1cb93a386Sopenharmony_ci/*
2cb93a386Sopenharmony_ci * Copyright 2017 Google Inc.
3cb93a386Sopenharmony_ci *
4cb93a386Sopenharmony_ci * Use of this source code is governed by a BSD-style license that can be
5cb93a386Sopenharmony_ci * found in the LICENSE file.
6cb93a386Sopenharmony_ci */
7cb93a386Sopenharmony_ci
8cb93a386Sopenharmony_ci#ifndef SKSL_REGEXPARSER
9cb93a386Sopenharmony_ci#define SKSL_REGEXPARSER
10cb93a386Sopenharmony_ci
11cb93a386Sopenharmony_ci#include "src/sksl/lex/RegexNode.h"
12cb93a386Sopenharmony_ci
13cb93a386Sopenharmony_ci#include <stack>
14cb93a386Sopenharmony_ci#include <string>
15cb93a386Sopenharmony_ci
16cb93a386Sopenharmony_ci/**
17cb93a386Sopenharmony_ci * Turns a simple regular expression into a parse tree. The regular expression syntax supports only
18cb93a386Sopenharmony_ci * the basic quantifiers ('*', '+', and '?'), alternation ('|'), character sets ('[a-z]'), and
19cb93a386Sopenharmony_ci * groups ('()').
20cb93a386Sopenharmony_ci */
21cb93a386Sopenharmony_ciclass RegexParser {
22cb93a386Sopenharmony_cipublic:
23cb93a386Sopenharmony_ci    RegexNode parse(std::string source);
24cb93a386Sopenharmony_ci
25cb93a386Sopenharmony_ciprivate:
26cb93a386Sopenharmony_ci    inline static constexpr char END = '\0';
27cb93a386Sopenharmony_ci
28cb93a386Sopenharmony_ci    char peek();
29cb93a386Sopenharmony_ci
30cb93a386Sopenharmony_ci    void expect(char c);
31cb93a386Sopenharmony_ci
32cb93a386Sopenharmony_ci    RegexNode pop();
33cb93a386Sopenharmony_ci
34cb93a386Sopenharmony_ci    /**
35cb93a386Sopenharmony_ci     * Matches a char literal, parenthesized group, character set, or dot ('.').
36cb93a386Sopenharmony_ci     */
37cb93a386Sopenharmony_ci    void term();
38cb93a386Sopenharmony_ci
39cb93a386Sopenharmony_ci    /**
40cb93a386Sopenharmony_ci     * Matches a term followed by an optional quantifier ('*', '+', or '?').
41cb93a386Sopenharmony_ci     */
42cb93a386Sopenharmony_ci    void quantifiedTerm();
43cb93a386Sopenharmony_ci
44cb93a386Sopenharmony_ci    /**
45cb93a386Sopenharmony_ci     * Matches a sequence of quantifiedTerms.
46cb93a386Sopenharmony_ci     */
47cb93a386Sopenharmony_ci    void sequence();
48cb93a386Sopenharmony_ci
49cb93a386Sopenharmony_ci    /**
50cb93a386Sopenharmony_ci     * Returns a node representing the given escape character (e.g. escapeSequence('n') returns a
51cb93a386Sopenharmony_ci     * node which matches a newline character).
52cb93a386Sopenharmony_ci     */
53cb93a386Sopenharmony_ci    RegexNode escapeSequence(char c);
54cb93a386Sopenharmony_ci
55cb93a386Sopenharmony_ci    /**
56cb93a386Sopenharmony_ci     * Matches a literal character or escape sequence.
57cb93a386Sopenharmony_ci     */
58cb93a386Sopenharmony_ci    void literal();
59cb93a386Sopenharmony_ci
60cb93a386Sopenharmony_ci    /**
61cb93a386Sopenharmony_ci     * Matches a dot ('.').
62cb93a386Sopenharmony_ci     */
63cb93a386Sopenharmony_ci    void dot();
64cb93a386Sopenharmony_ci
65cb93a386Sopenharmony_ci    /**
66cb93a386Sopenharmony_ci     * Matches a parenthesized group.
67cb93a386Sopenharmony_ci     */
68cb93a386Sopenharmony_ci    void group();
69cb93a386Sopenharmony_ci
70cb93a386Sopenharmony_ci    /**
71cb93a386Sopenharmony_ci     * Matches a literal character, escape sequence, or character range from a character set.
72cb93a386Sopenharmony_ci     */
73cb93a386Sopenharmony_ci    void setItem();
74cb93a386Sopenharmony_ci
75cb93a386Sopenharmony_ci    /**
76cb93a386Sopenharmony_ci     * Matches a character set.
77cb93a386Sopenharmony_ci     */
78cb93a386Sopenharmony_ci    void set();
79cb93a386Sopenharmony_ci
80cb93a386Sopenharmony_ci    void regex();
81cb93a386Sopenharmony_ci
82cb93a386Sopenharmony_ci    std::string fSource;
83cb93a386Sopenharmony_ci
84cb93a386Sopenharmony_ci    size_t fIndex;
85cb93a386Sopenharmony_ci
86cb93a386Sopenharmony_ci    std::stack<RegexNode> fStack;
87cb93a386Sopenharmony_ci};
88cb93a386Sopenharmony_ci
89cb93a386Sopenharmony_ci#endif
90