Error handling added

lib/utils.py

@@ -8,7 +8,6 @@ import multiprocessing as mp
 from math import ceil
 from typing import List
 from tweet_manager.lib import json2csv, format_csv
-from tweet_manager.lib.json2csv import NotFlattenedException
 
 from config import globals
 
@@ -58,6 +57,7 @@ def filesystem_writer(
             break
         elif csv_page == "ERROR":
             logger.error("Dumping recovery file and exiting")
+            if recovery_file_path >= 0:
                 dump_error_recovery_file(recovery_file_path, page_number)
             break
         elif csv_page is not None:
@@ -107,6 +107,7 @@ def process_page(
     :param queue: queue were processed data await to be written to FS
     """
 
+    try:
         logger.debug(
             "Worker {} launched: process_page executing".format(os.getpid()))
         if globals.timing:
@@ -117,20 +118,21 @@ def process_page(
         tweets_page = get_tweets_page(database_tweets, pagesize, page_number)
         buffer_tweets = {}
 
-    try:
         for tweet in tweets_page:
             csv_tweet_output_path =\
                 get_tweet_output_path(tweet, output_dir)
+            try:
                 csv_tweet_contents =\
                     "\n" + str(convert_tweet_to_csv(header, tweet))
-
+            except TweetConversionException as exc:
+                logger.error(exc.message)
+                logger.error("Origin tweet:\n" + str(exc.tweet))
+                logger.error("Discarding tweet and proceeding...")
+                continue
             if csv_tweet_output_path not in buffer_tweets.keys():
                 buffer_tweets[csv_tweet_output_path] = ""
             buffer_tweets[csv_tweet_output_path] += csv_tweet_contents
         client.close()
-    except NotFlattenedException as exc:
-        logger.critical(exc.args)
-        raise ExceptionAtPage("Corrupt tweets found", page_number)
 
         queue.put((page_number, buffer_tweets))
 
@@ -141,6 +143,10 @@ def process_page(
                 "Time processing & buffering tweet page: {}s"
                 .format(time1 - time0))
 
+    except Exception:
+        raise ExceptionAtPage(
+            "Something failed while processing page", page_number)
+
 
 def get_tweet_output_path(tweet: dict, output_dir: str) -> str:
     """
@@ -237,10 +243,22 @@ def convert_tweet_to_csv(header: str, tweet: dict) -> str:
               fields in CSV form
     """
 
+    try:
         flat_tweet = json2csv.flatten_dictionary(tweet)
+    except Exception:
+        raise TweetConversionException(
+            "Error when flattening tweet", tweet)
+    try:
         csv_tweet = json2csv.json2csv(flat_tweet, True, 5, False)
-    csv_appendable_tweet = format_csv.get_csv_line(header, csv_tweet)
-
+    except Exception:
+        raise TweetConversionException(
+            "Error when trying to convert tweet to CSV", flat_tweet)
+    try:
+        csv_appendable_tweet = format_csv.get_csv_line(
+            header, csv_tweet)
+    except Exception:
+        raise TweetConversionException(
+            "Error when formatting CSV tweet", csv_tweet)
     return csv_appendable_tweet
 
 
@@ -280,7 +298,7 @@ def dump_recovery_file(
     with open(recovery_file_path, "w") as f:
         json.dump(recovery_file_contents, f)
 
-    logger.error(
+    logger.debug(
         "Generated recovery file at {}".format(recovery_file_path))
 
 
@@ -451,3 +469,9 @@ class ExceptionAtPage(Exception):
     def __init__(self, message, error_page):
         self.message = message
         self.error_page = error_page
+
+
+class TweetConversionException(Exception):
+    def __init__(self, message, tweet):
+        self.message = message
+        self.tweet = tweet

pymongoexport_csv.py

@@ -58,6 +58,9 @@ if args.recovery:
     client.close()
     page_index = [page for page in full_page_index
                   if page not in recovery_data["dumped_pages"]]
+    if "error_page" in recovery_data:
+        logger.debug("Discarding corrupted page")
+        page_index.remove(recovery_data.pop("error_page"))
     logger.debug(
         "Resuming collection conversion. {} of {} pages left."
         .format(len(page_index), len(full_page_index)))
@@ -97,6 +100,11 @@ except utils.ExceptionAtPage as exc:
     logger.error("Error detected at page {}".format(exc.error_page))
     task_queue.put((exc.error_page, "ERROR"))
     sys.exit(1)
+except (Exception, KeyboardInterrupt):
+    logger.error("Error detected")
+    task_queue.put((-2, "ERROR"))
+    sys.exit(1)
+
 task_queue.put((-1, "END"))