Struct Tagger

Source

pub struct Tagger { /* private fields */ }

Implementations§

Source §

impl Tagger

Source

pub fn new<T: Into<Vec<u8>>>(arg: T) -> Tagger

Examples found in repository ?

examples/simple.rs (line 9)

5fn main() {
6    let input = "太郎は次郎が持っている本を花子に渡した。";
7    println!("INPUT: {}", input);
8
9    let mut tagger = Tagger::new("");
10
11    // gets tagged result as String
12    let mut result = tagger.parse_str(input);
13    println!("RESULT: {}", result);
14
15    // gets N best results as String
16    result = tagger.parse_nbest(3, input);
17    println!("NBEST:\n{}", result);
18
19    // gets N best in sequence
20    tagger.parse_nbest_init(input);
21    for i in 0..3 {
22        if let Some(res) = tagger.next() {
23            println!("{}:\n{}", i, res);
24        }
25    }
26
27    // gets Node object
28    for node in tagger.parse_to_node(input).iter_next() {
29        match node.stat as i32 {
30            mecab::MECAB_BOS_NODE => {
31                print!("{} BOS ", node.id);
32            }
33            mecab::MECAB_EOS_NODE => {
34                print!("{} EOS ", node.id);
35            }
36            _ => {
37                print!("{} {} ", node.id, &(node.surface)[..(node.length as usize)]);
38            }
39        }
40
41        println!("{} {} {} {} {} {} {} {} {} {} {} {} {}",
42                 node.feature,
43                 input.len() as isize - node.surface.len() as isize,
44                 input.len() as isize - node.surface.len() as isize + node.length as isize,
45                 node.rcattr,
46                 node.lcattr,
47                 node.posid,
48                 node.char_type,
49                 node.stat,
50                 node.isbest,
51                 node.alpha,
52                 node.beta,
53                 node.prob,
54                 node.cost);
55    }
56
57    // dictionary info
58    for dict in tagger.dictionary_info().iter() {
59        println!("\nfilename: {}", dict.filename);
60        println!("charset: {}", dict.charset);
61        println!("size: {}", dict.size);
62        println!("type: {}", dict.dict_type);
63        println!("lsize: {}", dict.lsize);
64        println!("rsize: {}", dict.rsize);
65        println!("version: {}", dict.version);
66    }
67}

Source

pub fn parse(&self, latice: &Lattice) -> bool

Examples found in repository ?

examples/multithreaded.rs (line 23)

6fn main() {
7    let input = "太郎は次郎が持っている本を花子に渡した。";
8
9    // create model object
10    let model = Arc::new(Model::new(""));
11
12    let handle = std::thread::spawn(move || {
13        // create tagger based on the model
14        let tagger = model.create_tagger();
15
16        // create lattice object per thread
17        let mut lattice = model.create_lattice();
18
19        // get tagged result as string
20        lattice.set_sentence(input);
21
22        // parse lattice
23        tagger.parse(&lattice);
24        println!("{}", lattice.to_string());
25
26        // iterate over node objects
27        for node in lattice.bos_node().iter_next() {
28            match node.stat as i32 {
29                mecab::MECAB_BOS_NODE => {
30                    print!("{} BOS ", node.id);
31                }
32                mecab::MECAB_EOS_NODE => {
33                    print!("{} EOS ", node.id);
34                }
35                _ => {
36                    print!("{} {} ", node.id, &(node.surface)[..(node.length as usize)]);
37                }
38            }
39
40            println!("{} {} {} {} {} {} {} {} {} {} {} {} {}",
41                     node.feature,
42                     input.len() as isize - node.surface.len() as isize,
43                     input.len() as isize - node.surface.len() as isize + node.length as isize,
44                     node.rcattr,
45                     node.lcattr,
46                     node.posid,
47                     node.char_type,
48                     node.stat,
49                     node.isbest,
50                     node.alpha,
51                     node.beta,
52                     node.prob,
53                     node.cost);
54        }
55
56        // iterate over begin and end nodes
57        let len = lattice.size();
58        for i in 0..len + 1 {
59            let b = lattice.begin_nodes(i);
60            let e = lattice.end_nodes(i);
61
62            if let Some(nodes) = b {
63                for node in nodes.iter_bnext() {
64                    println!("B[{}] {}\t{}", i, node.surface, node.feature);
65                }
66            }
67
68            if let Some(nodes) = e {
69                for node in nodes.iter_enext() {
70                    println!("E[{}] {}\t{}", i, node.surface, node.feature);
71                }
72            }
73        }
74
75        // get N best results
76        lattice.set_request_type(mecab::MECAB_NBEST);
77        lattice.set_sentence(input);
78        tagger.parse(&lattice);
79
80        for i in 0..10 {
81            println!("NBEST: {}", i);
82            println!("{}", lattice.to_string());
83
84            if !lattice.next() {
85                break;
86            }
87        }
88
89        // marginal probabilities
90        lattice.remove_request_type(mecab::MECAB_NBEST);
91        lattice.set_request_type(mecab::MECAB_MARGINAL_PROB);
92        lattice.set_sentence(input);
93        tagger.parse(&lattice);
94
95        println!("{}", lattice.theta());
96
97        for node in lattice.bos_node().iter_next() {
98            println!("{}\t{}\t{}",
99                     &(node.surface)[..(node.length as usize)],
100                     node.feature,
101                     node.prob);
102        }
103
104        // dictionary info
105        for dict in model.dictionary_info().iter() {
106            println!("\nfilename: {}", dict.filename);
107            println!("charset: {}", dict.charset);
108            println!("size: {}", dict.size);
109            println!("type: {}", dict.dict_type);
110            println!("lsize: {}", dict.lsize);
111            println!("rsize: {}", dict.rsize);
112            println!("version: {}", dict.version);
113        }
114    });
115
116    handle.join().unwrap();
117}

Source

pub fn parse_str<T: Into<Vec<u8>>>(&self, input: T) -> String

Examples found in repository ?

examples/simple.rs (line 12)

5fn main() {
6    let input = "太郎は次郎が持っている本を花子に渡した。";
7    println!("INPUT: {}", input);
8
9    let mut tagger = Tagger::new("");
10
11    // gets tagged result as String
12    let mut result = tagger.parse_str(input);
13    println!("RESULT: {}", result);
14
15    // gets N best results as String
16    result = tagger.parse_nbest(3, input);
17    println!("NBEST:\n{}", result);
18
19    // gets N best in sequence
20    tagger.parse_nbest_init(input);
21    for i in 0..3 {
22        if let Some(res) = tagger.next() {
23            println!("{}:\n{}", i, res);
24        }
25    }
26
27    // gets Node object
28    for node in tagger.parse_to_node(input).iter_next() {
29        match node.stat as i32 {
30            mecab::MECAB_BOS_NODE => {
31                print!("{} BOS ", node.id);
32            }
33            mecab::MECAB_EOS_NODE => {
34                print!("{} EOS ", node.id);
35            }
36            _ => {
37                print!("{} {} ", node.id, &(node.surface)[..(node.length as usize)]);
38            }
39        }
40
41        println!("{} {} {} {} {} {} {} {} {} {} {} {} {}",
42                 node.feature,
43                 input.len() as isize - node.surface.len() as isize,
44                 input.len() as isize - node.surface.len() as isize + node.length as isize,
45                 node.rcattr,
46                 node.lcattr,
47                 node.posid,
48                 node.char_type,
49                 node.stat,
50                 node.isbest,
51                 node.alpha,
52                 node.beta,
53                 node.prob,
54                 node.cost);
55    }
56
57    // dictionary info
58    for dict in tagger.dictionary_info().iter() {
59        println!("\nfilename: {}", dict.filename);
60        println!("charset: {}", dict.charset);
61        println!("size: {}", dict.size);
62        println!("type: {}", dict.dict_type);
63        println!("lsize: {}", dict.lsize);
64        println!("rsize: {}", dict.rsize);
65        println!("version: {}", dict.version);
66    }
67}

Source

pub fn parse_to_node<T: Into<Vec<u8>>>(&mut self, input: T) -> Node

Examples found in repository ?

examples/simple.rs (line 28)

5fn main() {
6    let input = "太郎は次郎が持っている本を花子に渡した。";
7    println!("INPUT: {}", input);
8
9    let mut tagger = Tagger::new("");
10
11    // gets tagged result as String
12    let mut result = tagger.parse_str(input);
13    println!("RESULT: {}", result);
14
15    // gets N best results as String
16    result = tagger.parse_nbest(3, input);
17    println!("NBEST:\n{}", result);
18
19    // gets N best in sequence
20    tagger.parse_nbest_init(input);
21    for i in 0..3 {
22        if let Some(res) = tagger.next() {
23            println!("{}:\n{}", i, res);
24        }
25    }
26
27    // gets Node object
28    for node in tagger.parse_to_node(input).iter_next() {
29        match node.stat as i32 {
30            mecab::MECAB_BOS_NODE => {
31                print!("{} BOS ", node.id);
32            }
33            mecab::MECAB_EOS_NODE => {
34                print!("{} EOS ", node.id);
35            }
36            _ => {
37                print!("{} {} ", node.id, &(node.surface)[..(node.length as usize)]);
38            }
39        }
40
41        println!("{} {} {} {} {} {} {} {} {} {} {} {} {}",
42                 node.feature,
43                 input.len() as isize - node.surface.len() as isize,
44                 input.len() as isize - node.surface.len() as isize + node.length as isize,
45                 node.rcattr,
46                 node.lcattr,
47                 node.posid,
48                 node.char_type,
49                 node.stat,
50                 node.isbest,
51                 node.alpha,
52                 node.beta,
53                 node.prob,
54                 node.cost);
55    }
56
57    // dictionary info
58    for dict in tagger.dictionary_info().iter() {
59        println!("\nfilename: {}", dict.filename);
60        println!("charset: {}", dict.charset);
61        println!("size: {}", dict.size);
62        println!("type: {}", dict.dict_type);
63        println!("lsize: {}", dict.lsize);
64        println!("rsize: {}", dict.rsize);
65        println!("version: {}", dict.version);
66    }
67}

Source

pub fn parse_nbest<T: Into<Vec<u8>>>(&self, n: usize, input: T) -> String

Examples found in repository ?

examples/simple.rs (line 16)

5fn main() {
6    let input = "太郎は次郎が持っている本を花子に渡した。";
7    println!("INPUT: {}", input);
8
9    let mut tagger = Tagger::new("");
10
11    // gets tagged result as String
12    let mut result = tagger.parse_str(input);
13    println!("RESULT: {}", result);
14
15    // gets N best results as String
16    result = tagger.parse_nbest(3, input);
17    println!("NBEST:\n{}", result);
18
19    // gets N best in sequence
20    tagger.parse_nbest_init(input);
21    for i in 0..3 {
22        if let Some(res) = tagger.next() {
23            println!("{}:\n{}", i, res);
24        }
25    }
26
27    // gets Node object
28    for node in tagger.parse_to_node(input).iter_next() {
29        match node.stat as i32 {
30            mecab::MECAB_BOS_NODE => {
31                print!("{} BOS ", node.id);
32            }
33            mecab::MECAB_EOS_NODE => {
34                print!("{} EOS ", node.id);
35            }
36            _ => {
37                print!("{} {} ", node.id, &(node.surface)[..(node.length as usize)]);
38            }
39        }
40
41        println!("{} {} {} {} {} {} {} {} {} {} {} {} {}",
42                 node.feature,
43                 input.len() as isize - node.surface.len() as isize,
44                 input.len() as isize - node.surface.len() as isize + node.length as isize,
45                 node.rcattr,
46                 node.lcattr,
47                 node.posid,
48                 node.char_type,
49                 node.stat,
50                 node.isbest,
51                 node.alpha,
52                 node.beta,
53                 node.prob,
54                 node.cost);
55    }
56
57    // dictionary info
58    for dict in tagger.dictionary_info().iter() {
59        println!("\nfilename: {}", dict.filename);
60        println!("charset: {}", dict.charset);
61        println!("size: {}", dict.size);
62        println!("type: {}", dict.dict_type);
63        println!("lsize: {}", dict.lsize);
64        println!("rsize: {}", dict.rsize);
65        println!("version: {}", dict.version);
66    }
67}

Source

pub fn parse_nbest_init<T: Into<Vec<u8>>>(&mut self, input: T) -> bool

Examples found in repository ?

examples/simple.rs (line 20)

5fn main() {
6    let input = "太郎は次郎が持っている本を花子に渡した。";
7    println!("INPUT: {}", input);
8
9    let mut tagger = Tagger::new("");
10
11    // gets tagged result as String
12    let mut result = tagger.parse_str(input);
13    println!("RESULT: {}", result);
14
15    // gets N best results as String
16    result = tagger.parse_nbest(3, input);
17    println!("NBEST:\n{}", result);
18
19    // gets N best in sequence
20    tagger.parse_nbest_init(input);
21    for i in 0..3 {
22        if let Some(res) = tagger.next() {
23            println!("{}:\n{}", i, res);
24        }
25    }
26
27    // gets Node object
28    for node in tagger.parse_to_node(input).iter_next() {
29        match node.stat as i32 {
30            mecab::MECAB_BOS_NODE => {
31                print!("{} BOS ", node.id);
32            }
33            mecab::MECAB_EOS_NODE => {
34                print!("{} EOS ", node.id);
35            }
36            _ => {
37                print!("{} {} ", node.id, &(node.surface)[..(node.length as usize)]);
38            }
39        }
40
41        println!("{} {} {} {} {} {} {} {} {} {} {} {} {}",
42                 node.feature,
43                 input.len() as isize - node.surface.len() as isize,
44                 input.len() as isize - node.surface.len() as isize + node.length as isize,
45                 node.rcattr,
46                 node.lcattr,
47                 node.posid,
48                 node.char_type,
49                 node.stat,
50                 node.isbest,
51                 node.alpha,
52                 node.beta,
53                 node.prob,
54                 node.cost);
55    }
56
57    // dictionary info
58    for dict in tagger.dictionary_info().iter() {
59        println!("\nfilename: {}", dict.filename);
60        println!("charset: {}", dict.charset);
61        println!("size: {}", dict.size);
62        println!("type: {}", dict.dict_type);
63        println!("lsize: {}", dict.lsize);
64        println!("rsize: {}", dict.rsize);
65        println!("version: {}", dict.version);
66    }
67}

Source

pub fn next(&self) -> Option<String>

Examples found in repository ?

examples/simple.rs (line 22)

5fn main() {
6    let input = "太郎は次郎が持っている本を花子に渡した。";
7    println!("INPUT: {}", input);
8
9    let mut tagger = Tagger::new("");
10
11    // gets tagged result as String
12    let mut result = tagger.parse_str(input);
13    println!("RESULT: {}", result);
14
15    // gets N best results as String
16    result = tagger.parse_nbest(3, input);
17    println!("NBEST:\n{}", result);
18
19    // gets N best in sequence
20    tagger.parse_nbest_init(input);
21    for i in 0..3 {
22        if let Some(res) = tagger.next() {
23            println!("{}:\n{}", i, res);
24        }
25    }
26
27    // gets Node object
28    for node in tagger.parse_to_node(input).iter_next() {
29        match node.stat as i32 {
30            mecab::MECAB_BOS_NODE => {
31                print!("{} BOS ", node.id);
32            }
33            mecab::MECAB_EOS_NODE => {
34                print!("{} EOS ", node.id);
35            }
36            _ => {
37                print!("{} {} ", node.id, &(node.surface)[..(node.length as usize)]);
38            }
39        }
40
41        println!("{} {} {} {} {} {} {} {} {} {} {} {} {}",
42                 node.feature,
43                 input.len() as isize - node.surface.len() as isize,
44                 input.len() as isize - node.surface.len() as isize + node.length as isize,
45                 node.rcattr,
46                 node.lcattr,
47                 node.posid,
48                 node.char_type,
49                 node.stat,
50                 node.isbest,
51                 node.alpha,
52                 node.beta,
53                 node.prob,
54                 node.cost);
55    }
56
57    // dictionary info
58    for dict in tagger.dictionary_info().iter() {
59        println!("\nfilename: {}", dict.filename);
60        println!("charset: {}", dict.charset);
61        println!("size: {}", dict.size);
62        println!("type: {}", dict.dict_type);
63        println!("lsize: {}", dict.lsize);
64        println!("rsize: {}", dict.rsize);
65        println!("version: {}", dict.version);
66    }
67}

Source

pub fn next_node(&self) -> Option<Node>

Source

pub fn format_node(&self, node: Node) -> String

Source

pub fn dictionary_info(&self) -> DictionaryInfo

Examples found in repository ?

examples/simple.rs (line 58)

5fn main() {
6    let input = "太郎は次郎が持っている本を花子に渡した。";
7    println!("INPUT: {}", input);
8
9    let mut tagger = Tagger::new("");
10
11    // gets tagged result as String
12    let mut result = tagger.parse_str(input);
13    println!("RESULT: {}", result);
14
15    // gets N best results as String
16    result = tagger.parse_nbest(3, input);
17    println!("NBEST:\n{}", result);
18
19    // gets N best in sequence
20    tagger.parse_nbest_init(input);
21    for i in 0..3 {
22        if let Some(res) = tagger.next() {
23            println!("{}:\n{}", i, res);
24        }
25    }
26
27    // gets Node object
28    for node in tagger.parse_to_node(input).iter_next() {
29        match node.stat as i32 {
30            mecab::MECAB_BOS_NODE => {
31                print!("{} BOS ", node.id);
32            }
33            mecab::MECAB_EOS_NODE => {
34                print!("{} EOS ", node.id);
35            }
36            _ => {
37                print!("{} {} ", node.id, &(node.surface)[..(node.length as usize)]);
38            }
39        }
40
41        println!("{} {} {} {} {} {} {} {} {} {} {} {} {}",
42                 node.feature,
43                 input.len() as isize - node.surface.len() as isize,
44                 input.len() as isize - node.surface.len() as isize + node.length as isize,
45                 node.rcattr,
46                 node.lcattr,
47                 node.posid,
48                 node.char_type,
49                 node.stat,
50                 node.isbest,
51                 node.alpha,
52                 node.beta,
53                 node.prob,
54                 node.cost);
55    }
56
57    // dictionary info
58    for dict in tagger.dictionary_info().iter() {
59        println!("\nfilename: {}", dict.filename);
60        println!("charset: {}", dict.charset);
61        println!("size: {}", dict.size);
62        println!("type: {}", dict.dict_type);
63        println!("lsize: {}", dict.lsize);
64        println!("rsize: {}", dict.rsize);
65        println!("version: {}", dict.version);
66    }
67}