@@ -420,7 +420,7 @@ def get_rich_pull_reviews(self, comments, eitem):
420420 # Copy data from the raw comment
421421 ecomment ['body' ] = comment ['body' ][:self .KEYWORD_MAX_LENGTH ]
422422 ecomment ['body_analyzed' ] = comment ['body' ]
423- ecomment ['url' ] = comment ['html_url' ]
423+ # ecomment['url'] = comment['html_url']
424424
425425 # extract reactions and add it to enriched item
426426 ecomment .update (self .__get_reactions (comment ))
@@ -524,6 +524,11 @@ def __get_rich_pull(self, item):
524524 self .copy_raw_fields (self .RAW_FIELDS_COPY , item , rich_pr )
525525 # The real data
526526 pull_request = item ['data' ]
527+
528+ if pull_request ['closed_at' ] == '' :
529+ pull_request ['closed_at' ] = None
530+ if pull_request ['merged_at' ] == '' :
531+ pull_request ['merged_at' ] = None
527532
528533 rich_pr ['time_to_close_days' ] = \
529534 get_time_diff_days (pull_request ['created_at' ], pull_request ['closed_at' ])
@@ -541,26 +546,26 @@ def __get_rich_pull(self, item):
541546 rich_pr ['user_name' ] = user ['name' ]
542547 rich_pr ['author_name' ] = user ['name' ]
543548 rich_pr ['user_email' ] = user .get ('email' , None )
544- rich_pr ["user_domain" ] = self .get_email_domain (user ['email' ]) if user [ 'email' ] else None
545- rich_pr ['user_org' ] = user [ 'company' ]
546- # rich_pr['user_location'] = user[ 'location']
549+ rich_pr ["user_domain" ] = self .get_email_domain (user ['email' ]) if user . get ( 'email' , None ) else None
550+ rich_pr ['user_org' ] = user . get ( 'company' , None )
551+ rich_pr ['user_location' ] = user . get ( 'location' , None )
547552 rich_pr ['user_geolocation' ] = None
548553 else :
549554 rich_pr ['user_name' ] = None
555+ rich_pr ['user_email' ] = None
550556 rich_pr ["user_domain" ] = None
551557 rich_pr ['user_org' ] = None
552558 rich_pr ['user_location' ] = None
553559 rich_pr ['user_geolocation' ] = None
554560 rich_pr ['author_name' ] = None
555- rich_pr ['user_email' ] = None
556561
557562 merged_by = pull_request .get ('merged_by_data' , None )
558563 if merged_by and merged_by != USER_NOT_AVAILABLE :
559564 rich_pr ['merge_author_login' ] = merged_by ['login' ]
560565 rich_pr ['merge_author_name' ] = merged_by ['name' ]
561- rich_pr ["merge_author_domain" ] = self .get_email_domain (merged_by ['email' ]) if merged_by [ 'email' ] else None
562- rich_pr ['merge_author_org' ] = merged_by [ 'company' ]
563- # rich_pr['merge_author_location'] = merged_by[ 'location']
566+ rich_pr ["merge_author_domain" ] = self .get_email_domain (merged_by ['email' ]) if merged_by . get ( 'email' , None ) else None
567+ rich_pr ['merge_author_org' ] = merged_by . get ( 'company' , None )
568+ rich_pr ['merge_author_location' ] = merged_by . get ( 'location' , None )
564569 rich_pr ['merge_author_geolocation' ] = None
565570 else :
566571 rich_pr ['merge_author_name' ] = None
@@ -635,12 +640,15 @@ def __get_rich_issue(self, item):
635640 self .copy_raw_fields (self .RAW_FIELDS_COPY , item , rich_issue )
636641 # The real data
637642 issue = item ['data' ]
643+
644+ if issue ['finished_at' ] == '' :
645+ issue ['finished_at' ] = None
638646
639647 rich_issue ['time_to_close_days' ] = \
640648 get_time_diff_days (issue ['created_at' ], issue ['finished_at' ])
641649
642- #issue have four status: open,progressing, closed, rejected .
643- if issue ['state' ] == 'open' or issue [ 'state' ] == 'progressing' :
650+ #issue have four status: open, closed.
651+ if issue ['state' ] == 'open' :
644652 rich_issue ['time_open_days' ] = \
645653 get_time_diff_days (issue ['created_at' ], datetime_utcnow ().replace (tzinfo = None ))
646654 else :
@@ -653,26 +661,26 @@ def __get_rich_issue(self, item):
653661 rich_issue ['user_name' ] = user ['name' ]
654662 rich_issue ['author_name' ] = user ['name' ]
655663 rich_issue ['user_email' ] = user .get ('email' , None )
656- rich_issue ["user_domain" ] = self .get_email_domain (user ['email' ]) if user [ 'email' ] else None
657- rich_issue ['user_org' ] = user [ 'company' ]
658- # rich_issue['user_location'] = user[ 'location']
664+ rich_issue ["user_domain" ] = self .get_email_domain (user ['email' ]) if user . get ( 'email' , None ) else None
665+ rich_issue ['user_org' ] = user . get ( 'company' , None )
666+ rich_issue ['user_location' ] = user . get ( 'location' , None )
659667 rich_issue ['user_geolocation' ] = None
660668 else :
661669 rich_issue ['user_name' ] = None
670+ rich_issue ['user_email' ] = None
662671 rich_issue ["user_domain" ] = None
663672 rich_issue ['user_org' ] = None
664673 rich_issue ['user_location' ] = None
665674 rich_issue ['user_geolocation' ] = None
666675 rich_issue ['author_name' ] = None
667- rich_issue ['user_email' ] = None
668676
669677 assignee = issue .get ('assignee_data' , None )
670678 if assignee and assignee != USER_NOT_AVAILABLE :
671679 rich_issue ['assignee_login' ] = assignee ['login' ]
672680 rich_issue ['assignee_name' ] = assignee ['name' ]
673- rich_issue ["assignee_domain" ] = self .get_email_domain (assignee ['email' ]) if assignee [ 'email' ] else None
674- rich_issue ['assignee_org' ] = assignee [ 'company' ]
675- # rich_issue['assignee_location'] = assignee[ 'location']
681+ rich_issue ["assignee_domain" ] = self .get_email_domain (assignee ['email' ]) if assignee . get ( 'email' , None ) else None
682+ rich_issue ['assignee_org' ] = assignee . get ( 'company' , None )
683+ rich_issue ['assignee_location' ] = assignee . get ( 'location' , None )
676684 rich_issue ['assignee_geolocation' ] = None
677685 else :
678686 rich_issue ['assignee_name' ] = None
@@ -703,9 +711,7 @@ def __get_rich_issue(self, item):
703711 rich_issue ['issue_labels' ] = labels
704712
705713 rich_issue ['item_type' ] = ISSUE_TYPE
706- rich_issue ['issue_pull_request' ] = True
707- if 'head' not in issue .keys () and 'pull_request' not in issue .keys ():
708- rich_issue ['issue_pull_request' ] = False
714+ rich_issue ['issue_pull_request' ] = False
709715
710716 rich_issue ['gitcode_repo' ] = rich_issue ['repository' ].replace (GITCODE , '' )
711717 rich_issue ['gitcode_repo' ] = re .sub ('.git$' , '' , rich_issue ['gitcode_repo' ])
0 commit comments