X-Git-Url: http://nitlanguage.org

diff --git a/lib/markdown/markdown.nit b/lib/markdown/markdown.nit
index 0d264cb..613e1a4 100644
--- a/lib/markdown/markdown.nit
+++ b/lib/markdown/markdown.nit
@@ -41,33 +41,39 @@ class MarkdownProcessor
 	#
 	#   In normal markdown the following:
 	#
-	#		This is a paragraph
-	#		* and this is not a list
+	# ~~~md
+	# This is a paragraph
+	# * and this is not a list
+	# ~~~
 	#
 	#   Will produce:
 	#
-	#		<p>This is a paragraph
-	#		* and this is not a list</p>
+	# ~~~html
+	# <p>This is a paragraph
+	# * and this is not a list</p>
+	# ~~~
 	#
-	#	When using extended mode this changes to:
+	#   When using extended mode this changes to:
 	#
-	#		<p>This is a paragraph</p>
-	#		<ul>
-	#		<li>and this is not a list</li>
-	#		</ul>
+	# ~~~html
+	# <p>This is a paragraph</p>
+	# <ul>
+	# <li>and this is not a list</li>
+	# </ul>
+	# ~~~
 	#
 	# * Fences code blocks
 	#
 	#   If you don't want to indent your all your code with 4 spaces,
 	#   you can wrap your code in ``` ``` ``` or `~~~`.
 	#
-	#	Here's an example:
+	#   Here's an example:
 	#
-	#		```
-	#		fun test do
-	#			print "Hello World!"
-	#		end
-	#		```
+	# ~~~md
+	# fun test do
+	#    print "Hello World!"
+	# end
+	# ~~~
 	#
 	# * Code blocks meta
 	#
@@ -76,43 +82,55 @@ class MarkdownProcessor
 	#   You can add an optional language identifier after the fence declaration to output
 	#   it in the HTML render.
 	#
-	#		```nit
-	#		import markdown
+	# ```nit
+	# import markdown
 	#
-	#		print "# Hello World!".md_to_html
-	#		```
+	# print "# Hello World!".md_to_html
+	# ```
 	#
 	#   Becomes
 	#
-	#		<pre class="nit"><code>import markdown
+	# ~~~html
+	# <pre class="nit"><code>import markdown
 	#
-	#		print "Hello World!".md_to_html
-	#		</code></pre>
+	# print "Hello World!".md_to_html
+	# </code></pre>
+	# ~~~
 	#
 	# * Underscores (Emphasis)
 	#
 	#   Underscores in the middle of a word like:
 	#
-	#		Con_cat_this
+	# ~~~md
+	# Con_cat_this
+	# ~~~
 	#
-	#	normally produces this:
+	#   normally produces this:
 	#
-	#		<p>Con<em>cat</em>this</p>
+	# ~~~html
+	# <p>Con<em>cat</em>this</p>
+	# ~~~
 	#
 	#   With extended mode they don't result in emphasis.
 	#
-	#		<p>Con_cat_this</p>
+	# ~~~html
+	# <p>Con_cat_this</p>
+	# ~~~
 	#
 	# * Strikethrough
 	#
 	#   Like in [GFM](https://help.github.com/articles/github-flavored-markdown),
 	#   strikethrought span is marked with `~~`.
 	#
-	#		~~Mistaken text.~~
+	# ~~~md
+	# ~~Mistaken text.~~
+	# ~~~
 	#
 	#   becomes
 	#
-	#		<del>Mistaken text.</del>
+	# ~~~html
+	# <del>Mistaken text.</del>
+	# ~~~
 	var ext_mode = true
 
 	init do self.emitter = new MarkdownEmitter(self)
@@ -151,7 +169,7 @@ class MarkdownProcessor
 				if c == '\n' then
 					eol = true
 				else if c == '\t' then
-					var np = pos + (4 - (pos.bin_and(3)))
+					var np = pos + (4 - (pos & 3))
 					while pos < np do
 						value.add ' '
 						pos += 1
@@ -250,8 +268,10 @@ class MarkdownProcessor
 	#
 	# Markdown allows link refs to be defined over two lines:
 	#
-	#	[id]: http://example.com/longish/path/to/resource/here
-	#		"Optional Title Here"
+	# ~~~md
+	# [id]: http://example.com/longish/path/to/resource/here
+	#	"Optional Title Here"
+	# ~~~
 	#
 	private var last_link_ref: nullable LinkRef = null
 
@@ -376,7 +396,11 @@ class MarkdownProcessor
 			c2 = ' '
 		end
 
-		var loc = text.pos_to_loc(pos)
+		var loc = new MDLocation(
+			current_loc.line_start,
+			current_loc.column_start + pos,
+			current_loc.line_start,
+			current_loc.column_start + pos)
 
 		if c == '*' then
 			if c1 == '*' then
@@ -456,6 +480,12 @@ class MarkdownProcessor
 		end
 		return -1
 	end
+
+	# Location used for next parsed token.
+	#
+	# This location can be changed by the emitter to adjust with `\n` found
+	# in the input.
+	private fun current_loc: MDLocation do return emitter.current_loc
 end
 
 # Emit output corresponding to blocks content.
@@ -499,15 +529,19 @@ class MarkdownEmitter
 	# Transform and emit mardown text
 	fun emit_text(text: Text) do emit_text_until(text, 0, null)
 
-	# Transform and emit mardown text starting at `from` and
+	# Transform and emit mardown text starting at `start` and
 	# until a token with the same type as `token` is found.
-	# Go until the end of text if `token` is null.
+	# Go until the end of `text` if `token` is null.
 	fun emit_text_until(text: Text, start: Int, token: nullable Token): Int do
 		var old_text = current_text
 		var old_pos = current_pos
 		current_text = text
 		current_pos = start
 		while current_pos < text.length do
+			if text[current_pos] == '\n' then
+				current_loc.line_start += 1
+				current_loc.column_start = -current_pos
+			end
 			var mt = processor.token_at(text, current_pos)
 			if (token != null and not token isa TokenNone) and
 			(mt.is_same_type(token) or
@@ -550,6 +584,21 @@ class MarkdownEmitter
 		return buffer_stack.last
 	end
 
+	# Stacked locations.
+	private var loc_stack = new List[MDLocation]
+
+	# Push a new MDLocation on the stack.
+	private fun push_loc(location: MDLocation) do loc_stack.add location
+
+	# Pop the last buffer.
+	private fun pop_loc: MDLocation do return loc_stack.pop
+
+	# Current output buffer.
+	private fun current_loc: MDLocation do
+		assert not loc_stack.is_empty
+		return loc_stack.last
+	end
+
 	# Append `e` to current buffer.
 	fun add(e: Writable) do
 		if e isa Text then
@@ -597,6 +646,11 @@ interface Decorator
 	# Kind of emitter used for decoration.
 	type EMITTER: MarkdownEmitter
 
+	# Render a single plain char.
+	#
+	# Redefine this method to add special escaping for plain text.
+	fun add_char(v: EMITTER, c: Char) do v.addc c
+
 	# Render a ruler block.
 	fun add_ruler(v: EMITTER, block: BlockRuler) is abstract
 
@@ -880,6 +934,11 @@ class MDLocation
 	var column_end: Int
 
 	redef fun to_s do return "{line_start},{column_start}--{line_end},{column_end}"
+
+	# Return a copy of `self`.
+	fun copy: MDLocation do
+		return new MDLocation(line_start, column_start, line_end, column_end)
+	end
 end
 
 # A block of markdown lines.
@@ -1108,10 +1167,32 @@ abstract class Block
 	fun emit_blocks(v: MarkdownEmitter) do
 		var block = self.block.first_block
 		while block != null do
+			v.push_loc(block.location)
 			block.kind.emit(v)
+			v.pop_loc
 			block = block.next
 		end
 	end
+
+	# The raw content of the block as a multi-line string.
+	fun raw_content: String do
+		var infence = self isa BlockFence
+		var text = new FlatBuffer
+		var line = self.block.first_line
+		while line != null do
+			if not line.is_empty then
+				var str = line.value
+				if not infence and str.has_prefix("    ") then
+					text.append str.substring(4, str.length - line.trailing)
+				else
+					text.append str
+				end
+			end
+			text.append "\n"
+			line = line.next
+		end
+		return text.write_to_string
+	end
 end
 
 # A block without any markdown specificities.
@@ -1152,6 +1233,9 @@ end
 class BlockCode
 	super Block
 
+	# Any string found after fence token.
+	var meta: nullable Text
+
 	# Number of char to skip at the beginning of the line.
 	#
 	# Block code lines start at 4 spaces.
@@ -1178,9 +1262,6 @@ end
 class BlockFence
 	super BlockCode
 
-	# Any string found after fence token.
-	var meta: nullable Text
-
 	# Fence code lines start at 0 spaces.
 	redef var line_start = 0
 end
@@ -1189,7 +1270,15 @@ end
 class BlockHeadline
 	super Block
 
-	redef fun emit(v) do v.decorator.add_headline(v, self)
+	redef fun emit(v) do
+		var loc = block.location.copy
+		loc.column_start += start
+		v.push_loc(loc)
+		v.decorator.add_headline(v, self)
+		v.pop_loc
+	end
+
+	private var start = 0
 
 	# Depth of the headline used to determine the headline level.
 	var depth = 0
@@ -1218,6 +1307,7 @@ class BlockHeadline
 			line.leading = 0
 			line.trailing = 0
 		end
+		self.start = start
 		depth = level.min(6)
 	end
 end
@@ -1703,6 +1793,7 @@ class LineFence
 		else
 			block = v.current_block.split(v.current_block.last_line.as(not null))
 		end
+		block.remove_surrounding_empty_lines
 		var meta = block.first_line.value.meta_from_fence
 		block.kind = new BlockFence(block, meta)
 		block.first_line.clear
@@ -1772,7 +1863,7 @@ end
 
 # A markdown list line.
 # Mainly used to factorize code between ordered and unordered lists.
-class LineList
+abstract class LineList
 	super Line
 
 	redef fun process(v) do
@@ -1852,7 +1943,7 @@ abstract class Token
 	var char: Char
 
 	# Output that token using `MarkdownEmitter::decorator`.
-	fun emit(v: MarkdownEmitter) do v.addc char
+	fun emit(v: MarkdownEmitter) do v.decorator.add_char(v, char)
 end
 
 # A token without a specific meaning.
@@ -2036,6 +2127,7 @@ abstract class TokenLinkOrImage
 					if pos == -1 then return -1
 				end
 			end
+			if pos < start then return -1
 			if md[pos] != ')' then return -1
 		else if md[pos] == '[' then
 			pos += 1
@@ -2474,24 +2566,6 @@ redef class Text
 		return null
 	end
 
-	# Init a `MDLocation` instance at `pos` in `self`.
-	private fun pos_to_loc(pos: Int): MDLocation do
-		assert pos <= length
-		var line = 1
-		var col = 0
-		var i = 0
-		while i <= pos do
-			col += 1
-			var c = self[i]
-			if c == '\n' then
-				line +=1
-				col = 0
-			end
-			i +=1
-		end
-		return new MDLocation(line, col, line, col)
-	end
-
 	# Is `self` an unsafe HTML element?
 	private fun is_html_unsafe: Bool do return html_unsafe_tags.has(self.write_to_string)