7 from lxml.builder import ElementMaker
9 from pkg_resources import resource_filename, resource_listdir
12 import treecutter.constants as const
13 from treecutter.image import Image
14 from treecutter.tools import warning
17 """Class representing a docbook document"""
18 def __init__(self,filename):
19 self._filename = filename
20 self._doc = etree.parse(self._filename)
21 self._dirname = os.path.dirname(self._filename)
24 t = self._doc.xpath(u'/db:article/db:info/db:title',namespaces=const.XPATH)
26 t = unicode(t[0].text)
27 ta = self._doc.xpath(u'/db:article/db:info/db:titleabbrev',namespaces=const.XPATH)
29 ta = unicode(ta[0].text)
32 def expand_imageobjects(self):
34 db = ElementMaker(namespace=const.DB_NS, nsmap=const.NSMAP)
35 images = self._doc.xpath(u"//db:imageobject/db:imagedata[@fileref]",namespaces=const.XPATH)
37 os.chdir(self._dirname)
40 caption = db.caption()
41 for p in img.caption().split('\n\n'):
42 caption.append(db.para(p))
43 link = db.para(db.link(img.infostr(),
44 **{const.XLINK+"href": img.filename()}))
46 mo = db.mediaobject(db.imageobject(
47 db.imagedata(fileref=img.resize(800,600))),caption)
50 mopp = mop.getparent()
51 mopp.insert(mopp.index(mop)+1,mo)
56 def parse_xincludes(self):
58 for c in self._doc.xpath(u"//xi:include[@parse='text']",namespaces=const.XPATH):
60 alang = c.get('accept-language')
61 xpointer = c.get('xpointer')
62 (p, ext) = os.path.splitext(href)
63 if ext in const.valid_scripts:
65 script = os.path.join(os.path.abspath(self._dirname)+'/'+href)
66 if os.path.isfile(script):
69 if href in resource_listdir('xinclude', ''):
70 script = resource_filename('xinclude', href)
73 print "Script "+href+" in "+self._filename+" missing"
75 exe.append("lang="+alang)
77 exe.append("xptr="+xpointer)
78 print " executing %15s" % (href),
80 os.chdir(self._dirname)
81 xml = subprocess.Popen(exe,stdout=subprocess.PIPE,
82 stderr=subprocess.PIPE)
83 (stdout, stderr) = xml.communicate()
84 #print xml.returnvalue
86 warning("%s : %s" % (" ".join(exe),stderr))
91 print " [%5.2f s] (%s)" % (round(te-ts,2),xpointer)
92 xstr = etree.fromstring(stdout)
93 # inserting the generated code and remove the xinclude reference
95 idp.insert(idp.index(c)+1,xstr)
98 def collect_links(self):
100 for r in self._doc.xpath(u"//db:link[@xlink:href]",namespaces=const.XPATH):
101 rf = os.path.join(self._dirname,r.get(const.XLINK+'href'))
102 if os.path.isfile(rf):
103 if r.get('security')=='encrypt':
104 with open(rf, 'rb') as f:
106 status = gpg.encrypt_file(
107 f, None, passphrase=getpass.getpass(rf+' password:'), symmetric=True,
109 r.set(const.XLINK+'href', r.get(const.XLINK+'href')+'.gpg')
114 def collect_images(self):
116 for i in self._doc.xpath(u"//db:imagedata[@fileref]",namespaces=const.XPATH):
117 im = os.path.join(self._dirname,i.get('fileref'))
118 if os.path.isfile(im):
121 print "WARNING: File "+im+" is missing!"
124 def collect_videos(self):
126 for i in self._doc.xpath(u"//db:videodata[@fileref]",namespaces=const.XPATH):
127 im = os.path.join(self._dirname,i.get('fileref'))
128 if os.path.isfile(im):
131 print "WARNING: File "+im+" is missing!"
134 def collect_forms(self):
136 for i in self._doc.xpath(u"//html:form[@action]",namespaces=const.XPATH):
137 pyscript = re.split('\.py',i.get('action'),1)[0]+'.py'
138 im = os.path.join(self._dirname,pyscript)
139 if os.path.isfile(im):
144 return etree.tostring(self._doc,encoding='UTF-8',pretty_print=False)
146 def xslt(self,transform):
147 return etree.tostring(transform(self._doc))
150 def recursively_empty(e):
153 return all((recursively_empty(c) for c in e.iterchildren()))
155 context = etree.iterwalk(self._doc)
156 for action, elem in context:
157 parent = elem.getparent()
158 if recursively_empty(elem):