Annotation of parser3/src/main/pa_common.C, revision 1.165
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:
1.165 ! paf 8: static const char* IDENT_COMMON_C="$Date: 2003/11/10 06:51:06 $";
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,
201: int timeout,
202: bool fail_on_status_ne_200) {
1.126 paf 203: if(!host)
204: throw Exception("http.host",
1.154 paf 205: 0,
1.126 paf 206: "zero hostname"); //never
207:
1.145 paf 208: #ifdef PA_USE_ALARM
1.146 paf 209: signal(SIGALRM, timeout_handler);
1.126 paf 210: #endif
211: int sock=-1;
1.145 paf 212: #ifdef PA_USE_ALARM
213: if(sigsetjmp(timeout_env, 1)) {
214: // stupid gcc [2.95.4] generated bad code
215: // which failed to handle sigsetjmp+throw: crashed inside of pre-throw code.
216: // rewritten simplier [though duplicating closesocket code]
217: if(sock>=0)
218: closesocket(sock);
219: throw Exception("http.timeout",
220: origin_string,
221: "timeout occured while retrieving document");
1.146 paf 222: return 0; // never
1.145 paf 223: } else {
224: alarm(timeout);
225: #endif
1.146 paf 226: try {
227: int result;
1.126 paf 228: struct sockaddr_in dest;
1.154 paf 229:
230: if(!set_addr(&dest, host, port))
1.126 paf 231: throw Exception("http.host",
1.154 paf 232: 0,
1.127 paf 233: "can not resolve hostname \"%s\"", host);
1.126 paf 234:
235: if((sock=socket(AF_INET, SOCK_STREAM, IPPROTO_TCP/*0*/))<0)
236: throw Exception("http.connect",
1.154 paf 237: 0,
1.127 paf 238: "can not make socket: %s (%d)", strerror(errno), errno);
1.126 paf 239: if(connect(sock, (struct sockaddr *)&dest, sizeof(dest)))
240: throw Exception("http.connect",
1.154 paf 241: 0,
1.127 paf 242: "can not connect to host \"%s\": %s (%d)", host, strerror(errno), errno);
1.126 paf 243: size_t request_size=strlen(request);
244: if(send(sock, request, request_size, 0)!=(ssize_t)request_size)
245: throw Exception("http.connect",
1.154 paf 246: 0,
1.127 paf 247: "error sending request: %s (%d)", strerror(errno), errno);
1.126 paf 248:
1.142 paf 249: result=http_read_response(response, sock, fail_on_status_ne_200);
250: closesocket(sock);
1.145 paf 251: #ifdef PA_USE_ALARM
1.142 paf 252: alarm(0);
1.126 paf 253: #endif
1.147 paf 254: return result;
1.146 paf 255: } catch(...) {
1.145 paf 256: #ifdef PA_USE_ALARM
1.146 paf 257: alarm(0);
1.126 paf 258: #endif
1.146 paf 259: if(sock>=0)
260: closesocket(sock);
1.154 paf 261: rethrow;
1.146 paf 262: }
1.148 paf 263: #ifdef PA_USE_ALARM
1.126 paf 264: }
1.148 paf 265: #endif
1.126 paf 266: }
267:
1.127 paf 268: #ifndef DOXYGEN
269: struct Http_pass_header_info {
1.154 paf 270: Request_charsets* charsets;
1.127 paf 271: String* request;
272: bool user_agent_specified;
273: };
274: #endif
1.154 paf 275: static void http_pass_header(HashStringValue::key_type key,
276: HashStringValue::value_type value,
277: Http_pass_header_info *info) {
278: *info->request <<key<<": "
279: << attributed_meaning_to_string(*value, String::L_HTTP_HEADER, false)
280: << CRLF;
1.135 paf 281:
1.154 paf 282: if(String(key, String::L_TAINTED).change_case(info->charsets->source(), String::CC_UPPER)=="USER-AGENT")
283: info->user_agent_specified=true;
1.126 paf 284: }
1.154 paf 285:
286:
287: static Charset* detect_charset(Charset& source_charset, const String& content_type_value) {
1.156 paf 288: const String::Body CONTENT_TYPE_VALUE=
1.154 paf 289: content_type_value.change_case(source_charset, String::CC_UPPER);
290: // content-type: xxx/xxx; source_charset=WE-NEED-THIS
291: // content-type: xxx/xxx; source_charset="WE-NEED-THIS"
292: // content-type: xxx/xxx; source_charset="WE-NEED-THIS";
293: size_t before_charseteq_pos=CONTENT_TYPE_VALUE.pos("CHARSET=");
294: if(before_charseteq_pos!=STRING_NOT_FOUND) {
295: size_t charset_begin=before_charseteq_pos+8/*CHARSET="*/;
296: size_t open_quote_pos=CONTENT_TYPE_VALUE.pos('"', charset_begin);
297: bool quoted=open_quote_pos==charset_begin;
298: if(quoted)
299: charset_begin++; // skip opening '"'
300: size_t charset_end=CONTENT_TYPE_VALUE.length();
301: if(quoted) {
302: size_t close_quote_pos=CONTENT_TYPE_VALUE.pos('"', charset_begin);
303: if(close_quote_pos!=STRING_NOT_FOUND)
304: charset_end=close_quote_pos;
305: } else {
306: size_t delim_pos=CONTENT_TYPE_VALUE.pos(';', charset_begin);
307: if(delim_pos!=STRING_NOT_FOUND)
308: charset_end=delim_pos;
309: }
1.156 paf 310: const String::Body CHARSET_NAME_BODY=
1.154 paf 311: CONTENT_TYPE_VALUE.mid(charset_begin, charset_end);
312:
313: return &charsets.get(CHARSET_NAME_BODY);
314: }
315:
316: return 0;
317: }
318:
319:
320: #ifndef DOXYGEN
321: struct File_read_http_result {
322: char *str; size_t length;
323: HashStringValue* headers;
324: };
325: #endif
1.155 paf 326: /// @todo build .cookies field. use ^file.tables.SET-COOKIES.menu{ for now
1.154 paf 327: static File_read_http_result file_read_http(Request_charsets& charsets,
328: const String& file_spec,
329: HashStringValue *options=0) {
330: File_read_http_result result;
1.126 paf 331: char host[MAX_STRING];
1.129 paf 332: const char* uri;
1.126 paf 333: int port;
1.154 paf 334: const char* method=0;
1.126 paf 335: int timeout=2;
1.142 paf 336: bool fail_on_status_ne_200=true;
1.154 paf 337: Value* vheaders=0;
338: Charset *asked_remote_charset=0;
1.126 paf 339:
1.127 paf 340: if(options) {
341: int valid_options=0;
1.154 paf 342: if(Value* vmethod=options->get(http_method_name)) {
1.127 paf 343: valid_options++;
1.154 paf 344: method=vmethod->as_string().cstr();
1.127 paf 345: }
1.154 paf 346: if(Value* vtimeout=options->get(http_timeout_name)) {
1.127 paf 347: valid_options++;
348: timeout=vtimeout->as_int();
349: }
1.154 paf 350: if(vheaders=options->get(http_headers_name)) {
1.127 paf 351: valid_options++;
352: }
1.154 paf 353: if(Value* vany_status=options->get(http_any_status_name)) {
1.142 paf 354: valid_options++;
355: fail_on_status_ne_200=!vany_status->as_bool();
356: }
1.154 paf 357: if(Value* vcharset_name=options->get(http_charset_name)) {
358: valid_options++;
359: asked_remote_charset=&::charsets.get(vcharset_name->as_string().
360: change_case(charsets.source(), String::CC_UPPER));
361: }
1.142 paf 362:
1.154 paf 363: if(valid_options!=options->count())
1.127 paf 364: throw Exception("parser.runtime",
365: 0,
366: "invalid option passed");
1.154 paf 367: }
368: if(!asked_remote_charset) // defaulting to $request:charset
369: asked_remote_charset=&charsets.source();
370:
371: //preparing request
372: String& connect_string=*new String;
373: // not in ^sql{... L_SQL ...} spirit, but closer to ^file::load one
374: connect_string.append(file_spec, String::L_URI); // tainted pieces -> URI pieces
375:
376: const char* request_cstr;
377: {
378: // influence URLencoding of tainted pieces to String::L_URI lang
379: Temp_client_charset temp(charsets, *asked_remote_charset);
380:
381: const char* connect_string_cstr=connect_string.cstr(String::L_UNSPECIFIED);
1.126 paf 382:
1.154 paf 383: const char* current=connect_string_cstr;
384: if(strncmp(current, "http://", 7)!=0)
385: throw Exception(0,
386: &connect_string,
387: "does not start with http://"); //never
388: current+=7;
389:
390: strncpy(host, current, sizeof(host)-1); host[sizeof(host)-1]=0;
391: char* host_uri=lsplit(host, '/');
392: uri=host_uri?current+(host_uri-1-host):"/";
393: char* port_cstr=lsplit(host, ':');
394: char* error_pos=0;
395: port=port_cstr?strtol(port_cstr, &error_pos, 0):80;
396:
397:
398: //making request
399: String request;
400: if(method)
401: request<<method;
402: else
403: request<<"GET";
404: request<< " "<< uri <<" HTTP/1.0" CRLF
405: "host: "<< host << CRLF;
406: bool user_agent_specified=false;
407: if(vheaders && !vheaders->is_string()) { // allow empty
408: if(HashStringValue *headers=vheaders->get_hash()) {
409: Http_pass_header_info info={&charsets, &request};
410: headers->for_each(http_pass_header, &info);
411: user_agent_specified=info.user_agent_specified;
412: } else
413: throw Exception("parser.runtime",
414: &connect_string,
415: "headers param must be hash");
416: };
417: if(!user_agent_specified) // defaulting
418: request << "user-agent: " DEFAULT_USER_AGENT CRLF;
419: request << CRLF;
420:
421: request_cstr=request.cstr(String::L_UNSPECIFIED);
422: }
423: // recode those pieces which are not in String::L_URI lang
424: // [those violating HTTP standard, but widly used]
425: request_cstr=Charset::transcode(
426: String::C(request_cstr, strlen(request_cstr)),
427: charsets.source(),
428: *asked_remote_charset);
1.126 paf 429:
430: //sending request
1.154 paf 431: String response;
1.142 paf 432: int status_code=http_request(response,
1.154 paf 433: host, port, request_cstr,
1.142 paf 434: timeout, fail_on_status_ne_200);
1.126 paf 435:
436: //processing results
1.154 paf 437: size_t pos=response.pos(CRLF CRLF, 4);
438: if(pos==STRING_NOT_FOUND || pos<1) {
1.126 paf 439: throw Exception("http.response",
1.133 paf 440: &connect_string,
1.126 paf 441: "bad response from host - no headers found");
442: }
1.154 paf 443: const String& header_block=response.mid(0, pos);
444: const String& raw_body=response.mid(pos+4, response.length());
1.126 paf 445:
1.154 paf 446: ArrayString aheaders;
447: result.headers=new HashStringValue;
1.155 paf 448: VHash* vtables=new VHash;
449: result.headers->put(http_tables_name, vtables);
450: HashStringValue& tables=vtables->hash();
451:
1.154 paf 452: size_t pos_after=0;
453: header_block.split(aheaders, pos_after, CRLF);
1.126 paf 454:
455: //processing headers
1.154 paf 456: size_t aheaders_count=aheaders.count();
457: Charset* real_remote_charset=0; // undetected, yet
458: for(size_t i=1; i<aheaders_count; i++) {
459: const String& line=*aheaders.get(i);
460: pos=line.pos(": ", 2);
461: if(pos==STRING_NOT_FOUND || pos<1)
1.126 paf 462: throw Exception("http.response",
1.154 paf 463: &connect_string,
464: "bad response from host - bad header \"%s\"", line.cstr());
1.156 paf 465: const String::Body HEADER_NAME=
1.154 paf 466: line.mid(0, pos).change_case(charsets.source(), String::CC_UPPER);
467: const String& header_value=line.mid(pos+2, line.length());
468: if(HEADER_NAME=="CONTENT-TYPE")
469: real_remote_charset=detect_charset(charsets.source(), header_value);
1.155 paf 470:
471: // tables
472: {
473: Value *valready=(Value *)tables.get(HEADER_NAME);
474: bool existed=valready!=0;
475: Table *table;
476: if(existed) {
477: // second+ appearence
478: table=valready->get_table();
479: } else {
480: // first appearence
481: Table::columns_type columns =new ArrayString(1);
482: *columns+=new String("value");
483: table=new Table(columns);
484: }
485: // this string becomes next row
486: ArrayString& row=*new ArrayString(1);
487: row+=&header_value;
488: *table+=&row;
489: // not existed before? add it
490: if(!existed)
491: tables.put(HEADER_NAME, new VTable(table));
492: }
493:
1.154 paf 494: result.headers->put(HEADER_NAME, new VString(header_value));
1.126 paf 495: }
1.154 paf 496: // defaulting to used-asked charset [it's never empty!]
497: if(!real_remote_charset)
498: real_remote_charset=asked_remote_charset;
1.126 paf 499:
1.165 ! paf 500: // todo: do that only if ^file::load[text
1.126 paf 501: // output response
1.154 paf 502: String::C real_body=Charset::transcode(
503: String::C(raw_body.cstrm()/*must be modifiable*/, raw_body.length()),
504: *real_remote_charset,
505: charsets.source());
506:
507: result.str=const_cast<char *>(real_body.str); // hacking a little
508: result.length=real_body.length;
509: result.headers->put(file_status_name, new VInt(status_code));
510: return result;
1.34 paf 511: }
1.123 paf 512:
1.154 paf 513: #endif
514:
1.123 paf 515: #ifndef DOXYGEN
516: struct File_read_action_info {
1.154 paf 517: char **data; size_t *data_size;
1.126 paf 518: };
1.123 paf 519: #endif
1.154 paf 520: static void file_read_action(
521: struct stat& finfo,
522: int f,
523: const String& file_spec, const char* fname, bool as_text,
524: void *context) {
1.126 paf 525: File_read_action_info& info=*static_cast<File_read_action_info *>(context);
1.123 paf 526: if(size_t to_read_size=(size_t)finfo.st_size) {
1.154 paf 527: *info.data=new(PointerFreeGC) char[to_read_size+(as_text?1:0)];
1.126 paf 528: *info.data_size=(size_t)read(f, *info.data, to_read_size);
1.123 paf 529:
530: if(ssize_t(*info.data_size)<0 || *info.data_size>to_read_size)
1.126 paf 531: throw Exception(0,
1.123 paf 532: &file_spec,
1.160 paf 533: "read failed: actually read %l bytes count not in [0..%lu] valid range",
1.126 paf 534: *info.data_size, to_read_size);
1.123 paf 535: } else { // empty file
536: if(as_text) {
1.154 paf 537: *info.data=new(PointerFreeGC) char[1];
1.123 paf 538: *(char*)(*info.data)=0;
539: } else
540: *info.data=0;
541: *info.data_size=0;
542: return;
543: }
1.126 paf 544: }
1.154 paf 545: File_read_result file_read(Request_charsets& charsets, const String& file_spec,
546: bool as_text, HashStringValue *params,
1.126 paf 547: bool fail_on_read_problem) {
1.154 paf 548: File_read_result result={false};
549: #ifdef PA_HTTP
550: if(file_spec.starts_with("http://")) {
1.126 paf 551: // fail on read problem
1.154 paf 552: File_read_http_result http=file_read_http(charsets, file_spec, params);
553: result.success=true;
554: result.str=http.str;
555: result.length=http.length;
556: result.headers=http.headers;
1.126 paf 557: } else {
1.154 paf 558: #endif
1.161 paf 559: if(params && params->count())
560: throw Exception("parser.runtime",
561: 0,
562: "invalid option passed");
563:
1.154 paf 564: File_read_action_info info={&result.str, &result.length};
565: result.success=file_read_action_under_lock(file_spec,
1.126 paf 566: "read", file_read_action, &info,
567: as_text, fail_on_read_problem);
1.154 paf 568: #ifdef PA_HTTP
1.126 paf 569: }
1.154 paf 570: #endif
1.123 paf 571:
1.154 paf 572: if(result.success && as_text) {
1.131 paf 573: // UTF-8 signature: EF BB BF
1.154 paf 574: if(result.length>=3) {
575: char *in=(char *)result.str;
1.159 paf 576: if(strncmp(in, "\xEF\xBB\xBF", 3)==0) {
1.154 paf 577: result.str=in+3; result.length-=3;// skip prefix
1.131 paf 578: }
579: }
580:
1.154 paf 581: fix_line_breaks((char *)(result.str), result.length);
1.123 paf 582: // note: after fixing
1.154 paf 583: ((char*&)(result.str))[result.length]=0;
1.123 paf 584: }
1.126 paf 585:
586: return result;
1.123 paf 587: }
588:
1.154 paf 589: #ifdef PA_SAFE_MODE
590: void check_safe_mode(struct stat finfo, const String& file_spec, const char* fname) {
591: if(finfo.st_uid/*foreign?*/!=geteuid()
592: && finfo.st_gid/*foreign?*/!=getegid())
593: throw Exception("parser.runtime",
594: &file_spec,
595: "parser is in safe mode: "
596: "reading files of foreign group and user disabled "
597: "[recompile parser with --disable-safe-mode configure option], "
598: "actual filename '%s', "
599: "fuid(%d)!=euid(%d) or fgid(%d)!=egid(%d)",
600: fname,
601: finfo.st_uid, geteuid(),
602: finfo.st_gid, getegid());
603: }
604: #endif
1.149 paf 605:
1.154 paf 606: bool file_read_action_under_lock(const String& file_spec,
1.126 paf 607: const char* action_name, File_read_action action, void *context,
608: bool as_text,
1.123 paf 609: bool fail_on_read_problem) {
1.154 paf 610: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.33 paf 611: int f;
612:
613: // first open, next stat:
1.45 paf 614: // directory update of NTFS hard links performed on open.
1.33 paf 615: // ex:
616: // a.html:^test[] and b.html hardlink to a.html
617: // user inserts ! before ^test in a.html
1.126 paf 618: // directory entry of b.html in NTFS not updated at once,
1.35 paf 619: // they delay update till open, so we would receive "!^test[" string
620: // if would do stat, next open.
1.123 paf 621: // later: it seems, even this does not help sometimes
1.98 paf 622: if((f=open(fname, O_RDONLY|(as_text?_O_TEXT:_O_BINARY)))>=0) {
1.123 paf 623: try {
1.162 paf 624: if(pa_lock_shared_blocking(f)!=0)
1.126 paf 625: throw Exception("file.lock",
1.123 paf 626: &file_spec,
627: "shared lock failed: %s (%d), actual filename '%s'",
1.154 paf 628: strerror(errno), errno, fname);
1.123 paf 629:
1.124 paf 630: struct stat finfo;
631: if(stat(fname, &finfo)!=0)
632: throw Exception("file.missing", // hardly possible: we just opened it OK
633: &file_spec,
634: "stat failed: %s (%d), actual filename '%s'",
1.154 paf 635: strerror(errno), errno, fname);
1.124 paf 636:
1.140 paf 637: #ifdef PA_SAFE_MODE
1.149 paf 638: check_safe_mode(finfo, file_spec, fname);
1.105 paf 639: #endif
1.32 paf 640:
1.154 paf 641: action(finfo, f, file_spec, fname, as_text, context);
1.123 paf 642: } catch(...) {
1.162 paf 643: pa_unlock(f);close(f);
1.123 paf 644: if(fail_on_read_problem)
1.154 paf 645: rethrow;
1.123 paf 646: return false;
647: }
1.87 paf 648:
1.162 paf 649: pa_unlock(f);close(f);
1.72 parser 650: return true;
1.118 paf 651: } else {
652: if(fail_on_read_problem)
1.126 paf 653: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.118 paf 654: &file_spec,
1.123 paf 655: "%s failed: %s (%d), actual filename '%s'",
1.154 paf 656: action_name, strerror(errno), errno, fname);
1.118 paf 657: return false;
658: }
1.8 paf 659: }
660:
1.63 parser 661: static void create_dir_for_file(const String& file_spec) {
662: size_t pos_after=1;
1.154 paf 663: size_t pos_before;
664: while((pos_before=file_spec.pos('/', pos_after))!=STRING_NOT_FOUND) {
665: mkdir(file_spec.mid(0, pos_before).cstr(String::L_FILE_SPEC), 0775);
1.63 parser 666: pos_after=pos_before+1;
667: }
668: }
669:
1.98 paf 670: bool file_write_action_under_lock(
1.28 paf 671: const String& file_spec,
1.126 paf 672: const char* action_name, File_write_action action, void *context,
673: bool as_text,
674: bool do_append,
675: bool do_block,
1.110 paf 676: bool fail_on_lock_problem) {
1.154 paf 677: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.28 paf 678: int f;
1.80 paf 679: if(access(fname, W_OK)!=0) // no
1.126 paf 680: create_dir_for_file(file_spec);
1.50 paf 681:
1.80 paf 682: if((f=open(fname,
683: O_CREAT|O_RDWR
684: |(as_text?_O_TEXT:_O_BINARY)
1.138 paf 685: |(do_append?O_APPEND:PA_O_TRUNC), 0664))>=0) {
1.162 paf 686: if((do_block?pa_lock_exclusive_blocking(f):pa_lock_exclusive_nonblocking(f))!=0) {
1.126 paf 687: Exception e("file.lock",
1.110 paf 688: &file_spec,
689: "shared lock failed: %s (%d), actual filename '%s'",
1.154 paf 690: strerror(errno), errno, fname);
1.126 paf 691: close(f);
1.110 paf 692: if(fail_on_lock_problem)
693: throw e;
1.98 paf 694: return false;
695: }
1.96 paf 696:
1.158 paf 697: try {
1.126 paf 698: action(f, context);
1.158 paf 699: } catch(...) {
1.138 paf 700: #ifdef HAVE_FTRUNCATE
1.104 paf 701: if(!do_append)
1.125 paf 702: ftruncate(f, lseek(f, 0, SEEK_CUR)); // one can not use O_TRUNC, read lower
1.138 paf 703: #endif
1.162 paf 704: pa_unlock(f);close(f);
1.154 paf 705: rethrow;
1.158 paf 706: }
1.80 paf 707:
1.138 paf 708: #ifdef HAVE_FTRUNCATE
1.104 paf 709: if(!do_append)
1.125 paf 710: 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 711: #endif
1.162 paf 712: pa_unlock(f);close(f);
1.98 paf 713: return true;
1.80 paf 714: } else
1.126 paf 715: throw Exception(errno==EACCES?"file.access":0,
1.80 paf 716: &file_spec,
1.96 paf 717: "%s failed: %s (%d), actual filename '%s'",
1.154 paf 718: action_name, strerror(errno), errno, fname);
1.96 paf 719: // here should be nothing, see rethrow above
720: }
721:
722: #ifndef DOXYGEN
723: struct File_write_action_info {
1.154 paf 724: const char* str; size_t length;
1.126 paf 725: };
1.96 paf 726: #endif
727: static void file_write_action(int f, void *context) {
1.126 paf 728: File_write_action_info& info=*static_cast<File_write_action_info *>(context);
1.154 paf 729: if(info.length) {
730: int written=write(f, info.str, info.length);
1.116 paf 731: if(written<0)
1.126 paf 732: throw Exception(0,
733: 0,
734: "write failed: %s (%d)", strerror(errno), errno);
1.113 paf 735: }
1.96 paf 736: }
737: void file_write(
738: const String& file_spec,
1.154 paf 739: const char* data, size_t size,
1.126 paf 740: bool as_text,
1.96 paf 741: bool do_append) {
1.126 paf 742: File_write_action_info info={data, size};
1.98 paf 743: file_write_action_under_lock(
1.154 paf 744: file_spec,
745: "write", file_write_action, &info,
746: as_text,
747: do_append);
1.30 paf 748: }
749:
1.63 parser 750: // throws nothing! [this is required in file_move & file_delete]
1.50 paf 751: static void rmdir(const String& file_spec, size_t pos_after) {
1.154 paf 752: size_t pos_before;
753: if((pos_before=file_spec.pos('/', pos_after))!=STRING_NOT_FOUND)
1.126 paf 754: rmdir(file_spec, pos_before+1);
1.50 paf 755:
1.154 paf 756: rmdir(file_spec.mid(0, pos_after-1/* / */).cstr(String::L_FILE_SPEC));
1.50 paf 757: }
1.164 paf 758: bool file_delete(const String& file_spec, bool fail_on_problem) {
1.154 paf 759: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.54 parser 760: if(unlink(fname)!=0)
1.164 paf 761: if(fail_on_problem)
1.126 paf 762: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.93 paf 763: &file_spec,
764: "unlink failed: %s (%d), actual filename '%s'",
1.154 paf 765: strerror(errno), errno, fname);
1.93 paf 766: else
767: return false;
1.50 paf 768:
1.126 paf 769: rmdir(file_spec, 1);
1.93 paf 770: return true;
1.60 parser 771: }
1.95 paf 772: void file_move(const String& old_spec, const String& new_spec) {
1.154 paf 773: const char* old_spec_cstr=old_spec.cstr(String::L_FILE_SPEC);
774: const char* new_spec_cstr=new_spec.cstr(String::L_FILE_SPEC);
1.63 parser 775:
1.126 paf 776: create_dir_for_file(new_spec);
1.63 parser 777:
1.60 parser 778: if(rename(old_spec_cstr, new_spec_cstr)!=0)
1.126 paf 779: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.60 parser 780: &old_spec,
781: "rename failed: %s (%d), actual filename '%s' to '%s'",
1.154 paf 782: strerror(errno), errno, old_spec_cstr, new_spec_cstr);
1.63 parser 783:
1.126 paf 784: rmdir(old_spec, 1);
1.31 paf 785: }
786:
1.51 paf 787:
1.126 paf 788: bool entry_exists(const char* fname, struct stat *afinfo) {
1.118 paf 789: struct stat lfinfo;
790: bool result=stat(fname, &lfinfo)==0;
791: if(afinfo)
792: *afinfo=lfinfo;
793: return result;
1.119 paf 794: }
795:
796: bool entry_exists(const String& file_spec) {
1.154 paf 797: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.126 paf 798: return entry_exists(fname, 0);
1.118 paf 799: }
800:
1.51 paf 801: static bool entry_readable(const String& file_spec, bool need_dir) {
1.154 paf 802: char* fname=file_spec.cstrm(String::L_FILE_SPEC);
1.120 paf 803: if(need_dir) {
1.126 paf 804: size_t size=strlen(fname);
1.120 paf 805: while(size) {
1.126 paf 806: char c=fname[size-1];
1.120 paf 807: if(c=='/' || c=='\\')
808: fname[--size]=0;
809: else
810: break;
811: }
812: }
1.51 paf 813: struct stat finfo;
1.118 paf 814: if(access(fname, R_OK)==0 && entry_exists(fname, &finfo)) {
1.109 paf 815: bool is_dir=(finfo.st_mode&S_IFDIR) != 0;
1.51 paf 816: return is_dir==need_dir;
817: }
818: return false;
819: }
1.31 paf 820: bool file_readable(const String& file_spec) {
1.126 paf 821: return entry_readable(file_spec, false);
1.51 paf 822: }
823: bool dir_readable(const String& file_spec) {
1.126 paf 824: return entry_readable(file_spec, true);
1.65 parser 825: }
1.154 paf 826: const String* file_readable(const String& path, const String& name) {
827: String& result=*new String(path);
828: result << "/";
829: result << name;
830: return file_readable(result)?&result:0;
1.43 paf 831: }
832: bool file_executable(const String& file_spec) {
1.154 paf 833: return access(file_spec.cstr(String::L_FILE_SPEC), X_OK)==0;
1.44 paf 834: }
835:
1.64 parser 836: bool file_stat(const String& file_spec,
1.58 parser 837: size_t& rsize,
1.126 paf 838: time_t& ratime,
839: time_t& rmtime,
840: time_t& rctime,
1.64 parser 841: bool fail_on_read_problem) {
1.154 paf 842: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
843: struct stat finfo;
1.44 paf 844: if(stat(fname, &finfo)!=0)
1.64 parser 845: if(fail_on_read_problem)
1.126 paf 846: throw Exception("file.missing",
1.67 parser 847: &file_spec,
848: "getting file size failed: %s (%d), real filename '%s'",
1.154 paf 849: strerror(errno), errno, fname);
1.64 parser 850: else
851: return false;
1.58 parser 852: rsize=finfo.st_size;
853: ratime=finfo.st_atime;
854: rmtime=finfo.st_mtime;
855: rctime=finfo.st_ctime;
1.64 parser 856: return true;
1.18 paf 857: }
858:
1.126 paf 859: char* getrow(char* *row_ref, char delim) {
860: char* result=*row_ref;
1.8 paf 861: if(result) {
1.126 paf 862: *row_ref=strchr(result, delim);
1.8 paf 863: if(*row_ref)
864: *((*row_ref)++)=0;
865: else if(!*result)
866: return 0;
867: }
868: return result;
869: }
870:
1.126 paf 871: char* lsplit(char* string, char delim) {
1.23 paf 872: if(string) {
1.126 paf 873: char* v=strchr(string, delim);
1.8 paf 874: if(v) {
875: *v=0;
876: return v+1;
877: }
878: }
879: return 0;
880: }
881:
1.126 paf 882: char* lsplit(char* *string_ref, char delim) {
883: char* result=*string_ref;
884: char* next=lsplit(*string_ref, delim);
1.8 paf 885: *string_ref=next;
886: return result;
1.9 paf 887: }
888:
1.126 paf 889: char* rsplit(char* string, char delim) {
1.18 paf 890: if(string) {
1.126 paf 891: char* v=strrchr(string, delim);
1.18 paf 892: if(v) {
1.9 paf 893: *v=0;
894: return v+1;
895: }
896: }
897: return NULL;
1.10 paf 898: }
899:
1.37 paf 900: /// @todo less stupid type detection
1.154 paf 901: const char* format(double value, char* fmt) {
1.126 paf 902: char local_buf[MAX_NUMBER];
1.108 paf 903: size_t size;
904:
1.10 paf 905: if(fmt)
906: if(strpbrk(fmt, "diouxX"))
907: if(strpbrk(fmt, "ouxX"))
1.126 paf 908: size=snprintf(local_buf, sizeof(local_buf), fmt, (uint)value);
1.10 paf 909: else
1.126 paf 910: size=snprintf(local_buf, sizeof(local_buf), fmt, (int)value);
1.10 paf 911: else
1.126 paf 912: size=snprintf(local_buf, sizeof(local_buf), fmt, value);
1.10 paf 913: else
1.126 paf 914: size=snprintf(local_buf, sizeof(local_buf), "%d", (int)value);
1.10 paf 915:
1.154 paf 916: return pa_strdup(local_buf, size);
1.12 paf 917: }
918:
1.36 paf 919: size_t stdout_write(const void *buf, size_t size) {
1.12 paf 920: #ifdef WIN32
921: do{
1.154 paf 922: int chunk_written=fwrite(buf, 1, min((size_t)8*0x400, size), stdout);
1.12 paf 923: if(chunk_written<=0)
924: break;
925: size-=chunk_written;
1.36 paf 926: buf=((const char*)buf)+chunk_written;
1.126 paf 927: } while(size>0);
1.12 paf 928:
929: return size;
930: #else
1.126 paf 931: return fwrite(buf, 1, size, stdout);
1.12 paf 932: #endif
1.2 paf 933: }
1.14 paf 934:
1.154 paf 935: char* unescape_chars(const char* cp, int len) {
936: char* s=new(PointerFreeGC) char[len + 1];
1.14 paf 937: enum EscapeState {
1.33 paf 938: EscapeRest,
939: EscapeFirst,
1.14 paf 940: EscapeSecond
941: } escapeState=EscapeRest;
942: int escapedValue=0;
943: int srcPos=0;
944: int dstPos=0;
945: while(srcPos < len) {
1.126 paf 946: int ch=cp[srcPos];
1.14 paf 947: switch(escapeState) {
948: case EscapeRest:
949: if(ch=='%') {
950: escapeState=EscapeFirst;
951: } else if(ch=='+') {
1.126 paf 952: s[dstPos++]=' ';
1.14 paf 953: } else {
954: s[dstPos++]=ch;
955: }
956: break;
957: case EscapeFirst:
958: escapedValue=hex_value[ch] << 4;
959: escapeState=EscapeSecond;
960: break;
961: case EscapeSecond:
1.126 paf 962: escapedValue +=hex_value[ch];
1.14 paf 963: s[dstPos++]=escapedValue;
964: escapeState=EscapeRest;
965: break;
966: }
1.126 paf 967: srcPos++;
1.14 paf 968: }
969: s[dstPos]=0;
970: return s;
1.24 paf 971: }
972:
973: #ifdef WIN32
1.126 paf 974: void back_slashes_to_slashes(char* s) {
1.24 paf 975: if(s)
976: for(; *s; s++)
977: if(*s=='\\')
1.126 paf 978: *s='/';
1.24 paf 979: }
1.42 paf 980: /*
1.126 paf 981: void slashes_to_back_slashes(char* s) {
1.42 paf 982: if(s)
983: for(; *s; s++)
984: if(*s=='/')
1.126 paf 985: *s='\\';
1.42 paf 986: }
987: */
1.24 paf 988: #endif
1.41 paf 989:
1.126 paf 990: bool StrEqNc(const char* s1, const char* s2, bool strict) {
1.41 paf 991: while(true) {
992: if(!(*s1)) {
993: if(!(*s2))
994: return true;
995: else
996: return !strict;
997: } else if(!(*s2))
998: return !strict;
999: if(isalpha(*s1)) {
1000: if(tolower(*s1) !=tolower(*s2))
1001: return false;
1002: } else if((*s1) !=(*s2))
1003: return false;
1.126 paf 1004: s1++;
1005: s2++;
1.41 paf 1006: }
1.57 parser 1007: }
1008:
1.84 paf 1009: static bool isLeap(int year) {
1.57 parser 1010: return !(
1011: (year % 4) || ((year % 400) && !(year % 100))
1.126 paf 1012: );
1.57 parser 1013: }
1014:
1015: int getMonthDays(int year, int month) {
1016: int monthDays[]={
1.126 paf 1017: 31,
1018: isLeap(year) ? 29 : 28,
1019: 31,
1020: 30,
1021: 31,
1022: 30,
1023: 31,
1024: 31,
1025: 30,
1026: 31,
1027: 30,
1.57 parser 1028: 31
1.126 paf 1029: };
1030: return monthDays[month];
1.41 paf 1031: }
1.69 parser 1032:
1.126 paf 1033: void remove_crlf(char* start, char* end) {
1034: for(char* p=start; p<end; p++)
1.69 parser 1035: switch(*p) {
1.126 paf 1036: case '\n': *p='|'; break;
1037: case '\r': *p=' '; break;
1.69 parser 1038: }
1.91 paf 1039: }
1040:
1041:
1042: /// must be last in this file
1043: #undef vsnprintf
1.126 paf 1044: int __vsnprintf(char* b, size_t s, const char* f, va_list l) {
1.91 paf 1045: if(!s)
1046: return 0;
1047:
1048: int r;
1049: // note: on win32& maybe somewhere else
1050: // vsnprintf do not writes terminating 0 in 'buffer full' case, reducing
1051: --s;
1052: #if _MSC_VER
1053: /*
1054: win32:
1055: mk:@MSITStore:C:\Program%20Files\Microsoft%20Visual%20Studio\MSDN\2001APR\1033\vccore.chm::/html/_crt__vsnprintf.2c_._vsnwprintf.htm
1056:
1.154 paf 1057: if the number of bytes to write exceeds buffer, then count bytes are written and Ö1 is returned
1.91 paf 1058: */
1.126 paf 1059: r=_vsnprintf(b, s, f, l);
1.91 paf 1060: if(r<0)
1061: r=s;
1062: #else
1.126 paf 1063: r=vsnprintf(b, s, f, l);
1.91 paf 1064: /*
1065: solaris:
1066: man vsnprintf
1067:
1068: The snprintf() function returns the number of characters
1069: formatted, that is, the number of characters that would have
1070: been written to the buffer if it were large enough. If the
1071: value of n is 0 on a call to snprintf(), an unspecified
1072: value less than 1 is returned.
1073: */
1074:
1075: if(r<0)
1076: r=0;
1077: else if(r>s)
1078: r=s;
1079: #endif
1080: b[r]=0;
1081: return r;
1082: }
1083:
1.126 paf 1084: int __snprintf(char* b, size_t s, const char* f, ...) {
1.91 paf 1085: va_list l;
1.126 paf 1086: va_start(l, f);
1087: int r=__vsnprintf(b, s, f, l);
1088: va_end(l);
1.91 paf 1089: return r;
1.98 paf 1090: }
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