X-Git-Url: http://www.chiark.greenend.org.uk/ucgi/~ian/git?p=chiark-tcl.git;a=blobdiff_plain;f=cdb%2Fwriteable.c;h=d5c5c5ba8272c4493aa20f0d4fcd62ad1c9fd165;hp=fb7a67f3ffe64781a4b4fe4ea7506331c4abfe39;hb=1896f839d37e20a5ec0d1eecb735d6a1fee1e3d2;hpb=06eaf80321d3b1a19358927c18d0b95523b9c74b

diff --git a/cdb/writeable.c b/cdb/writeable.c
index fb7a67f..d5c5c5b 100644
--- a/cdb/writeable.c
+++ b/cdb/writeable.c
@@ -15,10 +15,6 @@ static void maybe_close(int fd) {
   if (fd>=0) close(fd);
 }
 
-#define PE(m) do{						\
-    rc= cht_posixerr(ip, errno, "failed to " m); goto x_rc;	\
-  }while(0)
-
 /*==================== Subsystems and subtypes ====================*/
 
 /*---------- Pathbuf ----------*/
@@ -27,14 +23,13 @@ typedef struct Pathbuf {
   char *buf, *sfx;
 } Pathbuf;
 
-#define MAX_SUFFIX 4
+#define MAX_SUFFIX 5
 
 static void pathbuf_init(Pathbuf *pb, const char *pathb) {
   int l= strlen(pathb);
-  pb->buf= TALLOC(l + 4);
+  pb->buf= TALLOC(l + MAX_SUFFIX + 1);
   memcpy(pb->buf, pathb, l);
   pb->sfx= pb->buf + l;
-  *pb->sfx++= '.';
 }
 static const char *pathbuf_sfx(Pathbuf *pb, const char *suffix) {
   assert(strlen(suffix) <= MAX_SUFFIX);
@@ -94,9 +89,8 @@ static void ht_maybeupdate(HashTable *ht, const char *key,
 
 static const HashValue *ht_lookup(HashTable *ht, const char *key) {
   Tcl_HashEntry *he;
-  const HashValue *htv;
   
-  he= Tcl_FindHashEntry(ht, key);
+  he= Tcl_FindHashEntry(&ht->t, key);
   if (!he) return 0;
   
   return Tcl_GetHashValue(he);
@@ -171,11 +165,13 @@ static int rw_close(Tcl_Interp *ip, Rw *rw) {
   }
 
   pathbuf_free(&rw->pbsome); pathbuf_free(&rw->pbother);
-  TFREE(rw);
   return rc;
 }
 
-static void destroy_cdbrw_idtabcb(Tcl_Interp *ip, void *rw) { rw_close(0,rw); }
+static void destroy_cdbrw_idtabcb(Tcl_Interp *ip, void *rw) {
+  rw_close(0,rw);
+  TFREE(rw);
+}
 const IdDataSpec cdbtcl_rwdatabases= {
   "cdb-rwdb", "cdb-openrwdatabases-table", destroy_cdbrw_idtabcb
 };
@@ -196,7 +192,7 @@ static int acquire_lock(Tcl_Interp *ip, Pathbuf *pb, int *lockfd_r) {
   /* Remove r where umask would remove w;
    * eg umask intending 0664 here gives 0660 */
   
-  *lockfd_r= open(pathbuf_sfx(pb,".lock"), O_RDONLY|O_CREAT, lockmode);
+  *lockfd_r= open(pathbuf_sfx(pb,".lock"), O_RDWR|O_CREAT, lockmode);
   if (*lockfd_r < 0)
     return cht_posixerr(ip, errno, "could not open/create lockfile");
 
@@ -260,7 +256,7 @@ static int readstorelogrecord(FILE *f, HashTable *ht,
   int c, rc, r;
 
   c= getc(f);
-  if (c==EOF) { if (feof(f)) return -1; return -2; }
+  if (c==EOF) { return feof(f) ? -1 : -2; }
   if (c=='\n') return -3;
   if (c!='+') return -2;
 
@@ -281,6 +277,8 @@ static int readstorelogrecord(FILE *f, HashTable *ht,
   r= fread(htv_fillptr(val), 1,vallen, f);
   if (r!=vallen) goto x2_free_keyval;
 
+  c= getc(f);  if (c!='\n') goto x2_free_keyval;
+
   rc= omitfn ? omitfn(val, ctx) : TCL_OK;
   if (rc) { assert(rc>0); TFREE(val); }
   else updatefn(ht, key, val);
@@ -303,6 +301,9 @@ static int writerecord(FILE *f, const char *key, const HashValue *val) {
   r= fwrite(val->data, 1, val->len, f);
   if (r != val->len) return -1;
 
+  r= putc('\n', f);
+  if (r==EOF) return -1;
+
   return 0;
 }
 
@@ -310,32 +311,45 @@ static int writerecord(FILE *f, const char *key, const HashValue *val) {
 
 int cht_do_cdbwr_create_empty(ClientData cd, Tcl_Interp *ip,
 			      const char *pathb) {
-  static const char *const toremoves[]= {
-    ".main", ".cdb", ".log", ".tmp", 0
-  };
+  static const char *const toremoves[]= { ".cdb", ".log", ".tmp", 0 };
 
-  Pathbuf pb;
-  int lock_fd=-1, fd=-1, rc, r;
+  Pathbuf pb, pbmain;
+  int lock_fd=-1, rc, r;
+  FILE *f= 0;
   const char *const *toremove;
+  struct stat stab;
 
   pathbuf_init(&pb, pathb);
+  pathbuf_init(&pbmain, pathb);
+
   rc= acquire_lock(ip, &pb, &lock_fd);  if (rc) goto x_rc;
-  
-  fd= open(pathbuf_sfx(&pb, ".main"), O_RDWR|O_CREAT|O_EXCL, 0666);
-  if (fd <= 0) PE("create new database file");
+
+  r= lstat(pathbuf_sfx(&pbmain, ".main"), &stab);
+  if (!r) { rc= cht_staticerr(ip, "database already exists during creation",
+			      "CDB ALREADY-EXISTS");  goto x_rc; }
+  if (errno != ENOENT) PE("check for existing database .main during creation");
 
   for (toremove=toremoves; *toremove; toremove++) {
-    r= remove(*toremove);
+    r= remove(pathbuf_sfx(&pb, *toremove));
     if (r && errno != ENOENT)
       PE("delete possible spurious file during creation");
   }
   
+  f= fopen(pathbuf_sfx(&pb, ".tmp"), "w");
+  if (!f) PE("create new database .tmp");  
+  r= putc('\n', f);  if (r==EOF) PE("write sentinel to new database .tmp");
+  r= fclose(f);  f=0;  if (r) PE("close new database .tmp during creation");
+
+  r= rename(pb.buf, pbmain.buf);
+  if (r) PE("install new database .tmp as .main (finalising creation)");
+  
   rc= TCL_OK;
 
  x_rc:
-  maybe_close(fd);
+  if (f) fclose(f);
   maybe_close(lock_fd);
   pathbuf_free(&pb);
+  pathbuf_free(&pbmain);
   return rc;
 }
 
@@ -401,6 +415,7 @@ int cht_do_cdbwr_open(ClientData cd, Tcl_Interp *ip, const char *pathb,
   off_t logrecstart, logjunkpos;
 
   rw= TALLOC(sizeof(*rw));
+  rw->ix= -1;
   ht_setup(&rw->logincore);
   cht_scriptinv_init(&rw->on_info);
   cht_scriptinv_init(&rw->on_lexminval);
@@ -409,15 +424,14 @@ int cht_do_cdbwr_open(ClientData cd, Tcl_Interp *ip, const char *pathb,
   pathbuf_init(&rw->pbother, pathb);
   rw->autocompact= 1;
 
-  if (on_lexminval) {
-    rc= cht_scriptinv_set(&rw->on_lexminval, ip, on_lexminval, 0);
-    if (rc) goto x_rc;
-  } else {
-    rw->on_lexminval.llength= 0;
-  }
+  rc= cht_scriptinv_set(&rw->on_info, ip, on_info, 0);
+  if (rc) goto x_rc;
+
+  rc= cht_scriptinv_set(&rw->on_lexminval, ip, on_lexminval, 0);
+  if (rc) goto x_rc;
 
   mainfd= open(pathbuf_sfx(&rw->pbsome,".main"), O_RDONLY);
-  if (mainfd<0) PE("open exist3ing database file .main");
+  if (mainfd<0) PE("open existing database file .main");
   rc= acquire_lock(ip, &rw->pbsome, &rw->lock_fd);  if (rc) goto x_rc;
 
   r= fstat(mainfd, &stab);  if (r) PE("fstat .main");
@@ -427,8 +441,8 @@ int cht_do_cdbwr_open(ClientData cd, Tcl_Interp *ip, const char *pathb,
   if (rw->cdb_fd >=0) {
     rc= cdbinit(ip, rw);  if (rc) goto x_rc;
   } else if (errno == ENOENT) {
-    if (rw->mainsz) {
-      rc= cht_staticerr(ip, ".cdb does not exist but .main is nonempty -"
+    if (rw->mainsz > 1) {
+      rc= cht_staticerr(ip, ".cdb does not exist but .main is >1byte -"
 			" .cdb must have been accidentally deleted!",
 			"CDB CDBMISSING");
       goto x_rc;
@@ -499,6 +513,7 @@ int cht_do_cdbwr_open(ClientData cd, Tcl_Interp *ip, const char *pathb,
 
  x_rc:
   rw_close(0,rw);
+  TFREE(rw);
   maybe_close(mainfd);
   return rc;
 }
@@ -514,16 +529,16 @@ int cht_do_cdbwr_open_okjunk(ClientData cd, Tcl_Interp *ip, const char *pathb,
 struct ht_forall_ctx {
   struct cdb_make cdbm;
   FILE *mainfile;
-  int lexminvall;
   long *reccount;
-  const char *lexminval;
+  int lexminvall;
+  const char *lexminval; /* may be invalid if lexminvall <= 0 */
 };
 
 /*---------- helper functions ----------*/
 
 static int expiredp(const HashValue *val, struct ht_forall_ctx *a) {
   int r, l;
-  if (!val->len) return 0;
+  if (!val->len || a->lexminvall<=0) return 0;
   l= val->len < a->lexminvall ? val->len : a->lexminvall;
   r= memcmp(val->data, a->lexminval, l);
   if (r>0) return 0;
@@ -583,7 +598,7 @@ static int compact_core(Tcl_Interp *ip, Rw *rw, unsigned long logsz,
 	     logsz, (unsigned long)rw->mainsz);
   if (rc) goto x_rc;
 
-  if (rw->on_lexminval.llength) {
+  if (cht_scriptinv_interp(&rw->on_lexminval)) {
     rc= cht_scriptinv_invoke_fg(&rw->on_lexminval, 0,0);
     if (rc) goto x_rc;
 
@@ -597,7 +612,7 @@ static int compact_core(Tcl_Interp *ip, Rw *rw, unsigned long logsz,
     rc= ht_forall(&rw->logincore, delete_ifexpired, &a);
 
   } else {
-    a.lexminval= "";
+    a.lexminvall= 0;
   }
 
   /* merge unsuperseded records from main into hash table */
@@ -610,19 +625,23 @@ static int compact_core(Tcl_Interp *ip, Rw *rw, unsigned long logsz,
 			  expiredp, &a,
 			  ht_maybeupdate);
     if (ferror(a.mainfile)) { rc= cht_posixerr(ip, errno, "error reading"
-                         " .main during compact"); goto x_rc;
-    }
+                         " .main during compact"); goto x_rc; }
     if (r==-3) {
       break;
     } else if (r==-1 || r==-2) {
       errpos= ftello(a.mainfile);
       if (errpos<0) PE("ftello .main during report of syntax error");
-      snprintf(buf,sizeof(buf), "CDB SYNTAX MAIN %lu", (unsigned long)errpos);
+      snprintf(buf,sizeof(buf), "CDB %s MAIN %lu",
+	       r==-1 ? "TRUNCATED" : "SYNTAX", (unsigned long)errpos);
       Tcl_SetObjErrorCode(ip, Tcl_NewStringObj(buf,-1));
       snprintf(buf,sizeof(buf), "%lu", (unsigned long)errpos);
       Tcl_ResetResult(ip);
-      Tcl_AppendResult(ip, "syntax error in .main during"
-		       " compact, at file position ", buf, (char*)0);
+      Tcl_AppendResult(ip,
+		       r==-1 ? "unexpected eof (truncated file)"
+		       " in .main during compact, at file position "
+		       : "syntax error"
+		       " in .main during compact, at file position ",
+		       buf, (char*)0);
       rc= TCL_ERROR;
       goto x_rc;
     } else {
@@ -663,6 +682,9 @@ static int compact_core(Tcl_Interp *ip, Rw *rw, unsigned long logsz,
   r= ht_forall(&rw->logincore, addto_main, &a);
   if (r) { rc= cht_posixerr(ip, r, "error writing to new .main"
 			    " during compact");  goto x_rc; }
+
+  r= putc('\n', a.mainfile);
+  if (r==EOF) PE("write trailing \n to main during compact");
   
   r= fflush(a.mainfile);  if (r) PE("fflush new main during compact");
   r= fdatasync(fileno(a.mainfile));
@@ -681,7 +703,7 @@ static int compact_core(Tcl_Interp *ip, Rw *rw, unsigned long logsz,
 
   /* done! */
   
-  rc= infocb(ip, rw, "compact-end", "main=%luby nrecs=%l",
+  rc= infocb(ip, rw, "compact-end", "main=%luby nrecs=%ld",
 	     (unsigned long)rw->mainsz, *a.reccount);
   if (rc) goto x_rc;
 
@@ -741,6 +763,7 @@ int cht_do_cdbwr_close(ClientData cd, Tcl_Interp *ip, void *rw_v) {
   if (rc_close) rc= rc_close;
   
   cht_tabledataid_disposing(ip, rw_v, &cdbtcl_rwdatabases);
+  TFREE(rw);
   return rc;
 }
 
@@ -755,7 +778,16 @@ static int compact_keepopen(Tcl_Interp *ip, Rw *rw, int force) {
   if (logsz < 0) return cht_posixerr(ip, errno, "ftell .log"
 				       " during compact check or force");
 
-  if (!force && logsz < rw->mainsz / 10 + 1000) return TCL_OK;
+  if (!force && logsz < rw->mainsz / 3 + 1000) return TCL_OK;
+  /* Test case:                    ^^^ testing best value for this
+   *   main=9690434by nrecs=122803  read all in one go
+   *  no autocompact, :    6.96user 0.68system 0:08.93elapsed
+   *  auto, mulitplier 2:  7.10user 0.79system 0:09.54elapsed
+   *  auto, unity:         7.80user 0.98system 0:11.84elapsed
+   *  auto, divisor  2:    8.23user 1.05system 0:13.30elapsed
+   *  auto, divisor  3:    8.55user 1.12system 0:12.88elapsed
+   *  auto, divisor  5:    9.95user 1.43system 0:15.72elapsed
+   */
 
   rc= compact_core(ip, rw, logsz, &reccount);  if (rc) goto x_rc;
 
@@ -820,6 +852,8 @@ static int update(Tcl_Interp *ip, Rw *rw, const char *key,
   if (r) PE("write update to logfile");
 
   ht_update(&rw->logincore, key, val);
+
+  if (!rw->autocompact) return TCL_OK;
   return compact_keepopen(ip, rw, 0);
 
  x_rc:
@@ -847,11 +881,10 @@ int cht_do_cdbwr_delete(ClientData cd, Tcl_Interp *ip, void *rw_v,
 
 /*---------- Lookups ----------*/
 
-static int wrlookup(Tcl_Interp *ip, void *rw_v, const char *key,
+static int lookup_rw(Tcl_Interp *ip, void *rw_v, const char *key,
 		    const Byte **data_r, int *len_r /* -1 => notfound */) {
   Rw *rw= rw_v;
   const HashValue *val;
-  int r, l;
 
   val= ht_lookup(&rw->logincore, key);
   if (val) {
@@ -859,61 +892,33 @@ static int wrlookup(Tcl_Interp *ip, void *rw_v, const char *key,
     else { *data_r= 0; *len_r= -1; return TCL_OK; }
   }
 
-  r= cdb_find(&rw->cdb, key, strlen(key));
-  if (!r) { *data_r= 0; *len_r= -1; return TCL_OK; }
-  if (r<0) return cht_posixerr(ip, errno, "cdb_find failed");
-  assert(r==1);
-  *len_r= cdb_datalen(&rw->cdb);
-  assert(*len_r > 0);
-  *data_r= cdb_getdata(&rw->cdb);
-  if (!*data_r) return cht_posixerr(ip, errno, "cdb_getdata failed");
-  return TCL_OK;
+  return cht_cdb_lookup_cdb(ip, &rw->cdb, key, strlen(key), data_r, len_r);
 } 
 
-int cht_cdb_dosomelookup(ClientData cd, Tcl_Interp *ip, void *db_v,
-			 int (*somelookup)(Tcl_Interp *ip, void *db_v,
-					   const char *key,
-					   const Byte **data_r, int *len_r),
-			 const char *key, Tcl_Obj *def,
-			 Tcl_Obj **result) {
-  int r, len;
+int cht_do_cdbwr_lookup(ClientData cd, Tcl_Interp *ip, void *rw_v,
+			const char *key, Tcl_Obj *def,
+			Tcl_Obj **result) {
   const Byte *data;
+  int dlen, r;
   
-  r= somelookup(ip, db_v, key, &data, &len);
-  if (r) return r;
-  if (len>0) {
-    r= storeanswer();
-  }
-  if (def) {
-    *result= def;
-    return TCL_OK;
-  }
-  return cht_staticerr(ip, "cdbwr lookup key not found", "CDB NOTFOUND");
-  
-  if (len<0) *result= def;
-  else *result= Tcl_NewStringObj(
+  r= lookup_rw(ip, rw_v, key, &data, &dlen);  if (r) return r;
+  return cht_cdb_donesomelookup(ip, rw_v, def, result, data, dlen,
+				cht_cdb_storeanswer_string);
 }
-
-static int storeanswer
   
-int cht_do_cdbwr_lookup(ClientData cd, Tcl_Interp *ip, void *rw_v,
-			const char *key, HBytes_Value def,
-			HBytes_Value *result) {
-    *result= Tcl_NewStringObj(data, len);
-    if (!*result) return cht_staticerr(ip, "Tcl_NewStringObj failed for"
-          " lookup (utf-8 encoding problem?)", "CDB BADSTRING");
-    return TCL_OK;
 int cht_do_cdbwr_lookup_hb(ClientData cd, Tcl_Interp *ip, void *rw_v,
-			   const char *key, HBytes_Value def,
-			   HBytes_Value *result) {
-  Rw *rw= rw_v;
-  int r;
+			   const char *key, Tcl_Obj *def,
+			   Tcl_Obj **result) {
+  const Byte *data;
+  int dlen, r;
+  
+  r= lookup_rw(ip, rw_v, key, &data, &dlen);  if (r) return r;
+  return cht_cdb_donesomelookup(ip, rw_v, def, result, data, dlen,
+				cht_cdb_storeanswer_hb);
+}
 
-  r= lookup(ip, rw, key, &data, &len);
-  if (r) return r;
-  if (len>0) {
-    cht_hb_array(result, data, len);
-    return TCL_OK;
-  }
-  if (!HBYTES_ISSENTINEL(def)) {
-    cht_hb_array(result, cht_hb_data(&def), cht_hb
+int cht_do_cdbtoplevel_cdb_wr(ClientData cd, Tcl_Interp *ip,
+			      const Cdbwr_SubCommand* subcmd,
+			      int objc, Tcl_Obj *const *objv) {
+  return subcmd->func((void*)subcmd,ip,objc,objv);
+}