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