();
274:
275: if(fail_on_status_ne_200 && result!=200)
276: throw Exception("http.status",
277: &status_code,
278: "invalid HTTP response status");
279: }
1.142 paf 280: }
281: }
1.207 paf 282: done:
1.154 paf 283: if(result)
1.207 paf 284: {
285: response[response_size]=0;
1.154 paf 286: return result;
1.207 paf 287: }
1.142 paf 288: else
289: throw Exception("http.response",
290: 0,
1.173 paf 291: "bad response from host - no status found (size=%u)", response_size);
1.126 paf 292: }
293:
294: /* ********************** request *************************** */
295:
296: #if defined(SIGALRM) && defined(HAVE_SIGSETJMP) && defined(HAVE_SIGLONGJMP)
1.145 paf 297: # define PA_USE_ALARM
1.126 paf 298: #endif
299:
1.145 paf 300: #ifdef PA_USE_ALARM
1.126 paf 301: static sigjmp_buf timeout_env;
1.199 paf 302: static void timeout_handler(int /*sig*/){
1.126 paf 303: siglongjmp(timeout_env, 1);
304: }
305: #endif
306:
1.171 paf 307: static int http_request(char*& response, size_t& response_size,
1.193 paf 308: const char* host, short port,
1.152 paf 309: const char* request,
1.196 paf 310: int timeout_secs,
1.152 paf 311: bool fail_on_status_ne_200) {
1.126 paf 312: if(!host)
313: throw Exception("http.host",
1.154 paf 314: 0,
1.126 paf 315: "zero hostname"); //never
316:
1.200 paf 317: volatile // to prevent makeing it register variable, because it will be clobbered by longjmp [thanks gcc warning]
318: int sock=-1;
1.145 paf 319: #ifdef PA_USE_ALARM
1.146 paf 320: signal(SIGALRM, timeout_handler);
1.126 paf 321: #endif
1.145 paf 322: #ifdef PA_USE_ALARM
323: if(sigsetjmp(timeout_env, 1)) {
324: // stupid gcc [2.95.4] generated bad code
325: // which failed to handle sigsetjmp+throw: crashed inside of pre-throw code.
1.199 paf 326: // rewritten simplier [athough duplicating closesocket code]
1.145 paf 327: if(sock>=0)
328: closesocket(sock);
329: throw Exception("http.timeout",
1.199 paf 330: 0,
1.145 paf 331: "timeout occured while retrieving document");
1.146 paf 332: return 0; // never
1.145 paf 333: } else {
1.196 paf 334: alarm(timeout_secs);
1.145 paf 335: #endif
1.146 paf 336: try {
337: int result;
1.126 paf 338: struct sockaddr_in dest;
1.154 paf 339:
340: if(!set_addr(&dest, host, port))
1.126 paf 341: throw Exception("http.host",
1.154 paf 342: 0,
1.127 paf 343: "can not resolve hostname \"%s\"", host);
1.126 paf 344:
1.195 paf 345: if((sock=socket(AF_INET, SOCK_STREAM, IPPROTO_TCP/*0*/))<0) {
346: int no=pa_socks_errno();
1.126 paf 347: throw Exception("http.connect",
1.154 paf 348: 0,
1.195 paf 349: "can not make socket: %s (%d)", pa_socks_strerr(no), no);
350: }
1.196 paf 351:
352: // To enable SO_DONTLINGER (that is, disable SO_LINGER)
353: // l_onoff should be set to zero and setsockopt should be called
354: linger dont_linger={0,0};
355: setsockopt(sock, SOL_SOCKET, SO_LINGER, (const char *)&dont_linger, sizeof(dont_linger));
356:
1.201 paf 357: #ifdef WIN32
358: // SO_*TIMEO can be defined in .h but not implemlemented in protocol,
359: // failing subsequently with Option not supported by protocol (99) message
360: // could not suppress that, so leaving this only for win32
1.196 paf 361: int timeout_ms=timeout_secs*1000;
362: setsockopt(sock, SOL_SOCKET, SO_SNDTIMEO, (const char*)&timeout_ms, sizeof(timeout_ms));
363: setsockopt(sock, SOL_SOCKET, SO_RCVTIMEO, (const char*)&timeout_ms, sizeof(timeout_ms));
364: #endif
365:
1.195 paf 366: if(connect(sock, (struct sockaddr *)&dest, sizeof(dest))) {
367: int no=pa_socks_errno();
1.126 paf 368: throw Exception("http.connect",
1.154 paf 369: 0,
1.195 paf 370: "can not connect to host \"%s\": %s (%d)", host, pa_socks_strerr(no), no);
371: }
1.126 paf 372: size_t request_size=strlen(request);
1.195 paf 373: if(send(sock, request, request_size, 0)!=(ssize_t)request_size) {
374: int no=pa_socks_errno();
1.197 paf 375: throw Exception("http.timeout",
1.154 paf 376: 0,
1.195 paf 377: "error sending request: %s (%d)", pa_socks_strerr(no), no);
378: }
1.126 paf 379:
1.171 paf 380: result=http_read_response(response, response_size, sock, fail_on_status_ne_200);
1.142 paf 381: closesocket(sock);
1.145 paf 382: #ifdef PA_USE_ALARM
1.142 paf 383: alarm(0);
1.126 paf 384: #endif
1.147 paf 385: return result;
1.146 paf 386: } catch(...) {
1.145 paf 387: #ifdef PA_USE_ALARM
1.146 paf 388: alarm(0);
1.126 paf 389: #endif
1.146 paf 390: if(sock>=0)
391: closesocket(sock);
1.154 paf 392: rethrow;
1.146 paf 393: }
1.148 paf 394: #ifdef PA_USE_ALARM
1.126 paf 395: }
1.148 paf 396: #endif
1.126 paf 397: }
398:
1.127 paf 399: #ifndef DOXYGEN
400: struct Http_pass_header_info {
1.154 paf 401: Request_charsets* charsets;
1.127 paf 402: String* request;
403: bool user_agent_specified;
404: };
405: #endif
1.154 paf 406: static void http_pass_header(HashStringValue::key_type key,
407: HashStringValue::value_type value,
408: Http_pass_header_info *info) {
409: *info->request <<key<<": "
410: << attributed_meaning_to_string(*value, String::L_HTTP_HEADER, false)
411: << CRLF;
1.135 paf 412:
1.154 paf 413: if(String(key, String::L_TAINTED).change_case(info->charsets->source(), String::CC_UPPER)=="USER-AGENT")
414: info->user_agent_specified=true;
1.126 paf 415: }
1.154 paf 416:
417:
418: static Charset* detect_charset(Charset& source_charset, const String& content_type_value) {
1.156 paf 419: const String::Body CONTENT_TYPE_VALUE=
1.154 paf 420: content_type_value.change_case(source_charset, String::CC_UPPER);
421: // content-type: xxx/xxx; source_charset=WE-NEED-THIS
422: // content-type: xxx/xxx; source_charset="WE-NEED-THIS"
423: // content-type: xxx/xxx; source_charset="WE-NEED-THIS";
424: size_t before_charseteq_pos=CONTENT_TYPE_VALUE.pos("CHARSET=");
425: if(before_charseteq_pos!=STRING_NOT_FOUND) {
426: size_t charset_begin=before_charseteq_pos+8/*CHARSET="*/;
427: size_t open_quote_pos=CONTENT_TYPE_VALUE.pos('"', charset_begin);
428: bool quoted=open_quote_pos==charset_begin;
429: if(quoted)
430: charset_begin++; // skip opening '"'
431: size_t charset_end=CONTENT_TYPE_VALUE.length();
432: if(quoted) {
433: size_t close_quote_pos=CONTENT_TYPE_VALUE.pos('"', charset_begin);
434: if(close_quote_pos!=STRING_NOT_FOUND)
435: charset_end=close_quote_pos;
436: } else {
437: size_t delim_pos=CONTENT_TYPE_VALUE.pos(';', charset_begin);
438: if(delim_pos!=STRING_NOT_FOUND)
439: charset_end=delim_pos;
440: }
1.156 paf 441: const String::Body CHARSET_NAME_BODY=
1.154 paf 442: CONTENT_TYPE_VALUE.mid(charset_begin, charset_end);
443:
444: return &charsets.get(CHARSET_NAME_BODY);
445: }
446:
447: return 0;
448: }
449:
1.178 paf 450: static const String* basic_authorization_field(const char* user, const char* pass) {
451: if(!user&& !pass)
452: return 0;
453:
454: String combined;
455: if(user)
456: combined<<user;
457: combined<<":";
458: if(pass)
459: combined<<pass;
460:
461: String* result=new String("Basic "); *result<<pa_base64(combined.cstr(), combined.length());
462: return result;
463: }
1.154 paf 464:
1.181 paf 465: static void form_string_value2string(
466: HashStringValue::key_type key,
467: const String& value,
468: String& result)
469: {
470: result << String(key, String::L_TAINTED) << "=";
471: result.append(value, String::L_URI, true);
472: result<< "&";
473: }
474: #ifndef DOXYGEN
475: struct Form_table_value2string_info {
476: HashStringValue::key_type key;
477: String& result;
478:
479: Form_table_value2string_info(HashStringValue::key_type akey, String& aresult):
480: key(akey), result(aresult) {}
481: };
482: #endif
483: static void form_table_value2string(Table::element_type row, Form_table_value2string_info* info) {
484: form_string_value2string(info->key, *row->get(0), info->result);
485: }
1.180 paf 486: static void form_value2string(
487: HashStringValue::key_type key,
488: HashStringValue::value_type value,
489: String* result)
490: {
1.181 paf 491: if(const String* svalue=value->get_string())
492: form_string_value2string(key, *svalue, *result);
493: else if(Table* tvalue=value->get_table()) {
494: Form_table_value2string_info info(key, *result);
495: tvalue->for_each(form_table_value2string, &info);
496: } else
497: throw Exception(0,
498: new String(key, String::L_TAINTED),
499: "is %s, "HTTP_FORM_NAME" option value must either string or table", value->type());
1.180 paf 500: }
501: static const char* form2string(HashStringValue& form) {
502: String string;
503: form.for_each(form_value2string, &string);
1.181 paf 504: return string.cstr(String::L_UNSPECIFIED);
1.180 paf 505: }
1.154 paf 506: #ifndef DOXYGEN
507: struct File_read_http_result {
508: char *str; size_t length;
509: HashStringValue* headers;
510: };
511: #endif
1.194 paf 512: static void find_headers_end(char* p,
513: char*& headers_end_at,
514: char*& raw_body)
515: {
516: raw_body=p;
517: // \n\n
518: // \r\n\r\n
519: while((p=strchr(p, '\n'))) {
520: headers_end_at=++p; // \n>.<
521: if(*p=='\r') // \r\n>\r?<\n
522: p++;
523: if(*p=='\n') { // \r\n\r>\n?<
524: raw_body=p+1;
525: return;
526: }
527: }
528: headers_end_at=0;
529: }
530:
1.155 paf 531: /// @todo build .cookies field. use ^file.tables.SET-COOKIES.menu{ for now
1.154 paf 532: static File_read_http_result file_read_http(Request_charsets& charsets,
533: const String& file_spec,
1.169 paf 534: bool as_text,
535: HashStringValue *options=0) {
1.154 paf 536: File_read_http_result result;
1.126 paf 537: char host[MAX_STRING];
1.129 paf 538: const char* uri;
1.193 paf 539: short port;
1.180 paf 540: const char* method="GET"; bool method_is_get;
541: HashStringValue* form=0;
542: const char* body_cstr=0;
1.196 paf 543: int timeout_secs=2;
1.142 paf 544: bool fail_on_status_ne_200=true;
1.154 paf 545: Value* vheaders=0;
546: Charset *asked_remote_charset=0;
1.178 paf 547: const char* user_cstr=0;
548: const char* password_cstr=0;
1.126 paf 549:
1.127 paf 550: if(options) {
1.206 paf 551: int valid_options=get_valid_file_options_count(*options);
552:
1.177 paf 553: if(Value* vmethod=options->get(HTTP_METHOD_NAME)) {
1.127 paf 554: valid_options++;
1.154 paf 555: method=vmethod->as_string().cstr();
1.181 paf 556: }
557: if(Value* vform=options->get(HTTP_FORM_NAME)) {
1.180 paf 558: valid_options++;
559: form=vform->get_hash();
1.181 paf 560: }
561: if(Value* vbody=options->get(HTTP_BODY_NAME)) {
1.180 paf 562: valid_options++;
563: body_cstr=vbody->as_string().cstr(String::L_UNSPECIFIED);
1.181 paf 564: }
565: if(Value* vtimeout=options->get(HTTP_TIMEOUT_NAME)) {
1.127 paf 566: valid_options++;
1.196 paf 567: timeout_secs=vtimeout->as_int();
1.181 paf 568: }
569: if((vheaders=options->get(HTTP_HEADERS_NAME))) {
1.127 paf 570: valid_options++;
1.181 paf 571: }
572: if(Value* vany_status=options->get(HTTP_ANY_STATUS_NAME)) {
1.142 paf 573: valid_options++;
574: fail_on_status_ne_200=!vany_status->as_bool();
1.181 paf 575: }
576: if(Value* vcharset_name=options->get(HTTP_CHARSET_NAME)) {
1.154 paf 577: valid_options++;
578: asked_remote_charset=&::charsets.get(vcharset_name->as_string().
579: change_case(charsets.source(), String::CC_UPPER));
1.181 paf 580: }
581: if(Value* vuser=options->get(HTTP_USER)) {
1.178 paf 582: valid_options++;
583: user_cstr=vuser->as_string().cstr();
1.181 paf 584: }
585: if(Value* vpassword=options->get(HTTP_PASSWORD)) {
1.178 paf 586: valid_options++;
587: password_cstr=vpassword->as_string().cstr();
588: }
1.142 paf 589:
1.154 paf 590: if(valid_options!=options->count())
1.127 paf 591: throw Exception("parser.runtime",
592: 0,
593: "invalid option passed");
1.154 paf 594: }
595: if(!asked_remote_charset) // defaulting to $request:charset
596: asked_remote_charset=&charsets.source();
597:
1.180 paf 598: method_is_get=strcmp(method, "GET")==0;
599: if(method_is_get && body_cstr)
600: throw Exception("parser.runtime",
601: 0,
602: "you can not use $."HTTP_BODY_NAME" option with method GET");
603:
1.154 paf 604: //preparing request
605: String& connect_string=*new String;
606: // not in ^sql{... L_SQL ...} spirit, but closer to ^file::load one
607: connect_string.append(file_spec, String::L_URI); // tainted pieces -> URI pieces
608:
1.180 paf 609: String request_head_and_body;
1.154 paf 610: {
611: // influence URLencoding of tainted pieces to String::L_URI lang
612: Temp_client_charset temp(charsets, *asked_remote_charset);
613:
614: const char* connect_string_cstr=connect_string.cstr(String::L_UNSPECIFIED);
1.126 paf 615:
1.154 paf 616: const char* current=connect_string_cstr;
617: if(strncmp(current, "http://", 7)!=0)
618: throw Exception(0,
619: &connect_string,
620: "does not start with http://"); //never
621: current+=7;
622:
623: strncpy(host, current, sizeof(host)-1); host[sizeof(host)-1]=0;
624: char* host_uri=lsplit(host, '/');
625: uri=host_uri?current+(host_uri-1-host):"/";
626: char* port_cstr=lsplit(host, ':');
627: char* error_pos=0;
1.193 paf 628: port=port_cstr?(short)strtol(port_cstr, &error_pos, 0):80;
1.154 paf 629:
1.180 paf 630: if(strchr(uri, '?') && form)
631: throw Exception("parser.runtime",
632: 0,
633: "use either uri with ?params or $."HTTP_FORM_NAME" option");
634:
635: //making request head
636: String head;
637: head << method;
638: head << " " << uri;
639: if(form)
640: if(method_is_get)
1.181 paf 641: head << "?" << form2string(*form);
1.180 paf 642: head <<" HTTP/1.0" CRLF
1.154 paf 643: "host: "<< host << CRLF;
1.181 paf 644: if(form && !method_is_get) {
645: head << "content-type: application/x-www-form-urlencoded" CRLF;
646: body_cstr = form2string(*form);
647: }
1.178 paf 648:
1.179 paf 649: // http://www.ietf.org/rfc/rfc2617.txt
1.178 paf 650: if(const String* authorization_field_value=basic_authorization_field(user_cstr, password_cstr))
1.180 paf 651: head<<"authorization: "<<*authorization_field_value<<CRLF;
1.178 paf 652:
1.154 paf 653: bool user_agent_specified=false;
654: if(vheaders && !vheaders->is_string()) { // allow empty
655: if(HashStringValue *headers=vheaders->get_hash()) {
1.180 paf 656: Http_pass_header_info info={&charsets, &head, false};
1.154 paf 657: headers->for_each(http_pass_header, &info);
658: user_agent_specified=info.user_agent_specified;
659: } else
660: throw Exception("parser.runtime",
661: &connect_string,
662: "headers param must be hash");
663: };
664: if(!user_agent_specified) // defaulting
1.180 paf 665: head << "user-agent: " DEFAULT_USER_AGENT CRLF;
666:
667: if(body_cstr) {
668: // recode those pieces which are not in String::L_URI lang
669: // [those violating HTTP standard, but widly used]
670: body_cstr=Charset::transcode(
671: String::C(body_cstr, strlen(body_cstr)),
672: charsets.source(),
673: *asked_remote_charset);
1.181 paf 674:
675: head << "content-length: " << format(strlen(body_cstr), "%u") << CRLF;
1.180 paf 676: }
1.154 paf 677:
1.181 paf 678: const char* head_cstr=head.cstr(String::L_UNSPECIFIED);
679:
680: // recode those pieces which are not in String::L_URI lang
681: // [those violating HTTP standard, but widly used]
682: head_cstr=Charset::transcode(
683: String::C(head_cstr, strlen(head_cstr)),
684: charsets.source(),
685: *asked_remote_charset);
1.180 paf 686:
1.181 paf 687: // head + end of header
688: request_head_and_body << head_cstr << CRLF;
1.180 paf 689: // body
690: if(body_cstr)
691: request_head_and_body << body_cstr;
1.154 paf 692: }
1.126 paf 693:
694: //sending request
1.171 paf 695: char* response;
696: size_t response_size;
697: int status_code=http_request(response, response_size,
1.180 paf 698: host, port, request_head_and_body.cstr(),
1.196 paf 699: timeout_secs, fail_on_status_ne_200);
1.126 paf 700:
701: //processing results
1.171 paf 702: char* raw_body; size_t raw_body_size;
1.194 paf 703: char* headers_end_at;
704: find_headers_end(response,
705: headers_end_at,
706: raw_body);
1.191 paf 707: raw_body_size=response_size-(raw_body-response);
1.171 paf 708:
1.154 paf 709: result.headers=new HashStringValue;
1.155 paf 710: VHash* vtables=new VHash;
1.177 paf 711: result.headers->put(HTTP_TABLES_NAME, vtables);
1.194 paf 712: Charset* real_remote_charset=0; // undetected, yet
713:
714: if(headers_end_at) {
715: *headers_end_at=0;
716: const String header_block(String::C(response, headers_end_at-response), true);
717:
718: ArrayString aheaders;
719: HashStringValue& tables=vtables->hash();
1.155 paf 720:
1.194 paf 721: size_t pos_after=0;
722: header_block.split(aheaders, pos_after, "\n");
723:
724: //processing headers
725: size_t aheaders_count=aheaders.count();
726: for(size_t i=1; i<aheaders_count; i++) {
727: const String& line=*aheaders.get(i);
728: size_t pos=line.pos(':');
729: if(pos==STRING_NOT_FOUND || pos<1)
730: throw Exception("http.response",
731: &connect_string,
732: "bad response from host - bad header \"%s\"", line.cstr());
733: const String::Body HEADER_NAME=
734: line.mid(0, pos).change_case(charsets.source(), String::CC_UPPER);
735: const String& header_value=line.mid(pos+1, line.length()).trim(String::TRIM_BOTH, " \t\r");
736: if(as_text && HEADER_NAME=="CONTENT-TYPE")
737: real_remote_charset=detect_charset(charsets.source(), header_value);
738:
739: // tables
740: {
741: Value *valready=(Value *)tables.get(HEADER_NAME);
742: bool existed=valready!=0;
743: Table *table;
744: if(existed) {
745: // second+ appearence
746: table=valready->get_table();
747: } else {
748: // first appearence
749: Table::columns_type columns =new ArrayString(1);
750: *columns+=new String("value");
751: table=new Table(columns);
752: }
753: // this string becomes next row
754: ArrayString& row=*new ArrayString(1);
755: row+=&header_value;
756: *table+=&row;
757: // not existed before? add it
758: if(!existed)
759: tables.put(HEADER_NAME, new VTable(table));
1.155 paf 760: }
1.194 paf 761:
762: result.headers->put(HEADER_NAME, new VString(header_value));
1.155 paf 763: }
1.126 paf 764: }
765:
766: // output response
1.171 paf 767: String::C real_body=String::C(raw_body, raw_body_size);
1.192 paf 768: if(as_text && raw_body_size) { // must be checked because transcode returns CONST string in case length==0, which contradicts hacking few lines below
769: // defaulting to used-asked charset [it's never empty!]
770: if(!real_remote_charset)
771: real_remote_charset=asked_remote_charset;
1.169 paf 772: real_body=Charset::transcode(real_body, *real_remote_charset, charsets.source());
1.192 paf 773: }
1.154 paf 774:
775: result.str=const_cast<char *>(real_body.str); // hacking a little
776: result.length=real_body.length;
777: result.headers->put(file_status_name, new VInt(status_code));
778: return result;
1.34 paf 779: }
1.123 paf 780:
1.154 paf 781: #endif
782:
1.123 paf 783: #ifndef DOXYGEN
784: struct File_read_action_info {
1.154 paf 785: char **data; size_t *data_size;
1.188 paf 786: char* buf; size_t offset; size_t count;
1.126 paf 787: };
1.123 paf 788: #endif
1.154 paf 789: static void file_read_action(
790: struct stat& finfo,
791: int f,
1.166 paf 792: const String& file_spec, const char* /*fname*/, bool as_text,
1.154 paf 793: void *context) {
1.126 paf 794: File_read_action_info& info=*static_cast<File_read_action_info *>(context);
1.188 paf 795: size_t to_read_size=info.count;
796: if(!to_read_size)
797: to_read_size=(size_t)finfo.st_size;
798: assert( !(info.buf && as_text) );
799: if(to_read_size) {
800: if(info.offset)
801: lseek(f, info.offset, SEEK_SET);
802: *info.data=info.buf
803: ? info.buf
804: : new(PointerFreeGC) char[to_read_size+(as_text?1:0)];
1.126 paf 805: *info.data_size=(size_t)read(f, *info.data, to_read_size);
1.123 paf 806:
807: if(ssize_t(*info.data_size)<0 || *info.data_size>to_read_size)
1.126 paf 808: throw Exception(0,
1.123 paf 809: &file_spec,
1.173 paf 810: "read failed: actually read %u bytes count not in [0..%u] valid range",
1.126 paf 811: *info.data_size, to_read_size);
1.123 paf 812: } else { // empty file
1.209 ! paf 813: // for both, text and binary: for text we need that terminator, for binary we need nonzero pointer to be able to save such files
! 814: *info.data=new(PointerFreeGC) char[1];
! 815: *(char*)(*info.data)=0;
1.123 paf 816: *info.data_size=0;
817: return;
818: }
1.126 paf 819: }
1.154 paf 820: File_read_result file_read(Request_charsets& charsets, const String& file_spec,
821: bool as_text, HashStringValue *params,
1.188 paf 822: bool fail_on_read_problem,
823: char* buf, size_t offset, size_t count) {
1.167 paf 824: File_read_result result={false, 0, 0, 0};
1.154 paf 825: #ifdef PA_HTTP
826: if(file_spec.starts_with("http://")) {
1.203 paf 827: if(offset || count)
828: throw Exception("parser.runtime",
829: 0,
830: "offset and load options are not supported for HTTP:// file load");
831:
1.126 paf 832: // fail on read problem
1.169 paf 833: File_read_http_result http=file_read_http(charsets, file_spec, as_text, params);
1.154 paf 834: result.success=true;
835: result.str=http.str;
836: result.length=http.length;
837: result.headers=http.headers;
1.126 paf 838: } else {
1.154 paf 839: #endif
1.206 paf 840: if(params) {
841: int valid_options=get_valid_file_options_count(*params);
842: if(valid_options!=params->count())
843: throw Exception("parser.runtime",
844: 0,
845: "invalid option passed");
846: }
1.161 paf 847:
1.188 paf 848: File_read_action_info info={&result.str, &result.length,
849: buf, offset, count};
1.154 paf 850: result.success=file_read_action_under_lock(file_spec,
1.126 paf 851: "read", file_read_action, &info,
852: as_text, fail_on_read_problem);
1.154 paf 853: #ifdef PA_HTTP
1.126 paf 854: }
1.154 paf 855: #endif
1.123 paf 856:
1.154 paf 857: if(result.success && as_text) {
1.131 paf 858: // UTF-8 signature: EF BB BF
1.154 paf 859: if(result.length>=3) {
860: char *in=(char *)result.str;
1.159 paf 861: if(strncmp(in, "\xEF\xBB\xBF", 3)==0) {
1.154 paf 862: result.str=in+3; result.length-=3;// skip prefix
1.131 paf 863: }
864: }
865:
1.154 paf 866: fix_line_breaks((char *)(result.str), result.length);
1.123 paf 867: }
1.126 paf 868:
869: return result;
1.123 paf 870: }
871:
1.154 paf 872: #ifdef PA_SAFE_MODE
873: void check_safe_mode(struct stat finfo, const String& file_spec, const char* fname) {
874: if(finfo.st_uid/*foreign?*/!=geteuid()
875: && finfo.st_gid/*foreign?*/!=getegid())
876: throw Exception("parser.runtime",
877: &file_spec,
878: "parser is in safe mode: "
879: "reading files of foreign group and user disabled "
880: "[recompile parser with --disable-safe-mode configure option], "
881: "actual filename '%s', "
882: "fuid(%d)!=euid(%d) or fgid(%d)!=egid(%d)",
883: fname,
884: finfo.st_uid, geteuid(),
885: finfo.st_gid, getegid());
886: }
887: #endif
1.149 paf 888:
1.154 paf 889: bool file_read_action_under_lock(const String& file_spec,
1.126 paf 890: const char* action_name, File_read_action action, void *context,
891: bool as_text,
1.123 paf 892: bool fail_on_read_problem) {
1.154 paf 893: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.33 paf 894: int f;
895:
896: // first open, next stat:
1.45 paf 897: // directory update of NTFS hard links performed on open.
1.33 paf 898: // ex:
899: // a.html:^test[] and b.html hardlink to a.html
900: // user inserts ! before ^test in a.html
1.126 paf 901: // directory entry of b.html in NTFS not updated at once,
1.35 paf 902: // they delay update till open, so we would receive "!^test[" string
903: // if would do stat, next open.
1.123 paf 904: // later: it seems, even this does not help sometimes
1.98 paf 905: if((f=open(fname, O_RDONLY|(as_text?_O_TEXT:_O_BINARY)))>=0) {
1.123 paf 906: try {
1.162 paf 907: if(pa_lock_shared_blocking(f)!=0)
1.126 paf 908: throw Exception("file.lock",
1.123 paf 909: &file_spec,
910: "shared lock failed: %s (%d), actual filename '%s'",
1.154 paf 911: strerror(errno), errno, fname);
1.123 paf 912:
1.124 paf 913: struct stat finfo;
914: if(stat(fname, &finfo)!=0)
915: throw Exception("file.missing", // hardly possible: we just opened it OK
916: &file_spec,
917: "stat failed: %s (%d), actual filename '%s'",
1.154 paf 918: strerror(errno), errno, fname);
1.124 paf 919:
1.140 paf 920: #ifdef PA_SAFE_MODE
1.149 paf 921: check_safe_mode(finfo, file_spec, fname);
1.105 paf 922: #endif
1.32 paf 923:
1.154 paf 924: action(finfo, f, file_spec, fname, as_text, context);
1.123 paf 925: } catch(...) {
1.162 paf 926: pa_unlock(f);close(f);
1.123 paf 927: if(fail_on_read_problem)
1.154 paf 928: rethrow;
1.123 paf 929: return false;
930: }
1.87 paf 931:
1.162 paf 932: pa_unlock(f);close(f);
1.72 parser 933: return true;
1.118 paf 934: } else {
935: if(fail_on_read_problem)
1.126 paf 936: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.118 paf 937: &file_spec,
1.123 paf 938: "%s failed: %s (%d), actual filename '%s'",
1.154 paf 939: action_name, strerror(errno), errno, fname);
1.118 paf 940: return false;
941: }
1.8 paf 942: }
943:
1.202 paf 944: void create_dir_for_file(const String& file_spec) {
1.63 parser 945: size_t pos_after=1;
1.154 paf 946: size_t pos_before;
947: while((pos_before=file_spec.pos('/', pos_after))!=STRING_NOT_FOUND) {
948: mkdir(file_spec.mid(0, pos_before).cstr(String::L_FILE_SPEC), 0775);
1.63 parser 949: pos_after=pos_before+1;
950: }
951: }
952:
1.98 paf 953: bool file_write_action_under_lock(
1.28 paf 954: const String& file_spec,
1.126 paf 955: const char* action_name, File_write_action action, void *context,
956: bool as_text,
957: bool do_append,
958: bool do_block,
1.110 paf 959: bool fail_on_lock_problem) {
1.154 paf 960: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.28 paf 961: int f;
1.80 paf 962: if(access(fname, W_OK)!=0) // no
1.126 paf 963: create_dir_for_file(file_spec);
1.50 paf 964:
1.80 paf 965: if((f=open(fname,
966: O_CREAT|O_RDWR
967: |(as_text?_O_TEXT:_O_BINARY)
1.138 paf 968: |(do_append?O_APPEND:PA_O_TRUNC), 0664))>=0) {
1.162 paf 969: if((do_block?pa_lock_exclusive_blocking(f):pa_lock_exclusive_nonblocking(f))!=0) {
1.126 paf 970: Exception e("file.lock",
1.110 paf 971: &file_spec,
972: "shared lock failed: %s (%d), actual filename '%s'",
1.154 paf 973: strerror(errno), errno, fname);
1.126 paf 974: close(f);
1.110 paf 975: if(fail_on_lock_problem)
976: throw e;
1.98 paf 977: return false;
978: }
1.96 paf 979:
1.158 paf 980: try {
1.126 paf 981: action(f, context);
1.158 paf 982: } catch(...) {
1.138 paf 983: #ifdef HAVE_FTRUNCATE
1.104 paf 984: if(!do_append)
1.125 paf 985: ftruncate(f, lseek(f, 0, SEEK_CUR)); // one can not use O_TRUNC, read lower
1.138 paf 986: #endif
1.162 paf 987: pa_unlock(f);close(f);
1.154 paf 988: rethrow;
1.158 paf 989: }
1.80 paf 990:
1.138 paf 991: #ifdef HAVE_FTRUNCATE
1.104 paf 992: if(!do_append)
1.125 paf 993: 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 994: #endif
1.162 paf 995: pa_unlock(f);close(f);
1.98 paf 996: return true;
1.80 paf 997: } else
1.126 paf 998: throw Exception(errno==EACCES?"file.access":0,
1.80 paf 999: &file_spec,
1.96 paf 1000: "%s failed: %s (%d), actual filename '%s'",
1.154 paf 1001: action_name, strerror(errno), errno, fname);
1.96 paf 1002: // here should be nothing, see rethrow above
1003: }
1004:
1005: #ifndef DOXYGEN
1006: struct File_write_action_info {
1.154 paf 1007: const char* str; size_t length;
1.126 paf 1008: };
1.96 paf 1009: #endif
1010: static void file_write_action(int f, void *context) {
1.126 paf 1011: File_write_action_info& info=*static_cast<File_write_action_info *>(context);
1.154 paf 1012: if(info.length) {
1013: int written=write(f, info.str, info.length);
1.116 paf 1014: if(written<0)
1.126 paf 1015: throw Exception(0,
1016: 0,
1017: "write failed: %s (%d)", strerror(errno), errno);
1.113 paf 1018: }
1.96 paf 1019: }
1020: void file_write(
1021: const String& file_spec,
1.154 paf 1022: const char* data, size_t size,
1.126 paf 1023: bool as_text,
1.96 paf 1024: bool do_append) {
1.126 paf 1025: File_write_action_info info={data, size};
1.98 paf 1026: file_write_action_under_lock(
1.154 paf 1027: file_spec,
1028: "write", file_write_action, &info,
1029: as_text,
1030: do_append);
1.30 paf 1031: }
1032:
1.63 parser 1033: // throws nothing! [this is required in file_move & file_delete]
1.50 paf 1034: static void rmdir(const String& file_spec, size_t pos_after) {
1.154 paf 1035: size_t pos_before;
1036: if((pos_before=file_spec.pos('/', pos_after))!=STRING_NOT_FOUND)
1.126 paf 1037: rmdir(file_spec, pos_before+1);
1.50 paf 1038:
1.154 paf 1039: rmdir(file_spec.mid(0, pos_after-1/* / */).cstr(String::L_FILE_SPEC));
1.50 paf 1040: }
1.164 paf 1041: bool file_delete(const String& file_spec, bool fail_on_problem) {
1.154 paf 1042: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.54 parser 1043: if(unlink(fname)!=0)
1.164 paf 1044: if(fail_on_problem)
1.126 paf 1045: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.93 paf 1046: &file_spec,
1047: "unlink failed: %s (%d), actual filename '%s'",
1.154 paf 1048: strerror(errno), errno, fname);
1.93 paf 1049: else
1050: return false;
1.50 paf 1051:
1.126 paf 1052: rmdir(file_spec, 1);
1.93 paf 1053: return true;
1.60 parser 1054: }
1.95 paf 1055: void file_move(const String& old_spec, const String& new_spec) {
1.154 paf 1056: const char* old_spec_cstr=old_spec.cstr(String::L_FILE_SPEC);
1057: const char* new_spec_cstr=new_spec.cstr(String::L_FILE_SPEC);
1.63 parser 1058:
1.126 paf 1059: create_dir_for_file(new_spec);
1.63 parser 1060:
1.60 parser 1061: if(rename(old_spec_cstr, new_spec_cstr)!=0)
1.126 paf 1062: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.60 parser 1063: &old_spec,
1064: "rename failed: %s (%d), actual filename '%s' to '%s'",
1.154 paf 1065: strerror(errno), errno, old_spec_cstr, new_spec_cstr);
1.63 parser 1066:
1.126 paf 1067: rmdir(old_spec, 1);
1.31 paf 1068: }
1069:
1.51 paf 1070:
1.126 paf 1071: bool entry_exists(const char* fname, struct stat *afinfo) {
1.118 paf 1072: struct stat lfinfo;
1073: bool result=stat(fname, &lfinfo)==0;
1074: if(afinfo)
1075: *afinfo=lfinfo;
1076: return result;
1.119 paf 1077: }
1078:
1079: bool entry_exists(const String& file_spec) {
1.154 paf 1080: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.126 paf 1081: return entry_exists(fname, 0);
1.118 paf 1082: }
1083:
1.51 paf 1084: static bool entry_readable(const String& file_spec, bool need_dir) {
1.154 paf 1085: char* fname=file_spec.cstrm(String::L_FILE_SPEC);
1.120 paf 1086: if(need_dir) {
1.126 paf 1087: size_t size=strlen(fname);
1.120 paf 1088: while(size) {
1.126 paf 1089: char c=fname[size-1];
1.120 paf 1090: if(c=='/' || c=='\\')
1091: fname[--size]=0;
1092: else
1093: break;
1094: }
1095: }
1.51 paf 1096: struct stat finfo;
1.118 paf 1097: if(access(fname, R_OK)==0 && entry_exists(fname, &finfo)) {
1.109 paf 1098: bool is_dir=(finfo.st_mode&S_IFDIR) != 0;
1.51 paf 1099: return is_dir==need_dir;
1100: }
1101: return false;
1102: }
1.31 paf 1103: bool file_readable(const String& file_spec) {
1.126 paf 1104: return entry_readable(file_spec, false);
1.51 paf 1105: }
1106: bool dir_readable(const String& file_spec) {
1.126 paf 1107: return entry_readable(file_spec, true);
1.65 parser 1108: }
1.154 paf 1109: const String* file_readable(const String& path, const String& name) {
1110: String& result=*new String(path);
1111: result << "/";
1112: result << name;
1113: return file_readable(result)?&result:0;
1.43 paf 1114: }
1115: bool file_executable(const String& file_spec) {
1.154 paf 1116: return access(file_spec.cstr(String::L_FILE_SPEC), X_OK)==0;
1.44 paf 1117: }
1118:
1.64 parser 1119: bool file_stat(const String& file_spec,
1.58 parser 1120: size_t& rsize,
1.126 paf 1121: time_t& ratime,
1122: time_t& rmtime,
1123: time_t& rctime,
1.64 parser 1124: bool fail_on_read_problem) {
1.154 paf 1125: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1126: struct stat finfo;
1.44 paf 1127: if(stat(fname, &finfo)!=0)
1.64 parser 1128: if(fail_on_read_problem)
1.126 paf 1129: throw Exception("file.missing",
1.67 parser 1130: &file_spec,
1131: "getting file size failed: %s (%d), real filename '%s'",
1.154 paf 1132: strerror(errno), errno, fname);
1.64 parser 1133: else
1134: return false;
1.58 parser 1135: rsize=finfo.st_size;
1136: ratime=finfo.st_atime;
1137: rmtime=finfo.st_mtime;
1138: rctime=finfo.st_ctime;
1.64 parser 1139: return true;
1.18 paf 1140: }
1141:
1.126 paf 1142: char* getrow(char* *row_ref, char delim) {
1143: char* result=*row_ref;
1.8 paf 1144: if(result) {
1.126 paf 1145: *row_ref=strchr(result, delim);
1.8 paf 1146: if(*row_ref)
1147: *((*row_ref)++)=0;
1148: else if(!*result)
1149: return 0;
1150: }
1151: return result;
1152: }
1153:
1.126 paf 1154: char* lsplit(char* string, char delim) {
1.23 paf 1155: if(string) {
1.126 paf 1156: char* v=strchr(string, delim);
1.8 paf 1157: if(v) {
1158: *v=0;
1159: return v+1;
1160: }
1161: }
1162: return 0;
1163: }
1164:
1.126 paf 1165: char* lsplit(char* *string_ref, char delim) {
1166: char* result=*string_ref;
1167: char* next=lsplit(*string_ref, delim);
1.8 paf 1168: *string_ref=next;
1169: return result;
1.9 paf 1170: }
1171:
1.126 paf 1172: char* rsplit(char* string, char delim) {
1.18 paf 1173: if(string) {
1.126 paf 1174: char* v=strrchr(string, delim);
1.18 paf 1175: if(v) {
1.9 paf 1176: *v=0;
1177: return v+1;
1178: }
1179: }
1180: return NULL;
1.10 paf 1181: }
1182:
1.37 paf 1183: /// @todo less stupid type detection
1.154 paf 1184: const char* format(double value, char* fmt) {
1.126 paf 1185: char local_buf[MAX_NUMBER];
1.108 paf 1186: size_t size;
1187:
1.10 paf 1188: if(fmt)
1189: if(strpbrk(fmt, "diouxX"))
1190: if(strpbrk(fmt, "ouxX"))
1.126 paf 1191: size=snprintf(local_buf, sizeof(local_buf), fmt, (uint)value);
1.10 paf 1192: else
1.126 paf 1193: size=snprintf(local_buf, sizeof(local_buf), fmt, (int)value);
1.10 paf 1194: else
1.126 paf 1195: size=snprintf(local_buf, sizeof(local_buf), fmt, value);
1.10 paf 1196: else
1.126 paf 1197: size=snprintf(local_buf, sizeof(local_buf), "%d", (int)value);
1.10 paf 1198:
1.154 paf 1199: return pa_strdup(local_buf, size);
1.12 paf 1200: }
1201:
1.36 paf 1202: size_t stdout_write(const void *buf, size_t size) {
1.12 paf 1203: #ifdef WIN32
1.187 paf 1204: size_t to_write = size;
1.12 paf 1205: do{
1.154 paf 1206: int chunk_written=fwrite(buf, 1, min((size_t)8*0x400, size), stdout);
1.12 paf 1207: if(chunk_written<=0)
1208: break;
1209: size-=chunk_written;
1.36 paf 1210: buf=((const char*)buf)+chunk_written;
1.126 paf 1211: } while(size>0);
1.12 paf 1212:
1.187 paf 1213: return to_write-size;
1.12 paf 1214: #else
1.126 paf 1215: return fwrite(buf, 1, size, stdout);
1.12 paf 1216: #endif
1.2 paf 1217: }
1.14 paf 1218:
1.154 paf 1219: char* unescape_chars(const char* cp, int len) {
1220: char* s=new(PointerFreeGC) char[len + 1];
1.14 paf 1221: enum EscapeState {
1.33 paf 1222: EscapeRest,
1223: EscapeFirst,
1.14 paf 1224: EscapeSecond
1225: } escapeState=EscapeRest;
1.193 paf 1226: uchar escapedValue=0;
1.14 paf 1227: int srcPos=0;
1228: int dstPos=0;
1229: while(srcPos < len) {
1.193 paf 1230: uchar ch=(uchar)cp[srcPos];
1.14 paf 1231: switch(escapeState) {
1232: case EscapeRest:
1233: if(ch=='%') {
1234: escapeState=EscapeFirst;
1235: } else if(ch=='+') {
1.126 paf 1236: s[dstPos++]=' ';
1.14 paf 1237: } else {
1238: s[dstPos++]=ch;
1239: }
1240: break;
1241: case EscapeFirst:
1.193 paf 1242: escapedValue=(uchar)(hex_value[ch] << 4);
1.14 paf 1243: escapeState=EscapeSecond;
1244: break;
1245: case EscapeSecond:
1.126 paf 1246: escapedValue +=hex_value[ch];
1.14 paf 1247: s[dstPos++]=escapedValue;
1248: escapeState=EscapeRest;
1249: break;
1250: }
1.126 paf 1251: srcPos++;
1.14 paf 1252: }
1253: s[dstPos]=0;
1254: return s;
1.24 paf 1255: }
1256:
1257: #ifdef WIN32
1.126 paf 1258: void back_slashes_to_slashes(char* s) {
1.24 paf 1259: if(s)
1260: for(; *s; s++)
1261: if(*s=='\\')
1.126 paf 1262: *s='/';
1.24 paf 1263: }
1.42 paf 1264: /*
1.126 paf 1265: void slashes_to_back_slashes(char* s) {
1.42 paf 1266: if(s)
1267: for(; *s; s++)
1268: if(*s=='/')
1.126 paf 1269: *s='\\';
1.42 paf 1270: }
1271: */
1.24 paf 1272: #endif
1.41 paf 1273:
1.126 paf 1274: bool StrEqNc(const char* s1, const char* s2, bool strict) {
1.41 paf 1275: while(true) {
1276: if(!(*s1)) {
1277: if(!(*s2))
1278: return true;
1279: else
1280: return !strict;
1281: } else if(!(*s2))
1282: return !strict;
1.189 paf 1283: if(isalpha((unsigned char)*s1)) {
1.190 paf 1284: if(tolower((unsigned char)*s1) !=tolower((unsigned char)*s2))
1.41 paf 1285: return false;
1286: } else if((*s1) !=(*s2))
1287: return false;
1.126 paf 1288: s1++;
1289: s2++;
1.41 paf 1290: }
1.57 parser 1291: }
1292:
1.84 paf 1293: static bool isLeap(int year) {
1.57 parser 1294: return !(
1295: (year % 4) || ((year % 400) && !(year % 100))
1.126 paf 1296: );
1.57 parser 1297: }
1298:
1299: int getMonthDays(int year, int month) {
1300: int monthDays[]={
1.126 paf 1301: 31,
1302: isLeap(year) ? 29 : 28,
1303: 31,
1304: 30,
1305: 31,
1306: 30,
1307: 31,
1308: 31,
1309: 30,
1310: 31,
1311: 30,
1.57 parser 1312: 31
1.126 paf 1313: };
1314: return monthDays[month];
1.41 paf 1315: }
1.69 parser 1316:
1.126 paf 1317: void remove_crlf(char* start, char* end) {
1318: for(char* p=start; p<end; p++)
1.69 parser 1319: switch(*p) {
1.126 paf 1320: case '\n': *p='|'; break;
1321: case '\r': *p=' '; break;
1.69 parser 1322: }
1.91 paf 1323: }
1324:
1325:
1326: /// must be last in this file
1327: #undef vsnprintf
1.126 paf 1328: int __vsnprintf(char* b, size_t s, const char* f, va_list l) {
1.91 paf 1329: if(!s)
1330: return 0;
1331:
1332: int r;
1333: // note: on win32& maybe somewhere else
1334: // vsnprintf do not writes terminating 0 in 'buffer full' case, reducing
1335: --s;
1.172 paf 1336:
1337: // clients do not check for negative 's', feature: ignore such prints
1338: if((ssize_t)s<0)
1339: return 0;
1340:
1.91 paf 1341: #if _MSC_VER
1342: /*
1343: win32:
1344: mk:@MSITStore:C:\Program%20Files\Microsoft%20Visual%20Studio\MSDN\2001APR\1033\vccore.chm::/html/_crt__vsnprintf.2c_._vsnwprintf.htm
1345:
1.154 paf 1346: if the number of bytes to write exceeds buffer, then count bytes are written and Ö1 is returned
1.91 paf 1347: */
1.126 paf 1348: r=_vsnprintf(b, s, f, l);
1.91 paf 1349: if(r<0)
1350: r=s;
1351: #else
1.126 paf 1352: r=vsnprintf(b, s, f, l);
1.91 paf 1353: /*
1354: solaris:
1355: man vsnprintf
1356:
1357: The snprintf() function returns the number of characters
1358: formatted, that is, the number of characters that would have
1359: been written to the buffer if it were large enough. If the
1360: value of n is 0 on a call to snprintf(), an unspecified
1361: value less than 1 is returned.
1362: */
1363:
1364: if(r<0)
1365: r=0;
1.167 paf 1366: else if((size_t)r>s)
1.91 paf 1367: r=s;
1368: #endif
1369: b[r]=0;
1370: return r;
1371: }
1372:
1.126 paf 1373: int __snprintf(char* b, size_t s, const char* f, ...) {
1.91 paf 1374: va_list l;
1.126 paf 1375: va_start(l, f);
1376: int r=__vsnprintf(b, s, f, l);
1377: va_end(l);
1.91 paf 1378: return r;
1.178 paf 1379: }
1380:
1381: /* mime64 functions are from libgmime[http://spruce.sourceforge.net/gmime/] lib */
1382: /*
1383: * Authors: Michael Zucchi <notzed@helixcode.com>
1384: * Jeffrey Stedfast <fejj@helixcode.com>
1385: *
1386: * Copyright 2000 Helix Code, Inc. (www.helixcode.com)
1387: *
1388: * This program is free software; you can redistribute it and/or modify
1389: * it under the terms of the GNU General Public License as published by
1390: * the Free Software Foundation; either version 2 of the License, or
1391: * (at your option) any later version.
1392: *
1393: * This program is distributed in the hope that it will be useful,
1394: * but WITHOUT ANY WARRANTY; without even the implied warranty of
1395: * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
1396: * GNU General Public License for more details.
1397: *
1398: * You should have received a copy of the GNU General Public License
1399: * along with this program; if not, write to the Free Software
1400: * Foundation, Inc., 59 Temple Street #330, Boston, MA 02111-1307, USA.
1401: *
1402: */
1403: static char *base64_alphabet =
1404: "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";
1405:
1406: /**
1407: * g_mime_utils_base64_encode_step:
1408: * @in: input stream
1409: * @inlen: length of the input
1410: * @out: output string
1411: * @state: holds the number of bits that are stored in @save
1412: * @save: leftover bits that have not yet been encoded
1413: *
1414: * Base64 encodes a chunk of data. Performs an 'encode step', only
1415: * encodes blocks of 3 characters to the output at a time, saves
1416: * left-over state in state and save (initialise to 0 on first
1417: * invocation).
1418: *
1419: * Returns the number of bytes encoded.
1420: **/
1421: static size_t
1422: g_mime_utils_base64_encode_step (const unsigned char *in, size_t inlen, unsigned char *out, int *state, int *save)
1423: {
1.186 paf 1424: register const unsigned char *inptr;
1.178 paf 1425: register unsigned char *outptr;
1426:
1427: if (inlen <= 0)
1428: return 0;
1429:
1430: inptr = in;
1431: outptr = out;
1432:
1433: if (inlen + ((unsigned char *)save)[0] > 2) {
1434: const unsigned char *inend = in + inlen - 2;
1435: register int c1 = 0, c2 = 0, c3 = 0;
1436: register int already;
1437:
1438: already = *state;
1439:
1440: switch (((char *)save)[0]) {
1441: case 1: c1 = ((unsigned char *)save)[1]; goto skip1;
1442: case 2: c1 = ((unsigned char *)save)[1];
1443: c2 = ((unsigned char *)save)[2]; goto skip2;
1444: }
1445:
1446: /* yes, we jump into the loop, no i'm not going to change it, its beautiful! */
1447: while (inptr < inend) {
1448: c1 = *inptr++;
1449: skip1:
1450: c2 = *inptr++;
1451: skip2:
1452: c3 = *inptr++;
1453: *outptr++ = base64_alphabet [c1 >> 2];
1454: *outptr++ = base64_alphabet [(c2 >> 4) | ((c1 & 0x3) << 4)];
1455: *outptr++ = base64_alphabet [((c2 & 0x0f) << 2) | (c3 >> 6)];
1456: *outptr++ = base64_alphabet [c3 & 0x3f];
1457: /* this is a bit ugly ... */
1458: if ((++already) >= 19) {
1459: *outptr++ = '\n';
1460: already = 0;
1461: }
1462: }
1463:
1464: ((unsigned char *)save)[0] = 0;
1465: inlen = 2 - (inptr - inend);
1466: *state = already;
1467: }
1468:
1469: //d(printf ("state = %d, inlen = %d\n", (int)((char *)save)[0], inlen));
1470:
1471: if (inlen > 0) {
1472: register char *saveout;
1473:
1474: /* points to the slot for the next char to save */
1475: saveout = & (((char *)save)[1]) + ((char *)save)[0];
1476:
1477: /* inlen can only be 0 1 or 2 */
1478: switch (inlen) {
1479: case 2: *saveout++ = *inptr++;
1480: case 1: *saveout++ = *inptr++;
1481: }
1482: ((char *)save)[0] += inlen;
1483: }
1484:
1485: /*d(printf ("mode = %d\nc1 = %c\nc2 = %c\n",
1486: (int)((char *)save)[0],
1487: (int)((char *)save)[1],
1488: (int)((char *)save)[2]));*/
1489:
1490: return (outptr - out);
1491: }
1492:
1493: /**
1494: * g_mime_utils_base64_encode_close:
1495: * @in: input stream
1496: * @inlen: length of the input
1497: * @out: output string
1498: * @state: holds the number of bits that are stored in @save
1499: * @save: leftover bits that have not yet been encoded
1500: *
1501: * Base64 encodes the input stream to the output stream. Call this
1502: * when finished encoding data with g_mime_utils_base64_encode_step to
1503: * flush off the last little bit.
1504: *
1505: * Returns the number of bytes encoded.
1506: **/
1507: static size_t
1508: g_mime_utils_base64_encode_close (const unsigned char *in, size_t inlen, unsigned char *out, int *state, int *save)
1509: {
1510: unsigned char *outptr = out;
1511: int c1, c2;
1512:
1513: if (inlen > 0)
1514: outptr += g_mime_utils_base64_encode_step (in, inlen, outptr, state, save);
1515:
1516: c1 = ((unsigned char *)save)[1];
1517: c2 = ((unsigned char *)save)[2];
1518:
1519: switch (((unsigned char *)save)[0]) {
1520: case 2:
1521: outptr[2] = base64_alphabet [(c2 & 0x0f) << 2];
1522: goto skip;
1523: case 1:
1524: outptr[2] = '=';
1525: skip:
1526: outptr[0] = base64_alphabet [c1 >> 2];
1527: outptr[1] = base64_alphabet [c2 >> 4 | ((c1 & 0x3) << 4)];
1528: outptr[3] = '=';
1529: outptr += 4;
1530: break;
1531: }
1532:
1533: *outptr++ = 0;
1534:
1535: *save = 0;
1536: *state = 0;
1537:
1538: return (outptr - out);
1539: }
1540:
1541: char* pa_base64(const char *in, size_t len)
1542: {
1543: /* wont go to more than 2x size (overly conservative) */
1544: char* result=new(PointerFreeGC) char[len * 2 + 6];
1545: int state=0;
1546: int save=0;
1.183 paf 1547: #ifndef NDEBUG
1548: size_t filled=
1549: #endif
1550: g_mime_utils_base64_encode_close ((const unsigned char*)in, len,
1.178 paf 1551: (unsigned char*)result, &state, &save);
1552: assert(filled <= len * 2 + 6);
1553:
1554: return result;
1.98 paf 1555: }
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