Added ginsu.js for splitting TiddlyWiki files into separate tiddlers

ginsu.js

@@ -0,0 +1,30 @@
+// Break apart a TiddlyWiki HTML file into separate .tid files
+//
+// Usage: node ginsu.js <tiddlywikifile> <outputdir>
+//
+// The .html extension is optional
+//
+// Ginsu creates the specified places the .tid files in the specified directory (which must already exist)
+var sys = require("sys"),
+	fs = require("fs"),
+	path = require("path"),
+	tiddler = require("./js/Tiddler.js"),
+	tiddlyWikiInput = require("./js/TiddlyWikiInput.js"),
+	tiddlerOutput = require("./js/TiddlerOutput.js");
+
+var tiddlywikifilename = process.argv[2];
+var outputdir = process.argv[3];
+
+var tiddlywikidoc = fs.readFileSync(tiddlywikifilename,"utf8");
+var tiddlers = tiddlyWikiInput.parseTiddlyWiki(tiddlywikidoc);
+
+var recipe = [];
+for(var t=0; t<tiddlers.length; t++) {
+	var tid = new tiddler.Tiddler(tiddlers[t]);
+	var filename = tid.fields.title + ".tid";
+	fs.writeFileSync(path.join(outputdir,filename),tiddlerOutput.outputTiddler(tid));
+	recipe.push("tiddler: " + filename + "\n");
+}
+
+fs.writeFileSync(path.join(outputdir,"split.recipe"),recipe.join());
+

js/TiddlerOutput.js

@@ -7,6 +7,39 @@ var argParser = require("./ArgParser.js"),
 
 var tiddlerOutput = exports;
 
+/*
+Output a tiddler as a .tid file
+*/
+tiddlerOutput.outputTiddler = function(tid) {
+	var result = [];
+	var outputAttribute = function(name,value) {
+		result.push(name + ": " + value + "\n");
+	};
+	for(var t in tid.fields) {
+		switch(t) {
+			case "text":
+				// Ignore the text field
+				break;
+			case "tags":
+				// Output tags as a list
+				outputAttribute(t,tiddlerOutput.stringifyTags(tid.fields.tags));
+				break;
+			case "modified":
+			case "created":
+				// Output dates in YYYYMMDDHHMM
+				outputAttribute(t,utils.convertToYYYYMMDDHHMM(tid.fields[t]));
+				break;
+			default:
+				// Output other attributes raw
+				outputAttribute(t,tid.fields[t]);
+				break;
+		}
+	}
+	result.push("\n");
+	result.push(tid.fields.text);
+	return result.join("");
+}
+
 /*
 Output a tiddler as an HTML <DIV>
 out - array to push the output strings

js/TiddlyWikiInput.js

@@ -0,0 +1,50 @@
+/*
+Functions concerned with parsing TiddlyWiki files
+*/
+
+var tiddlerInput = require("./TiddlerInput.js"),
+	utils = require("./Utils.js");
+
+var tiddlyWikiInput = exports;
+
+/*
+Parses the text of a TiddlyWiki HTML file, and returns the tiddlers as an array of hashmaps of raw fields.
+
+*/
+tiddlyWikiInput.parseTiddlyWiki = function(tiddlywikidoc) {
+	var results = [];
+	var storeAreaPos = locateStoreArea(tiddlywikidoc);
+	if(storeAreaPos) {
+		var endOfDivRegExp = /(<\/div>\s*)/gi;
+		var startPos = storeAreaPos[0];
+		endOfDivRegExp.lastIndex = startPos;
+		var match = endOfDivRegExp.exec(tiddlywikidoc);
+		while(match && startPos < storeAreaPos[1]) {
+			var endPos = endOfDivRegExp.lastIndex;
+			var fields = tiddlerInput.parseTiddlerDiv(tiddlywikidoc.substring(startPos,endPos));
+			results.push(fields);
+			startPos = endPos;
+			match = endOfDivRegExp.exec(tiddlywikidoc);
+		}
+	}
+	return results;
+}
+
+function locateStoreArea(tiddlywikidoc)
+{
+	var startSaveArea = '<div id="' + 'storeArea">';
+	var startSaveAreaRegExp = /<div id=["']?storeArea['"]?>/gi;
+	var endSaveArea = '</d' + 'iv>';
+	var endSaveAreaCaps = '</D' + 'IV>';
+	var posOpeningDiv = tiddlywikidoc.search(startSaveAreaRegExp);
+	var limitClosingDiv = tiddlywikidoc.indexOf("<"+"!--POST-STOREAREA--"+">");
+	if(limitClosingDiv == -1) {
+		limitClosingDiv = tiddlywikidoc.indexOf("<"+"!--POST-BODY-START--"+">");
+	}
+	var start = limitClosingDiv == -1 ? tiddlywikidoc.length : limitClosingDiv;
+	var posClosingDiv = tiddlywikidoc.lastIndexOf(endSaveArea,start);
+	if(posClosingDiv == -1) {
+		posClosingDiv = tiddlywikidoc.lastIndexOf(endSaveAreaCaps,start);
+	}
+	return (posOpeningDiv != -1 && posClosingDiv != -1) ? [posOpeningDiv + startSaveArea.length,posClosingDiv] : null;
+}

test.sh

@@ -10,4 +10,9 @@ cook $PWD/test/data/tiddlywiki.com/index.html.recipe -d $PWD -o tmp/oldcooked.ht
 node cook.js $PWD/test/data/tiddlywiki.com/index.html.recipe > tmp/newcooked.html || exit 1
 
 # compare the two
-opendiff tmp/oldcooked.html tmp/newcooked.html
+# opendiff tmp/oldcooked.html tmp/newcooked.html
+
+# split the newly cooked tiddlywiki into tiddlers
+mkdir -p tmp/tiddlers
+rm -f tmp/tiddlers/*
+node ginsu.js tmp/newcooked.html tmp/tiddlers