| 1 | /* $Id$ |
|---|
| 2 | * |
|---|
| 3 | * Copyright (c) 2006-2008 Daisuke Maki <daisuke@endeworks.jp> |
|---|
| 4 | * All rights reserved. |
|---|
| 5 | */ |
|---|
| 6 | #ifndef __TEXT_MECAB_H__ |
|---|
| 7 | #define __TEXT_MECAB_H__ |
|---|
| 8 | #include "EXTERN.h" |
|---|
| 9 | #include "perl.h" |
|---|
| 10 | #include "XSUB.h" |
|---|
| 11 | #define NEED_newCONSTSUB |
|---|
| 12 | #define NEED_newRV_noinc |
|---|
| 13 | #define NEED_sv_2pv_nolen |
|---|
| 14 | #include "ppport.h" |
|---|
| 15 | #include <mecab.h> |
|---|
| 16 | |
|---|
| 17 | #ifndef TEXT_MECAB_DEBUG |
|---|
| 18 | #define TEXT_MECAB_DEBUG 0 |
|---|
| 19 | #endif |
|---|
| 20 | |
|---|
| 21 | #define XS_STATE(type, x) \ |
|---|
| 22 | INT2PTR(type, SvROK(x) ? SvIV(SvRV(x)) : SvIV(x)) |
|---|
| 23 | |
|---|
| 24 | #define XS_STRUCT2OBJ(sv, class, obj) \ |
|---|
| 25 | sv = newSViv(PTR2IV(obj)); \ |
|---|
| 26 | sv = newRV_noinc(sv); \ |
|---|
| 27 | sv_bless(sv, gv_stashpv(class, 1)); \ |
|---|
| 28 | SvREADONLY_on(sv); |
|---|
| 29 | |
|---|
| 30 | typedef mecab_t TextMeCab; |
|---|
| 31 | |
|---|
| 32 | typedef mecab_node_t TextMeCab_Node; |
|---|
| 33 | |
|---|
| 34 | typedef struct TextMeCab_Node_Cloned_Meta { |
|---|
| 35 | IV refcnt; |
|---|
| 36 | struct TextMeCab_Node_Cloned *first; |
|---|
| 37 | } TextMeCab_Node_Cloned_Meta; |
|---|
| 38 | |
|---|
| 39 | typedef struct TextMeCab_Node_Cloned { |
|---|
| 40 | struct TextMeCab_Node_Cloned *prev; |
|---|
| 41 | struct TextMeCab_Node_Cloned *next; |
|---|
| 42 | TextMeCab_Node_Cloned_Meta *meta; |
|---|
| 43 | TextMeCab_Node *actual; |
|---|
| 44 | } TextMeCab_Node_Cloned; |
|---|
| 45 | |
|---|
| 46 | |
|---|
| 47 | |
|---|
| 48 | #define XS_2MECAB(x) x |
|---|
| 49 | |
|---|
| 50 | #define MECAB_NODE_ID(x) x ? x->id : -1 |
|---|
| 51 | #define MECAB_NODE_LENGTH(x) x ? x->length : -1 |
|---|
| 52 | #define MECAB_NODE_RLENGTH(x) x ? x->rlength : -1 |
|---|
| 53 | #define MECAB_NODE_NEXT(x) x ? x->next : NULL |
|---|
| 54 | #define MECAB_NODE_PREV(x) x ? x->prev : NULL |
|---|
| 55 | #define MECAB_NODE_SURFACE(x) x ? x->surface : NULL |
|---|
| 56 | #define MECAB_NODE_FEATURE(x) x ? x->feature : NULL |
|---|
| 57 | #define MECAB_NODE_RCATTR(x) x ? x->rcAttr : -1 |
|---|
| 58 | #define MECAB_NODE_LCATTR(x) x ? x->lcAttr : -1 |
|---|
| 59 | #define MECAB_NODE_POSID(x) x ? x->posid : -1 |
|---|
| 60 | #define MECAB_NODE_CHAR_TYPE(x) x ? x->char_type : -1 |
|---|
| 61 | #define MECAB_NODE_STAT(x) x ? x->stat : -1 |
|---|
| 62 | #define MECAB_NODE_ISBEST(x) x ? x->isbest : -1 |
|---|
| 63 | #define MECAB_NODE_ALPHA(x) x ? x->alpha : -1 |
|---|
| 64 | #define MECAB_NODE_BETA(x) x ? x->beta : -1 |
|---|
| 65 | #define MECAB_NODE_PROB(x) x ? x->prob : -1 |
|---|
| 66 | #define MECAB_NODE_COST(x) x ? x->cost : -1 |
|---|
| 67 | #define MECAB_NODE_WCOST(x) x ? x->wcost : -1 |
|---|
| 68 | |
|---|
| 69 | /* Text::MeCab */ |
|---|
| 70 | void TextMeCab_bootstrap(); |
|---|
| 71 | TextMeCab *TextMeCab_new(char *class, char **argv, unsigned int argc); |
|---|
| 72 | TextMeCab *TextMeCab_new_from_av(char *class, AV *av); |
|---|
| 73 | TextMeCab_Node *TextMeCab_parse(TextMeCab *mecab, char *string); |
|---|
| 74 | void TextMeCab_DESTROY(TextMeCab *mecab); |
|---|
| 75 | |
|---|
| 76 | /* Text::MeCab::Node */ |
|---|
| 77 | unsigned int TextMeCab_Node_id(TextMeCab_Node *node); |
|---|
| 78 | unsigned int TextMeCab_Node_length(TextMeCab_Node *node); |
|---|
| 79 | unsigned int TextMeCab_Node_rlength(TextMeCab_Node *node); |
|---|
| 80 | TextMeCab_Node *TextMeCab_Node_next(TextMeCab_Node *node); |
|---|
| 81 | TextMeCab_Node *TextMeCab_Node_prev(TextMeCab_Node *node); |
|---|
| 82 | SV *TextMeCab_Node_surface(TextMeCab_Node *node); |
|---|
| 83 | const char *TextMeCab_Node_feature(TextMeCab_Node *node); |
|---|
| 84 | unsigned short TextMeCab_Node_rcattr(TextMeCab_Node *node); |
|---|
| 85 | unsigned short TextMeCab_Node_lcattr(TextMeCab_Node *node); |
|---|
| 86 | unsigned short TextMeCab_Node_posid(TextMeCab_Node *node); |
|---|
| 87 | unsigned char TextMeCab_Node_char_type(TextMeCab_Node *node); |
|---|
| 88 | unsigned char TextMeCab_Node_stat(TextMeCab_Node *node); |
|---|
| 89 | unsigned char TextMeCab_Node_isbest(TextMeCab_Node *node); |
|---|
| 90 | float TextMeCab_Node_alpha(TextMeCab_Node *node); |
|---|
| 91 | float TextMeCab_Node_beta(TextMeCab_Node *node); |
|---|
| 92 | float TextMeCab_Node_prob(TextMeCab_Node *node); |
|---|
| 93 | long TextMeCab_Node_cost(TextMeCab_Node *node); |
|---|
| 94 | short TextMeCab_Node_wcost(TextMeCab_Node *node); |
|---|
| 95 | const char *TextMeCab_Node_format(TextMeCab_Node *node, TextMeCab *mecab); |
|---|
| 96 | |
|---|
| 97 | TextMeCab_Node_Cloned *TextMeCab_Node_dclone(TextMeCab_Node *node); |
|---|
| 98 | TextMeCab_Node_Cloned *TextMeCab_Node_clone_single_node(TextMeCab_Node *node); |
|---|
| 99 | |
|---|
| 100 | /* Text::MeCab::Node::Cloned */ |
|---|
| 101 | void TextMeCab_Node_Cloned_free(TextMeCab_Node_Cloned *node); |
|---|
| 102 | unsigned int TextMeCab_Node_Cloned_id(TextMeCab_Node_Cloned *node); |
|---|
| 103 | unsigned int TextMeCab_Node_Cloned_length(TextMeCab_Node_Cloned *node); |
|---|
| 104 | unsigned int TextMeCab_Node_Cloned_rlength(TextMeCab_Node_Cloned *node); |
|---|
| 105 | TextMeCab_Node_Cloned *TextMeCab_Node_Cloned_next(TextMeCab_Node_Cloned *node); |
|---|
| 106 | TextMeCab_Node_Cloned *TextMeCab_Node_Cloned_prev(TextMeCab_Node_Cloned *node); |
|---|
| 107 | const char *TextMeCab_Node_Cloned_surface(TextMeCab_Node_Cloned *node); |
|---|
| 108 | const char *TextMeCab_Node_Cloned_feature(TextMeCab_Node_Cloned *node); |
|---|
| 109 | unsigned short TextMeCab_Node_Cloned_rcattr(TextMeCab_Node_Cloned *node); |
|---|
| 110 | unsigned short TextMeCab_Node_Cloned_lcattr(TextMeCab_Node_Cloned *node); |
|---|
| 111 | unsigned short TextMeCab_Node_Cloned_posid(TextMeCab_Node_Cloned *node); |
|---|
| 112 | unsigned char TextMeCab_Node_Cloned_char_type(TextMeCab_Node_Cloned *node); |
|---|
| 113 | unsigned char TextMeCab_Node_Cloned_stat(TextMeCab_Node_Cloned *node); |
|---|
| 114 | unsigned char TextMeCab_Node_Cloned_isbest(TextMeCab_Node_Cloned *node); |
|---|
| 115 | float TextMeCab_Node_Cloned_alpha(TextMeCab_Node_Cloned *node); |
|---|
| 116 | float TextMeCab_Node_Cloned_beta(TextMeCab_Node_Cloned *node); |
|---|
| 117 | float TextMeCab_Node_Cloned_prob(TextMeCab_Node_Cloned *node); |
|---|
| 118 | long TextMeCab_Node_Cloned_cost(TextMeCab_Node_Cloned *node); |
|---|
| 119 | short TextMeCab_Node_Cloned_wcost(TextMeCab_Node_Cloned *node); |
|---|
| 120 | const char *TextMeCab_Node_Cloned_format(TextMeCab_Node_Cloned *node, TextMeCab *mecab); |
|---|
| 121 | |
|---|
| 122 | #endif /* __TEXT_MECAB_H__ */ |
|---|