From a16c54771f2a80d07f124387b3aca7d11c70a27b Mon Sep 17 00:00:00 2001 From: Bryan Newbold Date: Mon, 29 Jun 2020 17:39:08 -0700 Subject: fixes for pdf_meta dict --- fatcat_scholar/transform.py | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'fatcat_scholar') diff --git a/fatcat_scholar/transform.py b/fatcat_scholar/transform.py index 5bedfd7..5bdd0c7 100644 --- a/fatcat_scholar/transform.py +++ b/fatcat_scholar/transform.py @@ -151,7 +151,7 @@ def _add_file_release_meta( fulltext.file_mimetype = fe.mimetype fulltext.access_url = best_url fulltext.access_type = best_url_type - if pdf_meta is not None and pdf_meta.get("has_page0_thumbnail"): + if pdf_meta is not None and pdf_meta["pdf_meta"].get("has_page0_thumbnail"): # eg: https://blobs.fatcat.wiki/thumbnail/pdf/32/29/322909fe57cef73b10a166996a4528d337026d16.180px.jpg fulltext.thumbnail_url = f"{ settings.THUMBNAIL_URL_PREFIX }{ fe.sha1[0:2] }/{ fe.sha1[2:4] }/{ fe.sha1 }.180px.jpg" return fulltext @@ -338,6 +338,7 @@ def run_transform(infile: Sequence) -> None: biblio_release_ident=obj.get("biblio_release_ident"), grobid_fulltext=obj.get("grobid_fulltext"), pdftotext_fulltext=obj.get("pdftotext_fulltext"), + pdf_meta=obj.get("pdf_meta"), sim_fulltext=obj.get("sim_fulltext"), ) es_doc = transform_heavy(heavy) -- cgit v1.2.3