Annotation of parser3/src/main/pa_common.C, revision 1.176
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:
1.176 ! paf 8: static const char * const IDENT_COMMON_C="$Date: 2004/03/01 11:43:01 $";
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: static const String http_method_name(HTTP_METHOD_NAME);
91: static const String http_timeout_name(HTTP_TIMEOUT_NAME);
92: static const String http_headers_name(HTTP_HEADERS_NAME);
93: static const String http_any_status_name(HTTP_ANY_STATUS_NAME);
94: static const String http_charset_name(HTTP_CHARSET_NAME);
1.155 paf 95: static const String http_tables_name(HTTP_TABLES_NAME);
1.154 paf 96:
97: // defines
98:
1.127 paf 99: #define DEFAULT_USER_AGENT "parser3"
100:
1.154 paf 101: // functions
1.127 paf 102:
1.154 paf 103: void fix_line_breaks(char *str, size_t& length) {
1.87 paf 104: //_asm int 3;
1.154 paf 105: const char* const eob=str+length;
106: char* dest=str;
1.72 parser 107: // fix DOS: \r\n -> \n
108: // fix Macintosh: \r -> \n
1.154 paf 109: char* bol=str;
1.137 paf 110: while(char* eol=(char*)memchr(bol, '\r', eob -bol)) {
1.72 parser 111: size_t len=eol-bol;
112: if(dest!=bol)
1.126 paf 113: memcpy(dest, bol, len);
1.72 parser 114: dest+=len;
1.126 paf 115: *dest++='\n';
1.72 parser 116:
1.126 paf 117: if(&eol[1]<eob && eol[1]=='\n') { // \r, \n = DOS
1.72 parser 118: bol=eol+2;
1.154 paf 119: length--;
1.126 paf 120: } else // \r, not \n = Macintosh
1.72 parser 121: bol=eol+1;
122: }
1.154 paf 123: // last piece without \r
1.72 parser 124: if(dest!=bol)
1.126 paf 125: memcpy(dest, bol, eob-bol);
1.154 paf 126: str[length]=0; // terminating
1.72 parser 127: }
1.18 paf 128:
1.154 paf 129: char* file_read_text(Request_charsets& charsets,
130: const String& file_spec,
131: bool fail_on_read_problem,
132: HashStringValue* params/*, HashStringValue* * out_fields*/) {
133: File_read_result file=
134: file_read(charsets, file_spec, true, params, fail_on_read_problem);
135: return file.success?file.str:0;
1.126 paf 136: }
137:
1.154 paf 138: #ifdef PA_HTTP
1.126 paf 139: //http request stuff
1.154 paf 140:
141: #undef CRLF
142: #define CRLF "\r\n"
1.126 paf 143:
144: static bool set_addr(struct sockaddr_in *addr, const char* host, const short port){
145: memset(addr, 0, sizeof(*addr));
146: addr->sin_family=AF_INET;
147: addr->sin_port=htons(port);
148: if(host) {
1.175 paf 149: ulong packed_ip=inet_addr(host);
150: struct hostent *hostIP = packed_ip==INADDR_NONE?
151: gethostbyname(host)
152: : gethostbyaddr((char *)&packed_ip, sizeof(packed_ip), AF_INET);
153: if(hostIP)
1.126 paf 154: memcpy(&addr->sin_addr, hostIP->h_addr, hostIP->h_length);
155: else
156: return false;
157: } else
158: addr->sin_addr.s_addr=INADDR_ANY;
159: return true;
160: }
161:
1.171 paf 162: static int http_read_response(char*& response, size_t& response_size, int sock, bool fail_on_status_ne_200){
163: response=(char*)pa_malloc_atomic(1/*terminator*/); // setting memory block type
164: response[(response_size=0)]=0;
1.154 paf 165: int result=0;
1.171 paf 166: char* EOLat=0;
1.126 paf 167: while(true) {
1.171 paf 168: char buf[MAX_STRING*10];
169: ssize_t received_size=recv(sock, buf, sizeof(buf), 0);
170: if(received_size<=0)
1.126 paf 171: break;
1.171 paf 172: response=(char*)pa_realloc(response, response_size+received_size+1/*terminator*/);
173: memcpy(response+response_size, buf, received_size);
174: response_size+=received_size;
175: response[response_size]=0;
176:
177: if(!result && (EOLat=strstr(response, CRLF))) { // checking status in first response
178: const String status_line(pa_strdup(response, EOLat-response));
1.154 paf 179: ArrayString astatus;
180: size_t pos_after=0;
181: status_line.split(astatus, pos_after, " ");
182: const String& status_code=*astatus.get(1);
183: result=status_code.as_int();
1.142 paf 184:
1.154 paf 185: if(fail_on_status_ne_200 && result!=200)
1.142 paf 186: throw Exception("http.status",
1.154 paf 187: &status_code,
1.142 paf 188: "invalid HTTP response status");
189: }
190: }
1.154 paf 191: if(result)
192: return result;
1.142 paf 193: else
194: throw Exception("http.response",
195: 0,
1.173 paf 196: "bad response from host - no status found (size=%u)", response_size);
1.126 paf 197: }
198:
199: /* ********************** request *************************** */
200:
201: #if defined(SIGALRM) && defined(HAVE_SIGSETJMP) && defined(HAVE_SIGLONGJMP)
1.145 paf 202: # define PA_USE_ALARM
1.126 paf 203: #endif
204:
1.145 paf 205: #ifdef PA_USE_ALARM
1.126 paf 206: static sigjmp_buf timeout_env;
207: static void timeout_handler(int sig){
208: siglongjmp(timeout_env, 1);
209: }
210: #endif
211:
1.171 paf 212: static int http_request(char*& response, size_t& response_size,
1.152 paf 213: const char* host, int port,
214: const char* request,
1.166 paf 215: int
216: #ifdef PA_USE_ALARM
217: timeout
218: #endif
219: ,
1.152 paf 220: bool fail_on_status_ne_200) {
1.126 paf 221: if(!host)
222: throw Exception("http.host",
1.154 paf 223: 0,
1.126 paf 224: "zero hostname"); //never
225:
1.145 paf 226: #ifdef PA_USE_ALARM
1.146 paf 227: signal(SIGALRM, timeout_handler);
1.126 paf 228: #endif
229: int sock=-1;
1.145 paf 230: #ifdef PA_USE_ALARM
231: if(sigsetjmp(timeout_env, 1)) {
232: // stupid gcc [2.95.4] generated bad code
233: // which failed to handle sigsetjmp+throw: crashed inside of pre-throw code.
234: // rewritten simplier [though duplicating closesocket code]
235: if(sock>=0)
236: closesocket(sock);
237: throw Exception("http.timeout",
238: origin_string,
239: "timeout occured while retrieving document");
1.146 paf 240: return 0; // never
1.145 paf 241: } else {
242: alarm(timeout);
243: #endif
1.146 paf 244: try {
245: int result;
1.126 paf 246: struct sockaddr_in dest;
1.154 paf 247:
248: if(!set_addr(&dest, host, port))
1.126 paf 249: throw Exception("http.host",
1.154 paf 250: 0,
1.127 paf 251: "can not resolve hostname \"%s\"", host);
1.126 paf 252:
253: if((sock=socket(AF_INET, SOCK_STREAM, IPPROTO_TCP/*0*/))<0)
254: throw Exception("http.connect",
1.154 paf 255: 0,
1.127 paf 256: "can not make socket: %s (%d)", strerror(errno), errno);
1.126 paf 257: if(connect(sock, (struct sockaddr *)&dest, sizeof(dest)))
258: throw Exception("http.connect",
1.154 paf 259: 0,
1.127 paf 260: "can not connect to host \"%s\": %s (%d)", host, strerror(errno), errno);
1.126 paf 261: size_t request_size=strlen(request);
262: if(send(sock, request, request_size, 0)!=(ssize_t)request_size)
263: throw Exception("http.connect",
1.154 paf 264: 0,
1.127 paf 265: "error sending request: %s (%d)", strerror(errno), errno);
1.126 paf 266:
1.171 paf 267: result=http_read_response(response, response_size, sock, fail_on_status_ne_200);
1.142 paf 268: closesocket(sock);
1.145 paf 269: #ifdef PA_USE_ALARM
1.142 paf 270: alarm(0);
1.126 paf 271: #endif
1.147 paf 272: return result;
1.146 paf 273: } catch(...) {
1.145 paf 274: #ifdef PA_USE_ALARM
1.146 paf 275: alarm(0);
1.126 paf 276: #endif
1.146 paf 277: if(sock>=0)
278: closesocket(sock);
1.154 paf 279: rethrow;
1.146 paf 280: }
1.148 paf 281: #ifdef PA_USE_ALARM
1.126 paf 282: }
1.148 paf 283: #endif
1.126 paf 284: }
285:
1.127 paf 286: #ifndef DOXYGEN
287: struct Http_pass_header_info {
1.154 paf 288: Request_charsets* charsets;
1.127 paf 289: String* request;
290: bool user_agent_specified;
291: };
292: #endif
1.154 paf 293: static void http_pass_header(HashStringValue::key_type key,
294: HashStringValue::value_type value,
295: Http_pass_header_info *info) {
296: *info->request <<key<<": "
297: << attributed_meaning_to_string(*value, String::L_HTTP_HEADER, false)
298: << CRLF;
1.135 paf 299:
1.154 paf 300: if(String(key, String::L_TAINTED).change_case(info->charsets->source(), String::CC_UPPER)=="USER-AGENT")
301: info->user_agent_specified=true;
1.126 paf 302: }
1.154 paf 303:
304:
305: static Charset* detect_charset(Charset& source_charset, const String& content_type_value) {
1.156 paf 306: const String::Body CONTENT_TYPE_VALUE=
1.154 paf 307: content_type_value.change_case(source_charset, String::CC_UPPER);
308: // content-type: xxx/xxx; source_charset=WE-NEED-THIS
309: // content-type: xxx/xxx; source_charset="WE-NEED-THIS"
310: // content-type: xxx/xxx; source_charset="WE-NEED-THIS";
311: size_t before_charseteq_pos=CONTENT_TYPE_VALUE.pos("CHARSET=");
312: if(before_charseteq_pos!=STRING_NOT_FOUND) {
313: size_t charset_begin=before_charseteq_pos+8/*CHARSET="*/;
314: size_t open_quote_pos=CONTENT_TYPE_VALUE.pos('"', charset_begin);
315: bool quoted=open_quote_pos==charset_begin;
316: if(quoted)
317: charset_begin++; // skip opening '"'
318: size_t charset_end=CONTENT_TYPE_VALUE.length();
319: if(quoted) {
320: size_t close_quote_pos=CONTENT_TYPE_VALUE.pos('"', charset_begin);
321: if(close_quote_pos!=STRING_NOT_FOUND)
322: charset_end=close_quote_pos;
323: } else {
324: size_t delim_pos=CONTENT_TYPE_VALUE.pos(';', charset_begin);
325: if(delim_pos!=STRING_NOT_FOUND)
326: charset_end=delim_pos;
327: }
1.156 paf 328: const String::Body CHARSET_NAME_BODY=
1.154 paf 329: CONTENT_TYPE_VALUE.mid(charset_begin, charset_end);
330:
331: return &charsets.get(CHARSET_NAME_BODY);
332: }
333:
334: return 0;
335: }
336:
337:
338: #ifndef DOXYGEN
339: struct File_read_http_result {
340: char *str; size_t length;
341: HashStringValue* headers;
342: };
343: #endif
1.155 paf 344: /// @todo build .cookies field. use ^file.tables.SET-COOKIES.menu{ for now
1.154 paf 345: static File_read_http_result file_read_http(Request_charsets& charsets,
346: const String& file_spec,
1.169 paf 347: bool as_text,
348: HashStringValue *options=0) {
1.154 paf 349: File_read_http_result result;
1.126 paf 350: char host[MAX_STRING];
1.129 paf 351: const char* uri;
1.126 paf 352: int port;
1.154 paf 353: const char* method=0;
1.126 paf 354: int timeout=2;
1.142 paf 355: bool fail_on_status_ne_200=true;
1.154 paf 356: Value* vheaders=0;
357: Charset *asked_remote_charset=0;
1.126 paf 358:
1.127 paf 359: if(options) {
360: int valid_options=0;
1.154 paf 361: if(Value* vmethod=options->get(http_method_name)) {
1.127 paf 362: valid_options++;
1.154 paf 363: method=vmethod->as_string().cstr();
1.127 paf 364: }
1.154 paf 365: if(Value* vtimeout=options->get(http_timeout_name)) {
1.127 paf 366: valid_options++;
367: timeout=vtimeout->as_int();
368: }
1.167 paf 369: if((vheaders=options->get(http_headers_name))) {
1.127 paf 370: valid_options++;
371: }
1.154 paf 372: if(Value* vany_status=options->get(http_any_status_name)) {
1.142 paf 373: valid_options++;
374: fail_on_status_ne_200=!vany_status->as_bool();
375: }
1.154 paf 376: if(Value* vcharset_name=options->get(http_charset_name)) {
377: valid_options++;
378: asked_remote_charset=&::charsets.get(vcharset_name->as_string().
379: change_case(charsets.source(), String::CC_UPPER));
380: }
1.142 paf 381:
1.154 paf 382: if(valid_options!=options->count())
1.127 paf 383: throw Exception("parser.runtime",
384: 0,
385: "invalid option passed");
1.154 paf 386: }
387: if(!asked_remote_charset) // defaulting to $request:charset
388: asked_remote_charset=&charsets.source();
389:
390: //preparing request
391: String& connect_string=*new String;
392: // not in ^sql{... L_SQL ...} spirit, but closer to ^file::load one
393: connect_string.append(file_spec, String::L_URI); // tainted pieces -> URI pieces
394:
395: const char* request_cstr;
396: {
397: // influence URLencoding of tainted pieces to String::L_URI lang
398: Temp_client_charset temp(charsets, *asked_remote_charset);
399:
400: const char* connect_string_cstr=connect_string.cstr(String::L_UNSPECIFIED);
1.126 paf 401:
1.154 paf 402: const char* current=connect_string_cstr;
403: if(strncmp(current, "http://", 7)!=0)
404: throw Exception(0,
405: &connect_string,
406: "does not start with http://"); //never
407: current+=7;
408:
409: strncpy(host, current, sizeof(host)-1); host[sizeof(host)-1]=0;
410: char* host_uri=lsplit(host, '/');
411: uri=host_uri?current+(host_uri-1-host):"/";
412: char* port_cstr=lsplit(host, ':');
413: char* error_pos=0;
414: port=port_cstr?strtol(port_cstr, &error_pos, 0):80;
415:
416:
417: //making request
418: String request;
419: if(method)
420: request<<method;
421: else
422: request<<"GET";
423: request<< " "<< uri <<" HTTP/1.0" CRLF
424: "host: "<< host << CRLF;
425: bool user_agent_specified=false;
426: if(vheaders && !vheaders->is_string()) { // allow empty
427: if(HashStringValue *headers=vheaders->get_hash()) {
1.167 paf 428: Http_pass_header_info info={&charsets, &request, false};
1.154 paf 429: headers->for_each(http_pass_header, &info);
430: user_agent_specified=info.user_agent_specified;
431: } else
432: throw Exception("parser.runtime",
433: &connect_string,
434: "headers param must be hash");
435: };
436: if(!user_agent_specified) // defaulting
437: request << "user-agent: " DEFAULT_USER_AGENT CRLF;
438: request << CRLF;
439:
440: request_cstr=request.cstr(String::L_UNSPECIFIED);
441: }
442: // recode those pieces which are not in String::L_URI lang
443: // [those violating HTTP standard, but widly used]
444: request_cstr=Charset::transcode(
445: String::C(request_cstr, strlen(request_cstr)),
446: charsets.source(),
447: *asked_remote_charset);
1.126 paf 448:
449: //sending request
1.171 paf 450: char* response;
451: size_t response_size;
452: int status_code=http_request(response, response_size,
1.154 paf 453: host, port, request_cstr,
1.142 paf 454: timeout, fail_on_status_ne_200);
1.126 paf 455:
456: //processing results
1.171 paf 457: char* raw_body; size_t raw_body_size;
458: char* headers_end_at=strstr(response, CRLF CRLF /*change '4' below along!*/);
459: if(headers_end_at) {
460: raw_body=headers_end_at+4;
461: raw_body_size=response_size-(raw_body-response);
462: } else
1.126 paf 463: throw Exception("http.response",
1.133 paf 464: &connect_string,
1.126 paf 465: "bad response from host - no headers found");
1.171 paf 466:
467: *headers_end_at=0;
468: const String header_block(response, headers_end_at-response, true);
1.126 paf 469:
1.154 paf 470: ArrayString aheaders;
471: result.headers=new HashStringValue;
1.155 paf 472: VHash* vtables=new VHash;
473: result.headers->put(http_tables_name, vtables);
474: HashStringValue& tables=vtables->hash();
475:
1.154 paf 476: size_t pos_after=0;
477: header_block.split(aheaders, pos_after, CRLF);
1.126 paf 478:
479: //processing headers
1.154 paf 480: size_t aheaders_count=aheaders.count();
481: Charset* real_remote_charset=0; // undetected, yet
482: for(size_t i=1; i<aheaders_count; i++) {
483: const String& line=*aheaders.get(i);
1.171 paf 484: size_t pos=line.pos(": ", 2);
1.154 paf 485: if(pos==STRING_NOT_FOUND || pos<1)
1.126 paf 486: throw Exception("http.response",
1.154 paf 487: &connect_string,
488: "bad response from host - bad header \"%s\"", line.cstr());
1.156 paf 489: const String::Body HEADER_NAME=
1.154 paf 490: line.mid(0, pos).change_case(charsets.source(), String::CC_UPPER);
491: const String& header_value=line.mid(pos+2, line.length());
492: if(HEADER_NAME=="CONTENT-TYPE")
493: real_remote_charset=detect_charset(charsets.source(), header_value);
1.155 paf 494:
495: // tables
496: {
497: Value *valready=(Value *)tables.get(HEADER_NAME);
498: bool existed=valready!=0;
499: Table *table;
500: if(existed) {
501: // second+ appearence
502: table=valready->get_table();
503: } else {
504: // first appearence
505: Table::columns_type columns =new ArrayString(1);
506: *columns+=new String("value");
507: table=new Table(columns);
508: }
509: // this string becomes next row
510: ArrayString& row=*new ArrayString(1);
511: row+=&header_value;
512: *table+=&row;
513: // not existed before? add it
514: if(!existed)
515: tables.put(HEADER_NAME, new VTable(table));
516: }
517:
1.154 paf 518: result.headers->put(HEADER_NAME, new VString(header_value));
1.126 paf 519: }
1.154 paf 520: // defaulting to used-asked charset [it's never empty!]
521: if(!real_remote_charset)
522: real_remote_charset=asked_remote_charset;
1.126 paf 523:
524: // output response
1.171 paf 525: String::C real_body=String::C(raw_body, raw_body_size);
526: 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 527: real_body=Charset::transcode(real_body, *real_remote_charset, charsets.source());
1.154 paf 528:
529: result.str=const_cast<char *>(real_body.str); // hacking a little
530: result.length=real_body.length;
531: result.headers->put(file_status_name, new VInt(status_code));
532: return result;
1.34 paf 533: }
1.123 paf 534:
1.154 paf 535: #endif
536:
1.123 paf 537: #ifndef DOXYGEN
538: struct File_read_action_info {
1.154 paf 539: char **data; size_t *data_size;
1.126 paf 540: };
1.123 paf 541: #endif
1.154 paf 542: static void file_read_action(
543: struct stat& finfo,
544: int f,
1.166 paf 545: const String& file_spec, const char* /*fname*/, bool as_text,
1.154 paf 546: void *context) {
1.126 paf 547: File_read_action_info& info=*static_cast<File_read_action_info *>(context);
1.123 paf 548: if(size_t to_read_size=(size_t)finfo.st_size) {
1.154 paf 549: *info.data=new(PointerFreeGC) char[to_read_size+(as_text?1:0)];
1.126 paf 550: *info.data_size=(size_t)read(f, *info.data, to_read_size);
1.123 paf 551:
552: if(ssize_t(*info.data_size)<0 || *info.data_size>to_read_size)
1.126 paf 553: throw Exception(0,
1.123 paf 554: &file_spec,
1.173 paf 555: "read failed: actually read %u bytes count not in [0..%u] valid range",
1.126 paf 556: *info.data_size, to_read_size);
1.123 paf 557: } else { // empty file
558: if(as_text) {
1.154 paf 559: *info.data=new(PointerFreeGC) char[1];
1.123 paf 560: *(char*)(*info.data)=0;
561: } else
562: *info.data=0;
563: *info.data_size=0;
564: return;
565: }
1.126 paf 566: }
1.154 paf 567: File_read_result file_read(Request_charsets& charsets, const String& file_spec,
568: bool as_text, HashStringValue *params,
1.126 paf 569: bool fail_on_read_problem) {
1.167 paf 570: File_read_result result={false, 0, 0, 0};
1.154 paf 571: #ifdef PA_HTTP
572: if(file_spec.starts_with("http://")) {
1.126 paf 573: // fail on read problem
1.169 paf 574: File_read_http_result http=file_read_http(charsets, file_spec, as_text, params);
1.154 paf 575: result.success=true;
576: result.str=http.str;
577: result.length=http.length;
578: result.headers=http.headers;
1.126 paf 579: } else {
1.154 paf 580: #endif
1.161 paf 581: if(params && params->count())
582: throw Exception("parser.runtime",
583: 0,
584: "invalid option passed");
585:
1.154 paf 586: File_read_action_info info={&result.str, &result.length};
587: result.success=file_read_action_under_lock(file_spec,
1.126 paf 588: "read", file_read_action, &info,
589: as_text, fail_on_read_problem);
1.154 paf 590: #ifdef PA_HTTP
1.126 paf 591: }
1.154 paf 592: #endif
1.123 paf 593:
1.154 paf 594: if(result.success && as_text) {
1.131 paf 595: // UTF-8 signature: EF BB BF
1.154 paf 596: if(result.length>=3) {
597: char *in=(char *)result.str;
1.159 paf 598: if(strncmp(in, "\xEF\xBB\xBF", 3)==0) {
1.154 paf 599: result.str=in+3; result.length-=3;// skip prefix
1.131 paf 600: }
601: }
602:
1.154 paf 603: fix_line_breaks((char *)(result.str), result.length);
1.123 paf 604: }
1.126 paf 605:
606: return result;
1.123 paf 607: }
608:
1.154 paf 609: #ifdef PA_SAFE_MODE
610: void check_safe_mode(struct stat finfo, const String& file_spec, const char* fname) {
611: if(finfo.st_uid/*foreign?*/!=geteuid()
612: && finfo.st_gid/*foreign?*/!=getegid())
613: throw Exception("parser.runtime",
614: &file_spec,
615: "parser is in safe mode: "
616: "reading files of foreign group and user disabled "
617: "[recompile parser with --disable-safe-mode configure option], "
618: "actual filename '%s', "
619: "fuid(%d)!=euid(%d) or fgid(%d)!=egid(%d)",
620: fname,
621: finfo.st_uid, geteuid(),
622: finfo.st_gid, getegid());
623: }
624: #endif
1.149 paf 625:
1.154 paf 626: bool file_read_action_under_lock(const String& file_spec,
1.126 paf 627: const char* action_name, File_read_action action, void *context,
628: bool as_text,
1.123 paf 629: bool fail_on_read_problem) {
1.154 paf 630: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.33 paf 631: int f;
632:
633: // first open, next stat:
1.45 paf 634: // directory update of NTFS hard links performed on open.
1.33 paf 635: // ex:
636: // a.html:^test[] and b.html hardlink to a.html
637: // user inserts ! before ^test in a.html
1.126 paf 638: // directory entry of b.html in NTFS not updated at once,
1.35 paf 639: // they delay update till open, so we would receive "!^test[" string
640: // if would do stat, next open.
1.123 paf 641: // later: it seems, even this does not help sometimes
1.98 paf 642: if((f=open(fname, O_RDONLY|(as_text?_O_TEXT:_O_BINARY)))>=0) {
1.123 paf 643: try {
1.162 paf 644: if(pa_lock_shared_blocking(f)!=0)
1.126 paf 645: throw Exception("file.lock",
1.123 paf 646: &file_spec,
647: "shared lock failed: %s (%d), actual filename '%s'",
1.154 paf 648: strerror(errno), errno, fname);
1.123 paf 649:
1.124 paf 650: struct stat finfo;
651: if(stat(fname, &finfo)!=0)
652: throw Exception("file.missing", // hardly possible: we just opened it OK
653: &file_spec,
654: "stat failed: %s (%d), actual filename '%s'",
1.154 paf 655: strerror(errno), errno, fname);
1.124 paf 656:
1.140 paf 657: #ifdef PA_SAFE_MODE
1.149 paf 658: check_safe_mode(finfo, file_spec, fname);
1.105 paf 659: #endif
1.32 paf 660:
1.154 paf 661: action(finfo, f, file_spec, fname, as_text, context);
1.123 paf 662: } catch(...) {
1.162 paf 663: pa_unlock(f);close(f);
1.123 paf 664: if(fail_on_read_problem)
1.154 paf 665: rethrow;
1.123 paf 666: return false;
667: }
1.87 paf 668:
1.162 paf 669: pa_unlock(f);close(f);
1.72 parser 670: return true;
1.118 paf 671: } else {
672: if(fail_on_read_problem)
1.126 paf 673: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.118 paf 674: &file_spec,
1.123 paf 675: "%s failed: %s (%d), actual filename '%s'",
1.154 paf 676: action_name, strerror(errno), errno, fname);
1.118 paf 677: return false;
678: }
1.8 paf 679: }
680:
1.63 parser 681: static void create_dir_for_file(const String& file_spec) {
682: size_t pos_after=1;
1.154 paf 683: size_t pos_before;
684: while((pos_before=file_spec.pos('/', pos_after))!=STRING_NOT_FOUND) {
685: mkdir(file_spec.mid(0, pos_before).cstr(String::L_FILE_SPEC), 0775);
1.63 parser 686: pos_after=pos_before+1;
687: }
688: }
689:
1.98 paf 690: bool file_write_action_under_lock(
1.28 paf 691: const String& file_spec,
1.126 paf 692: const char* action_name, File_write_action action, void *context,
693: bool as_text,
694: bool do_append,
695: bool do_block,
1.110 paf 696: bool fail_on_lock_problem) {
1.154 paf 697: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.28 paf 698: int f;
1.80 paf 699: if(access(fname, W_OK)!=0) // no
1.126 paf 700: create_dir_for_file(file_spec);
1.50 paf 701:
1.80 paf 702: if((f=open(fname,
703: O_CREAT|O_RDWR
704: |(as_text?_O_TEXT:_O_BINARY)
1.138 paf 705: |(do_append?O_APPEND:PA_O_TRUNC), 0664))>=0) {
1.162 paf 706: if((do_block?pa_lock_exclusive_blocking(f):pa_lock_exclusive_nonblocking(f))!=0) {
1.126 paf 707: Exception e("file.lock",
1.110 paf 708: &file_spec,
709: "shared lock failed: %s (%d), actual filename '%s'",
1.154 paf 710: strerror(errno), errno, fname);
1.126 paf 711: close(f);
1.110 paf 712: if(fail_on_lock_problem)
713: throw e;
1.98 paf 714: return false;
715: }
1.96 paf 716:
1.158 paf 717: try {
1.126 paf 718: action(f, context);
1.158 paf 719: } catch(...) {
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)); // one can not use O_TRUNC, read lower
1.138 paf 723: #endif
1.162 paf 724: pa_unlock(f);close(f);
1.154 paf 725: rethrow;
1.158 paf 726: }
1.80 paf 727:
1.138 paf 728: #ifdef HAVE_FTRUNCATE
1.104 paf 729: if(!do_append)
1.125 paf 730: 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 731: #endif
1.162 paf 732: pa_unlock(f);close(f);
1.98 paf 733: return true;
1.80 paf 734: } else
1.126 paf 735: throw Exception(errno==EACCES?"file.access":0,
1.80 paf 736: &file_spec,
1.96 paf 737: "%s failed: %s (%d), actual filename '%s'",
1.154 paf 738: action_name, strerror(errno), errno, fname);
1.96 paf 739: // here should be nothing, see rethrow above
740: }
741:
742: #ifndef DOXYGEN
743: struct File_write_action_info {
1.154 paf 744: const char* str; size_t length;
1.126 paf 745: };
1.96 paf 746: #endif
747: static void file_write_action(int f, void *context) {
1.126 paf 748: File_write_action_info& info=*static_cast<File_write_action_info *>(context);
1.154 paf 749: if(info.length) {
750: int written=write(f, info.str, info.length);
1.116 paf 751: if(written<0)
1.126 paf 752: throw Exception(0,
753: 0,
754: "write failed: %s (%d)", strerror(errno), errno);
1.113 paf 755: }
1.96 paf 756: }
757: void file_write(
758: const String& file_spec,
1.154 paf 759: const char* data, size_t size,
1.126 paf 760: bool as_text,
1.96 paf 761: bool do_append) {
1.126 paf 762: File_write_action_info info={data, size};
1.98 paf 763: file_write_action_under_lock(
1.154 paf 764: file_spec,
765: "write", file_write_action, &info,
766: as_text,
767: do_append);
1.30 paf 768: }
769:
1.63 parser 770: // throws nothing! [this is required in file_move & file_delete]
1.50 paf 771: static void rmdir(const String& file_spec, size_t pos_after) {
1.154 paf 772: size_t pos_before;
773: if((pos_before=file_spec.pos('/', pos_after))!=STRING_NOT_FOUND)
1.126 paf 774: rmdir(file_spec, pos_before+1);
1.50 paf 775:
1.154 paf 776: rmdir(file_spec.mid(0, pos_after-1/* / */).cstr(String::L_FILE_SPEC));
1.50 paf 777: }
1.164 paf 778: bool file_delete(const String& file_spec, bool fail_on_problem) {
1.154 paf 779: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.54 parser 780: if(unlink(fname)!=0)
1.164 paf 781: if(fail_on_problem)
1.126 paf 782: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.93 paf 783: &file_spec,
784: "unlink failed: %s (%d), actual filename '%s'",
1.154 paf 785: strerror(errno), errno, fname);
1.93 paf 786: else
787: return false;
1.50 paf 788:
1.126 paf 789: rmdir(file_spec, 1);
1.93 paf 790: return true;
1.60 parser 791: }
1.95 paf 792: void file_move(const String& old_spec, const String& new_spec) {
1.154 paf 793: const char* old_spec_cstr=old_spec.cstr(String::L_FILE_SPEC);
794: const char* new_spec_cstr=new_spec.cstr(String::L_FILE_SPEC);
1.63 parser 795:
1.126 paf 796: create_dir_for_file(new_spec);
1.63 parser 797:
1.60 parser 798: if(rename(old_spec_cstr, new_spec_cstr)!=0)
1.126 paf 799: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.60 parser 800: &old_spec,
801: "rename failed: %s (%d), actual filename '%s' to '%s'",
1.154 paf 802: strerror(errno), errno, old_spec_cstr, new_spec_cstr);
1.63 parser 803:
1.126 paf 804: rmdir(old_spec, 1);
1.31 paf 805: }
806:
1.51 paf 807:
1.126 paf 808: bool entry_exists(const char* fname, struct stat *afinfo) {
1.118 paf 809: struct stat lfinfo;
810: bool result=stat(fname, &lfinfo)==0;
811: if(afinfo)
812: *afinfo=lfinfo;
813: return result;
1.119 paf 814: }
815:
816: bool entry_exists(const String& file_spec) {
1.154 paf 817: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.126 paf 818: return entry_exists(fname, 0);
1.118 paf 819: }
820:
1.51 paf 821: static bool entry_readable(const String& file_spec, bool need_dir) {
1.154 paf 822: char* fname=file_spec.cstrm(String::L_FILE_SPEC);
1.120 paf 823: if(need_dir) {
1.126 paf 824: size_t size=strlen(fname);
1.120 paf 825: while(size) {
1.126 paf 826: char c=fname[size-1];
1.120 paf 827: if(c=='/' || c=='\\')
828: fname[--size]=0;
829: else
830: break;
831: }
832: }
1.51 paf 833: struct stat finfo;
1.118 paf 834: if(access(fname, R_OK)==0 && entry_exists(fname, &finfo)) {
1.109 paf 835: bool is_dir=(finfo.st_mode&S_IFDIR) != 0;
1.51 paf 836: return is_dir==need_dir;
837: }
838: return false;
839: }
1.31 paf 840: bool file_readable(const String& file_spec) {
1.126 paf 841: return entry_readable(file_spec, false);
1.51 paf 842: }
843: bool dir_readable(const String& file_spec) {
1.126 paf 844: return entry_readable(file_spec, true);
1.65 parser 845: }
1.154 paf 846: const String* file_readable(const String& path, const String& name) {
847: String& result=*new String(path);
848: result << "/";
849: result << name;
850: return file_readable(result)?&result:0;
1.43 paf 851: }
852: bool file_executable(const String& file_spec) {
1.154 paf 853: return access(file_spec.cstr(String::L_FILE_SPEC), X_OK)==0;
1.44 paf 854: }
855:
1.64 parser 856: bool file_stat(const String& file_spec,
1.58 parser 857: size_t& rsize,
1.126 paf 858: time_t& ratime,
859: time_t& rmtime,
860: time_t& rctime,
1.64 parser 861: bool fail_on_read_problem) {
1.154 paf 862: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
863: struct stat finfo;
1.44 paf 864: if(stat(fname, &finfo)!=0)
1.64 parser 865: if(fail_on_read_problem)
1.126 paf 866: throw Exception("file.missing",
1.67 parser 867: &file_spec,
868: "getting file size failed: %s (%d), real filename '%s'",
1.154 paf 869: strerror(errno), errno, fname);
1.64 parser 870: else
871: return false;
1.58 parser 872: rsize=finfo.st_size;
873: ratime=finfo.st_atime;
874: rmtime=finfo.st_mtime;
875: rctime=finfo.st_ctime;
1.64 parser 876: return true;
1.18 paf 877: }
878:
1.126 paf 879: char* getrow(char* *row_ref, char delim) {
880: char* result=*row_ref;
1.8 paf 881: if(result) {
1.126 paf 882: *row_ref=strchr(result, delim);
1.8 paf 883: if(*row_ref)
884: *((*row_ref)++)=0;
885: else if(!*result)
886: return 0;
887: }
888: return result;
889: }
890:
1.126 paf 891: char* lsplit(char* string, char delim) {
1.23 paf 892: if(string) {
1.126 paf 893: char* v=strchr(string, delim);
1.8 paf 894: if(v) {
895: *v=0;
896: return v+1;
897: }
898: }
899: return 0;
900: }
901:
1.126 paf 902: char* lsplit(char* *string_ref, char delim) {
903: char* result=*string_ref;
904: char* next=lsplit(*string_ref, delim);
1.8 paf 905: *string_ref=next;
906: return result;
1.9 paf 907: }
908:
1.126 paf 909: char* rsplit(char* string, char delim) {
1.18 paf 910: if(string) {
1.126 paf 911: char* v=strrchr(string, delim);
1.18 paf 912: if(v) {
1.9 paf 913: *v=0;
914: return v+1;
915: }
916: }
917: return NULL;
1.10 paf 918: }
919:
1.37 paf 920: /// @todo less stupid type detection
1.154 paf 921: const char* format(double value, char* fmt) {
1.126 paf 922: char local_buf[MAX_NUMBER];
1.108 paf 923: size_t size;
924:
1.10 paf 925: if(fmt)
926: if(strpbrk(fmt, "diouxX"))
927: if(strpbrk(fmt, "ouxX"))
1.126 paf 928: size=snprintf(local_buf, sizeof(local_buf), fmt, (uint)value);
1.10 paf 929: else
1.126 paf 930: size=snprintf(local_buf, sizeof(local_buf), fmt, (int)value);
1.10 paf 931: else
1.126 paf 932: size=snprintf(local_buf, sizeof(local_buf), fmt, value);
1.10 paf 933: else
1.126 paf 934: size=snprintf(local_buf, sizeof(local_buf), "%d", (int)value);
1.10 paf 935:
1.154 paf 936: return pa_strdup(local_buf, size);
1.12 paf 937: }
938:
1.36 paf 939: size_t stdout_write(const void *buf, size_t size) {
1.12 paf 940: #ifdef WIN32
941: do{
1.154 paf 942: int chunk_written=fwrite(buf, 1, min((size_t)8*0x400, size), stdout);
1.12 paf 943: if(chunk_written<=0)
944: break;
945: size-=chunk_written;
1.36 paf 946: buf=((const char*)buf)+chunk_written;
1.126 paf 947: } while(size>0);
1.12 paf 948:
949: return size;
950: #else
1.126 paf 951: return fwrite(buf, 1, size, stdout);
1.12 paf 952: #endif
1.2 paf 953: }
1.14 paf 954:
1.154 paf 955: char* unescape_chars(const char* cp, int len) {
956: char* s=new(PointerFreeGC) char[len + 1];
1.14 paf 957: enum EscapeState {
1.33 paf 958: EscapeRest,
959: EscapeFirst,
1.14 paf 960: EscapeSecond
961: } escapeState=EscapeRest;
962: int escapedValue=0;
963: int srcPos=0;
964: int dstPos=0;
965: while(srcPos < len) {
1.126 paf 966: int ch=cp[srcPos];
1.14 paf 967: switch(escapeState) {
968: case EscapeRest:
969: if(ch=='%') {
970: escapeState=EscapeFirst;
971: } else if(ch=='+') {
1.126 paf 972: s[dstPos++]=' ';
1.14 paf 973: } else {
974: s[dstPos++]=ch;
975: }
976: break;
977: case EscapeFirst:
978: escapedValue=hex_value[ch] << 4;
979: escapeState=EscapeSecond;
980: break;
981: case EscapeSecond:
1.126 paf 982: escapedValue +=hex_value[ch];
1.14 paf 983: s[dstPos++]=escapedValue;
984: escapeState=EscapeRest;
985: break;
986: }
1.126 paf 987: srcPos++;
1.14 paf 988: }
989: s[dstPos]=0;
990: return s;
1.24 paf 991: }
992:
993: #ifdef WIN32
1.126 paf 994: void back_slashes_to_slashes(char* s) {
1.24 paf 995: if(s)
996: for(; *s; s++)
997: if(*s=='\\')
1.126 paf 998: *s='/';
1.24 paf 999: }
1.42 paf 1000: /*
1.126 paf 1001: void slashes_to_back_slashes(char* s) {
1.42 paf 1002: if(s)
1003: for(; *s; s++)
1004: if(*s=='/')
1.126 paf 1005: *s='\\';
1.42 paf 1006: }
1007: */
1.24 paf 1008: #endif
1.41 paf 1009:
1.126 paf 1010: bool StrEqNc(const char* s1, const char* s2, bool strict) {
1.41 paf 1011: while(true) {
1012: if(!(*s1)) {
1013: if(!(*s2))
1014: return true;
1015: else
1016: return !strict;
1017: } else if(!(*s2))
1018: return !strict;
1019: if(isalpha(*s1)) {
1020: if(tolower(*s1) !=tolower(*s2))
1021: return false;
1022: } else if((*s1) !=(*s2))
1023: return false;
1.126 paf 1024: s1++;
1025: s2++;
1.41 paf 1026: }
1.57 parser 1027: }
1028:
1.84 paf 1029: static bool isLeap(int year) {
1.57 parser 1030: return !(
1031: (year % 4) || ((year % 400) && !(year % 100))
1.126 paf 1032: );
1.57 parser 1033: }
1034:
1035: int getMonthDays(int year, int month) {
1036: int monthDays[]={
1.126 paf 1037: 31,
1038: isLeap(year) ? 29 : 28,
1039: 31,
1040: 30,
1041: 31,
1042: 30,
1043: 31,
1044: 31,
1045: 30,
1046: 31,
1047: 30,
1.57 parser 1048: 31
1.126 paf 1049: };
1050: return monthDays[month];
1.41 paf 1051: }
1.69 parser 1052:
1.126 paf 1053: void remove_crlf(char* start, char* end) {
1054: for(char* p=start; p<end; p++)
1.69 parser 1055: switch(*p) {
1.126 paf 1056: case '\n': *p='|'; break;
1057: case '\r': *p=' '; break;
1.69 parser 1058: }
1.91 paf 1059: }
1060:
1061:
1062: /// must be last in this file
1063: #undef vsnprintf
1.126 paf 1064: int __vsnprintf(char* b, size_t s, const char* f, va_list l) {
1.91 paf 1065: if(!s)
1066: return 0;
1067:
1068: int r;
1069: // note: on win32& maybe somewhere else
1070: // vsnprintf do not writes terminating 0 in 'buffer full' case, reducing
1071: --s;
1.172 paf 1072:
1073: // clients do not check for negative 's', feature: ignore such prints
1074: if((ssize_t)s<0)
1075: return 0;
1076:
1.91 paf 1077: #if _MSC_VER
1078: /*
1079: win32:
1080: mk:@MSITStore:C:\Program%20Files\Microsoft%20Visual%20Studio\MSDN\2001APR\1033\vccore.chm::/html/_crt__vsnprintf.2c_._vsnwprintf.htm
1081:
1.154 paf 1082: if the number of bytes to write exceeds buffer, then count bytes are written and Ö1 is returned
1.91 paf 1083: */
1.126 paf 1084: r=_vsnprintf(b, s, f, l);
1.91 paf 1085: if(r<0)
1086: r=s;
1087: #else
1.126 paf 1088: r=vsnprintf(b, s, f, l);
1.91 paf 1089: /*
1090: solaris:
1091: man vsnprintf
1092:
1093: The snprintf() function returns the number of characters
1094: formatted, that is, the number of characters that would have
1095: been written to the buffer if it were large enough. If the
1096: value of n is 0 on a call to snprintf(), an unspecified
1097: value less than 1 is returned.
1098: */
1099:
1100: if(r<0)
1101: r=0;
1.167 paf 1102: else if((size_t)r>s)
1.91 paf 1103: r=s;
1104: #endif
1105: b[r]=0;
1106: return r;
1107: }
1108:
1.126 paf 1109: int __snprintf(char* b, size_t s, const char* f, ...) {
1.91 paf 1110: va_list l;
1.126 paf 1111: va_start(l, f);
1112: int r=__vsnprintf(b, s, f, l);
1113: va_end(l);
1.91 paf 1114: return r;
1.98 paf 1115: }
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