ferret 0.11.2 → 0.11.3

data/ext/analysis.c

@@ -925,9 +925,6 @@ static Token *std_next(TokenStream *ts)
         }
         ts->t = t + len;
         token[len] = 0;
-        Token *k = tk_set(&(CTS(ts)->token), token, len, (int)(start - ts->text),
-               (int)(ts->t - ts->text), 1);
-        return k;
         return tk_set(&(CTS(ts)->token), token, len, (int)(start - ts->text),
                (int)(ts->t - ts->text), 1);
     }

data/ext/compound_io.c

@@ -216,11 +216,10 @@ Store *open_cmpd_store(Store *store, const char *name)
     char *fname;
     FileEntry *entry = NULL;
     Store *new_store = NULL;
-    CompoundStore *cmpd = NULL;
-    InStream *is = NULL;
+    CompoundStore *volatile cmpd = NULL;
+    InStream *volatile is = NULL;
 
     TRY
-        new_store = store_new();
         cmpd = ALLOC_AND_ZERO(CompoundStore);
 
         cmpd->store       = store;
@@ -245,7 +244,6 @@ Store *open_cmpd_store(Store *store, const char *name)
             h_set(cmpd->entries, fname, entry);
         }
     XCATCHALL
-        store_destroy(new_store);
         if (is) is_close(is);
         if (cmpd->entries) h_destroy(cmpd->entries);
         free(cmpd);
@@ -256,6 +254,7 @@ Store *open_cmpd_store(Store *store, const char *name)
         entry->length = is_length(is) - entry->offset;
     }
 
+    new_store               = store_new();
     new_store->dir.cmpd     = cmpd;
     new_store->touch        = &cmpd_touch;
     new_store->exists       = &cmpd_exists;

data/ext/ferret.c

@@ -65,7 +65,8 @@ void
 object_add2(void *key, VALUE obj, const char *file, int line)
 {
     if (h_get(object_map, key))
-        printf("failed adding %d. %s:%d\n", (int)key, file, line);
+        printf("failed adding %x to %d; already contains %x. %s:%d\n",
+               (int)obj, (int)key, (int)h_get(object_map, key), file, line);
     //printf("adding %d. now contains %d %s:%d\n", (int)key, ++hash_cnt, file, line);
     h_set(object_map, key, (void *)obj);
 }

data/ext/index.c

@@ -6,6 +6,7 @@
 #include <string.h>
 #include <limits.h>
 #include <ctype.h>
+#include <unistd.h>
 
 #define GET_LOCK(lock, name, store, err_msg) do {\
     lock = store->open_lock(store, name);\
@@ -850,15 +851,9 @@ void sis_find_segments_file(Store *store, FindSegmentsFile *fsf,
                         is_close(gen_is);
                     XENDTRY
                     /* TODO:LOG "fallback check: " + gen0 + "; " + gen1 */
-                    if (gen0 == gen1) {
+                    if (gen0 == gen1 && gen0 > gen) {
                         /* The file is consistent. */
-                        if (gen0 > gen) {
-                            /* TODO:LOG "fallback to '" +
-                             * IndexFileNames.SEGMENTS_GEN + "' check: now
-                             * try generation " + gen0 + " > " + gen */
-                            gen = gen0;
-                        }
-                        goto method_two_loop_end;
+                        gen = gen0;
                     }
                     break;
                 }
@@ -866,7 +861,6 @@ void sis_find_segments_file(Store *store, FindSegmentsFile *fsf,
                 micro_sleep(50000);
             }
         }
-method_two_loop_end:
 
         /* Method 3 (fallback if Methods 2 & 3 are not reliable): since both
          * directory cache and file contents cache seem to be stale, just
@@ -888,9 +882,11 @@ method_two_loop_end:
                 /* OK, we've tried the same segments_N file twice in a row, so
                  * this must be a real error.  We throw the original exception
                  * we got. */
-                RAISE(IO_ERROR, "Error reading the segment infos");
+                RAISE(IO_ERROR,
+                      "Error reading the segment infos. Store listing was\n");
             }
             else {
+                micro_sleep(50000);
                 retry = true;
             }
         }
@@ -905,8 +901,34 @@ method_two_loop_end:
             run(store, fsf);
             RETURN_EARLY();
             return;
-        case IO_ERROR: case FILE_NOT_FOUND_ERROR:
+        case IO_ERROR: case FILE_NOT_FOUND_ERROR: case EOF_ERROR:
             HANDLED();
+            /*
+            if (gen != sis_current_segment_generation(store)) {
+                fprintf(stderr, "%lld != %lld\n",
+                        gen, sis_current_segment_generation(store));
+                fprintf(stderr, "%s\n", xcontext.msg);
+            }
+            else {
+                char *sl = store_to_s(store);
+                bool done = false;
+                fprintf(stderr, "%s\n>>>\n%s", xcontext.msg, sl);
+                free(sl);
+                while (!done) {
+                    TRY
+                        sis_put(sis_read(store), stderr);
+                        done = true;
+                    XCATCHALL
+                        HANDLED();
+                    XENDTRY
+                }
+            }
+
+            char *sl = store_to_s(store);
+            fprintf(stderr, "%s\n>>>\n%s", xcontext.msg, sl);
+            free(sl);
+            */
+
             /* Save the original root cause: */
             /* TODO:LOG "primary Exception on '" + segmentFileName + "': " +
              * err + "'; will retry: retry=" + retry + "; gen = " + gen */
@@ -931,7 +953,7 @@ method_two_loop_end:
                         RETURN_EARLY();
                         RETURN_EARLY();
                         return;
-                    case IO_ERROR: case FILE_NOT_FOUND_ERROR:
+                    case IO_ERROR: case FILE_NOT_FOUND_ERROR: case EOF_ERROR:
                         HANDLED();
                         /* TODO:LOG "secondary Exception on '" +
                          * prev_seg_file_name + "': " + err2 + "'; will retry"*/
@@ -1022,9 +1044,10 @@ void sis_read_i(Store *store, FindSegmentsFile *fsf)
     int i;
     bool success = false;
     char seg_file_name[SEGMENT_NAME_MAX_LENGTH];
-    InStream *is = NULL;
-    SegmentInfos *sis = ALLOC_AND_ZERO(SegmentInfos);
+    InStream *volatile is = NULL;
+    SegmentInfos *volatile sis = ALLOC_AND_ZERO(SegmentInfos);
     segfn_for_generation(seg_file_name, fsf->generation);
+    fsf->p_return = NULL;
     TRY
         is = store->open_input(store, seg_file_name);
         sis->store = store;
@@ -3727,10 +3750,6 @@ void ir_commit_i(IndexReader *ir)
             char curr_seg_fn[MAX_FILE_PATH];
             mutex_lock(&ir->store->mutex);
 
-            /* Should not be necessary: no prior commit should have left
-             * pending files, so just defensive: */
-            if (ir->deleter) deleter_clear_pending_deletions(ir->deleter);
-
             sis_curr_seg_file_name(curr_seg_fn, ir->store);
             
             ir->commit_i(ir);
@@ -4696,8 +4715,8 @@ IndexReader *mr_open(IndexReader **sub_readers, const int r_cnt)
 static void ir_open_i(Store *store, FindSegmentsFile *fsf)
 {
     volatile bool success = false;
-    IndexReader *ir = NULL;
-    SegmentInfos *sis = NULL;
+    IndexReader *volatile ir = NULL;
+    SegmentInfos *volatile sis = NULL;
     TRY
     do {
         FieldInfos *fis;
@@ -5820,7 +5839,6 @@ static void iw_commit_compound_file(IndexWriter *iw, SegmentInfo *si)
     sprintf(cfs_name, "%s.cfs", si->name);
 
     iw_create_compound_file(iw->store, iw->fis, si, cfs_name, iw->deleter);
-    deleter_commit_pending_deletions(iw->deleter);
 }
 
 static void iw_merge_segments(IndexWriter *iw, const int min_seg,
@@ -5841,7 +5859,6 @@ static void iw_merge_segments(IndexWriter *iw, const int min_seg,
     for (i = min_seg; i < max_seg; i++) {
         si_delete_files(sis->segs[i], iw->fis, iw->deleter);
     }
-    deleter_commit_pending_deletions(iw->deleter);
 
     sis_del_from_to(sis, min_seg, max_seg);
 
@@ -5851,6 +5868,7 @@ static void iw_merge_segments(IndexWriter *iw, const int min_seg,
     }
 
     sis_write(sis, iw->store, iw->deleter);
+    deleter_commit_pending_deletions(iw->deleter);
 
     mutex_unlock(&iw->store->mutex);
 
@@ -5910,6 +5928,7 @@ static void iw_flush_ram_segment(IndexWriter *iw)
     }
     /* commit the segments file and the fields file */
     sis_write(iw->sis, iw->store, iw->deleter);
+    deleter_commit_pending_deletions(iw->deleter);
 
     mutex_unlock(&iw->store->mutex);
 

data/ext/r_analysis.c

@@ -40,6 +40,7 @@ static VALUE cTokenStream;
 static ID id_next;
 static ID id_reset;
 static ID id_clone;
+static ID id_text;
 
 /* Analyzer Methods */
 static ID id_token_stream;
@@ -450,7 +451,9 @@ frt_ts_set_text(VALUE self, VALUE rtext)
     Data_Get_Struct(self, TokenStream, ts);
     StringValue(rtext);
     ts->reset(ts, rs2s(rtext));
-    object_set(&ts->text, rtext);
+    
+    /* prevent garbage collection */
+    rb_ivar_set(self, id_text, rtext);
 
     return rtext;
 }
@@ -538,7 +541,10 @@ typedef struct CWrappedTokenStream {
 static void
 cwrts_destroy_i(TokenStream *ts)
 {
-    rb_hash_delete(object_space, ((long)ts)|1);
+    if (object_get(&ts->text) != Qnil) {
+        object_del(&ts->text);
+    }
+    rb_hash_delete(object_space, ((VALUE)ts)|1);
     /*printf("rb_hash_size = %d\n", frt_rb_hash_size(object_space)); */
     free(ts);
 }
@@ -554,6 +560,7 @@ static TokenStream *
 cwrts_reset(TokenStream *ts, char *text)
 {
     ts->t = ts->text = text;
+    Xj
     rb_funcall(CWTS(ts)->rts, id_reset, 1, rb_str_new2(text));
     return ts;
 }
@@ -563,7 +570,7 @@ cwrts_clone_i(TokenStream *orig_ts)
 {
     TokenStream *new_ts = ts_clone_size(orig_ts, sizeof(CWrappedTokenStream));
     VALUE rts = CWTS(new_ts)->rts = rb_funcall(CWTS(orig_ts)->rts, id_clone, 0);
-    rb_hash_aset(object_space, ((long)new_ts)|1, rts);
+    rb_hash_aset(object_space, ((VALUE)new_ts)|1, rts);
     return new_ts;
 }
 
@@ -583,7 +590,7 @@ frt_get_cwrapped_rts(VALUE rts)
         ts->clone_i = &cwrts_clone_i;
         ts->destroy_i = &cwrts_destroy_i;
         /* prevent from being garbage collected */
-        rb_hash_aset(object_space, ((long)ts)|1, rts);
+        rb_hash_aset(object_space, ((VALUE)ts)|1, rts);
         ts->ref_cnt = 1;
     }
     return ts;
@@ -621,7 +628,10 @@ typedef struct RegExpTokenStream {
 static void
 rets_destroy_i(TokenStream *ts)
 {
-    rb_hash_delete(object_space, ((long)ts)|1);
+    if (object_get(&ts->text) != Qnil) {
+        object_del(&ts->text);
+    }
+    rb_hash_delete(object_space, ((VALUE)ts)|1);
     /*printf("rb_hash_size = %d\n", frt_rb_hash_size(object_space)); */
     free(ts);
 }
@@ -658,7 +668,7 @@ frt_rets_set_text(VALUE self, VALUE rtext)
     TokenStream *ts;
     GET_TS(ts, self);
 
-    rb_hash_aset(object_space, ((long)ts)|1, rtext);
+    rb_hash_aset(object_space, ((VALUE)ts)|1, rtext);
     StringValue(rtext);
     RETS(ts)->rtext = rtext;
     RETS(ts)->curr_ind = 0;
@@ -730,7 +740,7 @@ rets_new(VALUE rtext, VALUE regex, VALUE proc)
 
     if (rtext != Qnil) {
         rtext = StringValue(rtext);
-        rb_hash_aset(object_space, ((long)ts)|1, rtext);
+        rb_hash_aset(object_space, ((VALUE)ts)|1, rtext);
     }
     ts->reset = &rets_reset;
     ts->next = &rets_next;
@@ -1121,7 +1131,7 @@ typedef struct CWrappedAnalyzer
 static void
 cwa_destroy_i(Analyzer *a)
 {
-    rb_hash_delete(object_space, ((long)a)|1);
+    rb_hash_delete(object_space, ((VALUE)a)|1);
     /*printf("rb_hash_size = %d\n", frt_rb_hash_size(object_space)); */
     free(a);
 }
@@ -1149,7 +1159,7 @@ frt_get_cwrapped_analyzer(VALUE ranalyzer)
         a->ref_cnt   = 1;
         ((CWrappedAnalyzer *)a)->ranalyzer = ranalyzer;
         /* prevent from being garbage collected */
-        rb_hash_aset(object_space, ((long)a)|1, ranalyzer);
+        rb_hash_aset(object_space, ((VALUE)a)|1, ranalyzer);
     }
     return a;
 }
@@ -1509,11 +1519,11 @@ frt_re_analyzer_token_stream(VALUE self, VALUE rfield, VALUE rtext)
     object_set(&ts->text, rtext);
     if (ts->next == &rets_next) {
         RETS(ts)->rtext = rtext;
-        rb_hash_aset(object_space, ((long)ts)|1, rtext);
+        rb_hash_aset(object_space, ((VALUE)ts)|1, rtext);
     }
     else {
         RETS(((TokenFilter*)ts)->sub_ts)->rtext = rtext;
-        rb_hash_aset(object_space, ((long)((TokenFilter*)ts)->sub_ts)|1, rtext);
+        rb_hash_aset(object_space, ((VALUE)((TokenFilter*)ts)->sub_ts)|1, rtext);
     }
     return get_rb_token_stream(ts);
 }
@@ -2363,6 +2373,7 @@ Init_Analysis(void)
     id_next = rb_intern("next");
     id_reset = rb_intern("text=");
     id_clone = rb_intern("clone");
+    id_text = rb_intern("@text");
 
     /* Analyzer Methods */
     id_token_stream = rb_intern("token_stream");

data/ext/search.c

@@ -1040,7 +1040,7 @@ static TopDocs *isea_search_w(Searcher *self,
     sea_check_args(num_docs, first_doc);
 
     scorer = weight->scorer(weight, ISEA(self)->ir);
-    if (!scorer) {
+    if (!scorer || 0 == ISEA(self)->ir->num_docs(ISEA(self)->ir)) {
         return td_new(0, 0, NULL, 0.0);
     }
 

data/lib/ferret_version.rb

@@ -1,3 +1,3 @@
 module Ferret
-  VERSION = '0.11.2'
+  VERSION = '0.11.3'
 end

metadata

@@ -3,8 +3,8 @@ rubygems_version: 0.9.0
 specification_version: 1
 name: ferret
 version: !ruby/object:Gem::Version 
-  version: 0.11.2
-date: 2007-02-28 00:00:00 +11:00
+  version: 0.11.3
+date: 2007-03-03 00:00:00 +11:00
 summary: Ruby indexing library.
 require_paths: 
 - lib
@@ -36,6 +36,7 @@ files:
 - TODO
 - MIT-LICENSE
 - README
+- ext/q_multi_term.c
 - ext/r_qparser.c
 - ext/r_utils.c
 - ext/r_analysis.c
@@ -44,7 +45,6 @@ files:
 - ext/r_index.c
 - ext/ferret.h
 - ext/r_store.c
-- ext/q_multi_term.c
 - ext/hashset.c
 - ext/q_match_all.c
 - ext/bitvector.c