handle index out of range error

trim_tickerlist is called from load_datafile.
trim_tickerlist will throw an index out of range error when parsing a ticker json file for a timerange outside its current contents.
This patch to load_datafile checks the timerange is in the file or not handles the exception by downloading the file with complete range. If the Exchange does not have the full range a message is logged and the file not sent to ticker-parse.

In summary, is json file does not have full requested time range
1. download file
2. if still missing time range, log and continue without ticker trim.

Here is an example of the error prior to path and after patching.
This is from binannce using the pair "ZEN/BTC" and timerange "20180522-20180523"
"""
  File "/Users/creslin/PycharmProjects/freqtrade/freqtrade/optimize/__init__.py", line 107, in load_data
    pairdata = load_tickerdata_file(datadir, pair, ticker_interval, timerange=timerange)
  File "/Users/creslin/PycharmProjects/freqtrade/freqtrade/optimize/__init__.py", line 84, in load_tickerdata_file
    pairdata = trim_tickerlist(pairdata, timerange)
  File "/Users/creslin/PycharmProjects/freqtrade/freqtrade/optimize/__init__.py", line 36, in trim_tickerlist
while tickerlist[start_index][0] < start * 1000:
IndexError: list index out of range
""""

"""
2018-05-31 14:08:04,680 - freqtrade.optimize - INFO - Start timerange not in cached data
2018-05-31 14:08:04,680 - freqtrade.optimize - INFO - Download the pair: "ZEN/BTC", Interval: 5m
dumping json to "/Users/creslin/PycharmProjects/freqtrade/freqtrade/tests/testdata/ZEN_BTC-5m.json"
2018-05-31 14:08:08,225 - freqtrade.optimize - INFO - Start timerange unavailable from exchange
"""

freqtrade/optimize/__init__.py

@@ -29,7 +29,7 @@ def trim_tickerlist(tickerlist: List[Dict], timerange: Tuple[Tuple, int, int]) -
     if stype[0] == 'index':
         start_index = start
     elif stype[0] == 'date':
-        while start_index < len(tickerlist) and tickerlist[start_index][0] < start * 1000:
+        while tickerlist[start_index][0] < start * 1000:
             start_index += 1
 
     if stype[1] == 'line':
@@ -37,7 +37,7 @@ def trim_tickerlist(tickerlist: List[Dict], timerange: Tuple[Tuple, int, int]) -
     if stype[1] == 'index':
         stop_index = stop
     elif stype[1] == 'date':
-        while stop_index > 0 and tickerlist[stop_index-1][0] > stop * 1000:
+        while tickerlist[stop_index-1][0] > stop * 1000:
             stop_index -= 1
 
     if start_index > stop_index:
@@ -49,7 +49,8 @@ def trim_tickerlist(tickerlist: List[Dict], timerange: Tuple[Tuple, int, int]) -
 def load_tickerdata_file(
         datadir: str, pair: str,
         ticker_interval: str,
-        timerange: Optional[Tuple[Tuple, int, int]] = None) -> Optional[List[Dict]]:
+        timerange: Optional[Tuple[Tuple, int, int]] = None,
+        load_count=1 ) -> Optional[List[Dict]]:
     """
     Load a pair from file,
     :return dict OR empty if unsuccesful
@@ -74,6 +75,34 @@ def load_tickerdata_file(
             pairdata = json.load(tickerdata)
     else:
         return None
+    """
+    Check if timerange is in the pairdata loaded
+    Return None is not, which will then download the file.
+    This is to avoid trim_tickerlist throwing
+    "list index out of range" error else.
+    If we've been around the download loop and still missng a start 
+    or end date, return pair data there is without trimming and log 
+    exchange does not have the range requested
+    """
+    if timerange:
+        stype, start, stop = timerange
+        if stype[0] == 'date':
+            if ((pairdata[0][0]) > (start * 1000)):
+                if load_count > 1:
+                    logger.info('Start timerange unavailable from exchange')
+                    return pairdata
+                else:
+                    logger.info('Start timerange not in cached data')
+                    return None
+        if stype[1] == 'date':
+            if (pairdata[(len(pairdata) - 1)][0]) < (stop * 1000):
+                logger.info('End timerange not in cached data')
+                if load_count > 1:
+                    logger.info('End timerange for unavailable from exchange')
+                    return pairdata
+                else:
+                    logger.info('End timerange for not in cached data')
+                    return None
 
     if timerange:
         pairdata = trim_tickerlist(pairdata, timerange)
@@ -99,12 +128,18 @@ def load_data(datadir: str,
         download_pairs(datadir, _pairs, ticker_interval, timerange=timerange)
 
     for pair in _pairs:
-        pairdata = load_tickerdata_file(datadir, pair, ticker_interval, timerange=timerange)
-        if pairdata:
+        pairdata = load_tickerdata_file(datadir, pair, ticker_interval, timerange=timerange, load_count=1)
+        if not pairdata:
+            # download the tickerdata from exchange
+            download_backtesting_testdata(datadir,
+                                          pair=pair,
+                                          tick_interval=ticker_interval,
+                                          timerange=timerange)
+            # and retry reading the pair
+            # TODO if load_tickerdata returns None we're doing nothing with it.
+            # Added load_count argument
+            pairdata = load_tickerdata_file(datadir, pair, ticker_interval, timerange=timerange, load_count=2)
         result[pair] = pairdata
-        else:
-            logger.warn('No data for pair %s, use --update-pairs-cached to download the data', pair)
-
     return result