Fix overzealous heredoc highlighting

As the example Sudoku program demonstrates, the behaviour of heredocs in
GNU APL isn't synonymous with those of other languages. If used inside a
function, the heredoc won't start to collect lines until the function is
called - at which point, it begins collecting whatever's under the call.

Although this is useful, it also makes it impossible for a heredoc to be
used in a function without screwing the document's highlighting. Instead
of removing heredoc formatting altogether, this commit allows authors to
"opt-in" based on whether the terminator contains the language's name:

    text ← ⎕INP "END-OF-HTML"
        <h1> Heading </h1>
        <a href="somewhere.htm"> Link </a>
    END-OF-HTML

Only a handful of languages are recognised in this way: anything else is
regarded as APL code. Authors can enable highlighting for the following:

    ╔══════════════╦═══════════════════════════════════════════════╗
    ║   Language   ║                 Activated by                  ║
    ╟──────────────╫───────────────────────────────────────────────╢
    ║ HTML         ║ HTML, HTM                                     ║
    ║ XML          ║ XML, XSLT, SVG, RSS                           ║
    ║ CSS          ║ CSS, Stylesheet                               ║
    ║ JavaScript   ║ JS, ECMAScript, JavaScript, JScript           ║
    ║ JSON         ║ JSON                                          ║
    ║ Plain text   ║ Raw Text, Plain Text, Raw, Plain              ║
    ╚══════════════╩═══════════════════════════════════════════════╝

The matching is performed case-insensitively, and will also work against
substrings in addition to complete matches (raw text is the exception):

    text ← ⎕INP "HTML"
    text ← ⎕INP "EndHTMLDocument"
    text ← ⎕INP "CSS"

In addition to these changes, a few pattern-matching bugs were fixed:

  * Heredocs needed leading whitespace before they would highlight

  * The closing line was expected to have no other characters except the
    terminating string (whitespace included). GNU APL actually places no
    such restriction, though it discards anything sharing the terminator
    sequence's line.

README.md

@@ -3,18 +3,4 @@ APL language support in Atom
 
 This package adds Atom support for everybody's favourite array-wrangling crypto-language.
 
-
-
-Known Limitations
----------------------------------------------
-... which aren't worth fixing:
-
-### Heredocs (GNU APL)
-1. Strings with escaped quotes can't be used as terminators:
-```apl
-text ← ⎕INP 'This won''t work'
-```
-Obviously, the solution is simple: choose a sane identifier:
-```apl
-text ← ⎕INP 'EOF'
-```
+Currently being worked on; check back in a few days. =)

grammars/apl.cson

@@ -6,7 +6,8 @@ foldingStartMarker: '\\{'
 foldingStopMarker:  '\\}'
 patterns: [
 
-	{ include: "#main" }
+	{ include: "#heredocs" }
+	{ include: "#main"     }
 
 
 	# Shebang
@@ -66,25 +67,6 @@ patterns: [
 			{ include: "#main" }
 		]
 	}
-
-
-	# Heredoc (GNU APL)
-	{
-		name: "meta.heredoc.apl"
-		begin: "^\\s*(?:(?:⎕|[A-Za-z0-9∆⍙_]+[A-Za-z0-9∆⍙_¯]*)\\s*←\\s*)?.*?⎕INP\\s+(?:('|\")(.+?)\\1).*$"
-		end:   "^\\2$"
-		beginCaptures:
-			0:
-				name: "meta.heredoc.apl"
-				patterns: [{include: "#main"}]
-		endCaptures:
-			0: {name: "constant.other.apl"}
-		contentName: "text.embedded.html.basic"
-		patterns: [
-			{ include: "text.html.basic" }
-			{ include: "#embedded-apl"   }
-		]
-	}
 ]
 
 
@@ -100,8 +82,8 @@ repository:
 			{ include: "#int"     }
 			{ include: "#name"    }
 			{ include: "#lambda"  }
-			{ include: "#symbols" }
 			{ include: "#sysvars" }
+			{ include: "#symbols" }
 		]
 	}
 
@@ -332,8 +314,92 @@ repository:
 				3: {name: "punctuation.assignment.switch.apl"}
 			patterns: [{include: "#main"}]
 		}]
+
+
+	# Heredocs (GNU APL)
+	heredocs:
+		patterns: [{
+
+			# HTML
+			name: "meta.heredoc.apl"
+			begin: "^.*?⎕INP\\s+('|\")((?i).*?HTML?.*?)\\1.*$"
+			end:   "^.*?\\2.*?$"
+			beginCaptures: {0: {patterns: [{include: "#main"}]}}
+			endCaptures:   {0: {name: "constant.other.apl"}}
+			contentName: "text.embedded.html.basic"
+			patterns: [
+				{ include: "text.html.basic" }
+				{ include: "#embedded-apl"   }
+			]
+
+		}, {
+
+			# XML
+			name: "meta.heredoc.apl"
+			begin: "^.*?⎕INP\\s+('|\")((?i).*?(?:XML|XSLT|SVG|RSS).*?)\\1.*$"
+			end:   "^.*?\\2.*?$"
+			beginCaptures: {0: {patterns: [{include: "#main"}]}}
+			endCaptures:   {0: {name: "constant.other.apl"}}
+			contentName: "text.embedded.xml"
+			patterns: [
+				{ include: "text.xml"      }
+				{ include: "#embedded-apl" }
+			]
+
+		}, {
+
+			# CSS
+			name: "meta.heredoc.apl"
+			begin: "^.*?⎕INP\\s+('|\")((?i).*?(?:CSS|stylesheet).*?)\\1.*$"
+			end:   "^.*?\\2.*?$"
+			beginCaptures: {0: {patterns: [{include: "#main"}]}}
+			endCaptures:   {0: {name: "constant.other.apl"}}
+			contentName: "source.embedded.css"
+			patterns: [
+				{ include: "source.css"    }
+				{ include: "#embedded-apl" }
+			]
+
+		}, {
+
+			# JavaScript
+			name: "meta.heredoc.apl"
+			begin: "^.*?⎕INP\\s+('|\")((?i).*?(?:JS(?!ON)|(?:ECMA|J|Java).?Script).*?)\\1.*$"
+			end:   "^.*?\\2.*?$"
+			beginCaptures: {0: {patterns: [{include: "#main"}]}}
+			endCaptures:   {0: {name: "constant.other.apl"}}
+			contentName: "source.embedded.js"
+			patterns: [
+				{ include: "source.js"     }
+				{ include: "#embedded-apl" }
+			]
+
+		}, {
+
+			# JSON
+			name: "meta.heredoc.apl"
+			begin: "^.*?⎕INP\\s+('|\")((?i).*?(?:JSON).*?)\\1.*$"
+			end:   "^.*?\\2.*?$"
+			beginCaptures: {0: {patterns: [{include: "#main"}]}}
+			endCaptures:   {0: {name: "constant.other.apl"}}
+			contentName: "source.embedded.json"
+			patterns: [
+				{ include: "source.json"   }
+				{ include: "#embedded-apl" }
+			]
+		}, {
+
+			# Raw text
+			name: "meta.heredoc.apl"
+			begin: "^.*?⎕INP\\s+('|\")(?i)((?:Raw|Plain)?\\s*Te?xt)\\1.*$"
+			end:   "^.*?\\2.*?$"
+			beginCaptures: {0: {patterns: [{include: "#main"}]}}
+			endCaptures:   {0: {name: "constant.other.apl"}}
+			patterns: [{ include: "#embedded-apl" }]
+		}]
 
-
+
+
 	# Embedded APL sequences (for GNU APL heredocs)
 	"embedded-apl":
 		patterns: [{

tests/gnu-web.apl

@@ -126,7 +126,7 @@ yZ←⊃ HTML∆Ol I1, I2, I3, I4
       ⍝ We first give an example of ⎕INP in the style of PHP and another,
       ⍝ more compact, example further down below.
       ⍝
-      yBODY← '<?apl' '?>' ⎕INP 'END-OF-⎕INP'   ⍝ php style
+      yBODY← '<?apl' '?>' ⎕INP 'END-OF-HTML'   ⍝ PHP-style
 
 <DIV class="c1">
 <?apl HTML∆H1[''] xTITLE ?>
@@ -275,7 +275,7 @@ the creation of an own hosting account would be an overkill.
 
 </DIV>
 
-END-OF-⎕INP
+END-OF-HTML
 
 
       ⍝ the text above used an 'escape style' similar to PHP
@@ -286,7 +286,7 @@ END-OF-⎕INP
       ⍝ preferred style, for example the more compact { ... } style
       ⍝ as shown in the following example:
       ⍝
-      yBODY←yBODY, (,¨'{}') ⎕INP 'END-OF-⎕INP'   ⍝ more compact style
+      yBODY←yBODY, (,¨'{}') ⎕INP 'END-OF-HTML'   ⍝ more compact style
 <DIV class="c7">
 Return to {HTML∆x2y "http://www.gnu.org/home.html" HTML∆A "GNU's home page"}.
 <P>
@@ -309,7 +309,7 @@ Copyright (C) 2014 Free Software Foundation, Inc.,
 Verbatim copying and distribution of this entire article is
 permitted in any medium, provided this notice is preserved.<P>
 </DIV>
-END-OF-⎕INP
+END-OF-HTML
 
       HTML∆emit HTML∆Document
 

tests/heredocs.apl

@@ -0,0 +1,79 @@
+⍝ HTML
+text ← ⎕INP "END-OF-HTML"
+<header id="top">
+	Heading
+</header>
+END-OF-HTML
+
+
+  aaa text ← ⎕INP "END-OF-XHTML"
+<div id="top">
+	Heading
+</div>
+END-OF-XHTML
+
+
+
+ aatext ← ⎕INP "END-OF-HTML"
+<header id="top">Heading</header>
+========== END-OF-HTML ===============
+
+
+
+
+
+⍝ XML
+tree ← ⎕INP "END-OF-XML"
+<?xml version="1.0" encoding="utf-8"?>
+<svg><!-- Fancy lines, etc --></svg>
+END-OF-XML
+
+svg ← ⎕INP "SVG"
+<?xml version="1.0" encoding="utf-8"?>
+<svg><!-- Fancy lines, etc --></svg>
+End of SVG
+
+
+
+⍝ JavaScript
+js ← ⎕INP "ENDJS"
+"use strict";
+let light = "shine";
+function fn(){
+	return Math.max(...[5, 6, 8]);
+}
+JSON.stringify({});
+const $ = s => document.querySelector(s);
+ENDJS
+
+
+⍝ JSON
+json ← ⎕INP "JSON"
+{
+	"name": "language-apl",
+	"version": "0.0.1",
+	"description": "APL language support for Atom",
+	"keywords": ["APL", "Dyalog"],
+	"repository": "https://github.com/Alhadis/language-apl",
+	"license": "ISC",
+	"engines": {
+		"atom": "*"
+	},
+	"dependencies": {}
+}
+JSON
+
+
+⍝ CSS
+styles ← ⎕INP "END-CSS"
+html{
+	background: #f00;
+}
+END-CSS
+
+
+
+⍝ Plain text
+text ← ⎕INP "Plain Text"
+Z←Z←81 9⍴1 ◊ F1←0
+Plain Text