isomorfeus-ferret 0.17.2 → 0.17.4

data/ext/isomorfeus_ferret_ext/frt_in_stream.c

@@ -1,3 +1,8 @@
+/* prevent warning: #warning Please include winsock2.h before windows.h [-Wcpp] */
+#ifdef _WIN32
+#   include <winsock2.h>
+#endif
+
 #include "brotli_decode.h"
 #include "brotli_encode.h"
 #include "bzlib.h"
@@ -12,6 +17,7 @@
  */
 FrtInStream *frt_is_new(void) {
     FrtInStream *is = FRT_ALLOC(FrtInStream);
+    is->folder_name = NULL;
     is->f = FRT_ALLOC_AND_ZERO(FrtInStreamFile);
     is->f->ref_cnt = 1;
     is->buf.start = 0;
@@ -28,8 +34,7 @@ FrtInStream *frt_is_new(void) {
  * @raise FRT_IO_ERROR if there is a error reading from the filesystem
  * @raise FRT_EOF_ERROR if there is an attempt to read past the end of the file
  */
-static void frt_is_refill(FrtInStream *is)
-{
+static void frt_is_refill(FrtInStream *is) {
     frt_off_t start = is->buf.start + is->buf.pos;
     frt_off_t last = start + FRT_BUFFER_SIZE;
     frt_off_t flen = is->m->length_i(is);
@@ -65,22 +70,18 @@ static void frt_is_refill(FrtInStream *is)
  * @raise FRT_IO_ERROR if there is a error reading from the filesystem
  * @raise FRT_EOF_ERROR if there is an attempt to read past the end of the file
  */
-frt_uchar frt_is_read_byte(FrtInStream *is)
-{
+frt_uchar frt_is_read_byte(FrtInStream *is) {
     if (is->buf.pos >= is->buf.len) {
         frt_is_refill(is);
     }
-
     return read_byte(is);
 }
 
-off_t frt_is_pos(FrtInStream *is)
-{
+off_t frt_is_pos(FrtInStream *is) {
     return is->buf.start + is->buf.pos;
 }
 
-frt_uchar *frt_is_read_bytes(FrtInStream *is, frt_uchar *buf, int len)
-{
+frt_uchar *frt_is_read_bytes(FrtInStream *is, frt_uchar *buf, int len) {
     int i;
     frt_off_t start;
 
@@ -88,8 +89,7 @@ frt_uchar *frt_is_read_bytes(FrtInStream *is, frt_uchar *buf, int len)
         for (i = 0; i < len; i++) {
             buf[i] = read_byte(is);
         }
-    }
-    else {                              /* read all-at-once */
+    } else {                              /* read all-at-once */
         start = frt_is_pos(is);
         is->m->seek_i(is, start);
         is->m->read_i(is, buf, len);
@@ -122,23 +122,27 @@ void frt_is_close(FrtInStream *is) {
             is->m->close_i(is);
             free(is->f);
         }
+        if (is->folder_name) {
+            free((void *)is->folder_name);
+        }
         free(is);
     }
 }
 
-FrtInStream *frt_is_clone(FrtInStream *is)
-{
+FrtInStream *frt_is_clone(FrtInStream *is) {
     if (!(is->f))
         return NULL;
     FrtInStream *new_is = FRT_ALLOC(FrtInStream);
     memcpy(new_is, is, sizeof(FrtInStream));
+    if (is->folder_name) {
+        new_is->folder_name = frt_estrdup(is->folder_name);
+    }
     new_is->ref_cnt = 1;
     FRT_REF(new_is->f);
     return new_is;
 }
 
-frt_i32 frt_is_read_i32(FrtInStream *is)
-{
+frt_i32 frt_is_read_i32(FrtInStream *is) {
     return ((frt_i32)frt_is_read_byte(is) << 24) |
         ((frt_i32)frt_is_read_byte(is) << 16) |
         ((frt_i32)frt_is_read_byte(is) << 8) |
@@ -146,8 +150,7 @@ frt_i32 frt_is_read_i32(FrtInStream *is)
 }
 
 
-frt_i64 frt_is_read_i64(FrtInStream *is)
-{
+frt_i64 frt_is_read_i64(FrtInStream *is) {
     return ((frt_i64)frt_is_read_byte(is) << 56) |
         ((frt_i64)frt_is_read_byte(is) << 48) |
         ((frt_i64)frt_is_read_byte(is) << 40) |
@@ -159,16 +162,14 @@ frt_i64 frt_is_read_i64(FrtInStream *is)
 }
 
 
-frt_u32 frt_is_read_u32(FrtInStream *is)
-{
+frt_u32 frt_is_read_u32(FrtInStream *is) {
     return ((frt_u32)frt_is_read_byte(is) << 24) |
         ((frt_u32)frt_is_read_byte(is) << 16) |
         ((frt_u32)frt_is_read_byte(is) << 8) |
         ((frt_u32)frt_is_read_byte(is));
 }
 
-frt_u64 frt_is_read_u64(FrtInStream *is)
-{
+frt_u64 frt_is_read_u64(FrtInStream *is) {
     return ((frt_u64)frt_is_read_byte(is) << 56) |
         ((frt_u64)frt_is_read_byte(is) << 48) |
         ((frt_u64)frt_is_read_byte(is) << 40) |
@@ -181,8 +182,7 @@ frt_u64 frt_is_read_u64(FrtInStream *is)
 
 
 /* optimized to use unchecked read_byte if there is definitely space */
-unsigned int frt_is_read_vint(FrtInStream *is)
-{
+unsigned int frt_is_read_vint(FrtInStream *is) {
     register unsigned int res, b;
     register int shift = 7;
 
@@ -195,8 +195,7 @@ unsigned int frt_is_read_vint(FrtInStream *is)
             res |= (b & 0x7F) << shift;
             shift += 7;
         }
-    }
-    else {                              /* unchecked optimization */
+    } else {                              /* unchecked optimization */
         b = read_byte(is);
         res = b & 0x7F;                 /* 0x7F = 0b01111111 */
 
@@ -212,8 +211,7 @@ unsigned int frt_is_read_vint(FrtInStream *is)
 
 
 /* optimized to use unchecked read_byte if there is definitely space */
-off_t frt_is_read_voff_t(FrtInStream *is)
-{
+off_t frt_is_read_voff_t(FrtInStream *is) {
     register frt_off_t res, b;
     register int shift = 7;
 
@@ -226,8 +224,7 @@ off_t frt_is_read_voff_t(FrtInStream *is)
             res |= (b & 0x7F) << shift;
             shift += 7;
         }
-    }
-    else {                              /* unchecked optimization */
+    } else {                              /* unchecked optimization */
         b = read_byte(is);
         res = b & 0x7F;                 /* 0x7F = 0b01111111 */
 
@@ -242,8 +239,7 @@ off_t frt_is_read_voff_t(FrtInStream *is)
 }
 
 /* optimized to use unchecked read_byte if there is definitely space */
-frt_u64 frt_is_read_vll(FrtInStream *is)
-{
+frt_u64 frt_is_read_vll(FrtInStream *is) {
     register frt_u64 res, b;
     register int shift = 7;
 
@@ -256,8 +252,7 @@ frt_u64 frt_is_read_vll(FrtInStream *is)
             res |= (b & 0x7F) << shift;
             shift += 7;
         }
-    }
-    else {                              /* unchecked optimization */
+    } else {                              /* unchecked optimization */
         b = read_byte(is);
         res = b & 0x7F;                 /* 0x7F = 0b01111111 */
 
@@ -271,16 +266,14 @@ frt_u64 frt_is_read_vll(FrtInStream *is)
     return res;
 }
 
-void frt_is_skip_vints(FrtInStream *is, register int cnt)
-{
+void frt_is_skip_vints(FrtInStream *is, register int cnt) {
     for (; cnt > 0; cnt--) {
         while ((frt_is_read_byte(is) & 0x80) != 0) {
         }
     }
 }
 
-char *frt_is_read_string(FrtInStream *is)
-{
+char *frt_is_read_string(FrtInStream *is) {
     register int length = (int) frt_is_read_vint(is);
     char *str = FRT_ALLOC_N(char, length + 1);
     str[length] = '\0';
@@ -290,8 +283,7 @@ char *frt_is_read_string(FrtInStream *is)
         for (i = 0; i < length; i++) {
             str[i] = frt_is_read_byte(is);
         }
-    }
-    else {                      /* unchecked optimization */
+    } else {                      /* unchecked optimization */
         memcpy(str, is->buf.buf + is->buf.pos, length);
         is->buf.pos += length;
     }
@@ -299,8 +291,7 @@ char *frt_is_read_string(FrtInStream *is)
     return str;
 }
 
-char *frt_is_read_string_safe(FrtInStream *is)
-{
+char *frt_is_read_string_safe(FrtInStream *is) {
     register int length = (int) frt_is_read_vint(is);
     char *str = FRT_ALLOC_N(char, length + 1);
     str[length] = '\0';
@@ -311,8 +302,7 @@ char *frt_is_read_string_safe(FrtInStream *is)
             for (i = 0; i < length; i++) {
                 str[i] = frt_is_read_byte(is);
             }
-        }
-        else {                      /* unchecked optimization */
+        } else {                      /* unchecked optimization */
             memcpy(str, is->buf.buf + is->buf.pos, length);
             is->buf.pos += length;
         }
@@ -478,8 +468,8 @@ static char *frt_is_read_lz4_compressed_bytes(FrtInStream *is, int compressed_le
     return buf_out;
 }
 
-char *frt_is_read_compressed_bytes(FrtInStream *is, int compressed_len, int *len, FrtCompressionType compression) {
-    switch (compression) {
+char *frt_is_read_compressed_bytes(FrtInStream *is, int compressed_len, int *len, FrtCompressionType compression_type) {
+    switch (compression_type) {
         case FRT_COMPRESSION_BROTLI:
             return frt_is_read_brotli_compressed_bytes(is, compressed_len, len);
         case FRT_COMPRESSION_BZ2:

data/ext/isomorfeus_ferret_ext/frt_in_stream.h

@@ -56,6 +56,7 @@ typedef struct FrtInStreamFile {
 } FrtInStreamFile;
 
 struct FrtInStream {
+    const char *folder_name;
     FrtBuffer buf;
     struct FrtInStreamFile *f;
     union {
@@ -233,8 +234,8 @@ extern char *frt_is_read_string(FrtInStream *is);
  */
 extern char *frt_is_read_string_safe(FrtInStream *is);
 
-extern char *frt_is_read_compressed_bytes(FrtInStream *is, int compressed_len, int *len, FrtCompressionType compression);
+extern char *frt_is_read_compressed_bytes(FrtInStream *is, int compressed_len, int *len, FrtCompressionType compression_type);
 
 #define frt_is_length(mis) mis->m->length_i(mis)
 
-#endif
+#endif

data/ext/isomorfeus_ferret_ext/frt_ind.c

@@ -44,6 +44,7 @@ FrtIndex *frt_index_new(FrtStore *store, FrtAnalyzer *analyzer, FrtHashSet *def_
         FRT_REF(store);
     } else {
         self->store = frt_open_ram_store(NULL);
+        self->store->create_folder(self->store, segm_idx_name);
         create = true;
     }
     if (analyzer) {
@@ -79,8 +80,7 @@ FrtIndex *frt_index_new(FrtStore *store, FrtAnalyzer *analyzer, FrtHashSet *def_
     return self;
 }
 
-void frt_index_destroy(FrtIndex *self)
-{
+void frt_index_destroy(FrtIndex *self) {
     pthread_mutex_destroy(&self->mutex);
     INDEX_CLOSE_READER(self);
     if (self->iw) frt_iw_close(self->iw);
@@ -91,8 +91,7 @@ void frt_index_destroy(FrtIndex *self)
     free(self);
 }
 
-void frt_ensure_writer_open(FrtIndex *self)
-{
+void frt_ensure_writer_open(FrtIndex *self) {
     if (!self->iw) {
         INDEX_CLOSE_READER(self);
 
@@ -102,8 +101,7 @@ void frt_ensure_writer_open(FrtIndex *self)
     }
 }
 
-void frt_ensure_reader_open(FrtIndex *self)
-{
+void frt_ensure_reader_open(FrtIndex *self) {
     if (self->ir) {
         if (self->check_latest && !frt_ir_is_latest(self->ir)) {
             INDEX_CLOSE_READER(self);
@@ -118,16 +116,14 @@ void frt_ensure_reader_open(FrtIndex *self)
     self->ir = frt_ir_open(NULL, self->store);
 }
 
-void frt_ensure_searcher_open(FrtIndex *self)
-{
+void frt_ensure_searcher_open(FrtIndex *self) {
     frt_ensure_reader_open(self);
     if (!self->sea) {
         self->sea = frt_isea_new(self->ir);
     }
 }
 
-int frt_index_size(FrtIndex *self)
-{
+int frt_index_size(FrtIndex *self) {
     int size;
     pthread_mutex_lock(&self->mutex);
     {
@@ -138,8 +134,7 @@ int frt_index_size(FrtIndex *self)
     return size;
 }
 
-bool frt_index_is_deleted(FrtIndex *self, int doc_num)
-{
+bool frt_index_is_deleted(FrtIndex *self, int doc_num) {
     bool is_del;
     pthread_mutex_lock(&self->mutex);
     {
@@ -151,8 +146,7 @@ bool frt_index_is_deleted(FrtIndex *self, int doc_num)
 }
 
 static void index_del_doc_with_key_i(FrtIndex *self, FrtDocument *doc,
-                                            FrtHashSet *key)
-{
+                                            FrtHashSet *key) {
     FrtQuery *q;
     FrtTopDocs *td;
     FrtDocField *df;
@@ -182,14 +176,13 @@ static void index_del_doc_with_key_i(FrtIndex *self, FrtDocument *doc,
         frt_td_destroy(td);
         FRT_RAISE(FRT_ARG_ERROR, "%s", NON_UNIQUE_KEY_ERROR_MSG);
     } else if (td->total_hits == 1) {
-        frt_ir_delete_doc(self->ir, td->hits[0]->doc);
+        frt_ir_delete_doc(self->ir, td->hits[0]->doc_num);
     }
     frt_q_deref(q);
     frt_td_destroy(td);
 }
 
-static void index_add_doc_i(FrtIndex *self, FrtDocument *doc)
-{
+static void index_add_doc_i(FrtIndex *self, FrtDocument *doc) {
     if (self->key) {
         index_del_doc_with_key_i(self, doc, self->key);
     }
@@ -198,8 +191,7 @@ static void index_add_doc_i(FrtIndex *self, FrtDocument *doc)
     AUTOFLUSH_IW(self);
 }
 
-void frt_index_add_doc(FrtIndex *self, FrtDocument *doc)
-{
+void frt_index_add_doc(FrtIndex *self, FrtDocument *doc) {
     pthread_mutex_lock(&self->mutex);
     {
         index_add_doc_i(self, doc);
@@ -207,8 +199,7 @@ void frt_index_add_doc(FrtIndex *self, FrtDocument *doc)
     pthread_mutex_unlock(&self->mutex);
 }
 
-FrtQuery *frt_index_get_query(FrtIndex *self, char *qstr, rb_encoding *encoding)
-{
+FrtQuery *frt_index_get_query(FrtIndex *self, char *qstr, rb_encoding *encoding) {
     int i;
     FrtFieldInfos *fis;
     frt_ensure_searcher_open(self);
@@ -221,8 +212,7 @@ FrtQuery *frt_index_get_query(FrtIndex *self, char *qstr, rb_encoding *encoding)
 
 FrtTopDocs *frt_index_search_str(FrtIndex *self, char *qstr, int first_doc,
                           int num_docs, FrtFilter *filter, FrtSort *sort,
-                          FrtPostFilter *post_filter, rb_encoding *encoding)
-{
+                          FrtPostFilter *post_filter, rb_encoding *encoding) {
     FrtQuery *query;
     FrtTopDocs *td;
     query = frt_index_get_query(self, qstr, encoding); /* will ensure_searcher is open */
@@ -232,16 +222,14 @@ FrtTopDocs *frt_index_search_str(FrtIndex *self, char *qstr, int first_doc,
     return td;
 }
 
-FrtDocument *frt_index_get_doc(FrtIndex *self, int doc_num)
-{
+FrtDocument *frt_index_get_doc(FrtIndex *self, int doc_num) {
     FrtDocument *doc;
     frt_ensure_reader_open(self);
     doc = self->ir->get_doc(self->ir, doc_num);
     return doc;
 }
 
-FrtDocument *frt_index_get_doc_ts(FrtIndex *self, int doc_num)
-{
+FrtDocument *frt_index_get_doc_ts(FrtIndex *self, int doc_num) {
     FrtDocument *doc;
     pthread_mutex_lock(&self->mutex);
     {
@@ -267,13 +255,11 @@ FrtDocument *frt_index_get_doc_term(FrtIndex *self, ID field, const char *term)
     return doc;
 }
 
-FrtDocument *frt_index_get_doc_id(FrtIndex *self, const char *id)
-{
+FrtDocument *frt_index_get_doc_num(FrtIndex *self, const char *id) {
     return frt_index_get_doc_term(self, self->id_field, id);
 }
 
-void frt_index_delete(FrtIndex *self, int doc_num)
-{
+void frt_index_delete(FrtIndex *self, int doc_num) {
     pthread_mutex_lock(&self->mutex);
     {
         frt_ensure_reader_open(self);
@@ -305,20 +291,17 @@ void frt_index_delete_term(FrtIndex *self, ID field, const char *term) {
     pthread_mutex_unlock(&self->mutex);
 }
 
-void frt_index_delete_id(FrtIndex *self, const char *id)
-{
+void frt_index_delete_id(FrtIndex *self, const char *id) {
     frt_index_delete_term(self, self->id_field, id);
 }
 
-static void index_qdel_i(FrtSearcher *sea, int doc_num, float score, void *arg)
-{
+static void index_qdel_i(FrtSearcher *sea, int doc_num, float score, void *arg) {
     (void)score; (void)arg;
     frt_ir_delete_doc(((FrtIndexSearcher *)sea)->ir, doc_num);
 }
 
 void frt_index_delete_query(FrtIndex *self, FrtQuery *q, FrtFilter *f,
-                        FrtPostFilter *post_filter)
-{
+                        FrtPostFilter *post_filter) {
     pthread_mutex_lock(&self->mutex);
     {
         frt_ensure_searcher_open(self);
@@ -329,8 +312,7 @@ void frt_index_delete_query(FrtIndex *self, FrtQuery *q, FrtFilter *f,
 }
 
 void frt_index_delete_query_str(FrtIndex *self, char *qstr, FrtFilter *f,
-                            FrtPostFilter *post_filter, rb_encoding *encoding)
-{
+                            FrtPostFilter *post_filter, rb_encoding *encoding) {
     FrtQuery *q = frt_index_get_query(self, qstr, encoding);
     frt_index_delete_query(self, q, f, post_filter);
     frt_q_deref(q);

data/ext/isomorfeus_ferret_ext/frt_ind.h

@@ -38,7 +38,7 @@ extern FrtTopDocs *frt_index_search_str(FrtIndex *self, char *query, int first_d
 extern FrtQuery *frt_index_get_query(FrtIndex *self, char *qstr, rb_encoding *encoding);
 extern FrtDocument *frt_index_get_doc(FrtIndex *self, int doc_num);
 extern FrtDocument *frt_index_get_doc_ts(FrtIndex *self, int doc_num);
-extern FrtDocument *frt_index_get_doc_id(FrtIndex *self, const char *id);
+extern FrtDocument *frt_index_get_doc_num(FrtIndex *self, const char *id);
 extern FrtDocument *frt_index_get_doc_term(FrtIndex *self, ID field, const char *term);
 extern void frt_index_delete(FrtIndex *self, int doc_num);
 extern void frt_index_delete_term(FrtIndex *self, ID field, const char *term);