Annotation of parser3/src/main/pa_common.C, revision 1.170
1.15 paf 1: /** @file
1.16 paf 2: Parser: commonly functions.
3:
1.154 paf 4: Copyright(c) 2001-2003 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:
1.170 ! paf 8: static const char * const IDENT_COMMON_C="$Date: 2003/11/24 12:21:02 $";
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
66:
67: // defines for globals
68:
69: #define FILE_STATUS_NAME "status"
70:
71: // globals
72:
73: const String file_status_name(FILE_STATUS_NAME);
74:
75: // defines for statics
76:
77: #define HTTP_METHOD_NAME "method"
78: #define HTTP_TIMEOUT_NAME "timeout"
79: #define HTTP_HEADERS_NAME "headers"
80: #define HTTP_ANY_STATUS_NAME "any-status"
81: #define HTTP_CHARSET_NAME "charset"
1.155 paf 82: #define HTTP_TABLES_NAME "tables"
1.154 paf 83:
84: // statics
85:
86: static const String http_method_name(HTTP_METHOD_NAME);
87: static const String http_timeout_name(HTTP_TIMEOUT_NAME);
88: static const String http_headers_name(HTTP_HEADERS_NAME);
89: static const String http_any_status_name(HTTP_ANY_STATUS_NAME);
90: static const String http_charset_name(HTTP_CHARSET_NAME);
1.155 paf 91: static const String http_tables_name(HTTP_TABLES_NAME);
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:
1.154 paf 134: #ifdef PA_HTTP
1.126 paf 135: //http request stuff
1.154 paf 136:
137: #undef CRLF
138: #define CRLF "\r\n"
1.126 paf 139:
140: static bool set_addr(struct sockaddr_in *addr, const char* host, const short port){
141: memset(addr, 0, sizeof(*addr));
142: addr->sin_family=AF_INET;
143: addr->sin_port=htons(port);
144: if(host) {
145: if(struct hostent *hostIP=gethostbyname(host))
146: memcpy(&addr->sin_addr, hostIP->h_addr, hostIP->h_length);
147: else
148: return false;
149: } else
150: addr->sin_addr.s_addr=INADDR_ANY;
151: return true;
152: }
153:
1.142 paf 154: static int http_read_response(String& response, int sock, bool fail_on_status_ne_200){
1.154 paf 155: int result=0;
156: size_t EOLat=0;
1.126 paf 157: while(true) {
1.154 paf 158: char *buf=new(PointerFreeGC) char[MAX_STRING];
1.126 paf 159: ssize_t size=recv(sock, buf, MAX_STRING, 0);
160: if(size<=0)
161: break;
1.154 paf 162: response.append_strdup(buf, size, String::L_TAINTED);
163: if(!result && (EOLat=response.pos(CRLF, 2))!=STRING_NOT_FOUND) { // checking status in first response
1.142 paf 164: const String& status_line=response.mid(0, (size_t)EOLat);
1.154 paf 165: ArrayString astatus;
166: size_t pos_after=0;
167: status_line.split(astatus, pos_after, " ");
168: const String& status_code=*astatus.get(1);
169: result=status_code.as_int();
1.142 paf 170:
1.154 paf 171: if(fail_on_status_ne_200 && result!=200)
1.142 paf 172: throw Exception("http.status",
1.154 paf 173: &status_code,
1.142 paf 174: "invalid HTTP response status");
175: }
176: }
1.154 paf 177: if(result)
178: return result;
1.142 paf 179: else
180: throw Exception("http.response",
181: 0,
1.154 paf 182: "bad response from host - no status found (size=%lu)", response.length());
1.126 paf 183: }
184:
185: /* ********************** request *************************** */
186:
187: #if defined(SIGALRM) && defined(HAVE_SIGSETJMP) && defined(HAVE_SIGLONGJMP)
1.145 paf 188: # define PA_USE_ALARM
1.126 paf 189: #endif
190:
1.145 paf 191: #ifdef PA_USE_ALARM
1.126 paf 192: static sigjmp_buf timeout_env;
193: static void timeout_handler(int sig){
194: siglongjmp(timeout_env, 1);
195: }
196: #endif
197:
1.142 paf 198: static int http_request(String& response,
1.152 paf 199: const char* host, int port,
200: const char* request,
1.166 paf 201: int
202: #ifdef PA_USE_ALARM
203: timeout
204: #endif
205: ,
1.152 paf 206: bool fail_on_status_ne_200) {
1.126 paf 207: if(!host)
208: throw Exception("http.host",
1.154 paf 209: 0,
1.126 paf 210: "zero hostname"); //never
211:
1.145 paf 212: #ifdef PA_USE_ALARM
1.146 paf 213: signal(SIGALRM, timeout_handler);
1.126 paf 214: #endif
215: int sock=-1;
1.145 paf 216: #ifdef PA_USE_ALARM
217: if(sigsetjmp(timeout_env, 1)) {
218: // stupid gcc [2.95.4] generated bad code
219: // which failed to handle sigsetjmp+throw: crashed inside of pre-throw code.
220: // rewritten simplier [though duplicating closesocket code]
221: if(sock>=0)
222: closesocket(sock);
223: throw Exception("http.timeout",
224: origin_string,
225: "timeout occured while retrieving document");
1.146 paf 226: return 0; // never
1.145 paf 227: } else {
228: alarm(timeout);
229: #endif
1.146 paf 230: try {
231: int result;
1.126 paf 232: struct sockaddr_in dest;
1.154 paf 233:
234: if(!set_addr(&dest, host, port))
1.126 paf 235: throw Exception("http.host",
1.154 paf 236: 0,
1.127 paf 237: "can not resolve hostname \"%s\"", host);
1.126 paf 238:
239: if((sock=socket(AF_INET, SOCK_STREAM, IPPROTO_TCP/*0*/))<0)
240: throw Exception("http.connect",
1.154 paf 241: 0,
1.127 paf 242: "can not make socket: %s (%d)", strerror(errno), errno);
1.126 paf 243: if(connect(sock, (struct sockaddr *)&dest, sizeof(dest)))
244: throw Exception("http.connect",
1.154 paf 245: 0,
1.127 paf 246: "can not connect to host \"%s\": %s (%d)", host, strerror(errno), errno);
1.126 paf 247: size_t request_size=strlen(request);
248: if(send(sock, request, request_size, 0)!=(ssize_t)request_size)
249: throw Exception("http.connect",
1.154 paf 250: 0,
1.127 paf 251: "error sending request: %s (%d)", strerror(errno), errno);
1.126 paf 252:
1.142 paf 253: result=http_read_response(response, sock, fail_on_status_ne_200);
254: closesocket(sock);
1.145 paf 255: #ifdef PA_USE_ALARM
1.142 paf 256: alarm(0);
1.126 paf 257: #endif
1.147 paf 258: return result;
1.146 paf 259: } catch(...) {
1.145 paf 260: #ifdef PA_USE_ALARM
1.146 paf 261: alarm(0);
1.126 paf 262: #endif
1.146 paf 263: if(sock>=0)
264: closesocket(sock);
1.154 paf 265: rethrow;
1.146 paf 266: }
1.148 paf 267: #ifdef PA_USE_ALARM
1.126 paf 268: }
1.148 paf 269: #endif
1.126 paf 270: }
271:
1.127 paf 272: #ifndef DOXYGEN
273: struct Http_pass_header_info {
1.154 paf 274: Request_charsets* charsets;
1.127 paf 275: String* request;
276: bool user_agent_specified;
277: };
278: #endif
1.154 paf 279: static void http_pass_header(HashStringValue::key_type key,
280: HashStringValue::value_type value,
281: Http_pass_header_info *info) {
282: *info->request <<key<<": "
283: << attributed_meaning_to_string(*value, String::L_HTTP_HEADER, false)
284: << CRLF;
1.135 paf 285:
1.154 paf 286: if(String(key, String::L_TAINTED).change_case(info->charsets->source(), String::CC_UPPER)=="USER-AGENT")
287: info->user_agent_specified=true;
1.126 paf 288: }
1.154 paf 289:
290:
291: static Charset* detect_charset(Charset& source_charset, const String& content_type_value) {
1.156 paf 292: const String::Body CONTENT_TYPE_VALUE=
1.154 paf 293: content_type_value.change_case(source_charset, String::CC_UPPER);
294: // content-type: xxx/xxx; source_charset=WE-NEED-THIS
295: // content-type: xxx/xxx; source_charset="WE-NEED-THIS"
296: // content-type: xxx/xxx; source_charset="WE-NEED-THIS";
297: size_t before_charseteq_pos=CONTENT_TYPE_VALUE.pos("CHARSET=");
298: if(before_charseteq_pos!=STRING_NOT_FOUND) {
299: size_t charset_begin=before_charseteq_pos+8/*CHARSET="*/;
300: size_t open_quote_pos=CONTENT_TYPE_VALUE.pos('"', charset_begin);
301: bool quoted=open_quote_pos==charset_begin;
302: if(quoted)
303: charset_begin++; // skip opening '"'
304: size_t charset_end=CONTENT_TYPE_VALUE.length();
305: if(quoted) {
306: size_t close_quote_pos=CONTENT_TYPE_VALUE.pos('"', charset_begin);
307: if(close_quote_pos!=STRING_NOT_FOUND)
308: charset_end=close_quote_pos;
309: } else {
310: size_t delim_pos=CONTENT_TYPE_VALUE.pos(';', charset_begin);
311: if(delim_pos!=STRING_NOT_FOUND)
312: charset_end=delim_pos;
313: }
1.156 paf 314: const String::Body CHARSET_NAME_BODY=
1.154 paf 315: CONTENT_TYPE_VALUE.mid(charset_begin, charset_end);
316:
317: return &charsets.get(CHARSET_NAME_BODY);
318: }
319:
320: return 0;
321: }
322:
323:
324: #ifndef DOXYGEN
325: struct File_read_http_result {
326: char *str; size_t length;
327: HashStringValue* headers;
328: };
329: #endif
1.155 paf 330: /// @todo build .cookies field. use ^file.tables.SET-COOKIES.menu{ for now
1.154 paf 331: static File_read_http_result file_read_http(Request_charsets& charsets,
332: const String& file_spec,
1.169 paf 333: bool as_text,
334: HashStringValue *options=0) {
1.154 paf 335: File_read_http_result result;
1.126 paf 336: char host[MAX_STRING];
1.129 paf 337: const char* uri;
1.126 paf 338: int port;
1.154 paf 339: const char* method=0;
1.126 paf 340: int timeout=2;
1.142 paf 341: bool fail_on_status_ne_200=true;
1.154 paf 342: Value* vheaders=0;
343: Charset *asked_remote_charset=0;
1.126 paf 344:
1.127 paf 345: if(options) {
346: int valid_options=0;
1.154 paf 347: if(Value* vmethod=options->get(http_method_name)) {
1.127 paf 348: valid_options++;
1.154 paf 349: method=vmethod->as_string().cstr();
1.127 paf 350: }
1.154 paf 351: if(Value* vtimeout=options->get(http_timeout_name)) {
1.127 paf 352: valid_options++;
353: timeout=vtimeout->as_int();
354: }
1.167 paf 355: if((vheaders=options->get(http_headers_name))) {
1.127 paf 356: valid_options++;
357: }
1.154 paf 358: if(Value* vany_status=options->get(http_any_status_name)) {
1.142 paf 359: valid_options++;
360: fail_on_status_ne_200=!vany_status->as_bool();
361: }
1.154 paf 362: if(Value* vcharset_name=options->get(http_charset_name)) {
363: valid_options++;
364: asked_remote_charset=&::charsets.get(vcharset_name->as_string().
365: change_case(charsets.source(), String::CC_UPPER));
366: }
1.142 paf 367:
1.154 paf 368: if(valid_options!=options->count())
1.127 paf 369: throw Exception("parser.runtime",
370: 0,
371: "invalid option passed");
1.154 paf 372: }
373: if(!asked_remote_charset) // defaulting to $request:charset
374: asked_remote_charset=&charsets.source();
375:
376: //preparing request
377: String& connect_string=*new String;
378: // not in ^sql{... L_SQL ...} spirit, but closer to ^file::load one
379: connect_string.append(file_spec, String::L_URI); // tainted pieces -> URI pieces
380:
381: const char* request_cstr;
382: {
383: // influence URLencoding of tainted pieces to String::L_URI lang
384: Temp_client_charset temp(charsets, *asked_remote_charset);
385:
386: const char* connect_string_cstr=connect_string.cstr(String::L_UNSPECIFIED);
1.126 paf 387:
1.154 paf 388: const char* current=connect_string_cstr;
389: if(strncmp(current, "http://", 7)!=0)
390: throw Exception(0,
391: &connect_string,
392: "does not start with http://"); //never
393: current+=7;
394:
395: strncpy(host, current, sizeof(host)-1); host[sizeof(host)-1]=0;
396: char* host_uri=lsplit(host, '/');
397: uri=host_uri?current+(host_uri-1-host):"/";
398: char* port_cstr=lsplit(host, ':');
399: char* error_pos=0;
400: port=port_cstr?strtol(port_cstr, &error_pos, 0):80;
401:
402:
403: //making request
404: String request;
405: if(method)
406: request<<method;
407: else
408: request<<"GET";
409: request<< " "<< uri <<" HTTP/1.0" CRLF
410: "host: "<< host << CRLF;
411: bool user_agent_specified=false;
412: if(vheaders && !vheaders->is_string()) { // allow empty
413: if(HashStringValue *headers=vheaders->get_hash()) {
1.167 paf 414: Http_pass_header_info info={&charsets, &request, false};
1.154 paf 415: headers->for_each(http_pass_header, &info);
416: user_agent_specified=info.user_agent_specified;
417: } else
418: throw Exception("parser.runtime",
419: &connect_string,
420: "headers param must be hash");
421: };
422: if(!user_agent_specified) // defaulting
423: request << "user-agent: " DEFAULT_USER_AGENT CRLF;
424: request << CRLF;
425:
426: request_cstr=request.cstr(String::L_UNSPECIFIED);
427: }
428: // recode those pieces which are not in String::L_URI lang
429: // [those violating HTTP standard, but widly used]
430: request_cstr=Charset::transcode(
431: String::C(request_cstr, strlen(request_cstr)),
432: charsets.source(),
433: *asked_remote_charset);
1.126 paf 434:
435: //sending request
1.154 paf 436: String response;
1.142 paf 437: int status_code=http_request(response,
1.154 paf 438: host, port, request_cstr,
1.142 paf 439: timeout, fail_on_status_ne_200);
1.126 paf 440:
441: //processing results
1.154 paf 442: size_t pos=response.pos(CRLF CRLF, 4);
443: if(pos==STRING_NOT_FOUND || pos<1) {
1.126 paf 444: throw Exception("http.response",
1.133 paf 445: &connect_string,
1.126 paf 446: "bad response from host - no headers found");
447: }
1.154 paf 448: const String& header_block=response.mid(0, pos);
449: const String& raw_body=response.mid(pos+4, response.length());
1.126 paf 450:
1.154 paf 451: ArrayString aheaders;
452: result.headers=new HashStringValue;
1.155 paf 453: VHash* vtables=new VHash;
454: result.headers->put(http_tables_name, vtables);
455: HashStringValue& tables=vtables->hash();
456:
1.154 paf 457: size_t pos_after=0;
458: header_block.split(aheaders, pos_after, CRLF);
1.126 paf 459:
460: //processing headers
1.154 paf 461: size_t aheaders_count=aheaders.count();
462: Charset* real_remote_charset=0; // undetected, yet
463: for(size_t i=1; i<aheaders_count; i++) {
464: const String& line=*aheaders.get(i);
465: pos=line.pos(": ", 2);
466: if(pos==STRING_NOT_FOUND || pos<1)
1.126 paf 467: throw Exception("http.response",
1.154 paf 468: &connect_string,
469: "bad response from host - bad header \"%s\"", line.cstr());
1.156 paf 470: const String::Body HEADER_NAME=
1.154 paf 471: line.mid(0, pos).change_case(charsets.source(), String::CC_UPPER);
472: const String& header_value=line.mid(pos+2, line.length());
473: if(HEADER_NAME=="CONTENT-TYPE")
474: real_remote_charset=detect_charset(charsets.source(), header_value);
1.155 paf 475:
476: // tables
477: {
478: Value *valready=(Value *)tables.get(HEADER_NAME);
479: bool existed=valready!=0;
480: Table *table;
481: if(existed) {
482: // second+ appearence
483: table=valready->get_table();
484: } else {
485: // first appearence
486: Table::columns_type columns =new ArrayString(1);
487: *columns+=new String("value");
488: table=new Table(columns);
489: }
490: // this string becomes next row
491: ArrayString& row=*new ArrayString(1);
492: row+=&header_value;
493: *table+=&row;
494: // not existed before? add it
495: if(!existed)
496: tables.put(HEADER_NAME, new VTable(table));
497: }
498:
1.154 paf 499: result.headers->put(HEADER_NAME, new VString(header_value));
1.126 paf 500: }
1.154 paf 501: // defaulting to used-asked charset [it's never empty!]
502: if(!real_remote_charset)
503: real_remote_charset=asked_remote_charset;
1.126 paf 504:
505: // output response
1.169 paf 506: String::C real_body=String::C(raw_body.cstrm()/*must be modifiable*/, raw_body.length());
1.170 ! paf 507: if(as_text && raw_body.length()) // must be checked because transcode returns CONST string in case length==0, which contradicts hacking few lines below
1.169 paf 508: real_body=Charset::transcode(real_body, *real_remote_charset, charsets.source());
1.154 paf 509:
510: result.str=const_cast<char *>(real_body.str); // hacking a little
511: result.length=real_body.length;
512: result.headers->put(file_status_name, new VInt(status_code));
513: return result;
1.34 paf 514: }
1.123 paf 515:
1.154 paf 516: #endif
517:
1.123 paf 518: #ifndef DOXYGEN
519: struct File_read_action_info {
1.154 paf 520: char **data; size_t *data_size;
1.126 paf 521: };
1.123 paf 522: #endif
1.154 paf 523: static void file_read_action(
524: struct stat& finfo,
525: int f,
1.166 paf 526: const String& file_spec, const char* /*fname*/, bool as_text,
1.154 paf 527: void *context) {
1.126 paf 528: File_read_action_info& info=*static_cast<File_read_action_info *>(context);
1.123 paf 529: if(size_t to_read_size=(size_t)finfo.st_size) {
1.154 paf 530: *info.data=new(PointerFreeGC) char[to_read_size+(as_text?1:0)];
1.126 paf 531: *info.data_size=(size_t)read(f, *info.data, to_read_size);
1.123 paf 532:
533: if(ssize_t(*info.data_size)<0 || *info.data_size>to_read_size)
1.126 paf 534: throw Exception(0,
1.123 paf 535: &file_spec,
1.160 paf 536: "read failed: actually read %l bytes count not in [0..%lu] valid range",
1.126 paf 537: *info.data_size, to_read_size);
1.123 paf 538: } else { // empty file
539: if(as_text) {
1.154 paf 540: *info.data=new(PointerFreeGC) char[1];
1.123 paf 541: *(char*)(*info.data)=0;
542: } else
543: *info.data=0;
544: *info.data_size=0;
545: return;
546: }
1.126 paf 547: }
1.154 paf 548: File_read_result file_read(Request_charsets& charsets, const String& file_spec,
549: bool as_text, HashStringValue *params,
1.126 paf 550: bool fail_on_read_problem) {
1.167 paf 551: File_read_result result={false, 0, 0, 0};
1.154 paf 552: #ifdef PA_HTTP
553: if(file_spec.starts_with("http://")) {
1.126 paf 554: // fail on read problem
1.169 paf 555: File_read_http_result http=file_read_http(charsets, file_spec, as_text, params);
1.154 paf 556: result.success=true;
557: result.str=http.str;
558: result.length=http.length;
559: result.headers=http.headers;
1.126 paf 560: } else {
1.154 paf 561: #endif
1.161 paf 562: if(params && params->count())
563: throw Exception("parser.runtime",
564: 0,
565: "invalid option passed");
566:
1.154 paf 567: File_read_action_info info={&result.str, &result.length};
568: result.success=file_read_action_under_lock(file_spec,
1.126 paf 569: "read", file_read_action, &info,
570: as_text, fail_on_read_problem);
1.154 paf 571: #ifdef PA_HTTP
1.126 paf 572: }
1.154 paf 573: #endif
1.123 paf 574:
1.154 paf 575: if(result.success && as_text) {
1.131 paf 576: // UTF-8 signature: EF BB BF
1.154 paf 577: if(result.length>=3) {
578: char *in=(char *)result.str;
1.159 paf 579: if(strncmp(in, "\xEF\xBB\xBF", 3)==0) {
1.154 paf 580: result.str=in+3; result.length-=3;// skip prefix
1.131 paf 581: }
582: }
583:
1.154 paf 584: fix_line_breaks((char *)(result.str), result.length);
1.123 paf 585: }
1.126 paf 586:
587: return result;
1.123 paf 588: }
589:
1.154 paf 590: #ifdef PA_SAFE_MODE
591: void check_safe_mode(struct stat finfo, const String& file_spec, const char* fname) {
592: if(finfo.st_uid/*foreign?*/!=geteuid()
593: && finfo.st_gid/*foreign?*/!=getegid())
594: throw Exception("parser.runtime",
595: &file_spec,
596: "parser is in safe mode: "
597: "reading files of foreign group and user disabled "
598: "[recompile parser with --disable-safe-mode configure option], "
599: "actual filename '%s', "
600: "fuid(%d)!=euid(%d) or fgid(%d)!=egid(%d)",
601: fname,
602: finfo.st_uid, geteuid(),
603: finfo.st_gid, getegid());
604: }
605: #endif
1.149 paf 606:
1.154 paf 607: bool file_read_action_under_lock(const String& file_spec,
1.126 paf 608: const char* action_name, File_read_action action, void *context,
609: bool as_text,
1.123 paf 610: bool fail_on_read_problem) {
1.154 paf 611: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.33 paf 612: int f;
613:
614: // first open, next stat:
1.45 paf 615: // directory update of NTFS hard links performed on open.
1.33 paf 616: // ex:
617: // a.html:^test[] and b.html hardlink to a.html
618: // user inserts ! before ^test in a.html
1.126 paf 619: // directory entry of b.html in NTFS not updated at once,
1.35 paf 620: // they delay update till open, so we would receive "!^test[" string
621: // if would do stat, next open.
1.123 paf 622: // later: it seems, even this does not help sometimes
1.98 paf 623: if((f=open(fname, O_RDONLY|(as_text?_O_TEXT:_O_BINARY)))>=0) {
1.123 paf 624: try {
1.162 paf 625: if(pa_lock_shared_blocking(f)!=0)
1.126 paf 626: throw Exception("file.lock",
1.123 paf 627: &file_spec,
628: "shared lock failed: %s (%d), actual filename '%s'",
1.154 paf 629: strerror(errno), errno, fname);
1.123 paf 630:
1.124 paf 631: struct stat finfo;
632: if(stat(fname, &finfo)!=0)
633: throw Exception("file.missing", // hardly possible: we just opened it OK
634: &file_spec,
635: "stat failed: %s (%d), actual filename '%s'",
1.154 paf 636: strerror(errno), errno, fname);
1.124 paf 637:
1.140 paf 638: #ifdef PA_SAFE_MODE
1.149 paf 639: check_safe_mode(finfo, file_spec, fname);
1.105 paf 640: #endif
1.32 paf 641:
1.154 paf 642: action(finfo, f, file_spec, fname, as_text, context);
1.123 paf 643: } catch(...) {
1.162 paf 644: pa_unlock(f);close(f);
1.123 paf 645: if(fail_on_read_problem)
1.154 paf 646: rethrow;
1.123 paf 647: return false;
648: }
1.87 paf 649:
1.162 paf 650: pa_unlock(f);close(f);
1.72 parser 651: return true;
1.118 paf 652: } else {
653: if(fail_on_read_problem)
1.126 paf 654: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.118 paf 655: &file_spec,
1.123 paf 656: "%s failed: %s (%d), actual filename '%s'",
1.154 paf 657: action_name, strerror(errno), errno, fname);
1.118 paf 658: return false;
659: }
1.8 paf 660: }
661:
1.63 parser 662: static void create_dir_for_file(const String& file_spec) {
663: size_t pos_after=1;
1.154 paf 664: size_t pos_before;
665: while((pos_before=file_spec.pos('/', pos_after))!=STRING_NOT_FOUND) {
666: mkdir(file_spec.mid(0, pos_before).cstr(String::L_FILE_SPEC), 0775);
1.63 parser 667: pos_after=pos_before+1;
668: }
669: }
670:
1.98 paf 671: bool file_write_action_under_lock(
1.28 paf 672: const String& file_spec,
1.126 paf 673: const char* action_name, File_write_action action, void *context,
674: bool as_text,
675: bool do_append,
676: bool do_block,
1.110 paf 677: bool fail_on_lock_problem) {
1.154 paf 678: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.28 paf 679: int f;
1.80 paf 680: if(access(fname, W_OK)!=0) // no
1.126 paf 681: create_dir_for_file(file_spec);
1.50 paf 682:
1.80 paf 683: if((f=open(fname,
684: O_CREAT|O_RDWR
685: |(as_text?_O_TEXT:_O_BINARY)
1.138 paf 686: |(do_append?O_APPEND:PA_O_TRUNC), 0664))>=0) {
1.162 paf 687: if((do_block?pa_lock_exclusive_blocking(f):pa_lock_exclusive_nonblocking(f))!=0) {
1.126 paf 688: Exception e("file.lock",
1.110 paf 689: &file_spec,
690: "shared lock failed: %s (%d), actual filename '%s'",
1.154 paf 691: strerror(errno), errno, fname);
1.126 paf 692: close(f);
1.110 paf 693: if(fail_on_lock_problem)
694: throw e;
1.98 paf 695: return false;
696: }
1.96 paf 697:
1.158 paf 698: try {
1.126 paf 699: action(f, context);
1.158 paf 700: } catch(...) {
1.138 paf 701: #ifdef HAVE_FTRUNCATE
1.104 paf 702: if(!do_append)
1.125 paf 703: ftruncate(f, lseek(f, 0, SEEK_CUR)); // one can not use O_TRUNC, read lower
1.138 paf 704: #endif
1.162 paf 705: pa_unlock(f);close(f);
1.154 paf 706: rethrow;
1.158 paf 707: }
1.80 paf 708:
1.138 paf 709: #ifdef HAVE_FTRUNCATE
1.104 paf 710: if(!do_append)
1.125 paf 711: 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 712: #endif
1.162 paf 713: pa_unlock(f);close(f);
1.98 paf 714: return true;
1.80 paf 715: } else
1.126 paf 716: throw Exception(errno==EACCES?"file.access":0,
1.80 paf 717: &file_spec,
1.96 paf 718: "%s failed: %s (%d), actual filename '%s'",
1.154 paf 719: action_name, strerror(errno), errno, fname);
1.96 paf 720: // here should be nothing, see rethrow above
721: }
722:
723: #ifndef DOXYGEN
724: struct File_write_action_info {
1.154 paf 725: const char* str; size_t length;
1.126 paf 726: };
1.96 paf 727: #endif
728: static void file_write_action(int f, void *context) {
1.126 paf 729: File_write_action_info& info=*static_cast<File_write_action_info *>(context);
1.154 paf 730: if(info.length) {
731: int written=write(f, info.str, info.length);
1.116 paf 732: if(written<0)
1.126 paf 733: throw Exception(0,
734: 0,
735: "write failed: %s (%d)", strerror(errno), errno);
1.113 paf 736: }
1.96 paf 737: }
738: void file_write(
739: const String& file_spec,
1.154 paf 740: const char* data, size_t size,
1.126 paf 741: bool as_text,
1.96 paf 742: bool do_append) {
1.126 paf 743: File_write_action_info info={data, size};
1.98 paf 744: file_write_action_under_lock(
1.154 paf 745: file_spec,
746: "write", file_write_action, &info,
747: as_text,
748: do_append);
1.30 paf 749: }
750:
1.63 parser 751: // throws nothing! [this is required in file_move & file_delete]
1.50 paf 752: static void rmdir(const String& file_spec, size_t pos_after) {
1.154 paf 753: size_t pos_before;
754: if((pos_before=file_spec.pos('/', pos_after))!=STRING_NOT_FOUND)
1.126 paf 755: rmdir(file_spec, pos_before+1);
1.50 paf 756:
1.154 paf 757: rmdir(file_spec.mid(0, pos_after-1/* / */).cstr(String::L_FILE_SPEC));
1.50 paf 758: }
1.164 paf 759: bool file_delete(const String& file_spec, bool fail_on_problem) {
1.154 paf 760: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.54 parser 761: if(unlink(fname)!=0)
1.164 paf 762: if(fail_on_problem)
1.126 paf 763: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.93 paf 764: &file_spec,
765: "unlink failed: %s (%d), actual filename '%s'",
1.154 paf 766: strerror(errno), errno, fname);
1.93 paf 767: else
768: return false;
1.50 paf 769:
1.126 paf 770: rmdir(file_spec, 1);
1.93 paf 771: return true;
1.60 parser 772: }
1.95 paf 773: void file_move(const String& old_spec, const String& new_spec) {
1.154 paf 774: const char* old_spec_cstr=old_spec.cstr(String::L_FILE_SPEC);
775: const char* new_spec_cstr=new_spec.cstr(String::L_FILE_SPEC);
1.63 parser 776:
1.126 paf 777: create_dir_for_file(new_spec);
1.63 parser 778:
1.60 parser 779: if(rename(old_spec_cstr, new_spec_cstr)!=0)
1.126 paf 780: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.60 parser 781: &old_spec,
782: "rename failed: %s (%d), actual filename '%s' to '%s'",
1.154 paf 783: strerror(errno), errno, old_spec_cstr, new_spec_cstr);
1.63 parser 784:
1.126 paf 785: rmdir(old_spec, 1);
1.31 paf 786: }
787:
1.51 paf 788:
1.126 paf 789: bool entry_exists(const char* fname, struct stat *afinfo) {
1.118 paf 790: struct stat lfinfo;
791: bool result=stat(fname, &lfinfo)==0;
792: if(afinfo)
793: *afinfo=lfinfo;
794: return result;
1.119 paf 795: }
796:
797: bool entry_exists(const String& file_spec) {
1.154 paf 798: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.126 paf 799: return entry_exists(fname, 0);
1.118 paf 800: }
801:
1.51 paf 802: static bool entry_readable(const String& file_spec, bool need_dir) {
1.154 paf 803: char* fname=file_spec.cstrm(String::L_FILE_SPEC);
1.120 paf 804: if(need_dir) {
1.126 paf 805: size_t size=strlen(fname);
1.120 paf 806: while(size) {
1.126 paf 807: char c=fname[size-1];
1.120 paf 808: if(c=='/' || c=='\\')
809: fname[--size]=0;
810: else
811: break;
812: }
813: }
1.51 paf 814: struct stat finfo;
1.118 paf 815: if(access(fname, R_OK)==0 && entry_exists(fname, &finfo)) {
1.109 paf 816: bool is_dir=(finfo.st_mode&S_IFDIR) != 0;
1.51 paf 817: return is_dir==need_dir;
818: }
819: return false;
820: }
1.31 paf 821: bool file_readable(const String& file_spec) {
1.126 paf 822: return entry_readable(file_spec, false);
1.51 paf 823: }
824: bool dir_readable(const String& file_spec) {
1.126 paf 825: return entry_readable(file_spec, true);
1.65 parser 826: }
1.154 paf 827: const String* file_readable(const String& path, const String& name) {
828: String& result=*new String(path);
829: result << "/";
830: result << name;
831: return file_readable(result)?&result:0;
1.43 paf 832: }
833: bool file_executable(const String& file_spec) {
1.154 paf 834: return access(file_spec.cstr(String::L_FILE_SPEC), X_OK)==0;
1.44 paf 835: }
836:
1.64 parser 837: bool file_stat(const String& file_spec,
1.58 parser 838: size_t& rsize,
1.126 paf 839: time_t& ratime,
840: time_t& rmtime,
841: time_t& rctime,
1.64 parser 842: bool fail_on_read_problem) {
1.154 paf 843: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
844: struct stat finfo;
1.44 paf 845: if(stat(fname, &finfo)!=0)
1.64 parser 846: if(fail_on_read_problem)
1.126 paf 847: throw Exception("file.missing",
1.67 parser 848: &file_spec,
849: "getting file size failed: %s (%d), real filename '%s'",
1.154 paf 850: strerror(errno), errno, fname);
1.64 parser 851: else
852: return false;
1.58 parser 853: rsize=finfo.st_size;
854: ratime=finfo.st_atime;
855: rmtime=finfo.st_mtime;
856: rctime=finfo.st_ctime;
1.64 parser 857: return true;
1.18 paf 858: }
859:
1.126 paf 860: char* getrow(char* *row_ref, char delim) {
861: char* result=*row_ref;
1.8 paf 862: if(result) {
1.126 paf 863: *row_ref=strchr(result, delim);
1.8 paf 864: if(*row_ref)
865: *((*row_ref)++)=0;
866: else if(!*result)
867: return 0;
868: }
869: return result;
870: }
871:
1.126 paf 872: char* lsplit(char* string, char delim) {
1.23 paf 873: if(string) {
1.126 paf 874: char* v=strchr(string, delim);
1.8 paf 875: if(v) {
876: *v=0;
877: return v+1;
878: }
879: }
880: return 0;
881: }
882:
1.126 paf 883: char* lsplit(char* *string_ref, char delim) {
884: char* result=*string_ref;
885: char* next=lsplit(*string_ref, delim);
1.8 paf 886: *string_ref=next;
887: return result;
1.9 paf 888: }
889:
1.126 paf 890: char* rsplit(char* string, char delim) {
1.18 paf 891: if(string) {
1.126 paf 892: char* v=strrchr(string, delim);
1.18 paf 893: if(v) {
1.9 paf 894: *v=0;
895: return v+1;
896: }
897: }
898: return NULL;
1.10 paf 899: }
900:
1.37 paf 901: /// @todo less stupid type detection
1.154 paf 902: const char* format(double value, char* fmt) {
1.126 paf 903: char local_buf[MAX_NUMBER];
1.108 paf 904: size_t size;
905:
1.10 paf 906: if(fmt)
907: if(strpbrk(fmt, "diouxX"))
908: if(strpbrk(fmt, "ouxX"))
1.126 paf 909: size=snprintf(local_buf, sizeof(local_buf), fmt, (uint)value);
1.10 paf 910: else
1.126 paf 911: size=snprintf(local_buf, sizeof(local_buf), fmt, (int)value);
1.10 paf 912: else
1.126 paf 913: size=snprintf(local_buf, sizeof(local_buf), fmt, value);
1.10 paf 914: else
1.126 paf 915: size=snprintf(local_buf, sizeof(local_buf), "%d", (int)value);
1.10 paf 916:
1.154 paf 917: return pa_strdup(local_buf, size);
1.12 paf 918: }
919:
1.36 paf 920: size_t stdout_write(const void *buf, size_t size) {
1.12 paf 921: #ifdef WIN32
922: do{
1.154 paf 923: int chunk_written=fwrite(buf, 1, min((size_t)8*0x400, size), stdout);
1.12 paf 924: if(chunk_written<=0)
925: break;
926: size-=chunk_written;
1.36 paf 927: buf=((const char*)buf)+chunk_written;
1.126 paf 928: } while(size>0);
1.12 paf 929:
930: return size;
931: #else
1.126 paf 932: return fwrite(buf, 1, size, stdout);
1.12 paf 933: #endif
1.2 paf 934: }
1.14 paf 935:
1.154 paf 936: char* unescape_chars(const char* cp, int len) {
937: char* s=new(PointerFreeGC) char[len + 1];
1.14 paf 938: enum EscapeState {
1.33 paf 939: EscapeRest,
940: EscapeFirst,
1.14 paf 941: EscapeSecond
942: } escapeState=EscapeRest;
943: int escapedValue=0;
944: int srcPos=0;
945: int dstPos=0;
946: while(srcPos < len) {
1.126 paf 947: int ch=cp[srcPos];
1.14 paf 948: switch(escapeState) {
949: case EscapeRest:
950: if(ch=='%') {
951: escapeState=EscapeFirst;
952: } else if(ch=='+') {
1.126 paf 953: s[dstPos++]=' ';
1.14 paf 954: } else {
955: s[dstPos++]=ch;
956: }
957: break;
958: case EscapeFirst:
959: escapedValue=hex_value[ch] << 4;
960: escapeState=EscapeSecond;
961: break;
962: case EscapeSecond:
1.126 paf 963: escapedValue +=hex_value[ch];
1.14 paf 964: s[dstPos++]=escapedValue;
965: escapeState=EscapeRest;
966: break;
967: }
1.126 paf 968: srcPos++;
1.14 paf 969: }
970: s[dstPos]=0;
971: return s;
1.24 paf 972: }
973:
974: #ifdef WIN32
1.126 paf 975: void back_slashes_to_slashes(char* s) {
1.24 paf 976: if(s)
977: for(; *s; s++)
978: if(*s=='\\')
1.126 paf 979: *s='/';
1.24 paf 980: }
1.42 paf 981: /*
1.126 paf 982: void slashes_to_back_slashes(char* s) {
1.42 paf 983: if(s)
984: for(; *s; s++)
985: if(*s=='/')
1.126 paf 986: *s='\\';
1.42 paf 987: }
988: */
1.24 paf 989: #endif
1.41 paf 990:
1.126 paf 991: bool StrEqNc(const char* s1, const char* s2, bool strict) {
1.41 paf 992: while(true) {
993: if(!(*s1)) {
994: if(!(*s2))
995: return true;
996: else
997: return !strict;
998: } else if(!(*s2))
999: return !strict;
1000: if(isalpha(*s1)) {
1001: if(tolower(*s1) !=tolower(*s2))
1002: return false;
1003: } else if((*s1) !=(*s2))
1004: return false;
1.126 paf 1005: s1++;
1006: s2++;
1.41 paf 1007: }
1.57 parser 1008: }
1009:
1.84 paf 1010: static bool isLeap(int year) {
1.57 parser 1011: return !(
1012: (year % 4) || ((year % 400) && !(year % 100))
1.126 paf 1013: );
1.57 parser 1014: }
1015:
1016: int getMonthDays(int year, int month) {
1017: int monthDays[]={
1.126 paf 1018: 31,
1019: isLeap(year) ? 29 : 28,
1020: 31,
1021: 30,
1022: 31,
1023: 30,
1024: 31,
1025: 31,
1026: 30,
1027: 31,
1028: 30,
1.57 parser 1029: 31
1.126 paf 1030: };
1031: return monthDays[month];
1.41 paf 1032: }
1.69 parser 1033:
1.126 paf 1034: void remove_crlf(char* start, char* end) {
1035: for(char* p=start; p<end; p++)
1.69 parser 1036: switch(*p) {
1.126 paf 1037: case '\n': *p='|'; break;
1038: case '\r': *p=' '; break;
1.69 parser 1039: }
1.91 paf 1040: }
1041:
1042:
1043: /// must be last in this file
1044: #undef vsnprintf
1.126 paf 1045: int __vsnprintf(char* b, size_t s, const char* f, va_list l) {
1.91 paf 1046: if(!s)
1047: return 0;
1048:
1049: int r;
1050: // note: on win32& maybe somewhere else
1051: // vsnprintf do not writes terminating 0 in 'buffer full' case, reducing
1052: --s;
1053: #if _MSC_VER
1054: /*
1055: win32:
1056: mk:@MSITStore:C:\Program%20Files\Microsoft%20Visual%20Studio\MSDN\2001APR\1033\vccore.chm::/html/_crt__vsnprintf.2c_._vsnwprintf.htm
1057:
1.154 paf 1058: if the number of bytes to write exceeds buffer, then count bytes are written and Ö1 is returned
1.91 paf 1059: */
1.126 paf 1060: r=_vsnprintf(b, s, f, l);
1.91 paf 1061: if(r<0)
1062: r=s;
1063: #else
1.126 paf 1064: r=vsnprintf(b, s, f, l);
1.91 paf 1065: /*
1066: solaris:
1067: man vsnprintf
1068:
1069: The snprintf() function returns the number of characters
1070: formatted, that is, the number of characters that would have
1071: been written to the buffer if it were large enough. If the
1072: value of n is 0 on a call to snprintf(), an unspecified
1073: value less than 1 is returned.
1074: */
1075:
1076: if(r<0)
1077: r=0;
1.167 paf 1078: else if((size_t)r>s)
1.91 paf 1079: r=s;
1080: #endif
1081: b[r]=0;
1082: return r;
1083: }
1084:
1.126 paf 1085: int __snprintf(char* b, size_t s, const char* f, ...) {
1.91 paf 1086: va_list l;
1.126 paf 1087: va_start(l, f);
1088: int r=__vsnprintf(b, s, f, l);
1089: va_end(l);
1.91 paf 1090: return r;
1.98 paf 1091: }
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