Merge pull request #39 from lappis-unb/folder_inside_training_data

Add recursive strategy when reading the training corpus

Dockerfile

@@ -4,6 +4,7 @@ LABEL mantainer "Diego Dorgam <diego.dorgam@rocket.chat>"
 
 ENV HUBOT_LANG='en'                                                  \
     HUBOT_CORPUS='training_data/corpus.yml'                          \
+    HUBOT_RECURSIVE_TRAINING=false                                   \
     HUBOT_ADAPTER=rocketchat                                         \
     HUBOT_OWNER=RocketChat                                           \
     HUBOT_NAME=HubotNatural                                          \

docker-compose.yml

@@ -51,6 +51,7 @@ services:
       - HUBOT_DESCRIPTION='Hubot natural language processing'
       - HUBOT_LOG_LEVEL=debug
       - HUBOT_CORPUS=training_data/corpus.yml
+      - HUBOT_RECURSIVE_TRAINING=false
       - HUBOT_LANG=pt
       - RESPOND_TO_DM=true
       - RESPOND_TO_LIVECHAT=true

scripts/lib/common.coffee

@@ -31,14 +31,19 @@ common.sendMessages = (messages, msg, variables = {}) ->
   messages = messages.map (message) ->
     return common.msgVariables message, msg, variables
   msg.sendWithNaturalDelay messages
-
-getYAMLFiles = (filepath) ->
+  
+getYAMLFiles = (filepath, recursive = false) ->
   listFile = fs.readdirSync filepath
+
   dataFiles = []
-  if listFile.length > 0
-    dataFiles = listFile.map (filename) ->
-      return yaml.safeLoad fs.readFileSync filepath + '/' + filename, 'utf8'
-  else
+  for filename in listFile
+    file = filepath + '/' + filename
+    if fs.lstatSync(file).isFile()
+      dataFiles.push(yaml.safeLoad fs.readFileSync file, 'utf8')
+    else if recursive
+      dataFiles = dataFiles.concat(getYAMLFiles file, recursive)
+
+  if dataFiles.lenght is 0
     console.error('The directory: ' + filepath + ' is empty.')
   return dataFiles
 
@@ -68,7 +73,8 @@ common.loadConfigfile = (filepath) ->
       return yaml.safeLoad fs.readFileSync filepath, 'utf8'
 
     else if fs.lstatSync(filepath).isDirectory()
-      yamlFiles = getYAMLFiles(filepath)
+      recursiveTraining = process.env.HUBOT_RECURSIVE_TRAINING || false
+      yamlFiles = getYAMLFiles(filepath, recursiveTraining)
       return concatYAMLFiles(yamlFiles)
 
   catch err