1 | """
|
2 | alloc.py - Arena manages SourceLine and Token instances (could rename)
|
3 | """
|
4 |
|
5 | from _devbuild.gen.syntax_asdl import source_t, Token, SourceLine, loc
|
6 | from asdl import runtime
|
7 | from core import error
|
8 | from mycpp.mylib import log
|
9 |
|
10 | from typing import List, Dict, Any
|
11 |
|
12 | _ = log
|
13 |
|
14 |
|
15 | class ctx_SourceCode(object):
|
16 |
|
17 | def __init__(self, arena, src):
|
18 | # type: (Arena, source_t) -> None
|
19 | arena.PushSource(src)
|
20 | self.arena = arena
|
21 |
|
22 | def __enter__(self):
|
23 | # type: () -> None
|
24 | pass
|
25 |
|
26 | def __exit__(self, type, value, traceback):
|
27 | # type: (Any, Any, Any) -> None
|
28 | self.arena.PopSource()
|
29 |
|
30 |
|
31 | class Arena(object):
|
32 | """Manages source_t, SourceLine, Token."""
|
33 |
|
34 | def __init__(self, save_tokens=False):
|
35 | # type: (bool) -> None
|
36 |
|
37 | self.save_tokens = save_tokens
|
38 |
|
39 | # indexed by span_id
|
40 | self.tokens = [] # type: List[Token]
|
41 | self.num_tokens = 0
|
42 |
|
43 | # Only used in tools
|
44 | self.span_id_lookup = {} # type: Dict[Token, int]
|
45 |
|
46 | # All lines that haven't been discarded. For LST formatting.
|
47 | self.lines_list = [] # type: List[SourceLine]
|
48 |
|
49 | # reuse these instances in many line_span instances
|
50 | self.source_instances = [] # type: List[source_t]
|
51 |
|
52 | def SaveTokens(self):
|
53 | # type: () -> None
|
54 | """
|
55 | Used by --tool X. Do we need LosslessArena?
|
56 | """
|
57 | self.save_tokens = True
|
58 |
|
59 | def PushSource(self, src):
|
60 | # type: (source_t) -> None
|
61 | self.source_instances.append(src)
|
62 |
|
63 | def PopSource(self):
|
64 | # type: () -> None
|
65 | self.source_instances.pop()
|
66 |
|
67 | def AddLine(self, line, line_num):
|
68 | # type: (str, int) -> SourceLine
|
69 | """Save a physical line and return a line_id for later retrieval.
|
70 |
|
71 | The line number is 1-based.
|
72 | """
|
73 | src_line = SourceLine(line_num, line, self.source_instances[-1])
|
74 | self.lines_list.append(src_line)
|
75 | return src_line
|
76 |
|
77 | def DiscardLines(self):
|
78 | # type: () -> None
|
79 | """Remove references ot lines we've accumulated.
|
80 |
|
81 | - This makes the linear search in SnipCodeString() shorter.
|
82 | - It removes the ARENA's references to all lines. The TOKENS still
|
83 | reference some lines.
|
84 | """
|
85 | #log("discarding %d lines", len(self.lines_list))
|
86 | del self.lines_list[:]
|
87 |
|
88 | def SnipCodeString(self, left, right, inclusive=True):
|
89 | # type: (Token, Token, bool) -> str
|
90 | """Return the code string between left and right tokens, INCLUSIVE.
|
91 |
|
92 | Used for ALIAS expansion, which happens in the PARSER.
|
93 |
|
94 | The argument to aliases can span multiple lines, like this:
|
95 |
|
96 | $ myalias '1 2 3'
|
97 | """
|
98 | if inclusive:
|
99 | ileft = left.col
|
100 | iright = right.col + right.length
|
101 | else:
|
102 | ileft = left.col + left.length
|
103 | iright = right.col
|
104 |
|
105 | pieces = [] # type: List[str]
|
106 | if not inclusive:
|
107 | pieces.append(' ' * ileft)
|
108 |
|
109 | if left.line == right.line:
|
110 | for li in self.lines_list:
|
111 | if li == left.line:
|
112 | pieces.append(li.content[ileft:iright])
|
113 | return ''.join(pieces)
|
114 |
|
115 | saving = False
|
116 | found_left = False
|
117 | found_right = False
|
118 | for li in self.lines_list:
|
119 | if li == left.line:
|
120 | found_left = True
|
121 | saving = True
|
122 |
|
123 | # Save everything after the left token
|
124 | piece = li.content[ileft:]
|
125 | pieces.append(piece)
|
126 | #log(' %r', piece)
|
127 | continue
|
128 |
|
129 | if li == right.line:
|
130 | found_right = True
|
131 |
|
132 | piece = li.content[:iright]
|
133 | pieces.append(piece)
|
134 | #log(' %r', piece)
|
135 |
|
136 | saving = False
|
137 | break
|
138 |
|
139 | if saving:
|
140 | pieces.append(li.content)
|
141 | #log(' %r', li.content)
|
142 |
|
143 | assert found_left, "Couldn't find left token"
|
144 | assert found_right, "Couldn't find right token"
|
145 | return ''.join(pieces)
|
146 |
|
147 | def NewToken(self, id_, col, length, src_line):
|
148 | # type: (int, int, int, SourceLine) -> Token
|
149 |
|
150 | if length >= 65536:
|
151 | raise error.Parse(
|
152 | '', # ignored message
|
153 | loc.TokenTooLong(src_line, id_, length, col))
|
154 |
|
155 | tok = Token(id_, length, col, src_line, None)
|
156 | if self.save_tokens:
|
157 | span_id = self.num_tokens
|
158 | self.num_tokens += 1
|
159 |
|
160 | self.tokens.append(tok)
|
161 | self.span_id_lookup[tok] = span_id
|
162 | return tok
|
163 |
|
164 | def UnreadOne(self):
|
165 | # type: () -> None
|
166 | """Reuse the last span ID."""
|
167 | if self.save_tokens:
|
168 | self.tokens.pop()
|
169 | self.num_tokens -= 1
|
170 |
|
171 | def GetToken(self, span_id):
|
172 | # type: (int) -> Token
|
173 | assert span_id != runtime.NO_SPID, span_id
|
174 | assert span_id < len(self.tokens), \
|
175 | 'Span ID out of range: %d is greater than %d' % (span_id, len(self.tokens))
|
176 | return self.tokens[span_id]
|
177 |
|
178 | def GetSpanId(self, tok):
|
179 | # type: (Token) -> int
|
180 | """Given a Token, returns its a sequence number"""
|
181 | #return tok.span_id
|
182 | #return -1
|
183 | assert tok in self.span_id_lookup
|
184 | return self.span_id_lookup[tok]
|
185 |
|
186 | def LastSpanId(self):
|
187 | # type: () -> int
|
188 | """Return one past the last span ID."""
|
189 | return len(self.tokens)
|
190 |
|
191 |
|
192 | class LosslessArena(Arena):
|
193 | """
|
194 | TODO:
|
195 |
|
196 | Has lossless invariant, for
|
197 | --tool fmt
|
198 | --tool ysh-ify
|
199 |
|
200 | Retains all SourceLine and Token
|
201 |
|
202 | Somehow disallow re-parsing? Is that equivalent to ctx_SourceCode()?
|
203 | """
|
204 | pass
|
205 |
|
206 |
|
207 | class DynamicArena(Arena):
|
208 | """
|
209 | For batch and interactive shell
|
210 |
|
211 | TODO:
|
212 | - Test that SourceLine and Token are GC'd
|
213 |
|
214 | However, it should support:
|
215 | - SnipCodeString() for aliases
|
216 | - SnipCodeBlock() for Hay
|
217 |
|
218 | Neither of those are necessary in the LosslessArena? We might have
|
219 | different utilities there.
|
220 | """
|
221 | pass
|