adding comments

src/browser/worker.js

@@ -7,8 +7,6 @@ global.addEventListener('message', function(e){
     workerUtils.dispatchHandlers(packet, obj => postMessage(obj))
 })
 
-exports.getLanguageData = require('./lang.js')
-
 exports.getCore = function(req, res){
     if(!global.TesseractCore){
         res.progress({ status: 'loading tesseract core' })
@@ -18,4 +16,6 @@ exports.getCore = function(req, res){
     return TesseractCore
 }
 
+exports.getLanguageData = require('./lang.js')
+
 workerUtils.setAdapter(module.exports);

src/common/circularize.js

@@ -1,3 +1,10 @@
+// The result of dump.js is a big JSON tree
+// which can be easily serialized (for instance
+// to be sent from a webworker to the main app
+// or through Node's IPC), but we want
+// a (circular) DOM-like interface for walking
+// through the data. 
+
 module.exports = function circularize(page){
     page.paragraphs = []
     page.lines = []

src/common/desaturate.js

@@ -1,8 +1,11 @@
+// This converts an image to grayscale
+
 module.exports = function desaturate(image){
     var width, height;
     if(image.data){
         var src       = image.data;
-        width         = image.width, height = image.height;
+            width     = image.width, 
+            height    = image.height;
         var dst       = new Uint8Array(width * height);
         var srcLength = src.length | 0, srcLength_16 = (srcLength - 16) | 0;
         
@@ -16,8 +19,6 @@ module.exports = function desaturate(image){
         for (; i < srcLength; i += 4, ++j) //finish up
             dst[j]     = (((src[i] * 77 + src[i+1] * 151 + src[i+2] * 28) * src[i+3]) + ((255-src[i+3]) << 15) + 32768) >> 16
         image = dst;
-    } else {
+    } else { throw 'Invalid ImageData' }
-        throw 'Expected ImageData'
-    }
     return image
 }

src/common/dump.js

@@ -1,15 +1,3 @@
-function deindent(html){
-    var lines = html.split('\n')
-    if(lines[0].substring(0,2) === "  "){
-        for (var i = 0; i < lines.length; i++) {
-            if (lines[i].substring(0,2) === "  ") {
-                lines[i] = lines[i].slice(2)
-            }
-        };
-    }
-    return lines.join('\n')
-}
-
 module.exports = function DumpLiterallyEverything(Module, base){
     var ri = base.GetIterator();
     var blocks = [];
@@ -159,3 +147,18 @@ module.exports = function DumpLiterallyEverything(Module, base){
         version: base.Version(),
     }
 }
+
+// the generated HOCR is excessively indented, so
+// we get rid of that indentation
+
+function deindent(html){
+    var lines = html.split('\n')
+    if(lines[0].substring(0, 2) === "  "){
+        for (var i = 0; i < lines.length; i++) {
+            if (lines[i].substring(0,2) === "  ") {
+                lines[i] = lines[i].slice(2)
+            }
+        };
+    }
+    return lines.join('\n')
+}

src/node/index.js

@@ -1,4 +1,5 @@
 var path = require('path')
+
 exports.defaultOptions = {
     workerPath: path.join(__dirname, 'worker.js'),
     langPath: 'http://cdn.rawgit.com/naptha/tessdata/gh-pages/3.02/',

src/node/worker.js

@@ -6,9 +6,6 @@ process.on('message', function(packet){
     workerUtils.dispatchHandlers(packet, obj => process.send(obj))
 })
 
-exports.getLanguageData = require('./lang.js')
-
-
 var TesseractCore;
 exports.getCore = function(req, res){
     if(!TesseractCore){
@@ -19,4 +16,6 @@ exports.getCore = function(req, res){
     return TesseractCore
 }
 
+exports.getLanguageData = require('./lang.js')
+
 workerUtils.setAdapter(module.exports);