basic demo CSL/citeproc transform code

Needs tests

1 files changed, 60 insertions, 1 deletions

diff --git a/python/fatcat_export.py b/python/fatcat_export.py
index 33e23202..a9d46142 100755
--- a/python/fatcat_export.py
+++ b/python/fatcat_export.py
@@ -10,12 +10,19 @@ without permission by an third party.
 import sys
 import json
 import argparse
+
+from citeproc import CitationStylesStyle, CitationStylesBibliography
+from citeproc import Citation, CitationItem
+from citeproc import formatter
+from citeproc.source.json import CiteProcJSON
+from citeproc_styles import get_style_filepath
+
 import fatcat_client
 from fatcat_client.rest import ApiException
 from fatcat_client import ReleaseEntity, ContainerEntity, ChangelogEntry
 from fatcat_tools import uuid2fcid, entity_from_json, entity_to_dict, \
     release_to_elasticsearch, container_to_elasticsearch, \
-    changelog_to_elasticsearch, public_api
+    changelog_to_elasticsearch, public_api, release_to_csl
 
 
 def run_export_releases(args):
@@ -52,6 +59,43 @@ def run_transform_changelogs(args):
         args.json_output.write(
             json.dumps(changelog_to_elasticsearch(entity)) + '\n')
 
+def run_citeproc_releases(args):
+    for line in args.json_input:
+        line = line.strip()
+        if not line:
+            continue
+        entity = entity_from_json(line, ReleaseEntity, api_client=args.api.api_client)
+        csl_json = release_to_csl(entity)
+        # XXX:
+        csl_json['id'] = "release:" + (entity.ident or "unknown")
+        if args.style == "csl-json":
+            args.json_output.write(json.dumps(csl_json) + "\n")
+            continue
+        bib_src = CiteProcJSON([csl_json])
+        form = formatter.plain
+        if args.html:
+            form = formatter.html
+        style_path = get_style_filepath(args.style)
+        bib_style = CitationStylesStyle(style_path, validate=False)
+        bib = CitationStylesBibliography(bib_style, bib_src, form)
+        bib.register(Citation([CitationItem(csl_json['id'])]))
+        # XXX:
+        #args.json_output.write(
+        #    json.dumps(release_to_csl(entity)) + '\n')
+        lines = bib.bibliography()[0]
+        if args.style == "bibtex":
+            for l in lines:
+                if l.startswith(" @"):
+                    args.json_output.write("\n@")
+                elif l.startswith(" "):
+                    #print("line: START|{}|END".format(l))
+                    args.json_output.write("\n  " + l)
+                else:
+                    args.json_output.write(l)
+        else:
+            args.json_output.write(''.join(lines) + "\n")
+        print()
+
 def run_export_changelog(args):
     end = args.end
     if end is None:
@@ -109,6 +153,21 @@ def main():
         help="where to send output",
         default=sys.stdout, type=argparse.FileType('w'))
 
+    sub_citeproc_releases = subparsers.add_parser('citeproc-releases')
+    sub_citeproc_releases.set_defaults(func=run_citeproc_releases)
+    sub_citeproc_releases.add_argument('json_input',
+        help="JSON-per-line of release entities",
+        default=sys.stdin, type=argparse.FileType('r'))
+    sub_citeproc_releases.add_argument('json_output',
+        help="where to send output",
+        default=sys.stdout, type=argparse.FileType('w'))
+    sub_citeproc_releases.add_argument('--style',
+        help="citation style to output",
+        default='csl-json')
+    sub_citeproc_releases.add_argument('--html',
+        action='store_true',
+        help="output HTML, not plain text")
+
     sub_changelog = subparsers.add_parser('changelog')
     sub_changelog.set_defaults(func=run_export_changelog)
     sub_changelog.add_argument('--start',