isomorfeus-ferret 0.12.0

data/ext/isomorfeus_ferret_ext/frt_except.h

@@ -0,0 +1,188 @@
+/**
+ * Exception Handling Framework
+ *
+ * Exception Handling looks something like this;
+ *
+ * ### NOTE ###
+ * You must use a FRT_FINALLY block if you use "default:" block. Otherwise the
+ * default: block will get called in place of the FRT_FINALLY block.
+ *
+ *
+ * <pre>
+ *   FRT_TRY
+ *       FRT_RAISE(FRT_EXCEPTION, msg1);
+ *       break;
+ *   case FRT_EXCEPTION:
+ *       // This should be called
+ *       exception_handled = true;
+ *       FRT_HANDLED();
+ *       break;
+ *   default:
+ *       // shouldn't enter here
+ *       break;
+ *   FRT_XFINALLY
+ *       // this code will always be run
+ *       if (close_widget_one(arg) == 0) {
+ *           FRT_RAISE(FRT_EXCEPTION_CODE, msg);
+ *       }
+ *       // this code will also always run, even if the above exception is
+ *       // raised
+ *       if (close_widget_two(arg) == 0) {
+ *           FRT_RAISE(FRT_EXCEPTION_CODE, msg);
+ *       }
+ *   FRT_XENDTRY
+ * </pre>
+ *
+ * Basically exception handling uses the following macros;
+ *
+ * FRT_TRY
+ *   Sets up the exception handler and need be placed before any expected
+ *   exceptions would be raised.
+ *
+ * case <EXCEPTION_CODE>:
+ *   Internally the exception handling uses a switch statement so use the case
+ *   statement with the appropriate error code to catch Exceptions. Hence, if
+ *   you want to catch all exceptions, use the default keyword.
+ *
+ * FRT_HANDLED
+ *   If you catch and handle an exception you need to explicitely call
+ *   FRT_HANDLED(); or the exeption will be re-raised once the current exception
+ *   handling context is left.
+ *
+ * case FRT_FINALLY:
+ *   Code in this block is always called. Use this block to close any
+ *   resources opened in the Exception handling body.
+ *
+ * FRT_ENDTRY
+ *   Must be placed at the end of all exception handling code.
+ *
+ * FRT_XFINALLY
+ *   Similar to case FRT_FINALLY: except that it uses a fall through (ie, you must
+ *   not use a break before it) instead of a jump to get to it. This saves a
+ *   jump. It must be used in combination with FRT_XENDTRY and must not have any
+ *   other catches. This is an optimization so should probably be not be used
+ *   in most cases.
+ *
+ * FRT_XCATCHALL
+ *   Like FRT_XFINALLY but the block is only called when an exception is raised.
+ *   Must use in combination with FRT_XENDTRY and do not have any other FRT_FINALLY or
+ *   catch block.
+ *
+ * FRT_XENDTRY
+ *   Must use in combination with FRT_XFINALLY or FRT_XCATCHALL. Simply, it doesn't
+ *   jump to FRT_FINALLY, making it more efficient.
+ */
+#ifndef FRT_EXCEPT_H
+#define FRT_EXCEPT_H
+
+#include <setjmp.h>
+#include "frt_config.h"
+
+/* TODO make this an enum */
+#define FRT_BODY 0
+#define FRT_FINALLY 1
+#define FRT_EXCEPTION 2
+#define FRT_IO_ERROR 3
+#define FRT_FILE_NOT_FOUND_ERROR 4
+#define FRT_ARG_ERROR 5
+#define FRT_EOF_ERROR 6
+#define FRT_UNSUPPORTED_ERROR 7
+#define FRT_STATE_ERROR 8
+#define FRT_PARSE_ERROR 9
+#define FRT_MEM_ERROR 10
+#define FRT_INDEX_ERROR 11
+#define FRT_LOCK_ERROR 12
+
+#define FRT_EM_EXCEPTION "Exception"
+#define FRT_EM_IO_ERROR "IO Error"
+#define FRT_EM_FILE_NOT_FOUND_ERROR "File Not Found Error"
+#define FRT_EM_ARG_ERROR "Argument Error"
+#define FRT_EM_EOF_ERROR "End-of-File Error"
+#define FRT_EM_UNSUPPORTED_ERROR "Unsupported Function Error"
+#define FRT_EM_STATE_ERROR "State Error"
+#define FRT_EM_PARSE_ERROR "ParseError"
+#define FRT_EM_MEM_ERROR "Memory Error"
+#define FRT_EM_INDEX_ERROR "Index Error"
+#define FRT_EM_LOCK_ERROR "Lock Error"
+
+extern const char *const ERROR_TYPES[];
+extern const char *const FRT_UNSUPPORTED_ERROR_MSG;
+extern const char *const FRT_EOF_ERROR_MSG;
+extern const char *frt_err_code_to_type(const int err_code);
+
+extern void frb_rb_raise(const char *file, int line_num, const char *func, const char *err_type, const char *fmt, ...);
+
+typedef struct frt_xcontext_t
+{
+    jmp_buf jbuf;
+    struct frt_xcontext_t *next;
+    const char *msg;
+    int excode;
+    unsigned int handled : 1;
+    unsigned int in_finally : 1;
+} frt_xcontext_t;
+
+#define FRT_TRY\
+  do {\
+    frt_xcontext_t xcontext;\
+    xcontext.excode = 0;\
+    xcontext.msg = NULL;\
+    frt_xpush_context(&xcontext);\
+    switch (setjmp(xcontext.jbuf)) {\
+      case FRT_BODY:
+
+
+#define FRT_XENDTRY\
+    }\
+    frt_xpop_context();\
+  } while (0);
+
+#define FRT_ENDTRY\
+    }\
+    if (!xcontext.in_finally) {\
+      frt_xpop_context();\
+      xcontext.in_finally = 1;\
+      longjmp(xcontext.jbuf, FRT_FINALLY);\
+    }\
+  } while (0);
+
+#define FRT_RETURN_EARLY() frt_xpop_context()
+
+#define FRT_XFINALLY default: xcontext.in_finally = 1;
+
+#define FRT_XCATCHALL break; default: xcontext.in_finally = 1;
+
+#define FRT_HANDLED() xcontext.handled = 1; /* true */
+
+#define FRT_XMSG_BUFFER_SIZE 2048
+#define FRT_XMSG_BUFFER_FINAL_SIZE 2248
+
+#ifdef FRT_HAS_ISO_VARARGS
+# define FRT_RAISE(excode, ...) do {\
+  snprintf(frt_xmsg_buffer, FRT_XMSG_BUFFER_SIZE, __VA_ARGS__);\
+  snprintf(frt_xmsg_buffer_final, FRT_XMSG_BUFFER_FINAL_SIZE,\
+          "Error occurred in %s:%d - %s\n\t%s",\
+          __FILE__, __LINE__, __func__, frt_xmsg_buffer);\
+  frt_xraise(excode, frt_xmsg_buffer_final);\
+} while (0)
+#elif defined(FRT_HAS_GNUC_VARARGS)
+# define FRT_RAISE(excode, args...) do {\
+  snprintf(frt_xmsg_buffer, FRT_XMSG_BUFFER_SIZE, ##args);\
+  snprintf(frt_xmsg_buffer_final, FRT_XMSG_BUFFER_FINAL_SIZE,\
+          "Error occurred in %s:%d - %s\n\t%s\n",\
+          __FILE__, __LINE__, __func__, frt_xmsg_buffer);\
+  frt_xraise(excode, frt_xmsg_buffer_final);\
+} while (0)
+
+#else
+extern void FRT_RAISE(int excode, const char *fmt, ...);
+#endif
+
+extern void frt_xraise(int excode, const char *const msg);
+extern void frt_xpush_context(frt_xcontext_t *context);
+extern void frt_xpop_context();
+
+extern char frt_xmsg_buffer[FRT_XMSG_BUFFER_SIZE];
+extern char frt_xmsg_buffer_final[FRT_XMSG_BUFFER_FINAL_SIZE];
+
+#endif

data/ext/isomorfeus_ferret_ext/frt_field_index.c

@@ -0,0 +1,233 @@
+#include <string.h>
+#include "frt_field_index.h"
+
+/***************************************************************************
+ *
+ * FrtFieldIndex
+ *
+ ***************************************************************************/
+
+static unsigned long long field_index_hash(const void *p)
+{
+    FrtFieldIndex *self = (FrtFieldIndex *)p;
+    return frt_str_hash(rb_id2name(self->field)) ^ (unsigned long long)(self->klass);
+}
+
+static int field_index_eq(const void *p1, const void *p2)
+{
+    FrtFieldIndex *fi1 = (FrtFieldIndex *)p1;
+    FrtFieldIndex *fi2 = (FrtFieldIndex *)p2;
+    return (fi1->field == fi2->field) &&
+        (fi1->klass->type == fi2->klass->type);
+}
+
+static void field_index_destroy(void *p)
+{
+    FrtFieldIndex *self = (FrtFieldIndex *)p;
+    if (self->index) {
+        self->klass->destroy_index(self->index);
+    }
+    free(self);
+}
+
+FrtFieldIndex *frt_field_index_get(FrtIndexReader *ir, FrtSymbol field,
+                            const FrtFieldIndexClass *klass)
+{
+    int length = 0;
+    FrtTermEnum *volatile te = NULL;
+    FrtTermDocEnum *volatile tde = NULL;
+    FrtFieldInfo *fi = frt_fis_get_field(ir->fis, field);
+    const volatile int field_num = fi ? fi->number : -1;
+    FrtFieldIndex *volatile self = NULL;
+    FrtFieldIndex key;
+
+    if (field_num < 0) {
+        FRT_RAISE(FRT_ARG_ERROR,
+              "Cannot sort by field \"%s\". It doesn't exist in the index.",
+              rb_id2name(field));
+    }
+
+    if (!ir->field_index_cache) {
+        ir->field_index_cache = frt_h_new(&field_index_hash, &field_index_eq,
+                                      NULL, &field_index_destroy);
+    }
+
+    key.field = field;
+    key.klass = klass;
+    self = (FrtFieldIndex *)frt_h_get(ir->field_index_cache, &key);
+
+    if (self == NULL) {
+        self = FRT_ALLOC(FrtFieldIndex);
+        self->klass = klass;
+        /* FieldIndex only lives as long as the IndexReader lives so we can
+         * just use the field_infos field symbol */
+        self->field = fi->name;
+
+        length = ir->max_doc(ir);
+        if (length > 0) {
+            FRT_TRY
+            {
+                void *index;
+                tde = ir->term_docs(ir);
+                te = ir->terms(ir, field_num);
+                index = self->index = klass->create_index(length);
+                while (te->next(te)) {
+                    tde->seek_te(tde, te);
+                    klass->handle_term(index, tde, te->curr_term);
+                }
+            }
+            FRT_XFINALLY
+                tde->close(tde);
+                te->close(te);
+            FRT_XENDTRY
+        }
+        frt_h_set(ir->field_index_cache, self, self);
+    }
+
+    return self;
+}
+
+/******************************************************************************
+ * ByteFieldIndex < FieldIndex
+ *
+ * The ByteFieldIndex holds an array of integers for each document in the
+ * index where the integer represents the sort value for the document.  This
+ * index should only be used for sorting and not as a field cache of the
+ * column's value.
+ ******************************************************************************/
+static void byte_handle_term(void *index_ptr,
+                             FrtTermDocEnum *tde,
+                             const char *text)
+{
+    long *index = (long *)index_ptr;
+    long val = index[-1]++;
+    (void)text;
+    while (tde->next(tde)) {
+        index[tde->doc_num(tde)] = val;
+    }
+}
+
+static void *byte_create_index(int size)
+{
+    long *index = FRT_ALLOC_AND_ZERO_N(long, size + 1);
+    index[0] = 1;
+    return &index[1];
+}
+
+static void byte_destroy_index(void *p)
+{
+    long *index = (long *)p;
+    free(&index[-1]);
+}
+
+const FrtFieldIndexClass FRT_BYTE_FIELD_INDEX_CLASS = {
+    "byte",
+    &byte_create_index,
+    &byte_destroy_index,
+    &byte_handle_term
+};
+
+/******************************************************************************
+ * IntegerFieldIndex < FieldIndex
+ ******************************************************************************/
+static void *integer_create_index(int size)
+{
+    return FRT_ALLOC_AND_ZERO_N(long, size);
+}
+
+static void integer_handle_term(void *index_ptr,
+                                FrtTermDocEnum *tde,
+                                const char *text)
+{
+    long *index = (long *)index_ptr;
+    long val;
+    sscanf(text, "%ld", &val);
+    while (tde->next(tde)) {
+        index[tde->doc_num(tde)] = val;
+    }
+}
+
+const FrtFieldIndexClass FRT_INTEGER_FIELD_INDEX_CLASS = {
+    "integer",
+    &integer_create_index,
+    &free,
+    &integer_handle_term
+};
+
+/******************************************************************************
+ * FloatFieldIndex < FieldIndex
+ ******************************************************************************/
+#define VALUES_ARRAY_START_SIZE 8
+static void *float_create_index(int size)
+{
+    return FRT_ALLOC_AND_ZERO_N(float, size);
+}
+
+static void float_handle_term(void *index_ptr,
+                              FrtTermDocEnum *tde,
+                              const char *text)
+{
+    float *index = (float *)index_ptr;
+    float val;
+    sscanf(text, "%g", &val);
+    while (tde->next(tde)) {
+        index[tde->doc_num(tde)] = val;
+    }
+}
+
+const FrtFieldIndexClass FRT_FLOAT_FIELD_INDEX_CLASS = {
+    "float",
+    &float_create_index,
+    &free,
+    &float_handle_term
+};
+
+/******************************************************************************
+ * StringFieldIndex < FieldIndex
+ ******************************************************************************/
+
+static void *string_create_index(int size)
+{
+    FrtStringIndex *self = FRT_ALLOC_AND_ZERO(FrtStringIndex);
+    self->size = size;
+    self->index = FRT_ALLOC_AND_ZERO_N(long, size);
+    self->v_capa = VALUES_ARRAY_START_SIZE;
+    self->v_size = 1; /* leave the first value as NULL */
+    self->values = FRT_ALLOC_AND_ZERO_N(char *, VALUES_ARRAY_START_SIZE);
+    return self;
+}
+
+static void string_destroy_index(void *p)
+{
+    FrtStringIndex *self = (FrtStringIndex *)p;
+    int i;
+    free(self->index);
+    for (i = 0; i < self->v_size; i++) {
+        free(self->values[i]);
+    }
+    free(self->values);
+    free(self);
+}
+
+static void string_handle_term(void *index_ptr,
+                               FrtTermDocEnum *tde,
+                               const char *text)
+{
+    FrtStringIndex *index = (FrtStringIndex *)index_ptr;
+    if (index->v_size >= index->v_capa) {
+        index->v_capa *= 2;
+        FRT_REALLOC_N(index->values, char *, index->v_capa);
+    }
+    index->values[index->v_size] = frt_estrdup(text);
+    while (tde->next(tde)) {
+        index->index[tde->doc_num(tde)] = index->v_size;
+    }
+    index->v_size++;
+}
+
+const FrtFieldIndexClass FRT_STRING_FIELD_INDEX_CLASS = {
+    "string",
+    &string_create_index,
+    &string_destroy_index,
+    &string_handle_term
+};

data/ext/isomorfeus_ferret_ext/frt_field_index.h

@@ -0,0 +1,42 @@
+#ifndef FRT_FIELD_INDEX_H
+#define FRT_FIELD_INDEX_H
+
+#include "frt_index.h"
+
+/***************************************************************************
+ *
+ * FrtFieldIndex
+ *
+ ***************************************************************************/
+
+typedef struct FrtStringIndex {
+    int size;
+    long *index;
+    char **values;
+    int v_size;
+    int v_capa;
+} FrtStringIndex;
+
+typedef struct FrtFieldIndexClass FrtFieldIndexClass;
+struct FrtFieldIndexClass {
+    const char *type;
+    void *(*create_index)(int size);
+    void  (*destroy_index)(void *p);
+    void  (*handle_term)(void *index, FrtTermDocEnum *tde, const char *text);
+};
+
+typedef struct FrtFieldIndex {
+    FrtSymbol field;
+    const FrtFieldIndexClass *klass;
+    void *index;
+} FrtFieldIndex;
+
+extern const FrtFieldIndexClass FRT_INTEGER_FIELD_INDEX_CLASS;
+extern const FrtFieldIndexClass   FRT_FLOAT_FIELD_INDEX_CLASS;
+extern const FrtFieldIndexClass  FRT_STRING_FIELD_INDEX_CLASS;
+extern const FrtFieldIndexClass    FRT_BYTE_FIELD_INDEX_CLASS;
+
+extern FrtFieldIndex *frt_field_index_get(FrtIndexReader *ir, FrtSymbol field,
+                                   const FrtFieldIndexClass *klass);
+
+#endif

data/ext/isomorfeus_ferret_ext/frt_filter.c

@@ -0,0 +1,157 @@
+#include "frt_global.h"
+#include "frt_search.h"
+#include <string.h>
+
+/***************************************************************************
+ *
+ * Filter
+ *
+ ***************************************************************************/
+
+void frt_filt_destroy_i(FrtFilter *filt)
+{
+    frt_h_destroy(filt->cache);
+    free(filt);
+}
+
+void frt_filt_deref(FrtFilter *filt)
+{
+    if (--(filt->ref_cnt) == 0) {
+        filt->destroy_i(filt);
+    }
+}
+
+FrtBitVector *frt_filt_get_bv(FrtFilter *filt, FrtIndexReader *ir)
+{
+    FrtCacheObject *co = (FrtCacheObject *)frt_h_get(filt->cache, ir);
+
+    if (!co) {
+        FrtBitVector *bv;
+        if (!ir->cache) {
+            frt_ir_add_cache(ir);
+        }
+        bv = filt->get_bv_i(filt, ir);
+        co = frt_co_create(filt->cache, ir->cache, filt, ir,
+                       (frt_free_ft)&frt_bv_destroy, (void *)bv);
+    }
+    return (FrtBitVector *)co->obj;
+}
+
+static char *filt_to_s_i(FrtFilter *filt)
+{
+    return frt_estrdup(rb_id2name(filt->name));
+}
+
+static unsigned long long frt_filt_hash_default(FrtFilter *filt)
+{
+    (void)filt;
+    return 0;
+}
+
+static int frt_filt_eq_default(FrtFilter *filt, FrtFilter *o)
+{
+    (void)filt; (void)o;
+    return false;
+}
+
+FrtFilter *frt_filt_create(size_t size, FrtSymbol name)
+{
+    FrtFilter *filt    = (FrtFilter *)frt_emalloc(size);
+    filt->cache     = frt_co_hash_create();
+    filt->name      = name;
+    filt->to_s      = &filt_to_s_i;
+    filt->hash      = &frt_filt_hash_default;
+    filt->eq        = &frt_filt_eq_default;
+    filt->destroy_i = &frt_filt_destroy_i;
+    filt->ref_cnt   = 1;
+    return filt;
+}
+
+unsigned long long frt_filt_hash(FrtFilter *filt)
+{
+    return frt_str_hash(rb_id2name(filt->name)) ^ filt->hash(filt);
+}
+
+int frt_filt_eq(FrtFilter *filt, FrtFilter *o)
+{
+    return ((filt == o)
+            || ((filt->name == o->name)
+                && (filt->eq == o->eq)
+                && (filt->eq(filt, o))));
+}
+
+/***************************************************************************
+ *
+ * QueryFilter
+ *
+ ***************************************************************************/
+
+#define QF(filt) ((QueryFilter *)(filt))
+typedef struct QueryFilter
+{
+    FrtFilter super;
+    FrtQuery *query;
+} QueryFilter;
+
+static char *qfilt_to_s(FrtFilter *filt)
+{
+    FrtQuery *query = QF(filt)->query;
+    char *query_str = query->to_s(query, (FrtSymbol)NULL);
+    char *filter_str = frt_strfmt("QueryFilter< %s >", query_str);
+    free(query_str);
+    return filter_str;
+}
+
+static FrtBitVector *qfilt_get_bv_i(FrtFilter *filt, FrtIndexReader *ir)
+{
+    FrtBitVector *bv = frt_bv_new_capa(ir->max_doc(ir));
+    FrtSearcher *sea = frt_isea_new(ir);
+    FrtWeight *weight = frt_q_weight(QF(filt)->query, sea);
+    FrtScorer *scorer = weight->scorer(weight, ir);
+    if (scorer) {
+        while (scorer->next(scorer)) {
+            frt_bv_set(bv, scorer->doc);
+        }
+        scorer->destroy(scorer);
+    }
+    weight->destroy(weight);
+    free(sea);
+    return bv;
+}
+
+static unsigned long long qfilt_hash(FrtFilter *filt)
+{
+    return frt_q_hash(QF(filt)->query);
+}
+
+static int qfilt_eq(FrtFilter *filt, FrtFilter *o)
+{
+    return frt_q_eq(QF(filt)->query, QF(o)->query);
+}
+
+static void qfilt_destroy_i(FrtFilter *filt)
+{
+    FrtQuery *query = QF(filt)->query;
+    frt_q_deref(query);
+    frt_filt_destroy_i(filt);
+}
+
+FrtFilter *frt_qfilt_new_nr(FrtQuery *query)
+{
+    FrtFilter *filt = filt_new(QueryFilter);
+
+    QF(filt)->query = query;
+
+    filt->get_bv_i  = &qfilt_get_bv_i;
+    filt->hash      = &qfilt_hash;
+    filt->eq        = &qfilt_eq;
+    filt->to_s      = &qfilt_to_s;
+    filt->destroy_i = &qfilt_destroy_i;
+    return filt;
+}
+
+FrtFilter *frt_qfilt_new(FrtQuery *query)
+{
+    FRT_REF(query);
+    return frt_qfilt_new_nr(query);
+}