Cutlery/calibre
1
0
Fork 0
mirror of https://github.com/kovidgoyal/calibre.git synced 2025-07-09 03:04:10 -04:00
Code Issues Packages Projects Releases Wiki Activity

cleaning RTF footnote

Browse Source
This commit is contained in:
Sengian 2011-01-08 08:35:59 +01:00
parent 182d0915b9
commit 3356fe6a8b
1 changed files with 47 additions and 37 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

84
src/calibre/ebooks/rtf2xml/footnote.py
View File

@ -16,7 +16,9 @@
# # # #
######################################################################### #########################################################################
import os, tempfile import os, tempfile
from calibre.ebooks.rtf2xml import copy from calibre.ebooks.rtf2xml import copy
class Footnote: class Footnote:
""" """
Two public methods are available. The first separates all of the Two public methods are available. The first separates all of the
@ -35,6 +37,7 @@ class Footnote:
self.__copy = copy self.__copy = copy
self.__write_to = tempfile.mktemp() self.__write_to = tempfile.mktemp()
self.__found_a_footnote = 0 self.__found_a_footnote = 0
def __first_line_func(self, line): def __first_line_func(self, line):
""" """
Print the tag info for footnotes. Check whether footnote is an Print the tag info for footnotes. Check whether footnote is an
@ -47,6 +50,7 @@ class Footnote:
self.__write_to_foot_obj.write( self.__write_to_foot_obj.write(
'mi<tg<open-att__<footnote<num>%s\n' % self.__footnote_count) 'mi<tg<open-att__<footnote<num>%s\n' % self.__footnote_count)
self.__first_line = 0 self.__first_line = 0
def __in_footnote_func(self, line): def __in_footnote_func(self, line):
"""Handle all tokens that are part of footnote""" """Handle all tokens that are part of footnote"""
if self.__first_line: if self.__first_line:
@ -68,6 +72,7 @@ class Footnote:
'mi<mk<footnt-clo\n') 'mi<mk<footnt-clo\n')
else: else:
self.__write_to_foot_obj.write(line) self.__write_to_foot_obj.write(line)
def __found_footnote(self, line): def __found_footnote(self, line):
""" Found a footnote""" """ Found a footnote"""
self.__found_a_footnote = 1 self.__found_a_footnote = 1
@ -81,6 +86,7 @@ class Footnote:
'mi<mk<footnt-ind<%04d\n' % self.__footnote_count) 'mi<mk<footnt-ind<%04d\n' % self.__footnote_count)
self.__write_to_foot_obj.write( self.__write_to_foot_obj.write(
'mi<mk<footnt-ope<%04d\n' % self.__footnote_count) 'mi<mk<footnt-ope<%04d\n' % self.__footnote_count)
def __default_sep(self, line): def __default_sep(self, line):
"""Handle all tokens that are not footnote tokens""" """Handle all tokens that are not footnote tokens"""
if self.__token_info == 'cw<nt<footnote__': if self.__token_info == 'cw<nt<footnote__':
@ -91,6 +97,7 @@ class Footnote:
self.__write_obj.write( self.__write_obj.write(
'tx<nu<__________<%s\n' % num 'tx<nu<__________<%s\n' % num
) )
def __initiate_sep_values(self): def __initiate_sep_values(self):
""" """
initiate counters for separate_footnotes method. initiate counters for separate_footnotes method.
@ -102,6 +109,7 @@ class Footnote:
self.__in_footnote = 0 self.__in_footnote = 0
self.__first_line = 0 #have not processed the first line of footnote self.__first_line = 0 #have not processed the first line of footnote
self.__footnote_count = 0 self.__footnote_count = 0
def separate_footnotes(self): def separate_footnotes(self):
""" """
Separate all the footnotes in an RTF file and put them at the bottom, Separate all the footnotes in an RTF file and put them at the bottom,
@ -111,58 +119,55 @@ class Footnote:
bottom of the main file. bottom of the main file.
""" """
self.__initiate_sep_values() self.__initiate_sep_values()
read_obj = open(self.__file)
self.__write_obj = open(self.__write_to, 'w') self.__write_obj = open(self.__write_to, 'w')
self.__footnote_holder = tempfile.mktemp() with open(self.__file) as read_obj:
self.__write_to_foot_obj = open(self.__footnote_holder, 'w') self.__footnote_holder = tempfile.mktemp()
line_to_read = 1 self.__write_to_foot_obj = open(self.__footnote_holder, 'w')
while line_to_read: line_to_read = 1
line_to_read = read_obj.readline() while line_to_read:
line = line_to_read line_to_read = read_obj.readline()
self.__token_info = line[:16] line = line_to_read
# keep track of opening and closing brackets self.__token_info = line[:16]
if self.__token_info == 'ob<nu<open-brack': # keep track of opening and closing brackets
self.__ob_count = line[-5:-1] if self.__token_info == 'ob<nu<open-brack':
if self.__token_info == 'cb<nu<clos-brack': self.__ob_count = line[-5:-1]
self.__cb_count = line[-5:-1] if self.__token_info == 'cb<nu<clos-brack':
# In the middle of footnote text self.__cb_count = line[-5:-1]
if self.__in_footnote: # In the middle of footnote text
self.__in_footnote_func(line) if self.__in_footnote:
# not in the middle of footnote text self.__in_footnote_func(line)
else: # not in the middle of footnote text
self.__default_sep(line) else:
self.__default_sep(line)
self.__write_obj.close() self.__write_obj.close()
read_obj.close()
self.__write_to_foot_obj.close() self.__write_to_foot_obj.close()
read_obj = open(self.__footnote_holder, 'r') with open(self.__footnote_holder, 'r') as read_obj,
write_obj = open(self.__write_to, 'a') open(self.__write_to, 'a') as write_obj:
write_obj.write( write_obj.write(
'mi<mk<sect-close\n' 'mi<mk<sect-close\n'
'mi<mk<body-close\n' 'mi<mk<body-close\n'
'mi<tg<close_____<section\n' 'mi<tg<close_____<section\n'
'mi<tg<close_____<body\n' 'mi<tg<close_____<body\n'
'mi<tg<close_____<doc\n' 'mi<tg<close_____<doc\n'
'mi<mk<footnt-beg\n') 'mi<mk<footnt-beg\n')
line = 1 for line in read_obj:
while line: write_obj.write(line)
line = read_obj.readline() write_obj.write(
write_obj.write(line) 'mi<mk<footnt-end\n')
write_obj.write(
'mi<mk<footnt-end\n')
read_obj.close()
write_obj.close()
os.remove(self.__footnote_holder) os.remove(self.__footnote_holder)
copy_obj = copy.Copy(bug_handler = self.__bug_handler) copy_obj = copy.Copy(bug_handler = self.__bug_handler)
if self.__copy: if self.__copy:
copy_obj.copy_file(self.__write_to, "footnote_separate.data") copy_obj.copy_file(self.__write_to, "footnote_separate.data")
copy_obj.rename(self.__write_to, self.__file) copy_obj.rename(self.__write_to, self.__file)
os.remove(self.__write_to) os.remove(self.__write_to)
def update_info(self, file, copy): def update_info(self, file, copy):
""" """
Unused method Unused method
""" """
self.__file = file self.__file = file
self.__copy = copy self.__copy = copy
def __get_foot_body_func(self, line): def __get_foot_body_func(self, line):
""" """
Process lines in main body and look for beginning of footnotes. Process lines in main body and look for beginning of footnotes.
@ -172,6 +177,7 @@ class Footnote:
self.__state = 'foot' self.__state = 'foot'
else: else:
self.__write_obj.write(line) self.__write_obj.write(line)
def __get_foot_foot_func(self, line): def __get_foot_foot_func(self, line):
""" """
Copy footnotes from bottom of file to a separate, temporary file. Copy footnotes from bottom of file to a separate, temporary file.
@ -180,6 +186,7 @@ class Footnote:
self.__state = 'body' self.__state = 'body'
else: else:
self.__write_to_foot_obj.write(line) self.__write_to_foot_obj.write(line)
def __get_footnotes(self): def __get_footnotes(self):
""" """
Private method to remove footnotes from main file. Read one line from Private method to remove footnotes from main file. Read one line from
@ -203,6 +210,7 @@ class Footnote:
read_obj.close() read_obj.close()
self.__write_obj.close() self.__write_obj.close()
self.__write_to_foot_obj.close() self.__write_to_foot_obj.close()
def __get_foot_from_temp(self, num): def __get_foot_from_temp(self, num):
""" """
Private method for joining footnotes to body. This method reads from Private method for joining footnotes to body. This method reads from
@ -223,6 +231,7 @@ class Footnote:
else: else:
if line == look_for: if line == look_for:
found_foot = 1 found_foot = 1
def __join_from_temp(self): def __join_from_temp(self):
""" """
Private method for rejoining footnotes to body. Read from the Private method for rejoining footnotes to body. Read from the
@ -242,6 +251,7 @@ class Footnote:
line = self.__get_foot_from_temp(line[17:-1]) line = self.__get_foot_from_temp(line[17:-1])
self.__write_obj.write(line) self.__write_obj.write(line)
read_obj.close() read_obj.close()
def join_footnotes(self): def join_footnotes(self):
""" """
Join the footnotes from the bottom of the file and put them in their Join the footnotes from the bottom of the file and put them in their