cb22561995-10-11Fredrik Hübinette (Hubbe) /*\
06983f1996-09-22Fredrik Hübinette (Hubbe) ||| This file a part of Pike, and is copyright by Fredrik Hubinette ||| Pike is distributed as GPL (General Public License)
cb22561995-10-11Fredrik Hübinette (Hubbe) ||| See the files COPYING and DISCLAIMER for more information. \*/
5267b71995-08-09Fredrik Hübinette (Hubbe) #include "global.h" #include "stralloc.h"
bb55f81997-03-16Fredrik Hübinette (Hubbe) #include "pike_macros.h"
5267b71995-08-09Fredrik Hübinette (Hubbe) #include "dynamic_buffer.h"
bb55f81997-03-16Fredrik Hübinette (Hubbe) #include "pike_macros.h"
9aa6fa1997-05-19Fredrik Hübinette (Hubbe) #include "pike_memory.h"
5267b71995-08-09Fredrik Hübinette (Hubbe) #include "error.h"
9c6f7d1997-04-15Fredrik Hübinette (Hubbe) #include "gc.h"
75920f1997-12-28Fredrik Hübinette (Hubbe) #include "stuff.h"
5267b71995-08-09Fredrik Hübinette (Hubbe) 
2043ba1998-02-10Fredrik Hübinette (Hubbe) #include <ctype.h>
49ed181998-10-15Henrik Grubbström (Grubba) RCSID("$Id: stralloc.c,v 1.45 1998/10/15 02:34:37 grubba Exp $");
24ddc71998-03-28Henrik Grubbström (Grubba) 
af93211996-10-12Fredrik Hübinette (Hubbe) #define BEGIN_HASH_SIZE 997 #define MAX_AVG_LINK_LENGTH 3
063fe31998-03-10Per Hedbor #define HASH_PREFIX 64
af93211996-10-12Fredrik Hübinette (Hubbe)  unsigned INT32 htable_size=0;
75920f1997-12-28Fredrik Hübinette (Hubbe) static unsigned int hashprimes_entry=0;
af93211996-10-12Fredrik Hübinette (Hubbe) static struct pike_string **base_table=0;
5267b71995-08-09Fredrik Hübinette (Hubbe) static unsigned INT32 full_hash_value;
af93211996-10-12Fredrik Hübinette (Hubbe) unsigned INT32 num_strings=0;
5267b71995-08-09Fredrik Hübinette (Hubbe) 
af93211996-10-12Fredrik Hübinette (Hubbe) /*** Main string hash function ***/
db4a401998-10-09Fredrik Hübinette (Hubbe)  #define StrHash(s,len) low_do_hash(s,len,0) static unsigned int low_do_hash(const void *s, int len, int size_shift)
5267b71995-08-09Fredrik Hübinette (Hubbe) {
db4a401998-10-09Fredrik Hübinette (Hubbe)  full_hash_value=hashmem(s,len<<size_shift,HASH_PREFIX<<size_shift);
af93211996-10-12Fredrik Hübinette (Hubbe)  return full_hash_value % htable_size;
5267b71995-08-09Fredrik Hübinette (Hubbe) }
db4a401998-10-09Fredrik Hübinette (Hubbe) static INLINE unsigned int do_hash(struct pike_string *s) {
3e625c1998-10-11Fredrik Hübinette (Hubbe)  return low_do_hash(s->str, s->len, s->size_shift);
db4a401998-10-09Fredrik Hübinette (Hubbe) } static INLINE int find_magnitude1(const unsigned INT16 *s, int len) { while(--len>=0) if(s[len]>=256) return 1; return 0; } static INLINE int find_magnitude2(const unsigned INT32 *s, int len) { while(--len>=0) { if(s[len]>=256) { do { if(s[len]>=65536) return 2; }while(--len>=0); return 1; } } return 0; } static INLINE int min_magnitude(const unsigned INT32 c) { if(c<256) return 0; if(c<65536) return 1; return 2; } static INLINE unsigned INT32 generic_extract (const void *str, int size, int pos) { switch(size) { case 0: return ((unsigned char *)str)[pos]; case 1: return ((unsigned INT16 *)str)[pos]; case 2: return ((unsigned INT32 *)str)[pos]; default: fatal("Illegal shift size!\n"); } } INLINE unsigned INT32 index_shared_string(struct pike_string *s, int pos) { #ifdef DEBUG
49ed181998-10-15Henrik Grubbström (Grubba)  if(pos >= s->len || pos<0) { if (s->len) { fatal("String index %d is out of range [0 - %d]!\n", pos, s->len-1); } else { fatal("Attempt to index the empty string with %d!\n", pos); } }
db4a401998-10-09Fredrik Hübinette (Hubbe) #endif return generic_extract(s->str,s->size_shift,pos); }
3e625c1998-10-11Fredrik Hübinette (Hubbe) INLINE void low_set_index(struct pike_string *s, int pos, int value)
db4a401998-10-09Fredrik Hübinette (Hubbe) { #ifdef DEBUG if(pos > s->len || pos<0) fatal("string index out of range!\n"); if(pos == s->len && value) fatal("string zero termination foul!\n"); #endif switch(s->size_shift) { case 0: STR0(s)[pos]=value; break; case 1: STR1(s)[pos]=value; break; case 2: STR2(s)[pos]=value; break; default: fatal("Illegal shift size!\n"); } } #ifdef DEBUG struct INLINE pike_string *debug_check_size_shift(struct pike_string *a,int shift) { if(a->size_shift != shift) fatal("Wrong STRX macro used!\n"); return a; } #endif
3e625c1998-10-11Fredrik Hübinette (Hubbe) #define CONVERT(FROM,TO) \ INLINE void PIKE_CONCAT4(convert_,FROM,_to_,TO)(PIKE_CONCAT(p_wchar,TO) *to, const PIKE_CONCAT(p_wchar,FROM) *from, int len) \ { while(--len>=0) *(to++)=*(from++); } \ INLINE INT32 PIKE_CONCAT4(compare_,FROM,_to_,TO)(const PIKE_CONCAT(p_wchar,TO) *to, const PIKE_CONCAT(p_wchar,FROM) *from, int len) \ { int tmp; while(--len>=0) if((tmp=*(to++)-*(from++))) return tmp; return 0; } CONVERT(0,1) CONVERT(0,2) CONVERT(1,0) CONVERT(1,2) CONVERT(2,0) CONVERT(2,1)
db4a401998-10-09Fredrik Hübinette (Hubbe) int generic_compare_strings(const void *a,int alen, int asize, const void *b,int blen, int bsize) { #define TWO_SIZES(X,Y) (((X)<<2)+(Y)) if(alen != blen) return 0; if(asize==bsize) { return !MEMCPY(a,b,alen<<asize); }else{ INT32 pos; for(pos=0;pos< alen ;pos++) if(generic_extract(a,asize,pos) != generic_extract(b,bsize,pos)) return 0; return 1; } }
0bc4cf1998-10-13Fredrik Hübinette (Hubbe) void generic_memcpy(PCHARP to, PCHARP from,
3e625c1998-10-11Fredrik Hübinette (Hubbe)  int len)
db4a401998-10-09Fredrik Hübinette (Hubbe) {
0bc4cf1998-10-13Fredrik Hübinette (Hubbe)  switch(TWO_SIZES(from.shift,to.shift))
db4a401998-10-09Fredrik Hübinette (Hubbe)  { case TWO_SIZES(0,0):
0bc4cf1998-10-13Fredrik Hübinette (Hubbe)  convert_0_to_0((p_wchar0 *)to.ptr,(p_wchar0 *)from.ptr,len);
db4a401998-10-09Fredrik Hübinette (Hubbe)  break; case TWO_SIZES(0,1):
0bc4cf1998-10-13Fredrik Hübinette (Hubbe)  convert_0_to_1((p_wchar1 *)to.ptr,(p_wchar0 *)from.ptr,len);
db4a401998-10-09Fredrik Hübinette (Hubbe)  break; case TWO_SIZES(0,2):
0bc4cf1998-10-13Fredrik Hübinette (Hubbe)  convert_0_to_2((p_wchar2 *)to.ptr,(p_wchar0 *)from.ptr,len);
db4a401998-10-09Fredrik Hübinette (Hubbe)  break; case TWO_SIZES(1,0):
0bc4cf1998-10-13Fredrik Hübinette (Hubbe)  convert_1_to_0((p_wchar0 *)to.ptr,(p_wchar1 *)from.ptr,len);
db4a401998-10-09Fredrik Hübinette (Hubbe)  break; case TWO_SIZES(1,1):
0bc4cf1998-10-13Fredrik Hübinette (Hubbe)  convert_1_to_1((p_wchar1 *)to.ptr,(p_wchar1 *)from.ptr,len);
db4a401998-10-09Fredrik Hübinette (Hubbe)  break; case TWO_SIZES(1,2):
0bc4cf1998-10-13Fredrik Hübinette (Hubbe)  convert_1_to_2((p_wchar2 *)to.ptr,(p_wchar1 *)from.ptr,len);
db4a401998-10-09Fredrik Hübinette (Hubbe)  break; case TWO_SIZES(2,0):
0bc4cf1998-10-13Fredrik Hübinette (Hubbe)  convert_2_to_0((p_wchar0 *)to.ptr,(p_wchar2 *)from.ptr,len);
db4a401998-10-09Fredrik Hübinette (Hubbe)  break; case TWO_SIZES(2,1):
0bc4cf1998-10-13Fredrik Hübinette (Hubbe)  convert_2_to_1((p_wchar1 *)to.ptr,(p_wchar2 *)from.ptr,len);
db4a401998-10-09Fredrik Hübinette (Hubbe)  break; case TWO_SIZES(2,2):
0bc4cf1998-10-13Fredrik Hübinette (Hubbe)  convert_2_to_2((p_wchar2 *)to.ptr,(p_wchar2 *)from.ptr,len);
db4a401998-10-09Fredrik Hübinette (Hubbe)  break; } }
0bc4cf1998-10-13Fredrik Hübinette (Hubbe) INLINE void pike_string_cpy(PCHARP to, struct pike_string *from)
3e625c1998-10-11Fredrik Hübinette (Hubbe) {
0bc4cf1998-10-13Fredrik Hübinette (Hubbe)  generic_memcpy(to,MKPCHARP_STR(from),from->len);
3e625c1998-10-11Fredrik Hübinette (Hubbe) }
5014211998-10-14Fredrik Hübinette (Hubbe)  #ifdef DEBUG #ifdef DEBUG_MALLOC #define DM(X) X #else #define DM(X) #endif
2043ba1998-02-10Fredrik Hübinette (Hubbe) static void locate_problem(int (*isproblem)(struct pike_string *)) { unsigned INT32 e; struct pike_string *s;
9e52381998-03-01Fredrik Hübinette (Hubbe)  DM(struct memhdr *yes=alloc_memhdr()); DM(struct memhdr *no=alloc_memhdr());
2043ba1998-02-10Fredrik Hübinette (Hubbe)  for(e=0;e<htable_size;e++)
9e52381998-03-01Fredrik Hübinette (Hubbe)  {
2043ba1998-02-10Fredrik Hübinette (Hubbe)  for(s=base_table[e];s;s=s->next)
9e52381998-03-01Fredrik Hübinette (Hubbe)  { if(isproblem(s)) { fprintf(stderr,"***Guilty string:\n"); debug_dump_pike_string(s, 70); DM(add_marks_to_memhdr(yes,s)); }else{ DM(add_marks_to_memhdr(no,s)); } } }
2043ba1998-02-10Fredrik Hübinette (Hubbe) 
9e52381998-03-01Fredrik Hübinette (Hubbe)  DM(fprintf(stderr,"Plausible problem location(s):\n")); DM(dump_memhdr_locations(yes,0)); DM(fprintf(stderr,"More Plausible problem location(s):\n")); DM(dump_memhdr_locations(yes,no));
2043ba1998-02-10Fredrik Hübinette (Hubbe) } static int has_zero_refs(struct pike_string *s) { return s->refs<=0; } static int wrong_hash(struct pike_string *s) {
db4a401998-10-09Fredrik Hübinette (Hubbe)  return (s->hval % htable_size) != do_hash(s);
2043ba1998-02-10Fredrik Hübinette (Hubbe) } static int improper_zero_termination(struct pike_string *s) {
db4a401998-10-09Fredrik Hübinette (Hubbe)  return index_shared_string(s,s->len);
2043ba1998-02-10Fredrik Hübinette (Hubbe) } #else #define locate_problem(X) #endif
5267b71995-08-09Fredrik Hübinette (Hubbe) 
db4a401998-10-09Fredrik Hübinette (Hubbe) /*\ find a string in the shared string table. ||| This assumes that the string is minimized!!!! \*/ static struct pike_string *internal_findstring(const char *s, int len, int size_shift, int h)
5267b71995-08-09Fredrik Hübinette (Hubbe) {
06983f1996-09-22Fredrik Hübinette (Hubbe)  struct pike_string *curr,**prev, **base;
5267b71995-08-09Fredrik Hübinette (Hubbe)  for(base = prev = base_table + h;( curr=*prev ); prev=&curr->next) { #ifdef DEBUG if(curr->refs<1)
38d6081998-02-07Fredrik Hübinette (Hubbe)  { debug_dump_pike_string(curr, 70);
2043ba1998-02-10Fredrik Hübinette (Hubbe)  locate_problem(has_zero_refs);
5267b71995-08-09Fredrik Hübinette (Hubbe)  fatal("String with no references.\n");
38d6081998-02-07Fredrik Hübinette (Hubbe)  }
5267b71995-08-09Fredrik Hübinette (Hubbe) #endif if (full_hash_value == curr->hval && len==curr->len &&
db4a401998-10-09Fredrik Hübinette (Hubbe)  size_shift==curr->size_shift &&
3e625c1998-10-11Fredrik Hübinette (Hubbe)  !MEMCMP(curr->str, s,len<<size_shift)) /* found it */
5267b71995-08-09Fredrik Hübinette (Hubbe)  { *prev = curr->next; curr->next = *base; *base = curr; return curr; /* pointer to string */ } } return 0; /* not found */ }
ca74dd1996-10-08Fredrik Hübinette (Hubbe) struct pike_string *binary_findstring(const char *foo, INT32 l)
5267b71995-08-09Fredrik Hübinette (Hubbe) {
db4a401998-10-09Fredrik Hübinette (Hubbe)  return internal_findstring(foo, l, 0, StrHash(foo,l));
5267b71995-08-09Fredrik Hübinette (Hubbe) }
ca74dd1996-10-08Fredrik Hübinette (Hubbe) struct pike_string *findstring(const char *foo) { return binary_findstring(foo, strlen(foo)); }
5267b71995-08-09Fredrik Hübinette (Hubbe) /* * find a string that is already shared and move it to the head * of that list in the hastable */
db4a401998-10-09Fredrik Hübinette (Hubbe) static struct pike_string *propagate_shared_string(const struct pike_string *s, int h)
5267b71995-08-09Fredrik Hübinette (Hubbe) {
06983f1996-09-22Fredrik Hübinette (Hubbe)  struct pike_string *curr, **prev, **base;
5267b71995-08-09Fredrik Hübinette (Hubbe)  for(base = prev = base_table + h;( curr=*prev ); prev=&curr->next) { if (curr == s) /* found it */ { *prev=curr->next; curr->next=*base; *base=curr; return curr; } #ifdef DEBUG if(curr->refs<1)
38d6081998-02-07Fredrik Hübinette (Hubbe)  { debug_dump_pike_string(curr, 70);
2043ba1998-02-10Fredrik Hübinette (Hubbe)  locate_problem(has_zero_refs);
5267b71995-08-09Fredrik Hübinette (Hubbe)  fatal("String with no references.\n");
38d6081998-02-07Fredrik Hübinette (Hubbe)  }
5267b71995-08-09Fredrik Hübinette (Hubbe) #endif } return 0; /* not found */ }
af93211996-10-12Fredrik Hübinette (Hubbe) /*** rehash ***/ static void rehash_string_backwards(struct pike_string *s)
5267b71995-08-09Fredrik Hübinette (Hubbe) {
af93211996-10-12Fredrik Hübinette (Hubbe)  int h; if(!s) return; rehash_string_backwards(s->next); h=s->hval % htable_size; s->next=base_table[h]; base_table[h]=s; }
aef30b1996-10-11Fredrik Hübinette (Hubbe) 
be478c1997-08-30Henrik Grubbström (Grubba) static void rehash(void)
af93211996-10-12Fredrik Hübinette (Hubbe) { int h,old; struct pike_string **old_base;
aef30b1996-10-11Fredrik Hübinette (Hubbe) 
af93211996-10-12Fredrik Hübinette (Hubbe)  old=htable_size; old_base=base_table;
75920f1997-12-28Fredrik Hübinette (Hubbe)  htable_size=hashprimes[++hashprimes_entry];
af93211996-10-12Fredrik Hübinette (Hubbe)  base_table=(struct pike_string **)xalloc(sizeof(struct pike_string *)*htable_size); MEMSET((char *)base_table,0,sizeof(struct pike_string *)*htable_size); for(h=0;h<old;h++) rehash_string_backwards(old_base[h]); if(old_base) free((char *)old_base);
5267b71995-08-09Fredrik Hübinette (Hubbe) }
af93211996-10-12Fredrik Hübinette (Hubbe)  /*** Make new strings ***/
5267b71995-08-09Fredrik Hübinette (Hubbe)  /* note that begin_shared_string expects the _exact_ size of the string, * not the maximum size */
61e9a01998-01-25Fredrik Hübinette (Hubbe) struct pike_string *debug_begin_shared_string(int len)
5267b71995-08-09Fredrik Hübinette (Hubbe) {
06983f1996-09-22Fredrik Hübinette (Hubbe)  struct pike_string *t;
7abf491998-04-17Fredrik Hübinette (Hubbe) #ifdef DEBUG
6a13711998-04-17Fredrik Hübinette (Hubbe)  extern int d_flag;
7abf491998-04-17Fredrik Hübinette (Hubbe)  if(d_flag>10) verify_shared_strings_tables(); #endif
06983f1996-09-22Fredrik Hübinette (Hubbe)  t=(struct pike_string *)xalloc(len + sizeof(struct pike_string));
ca74dd1996-10-08Fredrik Hübinette (Hubbe)  t->str[len]=0;
5267b71995-08-09Fredrik Hübinette (Hubbe)  t->len=len;
db4a401998-10-09Fredrik Hübinette (Hubbe)  t->size_shift=0;
5267b71995-08-09Fredrik Hübinette (Hubbe)  return t; }
af93211996-10-12Fredrik Hübinette (Hubbe) static void link_pike_string(struct pike_string *s, int h) { s->refs = 0; s->next = base_table[h]; base_table[h] = s; s->hval=full_hash_value; num_strings++; if(num_strings > MAX_AVG_LINK_LENGTH * htable_size) rehash(); }
db4a401998-10-09Fredrik Hübinette (Hubbe) struct pike_string *debug_begin_wide_shared_string(int len, int shift) { struct pike_string *t; #ifdef DEBUG extern int d_flag; if(d_flag>10) verify_shared_strings_tables(); #endif t=(struct pike_string *)xalloc((len<<shift) + sizeof(struct pike_string)); t->len=len; t->size_shift=shift; low_set_index(t,len,0); return t; }
3e625c1998-10-11Fredrik Hübinette (Hubbe) /* * This function assumes that the shift size is already the minimum it * can be. */ struct pike_string *low_end_shared_string(struct pike_string *s)
5267b71995-08-09Fredrik Hübinette (Hubbe) { int len,h;
06983f1996-09-22Fredrik Hübinette (Hubbe)  struct pike_string *s2;
5267b71995-08-09Fredrik Hübinette (Hubbe) 
3e625c1998-10-11Fredrik Hübinette (Hubbe)  len=s->len; h=do_hash(s); s2=internal_findstring(s->str,len,s->size_shift,h); #ifdef DEBUG if(s2==s) fatal("end_shared_string called twice! (or something like that)\n"); #endif if(s2) { free((char *)s); s=s2; }else{ link_pike_string(s, h); } add_ref(s); return s; } /* * This function checks if the shift size can be decreased before * entering the string in the shared string table */ struct pike_string *end_shared_string(struct pike_string *s) { struct pike_string *s2;
db4a401998-10-09Fredrik Hübinette (Hubbe)  switch(s->size_shift) { default: fatal("ARGHEL!\n"); case 2: switch(find_magnitude2(STR2(s),s->len)) { case 0: s2=begin_shared_string(s->len); convert_2_to_0(STR0(s2),STR2(s),s->len); free((char *)s); s=s2; break; case 1: s2=begin_wide_shared_string(s->len,2); convert_2_to_1(STR1(s2),STR2(s),s->len); free((char *)s); s=s2; /* Fall though */ } break; case 1: if(!find_magnitude1(STR1(s),s->len)) { s2=begin_shared_string(s->len); convert_1_to_0(STR0(s2),STR1(s),s->len); free((char *)s); s=s2; } break; case 0: break; }
3e625c1998-10-11Fredrik Hübinette (Hubbe)  return low_end_shared_string(s);
5267b71995-08-09Fredrik Hübinette (Hubbe) }
db4a401998-10-09Fredrik Hübinette (Hubbe) 
61e9a01998-01-25Fredrik Hübinette (Hubbe) struct pike_string * debug_make_shared_binary_string(const char *str,int len)
5267b71995-08-09Fredrik Hübinette (Hubbe) {
06983f1996-09-22Fredrik Hübinette (Hubbe)  struct pike_string *s;
5267b71995-08-09Fredrik Hübinette (Hubbe)  int h=StrHash(str,len);
db4a401998-10-09Fredrik Hübinette (Hubbe)  s = internal_findstring(str,len,0,h);
5267b71995-08-09Fredrik Hübinette (Hubbe)  if (!s) { s=begin_shared_string(len); MEMCPY(s->str, str, len);
af93211996-10-12Fredrik Hübinette (Hubbe)  link_pike_string(s, h);
5267b71995-08-09Fredrik Hübinette (Hubbe)  }
d6ac731998-04-20Henrik Grubbström (Grubba)  add_ref(s);
5267b71995-08-09Fredrik Hübinette (Hubbe)  return s; }
db4a401998-10-09Fredrik Hübinette (Hubbe) struct pike_string * debug_make_shared_binary_string1(const INT16 *str,int len) { struct pike_string *s; int h; if(!find_magnitude1(str,len)) { /* Wrong size, convert */ s=begin_shared_string(len); convert_1_to_0(s->str,str,len); return end_shared_string(s); }
3e625c1998-10-11Fredrik Hübinette (Hubbe)  h=low_do_hash(str, len, 1);
db4a401998-10-09Fredrik Hübinette (Hubbe)  s = internal_findstring((char *)str,len,1,h); if (!s) { s=begin_wide_shared_string(len,1); MEMCPY(s->str, str, len<<1); link_pike_string(s, h); } add_ref(s); return s; } struct pike_string * debug_make_shared_binary_string2(const INT32 *str,int len) { struct pike_string *s; int h; switch(find_magnitude2(str,len)) { case 0: /* Wrong size, convert */ s=begin_shared_string(len); convert_2_to_0(s->str,str,len); return end_shared_string(s); case 1: /* Wrong size, convert */ s=begin_wide_shared_string(len,1); convert_2_to_1(STR1(s),str,len); return end_shared_string(s); /* not entirely optimal */ }
3e625c1998-10-11Fredrik Hübinette (Hubbe)  h=low_do_hash(str, len, 2);
db4a401998-10-09Fredrik Hübinette (Hubbe)  s = internal_findstring((char *)str,len,2,h); if (!s) { s=begin_wide_shared_string(len,2); MEMCPY(s->str, str, len<<2); link_pike_string(s, h); } add_ref(s); return s; }
61e9a01998-01-25Fredrik Hübinette (Hubbe) struct pike_string *debug_make_shared_string(const char *str)
5267b71995-08-09Fredrik Hübinette (Hubbe) { return make_shared_binary_string(str, strlen(str)); }
3e625c1998-10-11Fredrik Hübinette (Hubbe) struct pike_string *debug_make_shared_string1(const INT16 *str)
4d1ed11998-09-18Fredrik Hübinette (Hubbe) {
db4a401998-10-09Fredrik Hübinette (Hubbe)  INT32 len;
4d1ed11998-09-18Fredrik Hübinette (Hubbe)  for(len=0;str[len];len++);
db4a401998-10-09Fredrik Hübinette (Hubbe)  return debug_make_shared_binary_string1(str,len); }
3e625c1998-10-11Fredrik Hübinette (Hubbe) struct pike_string *debug_make_shared_string2(const INT32 *str)
db4a401998-10-09Fredrik Hübinette (Hubbe) { INT32 len; for(len=0;str[len];len++); return debug_make_shared_binary_string2(str,len);
4d1ed11998-09-18Fredrik Hübinette (Hubbe) }
af93211996-10-12Fredrik Hübinette (Hubbe) /*** Free strings ***/
5267b71995-08-09Fredrik Hübinette (Hubbe) 
0a3d601996-10-09Fredrik Hübinette (Hubbe) void unlink_pike_string(struct pike_string *s)
5267b71995-08-09Fredrik Hübinette (Hubbe) { int h;
db4a401998-10-09Fredrik Hübinette (Hubbe)  h=do_hash(s);
5267b71995-08-09Fredrik Hübinette (Hubbe)  propagate_shared_string(s,h); base_table[h]=s->next;
2302231998-04-23Fredrik Hübinette (Hubbe) #ifdef DEBUG s->next=(struct pike_string *)-1; #endif
760b261996-12-03Fredrik Hübinette (Hubbe)  num_strings--;
0a3d601996-10-09Fredrik Hübinette (Hubbe) } void really_free_string(struct pike_string *s) {
9367351997-01-27Fredrik Hübinette (Hubbe) #ifdef DEBUG extern int d_flag; if(d_flag > 2) { if(s->next == (struct pike_string *)-1) fatal("Freeing shared string again!\n");
7abf491998-04-17Fredrik Hübinette (Hubbe)  if(((long)s->next) & 1) fatal("Freeing shared string again, memory corrupt or other bug!\n");
9367351997-01-27Fredrik Hübinette (Hubbe)  } #endif
0a3d601996-10-09Fredrik Hübinette (Hubbe)  unlink_pike_string(s);
5267b71995-08-09Fredrik Hübinette (Hubbe)  free((char *)s); }
61e9a01998-01-25Fredrik Hübinette (Hubbe) void debug_free_string(struct pike_string *s) { if(--s->refs<=0) really_free_string(s); }
af93211996-10-12Fredrik Hübinette (Hubbe) 
5267b71995-08-09Fredrik Hübinette (Hubbe) /*
af93211996-10-12Fredrik Hübinette (Hubbe)  * String table status
5267b71995-08-09Fredrik Hübinette (Hubbe)  */
06983f1996-09-22Fredrik Hübinette (Hubbe) struct pike_string *add_string_status(int verbose)
5267b71995-08-09Fredrik Hübinette (Hubbe) { char b[200]; init_buf(); if (verbose) { int allocd_strings=0; int allocd_bytes=0; int num_distinct_strings=0; int bytes_distinct_strings=0; int overhead_bytes=0;
af93211996-10-12Fredrik Hübinette (Hubbe)  unsigned INT32 e;
06983f1996-09-22Fredrik Hübinette (Hubbe)  struct pike_string *p;
af93211996-10-12Fredrik Hübinette (Hubbe)  for(e=0;e<htable_size;e++)
5267b71995-08-09Fredrik Hübinette (Hubbe)  { for(p=base_table[e];p;p=p->next) { num_distinct_strings++;
f90e541995-08-17Fredrik Hübinette (Hubbe)  bytes_distinct_strings+=MY_ALIGN(p->len);
5267b71995-08-09Fredrik Hübinette (Hubbe)  allocd_strings+=p->refs;
f90e541995-08-17Fredrik Hübinette (Hubbe)  allocd_bytes+=p->refs*MY_ALIGN(p->len+3);
5267b71995-08-09Fredrik Hübinette (Hubbe)  } }
06983f1996-09-22Fredrik Hübinette (Hubbe)  overhead_bytes=(sizeof(struct pike_string)-1)*num_distinct_strings;
5267b71995-08-09Fredrik Hübinette (Hubbe)  my_strcat("\nShared string hash table:\n"); my_strcat("-------------------------\t Strings Bytes\n"); sprintf(b,"Total asked for\t\t\t%8ld %8ld\n", (long)allocd_strings, (long)allocd_bytes); my_strcat(b); sprintf(b,"Strings malloced\t\t%8ld %8ld + %ld overhead\n", (long)num_distinct_strings, (long)bytes_distinct_strings, (long)overhead_bytes); my_strcat(b); sprintf(b,"Space actually required/total string bytes %d%%\n", (bytes_distinct_strings + overhead_bytes)*100 / allocd_bytes); my_strcat(b); } /* sprintf(b,"Searches: %ld Average search length: %6.3f\n", (long)num_str_searches, (double)search_len / num_str_searches); my_strcat(b); */ return free_buf(); }
af93211996-10-12Fredrik Hübinette (Hubbe) /*** DEBUG ***/ #ifdef DEBUG void check_string(struct pike_string *s) {
db4a401998-10-09Fredrik Hübinette (Hubbe)  do_hash(s);
af93211996-10-12Fredrik Hübinette (Hubbe)  if(full_hash_value != s->hval)
2043ba1998-02-10Fredrik Hübinette (Hubbe)  { locate_problem(wrong_hash);
af93211996-10-12Fredrik Hübinette (Hubbe)  fatal("Hash value changed?\n");
2043ba1998-02-10Fredrik Hübinette (Hubbe)  }
af93211996-10-12Fredrik Hübinette (Hubbe)  if(debug_findstring(s) !=s) fatal("Shared string not shared.\n");
db4a401998-10-09Fredrik Hübinette (Hubbe)  if(index_shared_string(s,s->len))
2043ba1998-02-10Fredrik Hübinette (Hubbe)  { locate_problem(improper_zero_termination);
af93211996-10-12Fredrik Hübinette (Hubbe)  fatal("Shared string is not zero terminated properly.\n");
2043ba1998-02-10Fredrik Hübinette (Hubbe)  }
af93211996-10-12Fredrik Hübinette (Hubbe) }
be478c1997-08-30Henrik Grubbström (Grubba) void verify_shared_strings_tables(void)
af93211996-10-12Fredrik Hübinette (Hubbe) {
beac721998-04-16Fredrik Hübinette (Hubbe)  unsigned INT32 e, h, num=0;
95fd021998-05-25Henrik Grubbström (Grubba)  unsigned INT32 orig_full_hash = full_hash_value;
af93211996-10-12Fredrik Hübinette (Hubbe)  struct pike_string *s; for(e=0;e<htable_size;e++) { h=0; for(s=base_table[e];s;s=s->next) {
beac721998-04-16Fredrik Hübinette (Hubbe)  num++;
af93211996-10-12Fredrik Hübinette (Hubbe)  h++; if(s->len < 0) fatal("Shared string shorter than zero bytes.\n"); if(s->refs <= 0)
2043ba1998-02-10Fredrik Hübinette (Hubbe)  { locate_problem(has_zero_refs);
af93211996-10-12Fredrik Hübinette (Hubbe)  fatal("Shared string had too few references.\n");
2043ba1998-02-10Fredrik Hübinette (Hubbe)  }
af93211996-10-12Fredrik Hübinette (Hubbe)  if(s->str[s->len])
2043ba1998-02-10Fredrik Hübinette (Hubbe)  { locate_problem(improper_zero_termination);
af93211996-10-12Fredrik Hübinette (Hubbe)  fatal("Shared string didn't end with a zero.\n");
2043ba1998-02-10Fredrik Hübinette (Hubbe)  }
af93211996-10-12Fredrik Hübinette (Hubbe) 
db4a401998-10-09Fredrik Hübinette (Hubbe)  if(do_hash(s) != e)
2043ba1998-02-10Fredrik Hübinette (Hubbe)  { locate_problem(wrong_hash);
af93211996-10-12Fredrik Hübinette (Hubbe)  fatal("Shared string hashed to wrong place.\n");
2043ba1998-02-10Fredrik Hübinette (Hubbe)  }
af93211996-10-12Fredrik Hübinette (Hubbe)  if(s->hval != full_hash_value) fatal("Shared string hashed to other number.\n"); if(h>10000) { struct pike_string *s2; for(s2=s;s2;s2=s2->next) if(s2 == s) fatal("Shared string table is cyclic.\n"); h=0; } } }
beac721998-04-16Fredrik Hübinette (Hubbe)  if(num != num_strings) fatal("Num strings is wrong %d!=%d\n",num,num_strings);
95fd021998-05-25Henrik Grubbström (Grubba)  full_hash_value = orig_full_hash;
af93211996-10-12Fredrik Hübinette (Hubbe) }
62971d1998-01-19Fredrik Hübinette (Hubbe) int safe_debug_findstring(struct pike_string *foo) { unsigned INT32 e; if(!base_table) return 0; for(e=0;e<htable_size;e++) { struct pike_string *p; for(p=base_table[e];p;p=p->next) if(p==foo) return 1; } return 0; }
af93211996-10-12Fredrik Hübinette (Hubbe) struct pike_string *debug_findstring(const struct pike_string *foo) { struct pike_string *tmp; tmp=propagate_shared_string(foo, foo->hval % htable_size); #if 0 if(!tmp) { unsigned INT32 e; struct pike_string *tmp2; fprintf(stderr,"String %p %ld %ld %s\n", foo, (long)foo->hval, (long)foo->len, foo->str); StrHash(foo->str,foo->len); fprintf(stderr,"------ %p %ld\n", base_table[foo->hval %htable_size], (long)full_hash_value); for(tmp2=base_table[foo->hval % htable_size];tmp2;tmp2=tmp2->next) { if(tmp2 == tmp) fprintf(stderr,"!!%p!!->",tmp2); else fprintf(stderr,"%p->",tmp2); } fprintf(stderr,"0\n"); for(e=0;e<htable_size;e++) { for(tmp2=base_table[e];tmp2;tmp2=tmp2->next) { if(tmp2 == tmp) fprintf(stderr,"String found in hashbin %ld (not %ld)\n", (long)e, (long)(foo->hval % htable_size)); } } } #endif return tmp; }
38d6081998-02-07Fredrik Hübinette (Hubbe) void debug_dump_pike_string(struct pike_string *s, INT32 max) { INT32 e;
a737441998-10-11Henrik Grubbström (Grubba)  fprintf(stderr,"0x%p: %ld refs, len=%ld, size_shift=%d, hval=%lux (%lx)\n",
38d6081998-02-07Fredrik Hübinette (Hubbe)  s, (long)s->refs, (long)s->len,
a737441998-10-11Henrik Grubbström (Grubba)  s->size_shift,
38d6081998-02-07Fredrik Hübinette (Hubbe)  (unsigned long)s->hval, (unsigned long)StrHash(s->str, s->len)); fprintf(stderr," \""); for(e=0;e<s->len && max>0;e++) { int c=EXTRACT_UCHAR(s->str+e); switch(c) { case '\t': fprintf(stderr,"\\t"); max-=2; break; case '\n': fprintf(stderr,"\\n"); max-=2; break; case '\r': fprintf(stderr,"\\r"); max-=2; break; case '\b': fprintf(stderr,"\\b"); max-=2; break; default: if(is8bitalnum(c) || c==' ' || isgraph(c)) { putc(c,stderr); max--; }else{ fprintf(stderr,"\\%03o",c); max-=4; } } } if(!max) fprintf(stderr,"...\n"); else fprintf(stderr,"\"\n"); }
be478c1997-08-30Henrik Grubbström (Grubba) void dump_stralloc_strings(void)
5267b71995-08-09Fredrik Hübinette (Hubbe) {
af93211996-10-12Fredrik Hübinette (Hubbe)  unsigned INT32 e;
06983f1996-09-22Fredrik Hübinette (Hubbe)  struct pike_string *p;
af93211996-10-12Fredrik Hübinette (Hubbe)  for(e=0;e<htable_size;e++)
5267b71995-08-09Fredrik Hübinette (Hubbe)  for(p=base_table[e];p;p=p->next)
38d6081998-02-07Fredrik Hübinette (Hubbe)  debug_dump_pike_string(p, 70);
5267b71995-08-09Fredrik Hübinette (Hubbe) }
af93211996-10-12Fredrik Hübinette (Hubbe) #endif /*** String compare functions ***/ /* does not take locale into account */ int low_quick_binary_strcmp(char *a,INT32 alen, char *b,INT32 blen) { int tmp; if(alen > blen) { tmp=MEMCMP(a, b, blen); if(tmp) return tmp; return 1; }else if(alen < blen){ tmp=MEMCMP(a, b, alen); if(tmp) return tmp; return -1; }else{ return MEMCMP(a, b, alen); } }
3e625c1998-10-11Fredrik Hübinette (Hubbe) /* does not take locale into account */ int generic_quick_binary_strcmp(const char *a,INT32 alen, int asize, const char *b,INT32 blen, int bsize) { if(!asize && !bsize) { int tmp; if(alen > blen) { tmp=MEMCMP(a, b, blen); if(tmp) return tmp; return 1; }else if(alen < blen){ tmp=MEMCMP(a, b, alen); if(tmp) return tmp; return -1; }else{ return MEMCMP(a, b, alen); } }else{ INT32 pos; for(pos=0;pos< MINIMUM(alen,blen) ;pos++) { INT32 ac=generic_extract(a,asize,pos); INT32 bc=generic_extract(b,bsize,pos); if(ac-bc) return ac-bc; } return alen-blen; } }
af93211996-10-12Fredrik Hübinette (Hubbe) #ifndef HAVE_STRCOLL /* No locale function available */ static int low_binary_strcmp(char *a,INT32 alen, char *b,INT32 blen) { low_quick_binary_strcmp(a,alen,b,blen); } #else /* takes locale into account */ static int low_binary_strcmp(char *a,INT32 alen, char *b,INT32 blen) { INT32 tmp; while(alen>0 && blen>0) { tmp=strcoll(a,b); if(tmp) return (int)tmp; tmp=strlen(a)+1; a+=tmp; b+=tmp; alen-=tmp; blen-=tmp; } if(alen==blen) return 0; if(alen > blen) return 1; return -1; } #endif /* Does not take locale into account */
3e625c1998-10-11Fredrik Hübinette (Hubbe) 
af93211996-10-12Fredrik Hübinette (Hubbe) int my_quick_strcmp(struct pike_string *a,struct pike_string *b) { if(a==b) return 0;
3e625c1998-10-11Fredrik Hübinette (Hubbe)  return generic_quick_binary_strcmp(a->str, a->len, a->size_shift, b->str, b->len, b->size_shift);
af93211996-10-12Fredrik Hübinette (Hubbe) } /* Does take locale into account */ int my_strcmp(struct pike_string *a,struct pike_string *b) { if(a==b) return 0;
3e625c1998-10-11Fredrik Hübinette (Hubbe)  switch(TWO_SIZES(a->size_shift,b->size_shift)) { case TWO_SIZES(0,0): return low_binary_strcmp(a->str,a->len,b->str,b->len); default: { INT32 e,l=MINIMUM(a->len,b->len); for(e=0;e<l;e++) { INT32 ac=index_shared_string(a,e); INT32 bc=index_shared_string(b,e); #ifdef HAVE_STRCOLL if(ac < 256 && bc < 256) { char atmp[2],btmp[2]; int tmp; atmp[0]=ac; btmp[0]=bc; atmp[1]=0; btmp[1]=0; if((tmp=strcoll(atmp,btmp))) return tmp; }else #endif if(ac-bc) return ac-bc; } return a->len - b->len; } }
af93211996-10-12Fredrik Hübinette (Hubbe) }
8d28be1997-02-10Fredrik Hübinette (Hubbe) struct pike_string *realloc_unlinked_string(struct pike_string *a, INT32 size) { struct pike_string *r; r=(struct pike_string *)realloc((char *)a,
3e625c1998-10-11Fredrik Hübinette (Hubbe)  sizeof(struct pike_string)+ ((size+1)<<a->size_shift)); /* FIXME !! */
8d28be1997-02-10Fredrik Hübinette (Hubbe)  if(!r) { r=begin_shared_string(size);
db4a401998-10-09Fredrik Hübinette (Hubbe)  MEMCPY(r->str, a->str, a->len<<a->size_shift);
8d28be1997-02-10Fredrik Hübinette (Hubbe)  free((char *)a); } r->len=size;
db4a401998-10-09Fredrik Hübinette (Hubbe)  low_set_index(r,size,0);
8d28be1997-02-10Fredrik Hübinette (Hubbe)  return r; } /* Returns an unlinked string ready for end_shared_string */ struct pike_string *realloc_shared_string(struct pike_string *a, INT32 size) { struct pike_string *r; if(a->refs==1) { unlink_pike_string(a); return realloc_unlinked_string(a, size); }else{
db4a401998-10-09Fredrik Hübinette (Hubbe)  r=begin_wide_shared_string(size,a->size_shift); MEMCPY(r->str, a->str, a->len<<a->size_shift);
7094631997-02-24Fredrik Hübinette (Hubbe)  free_string(a);
8d28be1997-02-10Fredrik Hübinette (Hubbe)  return r; } }
3e625c1998-10-11Fredrik Hübinette (Hubbe) struct pike_string *new_realloc_shared_string(struct pike_string *a, INT32 size, int shift) { struct pike_string *r; if(shift == a->size_shift) return realloc_shared_string(a,size); r=begin_wide_shared_string(size,shift);
0bc4cf1998-10-13Fredrik Hübinette (Hubbe)  pike_string_cpy(MKPCHARP_STR(r),a);
3e625c1998-10-11Fredrik Hübinette (Hubbe)  free_string(a); return r; }
0d3ea51998-01-19Fredrik Hübinette (Hubbe) /* Modify one index in a shared string * Not suitable for building new strings or changing multiple characters * within a string!
db4a401998-10-09Fredrik Hübinette (Hubbe)  * * Phew, this function become complicated when I inserted magic for wide * characters...
0d3ea51998-01-19Fredrik Hübinette (Hubbe)  */ struct pike_string *modify_shared_string(struct pike_string *a, INT32 index,
db4a401998-10-09Fredrik Hübinette (Hubbe)  INT32 c)
0d3ea51998-01-19Fredrik Hübinette (Hubbe) {
db4a401998-10-09Fredrik Hübinette (Hubbe)  INT32 old_value;
0d3ea51998-01-19Fredrik Hübinette (Hubbe) #ifdef DEBUG if(index<0 || index>=a->len) fatal("Index out of range in modify_shared_string()\n"); #endif
db4a401998-10-09Fredrik Hübinette (Hubbe)  old_value=index_shared_string(a,index); if(old_value==c) return a; /* First test if the string needs to be grown: * ie; the new value does not fit in the char size of * the old string */ if(min_magnitude(c) > a->size_shift) { /* String must be grown */ struct pike_string *b; switch(TWO_SIZES(min_magnitude(c),a->size_shift)) { case TWO_SIZES(1,0): b=begin_wide_shared_string(a->len,1); convert_0_to_1(STR1(b),a->str,a->len); STR1(b)[index]=c;
7238061998-10-11Fredrik Hübinette (Hubbe)  free_string(a);
db4a401998-10-09Fredrik Hübinette (Hubbe)  return end_shared_string(b); case TWO_SIZES(2,0): b=begin_wide_shared_string(a->len,2); convert_0_to_2(STR2(b),a->str,a->len); STR2(b)[index]=c;
7238061998-10-11Fredrik Hübinette (Hubbe)  free_string(a);
db4a401998-10-09Fredrik Hübinette (Hubbe)  return end_shared_string(b); case TWO_SIZES(2,1): b=begin_wide_shared_string(a->len,2); convert_1_to_2(STR2(b),STR1(a),a->len); STR2(b)[index]=c;
7238061998-10-11Fredrik Hübinette (Hubbe)  free_string(a);
db4a401998-10-09Fredrik Hübinette (Hubbe)  return end_shared_string(b); default: fatal("Odd wide string conversion!\n"); } } /* Next we test if the new string can be shrunk * if all characters in the new string can fit in a string * of a lower magnitude, it must be shrunk */ if(min_magnitude(old_value) == a->size_shift && min_magnitude(c) < min_magnitude(old_value)) { /* We *might* need to shrink the string */ struct pike_string *b; int size,tmp; switch(a->size_shift) { case 0: fatal("Unshrinkable!\n"); case 1: /* Test if we *actually* can shrink it.. */ if(find_magnitude1(STR1(a),index)) break; if(find_magnitude1(STR1(a)+index+1,a->len-index-1)) break; b=begin_shared_string(a->len); convert_1_to_0(b->str,STR1(a),a->len); b->str[index]=c; free_string(a); return end_shared_string(b); case 2: /* Test if we *actually* can shrink it.. */ size=find_magnitude2(STR2(a),index); if(size==2) break; /* nope */ tmp=find_magnitude2(STR2(a)+index+1,a->len-index-1); if(tmp==2) break; /* nope */ size=MAXIMUM(MAXIMUM(size,tmp),min_magnitude(c)); switch(size) { case 0: b=begin_shared_string(a->len); convert_2_to_0(b->str,STR2(a),a->len); b->str[index]=c; free_string(a); return end_shared_string(b); case 1: b=begin_wide_shared_string(a->len,1); convert_2_to_1((unsigned INT16 *)b->str,STR2(a),a->len); STR1(b)[index]=c; free_string(a); return end_shared_string(b); } } } /* We now know that the string has the right character size */
0d3ea51998-01-19Fredrik Hübinette (Hubbe)  if(a->refs==1) {
db4a401998-10-09Fredrik Hübinette (Hubbe)  /* One ref - destructive mode */
0d3ea51998-01-19Fredrik Hübinette (Hubbe)  if(index>=HASH_PREFIX && index<a->len-8) {
db4a401998-10-09Fredrik Hübinette (Hubbe)  /* Doesn't change hash value - sneak it in there */ low_set_index(a,index,c);
0d3ea51998-01-19Fredrik Hübinette (Hubbe)  return a; }else{ unlink_pike_string(a);
db4a401998-10-09Fredrik Hübinette (Hubbe)  low_set_index(a,index,c);
0d3ea51998-01-19Fredrik Hübinette (Hubbe)  return end_shared_string(a); } }else{ struct pike_string *r;
db4a401998-10-09Fredrik Hübinette (Hubbe)  r=begin_wide_shared_string(a->len,a->size_shift); MEMCPY(r->str, a->str, a->len << a->size_shift); low_set_index(r,index,c);
0d3ea51998-01-19Fredrik Hübinette (Hubbe)  free_string(a); return end_shared_string(r); } }
af93211996-10-12Fredrik Hübinette (Hubbe) /*** Add strings ***/
06983f1996-09-22Fredrik Hübinette (Hubbe) struct pike_string *add_shared_strings(struct pike_string *a, struct pike_string *b)
5267b71995-08-09Fredrik Hübinette (Hubbe) {
06983f1996-09-22Fredrik Hübinette (Hubbe)  struct pike_string *ret;
0bc4cf1998-10-13Fredrik Hübinette (Hubbe)  PCHARP tmp;
db4a401998-10-09Fredrik Hübinette (Hubbe)  int target_size=MAXIMUM(a->size_shift,b->size_shift);
5267b71995-08-09Fredrik Hübinette (Hubbe) 
db4a401998-10-09Fredrik Hübinette (Hubbe)  ret=begin_wide_shared_string(a->len+b->len,target_size);
0bc4cf1998-10-13Fredrik Hübinette (Hubbe)  tmp=MKPCHARP_STR(ret); pike_string_cpy(tmp,a); INC_PCHARP(tmp,a->len); pike_string_cpy(tmp,b);
db4a401998-10-09Fredrik Hübinette (Hubbe)  return end_shared_string(ret);
5267b71995-08-09Fredrik Hübinette (Hubbe) }
b1f4eb1998-01-13Fredrik Hübinette (Hubbe) struct pike_string *add_and_free_shared_strings(struct pike_string *a, struct pike_string *b) { INT32 alen=a->len;
db4a401998-10-09Fredrik Hübinette (Hubbe)  if(a->size_shift == b->size_shift) { a=realloc_shared_string(a,alen + b->len); MEMCPY(a->str+(alen<<a->size_shift),b->str,b->len<<b->size_shift); free_string(b); return end_shared_string(a); }else{ struct pike_string *ret=add_shared_strings(a,b); free_string(a); free_string(b); return ret; } } int string_search(struct pike_string *haystack, struct pike_string *needle, int start) { struct generic_mem_searcher s; char *r; if(needle->size_shift > haystack->size_shift) return -1; init_generic_memsearcher(&s, needle->str, needle->len, needle->size_shift,
0bc4cf1998-10-13Fredrik Hübinette (Hubbe)  haystack->len-start,
db4a401998-10-09Fredrik Hübinette (Hubbe)  haystack->size_shift); r=(char *)generic_memory_search(&s,
0bc4cf1998-10-13Fredrik Hübinette (Hubbe)  haystack->str+(start<<haystack->size_shift), haystack->len-start,
db4a401998-10-09Fredrik Hübinette (Hubbe)  haystack->size_shift); if(!r) return -1; return (r-haystack->str)>>haystack->size_shift;
b1f4eb1998-01-13Fredrik Hübinette (Hubbe) }
3e625c1998-10-11Fredrik Hübinette (Hubbe) struct pike_string *string_slice(struct pike_string *s, INT32 start, INT32 len) { #ifdef DEBUG if(start < 0 || len<0 || start+len>s->len ) { fatal("string_slice, start = %d, len = %d, s->len = %d\n",start,len,s->len); } #endif if(start==0 && len==s->len) { add_ref(s); return s; } switch(s->size_shift) { case 0: return make_shared_binary_string(STR0(s)+start,len); case 1: return make_shared_binary_string1(STR1(s)+start,len); case 2: return make_shared_binary_string2(STR2(s)+start,len); default: fatal("Illegal shift size!\n"); } }
db4a401998-10-09Fredrik Hübinette (Hubbe) 
af93211996-10-12Fredrik Hübinette (Hubbe) /*** replace function ***/
06983f1996-09-22Fredrik Hübinette (Hubbe) struct pike_string *string_replace(struct pike_string *str,
db4a401998-10-09Fredrik Hübinette (Hubbe)  struct pike_string *del, struct pike_string *to)
5267b71995-08-09Fredrik Hübinette (Hubbe) {
06983f1996-09-22Fredrik Hübinette (Hubbe)  struct pike_string *ret;
0bc4cf1998-10-13Fredrik Hübinette (Hubbe)  char *s,*tmp,*end; PCHARP r;
3e625c1998-10-11Fredrik Hübinette (Hubbe)  int shift; struct generic_mem_searcher searcher;
5267b71995-08-09Fredrik Hübinette (Hubbe) 
a991451997-07-08Fredrik Hübinette (Hubbe)  if(!str->len) {
d6ac731998-04-20Henrik Grubbström (Grubba)  add_ref(str);
a991451997-07-08Fredrik Hübinette (Hubbe)  return str; }
3e625c1998-10-11Fredrik Hübinette (Hubbe)  shift=MAXIMUM(str->size_shift,to->size_shift);
a991451997-07-08Fredrik Hübinette (Hubbe)  if(!del->len) {
0bc4cf1998-10-13Fredrik Hübinette (Hubbe)  int e,pos;
3e625c1998-10-11Fredrik Hübinette (Hubbe)  ret=begin_wide_shared_string(str->len + to->len * (str->len -1),shift); low_set_index(ret,0,index_shared_string(str,0));
0bc4cf1998-10-13Fredrik Hübinette (Hubbe)  for(pos=e=1;e<str->len;e++)
a991451997-07-08Fredrik Hübinette (Hubbe)  {
0bc4cf1998-10-13Fredrik Hübinette (Hubbe)  pike_string_cpy(MKPCHARP_STR_OFF(ret,pos),to);
3e625c1998-10-11Fredrik Hübinette (Hubbe)  pos+=to->len; low_set_index(ret,pos++,index_shared_string(str,e));
a991451997-07-08Fredrik Hübinette (Hubbe)  } return end_shared_string(ret); }
5267b71995-08-09Fredrik Hübinette (Hubbe)  s=str->str;
3e625c1998-10-11Fredrik Hübinette (Hubbe)  end=s+(str->len<<str->size_shift);
5267b71995-08-09Fredrik Hübinette (Hubbe) 
1f515a1997-02-15Fredrik Hübinette (Hubbe)  if(del->len == to->len)
5267b71995-08-09Fredrik Hübinette (Hubbe)  {
3e625c1998-10-11Fredrik Hübinette (Hubbe)  init_generic_memsearcher(&searcher, del->str, del->len, del->size_shift, str->len, str->size_shift); ret=begin_wide_shared_string(str->len,shift);
1f515a1997-02-15Fredrik Hübinette (Hubbe)  }else{ INT32 delimeters=0;
3e625c1998-10-11Fredrik Hübinette (Hubbe)  init_generic_memsearcher(&searcher, del->str, del->len, del->size_shift, str->len*2, str->size_shift); while((s=generic_memory_search(&searcher, s, (end-s)>>str->size_shift, str->size_shift)))
1f515a1997-02-15Fredrik Hübinette (Hubbe)  { delimeters++;
3e625c1998-10-11Fredrik Hübinette (Hubbe)  s+=del->len << str->size_shift;
1f515a1997-02-15Fredrik Hübinette (Hubbe)  } if(!delimeters) {
d6ac731998-04-20Henrik Grubbström (Grubba)  add_ref(str);
1f515a1997-02-15Fredrik Hübinette (Hubbe)  return str; }
5267b71995-08-09Fredrik Hübinette (Hubbe) 
3e625c1998-10-11Fredrik Hübinette (Hubbe)  ret=begin_wide_shared_string(str->len + (to->len-del->len)*delimeters, shift);
5267b71995-08-09Fredrik Hübinette (Hubbe)  } s=str->str;
0bc4cf1998-10-13Fredrik Hübinette (Hubbe)  r=MKPCHARP_STR(ret);
5267b71995-08-09Fredrik Hübinette (Hubbe) 
3e625c1998-10-11Fredrik Hübinette (Hubbe)  while((tmp=(char *)generic_memory_search(&searcher, s, (end-s)>>str->size_shift, str->size_shift)))
5267b71995-08-09Fredrik Hübinette (Hubbe)  {
0bc4cf1998-10-13Fredrik Hübinette (Hubbe)  generic_memcpy(r,MKPCHARP(s,str->size_shift),(tmp-s)>>str->size_shift); INC_PCHARP(r,(tmp-s)>>str->size_shift); pike_string_cpy(r,to); INC_PCHARP(r,to->len);
3e625c1998-10-11Fredrik Hübinette (Hubbe)  s=tmp+(del->len << str->size_shift);
5267b71995-08-09Fredrik Hübinette (Hubbe)  }
0bc4cf1998-10-13Fredrik Hübinette (Hubbe)  generic_memcpy(r,MKPCHARP(s,str->size_shift),(end-s)>>str->size_shift);
5267b71995-08-09Fredrik Hübinette (Hubbe)  return end_shared_string(ret); }
af93211996-10-12Fredrik Hübinette (Hubbe) /*** init/exit memory ***/
be478c1997-08-30Henrik Grubbström (Grubba) void init_shared_string_table(void)
af93211996-10-12Fredrik Hübinette (Hubbe) {
85f59e1998-01-08Fredrik Hübinette (Hubbe)  for(hashprimes_entry=0;hashprimes[hashprimes_entry]<BEGIN_HASH_SIZE;hashprimes_entry++);
75920f1997-12-28Fredrik Hübinette (Hubbe)  htable_size=hashprimes[hashprimes_entry];
af93211996-10-12Fredrik Hübinette (Hubbe)  base_table=(struct pike_string **)xalloc(sizeof(struct pike_string *)*htable_size); MEMSET((char *)base_table,0,sizeof(struct pike_string *)*htable_size); }
61e9a01998-01-25Fredrik Hübinette (Hubbe) #ifdef DEBUG_MALLOC struct shared_string_location *all_shared_string_locations; #endif
be478c1997-08-30Henrik Grubbström (Grubba) void cleanup_shared_string_table(void)
5267b71995-08-09Fredrik Hübinette (Hubbe) {
af93211996-10-12Fredrik Hübinette (Hubbe)  unsigned INT32 e;
06983f1996-09-22Fredrik Hübinette (Hubbe)  struct pike_string *s,*next;
61e9a01998-01-25Fredrik Hübinette (Hubbe)  #if defined(DEBUG) && defined(DEBUG_MALLOC) while(all_shared_string_locations) { struct shared_string_location *x=all_shared_string_locations; all_shared_string_locations=x->next; free_string(x->s); x->s=0; } if(verbose_debug_exit) {
3c0c281998-01-26Fredrik Hübinette (Hubbe)  INT32 num,size; count_memory_in_strings(&num,&size); if(num) { fprintf(stderr,"Strings left: %d (%d bytes) (zapped)\n",num,size); dump_stralloc_strings(); }
61e9a01998-01-25Fredrik Hübinette (Hubbe)  } #endif
af93211996-10-12Fredrik Hübinette (Hubbe)  for(e=0;e<htable_size;e++)
5267b71995-08-09Fredrik Hübinette (Hubbe)  { for(s=base_table[e];s;s=next) { next=s->next; #ifdef REALLY_FREE free((char *)s); #else s->next=0; #endif } base_table[e]=0; }
b906cf1996-11-27Fredrik Hübinette (Hubbe)  free((char *)base_table);
61e9a01998-01-25Fredrik Hübinette (Hubbe)  base_table=0; num_strings=0;
5267b71995-08-09Fredrik Hübinette (Hubbe) }
c3c7031996-12-04Fredrik Hübinette (Hubbe)  void count_memory_in_strings(INT32 *num, INT32 *size) { unsigned INT32 e, num_=0, size_=0;
61e9a01998-01-25Fredrik Hübinette (Hubbe)  if(!base_table) { *num=*size=0; return; }
c3c7031996-12-04Fredrik Hübinette (Hubbe)  size_+=htable_size * sizeof(struct pike_string *); for(e=0;e<htable_size;e++) { struct pike_string *p; for(p=base_table[e];p;p=p->next) { num_++;
db4a401998-10-09Fredrik Hübinette (Hubbe)  size_+=sizeof(struct pike_string)+(p->len<<p->size_shift);
c3c7031996-12-04Fredrik Hübinette (Hubbe)  } } #ifdef DEBUG if(num_strings != num_)
beac721998-04-16Fredrik Hübinette (Hubbe)  fatal("Num strings is wrong! %d!=%d.\n",num_strings, num_);
c3c7031996-12-04Fredrik Hübinette (Hubbe) #endif num[0]=num_; size[0]=size_; }
9367351997-01-27Fredrik Hübinette (Hubbe) 
be478c1997-08-30Henrik Grubbström (Grubba) void gc_mark_all_strings(void)
9367351997-01-27Fredrik Hübinette (Hubbe) { unsigned INT32 e; if(!base_table) return; for(e=0;e<htable_size;e++) { struct pike_string *p; for(p=base_table[e];p;p=p->next) gc_is_referenced(p); } }
db4a401998-10-09Fredrik Hübinette (Hubbe)  void init_string_builder(struct string_builder *s, int mag) { s->malloced=256; s->s=begin_wide_shared_string(256,mag); s->s->len=0;
3e625c1998-10-11Fredrik Hübinette (Hubbe)  s->known_shift=0;
db4a401998-10-09Fredrik Hübinette (Hubbe) }
3e625c1998-10-11Fredrik Hübinette (Hubbe) static void string_build_mkspace(struct string_builder *s, int chars, int mag)
db4a401998-10-09Fredrik Hübinette (Hubbe) { if(mag > s->s->size_shift) { struct pike_string *n; int l=s->s->len+chars+s->malloced; n=begin_wide_shared_string(l,mag);
0bc4cf1998-10-13Fredrik Hübinette (Hubbe)  pike_string_cpy(MKPCHARP_STR(n),s->s);
db4a401998-10-09Fredrik Hübinette (Hubbe)  n->len=s->s->len; s->malloced=l; free((char *)s->s); s->s=n; } else if(s->s->len+chars > s->malloced) { int newlen=MAXIMUM(s->malloced*2,s->s->len+chars); s->s=(struct pike_string *)realloc((char *)s->s, sizeof(struct pike_string)+ ((newlen+1)<<s->s->size_shift)); if(!s->s) fatal("Out of memory.\n"); s->malloced=newlen; } }
3e625c1998-10-11Fredrik Hübinette (Hubbe) static void *string_builder_allocate(struct string_builder *s, int chars, int mag) { void *ret; string_build_mkspace(s,chars,mag); if(chars<0) s->known_shift=0; ret=s->s->str + (s->s->len<<s->s->size_shift); s->s->len+=chars; return ret; }
db4a401998-10-09Fredrik Hübinette (Hubbe) void string_builder_putchar(struct string_builder *s, int ch) { INT32 i; string_build_mkspace(s,1,min_magnitude(ch));
3e625c1998-10-11Fredrik Hübinette (Hubbe)  s->known_shift=MAXIMUM(min_magnitude(ch),s->known_shift);
db4a401998-10-09Fredrik Hübinette (Hubbe)  i=s->s->len++; low_set_index(s->s,i,ch); } void string_builder_binary_strcat(struct string_builder *s, char *str, INT32 len) {
0bc4cf1998-10-13Fredrik Hübinette (Hubbe)  string_build_mkspace(s,len,0);
db4a401998-10-09Fredrik Hübinette (Hubbe)  switch(s->s->size_shift) { case 0: convert_0_to_0(STR0(s->s)+s->s->len,str,len); break; case 1: convert_0_to_1(STR1(s->s)+s->s->len,str,len); break; case 2: convert_0_to_2(STR2(s->s)+s->s->len,str,len); break; default: fatal("Illegal magnitude!\n"); } s->s->len+=len; }
0bc4cf1998-10-13Fredrik Hübinette (Hubbe) void string_builder_append(struct string_builder *s, PCHARP from, INT32 len) { string_build_mkspace(s,len,from.shift); generic_memcpy(MKPCHARP_STR_OFF(s->s,s->s->len), from, len); s->s->len+=len; } void string_builder_fill(struct string_builder *s, int howmany, PCHARP from, INT32 len, INT32 offset) { INT32 tmp; #ifdef DEBUG if(len<=0) fatal("Cannot fill with zero length strings!\n"); #endif if(howmany<=0) return; if(!s->s->size_shift && len == 1 && (!from.shift || !min_magnitude(EXTRACT_PCHARP(from)))) { MEMSET(string_builder_allocate(s,howmany,0), EXTRACT_PCHARP(from), howmany); return; } string_build_mkspace(s,howmany,from.shift); tmp=MINIMUM(howmany, len - offset); generic_memcpy(MKPCHARP_STR_OFF(s->s,s->s->len), ADD_PCHARP(from,offset), tmp); s->s->len+=tmp; howmany-=tmp; if(howmany > 0) { void *new_from; PCHARP to; tmp=MINIMUM(howmany, len); to=MKPCHARP_STR_OFF(s->s,s->s->len); generic_memcpy(to,from, tmp); s->s->len+=tmp; howmany-=tmp; while(howmany > 0) { tmp=MINIMUM(len, howmany); MEMCPY(s->s->str + (s->s->len << s->s->size_shift), to.ptr, tmp << s->s->size_shift); len+=tmp; howmany-=tmp; s->s->len+=tmp; } } }
db4a401998-10-09Fredrik Hübinette (Hubbe) void string_builder_strcat(struct string_builder *s, char *str) { string_builder_binary_strcat(s,str,strlen(str)); } void string_builder_shared_strcat(struct string_builder *s, struct pike_string *str) { string_build_mkspace(s,str->len,s->s->size_shift);
0bc4cf1998-10-13Fredrik Hübinette (Hubbe)  pike_string_cpy(MKPCHARP_STR_OFF(s->s,s->s->len), str);
3e625c1998-10-11Fredrik Hübinette (Hubbe)  s->known_shift=MAXIMUM(s->known_shift,str->size_shift);
db4a401998-10-09Fredrik Hübinette (Hubbe)  s->s->len+=str->len; }
7238061998-10-11Fredrik Hübinette (Hubbe)  void reset_string_builder(struct string_builder *s) { s->known_shift=0; s->s->len=0; }
0bc4cf1998-10-13Fredrik Hübinette (Hubbe) void free_string_builder(struct string_builder *s) { free((char *)s->s); }
db4a401998-10-09Fredrik Hübinette (Hubbe) struct pike_string *finish_string_builder(struct string_builder *s) { low_set_index(s->s,s->s->len,0);
3e625c1998-10-11Fredrik Hübinette (Hubbe)  if(s->known_shift == s->s->size_shift) return low_end_shared_string(s->s);
db4a401998-10-09Fredrik Hübinette (Hubbe)  return end_shared_string(s->s); }
0bc4cf1998-10-13Fredrik Hübinette (Hubbe) PCHARP MEMCHR_PCHARP(PCHARP ptr, int chr, int len) { switch(ptr.shift) { case 0: return MKPCHARP(MEMCHR0(ptr.ptr,chr,len),0); case 1: return MKPCHARP(MEMCHR1(ptr.ptr,chr,len),1); case 2: return MKPCHARP(MEMCHR2(ptr.ptr,chr,len),2); default: fatal("Illegal shift in MEMCHR_PCHARP.\n"); } } #define DIGIT(x) (isdigit(x) ? (x) - '0' : \ islower(x) ? (x) + 10 - 'a' : (x) + 10 - 'A') #define MBASE ('z' - 'a' + 1 + 10) long STRTOL_PCHARP(PCHARP str, PCHARP *ptr, int base) { register long val; register int c; int xx, neg = 0; if (ptr) *ptr = str; if (base < 0 || base > MBASE) return 0; if (!isalnum(c = EXTRACT_PCHARP(str))) { while (ISSPACE(c)) { INC_PCHARP(str,1); c=EXTRACT_PCHARP(str); } switch (c) { case '-': neg++; case '+': /* fall-through */ INC_PCHARP(str,1); c=EXTRACT_PCHARP(str); } } if (!base) { if (c != '0') base = 10; else if (INDEX_PCHARP(str,1) == 'x' || INDEX_PCHARP(str,1) == 'X') base = 16; else base = 8; } if (!isalnum(c) || (xx = DIGIT(c)) >= base) return 0; /* no number formed */ if (base == 16 && c == '0' && isxdigit(INDEX_PCHARP(str,2)) && (INDEX_PCHARP(str,1) == 'x' || INDEX_PCHARP(str,1) == 'X')) { INC_PCHARP(str,2); c = EXTRACT_PCHARP(str); /* skip over leading "0x" or "0X" */ } val=-DIGIT(c); while(1) { INC_PCHARP(str,1); c=EXTRACT_PCHARP(str); if(!(isalnum(c) && (xx=DIGIT(c)) < base)) break; val = base * val - xx; } if (ptr) *ptr = str; return (neg ? val : -val); }