From fa1ad78cd0e00f524221f972889ee32373d7b94e Mon Sep 17 00:00:00 2001 From: Bryan Newbold Date: Mon, 7 Feb 2022 19:51:56 -0800 Subject: container: scholars portal kbart link; unknown type display --- python/fatcat_web/templates/container_view.html | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'python/fatcat_web') diff --git a/python/fatcat_web/templates/container_view.html b/python/fatcat_web/templates/container_view.html index c5f68367..d88b44ab 100644 --- a/python/fatcat_web/templates/container_view.html +++ b/python/fatcat_web/templates/container_view.html @@ -76,7 +76,7 @@ {% if type_row == "_unknown" %} - Unknown + unknown-type {% else %} {{ type_row }} {% endif %} @@ -158,6 +158,8 @@ HathiTrust: {% elif k == "portico" and container.issnl %} Portico: + {% elif k == "scholarsportal" and container.issnl %} + Scholars Portal: {% else %} {{ k }}: {% endif %} -- cgit v1.2.3 From 6244c06abf8488fff87b30cb0a8433592f1f5d24 Mon Sep 17 00:00:00 2001 From: Bryan Newbold Date: Tue, 8 Feb 2022 14:35:39 -0800 Subject: container search: iterate on SERP page (including stats) --- python/fatcat_web/search.py | 16 +++++- python/fatcat_web/templates/container_search.html | 3 +- python/fatcat_web/templates/entity_macros.html | 63 ++++++++++++++++++++--- 3 files changed, 72 insertions(+), 10 deletions(-) (limited to 'python/fatcat_web') diff --git a/python/fatcat_web/search.py b/python/fatcat_web/search.py index 8cbe09f6..2a3515d4 100644 --- a/python/fatcat_web/search.py +++ b/python/fatcat_web/search.py @@ -178,7 +178,7 @@ def do_container_search(query: GenericQuery, deep_page_limit: int = 2000) -> Sea search = Search(using=app.es_client, index=app.config["ELASTICSEARCH_CONTAINER_INDEX"]) - search = search.query( + basic_query = Q( "query_string", query=query.q, default_operator="AND", @@ -188,6 +188,20 @@ def do_container_search(query: GenericQuery, deep_page_limit: int = 2000) -> Sea fields=["biblio"], ) + search = search.query( + "boosting", + positive=Q( + "bool", + must=basic_query, + should=[ + Q("range", releases_total={"gte": 500}), + Q("range", releases_total={"gte": 5000}), + ], + ), + negative=Q("term", releases_total=0), + negative_boost=0.5, + ) + # Sanity checks limit = min((int(query.limit or 25), 100)) offset = max((int(query.offset or 0), 0)) diff --git a/python/fatcat_web/templates/container_search.html b/python/fatcat_web/templates/container_search.html index 0eeed55e..f091fc3b 100644 --- a/python/fatcat_web/templates/container_search.html +++ b/python/fatcat_web/templates/container_search.html @@ -34,9 +34,10 @@ {% if found.results %} {{ search_macros.top_results(query, found) }} +
{% for entity in found.results %} - {{ entity_macros.container_search_result_row(entity) }} + {{ entity_macros.container_search_result_row(entity, show_stats=True) }} {% endfor %} {% if found.results|length > 8 %} diff --git a/python/fatcat_web/templates/entity_macros.html b/python/fatcat_web/templates/entity_macros.html index 9b419c41..b801f3c5 100644 --- a/python/fatcat_web/templates/entity_macros.html +++ b/python/fatcat_web/templates/entity_macros.html @@ -267,23 +267,70 @@ {% endmacro %} -{% macro container_search_result_row(entity) -%} +{% macro container_search_result_row(entity, show_stats=False) -%}
-

+ {% if show_stats %} +
+ {% if entity.releases_total %} + ~{{ "{:,}".format(entity.releases_total) }} releases + {% elif entity.releases_total == 0 %} + (not indexed) + {% endif %} + {% if entity.releases_total %} + {{ preservation_bar({'bright': entity.preservation_bright, 'dark': entity.preservation_dark, 'none': entity.preservation_none, 'total': entity.releases_total}) }} + {% endif %} +
+ {% endif %} +

{{ entity['name'] }} - {% if entity.is_oa %}{% endif %}

+ {% if entity.original_name %} + {{ entity.original_name }}
+ {% endif %} {% if entity.publisher %} -
{{ entity.publisher }}
+ {{ entity.publisher }} +
{% endif %} {% if entity.issnl %} - issn:{{ entity.issnl }} + issnl:{{ entity.issnl }}   + {% endif %} + {% if entity.wikidata_qid %} + wikidata:{{ entity.wikidata_qid }}   + {% endif %} + {% if entity.dblp_prefix %} + dblp:{{ entity.dblp_prefix }}   + {% endif %} + {% if entity.ia_sim_collection %} + [archive.org]   + {% elif entity.sim_pubid %} + [archive.org]   + {% endif %} + {# too much clutter + {% if entity.country_code %} + country:{{ entity.country_code }}   + {% endif %} + {% for lang in entity.languages %} + lang:{{ lang }}   + {% endfor %} + {% if entity.any_kbart %} + [KBART]   + {% endif %} +
+ #} + {% if entity.in_doaj %} + [DOAJ]   + {% endif %} + {% if entity.in_road %} + [ROAD]   + {% endif %} + {% if entity.is_oa and not (entity.in_doaj or entity.in_road) %} + [open-access]   {% endif %} - {% if entity.container_type %} -  {{ entity.container_type }} + {% if entity.container_type and entity.container_type != "journal" %} + [{{ entity.container_type }}]   {% endif %} {% if entity.publication_status and entity.publication_status != "active" %} -  {{ entity.publication_status }} + {{ entity.publication_status }}   {% endif %}
{% endmacro %} -- cgit v1.2.3 From 6976b6868cdf0628aa79d47aab4e889a9ccfc0dc Mon Sep 17 00:00:00 2001 From: Bryan Newbold Date: Tue, 8 Feb 2022 19:02:34 -0800 Subject: container SERP: fix layout on mobile --- python/fatcat_web/templates/container_search.html | 7 +- python/fatcat_web/templates/entity_macros.html | 109 +++++++++++----------- 2 files changed, 62 insertions(+), 54 deletions(-) (limited to 'python/fatcat_web') diff --git a/python/fatcat_web/templates/container_search.html b/python/fatcat_web/templates/container_search.html index f091fc3b..0031a0a8 100644 --- a/python/fatcat_web/templates/container_search.html +++ b/python/fatcat_web/templates/container_search.html @@ -33,7 +33,12 @@ {% if found %} {% if found.results %} - {{ search_macros.top_results(query, found) }} +
+
+ Content Status +
+ {{ search_macros.top_results(query, found) }} +

{% for entity in found.results %} diff --git a/python/fatcat_web/templates/entity_macros.html b/python/fatcat_web/templates/entity_macros.html index b801f3c5..6ad63a14 100644 --- a/python/fatcat_web/templates/entity_macros.html +++ b/python/fatcat_web/templates/entity_macros.html @@ -268,9 +268,63 @@ {% macro container_search_result_row(entity, show_stats=False) -%} -
+
+
+

+ {{ entity['name'] }} +

+ {% if entity.original_name %} + {{ entity.original_name }}
+ {% endif %} + {% if entity.publisher %} + {{ entity.publisher }} +
+ {% endif %} + {% if entity.issnl %} + issnl:{{ entity.issnl }}   + {% endif %} + {% if entity.wikidata_qid %} + wikidata:{{ entity.wikidata_qid }}   + {% endif %} + {% if entity.dblp_prefix %} + dblp:{{ entity.dblp_prefix }}   + {% endif %} + {% if entity.ia_sim_collection %} + [archive.org]   + {% elif entity.sim_pubid %} + [archive.org]   + {% endif %} + {# too much clutter + {% if entity.country_code %} + country:{{ entity.country_code }}   + {% endif %} + {% for lang in entity.languages %} + lang:{{ lang }}   + {% endfor %} + {% if entity.any_kbart %} + [KBART]   + {% endif %} +
+ #} + {% if entity.in_doaj %} + [DOAJ]   + {% endif %} + {% if entity.in_road %} + [ROAD]   + {% endif %} + {% if entity.is_oa and not (entity.in_doaj or entity.in_road) %} + [open-access]   + {% endif %} + {% if entity.container_type and entity.container_type != "journal" %} + [{{ entity.container_type }}]   + {% endif %} + {% if entity.publication_status and entity.publication_status != "active" %} + {{ entity.publication_status }}   + {% endif %} +
+ {% if show_stats %} -
+
{% if entity.releases_total %} ~{{ "{:,}".format(entity.releases_total) }} releases {% elif entity.releases_total == 0 %} @@ -281,57 +335,6 @@ {% endif %}
{% endif %} -

- {{ entity['name'] }} -

- {% if entity.original_name %} - {{ entity.original_name }}
- {% endif %} - {% if entity.publisher %} - {{ entity.publisher }} -
- {% endif %} - {% if entity.issnl %} - issnl:{{ entity.issnl }}   - {% endif %} - {% if entity.wikidata_qid %} - wikidata:{{ entity.wikidata_qid }}   - {% endif %} - {% if entity.dblp_prefix %} - dblp:{{ entity.dblp_prefix }}   - {% endif %} - {% if entity.ia_sim_collection %} - [archive.org]   - {% elif entity.sim_pubid %} - [archive.org]   - {% endif %} - {# too much clutter - {% if entity.country_code %} - country:{{ entity.country_code }}   - {% endif %} - {% for lang in entity.languages %} - lang:{{ lang }}   - {% endfor %} - {% if entity.any_kbart %} - [KBART]   - {% endif %} -
- #} - {% if entity.in_doaj %} - [DOAJ]   - {% endif %} - {% if entity.in_road %} - [ROAD]   - {% endif %} - {% if entity.is_oa and not (entity.in_doaj or entity.in_road) %} - [open-access]   - {% endif %} - {% if entity.container_type and entity.container_type != "journal" %} - [{{ entity.container_type }}]   - {% endif %} - {% if entity.publication_status and entity.publication_status != "active" %} - {{ entity.publication_status }}   - {% endif %}
{% endmacro %} -- cgit v1.2.3 From 9ab3cd54aa039393d294cdc85871353651c35576 Mon Sep 17 00:00:00 2001 From: Bryan Newbold Date: Wed, 9 Feb 2022 00:23:04 -0800 Subject: search: improve container_id handling --- python/fatcat_web/search.py | 33 ++++++--------------------------- 1 file changed, 6 insertions(+), 27 deletions(-) (limited to 'python/fatcat_web') diff --git a/python/fatcat_web/search.py b/python/fatcat_web/search.py index 2a3515d4..9e9376cc 100644 --- a/python/fatcat_web/search.py +++ b/python/fatcat_web/search.py @@ -44,16 +44,6 @@ class ReleaseQuery: query_str = args.get("q") or "*" - container_id = args.get("container_id") - # TODO: as filter, not in query string - if container_id: - query_str += ' container_id:"{}"'.format(container_id) - - # TODO: where are container_issnl queries actually used? - issnl = args.get("container_issnl") - if issnl and query_str: - query_str += ' container_issnl:"{}"'.format(issnl) - offset = args.get("offset", "0") offset = max(0, int(offset)) if offset.isnumeric() else 0 @@ -61,7 +51,7 @@ class ReleaseQuery: q=query_str, offset=offset, fulltext_only=bool(args.get("fulltext_only")), - container_id=container_id, + container_id=args.get("container_id"), recent=bool(args.get("recent")), exclude_stubs=bool(args.get("exclude_stubs")), ) @@ -263,6 +253,9 @@ def do_release_search(query: ReleaseQuery, deep_page_limit: int = 2000) -> Searc ], ) + if query.container_id: + search = search.filter("term", container_id=query.container_id) + search = search.query( "boosting", positive=Q( @@ -657,11 +650,7 @@ def get_elastic_preservation_by_year(query: ReleaseQuery) -> List[Dict[str, Any] "biblio", ], ) - if query.container_id: - search = search.filter( - "term", - container_id=query.container_id, - ) + search = search.filter("term", container_id=query.container_id) if query.exclude_stubs: search = search.query( "bool", @@ -923,17 +912,7 @@ def get_elastic_preservation_by_type(query: ReleaseQuery) -> List[dict]: ], ) if query.container_id: - search = search.query( - "bool", - filter=[ - Q( - "bool", - must=[ - Q("match", container_id=query.container_id), - ], - ), - ], - ) + search = search.filter("term", container_id=query.container_id) if query.recent: date_today = datetime.date.today() start_date = str(date_today - datetime.timedelta(days=60)) -- cgit v1.2.3 From 5bc77c47eed20676cd3db162c9675311f77c6cf9 Mon Sep 17 00:00:00 2001 From: Bryan Newbold Date: Wed, 9 Feb 2022 00:24:46 -0800 Subject: web: move search-in-container to dedicated tab --- python/fatcat_web/routes.py | 46 ++++++++++++++ python/fatcat_web/templates/container_view.html | 5 +- .../templates/container_view_search.html | 70 ++++++++++++++++++++++ python/fatcat_web/templates/entity_base.html | 1 + 4 files changed, 119 insertions(+), 3 deletions(-) create mode 100644 python/fatcat_web/templates/container_view_search.html (limited to 'python/fatcat_web') diff --git a/python/fatcat_web/routes.py b/python/fatcat_web/routes.py index 186166bd..b25cd37c 100644 --- a/python/fatcat_web/routes.py +++ b/python/fatcat_web/routes.py @@ -1079,6 +1079,52 @@ def coverage_search() -> AnyResponse: ) +@app.route("/container//search", methods=["GET", "POST"]) +def container_view_search(ident: str) -> AnyResponse: + entity = generic_get_entity("container", ident) + + if entity.state == "redirect": + return redirect(f"/container/{entity.redirect}") + elif entity.state == "deleted": + return render_template("deleted_entity.html", entity_type="container", entity=entity) + + if "q" not in request.args.keys(): + return render_template( + "container_view_search.html", + query=ReleaseQuery(), + found=None, + entity_type="container", + entity=entity, + editgroup_id=None, + ) + + query = ReleaseQuery.from_args(request.args) + query.container_id = ident + try: + found = do_release_search(query) + except FatcatSearchError as fse: + return ( + render_template( + "container_view_search.html", + query=query, + es_error=fse, + entity_type="container", + entity=entity, + editgroup_id=None, + ), + fse.status_code, + ) + + return render_template( + "container_view_search.html", + query=query, + found=found, + entity_type="container", + entity=entity, + editgroup_id=None, + ) + + def get_changelog_stats() -> Dict[str, Any]: stats = {} latest_changelog = api.get_changelog(limit=1)[0] diff --git a/python/fatcat_web/templates/container_view.html b/python/fatcat_web/templates/container_view.html index d88b44ab..abb31e06 100644 --- a/python/fatcat_web/templates/container_view.html +++ b/python/fatcat_web/templates/container_view.html @@ -24,12 +24,11 @@ {% endif %} {% if container.state == "active" %} -

Search Releases from this Container

-
+

Search Content

+
-
diff --git a/python/fatcat_web/templates/container_view_search.html b/python/fatcat_web/templates/container_view_search.html new file mode 100644 index 00000000..289c8dad --- /dev/null +++ b/python/fatcat_web/templates/container_view_search.html @@ -0,0 +1,70 @@ +{% set container = entity %} +{% set entity_view = "search" %} +{% set entity_type = "container" %} +{% import "entity_macros.html" as entity_macros %} +{% import "search_macros.html" as search_macros %} +{% extends "entity_base.html" %} + +{% block entity_main %} +
+ +
+

Search inside Container

+ +
+
+ + +
+
Can also search all releases. +
+ +
+ +
+{% if found %} + {% if found.results %} + + {{ search_macros.top_results(query, found) }} + + {% for paper in found.results %} + {{ entity_macros.release_search_result_row(paper) }} + {% endfor %} + + {% if found.results|length > 8 %} +
+
+ {{ search_macros.bottom_results(query, found, endpoint='release_search') }} +
+ {% endif %} + + {% else %} + + Raw query was: {{ query.q }} + +
+
+
+ confused paper man +
+
+

No results found!

+

You could try elsewhere:

+ +
+
+
+ + {% endif %} + +{% elif es_error %} + {{ search_macros.es_error_msg(es_error) }} +{% endif %} + +
+{% endblock %} + diff --git a/python/fatcat_web/templates/entity_base.html b/python/fatcat_web/templates/entity_base.html index c3d6096b..2782edd5 100644 --- a/python/fatcat_web/templates/entity_base.html +++ b/python/fatcat_web/templates/entity_base.html @@ -83,6 +83,7 @@ {{ entity_tab("overview", "Overview", "") }} {% if entity_type == "container" and entity.state == 'active' and not editgroup %} {{ entity_tab("coverage", "Preservation", "/coverage") }} + {{ entity_tab("search", "Search", "/search") }} {% elif entity_type == "release" and entity.state != 'deleted' %} {{ entity_tab("contribs", "Authors", "/contribs", entity._authors|count ) }} {% if entity.state == 'active' %} -- cgit v1.2.3 From d73ab1f7cc45c122f321f0e717de2067554baabb Mon Sep 17 00:00:00 2001 From: Bryan Newbold Date: Wed, 9 Feb 2022 17:16:11 -0800 Subject: containers: initial work on 'browse' feature --- python/fatcat_web/routes.py | 11 +++- python/fatcat_web/search.py | 64 ++++++++++++++++++++++ .../templates/container_view_browse.html | 32 +++++++++++ python/fatcat_web/templates/entity_base.html | 1 + 4 files changed, 107 insertions(+), 1 deletion(-) create mode 100644 python/fatcat_web/templates/container_view_browse.html (limited to 'python/fatcat_web') diff --git a/python/fatcat_web/routes.py b/python/fatcat_web/routes.py index b25cd37c..9f46c674 100644 --- a/python/fatcat_web/routes.py +++ b/python/fatcat_web/routes.py @@ -60,6 +60,7 @@ from fatcat_web.search import ( ReleaseQuery, do_container_search, do_release_search, + get_elastic_container_browse_year_volume, get_elastic_container_histogram_legacy, get_elastic_container_preservation_by_volume, get_elastic_container_random_releases, @@ -286,6 +287,8 @@ def generic_entity_view(entity_type: str, ident: str, view_template: str) -> Any entity._type_preservation = get_elastic_preservation_by_type( ReleaseQuery(container_id=ident), ) + if view_template == "container_view_browse.html": + entity._browse_volume_year = get_elastic_container_browse_year_volume(entity.ident) return render_template( view_template, entity_type=entity_type, entity=entity, editgroup_id=None @@ -346,6 +349,12 @@ def container_view_coverage(ident: str) -> AnyResponse: return generic_entity_view("container", ident, "container_view_coverage.html") +@app.route("/container//browse", methods=["GET"]) +def container_view_browser(ident: str) -> AnyResponse: + # note: there is a special hack to add entity._type_preservation for this endpoint + return generic_entity_view("container", ident, "container_view_browse.html") + + @app.route("/container//metadata", methods=["GET"]) def container_view_metadata(ident: str) -> AnyResponse: return generic_entity_view("container", ident, "entity_view_metadata.html") @@ -1079,7 +1088,7 @@ def coverage_search() -> AnyResponse: ) -@app.route("/container//search", methods=["GET", "POST"]) +@app.route("/container//search", methods=["GET", "POST"]) def container_view_search(ident: str) -> AnyResponse: entity = generic_get_entity("container", ident) diff --git a/python/fatcat_web/search.py b/python/fatcat_web/search.py index 9e9376cc..ac4dc34e 100644 --- a/python/fatcat_web/search.py +++ b/python/fatcat_web/search.py @@ -327,6 +327,70 @@ def get_elastic_container_random_releases(ident: str, limit: int = 5) -> List[Di return results +def get_elastic_container_browse_year_volume(ident: str) -> List[Dict[int, Any]]: + """ + Returns a set of histogram buckets: + + container_ident: str + years{} + volumes{} + """ + + search = Search(using=app.es_client, index=app.config["ELASTICSEARCH_RELEASE_INDEX"]) + search = search.query( + "bool", + filter=[Q("bool", must_not=[Q("match", release_type="stub")])], + ) + search = search.filter("term", container_id=ident) + search.aggs.bucket( + "year_volume", + "composite", + size=1500, + sources=[ + { + "year": { + "histogram": { + "field": "release_year", + "interval": 1, + "missing_bucket": True, + # TODO: es-public-proxy support? + # "order": "asc", + # "missing_order": "last", + }, + } + }, + { + "volume": { + "terms": { + "field": "volume", + "missing_bucket": True, + # TODO: es-public-proxy support? + # "order": "asc", + # "missing_order": "last", + }, + } + }, + ], + ) + search = search[:0] + search = search.params(request_cache=True) + resp = wrap_es_execution(search) + buckets = resp.aggregations.year_volume.buckets + # print(buckets) + buckets = [h for h in buckets if h["key"]["year"]] + year_nums = set([int(h["key"]["year"]) for h in buckets]) + year_dicts: Dict[int, Dict[str, Any]] = dict() + if year_nums: + for year in year_nums: + year_dicts[year] = {} + for row in buckets: + year_dicts[int(row["key"]["year"])][row["key"]["volume"] or "_unknown"] = int( + row["doc_count"] + ) + # return sorted(year_dicts.values(), key=lambda x: x["year"]) + return year_dicts + + def get_elastic_entity_stats() -> dict: """ TODO: files, filesets, webcaptures (no schema yet) diff --git a/python/fatcat_web/templates/container_view_browse.html b/python/fatcat_web/templates/container_view_browse.html new file mode 100644 index 00000000..b5691899 --- /dev/null +++ b/python/fatcat_web/templates/container_view_browse.html @@ -0,0 +1,32 @@ +{% set container = entity %} +{% set entity_view = "browse" %} +{% set entity_type = "container" %} +{% import "entity_macros.html" as entity_macros %} +{% extends "entity_base.html" %} + +{% block entity_main %} + +{% if entity._browse_volume_year %} +

Browse by Year and Volume

+
    +{% for year in entity._browse_volume_year.keys()|sort|reverse %} + {% for volume in entity._browse_volume_year[year].keys()|sort|reverse %} + {% if volume == '_unknown' %} +
  • {{ year }} ({{ entity._browse_volume_year[year][volume] }} releases) + {% else %} +
  • {{ year }} | Vol. {{ volume }} ({{ entity._browse_volume_year[year][volume] }} releases) + {% endif %} + {% endfor %} +{% endfor %} +
+{% elif entity._browse_issues %} +{% for issue in entity._browse_issues.keys()|sort|reverse %} +

{{ issue }}

+ {% for paper in entity._browse_issues[issue] %} + {{ paper.title }}
+ {% endfor %} +{% endfor %} +{% endif %} + +{% endblock %} + diff --git a/python/fatcat_web/templates/entity_base.html b/python/fatcat_web/templates/entity_base.html index 2782edd5..633f3aee 100644 --- a/python/fatcat_web/templates/entity_base.html +++ b/python/fatcat_web/templates/entity_base.html @@ -82,6 +82,7 @@
{{ entity_tab("overview", "Overview", "") }} {% if entity_type == "container" and entity.state == 'active' and not editgroup %} + {{ entity_tab("browse", "Browse", "/browse") }} {{ entity_tab("coverage", "Preservation", "/coverage") }} {{ entity_tab("search", "Search", "/search") }} {% elif entity_type == "release" and entity.state != 'deleted' %} -- cgit v1.2.3 From ef7d24cd08b98d58779335acd1d655bd342cd9ec Mon Sep 17 00:00:00 2001 From: Bryan Newbold Date: Wed, 9 Feb 2022 18:03:23 -0800 Subject: move container_status ES query code from fatcat_web to fatcat_tools The main motivation is to never have fatcat_tools import from fatcat_web, only vica-versa. Some code in fatcat_tools needs container stats, so starting with that code path (plus some generic helpers). --- python/fatcat_tools/search/common.py | 96 +++++++++++++++ python/fatcat_tools/search/stats.py | 87 ++++++++++++++ python/fatcat_tools/workers/elasticsearch.py | 4 +- python/fatcat_transform.py | 4 +- python/fatcat_web/routes.py | 2 +- python/fatcat_web/search.py | 169 +++------------------------ 6 files changed, 205 insertions(+), 157 deletions(-) create mode 100644 python/fatcat_tools/search/common.py create mode 100644 python/fatcat_tools/search/stats.py (limited to 'python/fatcat_web') diff --git a/python/fatcat_tools/search/common.py b/python/fatcat_tools/search/common.py new file mode 100644 index 00000000..584757fd --- /dev/null +++ b/python/fatcat_tools/search/common.py @@ -0,0 +1,96 @@ +import sys +from typing import Any, Dict, List, Union + +import elasticsearch +import elasticsearch_dsl.response +from elasticsearch_dsl import Search + + +class FatcatSearchError(Exception): + def __init__(self, status_code: Union[int, str], name: str, description: str = None): + if status_code == "TIMEOUT": + status_code = 504 + elif isinstance(status_code, str): + try: + status_code = int(status_code) + except ValueError: + status_code = 503 + self.status_code = status_code + self.name = name + self.description = description + + +def _hits_total_int(val: Any) -> int: + """ + Compatibility hack between ES 6.x and 7.x. In ES 6x, total is returned as + an int in many places, in ES 7 as a dict (JSON object) with 'value' key + """ + if isinstance(val, int): + return val + else: + return int(val["value"]) + + +def results_to_dict(response: elasticsearch_dsl.response.Response) -> List[dict]: + """ + Takes a response returns all the hits as JSON objects. + + Also handles surrogate strings that elasticsearch returns sometimes, + probably due to mangled data processing in some pipeline. "Crimes against + Unicode"; production workaround + """ + + results = [] + for h in response: + r = h._d_ + # print(h.meta._d_) + results.append(r) + + for h in results: + for key in h: + if type(h[key]) is str: + h[key] = h[key].encode("utf8", "ignore").decode("utf8") + return results + + +def wrap_es_execution(search: Search) -> Any: + """ + Executes a Search object, and converts various ES error types into + something we can pretty print to the user. + """ + try: + resp = search.execute() + except elasticsearch.exceptions.RequestError as e: + # this is a "user" error + print("elasticsearch 400: " + str(e.info), file=sys.stderr) + description = None + assert isinstance(e.info, dict) + if e.info.get("error", {}).get("root_cause", {}): + description = str(e.info["error"]["root_cause"][0].get("reason")) + raise FatcatSearchError(e.status_code, str(e.error), description) + except elasticsearch.exceptions.ConnectionError as e: + raise FatcatSearchError(e.status_code, "ConnectionError: search engine not available") + except elasticsearch.exceptions.TransportError as e: + # all other errors + print("elasticsearch non-200 status code: {}".format(e.info), file=sys.stderr) + description = None + assert isinstance(e.info, dict) + if e.info and e.info.get("error", {}).get("root_cause", {}): + description = str(e.info["error"]["root_cause"][0].get("reason")) + raise FatcatSearchError(e.status_code, str(e.error), description) + return resp + + +def agg_to_dict(agg: Any) -> Dict[str, Any]: + """ + Takes a simple term aggregation result (with buckets) and returns a simple + dict with keys as terms and counts as values. Includes an extra value + '_other', and by convention aggregations should be written to have "missing" + values as '_unknown'. + """ + result = dict() + for bucket in agg.buckets: + result[bucket.key] = bucket.doc_count + if agg.sum_other_doc_count: + result["_other"] = agg.sum_other_doc_count + return result diff --git a/python/fatcat_tools/search/stats.py b/python/fatcat_tools/search/stats.py new file mode 100644 index 00000000..5496b94a --- /dev/null +++ b/python/fatcat_tools/search/stats.py @@ -0,0 +1,87 @@ +from typing import Any, Dict + +import elasticsearch +from elasticsearch_dsl import Search + +from fatcat_tools.search.common import _hits_total_int, agg_to_dict, wrap_es_execution + + +def query_es_container_stats( + ident: str, + es_client: elasticsearch.Elasticsearch, + es_index: str = "fatcat_release", + merge_shadows: bool = False, +) -> Dict[str, Any]: + """ + Returns dict: + ident + total: count + in_web: count + in_kbart: count + is_preserved: count + preservation{} + "histogram" by preservation status + release_type{} + "histogram" by release type + """ + + search = Search(using=es_client, index=es_index) + search = search.query( + "term", + container_id=ident, + ) + search.aggs.bucket( + "container_stats", + "filters", + filters={ + "in_web": { + "term": {"in_web": True}, + }, + "in_kbart": { + "term": {"in_kbart": True}, + }, + "is_preserved": { + "term": {"is_preserved": True}, + }, + }, + ) + search.aggs.bucket( + "preservation", + "terms", + field="preservation", + missing="_unknown", + ) + search.aggs.bucket( + "release_type", + "terms", + field="release_type", + missing="_unknown", + ) + + search = search[:0] + + search = search.params(request_cache=True) + search = search.params(track_total_hits=True) + resp = wrap_es_execution(search) + + container_stats = resp.aggregations.container_stats.buckets + preservation_bucket = agg_to_dict(resp.aggregations.preservation) + preservation_bucket["total"] = _hits_total_int(resp.hits.total) + for k in ("bright", "dark", "shadows_only", "none"): + if k not in preservation_bucket: + preservation_bucket[k] = 0 + if merge_shadows: + preservation_bucket["none"] += preservation_bucket["shadows_only"] + preservation_bucket["shadows_only"] = 0 + release_type_bucket = agg_to_dict(resp.aggregations.release_type) + stats = { + "ident": ident, + "total": _hits_total_int(resp.hits.total), + "in_web": container_stats["in_web"]["doc_count"], + "in_kbart": container_stats["in_kbart"]["doc_count"], + "is_preserved": container_stats["is_preserved"]["doc_count"], + "preservation": preservation_bucket, + "release_type": release_type_bucket, + } + + return stats diff --git a/python/fatcat_tools/workers/elasticsearch.py b/python/fatcat_tools/workers/elasticsearch.py index bfadea64..79071810 100644 --- a/python/fatcat_tools/workers/elasticsearch.py +++ b/python/fatcat_tools/workers/elasticsearch.py @@ -14,13 +14,13 @@ from fatcat_openapi_client import ( ) from fatcat_tools import entity_from_json, public_api +from fatcat_tools.search.stats import query_es_container_stats from fatcat_tools.transforms import ( changelog_to_elasticsearch, container_to_elasticsearch, file_to_elasticsearch, release_to_elasticsearch, ) -from fatcat_web.search import get_elastic_container_stats from .worker_common import FatcatWorker @@ -156,7 +156,7 @@ class ElasticsearchReleaseWorker(FatcatWorker): continue if self.entity_type == ContainerEntity and self.query_stats: - stats = get_elastic_container_stats( + stats = query_es_container_stats( entity.ident, es_client=es_client, es_index=self.elasticsearch_release_index, diff --git a/python/fatcat_transform.py b/python/fatcat_transform.py index 67bf56c5..4f28951c 100755 --- a/python/fatcat_transform.py +++ b/python/fatcat_transform.py @@ -13,6 +13,7 @@ import elasticsearch from fatcat_openapi_client import ChangelogEntry, ContainerEntity, FileEntity, ReleaseEntity from fatcat_tools import public_api +from fatcat_tools.search.stats import query_es_container_stats from fatcat_tools.transforms import ( changelog_to_elasticsearch, citeproc_csl, @@ -22,7 +23,6 @@ from fatcat_tools.transforms import ( release_to_csl, release_to_elasticsearch, ) -from fatcat_web.search import get_elastic_container_stats def run_elasticsearch_releases(args: argparse.Namespace) -> None: @@ -50,7 +50,7 @@ def run_elasticsearch_containers(args: argparse.Namespace) -> None: if args.query_stats: es_doc = container_to_elasticsearch( entity, - stats=get_elastic_container_stats( + stats=query_es_container_stats( entity.ident, es_client=es_client, es_index=es_release_index, diff --git a/python/fatcat_web/routes.py b/python/fatcat_web/routes.py index 9f46c674..0afc189f 100644 --- a/python/fatcat_web/routes.py +++ b/python/fatcat_web/routes.py @@ -30,6 +30,7 @@ from fatcat_tools.normal import ( clean_sha1, clean_sha256, ) +from fatcat_tools.search.common import FatcatSearchError from fatcat_tools.transforms import citeproc_csl, release_to_csl from fatcat_web import AnyResponse, Config, api, app, auth_api, mwoauth, priv_api from fatcat_web.auth import ( @@ -55,7 +56,6 @@ from fatcat_web.graphics import ( ) from fatcat_web.kafka import kafka_pixy_produce from fatcat_web.search import ( - FatcatSearchError, GenericQuery, ReleaseQuery, do_container_search, diff --git a/python/fatcat_web/search.py b/python/fatcat_web/search.py index ac4dc34e..99c8ee77 100644 --- a/python/fatcat_web/search.py +++ b/python/fatcat_web/search.py @@ -4,31 +4,22 @@ the formal API) """ import datetime -import sys from dataclasses import dataclass -from typing import Any, Dict, List, Optional, Tuple, Union +from typing import Any, Dict, List, Optional, Tuple import elasticsearch -import elasticsearch_dsl.response from elasticsearch_dsl import Q, Search +from fatcat_tools.search.common import ( + _hits_total_int, + agg_to_dict, + results_to_dict, + wrap_es_execution, +) +from fatcat_tools.search.stats import query_es_container_stats from fatcat_web import app -class FatcatSearchError(Exception): - def __init__(self, status_code: Union[int, str], name: str, description: str = None): - if status_code == "TIMEOUT": - status_code = 504 - elif isinstance(status_code, str): - try: - status_code = int(status_code) - except ValueError: - status_code = 503 - self.status_code = status_code - self.name = name - self.description = description - - @dataclass class ReleaseQuery: q: Optional[str] = None @@ -88,82 +79,6 @@ class SearchHits: results: List[Any] -def _hits_total_int(val: Any) -> int: - """ - Compatibility hack between ES 6.x and 7.x. In ES 6x, total is returned as - an int in many places, in ES 7 as a dict (JSON object) with 'value' key - """ - if isinstance(val, int): - return val - else: - return int(val["value"]) - - -def results_to_dict(response: elasticsearch_dsl.response.Response) -> List[dict]: - """ - Takes a response returns all the hits as JSON objects. - - Also handles surrogate strings that elasticsearch returns sometimes, - probably due to mangled data processing in some pipeline. "Crimes against - Unicode"; production workaround - """ - - results = [] - for h in response: - r = h._d_ - # print(h.meta._d_) - results.append(r) - - for h in results: - for key in h: - if type(h[key]) is str: - h[key] = h[key].encode("utf8", "ignore").decode("utf8") - return results - - -def wrap_es_execution(search: Search) -> Any: - """ - Executes a Search object, and converts various ES error types into - something we can pretty print to the user. - """ - try: - resp = search.execute() - except elasticsearch.exceptions.RequestError as e: - # this is a "user" error - print("elasticsearch 400: " + str(e.info), file=sys.stderr) - description = None - assert isinstance(e.info, dict) - if e.info.get("error", {}).get("root_cause", {}): - description = str(e.info["error"]["root_cause"][0].get("reason")) - raise FatcatSearchError(e.status_code, str(e.error), description) - except elasticsearch.exceptions.ConnectionError as e: - raise FatcatSearchError(e.status_code, "ConnectionError: search engine not available") - except elasticsearch.exceptions.TransportError as e: - # all other errors - print("elasticsearch non-200 status code: {}".format(e.info), file=sys.stderr) - description = None - assert isinstance(e.info, dict) - if e.info and e.info.get("error", {}).get("root_cause", {}): - description = str(e.info["error"]["root_cause"][0].get("reason")) - raise FatcatSearchError(e.status_code, str(e.error), description) - return resp - - -def agg_to_dict(agg: Any) -> Dict[str, Any]: - """ - Takes a simple term aggregation result (with buckets) and returns a simple - dict with keys as terms and counts as values. Includes an extra value - '_other', and by convention aggregations should be written to have "missing" - values as '_unknown'. - """ - result = dict() - for bucket in agg.buckets: - result[bucket.key] = bucket.doc_count - if agg.sum_other_doc_count: - result["_other"] = agg.sum_other_doc_count - return result - - def do_container_search(query: GenericQuery, deep_page_limit: int = 2000) -> SearchHits: search = Search(using=app.es_client, index=app.config["ELASTICSEARCH_CONTAINER_INDEX"]) @@ -536,6 +451,9 @@ def get_elastic_container_stats( merge_shadows: Optional[bool] = None, ) -> Dict[str, Any]: """ + This is a DEPRECATED backwards-compatability wrapper around the new + query_es_container_stats() method from fatcat_tools. + Returns dict: ident issnl (optional) @@ -556,66 +474,13 @@ def get_elastic_container_stats( if merge_shadows is None: merge_shadows = app.config["FATCAT_MERGE_SHADOW_PRESERVATION"] - search = Search(using=es_client, index=es_index) - search = search.query( - "term", - container_id=ident, - ) - search.aggs.bucket( - "container_stats", - "filters", - filters={ - "in_web": { - "term": {"in_web": True}, - }, - "in_kbart": { - "term": {"in_kbart": True}, - }, - "is_preserved": { - "term": {"is_preserved": True}, - }, - }, - ) - search.aggs.bucket( - "preservation", - "terms", - field="preservation", - missing="_unknown", + stats = query_es_container_stats( + ident=ident, + es_client=es_client, + es_index=es_index, + merge_shadows=merge_shadows, ) - search.aggs.bucket( - "release_type", - "terms", - field="release_type", - missing="_unknown", - ) - - search = search[:0] - - search = search.params(request_cache=True) - search = search.params(track_total_hits=True) - resp = wrap_es_execution(search) - - container_stats = resp.aggregations.container_stats.buckets - preservation_bucket = agg_to_dict(resp.aggregations.preservation) - preservation_bucket["total"] = _hits_total_int(resp.hits.total) - for k in ("bright", "dark", "shadows_only", "none"): - if k not in preservation_bucket: - preservation_bucket[k] = 0 - if merge_shadows: - preservation_bucket["none"] += preservation_bucket["shadows_only"] - preservation_bucket["shadows_only"] = 0 - release_type_bucket = agg_to_dict(resp.aggregations.release_type) - stats = { - "ident": ident, - "issnl": issnl, - "total": _hits_total_int(resp.hits.total), - "in_web": container_stats["in_web"]["doc_count"], - "in_kbart": container_stats["in_kbart"]["doc_count"], - "is_preserved": container_stats["is_preserved"]["doc_count"], - "preservation": preservation_bucket, - "release_type": release_type_bucket, - } - + stats["issnl"] = issnl return stats -- cgit v1.2.3 From ebd467fba4773a6dd027ca118124e4c6f80a7e07 Mon Sep 17 00:00:00 2001 From: Bryan Newbold Date: Wed, 9 Feb 2022 18:08:29 -0800 Subject: container browse: quick fix type annotation --- python/fatcat_web/search.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'python/fatcat_web') diff --git a/python/fatcat_web/search.py b/python/fatcat_web/search.py index 99c8ee77..b19b57e6 100644 --- a/python/fatcat_web/search.py +++ b/python/fatcat_web/search.py @@ -242,7 +242,7 @@ def get_elastic_container_random_releases(ident: str, limit: int = 5) -> List[Di return results -def get_elastic_container_browse_year_volume(ident: str) -> List[Dict[int, Any]]: +def get_elastic_container_browse_year_volume(ident: str) -> Dict[int, Dict[str, Any]]: """ Returns a set of histogram buckets: -- cgit v1.2.3 From 9b9ede257753b1e3fd19bbf90a5895de5db18504 Mon Sep 17 00:00:00 2001 From: Bryan Newbold Date: Fri, 11 Feb 2022 11:14:07 -0800 Subject: containers: progress on browse interface --- python/fatcat_web/routes.py | 92 ++++++++++++++++++++-- python/fatcat_web/search.py | 7 +- .../templates/container_view_browse.html | 71 ++++++++++++----- python/fatcat_web/templates/entity_macros.html | 4 +- 4 files changed, 146 insertions(+), 28 deletions(-) (limited to 'python/fatcat_web') diff --git a/python/fatcat_web/routes.py b/python/fatcat_web/routes.py index 0afc189f..6e3e9b57 100644 --- a/python/fatcat_web/routes.py +++ b/python/fatcat_web/routes.py @@ -287,8 +287,6 @@ def generic_entity_view(entity_type: str, ident: str, view_template: str) -> Any entity._type_preservation = get_elastic_preservation_by_type( ReleaseQuery(container_id=ident), ) - if view_template == "container_view_browse.html": - entity._browse_volume_year = get_elastic_container_browse_year_volume(entity.ident) return render_template( view_template, entity_type=entity_type, entity=entity, editgroup_id=None @@ -350,9 +348,93 @@ def container_view_coverage(ident: str) -> AnyResponse: @app.route("/container//browse", methods=["GET"]) -def container_view_browser(ident: str) -> AnyResponse: - # note: there is a special hack to add entity._type_preservation for this endpoint - return generic_entity_view("container", ident, "container_view_browse.html") +def container_view_browse(ident: str) -> AnyResponse: + entity = generic_get_entity("container", ident) + + if entity.state == "redirect": + return redirect(f"/container/{entity.redirect}") + elif entity.state == "deleted": + return render_template("deleted_entity.html", entity_type="container", entity=entity) + + query_sort: Optional[List[str]] + if request.args.get('year') and 'volume' in request.args and 'issue' in request.args: + # year, volume, issue specified; browse-by-page + year = int(request.args.get('year')) + volume = request.args.get('volume', '') + issue = request.args.get('issue', '') + if volume: + volume = f'volume:"{volume}"' + else: + volume = "!volume:*" + if issue: + issue = f'issue:"{issue}"' + else: + issue = "!issue:*" + query_string = f'year:{year} {volume} {issue}' + query_sort = ["first_page", "release_date"] + elif request.args.get('year') and 'volume' in request.args: + # year, volume specified (no issue); browse-by-page + year = int(request.args.get('year')) + volume = request.args.get('volume', '') + if volume: + volume = f'volume:"{volume}"' + else: + volume = "!volume:*" + query_string = f'year:{year} {volume}' + query_sort = ["issue", "first_page", "release_date"] + elif request.args.get('year'): + # year specified, not anything else; browse-by-date + year = int(request.args.get('year')) + query_string = f"year:{year}" + query_sort = ["release_date"] + else: + entity._browse_volume_year = get_elastic_container_browse_year_volume(entity.ident) + return render_template( + "container_view_browse.html", entity_type="container", entity=entity, editgroup_id=None + ) + + print(query_string) + query = ReleaseQuery( + q=query_string, + limit=200, + offset=0, + container_id=ident, + fulltext_only=False, + recent=False, + exclude_stubs=True, + sort=query_sort, + ) + + try: + found = do_release_search(query) + except FatcatSearchError as fse: + return ( + render_template( + "container_view_search.html", + query=query, + es_error=fse, + entity_type="container", + entity=entity, + editgroup_id=None, + ), + fse.status_code, + ) + + # HACK: re-sort by first page *numerically* + if found.results and query_sort and 'first_page' in query_sort: + for doc in found.results: + if doc.get('first_page') and doc['first_page'].isdigit(): + doc['first_page'] = int(doc['first_page']) + found.results = sorted(found.results, key=lambda d: d.get('first_page') or 99999999) + + return render_template( + "container_view_browse.html", + query=query, + releases_found=found, + entity_type="container", + entity=entity, + editgroup_id=None, + ) @app.route("/container//metadata", methods=["GET"]) diff --git a/python/fatcat_web/search.py b/python/fatcat_web/search.py index b19b57e6..3f4a216c 100644 --- a/python/fatcat_web/search.py +++ b/python/fatcat_web/search.py @@ -29,6 +29,7 @@ class ReleaseQuery: container_id: Optional[str] = None recent: bool = False exclude_stubs: bool = False + sort: Optional[List[str]] = None @staticmethod def from_args(args: Dict[str, Any]) -> "ReleaseQuery": @@ -45,6 +46,7 @@ class ReleaseQuery: container_id=args.get("container_id"), recent=bool(args.get("recent")), exclude_stubs=bool(args.get("exclude_stubs")), + sort=None, ) @@ -182,6 +184,9 @@ def do_release_search(query: ReleaseQuery, deep_page_limit: int = 2000) -> Searc negative_boost=0.5, ) + if query.sort: + search = search.sort(*query.sort) + # Sanity checks limit = min((int(query.limit or 25), 100)) offset = max((int(query.offset or 0), 0)) @@ -299,7 +304,7 @@ def get_elastic_container_browse_year_volume(ident: str) -> Dict[int, Dict[str, for year in year_nums: year_dicts[year] = {} for row in buckets: - year_dicts[int(row["key"]["year"])][row["key"]["volume"] or "_unknown"] = int( + year_dicts[int(row["key"]["year"])][row["key"]["volume"] or "000_unknown"] = int( row["doc_count"] ) # return sorted(year_dicts.values(), key=lambda x: x["year"]) diff --git a/python/fatcat_web/templates/container_view_browse.html b/python/fatcat_web/templates/container_view_browse.html index b5691899..aa88b666 100644 --- a/python/fatcat_web/templates/container_view_browse.html +++ b/python/fatcat_web/templates/container_view_browse.html @@ -4,28 +4,59 @@ {% import "entity_macros.html" as entity_macros %} {% extends "entity_base.html" %} -{% block entity_main %} - -{% if entity._browse_volume_year %} -

Browse by Year and Volume

-