| author | amit@thunder |
| Wed, 10 Mar 2010 00:04:25 +0530 | |
| changeset 44 | d0e9b52bda73 |
| parent 41 | e54725be4df6 |
| child 45 | b5bff924ef69 |
| permissions | -rw-r--r-- |
|
26
1846ab4ebdda
Bug fixes and added a script for changes in final html
amit@thunder
parents:
diff
changeset
|
1 |
import glob |
|
39
bc65d8802897
Bug fix in myrst .. so that it does not fail at not finding a file
amit@thunder
parents:
32
diff
changeset
|
2 |
#import lxml |
|
26
1846ab4ebdda
Bug fixes and added a script for changes in final html
amit@thunder
parents:
diff
changeset
|
3 |
import re |
|
1846ab4ebdda
Bug fixes and added a script for changes in final html
amit@thunder
parents:
diff
changeset
|
4 |
import os |
|
44
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
5 |
from BeautifulSoup import BeautifulSoup ,NavigableString |
|
26
1846ab4ebdda
Bug fixes and added a script for changes in final html
amit@thunder
parents:
diff
changeset
|
6 |
import time |
|
41
e54725be4df6
Changed paths dependent on repo location to be taken from the script also changed how the soup is printed
amit@thunder
parents:
40
diff
changeset
|
7 |
import sys |
|
44
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
8 |
import xml.etree.ElementTree as ET |
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
9 |
import xml |
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
10 |
|
|
41
e54725be4df6
Changed paths dependent on repo location to be taken from the script also changed how the soup is printed
amit@thunder
parents:
40
diff
changeset
|
11 |
repo='/home/hg/repos/SEES-hacks/temp/' |
|
44
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
12 |
#repo='/home/amit/testdocbook2/' |
|
26
1846ab4ebdda
Bug fixes and added a script for changes in final html
amit@thunder
parents:
diff
changeset
|
13 |
|
|
44
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
14 |
def sort_doubledigit(chapter_names): |
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
15 |
for item in chapter_names: |
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
16 |
reg_obj=re.compile(os.path.join(repo,'ch1[0-9].*.html')) |
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
17 |
if (reg_obj.match(item)): |
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
18 |
item_tmp=item |
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
19 |
chapter_names.remove(item) |
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
20 |
chapter_names.append(item_tmp) |
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
21 |
return chapter_names |
|
26
1846ab4ebdda
Bug fixes and added a script for changes in final html
amit@thunder
parents:
diff
changeset
|
22 |
|
|
1846ab4ebdda
Bug fixes and added a script for changes in final html
amit@thunder
parents:
diff
changeset
|
23 |
def finalchanges(file_name,html_string): |
|
1846ab4ebdda
Bug fixes and added a script for changes in final html
amit@thunder
parents:
diff
changeset
|
24 |
"""some of the final changes that need to do be done on the html before creating the final usable page in the hgbook project""" |
|
1846ab4ebdda
Bug fixes and added a script for changes in final html
amit@thunder
parents:
diff
changeset
|
25 |
# print html_string |
|
1846ab4ebdda
Bug fixes and added a script for changes in final html
amit@thunder
parents:
diff
changeset
|
26 |
replace_string="""<head><meta http-equiv="Content-Type" content="text/html; charset=utf-8"><title>Chapter 2. Basic Python</title><link rel="stylesheet" href="/review/support/styles.css" type="text/css"><meta name="generator" content="DocBook XSL Stylesheets V1.74.3"><link rel="shortcut icon" type="image/png" href="/review/support/figs/favicon.png"><script type="text/javascript" src="/review/support/jquery-min.js"></script><script type="text/javascript" src="/review/support/form.js"></script><script type="text/javascript" src="/review/support/hsbook.js"></script></head>""" |
|
44
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
27 |
ch_name=os.path.split(file_name)[1].split('.')[0]
|
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
28 |
chapter_names_unsorted=glob.glob(os.path.join(repo,'ch*.html')) |
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
29 |
chapter_names_unsorted.sort() |
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
30 |
chapter_names_sorted=chapter_names_unsorted |
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
31 |
# print chapter_names_sorted |
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
32 |
chapter_names_sorted=sort_doubledigit(chapter_names_sorted) |
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
33 |
chapter_names=chapter_names_sorted |
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
34 |
previous_string='<<<' |
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
35 |
next_string='>>>' |
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
36 |
html_src_folder="review/html/" |
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
37 |
current_chapter_index=chapter_names.index(file_name) |
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
38 |
|
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
39 |
current_chapter=os.path.join(html_src_folder,chapter_names[current_chapter_index].split('/')[-1])
|
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
40 |
if (current_chapter_index-1>0): |
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
41 |
previous_chapter=os.path.join(html_src_folder,chapter_names[current_chapter_index-1].split('/')[-1])
|
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
42 |
else: |
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
43 |
previous_chapter='' |
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
44 |
previous_string='' |
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
45 |
try : |
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
46 |
next_chapter=os.path.join(html_src_folder,chapter_names[current_chapter_index+1].split('/')[-1])
|
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
47 |
except: |
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
48 |
next_string='' |
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
49 |
next_chapter='' |
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
50 |
|
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
51 |
ch_name_tmp=file_name.split('.')[0]
|
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
52 |
chapter_xml=ch_name_tmp+'.xml' |
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
53 |
|
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
54 |
|
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
55 |
try: |
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
56 |
xml_file =open(chapter_xml,'r').read() |
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
57 |
xml_tree=ET.fromstring(xml_file) |
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
58 |
try: |
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
59 |
title_tag=xml_tree.find('title')
|
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
60 |
current_chapter_title=title_tag.text |
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
61 |
except: |
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
62 |
section=xml_tree.getchildren()[0] |
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
63 |
title_tag=section.find('title')
|
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
64 |
current_chapter_title=title_tag.text |
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
65 |
|
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
66 |
print current_chapter_title |
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
67 |
|
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
68 |
# soup.html.body.insert(0,NavigableString(body_add_string)) |
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
69 |
|
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
70 |
|
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
71 |
except : |
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
72 |
ch_title=re.split('[0-9]*',ch_name)[1]
|
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
73 |
title_string='Chapter. '+ ch_title |
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
74 |
current_chapter_title=title_string |
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
75 |
|
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
76 |
|
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
77 |
|
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
78 |
body_add_string="""<div><table width="100%%" summary="Navigation header"><tr><th colspan="3" align="center">Chapter %s</th></tr><tr><td width="20%%" align="left"><a accesskey="p" href="%s">%s</a></td><th width="60%%" align="center"> </th><td width="20%%" align="right"> <a accesskey="n" href="%s">%s</a></td></tr></table></div>"""%(current_chapter_title,previous_chapter,previous_string,next_chapter,next_string) |
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
79 |
|
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
80 |
|
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
81 |
|
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
82 |
|
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
83 |
|
|
26
1846ab4ebdda
Bug fixes and added a script for changes in final html
amit@thunder
parents:
diff
changeset
|
84 |
reg_obj=re.compile('<head>.*</head>',re.DOTALL)
|
|
1846ab4ebdda
Bug fixes and added a script for changes in final html
amit@thunder
parents:
diff
changeset
|
85 |
html_string=reg_obj.sub(replace_string, html_string,re.DOTALL) |
|
1846ab4ebdda
Bug fixes and added a script for changes in final html
amit@thunder
parents:
diff
changeset
|
86 |
html_string=re.sub('><a name',' id', html_string)
|
|
1846ab4ebdda
Bug fixes and added a script for changes in final html
amit@thunder
parents:
diff
changeset
|
87 |
soup=BeautifulSoup(html_string.decode('ascii','ignore'))
|
|
44
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
88 |
soup.html.head.title.string.replaceWith(current_chapter_title) |
|
26
1846ab4ebdda
Bug fixes and added a script for changes in final html
amit@thunder
parents:
diff
changeset
|
89 |
div=soup.html.div |
|
1846ab4ebdda
Bug fixes and added a script for changes in final html
amit@thunder
parents:
diff
changeset
|
90 |
|
|
1846ab4ebdda
Bug fixes and added a script for changes in final html
amit@thunder
parents:
diff
changeset
|
91 |
try: |
|
1846ab4ebdda
Bug fixes and added a script for changes in final html
amit@thunder
parents:
diff
changeset
|
92 |
del(div['title']) |
|
1846ab4ebdda
Bug fixes and added a script for changes in final html
amit@thunder
parents:
diff
changeset
|
93 |
div['id'] = ch_name |
|
1846ab4ebdda
Bug fixes and added a script for changes in final html
amit@thunder
parents:
diff
changeset
|
94 |
except TypeError: |
|
1846ab4ebdda
Bug fixes and added a script for changes in final html
amit@thunder
parents:
diff
changeset
|
95 |
print file_name |
|
44
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
96 |
|
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
97 |
soup.html.body.insert(0,NavigableString(body_add_string)) |
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
98 |
|
|
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
99 |
|
|
26
1846ab4ebdda
Bug fixes and added a script for changes in final html
amit@thunder
parents:
diff
changeset
|
100 |
return soup |
|
1846ab4ebdda
Bug fixes and added a script for changes in final html
amit@thunder
parents:
diff
changeset
|
101 |
|
|
1846ab4ebdda
Bug fixes and added a script for changes in final html
amit@thunder
parents:
diff
changeset
|
102 |
if __name__=='__main__': |
|
41
e54725be4df6
Changed paths dependent on repo location to be taken from the script also changed how the soup is printed
amit@thunder
parents:
40
diff
changeset
|
103 |
file_names=glob.glob(os.path.join(repo,'ch*.html')) |
|
26
1846ab4ebdda
Bug fixes and added a script for changes in final html
amit@thunder
parents:
diff
changeset
|
104 |
for file_name in file_names: |
|
1846ab4ebdda
Bug fixes and added a script for changes in final html
amit@thunder
parents:
diff
changeset
|
105 |
file_obj=open(file_name,'r') |
|
1846ab4ebdda
Bug fixes and added a script for changes in final html
amit@thunder
parents:
diff
changeset
|
106 |
soup=finalchanges(file_name,file_obj.read()) |
|
1846ab4ebdda
Bug fixes and added a script for changes in final html
amit@thunder
parents:
diff
changeset
|
107 |
time.sleep(1) |
|
1846ab4ebdda
Bug fixes and added a script for changes in final html
amit@thunder
parents:
diff
changeset
|
108 |
file_obj=open(file_name,'w') |
|
40
ef147a79b098
Added change names to do required changes in names such that we get the chapter names beyond 10 in proper list
amit@thunder
parents:
39
diff
changeset
|
109 |
print >>file_obj ,soup |
|
44
d0e9b52bda73
Changed the algorithm for getting the titles ... Also added the ability to navigate to the next chapters
amit@thunder
parents:
41
diff
changeset
|
110 |
print file_name |
|
26
1846ab4ebdda
Bug fixes and added a script for changes in final html
amit@thunder
parents:
diff
changeset
|
111 |
|
|
1846ab4ebdda
Bug fixes and added a script for changes in final html
amit@thunder
parents:
diff
changeset
|
112 |
|
|
1846ab4ebdda
Bug fixes and added a script for changes in final html
amit@thunder
parents:
diff
changeset
|
113 |