ferret 0.9.6 → 0.10.0

data/ext/extconf.rb

@@ -1,14 +1,7 @@
 # extconf.rb for Ferret extensions
 if (/mswin/ =~ RUBY_PLATFORM) and ENV['make'].nil?
-  File.open('Makefile', "w") {}
-  begin
-    `nmake`
-    require 'mkmf'
-    create_makefile("ferret_ext")
-  rescue => error
-    require 'fileutils'
-    FileUtils.copy('dummy.exe', 'nmake.exe')
-  end
+  require 'mkmf'
+  create_makefile("ferret_ext")
 else
   require 'mkmf'
   $CFLAGS += " -fno-common"

data/ext/ferret.c

@@ -1,53 +1,57 @@
+#include <errno.h>
 #include "ferret.h"
 #include "except.h"
 #include "hash.h"
 #include "hashset.h"
+#include "threading.h"
 
 /* Object Map */
-static HshTable *object_map;
+static HashTable *object_map;
 
 /* IDs */
 ID id_new;
+ID id_capacity;
+ID id_less_than;
+ID id_lt;
 ID id_call;
 ID id_is_directory;
+ID id_data;
 
 static ID id_mkdir_p;
 
+/* Symbols */
+VALUE sym_yes;
+VALUE sym_no;
+VALUE sym_true;
+VALUE sym_false;
+VALUE sym_path;
+VALUE sym_dir;
+
 /* Modules */
 VALUE mFerret;
-VALUE mAnalysis;
-VALUE mDocument;
-VALUE mIndex;
-VALUE mSearch;
 VALUE mStore;
 VALUE mStringHelper;
-VALUE mUtils;
 VALUE mSpans;
 
 /* Classes */
 /*
 */
 
-xcontext_t *xtop_context = NULL;
-
-unsigned int
-value_hash(const void *key)
+unsigned long value_hash(const void *key)
 {
-  return (unsigned int)key;
+    return (unsigned long)key;
 }
 
-int
-value_eq(const void *key1, const void *key2)
+int value_eq(const void *key1, const void *key2)
 {
-  return key1 == key2;
+    return key1 == key2;
 }
 
-VALUE
-object_get(void *key)
+VALUE object_get(void *key)
 {
-  VALUE val = (VALUE)h_get(object_map, key);
-  if (!val) val = Qnil;
-  return val;
+    VALUE val = (VALUE)h_get(object_map, key);
+    if (!val) val = Qnil;
+    return val;
 }
 
 //static int hash_cnt = 0;
@@ -55,133 +59,236 @@ void
 //object_add(void *key, VALUE obj)
 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("adding %d. now contains %d %s:%d\n", (int)key, ++hash_cnt, file, line);
-  h_set(object_map, key, (void *)obj);
+    if (h_get(object_map, key))
+        printf("failed adding %d. %s:%d\n", (int)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);
 }
 
 void
 //object_set(void *key, VALUE obj)
 object_set2(void *key, VALUE obj, const char *file, int line)
 {
-  //if (!h_get(object_map, key))
-    //printf("seting %d. now contains %d %s:%d\n", (int)key, ++hash_cnt, file, line);
-  h_set(object_map, key, (void *)obj);
+    //if (!h_get(object_map, key))
+      //printf("adding %d. now contains %d %s:%d\n", (int)key, ++hash_cnt, file, line);
+    h_set(object_map, key, (void *)obj);
 }
 
 void
 //object_del(void *key)
 object_del2(void *key, const char *file, int line)
 {
-  if (object_get(key) == Qnil) 
-    printf("failed deleting %d. %s:%d\n", (int)key, file, line);
-  //printf("deleting %d. now contains %d, %s:%d\n", (int)key, --hash_cnt, file, line);
-  h_del(object_map, key);
+    if (object_get(key) == Qnil) 
+        printf("failed deleting %d. %s:%d\n", (int)key, file, line);
+    //printf("deleting %d. now contains %d, %s:%d\n", (int)key, --hash_cnt, file, line);
+    h_del(object_map, key);
 }
 
-void
-frt_gc_mark(void *key)
+void frt_gc_mark(void *key)
 {
-  VALUE val = (VALUE)h_get(object_map, key);
-  if (val)
-    rb_gc_mark(val);
+    VALUE val = (VALUE)h_get(object_map, key);
+    if (val)
+        rb_gc_mark(val);
 }
 
-VALUE
-frt_data_alloc(VALUE klass)
+VALUE frt_data_alloc(VALUE klass)
 {
-  return Frt_Make_Struct(klass);
+    return Frt_Make_Struct(klass);
 }
 
-void
-frt_deref_free(void *p)
+void frt_deref_free(void *p)
 {
-  object_del(p);
+    object_del(p);
 }
 
-void
-frt_thread_once(int *once_control, void (*init_routine) (void))
+void frt_thread_once(int *once_control, void (*init_routine) (void))
 {
-  if (*once_control) {
-    init_routine();
-    *once_control = 0;
-  }
+    if (*once_control) {
+        init_routine();
+        *once_control = 0;
+    }
 }
 
-void
-frt_thread_key_create(thread_key_t *key, void (*destr_function) (void *))
+void frt_thread_key_create(thread_key_t *key, void (*destr_function)(void *))
 {
-  *key = h_new(&value_hash, &value_eq, NULL, destr_function);
+    *key = h_new(&value_hash, &value_eq, NULL, destr_function);
 }
 
-void
-frt_thread_key_delete(thread_key_t key)
+void frt_thread_key_delete(thread_key_t key)
 {
-  h_destroy(key);
+    h_destroy(key);
 }
 
-void
-frt_thread_setspecific(thread_key_t key, const void *pointer)
+void frt_thread_setspecific(thread_key_t key, const void *pointer)
 {
-  h_set(key, (void *)rb_thread_current(), (void *)pointer);
+    h_set(key, (void *)rb_thread_current(), (void *)pointer);
 }
 
-void *
-frt_thread_getspecific(thread_key_t key)
+void *frt_thread_getspecific(thread_key_t key)
 {
-  return h_get(key, (void *)rb_thread_current());
+    return h_get(key, (void *)rb_thread_current());
 }
 
-void
-frt_create_dir(VALUE rpath)
+void frt_create_dir(VALUE rpath)
 {
-  VALUE mFileUtils;
-  rb_require("fileutils");
-  mFileUtils = rb_define_module("FileUtils");
-  rb_funcall(mFileUtils, id_mkdir_p, 1, rpath);
+    VALUE mFileUtils;
+    rb_require("fileutils");
+    mFileUtils = rb_define_module("FileUtils");
+    rb_funcall(mFileUtils, id_mkdir_p, 1, rpath);
 }
 
-VALUE
-frt_hs_to_rb_ary(HashSet *hs)
+VALUE frt_hs_to_rb_ary(HashSet *hs)
 {
-  int i;
-  VALUE ary = rb_ary_new();
-  for (i = 0; i < hs->size; i++) {
-    rb_ary_push(ary, rb_str_new2(hs->elems[i]));
-  }
-  return ary;
+    int i;
+    VALUE ary = rb_ary_new();
+    for (i = 0; i < hs->size; i++) {
+        rb_ary_push(ary, rb_str_new2(hs->elems[i]));
+    }
+    return ary;
 }
 
-void
-Init_ferret_ext(void)
-{
-  /* initialize object map */
-  object_map = h_new(&value_hash, &value_eq, NULL, NULL);
-
-  /* IDs */
-	id_new = rb_intern("new");
-	id_call = rb_intern("call");
-
-  id_mkdir_p = rb_intern("mkdir_p");
-  id_is_directory = rb_intern("directory?");
-
-  /* Modules */
-  mFerret = rb_define_module("Ferret");
-  mAnalysis = rb_define_module_under(mFerret, "Analysis");
-  mDocument = rb_define_module_under(mFerret, "Document");
-  mIndex = rb_define_module_under(mFerret, "Index");
-  mSearch = rb_define_module_under(mFerret, "Search");
-  mStore = rb_define_module_under(mFerret, "Store");
-  mUtils = rb_define_module_under(mFerret, "Utils");
-  mSpans = rb_define_module_under(mSearch, "Spans");
-
-  /* Inits */
-  Init_term();
-  Init_analysis();
-  Init_doc();
-  Init_dir();
-  Init_index_io();
-  Init_search();
-  Init_qparser();
+void *frt_rb_data_ptr(VALUE val)
+{
+    Check_Type(val, T_DATA);
+    return DATA_PTR(val);
+}
+
+char *
+frt_field(VALUE rfield)
+{
+    switch (TYPE(rfield)) {
+        case T_SYMBOL:
+            return rb_id2name(SYM2ID(rfield));
+        case T_STRING:
+            return RSTRING(rfield)->ptr;
+        default:
+            rb_raise(rb_eArgError, "field name must be a symbol");
+    }
+    return NULL;
+}
+
+static VALUE error_map;
+
+VALUE frt_get_error(const char *err_type)
+{
+    VALUE error_class;
+    if (Qnil != (error_class = rb_hash_aref(error_map, rb_intern(err_type)))) {
+        return error_class;
+    }
+    return rb_eStandardError;
+}
+
+#define FRT_BUF_SIZ 2046
+#ifdef FRT_HAS_VARARGS
+void vfrt_rb_raise(const char *file, int line_num, const char *func,
+                   const char *err_type, const char *fmt, va_list args)
+#endif
+{
+    char buf[FRT_BUF_SIZ];
+    size_t so_far = 0;
+#ifdef FRT_HAS_VARARGS
+    snprintf(buf, FRT_BUF_SIZ, "%s occured at <%s>:%d in %s\n",
+            err_type, file, line_num, func);
+#else
+    snprintf(buf, FRT_BUF_SIZ, "%s occured:\n", err_type);
+#endif
+    so_far = strlen(buf);
+    vsnprintf(buf + so_far, FRT_BUF_SIZ - so_far, fmt, args);
+
+    so_far = strlen(buf);
+    if (fmt[0] != '\0' && fmt[strlen(fmt) - 1] == ':') {
+        snprintf(buf + so_far, FRT_BUF_SIZ - so_far, " %s", strerror(errno));
+        so_far = strlen(buf);
+    }
+
+    snprintf(buf + so_far, FRT_BUF_SIZ - so_far, "\n");
+    rb_raise(frt_get_error(err_type), buf);
+}
+
+#ifdef FRT_HAS_VARARGS
+void frt_rb_raise(const char *file, int line_num, const char *func,
+                  const char *err_type, const char *fmt, ...)
+#else
+void FRT_EXIT(const char *err_type, const char *fmt, ...)
+#endif
+{
+    va_list args;
+    va_start(args, fmt);
+#ifdef FRT_HAS_VARARGS
+    vfrt_rb_raise(file, line_num, func, err_type, fmt, args);
+#else
+    V_FRT_EXIT(err_type, fmt, args);
+#endif
+    va_end(args);
+}
+
+/*
+ *  Document-module: Ferret
+ *
+ *  See the README
+ */
+void Init_Ferret(void)
+{
+    mFerret = rb_define_module("Ferret");
+}
+
+void Init_ferret_ext(void)
+{
+    VALUE cParseError;
+    VALUE cStateError;
+
+    /* initialize object map */
+    object_map = h_new(&value_hash, &value_eq, NULL, NULL);
+
+    /* IDs */
+    id_new = rb_intern("new");
+    id_call = rb_intern("call");
+
+    id_capacity = rb_intern("capacity");
+    id_less_than = rb_intern("less_than");
+    id_lt = rb_intern("<");
+
+    id_mkdir_p = rb_intern("mkdir_p");
+    id_is_directory = rb_intern("directory?");
+
+    id_data = rb_intern("@data");
+
+    /* Symbols */
+    sym_yes = ID2SYM(rb_intern("yes"));;
+    sym_no = ID2SYM(rb_intern("no"));;
+    sym_true = ID2SYM(rb_intern("true"));;
+    sym_false = ID2SYM(rb_intern("false"));;
+    sym_path = ID2SYM(rb_intern("path"));;
+    sym_dir = ID2SYM(rb_intern("dir"));;
+
+    /* Inits */
+    Init_Ferret();
+    Init_Utils();
+    Init_Analysis();
+    Init_Store();
+    Init_Index();
+    Init_Search();
+    Init_QueryParser();
+
+    /* Error Classes */
+    cParseError =
+        rb_define_class_under(mFerret, "ParseError", rb_eStandardError);
+    cStateError =
+        rb_define_class_under(mFerret, "StateError", rb_eStandardError);
+
+    error_map = rb_hash_new();
+    rb_hash_aset(error_map, rb_intern("Exception"),         rb_eStandardError);
+    rb_hash_aset(error_map, rb_intern("IO Error"),          rb_eIOError);
+    rb_hash_aset(error_map, rb_intern("Argument Error"),    rb_eArgError);
+    rb_hash_aset(error_map, rb_intern("End-of-File Error"), rb_eEOFError);
+    rb_hash_aset(error_map, rb_intern("Unsupported Function Error"),
+                                                            rb_eNotImpError);
+    rb_hash_aset(error_map, rb_intern("State Error"),       cStateError);
+    rb_hash_aset(error_map, rb_intern("ParseError"),        cParseError);
+    rb_hash_aset(error_map, rb_intern("Memory Error"),      rb_eNoMemError);
+    rb_hash_aset(error_map, rb_intern("Index Error"),       rb_eIndexError);
+    rb_hash_aset(error_map, rb_intern("Lock Error"),        cLockError);
+
+    rb_define_const(mFerret, "EXCEPTION_MAP", error_map);
+    rb_define_const(mFerret, "FIX_INT_MAX", INT2FIX(INT_MAX >> 1));
 }

data/ext/ferret.h

@@ -7,31 +7,41 @@
 
 /* IDs */
 extern ID id_new;
+extern ID id_capacity;
+extern ID id_less_than;
+extern ID id_lt;
 extern ID id_call;
 extern ID id_is_directory;
+extern ID id_data;
+
+/* Symbols */
+extern VALUE sym_yes;
+extern VALUE sym_no;
+extern VALUE sym_true;
+extern VALUE sym_false;
+extern VALUE sym_path;
+extern VALUE sym_dir;
 
 /* Modules */
 extern VALUE mFerret;
-extern VALUE mAnalysis;
-extern VALUE mDocument;
 extern VALUE mIndex;
 extern VALUE mSearch;
 extern VALUE mStore;
 extern VALUE mStringHelper;
-extern VALUE mUtils;
 extern VALUE mSpans;
 
 /* Classes */
 extern VALUE cDirectory;
+extern VALUE cLockError;
 
 /* Ferret Inits */
-extern void Init_term();
-extern void Init_dir();
-extern void Init_analysis();
-extern void Init_doc();
-extern void Init_index_io();
-extern void Init_search();
-extern void Init_qparser();
+extern void Init_Utils();
+extern void Init_Analysis();
+extern void Init_Store();
+extern void Init_Index();
+extern void Init_Search();
+extern void Init_QueryParser();
+
 //extern void object_add(void *key, VALUE obj);
 #define object_add(key, obj) object_add2(key, obj,  __FILE__, __LINE__)
 extern void object_add2(void *key, VALUE obj, const char *file, int line);
@@ -44,10 +54,11 @@ extern void object_del2(void *key, const char *file, int line);
 extern void frt_gc_mark(void *key);
 extern VALUE object_get(void *key);
 extern VALUE frt_data_alloc(VALUE klass);
-extern VALUE frt_get_doc(Document *doc);
 extern void frt_deref_free(void *p);
 extern void frt_create_dir(VALUE rpath);
 extern VALUE frt_hs_to_rb_ary(HashSet *hs);
+extern void *frt_rb_data_ptr(VALUE val);
+extern char * frt_field(VALUE rfield);
 
 #define Frt_Make_Struct(klass)\
   rb_data_object_alloc(klass,NULL,(RUBY_DATA_FUNC)NULL,(RUBY_DATA_FUNC)NULL)