Modify mwlib.rl to optionally include article headings as TOC entries

mwlib/rl/_version.py

@@ -1,2 +1,2 @@
-version = __version__ = "0.14.6"
-__version_info__ = (0, 14, 6)
+version = __version__ = "0.14.6.1"
+__version_info__ = (0, 14, 6, 1)

mwlib/rl/pagetemplates.py

@@ -202,9 +202,10 @@ def _startBuild(self, filename=None, canvasmaker=canvas.Canvas):
 
         type2lvl = {'chapter': 0,
                     'article': 1,
-                    'heading2': 2,
-                    'heading3': 3,
-                    'heading4': 4,
+                    'heading1': 2,
+                    'heading2': 3,
+                    'heading3': 4,
+                    'heading4': 5,
                     }
         got_chapter = False
         last_lvl =  0

mwlib/rl/pdfstyles.py

@@ -62,7 +62,7 @@
 
 show_title_page = True
 show_title_page_footer = True
-show_page_header = True 
+show_page_header = True
 show_page_footer = True
 page_break_after_article = False
 
@@ -100,9 +100,13 @@
 # note that no TOC is generated if only one article is rendered
 render_toc = True
 
+# indicates the heading levels that should be included in the toc
+# [0] = Chapter, [1] = Article, [2-5] = Heading 1-4
+render_toc_levels = (1,1,0,1,1,1)
+
 ### TABLE CONFIG
 
-tableOverflowTolerance = 20  # max width overflow for tables    unit: pt 
+tableOverflowTolerance = 20  # max width overflow for tables    unit: pt
 cell_padding = 3
 min_rows_for_break = 3 # page breaks before tables are only forced if more than n rows are present
 
@@ -198,7 +202,7 @@ def __init__(self, name, parent=None, **kw):
         self.backColor = None
         self.wordWrap = None
         self.textTransform = None
-        
+
 def text_style(mode='p', indent_lvl=0, in_table=0, relsize='normal', text_align=None):
     """
     mode: p (normal paragraph), blockquote, center (centered paragraph), footer, figure (figure caption text),
@@ -252,7 +256,7 @@ def text_style(mode='p', indent_lvl=0, in_table=0, relsize='normal', text_align=
 
     if mode == 'img_attribution':
         style.spaceBefore = 2
-        
+
     if mode in ['articlefoot', 'references']:
         style.fontSize=small_font_size
         style.leading=small_leading
@@ -263,11 +267,11 @@ def text_style(mode='p', indent_lvl=0, in_table=0, relsize='normal', text_align=
         style.borderPadding = 3 # borderPadding is not calculated onto the box dimensions.
         style.spaceBefore = 6 # therefore spaceBefore = 3 + borderPadding
         style.spaceAfter = 9 # add an extra 3 to spaceAfter, b/c spacing seems to small otherwise
-    
+
     if mode == 'source' or mode == 'preformatted':
-        style.fontName = mono_font   
+        style.fontName = mono_font
         style.flowable = False
-        
+
     if mode == 'list' or mode == 'references':
         style.spaceBefore = 0
         style.bulletIndent = list_left_indent * max(0, indent_lvl-1)
@@ -281,7 +285,7 @@ def text_style(mode='p', indent_lvl=0, in_table=0, relsize='normal', text_align=
         style.spaceBefore = 16
         style.fontName= sans_font
         style.alignment = TA_LEFT
-        
+
     if mode == 'booksubtitle':
         style.fontSize = 24
         style.leading = 30
@@ -313,16 +317,36 @@ def text_style(mode='p', indent_lvl=0, in_table=0, relsize='normal', text_align=
     if mode == 'toc_group':
         style.fontSize = 18
         style.leading = 22
-       
+
     if mode == 'toc_chapter':
         style.fontSize = 14
         style.leading = 18
-        
+
     if mode == 'toc_article':
         style.fontSize = 10
         style.leading = 12
         style.leftIndent = para_left_indent
 
+    if mode == 'toc_heading1':
+        style.fontSize = 10
+        style.leading = 12
+        style.leftIndent = para_left_indent * sum(render_toc_levels[1:3])
+
+    if mode == 'toc_heading2':
+        style.fontSize = 10
+        style.leading = 12
+        style.leftIndent = para_left_indent * sum(render_toc_levels[1:4])
+
+    if mode == 'toc_heading3':
+        style.fontSize = 10
+        style.leading = 12
+        style.leftIndent = para_left_indent * sum(render_toc_levels[1:5])
+
+    if mode == 'toc_heading4':
+        style.fontSize = 10
+        style.leading = 12
+        style.leftIndent = para_left_indent * sum(render_toc_levels[1:6])
+
     return style
 
 table_style = {'spaceBefore': 0.25*cm,
@@ -340,7 +364,7 @@ def __init__(self, name, parent=None, **kw):
         self.leftIndent = 0
         self.rightIndent = 0
         self.firstLineIndent = 0
-        self.alignment = TA_LEFT        
+        self.alignment = TA_LEFT
         self.spaceBefore = 12
         self.spaceAfter = 6
         self.bulletFontName = serif_font
@@ -352,7 +376,7 @@ def __init__(self, name, parent=None, **kw):
         self.textTransform = None
         #self.allowWidows = 0
         #self.allowOrphans = 0
-        
+
 def heading_style(mode='chapter', lvl=1, text_align=None):
 
     style = BaseHeadingStyle(name='heading_style_%s_%d' % (mode, lvl))
@@ -372,7 +396,7 @@ def heading_style(mode='chapter', lvl=1, text_align=None):
         style.spaceBefore = 20
         style.spaceAfter = 2
     elif mode == 'section':
-        lvl = max(min(5,lvl), 1)  
+        lvl = max(min(5,lvl), 1)
         style.fontSize = 18 - (lvl - 1) * 2
         style.leading = style.fontSize + max(2, min(int(style.fontSize / 5), 3)) # magic: increase in leading is between 2 and 3 depending on fontsize...
         style.spaceBefore = min(style.leading, 20)
@@ -398,10 +422,10 @@ def heading_style(mode='chapter', lvl=1, text_align=None):
         style.alignment = TA_RIGHT
     elif text_align == 'justify':
         style.alignment = TA_JUSTIFY
-    
+
     style.prevent_post_pagebreak = True
     return style
-    
+
 
 # import custom configuration to override configuration values
 # if doing so, you need to be careful not to break things...

mwlib/rl/rlwriter.py

@@ -658,7 +658,8 @@ def writeChapter(self, chapter):
 
         elements.append(self._getPageTemplate(''))
         elements.extend([NotAtTopPageBreak(), hr, chapter_para, hr])
-        elements.append(TocEntry(txt=title, lvl='chapter'))
+        if pdfstyles.render_toc_levels[0]==1:
+            elements.append(TocEntry(txt=title, lvl='chapter'))
         elements.append(self._getPageTemplate(chapter.next_article_title))
         elements.extend(self.renderChildren(chapter))
 
@@ -680,14 +681,23 @@ def writeSection(self, obj):
             heading_txt = ''
         self.formatter.sectiontitle_mode = False
 
-        if 1 <= lvl <= 4 and self.inline_mode == 0 and self.table_nesting==0:
+        try:
+            render_toc_level = (pdfstyles.render_toc_levels[lvl+1]==1)
+        except IndexError:
+            render_toc_level = False
+
+
+        if 1 <= lvl <= 5 and self.inline_mode == 0 and self.table_nesting==0 and render_toc_level is True:
             anchor = '<a name="%d"/>' % len(self.bookmarks)
             bm_type = 'article' if lvl==1 else 'heading%s' % lvl
             self.bookmarks.append((obj.children[0].getAllDisplayText(), bm_type))
         else:
             anchor = ''
         elements = [Paragraph('<font name="%s"><b>%s</b></font>%s' % (headingStyle.fontName, heading_txt, anchor), headingStyle)]
 
+        if 2 <= lvl <= 5 and render_toc_level is True:
+            elements.append(TocEntry(txt=heading_txt, lvl='heading%s' % lvl))
+
         if self.table_size_calc == 0:
             obj.removeChild(obj.children[0])
         elements.extend(self.renderMixed(obj))
@@ -727,7 +737,8 @@ def writeArticleMetainfo(self):
         elements = []
         title = self.formatter.cleanText(_('Article Sources and Contributors'))
         elements.append(Paragraph('<b>%s</b>' % title, heading_style(mode='article')))
-        elements.append(TocEntry(txt=title, lvl='article'))
+        if pdfstyles.render_toc_levels[1]==1:
+            elements.append(TocEntry(txt=title, lvl='article'))
         for title, url, authors in self.article_meta_info:
             authors_text = self._filterAnonIpEdits(authors)
             txt = '<b>%(title)s</b> &nbsp;<i>%(source_label)s</i>: %(source)s &nbsp;<i>%(contribs_label)s</i>: %(contribs)s ' % {
@@ -746,7 +757,8 @@ def writeImageMetainfo(self):
         elements = []
         title = self.formatter.cleanText(_('Image Sources, Licenses and Contributors'))
         elements.append(Paragraph('<b>%s</b>' % title, heading_style(mode='article')))
-        elements.append(TocEntry(txt=title, lvl='article'))
+        if pdfstyles.render_toc_levels[1]==1:
+            elements.append(TocEntry(txt=title, lvl='article'))
         for _id, title, url, license, authors in sorted(self.img_meta_info.values()):
             authors_text = self._filterAnonIpEdits(authors)
             if not license:
@@ -830,7 +842,8 @@ def writeArticle(self, article):
 
         heading_para = Paragraph('<b>%s</b>%s' % (title, heading_anchor), heading_style("article"))
         elements.append(heading_para)
-        elements.append(TocEntry(txt=title, lvl='article'))
+        if pdfstyles.render_toc_levels[1]==1:
+            elements.append(TocEntry(txt=title or article.caption, lvl='article'))
 
         if pdfstyles.show_article_hr:
             elements.append(HRFlowable(width='100%', hAlign='LEFT', thickness=1, spaceBefore=0, spaceAfter=10, color=colors.black))

mwlib/rl/toc.py

@@ -26,7 +26,7 @@ def __init__(self):
         font_switcher.registerFontDefinitionList(fontconfig.fonts)
         font_switcher.registerReportlabFonts(fontconfig.fonts)
 
-        
+
     def build(self, pdfpath, toc_entries, has_title_page=False, rtl=False):
         outpath = os.path.dirname(pdfpath)
         tocpath = os.path.join(outpath, 'toc.pdf')
@@ -35,11 +35,11 @@ def build(self, pdfpath, toc_entries, has_title_page=False, rtl=False):
         return self.combinePdfs(pdfpath, tocpath, finalpath, has_title_page)
 
     def _getColWidths(self):
-        p = Paragraph('<b>%d</b>' % 9999, pdfstyles.text_style(mode='toc_article', text_align='right'))        
+        p = Paragraph('<b>%d</b>' % 9999, pdfstyles.text_style(mode='toc_article', text_align='right'))
         w, h = p.wrap(0, pdfstyles.print_height)
         # subtracting 30pt below is *probably* necessary b/c of the table margins
         return [pdfstyles.print_width - w - 30, w]
-    
+
     def renderToc(self, tocpath, toc_entries, rtl):
         doc = SimpleDocTemplate(tocpath, pagesize=(pdfstyles.page_width, pdfstyles.page_height))
         elements = []
@@ -48,7 +48,7 @@ def renderToc(self, tocpath, toc_entries, rtl):
         styles = []
         col_widths = self._getColWidths()
         for row_idx, (lvl, txt, page_num) in enumerate(toc_entries):
-            if lvl == 'article':
+            if lvl == 'article' or lvl.startswith('heading'):
                 page_num = str(page_num)
             elif lvl == 'chapter':
                 page_num = '<b>%d</b>' % page_num