Correction of hard line breaking bug in RTF (introduce a <br/> for \line)

resources/templates/rtf.xsl

@@ -413,6 +413,10 @@
         </xsl:element>
     </xsl:template>
     
+    <xsl:template match="rtf:hardline-break">
+        <xsl:element name="br"/>
+    </xsl:template>
+
     <xsl:template match="rtf:rtf-definition|rtf:font-table|rtf:color-table|rtf:style-table|rtf:page-definition|rtf:list-table|rtf:override-table|rtf:override-list|rtf:list-text"/>
 
     <xsl:template match="rtf:body">

src/calibre/ebooks/rtf2xml/ParseRtf.py

@@ -90,7 +90,7 @@ class ParseRtf:
                 out_file = '',
                 out_dir = None,
                 dtd = '',
-                debug = 0,
+                #debug = 0, #why? calibre
                 deb_dir = None,
                 convert_symbol = None,
                 convert_wingdings = None,
@@ -132,7 +132,7 @@ class ParseRtf:
         self.__dtd_path = dtd
         self.__check_file(in_file,"file_to_parse")
         self.__char_data = char_data
-        self.__debug_dir = debug
+        self.__debug_dir = deb_dir #self.__debug_dir = debug calibre
         self.__check_dir(self.__temp_dir)
         self.__copy = self.__check_dir(self.__debug_dir)
         self.__convert_caps = convert_caps

src/calibre/ebooks/rtf2xml/inline.py

@@ -51,6 +51,7 @@ class Inline:
             'tx<ut<__________'  :       self.__found_text_func,
             'mi<mk<inline-fld'  :       self.__found_text_func,
             'text'              :       self.__found_text_func,
+            'cw<nu<hard-lineb'  :       self.__found_text_func, #calibre
             'cb<nu<clos-brack'  :       self.__close_bracket_func,
             'mi<mk<par-end___'  :       self.__end_para_func,
             'mi<mk<footnt-ope'  :       self.__end_para_func,
@@ -62,6 +63,7 @@ class Inline:
             'tx<hx<__________'  :       self.__found_text_func,
             'tx<ut<__________'  :       self.__found_text_func,
             'text'              :       self.__found_text_func,
+            'cw<nu<hard-lineb'  :       self.__found_text_func, #calibre
             'mi<mk<inline-fld'  :       self.__found_text_func,
             'ob<nu<open-brack':         self.__found_open_bracket_func,
             'mi<mk<par-end___'  :       self.__end_para_func,
@@ -133,10 +135,12 @@ class Inline:
         Returns:
             nothing
         Logic:
+            Write if not hardline break
         """
         action = self.__default_dict.get(self.__token_info)
         if action:
             action(line)
+        if self.__token_info != 'cw<nu<hard-lineb': #calibre
             self.__write_obj.write(line)
     def __found_open_bracket_func(self, line):
         """
@@ -164,7 +168,7 @@ class Inline:
             Use the dictionary to get the approriate function.
             Always print out the line.
         """
-        if line[0:2] == 'cw':
+        if line[0:5] == 'cw<ci': #calibre: bug in original function no diff between cw<ci and cw<pf
             self.__handle_control_word(line)
         else:
             action = self.__after_open_bracket_dict.get(self.__token_info)
@@ -247,12 +251,13 @@ class Inline:
         Return:
             nothing
         Logic:
-            Two cases:
+            Three cases:
             1. in a list. Simply write inline
             2. Not in a list
                 Text can mark the start of a paragraph.
                 If already in a paragraph, check to see if any groups are waiting
                 to be added. If so, use another method to write these groups.
+            3. If not check if hardline break, then write
         """
         if self.__place == 'in_list':
             self.__write_inline()
@@ -261,8 +266,11 @@ class Inline:
                 self.__in_para = 1
                 self.__start_para_func(line)
             else:
+                if self.__token_info == 'cw<nu<hard-lineb': #calibre
+                    self.__write_obj.write('mi<tg<empty_____<hardline-break\n')
                 if self.__groups_in_waiting[0] != 0:
                     self.__write_inline()
+                
     def __write_inline(self):
         """
         Required:
@@ -279,7 +287,7 @@ class Inline:
             Get the keys in each dictionary. If 'font-style' is in the keys,
             write a marker tag. (I will use this marker tag later when conerting
             hext text to utf8.)
-            Write a tag for the inline vaues.
+            Write a tag for the inline values.
         """
         if self.__groups_in_waiting[0] != 0:
             last_index = -1 * self.__groups_in_waiting[0]

src/calibre/ebooks/rtf2xml/process_tokens.py

@@ -73,7 +73,8 @@ class ProcessTokens:
         'backslash'          :	('nu', '\\', self.text_func),
         'ob'                 :	('nu', '{', self.text_func),
         'cb'                 :	('nu', '}', self.text_func),
-        'line'               :  ('nu', ' ', self.text_func),
+        'line'               :  ('nu', 'hard-lineb', self.default_func), #calibre
+        #'line'               :  ('nu', ' ', self.text_func), calibre
         # paragraph formatting => pf
         'page'               :  ('pf', 'page-break', self.default_func),
         'par'                :	('pf', 'par-end___', self.default_func),