--- parser3/src/main/untaint.C 2001/03/25 09:40:55 1.18 +++ parser3/src/main/untaint.C 2001/04/03 17:01:03 1.31 @@ -5,7 +5,7 @@ Author: Alexander Petrosyan (http://design.ru/paf) - $Id: untaint.C,v 1.18 2001/03/25 09:40:55 paf Exp $ + $Id: untaint.C,v 1.31 2001/04/03 17:01:03 paf Exp $ */ #include "pa_config_includes.h" @@ -43,7 +43,11 @@ inline bool need_file_encode(unsigned ch if((c>='0') &&(c<='9') ||(c>='A') &&(c<='Z') ||(c>='a') &&(c<='z')) return false; - return !strchr("./", c); + return !strchr( +#ifdef WIN32 + ":\\" +#endif + "./", c); } inline bool need_uri_encode(unsigned char c){ if((c>='0') &&(c<='9') ||(c>='A') &&(c<='Z') ||(c>='a') &&(c<='z')) @@ -65,32 +69,36 @@ static bool typo_present(Array::Item *va const char *src=static_cast(info); int partial; - row->get_string(0)->cmp(src, partial); + row->get_string(0)->cmp(partial, src); return partial==0 || // full match partial==1; // typo left column starts 'src' } -/// @todo optimize whitespaces for all but 'html' -char *String::store_to(char *dest) const { +/** + @test optimize whitespaces for all but 'html' + @todo fix theoretical \n mem overrun in TYPO replacements +*/ +char *String::store_to(char *dest, Untaint_lang lang) const { // $MAIN:html-typo table - Table *typo_table=static_cast(pool().tag()); + Table *user_typo_table=static_cast
(pool().tag()); + Table *typo_table=user_typo_table?user_typo_table:default_typo_table; const Chunk *chunk=&head; do { const Chunk::Row *row=chunk->rows; - for(int i=0; icount; i++) { + for(size_t i=0; icount; i++, row++) { if(row==append_here) goto break2; // WARNING: // string can grow only UNTAINT_TIMES_BIGGER - switch(row->item.lang) { - case UL_NO: + switch(lang==UL_UNKNOWN?row->item.lang:lang) { + case UL_CLEAN: // clean piece - case UL_YES: + case UL_TAINTED: // tainted piece, but undefined untaint language - // for VString.get_double of tainted values + // for VString.as_double of tainted values // for ^process{body} evaluation case UL_AS_IS: // tainted, untaint language: as-is @@ -147,82 +155,78 @@ char *String::store_to(char *dest) const case '>': to_string(">", 4); break; case '<': to_string("<", 4); break; case '"': to_string(""", 6); break; - //TODO: XSLT to_string!'\'', "'", 6) + //TODO: XSLT case '\'': to_string("'", 6); break; _default; }); break; case UL_HTML_TYPO: { // tainted, untaint language: html-typo - char *html_for_typo= - (char *)malloc(size()*6/*""" the longest possible*/+1); + char *html_for_typo=(char *)malloc(size()*2/* '\n' -> '\' 'n' */+1); + // note: + // there still is a possibility that user + // would not replace \n as she supposed to + // and rather replace \ and n into huge strings + // thus causing memory overrun + // this can be dealed by allocating *2 memory, but that's too expensive size_t html_for_typo_size; { // local dest char *dest=html_for_typo; escape(switch(*src) { - // BEWARE: check maximum replacement length in malloc above - case '&': to_string("&", 5); break; - case '>': to_string(">", 4); break; - case '<': to_string("<", 4); break; - case '"': to_string(""", 6); break; // convinient name for typo match "\n" case '\r': if(typo_table) { *dest++='\\'; *dest++='n'; // \r -> \n - if(src[1]=='\n') // \r\n -> remove \n - src++; + if(src[1]=='\n') { // \r\n -> remove \n + size--; src++; + } } break; case '\n': if(typo_table) to_string("\\n", 2); break; - //TODO: XSLT to_string!'\'', "'", 6) + //TODO: XSLT case '\'': to_string("'", 6); break; _default; }); *dest=0; html_for_typo_size=dest-html_for_typo; } // typo table replacements - if(typo_table) { - const char *src=html_for_typo; - do { - // there is a row where first column starts 'src' - if(Table::Item *item=typo_table->first_that(typo_present, src)) { - // get a=>b values - const String& a=*static_cast(item)->get_string(0); - const String& b=*static_cast(item)->get_string(1); - // empty 'a' | 'b' checks - if(a.size()==0 || b.size()==0) { - pool().set_tag(0); // avoid recursion - THROW(0, 0, - typo_table->origin_string(), - "typo table column elements must not be empty"); - } - // overflow check: - // b allowed to be max UNTAINT_TIMES_BIGGER then a - if(b.size()>UNTAINT_TIMES_BIGGER*a.size()) { - pool().set_tag(0); // avoid recursion - THROW(0, 0, - &b, - "is %g times longer then '%s', " - "while maximum, handled by Parser, is %d", - ((double)b.size())/a.size(), - a.cstr(), - UNTAINT_TIMES_BIGGER); - } - - // skip 'a' in 'src' - src+=a.size(); - // write 'b' to 'dest' - b.store_to(dest); - dest+=b.size(); - } else - *dest++=*src++; - } while(*src); - } else { - memcpy(dest, html_for_typo, html_for_typo_size); - dest+=html_for_typo_size; - } + const char *src=html_for_typo; + do { + // there is a row where first column starts 'src' + if(Table::Item *item=typo_table->first_that(typo_present, src)) { + // get a=>b values + const String& a=*static_cast(item)->get_string(0); + const String& b=*static_cast(item)->get_string(1); + // empty 'a' | 'b' checks + if(a.size()==0 || b.size()==0) { + pool().set_tag(default_typo_table); // avoid recursion + THROW(0, 0, + typo_table->origin_string(), + "typo table column elements must not be empty"); + } + // overflow check: + // b allowed to be max UNTAINT_TIMES_BIGGER then a + if(b.size()>UNTAINT_TIMES_BIGGER*a.size()) { + pool().set_tag(default_typo_table); // avoid recursion + THROW(0, 0, + &b, + "is %g times longer then '%s', " + "while maximum, handled by Parser, is %d", + ((double)b.size())/a.size(), + a.cstr(), + UNTAINT_TIMES_BIGGER); + } + + // skip 'a' in 'src' + src+=a.size(); + // write 'b' to 'dest' + b.store_to(dest); + dest+=b.size(); + } else + *dest++=*src++; + } while(*src); break; } default: @@ -232,7 +236,6 @@ char *String::store_to(char *dest) const static_cast(row->item.lang), i); } - row++; } chunk=row->link; } while(chunk);