Source code for lupyne.engine.documents

Wrappers for lucene Fields and Documents.

from future_builtins import map
import datetime
import calendar
import operator
import warnings
import lucene  # noqa
from java.lang import Double, Float, Long, Number, Object
from java.util import Arrays, HashSet
from org.apache.lucene import document, index, search, util
from import grouping
from .queries import Query

[docs]class Field(document.FieldType): """Saved parameters which can generate lucene Fields given values. :param name: name of field :param boost: boost factor :param stored, indexed, settings: lucene FieldType attributes """ attrs = {name[3].lower() + name[4:] for name in dir(document.FieldType) if name.startswith('set')} def __init__(self, name, stored=False, indexed=True, boost=1.0, **settings): document.FieldType.__init__(self), self.boost = name, boost self.update(stored=stored, indexed=indexed, **settings) def update(self, docValueType='', indexOptions='', numericType='', **settings): if docValueType: self.setDocValueType(getattr(index.FieldInfo.DocValuesType, docValueType.upper())) if indexOptions: self.setIndexOptions(getattr(index.FieldInfo.IndexOptions, indexOptions.upper())) if numericType: self.setNumericType(getattr(document.FieldType.NumericType, numericType.upper())) for name in settings: getattr(self, 'set' + name[:1].upper() + name[1:])(settings[name]) @property def settings(self): "dict representation of settings" defaults = document.FieldType() result = {'indexed': self.indexed()} for name in Field.attrs: value = getattr(self, name)() if value != getattr(defaults, name)(): result[name] = value if isinstance(value, int) else str(value) return result
[docs] def items(self, *values): "Generate lucene Fields suitable for adding to a document." for value in values: field = document.Field(, value, self) field.setBoost(self.boost) yield field
[docs]class MapField(Field): """Field which applies a function across its values. :param func: callable """ def __init__(self, name, func, **kwargs): Field.__init__(self, name, **kwargs) self.func = func
[docs] def items(self, *values): "Generate fields with mapped values." return Field.items(self, *map(self.func, values))
[docs]class NestedField(Field): """Field which indexes every component into its own field. Original value may be stored for convenience. :param sep: field separator used on name and values """ def __init__(self, name, sep='.', tokenized=False, **kwargs): Field.__init__(self, name, tokenized=tokenized, **kwargs) self.sep = sep self.names = tuple(self.values(name))
[docs] def values(self, value): "Generate component field values in order." value = value.split(self.sep) for index in range(1, len(value) + 1): yield self.sep.join(value[:index])
[docs] def items(self, *values): "Generate indexed component fields." for value in values: for index, text in enumerate(self.values(value)): yield document.Field(self.names[index], text, self)
[docs] def prefix(self, value): "Return prefix query of the closest possible prefixed field." index = value.count(self.sep) return Query.prefix(self.names[index], value)
[docs] def range(self, start, stop, lower=True, upper=False): "Return range query of the closest possible prefixed field." index = max(value.count(self.sep) for value in (start, stop) if value is not None) return Query.range(self.names[index], start, stop, lower, upper)
[docs]class DocValuesField(Field): """Field which stores a per-document values, used for efficient sorting. :param name: name of field :param type: lucene DocValuesType string """ def __init__(self, name, type): Field.__init__(self, name, indexed=False, docValueType=type) self.cls = getattr(document, type.title().replace('_', '') + 'DocValuesField')
[docs] def items(self, *values): "Generate lucene DocValuesFields suitable for adding to a document." for value in values: yield self.cls(, long(value) if isinstance(value, int) else util.BytesRef(value))
[docs]class NumericField(Field): """Field which indexes numbers in a prefix tree. :param name: name of field :param type: optional int, float, or lucene NumericType string """ def __init__(self, name, type=None, tokenized=False, **kwargs): if type: kwargs['numericType'] = {int: 'long', float: 'double'}.get(type, str(type)) Field.__init__(self, name, tokenized=tokenized, **kwargs)
[docs] def items(self, *values): "Generate lucene NumericFields suitable for adding to a document." if not self.numericType(): cls, = set(map(type, values)) self.update(numericType='double' if issubclass(cls, float) else 'long') for value in values: if isinstance(value, float): yield document.DoubleField(, value, self) else: yield document.LongField(, long(value), self)
[docs] def range(self, start, stop, lower=True, upper=False): "Return lucene NumericRangeQuery." step = self.numericPrecisionStep() if isinstance(start, float) or isinstance(stop, float): start, stop = (value if value is None else Double(value) for value in (start, stop)) return search.NumericRangeQuery.newDoubleRange(, step, start, stop, lower, upper) if start is not None: start = None if start < Long.MIN_VALUE else Long(long(start)) if stop is not None: stop = None if stop > Long.MAX_VALUE else Long(long(stop)) return search.NumericRangeQuery.newLongRange(, step, start, stop, lower, upper)
[docs] def term(self, value): "Return range query to match single term." return self.range(value, value, upper=True)
[docs] def filter(self, *args, **kwargs): ".. deprecated:: 1.9 convert NumericRangeQuery with `Query`_.filter instead" warnings.warn(NumericField.filter.__doc__, DeprecationWarning) return Query.filter(self.range(*args, **kwargs), cache=False)
[docs]class DateTimeField(NumericField): """Field which indexes datetimes as a NumericField of timestamps. Supports datetimes, dates, and any prefix of time tuples. """ def __init__(self, name, **kwargs): NumericField.__init__(self, name, type=float, **kwargs) @classmethod
[docs] def timestamp(cls, date): "Return utc timestamp from date or time tuple." if isinstance(date, return calendar.timegm(date.timetuple()) + getattr(date, 'microsecond', 0) * 1e-6 return float(calendar.timegm(tuple(date) + (None, 1, 1, 0, 0, 0)[len(date):]))
[docs] def items(self, *dates): "Generate lucene NumericFields of timestamps." return NumericField.items(self, *map(self.timestamp, dates))
[docs] def range(self, start, stop, lower=True, upper=False): "Return NumericRangeQuery of timestamps." start, stop = (date and self.timestamp(date) for date in (start, stop)) return NumericField.range(self, start, stop, lower, upper)
[docs] def prefix(self, date): "Return range query which matches the date prefix." if isinstance(date, date = date.timetuple()[:6 if isinstance(date, datetime.datetime) else 3] if len(date) == 2 and date[1] == 12: # month must be valid return self.range(date, (date[0] + 1, 1)) return self.range(date, tuple(date[:-1]) + (date[-1] + 1,))
[docs] def duration(self, date, days=0, **delta): """Return date range query within time span of date. :param date: origin date or tuple :param days,delta: timedelta parameters """ if not isinstance(date, date = datetime.datetime(*(tuple(date) + (None, 1, 1)[len(date):])) delta = datetime.timedelta(days, **delta) return self.range(*sorted([date, date + delta]), upper=True)
[docs] def within(self, days=0, weeks=0, utc=True, **delta): """Return date range query within current time and delta. If the delta is an exact number of days, then dates will be used. :param days,weeks: number of days to offset from today :param utc: optionally use utc instead of local time :param delta: additional timedelta parameters """ date = datetime.datetime.utcnow() if utc else if not (isinstance(days + weeks, float) or delta): date = return self.duration(date, days, weeks=weeks, **delta)
[docs]class Document(dict): "Multimapping of field names to values, but default getters return the first value." def __init__(self, doc): for field in doc.iterator(): value = convert(field.numericValue() or field.stringValue() or field.binaryValue()) self.setdefault(, []).append(value) def __getitem__(self, name): return dict.__getitem__(self, name)[0] def get(self, name, default=None): return dict.get(self, name, [default])[0]
[docs] def getlist(self, name): "Return list of all values for given field." return dict.get(self, name, [])
[docs] def dict(self, *names, **defaults): """Return dict representation of document. :param names: names of multi-valued fields to return as a list :param defaults: include only given fields, using default values as necessary """ defaults.update((name, self[name]) for name in (defaults or self) if name in self) defaults.update((name, self.getlist(name)) for name in names) return defaults
def convert(value): "Return python object from java Object." if util.BytesRef.instance_(value): return util.BytesRef.cast_(value).utf8ToString() if not Number.instance_(value): return value.toString() if Object.instance_(value) else value value = Number.cast_(value) return value.doubleValue() if Float.instance_(value) or Double.instance_(value) else int(value.longValue())
[docs]class Hit(Document): "A Document from a search result, with :attr:`id`, :attr:`score`, and optional sort :attr:`keys`." def __init__(self, doc, id, score, keys=()): Document.__init__(self, doc), self.score = id, score self.keys = tuple(map(convert, keys))
[docs] def dict(self, *names, **defaults): "Return dict representation of document with __id__, __score__, and any sort __keys__." result = Document.dict(self, *names, **defaults) result.update(, __score__=self.score) if self.keys: result['__keys__'] = self.keys return result
[docs]class Hits(object): """Search results: lazily evaluated and memory efficient. Provides a read-only sequence interface to hit objects. :param searcher: `IndexSearcher`_ which can retrieve documents :param scoredocs: lucene ScoreDocs :param count: total number of hits :param maxscore: maximum score :param fields: optional field selectors """ def __init__(self, searcher, scoredocs, count=None, maxscore=None, fields=None): self.searcher, self.scoredocs = searcher, scoredocs self.count, self.maxscore = count, maxscore self.fields = fields
[docs] def select(self, *fields): "Only load selected fields." self.fields = HashSet(Arrays.asList(fields))
[docs] def __len__(self): return len(self.scoredocs)
[docs] def __getitem__(self, index): if isinstance(index, slice): start, stop, step = index.indices(len(self)) assert step == 1, 'slice step is not supported' scoredocs = self.scoredocs[start:stop] if stop - start < len(self) else self.scoredocs return type(self)(self.searcher, scoredocs, self.count, self.maxscore, self.fields) scoredoc = self.scoredocs[index] keys = search.FieldDoc.cast_(scoredoc).fields if search.FieldDoc.instance_(scoredoc) else () doc = self.searcher.doc(scoredoc.doc, self.fields) return Hit(doc, scoredoc.doc, scoredoc.score, keys)
@property def ids(self): return map(operator.attrgetter('doc'), self.scoredocs) @property def scores(self): return map(operator.attrgetter('score'), self.scoredocs)
[docs] def items(self): "Generate zipped ids and scores." return map(operator.attrgetter('doc', 'score'), self.scoredocs)
[docs] def groupby(self, func, count=None, docs=None): """Return ordered list of `Hits`_ grouped by value of function applied to doc ids. Optionally limit the number of groups and docs per group.""" groups = {} for scoredoc in self.scoredocs: value = func(scoredoc.doc) try: group = groups[value] except KeyError: group = groups[value] = type(self)(self.searcher, [], fields=self.fields) group.index, group.value = len(groups), value group.scoredocs.append(scoredoc) groups = sorted(groups.values(), key=lambda group: group.__dict__.pop('index')) for group in groups: group.count, group.maxscore = len(group), max(group.scores) group.scoredocs = group.scoredocs[:docs] return Groups(self.searcher, groups[:count], len(groups), self.maxscore, self.fields)
[docs] def filter(self, func): "Return `Hits`_ filtered by function applied to doc ids." scoredocs = [scoredoc for scoredoc in self.scoredocs if func(scoredoc.doc)] return type(self)(self.searcher, scoredocs, fields=self.fields)
[docs] def sorted(self, key, reverse=False): "Return `Hits`_ sorted by key function applied to doc ids." scoredocs = sorted(self.scoredocs, key=lambda scoredoc: key(scoredoc.doc), reverse=reverse) return type(self)(self.searcher, scoredocs, self.count, self.maxscore, self.fields)
[docs]class Groups(object): "Sequence of grouped `Hits`_." select = Hits.__dict__['select'] def __init__(self, searcher, groupdocs, count=None, maxscore=None, fields=None): self.searcher, self.groupdocs = searcher, groupdocs self.count, self.maxscore = count, maxscore self.fields = fields
[docs] def __len__(self): return len(self.groupdocs)
[docs] def __getitem__(self, index): hits = groupdocs = self.groupdocs[index] if isinstance(groupdocs, grouping.GroupDocs): hits = Hits(self.searcher, groupdocs.scoreDocs, groupdocs.totalHits, groupdocs.maxScore) hits.value = convert(groupdocs.groupValue) hits.fields = self.fields return hits
@property def facets(self): "mapping of field values and counts" return {hits.value: hits.count for hits in self}
[docs]class GroupingSearch(grouping.GroupingSearch): """Inherited lucene GroupingSearch with optimized faceting. :param field: unique field name to group by :param sort: lucene Sort to order groups and docs :param cache: use unlimited caching :param attrs: additional attributes to set """ def __init__(self, field, sort=None, cache=True, **attrs): grouping.GroupingSearch.__init__(self, field) self.field = field if sort: self.groupSort = self.sortWithinGroup = sort self.fillSortFields = True if cache: self.setCachingInMB(float('inf'), True) for name in attrs: getattr(type(self), name).__set__(self, attrs[name])
[docs] def __len__(self): return self.allMatchingGroups.size()
[docs] def __iter__(self): return map(convert, self.allMatchingGroups)
[docs] def search(self, searcher, query, filter=None, count=None, start=0): "Run query and return `Groups`_." if count is None: count = sum(search.FieldCache.DEFAULT.getTermsIndex(reader, self.field).valueCount for reader in searcher.readers) topgroups =, searcher, filter, query, start, count - start) return Groups(searcher, topgroups.groups, topgroups.totalHitCount, topgroups.maxScore)