Tidy HTML before trying to convert it with abiword

2 files changed, 42 insertions, 2 deletions

diff --git a/src/node/utils/Settings.js b/src/node/utils/Settings.js
index b7d1f0bc..2cc6a926 100644
--- a/src/node/utils/Settings.js
+++ b/src/node/utils/Settings.js
@@ -153,6 +153,11 @@ exports.minify = true;
 exports.abiword = null;
 
 /**
+ * The path of the tidy executable
+ */
+exports.tidyHtml = null;
+
+/**
  * Should we support none natively supported file types on import?
  */
 exports.allowUnknownFileEnds = true;
@@ -167,7 +172,7 @@ exports.loglevel = "INFO";
  */
 exports.disableIPlogging = false;
 
-/** 
+/**
  * Disable Load Testing
  */
 exports.loadTest = false;
@@ -239,7 +244,7 @@ exports.reloadSettings = function reloadSettings() {
   } else {
     settingsFilename = path.resolve(path.join(exports.root, settingsFilename));
   }
-  
+
   var settingsStr;
   try{
     //read the settings sync
diff --git a/src/node/utils/TidyHtml.js b/src/node/utils/TidyHtml.js
new file mode 100644
index 00000000..13dc2ece
--- /dev/null
+++ b/src/node/utils/TidyHtml.js
@@ -0,0 +1,35 @@
+/**
+ * Tidy up the HTML in a given file
+ */
+
+var settings = require("./Settings");
+var spawn = require('child_process').spawn;
+
+exports.tidy = function(srcFile, callback) {
+  // Don't do anything if Tidy hasn't been enabled
+  if (!settings.tidyHtml) {
+    return callback(null);
+  }
+
+  var errMessage = '';
+
+  // Spawn a new tidy instance that cleans up the file inline
+  var tidy = spawn(settings.tidyHtml, ['-modify', srcFile]);
+
+  // Keep track of any error messages
+  tidy.stderr.on('data', function (data) {
+    errMessage += data.toString();
+  });
+
+  // Wait until Tidy is done
+  tidy.on('close', function(code) {
+    // Tidy returns a 0 when no errors occur and a 1 exit code when
+    // the file could be tidied but a few warnings were generated
+    if (code === 0 || code === 1) {
+      return callback(null);
+    } else {
+      console.error(errMessage);
+      return callback('Tidy died with exit code ' + code);
+    }
+  });
+};