X-Git-Url: https://wannabe.guru.org/gitweb/?a=blobdiff_plain;f=logical_search.py;h=e710d0b65be57912bcd7781d6c9cc80eeffe2249;hb=53d8e09eb7aded181479fda535349b4583e49d3f;hp=b6d7479879010d6ea40ef813d03e84574ead7e55;hpb=532df2c5b57c7517dfb3dddd8c1358fbadf8baf3;p=python_utils.git diff --git a/logical_search.py b/logical_search.py index b6d7479..e710d0b 100644 --- a/logical_search.py +++ b/logical_search.py @@ -3,9 +3,7 @@ # © Copyright 2021-2022, Scott Gasch """This is a module concerned with the creation of and searching of a -corpus of documents. The corpus is held in memory for fast -searching. - +corpus of documents. The corpus and index are held in memory. """ from __future__ import annotations @@ -28,18 +26,27 @@ class ParseError(Exception): class Document: """A class representing a searchable document.""" - # A unique identifier for each document. docid: str = '' + """A unique identifier for each document -- must be provided + by the caller. See :meth:`python_modules.id_generator.get` or + :meth:`python_modules.string_utils.generate_uuid` for potential + sources.""" - # A set of tag strings for this document. May be empty. tags: Set[str] = field(default_factory=set) + """A set of tag strings for this document. May be empty. Tags + are simply text labels that are associated with a document and + may be used to search for it later. + """ - # A list of key->value strings for this document. May be empty. properties: List[Tuple[str, str]] = field(default_factory=list) + """A list of key->value strings for this document. May be empty. + Properties are more flexible tags that have both a label and a + value. e.g. "category:mystery" or "author:smith".""" - # An optional reference to something else; interpreted only by - # caller code, ignored here. reference: Optional[Any] = None + """An optional reference to something else for convenience; + interpreted only by caller code, ignored here. + """ class Operation(enum.Enum): @@ -69,7 +76,11 @@ class Operation(enum.Enum): class Corpus(object): - """A collection of searchable documents. + """A collection of searchable documents. The caller can + add documents to it (or edit existing docs) via :meth:`add_doc`, + retrieve a document given its docid via :meth:`get_doc`, and + perform various lookups of documents. The most interesting + lookup is implemented in :meth:`query`. >>> c = Corpus() >>> c.add_doc(Document( @@ -104,6 +115,12 @@ class Corpus(object): ... ) >>> c.query('author:Scott and important') {1} + >>> c.query('*') + {1, 2, 3} + >>> c.query('*:*') + {1, 2, 3} + >>> c.query('*:Scott') + {1, 3} """ def __init__(self) -> None: @@ -117,11 +134,14 @@ class Corpus(object): distinct docid that will serve as its primary identifier. If the same Document is added multiple times, only the most recent addition is indexed. If two distinct documents with - the same docid are added, the latter klobbers the former in the - indexes. + the same docid are added, the latter klobbers the former in + the indexes. See :meth:`python_modules.id_generator.get` or + :meth:`python_modules.string_utils.generate_uuid` for potential + sources of docids. Each Document may have an optional set of tags which can be - used later in expressions to the query method. + used later in expressions to the query method. These are simple + text labels. Each Document may have an optional list of key->value tuples which can be used later in expressions to the query method. @@ -130,6 +150,9 @@ class Corpus(object): never interpreted by this module. This is meant to allow easy mapping between Documents in this corpus and external objects they may represent. + + Args: + doc: the document to add or edit """ if doc.docid in self.documents_by_docid: @@ -155,13 +178,27 @@ class Corpus(object): self.docids_with_property[key].add(doc.docid) def get_docids_by_exact_tag(self, tag: str) -> Set[str]: - """Return the set of docids that have a particular tag.""" + """Return the set of docids that have a particular tag. + Args: + tag: the tag for which to search + + Returns: + A set containing docids with the provided tag which + may be empty.""" return self.docids_by_tag[tag] def get_docids_by_searching_tags(self, tag: str) -> Set[str]: - """Return the set of docids with a tag that contains a str""" + """Return the set of docids with a tag that contains a str. + + Args: + tag: the tag pattern for which to search + Returns: + A set containing docids with tags that match the pattern + provided. e.g., if the arg was "foo" tags "football", "foobar", + and "food" all match. + """ ret = set() for search_tag in self.docids_by_tag: if tag in search_tag: @@ -173,42 +210,65 @@ class Corpus(object): """Return the set of docids that have a particular property no matter what that property's value. + Args: + key: the key value to search for. + + Returns: + A set of docids that contain the key (no matter what value) + which may be empty. """ return self.docids_with_property[key] def get_docids_by_property(self, key: str, value: str) -> Set[str]: """Return the set of docids that have a particular property with a - particular value.. + particular value. + Args: + key: the key to search for + value: the value that key must have in order to match a doc. + + Returns: + A set of docids that contain key with value which may be empty. """ return self.docids_by_property[(key, value)] def invert_docid_set(self, original: Set[str]) -> Set[str]: """Invert a set of docids.""" - return {docid for docid in self.documents_by_docid if docid not in original} def get_doc(self, docid: str) -> Optional[Document]: - """Given a docid, retrieve the previously added Document.""" + """Given a docid, retrieve the previously added Document. + Args: + docid: the docid to retrieve + + Returns: + The Document with docid or None to indicate no match. + """ return self.documents_by_docid.get(docid, None) def query(self, query: str) -> Optional[Set[str]]: """Query the corpus for documents that match a logical expression. - Returns a (potentially empty) set of docids for the matching - (previously added) documents or None on error. - e.g. + Args: + query: the logical query expressed using a simple language + that understands conjunction (and operator), disjunction + (or operator) and inversion (not operator) as well as + parenthesis. Here are some legal sample queries:: - tag1 and tag2 and not tag3 + tag1 and tag2 and not tag3 - (tag1 or tag2) and (tag3 or tag4) + (tag1 or tag2) and (tag3 or tag4) - (tag1 and key2:value2) or (tag2 and key1:value1) + (tag1 and key2:value2) or (tag2 and key1:value1) - key:* + key:* - tag1 and key:* + tag1 and key:* + + Returns: + A (potentially empty) set of docids for the matching + (previously added) documents or None on error. """ try: @@ -357,12 +417,24 @@ class Node(object): key, value = tag.split(":") except ValueError as v: raise ParseError(f'Invalid key:value syntax at "{tag}"') from v - if value == "*": - r = self.corpus.get_docids_with_property(key) + + if key == '*': + r = set() + for kv, s in self.corpus.docids_by_property.items(): + if value in ('*', kv[1]): + r.update(s) else: - r = self.corpus.get_docids_by_property(key, value) + if value == '*': + r = self.corpus.get_docids_with_property(key) + else: + r = self.corpus.get_docids_by_property(key, value) else: - r = self.corpus.get_docids_by_exact_tag(tag) + if tag == '*': + r = set() + for s in self.corpus.docids_by_tag.values(): + r.update(s) + else: + r = self.corpus.get_docids_by_exact_tag(tag) retval.update(r) else: raise ParseError(f"Unexpected query {tag}")