def articleExists(dict_tree, title, start_page, end_page):
'''
Returns true if an article exists in dict_tree.
The test is done by checking title, start page and end page
:param dict_tree:
:param title:
:param start_page:
:param end_page:
'''
def getPages(ps,i):
r = []
t = sorted([int(p[0]) for p in pages if p[0].isdigit()])
if t: r.append(t[i])
t = sorted([int(p[1]) for p in pages if p[1].isdigit()])
if t: r.append(t[i])
return r
named_articles = dmd_sec_tools.getDmdsWithContent(dict_tree,
'Article',
'TitleDocMain',
title)
for a in named_articles:
#=======================================================================
# If there are article in dict_tree with same name as the one being
# added, compare start and end pages.
#=======================================================================
phys_pages = struct_link_tools.getPhysListByLogIds(dict_tree, [a])
physical_struct_map = phys_struct_map_tools.get(dict_tree)
pages = phys_struct_map_tools.getSelectedPages(physical_struct_map, phys_pages)
# pages: a list of (str,str) tuples, where first elem is ORDER and
# second elem is ORDERLBEL
# 1: Find first page in pages (both ORDER and ORDERLABEL) and see if
# it is equal to start_page.
# We need to check both because possible offset difference
start_pages = getPages(pages,0)
same_start_page = (start_page in start_pages)
# 2: Same with end_page
end_pages = getPages(pages,-1)
same_end_page = (end_page in end_pages)
if (same_start_page and same_end_page):
return True
return False
def addPagesOnLevel(logical_struct_map,dict_tree):
div_key = '{http://www.loc.gov/METS/}div'
if isinstance(logical_struct_map, dict): # Only one root
parent_id = logical_struct_map['@ID']
if struct_link_tools.exists(dict_tree, parent_id):
# Current element already has pages
return
children_ids = log_struct_map_tools.getChildrenLogIds(logical_struct_map)
if not children_ids: # No children - dammit
return
if div_key in logical_struct_map:
# Recursively handle subtree before adding pages to this element
addPagesOnLevel(logical_struct_map[div_key],dict_tree)
phys_list = struct_link_tools.getPhysListByLogIds(dict_tree, children_ids)
# Add children pages (phys_list) to parent (parent_id)
dict_tree = struct_link_tools.addLinks(dict_tree, parent_id, phys_list)
#pprint.pprint(struct_link)
elif isinstance(logical_struct_map, list): # Multiple branches
for branch in logical_struct_map:
addPagesOnLevel(branch,dict_tree)
return dict_tree
