gdritter repos rrecutils / 4a0ba3e
Basic, hacky impl of limited parser Getty Ritter 7 years ago
7 changed file(s) with 303 addition(s) and 0 deletion(s). Collapse all Expand all
+4
-0
.gitignore less more
1 /target/
2 **/*.rs.bk
3 Cargo.lock
4 *~
+22
-0
Cargo.toml less more
1 [package]
2 name = "rrecutils"
3 version = "0.1.0"
4 authors = ["Getty Ritter <gettylefou@gmail.com>"]
5
6 [lib]
7 name = "rrecutils"
8 path = "src/lib.rs"
9
10 [dependencies]
11 regex = "0.2"
12 serde = "*"
13 serde_json = "*"
14 clap = "2.27.1"
15
16 [[bin]]
17 name = "rr-pretty"
18 path = "src/tools/pretty.rs"
19
20 [[bin]]
21 name = "rr-to-json"
22 path = "src/tools/tojson.rs"
+17
-0
samples/entries.rec less more
1 %rec: Article
2
3 Id: 1
4 Title: Article 1
5
6 Id: 2
7 Title: Article 2
8
9 %rec: Stock
10
11 Id: 1
12 Type: sell
13 Date: 20 April 2011
14
15 Id: 2
16 Type: stock
17 Date: 21 April 2011
+13
-0
samples/has_default.rec less more
1 Id: 1
2 Title: Blah
3
4 Id: 2
5 Title: Bleh
6
7 %rec: Movement
8
9 Date: 13-Aug-2012
10 Concept: 20
11
12 Date: 24-Sept-2012
13 Concept: 12
+154
-0
src/lib.rs less more
1 struct ParsingContext {
2 continuation_line: bool,
3 current_record_type: Option<String>,
4 }
5
6 #[derive(Eq, PartialEq, Debug)]
7 pub struct Record {
8 pub rec_type: Option<String>,
9 pub fields: Vec<(String, String)>,
10 }
11
12 #[derive(Eq, PartialEq, Debug)]
13 pub struct Recfile {
14 pub records: Vec<Record>,
15 }
16
17
18 impl Recfile {
19 pub fn parse<I>(i: I) -> Result<Recfile, String>
20 where I: std::io::BufRead
21 {
22 let mut iter = i.lines();
23 let mut current = Record {
24 fields: vec![],
25 rec_type: None,
26 };
27 let mut buf = vec![];
28 let mut ctx = ParsingContext {
29 continuation_line: false,
30 current_record_type: None,
31 };
32
33 while let Some(Ok(ln)) = iter.next() {
34 let ln = ln.trim_left_matches(' ');
35
36 if ln.starts_with('#') {
37 // skip comment lines
38 } else if ln.is_empty() {
39 if !current.fields.is_empty() {
40 buf.push(current);
41 current = Record {
42 rec_type: ctx.current_record_type.clone(),
43 fields: vec![],
44 };
45 }
46 } else if ln.starts_with('+') {
47 if let Some(val) = current.fields.last_mut() {
48 val.1.push_str("\n");
49 val.1.push_str(
50 if ln[1..].starts_with(' ') {
51 &ln[2..]
52 } else {
53 &ln[1..]
54 });
55 } else {
56 return Err(format!(
57 "Found continuation line in nonsensical place: {}",
58 ln));
59 }
60 } else if let Some(pos) = ln.find(':') {
61 let (key, val) = ln.split_at(pos);
62 current.fields.push((
63 key.to_owned(),
64 val[1..].trim_left().to_owned()));
65 if key == "%rec" {
66 ctx.current_record_type = Some(val[1..].trim_left().to_owned());
67 }
68 } else {
69 return Err(format!("Invalid line: {:?}", ln));
70 }
71 }
72
73 if !current.fields.is_empty() {
74 buf.push(current);
75 }
76
77 Ok(Recfile { records: buf })
78 }
79 }
80
81 #[cfg(test)]
82 mod tests {
83 use ::{Recfile,Record};
84
85 fn test_parse(input: &[u8], expected: Vec<Vec<(&str, &str)>>) {
86 let file = Recfile {
87 records: expected.iter().map( |v| {
88 Record {
89 fields: v.iter().map( |&(k, v)| {
90 (k.to_owned(), v.to_owned())
91 }).collect(),
92 }
93 }).collect(),
94 };
95 assert_eq!(Recfile::parse(input), Ok(file));
96 }
97
98 #[test]
99 fn empty_file() {
100 test_parse(b"\n", vec![]);
101 }
102
103 #[test]
104 fn only_comments() {
105 test_parse(b"# an empty file\n", vec![]);
106 }
107
108 #[test]
109 fn one_section() {
110 test_parse(b"hello: yes\n", vec![ vec![ ("hello", "yes") ] ]);
111 }
112
113 #[test]
114 fn two_sections() {
115 test_parse(
116 b"hello: yes\n\ngoodbye: no\n",
117 vec![
118 vec![ ("hello", "yes") ],
119 vec![ ("goodbye", "no") ],
120 ],
121 );
122 }
123
124 #[test]
125 fn continuation_with_space() {
126 test_parse(
127 b"hello: yes\n+ but also no\n",
128 vec![
129 vec![ ("hello", "yes\nbut also no") ],
130 ],
131 );
132 }
133
134 #[test]
135 fn continuation_without_space() {
136 test_parse(
137 b"hello: yes\n+but also no\n",
138 vec![
139 vec![ ("hello", "yes\nbut also no") ],
140 ],
141 );
142 }
143
144 #[test]
145 fn continuation_with_two_spaces() {
146 test_parse(
147 b"hello: yes\n+ but also no\n",
148 vec![
149 vec![ ("hello", "yes\n but also no") ],
150 ],
151 );
152 }
153
154 }
+13
-0
src/tools/pretty.rs less more
1 extern crate clap;
2 extern crate rrecutils;
3
4 fn main() {
5 let matches = clap::App::new("rr-pretty")
6 .version("0.0")
7 .author("Getty Ritter <rrecutils@infinitenegativeutility.com>")
8 .about("Display the Rust AST for a Recutils file")
9 .get_matches();
10 let source = std::io::stdin();
11 let records = rrecutils::Recfile::parse(source.lock());
12 println!("{:#?}", records);
13 }
+80
-0
src/tools/tojson.rs less more
1 extern crate clap;
2 extern crate rrecutils;
3 extern crate serde_json;
4
5 use std::{fmt,fs,io};
6
7 use serde_json::Value;
8 use serde_json::map::Map;
9
10 fn record_to_json(rec: &rrecutils::Record) -> Value {
11 let mut m = Map::new();
12 for tup in rec.fields.iter() {
13 let k = tup.0.clone();
14 let v = tup.1.clone();
15 m.insert(k, Value::String(v));
16 }
17 Value::Object(m)
18 }
19
20 fn unwrap_err<L, R: fmt::Debug>(value: Result<L, R>) -> L {
21 match value {
22 Ok(v) => v,
23 Err(err) => {
24 println!("{:?}", err);
25 std::process::exit(99)
26 }
27 }
28 }
29
30 fn main() {
31 let matches = clap::App::new("rr-to-json")
32 .version("0.0")
33 .author("Getty Ritter <rrecutils@infinitenegativeutility.com>")
34 .about("Display the Rust AST for a Recutils file")
35 .arg(clap::Arg::with_name("pretty")
36 .short("p")
37 .long("pretty")
38 .help("Pretty-print the resulting JSON"))
39 .arg(clap::Arg::with_name("input")
40 .short("i")
41 .long("input")
42 .value_name("FILE")
43 .help("The input recfile (or - for stdin)"))
44 .arg(clap::Arg::with_name("output")
45 .short("o")
46 .long("output")
47 .value_name("FILE")
48 .help("The desired output location (or - for stdout)"))
49 .get_matches();
50
51 let stdin = io::stdin();
52
53 let input: Box<io::BufRead> =
54 match matches.value_of("input").unwrap_or("-") {
55 "-" => Box::new(stdin.lock()),
56 path =>
57 Box::new(io::BufReader::new(unwrap_err(fs::File::open(path)))),
58 };
59
60 let json = Value::Array(unwrap_err(rrecutils::Recfile::parse(input))
61 .records
62 .iter()
63 .map(|x| record_to_json(x))
64 .collect());
65
66 let mut output: Box<io::Write> =
67 match matches.value_of("output").unwrap_or("-") {
68 "-" => Box::new(io::stdout()),
69 path => Box::new(unwrap_err(fs::File::open(path))),
70 };
71
72 let serialized = if matches.is_present("pretty") {
73 unwrap_err(serde_json::to_string_pretty(&json))
74 } else {
75 json.to_string()
76 };
77
78 unwrap_err(writeln!(output, "{}", serialized));
79
80 }