X-Git-Url: https://git.mdrn.pl/librarian.git/blobdiff_plain/33165a21eb5b83f96d73e616414a1ee9e7227013..e4836157dee23cebc8e69e9dafab3cac41f20035:/librarian/pdf.py?ds=sidebyside

diff --git a/librarian/pdf.py b/librarian/pdf.py
index 32fa571..2d45372 100644
--- a/librarian/pdf.py
+++ b/librarian/pdf.py
@@ -36,7 +36,7 @@ STYLESHEETS = {
 }
 
 
-def insert_tags(doc, split_re, tagname):
+def insert_tags(doc, split_re, tagname, exclude=None):
     """ inserts <tagname> for every occurence of `split_re' in text nodes in the `doc' tree 
 
     >>> t = etree.fromstring('<a><b>A-B-C</b>X-Y-Z</a>');
@@ -46,6 +46,8 @@ def insert_tags(doc, split_re, tagname):
     """
 
     for elem in doc.iter(tag=etree.Element):
+        if exclude and elem.tag in exclude:
+            continue
         if elem.text:
             chunks = split_re.split(elem.text)
             while len(chunks) > 1:
@@ -67,13 +69,17 @@ def insert_tags(doc, split_re, tagname):
 def substitute_hyphens(doc):
     insert_tags(doc, 
                 re.compile("(?<=[^-\s])-(?=[^-\s])"),
-                "dywiz")
+                "dywiz",
+                exclude=[DCNS("identifier.url"), DCNS("rights.license")]
+                )
 
 
 def fix_hanging(doc):
     insert_tags(doc, 
                 re.compile("(?<=\s\w)\s+"),
-                "nbsp")
+                "nbsp",
+                exclude=[DCNS("identifier.url"), DCNS("rights.license")]
+                )
 
 
 def move_motifs_inside(doc):
@@ -158,7 +164,8 @@ def package_available(package, args='', verbose=False):
     return p == 0
 
 
-def transform(provider, slug=None, file_path=None, output_file=None, output_dir=None, make_dir=False, verbose=False, save_tex=None):
+def transform(provider, slug=None, file_path=None, 
+              output_file=None, output_dir=None, make_dir=False, verbose=False, save_tex=None, morefloats=None):
     """ produces a PDF file with XeLaTeX
 
     provider: a DocProvider
@@ -169,6 +176,7 @@ def transform(provider, slug=None, file_path=None, output_file=None, output_dir=
     make_dir: writes output to <output_dir>/<author>/<slug>.pdf istead of <output_dir>/<slug>.pdf
     verbose: prints all output from LaTeX
     save_tex: path to save the intermediary LaTeX file to
+    morefloats (old/new/none): force specific morefloats
     """
 
     # Parse XSLT
@@ -183,9 +191,10 @@ def transform(provider, slug=None, file_path=None, output_file=None, output_dir=
             document = load_including_children(provider, slug=slug)
 
         # check for LaTeX packages
-        if not package_available('morefloats', 'maxfloats=19'):
-            # using old morefloats or none at all
-            document.edoc.getroot().set('old-morefloats', 'yes')
+        if morefloats:
+            document.edoc.getroot().set('morefloats', morefloats.lower())
+        elif package_available('morefloats', 'maxfloats=19'):
+            document.edoc.getroot().set('morefloats', 'new')
 
         # hack the tree
         move_motifs_inside(document.edoc)
@@ -274,8 +283,7 @@ def load_including_children(provider, slug=None, uri=None, file_path=None):
         raise ValueError('Neither slug, URI nor file path provided for a book.')
 
     document = WLDocument.from_file(f, True,
-        parse_dublincore=True,
-        preserve_lines=False)
+        parse_dublincore=True)
 
     f.close()