1 # SPDX-License-Identifier: GPL-3.0-or-later
3 # This file is part of Nominatim. (https://nominatim.org)
5 # Copyright (C) 2023 by the Nominatim developer community.
6 # For a full list of authors see the git log.
8 Create query interpretations where each vertice in the query is assigned
9 a specific function (expressed as a token type).
11 from typing import Optional, List, Iterator
14 import nominatim.api.search.query as qmod
15 from nominatim.api.logging import log
17 # pylint: disable=too-many-return-statements,too-many-branches
19 @dataclasses.dataclass
21 """ A token range for a specific type of tokens.
24 trange: qmod.TokenRange
27 PENALTY_TOKENCHANGE = {
28 qmod.BreakType.START: 0.0,
29 qmod.BreakType.END: 0.0,
30 qmod.BreakType.PHRASE: 0.0,
31 qmod.BreakType.WORD: 0.1,
32 qmod.BreakType.PART: 0.2,
33 qmod.BreakType.TOKEN: 0.4
36 TypedRangeSeq = List[TypedRange]
38 @dataclasses.dataclass
39 class TokenAssignment: # pylint: disable=too-many-instance-attributes
40 """ Representation of a possible assignment of token types
41 to the tokens in a tokenized query.
44 name: Optional[qmod.TokenRange] = None
45 address: List[qmod.TokenRange] = dataclasses.field(default_factory=list)
46 housenumber: Optional[qmod.TokenRange] = None
47 postcode: Optional[qmod.TokenRange] = None
48 country: Optional[qmod.TokenRange] = None
49 category: Optional[qmod.TokenRange] = None
50 qualifier: Optional[qmod.TokenRange] = None
54 def from_ranges(ranges: TypedRangeSeq) -> 'TokenAssignment':
55 """ Create a new token assignment from a sequence of typed spans.
57 out = TokenAssignment()
59 if token.ttype == qmod.TokenType.PARTIAL:
60 out.address.append(token.trange)
61 elif token.ttype == qmod.TokenType.HOUSENUMBER:
62 out.housenumber = token.trange
63 elif token.ttype == qmod.TokenType.POSTCODE:
64 out.postcode = token.trange
65 elif token.ttype == qmod.TokenType.COUNTRY:
66 out.country = token.trange
67 elif token.ttype == qmod.TokenType.CATEGORY:
68 out.category = token.trange
69 elif token.ttype == qmod.TokenType.QUALIFIER:
70 out.qualifier = token.trange
75 """ Working state used to put together the token assignements.
77 Represents an intermediate state while traversing the tokenized
80 def __init__(self, seq: TypedRangeSeq,
81 direction: int = 0, penalty: float = 0.0) -> None:
83 self.direction = direction
84 self.penalty = penalty
87 def __str__(self) -> str:
88 seq = ''.join(f'[{r.trange.start} - {r.trange.end}: {r.ttype.name}]' for r in self.seq)
89 return f'{seq} (dir: {self.direction}, penalty: {self.penalty})'
93 def end_pos(self) -> int:
94 """ Return the index of the global end of the current sequence.
96 return self.seq[-1].trange.end if self.seq else 0
99 def has_types(self, *ttypes: qmod.TokenType) -> bool:
100 """ Check if the current sequence contains any typed ranges of
103 return any(s.ttype in ttypes for s in self.seq)
106 def is_final(self) -> bool:
107 """ Return true when the sequence cannot be extended by any
108 form of token anymore.
110 # Country and category must be the final term for left-to-right
111 return len(self.seq) > 1 and \
112 self.seq[-1].ttype in (qmod.TokenType.COUNTRY, qmod.TokenType.CATEGORY)
115 def appendable(self, ttype: qmod.TokenType) -> Optional[int]:
116 """ Check if the give token type is appendable to the existing sequence.
118 Returns None if the token type is not appendable, otherwise the
119 new direction of the sequence after adding such a type. The
122 if ttype == qmod.TokenType.WORD:
126 # Append unconditionally to the empty list
127 if ttype == qmod.TokenType.COUNTRY:
129 if ttype in (qmod.TokenType.HOUSENUMBER, qmod.TokenType.QUALIFIER):
131 return self.direction
133 # Name tokens are always acceptable and don't change direction
134 if ttype == qmod.TokenType.PARTIAL:
135 return self.direction
137 # Other tokens may only appear once
138 if self.has_types(ttype):
141 if ttype == qmod.TokenType.HOUSENUMBER:
142 if self.direction == 1:
143 if len(self.seq) == 1 and self.seq[0].ttype == qmod.TokenType.QUALIFIER:
145 if len(self.seq) > 2 \
146 or self.has_types(qmod.TokenType.POSTCODE, qmod.TokenType.COUNTRY):
147 return None # direction left-to-right: housenumber must come before anything
148 elif self.direction == -1 \
149 or self.has_types(qmod.TokenType.POSTCODE, qmod.TokenType.COUNTRY):
150 return -1 # force direction right-to-left if after other terms
152 return self.direction
154 if ttype == qmod.TokenType.POSTCODE:
155 if self.direction == -1:
156 if self.has_types(qmod.TokenType.HOUSENUMBER, qmod.TokenType.QUALIFIER):
159 if self.direction == 1:
160 return None if self.has_types(qmod.TokenType.COUNTRY) else 1
161 if self.has_types(qmod.TokenType.HOUSENUMBER, qmod.TokenType.QUALIFIER):
163 return self.direction
165 if ttype == qmod.TokenType.COUNTRY:
166 return None if self.direction == -1 else 1
168 if ttype == qmod.TokenType.CATEGORY:
169 return self.direction
171 if ttype == qmod.TokenType.QUALIFIER:
172 if self.direction == 1:
173 if (len(self.seq) == 1
174 and self.seq[0].ttype in (qmod.TokenType.PARTIAL, qmod.TokenType.CATEGORY)) \
175 or (len(self.seq) == 2
176 and self.seq[0].ttype == qmod.TokenType.CATEGORY
177 and self.seq[1].ttype == qmod.TokenType.PARTIAL):
180 if self.direction == -1:
183 tempseq = self.seq[1:] if self.seq[0].ttype == qmod.TokenType.CATEGORY else self.seq
184 if len(tempseq) == 0:
186 if len(tempseq) == 1 and self.seq[0].ttype == qmod.TokenType.HOUSENUMBER:
188 if len(tempseq) > 1 or self.has_types(qmod.TokenType.POSTCODE, qmod.TokenType.COUNTRY):
195 def advance(self, ttype: qmod.TokenType, end_pos: int,
196 btype: qmod.BreakType) -> Optional['_TokenSequence']:
197 """ Return a new token sequence state with the given token type
200 newdir = self.appendable(ttype)
205 newseq = [TypedRange(ttype, qmod.TokenRange(0, end_pos))]
209 if btype != qmod.BreakType.PHRASE and last.ttype == ttype:
210 # extend the existing range
211 newseq = self.seq[:-1] + [TypedRange(ttype, last.trange.replace_end(end_pos))]
215 newseq = list(self.seq) + [TypedRange(ttype,
216 qmod.TokenRange(last.trange.end, end_pos))]
217 new_penalty = PENALTY_TOKENCHANGE[btype]
219 return _TokenSequence(newseq, newdir, self.penalty + new_penalty)
222 def _adapt_penalty_from_priors(self, priors: int, new_dir: int) -> bool:
226 if self.direction == 0:
227 self.direction = new_dir
234 def recheck_sequence(self) -> bool:
235 """ Check that the sequence is a fully valid token assignment
236 and addapt direction and penalties further if necessary.
238 This function catches some impossible assignments that need
239 forward context and can therefore not be exluded when building
242 # housenumbers may not be further than 2 words from the beginning.
243 # If there are two words in front, give it a penalty.
244 hnrpos = next((i for i, tr in enumerate(self.seq)
245 if tr.ttype == qmod.TokenType.HOUSENUMBER),
247 if hnrpos is not None:
248 if self.direction != -1:
249 priors = sum(1 for t in self.seq[:hnrpos] if t.ttype == qmod.TokenType.PARTIAL)
250 if not self._adapt_penalty_from_priors(priors, -1):
252 if self.direction != 1:
253 priors = sum(1 for t in self.seq[hnrpos+1:] if t.ttype == qmod.TokenType.PARTIAL)
254 if not self._adapt_penalty_from_priors(priors, 1):
260 def _get_assignments_postcode(self, base: TokenAssignment,
261 query_len: int) -> Iterator[TokenAssignment]:
262 """ Yield possible assignments of Postcode searches with an
265 assert base.postcode is not None
267 if (base.postcode.start == 0 and self.direction != -1)\
268 or (base.postcode.end == query_len and self.direction != 1):
269 log().comment('postcode search')
270 # <address>,<postcode> should give preference to address search
271 if base.postcode.start == 0:
272 penalty = self.penalty
273 self.direction = -1 # name searches are only possbile backwards
275 penalty = self.penalty + 0.1
276 self.direction = 1 # name searches are only possbile forwards
277 yield dataclasses.replace(base, penalty=penalty)
280 def _get_assignments_address_forward(self, base: TokenAssignment,
281 query: qmod.QueryStruct) -> Iterator[TokenAssignment]:
282 """ Yield possible assignments of address searches with
283 left-to-right reading.
285 first = base.address[0]
287 log().comment('first word = name')
288 yield dataclasses.replace(base, penalty=self.penalty,
289 name=first, address=base.address[1:])
292 # * if another name term comes after the first one and before the
294 # * a qualifier comes after the name
295 # * the containing phrase is strictly typed
296 if (base.housenumber and first.end < base.housenumber.start)\
297 or (base.qualifier and base.qualifier > first)\
298 or (query.nodes[first.start].ptype != qmod.PhraseType.NONE):
301 penalty = self.penalty
304 # * <name>, <street>, <housenumber> , ...
305 # * queries that are comma-separated
306 if (base.housenumber and base.housenumber > first) or len(query.source) > 1:
309 for i in range(first.start + 1, first.end):
310 name, addr = first.split(i)
311 log().comment(f'split first word = name ({i - first.start})')
312 yield dataclasses.replace(base, name=name, address=[addr] + base.address[1:],
313 penalty=penalty + PENALTY_TOKENCHANGE[query.nodes[i].btype])
316 def _get_assignments_address_backward(self, base: TokenAssignment,
317 query: qmod.QueryStruct) -> Iterator[TokenAssignment]:
318 """ Yield possible assignments of address searches with
319 right-to-left reading.
321 last = base.address[-1]
323 if self.direction == -1 or len(base.address) > 1:
324 log().comment('last word = name')
325 yield dataclasses.replace(base, penalty=self.penalty,
326 name=last, address=base.address[:-1])
329 # * if another name term comes before the last one and after the
331 # * a qualifier comes before the name
332 # * the containing phrase is strictly typed
333 if (base.housenumber and last.start > base.housenumber.end)\
334 or (base.qualifier and base.qualifier < last)\
335 or (query.nodes[last.start].ptype != qmod.PhraseType.NONE):
338 penalty = self.penalty
339 if base.housenumber and base.housenumber < last:
341 if len(query.source) > 1:
344 for i in range(last.start + 1, last.end):
345 addr, name = last.split(i)
346 log().comment(f'split last word = name ({i - last.start})')
347 yield dataclasses.replace(base, name=name, address=base.address[:-1] + [addr],
348 penalty=penalty + PENALTY_TOKENCHANGE[query.nodes[i].btype])
351 def get_assignments(self, query: qmod.QueryStruct) -> Iterator[TokenAssignment]:
352 """ Yield possible assignments for the current sequence.
354 This function splits up general name assignments into name
355 and address and yields all possible variants of that.
357 base = TokenAssignment.from_ranges(self.seq)
359 num_addr_tokens = sum(t.end - t.start for t in base.address)
360 if num_addr_tokens > 50:
363 # Postcode search (postcode-only search is covered in next case)
364 if base.postcode is not None and base.address:
365 yield from self._get_assignments_postcode(base, query.num_token_slots())
367 # Postcode or country-only search
369 if not base.housenumber and (base.postcode or base.country or base.category):
370 log().comment('postcode/country search')
371 yield dataclasses.replace(base, penalty=self.penalty)
373 # <postcode>,<address> should give preference to postcode search
374 if base.postcode and base.postcode.start == 0:
377 # Right-to-left reading of the address
378 if self.direction != -1:
379 yield from self._get_assignments_address_forward(base, query)
381 # Left-to-right reading of the address
382 if self.direction != 1:
383 yield from self._get_assignments_address_backward(base, query)
385 # variant for special housenumber searches
387 yield dataclasses.replace(base, penalty=self.penalty)
390 def yield_token_assignments(query: qmod.QueryStruct) -> Iterator[TokenAssignment]:
391 """ Return possible word type assignments to word positions.
393 The assignments are computed from the concrete tokens listed
394 in the tokenized query.
396 The result includes the penalty for transitions from one word type to
397 another. It does not include penalties for transitions within a
400 todo = [_TokenSequence([], direction=0 if query.source[0].ptype == qmod.PhraseType.NONE else 1)]
404 node = query.nodes[state.end_pos]
406 for tlist in node.starting:
407 newstate = state.advance(tlist.ttype, tlist.end, node.btype)
408 if newstate is not None:
409 if newstate.end_pos == query.num_token_slots():
410 if newstate.recheck_sequence():
411 log().var_dump('Assignment', newstate)
412 yield from newstate.get_assignments(query)
413 elif not newstate.is_final():
414 todo.append(newstate)