X-Git-Url: https://git.mdrn.pl/librarian.git/blobdiff_plain/5175c4cb9704f442c20abffbd575fb588278b3b2..1ef7c1807a866e5d8fb99238499be4ea240dceae:/librarian/pdf.py?ds=sidebyside diff --git a/librarian/pdf.py b/librarian/pdf.py index c02e98c..44ec644 100644 --- a/librarian/pdf.py +++ b/librarian/pdf.py @@ -70,19 +70,30 @@ def insert_tags(doc, split_re, tagname, exclude=None): def substitute_hyphens(doc): - insert_tags(doc, - re.compile("(?<=[^-\s])-(?=[^-\s])"), - "dywiz", - exclude=[DCNS("identifier.url"), DCNS("rights.license"), 'www'] - ) + insert_tags( + doc, re.compile("(?<=[^-\s])-(?=[^-\s])"), "dywiz", + exclude=[ + DCNS("identifier.url"), + DCNS("rights.license"), + DCNS("title"), + DCNS("description"), + DCNS("subject.curriculum"), + 'www', + ] + ) def fix_hanging(doc): - insert_tags(doc, - re.compile("(?<=\s\w)\s+"), - "nbsp", - exclude=[DCNS("identifier.url"), DCNS("rights.license"), DCNS("title"), DCNS("description")] - ) + insert_tags( + doc, re.compile("(?<=\s\w)\s+"), "nbsp", + exclude=[ + DCNS("identifier.url"), + DCNS("rights.license"), + DCNS("title"), + DCNS("description"), + DCNS("subject.curriculum"), + ] + ) def move_motifs_inside(doc):