Implement cross-references

example.md

@@ -22,7 +22,7 @@ This is the *blah blah*.
 	
 		hijkl
 	
-	Continued
+	Cross reference to `rule 1`_
 
 (2) DEF
 

example.rtf

@@ -1 +1 @@
-{\rtf1\deff0{\fonttbl{\f0 TeX Gyre Heros{\*\falt FreeSans}{\*\falt Liberation Sans}{\*\falt Arial};}}\paperw11905\paperh16837\margl1133\margr1133\margt1133\margb1133{\header\f0\fs16\tqr\tx9637 Example Legal Document\tab }{\footer\f0\fs16\tqr\tx9637 \tab\chpgn}\sa198\fs1\~\fs24 {\fs20\li566{\b Note:} This is a note.\par}{\sb566\keepn\b\fs26\qc\caps Part 1\u8212?Preliminary\par}{\keepn\b\fi-566\li566 1\tab Short title\par}{\li566 This is the {\i blah blah}.\par}{\keepn\b\fi-566\li566 2\tab Another heading\par}{\fi-566\li1133 (1)\tab ABC\par}{\li1133 Continued\par}{\fi-566\li1700 (a)\tab asdf\par}{\fi-566\li1700 (b)\tab abcdefg\par}{\li1700 hijkl\par}{\li1133 Continued\par}{\fi-566\li1133 (2)\tab DEF\par}{\keepn\b\fi-566\li566 3\tab Table example\par}{\sa0\trowd\trgaph120\trleft566\clbrdrt\brdrs\clbrdrb\brdrs\cellx2380\clbrdrt\brdrs\clbrdrb\brdrs\cellx9636 {\b Column 1\intbl\cell}{\b Column 2\intbl\cell}\row{\b This is a long column header that will overflow onto multiple lines\intbl\cell}{\b Second column\intbl\cell}\row {1. First row\intbl\cell}{Foo bar\intbl\cell}\row {2. Second row\intbl\cell}{Baz qux\intbl\cell}\row }}
+{\rtf1\deff0{\fonttbl{\f0 TeX Gyre Heros{\*\falt FreeSans}{\*\falt Liberation Sans}{\*\falt Arial};}}\paperw11905\paperh16837\margl1133\margr1133\margt1133\margb1133{\header\f0\fs16\tqr\tx9637 Example Legal Document\tab }{\footer\f0\fs16\tqr\tx9637 \tab\chpgn}\sa198\fs1\~\fs24 {\fs20\li566{\b Note:} This is a note.\par}{\sb566\keepn\b\fs26\qc\caps Part 1\u8212?Preliminary\par}{\keepn\b\fi-566\li566 1\tab Short title\par}{\li566 This is the {\i blah blah}.\par}{\keepn\b\fi-566\li566 2\tab Another heading\par}{\fi-566\li1133 (1)\tab ABC\par}{\li1133 Continued\par}{\fi-566\li1700 (a)\tab asdf\par}{\fi-566\li1700 (b)\tab abcdefg\par}{\li1700 hijkl\par}{\li1133 Cross reference to rule\u160?1\par}{\fi-566\li1133 (2)\tab DEF\par}{\keepn\b\fi-566\li566 3\tab Table example\par}{\sa0\trowd\trgaph120\trleft566\clbrdrt\brdrs\clbrdrb\brdrs\cellx2380\clbrdrt\brdrs\clbrdrb\brdrs\cellx9636 {\b Column 1\intbl\cell}{\b Column 2\intbl\cell}\row{\b This is a long column header that will overflow onto multiple lines\intbl\cell}{\b Second column\intbl\cell}\row {1. First row\intbl\cell}{Foo bar\intbl\cell}\row {2. Second row\intbl\cell}{Baz qux\intbl\cell}\row }}

legalmd/__main__.py

@@ -40,4 +40,22 @@ with renderer_cls() as renderer:
 	doc.author = rawdoc.get('author', '')
 	doc.footer = rawdoc.get('footer', '')
 	
+	doc.full_label_map = {}
+	
+	# Preprocess custom tokens
+	def walk_elem(elem, parent=None):
+		elem.doc = doc
+		elem.parent = parent
+		
+		if isinstance(elem, legal_token.NumberedHeading) or isinstance(elem, legal_token.SubrulesItem):
+			full_label = elem.full_label()
+			if full_label:
+				doc.full_label_map[full_label] = elem
+		
+		if hasattr(elem, 'children'):
+			for child in elem.children:
+				walk_elem(child, elem)
+	
+	walk_elem(doc)
+	
 	print(renderer.render(doc))

legalmd/latex_renderer.py

@@ -14,6 +14,8 @@
 #    You should have received a copy of the GNU Affero General Public License
 #    along with this program.  If not, see <https://www.gnu.org/licenses/>.
 
+import hashlib
+
 import mistletoe
 import mistletoe.latex_renderer
 
@@ -35,6 +37,8 @@ class LaTeXRenderer(mistletoe.latex_renderer.LaTeXRenderer):
 	def __init__(self, *extras):
 		super().__init__(*extras)
 		
+		self.render_map['CrossReference'] = self.render_cross_reference
+		
 		self.render_map['NumberedHeading'] = self.render_numbered_heading
 		self.render_map['Subrules'] = self.render_subrules
 		self.render_map['SubrulesItem'] = self.render_subrules_item
@@ -48,6 +52,20 @@ class LaTeXRenderer(mistletoe.latex_renderer.LaTeXRenderer):
 		result = result.replace('★★★', r'\texorpdfstring{\freeserif ★★★}{★★★}')
 		return result
 	
+	def render_cross_reference(self, token):
+		reference = token.get_reference()
+		if not reference:
+			raise Exception('Unable to resolve reference "{}"'.format(token.reference_num))
+		
+		sha = hashlib.sha256()
+		sha.update(reference.full_label().encode('utf-8'))
+		
+		return format(r'\hyperlink{<linkname>}{<reference_type><reference_num>}',
+			linkname=sha.hexdigest(),
+			reference_type=(token.reference_type + '~') if token.reference_type else '',
+			reference_num=token.reference_num
+		)
+	
 	def render_heading(self, token):
 		if token.level == 1:
 			heading_last, self.heading_last = self.heading_last, True
@@ -69,10 +87,17 @@ class LaTeXRenderer(mistletoe.latex_renderer.LaTeXRenderer):
 			)
 	
 	def render_numbered_heading(self, token):
+		sha = hashlib.sha256()
+		sha.update(token.full_label().encode('utf-8'))
+		hyperlink = format(r'\hypertarget{<linkname>}{}',
+			linkname=sha.hexdigest()
+		)
+		
 		if token.level == 1:
 			# Part
 			heading_last, self.heading_last = self.heading_last, True
-			return format(r'{\par\vspace{1cm plus 0.3cm minus 0.3cm}\bfseries\fontsize{13pt}{15pt}\selectfont\centering\uppercase{Part <label>—<content>}\phantomsection\addcontentsline{toc}{section}{Part <label>—<content>}\nopagebreak\par}',
+			return format(r'{\par\vspace{1cm plus 0.3cm minus 0.3cm}<hyperlink>\bfseries\fontsize{13pt}{15pt}\selectfont\centering\uppercase{Part <label>—<content>}\phantomsection\addcontentsline{toc}{section}{Part <label>—<content>}\nopagebreak\par}',
+				hyperlink=hyperlink,
 				label=token.label,
 				content=self.render_inner(token)
 			)
@@ -80,7 +105,8 @@ class LaTeXRenderer(mistletoe.latex_renderer.LaTeXRenderer):
 		if token.level == 2:
 			# Division
 			heading_last, self.heading_last = self.heading_last, True
-			return format(r'{\par\vspace{<space_above>}\bfseries\fontsize{13pt}{15pt}\selectfont\centering Division <label>—<content>\phantomsection\addcontentsline{toc}{subsection}{Division <label>—<content>}\nopagebreak\par}',
+			return format(r'{\par\vspace{<space_above>}<hyperlink>\bfseries\fontsize{13pt}{15pt}\selectfont\centering Division <label>—<content>\phantomsection\addcontentsline{toc}{subsection}{Division <label>—<content>}\nopagebreak\par}',
+				hyperlink=hyperlink,
 				space_above='1cm plus 0.3cm minus 0.3cm' if not heading_last else '0cm',
 				label=token.label,
 				content=self.render_inner(token)
@@ -89,7 +115,8 @@ class LaTeXRenderer(mistletoe.latex_renderer.LaTeXRenderer):
 		if token.level == 3:
 			# Section
 			heading_last, self.heading_last = self.heading_last, False
-			return format(r'{\par\leftskip=\quotemargin\bfseries\makebox[<lmarg>][l]{<label>}<content>\phantomsection\addcontentsline{toc}{subsubsection}{\protect\numberline{<label>} <content>}\nopagebreak\par}',
+			return format(r'{\par<hyperlink>\leftskip=\quotemargin\bfseries\makebox[<lmarg>][l]{<label>}<content>\phantomsection\addcontentsline{toc}{subsubsection}{\protect\numberline{<label>} <content>}\nopagebreak\par}',
+				hyperlink=hyperlink,
 				label=token.label,
 				content=self.render_inner(token)
 			)
@@ -122,7 +149,14 @@ class LaTeXRenderer(mistletoe.latex_renderer.LaTeXRenderer):
 	
 	def render_subrules_item(self, token):
 		if token.label:
-			return format(r'\N{\par\leftskip=\dimexpr\quotemargin+<lmarg>+<level>cm\relax\hangindent=1cm\parskip=<parskip>\makebox[<lmarg>][l]{<label>}<content>\par}\N',
+			sha = hashlib.sha256()
+			sha.update(token.full_label().encode('utf-8'))
+			hyperlink = format(r'\hypertarget{<linkname>}{}',
+				linkname=sha.hexdigest()
+			)
+			
+			return format(r'\N{\par<hyperlink>\leftskip=\dimexpr\quotemargin+<lmarg>+<level>cm\relax\hangindent=1cm\parskip=<parskip>\makebox[<lmarg>][l]{<label>}<content>\par}\N',
+				hyperlink=hyperlink,
 				parskip=r'\parskip',# if token.level <= 1 else '0cm',
 				label=token.label,
 				level=token.level,

legalmd/legal_token.py

@@ -46,6 +46,14 @@ class NumberedHeading(mistletoe.block_token.BlockToken):
 		
 		return level, label, content
 	
+	def full_label(self):
+		if not self.label:
+			return None
+		if not isinstance(self.parent, mistletoe.block_token.Document):
+			return None
+		
+		return self.label
+
 mistletoe.block_token.add_token(NumberedHeading)
 
 class Subrules(mistletoe.block_token.BlockToken):
@@ -116,6 +124,27 @@ class SubrulesItem(mistletoe.block_token.BlockToken):
 		
 		return level, label, children
 	
+	def full_label(self):
+		if not self.label:
+			return None
+		if not isinstance(self.parent.parent, mistletoe.block_token.Document):
+			return None
+		
+		labels = [self.label]
+		
+		# Subrules items
+		cur_level = self.level
+		for child in reversed(self.parent.children[0:self.parent.children.index(self)]):
+			if child.level < cur_level and child.label:
+				labels.append(child.label)
+				cur_level = child.level
+		
+		# Section
+		section = next(x for x in reversed(self.parent.parent.children[0:self.parent.parent.children.index(self.parent)]) if isinstance(x, NumberedHeading))
+		labels.append(section.label)
+		
+		return ''.join(reversed(labels))
+
 class Note(mistletoe.block_token.BlockToken):
 	pattern = re.compile(r'(\t*)([0-9A-Z ]+):\s+(.+)')
 	
@@ -213,3 +242,43 @@ class TableCell(mistletoe.block_token.BlockToken):
 
 mistletoe.block_token.remove_token(mistletoe.block_token.Table)
 mistletoe.block_token.add_token(Table)
+
+class CrossReference(mistletoe.span_token.SpanToken):
+	pattern = re.compile(r'`(?:([A-Za-z]+?)\s+)?([0-9A-Za-z\(\)]+?)`_')
+	pattern_parts = re.compile(r'^[0-9A-Za-z]+|\([0-9A-Za-z]+\)')
+	
+	def __init__(self, match):
+		self.reference_type = match.group(1)
+		self.reference_num = match.group(2)
+		self.reference_parts = re.findall(self.pattern_parts, self.reference_num)
+	
+	def get_reference(self):
+		parent = None
+		
+		elem = self.parent
+		while elem is not None:
+			if isinstance(elem, SubrulesItem):
+				parent = elem
+				break
+			elem = elem.parent
+		
+		if parent is None:
+			# Might be a freestanding rule
+			section = next((x for x in reversed(self.parent.parent.children[0:self.parent.parent.children.index(self.parent)]) if isinstance(x, NumberedHeading)), None)
+			if section:
+				parent = section
+		
+		if parent and parent.full_label():
+			basenum = re.findall(self.pattern_parts, parent.full_label())
+		else:
+			basenum = []
+		
+		# Try possibilities
+		for index in reversed(range(0, len(basenum) + 1)):
+			try_ref = ''.join(basenum[:index] + self.reference_parts)
+			if try_ref in self.doc.full_label_map:
+				return self.doc.full_label_map[try_ref]
+		
+		return None
+
+mistletoe.span_token.add_token(CrossReference)

legalmd/rtf_renderer.py

@@ -38,6 +38,8 @@ class RTFRenderer(mistletoe.base_renderer.BaseRenderer):
 	def __init__(self, *extras):
 		super().__init__(*extras)
 		
+		self.render_map['CrossReference'] = self.render_cross_reference
+		
 		self.render_map['NumberedHeading'] = self.render_numbered_heading
 		self.render_map['Subrules'] = self.render_subrules
 		self.render_map['SubrulesItem'] = self.render_subrules_item
@@ -74,6 +76,12 @@ class RTFRenderer(mistletoe.base_renderer.BaseRenderer):
 	def render_escape_sequence(self, token):
 		raise Exception('NYI')
 	
+	def render_cross_reference(self, token):
+		return '{reference_type}{reference_num}'.format(
+			reference_type=(token.reference_type + ' ') if token.reference_type else '',
+			reference_num=token.reference_num
+		)
+	
 	def render_heading(self, token):
 		if token.level == 1:
 			# Part