5   def create(self, i_url, i_root, i_re_neg = "", i_title = ""):  argument
9     self.relative_url = i_url
10     self.title = i_title    
11     self.full_url = self.merge_url(i_url, i_root)    
12     if self.full_url is None:
18       if not(r.match(self.full_url) is None):
24     self.netloc = urlparse.urlparse(self.full_url).netloc        
27   def merge_url(self, i_url, i_root):  argument
71   def __init__(self):  argument
73     self.place_from = place()
74     self.place_to   = place()
75     self.caption = ""
81   def __init__(self):  argument
83     self.places = []
84     self.links  = []
86   def add_place(self, i_place):  argument
87     ret_place = self.find_place(i_place)    
89       self.places.append(i_place)
94   def find_place(self, i_place):  argument
95     for i in self.places:    
100   def add_link(self, i_place_from, i_place_to, i_caption, i_root_url = ""):  argument
115       i_caption = self.escape_html(i_place_to.full_url)
118     lnk.place_from=self.add_place(i_place_from)
119     lnk.place_to  =self.add_place(i_place_to)
121     self.links.append(lnk)
124   def has_page_from_role(self, i_page):  argument
126     for lnk in self.links:      
131   def escape_html(self, i_link):  argument
139     def process_site(self, i_url, i_counter = 100, i_re_neg = ""):        argument
143       self.loop_counter = self.loop_counter + 1      
145       if self.root_netloc == "":
147         self.root_netloc = p.netloc
148         self.root_url = i_url
150       self.url = i_url      
151       self.negative_regexp = i_re_neg
157         self.parse(s)
164       next_url = self.get_next_url()
166       if (next_url is None) or (self.loop_counter > i_counter):        
171         self.process_site(next_url, i_counter, i_re_neg)
175     def parse(self, s):  argument
177       self.feed(s)
178       self.close()
180     def __init__(self, verbose=0):  argument
182       sgmllib.SGMLParser.__init__(self, verbose)
183       self.network = network()
184       self.page_counter = 0
185       self.root_netloc = ""
186       self.exp_branch = []
187       self.str_indent = ""
188       self.tag_value = []
189       self.curr_ahref = ""
190       self.curr_title = ""
191       self.curr_ahref_title = ""      
192       self.interesting_tag = 0
193       self.loop_counter = 0
194       self.negative_regexp = ""
195       self.root_url = ""
197     def start_a(self, attributes):  argument
199       self.curr_ahref   = ""
200       self.curr_ahref_title = ""                
201       self.tag_value = []      
202       self.interesting_tag = 0
206           self.curr_ahref = value
207           if self.curr_ahref[0] == "#":
209             self.interesting_tag = 0
211             self.interesting_tag = 1
214           self.curr_ahref_title = value                                      
216     def end_a(self):  argument
218       if self.interesting_tag == 0:
222       if len(self.curr_ahref) > 0:
224         if self.curr_ahref[0] == "#":
225           self.curr_ahref = self.url + self.curr_ahref
227       title = string.join(self.tag_value, "").strip()
229         title = self.curr_ahref_title 
233       res_from = p_from.create("", self.url, self.negative_regexp, self.curr_title)
235       res_to   = p_to.create(self.curr_ahref, self.url, self.negative_regexp)
237       self.interesting_tag = 0
241         self.network.add_link(p_from, p_to, title, self.root_url)                               
243     def start_title(self, attributes):      argument
245       self.interesting_tag = 1
246       self.tag_value = []
249     def end_title(self):  argument
251       self.interesting_tag = 0
252       self.curr_title = string.join(self.tag_value, "").strip()
255     def handle_data(self, data):  argument
257       if self.interesting_tag == 1:
258          self.tag_value.append(data)
261     def print_hyperlinks(self):  argument
263       for lnk in self.network.links:            
268     def export_item(self, i_place, i_caption, i_indent, i_parent_url = "", i_root_url = ""):  argument
279         i_root_url = self.root_url
281       is_branch = self.network.has_page_from_role(i_place)
284       self.str_indent = "  " * i_indent
285       for exp in self.exp_branches:
292 …self.printx("<branchLink id=\"" + self.network.escape_html(exp.full_url) + "\" link=\"" + self.net…
293         self.printx("  <branchText>" + i_caption + "</branchText>")
294         self.printx("</branchLink>")
303         self.exp_branches.append(i_place)
304 …self.printx("<branch id=\"" + self.network.escape_html(i_place.full_url) + "\" link=\"" + self.net…
305         self.printx("  <branchText>" + i_caption + "</branchText>")
306         freeze_indent = self.str_indent
308         for lnk in self.network.links:
310             self.export_item(lnk.place_to, lnk.caption, i_indent + 1, i_place.full_url, i_root_url)
312         self.str_indent = freeze_indent
313         self.printx("</branch>")            
315         self.printx("<leaf link=\"" + self.network.escape_html(i_place.full_url) + "\">")
316         self.printx("  <leafText>" + i_caption + "</leafText>")
317         self.printx("</leaf>")
320     def export_hyperlinks(self):  argument
322       self.exp_branches = []
323       self.printx("""<?xml version='1.0' encoding='windows-1250'?>
327       self.export_item(self.network.links[0].place_from, self.network.links[0].caption, 1)
328       self.str_indent = ''
329       self.printx("</tree>")
332     def get_next_url(self):  argument
334       start_cnt = self.page_counter
335       for pg in self.network.places[start_cnt:]:
336         self.page_counter = self.page_counter + 1        
338         if not self.network.has_page_from_role(pg):          
345     def printx(self, i_str):      argument
347       print self.str_indent + i_str