pike.git/ src/ stralloc.c

1996-10-12

1996-10-12 18:04:53 by Fredrik Hübinette (Hubbe) <hubbe@hubbe.net>

• af9321075140fcfafb0eb4b1a3f62493f1aa15eb (416 lines) (+237/-179) ^{[ Show | Annotate ]}
  Branch: 7.9

shared string table now grows when needed

Rev: src/main.c:1.9
Rev: src/stralloc.c:1.5
Rev: src/stralloc.h:1.3

11:   #include "memory.h"   #include "error.h"    - static struct pike_string *base_table[HTABLE_SIZE]; + #define BEGIN_HASH_SIZE 997 + #define MAX_AVG_LINK_LENGTH 3 +  + unsigned INT32 htable_size=0; + static struct pike_string **base_table=0;   static unsigned INT32 full_hash_value; -  + unsigned INT32 num_strings=0;    - /* -  */ + /*** Main string hash function ***/   static unsigned int StrHash(const char *s,int len)   {    full_hash_value=hashmem((unsigned char *)s, len, 20); -  return full_hash_value % HTABLE_SIZE; +  return full_hash_value % htable_size;   }    - #ifdef DEBUG - void check_string(struct pike_string *s) - { -  StrHash(s->str, s->len); -  if(full_hash_value != s->hval) -  fatal("Hash value changed?\n"); +     -  if(debug_findstring(s) !=s) -  fatal("Shared string not shared.\n"); -  -  if(s->str[s->len]) -  fatal("Shared string is not zero terminated properly.\n"); - } -  - void verify_shared_strings_tables() - { -  unsigned int e, h; -  struct pike_string *s; -  -  for(e=0;e<HTABLE_SIZE;e++) -  { -  h=0; -  for(s=base_table[e];s;s=s->next) -  { -  h++; -  if(s->len < 0) -  fatal("Shared string shorter than zero bytes.\n"); -  -  if(s->refs <= 0) -  fatal("Shared string had too few references.\n"); -  -  if(s->str[s->len]) -  fatal("Shared string didn't end with a zero.\n"); -  -  if(StrHash(s->str, s->len) != e) -  fatal("Shared string hashed to wrong place.\n"); -  -  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; -  } -  } -  } - } - #endif -  - /* -  * find a string in the shared string table. -  */ + /*** find a string in the shared string table. ***/   static struct pike_string *internal_findstring(const char *s,int len,int h)   {    struct pike_string *curr,**prev, **base;
137:    return 0; /* not found */   }    - #ifdef DEBUG - struct pike_string *debug_findstring(const struct pike_string *foo) - { -  struct pike_string *tmp; -  tmp=propagate_shared_string(foo, foo->hval % HTABLE_SIZE); + /*** rehash ***/    - #if 0 -  if(!tmp) + static void rehash_string_backwards(struct pike_string *s)   { -  int 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); +  int h; +  if(!s) return; +  rehash_string_backwards(s->next); +  h=s->hval % htable_size; +  s->next=base_table[h]; +  base_table[h]=s;   } -  fprintf(stderr,"0\n"); +     -  for(e=0;e<HTABLE_SIZE;e++) + static void rehash()   { -  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)); +  int h,old; +  struct pike_string **old_base; +  +  old=htable_size; +  old_base=base_table; +  +  htable_size=htable_size*2 +1; +  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);   } -  } -  } - #endif -  return tmp; - } - #endif +     -  +  + /*** Make new strings ***/ +    /* note that begin_shared_string expects the _exact_ size of the string,    * not the maximum size    */
194:    return t;   }    + 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(); + } +    struct pike_string *end_shared_string(struct pike_string *s)   {    int len,h;
208:    free((char *)s);    s=s2;    }else{ -  s->refs = 0; -  s->next = base_table[h]; -  base_table[h] = s; -  s->hval=full_hash_value; +  link_pike_string(s, h);    }    s->refs++;   
228:    {    s=begin_shared_string(len);    MEMCPY(s->str, str, len); -  s->str[len] = 0; -  s->refs = 0; -  s->next = base_table[h]; -  base_table[h] = s; -  s->hval=full_hash_value; +  link_pike_string(s, h);    }       s->refs++;
245:    return make_shared_binary_string(str, strlen(str));   }    - /* 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); -  } - } + /*** Free strings ***/    - #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 */ - int my_quick_strcmp(struct pike_string *a,struct pike_string *b) - { -  if(a==b) return 0; -  -  return low_quick_binary_strcmp(a->str,a->len,b->str,b->len); - } -  - /* Does take locale into account */ - int my_strcmp(struct pike_string *a,struct pike_string *b) - { -  if(a==b) return 0; -  -  return low_binary_strcmp(a->str,a->len,b->str,b->len); - } -  +    void unlink_pike_string(struct pike_string *s)   {    int h;
325:    free((char *)s);   }    +    /* -  * +  * String table status    */   struct pike_string *add_string_status(int verbose)   {
341:    int num_distinct_strings=0;    int bytes_distinct_strings=0;    int overhead_bytes=0; -  int e; +  unsigned INT32 e;    struct pike_string *p; -  for(e=0;e<HTABLE_SIZE;e++) +  for(e=0;e<htable_size;e++)    {    for(p=base_table[e];p;p=p->next)    {
378:    return free_buf();   }    + /*** DEBUG ***/ + #ifdef DEBUG +  + void check_string(struct pike_string *s) + { +  StrHash(s->str, s->len); +  if(full_hash_value != s->hval) +  fatal("Hash value changed?\n"); +  +  if(debug_findstring(s) !=s) +  fatal("Shared string not shared.\n"); +  +  if(s->str[s->len]) +  fatal("Shared string is not zero terminated properly.\n"); + } +  + void verify_shared_strings_tables() + { +  unsigned INT32 e, h; +  struct pike_string *s; +  +  for(e=0;e<htable_size;e++) +  { +  h=0; +  for(s=base_table[e];s;s=s->next) +  { +  h++; +  if(s->len < 0) +  fatal("Shared string shorter than zero bytes.\n"); +  +  if(s->refs <= 0) +  fatal("Shared string had too few references.\n"); +  +  if(s->str[s->len]) +  fatal("Shared string didn't end with a zero.\n"); +  +  if(StrHash(s->str, s->len) != e) +  fatal("Shared string hashed to wrong place.\n"); +  +  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; +  } +  } +  } + } +  + 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; + } +    void dump_stralloc_strings()   { -  int e; +  unsigned INT32 e;    struct pike_string *p; -  for(e=0;e<HTABLE_SIZE;e++) +  for(e=0;e<htable_size;e++)    for(p=base_table[e];p;p=p->next)    printf("%ld refs \"%s\"\n",(long)p->refs,p->str);   }    -  + #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); +  } + } +  + #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 */ + int my_quick_strcmp(struct pike_string *a,struct pike_string *b) + { +  if(a==b) return 0; +  +  return low_quick_binary_strcmp(a->str,a->len,b->str,b->len); + } +  + /* Does take locale into account */ + int my_strcmp(struct pike_string *a,struct pike_string *b) + { +  if(a==b) return 0; +  +  return low_binary_strcmp(a->str,a->len,b->str,b->len); + } +  + /*** Add strings ***/   struct pike_string *add_shared_strings(struct pike_string *a,    struct pike_string *b)   {
405:    return ret;   }    + /*** replace function ***/   struct pike_string *string_replace(struct pike_string *str,    struct pike_string *del,    struct pike_string *to)
449:    return end_shared_string(ret);   }    + /*** init/exit memory ***/ + void init_shared_string_table() + { +  htable_size=BEGIN_HASH_SIZE; +  base_table=(struct pike_string **)xalloc(sizeof(struct pike_string *)*htable_size); +  MEMSET((char *)base_table,0,sizeof(struct pike_string *)*htable_size); + } +    void cleanup_shared_string_table()   { -  int e; +  unsigned INT32 e;    struct pike_string *s,*next; -  for(e=0;e<HTABLE_SIZE;e++) +  for(e=0;e<htable_size;e++)    {    for(s=base_table[e];s;s=next)    {