diff --git a/schema/areas_of_code.csv b/schema/areas_of_code.csv index 176e1891a..8d6a2b9f6 100644 --- a/schema/areas_of_code.csv +++ b/schema/areas_of_code.csv @@ -1,37 +1,37 @@ -name,type,aggregatable,description -addedlines,long,true,"Number of lines added in this file by this commit (only in this file, not the whole commit)." -author_bot,boolean,true,"True if the given author is identified as a bot." -author_domain,keyword,true,"Domain associated to the author in SortingHat profile." -author_id,keyword,true,"Author Id from SortingHat." -author_name,keyword,true,"Author name." -author_org_name,keyword,true,"Organization name." -author_user_name,keyword,true,"Author user name." -author_uuid,keyword,true,"Author UUID from SortingHat." -committer,keyword,true,"Committer name as it appears in commit (including e-mail)." -committer_date,date,true,"Date when committer made this commit." -date,date,true,"Author date (when the original author made the commit)." -eventtype,keyword,true,"COMMIT." -file_dir_name,keyword,true,"Path in which the file is located, not including file name." -file_ext,keyword,true,"File extension." -file_name,keyword,true,"File name with extension." -file_path_list,keyword,true,"List of split path parts." -fileaction,keyword,true,"Action performed by the commit over the file." -filepath,keyword,true,"Complete file path." -files,long,true,"Number of files touched by the same commit this file is included in." -filetype,keyword,true,"Code or Other, based on file extension." -git_author_domain,keyword,true,"Domain extracted from author email included within the commit, if any." -grimoire_creation_date,date,true,"Author date (when the original author made the commit)." -hash,keyword,true,"Commit hash." -id,keyword,true,"Commit hash." -message,text,false,"Commit message split by terms." -message.keyword,keyword,true,"Commit message as a single String." -metadata__enriched_on,date,true,"Date when the item was enriched." -metadata__timestamp,date,true,"Date when the item was stored in RAW index." -metadata__updated_on,date,true,"Date when the item was updated in its original data source." -owner,keyword,true,"Owner (code author) name as it appears in commit (including e-mail)." -perceval_uuid,keyword,true,"Perceval UUID." -project,keyword,true,"Project." -project_1,keyword,true,"Project (if more than one level is allowed in project hierarchy)." -removedlines,long,true,"Number of lines removed in this file by this commit (only in this file, not the whole commit)." -repository,keyword,true,"Repository name." -uuid,keyword,true,"Item unique identifier. Same as '_id'" +name,type,aggregatable,description,Raw/Calculated +addedlines,long,TRUE,"Number of lines added in this file by this commit (only in this file, not the whole commit).",GrimoireLab +author_bot,boolean,TRUE,True if the given author is identified as a bot.,GrimoireLab / SortingHat +author_domain,keyword,TRUE,Domain associated to the author in SortingHat profile.,GrimoireLab / SortingHat +author_id,keyword,TRUE,Author Id from SortingHat.,GrimoireLab / SortingHat +author_name,keyword,TRUE,Author name.,GrimoireLab / SortingHat +author_org_name,keyword,TRUE,Organization name.,GrimoireLab / SortingHat +author_user_name,keyword,TRUE,Author user name.,GrimoireLab / SortingHat +author_uuid,keyword,TRUE,Author UUID from SortingHat.,GrimoireLab / SortingHat +committer,keyword,TRUE,Committer name as it appears in commit (including e-mail).,GrimoireLab / SortingHat +committer_date,date,TRUE,Date when committer made this commit.,Data Source / Parsed +date,date,TRUE,Author date (when the original author made the commit).,Data Source / Parsed +eventtype,keyword,TRUE,COMMIT.,GrimoireLab +file_dir_name,keyword,TRUE,"Path in which the file is located, not including file name.",Data Source / Parsed +file_ext,keyword,TRUE,File extension.,Data Source / Parsed +file_name,keyword,TRUE,File name with extension.,Data Source / Parsed +file_path_list,keyword,TRUE,List of split path parts.,Data Source / Parsed +fileaction,keyword,TRUE,Action performed by the commit over the file.,Data Source / Parsed +filepath,keyword,TRUE,Complete file path.,Data Source / Parsed +files,long,TRUE,Number of files touched by the same commit this file is included in.,Data Source / Parsed +filetype,keyword,TRUE,"Code or Other, based on file extension.",GrimoireLab +git_author_domain,keyword,TRUE,"Domain extracted from author email included within the commit, if any.",Data Source / Parsed +grimoire_creation_date,date,TRUE,Author date (when the original author made the commit).,Data Source / Parsed +hash,keyword,TRUE,Commit hash.,Data Source / Parsed +id,keyword,TRUE,Commit hash.,Data Source / Parsed +message,text,FALSE,Commit message split by terms.,Data Source / Parsed +message.keyword,keyword,TRUE,Commit message as a single String.,Data Source / Parsed +metadata__enriched_on,date,TRUE,Date when the item was enriched.,GrimoireLab +metadata__timestamp,date,TRUE,Date when the item was stored in RAW index.,GrimoireLab +metadata__updated_on,date,TRUE,Date when the item was updated in its original data source.,GrimoireLab +owner,keyword,TRUE,Owner (code author) name as it appears in commit (including e-mail).,Data Source / Parsed +perceval_uuid,keyword,TRUE,Perceval UUID.,GrimoireLab +project,keyword,TRUE,Project.,GrimoireLab +project_1,keyword,TRUE,Project (if more than one level is allowed in project hierarchy).,GrimoireLab +removedlines,long,TRUE,"Number of lines removed in this file by this commit (only in this file, not the whole commit).",GrimoireLab +repository,keyword,TRUE,Repository name.,Data Source / Parsed +uuid,keyword,TRUE,Item unique identifier. Same as '_id',GrimoireLab \ No newline at end of file diff --git a/schema/git_survival.csv b/schema/git_survival.csv index ddaade651..6c27e8e43 100644 --- a/schema/git_survival.csv +++ b/schema/git_survival.csv @@ -1,25 +1,37 @@ -name,type,aggregatable,description -uuid,keyword,true,"Perceval UUID." -origin,keyword,true,"Original URL where the repository was retrieved from." -repository,keyword,true,"Repository URL." -interval_months,long,true,"Time frame of each analysis." -from_date,date,true,"Starting date of the time frame." -to_date,date,true,"End date of the time frame." -study_creation_date,date,true,"Date when the study was executed." -author_uuid,keyword,true,"Author UUID from SortingHat." -author_name,keyword,true,"Author name." -author_bot,boolean,true,"True if the given author is identified as a bot." -author_user_name,keyword,true,"Username of the user." -author_org_name,keyword,true,"Organization name." -author_domain,keyword,true,"Domain associated to the author in SortingHat profile." -metadata__enriched_on,date,true,"Date when the item was enriched." -metadata__gelk_backend_name,keyword,true,"Name of the backend used to enrich information." -metadata__gelk_version,keyword,true,"Version of the backend used to enrich information." -grimoire_creation_date,date,true,"Commit date (when the original author made the commit)." -is_git_survived,long,true,"Field containing '1' that allows to sum fields when concatenating with other indexes." -prediction_09,long,true,"Number of days until the next predicted activity (90% of probability)." -prediction_07,long,true,"Number of days until the next predicted activity (70% of probability)." -prediction_05,long,true,"Number of days until the next predicted activity (50% of probability)" -next_activity_09,date,true,"Date of the next predicted activity (90% of probability)" -next_activity_07,date,true,"Date of the next predicted activity (70% of probability)" -next_activity_05,date,true,"Date of the next predicted activity (50% of probability)" \ No newline at end of file +name,type,aggregatable,description,Raw/Calculated +addedlines,long,TRUE,"Number of lines added in this file by this commit (only in this file, not the whole commit).",GrimoireLab +author_bot,boolean,TRUE,True if the given author is identified as a bot.,GrimoireLab / SortingHat +author_domain,keyword,TRUE,Domain associated to the author in SortingHat profile.,GrimoireLab / SortingHat +author_id,keyword,TRUE,Author Id from SortingHat.,GrimoireLab / SortingHat +author_name,keyword,TRUE,Author name.,GrimoireLab / SortingHat +author_org_name,keyword,TRUE,Organization name.,GrimoireLab / SortingHat +author_user_name,keyword,TRUE,Author user name.,GrimoireLab / SortingHat +author_uuid,keyword,TRUE,Author UUID from SortingHat.,GrimoireLab / SortingHat +committer,keyword,TRUE,Committer name as it appears in commit (including e-mail).,GrimoireLab / SortingHat +committer_date,date,TRUE,Date when committer made this commit.,Data Source / Parsed +date,date,TRUE,Author date (when the original author made the commit).,Data Source / Parsed +eventtype,keyword,TRUE,COMMIT.,GrimoireLab +file_dir_name,keyword,TRUE,"Path in which the file is located, not including file name.",Data Source / Parsed +file_ext,keyword,TRUE,File extension.,Data Source / Parsed +file_name,keyword,TRUE,File name with extension.,Data Source / Parsed +file_path_list,keyword,TRUE,List of split path parts.,Data Source / Parsed +fileaction,keyword,TRUE,Action performed by the commit over the file.,Data Source / Parsed +filepath,keyword,TRUE,Complete file path.,Data Source / Parsed +files,long,TRUE,Number of files touched by the same commit this file is included in.,Data Source / Parsed +filetype,keyword,TRUE,"Code or Other, based on file extension.",GrimoireLab +git_author_domain,keyword,TRUE,"Domain extracted from author email included within the commit, if any.",Data Source / Parsed +grimoire_creation_date,date,TRUE,Author date (when the original author made the commit).,Data Source / Parsed +hash,keyword,TRUE,Commit hash.,Data Source / Parsed +id,keyword,TRUE,Commit hash.,Data Source / Parsed +message,text,FALSE,Commit message split by terms.,Data Source / Parsed +message.keyword,keyword,TRUE,Commit message as a single String.,Data Source / Parsed +metadata__enriched_on,date,TRUE,Date when the item was enriched.,GrimoireLab +metadata__timestamp,date,TRUE,Date when the item was stored in RAW index.,GrimoireLab +metadata__updated_on,date,TRUE,Date when the item was updated in its original data source.,GrimoireLab +owner,keyword,TRUE,Owner (code author) name as it appears in commit (including e-mail).,Data Source / Parsed +perceval_uuid,keyword,TRUE,Perceval UUID.,GrimoireLab +project,keyword,TRUE,Project.,GrimoireLab +project_1,keyword,TRUE,Project (if more than one level is allowed in project hierarchy).,GrimoireLab +removedlines,long,TRUE,"Number of lines removed in this file by this commit (only in this file, not the whole commit).",GrimoireLab +repository,keyword,TRUE,Repository name.,Data Source / Parsed +uuid,keyword,TRUE,Item unique identifier. Same as '_id',GrimoireLab \ No newline at end of file diff --git a/schema/github2_pull_requests.csv b/schema/github2_pull_requests.csv index e29f39738..f26253d13 100644 --- a/schema/github2_pull_requests.csv +++ b/schema/github2_pull_requests.csv @@ -1,98 +1,96 @@ -name,type,aggregatable,description -,NA,NA,"Fields coming from cross references study (available only when it is active), see cross_references.csv." -assignee_geolocation,geo_point,true,"Pull request assignee geolocation from GitHub." -author_bot,boolean,true,"True/False if the Pull request author is a bot or not from SortingHat profile." -author_domain,keyword,true,"Pull request author domain name from SortingHat profile." -author_gender,keyword,true,"Pull request author gender, based on her name, from SortingHat (disabled by default)." -author_gender_acc,float,true,"Pull request author gender accuracy from SortingHat (disabled by default)." -author_id,keyword,true,"Pull request author ID from SortingHat profile." -author_name,keyword,true,"Pull request author name from SortingHat profile." -author_org_name,keyword,true,"Pull request author organization name from SortingHat profile." -author_user_name,keyword,true,"Pull request author username from SortingHat profile." -author_uuid,keyword,true,"Pull request author UUID from SortingHat profile." -body,keyword,true,"Body of the pull request/comment." -body_analyzed,text,false,"Body of the pull request/comment." -code_merge_duration,float,true,"Difference in days between creation and merging dates." -comment_created_at,date,true,"Date when the comment was created." -comment_updated_at,date,true,"Date when the comment was updated." -demography_max_date,date,true,"Date of the latest pull request of the corresponding author. Available only when demography study is active." -demography_min_date,date,true,"Date of the first (oldest) pull request of the corresponding author. Available only when demography study is active." -forks,long,true,"Number of repository forks." -github_repo,keyword,true,"GitHub repository name." -grimoire_creation_date,date,true,"Pull request/comment creation date." -id,keyword,true,"Pull request/comment ID." -is_github_pull_request,long,true,"Used to separate pull requests from comments." -is_github_review_comment,long,true,"Used to separate pull requests from comments." -is_github_comment,long,true,"Used to unify pull request and issue comments." -item_type,keyword,true,"The type of the item (pull request/comment)." -merge_author_domain,keyword,true,"Merge author domain from GitHub." -merge_author_geolocation,geo_point,true,"Merge author geolocation from GitHub." -merge_author_location,keyword,true,"Merge author location as string from GitHub." -merge_author_login,keyword,true,"Merge author login from GitHub." -merge_author_name,keyword,true,"Merge author name from GitHub." -merge_author_org,keyword,true,"Merge author organization from GitHub." -merged_by_data_bot,boolean,true,"True/False if the merge author is a bot or not, from SortingHat profile." -merged_by_data_domain,keyword,true,"Merge author domain from SortingHat profile." -merged_by_data_gender,keyword,true,"Merge author gender, based on her name, from SortingHat profile(disabled by default)." -merged_by_data_gender_acc,float,true,"Merge author gender accuracy from SortingHat profile(disabled by default)." -merged_by_data_id,keyword,true,"Merge author's ID from SortingHat profile." -merged_by_data_name,keyword,true,"Merge author name from SortingHat profile." -merged_by_data_org_name,keyword,true,"Merge author organization from SortingHat profile." -merged_by_data_user_name,keyword,true,"Merge author username from SortingHat profile." -merged_by_data_uuid,keyword,true,"Merge author UUID from SortingHat profile." -metadata__enriched_on,date,true,"Date when the item was enriched." -metadata__gelk_backend_name,keyword,true,"Name of the backend used to enrich information." -metadata__gelk_version,keyword,true,"Version of the backend used to enrich information." -metadata__timestamp,date,true,"Date when the item was stored in RAW index." -metadata__updated_on,date,true,"Date when the item was updated on its original data source." -num_review_comments,long,true,"Number of review comments." -origin,keyword,true,"Original URL where the repository was retrieved from." -project,keyword,true,"Project." -project_1,keyword,true,"Project (if more than one level is allowed in project hierarchy)." -pull_closed_at,date,true,"Date in which the Issue was closed." -pull_created_at,date,true,"Date in which the Issue was created." -pull_id,keyword,true,"Pull request ID on GitHub." -pull_id_in_repo,keyword,true,"Pull request ID in the GitHub repository." -pull_labels,keyword,true,"Pull request assigned labels." -pull_merged,boolean,true,"True if the pull request was already merged." -pull_merged_at,date,true,"Date when the pull request was merged." -pull_state,keyword,true,"State of the pull request (open/closed/merged)." -pull_updated_at,date,true,"Date when the pull request was last updated." -pull_url,keyword,true,"Full URL of the pull request." -reaction_confused,long,true,"Number of reactions 'confused'." -reaction_eyes,long,true,"Number of reactions 'eyes'." -reaction_heart,long,true,"Number of reactions 'heart'." -reaction_hooray,long,true,"Number of reactions 'hooray'." -reaction_rocket,long,true,"Number of reactions 'rocket'." -reaction_thumb_down,long,true,"Number of reactions '-1'." -reaction_thumb_up,long,true,"Number of reactions '+1'." -reaction_total_count,long,true,"Number of total reactions." -repository,keyword,true,"Repository name." -repository_labels,keyword,true,"Custom repository labels defined by the user." -review_state,keyword,true,"Review type APPROVED, COMMENTED, CHANGES_REQUESTED, or empty." -sub_type,keyword,true,"Type of the comment (pull review comment)." -tag,keyword,true,"Perceval tag." -time_open_days,float,true,"Time the pull request is open counted in days." -time_to_close_days,float,true,"Time to close a pull request counted in days." -time_to_merge_request_response,float,true,"Time to merge a pull request in days." -issue_title,keyword,true,"The title of the pull request." -issue_title_analyzed,text,false,"Pull request title split by terms to allow searching." -issue_id_in_repo,keyword,true,"The issue's ID in the repository." -issue_url,keyword,true,"Full URL of the issue." -url,keyword,true,"Url of the pull request/comment." -user_data_bot,boolean,true,"True/False if the Pull Request author is a bot or not from SortingHat profile." -user_data_domain,keyword,true,"Pull request author domain name from SortingHat profile." -user_data_gender,keyword,true,"Pull Request author gender, based on her name, from SortingHat (disabled by default)." -user_data_gender_acc,float,true,"Pull request author gender accuracy from SortingHat (disabled by default)." -user_data_id,keyword,true,"Pull request author ID from SortingHat profile." -user_data_name,keyword,true,"Pull request author name from SortingHat profile." -user_data_org_name,keyword,true,"Pull request author organization name from SortingHat profile." -user_data_user_name,keyword,true,"Pull request author username from SortingHat profile." -user_data_uuid,keyword,true,"Pull request author UUID from SortingHat profile." -user_domain,keyword,true,"Pull request author domain name from GitHub." -user_geolocation,geo_point,true,"Pull request author geolocation from GitHub." -user_location,keyword,true,"Pull request author location as string from GitHub." -user_login,keyword,true,"Pull request author login from GitHub." -user_name,keyword,true,"Pull request author username from GitHub." -user_org,keyword,true,"Pull request author organization from GitHub." -uuid,keyword,true,"Perceval UUID." +name,type,aggregatable,description,Raw / Calculated +,NA,NA,"Fields coming from cross references study (available only when it is active), see cross_references.csv.", +assignee_geolocation,geo_point,TRUE,Pull request assignee geolocation from GitHub.,Data Source / Parsed +author_bot,boolean,TRUE,True/False if the Pull request author is a bot or not from SortingHat profile.,GrimoireLab / SortingHat +author_domain,keyword,TRUE,Pull request author domain name from SortingHat profile.,GrimoireLab / SortingHat +author_gender,keyword,TRUE,"Pull request author gender, based on her name, from SortingHat (disabled by default).",GrimoireLab / SortingHat +author_gender_acc,float,TRUE,Pull request author gender accuracy from SortingHat (disabled by default).,GrimoireLab / SortingHat +author_id,keyword,TRUE,Pull request author ID from SortingHat profile.,GrimoireLab / SortingHat +author_name,keyword,TRUE,Pull request author name from SortingHat profile.,GrimoireLab / SortingHat +author_org_name,keyword,TRUE,Pull request author organization name from SortingHat profile.,GrimoireLab / SortingHat +author_user_name,keyword,TRUE,Pull request author username from SortingHat profile.,GrimoireLab / SortingHat +author_uuid,keyword,TRUE,Pull request author UUID from SortingHat profile.,GrimoireLab / SortingHat +body,keyword,TRUE,Body of the pull request/comment.,Data Source / Parsed +body_analyzed,text,FALSE,Body of the pull request/comment.,Data Source / Parsed +code_merge_duration,float,TRUE,Difference in days between creation and merging dates.,GrimoireLab +comment_updated_at,date,TRUE,Date when the comment was updated.,Data Source / Parsed +demography_max_date,date,TRUE,Date of the latest pull request of the corresponding author. Available only when demography study is active.,GrimoireLab +demography_min_date,date,TRUE,Date of the first (oldest) pull request of the corresponding author. Available only when demography study is active.,GrimoireLab +forks,long,TRUE,Number of repository forks.,Data Source / Parsed +github_repo,keyword,TRUE,GitHub repository name.,Data Source / Parsed +grimoire_creation_date,date,TRUE,Pull request/comment creation date.,Data Source / Parsed +id,keyword,TRUE,Pull request/comment ID.,Data Source / Parsed +is_github_pull_request,long,TRUE,Used to separate pull requests from comments.,GrimoireLab +is_github_review_comment,long,TRUE,Used to separate pull requests from comments.,GrimoireLab +is_github_comment,long,TRUE,Used to unify pull request and issue comments.,GrimoireLab +item_type,keyword,TRUE,The type of the item (pull request/comment).,Data Source / Parsed +merge_author_domain,keyword,TRUE,Merge author domain from GitHub.,Data Source / Parsed +merge_author_geolocation,geo_point,TRUE,Merge author geolocation from GitHub.,Data Source / Parsed +merge_author_location,keyword,TRUE,Merge author location as string from GitHub.,Data Source / Parsed +merge_author_login,keyword,TRUE,Merge author login from GitHub.,Data Source / Parsed +merge_author_name,keyword,TRUE,Merge author name from GitHub.,Data Source / Parsed +merge_author_org,keyword,TRUE,Merge author organization from GitHub.,Data Source / Parsed +merged_by_data_bot,boolean,TRUE,"True/False if the merge author is a bot or not, from SortingHat profile.",GrimoireLab / SortingHat +merged_by_data_domain,keyword,TRUE,Merge author domain from SortingHat profile.,GrimoireLab / SortingHat +merged_by_data_gender,keyword,TRUE,"Merge author gender, based on her name, from SortingHat profile(disabled by default).",GrimoireLab / SortingHat +merged_by_data_gender_acc,float,TRUE,Merge author gender accuracy from SortingHat profile(disabled by default).,GrimoireLab / SortingHat +merged_by_data_id,keyword,TRUE,Merge author's ID from SortingHat profile.,GrimoireLab / SortingHat +merged_by_data_name,keyword,TRUE,Merge author name from SortingHat profile.,GrimoireLab / SortingHat +merged_by_data_org_name,keyword,TRUE,Merge author organization from SortingHat profile.,GrimoireLab / SortingHat +merged_by_data_user_name,keyword,TRUE,Merge author username from SortingHat profile.,GrimoireLab / SortingHat +merged_by_data_uuid,keyword,TRUE,Merge author UUID from SortingHat profile.,GrimoireLab / SortingHat +metadata__enriched_on,date,TRUE,Date when the item was enriched.,GrimoireLab +metadata__gelk_backend_name,keyword,TRUE,Name of the backend used to enrich information.,GrimoireLab +metadata__gelk_version,keyword,TRUE,Version of the backend used to enrich information.,GrimoireLab +metadata__timestamp,date,TRUE,Date when the item was stored in RAW index.,GrimoireLab +metadata__updated_on,date,TRUE,Date when the item was updated on its original data source.,GrimoireLab +num_review_comments,long,TRUE,Number of review comments.,GrimoireLab +origin,keyword,TRUE,Original URL where the repository was retrieved from.,Data Source / Parsed +project,keyword,TRUE,Project.,GrimoireLab +project_1,keyword,TRUE,Project (if more than one level is allowed in project hierarchy).,GrimoireLab +pull_closed_at,date,TRUE,Date in which the Issue was closed.,Data Source / Parsed +pull_created_at,date,TRUE,Date in which the Issue was created.,Data Source / Parsed +pull_id,keyword,TRUE,Pull request ID on GitHub.,Data Source / Parsed +pull_id_in_repo,keyword,TRUE,Pull request ID in the GitHub repository.,Data Source / Parsed +pull_labels,keyword,TRUE,Pull request assigned labels.,Data Source / Parsed +pull_merged,boolean,TRUE,True if the pull request was already merged.,Data Source / Parsed +pull_merged_at,date,TRUE,Date when the pull request was merged.,Data Source / Parsed +pull_state,keyword,TRUE,State of the pull request (open/closed/merged).,Data Source / Parsed +pull_updated_at,date,TRUE,Date when the pull request was last updated.,Data Source / Parsed +pull_url,keyword,TRUE,Full URL of the pull request.,Data Source / Parsed +reaction_confused,long,TRUE,Number of reactions 'confused'.,GrimoireLab +reaction_eyes,long,TRUE,Number of reactions 'eyes'.,GrimoireLab +reaction_heart,long,TRUE,Number of reactions 'heart'.,GrimoireLab +reaction_hooray,long,TRUE,Number of reactions 'hooray'.,GrimoireLab +reaction_rocket,long,TRUE,Number of reactions 'rocket'.,GrimoireLab +reaction_thumb_down,long,TRUE,Number of reactions '-1'.,GrimoireLab +reaction_thumb_up,long,TRUE,Number of reactions '+1'.,GrimoireLab +reaction_total_count,long,TRUE,Number of total reactions.,GrimoireLab +repository,keyword,TRUE,Repository name.,Data Source / Parsed +repository_labels,keyword,TRUE,Custom repository labels defined by the user.,Data Source / Parsed +sub_type,keyword,TRUE,Type of the comment (pull review comment).,Data Source / Parsed +tag,keyword,TRUE,Perceval tag.,Data Source / Parsed +time_open_days,float,TRUE,Time the pull request is open counted in days.,GrimoireLab +time_to_close_days,float,TRUE,Time to close a pull request counted in days.,GrimoireLab +time_to_merge_request_response,float,TRUE,Time to merge a pull request in days.,GrimoireLab +issue_title,keyword,TRUE,The title of the pull request.,Data Source / Parsed +issue_title_analyzed,text,FALSE,Pull request title split by terms to allow searching.,Data Source / Parsed +issue_id_in_repo,keyword,TRUE,The issue's ID in the repository.,Data Source / Parsed +issue_url,keyword,TRUE,Full URL of the issue.,Data Source / Parsed +url,keyword,TRUE,Url of the pull request/comment.,Data Source / Parsed +user_data_bot,boolean,TRUE,True/False if the Pull Request author is a bot or not from SortingHat profile.,GrimoireLab / SortingHat +user_data_domain,keyword,TRUE,Pull request author domain name from SortingHat profile.,GrimoireLab / SortingHat +user_data_gender,keyword,TRUE,"Pull Request author gender, based on her name, from SortingHat (disabled by default).",GrimoireLab / SortingHat +user_data_gender_acc,float,TRUE,Pull request author gender accuracy from SortingHat (disabled by default).,GrimoireLab / SortingHat +user_data_id,keyword,TRUE,Pull request author ID from SortingHat profile.,GrimoireLab / SortingHat +user_data_name,keyword,TRUE,Pull request author name from SortingHat profile.,GrimoireLab / SortingHat +user_data_org_name,keyword,TRUE,Pull request author organization name from SortingHat profile.,GrimoireLab / SortingHat +user_data_user_name,keyword,TRUE,Pull request author username from SortingHat profile.,GrimoireLab / SortingHat +user_data_uuid,keyword,TRUE,Pull request author UUID from SortingHat profile.,GrimoireLab / SortingHat +user_domain,keyword,TRUE,Pull request author domain name from GitHub.,Data Source / Parsed +user_geolocation,geo_point,TRUE,Pull request author geolocation from GitHub.,Data Source / Parsed +user_location,keyword,TRUE,Pull request author location as string from GitHub.,Data Source / Parsed +user_login,keyword,TRUE,Pull request author login from GitHub.,Data Source / Parsed +user_name,keyword,TRUE,Pull request author username from GitHub.,Data Source / Parsed +user_org,keyword,TRUE,Pull request author organization from GitHub.,Data Source / Parsed +uuid,keyword,TRUE,Perceval UUID.,GrimoireLab \ No newline at end of file diff --git a/schema/github_backlog.csv b/schema/github_backlog.csv index 0548165f3..88e875461 100644 --- a/schema/github_backlog.csv +++ b/schema/github_backlog.csv @@ -1,13 +1,13 @@ -name,type,aggregatable,description -average_opened_time,long,true,"Average of issue opened time." -grimoire_creation_date,date,true,"Issue creation date." -interval_days,long,true,"Number of days between each analysis." -is_github_stats,boolean,true,"True if is Github Stats." -labels,keyword,true,"The labels assigned to an issue." -metadata__enriched_on,date,true,"Date when the data were enriched." -opened,long,true,"Number of issues opens." -organization,keyword,true,"Organization name of the repository uthor" -origin,keyword,true,"The original URL from which the repository was retrieved from." -project,keyword,true,"Project name." -study_creation_date,date,true,"Date when the study is created." -uuid,keyword,true,"Perceval UUID." +name,type,aggregatable,description,Raw / Calculated +average_opened_time,long,TRUE,Average of issue opened time.,GrimoireLab +grimoire_creation_date,date,TRUE,Issue creation date.,Data Source / Parsed +interval_days,long,TRUE,Number of days between each analysis.,GrimoireLab +is_github_stats,boolean,TRUE,True if is Github Stats.,GrimoireLab +labels,keyword,TRUE,The labels assigned to an issue.,Data Source / Parsed +metadata__enriched_on,date,TRUE,Date when the data were enriched.,GrimoireLab +opened,long,TRUE,Number of issues opens.,GrimoireLab +organization,keyword,TRUE,Organization name of the repository uthor,Data Source / Parsed +origin,keyword,TRUE,The original URL from which the repository was retrieved from.,Data Source / Parsed +project,keyword,TRUE,Project name.,GrimoireLab +study_creation_date,date,TRUE,Date when the study is created.,GrimoireLab +uuid,keyword,TRUE,Perceval UUID.,GrimoireLab \ No newline at end of file diff --git a/schema/github_events.csv b/schema/github_events.csv index 6fb0ee924..3b1af3dc1 100644 --- a/schema/github_events.csv +++ b/schema/github_events.csv @@ -1,117 +1,117 @@ -name,type,aggregatable,description -,NA,NA,"Fields coming from cross references study (available only when it is active), see cross_references.csv." -actor_bot,boolean,true,"True/False if the event actor is a bot or not from SortingHat profile." -actor_domain,keyword,true,"Event actor domain name from SortingHat profile." -actor_gender,keyword,true,"Event actor gender, based on her name, from SortingHat (disabled by default)." -actor_gender_acc,float,true,"Event actor gender accuracy from SortingHat (disabled by default)." -actor_id,keyword,true,"Event actor ID from SortingHat profile." -actor_name,keyword,true,"Event actor name from SortingHat profile." -actor_org_name,keyword,true,"Event actor organization name from SortingHat profile." -actor_multi_org_names,keyword,true,"List of the actor organizations from SortingHat profile." -actor_user_name,keyword,true,"Event actor username from SortingHat profile." -actor_uuid,keyword,true,"Event actor UUID from SortingHat profile." -author_bot,boolean,true,"True/False if the event author is a bot or not from SortingHat profile." -author_domain,keyword,true,"Event actor domain name from SortingHat profile." -author_gender,keyword,true,"Event actor gender, based on her name, from SortingHat (disabled by default)." -author_gender_acc,float,true,"Event actor gender accuracy from SortingHat (disabled by default)." -author_id,keyword,true,"Event actor ID from SortingHat profile." -author_name,keyword,true,"Event actor name from SortingHat profile." -author_org_name,keyword,true,"Event actor organization name from SortingHat profile." -author_multi_org_names,keyword,true,"List of the actor organizations from SortingHat profile." -author_user_name,keyword,true,"Event actor username from SortingHat profile." -author_uuid,keyword,true,"Event actor UUID from SortingHat profile." -board_closed_at,date,true,"Date when the board was closed." -board_column,keyword,true,"Name of the column where the issue is." -board_created_at,date,true,"Date when the board was created." -board_name,keyword,true,"Name of the board." -board_previous_column,keyword,true,"Name of the previous column where the issue was." -board_state,keyword,true,"State of the board (open/closed)." -board_updated_at,date,true,"Date when the board was updated." -board_url,keyword,true,"URL of the board." -closer_closed,boolean,true,"True/False if the pull request is closed or not." -closer_closed_at,date,true,"Date when the pull request was closed." -closer_created_at,date,true,"Date when the pull request was created." -closer_event_url,keyword,true,"URL where the closed event occurred." -closer_event_url,boolean,true,"True/False if the pull request is merged or not." -closer_number,long,true,"Number of the pull request." -closer_pull_submitter,keyword,true,"Pull request submitter login." -closer_repo,keyword,true,"Repository of the pull request." -closer_type,keyword,true,"Type of the closer object (currently only pull requests are processed)." -closer_updated_at,date,true,"Date when the pull request was updated." -closer_url,keyword,true,"URL of the pull request." -created_at,date,true,"Date when the event was created." -duration_from_previous_event,long,true,"Duration in days from the previous event (added by duration analysis study)." -event_type,keyword,true,"Type of the event processed." -github_repo,keyword,true,"GitHub repository name of the issue." -grimoire_creation_date,date,true,"Pull Request creation date." -is_github_issue,long,true,"Used to separate issues from other items such as pull requests." -issue_closed_at,date,true,"Date when the issue was closed." -issue_created_at,date,true,"Date when the issue was opened." -issue_id,long,true,"Issue's ID in GitHub." -id_in_repo,keyword,true,"The issue's ID in the repository it was created." -issue_labels,keyword,true,"The labels assigned to an issue." -issue_state,keyword,true,"State of the item (open/closed)." -issue_updated_at,date,true,"Date when the issue was last updated." -issue_url_id,keyword,true,"Consists of the project path and the issue's id." -issue_url,keyword,true,"Full URL of the issue." -item_type,keyword,true,"The type of the item (issue/pull request)." -label,keyword,true,"Label name." -label_description,keyword,true,"Label description." -label_created_at,date,true,"Date when the label was created." -label_is_default,boolean,true,"True/False whether the label is a GitHub default one." -label_updated_at,date,true,"Date when the label was updated." -merge_closed,boolean,true,"True/False if the pull request is closed or not." -merge_closed_at,date,true,"Date when the pull request was closed." -merge_created_at,date,true,"Date when the pull request was created." -merge_merged,boolean,true,"True/False if the pull request is merged or not." -merge_merged_at,date,true,"Date when the pull request was merged." -merge_updated_at,date,true,"Date when the pull request was updated." -merge_url,string,true,"URL of the pull request." -metadata__enriched_on,date,true,"Date when the data were enriched." -metadata__gelk_backend_name,keyword,true,"Name of the backend used to enrich the data." -metadata__gelk_version,keyword,true,"Version of the backend used to enrich the data." -metadata__timestamp,date,true,"Date when the item was stored in ElasticSearch raw index." -metadata__updated_on,date,true,"Date when the item was updated on its original data source." -origin,keyword,true,"The original URL from which the repository was retrieved from." -previous_event_uuid,keyword,true,"Previous event uuid (added by duration analysis study)." -project_1,keyword,true,"Used if more than one project levels are allowed in the project hierarchy." -project,keyword,true,"Project name." -pull_request,boolean,true,"True/False if the item is a pull request or not." -reference_will_close_target,boolean,true,"True/False if the target will be closed when the source is merged." -reference_cross_repo,boolean,true,"True/False if the reference originated in a different repository." -reference_event_url,keyword,true,"URL of the reference event." -reference_source_closed,boolean,true,"True/False if the source is closed or not." -reference_source_closed_at,date,true,"Date when the source was closed." -reference_source_created_at,date,true,"Date when the source was created." -reference_source_merged,boolean,true,"True/False if the source was merged or not." -reference_source_number,long,true,"Number of the source." -reference_source_repo,keyword,true,"Repo of the source." -reference_source_type,keyword,true,"Type of the source (issue/pull request)." -reference_source_updated_at,date,true,"Date when the source was updated." -reference_source_url,keyword,true,"URL of the source." -reporter_bot,boolean,true,"True/False if the issue reporter is a bot or not from SortingHat profile." -reporter_domain,keyword,true,"Issue reporter domain name from SortingHat profile." -reporter_gender,keyword,true,"Issue reporter gender, based on her name, from SortingHat (disabled by default)." -reporter_gender_acc,float,true,"Issue reporter gender accuracy from SortingHat (disabled by default)." -reporter_id,keyword,true,"Issue reporter ID from SortingHat profile." -reporter_name,keyword,true,"Issue reporter name from SortingHat profile." -reporter_org_name,keyword,true,"Issue reporter organization name from SortingHat profile." -reporter_multi_org_names,keyword,true,"List of the reporter organizations from SortingHat profile." -reporter_user_name,keyword,true,"Issue reporter username from SortingHat profile." -reporter_uuid,keyword,true,"Issue reporter UUID from SortingHat profile." -repository,keyword,true,"Repository name." -repository_labels,keyword,true,"Custom repository labels defined by the user." -submitter_bot,boolean,true,"True/False if the pull request submitter is a bot or not from SortingHat profile." -submitter_domain,keyword,true,"Pull request submitter domain name from SortingHat profile." -submitter_gender,keyword,true,"Pull request submitter gender, based on her name, from SortingHat (disabled by default)." -submitter_gender_acc,float,true,"Pull request submitter gender accuracy from SortingHat (disabled by default)." -submitter_id,keyword,true,"Pull request submitter ID from SortingHat profile." -submitter_name,keyword,true,"Pull request submitter name from SortingHat profile." -submitter_org_name,keyword,true,"Pull request submitter organization name from SortingHat profile." -submitter_multi_org_names,keyword,true,"List of the pull request submitter organizations from SortingHat profile." -submitter_user_name,keyword,true,"Pull request submitter username from SortingHat profile." -submitter_uuid,keyword,true,"Pull request submitter UUID from SortingHat profile." -tag,keyword,true,"Perceval tag." -title,keyword,true,"The title of the Pull Request." -title_analyzed,text,false,"Pull Request title split by terms to allow searching." -uuid,keyword,true,"Perceval UUID." +name,type,aggregatable,description,Raw / Calculated +,NA,NA,"Fields coming from cross references study (available only when it is active), see cross_references.csv.", +actor_bot,boolean,TRUE,True/False if the event actor is a bot or not from SortingHat profile.,GrimoireLab / SortingHat +actor_domain,keyword,TRUE,Event actor domain name from SortingHat profile.,GrimoireLab / SortingHat +actor_gender,keyword,TRUE,"Event actor gender, based on her name, from SortingHat (disabled by default).",GrimoireLab / SortingHat +actor_gender_acc,float,TRUE,Event actor gender accuracy from SortingHat (disabled by default).,GrimoireLab / SortingHat +actor_id,keyword,TRUE,Event actor ID from SortingHat profile.,GrimoireLab / SortingHat +actor_name,keyword,TRUE,Event actor name from SortingHat profile.,GrimoireLab / SortingHat +actor_org_name,keyword,TRUE,Event actor organization name from SortingHat profile.,GrimoireLab / SortingHat +actor_multi_org_names,keyword,TRUE,List of the actor organizations from SortingHat profile.,GrimoireLab / SortingHat +actor_user_name,keyword,TRUE,Event actor username from SortingHat profile.,GrimoireLab / SortingHat +actor_uuid,keyword,TRUE,Event actor UUID from SortingHat profile.,GrimoireLab / SortingHat +author_bot,boolean,TRUE,True/False if the event author is a bot or not from SortingHat profile.,GrimoireLab / SortingHat +author_domain,keyword,TRUE,Event actor domain name from SortingHat profile.,GrimoireLab / SortingHat +author_gender,keyword,TRUE,"Event actor gender, based on her name, from SortingHat (disabled by default).",GrimoireLab / SortingHat +author_gender_acc,float,TRUE,Event actor gender accuracy from SortingHat (disabled by default).,GrimoireLab / SortingHat +author_id,keyword,TRUE,Event actor ID from SortingHat profile.,GrimoireLab / SortingHat +author_name,keyword,TRUE,Event actor name from SortingHat profile.,GrimoireLab / SortingHat +author_org_name,keyword,TRUE,Event actor organization name from SortingHat profile.,GrimoireLab / SortingHat +author_multi_org_names,keyword,TRUE,List of the actor organizations from SortingHat profile.,GrimoireLab / SortingHat +author_user_name,keyword,TRUE,Event actor username from SortingHat profile.,GrimoireLab / SortingHat +author_uuid,keyword,TRUE,Event actor UUID from SortingHat profile.,GrimoireLab / SortingHat +board_closed_at,date,TRUE,Date when the board was closed.,Data Source / Parsed +board_column,keyword,TRUE,Name of the column where the issue is.,Data Source / Parsed +board_created_at,date,TRUE,Date when the board was created.,Data Source / Parsed +board_name,keyword,TRUE,Name of the board.,Data Source / Parsed +board_previous_column,keyword,TRUE,Name of the previous column where the issue was.,Data Source / Parsed +board_state,keyword,TRUE,State of the board (open/closed).,Data Source / Parsed +board_updated_at,date,TRUE,Date when the board was updated.,Data Source / Parsed +board_url,keyword,TRUE,URL of the board.,Data Source / Parsed +closer_closed,boolean,TRUE,True/False if the pull request is closed or not.,Data Source / Parsed +closer_closed_at,date,TRUE,Date when the pull request was closed.,Data Source / Parsed +closer_created_at,date,TRUE,Date when the pull request was created.,Data Source / Parsed +closer_event_url,keyword,TRUE,URL where the closed event occurred.,Data Source / Parsed +closer_event_url,boolean,TRUE,True/False if the pull request is merged or not.,Data Source / Parsed +closer_number,long,TRUE,Number of the pull request.,Data Source / Parsed +closer_pull_submitter,keyword,TRUE,Pull request submitter login.,Data Source / Parsed +closer_repo,keyword,TRUE,Repository of the pull request.,Data Source / Parsed +closer_type,keyword,TRUE,Type of the closer object (currently only pull requests are processed).,Data Source / Parsed +closer_updated_at,date,TRUE,Date when the pull request was updated.,Data Source / Parsed +closer_url,keyword,TRUE,URL of the pull request.,Data Source / Parsed +created_at,date,TRUE,Date when the event was created.,Data Source / Parsed +duration_from_previous_event,long,TRUE,Duration in days from the previous event (added by duration analysis study).,GrimoireLab +event_type,keyword,TRUE,Type of the event processed.,Data Source / Parsed +github_repo,keyword,TRUE,GitHub repository name of the issue.,Data Source / Parsed +grimoire_creation_date,date,TRUE,Pull Request creation date.,Data Source / Parsed +is_github_issue,long,TRUE,Used to separate issues from other items such as pull requests.,Data Source / Parsed +issue_closed_at,date,TRUE,Date when the issue was closed.,Data Source / Parsed +issue_created_at,date,TRUE,Date when the issue was opened.,Data Source / Parsed +issue_id,long,TRUE,Issue's ID in GitHub.,Data Source / Parsed +id_in_repo,keyword,TRUE,The issue's ID in the repository it was created.,Data Source / Parsed +issue_labels,keyword,TRUE,The labels assigned to an issue.,Data Source / Parsed +issue_state,keyword,TRUE,State of the item (open/closed).,Data Source / Parsed +issue_updated_at,date,TRUE,Date when the issue was last updated.,Data Source / Parsed +issue_url_id,keyword,TRUE,Consists of the project path and the issue's id.,Data Source / Parsed +issue_url,keyword,TRUE,Full URL of the issue.,Data Source / Parsed +item_type,keyword,TRUE,The type of the item (issue/pull request).,Data Source / Parsed +label,keyword,TRUE,Label name.,Data Source / Parsed +label_description,keyword,TRUE,Label description.,Data Source / Parsed +label_created_at,date,TRUE,Date when the label was created.,Data Source / Parsed +label_is_default,boolean,TRUE,True/False whether the label is a GitHub default one.,Data Source / Parsed +label_updated_at,date,TRUE,Date when the label was updated.,Data Source / Parsed +merge_closed,boolean,TRUE,True/False if the pull request is closed or not.,Data Source / Parsed +merge_closed_at,date,TRUE,Date when the pull request was closed.,Data Source / Parsed +merge_created_at,date,TRUE,Date when the pull request was created.,Data Source / Parsed +merge_merged,boolean,TRUE,True/False if the pull request is merged or not.,Data Source / Parsed +merge_merged_at,date,TRUE,Date when the pull request was merged.,Data Source / Parsed +merge_updated_at,date,TRUE,Date when the pull request was updated.,Data Source / Parsed +merge_url,string,TRUE,URL of the pull request.,Data Source / Parsed +metadata__enriched_on,date,TRUE,Date when the data were enriched.,Data Source / Parsed +metadata__gelk_backend_name,keyword,TRUE,Name of the backend used to enrich the data.,GrimoireLab +metadata__gelk_version,keyword,TRUE,Version of the backend used to enrich the data.,GrimoireLab +metadata__timestamp,date,TRUE,Date when the item was stored in ElasticSearch raw index.,GrimoireLab +metadata__updated_on,date,TRUE,Date when the item was updated on its original data source.,GrimoireLab +origin,keyword,TRUE,The original URL from which the repository was retrieved from.,Data Source / Parsed +previous_event_uuid,keyword,TRUE,Previous event uuid (added by duration analysis study).,GrimoireLab +project_1,keyword,TRUE,Used if more than one project levels are allowed in the project hierarchy.,GrimoireLab +project,keyword,TRUE,Project name.,GrimoireLab +pull_request,boolean,TRUE,True/False if the item is a pull request or not.,GrimoireLab +reference_will_close_target,boolean,TRUE,True/False if the target will be closed when the source is merged.,GrimoireLab +reference_cross_repo,boolean,TRUE,True/False if the reference originated in a different repository.,GrimoireLab +reference_event_url,keyword,TRUE,URL of the reference event.,Data Source / Parsed +reference_source_closed,boolean,TRUE,True/False if the source is closed or not.,Data Source / Parsed +reference_source_closed_at,date,TRUE,Date when the source was closed.,Data Source / Parsed +reference_source_created_at,date,TRUE,Date when the source was created.,Data Source / Parsed +reference_source_merged,boolean,TRUE,True/False if the source was merged or not.,Data Source / Parsed +reference_source_number,long,TRUE,Number of the source.,Data Source / Parsed +reference_source_repo,keyword,TRUE,Repo of the source.,Data Source / Parsed +reference_source_type,keyword,TRUE,Type of the source (issue/pull request).,Data Source / Parsed +reference_source_updated_at,date,TRUE,Date when the source was updated.,Data Source / Parsed +reference_source_url,keyword,TRUE,URL of the source.,Data Source / Parsed +reporter_bot,boolean,TRUE,True/False if the issue reporter is a bot or not from SortingHat profile.,GrimoireLab / SortingHat +reporter_domain,keyword,TRUE,Issue reporter domain name from SortingHat profile.,GrimoireLab / SortingHat +reporter_gender,keyword,TRUE,"Issue reporter gender, based on her name, from SortingHat (disabled by default).",GrimoireLab / SortingHat +reporter_gender_acc,float,TRUE,Issue reporter gender accuracy from SortingHat (disabled by default).,GrimoireLab / SortingHat +reporter_id,keyword,TRUE,Issue reporter ID from SortingHat profile.,GrimoireLab / SortingHat +reporter_name,keyword,TRUE,Issue reporter name from SortingHat profile.,GrimoireLab / SortingHat +reporter_org_name,keyword,TRUE,Issue reporter organization name from SortingHat profile.,GrimoireLab / SortingHat +reporter_multi_org_names,keyword,TRUE,List of the reporter organizations from SortingHat profile.,GrimoireLab / SortingHat +reporter_user_name,keyword,TRUE,Issue reporter username from SortingHat profile.,GrimoireLab / SortingHat +reporter_uuid,keyword,TRUE,Issue reporter UUID from SortingHat profile.,GrimoireLab / SortingHat +repository,keyword,TRUE,Repository name.,Data Source / Parsed +repository_labels,keyword,TRUE,Custom repository labels defined by the user.,Data Source / Parsed +submitter_bot,boolean,TRUE,True/False if the pull request submitter is a bot or not from SortingHat profile.,GrimoireLab / SortingHat +submitter_domain,keyword,TRUE,Pull request submitter domain name from SortingHat profile.,GrimoireLab / SortingHat +submitter_gender,keyword,TRUE,"Pull request submitter gender, based on her name, from SortingHat (disabled by default).",GrimoireLab / SortingHat +submitter_gender_acc,float,TRUE,Pull request submitter gender accuracy from SortingHat (disabled by default).,GrimoireLab / SortingHat +submitter_id,keyword,TRUE,Pull request submitter ID from SortingHat profile.,GrimoireLab / SortingHat +submitter_name,keyword,TRUE,Pull request submitter name from SortingHat profile.,GrimoireLab / SortingHat +submitter_org_name,keyword,TRUE,Pull request submitter organization name from SortingHat profile.,GrimoireLab / SortingHat +submitter_multi_org_names,keyword,TRUE,List of the pull request submitter organizations from SortingHat profile.,GrimoireLab / SortingHat +submitter_user_name,keyword,TRUE,Pull request submitter username from SortingHat profile.,GrimoireLab / SortingHat +submitter_uuid,keyword,TRUE,Pull request submitter UUID from SortingHat profile.,GrimoireLab / SortingHat +tag,keyword,TRUE,Perceval tag.,GrimoireLab +title,keyword,TRUE,The title of the Pull Request.,Data Source / Parsed +title_analyzed,text,FALSE,Pull Request title split by terms to allow searching.,Data Source / Parsed +uuid,keyword,TRUE,Perceval UUID.,GrimoireLab \ No newline at end of file diff --git a/schema/github_issues.csv b/schema/github_issues.csv index 8fd9d77a8..6ba371138 100644 --- a/schema/github_issues.csv +++ b/schema/github_issues.csv @@ -1,72 +1,72 @@ -name,type,aggregatable,description -,NA,NA,"Fields coming from cross references study (available only when it is active), see cross_references.csv." -assignee_data_bot,boolean,true,"True/False if the assignee is a bot or not." -assignee_data_domain,keyword,true,"Assignee's domain name from SortingHat profile." -assignee_data_id,keyword,true,"Assignee's id from SortingHat profile." -assignee_data_name,keyword,true,"Assignee's name." -assignee_data_org_name,keyword,true,"Assignee's organization name from SortingHat profile." -assignee_multi_org_names,keyword,true,"List of the assignee organizations from SortingHat profile." -assignee_data_user_name,keyword,true,"Assignee's username from SortingHat profile." -assignee_data_uuid,keyword,true,"Assignee's UUID from SortingHat profile." -assignee_domain,keyword,true,"Assignee's domain name from GitHub." -assignee_geolocation,geo_point,true,"Assignee's global location using coordinates." -assignee_location,keyword,true,"Assignee's geographical location." -assignee_login,keyword,true,"Assignee's login name from GitHub." -assignee_name,keyword,true,"Assignee's name from GitHub." -assignee_org,keyword,true,"Assignee's organization name from GitHub." -author_bot,boolean,true,"True/False if the author is a bot or not." -author_domain,keyword,true,"Author's domain name from SortingHat profile." -author_id,keyword,true,"Author's ID from SortingHat profile." -author_name,keyword,true,"Author's name." -author_org_name,keyword,true,"Author's organization name from SortingHat profile." -author_multi_org_names,keyword,true,"List of the author organizations from SortingHat profile." -author_user_name,keyword,true,"Author's username from SortingHat profile." -author_uuid,keyword,true,"Author's UUID from SortingHat profile." -closed_at,date,true,"Date when an issue was closed." -created_at,date,true,"Date when an issue was opened." -demography_max_date,date,true,"Date of the latest issue of the corresponding author. Available only when demography study is active." -demography_min_date,date,true,"Date of the first (oldest) issue of the corresponding author. Available only when demography study is active." -github_repo,keyword,true,"The name of the GitHub repository." -grimoire_creation_date,date,true,"Issue creation date." -id_in_repo,keyword,true,"The issue's ID in the repository it was created." -id,long,true,"Issue's ID in GitHub." -is_github_issue,long,true,"Used to separate issues from other items such as pull requests." -issue_url,keyword,true,"Full URL of the issue." -item_type,keyword,true,"The type of the item (issue/pull request)." -labels,keyword,true,"The labels assigned to an issue." -metadata__enriched_on,date,true,"Date when the data were enriched." -metadata__gelk_backend_name,keyword,true,"Name of the backend used to enrich the data." -metadata__gelk_version,keyword,true,"Version of the backend used to enrich the data." -metadata__timestamp,date,true,"Date when the item was stored in ElasticSearch raw index." -metadata__updated_on,date,true,"Date when the item was updated on its original data source." -origin,keyword,true,"The original URL from which the repository was retrieved from." -project_1,keyword,true,"Used if more than one project levels are allowed in the project hierarchy." -project,keyword,true,"Project name." -pull_request,boolean,true,"True/False if the item is a pull request or not." -repository,keyword,true,"Repository name." -repository_labels,keyword,true,"Custom repository labels defined by the user." -state,keyword,true,"State of the item (open/closed)." -tag,keyword,true,"Perceval tag." -time_open_days,float,true,"Time the item is open counted in days." -time_to_close_days,float,true,"Time to close an issue counted in days." -time_to_first_attention,float,true,"Time to first attention to an issue counted in days." -title_analyzed,text,false,"Issue title split by terms to allow searching." -title,keyword,true,"The title of the issue." -updated_at,date,true,"Date when the issue was last updated." -url_id,keyword,true,"Consists of the project path and the issue's id." -url,keyword,true,"Full URL of the issue." -user_data_bot,boolean,true,"True/False if the user is a bot or not." -user_data_domain,keyword,true,"User's domain name from SortingHat profile." -user_data_id,keyword,true,"User's ID from SortingHat profile." -user_data_name,keyword,true,"User's name from SortingHat profile." -user_data_org_name,keyword,true,"Author's organization name from SortingHat profile." -user_data_multi_org_names,keyword,true,"List of the author organizations from SortingHat profile." -user_data_user_name,keyword,true,"User's username from SortingHat profile." -user_data_uuid,keyword,true,"User's UUID from SortingHat profile." -user_domain,keyword,true,"User's domain name from GitHub." -user_geolocation,geo_point,true,"User's global location using coordinates." -user_location,keyword,true,"User's geographical location." -user_login,keyword,true,"User's login name from GitHub." -user_name,keyword,true,"User's name." -user_org,keyword,true,"User's organization name." -uuid,keyword,true,"Perceval UUID." +name,type,aggregatable,description,Raw / Calculated +,NA,NA,"Fields coming from cross references study (available only when it is active), see cross_references.csv.", +assignee_data_bot,boolean,TRUE,True/False if the assignee is a bot or not.,GrimoireLab / SortingHat +assignee_data_domain,keyword,TRUE,Assignee's domain name from SortingHat profile.,GrimoireLab / SortingHat +assignee_data_id,keyword,TRUE,Assignee's id from SortingHat profile.,GrimoireLab / SortingHat +assignee_data_name,keyword,TRUE,Assignee's name.,GrimoireLab / SortingHat +assignee_data_org_name,keyword,TRUE,Assignee's organization name from SortingHat profile.,GrimoireLab / SortingHat +assignee_multi_org_names,keyword,TRUE,List of the assignee organizations from SortingHat profile.,GrimoireLab / SortingHat +assignee_data_user_name,keyword,TRUE,Assignee's username from SortingHat profile.,GrimoireLab / SortingHat +assignee_data_uuid,keyword,TRUE,Assignee's UUID from SortingHat profile.,GrimoireLab / SortingHat +assignee_domain,keyword,TRUE,Assignee's domain name from GitHub.,Data Source / Parsed +assignee_geolocation,geo_point,TRUE,Assignee's global location using coordinates.,Data Source / Parsed +assignee_location,keyword,TRUE,Assignee's geographical location.,Data Source / Parsed +assignee_login,keyword,TRUE,Assignee's login name from GitHub.,Data Source / Parsed +assignee_name,keyword,TRUE,Assignee's name from GitHub.,Data Source / Parsed +assignee_org,keyword,TRUE,Assignee's organization name from GitHub.,Data Source / Parsed +author_bot,boolean,TRUE,True/False if the author is a bot or not.,GrimoireLab / SortingHat +author_domain,keyword,TRUE,Author's domain name from SortingHat profile.,GrimoireLab / SortingHat +author_id,keyword,TRUE,Author's ID from SortingHat profile.,GrimoireLab / SortingHat +author_name,keyword,TRUE,Author's name.,GrimoireLab / SortingHat +author_org_name,keyword,TRUE,Author's organization name from SortingHat profile.,GrimoireLab / SortingHat +author_multi_org_names,keyword,TRUE,List of the author organizations from SortingHat profile.,GrimoireLab / SortingHat +author_user_name,keyword,TRUE,Author's username from SortingHat profile.,GrimoireLab / SortingHat +author_uuid,keyword,TRUE,Author's UUID from SortingHat profile.,GrimoireLab / SortingHat +closed_at,date,TRUE,Date when an issue was closed.,Data Source / Parsed +created_at,date,TRUE,Date when an issue was opened.,Data Source / Parsed +demography_max_date,date,TRUE,Date of the latest issue of the corresponding author. Available only when demography study is active.,Data Source / Parsed +demography_min_date,date,TRUE,Date of the first (oldest) issue of the corresponding author. Available only when demography study is active.,Data Source / Parsed +github_repo,keyword,TRUE,The name of the GitHub repository.,Data Source / Parsed +grimoire_creation_date,date,TRUE,Issue creation date.,Data Source / Parsed +id_in_repo,keyword,TRUE,The issue's ID in the repository it was created.,Data Source / Parsed +id,long,TRUE,Issue's ID in GitHub.,Data Source / Parsed +is_github_issue,long,TRUE,Used to separate issues from other items such as pull requests.,Data Source / Parsed +issue_url,keyword,TRUE,Full URL of the issue.,Data Source / Parsed +item_type,keyword,TRUE,The type of the item (issue/pull request).,Data Source / Parsed +labels,keyword,TRUE,The labels assigned to an issue.,Data Source / Parsed +metadata__enriched_on,date,TRUE,Date when the data were enriched.,GrimoireLab +metadata__gelk_backend_name,keyword,TRUE,Name of the backend used to enrich the data.,GrimoireLab +metadata__gelk_version,keyword,TRUE,Version of the backend used to enrich the data.,GrimoireLab +metadata__timestamp,date,TRUE,Date when the item was stored in ElasticSearch raw index.,GrimoireLab +metadata__updated_on,date,TRUE,Date when the item was updated on its original data source.,GrimoireLab +origin,keyword,TRUE,The original URL from which the repository was retrieved from.,Data Source / Parsed +project_1,keyword,TRUE,Used if more than one project levels are allowed in the project hierarchy.,GrimoireLab +project,keyword,TRUE,Project name.,GrimoireLab +pull_request,boolean,TRUE,True/False if the item is a pull request or not.,Data Source / Parsed +repository,keyword,TRUE,Repository name.,Data Source / Parsed +repository_labels,keyword,TRUE,Custom repository labels defined by the user.,Data Source / Parsed +state,keyword,TRUE,State of the item (open/closed).,Data Source / Parsed +tag,keyword,TRUE,Perceval tag.,GrimoireLab +time_open_days,float,TRUE,Time the item is open counted in days.,GrimoireLab +time_to_close_days,float,TRUE,Time to close an issue counted in days.,GrimoireLab +time_to_first_attention,float,TRUE,Time to first attention to an issue counted in days.,GrimoireLab +title_analyzed,text,FALSE,Issue title split by terms to allow searching.,Data Source / Parsed +title,keyword,TRUE,The title of the issue.,Data Source / Parsed +updated_at,date,TRUE,Date when the issue was last updated.,Data Source / Parsed +url_id,keyword,TRUE,Consists of the project path and the issue's id.,Data Source / Parsed +url,keyword,TRUE,Full URL of the issue.,Data Source / Parsed +user_data_bot,boolean,TRUE,True/False if the user is a bot or not.,GrimoireLab / SortingHat +user_data_domain,keyword,TRUE,User's domain name from SortingHat profile.,GrimoireLab / SortingHat +user_data_id,keyword,TRUE,User's ID from SortingHat profile.,GrimoireLab / SortingHat +user_data_name,keyword,TRUE,User's name from SortingHat profile.,GrimoireLab / SortingHat +user_data_org_name,keyword,TRUE,Author's organization name from SortingHat profile.,GrimoireLab / SortingHat +user_data_multi_org_names,keyword,TRUE,List of the author organizations from SortingHat profile.,GrimoireLab / SortingHat +user_data_user_name,keyword,TRUE,User's username from SortingHat profile.,GrimoireLab / SortingHat +user_data_uuid,keyword,TRUE,User's UUID from SortingHat profile.,GrimoireLab / SortingHat +user_domain,keyword,TRUE,User's domain name from GitHub.,Data Source / Parsed +user_geolocation,geo_point,TRUE,User's global location using coordinates.,Data Source / Parsed +user_location,keyword,TRUE,User's geographical location.,Data Source / Parsed +user_login,keyword,TRUE,User's login name from GitHub.,Data Source / Parsed +user_name,keyword,TRUE,User's name.,Data Source / Parsed +user_org,keyword,TRUE,User's organization name.,Data Source / Parsed +uuid,keyword,TRUE,Perceval UUID.,GrimoireLab \ No newline at end of file diff --git a/schema/github_pull_requests.csv b/schema/github_pull_requests.csv index 8bc68e989..2ecc6d292 100644 --- a/schema/github_pull_requests.csv +++ b/schema/github_pull_requests.csv @@ -1,85 +1,85 @@ -name,type,aggregatable,description -,NA,NA,"Fields coming from cross references study (available only when it is active), see cross_references.csv." -additions,long,true,"Number of additions in the Pull Request." -assignee_geolocation,geo_point,true,"Pull Request assignee geolocation from GitHub." -author_bot,boolean,true,"True/False if the Pull Request author is a bot or not from SortingHat profile." -author_domain,keyword,true,"Pull Request author domain name from SortingHat profile." -author_gender,keyword,true,"Pull Request author gender, based on her name, from SortingHat (disabled by default)." -author_gender_acc,float,true,"Pull Request author gender accuracy from SortingHat (disabled by default)." -author_id,keyword,true,"Pull Request author ID from SortingHat profile." -author_name,keyword,true,"Pull Request author name from SortingHat profile." -author_org_name,keyword,true,"Pull Request author organization name from SortingHat profile." -author_multi_org_names,keyword,true,"List of the author organizations from SortingHat profile." -author_user_name,keyword,true,"Pull Request author username from SortingHat profile." -author_uuid,keyword,true,"Pull Request author UUID from SortingHat profile." -changed_files,long,true,"Number of changed files in the Pull Request." -closed_at,date,true,"Date in which the Issue was closed." -code_merge_duration,float,true,"Difference in days between creation and merging dates." -created_at,date,true,"Date in which the Issue was created." -deletions,long,true,"Number of deletions in the Pull Request." -forks,long,true,"Number of repository forks." -github_repo,keyword,true,"GitHub repository name." -grimoire_creation_date,date,true,"Pull Request creation date." -id,long,true,"ID in GitHub." -id_in_repo,keyword,true,"ID in the GitHub repository it was created." -issue_url,keyword,true,"Full URL of the issue." -is_github_pull_request,long,true,"1 indicating this is a Pull Request, used for counting in case other kind of items could be stored in the same index or alias." -item_type,keyword,true,"Item type, in this case 'pull request'" -labels,keyword,true,"Pull Request assigned labels." -merge_author_domain,keyword,true,"Merge author domain from GitHub." -merge_author_geolocation,geo_point,true,"Merge author geolocation from GitHub." -merge_author_location,keyword,true,"Merge author location as string from GitHub." -merge_author_login,keyword,true,"Merge author login from GitHub." -merge_author_name,keyword,true,"Merge author name from GitHub." -merge_author_org,keyword,true,"Merge author organization from GitHub." -merged,boolean,true,"True if the Pull Request was already merged." -merged_at,date,true,"Date when the Pull Request was merged." -merged_by_data_bot,boolean,true,"True/False if the merge author is a bot or not, from SortingHat profile." -merged_by_data_domain,keyword,true,"Merge author domain from SortingHat profile." -merged_by_data_gender,keyword,true,"Merge author gender, based on her name, from SortingHat profile(disabled by default)." -merged_by_data_gender_acc,float,true,"Merge author gender accuracy from SortingHat profile(disabled by default)." -merged_by_data_id,keyword,true,"Merge author's ID from SortingHat profile." -merged_by_data_name,keyword,true,"Merge author name from SortingHat profile." -merged_by_data_org_name,keyword,true,"Merge author organization from SortingHat profile." -merged_by_multi_org_names,keyword,true,"List of the merge author organizations from SortingHat profile." -merged_by_data_user_name,keyword,true,"Merge author username from SortingHat profile." -merged_by_data_uuid,keyword,true,"Merge author UUID from SortingHat profile." -metadata__enriched_on,date,true,"Date when the item was enriched." -metadata__gelk_backend_name,keyword,true,"Name of the backend used to enrich information." -metadata__gelk_version,keyword,true,"Version of the backend used to enrich information." -metadata__timestamp,date,true,"Date when the item was stored in RAW index." -metadata__updated_on,date,true,"Date when the item was updated on its original data source." -num_review_comments,long,true,"Number of comments." -origin,keyword,true,"Original URL where the repository was retrieved from." -project,keyword,true,"Project." -project_1,keyword,true,"Project (if more than one level is allowed in project hierarchy)." -pull_request,boolean,true,"True indicating this item is a pull request or not (to be used when the index is queried through aliases including other indexes)." -repository,keyword,true,"Repository name." -repository_labels,keyword,true,"Custom repository labels defined by the user." -state,keyword,true,"State of the Pull Request (GitHub has only 2 possible states: open or closed)." -tag,keyword,true,"Perceval tag." -time_open_days,float,true,"Time the Pull Request is open counted in days." -time_to_close_days,float,true,"Time to close a Pull Request counted in days." -time_to_merge_request_response,float,true,"Time to get a response on a Pull Request in days." -title,keyword,true,"The title of the Pull Request." -title_analyzed,text,false,"Pull Request title split by terms to allow searching." -updated_at,date,true,"Date when the Pull Request was last updated." -url,keyword,true,"Full URL of the Pull Request." -url_id,keyword,true,"Consists of the project path and the Pull Request id." -user_data_bot,boolean,true,"True/False if the Pull Request author is a bot or not from SortingHat profile." -user_data_domain,keyword,true,"Pull Request author domain name from SortingHat profile." -user_data_gender,keyword,true,"Pull Request author gender, based on her name, from SortingHat (disabled by default)." -user_data_gender_acc,float,true,"Pull Request author gender accuracy from SortingHat (disabled by default)." -user_data_id,keyword,true,"Pull Request author ID from SortingHat profile." -user_data_name,keyword,true,"Pull Request author name from SortingHat profile." -user_data_org_name,keyword,true,"Pull Request author organization name from SortingHat profile." -user_data_multi_org_names,keyword,true,"List of the Pull Request author organizations from SortingHat profile." -user_data_user_name,keyword,true,"Pull Request author username from SortingHat profile." -user_data_uuid,keyword,true,"Pull Request author UUID from SortingHat profile." -user_domain,keyword,true,"Pull Request author domain name from GitHub." -user_geolocation,geo_point,true,"Pull Request author geolocation from GitHub." -user_location,keyword,true,"Pull Request author location as string from GitHub." -user_login,keyword,true,"Pull Request author login from GitHub." -user_name,keyword,true,"Pull Request author username from GitHub." -user_org,keyword,true,"Pull Request author organization from GitHub." -uuid,keyword,true,"Perceval UUID." \ No newline at end of file +name,type,aggregatable,description,Raw / Calculated +,NA,NA,"Fields coming from cross references study (available only when it is active), see cross_references.csv.", +additions,long,TRUE,Number of additions in the Pull Request.,GrimoireLab +assignee_geolocation,geo_point,TRUE,Pull Request assignee geolocation from GitHub.,Data Source / Parsed +author_bot,boolean,TRUE,True/False if the Pull Request author is a bot or not from SortingHat profile.,GrimoireLab / SortingHat +author_domain,keyword,TRUE,Pull Request author domain name from SortingHat profile.,GrimoireLab / SortingHat +author_gender,keyword,TRUE,"Pull Request author gender, based on her name, from SortingHat (disabled by default).",GrimoireLab / SortingHat +author_gender_acc,float,TRUE,Pull Request author gender accuracy from SortingHat (disabled by default).,GrimoireLab / SortingHat +author_id,keyword,TRUE,Pull Request author ID from SortingHat profile.,GrimoireLab / SortingHat +author_name,keyword,TRUE,Pull Request author name from SortingHat profile.,GrimoireLab / SortingHat +author_org_name,keyword,TRUE,Pull Request author organization name from SortingHat profile.,GrimoireLab / SortingHat +author_multi_org_names,keyword,TRUE,List of the author organizations from SortingHat profile.,GrimoireLab / SortingHat +author_user_name,keyword,TRUE,Pull Request author username from SortingHat profile.,GrimoireLab / SortingHat +author_uuid,keyword,TRUE,Pull Request author UUID from SortingHat profile.,GrimoireLab / SortingHat +changed_files,long,TRUE,Number of changed files in the Pull Request.,GrimoireLab +closed_at,date,TRUE,Date in which the Issue was closed.,Data Source / Parsed +code_merge_duration,float,TRUE,Difference in days between creation and merging dates.,GrimoireLab +created_at,date,TRUE,Date in which the Issue was created.,Data Source / Parsed +deletions,long,TRUE,Number of deletions in the Pull Request.,GrimoireLab +forks,long,TRUE,Number of repository forks.,Data Source / Parsed +github_repo,keyword,TRUE,GitHub repository name.,Data Source / Parsed +grimoire_creation_date,date,TRUE,Pull Request creation date.,Data Source / Parsed +id,long,TRUE,ID in GitHub.,Data Source / Parsed +id_in_repo,keyword,TRUE,ID in the GitHub repository it was created.,Data Source / Parsed +issue_url,keyword,TRUE,Full URL of the issue.,Data Source / Parsed +is_github_pull_request,long,TRUE,"1 indicating this is a Pull Request, used for counting in case other kind of items could be stored in the same index or alias.",Data Source / Parsed +item_type,keyword,TRUE,"Item type, in this case 'pull request'",Data Source / Parsed +labels,keyword,TRUE,Pull Request assigned labels.,Data Source / Parsed +merge_author_domain,keyword,TRUE,Merge author domain from GitHub.,Data Source / Parsed +merge_author_geolocation,geo_point,TRUE,Merge author geolocation from GitHub.,Data Source / Parsed +merge_author_location,keyword,TRUE,Merge author location as string from GitHub.,Data Source / Parsed +merge_author_login,keyword,TRUE,Merge author login from GitHub.,Data Source / Parsed +merge_author_name,keyword,TRUE,Merge author name from GitHub.,Data Source / Parsed +merge_author_org,keyword,TRUE,Merge author organization from GitHub.,Data Source / Parsed +merged,boolean,TRUE,True if the Pull Request was already merged.,Data Source / Parsed +merged_at,date,TRUE,Date when the Pull Request was merged.,Data Source / Parsed +merged_by_data_bot,boolean,TRUE,"True/False if the merge author is a bot or not, from SortingHat profile.",GrimoireLab / SortingHat +merged_by_data_domain,keyword,TRUE,Merge author domain from SortingHat profile.,GrimoireLab / SortingHat +merged_by_data_gender,keyword,TRUE,"Merge author gender, based on her name, from SortingHat profile(disabled by default).",GrimoireLab / SortingHat +merged_by_data_gender_acc,float,TRUE,Merge author gender accuracy from SortingHat profile(disabled by default).,GrimoireLab / SortingHat +merged_by_data_id,keyword,TRUE,Merge author's ID from SortingHat profile.,GrimoireLab / SortingHat +merged_by_data_name,keyword,TRUE,Merge author name from SortingHat profile.,GrimoireLab / SortingHat +merged_by_data_org_name,keyword,TRUE,Merge author organization from SortingHat profile.,GrimoireLab / SortingHat +merged_by_multi_org_names,keyword,TRUE,List of the merge author organizations from SortingHat profile.,GrimoireLab / SortingHat +merged_by_data_user_name,keyword,TRUE,Merge author username from SortingHat profile.,GrimoireLab / SortingHat +merged_by_data_uuid,keyword,TRUE,Merge author UUID from SortingHat profile.,GrimoireLab / SortingHat +metadata__enriched_on,date,TRUE,Date when the item was enriched.,GrimoireLab +metadata__gelk_backend_name,keyword,TRUE,Name of the backend used to enrich information.,GrimoireLab +metadata__gelk_version,keyword,TRUE,Version of the backend used to enrich information.,GrimoireLab +metadata__timestamp,date,TRUE,Date when the item was stored in RAW index.,GrimoireLab +metadata__updated_on,date,TRUE,Date when the item was updated on its original data source.,GrimoireLab +num_review_comments,long,TRUE,Number of comments.,GrimoireLab +origin,keyword,TRUE,Original URL where the repository was retrieved from.,Data Source / Parsed +project,keyword,TRUE,Project.,GrimoireLab +project_1,keyword,TRUE,Project (if more than one level is allowed in project hierarchy).,GrimoireLab +pull_request,boolean,TRUE,True indicating this item is a pull request or not (to be used when the index is queried through aliases including other indexes).,Data Source / Parsed +repository,keyword,TRUE,Repository name.,Data Source / Parsed +repository_labels,keyword,TRUE,Custom repository labels defined by the user.,Data Source / Parsed +state,keyword,TRUE,State of the Pull Request (GitHub has only 2 possible states: open or closed).,Data Source / Parsed +tag,keyword,TRUE,Perceval tag.,GrimoireLab +time_open_days,float,TRUE,Time the Pull Request is open counted in days.,GrimoireLab +time_to_close_days,float,TRUE,Time to close a Pull Request counted in days.,GrimoireLab +time_to_merge_request_response,float,TRUE,Time to get a response on a Pull Request in days.,GrimoireLab +title,keyword,TRUE,The title of the Pull Request.,Data Source / Parsed +title_analyzed,text,FALSE,Pull Request title split by terms to allow searching.,Data Source / Parsed +updated_at,date,TRUE,Date when the Pull Request was last updated.,Data Source / Parsed +url,keyword,TRUE,Full URL of the Pull Request.,Data Source / Parsed +url_id,keyword,TRUE,Consists of the project path and the Pull Request id.,Data Source / Parsed +user_data_bot,boolean,TRUE,True/False if the Pull Request author is a bot or not from SortingHat profile.,GrimoireLab / SortingHat +user_data_domain,keyword,TRUE,Pull Request author domain name from SortingHat profile.,GrimoireLab / SortingHat +user_data_gender,keyword,TRUE,"Pull Request author gender, based on her name, from SortingHat (disabled by default).",GrimoireLab / SortingHat +user_data_gender_acc,float,TRUE,Pull Request author gender accuracy from SortingHat (disabled by default).,GrimoireLab / SortingHat +user_data_id,keyword,TRUE,Pull Request author ID from SortingHat profile.,GrimoireLab / SortingHat +user_data_name,keyword,TRUE,Pull Request author name from SortingHat profile.,GrimoireLab / SortingHat +user_data_org_name,keyword,TRUE,Pull Request author organization name from SortingHat profile.,GrimoireLab / SortingHat +user_data_multi_org_names,keyword,TRUE,List of the Pull Request author organizations from SortingHat profile.,GrimoireLab / SortingHat +user_data_user_name,keyword,TRUE,Pull Request author username from SortingHat profile.,GrimoireLab / SortingHat +user_data_uuid,keyword,TRUE,Pull Request author UUID from SortingHat profile.,GrimoireLab / SortingHat +user_domain,keyword,TRUE,Pull Request author domain name from GitHub.,Data Source / Parsed +user_geolocation,geo_point,TRUE,Pull Request author geolocation from GitHub.,Data Source / Parsed +user_location,keyword,TRUE,Pull Request author location as string from GitHub.,Data Source / Parsed +user_login,keyword,TRUE,Pull Request author login from GitHub.,Data Source / Parsed +user_name,keyword,TRUE,Pull Request author username from GitHub.,Data Source / Parsed +user_org,keyword,TRUE,Pull Request author organization from GitHub.,Data Source / Parsed +uuid,keyword,TRUE,Perceval UUID.,GrimoireLab \ No newline at end of file diff --git a/schema/github_repos.csv b/schema/github_repos.csv index c7b802e03..9f9c29cee 100644 --- a/schema/github_repos.csv +++ b/schema/github_repos.csv @@ -1,17 +1,17 @@ -name,type,aggregatable,description -fetched_on,long,true,"Timestamp when the item was fetched, equivalent to metadata__updated_on." -forks_count,long,true,"Number of forks of the repository." -grimoire_creation_date,date,true,"Date when the repo was fetched." -is_github_repository,long,true,"Used to separate repositories from other items such as pull requests and issues." -metadata__enriched_on,date,true,"Date when the data were enriched." -metadata__gelk_backend_name,keyword,true,"Name of the backend used to enrich the data." -metadata__gelk_version,keyword,true,"Version of the backend used to enrich the data." -metadata__timestamp,date,true,"Date when the item was stored in ElasticSearch raw index." -metadata__updated_on,date,true,"Date when the item was updated on its original data source." -origin,keyword,true,"The original URL from which the repository was retrieved from." -repository_labels,keyword,true,"Custom repository labels defined by the user." -stargazers_count,long,true,"Number of stars of the repository." -subscribers_count,long,true,"Number of watchers of the repository." -tag,keyword,true,"Perceval tag." -url,keyword,true,"Full URL of the repository" -uuid,keyword,true,"Perceval UUID." +name,type,aggregatable,description,Raw / Calculated +fetched_on,long,TRUE,"Timestamp when the item was fetched, equivalent to metadata__updated_on.",GrimoireLab +forks_count,long,TRUE,Number of forks of the repository.,Data Source / Parsed +grimoire_creation_date,date,TRUE,Date when the repo was fetched.,Data Source / Parsed +is_github_repository,long,TRUE,Used to separate repositories from other items such as pull requests and issues.,GrimoireLab +metadata__enriched_on,date,TRUE,Date when the data were enriched.,GrimoireLab +metadata__gelk_backend_name,keyword,TRUE,Name of the backend used to enrich the data.,GrimoireLab +metadata__gelk_version,keyword,TRUE,Version of the backend used to enrich the data.,GrimoireLab +metadata__timestamp,date,TRUE,Date when the item was stored in ElasticSearch raw index.,GrimoireLab +metadata__updated_on,date,TRUE,Date when the item was updated on its original data source.,GrimoireLab +origin,keyword,TRUE,The original URL from which the repository was retrieved from.,Data Source / Parsed +repository_labels,keyword,TRUE,Custom repository labels defined by the user.,Data Source / Parsed +stargazers_count,long,TRUE,Number of stars of the repository.,Data Source / Parsed +subscribers_count,long,TRUE,Number of watchers of the repository.,Data Source / Parsed +tag,keyword,TRUE,Perceval tag.,GrimoireLab +url,keyword,TRUE,Full URL of the repository,Data Source / Parsed +uuid,keyword,TRUE,Perceval UUID.,GrimoireLab \ No newline at end of file