update journal meta import/transform

2 files changed, 194 insertions, 132 deletions

diff --git a/python/fatcat_tools/importers/journal_metadata.py b/python/fatcat_tools/importers/journal_metadata.py
index cf3971b5..7f6b1ee8 100644
--- a/python/fatcat_tools/importers/journal_metadata.py
+++ b/python/fatcat_tools/importers/journal_metadata.py
@@ -28,90 +28,9 @@ def truthy(s):
 class JournalMetadataImporter(EntityImporter):
     """
     Imports journal metadata ("containers") by ISSN, currently from a custom
-    (data munged) .csv file format
-
-    CSV format (generated from git.archive.org/webgroup/oa-journal-analysis):
-
-        ISSN-L,in_doaj,in_road,in_norwegian,in_crossref,title,publisher,url,lang,ISSN-print,ISSN-electronic,doi_count,has_doi,is_oa,is_kept,publisher_size,url_live,url_live_status,url_live_final_status,url_live_final_url,url_live_status_simple,url_live_final_status_simple,url_domain,gwb_pdf_count
-
-
-    'extra' fields:
-
-        doaj
-            as_of: datetime of most recent check; if not set, not actually in DOAJ
-            seal: bool
-            work_level: bool (are work-level publications deposited with DOAJ?)
-            archiving: array, can include 'library' or 'other'
-        road
-            as_of: datetime of most recent check; if not set, not actually in ROAD
-        pubmed (TODO: delete?)
-            as_of: datetime of most recent check; if not set, not actually indexed in pubmed
-        norwegian (TODO: drop this?)
-            as_of: datetime of most recent check; if not set, not actually indexed in pubmed
-            id (integer)
-            level (integer; 0-2)
-        kbart
-            lockss
-                year_rle
-                volume_rle
-            portico
-                ...
-            clockss
-                ...
-        sherpa_romeo
-            color
-        jstor
-            year_rle
-            volume_rle
-        scopus
-            id
-            TODO: print/electronic distinction?
-        wos
-            id
-        doi
-            crossref_doi: DOI of the title in crossref (if exists)
-            prefixes: array of strings (DOI prefixes, up to the '/'; any registrar, not just Crossref)
-        ia
-            sim
-                nap_id
-                year_rle
-                volume_rle
-            longtail: boolean
-            homepage
-                as_of: datetime of last attempt
-                url
-                status: HTTP/heritrix status of homepage crawl
-
-        issnp: string
-        issne: string
-        coden: string
-        abbrev: string
-        oclc_id: string (TODO: lookup?)
-        lccn_id: string (TODO: lookup?)
-        dblb_id: string
-        default_license: slug
-        original_name: native name (if name is translated)
-        platform: hosting platform: OJS, wordpress, scielo, etc
-        mimetypes: array of strings (eg, 'application/pdf', 'text/html')
-        first_year: year (integer)
-        last_year: if publishing has stopped
-        primary_language: single ISO code, or 'mixed'
-        languages: array of ISO codes
-        region: TODO: continent/world-region
-        nation: shortcode of nation
-        discipline: TODO: highest-level subject; "life science", "humanities", etc
-        field: TODO: narrower description of field
-        subjects: TODO?
-        url: homepage
-        is_oa: boolean. If true, can assume all releases under this container are "Open Access"
-        TODO: domains, if exclusive?
-        TODO: fulltext_regex, if a known pattern?
-
-    For KBART, etc:
-        We "over-count" on the assumption that "in-progress" status works will soon actually be preserved.
-        year and volume spans are run-length-encoded arrays, using integers:
-            - if an integer, means that year is preserved
-            - if an array of length 2, means everything between the two numbers (inclusive) is preserved
+    munged JSON format (see ../extra/journal_metadata/).
+
+    See guide for details on the many 'extra' fields used here.
     """
 
     def __init__(self, api, **kwargs):
@@ -125,34 +44,50 @@ class JournalMetadataImporter(EntityImporter):
             editgroup_extra=eg_extra)
 
     def want(self, raw_record):
-        if raw_record.get('ISSN-L'):
+        if raw_record.get('issnl'):
             return True
         return False
 
     def parse_record(self, row):
         """
-        row is a python dict (parsed from CSV).
+        row is a python dict (parsed from JSON).
+
         returns a ContainerEntity (or None if invalid or couldn't parse)
         """
-        title = or_none(row['title'])
-        issnl = or_none(row['ISSN-L'])
-        if title is None or issnl is None:
-            return None
-        extra = dict(
-            in_doaj=truthy(row['in_doaj']),
-            in_road=truthy(row['in_road']),
-            in_norwegian=truthy(row['in_norwegian']),
-            language=or_none(row['lang']),
-            url=or_none(row['url']),
-            ISSNp=or_none(row['ISSN-print']),
-            ISSNe=or_none(row['ISSN-electronic']),
-            is_oa=truthy(row['is_oa']),
-            is_kept=truthy(row['is_kept']),
-        )
+
+        extra = dict()
+        for key in ('issne', 'issnp', 'languages', 'country', 'urls', 'abbrev',
+            'coden', 'aliases', 'original_name', 'first_year', 'last_year',
+            'platform', 'default_license', 'road', 'mimetypes',
+            'sherpa_romeo', 'kbart'):
+            if row.get(key):
+                extra[key] = row[key]
+        # TODO: not including for now: norwegian, dois/crossref, ia
+
+        extra_doaj = dict()
+        if row.get('doaj'):
+            if row['doaj'].get('as_of'):
+                extra_doaj['as_of'] = row['doaj']['as_of']
+            if row['doaj'].get('works'):
+                extra_doaj['works'] = row['doaj']['works']
+        if extra_doaj:
+            extra['doaj'] = extra_doaj
+
+        extra_ia = dict()
+        # TODO: would like an ia.longtail_ia flag
+        if row.get('sim'):
+            extra_ia['sim'] = {
+                'year_spans': row['sim']['year_spans'],
+            }
+        if extra_ia:
+            extra['ia'] = extra_ia
+
         ce = fatcat_client.ContainerEntity(
-            issnl=issnl,
-            name=clean(title),
-            publisher=or_none(clean(row['publisher'])),
+            issnl=row['issnl'],
+            container_type=None, # TODO
+            name=clean(row.get('name')),
+            publisher=clean(row.get('publisher')),
+            wikidata_qid=None, # TODO
             extra=extra)
         return ce
 
diff --git a/python/fatcat_tools/transforms.py b/python/fatcat_tools/transforms.py
index 2493b1ab..a85c877c 100644
--- a/python/fatcat_tools/transforms.py
+++ b/python/fatcat_tools/transforms.py
@@ -19,7 +19,22 @@ def entity_from_json(json_str, entity_type):
     thing.data = json_str
     return ac.deserialize(thing, entity_type)
 
-def release_to_elasticsearch(release):
+def check_kbart(year, archive):
+    if not archive or not archive.get('year_spans'):
+        return None
+    for span in archive['year_spans']:
+        if year >= span[0] and year <= span[1]:
+            return True
+    return False
+
+def test_check_kbart():
+
+    assert check_kbart(1990, dict(year_spans=[[2000, 2000]])) == False
+    assert check_kbart(2000, dict(year_spans=[[2000, 2000]])) == True
+    assert check_kbart(1950, dict(year_spans=[[1900, 1920], [1990, 2000]])) == False
+    assert check_kbart(1950, dict(year_spans=[[1900, 1920], [1930, 2000]])) == True
+
+def release_to_elasticsearch(entity):
     """
     Converts from an entity model/schema to elasticsearch oriented schema.
 
@@ -27,15 +42,16 @@ def release_to_elasticsearch(release):
     Raises exception on error (never returns None)
     """
 
-    if release.state in ('redirect', 'deleted'):
+    if entity.state in ('redirect', 'deleted'):
         return dict(
-            ident = release.ident,
-            state = release.state,
+            ident = entity.ident,
+            state = entity.state,
         )
-    elif release.state != 'active':
-        raise ValueError("Unhandled release state: {}".format(release.state))
+    elif entity.state != 'active':
+        raise ValueError("Unhandled entity state: {}".format(entity.state))
 
     # First, the easy ones (direct copy)
+    release = entity
     t = dict(
         ident = release.ident,
         state = release.state,
@@ -57,11 +73,14 @@ def release_to_elasticsearch(release):
     )
 
     is_oa = None
+    is_preserved = None
     is_longtail_oa = None
     in_kbart = None
+    in_jstor = False
     in_web = False
     in_dweb = False
     in_ia = False
+    in_ia_sim = False
     in_shadow = False
 
     if release.release_date:
@@ -88,19 +107,35 @@ def release_to_elasticsearch(release):
         t['container_issnl'] = container.issnl
         t['container_type'] = container.container_type
         if container.extra:
-            if container.extra.get('is_oa') or container.extra.get('in_doaj'):
+            c_extra = container.extra
+            if c_extra.get('kbart') and release.year:
+                in_jstor = check_kbart(release.year, c_extra['kbart'].get('jstor'))
+                in_kbart = in_jstor
+                for archive in ('portico', 'lockss', 'clockss'):
+                    in_kbart = in_kbart or check_kbart(release.year, c_extra['kbart'].get(archive))
+
+            if c_extra.get('ia'):
+                if c_extra['ia'].get('sim') and release.year:
+                    in_ia_sim = check_kbart(release, c_extra['ia']['sim'].get('year_spans'))
+                if c_extra['ia'].get('longtail_oa'):
+                    is_longtail_oa = True
+            if c_extra.get('sherpa_romeo'):
+                if c_extra['sherpa_romeo'].get('color') == 'white':
+                    is_oa = False
+            if c_extra.get('default_license') and c_extra.get('default_license').startswith('CC-'):
                 is_oa = True
-            if container.extra.get('in_kbart'):
-                # TODO: better KBART check goes here
-                in_kbart = True
-            if container.extra.get('ia'):
-                # TODO: container longtail check goes here
-                # TODO: sim/microfilm check goes here
-                pass
-            # TODO: SHERPA/Romeo goes here
+            if c_extra.get('doaj'):
+                if c_extra['doaj'].get('as_of'):
+                    is_oa = True
+            if c_extra.get('road'):
+                if c_extra['road'].get('as_of'):
+                    is_oa = True
     else:
         t['publisher'] = release.publisher
 
+    if release.jstor_id or (release.doi and release.doi.startswith('10.2307/')):
+        in_jstor = True
+
     files = release.files or []
     t['file_count'] = len(files)
     t['fileset_count'] = len(release.filesets or [])
@@ -118,13 +153,15 @@ def release_to_elasticsearch(release):
             if url.url.lower().startswith('http'):
                 in_web = True
             if url.rel in ('dweb', 'p2p', 'ipfs', 'dat', 'torrent'):
-                # TODO: not sure what rel will be
+                # not sure what rel will be for this stuff
                 in_dweb = True
             if is_pdf:
                 any_pdf_url = url.url
             if is_pdf and url.rel in ('webarchive', 'repository') and is_pdf:
                 is_preserved = True
                 good_pdf_url = url.url
+            if '//www.jstor.org/' in url.url:
+                in_jstor = True
             if '//web.archive.org/' in url.url or '//archive.org/' in url.url:
                 in_ia = True
                 if is_pdf:
@@ -141,18 +178,15 @@ def release_to_elasticsearch(release):
 
     extra = release.extra or dict()
     if extra:
-        # TODO: longtail OA check from GROBID here
-        if extra.get('in_kbart'):
-            # NOTE: not actually setting this anywhere
-            in_kbart = True
         if extra.get('is_oa'):
-            # NOTE: not actually setting this anywhere
+            # NOTE: not actually setting this anywhere... but could
             is_oa = True
-        if extra.get('grobid'):
-            if not t.get('container_name'):
-                t['container_name'] = extra['grobid'].get('container_name')
-            if extra['grobid'].get('longtail_oa'):
-                is_longtail_oa = True
+        if extra.get('longtail_oa'):
+            # sometimes set by GROBID/matcher
+            is_oa = True
+            is_longtail_oa = True
+        if not t.get('container_name'):
+            t['container_name'] = extra.get('container_name')
         if extra.get('crossref'):
             if extra['crossref'].get('archive'):
                 # all crossref archives are KBART, I believe
@@ -163,8 +197,101 @@ def release_to_elasticsearch(release):
     t['is_oa'] = is_oa
     t['is_longtail_oa'] = is_longtail_oa
     t['in_kbart'] = in_kbart
+    t['in_jstor'] = in_jstor
     t['in_web'] = in_web
     t['in_dweb'] = in_dweb
-    t['in_ia'] = in_ia
-    t['is_preserved'] = in_ia or in_kbart
+    t['in_ia'] = bool(in_ia)
+    t['is_preserved'] = bool(is_preserved or in_ia or in_kbart or in_jstor)
+    return t
+
+def container_to_elasticsearch(entity):
+    """
+    Converts from an entity model/schema to elasticsearch oriented schema.
+
+    Returns: dict
+    Raises exception on error (never returns None)
+    """
+
+    if entity.state in ('redirect', 'deleted'):
+        return dict(
+            ident = entity.ident,
+            state = entity.state,
+        )
+    elif entity.state != 'active':
+        raise ValueError("Unhandled entity state: {}".format(entity.state))
+
+    # First, the easy ones (direct copy)
+    t = dict(
+        ident = entity.ident,
+        state = entity.state,
+        revision = entity.revision,
+
+        name = entity.name,
+        publisher = entity.publisher,
+        container_type = entity.container_type,
+        issnl = entity.issnl,
+        wikidata_qid = entity.wikidata_qid,
+
+        entity_status = entity.entity_status,
+        language = entity.language,
+        license = entity.license_slug,
+        doi = entity.doi,
+        pmid = entity.pmid,
+        isbn13 = entity.isbn13,
+        core_id = entity.core_id,
+        arxiv_id = entity.core_id,
+        jstor_id = entity.jstor_id,
+    )
+
+    # TODO: region, discipline
+    # TODO: single primary language?
+    for key in ('country', 'languages', 'mimetypes', 'first_year', 'last_year'):
+        if entity.extra.get(key):
+            t[key] = entity.extra[key]
+
+    in_doaj = None
+    in_road = None
+    # TODO: not currently implemented
+    in_doi = None
+    # TODO: would be nice to have 'in_doaj_works', or maybe just "any_pid"
+    #in_doaj_works = None
+    in_sherpa_romeo = None
+    is_oa = None
+    # TODO: not actually set/stored anywhere?
+    is_longtail_oa = None
+    any_kbart = None
+    any_jstor = None
+    any_ia_sim = None
+
+    extra = entity.extra
+    if extra.get('doaj'):
+        if extra['doaj'].get('as_of'):
+            in_doaj = True
+    if extra.get('road'):
+        if extra['road'].get('as_of'):
+            in_road = True
+    if extra.get('default_license'):
+        if extra['default_license'].startswith('CC-'):
+            is_oa = True
+    if extra.get('sherpa_romeo'):
+        in_sherpa_romeo = True
+        if extra['sherpa_romeo'].get('color') == 'white':
+            is_oa = False
+    if extra.get('kbart'):
+        any_kbart = True
+        if extra['kbart'].get('jstor'):
+            any_jstor = True
+    if extra.get('ia'):
+        if extra['ia'].get('sim'):
+            any_ia_sim = True
+
+    t['in_doaj'] = is_doaj
+    t['in_road'] = is_road
+    t['in_doi'] = in_doi
+    t['in_sherpa_romeo'] = in_sherpa_romeo
+    t['is_oa'] = in_doaj or in_road or is_longtail_oa or ia_oa
+    t['is_longtail_oa'] = is_longtail_oa
+    t['any_kbart'] = any_ia_sim
+    t['any_jstor'] = any_ia_sim
+    t['any_ia_sim'] = bool(any_ia_sim)
     return t