From a175314c3e5827eb193872241446f2f8f5c9d33c Mon Sep 17 00:00:00 2001 From: Daniel Baumann Date: Sat, 4 May 2024 20:07:14 +0200 Subject: Adding upstream version 1:10.5.12. Signed-off-by: Daniel Baumann --- sql/gen_lex_hash.cc | 482 ++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 482 insertions(+) create mode 100644 sql/gen_lex_hash.cc (limited to 'sql/gen_lex_hash.cc') diff --git a/sql/gen_lex_hash.cc b/sql/gen_lex_hash.cc new file mode 100644 index 00000000..98e6205e --- /dev/null +++ b/sql/gen_lex_hash.cc @@ -0,0 +1,482 @@ +/* + Copyright (c) 2000, 2012, Oracle and/or its affiliates. + + This program is free software; you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation; version 2 of the License. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License + along with this program; if not, write to the Free Software + Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1335 USA */ + +/** + @file + + @details +@verbatim +The idea of presented algorithm see in +"The Art of Computer Programming" by Donald E. Knuth +Volume 3 "Sorting and searching" +(chapter 6.3 "Digital searching" - name and number of chapter + is back translation from Russian edition :)) + +as illustration of data structures, imagine next table: + +static SYMBOL symbols[] = { + { "ADD", SYM(ADD),0,0}, + { "AND", SYM(AND),0,0}, + { "DAY", SYM(DAY_SYM),0,0}, +}; + +for this structure, presented program generate next searching-structure: + ++-----------+-+-+-+ +| len |1|2|3| ++-----------+-+-+-+ +|first_char |0|0|a| +|last_char |0|0|d| +|link |0|0|+| + | + V + +----------+-+-+-+--+ + | 1 char|a|b|c|d | + +----------+-+-+-+--+ + |first_char|b|0|0|0 | + |last_char |n|0|0|-1| + |link |+|0|0|+ | + | | + | V + | symbols[2] ( "DAY" ) + V ++----------+--+-+-+-+-+-+-+-+-+-+--+ +| 2 char|d |e|f|j|h|i|j|k|l|m|n | ++----------+--+-+-+-+-+-+-+-+-+-+--+ +|first_char|0 |0|0|0|0|0|0|0|0|0|0 | +|last_char |-1|0|0|0|0|0|0|0|0|0|-1| +|link |+ |0|0|0|0|0|0|0|0|0|+ | + | | + V V + symbols[0] ( "ADD" ) symbols[1] ( "AND" ) + +for optimization, link is the 16-bit index in 'symbols' or 'sql_functions' +or search-array.. + +So, we can read full search-structure as 32-bit word +@endverbatim + +@todo + use instead to_upper_lex, special array + (substitute chars) without skip codes.. +@todo + try use reverse order of comparing.. + +*/ + +#define NO_YACC_SYMBOLS +#undef CHECK_UNLIKELY +#include "mariadb.h" +#include "mysql_version.h" +#include "lex.h" +#include + +#include /* ORACLE_WELCOME_COPYRIGHT_NOTICE */ + +struct hash_lex_struct +{ + int first_char; + char last_char; + union{ + hash_lex_struct *char_tails; + int iresult; + }; + int ithis; +}; + +hash_lex_struct *get_hash_struct_by_len(hash_lex_struct **root_by_len, + int len, int *max_len) +{ + if (*max_lenfirst_char= 0; + *max_len= len; + } + return (*root_by_len)+(len-1); +} + +void insert_into_hash(hash_lex_struct *root, const char *name, + int len_from_begin, int index, int function) +{ + hash_lex_struct *end, *cur, *tails; + + if (!root->first_char) + { + root->first_char= -1; + root->iresult= index; + return; + } + + if (root->first_char == -1) + { + int index2= root->iresult; + const char *name2= (index2 < 0 ? sql_functions[-index2-1] : + symbols[index2]).name + len_from_begin; + root->first_char= (int) (uchar) name2[0]; + root->last_char= (char) root->first_char; + tails= (hash_lex_struct*)malloc(sizeof(hash_lex_struct)); + root->char_tails= tails; + tails->first_char= -1; + tails->iresult= index2; + } + + size_t real_size= (root->last_char-root->first_char+1); + + if (root->first_char>(*name)) + { + size_t new_size= root->last_char-(*name)+1; + if (unlikely(new_sizechar_tails; + tails= (hash_lex_struct*)realloc((char*)tails, + sizeof(hash_lex_struct)*new_size); + root->char_tails= tails; + memmove(tails+(new_size-real_size),tails,real_size*sizeof(hash_lex_struct)); + end= tails + new_size - real_size; + for (cur= tails; curfirst_char= 0; + root->first_char= (int) (uchar) *name; + } + + if (root->last_char<(*name)) + { + size_t new_size= (*name)-root->first_char+1; + if (unlikely(new_sizechar_tails; + tails= (hash_lex_struct*)realloc((char*)tails, + sizeof(hash_lex_struct)*new_size); + root->char_tails= tails; + end= tails + new_size; + for (cur= tails+real_size; curfirst_char= 0; + root->last_char= (*name); + } + + insert_into_hash(root->char_tails+(*name)-root->first_char, + name+1,len_from_begin+1,index,function); +} + + +hash_lex_struct *root_by_len= 0; +int max_len=0; + +hash_lex_struct *root_by_len2= 0; +int max_len2=0; + +void insert_symbols() +{ + size_t i= 0; + SYMBOL *cur; + for (cur= symbols; ilength,&max_len); + insert_into_hash(root,cur->name,0,(uint) i,0); + } +} + +void insert_sql_functions() +{ + int i= 0; + SYMBOL *cur; + for (cur= sql_functions; i < (int) array_elements(sql_functions); cur++, i++) + { + hash_lex_struct *root= + get_hash_struct_by_len(&root_by_len,cur->length,&max_len); + insert_into_hash(root,cur->name,0,-i-1,1); + } +} + +void calc_length() +{ + SYMBOL *cur, *end= symbols + array_elements(symbols); + for (cur= symbols; cur < end; cur++) + cur->length=(uchar) strlen(cur->name); + end= sql_functions + array_elements(sql_functions); + for (cur= sql_functions; curlength=(uchar) strlen(cur->name); +} + +void generate_find_structs() +{ + root_by_len= 0; + max_len=0; + + insert_symbols(); + + root_by_len2= root_by_len; + max_len2= max_len; + + root_by_len= 0; + max_len= 0; + + insert_symbols(); + insert_sql_functions(); +} + +char *hash_map= 0; +int size_hash_map= 0; + +void add_struct_to_map(hash_lex_struct *st) +{ + st->ithis= size_hash_map/4; + size_hash_map+= 4; + hash_map= (char*)realloc((char*)hash_map,size_hash_map); + hash_map[size_hash_map-4]= (char) (st->first_char == -1 ? 0 : + st->first_char); + hash_map[size_hash_map-3]= (char) (st->first_char == -1 || + st->first_char == 0 ? 0 : st->last_char); + if (st->first_char == -1) + { + hash_map[size_hash_map-2]= ((unsigned int)(int16)st->iresult)&255; + hash_map[size_hash_map-1]= ((unsigned int)(int16)st->iresult)>>8; + } + else if (st->first_char == 0) + { + hash_map[size_hash_map-2]= ((unsigned int)(int16)array_elements(symbols))&255; + hash_map[size_hash_map-1]= ((unsigned int)(int16)array_elements(symbols))>>8; + } +} + + +void add_structs_to_map(hash_lex_struct *st, int len) +{ + hash_lex_struct *cur, *end= st+len; + for (cur= st; curfirst_char && cur->first_char != -1) + add_structs_to_map(cur->char_tails,cur->last_char-cur->first_char+1); + } +} + +void set_links(hash_lex_struct *st, int len) +{ + hash_lex_struct *cur, *end= st+len; + for (cur= st; curfirst_char != 0 && cur->first_char != -1) + { + int ilink= cur->char_tails->ithis; + hash_map[cur->ithis*4+2]= ilink%256; + hash_map[cur->ithis*4+3]= ilink/256; + set_links(cur->char_tails,cur->last_char-cur->first_char+1); + } + } +} + + +void print_hash_map(const char *name) +{ + char *cur; + int i; + + printf("static uchar %s[%d]= {\n",name,size_hash_map); + for (i=0, cur= hash_map; ilength!=s2->length || strncmp(s1->name,s2->name,s1->length)) + return 0; + + const char *err_tmpl= "\ngen_lex_hash fatal error : \ +Unfortunately gen_lex_hash can not generate a hash,\n since \ +your lex.h has duplicate definition for a symbol \"%s\"\n\n"; + printf (err_tmpl,s1->name); + fprintf (stderr,err_tmpl,s1->name); + + return 1; +} + + +int check_duplicates() +{ + SYMBOL *cur1, *cur2, *s_end, *f_end; + + s_end= symbols + array_elements(symbols); + f_end= sql_functions + array_elements(sql_functions); + + for (cur1= symbols; cur1sql_functions_max_len) return 0;\n\ + hash_map= sql_functions_map;\n\ + uint32 cur_struct= uint4korr(hash_map+((len-1)*4));\n\ +\n\ + for (;;){\n\ + uchar first_char= (uchar)cur_struct;\n\ +\n\ + if (first_char == 0)\n\ + {\n\ + int16 ires= (int16)(cur_struct>>16);\n\ + if (ires==array_elements(symbols)) return 0;\n\ + SYMBOL *res;\n\ + if (ires>=0) \n\ + res= symbols+ires;\n\ + else\n\ + res= sql_functions-ires-1;\n\ + uint count= (uint) (cur_str - s);\n\ + return lex_casecmp(cur_str,res->name+count,len-count) ? 0 : res;\n\ + }\n\ +\n\ + uchar cur_char= (uchar)to_upper_lex[(uchar)*cur_str];\n\ + if (cur_char>=8;\n\ + if (cur_char>(uchar)cur_struct) return 0;\n\ +\n\ + cur_struct>>=8;\n\ + cur_struct= uint4korr(hash_map+\n\ + (((uint16)cur_struct + cur_char - first_char)*4));\n\ + cur_str++;\n\ + }\n" +); + + printf("\ + }else{\n\ + if (len>symbols_max_len) return 0;\n\ + hash_map= symbols_map;\n\ + uint32 cur_struct= uint4korr(hash_map+((len-1)*4));\n\ +\n\ + for (;;){\n\ + uchar first_char= (uchar)cur_struct;\n\ +\n\ + if (first_char==0) {\n\ + int16 ires= (int16)(cur_struct>>16);\n\ + if (ires==array_elements(symbols)) return 0;\n\ + SYMBOL *res= symbols+ires;\n\ + uint count= (uint) (cur_str - s);\n\ + return lex_casecmp(cur_str,res->name+count,len-count)!=0 ? 0 : res;\n\ + }\n\ +\n\ + uchar cur_char= (uchar)to_upper_lex[(uchar)*cur_str];\n\ + if (cur_char>=8;\n\ + if (cur_char>(uchar)cur_struct) return 0;\n\ +\n\ + cur_struct>>=8;\n\ + cur_struct= uint4korr(hash_map+\n\ + (((uint16)cur_struct + cur_char - first_char)*4));\n\ + cur_str++;\n\ + }\n\ + }\n\ +}\n" +); + exit(0); +} + -- cgit v1.2.3