Annotation of parser3/src/main/pa_common.C, revision 1.175
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:
1.175 ! paf 8: static const char * const IDENT_COMMON_C="$Date: 2004/02/11 15:33:15 $";
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) {
1.175 ! paf 145: ulong packed_ip=inet_addr(host);
! 146: struct hostent *hostIP = packed_ip==INADDR_NONE?
! 147: gethostbyname(host)
! 148: : gethostbyaddr((char *)&packed_ip, sizeof(packed_ip), AF_INET);
! 149: if(hostIP)
1.126 paf 150: memcpy(&addr->sin_addr, hostIP->h_addr, hostIP->h_length);
151: else
152: return false;
153: } else
154: addr->sin_addr.s_addr=INADDR_ANY;
155: return true;
156: }
157:
1.171 paf 158: static int http_read_response(char*& response, size_t& response_size, int sock, bool fail_on_status_ne_200){
159: response=(char*)pa_malloc_atomic(1/*terminator*/); // setting memory block type
160: response[(response_size=0)]=0;
1.154 paf 161: int result=0;
1.171 paf 162: char* EOLat=0;
1.126 paf 163: while(true) {
1.171 paf 164: char buf[MAX_STRING*10];
165: ssize_t received_size=recv(sock, buf, sizeof(buf), 0);
166: if(received_size<=0)
1.126 paf 167: break;
1.171 paf 168: response=(char*)pa_realloc(response, response_size+received_size+1/*terminator*/);
169: memcpy(response+response_size, buf, received_size);
170: response_size+=received_size;
171: response[response_size]=0;
172:
173: if(!result && (EOLat=strstr(response, CRLF))) { // checking status in first response
174: const String status_line(pa_strdup(response, EOLat-response));
1.154 paf 175: ArrayString astatus;
176: size_t pos_after=0;
177: status_line.split(astatus, pos_after, " ");
178: const String& status_code=*astatus.get(1);
179: result=status_code.as_int();
1.142 paf 180:
1.154 paf 181: if(fail_on_status_ne_200 && result!=200)
1.142 paf 182: throw Exception("http.status",
1.154 paf 183: &status_code,
1.142 paf 184: "invalid HTTP response status");
185: }
186: }
1.154 paf 187: if(result)
188: return result;
1.142 paf 189: else
190: throw Exception("http.response",
191: 0,
1.173 paf 192: "bad response from host - no status found (size=%u)", response_size);
1.126 paf 193: }
194:
195: /* ********************** request *************************** */
196:
197: #if defined(SIGALRM) && defined(HAVE_SIGSETJMP) && defined(HAVE_SIGLONGJMP)
1.145 paf 198: # define PA_USE_ALARM
1.126 paf 199: #endif
200:
1.145 paf 201: #ifdef PA_USE_ALARM
1.126 paf 202: static sigjmp_buf timeout_env;
203: static void timeout_handler(int sig){
204: siglongjmp(timeout_env, 1);
205: }
206: #endif
207:
1.171 paf 208: static int http_request(char*& response, size_t& response_size,
1.152 paf 209: const char* host, int port,
210: const char* request,
1.166 paf 211: int
212: #ifdef PA_USE_ALARM
213: timeout
214: #endif
215: ,
1.152 paf 216: bool fail_on_status_ne_200) {
1.126 paf 217: if(!host)
218: throw Exception("http.host",
1.154 paf 219: 0,
1.126 paf 220: "zero hostname"); //never
221:
1.145 paf 222: #ifdef PA_USE_ALARM
1.146 paf 223: signal(SIGALRM, timeout_handler);
1.126 paf 224: #endif
225: int sock=-1;
1.145 paf 226: #ifdef PA_USE_ALARM
227: if(sigsetjmp(timeout_env, 1)) {
228: // stupid gcc [2.95.4] generated bad code
229: // which failed to handle sigsetjmp+throw: crashed inside of pre-throw code.
230: // rewritten simplier [though duplicating closesocket code]
231: if(sock>=0)
232: closesocket(sock);
233: throw Exception("http.timeout",
234: origin_string,
235: "timeout occured while retrieving document");
1.146 paf 236: return 0; // never
1.145 paf 237: } else {
238: alarm(timeout);
239: #endif
1.146 paf 240: try {
241: int result;
1.126 paf 242: struct sockaddr_in dest;
1.154 paf 243:
244: if(!set_addr(&dest, host, port))
1.126 paf 245: throw Exception("http.host",
1.154 paf 246: 0,
1.127 paf 247: "can not resolve hostname \"%s\"", host);
1.126 paf 248:
249: if((sock=socket(AF_INET, SOCK_STREAM, IPPROTO_TCP/*0*/))<0)
250: throw Exception("http.connect",
1.154 paf 251: 0,
1.127 paf 252: "can not make socket: %s (%d)", strerror(errno), errno);
1.126 paf 253: if(connect(sock, (struct sockaddr *)&dest, sizeof(dest)))
254: throw Exception("http.connect",
1.154 paf 255: 0,
1.127 paf 256: "can not connect to host \"%s\": %s (%d)", host, strerror(errno), errno);
1.126 paf 257: size_t request_size=strlen(request);
258: if(send(sock, request, request_size, 0)!=(ssize_t)request_size)
259: throw Exception("http.connect",
1.154 paf 260: 0,
1.127 paf 261: "error sending request: %s (%d)", strerror(errno), errno);
1.126 paf 262:
1.171 paf 263: result=http_read_response(response, response_size, sock, fail_on_status_ne_200);
1.142 paf 264: closesocket(sock);
1.145 paf 265: #ifdef PA_USE_ALARM
1.142 paf 266: alarm(0);
1.126 paf 267: #endif
1.147 paf 268: return result;
1.146 paf 269: } catch(...) {
1.145 paf 270: #ifdef PA_USE_ALARM
1.146 paf 271: alarm(0);
1.126 paf 272: #endif
1.146 paf 273: if(sock>=0)
274: closesocket(sock);
1.154 paf 275: rethrow;
1.146 paf 276: }
1.148 paf 277: #ifdef PA_USE_ALARM
1.126 paf 278: }
1.148 paf 279: #endif
1.126 paf 280: }
281:
1.127 paf 282: #ifndef DOXYGEN
283: struct Http_pass_header_info {
1.154 paf 284: Request_charsets* charsets;
1.127 paf 285: String* request;
286: bool user_agent_specified;
287: };
288: #endif
1.154 paf 289: static void http_pass_header(HashStringValue::key_type key,
290: HashStringValue::value_type value,
291: Http_pass_header_info *info) {
292: *info->request <<key<<": "
293: << attributed_meaning_to_string(*value, String::L_HTTP_HEADER, false)
294: << CRLF;
1.135 paf 295:
1.154 paf 296: if(String(key, String::L_TAINTED).change_case(info->charsets->source(), String::CC_UPPER)=="USER-AGENT")
297: info->user_agent_specified=true;
1.126 paf 298: }
1.154 paf 299:
300:
301: static Charset* detect_charset(Charset& source_charset, const String& content_type_value) {
1.156 paf 302: const String::Body CONTENT_TYPE_VALUE=
1.154 paf 303: content_type_value.change_case(source_charset, String::CC_UPPER);
304: // content-type: xxx/xxx; source_charset=WE-NEED-THIS
305: // content-type: xxx/xxx; source_charset="WE-NEED-THIS"
306: // content-type: xxx/xxx; source_charset="WE-NEED-THIS";
307: size_t before_charseteq_pos=CONTENT_TYPE_VALUE.pos("CHARSET=");
308: if(before_charseteq_pos!=STRING_NOT_FOUND) {
309: size_t charset_begin=before_charseteq_pos+8/*CHARSET="*/;
310: size_t open_quote_pos=CONTENT_TYPE_VALUE.pos('"', charset_begin);
311: bool quoted=open_quote_pos==charset_begin;
312: if(quoted)
313: charset_begin++; // skip opening '"'
314: size_t charset_end=CONTENT_TYPE_VALUE.length();
315: if(quoted) {
316: size_t close_quote_pos=CONTENT_TYPE_VALUE.pos('"', charset_begin);
317: if(close_quote_pos!=STRING_NOT_FOUND)
318: charset_end=close_quote_pos;
319: } else {
320: size_t delim_pos=CONTENT_TYPE_VALUE.pos(';', charset_begin);
321: if(delim_pos!=STRING_NOT_FOUND)
322: charset_end=delim_pos;
323: }
1.156 paf 324: const String::Body CHARSET_NAME_BODY=
1.154 paf 325: CONTENT_TYPE_VALUE.mid(charset_begin, charset_end);
326:
327: return &charsets.get(CHARSET_NAME_BODY);
328: }
329:
330: return 0;
331: }
332:
333:
334: #ifndef DOXYGEN
335: struct File_read_http_result {
336: char *str; size_t length;
337: HashStringValue* headers;
338: };
339: #endif
1.155 paf 340: /// @todo build .cookies field. use ^file.tables.SET-COOKIES.menu{ for now
1.154 paf 341: static File_read_http_result file_read_http(Request_charsets& charsets,
342: const String& file_spec,
1.169 paf 343: bool as_text,
344: HashStringValue *options=0) {
1.154 paf 345: File_read_http_result result;
1.126 paf 346: char host[MAX_STRING];
1.129 paf 347: const char* uri;
1.126 paf 348: int port;
1.154 paf 349: const char* method=0;
1.126 paf 350: int timeout=2;
1.142 paf 351: bool fail_on_status_ne_200=true;
1.154 paf 352: Value* vheaders=0;
353: Charset *asked_remote_charset=0;
1.126 paf 354:
1.127 paf 355: if(options) {
356: int valid_options=0;
1.154 paf 357: if(Value* vmethod=options->get(http_method_name)) {
1.127 paf 358: valid_options++;
1.154 paf 359: method=vmethod->as_string().cstr();
1.127 paf 360: }
1.154 paf 361: if(Value* vtimeout=options->get(http_timeout_name)) {
1.127 paf 362: valid_options++;
363: timeout=vtimeout->as_int();
364: }
1.167 paf 365: if((vheaders=options->get(http_headers_name))) {
1.127 paf 366: valid_options++;
367: }
1.154 paf 368: if(Value* vany_status=options->get(http_any_status_name)) {
1.142 paf 369: valid_options++;
370: fail_on_status_ne_200=!vany_status->as_bool();
371: }
1.154 paf 372: if(Value* vcharset_name=options->get(http_charset_name)) {
373: valid_options++;
374: asked_remote_charset=&::charsets.get(vcharset_name->as_string().
375: change_case(charsets.source(), String::CC_UPPER));
376: }
1.142 paf 377:
1.154 paf 378: if(valid_options!=options->count())
1.127 paf 379: throw Exception("parser.runtime",
380: 0,
381: "invalid option passed");
1.154 paf 382: }
383: if(!asked_remote_charset) // defaulting to $request:charset
384: asked_remote_charset=&charsets.source();
385:
386: //preparing request
387: String& connect_string=*new String;
388: // not in ^sql{... L_SQL ...} spirit, but closer to ^file::load one
389: connect_string.append(file_spec, String::L_URI); // tainted pieces -> URI pieces
390:
391: const char* request_cstr;
392: {
393: // influence URLencoding of tainted pieces to String::L_URI lang
394: Temp_client_charset temp(charsets, *asked_remote_charset);
395:
396: const char* connect_string_cstr=connect_string.cstr(String::L_UNSPECIFIED);
1.126 paf 397:
1.154 paf 398: const char* current=connect_string_cstr;
399: if(strncmp(current, "http://", 7)!=0)
400: throw Exception(0,
401: &connect_string,
402: "does not start with http://"); //never
403: current+=7;
404:
405: strncpy(host, current, sizeof(host)-1); host[sizeof(host)-1]=0;
406: char* host_uri=lsplit(host, '/');
407: uri=host_uri?current+(host_uri-1-host):"/";
408: char* port_cstr=lsplit(host, ':');
409: char* error_pos=0;
410: port=port_cstr?strtol(port_cstr, &error_pos, 0):80;
411:
412:
413: //making request
414: String request;
415: if(method)
416: request<<method;
417: else
418: request<<"GET";
419: request<< " "<< uri <<" HTTP/1.0" CRLF
420: "host: "<< host << CRLF;
421: bool user_agent_specified=false;
422: if(vheaders && !vheaders->is_string()) { // allow empty
423: if(HashStringValue *headers=vheaders->get_hash()) {
1.167 paf 424: Http_pass_header_info info={&charsets, &request, false};
1.154 paf 425: headers->for_each(http_pass_header, &info);
426: user_agent_specified=info.user_agent_specified;
427: } else
428: throw Exception("parser.runtime",
429: &connect_string,
430: "headers param must be hash");
431: };
432: if(!user_agent_specified) // defaulting
433: request << "user-agent: " DEFAULT_USER_AGENT CRLF;
434: request << CRLF;
435:
436: request_cstr=request.cstr(String::L_UNSPECIFIED);
437: }
438: // recode those pieces which are not in String::L_URI lang
439: // [those violating HTTP standard, but widly used]
440: request_cstr=Charset::transcode(
441: String::C(request_cstr, strlen(request_cstr)),
442: charsets.source(),
443: *asked_remote_charset);
1.126 paf 444:
445: //sending request
1.171 paf 446: char* response;
447: size_t response_size;
448: int status_code=http_request(response, response_size,
1.154 paf 449: host, port, request_cstr,
1.142 paf 450: timeout, fail_on_status_ne_200);
1.126 paf 451:
452: //processing results
1.171 paf 453: char* raw_body; size_t raw_body_size;
454: char* headers_end_at=strstr(response, CRLF CRLF /*change '4' below along!*/);
455: if(headers_end_at) {
456: raw_body=headers_end_at+4;
457: raw_body_size=response_size-(raw_body-response);
458: } else
1.126 paf 459: throw Exception("http.response",
1.133 paf 460: &connect_string,
1.126 paf 461: "bad response from host - no headers found");
1.171 paf 462:
463: *headers_end_at=0;
464: const String header_block(response, headers_end_at-response, true);
1.126 paf 465:
1.154 paf 466: ArrayString aheaders;
467: result.headers=new HashStringValue;
1.155 paf 468: VHash* vtables=new VHash;
469: result.headers->put(http_tables_name, vtables);
470: HashStringValue& tables=vtables->hash();
471:
1.154 paf 472: size_t pos_after=0;
473: header_block.split(aheaders, pos_after, CRLF);
1.126 paf 474:
475: //processing headers
1.154 paf 476: size_t aheaders_count=aheaders.count();
477: Charset* real_remote_charset=0; // undetected, yet
478: for(size_t i=1; i<aheaders_count; i++) {
479: const String& line=*aheaders.get(i);
1.171 paf 480: size_t pos=line.pos(": ", 2);
1.154 paf 481: if(pos==STRING_NOT_FOUND || pos<1)
1.126 paf 482: throw Exception("http.response",
1.154 paf 483: &connect_string,
484: "bad response from host - bad header \"%s\"", line.cstr());
1.156 paf 485: const String::Body HEADER_NAME=
1.154 paf 486: line.mid(0, pos).change_case(charsets.source(), String::CC_UPPER);
487: const String& header_value=line.mid(pos+2, line.length());
488: if(HEADER_NAME=="CONTENT-TYPE")
489: real_remote_charset=detect_charset(charsets.source(), header_value);
1.155 paf 490:
491: // tables
492: {
493: Value *valready=(Value *)tables.get(HEADER_NAME);
494: bool existed=valready!=0;
495: Table *table;
496: if(existed) {
497: // second+ appearence
498: table=valready->get_table();
499: } else {
500: // first appearence
501: Table::columns_type columns =new ArrayString(1);
502: *columns+=new String("value");
503: table=new Table(columns);
504: }
505: // this string becomes next row
506: ArrayString& row=*new ArrayString(1);
507: row+=&header_value;
508: *table+=&row;
509: // not existed before? add it
510: if(!existed)
511: tables.put(HEADER_NAME, new VTable(table));
512: }
513:
1.154 paf 514: result.headers->put(HEADER_NAME, new VString(header_value));
1.126 paf 515: }
1.154 paf 516: // defaulting to used-asked charset [it's never empty!]
517: if(!real_remote_charset)
518: real_remote_charset=asked_remote_charset;
1.126 paf 519:
520: // output response
1.171 paf 521: String::C real_body=String::C(raw_body, raw_body_size);
522: 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 523: real_body=Charset::transcode(real_body, *real_remote_charset, charsets.source());
1.154 paf 524:
525: result.str=const_cast<char *>(real_body.str); // hacking a little
526: result.length=real_body.length;
527: result.headers->put(file_status_name, new VInt(status_code));
528: return result;
1.34 paf 529: }
1.123 paf 530:
1.154 paf 531: #endif
532:
1.123 paf 533: #ifndef DOXYGEN
534: struct File_read_action_info {
1.154 paf 535: char **data; size_t *data_size;
1.126 paf 536: };
1.123 paf 537: #endif
1.154 paf 538: static void file_read_action(
539: struct stat& finfo,
540: int f,
1.166 paf 541: const String& file_spec, const char* /*fname*/, bool as_text,
1.154 paf 542: void *context) {
1.126 paf 543: File_read_action_info& info=*static_cast<File_read_action_info *>(context);
1.123 paf 544: if(size_t to_read_size=(size_t)finfo.st_size) {
1.154 paf 545: *info.data=new(PointerFreeGC) char[to_read_size+(as_text?1:0)];
1.126 paf 546: *info.data_size=(size_t)read(f, *info.data, to_read_size);
1.123 paf 547:
548: if(ssize_t(*info.data_size)<0 || *info.data_size>to_read_size)
1.126 paf 549: throw Exception(0,
1.123 paf 550: &file_spec,
1.173 paf 551: "read failed: actually read %u bytes count not in [0..%u] valid range",
1.126 paf 552: *info.data_size, to_read_size);
1.123 paf 553: } else { // empty file
554: if(as_text) {
1.154 paf 555: *info.data=new(PointerFreeGC) char[1];
1.123 paf 556: *(char*)(*info.data)=0;
557: } else
558: *info.data=0;
559: *info.data_size=0;
560: return;
561: }
1.126 paf 562: }
1.154 paf 563: File_read_result file_read(Request_charsets& charsets, const String& file_spec,
564: bool as_text, HashStringValue *params,
1.126 paf 565: bool fail_on_read_problem) {
1.167 paf 566: File_read_result result={false, 0, 0, 0};
1.154 paf 567: #ifdef PA_HTTP
568: if(file_spec.starts_with("http://")) {
1.126 paf 569: // fail on read problem
1.169 paf 570: File_read_http_result http=file_read_http(charsets, file_spec, as_text, params);
1.154 paf 571: result.success=true;
572: result.str=http.str;
573: result.length=http.length;
574: result.headers=http.headers;
1.126 paf 575: } else {
1.154 paf 576: #endif
1.161 paf 577: if(params && params->count())
578: throw Exception("parser.runtime",
579: 0,
580: "invalid option passed");
581:
1.154 paf 582: File_read_action_info info={&result.str, &result.length};
583: result.success=file_read_action_under_lock(file_spec,
1.126 paf 584: "read", file_read_action, &info,
585: as_text, fail_on_read_problem);
1.154 paf 586: #ifdef PA_HTTP
1.126 paf 587: }
1.154 paf 588: #endif
1.123 paf 589:
1.154 paf 590: if(result.success && as_text) {
1.131 paf 591: // UTF-8 signature: EF BB BF
1.154 paf 592: if(result.length>=3) {
593: char *in=(char *)result.str;
1.159 paf 594: if(strncmp(in, "\xEF\xBB\xBF", 3)==0) {
1.154 paf 595: result.str=in+3; result.length-=3;// skip prefix
1.131 paf 596: }
597: }
598:
1.154 paf 599: fix_line_breaks((char *)(result.str), result.length);
1.123 paf 600: }
1.126 paf 601:
602: return result;
1.123 paf 603: }
604:
1.154 paf 605: #ifdef PA_SAFE_MODE
606: void check_safe_mode(struct stat finfo, const String& file_spec, const char* fname) {
607: if(finfo.st_uid/*foreign?*/!=geteuid()
608: && finfo.st_gid/*foreign?*/!=getegid())
609: throw Exception("parser.runtime",
610: &file_spec,
611: "parser is in safe mode: "
612: "reading files of foreign group and user disabled "
613: "[recompile parser with --disable-safe-mode configure option], "
614: "actual filename '%s', "
615: "fuid(%d)!=euid(%d) or fgid(%d)!=egid(%d)",
616: fname,
617: finfo.st_uid, geteuid(),
618: finfo.st_gid, getegid());
619: }
620: #endif
1.149 paf 621:
1.154 paf 622: bool file_read_action_under_lock(const String& file_spec,
1.126 paf 623: const char* action_name, File_read_action action, void *context,
624: bool as_text,
1.123 paf 625: bool fail_on_read_problem) {
1.154 paf 626: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.33 paf 627: int f;
628:
629: // first open, next stat:
1.45 paf 630: // directory update of NTFS hard links performed on open.
1.33 paf 631: // ex:
632: // a.html:^test[] and b.html hardlink to a.html
633: // user inserts ! before ^test in a.html
1.126 paf 634: // directory entry of b.html in NTFS not updated at once,
1.35 paf 635: // they delay update till open, so we would receive "!^test[" string
636: // if would do stat, next open.
1.123 paf 637: // later: it seems, even this does not help sometimes
1.98 paf 638: if((f=open(fname, O_RDONLY|(as_text?_O_TEXT:_O_BINARY)))>=0) {
1.123 paf 639: try {
1.162 paf 640: if(pa_lock_shared_blocking(f)!=0)
1.126 paf 641: throw Exception("file.lock",
1.123 paf 642: &file_spec,
643: "shared lock failed: %s (%d), actual filename '%s'",
1.154 paf 644: strerror(errno), errno, fname);
1.123 paf 645:
1.124 paf 646: struct stat finfo;
647: if(stat(fname, &finfo)!=0)
648: throw Exception("file.missing", // hardly possible: we just opened it OK
649: &file_spec,
650: "stat failed: %s (%d), actual filename '%s'",
1.154 paf 651: strerror(errno), errno, fname);
1.124 paf 652:
1.140 paf 653: #ifdef PA_SAFE_MODE
1.149 paf 654: check_safe_mode(finfo, file_spec, fname);
1.105 paf 655: #endif
1.32 paf 656:
1.154 paf 657: action(finfo, f, file_spec, fname, as_text, context);
1.123 paf 658: } catch(...) {
1.162 paf 659: pa_unlock(f);close(f);
1.123 paf 660: if(fail_on_read_problem)
1.154 paf 661: rethrow;
1.123 paf 662: return false;
663: }
1.87 paf 664:
1.162 paf 665: pa_unlock(f);close(f);
1.72 parser 666: return true;
1.118 paf 667: } else {
668: if(fail_on_read_problem)
1.126 paf 669: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.118 paf 670: &file_spec,
1.123 paf 671: "%s failed: %s (%d), actual filename '%s'",
1.154 paf 672: action_name, strerror(errno), errno, fname);
1.118 paf 673: return false;
674: }
1.8 paf 675: }
676:
1.63 parser 677: static void create_dir_for_file(const String& file_spec) {
678: size_t pos_after=1;
1.154 paf 679: size_t pos_before;
680: while((pos_before=file_spec.pos('/', pos_after))!=STRING_NOT_FOUND) {
681: mkdir(file_spec.mid(0, pos_before).cstr(String::L_FILE_SPEC), 0775);
1.63 parser 682: pos_after=pos_before+1;
683: }
684: }
685:
1.98 paf 686: bool file_write_action_under_lock(
1.28 paf 687: const String& file_spec,
1.126 paf 688: const char* action_name, File_write_action action, void *context,
689: bool as_text,
690: bool do_append,
691: bool do_block,
1.110 paf 692: bool fail_on_lock_problem) {
1.154 paf 693: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.28 paf 694: int f;
1.80 paf 695: if(access(fname, W_OK)!=0) // no
1.126 paf 696: create_dir_for_file(file_spec);
1.50 paf 697:
1.80 paf 698: if((f=open(fname,
699: O_CREAT|O_RDWR
700: |(as_text?_O_TEXT:_O_BINARY)
1.138 paf 701: |(do_append?O_APPEND:PA_O_TRUNC), 0664))>=0) {
1.162 paf 702: if((do_block?pa_lock_exclusive_blocking(f):pa_lock_exclusive_nonblocking(f))!=0) {
1.126 paf 703: Exception e("file.lock",
1.110 paf 704: &file_spec,
705: "shared lock failed: %s (%d), actual filename '%s'",
1.154 paf 706: strerror(errno), errno, fname);
1.126 paf 707: close(f);
1.110 paf 708: if(fail_on_lock_problem)
709: throw e;
1.98 paf 710: return false;
711: }
1.96 paf 712:
1.158 paf 713: try {
1.126 paf 714: action(f, context);
1.158 paf 715: } catch(...) {
1.138 paf 716: #ifdef HAVE_FTRUNCATE
1.104 paf 717: if(!do_append)
1.125 paf 718: ftruncate(f, lseek(f, 0, SEEK_CUR)); // one can not use O_TRUNC, read lower
1.138 paf 719: #endif
1.162 paf 720: pa_unlock(f);close(f);
1.154 paf 721: rethrow;
1.158 paf 722: }
1.80 paf 723:
1.138 paf 724: #ifdef HAVE_FTRUNCATE
1.104 paf 725: if(!do_append)
1.125 paf 726: 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 727: #endif
1.162 paf 728: pa_unlock(f);close(f);
1.98 paf 729: return true;
1.80 paf 730: } else
1.126 paf 731: throw Exception(errno==EACCES?"file.access":0,
1.80 paf 732: &file_spec,
1.96 paf 733: "%s failed: %s (%d), actual filename '%s'",
1.154 paf 734: action_name, strerror(errno), errno, fname);
1.96 paf 735: // here should be nothing, see rethrow above
736: }
737:
738: #ifndef DOXYGEN
739: struct File_write_action_info {
1.154 paf 740: const char* str; size_t length;
1.126 paf 741: };
1.96 paf 742: #endif
743: static void file_write_action(int f, void *context) {
1.126 paf 744: File_write_action_info& info=*static_cast<File_write_action_info *>(context);
1.154 paf 745: if(info.length) {
746: int written=write(f, info.str, info.length);
1.116 paf 747: if(written<0)
1.126 paf 748: throw Exception(0,
749: 0,
750: "write failed: %s (%d)", strerror(errno), errno);
1.113 paf 751: }
1.96 paf 752: }
753: void file_write(
754: const String& file_spec,
1.154 paf 755: const char* data, size_t size,
1.126 paf 756: bool as_text,
1.96 paf 757: bool do_append) {
1.126 paf 758: File_write_action_info info={data, size};
1.98 paf 759: file_write_action_under_lock(
1.154 paf 760: file_spec,
761: "write", file_write_action, &info,
762: as_text,
763: do_append);
1.30 paf 764: }
765:
1.63 parser 766: // throws nothing! [this is required in file_move & file_delete]
1.50 paf 767: static void rmdir(const String& file_spec, size_t pos_after) {
1.154 paf 768: size_t pos_before;
769: if((pos_before=file_spec.pos('/', pos_after))!=STRING_NOT_FOUND)
1.126 paf 770: rmdir(file_spec, pos_before+1);
1.50 paf 771:
1.154 paf 772: rmdir(file_spec.mid(0, pos_after-1/* / */).cstr(String::L_FILE_SPEC));
1.50 paf 773: }
1.164 paf 774: bool file_delete(const String& file_spec, bool fail_on_problem) {
1.154 paf 775: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.54 parser 776: if(unlink(fname)!=0)
1.164 paf 777: if(fail_on_problem)
1.126 paf 778: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.93 paf 779: &file_spec,
780: "unlink failed: %s (%d), actual filename '%s'",
1.154 paf 781: strerror(errno), errno, fname);
1.93 paf 782: else
783: return false;
1.50 paf 784:
1.126 paf 785: rmdir(file_spec, 1);
1.93 paf 786: return true;
1.60 parser 787: }
1.95 paf 788: void file_move(const String& old_spec, const String& new_spec) {
1.154 paf 789: const char* old_spec_cstr=old_spec.cstr(String::L_FILE_SPEC);
790: const char* new_spec_cstr=new_spec.cstr(String::L_FILE_SPEC);
1.63 parser 791:
1.126 paf 792: create_dir_for_file(new_spec);
1.63 parser 793:
1.60 parser 794: if(rename(old_spec_cstr, new_spec_cstr)!=0)
1.126 paf 795: throw Exception(errno==EACCES?"file.access":errno==ENOENT?"file.missing":0,
1.60 parser 796: &old_spec,
797: "rename failed: %s (%d), actual filename '%s' to '%s'",
1.154 paf 798: strerror(errno), errno, old_spec_cstr, new_spec_cstr);
1.63 parser 799:
1.126 paf 800: rmdir(old_spec, 1);
1.31 paf 801: }
802:
1.51 paf 803:
1.126 paf 804: bool entry_exists(const char* fname, struct stat *afinfo) {
1.118 paf 805: struct stat lfinfo;
806: bool result=stat(fname, &lfinfo)==0;
807: if(afinfo)
808: *afinfo=lfinfo;
809: return result;
1.119 paf 810: }
811:
812: bool entry_exists(const String& file_spec) {
1.154 paf 813: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
1.126 paf 814: return entry_exists(fname, 0);
1.118 paf 815: }
816:
1.51 paf 817: static bool entry_readable(const String& file_spec, bool need_dir) {
1.154 paf 818: char* fname=file_spec.cstrm(String::L_FILE_SPEC);
1.120 paf 819: if(need_dir) {
1.126 paf 820: size_t size=strlen(fname);
1.120 paf 821: while(size) {
1.126 paf 822: char c=fname[size-1];
1.120 paf 823: if(c=='/' || c=='\\')
824: fname[--size]=0;
825: else
826: break;
827: }
828: }
1.51 paf 829: struct stat finfo;
1.118 paf 830: if(access(fname, R_OK)==0 && entry_exists(fname, &finfo)) {
1.109 paf 831: bool is_dir=(finfo.st_mode&S_IFDIR) != 0;
1.51 paf 832: return is_dir==need_dir;
833: }
834: return false;
835: }
1.31 paf 836: bool file_readable(const String& file_spec) {
1.126 paf 837: return entry_readable(file_spec, false);
1.51 paf 838: }
839: bool dir_readable(const String& file_spec) {
1.126 paf 840: return entry_readable(file_spec, true);
1.65 parser 841: }
1.154 paf 842: const String* file_readable(const String& path, const String& name) {
843: String& result=*new String(path);
844: result << "/";
845: result << name;
846: return file_readable(result)?&result:0;
1.43 paf 847: }
848: bool file_executable(const String& file_spec) {
1.154 paf 849: return access(file_spec.cstr(String::L_FILE_SPEC), X_OK)==0;
1.44 paf 850: }
851:
1.64 parser 852: bool file_stat(const String& file_spec,
1.58 parser 853: size_t& rsize,
1.126 paf 854: time_t& ratime,
855: time_t& rmtime,
856: time_t& rctime,
1.64 parser 857: bool fail_on_read_problem) {
1.154 paf 858: const char* fname=file_spec.cstr(String::L_FILE_SPEC);
859: struct stat finfo;
1.44 paf 860: if(stat(fname, &finfo)!=0)
1.64 parser 861: if(fail_on_read_problem)
1.126 paf 862: throw Exception("file.missing",
1.67 parser 863: &file_spec,
864: "getting file size failed: %s (%d), real filename '%s'",
1.154 paf 865: strerror(errno), errno, fname);
1.64 parser 866: else
867: return false;
1.58 parser 868: rsize=finfo.st_size;
869: ratime=finfo.st_atime;
870: rmtime=finfo.st_mtime;
871: rctime=finfo.st_ctime;
1.64 parser 872: return true;
1.18 paf 873: }
874:
1.126 paf 875: char* getrow(char* *row_ref, char delim) {
876: char* result=*row_ref;
1.8 paf 877: if(result) {
1.126 paf 878: *row_ref=strchr(result, delim);
1.8 paf 879: if(*row_ref)
880: *((*row_ref)++)=0;
881: else if(!*result)
882: return 0;
883: }
884: return result;
885: }
886:
1.126 paf 887: char* lsplit(char* string, char delim) {
1.23 paf 888: if(string) {
1.126 paf 889: char* v=strchr(string, delim);
1.8 paf 890: if(v) {
891: *v=0;
892: return v+1;
893: }
894: }
895: return 0;
896: }
897:
1.126 paf 898: char* lsplit(char* *string_ref, char delim) {
899: char* result=*string_ref;
900: char* next=lsplit(*string_ref, delim);
1.8 paf 901: *string_ref=next;
902: return result;
1.9 paf 903: }
904:
1.126 paf 905: char* rsplit(char* string, char delim) {
1.18 paf 906: if(string) {
1.126 paf 907: char* v=strrchr(string, delim);
1.18 paf 908: if(v) {
1.9 paf 909: *v=0;
910: return v+1;
911: }
912: }
913: return NULL;
1.10 paf 914: }
915:
1.37 paf 916: /// @todo less stupid type detection
1.154 paf 917: const char* format(double value, char* fmt) {
1.126 paf 918: char local_buf[MAX_NUMBER];
1.108 paf 919: size_t size;
920:
1.10 paf 921: if(fmt)
922: if(strpbrk(fmt, "diouxX"))
923: if(strpbrk(fmt, "ouxX"))
1.126 paf 924: size=snprintf(local_buf, sizeof(local_buf), fmt, (uint)value);
1.10 paf 925: else
1.126 paf 926: size=snprintf(local_buf, sizeof(local_buf), fmt, (int)value);
1.10 paf 927: else
1.126 paf 928: size=snprintf(local_buf, sizeof(local_buf), fmt, value);
1.10 paf 929: else
1.126 paf 930: size=snprintf(local_buf, sizeof(local_buf), "%d", (int)value);
1.10 paf 931:
1.154 paf 932: return pa_strdup(local_buf, size);
1.12 paf 933: }
934:
1.36 paf 935: size_t stdout_write(const void *buf, size_t size) {
1.12 paf 936: #ifdef WIN32
937: do{
1.154 paf 938: int chunk_written=fwrite(buf, 1, min((size_t)8*0x400, size), stdout);
1.12 paf 939: if(chunk_written<=0)
940: break;
941: size-=chunk_written;
1.36 paf 942: buf=((const char*)buf)+chunk_written;
1.126 paf 943: } while(size>0);
1.12 paf 944:
945: return size;
946: #else
1.126 paf 947: return fwrite(buf, 1, size, stdout);
1.12 paf 948: #endif
1.2 paf 949: }
1.14 paf 950:
1.154 paf 951: char* unescape_chars(const char* cp, int len) {
952: char* s=new(PointerFreeGC) char[len + 1];
1.14 paf 953: enum EscapeState {
1.33 paf 954: EscapeRest,
955: EscapeFirst,
1.14 paf 956: EscapeSecond
957: } escapeState=EscapeRest;
958: int escapedValue=0;
959: int srcPos=0;
960: int dstPos=0;
961: while(srcPos < len) {
1.126 paf 962: int ch=cp[srcPos];
1.14 paf 963: switch(escapeState) {
964: case EscapeRest:
965: if(ch=='%') {
966: escapeState=EscapeFirst;
967: } else if(ch=='+') {
1.126 paf 968: s[dstPos++]=' ';
1.14 paf 969: } else {
970: s[dstPos++]=ch;
971: }
972: break;
973: case EscapeFirst:
974: escapedValue=hex_value[ch] << 4;
975: escapeState=EscapeSecond;
976: break;
977: case EscapeSecond:
1.126 paf 978: escapedValue +=hex_value[ch];
1.14 paf 979: s[dstPos++]=escapedValue;
980: escapeState=EscapeRest;
981: break;
982: }
1.126 paf 983: srcPos++;
1.14 paf 984: }
985: s[dstPos]=0;
986: return s;
1.24 paf 987: }
988:
989: #ifdef WIN32
1.126 paf 990: void back_slashes_to_slashes(char* s) {
1.24 paf 991: if(s)
992: for(; *s; s++)
993: if(*s=='\\')
1.126 paf 994: *s='/';
1.24 paf 995: }
1.42 paf 996: /*
1.126 paf 997: void slashes_to_back_slashes(char* s) {
1.42 paf 998: if(s)
999: for(; *s; s++)
1000: if(*s=='/')
1.126 paf 1001: *s='\\';
1.42 paf 1002: }
1003: */
1.24 paf 1004: #endif
1.41 paf 1005:
1.126 paf 1006: bool StrEqNc(const char* s1, const char* s2, bool strict) {
1.41 paf 1007: while(true) {
1008: if(!(*s1)) {
1009: if(!(*s2))
1010: return true;
1011: else
1012: return !strict;
1013: } else if(!(*s2))
1014: return !strict;
1015: if(isalpha(*s1)) {
1016: if(tolower(*s1) !=tolower(*s2))
1017: return false;
1018: } else if((*s1) !=(*s2))
1019: return false;
1.126 paf 1020: s1++;
1021: s2++;
1.41 paf 1022: }
1.57 parser 1023: }
1024:
1.84 paf 1025: static bool isLeap(int year) {
1.57 parser 1026: return !(
1027: (year % 4) || ((year % 400) && !(year % 100))
1.126 paf 1028: );
1.57 parser 1029: }
1030:
1031: int getMonthDays(int year, int month) {
1032: int monthDays[]={
1.126 paf 1033: 31,
1034: isLeap(year) ? 29 : 28,
1035: 31,
1036: 30,
1037: 31,
1038: 30,
1039: 31,
1040: 31,
1041: 30,
1042: 31,
1043: 30,
1.57 parser 1044: 31
1.126 paf 1045: };
1046: return monthDays[month];
1.41 paf 1047: }
1.69 parser 1048:
1.126 paf 1049: void remove_crlf(char* start, char* end) {
1050: for(char* p=start; p<end; p++)
1.69 parser 1051: switch(*p) {
1.126 paf 1052: case '\n': *p='|'; break;
1053: case '\r': *p=' '; break;
1.69 parser 1054: }
1.91 paf 1055: }
1056:
1057:
1058: /// must be last in this file
1059: #undef vsnprintf
1.126 paf 1060: int __vsnprintf(char* b, size_t s, const char* f, va_list l) {
1.91 paf 1061: if(!s)
1062: return 0;
1063:
1064: int r;
1065: // note: on win32& maybe somewhere else
1066: // vsnprintf do not writes terminating 0 in 'buffer full' case, reducing
1067: --s;
1.172 paf 1068:
1069: // clients do not check for negative 's', feature: ignore such prints
1070: if((ssize_t)s<0)
1071: return 0;
1072:
1.91 paf 1073: #if _MSC_VER
1074: /*
1075: win32:
1076: mk:@MSITStore:C:\Program%20Files\Microsoft%20Visual%20Studio\MSDN\2001APR\1033\vccore.chm::/html/_crt__vsnprintf.2c_._vsnwprintf.htm
1077:
1.154 paf 1078: if the number of bytes to write exceeds buffer, then count bytes are written and Ö1 is returned
1.91 paf 1079: */
1.126 paf 1080: r=_vsnprintf(b, s, f, l);
1.91 paf 1081: if(r<0)
1082: r=s;
1083: #else
1.126 paf 1084: r=vsnprintf(b, s, f, l);
1.91 paf 1085: /*
1086: solaris:
1087: man vsnprintf
1088:
1089: The snprintf() function returns the number of characters
1090: formatted, that is, the number of characters that would have
1091: been written to the buffer if it were large enough. If the
1092: value of n is 0 on a call to snprintf(), an unspecified
1093: value less than 1 is returned.
1094: */
1095:
1096: if(r<0)
1097: r=0;
1.167 paf 1098: else if((size_t)r>s)
1.91 paf 1099: r=s;
1100: #endif
1101: b[r]=0;
1102: return r;
1103: }
1104:
1.126 paf 1105: int __snprintf(char* b, size_t s, const char* f, ...) {
1.91 paf 1106: va_list l;
1.126 paf 1107: va_start(l, f);
1108: int r=__vsnprintf(b, s, f, l);
1109: va_end(l);
1.91 paf 1110: return r;
1.98 paf 1111: }
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