about summary refs log tree commit diff stats
path: root/013literal_string.cc
blob: 29a185b2ac8a9431a873a858815087a52e1e3862 (plain) (blame)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
//: For convenience, some instructions will take literal arrays of characters (strings).
//:
//: Instead of quotes, we'll use [] to delimit strings. That'll reduce the
//: need for escaping since we can support nested brackets. And we can also
//: imagine that 'recipe' might one day itself be defined in mu, doing its own
//: parsing.

:(scenarios load)
:(scenario string_literal)
recipe main [
  1:address:array:character <- copy [abc def]  # copy can't really take a string
]
+parse:   ingredient: {name: "abc def", properties: [_: "literal-string"]}

:(scenario string_literal_with_colons)
recipe main [
  1:address:array:character <- copy [abc:def/ghi]
]
+parse:   ingredient: {name: "abc:def/ghi", properties: [_: "literal-string"]}

:(before "End Mu Types Initialization")
Type_ordinal["literal-string"] = 0;

:(after "string next_word(istream& in)")
  if (in.peek() == '[') {
    string result = slurp_quoted(in);
    skip_whitespace(in);
    skip_comment(in);
//?     cerr << '^' << result << "$\n"; //? 1
    return result;
  }

:(code)
string slurp_quoted(istream& in) {
  ostringstream out;
  assert(!in.eof());  assert(in.peek() == '[');  out << static_cast<char>(in.get());  // slurp the '['
  if (code_string(in, out))
    slurp_quoted_comment_aware(in, out);
  else
    slurp_quoted_comment_oblivious(in, out);
  return out.str();
}

// A string is a code string if it contains a newline before any non-whitespace
// todo: support comments before the newline. But that gets messy.
bool code_string(istream& in, ostringstream& out) {
  while (!in.eof()) {
    char c = in.get();
    if (!isspace(c)) {
      in.putback(c);
//?       cerr << "code_string: " << out.str() << '\n'; //? 1
      return false;
    }
    out << c;
    if (c == '\n') {
//?       cerr << "code_string: " << out.str() << '\n'; //? 1
      return true;
    }
  }
  return false;
}

// Read a regular string. Regular strings can only contain other regular
// strings.
void slurp_quoted_comment_oblivious(istream& in, ostringstream& out) {
//?   cerr << "comment oblivious\n"; //? 1
  int brace_depth = 1;
  while (!in.eof()) {
    char c = in.get();
//?     cerr << '%' << (int)c << ' ' << brace_depth << ": " << out.str() << "%$\n"; //? 1
//?     cout << (int)c << ": " << brace_depth << '\n'; //? 2
    if (c == '\\') {
      out << static_cast<char>(in.get());
      continue;
    }
    out << c;
//?     cout << out.str() << "$\n"; //? 1
    if (c == '[') ++brace_depth;
    if (c == ']') --brace_depth;
    if (brace_depth == 0) break;
  }
  if (in.eof() && brace_depth > 0) {
    raise << "unbalanced '['\n" << end();
    out.clear();
  }
}

// Read a code string. Code strings can contain either code or regular strings.
void slurp_quoted_comment_aware(istream& in, ostringstream& out) {
//?   cerr << "comment aware\n"; //? 1
  char c;
  while (in >> c) {
//?     cerr << '^' << (int)c << ": " << out.str() << "$\n"; //? 1
    if (c == '\\') {
      out << static_cast<char>(in.get());
      continue;
    }
    if (c == '#') {
      out << c;
      while (!in.eof() && in.peek() != '\n') out << static_cast<char>(in.get());
      continue;
    }
    if (c == '[') {
      in.putback(c);
      // recurse
      out << slurp_quoted(in);
      continue;
    }
    out << c;
    if (c == ']') return;
  }
  raise << "unbalanced '['\n" << end();
  out.clear();
}

:(after "reagent::reagent(string s)")
//?   cout << s.at(0) << '\n'; //? 1
  if (s.at(0) == '[') {
    assert(*s.rbegin() == ']');
    // delete [] delimiters
    s.erase(0, 1);
    s.erase(SIZE(s)-1);
    name = s;
    types.push_back(0);
    properties.push_back(pair<string, vector<string> >(name, vector<string>()));
    properties.back().second.push_back("literal-string");
    return;
  }

//: Two tweaks to printing literal strings compared to other reagents:
//:   a) Don't print the string twice in the representation, just put '_' in
//:   the property list.
//:   b) Escape newlines in the string to make it more friendly to trace().

:(after "string reagent::to_string()")
  if (!properties.at(0).second.empty() && properties.at(0).second.at(0) == "literal-string") {
    return emit_literal_string(name);
  }

:(code)
string emit_literal_string(string name) {
  size_t pos = 0;
  while (pos != string::npos)
    pos = replace(name, "\n", "\\n", pos);
  return "{name: \""+name+"\", properties: [_: \"literal-string\"]}";
}

size_t replace(string& str, const string& from, const string& to, size_t n) {
  size_t result = str.find(from, n);
  if (result != string::npos)
    str.replace(result, from.length(), to);
  return result;
}

:(scenario string_literal_nested)
recipe main [
  1:address:array:character <- copy [abc [def]]
]
+parse:   ingredient: {name: "abc [def]", properties: [_: "literal-string"]}

:(scenario string_literal_escaped)
recipe main [
  1:address:array:character <- copy [abc \[def]
]
+parse:   ingredient: {name: "abc [def", properties: [_: "literal-string"]}

:(scenario string_literal_escaped_comment_aware)
recipe main [
  1:address:array:character <- copy [
abc \\\[def]
]
+parse:   ingredient: {name: "\nabc \[def", properties: [_: "literal-string"]}

:(scenario string_literal_and_comment)
recipe main [
  1:address:array:character <- copy [abc]  # comment
]
+parse: instruction: copy
+parse:   ingredient: {name: "abc", properties: [_: "literal-string"]}
+parse:   product: {name: "1", properties: ["1": "address":"array":"character"]}
# no other ingredients
$parse: 3

:(scenario string_literal_escapes_newlines_in_trace)
recipe main [
  copy [abc
def]
]
+parse:   ingredient: {name: "abc\ndef", properties: [_: "literal-string"]}

:(scenario string_literal_can_skip_past_comments)
recipe main [
  copy [
    # ']' inside comment
    bar
  ]
]
+parse:   ingredient: {name: "\n    # ']' inside comment\n    bar\n  ", properties: [_: "literal-string"]}

:(scenario string_literal_empty)
recipe main [
  copy []
]
+parse:   ingredient: {name: "", properties: [_: "literal-string"]}