grammar.rs 27.9 KB
Newer Older
1
use std::collections::BTreeMap as Map;
P
Phodal Huang 已提交
2

P
Phodal Huang 已提交
3
use crate::grammar::line_tokens::{LineTokens, TokenTypeMatcher};
P
Phodal Huang 已提交
4 5
use crate::grammar::local_stack_element::LocalStackElement;
use crate::grammar::{MatchRuleResult, ScopeListElement, StackElement};
P
Phodal Huang 已提交
6
use crate::inter::{IRawGrammar, IRawRepository, IRawRepositoryMap, IRawRule};
P
Phodal Huang 已提交
7
use crate::rule::abstract_rule::RuleEnum;
P
Phodal Huang 已提交
8
use crate::rule::rule_factory::RuleFactory;
P
Phodal Huang 已提交
9 10 11
use crate::rule::{
    AbstractRule, BeginWhileRule, EmptyRule, IGrammarRegistry, IRuleFactoryHelper, IRuleRegistry,
};
P
Phodal Huang 已提交
12
use core::cmp;
P
Phodal Huang 已提交
13
use scie_scanner::scanner::scanner::IOnigCaptureIndex;
P
Phodal Huang 已提交
14

P
Phodal Huang 已提交
15 16 17 18 19 20 21 22 23 24 25 26 27
pub struct IToken {
    pub start_index: i32,
    pub end_index: i32,
    pub scopes: Vec<String>,
}

pub struct ITokenizeLineResult {
    pub tokens: Vec<IToken>,
    pub rule_stack: Box<StackElement>,
}

pub struct ITokenizeLineResult2 {
    pub tokens: Vec<i32>,
P
Phodal Huang 已提交
28
    pub rule_stack: Box<StackElement>,
P
Phodal Huang 已提交
29 30 31 32 33
}

pub trait IGrammar {
    fn tokenize_line(line_text: String, prev_state: Option<StackElement>) -> ITokenizeLineResult;
    /**
P
Phodal Huang 已提交
34 35 36 37 38 39 40 41 42
     * Tokenize `lineText` using previous line state `prevState`.
     * The result contains the tokens in binary format, resolved with the following information:
     *  - language
     *  - token type (regex, string, comment, other)
     *  - font style
     *  - foreground color
     *  - background color
     * e.g. for getting the languageId: `(metadata & MetadataConsts.LANGUAGEID_MASK) >>> MetadataConsts.LANGUAGEID_OFFSET`
     */
P
Phodal Huang 已提交
43
    fn tokenize_line2(line_text: String, prev_state: Option<StackElement>) -> ITokenizeLineResult2;
P
Phodal Huang 已提交
44 45
}

P
Phodal Huang 已提交
46
pub trait Matcher {}
P
Phodal Huang 已提交
47

P
Phodal Huang 已提交
48 49 50
#[derive(Debug, Clone)]
pub struct CheckWhileRuleResult {
    pub rule: Box<BeginWhileRule>,
51 52 53 54 55 56
    pub stack: Box<StackElement>,
}

#[derive(Debug, Clone)]
pub struct CheckWhileConditionResult {
    pub stack: Box<StackElement>,
P
Phodal Huang 已提交
57
    pub line_pos: i32,
58
    pub anchor_position: i32,
P
Phodal Huang 已提交
59
    pub is_first_line: bool,
P
Phodal Huang 已提交
60 61
}

P
Phodal Huang 已提交
62 63 64 65
#[derive(Debug, Clone)]
pub struct TokenizeResult {
    line_length: usize,
    line_tokens: Box<LineTokens>,
P
Phodal Huang 已提交
66
    rule_stack: Box<Option<StackElement>>,
P
Phodal Huang 已提交
67 68
}

P
Phodal Huang 已提交
69
#[derive(Debug, Clone)]
P
Phodal Huang 已提交
70
pub struct Grammar {
71
    root_id: i32,
P
Phodal Huang 已提交
72
    grammar: IRawGrammar,
73
    pub last_rule_id: i32,
74
    pub rule_id2desc: Map<i32, Box<dyn AbstractRule>>,
P
Phodal Huang 已提交
75
    pub _token_type_matchers: Vec<TokenTypeMatcher>,
P
Phodal Huang 已提交
76 77
}

P
Phodal Huang 已提交
78
pub fn init_grammar(grammar: IRawGrammar, _base: Option<IRawRule>) -> IRawGrammar {
P
Phodal Huang 已提交
79 80 81
    let mut _grammar = grammar.clone();

    let mut new_based: IRawRule = IRawRule::new();
P
Phodal Huang 已提交
82 83 84
    if let Some(repo) = grammar.clone().repository {
        new_based.location = repo.clone().location;
    }
P
Phodal Huang 已提交
85 86
    new_based.patterns = Some(grammar.clone().patterns.clone());
    new_based.name = grammar.clone().name;
P
Phodal Huang 已提交
87 88 89 90

    let mut repository_map = IRawRepositoryMap::new();
    repository_map.base_s = Some(new_based.clone());
    repository_map.self_s = Some(new_based.clone());
P
Phodal Huang 已提交
91 92 93
    if let Some(repo) = grammar.clone().repository {
        repository_map.name_map = repo.clone().map.name_map.clone();
    }
P
Phodal Huang 已提交
94 95 96

    _grammar.repository = Some(IRawRepository {
        map: Box::new(repository_map.clone()),
97
        location: None,
P
Phodal Huang 已提交
98 99 100 101 102
    });

    _grammar
}

P
Phodal Huang 已提交
103
impl Grammar {
P
Phodal Huang 已提交
104
    pub fn new(grammar: IRawGrammar) -> Grammar {
P
Phodal Huang 已提交
105
        let _grammar = init_grammar(grammar.clone(), None);
P
Phodal Huang 已提交
106
        Grammar {
107
            last_rule_id: 0,
P
Phodal Huang 已提交
108
            grammar: _grammar,
P
Phodal Huang 已提交
109
            root_id: -1,
110
            rule_id2desc: Map::new(),
P
Phodal Huang 已提交
111
            _token_type_matchers: vec![],
P
Phodal Huang 已提交
112 113 114
        }
    }

P
Phodal Huang 已提交
115
    fn tokenize(
116
        &mut self,
P
Phodal Huang 已提交
117
        line_text: String,
P
Phodal Huang 已提交
118
        prev_state: &mut Option<StackElement>,
P
Phodal Huang 已提交
119
        emit_binary_tokens: bool,
P
Phodal Huang 已提交
120
    ) -> TokenizeResult {
121
        if self.root_id.clone() == -1 {
P
Phodal Huang 已提交
122
            let repository = self.grammar.repository.clone().unwrap();
P
Phodal Huang 已提交
123
            let based = repository.clone().map.self_s.unwrap();
P
Phodal Huang 已提交
124 125 126 127 128 129
            self.root_id = RuleFactory::get_compiled_rule_id(
                based.clone(),
                self,
                &mut repository.clone(),
                String::from(""),
            );
130
        }
P
Phodal Huang 已提交
131

P
Phodal Huang 已提交
132
        let mut is_first_line: bool = false;
133 134

        let mut current_state = StackElement::null();
P
Phodal Huang 已提交
135
        match prev_state.clone() {
P
Phodal Huang 已提交
136
            None => is_first_line = true,
137 138 139 140
            Some(state) => {
                if state == StackElement::null() {
                    is_first_line = true
                }
141 142

                current_state = state;
P
Phodal Huang 已提交
143
            }
144
        }
P
Phodal Huang 已提交
145

P
Phodal Huang 已提交
146
        if is_first_line {
P
Phodal Huang 已提交
147
            let _root_scope_name = self.get_rule(self.root_id.clone()).get_name(None, None);
P
Phodal Huang 已提交
148 149 150 151 152
            let mut root_scope_name = String::from("unknown");
            if let Some(name) = _root_scope_name {
                root_scope_name = name
            }

P
Phodal Huang 已提交
153
            let scope_list = ScopeListElement::new(None, root_scope_name);
154
            let state = StackElement::new(
P
Phodal Huang 已提交
155 156 157 158 159 160 161 162
                None,
                self.root_id.clone(),
                -1,
                -1,
                false,
                None,
                scope_list.clone(),
                scope_list.clone(),
163 164 165
            );

            current_state = state;
P
Phodal Huang 已提交
166 167
        } else {
            is_first_line = false;
P
Phodal Huang 已提交
168
            current_state.reset();
P
Phodal Huang 已提交
169 170
        }

171
        let format_line_text = line_text.clone() + "\n";
P
Phodal Huang 已提交
172
        let mut line_tokens = LineTokens::new(
P
Phodal Huang 已提交
173 174 175 176
            emit_binary_tokens,
            line_text,
            self._token_type_matchers.clone(),
        );
P
Phodal Huang 已提交
177 178
        let next_state = self.tokenize_string(
            format_line_text.clone(),
P
Phodal Huang 已提交
179 180
            is_first_line,
            0,
P
Phodal Huang 已提交
181
            current_state,
182
            &mut line_tokens,
P
Phodal Huang 已提交
183
            true,
184
        );
P
Phodal Huang 已提交
185 186 187 188

        TokenizeResult {
            line_length: format_line_text.clone().len(),
            line_tokens: Box::new(line_tokens),
P
Phodal Huang 已提交
189
            rule_stack: Box::new(next_state),
P
Phodal Huang 已提交
190
        }
P
Phodal Huang 已提交
191 192
    }

P
Phodal Huang 已提交
193 194 195
    pub fn tokenize_string(
        &mut self,
        line_text: String,
196 197
        origin_is_first: bool,
        origin_line_pos: i32,
P
Phodal Huang 已提交
198
        mut stack: StackElement,
P
Phodal Huang 已提交
199
        line_tokens: &mut LineTokens,
P
Phodal Huang 已提交
200
        check_while_conditions: bool,
201
    ) -> Option<StackElement> {
P
Phodal Huang 已提交
202
        let _line_length = line_text.len();
203
        let mut _stop = false;
204
        let mut anchor_position = -1;
205 206
        let mut line_pos = origin_line_pos.clone();
        let mut is_first_line = origin_is_first.clone();
P
Phodal Huang 已提交
207 208

        if check_while_conditions {
209
            let while_check_result = self.check_while_conditions(
P
Phodal Huang 已提交
210
                line_text.clone(),
211 212
                origin_is_first.clone(),
                origin_line_pos.clone(),
213
                stack.clone(),
P
Phodal Huang 已提交
214 215
                line_tokens.clone(),
            );
216 217 218 219
            stack = *while_check_result.stack;
            line_pos = while_check_result.line_pos;
            is_first_line = while_check_result.is_first_line;
            anchor_position = while_check_result.anchor_position;
P
Phodal Huang 已提交
220 221
        }

222
        while !_stop {
P
Phodal Huang 已提交
223 224 225 226
            let r = self.match_rule(
                line_text.clone(),
                is_first_line,
                line_pos,
P
Phodal Huang 已提交
227
                &mut stack,
P
Phodal Huang 已提交
228 229
                anchor_position,
            );
230
            if let None = r {
P
Phodal Huang 已提交
231
                line_tokens.produce(&mut stack, _line_length as i32);
232
                _stop = true;
P
Phodal Huang 已提交
233
                return Some(stack.clone());
234 235
            }

P
Phodal Huang 已提交
236 237 238 239
            let capture_result = r.unwrap();
            let capture_indices = capture_result.capture_indices;
            let matched_rule_id = capture_result.matched_rule_id;
            if matched_rule_id == -1 {
240 241 242
                let _popped_rule = stack.get_rule(self);
                if let RuleEnum::BeginEndRule(popped_rule) = _popped_rule.get_rule_instance() {
                    let name_scopes_list = stack.clone().name_scopes_list;
P
Phodal Huang 已提交
243
                    line_tokens.produce(&mut stack, capture_indices[0].start.clone() as i32);
244 245 246
                    stack = stack.set_content_name_scopes_list(name_scopes_list);
                    Grammar::handle_captures(
                        self,
247
                        line_text.clone(),
248 249 250 251 252 253 254 255 256 257 258 259 260
                        is_first_line,
                        &mut stack,
                        line_tokens,
                        popped_rule.end_captures,
                        capture_indices.clone(),
                    );

                    line_tokens.produce(&mut stack, capture_indices[0].end as i32);
                    let popped = stack.clone();
                    if let Some(_stack) = stack.pop() {
                        stack = _stack;
                    }
                    anchor_position = popped.anchor_pos;
P
Phodal Huang 已提交
261 262
                } else {
                    println!("_popped_rule {:?}", _popped_rule.clone());
263
                }
P
Phodal Huang 已提交
264 265
            } else {
                let rule = self.get_rule(matched_rule_id);
P
Phodal Huang 已提交
266
                line_tokens.produce(&mut stack, capture_indices[0].start as i32);
P
Phodal Huang 已提交
267
                // let before_push = stack.clone();
P
Phodal Huang 已提交
268 269
                let scope_name =
                    rule.get_name(Some(line_text.clone()), Some(capture_indices.clone()));
P
Phodal Huang 已提交
270
                let name_scopes_list = stack.content_name_scopes_list.push(self, scope_name);
P
Phodal Huang 已提交
271 272 273 274
                let mut begin_rule_capture_eol = false;
                if capture_indices[0].end == _line_length {
                    begin_rule_capture_eol = true;
                }
P
Phodal Huang 已提交
275
                stack = stack.push(
P
Phodal Huang 已提交
276 277 278 279 280 281
                    matched_rule_id,
                    line_pos,
                    anchor_position,
                    begin_rule_capture_eol,
                    None,
                    name_scopes_list.clone(),
P
Phodal Huang 已提交
282
                    name_scopes_list.clone(),
P
Phodal Huang 已提交
283 284
                );

P
Phodal Huang 已提交
285 286
                match rule.get_rule_instance() {
                    RuleEnum::BeginEndRule(begin_rule) => {
287
                        let push_rule = begin_rule.clone();
P
Phodal Huang 已提交
288
                        Grammar::handle_captures(
P
Phodal Huang 已提交
289 290 291
                            self,
                            line_text.clone(),
                            is_first_line,
P
Phodal Huang 已提交
292
                            &mut stack,
293
                            line_tokens,
P
Phodal Huang 已提交
294 295 296
                            begin_rule.begin_captures,
                            capture_indices.clone(),
                        );
P
Phodal Huang 已提交
297

P
Phodal Huang 已提交
298
                        line_tokens.produce(&mut stack, capture_indices[0].end.clone() as i32);
299
                        anchor_position = capture_indices[0].end.clone() as i32;
P
Phodal Huang 已提交
300 301 302 303
                        let content_name = push_rule.get_content_name(
                            Some(line_text.clone()),
                            Some(capture_indices.clone()),
                        );
P
Phodal Huang 已提交
304
                        let _content_name_scopes_list = name_scopes_list.push(self, content_name);
305
                        stack = stack.set_content_name_scopes_list(_content_name_scopes_list);
306 307 308 309
                        // todo: not used
                        // if push_rule.endHasBackReferences {
                        //
                        // }
310

311 312 313
                        // if (!hasAdvanced && beforePush.hasSameRuleAs(stack)) {
                        // _stop = true;
                        // return None;
314
                    }
P
Phodal Huang 已提交
315
                    RuleEnum::BeginWhileRule(_while_rule) => {
P
Phodal Huang 已提交
316
                        println!("todo: RuleEnum - BeginWhileRule");
317
                        _stop = true;
P
Phodal Huang 已提交
318
                        return Some(stack.clone());
319
                    }
P
Phodal Huang 已提交
320 321 322 323 324 325 326 327 328 329 330 331 332 333 334
                    RuleEnum::MatchRule(match_rule) => {
                        Grammar::handle_captures(
                            self,
                            line_text.clone(),
                            is_first_line,
                            &mut stack,
                            line_tokens,
                            match_rule.captures,
                            capture_indices.clone(),
                        );
                        line_tokens.produce(&mut stack, capture_indices[0].end.clone() as i32);
                        if let Some(_stack) = stack.pop() {
                            stack = _stack;
                        }
                    }
335
                    _ => {
P
Phodal Huang 已提交
336
                        println!("todo: RuleEnum - Others");
337
                        _stop = true;
P
Phodal Huang 已提交
338
                        return Some(stack.clone());
P
Phodal Huang 已提交
339 340
                    }
                }
P
Phodal Huang 已提交
341
            }
342 343 344 345 346

            if capture_indices[0].end > line_pos as usize {
                line_pos = capture_indices[0].end as i32;
                is_first_line = false;
            }
347
        }
348
        Some(stack.clone())
P
Phodal Huang 已提交
349 350
    }

P
Phodal Huang 已提交
351 352 353 354
    pub fn handle_captures(
        grammar: &mut Grammar,
        line_text: String,
        is_first_line: bool,
355 356
        stack: &mut StackElement,
        line_tokens: &mut LineTokens,
P
Phodal Huang 已提交
357 358
        captures: Vec<Box<dyn AbstractRule>>,
        capture_indices: Vec<IOnigCaptureIndex>,
P
Phodal Huang 已提交
359
    ) -> Option<LineTokens> {
P
Phodal Huang 已提交
360 361
        let captures_len = captures.clone().len();
        if captures_len == 0 {
P
Phodal Huang 已提交
362
            return None;
P
Phodal Huang 已提交
363 364 365 366 367 368 369
        }

        let len = cmp::min(captures_len, capture_indices.len());
        let mut local_stack: Vec<LocalStackElement> = vec![];
        let max_end = capture_indices[0].end;
        for i in 0..len {
            let capture_rule = captures[i].clone();
P
Phodal Huang 已提交
370 371 372 373
            if let RuleEnum::CaptureRule(capture) = capture_rule.get_rule_instance() {
                if capture.clone().rule._type == "" {
                    continue;
                }
P
Phodal Huang 已提交
374

P
Phodal Huang 已提交
375 376 377 378
                let capture_index = capture_indices[i].clone();
                if capture_index.length == 0 {
                    continue;
                }
P
Phodal Huang 已提交
379

P
Phodal Huang 已提交
380 381 382
                if capture_index.start > max_end {
                    continue;
                }
383

P
Phodal Huang 已提交
384 385 386 387 388 389 390 391 392 393
                while local_stack.len() > 0
                    && local_stack[local_stack.len() - 1].end_pos <= capture_index.start as i32
                {
                    let mut local_stack_element = local_stack[local_stack.len() - 1].clone();
                    line_tokens.produce_from_scopes(
                        &mut local_stack_element.scopes,
                        local_stack_element.end_pos,
                    );
                    local_stack.pop();
                }
394

P
Phodal Huang 已提交
395 396 397 398 399 400 401 402 403
                if local_stack.len() > 0 {
                    let mut local_stack_element = local_stack[local_stack.len() - 1].clone();
                    line_tokens.produce_from_scopes(
                        &mut local_stack_element.scopes,
                        capture_index.start as i32,
                    );
                } else {
                    line_tokens.produce(stack, capture_index.start as i32);
                }
404

P
Phodal Huang 已提交
405
                if capture.retokenize_captured_with_rule_id != 0 {
P
Phodal Huang 已提交
406 407 408 409 410
                    let scope_name =
                        capture.get_name(Some(line_text.clone()), Some(capture_indices.clone()));
                    let name_scopes_list = stack.content_name_scopes_list.push(grammar, scope_name);
                    let content_name = capture
                        .get_content_name(Some(line_text.clone()), Some(capture_indices.clone()));
P
Phodal Huang 已提交
411 412
                    let content_name_scopes_list = name_scopes_list.push(grammar, content_name);

413
                    let stack_clone = stack.clone().push(
P
Phodal Huang 已提交
414 415 416 417 418 419 420 421 422 423 424 425 426
                        capture.retokenize_captured_with_rule_id,
                        capture_index.start.clone() as i32,
                        -1,
                        false,
                        None,
                        name_scopes_list,
                        content_name_scopes_list,
                    );

                    let sub_text = line_text.split_at(capture_index.end).0;
                    let mut sub_is_first_line = false;
                    if is_first_line && capture_index.start == 0 {
                        sub_is_first_line = true;
427
                    }
P
Phodal Huang 已提交
428 429 430 431 432 433 434 435 436 437 438
                    Grammar::tokenize_string(
                        grammar,
                        String::from(sub_text),
                        sub_is_first_line,
                        capture_index.start as i32,
                        stack_clone,
                        line_tokens,
                        false,
                    );
                    // todo: find the next_text_not_switch_issues
                    continue;
439
                }
P
Phodal Huang 已提交
440

P
Phodal Huang 已提交
441 442
                let capture_scope_name =
                    capture_rule.get_name(Some(line_text.clone()), Some(capture_indices.clone()));
P
Phodal Huang 已提交
443
                if let Some(_name) = capture_scope_name.clone() {
P
Phodal Huang 已提交
444 445 446 447 448 449 450 451 452
                    let mut base = stack.clone().content_name_scopes_list;
                    if local_stack.len() > 0 {
                        base = local_stack[local_stack.len() - 1].clone().scopes;
                    }
                    let capture_rule_scopes_list = base.push(grammar, capture_scope_name.clone());
                    local_stack.push(LocalStackElement::new(
                        capture_rule_scopes_list,
                        capture_index.end as i32,
                    ));
P
Phodal Huang 已提交
453
                }
P
Phodal Huang 已提交
454 455
            } else {
                println!("lose rule: {:?}", capture_rule.clone());
P
Phodal Huang 已提交
456 457 458 459
            }
        }

        while local_stack.len() > 0 {
P
Phodal Huang 已提交
460 461
            let mut last_stack = local_stack[local_stack.len() - 1].clone();
            line_tokens.produce_from_scopes(&mut last_stack.scopes, last_stack.end_pos);
P
Phodal Huang 已提交
462
            local_stack.pop();
P
Phodal Huang 已提交
463
        }
P
Phodal Huang 已提交
464 465

        return Some(line_tokens.to_owned());
P
Phodal Huang 已提交
466
    }
P
Phodal Huang 已提交
467 468 469 470 471
    /**
     * Walk the stack from bottom to top, and check each while condition in this order.
     * If any fails, cut off the entire stack above the failed while condition. While conditions
     * may also advance the linePosition.
     */
P
Phodal Huang 已提交
472 473 474 475 476
    pub fn check_while_conditions(
        &mut self,
        line_text: String,
        is_first_line: bool,
        line_pos: i32,
477
        mut stack: StackElement,
P
Phodal Huang 已提交
478
        _line_tokens: LineTokens,
479
    ) -> CheckWhileConditionResult {
P
Phodal Huang 已提交
480
        let mut anchor_position = -1;
P
Phodal Huang 已提交
481
        if stack.begin_rule_captured_eol {
P
Phodal Huang 已提交
482 483
            anchor_position = 0
        }
P
Phodal Huang 已提交
484 485 486 487 488 489 490 491
        let mut while_rules = vec![];
        let mut has_node = true;
        let mut node = stack.clone();
        while has_node {
            let rule = node.clone().get_rule(self);
            if let RuleEnum::BeginWhileRule(begin_rule) = rule.get_rule_instance() {
                while_rules.push(CheckWhileRuleResult {
                    rule: Box::from(begin_rule),
492
                    stack: Box::from(node.clone()),
P
Phodal Huang 已提交
493 494 495 496
                })
            }

            match node.pop() {
P
Phodal Huang 已提交
497
                None => has_node = false,
P
Phodal Huang 已提交
498 499
                Some(n) => {
                    node = n;
500
                }
P
Phodal Huang 已提交
501 502 503
            }
        }

504 505 506 507 508 509 510 511
        for while_rule in while_rules.clone() {
            let allow_g = anchor_position == line_pos;
            let mut rule_scanner = while_rule.clone().rule.compile_while(
                self,
                while_rule.clone().stack.end_rule,
                is_first_line,
                allow_g,
            );
P
Phodal Huang 已提交
512 513 514
            let match_result = rule_scanner
                .scanner
                .find_next_match_sync(line_text.clone(), line_pos);
515 516 517 518 519 520 521 522 523
            match match_result {
                None => {
                    stack = while_rule.stack.pop().unwrap();
                    break;
                }
                Some(_) => {
                    println!("todo: check_while_conditions");
                }
            }
P
Phodal Huang 已提交
524 525
        }

526 527 528 529 530
        // println!("{:?}", while_rules);
        CheckWhileConditionResult {
            stack: Box::new(stack),
            line_pos,
            anchor_position,
P
Phodal Huang 已提交
531
            is_first_line,
532
        }
P
Phodal Huang 已提交
533
    }
P
Phodal Huang 已提交
534

P
Phodal Huang 已提交
535 536 537 538 539
    pub fn match_rule_or_injections(
        &mut self,
        line_text: String,
        is_first_line: bool,
        line_pos: i32,
P
Phodal Huang 已提交
540
        stack: &mut StackElement,
P
Phodal Huang 已提交
541
        anchor_position: i32,
P
Phodal Huang 已提交
542
    ) {
P
Phodal Huang 已提交
543 544
        let match_result =
            self.match_rule(line_text, is_first_line, line_pos, stack, anchor_position);
P
Phodal Huang 已提交
545 546
        if let Some(_result) = match_result {
        } else {
547 548 549
            // None
        };
        // todo: get injections logic
P
Phodal Huang 已提交
550 551 552 553 554 555 556
    }

    pub fn match_rule(
        &mut self,
        line_text: String,
        is_first_line: bool,
        line_pos: i32,
P
Phodal Huang 已提交
557
        stack: &mut StackElement,
P
Phodal Huang 已提交
558
        anchor_position: i32,
559
    ) -> Option<MatchRuleResult> {
560
        let mut rule = stack.get_rule(self);
P
Phodal Huang 已提交
561
        let rule_info = rule.clone().get_rule_instance();
P
Phodal Huang 已提交
562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582

        let mut rule_scanner;
        match rule_info {
            RuleEnum::BeginEndRule(mut begin_end_rule) => {
                rule_scanner = begin_end_rule.compile(
                    self,
                    stack.end_rule.clone(),
                    is_first_line,
                    line_pos == anchor_position,
                );
            }
            _ => {
                rule_scanner = rule.compile(
                    self,
                    stack.end_rule.clone(),
                    is_first_line,
                    line_pos == anchor_position,
                );
            }
        }

P
Phodal Huang 已提交
583 584 585
        let r = rule_scanner
            .scanner
            .find_next_match_sync(line_text, line_pos);
P
Phodal Huang 已提交
586
        if let Some(result) = r {
587 588
            let match_rule_result = MatchRuleResult {
                capture_indices: result.capture_indices,
589
                matched_rule_id: rule_scanner.rules[result.index],
590 591 592
            };

            Some(match_rule_result)
P
Phodal Huang 已提交
593 594 595
        } else {
            None
        }
P
Phodal Huang 已提交
596
    }
P
Phodal Huang 已提交
597

P
Phodal Huang 已提交
598 599 600
    pub fn tokenize_line(
        &mut self,
        line_text: String,
P
Phodal Huang 已提交
601
        prev_state: &mut Option<StackElement>,
P
Phodal Huang 已提交
602
    ) -> TokenizeResult {
P
Phodal Huang 已提交
603 604 605
        self.tokenize(line_text, prev_state, false)
    }

P
Phodal Huang 已提交
606
    pub fn tokenize_line2(&self, _line_text: String, _prev_state: Option<StackElement>) {}
P
Phodal Huang 已提交
607
}
P
Phodal Huang 已提交
608 609 610 611

impl IRuleFactoryHelper for Grammar {}

impl IGrammarRegistry for Grammar {
P
Phodal Huang 已提交
612 613
    fn get_external_grammar(
        &self,
P
Phodal Huang 已提交
614 615
        _scope_name: String,
        _repository: IRawRepository,
P
Phodal Huang 已提交
616
    ) -> Option<IRawGrammar> {
P
Phodal Huang 已提交
617 618 619 620 621
        None
    }
}

impl IRuleRegistry for Grammar {
P
Phodal Huang 已提交
622 623
    fn register_id(&mut self) -> i32 {
        self.last_rule_id = self.last_rule_id + 1;
P
Phodal Huang 已提交
624
        self.last_rule_id.clone()
P
Phodal Huang 已提交
625 626
    }

P
Phodal Huang 已提交
627 628 629
    fn get_rule(&mut self, pattern_id: i32) -> Box<dyn AbstractRule> {
        if let Some(rule) = self.rule_id2desc.get_mut(&pattern_id) {
            return rule.to_owned();
P
Phodal Huang 已提交
630
        }
P
Phodal Huang 已提交
631
        Box::from(EmptyRule {})
P
Phodal Huang 已提交
632
    }
P
Phodal Huang 已提交
633

P
Phodal Huang 已提交
634
    fn register_rule(&mut self, result: Box<dyn AbstractRule>) -> Box<dyn AbstractRule> {
P
Phodal Huang 已提交
635
        self.rule_id2desc
P
Phodal Huang 已提交
636
            .insert(result.id().clone(), result.clone());
637
        result
P
Phodal Huang 已提交
638
    }
P
Phodal Huang 已提交
639 640 641 642
}

#[cfg(test)]
mod tests {
P
Phodal Huang 已提交
643
    use std::fs::File;
644
    use std::io::{Read, Write};
P
Phodal Huang 已提交
645
    use std::path::Path;
P
Phodal Huang 已提交
646

P
Phodal Huang 已提交
647
    use crate::grammar::{Grammar, StackElement};
P
Phodal Huang 已提交
648
    use crate::inter::IRawGrammar;
P
Phodal Huang 已提交
649
    use crate::rule::abstract_rule::RuleEnum;
P
Phodal Huang 已提交
650
    use crate::rule::IRuleRegistry;
P
Phodal Huang 已提交
651

P
Phodal Huang 已提交
652
    #[test]
P
Phodal Huang 已提交
653
    fn should_build_json_code() {
654 655 656 657 658 659 660
        let code = "
#include <stdio.h>
int main() {
printf(\"Hello, World!\");
return 0;
}
";
P
Phodal Huang 已提交
661
        let grammar = to_grammar_with_code("test-cases/first-mate/fixtures/c.json", code);
P
Phodal Huang 已提交
662
        assert_eq!(28, grammar.rule_id2desc.get(&1).unwrap().patterns_length());
P
Phodal Huang 已提交
663 664 665 666 667 668 669
        debug_output(&grammar, String::from("program.json"));
    }

    #[test]
    fn should_identify_c_include() {
        let code = "#include <stdio.h>";
        let grammar = to_grammar_with_code("test-cases/first-mate/fixtures/c.json", code);
P
Phodal Huang 已提交
670
        assert_eq!(grammar.rule_id2desc.len(), 161);
P
Phodal Huang 已提交
671
        debug_output(&grammar, String::from("program.json"));
672 673
    }

P
Phodal Huang 已提交
674 675 676
    #[test]
    fn should_build_text_grammar() {
        let code = "
P
Phodal Huang 已提交
677
GitHub 漫游指南
P
Phodal Huang 已提交
678
";
P
Phodal Huang 已提交
679
        let grammar = to_grammar_with_code("test-cases/first-mate/fixtures/text.json", code);
680
        assert_eq!(grammar.rule_id2desc.len(), 8);
681 682 683
    }

    fn debug_output(grammar: &Grammar, path: String) {
P
Phodal Huang 已提交
684
        let j = serde_json::to_string(&grammar.rule_id2desc).unwrap();
685
        let mut file = File::create(path).unwrap();
P
Phodal Huang 已提交
686
        match file.write_all(j.as_bytes()) {
P
Phodal Huang 已提交
687 688
            Ok(_) => {}
            Err(_) => {}
P
Phodal Huang 已提交
689
        };
P
Phodal Huang 已提交
690 691
    }

692 693 694
    #[test]
    fn should_build_json_grammar() {
        let code = "{}";
P
Phodal Huang 已提交
695
        let grammar = to_grammar_with_code("test-cases/first-mate/fixtures/json.json", code);
696 697 698 699 700 701 702
        assert_eq!(grammar.rule_id2desc.len(), 22);
        debug_output(&grammar, String::from("program.json"));
    }

    #[test]
    fn should_build_html_grammar() {
        let code = "{}";
P
Phodal Huang 已提交
703
        let grammar = to_grammar_with_code("test-cases/first-mate/fixtures/html.json", code);
P
Phodal Huang 已提交
704
        assert_eq!(grammar.rule_id2desc.len(), 101);
705 706 707
        debug_output(&grammar, String::from("program.json"));
    }

P
Phodal Huang 已提交
708 709 710 711 712 713 714
    #[test]
    fn should_build_correct_end_rule_id_for_makefile() {
        let code = "CC=gcc
CFLAGS=-I.
DEPS = hellomake.h
OBJ = hellomake.o hellofunc.o
";
P
Phodal Huang 已提交
715
        let mut grammar = to_grammar_with_code("test-cases/first-mate/fixtures/makefile.json", code);
P
Phodal Huang 已提交
716
        let mut end_rule_count = 0;
P
Phodal Huang 已提交
717
        for (_x, rule) in grammar.rule_id2desc.clone() {
P
Phodal Huang 已提交
718 719
            let rule_instance = rule.get_rule_instance();
            if let RuleEnum::BeginEndRule(rule) = rule_instance {
P
Phodal Huang 已提交
720
                assert_eq!(rule._end.rule_id, -1);
P
Phodal Huang 已提交
721
                end_rule_count = end_rule_count + 1;
P
Phodal Huang 已提交
722 723
            }
        }
P
Phodal Huang 已提交
724
        assert_eq!(grammar.get_rule(1).patterns_length(), 4);
P
Phodal Huang 已提交
725
        assert_eq!(end_rule_count, 24);
P
Phodal Huang 已提交
726 727 728
        debug_output(&grammar, String::from("program.json"));
    }

P
Phodal Huang 已提交
729 730
    #[test]
    fn should_build_makefile_grammar() {
731 732 733 734 735 736 737 738 739 740
        let code = "CC=gcc
CFLAGS=-I.
DEPS = hellomake.h
OBJ = hellomake.o hellofunc.o

%.o: %.c $(DEPS)
	$(CC) -c -o $@ $< $(CFLAGS)

hellomake: $(OBJ)
	$(CC) -o $@ $^ $(CFLAGS)
P
Phodal Huang 已提交
741
";
P
Phodal Huang 已提交
742
        let mut grammar = to_grammar_with_code("test-cases/first-mate/fixtures/makefile.json", code);
P
Phodal Huang 已提交
743
        assert_eq!(grammar.rule_id2desc.len(), 82);
744
        assert_eq!(grammar.get_rule(1).patterns_length(), 4);
P
Phodal Huang 已提交
745 746 747
        debug_output(&grammar, String::from("program.json"));
    }

P
Phodal Huang 已提交
748 749
    #[test]
    fn should_resolve_make_file_error_issues() {
750
        let code = "%.o: %.c $(DEPS)
P
Phodal Huang 已提交
751
";
P
Phodal Huang 已提交
752
        let mut grammar = to_grammar_with_code("test-cases/first-mate/fixtures/makefile.json", code);
P
Phodal Huang 已提交
753
        let result = grammar.tokenize_line(String::from("%.o: %.c $(DEPS)"), &mut None);
P
Phodal Huang 已提交
754 755 756 757 758 759 760 761 762 763 764
        let tokens = result.line_tokens._tokens.clone();
        assert_eq!(8, tokens.len());
        assert_eq!("Makefile,meta.scope.target.makefile,entity.name.function.target.makefile,constant.other.placeholder.makefile", tokens[0].scopes.join(","));
        assert_eq!(0, tokens[0].start_index);
        assert_eq!(1, tokens[1].start_index);
        assert_eq!(3, tokens[2].start_index);
        assert_eq!(4, tokens[3].start_index);
        assert_eq!(9, tokens[4].start_index);
        assert_eq!(11, tokens[5].start_index);
        assert_eq!(15, tokens[6].start_index);
        assert_eq!(16, tokens[7].start_index);
P
Phodal Huang 已提交
765 766 767
        debug_output(&grammar, String::from("program.json"));
    }

P
Phodal Huang 已提交
768 769 770
    #[test]
    fn should_resolve_make_file_error_issues2() {
        let code = "hellomake: $(OBJ)
P
Phodal Huang 已提交
771
\t$(CC) -o $@ $^ $(CFLAGS)";
P
Phodal Huang 已提交
772
        let grammar = to_grammar_with_code("test-cases/first-mate/fixtures/makefile.json", code);
P
Phodal Huang 已提交
773 774 775
        debug_output(&grammar, String::from("program.json"));
    }

P
Phodal Huang 已提交
776 777
    fn to_grammar_with_code(grammar_path: &str, code: &str) -> Grammar {
        let mut grammar = to_grammar(grammar_path);
778
        let c_code = String::from(code);
P
Phodal Huang 已提交
779
        let mut rule_stack = Some(StackElement::null());
P
Phodal Huang 已提交
780
        for line in c_code.lines() {
P
Phodal Huang 已提交
781 782
            let result = grammar.tokenize_line(String::from(line), &mut rule_stack);
            rule_stack = *result.rule_stack;
P
Phodal Huang 已提交
783 784 785
            for token in result.line_tokens._tokens {
                let start = token.start_index.clone() as usize;
                let end = token.end_index.clone() as usize;
P
Phodal Huang 已提交
786 787 788 789 790
                let new_line: String = String::from(line)
                    .chars()
                    .skip(start)
                    .take(end - start)
                    .collect();
P
Phodal Huang 已提交
791
                let token_str: String = token.scopes.join(", ");
P
Phodal Huang 已提交
792 793 794 795
                println!(
                    " - token from {:?} to {:?} ({:?}) with scopes {:?}",
                    token.start_index, token.end_index, new_line, token_str
                )
P
Phodal Huang 已提交
796
            }
P
Phodal Huang 已提交
797 798
        };

799
        grammar
P
Phodal Huang 已提交
800
    }
P
Phodal Huang 已提交
801 802 803 804 805 806 807 808 809 810

    fn to_grammar(grammar_path: &str) -> Grammar {
        let path = Path::new(grammar_path);
        let mut file = File::open(path).unwrap();
        let mut data = String::new();
        file.read_to_string(&mut data).unwrap();

        let g: IRawGrammar = serde_json::from_str(&data).unwrap();
        Grammar::new(g)
    }
P
Phodal Huang 已提交
811
}