Add raw text option to text conversion (lesmianator will need it).

[librarian.git] / librarian / pdf.py

diff --git a/librarian/pdf.py b/librarian/pdf.py
index f8a4d73..5f6f0a2 100644 (file)
--- a/librarian/pdf.py
+++ b/librarian/pdf.py
@@ -37,7 +37,7 @@ STYLESHEETS = {
 
 
 def insert_tags(doc, split_re, tagname, exclude=None):
-    """ inserts <tagname> for every occurence of `split_re' in text nodes in the `doc' tree 
+    """ 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>');
     >>> insert_tags(t, re.compile('-'), 'd');
@@ -67,7 +67,7 @@ def insert_tags(doc, split_re, tagname, exclude=None):
 
 
 def substitute_hyphens(doc):
-    insert_tags(doc, 
+    insert_tags(doc,
                 re.compile("(?<=[^-\s])-(?=[^-\s])"),
                 "dywiz",
                 exclude=[DCNS("identifier.url"), DCNS("rights.license")]
@@ -75,7 +75,7 @@ def substitute_hyphens(doc):
 
 
 def fix_hanging(doc):
-    insert_tags(doc, 
+    insert_tags(doc,
                 re.compile("(?<=\s\w)\s+"),
                 "nbsp",
                 exclude=[DCNS("identifier.url"), DCNS("rights.license")]
@@ -132,9 +132,12 @@ def parse_creator(doc):
     for person in doc.xpath("|".join('//dc:'+(tag) for tag in (
                     'creator', 'contributor.translator', 'contributor.editor', 'contributor.technical_editor')),
                     namespaces = {'dc': str(DCNS)})[::-1]:
+        if not person.text:
+            continue
         p = Person.from_text(person.text)
         person_parsed = deepcopy(person)
         person_parsed.tag = person.tag + '_parsed'
+        person_parsed.set('sortkey', person.text)
         person_parsed.text = p.readable()
         person.getparent().insert(0, person_parsed)
 
@@ -144,7 +147,7 @@ def get_stylesheet(name):
 
 
 def package_available(package, args='', verbose=False):
-    """ check if a verion of a latex package accepting given args is available """  
+    """ check if a verion of a latex package accepting given args is available """
     tempdir = mkdtemp('-wl2pdf-test')
     fpath = os.path.join(tempdir, 'test.tex')
     f = open(fpath, 'w')
@@ -163,7 +166,7 @@ def package_available(package, args='', verbose=False):
     return p == 0
 
 
-def transform(provider, slug=None, file_path=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
 
@@ -268,7 +271,7 @@ def transform(provider, slug=None, file_path=None,
 
 
 def load_including_children(provider, slug=None, uri=None, file_path=None):
-    """ makes one big xml file with children inserted at end 
+    """ makes one big xml file with children inserted at end
     either slug or uri must be provided
     """
 
@@ -288,7 +291,6 @@ def load_including_children(provider, slug=None, uri=None, file_path=None):
         parse_dublincore=True)
 
     f.close()
-    print document.book_info
     for child_uri in document.book_info.parts:
         print child_uri
         child = load_including_children(provider, uri=child_uri)