Combines search formatter and parser function

The parser was extracting the list of search results from the json
object returned by the search endpoint, and the formatter was converting
an individual json entry into a SearchResult object. This just merged
them into one function, because they are never used separately.
This commit is contained in:
Mouse Reeve 2022-05-30 12:52:31 -07:00
parent 525e2a591d
commit 87fe984462
4 changed files with 60 additions and 85 deletions

View file

@ -52,11 +52,9 @@ class AbstractMinimalConnector(ABC):
"""Format the search results based on the formt of the query"""
# TODO: inventaire min confidence
parser = self.parse_search_data
formatter = self.format_search_result
if maybe_isbn(query):
parser = self.parse_isbn_search_data
formatter = self.format_isbn_search_result
return [formatter(doc) for doc in parser(data)[:10]]
return list(parser(data))[:10]
@abstractmethod
def get_or_create_book(self, remote_id):
@ -66,18 +64,10 @@ class AbstractMinimalConnector(ABC):
def parse_search_data(self, data):
"""turn the result json from a search into a list"""
@abstractmethod
def format_search_result(self, search_result):
"""create a SearchResult obj from json"""
@abstractmethod
def parse_isbn_search_data(self, data):
"""turn the result json from a search into a list"""
@abstractmethod
def format_isbn_search_result(self, search_result):
"""create a SearchResult obj from json"""
class AbstractConnector(AbstractMinimalConnector):
"""generic book data connector"""

View file

@ -11,14 +11,10 @@ class Connector(AbstractMinimalConnector):
return activitypub.resolve_remote_id(remote_id, model=models.Edition)
def parse_search_data(self, data):
return data
def format_search_result(self, search_result):
search_result["connector"] = self
return SearchResult(**search_result)
for search_result in data:
search_result["connector"] = self
yield SearchResult(**search_result)
def parse_isbn_search_data(self, data):
return data
def format_isbn_search_result(self, search_result):
return self.format_search_result(search_result)
for search_result in data:
yield self.format_search_result(search_result)

View file

@ -78,44 +78,37 @@ class Connector(AbstractConnector):
}
def parse_search_data(self, data):
return data.get("results")
def format_search_result(self, search_result):
images = search_result.get("image")
cover = f"{self.covers_url}/img/entities/{images[0]}" if images else None
# a deeply messy translation of inventaire's scores
confidence = float(search_result.get("_score", 0.1))
confidence = 0.1 if confidence < 150 else 0.999
return SearchResult(
title=search_result.get("label"),
key=self.get_remote_id(search_result.get("uri")),
author=search_result.get("description"),
view_link=f"{self.base_url}/entity/{search_result.get('uri')}",
cover=cover,
confidence=confidence,
connector=self,
)
for search_result in data.get("results"):
images = search_result.get("image")
cover = f"{self.covers_url}/img/entities/{images[0]}" if images else None
# a deeply messy translation of inventaire's scores
confidence = float(search_result.get("_score", 0.1))
confidence = 0.1 if confidence < 150 else 0.999
yield SearchResult(
title=search_result.get("label"),
key=self.get_remote_id(search_result.get("uri")),
author=search_result.get("description"),
view_link=f"{self.base_url}/entity/{search_result.get('uri')}",
cover=cover,
confidence=confidence,
connector=self,
)
def parse_isbn_search_data(self, data):
"""got some daaaata"""
results = data.get("entities")
if not results:
return []
return list(results.values())
def format_isbn_search_result(self, search_result):
"""totally different format than a regular search result"""
title = search_result.get("claims", {}).get("wdt:P1476", [])
if not title:
return None
return SearchResult(
title=title[0],
key=self.get_remote_id(search_result.get("uri")),
author=search_result.get("description"),
view_link=f"{self.base_url}/entity/{search_result.get('uri')}",
cover=self.get_cover_url(search_result.get("image")),
connector=self,
)
results = data.get("entities", [])
for search_result in list(results.values()):
title = search_result.get("claims", {}).get("wdt:P1476", [])
if not title:
continue
yield SearchResult(
title=title[0],
key=self.get_remote_id(search_result.get("uri")),
author=search_result.get("description"),
view_link=f"{self.base_url}/entity/{search_result.get('uri')}",
cover=self.get_cover_url(search_result.get("image")),
connector=self,
)
def is_work_data(self, data):
return data.get("type") == "work"

View file

@ -153,38 +153,34 @@ class Connector(AbstractConnector):
return f"{self.covers_url}/b/id/{image_name}"
def parse_search_data(self, data):
return data.get("docs")
def format_search_result(self, search_result):
# build the remote id from the openlibrary key
key = self.books_url + search_result["key"]
author = search_result.get("author_name") or ["Unknown"]
cover_blob = search_result.get("cover_i")
cover = self.get_cover_url([cover_blob], size="M") if cover_blob else None
return SearchResult(
title=search_result.get("title"),
key=key,
author=", ".join(author),
connector=self,
year=search_result.get("first_publish_year"),
cover=cover,
)
for search_result in data.get("docs"):
# build the remote id from the openlibrary key
key = self.books_url + search_result["key"]
author = search_result.get("author_name") or ["Unknown"]
cover_blob = search_result.get("cover_i")
cover = self.get_cover_url([cover_blob], size="M") if cover_blob else None
yield SearchResult(
title=search_result.get("title"),
key=key,
author=", ".join(author),
connector=self,
year=search_result.get("first_publish_year"),
cover=cover,
)
def parse_isbn_search_data(self, data):
return list(data.values())
def format_isbn_search_result(self, search_result):
# build the remote id from the openlibrary key
key = self.books_url + search_result["key"]
authors = search_result.get("authors") or [{"name": "Unknown"}]
author_names = [author.get("name") for author in authors]
return SearchResult(
title=search_result.get("title"),
key=key,
author=", ".join(author_names),
connector=self,
year=search_result.get("publish_date"),
)
for search_result in list(data.values()):
# build the remote id from the openlibrary key
key = self.books_url + search_result["key"]
authors = search_result.get("authors") or [{"name": "Unknown"}]
author_names = [author.get("name") for author in authors]
yield SearchResult(
title=search_result.get("title"),
key=key,
author=", ".join(author_names),
connector=self,
year=search_result.get("publish_date"),
)
def load_edition_data(self, olkey):
"""query openlibrary for editions of a work"""