v4.0.7

HISTORY.rst

@@ -3,6 +3,59 @@ History
 -------
 
 
+4.0.6 (2015-06-21)
+++++++++++++++++++
+
+- allow customizing status bar message in sublime-settings file
+- guess language using multiple methods, then use most accurate guess
+- use entity and type for new heartbeats api resource schema
+- correctly log message from py.warnings module
+- upgrade wakatime cli to v4.0.15
+
+
+4.0.6 (2015-05-16)
+++++++++++++++++++
+
+- fix bug with auto detecting project name
+- upgrade wakatime cli to v4.0.13
+
+
+4.0.5 (2015-05-15)
+++++++++++++++++++
+
+- correctly display caller and lineno in log file when debug is true
+- project passed with --project argument will always be used
+- new --alternate-project argument
+- upgrade wakatime cli to v4.0.12
+
+
+4.0.4 (2015-05-12)
+++++++++++++++++++
+
+- reuse SSL connection over multiple processes for improved performance
+- upgrade wakatime cli to v4.0.11
+
+
+4.0.3 (2015-05-06)
+++++++++++++++++++
+
+- send cursorpos to wakatime cli
+- upgrade wakatime cli to v4.0.10
+
+
+4.0.2 (2015-05-06)
+++++++++++++++++++
+
+- only send heartbeats for the currently active buffer
+
+
+4.0.1 (2015-05-06)
+++++++++++++++++++
+
+- ignore git temporary files
+- don't send two write heartbeats within 2 seconds of eachother
+
+
 4.0.0 (2015-04-12)
 ++++++++++++++++++
 

WakaTime.py

@@ -7,7 +7,7 @@ Website:     https://wakatime.com/
 ==========================================================="""
 
 
-__version__ = '4.0.0'
+__version__ = '4.0.7'
 
 
 import sublime
@@ -25,13 +25,13 @@ from subprocess import Popen
 
 
 # globals
-ACTION_FREQUENCY = 2
+HEARTBEAT_FREQUENCY = 2
 ST_VERSION = int(sublime.version())
 PLUGIN_DIR = os.path.dirname(os.path.realpath(__file__))
 API_CLIENT = os.path.join(PLUGIN_DIR, 'packages', 'wakatime', 'cli.py')
 SETTINGS_FILE = 'WakaTime.sublime-settings'
 SETTINGS = {}
-LAST_ACTION = {
+LAST_HEARTBEAT = {
     'time': 0,
     'file': None,
     'is_write': False,
@@ -136,8 +136,10 @@ def obfuscate_apikey(command_list):
     return cmd
 
 
-def enough_time_passed(now, last_time):
-    if now - last_time > ACTION_FREQUENCY * 60:
+def enough_time_passed(now, last_heartbeat, is_write):
+    if now - last_heartbeat['time'] > HEARTBEAT_FREQUENCY * 60:
+        return True
+    if is_write and now - last_heartbeat['time'] > 2:
         return True
     return False
 
@@ -171,17 +173,27 @@ def find_project_from_folders(folders, current_file):
     return os.path.basename(folder) if folder else None
 
 
-def handle_action(view, is_write=False):
+def is_view_active(view):
+    if view:
+        active_window = sublime.active_window()
+        if active_window:
+            active_view = active_window.active_view()
+            if active_view:
+                return active_view.buffer_id() == view.buffer_id()
+    return False
+
+
+def handle_heartbeat(view, is_write=False):
     window = view.window()
     if window is not None:
         target_file = view.file_name()
         project = window.project_data() if hasattr(window, 'project_data') else None
         folders = window.folders()
-        thread = SendActionThread(target_file, view, is_write=is_write, project=project, folders=folders)
+        thread = SendHeartbeatThread(target_file, view, is_write=is_write, project=project, folders=folders)
         thread.start()
 
 
-class SendActionThread(threading.Thread):
+class SendHeartbeatThread(threading.Thread):
 
     def __init__(self, target_file, view, is_write=False, project=None, folders=None, force=False):
         threading.Thread.__init__(self)
@@ -194,14 +206,15 @@ class SendActionThread(threading.Thread):
         self.debug = SETTINGS.get('debug')
         self.api_key = SETTINGS.get('api_key', '')
         self.ignore = SETTINGS.get('ignore', [])
-        self.last_action = LAST_ACTION.copy()
+        self.last_heartbeat = LAST_HEARTBEAT.copy()
+        self.cursorpos = view.sel()[0].begin() if view.sel() else None
         self.view = view
 
     def run(self):
         with self.lock:
             if self.target_file:
                 self.timestamp = time.time()
-                if self.force or (self.is_write and not self.last_action['is_write']) or self.target_file != self.last_action['file'] or enough_time_passed(self.timestamp, self.last_action['time']):
+                if self.force or self.target_file != self.last_heartbeat['file'] or enough_time_passed(self.timestamp, self.last_heartbeat, self.is_write):
                     self.send_heartbeat()
 
     def send_heartbeat(self):
@@ -219,11 +232,13 @@ class SendActionThread(threading.Thread):
         if self.is_write:
             cmd.append('--write')
         if self.project and self.project.get('name'):
-            cmd.extend(['--project', self.project.get('name')])
+            cmd.extend(['--alternate-project', self.project.get('name')])
         elif self.folders:
             project_name = find_project_from_folders(self.folders, self.target_file)
             if project_name:
-                cmd.extend(['--project', project_name])
+                cmd.extend(['--alternate-project', project_name])
+        if self.cursorpos is not None:
+            cmd.extend(['--cursorpos', '{0}'.format(self.cursorpos)])
         for pattern in self.ignore:
             cmd.extend(['--ignore', pattern])
         if self.debug:
@@ -243,15 +258,15 @@ class SendActionThread(threading.Thread):
 
     def sent(self):
         sublime.set_timeout(self.set_status_bar, 0)
-        sublime.set_timeout(self.set_last_action, 0)
+        sublime.set_timeout(self.set_last_heartbeat, 0)
 
     def set_status_bar(self):
         if SETTINGS.get('status_bar_message'):
-            self.view.set_status('wakatime', 'WakaTime active {0}'.format(datetime.now().strftime('%I:%M %p')))
+            self.view.set_status('wakatime', datetime.now().strftime(SETTINGS.get('status_bar_message_fmt')))
 
-    def set_last_action(self):
-        global LAST_ACTION
-        LAST_ACTION = {
+    def set_last_heartbeat(self):
+        global LAST_HEARTBEAT
+        LAST_HEARTBEAT = {
             'file': self.target_file,
             'time': self.timestamp,
             'is_write': self.is_write,
@@ -283,13 +298,15 @@ if ST_VERSION < 3000:
 class WakatimeListener(sublime_plugin.EventListener):
 
     def on_post_save(self, view):
-        handle_action(view, is_write=True)
+        handle_heartbeat(view, is_write=True)
 
     def on_selection_modified(self, view):
-        handle_action(view)
+        if is_view_active(view):
+            handle_heartbeat(view)
 
     def on_modified(self, view):
-        handle_action(view)
+        if is_view_active(view):
+            handle_heartbeat(view)
 
 
 class WakatimeDashboardCommand(sublime_plugin.ApplicationCommand):

WakaTime.sublime-settings

@@ -9,12 +9,15 @@
     
     // Ignore files; Files (including absolute paths) that match one of these
     // POSIX regular expressions will not be logged.
-    "ignore": ["^/tmp/", "^/etc/", "^/var/"],
+    "ignore": ["^/tmp/", "^/etc/", "^/var/", "COMMIT_EDITMSG$", "PULLREQ_EDITMSG$", "MERGE_MSG$", "TAG_EDITMSG$"],
     
     // Debug mode. Set to true for verbose logging. Defaults to false.
     "debug": false,
     
     // Status bar message. Set to false to hide status bar message.
     // Defaults to true.
-    "status_bar_message": true
+    "status_bar_message": true,
+    
+    // Status bar message format.
+    "status_bar_message_fmt": "WakaTime active %I:%M %p"
 }

packages/wakatime/__about__.py

@@ -1,7 +1,7 @@
 __title__ = 'wakatime'
 __description__ = 'Common interface to the WakaTime api.'
 __url__ = 'https://github.com/wakatime/wakatime'
-__version_info__ = ('4', '0', '8')
+__version_info__ = ('4', '0', '15')
 __version__ = '.'.join(__version_info__)
 __author__ = 'Alan Hamlett'
 __author_email__ = 'alan@wakatime.com'

packages/wakatime/base.py

@@ -29,14 +29,14 @@ sys.path.insert(0, os.path.join(os.path.dirname(os.path.abspath(__file__)), 'pac
 
 from .__about__ import __version__
 from .compat import u, open, is_py3
-from .offlinequeue import Queue
 from .logger import setup_logging
-from .project import find_project
-from .stats import get_file_stats
+from .offlinequeue import Queue
 from .packages import argparse
 from .packages import simplejson as json
-from .packages import requests
 from .packages.requests.exceptions import RequestException
+from .project import find_project
+from .session_cache import SessionCache
+from .stats import get_file_stats
 try:
     from .packages import tzlocal
 except:
@@ -147,14 +147,20 @@ def parseArguments(argv):
             type=float,
             help='optional floating-point unix epoch timestamp; '+
                  'uses current time by default')
+    parser.add_argument('--lineno', dest='lineno',
+            help='optional line number; current line being edited')
+    parser.add_argument('--cursorpos', dest='cursorpos',
+            help='optional cursor position in the current file')
     parser.add_argument('--notfile', dest='notfile', action='store_true',
             help='when set, will accept any value for the file. for example, '+
                  'a domain name or other item you want to log time towards.')
     parser.add_argument('--proxy', dest='proxy',
                         help='optional https proxy url; for example: '+
                         'https://user:pass@localhost:8080')
-    parser.add_argument('--project', dest='project_name',
-            help='optional project name; auto-discovered project takes priority')
+    parser.add_argument('--project', dest='project',
+            help='optional project name')
+    parser.add_argument('--alternate-project', dest='alternate_project',
+            help='optional alternate project name; auto-discovered project takes priority')
     parser.add_argument('--disableoffline', dest='offline',
             action='store_false',
             help='disables offline time logging instead of queuing logged time')
@@ -308,20 +314,25 @@ def send_heartbeat(project=None, branch=None, stats={}, key=None, targetFile=Non
     log.debug('Sending heartbeat to api at %s' % api_url)
     data = {
         'time': timestamp,
-        'file': targetFile,
+        'entity': targetFile,
+        'type': 'file',
     }
     if hidefilenames and targetFile is not None and not notfile:
-        data['file'] = data['file'].rsplit('/', 1)[-1].rsplit('\\', 1)[-1]
-        if len(data['file'].strip('.').split('.', 1)) > 1:
-            data['file'] = u('HIDDEN.{ext}').format(ext=u(data['file'].strip('.').rsplit('.', 1)[-1]))
+        data['entity'] = data['entity'].rsplit('/', 1)[-1].rsplit('\\', 1)[-1]
+        if len(data['entity'].strip('.').split('.', 1)) > 1:
+            data['entity'] = u('HIDDEN.{ext}').format(ext=u(data['entity'].strip('.').rsplit('.', 1)[-1]))
         else:
-            data['file'] = u('HIDDEN')
+            data['entity'] = u('HIDDEN')
     if stats.get('lines'):
         data['lines'] = stats['lines']
     if stats.get('language'):
         data['language'] = stats['language']
     if stats.get('dependencies'):
         data['dependencies'] = stats['dependencies']
+    if stats.get('lineno'):
+        data['lineno'] = stats['lineno']
+    if stats.get('cursorpos'):
+        data['cursorpos'] = stats['cursorpos']
     if isWrite:
         data['is_write'] = isWrite
     if project:
@@ -352,10 +363,13 @@ def send_heartbeat(project=None, branch=None, stats={}, key=None, targetFile=Non
     if tz:
         headers['TimeZone'] = u(tz.zone)
 
+    session_cache = SessionCache()
+    session = session_cache.get()
+
     # log time to api
     response = None
     try:
-        response = requests.post(api_url, data=request_body, headers=headers,
+        response = session.post(api_url, data=request_body, headers=headers,
                                  proxies=proxies)
     except RequestException:
         exception_data = {
@@ -377,6 +391,7 @@ def send_heartbeat(project=None, branch=None, stats={}, key=None, targetFile=Non
             log.debug({
                 'response_code': response_code,
             })
+            session_cache.save(session)
             return True
         if offline:
             if response_code != 400:
@@ -402,6 +417,7 @@ def send_heartbeat(project=None, branch=None, stats={}, key=None, targetFile=Non
                 'response_code': response_code,
                 'response_content': response_content,
             })
+    session_cache.delete()
     return False
 
 
@@ -424,23 +440,27 @@ def main(argv=None):
 
     if os.path.isfile(args.targetFile) or args.notfile:
 
-        stats = get_file_stats(args.targetFile, notfile=args.notfile)
+        stats = get_file_stats(args.targetFile, notfile=args.notfile,
+                               lineno=args.lineno, cursorpos=args.cursorpos)
 
         project = None
         if not args.notfile:
             project = find_project(args.targetFile, configs=configs)
         branch = None
-        project_name = args.project_name
+        project_name = args.project
         if project:
             branch = project.branch()
-            project_name = project.name()
+            if not project_name:
+                project_name = project.name()
+        if not project_name:
+            project_name = args.alternate_project
 
-        if send_heartbeat(
-                project=project_name,
-                branch=branch,
-                stats=stats,
-                **vars(args)
-            ):
+        kwargs = vars(args)
+        kwargs['project'] = project_name
+        kwargs['branch'] = branch
+        kwargs['stats'] = stats
+
+        if send_heartbeat(**kwargs):
             queue = Queue()
             while True:
                 heartbeat = queue.pop()

packages/wakatime/logger.py

@@ -37,28 +37,34 @@ class CustomEncoder(json.JSONEncoder):
 
 class JsonFormatter(logging.Formatter):
 
-    def setup(self, timestamp, isWrite, targetFile, version, plugin):
+    def setup(self, timestamp, isWrite, targetFile, version, plugin, verbose,
+              warnings=False):
         self.timestamp = timestamp
         self.isWrite = isWrite
         self.targetFile = targetFile
         self.version = version
         self.plugin = plugin
+        self.verbose = verbose
+        self.warnings = warnings
 
-    def format(self, record):
+    def format(self, record, *args):
         data = OrderedDict([
             ('now', self.formatTime(record, self.datefmt)),
-            ('version', self.version),
-            ('plugin', self.plugin),
-            ('time', self.timestamp),
-            ('isWrite', self.isWrite),
-            ('file', self.targetFile),
-            ('level', record.levelname),
-            ('message', record.msg),
         ])
+        data['version'] = self.version
+        data['plugin'] = self.plugin
+        data['time'] = self.timestamp
+        if self.verbose:
+            data['caller'] = record.pathname
+            data['lineno'] = record.lineno
+            data['isWrite'] = self.isWrite
+            data['file'] = self.targetFile
+            if not self.isWrite:
+                del data['isWrite']
+        data['level'] = record.levelname
+        data['message'] = record.getMessage() if self.warnings else record.msg
         if not self.plugin:
             del data['plugin']
-        if not self.isWrite:
-            del data['isWrite']
         return CustomEncoder().encode(data)
 
     def formatException(self, exc_info):
@@ -73,7 +79,6 @@ def set_log_level(logger, args):
 
 
 def setup_logging(args, version):
-    logging.captureWarnings(True)
     logger = logging.getLogger('WakaTime')
     set_log_level(logger, args)
     if len(logger.handlers) > 0:
@@ -84,6 +89,7 @@ def setup_logging(args, version):
             targetFile=args.targetFile,
             version=version,
             plugin=args.plugin,
+            verbose=args.verbose,
         )
         logger.handlers[0].setFormatter(formatter)
         return logger
@@ -98,7 +104,27 @@ def setup_logging(args, version):
         targetFile=args.targetFile,
         version=version,
         plugin=args.plugin,
+        verbose=args.verbose,
     )
     handler.setFormatter(formatter)
     logger.addHandler(handler)
+
+    warnings_formatter = JsonFormatter(datefmt='%Y/%m/%d %H:%M:%S %z')
+    warnings_formatter.setup(
+        timestamp=args.timestamp,
+        isWrite=args.isWrite,
+        targetFile=args.targetFile,
+        version=version,
+        plugin=args.plugin,
+        verbose=args.verbose,
+        warnings=True,
+    )
+    warnings_handler = logging.FileHandler(os.path.expanduser(logfile))
+    warnings_handler.setFormatter(warnings_formatter)
+    logging.getLogger('py.warnings').addHandler(warnings_handler)
+    try:
+        logging.captureWarnings(True)
+    except AttributeError:
+        pass  # Python >= 2.7 is needed to capture warnings
+
     return logger

packages/wakatime/offlinequeue.py

@@ -1,10 +1,9 @@
 # -*- coding: utf-8 -*-
 """
-    wakatime.queue
-    ~~~~~~~~~~~~~~
+    wakatime.offlinequeue
+    ~~~~~~~~~~~~~~~~~~~~~
 
-    Queue for offline time logging.
-    http://wakatime.com
+    Queue for saving heartbeats while offline.
 
     :copyright: (c) 2014 Alan Hamlett.
     :license: BSD, see LICENSE for more details.

packages/wakatime/session_cache.py

@@ -0,0 +1,109 @@
+# -*- coding: utf-8 -*-
+"""
+    wakatime.session_cache
+    ~~~~~~~~~~~~~~~~~~~~~~
+
+    Persist requests.Session for multiprocess SSL handshake pooling.
+
+    :copyright: (c) 2015 Alan Hamlett.
+    :license: BSD, see LICENSE for more details.
+"""
+
+
+import logging
+import os
+import pickle
+import sys
+import traceback
+
+try:
+    import sqlite3
+    HAS_SQL = True
+except ImportError:
+    HAS_SQL = False
+
+sys.path.insert(0, os.path.join(os.path.dirname(os.path.abspath(__file__)), 'packages'))
+
+from .packages import requests
+
+
+log = logging.getLogger('WakaTime')
+
+
+class SessionCache(object):
+    DB_FILE = os.path.join(os.path.expanduser('~'), '.wakatime.db')
+
+    def connect(self):
+        conn = sqlite3.connect(self.DB_FILE)
+        c = conn.cursor()
+        c.execute('''CREATE TABLE IF NOT EXISTS session (
+            value BLOB)
+        ''')
+        return (conn, c)
+
+
+    def save(self, session):
+        """Saves a requests.Session object for the next heartbeat process.
+        """
+
+        if not HAS_SQL:
+            return
+        try:
+            conn, c = self.connect()
+            c.execute('DELETE FROM session')
+            values = {
+                'value': pickle.dumps(session),
+            }
+            c.execute('INSERT INTO session VALUES (:value)', values)
+            conn.commit()
+            conn.close()
+        except:
+            log.error(traceback.format_exc())
+
+
+    def get(self):
+        """Returns a requests.Session object.
+
+        Gets Session from sqlite3 cache or creates a new Session.
+        """
+
+        if not HAS_SQL:
+            return requests.session()
+
+        try:
+            conn, c = self.connect()
+        except:
+            log.error(traceback.format_exc())
+            return requests.session()
+
+        session = None
+        try:
+            c.execute('BEGIN IMMEDIATE')
+            c.execute('SELECT value FROM session LIMIT 1')
+            row = c.fetchone()
+            if row is not None:
+                session = pickle.loads(row[0])
+        except:
+            log.error(traceback.format_exc())
+
+        try:
+            conn.close()
+        except:
+            log.error(traceback.format_exc())
+
+        return session if session is not None else requests.session()
+
+
+    def delete(self):
+        """Clears all cached Session objects.
+        """
+
+        if not HAS_SQL:
+            return
+        try:
+            conn, c = self.connect()
+            c.execute('DELETE FROM session')
+            conn.commit()
+            conn.close()
+        except:
+            log.error(traceback.format_exc())

packages/wakatime/stats.py

@@ -20,13 +20,15 @@ if sys.version_info[0] == 2:
     sys.path.insert(0, os.path.join(os.path.dirname(os.path.abspath(__file__)), 'packages', 'pygments_py2'))
 else:
     sys.path.insert(0, os.path.join(os.path.dirname(os.path.abspath(__file__)), 'packages', 'pygments_py3'))
-from pygments.lexers import guess_lexer_for_filename
+from pygments.lexers import get_lexer_by_name, guess_lexer_for_filename
+from pygments.modeline import get_filetype_from_buffer
+from pygments.util import ClassNotFound
 
 
 log = logging.getLogger('WakaTime')
 
 
-# force file name extensions to be recognized as a certain language
+# extensions taking priority over lexer
 EXTENSIONS = {
     'j2': 'HTML',
     'markdown': 'Markdown',
@@ -34,6 +36,8 @@ EXTENSIONS = {
     'mdown': 'Markdown',
     'twig': 'Twig',
 }
+
+# lexers to human readable languages
 TRANSLATIONS = {
     'CSS+Genshi Text': 'CSS',
     'CSS+Lasso': 'CSS',
@@ -45,31 +49,127 @@ TRANSLATIONS = {
     'RHTML': 'HTML',
 }
 
+# extensions for when no lexer is found
+AUXILIARY_EXTENSIONS = {
+    'vb': 'VB.net',
+}
+
 
 def guess_language(file_name):
-    language, lexer = None, None
-    try:
-        with open(file_name, 'r', encoding='utf-8') as fh:
-            lexer = guess_lexer_for_filename(file_name, fh.read(512000))
-    except:
-        pass
+    """Guess lexer and language for a file.
+
+    Returns (language, lexer) tuple where language is a unicode string.
+    """
+
+    lexer = smart_guess_lexer(file_name)
+
+    language = None
+
+    # guess language from file extension
     if file_name:
-        language = guess_language_from_extension(file_name.rsplit('.', 1)[-1])
-    if lexer and language is None:
-        language = translate_language(u(lexer.name))
+        language = get_language_from_extension(file_name, EXTENSIONS)
+
+    # get language from lexer if we didn't have a hard-coded extension rule
+    if language is None and lexer:
+        language = u(lexer.name)
+
+    if language is None:
+        language = get_language_from_extension(file_name, AUXILIARY_EXTENSIONS)
+
+    if language is not None:
+        language = translate_language(language)
+
     return language, lexer
 
 
-def guess_language_from_extension(extension):
+def smart_guess_lexer(file_name):
+    """Guess Pygments lexer for a file.
+
+    Looks for a vim modeline in file contents, then compares the accuracy
+    of that lexer with a second guess. The second guess looks up all lexers
+    matching the file name, then runs a text analysis for the best choice.
+    """
+    lexer = None
+
+    text = get_file_contents(file_name)
+
+    lexer_1, accuracy_1 = guess_lexer_using_filename(file_name, text)
+    lexer_2, accuracy_2 = guess_lexer_using_modeline(text)
+
+    if lexer_1:
+        lexer = lexer_1
+    if (lexer_2 and accuracy_2 and
+        (not accuracy_1 or accuracy_2 > accuracy_1)):
+        lexer = lexer_2
+
+    return lexer
+
+
+def guess_lexer_using_filename(file_name, text):
+    """Guess lexer for given text, limited to lexers for this file's extension.
+
+    Returns a tuple of (lexer, accuracy).
+    """
+
+    lexer, accuracy = None, None
+
+    try:
+        lexer = guess_lexer_for_filename(file_name, text)
+    except:
+        pass
+
+    if lexer is not None:
+        try:
+            accuracy = lexer.analyse_text(text)
+        except:
+            pass
+
+    return lexer, accuracy
+
+
+def guess_lexer_using_modeline(text):
+    """Guess lexer for given text using Vim modeline.
+
+    Returns a tuple of (lexer, accuracy).
+    """
+
+    lexer, accuracy = None, None
+
+    file_type = get_filetype_from_buffer(text)
+    if file_type is not None:
+        try:
+            lexer = get_lexer_by_name(file_type)
+        except ClassNotFound:
+            pass
+
+    if lexer is not None:
+        try:
+            accuracy = lexer.analyse_text(text)
+        except:
+            pass
+
+    return lexer, accuracy
+
+
+def get_language_from_extension(file_name, extension_map):
+    """Returns a matching language for the given file_name using extension_map.
+    """
+
+    extension = file_name.rsplit('.', 1)[-1] if len(file_name.rsplit('.', 1)) > 1 else None
+
     if extension:
-        if extension in EXTENSIONS:
-            return EXTENSIONS[extension]
-        if extension.lower() in EXTENSIONS:
-            return EXTENSIONS[extension.lower()]
+        if extension in extension_map:
+            return extension_map[extension]
+        if extension.lower() in extension_map:
+            return extension_map[extension.lower()]
+
     return None
 
 
 def translate_language(language):
+    """Turns Pygments lexer class name string into human-readable language.
+    """
+
     if language in TRANSLATIONS:
         language = TRANSLATIONS[language]
     return language
@@ -86,12 +186,14 @@ def number_lines_in_file(file_name):
     return lines
 
 
-def get_file_stats(file_name, notfile=False):
+def get_file_stats(file_name, notfile=False, lineno=None, cursorpos=None):
     if notfile:
         stats = {
             'language': None,
             'dependencies': [],
             'lines': None,
+            'lineno': lineno,
+            'cursorpos': cursorpos,
         }
     else:
         language, lexer = guess_language(file_name)
@@ -101,5 +203,20 @@ def get_file_stats(file_name, notfile=False):
             'language': language,
             'dependencies': dependencies,
             'lines': number_lines_in_file(file_name),
+            'lineno': lineno,
+            'cursorpos': cursorpos,
         }
     return stats
+
+
+def get_file_contents(file_name):
+    """Returns the first 512000 bytes of the file's contents.
+    """
+
+    text = None
+    try:
+        with open(file_name, 'r', encoding='utf-8') as fh:
+            text = fh.read(512000)
+    except:
+        pass
+    return text