Annotation of parser3/src/main/pa_common.C, revision 1.177
1.15 paf 1: /** @file
1.16 paf 2: Parser: commonly functions.
3:
1.174 paf 4: Copyright(c) 2001-2004 ArtLebedev Group (http://www.artlebedev.com)
1.101 paf 5: Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru)
1.111 paf 6: */
1.16 paf 7:
1.177 ! paf 8: static const char * const IDENT_COMMON_C="$Date: 2004/03/01 12:16:00 $";
1.1 paf 9:
10: #include "pa_common.h"
1.4 paf 11: #include "pa_exception.h"
1.154 paf 12: #include "pa_hash.h"
1.14 paf 13: #include "pa_globals.h"
1.154 paf 14: #include "pa_request_charsets.h"
15: #include "pa_charsets.h"
16:
17: #define PA_HTTP
18:
19: #ifdef PA_HTTP
1.126 paf 20: #include "pa_vstring.h"
1.154 paf 21: #include "pa_vint.h"
1.155 paf 22: #include "pa_vhash.h"
23: #include "pa_vtable.h"
1.154 paf 24:
25: #ifdef CYGWIN
26: #define _GNU_H_WINDOWS32_SOCKETS
27: // for PASCAL
28: #include <windows.h>
29: // SOCKET
30: typedef u_int SOCKET;
31: int PASCAL closesocket(SOCKET);
32: #else
33: # if defined(WIN32)
34: # include <windows.h>
35: # else
36: # define closesocket close
37: # endif
38: #endif
1.1 paf 39:
1.126 paf 40: #else
1.154 paf 41:
42: # if defined(WIN32)
43: # include <windows.h>
44: # endif
45:
1.98 paf 46: #endif
47:
1.93 paf 48: // some maybe-undefined constants
49:
1.82 paf 50: #ifndef _O_TEXT
51: # define _O_TEXT 0
52: #endif
53: #ifndef _O_BINARY
54: # define _O_BINARY 0
1.47 paf 55: #endif
1.80 paf 56:
1.138 paf 57: #ifdef HAVE_FTRUNCATE
58: # define PA_O_TRUNC 0
59: #else
60: # ifdef _O_TRUNC
61: # define PA_O_TRUNC _O_TRUNC
62: # else
63: # error you must have either ftruncate function or _O_TRUNC bit declared
64: # endif
1.154 paf 65: #endif
1.176 paf 66:
67: # ifndef INADDR_NONE
68: # define INADDR_NONE ((ulong) -1)
69: # endif
1.154 paf 70:
71: // defines for globals
72:
73: #define FILE_STATUS_NAME "status"
74:
75: // globals
76:
77: const String file_status_name(FILE_STATUS_NAME);
78:
79: // defines for statics
80:
81: #define HTTP_METHOD_NAME "method"
82: #define HTTP_TIMEOUT_NAME "timeout"
83: #define HTTP_HEADERS_NAME "headers"
84: #define HTTP_ANY_STATUS_NAME "any-status"
85: #define HTTP_CHARSET_NAME "charset"
1.155 paf 86: #define HTTP_TABLES_NAME "tables"
1.154 paf 87:
88: // statics
89:
90: // defines
91:
1.127 paf 92: #define DEFAULT_USER_AGENT "parser3"
93:
1.154 paf 94: // functions
1.127 paf 95:
1.154 paf 96: void fix_line_breaks(char *str, size_t& length) {
1.87 paf 97: //_asm int 3;
1.154 paf 98: const char* const eob=str+length;
99: char* dest=str;
1.72 parser 100: // fix DOS: \r\n -> \n
101: // fix Macintosh: \r -> \n
1.154 paf 102: char* bol=str;
1.137 paf 103: while(char* eol=(char*)memchr(bol, '\r', eob -bol)) {
1.72 parser 104: size_t len=eol-bol;
105: if(dest!=bol)
1.126 paf 106: memcpy(dest, bol, len);
1.72 parser 107: dest+=len;
1.126 paf 108: *dest++='\n';
1.72 parser 109:
1.126 paf 110: if(&eol[1]<eob && eol[1]=='\n') { // \r, \n = DOS
1.72 parser 111: bol=eol+2;
1.154 paf 112: length--;
1.126 paf 113: } else // \r, not \n = Macintosh
1.72 parser 114: bol=eol+1;
115: }
1.154 paf 116: // last piece without \r
1.72 parser 117: if(dest!=bol)
1.126 paf 118: memcpy(dest, bol, eob-bol);
1.154 paf 119: str[length]=0; // terminating
1.72 parser 120: }
1.18 paf 121:
1.154 paf 122: char* file_read_text(Request_charsets& charsets,
123: const String& file_spec,
124: bool fail_on_read_problem,
125: HashStringValue* params/*, HashStringValue* * out_fields*/) {
126: File_read_result file=
127: file_read(charsets, file_spec, true, params, fail_on_read_problem);
128: return file.success?file.str:0;
1.126 paf 129: }
130:
1.154 paf 131: #ifdef PA_HTTP
1.126 paf 132: //http request stuff
1.154 paf 133:
134: #undef CRLF
135: #define CRLF "\r\n"
1.126 paf 136:
137: static bool set_addr(struct sockaddr_in *addr, const char* host, const short port){
138: memset(addr, 0, sizeof(*addr));
139: addr->sin_family=AF_INET;
140: addr->sin_port=htons(port);
141: if(host) {
1.175 paf 142: ulong packed_ip=inet_addr(host);
143: struct hostent *hostIP = packed_ip==INADDR_NONE?
144: gethostbyname(host)
145: : gethostbyaddr((char *)&packed_ip, sizeof(packed_ip), AF_INET);
146: if(hostIP)
1.126 paf 147: memcpy(&addr->sin_addr, hostIP->h_addr, hostIP->h_length);
148: else
149: return false;
150: } else
151: addr->sin_addr.s_addr=INADDR_ANY;
152: return true;
153: }
154:
1.171 paf 155: static int http_read_response(char*& response, size_t& response_size, int sock, bool fail_on_status_ne_200){
156: response=(char*)pa_malloc_atomic(1/*terminator*/); // setting memory block type
157: response[(response_size=0)]=0;
1.154 paf 158: int result=0;
1.171 paf 159: char* EOLat=0;
1.126 paf 160: while(true) {
1.171 paf 161: char buf[MAX_STRING*10];
162: ssize_t received_size=recv(sock, buf, sizeof(buf), 0);
163: if(received_size<=0)
1.126 paf 164: break;
1.171 paf 165: response=(char*)pa_realloc(response, response_size+received_size+1/*terminator*/);
166: memcpy(response+response_size, buf, received_size);
167: response_size+=received_size;
168: response[response_size]=0;
169:
170: if(!result && (EOLat=strstr(response, CRLF))) { // checking status in first response
171: const String status_line(pa_strdup(response, EOLat-response));
1.154 paf 172: ArrayString astatus;
173: size_t pos_after=0;
174: status_line.split(astatus, pos_after, " ");
175: const String& status_code=*astatus.get(1);
176: result=status_code.as_int();
1.142 paf 177:
1.154 paf 178: if(fail_on_status_ne_200 && result!=200)
1.142 paf 179: throw Exception("http.status",
1.154 paf 180: &status_code,
1.142 paf 181: "invalid HTTP response status");
182: }
183: }
1.154 paf 184: if(result)
185: return result;
1.142 paf 186: else
187: throw Exception("http.response",
188: 0,
1.173 paf 189: "bad response from host - no status found (size=%u)", response_size);
1.126 paf 190: }
191:
192: /* ********************** request *************************** */
193:
194: #if defined(SIGALRM) && defined(HAVE_SIGSETJMP) && defined(HAVE_SIGLONGJMP)
1.145 paf 195: # define PA_USE_ALARM
1.126 paf 196: #endif
197:
1.145 paf 198: #ifdef PA_USE_ALARM
1.126 paf 199: static sigjmp_buf timeout_env;
200: static void timeout_handler(int sig){
201: siglongjmp(timeout_env, 1);
202: }
203: #endif
204:
1.171 paf 205: static int http_request(char*& response, size_t& response_size,
1.152 paf 206: const char* host, int port,
207: const char* request,
1.166 paf 208: int
209: #ifdef PA_USE_ALARM
210: timeout
211: #endif
212: ,
1.152 paf 213: bool fail_on_status_ne_200) {
1.126 paf 214: if(!host)
215: throw Exception("http.host",
1.154 paf 216: 0,
1.126 paf 217: "zero hostname"); //never
218:
1.145 paf 219: #ifdef PA_USE_ALARM
1.146 paf 220: signal(SIGALRM, timeout_handler);
1.126 paf 221: #endif
222: int sock=-1;
1.145 paf 223: #ifdef PA_USE_ALARM
224: if(sigsetjmp(timeout_env, 1)) {
225: // stupid gcc [2.95.4] generated bad code
226: // which failed to handle sigsetjmp+throw: crashed inside of pre-throw code.
227: // rewritten simplier [though duplicating closesocket code]
228: if(sock>=0)
229: closesocket(sock);
230: throw Exception("http.timeout",
231: origin_string,
232: "timeout occured while retrieving document");
1.146 paf 233: return 0; // never
1.145 paf 234: } else {
235: alarm(timeout);
236: #endif
1.146 paf 237: try {
238: int result;
1.126 paf 239: struct sockaddr_in dest;
1.154 paf 240:
241: if(!set_addr(&dest, host, port))
1.126 paf 242: throw Exception("http.host",
1.154 paf 243: 0,
1.127 paf 244: "can not resolve hostname \"%s\"", host);
1.126 paf 245:
246: if((sock=socket(AF_INET, SOCK_STREAM, IPPROTO_TCP/*0*/))<0)
247: throw Exception("http.connect",
1.154 paf 248: 0,
1.127 paf 249: "can not make socket: %s (%d)", strerror(errno), errno);
1.126 paf 250: if(connect(sock, (struct sockaddr *)&dest, sizeof(dest)))
251: throw Exception("http.connect",
1.154 paf 252: 0,
1.127 paf 253: "can not connect to host \"%s\": %s (%d)", host, strerror(errno), errno);
1.126 paf 254: size_t request_size=strlen(request);
255: if(send(sock, request, request_size, 0)!=(ssize_t)request_size)
256: throw Exception("http.connect",
1.154 paf 257: 0,
1.127 paf 258: "error sending request: %s (%d)", strerror(errno), errno);
1.126 paf 259:
1.171 paf 260: result=http_read_response(response, response_size, sock, fail_on_status_ne_200);
1.142 paf 261: closesocket(sock);
1.145 paf 262: #ifdef PA_USE_ALARM
1.142 paf 263: alarm(0);
1.126 paf 264: #endif
1.147 paf 265: return result;
1.146 paf 266: } catch(...) {
1.145 paf 267: #ifdef PA_USE_ALARM
1.146 paf 268: alarm(0);
1.126 paf 269: #endif
1.146 paf 270: if(sock>=0)
271: closesocket(sock);
1.154 paf 272: rethrow;
1.146 paf 273: }
1.148 paf 274: #ifdef PA_USE_ALARM
1.126 paf 275: }
1.148 paf 276: #endif
1.126 paf 277: }
278:
1.127 paf 279: #ifndef DOXYGEN
280: struct Http_pass_header_info {
1.154 paf 281: Request_charsets* charsets;
1.127 paf 282: String* request;
283: bool user_agent_specified;
284: };
285: #endif
1.154 paf 286: static void http_pass_header(HashStringValue::key_type key,
287: HashStringValue::value_type value,
288: Http_pass_header_info *info) {
289: *info->request <<key<<": "
290: << attributed_meaning_to_string(*value, String::L_HTTP_HEADER, false)
291: << CRLF;
1.135 paf 292:
1.154 paf 293: if(String(key, String::L_TAINTED).change_case(info->charsets->source(), String::CC_UPPER)=="USER-AGENT")
294: info->user_agent_specified=true;
1.126 paf 295: }
1.154 paf 296:
297:
298: static Charset* detect_charset(Charset& source_charset, const String& content_type_value) {
1.156 paf 299: const String::Body CONTENT_TYPE_VALUE=
1.154 paf 300: content_type_value.change_case(source_charset, String::CC_UPPER);
301: // content-type: xxx/xxx; source_charset=WE-NEED-THIS
302: // content-type: xxx/xxx; source_charset="WE-NEED-THIS"
303: // content-type: xxx/xxx; source_charset="WE-NEED-THIS";
304: size_t before_charseteq_pos=CONTENT_TYPE_VALUE.pos("CHARSET=");
305: if(before_charseteq_pos!=STRING_NOT_FOUND) {
306: size_t charset_begin=before_charseteq_pos+8/*CHARSET="*/;
307: size_t open_quote_pos=CONTENT_TYPE_VALUE.pos('"', charset_begin);
308: bool quoted=open_quote_pos==charset_begin;
309: if(quoted)
310: charset_begin++; // skip opening '"'
311: size_t charset_end=CONTENT_TYPE_VALUE.length();
312: if(quoted) {
313: size_t close_quote_pos=CONTENT_TYPE_VALUE.pos('"', charset_begin);
314: if(close_quote_pos!=STRING_NOT_FOUND)
315: charset_end=close_quote_pos;
316: } else {
317: size_t delim_pos=CONTENT_TYPE_VALUE.pos(';', charset_begin);
318: if(delim_pos!=STRING_NOT_FOUND)
319: charset_end=delim_pos;
320: }
1.156 paf 321: const String::Body CHARSET_NAME_BODY=
1.154 paf 322: CONTENT_TYPE_VALUE.mid(charset_begin, charset_end);
323:
324: return &charsets.get(CHARSET_NAME_BODY);
325: }
326:
327: return 0;
328: }
329:
330:
331: #ifndef DOXYGEN
332: struct File_read_http_result {
333: char *str; size_t length;
334: HashStringValue* headers;
335: };
336: #endif
1.155 paf 337: /// @todo build .cookies field. use ^file.tables.SET-COOKIES.menu{ for now
1.154 paf 338: static File_read_http_result file_read_http(Request_charsets& charsets,
339: const String& file_spec,
1.169 paf 340: bool as_text,
341: HashStringValue *options=0) {
1.154 paf 342: File_read_http_result result;
1.126 paf 343: char host[MAX_STRING];
1.129 paf 344: const char* uri;
1.126 paf 345: int port;
1.154 paf 346: const char* method=0;
1.126 paf 347: int timeout=2;
1.142 paf 348: bool fail_on_status_ne_200=true;
1.154 paf 349: Value* vheaders=0;
350: Charset *asked_remote_charset=0;
1.126 paf 351:
1.127 paf 352: if(options) {
353: int valid_options=0;
1.177 ! paf 354: if(Value* vmethod=options->get(HTTP_METHOD_NAME)) {
1.127 paf 355: valid_options++;
1.154 paf 356: method=vmethod->as_string().cstr();
1.127 paf 357: }
1.177 ! paf 358: if(Value* vtimeout=options->get(HTTP_TIMEOUT_NAME)) {
1.127 paf 359: valid_options++;
360: timeout=vtimeout->as_int();
361: }
1.177 ! paf 362: if((vheaders=options->get(HTTP_HEADERS_NAME))) {
1.127 paf 363: valid_options++;
364: }
1.177 ! paf 365: if(Value* vany_status=options->get(HTTP_ANY_STATUS_NAME)) {
1.142 paf 366: valid_options++;
367: fail_on_status_ne_200=!vany_status->as_bool();
368: }
1.177 ! paf 369: if(Value* vcharset_name=options->get(HTTP_CHARSET_NAME)) {
1.154 paf 370: valid_options++;
371: asked_remote_charset=&::charsets.get(vcharset_name->as_string().
372: change_case(charsets.source(), String::CC_UPPER));
373: }
1.142 paf 374:
1.154 paf 375: if(valid_options!=options->count())
1.127 paf 376: throw Exception("parser.runtime",
377: 0,
378: "invalid option passed");
1.154 paf 379: }
380: if(!asked_remote_charset) // defaulting to $request:charset
381: asked_remote_charset=&charsets.source();
382:
383: //preparing request
384: String& connect_string=*new String;
385: // not in ^sql{... L_SQL ...} spirit, but closer to ^file::load one
386: connect_string.append(file_spec, String::L_URI); // tainted pieces -> URI pieces
387:
388: const char* request_cstr;
389: {
390: // influence URLencoding of tainted pieces to String::L_URI lang
391: Temp_client_charset temp(charsets, *asked_remote_charset);
392:
393: const char* connect_string_cstr=connect_string.cstr(String::L_UNSPECIFIED);
1.126 paf 394:
1.154 paf 395: const char* current=connect_string_cstr;
396: if(strncmp(current, "http://", 7)!=0)
397: throw Exception(0,
398: &connect_string,
399: "does not start with http://"); //never
400: current+=7;
401:
402: strncpy(host, current, sizeof(host)-1); host[sizeof(host)-1]=0;
403: char* host_uri=lsplit(host, '/');
404: uri=host_uri?current+(host_uri-1-host):"/";
405: char* port_cstr=lsplit(host, ':');
406: char* error_pos=0;
407: port=port_cstr?strtol(port_cstr, &error_pos, 0):80;
408:
409: //making request
410: String request;
411: if(method)
412: request<<method;
413: else
414: request<<"GET";
415: request<< " "<< uri <<" HTTP/1.0" CRLF
416: "host: "<< host << CRLF;
417: bool user_agent_specified=false;
418: if(vheaders && !vheaders->is_string()) { // allow empty
419: if(HashStringValue *headers=vheaders->get_hash()) {
1.167 paf 420: Http_pass_header_info info={&charsets, &request, false};
1.154 paf 421: headers->for_each(http_pass_header, &info);
422: user_agent_specified=info.user_agent_specified;
423: } else
424: throw Exception("parser.runtime",
425: &connect_string,
426: "headers param must be hash");
427: };
428: if(!user_agent_specified) // defaulting
429: request << "user-agent: " DEFAULT_USER_AGENT CRLF;
430: request << CRLF;
431:
432: request_cstr=request.cstr(String::L_UNSPECIFIED);
433: }
434: // recode those pieces which are not in String::L_URI lang
435: // [those violating HTTP standard, but widly used]
436: request_cstr=Charset::transcode(
437: String::C(request_cstr, strlen(request_cstr)),
438: charsets.source(),
439: *asked_remote_charset);
1.126 paf 440:
441: //sending request
1.171 paf 442: char* response;
443: size_t response_size;
444: int status_code=http_request(response, response_size,
1.154 paf 445: host, port, request_cstr,
1.142 paf 446: timeout, fail_on_status_ne_200);
1.126 paf 447:
448: //processing results
1.171 paf 449: char* raw_body; size_t raw_body_size;
450: char* headers_end_at=strstr(response, CRLF CRLF /*change '4' below along!*/);
451: if(headers_end_at) {
452: raw_body=headers_end_at+4;
453: raw_body_size=response_size-(raw_body-response);
454: } else
1.126 paf 455: throw Exception("http.response",
1.133 paf 456: &connect_string,
1.126 paf 457: "bad response from host - no headers found");
1.171 paf 458:
459: *headers_end_at=0;
460: const String header_block(response, headers_end_at-response, true);
1.126 paf 461:
1.154 paf 462: ArrayString aheaders;
463: result.headers=new HashStringValue;
1.155 paf 464: VHash* vtables=new VHash;
1.177 ! paf 465: result.headers->put(HTTP_TABLES_NAME, vtables);
1.155 paf 466: HashStringValue& tables=vtables->hash();
467:
1.154 paf 468: size_t pos_after=0;
469: header_block.split(aheaders, pos_after, CRLF);
1.126 paf 470:
471: //processing headers
1.154 paf 472: size_t aheaders_count=aheaders.count();
473: Charset* real_remote_charset=0; // undetected, yet
474: for(size_t i=1; i<aheaders_count; i++) {
475: const String& line=*aheaders.get(i);
1.171 paf 476: size_t pos=line.pos(": ", 2);
1.154 paf 477: if(pos==STRING_NOT_FOUND || pos<1)
1.126 paf 478: throw Exception("http.response",
1.154 paf 479: &connect_string,
480: "bad response from host - bad header \"%s\"", line.cstr());
1.156 paf 481: const String::Body HEADER_NAME=
1.154 paf 482: line.mid(0, pos).change_case(charsets.source(), String::CC_UPPER);
483: const String& header_value=line.mid(pos+2, line.length());
484: if(HEADER_NAME=="CONTENT-TYPE")
485: real_remote_charset=detect_charset(charsets.source(), header_value);
1.155 paf 486:
487: // tables
488: {
489: Value *valready=(Value *)tables.get(HEADER_NAME);
490: bool existed=valready!=0;
491: Table *table;
492: if(existed) {
493: // second+ appearence
494: table=valready->get_table();
495: } else {
496: // first appearence
497: Table::columns_type columns =new ArrayString(1);
498: *columns+=new String("value");
499: table=new Table(columns);
500: }
501: // this string becomes next row
502: ArrayString& row=*new ArrayString(1);
503: row+=&header_value;
504: *table+=&row;
505: // not existed before? add it
506: if(!existed)
507: tables.put(HEADER_NAME, new VTable(table));
508: }
509:
1.154 paf 510: result.headers->put(HEADER_NAME, new VString(header_value));
1.126 paf 511: }
1.154 paf 512: // defaulting to used-asked charset [it's never empty!]
513: if(!real_remote_charset)
514: real_remote_charset=asked_remote_charset;
1.126 paf 515:
516: // output response
1.171 paf 517: String::C real_body=String::C(raw_body, raw_body_size);
518: if(as_text && raw_body_size) // must be checked because transcode returns CONST string in case length==0, which contradicts hacking few lines below
1.169 paf 519: real_body=Charset::transcode(real_body, *real_remote_charset, charsets.source());
1.154 paf 520:
521: result.str=const_cast<char *>(real_body.str); // hacking a little
522: result.length=real_body.length;
523: result.headers->put(file_status_name, new VInt(status_code));
524: return result;
1.34 paf 525: }
1.123 paf 526:
1.154 paf 527: #endif
528:
1.123 paf 529: #ifndef DOXYGEN
530: struct File_read_action_info {
1.154 paf 531: char **data; size_t *data_size;
1.126 paf 532: };
1.123 paf 533: #endif
1.154 paf 534: static void file_read_action(
535: struct stat& finfo,
536: int f,
1.166 paf 537: const String& file_spec, const char* /*fname*/, bool as_text,
1.154 paf 538: void *context) {
1.126 paf 539: File_read_action_info& info=*static_cast<File_read_action_info *>(context);
1.123 paf 540: if(size_t to_read_size=(size_t)finfo.st_size) {
1.154 paf 541: *info.data=new(PointerFreeGC) char[to_read_size+(as_text?1:0)];
1.126 paf 542: *info.data_size=(size_t)read(f, *info.data, to_read_size);
1.123 paf 543:
544: if(ssize_t(*info.data_size)<0 || *info.data_size>to_read_size)
1.126 paf 545: throw Exception(0,
1.123 paf 546: &file_spec,
1.173 paf 547: "read failed: actually read %u bytes count not in [0..%u] valid range",
1.126 paf 548: *info.data_size, to_read_size);
1.123 paf 549: } else { // empty file
550: if(as_text) {
1.154 paf 551: *info.data=new(PointerFreeGC) char[1];
1.123 paf 552: *(char*)(*info.data)=0;
553: } else
554: *info.data=0;
555: *info.data_size=0;
556: return;
557: }
1.126 paf 558: }
1.154 paf 559: File_read_result file_read(Request_charsets& charsets, const String& file_spec,
560: bool as_text, HashStringValue *params,
1.126 paf 561: bool fail_on_read_problem) {
1.167 paf 562: File_read_result result={false, 0, 0, 0};
1.154 paf 563: #ifdef PA_HTTP
564: if(file_spec.starts_with("http://")) {
1.126 paf 565: // fail on read problem
1.169 paf 566: File_read_http_result http=file_read_http(charsets, file_spec, as_text, params);
1.154 paf 567: result.success=true;
568: result.str=http.str;
569: result.length=http.length;
570: result.headers=http.headers;
1.126 paf 571: } else {
1.154 paf 572: #endif
1.161 paf 573: if(params && params->count())
574: throw Exception("parser.runtime",
575: 0,
576: "invalid option passed");
577:
1.154 paf 578: File_read_action_info info={&result.str, &result.length};
579: result.success=file_read_action_under_lock(file_spec,
1.126 paf 580: "read", file_read_action, &info,
581: as_text, fail_on_read_problem);
1.154 paf 582: #ifdef PA_HTTP
1.126 paf 583: }
1.154 paf 584: #endif
1.123 paf 585:
1.154 paf 586: if(result.success && as_text) {
1.131 paf 587: // UTF-8 signature: EF BB BF
1.154 paf 588: if(result.length>=3) {
589: char *in=(char *)result.str;
1.159 paf 590: if(strncmp(in, "\xEF\xBB\xBF", 3)==0) {
1.154 paf 591: result.str=in+3; result.length-=3;// skip prefix
1.131 paf 592: }
593: }
594:
1.154 paf 595: fix_line_breaks((char *)(result.str), result.length);
1.123 paf 596: }
1.126 paf 597:
598: return result;
1.123 paf 599: }
600:
1.154 paf 601: #ifdef PA_SAFE_MODE
602: void check_safe_mode(struct stat finfo, const String& file_spec, const char* fname) {
603: if(finfo.st_uid/*foreign?*/!=geteuid()
604: && finfo.st_gid/*foreign?*/!=getegid())
605: throw Exception("parser.runtime",
606: &file_spec,
607: "parser is in safe mode: "
608: "reading files of foreign group and user disabled "
609: "[recompile parser with --disable-safe-mode configure option], "
610: "actual filename '%s', "
611: "fuid(%d)!=euid(%d) or fgid(%d)!=egid(%d)",
612: fname,
613: finfo.st_uid, geteuid(),
614: finfo.st_gid, getegid());
615: }
616: #endif
1.149 paf 617:
1.154 paf 618: bool file_read_action_under_lock(const String& file_spec,
1.126 paf 619: const char* action_name, File_read_action action, void *context,
620: bool as_text,
1.123 paf 621: bool fail_on_read_problem) {
1.154 paf 622: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.33 paf 623: int f;
624:
625: // first open, next stat:
1.45 paf 626: // directory update of NTFS hard links performed on open.
1.33 paf 627: // ex:
628: // a.html:^test[] and b.html hardlink to a.html
629: // user inserts ! before ^test in a.html
1.126 paf 630: // directory entry of b.html in NTFS not updated at once,
1.35 paf 631: // they delay update till open, so we would receive "!^test[" string
632: // if would do stat, next open.
1.123 paf 633: // later: it seems, even this does not help sometimes
1.98 paf 634: if((f=open(fname, O_RDONLY|(as_text?_O_TEXT:_O_BINARY)))>=0) {
1.123 paf 635: try {
1.162 paf 636: if(pa_lock_shared_blocking(f)!=0)
1.126 paf 637: throw Exception("file.lock",
1.123 paf 638: &file_spec,
639: "shared lock failed: %s (%d), actual filename '%s'",
1.154 paf 640: strerror(errno), errno, fname);
1.123 paf 641:
1.124 paf 642: struct stat finfo;
643: if(stat(fname, &finfo)!=0)
644: throw Exception("file.missing", // hardly possible: we just opened it OK
645: &file_spec,
646: "stat failed: %s (%d), actual filename '%s'",
1.154 paf 647: strerror(errno), errno, fname);
1.124 paf 648:
1.140 paf 649: #ifdef PA_SAFE_MODE
1.149 paf 650: check_safe_mode(finfo, file_spec, fname);
1.105 paf 651: #endif
1.32 paf 652:
1.154 paf 653: action(finfo, f, file_spec, fname, as_text, context);
1.123 paf 654: } catch(...) {
1.162 paf 655: pa_unlock(f);close(f);
1.123 paf 656: if(fail_on_read_problem)
1.154 paf 657: rethrow;
1.123 paf 658: return false;
659: }
1.87 paf 660:
1.162 paf 661: pa_unlock(f);close(f);
1.72 parser 662: return true;
1.118 paf 663: } else {
664: if(fail_on_read_problem)
1.126 paf 665: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.118 paf 666: &file_spec,
1.123 paf 667: "%s failed: %s (%d), actual filename '%s'",
1.154 paf 668: action_name, strerror(errno), errno, fname);
1.118 paf 669: return false;
670: }
1.8 paf 671: }
672:
1.63 parser 673: static void create_dir_for_file(const String& file_spec) {
674: size_t pos_after=1;
1.154 paf 675: size_t pos_before;
676: while((pos_before=file_spec.pos('/', pos_after))!=STRING_NOT_FOUND) {
677: mkdir(file_spec.mid(0, pos_before).cstr(String::L_FILE_SPEC), 0775);
1.63 parser 678: pos_after=pos_before+1;
679: }
680: }
681:
1.98 paf 682: bool file_write_action_under_lock(
1.28 paf 683: const String& file_spec,
1.126 paf 684: const char* action_name, File_write_action action, void *context,
685: bool as_text,
686: bool do_append,
687: bool do_block,
1.110 paf 688: bool fail_on_lock_problem) {
1.154 paf 689: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.28 paf 690: int f;
1.80 paf 691: if(access(fname, W_OK)!=0) // no
1.126 paf 692: create_dir_for_file(file_spec);
1.50 paf 693:
1.80 paf 694: if((f=open(fname,
695: O_CREAT|O_RDWR
696: |(as_text?_O_TEXT:_O_BINARY)
1.138 paf 697: |(do_append?O_APPEND:PA_O_TRUNC), 0664))>=0) {
1.162 paf 698: if((do_block?pa_lock_exclusive_blocking(f):pa_lock_exclusive_nonblocking(f))!=0) {
1.126 paf 699: Exception e("file.lock",
1.110 paf 700: &file_spec,
701: "shared lock failed: %s (%d), actual filename '%s'",
1.154 paf 702: strerror(errno), errno, fname);
1.126 paf 703: close(f);
1.110 paf 704: if(fail_on_lock_problem)
705: throw e;
1.98 paf 706: return false;
707: }
1.96 paf 708:
1.158 paf 709: try {
1.126 paf 710: action(f, context);
1.158 paf 711: } catch(...) {
1.138 paf 712: #ifdef HAVE_FTRUNCATE
1.104 paf 713: if(!do_append)
1.125 paf 714: ftruncate(f, lseek(f, 0, SEEK_CUR)); // one can not use O_TRUNC, read lower
1.138 paf 715: #endif
1.162 paf 716: pa_unlock(f);close(f);
1.154 paf 717: rethrow;
1.158 paf 718: }
1.80 paf 719:
1.138 paf 720: #ifdef HAVE_FTRUNCATE
1.104 paf 721: if(!do_append)
1.125 paf 722: 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 723: #endif
1.162 paf 724: pa_unlock(f);close(f);
1.98 paf 725: return true;
1.80 paf 726: } else
1.126 paf 727: throw Exception(errno==EACCES?"file.access":0,
1.80 paf 728: &file_spec,
1.96 paf 729: "%s failed: %s (%d), actual filename '%s'",
1.154 paf 730: action_name, strerror(errno), errno, fname);
1.96 paf 731: // here should be nothing, see rethrow above
732: }
733:
734: #ifndef DOXYGEN
735: struct File_write_action_info {
1.154 paf 736: const char* str; size_t length;
1.126 paf 737: };
1.96 paf 738: #endif
739: static void file_write_action(int f, void *context) {
1.126 paf 740: File_write_action_info& info=*static_cast<File_write_action_info *>(context);
1.154 paf 741: if(info.length) {
742: int written=write(f, info.str, info.length);
1.116 paf 743: if(written<0)
1.126 paf 744: throw Exception(0,
745: 0,
746: "write failed: %s (%d)", strerror(errno), errno);
1.113 paf 747: }
1.96 paf 748: }
749: void file_write(
750: const String& file_spec,
1.154 paf 751: const char* data, size_t size,
1.126 paf 752: bool as_text,
1.96 paf 753: bool do_append) {
1.126 paf 754: File_write_action_info info={data, size};
1.98 paf 755: file_write_action_under_lock(
1.154 paf 756: file_spec,
757: "write", file_write_action, &info,
758: as_text,
759: do_append);
1.30 paf 760: }
761:
1.63 parser 762: // throws nothing! [this is required in file_move & file_delete]
1.50 paf 763: static void rmdir(const String& file_spec, size_t pos_after) {
1.154 paf 764: size_t pos_before;
765: if((pos_before=file_spec.pos('/', pos_after))!=STRING_NOT_FOUND)
1.126 paf 766: rmdir(file_spec, pos_before+1);
1.50 paf 767:
1.154 paf 768: rmdir(file_spec.mid(0, pos_after-1/* / */).cstr(String::L_FILE_SPEC));
1.50 paf 769: }
1.164 paf 770: bool file_delete(const String& file_spec, bool fail_on_problem) {
1.154 paf 771: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.54 parser 772: if(unlink(fname)!=0)
1.164 paf 773: if(fail_on_problem)
1.126 paf 774: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.93 paf 775: &file_spec,
776: "unlink failed: %s (%d), actual filename '%s'",
1.154 paf 777: strerror(errno), errno, fname);
1.93 paf 778: else
779: return false;
1.50 paf 780:
1.126 paf 781: rmdir(file_spec, 1);
1.93 paf 782: return true;
1.60 parser 783: }
1.95 paf 784: void file_move(const String& old_spec, const String& new_spec) {
1.154 paf 785: const char* old_spec_cstr=old_spec.cstr(String::L_FILE_SPEC);
786: const char* new_spec_cstr=new_spec.cstr(String::L_FILE_SPEC);
1.63 parser 787:
1.126 paf 788: create_dir_for_file(new_spec);
1.63 parser 789:
1.60 parser 790: if(rename(old_spec_cstr, new_spec_cstr)!=0)
1.126 paf 791: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.60 parser 792: &old_spec,
793: "rename failed: %s (%d), actual filename '%s' to '%s'",
1.154 paf 794: strerror(errno), errno, old_spec_cstr, new_spec_cstr);
1.63 parser 795:
1.126 paf 796: rmdir(old_spec, 1);
1.31 paf 797: }
798:
1.51 paf 799:
1.126 paf 800: bool entry_exists(const char* fname, struct stat *afinfo) {
1.118 paf 801: struct stat lfinfo;
802: bool result=stat(fname, &lfinfo)==0;
803: if(afinfo)
804: *afinfo=lfinfo;
805: return result;
1.119 paf 806: }
807:
808: bool entry_exists(const String& file_spec) {
1.154 paf 809: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.126 paf 810: return entry_exists(fname, 0);
1.118 paf 811: }
812:
1.51 paf 813: static bool entry_readable(const String& file_spec, bool need_dir) {
1.154 paf 814: char* fname=file_spec.cstrm(String::L_FILE_SPEC);
1.120 paf 815: if(need_dir) {
1.126 paf 816: size_t size=strlen(fname);
1.120 paf 817: while(size) {
1.126 paf 818: char c=fname[size-1];
1.120 paf 819: if(c=='/' || c=='\\')
820: fname[--size]=0;
821: else
822: break;
823: }
824: }
1.51 paf 825: struct stat finfo;
1.118 paf 826: if(access(fname, R_OK)==0 && entry_exists(fname, &finfo)) {
1.109 paf 827: bool is_dir=(finfo.st_mode&S_IFDIR) != 0;
1.51 paf 828: return is_dir==need_dir;
829: }
830: return false;
831: }
1.31 paf 832: bool file_readable(const String& file_spec) {
1.126 paf 833: return entry_readable(file_spec, false);
1.51 paf 834: }
835: bool dir_readable(const String& file_spec) {
1.126 paf 836: return entry_readable(file_spec, true);
1.65 parser 837: }
1.154 paf 838: const String* file_readable(const String& path, const String& name) {
839: String& result=*new String(path);
840: result << "/";
841: result << name;
842: return file_readable(result)?&result:0;
1.43 paf 843: }
844: bool file_executable(const String& file_spec) {
1.154 paf 845: return access(file_spec.cstr(String::L_FILE_SPEC), X_OK)==0;
1.44 paf 846: }
847:
1.64 parser 848: bool file_stat(const String& file_spec,
1.58 parser 849: size_t& rsize,
1.126 paf 850: time_t& ratime,
851: time_t& rmtime,
852: time_t& rctime,
1.64 parser 853: bool fail_on_read_problem) {
1.154 paf 854: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
855: struct stat finfo;
1.44 paf 856: if(stat(fname, &finfo)!=0)
1.64 parser 857: if(fail_on_read_problem)
1.126 paf 858: throw Exception("file.missing",
1.67 parser 859: &file_spec,
860: "getting file size failed: %s (%d), real filename '%s'",
1.154 paf 861: strerror(errno), errno, fname);
1.64 parser 862: else
863: return false;
1.58 parser 864: rsize=finfo.st_size;
865: ratime=finfo.st_atime;
866: rmtime=finfo.st_mtime;
867: rctime=finfo.st_ctime;
1.64 parser 868: return true;
1.18 paf 869: }
870:
1.126 paf 871: char* getrow(char* *row_ref, char delim) {
872: char* result=*row_ref;
1.8 paf 873: if(result) {
1.126 paf 874: *row_ref=strchr(result, delim);
1.8 paf 875: if(*row_ref)
876: *((*row_ref)++)=0;
877: else if(!*result)
878: return 0;
879: }
880: return result;
881: }
882:
1.126 paf 883: char* lsplit(char* string, char delim) {
1.23 paf 884: if(string) {
1.126 paf 885: char* v=strchr(string, delim);
1.8 paf 886: if(v) {
887: *v=0;
888: return v+1;
889: }
890: }
891: return 0;
892: }
893:
1.126 paf 894: char* lsplit(char* *string_ref, char delim) {
895: char* result=*string_ref;
896: char* next=lsplit(*string_ref, delim);
1.8 paf 897: *string_ref=next;
898: return result;
1.9 paf 899: }
900:
1.126 paf 901: char* rsplit(char* string, char delim) {
1.18 paf 902: if(string) {
1.126 paf 903: char* v=strrchr(string, delim);
1.18 paf 904: if(v) {
1.9 paf 905: *v=0;
906: return v+1;
907: }
908: }
909: return NULL;
1.10 paf 910: }
911:
1.37 paf 912: /// @todo less stupid type detection
1.154 paf 913: const char* format(double value, char* fmt) {
1.126 paf 914: char local_buf[MAX_NUMBER];
1.108 paf 915: size_t size;
916:
1.10 paf 917: if(fmt)
918: if(strpbrk(fmt, "diouxX"))
919: if(strpbrk(fmt, "ouxX"))
1.126 paf 920: size=snprintf(local_buf, sizeof(local_buf), fmt, (uint)value);
1.10 paf 921: else
1.126 paf 922: size=snprintf(local_buf, sizeof(local_buf), fmt, (int)value);
1.10 paf 923: else
1.126 paf 924: size=snprintf(local_buf, sizeof(local_buf), fmt, value);
1.10 paf 925: else
1.126 paf 926: size=snprintf(local_buf, sizeof(local_buf), "%d", (int)value);
1.10 paf 927:
1.154 paf 928: return pa_strdup(local_buf, size);
1.12 paf 929: }
930:
1.36 paf 931: size_t stdout_write(const void *buf, size_t size) {
1.12 paf 932: #ifdef WIN32
933: do{
1.154 paf 934: int chunk_written=fwrite(buf, 1, min((size_t)8*0x400, size), stdout);
1.12 paf 935: if(chunk_written<=0)
936: break;
937: size-=chunk_written;
1.36 paf 938: buf=((const char*)buf)+chunk_written;
1.126 paf 939: } while(size>0);
1.12 paf 940:
941: return size;
942: #else
1.126 paf 943: return fwrite(buf, 1, size, stdout);
1.12 paf 944: #endif
1.2 paf 945: }
1.14 paf 946:
1.154 paf 947: char* unescape_chars(const char* cp, int len) {
948: char* s=new(PointerFreeGC) char[len + 1];
1.14 paf 949: enum EscapeState {
1.33 paf 950: EscapeRest,
951: EscapeFirst,
1.14 paf 952: EscapeSecond
953: } escapeState=EscapeRest;
954: int escapedValue=0;
955: int srcPos=0;
956: int dstPos=0;
957: while(srcPos < len) {
1.126 paf 958: int ch=cp[srcPos];
1.14 paf 959: switch(escapeState) {
960: case EscapeRest:
961: if(ch=='%') {
962: escapeState=EscapeFirst;
963: } else if(ch=='+') {
1.126 paf 964: s[dstPos++]=' ';
1.14 paf 965: } else {
966: s[dstPos++]=ch;
967: }
968: break;
969: case EscapeFirst:
970: escapedValue=hex_value[ch] << 4;
971: escapeState=EscapeSecond;
972: break;
973: case EscapeSecond:
1.126 paf 974: escapedValue +=hex_value[ch];
1.14 paf 975: s[dstPos++]=escapedValue;
976: escapeState=EscapeRest;
977: break;
978: }
1.126 paf 979: srcPos++;
1.14 paf 980: }
981: s[dstPos]=0;
982: return s;
1.24 paf 983: }
984:
985: #ifdef WIN32
1.126 paf 986: void back_slashes_to_slashes(char* s) {
1.24 paf 987: if(s)
988: for(; *s; s++)
989: if(*s=='\\')
1.126 paf 990: *s='/';
1.24 paf 991: }
1.42 paf 992: /*
1.126 paf 993: void slashes_to_back_slashes(char* s) {
1.42 paf 994: if(s)
995: for(; *s; s++)
996: if(*s=='/')
1.126 paf 997: *s='\\';
1.42 paf 998: }
999: */
1.24 paf 1000: #endif
1.41 paf 1001:
1.126 paf 1002: bool StrEqNc(const char* s1, const char* s2, bool strict) {
1.41 paf 1003: while(true) {
1004: if(!(*s1)) {
1005: if(!(*s2))
1006: return true;
1007: else
1008: return !strict;
1009: } else if(!(*s2))
1010: return !strict;
1011: if(isalpha(*s1)) {
1012: if(tolower(*s1) !=tolower(*s2))
1013: return false;
1014: } else if((*s1) !=(*s2))
1015: return false;
1.126 paf 1016: s1++;
1017: s2++;
1.41 paf 1018: }
1.57 parser 1019: }
1020:
1.84 paf 1021: static bool isLeap(int year) {
1.57 parser 1022: return !(
1023: (year % 4) || ((year % 400) && !(year % 100))
1.126 paf 1024: );
1.57 parser 1025: }
1026:
1027: int getMonthDays(int year, int month) {
1028: int monthDays[]={
1.126 paf 1029: 31,
1030: isLeap(year) ? 29 : 28,
1031: 31,
1032: 30,
1033: 31,
1034: 30,
1035: 31,
1036: 31,
1037: 30,
1038: 31,
1039: 30,
1.57 parser 1040: 31
1.126 paf 1041: };
1042: return monthDays[month];
1.41 paf 1043: }
1.69 parser 1044:
1.126 paf 1045: void remove_crlf(char* start, char* end) {
1046: for(char* p=start; p<end; p++)
1.69 parser 1047: switch(*p) {
1.126 paf 1048: case '\n': *p='|'; break;
1049: case '\r': *p=' '; break;
1.69 parser 1050: }
1.91 paf 1051: }
1052:
1053:
1054: /// must be last in this file
1055: #undef vsnprintf
1.126 paf 1056: int __vsnprintf(char* b, size_t s, const char* f, va_list l) {
1.91 paf 1057: if(!s)
1058: return 0;
1059:
1060: int r;
1061: // note: on win32& maybe somewhere else
1062: // vsnprintf do not writes terminating 0 in 'buffer full' case, reducing
1063: --s;
1.172 paf 1064:
1065: // clients do not check for negative 's', feature: ignore such prints
1066: if((ssize_t)s<0)
1067: return 0;
1068:
1.91 paf 1069: #if _MSC_VER
1070: /*
1071: win32:
1072: mk:@MSITStore:C:\Program%20Files\Microsoft%20Visual%20Studio\MSDN\2001APR\1033\vccore.chm::/html/_crt__vsnprintf.2c_._vsnwprintf.htm
1073:
1.154 paf 1074: if the number of bytes to write exceeds buffer, then count bytes are written and Ö1 is returned
1.91 paf 1075: */
1.126 paf 1076: r=_vsnprintf(b, s, f, l);
1.91 paf 1077: if(r<0)
1078: r=s;
1079: #else
1.126 paf 1080: r=vsnprintf(b, s, f, l);
1.91 paf 1081: /*
1082: solaris:
1083: man vsnprintf
1084:
1085: The snprintf() function returns the number of characters
1086: formatted, that is, the number of characters that would have
1087: been written to the buffer if it were large enough. If the
1088: value of n is 0 on a call to snprintf(), an unspecified
1089: value less than 1 is returned.
1090: */
1091:
1092: if(r<0)
1093: r=0;
1.167 paf 1094: else if((size_t)r>s)
1.91 paf 1095: r=s;
1096: #endif
1097: b[r]=0;
1098: return r;
1099: }
1100:
1.126 paf 1101: int __snprintf(char* b, size_t s, const char* f, ...) {
1.91 paf 1102: va_list l;
1.126 paf 1103: va_start(l, f);
1104: int r=__vsnprintf(b, s, f, l);
1105: va_end(l);
1.91 paf 1106: return r;
1.98 paf 1107: }
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