Merge branch 'feature/fetcher' into develop

This commit is contained in:
Petr Korolev 2015-04-22 12:28:10 +03:00
commit 6a7dbeb450
2 changed files with 269 additions and 193 deletions

View File

@ -9,6 +9,7 @@ require_relative "github_changelog_generator/parser"
require_relative "github_changelog_generator/generator" require_relative "github_changelog_generator/generator"
require_relative "github_changelog_generator/version" require_relative "github_changelog_generator/version"
require_relative "github_changelog_generator/reader" require_relative "github_changelog_generator/reader"
require_relative "github_changelog_generator/fetcher"
module GitHubChangelogGenerator module GitHubChangelogGenerator
# Default error for ChangelogGenerator # Default error for ChangelogGenerator
@ -19,33 +20,19 @@ module GitHubChangelogGenerator
class ChangelogGenerator class ChangelogGenerator
attr_accessor :options, :all_tags, :github attr_accessor :options, :all_tags, :github
PER_PAGE_NUMBER = 30
GH_RATE_LIMIT_EXCEEDED_MSG = "Warning: GitHub API rate limit (5000 per hour) exceeded, change log may be " \
"missing some issues. You can limit the number of issues fetched using the `--max-issues NUM` argument."
# Class, responsible for whole change log generation cycle # Class, responsible for whole change log generation cycle
# @return initialised insance of ChangelogGenerator # @return initialised instance of ChangelogGenerator
def initialize def initialize
@options = Parser.parse_options @options = Parser.parse_options
fetch_github_token @fetcher = GitHubChangelogGenerator::Fetcher.new @options
github_options = { per_page: PER_PAGE_NUMBER } @generator = Generator.new @options
github_options[:oauth_token] = @github_token unless @github_token.nil?
github_options[:endpoint] = options[:github_endpoint] unless options[:github_endpoint].nil?
github_options[:site] = options[:github_endpoint] unless options[:github_site].nil?
begin # @all_tags = get_filtered_tags
@github = Github.new github_options @all_tags = @fetcher.get_all_tags
rescue
puts GH_RATE_LIMIT_EXCEEDED_MSG.yellow
end
@generator = Generator.new(@options) @issues, @pull_requests = @fetcher.fetch_issues_and_pull_requests
@all_tags = get_all_tags
@issues, @pull_requests = fetch_issues_and_pull_requests
@pull_requests = @options[:pulls] ? get_filtered_pull_requests : [] @pull_requests = @options[:pulls] ? get_filtered_pull_requests : []
@ -53,7 +40,23 @@ module GitHubChangelogGenerator
fetch_event_for_issues_and_pr fetch_event_for_issues_and_pr
detect_actual_closed_dates detect_actual_closed_dates
@tag_times_hash = {} end
# Return tags after filtering tags in lists provided by option: --between-tags & --exclude-tags
#
# @return [Array]
def get_filtered_tags
all_tags = @fetcher.get_all_tags
filtered_tags = []
if @options[:between_tags]
@options[:between_tags].each do |tag|
unless all_tags.include? tag
puts "Warning: can't find tag #{tag}, specified with --between-tags option.".yellow
end
end
filtered_tags = all_tags.select { |tag| @options[:between_tags].include? tag }
end
filtered_tags
end end
def detect_actual_closed_dates def detect_actual_closed_dates
@ -81,6 +84,8 @@ module GitHubChangelogGenerator
end end
end end
# Fill :actual_date parameter of specified issue by closed date of the commit, it it was closed by commit.
# @param [Hash] issue
def find_closed_date_by_commit(issue) def find_closed_date_by_commit(issue)
unless issue["events"].nil? unless issue["events"].nil?
# if it's PR -> then find "merged event", in case of usual issue -> fond closed date # if it's PR -> then find "merged event", in case of usual issue -> fond closed date
@ -92,7 +97,7 @@ module GitHubChangelogGenerator
issue[:actual_date] = issue[:closed_at] issue[:actual_date] = issue[:closed_at]
else else
begin begin
commit = @github.git_data.commits.get @options[:user], @options[:project], event[:commit_id] commit = @fetcher.fetch_commit(event)
issue[:actual_date] = commit[:author][:date] issue[:actual_date] = commit[:author][:date]
rescue rescue
puts "Warning: Can't fetch commit #{event[:commit_id]}. It is probably referenced from another repo.".yellow puts "Warning: Can't fetch commit #{event[:commit_id]}. It is probably referenced from another repo.".yellow
@ -110,42 +115,6 @@ module GitHubChangelogGenerator
puts JSON.pretty_generate(json) puts JSON.pretty_generate(json)
end end
# This method fetch missing required attributes for pull requests
# :merged_at - is a date, when issue PR was merged.
# More correct to use this date, not closed date.
def fetch_merged_at_pull_requests
if @options[:verbose]
print "Fetching merged dates...\r"
end
pull_requests = []
begin
response = @github.pull_requests.list @options[:user], @options[:project], state: "closed"
page_i = 0
response.each_page do |page|
page_i += PER_PAGE_NUMBER
count_pages = response.count_pages
print "Fetching merged dates... #{page_i}/#{count_pages * PER_PAGE_NUMBER}\r"
pull_requests.concat(page)
end
rescue
puts GH_RATE_LIMIT_EXCEEDED_MSG.yellow
end
print " \r"
@pull_requests.each { |pr|
fetched_pr = pull_requests.find { |fpr|
fpr.number == pr.number
}
pr[:merged_at] = fetched_pr[:merged_at]
pull_requests.delete(fetched_pr)
}
if @options[:verbose]
puts "Fetching merged dates: Done!"
end
end
# This method fetches missing params for PR and filter them by specified options # This method fetches missing params for PR and filter them by specified options
# It include add all PR's with labels from @options[:include_labels] array # It include add all PR's with labels from @options[:include_labels] array
# And exclude all from :exclude_labels array. # And exclude all from :exclude_labels array.
@ -164,6 +133,28 @@ module GitHubChangelogGenerator
filtered_pull_requests filtered_pull_requests
end end
# This method fetch missing required attributes for pull requests
# :merged_at - is a date, when issue PR was merged.
# More correct to use this date, not closed date.
def fetch_merged_at_pull_requests
if @options[:verbose]
print "Fetching merged dates...\r"
end
pull_requests = @fetcher.fetch_pull_requests
@pull_requests.each { |pr|
fetched_pr = pull_requests.find { |fpr|
fpr.number == pr.number
}
pr[:merged_at] = fetched_pr[:merged_at]
pull_requests.delete(fetched_pr)
}
if @options[:verbose]
puts "Fetching merged dates: Done!"
end
end
# Include issues with labels, specified in :include_labels # Include issues with labels, specified in :include_labels
# @param [Array] issues to filter # @param [Array] issues to filter
# @return [Array] filtered array of issues # @return [Array] filtered array of issues
@ -238,7 +229,7 @@ module GitHubChangelogGenerator
puts "Sorting tags..." puts "Sorting tags..."
end end
@all_tags.sort_by! { |x| get_time_of_tag(x) }.reverse! @all_tags.sort_by! { |x| @fetcher.get_time_of_tag(x) }.reverse!
if @options[:verbose] if @options[:verbose]
puts "Generating log..." puts "Generating log..."
@ -274,9 +265,8 @@ module GitHubChangelogGenerator
i = 0 i = 0
all = @all_tags.count all = @all_tags.count
@all_tags.each { |tag| @all_tags.each { |tag|
# explicit set @tag_times_hash to write data safety.
threads << Thread.new { threads << Thread.new {
get_time_of_tag(tag, @tag_times_hash) @fetcher.get_time_of_tag(tag)
if @options[:verbose] if @options[:verbose]
print "Fetching tags dates: #{i + 1}/#{all}\r" print "Fetching tags dates: #{i + 1}/#{all}\r"
i += 1 i += 1
@ -293,48 +283,6 @@ module GitHubChangelogGenerator
end end
end end
def get_all_tags
if @options[:verbose]
print "Fetching tags...\r"
end
tags = []
begin
response = @github.repos.tags @options[:user], @options[:project]
page_i = 0
count_pages = response.count_pages
response.each_page do |page|
page_i += PER_PAGE_NUMBER
print "Fetching tags... #{page_i}/#{count_pages * PER_PAGE_NUMBER}\r"
tags.concat(page)
end
print " \r"
if tags.count == 0
puts "Warning: Can't find any tags in repo. Make sure, that you push tags to remote repo via 'git push --tags'".yellow
elsif @options[:verbose]
puts "Found #{tags.count} tags"
end
rescue
puts GH_RATE_LIMIT_EXCEEDED_MSG.yellow
end
tags
end
def fetch_github_token
env_var = @options[:token] ? @options[:token] : (ENV.fetch "CHANGELOG_GITHUB_TOKEN", nil)
unless env_var
puts "Warning: No token provided (-t option) and variable $CHANGELOG_GITHUB_TOKEN was not found.".yellow
puts "This script can make only 50 requests per hour to GitHub API without a token!".yellow
end
@github_token ||= env_var
end
# Generate log only between 2 specified tags # Generate log only between 2 specified tags
# @param [String] older_tag all issues before this tag date will be excluded. May be nil, if it's first tag # @param [String] older_tag all issues before this tag date will be excluded. May be nil, if it's first tag
# @param [String] newer_tag all issue after this tag will be excluded. May be nil for unreleased section # @param [String] newer_tag all issue after this tag will be excluded. May be nil for unreleased section
@ -403,8 +351,8 @@ module GitHubChangelogGenerator
def delete_by_time(array, hash_key = :actual_date, older_tag = nil, newer_tag = nil) def delete_by_time(array, hash_key = :actual_date, older_tag = nil, newer_tag = nil)
fail ChangelogGeneratorError, "At least one of the tags should be not nil!".red if older_tag.nil? && newer_tag.nil? fail ChangelogGeneratorError, "At least one of the tags should be not nil!".red if older_tag.nil? && newer_tag.nil?
newer_tag_time = newer_tag && get_time_of_tag(newer_tag) newer_tag_time = newer_tag && @fetcher.get_time_of_tag(newer_tag)
older_tag_time = older_tag && get_time_of_tag(older_tag) older_tag_time = older_tag && @fetcher.get_time_of_tag(older_tag)
array.select { |req| array.select { |req|
if req[hash_key] if req[hash_key]
@ -439,9 +387,9 @@ module GitHubChangelogGenerator
# @param [String] older_tag_name Older tag, used for the links. Could be nil for last tag. # @param [String] older_tag_name Older tag, used for the links. Could be nil for last tag.
# @return [String] Ready and parsed section # @return [String] Ready and parsed section
def create_log(pull_requests, issues, newer_tag, older_tag_name = nil) def create_log(pull_requests, issues, newer_tag, older_tag_name = nil)
newer_tag_time = newer_tag.nil? ? Time.new : get_time_of_tag(newer_tag) newer_tag_time = newer_tag.nil? ? Time.new : @fetcher.get_time_of_tag(newer_tag)
newer_tag_name = newer_tag.nil? ? @options[:unreleased_label] : newer_tag["name"] newer_tag_name = newer_tag.nil? ? @options[:unreleased_label] : newer_tag["name"]
newer_tag_link = newer_tag.nil? ? "HEAD" : newer_tag_name newer_tag_link = newer_tag.nil? ? "HEAD" : newer_tag_name
github_site = options[:github_site] || "https://github.com" github_site = options[:github_site] || "https://github.com"
project_url = "#{github_site}/#{@options[:user]}/#{@options[:project]}" project_url = "#{github_site}/#{@options[:user]}/#{@options[:project]}"
@ -534,27 +482,8 @@ module GitHubChangelogGenerator
log log
end end
# Try to find tag date in local hash. # Filter issues according labels
# Otherwise fFetch tag time and put it to local hash file. # @return [Array] Filtered issues
# @param [String] tag_name name of the tag
# @param [Hash] tag_times_hash the hash of tag times
# @return [Time] time of specified tag
def get_time_of_tag(tag_name, tag_times_hash = @tag_times_hash)
fail ChangelogGeneratorError, "tag_name is nil".red if tag_name.nil?
if tag_times_hash[tag_name["name"]]
return @tag_times_hash[tag_name["name"]]
end
begin
github_git_data_commits_get = @github.git_data.commits.get @options[:user], @options[:project], tag_name["commit"]["sha"]
rescue
puts GH_RATE_LIMIT_EXCEEDED_MSG.yellow
end
time_string = github_git_data_commits_get["committer"]["date"]
@tag_times_hash[tag_name["name"]] = Time.parse(time_string)
end
def get_filtered_issues def get_filtered_issues
filtered_issues = include_issues_by_labels(@issues) filtered_issues = include_issues_by_labels(@issues)
@ -567,41 +496,8 @@ module GitHubChangelogGenerator
filtered_issues filtered_issues
end end
# This method fetch all closed issues and separate them to pull requests and pure issues # Fetch event for issues and pull requests
# (pull request is kind of issue in term of GitHub) # @return [Array] array of fetched issues
# @return [Tuple] with issues and pull requests
def fetch_issues_and_pull_requests
if @options[:verbose]
print "Fetching closed issues...\r"
end
issues = []
begin
response = @github.issues.list user: @options[:user], repo: @options[:project], state: "closed", filter: "all", labels: nil
page_i = 0
count_pages = response.count_pages
response.each_page do |page|
page_i += PER_PAGE_NUMBER
print "Fetching issues... #{page_i}/#{count_pages * PER_PAGE_NUMBER}\r"
issues.concat(page)
break if @options[:max_issues] && issues.length >= @options[:max_issues]
end
rescue
puts GH_RATE_LIMIT_EXCEEDED_MSG.yellow
end
print " \r"
if @options[:verbose]
puts "Received issues: #{issues.count}"
end
# remove pull request from issues:
issues.partition { |x|
x[:pull_request].nil?
}
end
def fetch_event_for_issues_and_pr def fetch_event_for_issues_and_pr
if @options[:verbose] if @options[:verbose]
print "Fetching events for issues and PR: 0/#{@issues.count + @pull_requests.count}\r" print "Fetching events for issues and PR: 0/#{@issues.count + @pull_requests.count}\r"
@ -609,36 +505,7 @@ module GitHubChangelogGenerator
# Async fetching events: # Async fetching events:
fetch_events_async(@issues + @pull_requests) @fetcher.fetch_events_async(@issues + @pull_requests)
end
def fetch_events_async(issues)
i = 0
max_thread_number = 50
threads = []
issues.each_slice(max_thread_number) { |issues_slice|
issues_slice.each { |issue|
threads << Thread.new {
begin
obj = @github.issues.events.list user: @options[:user], repo: @options[:project], issue_number: issue["number"]
rescue
puts GH_RATE_LIMIT_EXCEEDED_MSG.yellow
end
issue[:events] = obj.body
print "Fetching events for issues and PR: #{i + 1}/#{@issues.count + @pull_requests.count}\r"
i += 1
}
}
threads.each(&:join)
threads = []
}
# to clear line from prev print
print " \r"
if @options[:verbose]
puts "Fetching events for issues and PR: #{i} Done!"
end
end end
end end

View File

@ -0,0 +1,209 @@
require "logger"
module GitHubChangelogGenerator
# A Fetcher responsible for all requests to GitHub and all basic manipulation with related data
# (such as filtering, validating, e.t.c)
#
# Example:
# fetcher = GitHubChangelogGenerator::Fetcher.new options
class Fetcher
PER_PAGE_NUMBER = 30
GH_RATE_LIMIT_EXCEEDED_MSG = "Warning: GitHub API rate limit (5000 per hour) exceeded, change log may be " \
"missing some issues. You can limit the number of issues fetched using the `--max-issues NUM` argument."
def initialize(options = {})
@options = options
@user = @options[:user]
@project = @options[:project]
@github_token = fetch_github_token
@tag_times_hash = {}
@logger = Logger.new(STDOUT)
@logger.formatter = proc do |_severity, _datetime, _progname, msg|
"#{msg}\n"
end
github_options = { per_page: PER_PAGE_NUMBER }
github_options[:oauth_token] = @github_token unless @github_token.nil?
github_options[:endpoint] = options[:github_endpoint] unless options[:github_endpoint].nil?
github_options[:site] = options[:github_endpoint] unless options[:github_site].nil?
begin
@github = Github.new github_options
rescue
@logger.warn GH_RATE_LIMIT_EXCEEDED_MSG.yellow
end
end
# Returns GitHub token. First try to use variable, provided by --token option,
# otherwise try to fetch it from CHANGELOG_GITHUB_TOKEN env variable.
#
# @return [String]
def fetch_github_token
env_var = @options[:token] ? @options[:token] : (ENV.fetch "CHANGELOG_GITHUB_TOKEN", nil)
unless env_var
@logger.warn "Warning: No token provided (-t option) and variable $CHANGELOG_GITHUB_TOKEN was not found.".yellow
@logger.warn "This script can make only 50 requests to GitHub API per hour without token!".yellow
end
env_var
end
# Fetch all tags from repo
# @return [Array] array of tags
def get_all_tags
if @options[:verbose]
print "Fetching tags...\r"
end
tags = []
begin
response = @github.repos.tags @options[:user], @options[:project]
page_i = 0
count_pages = response.count_pages
response.each_page do |page|
page_i += PER_PAGE_NUMBER
print "Fetching tags... #{page_i}/#{count_pages * PER_PAGE_NUMBER}\r"
tags.concat(page)
end
print " \r"
if tags.count == 0
@logger.warn "Warning: Can't find any tags in repo.\
Make sure, that you push tags to remote repo via 'git push --tags'".yellow
elsif @options[:verbose]
@logger.info "Found #{tags.count} tags"
end
rescue
@logger.warn GH_RATE_LIMIT_EXCEEDED_MSG.yellow
end
tags
end
# This method fetch all closed issues and separate them to pull requests and pure issues
# (pull request is kind of issue in term of GitHub)
# @return [Tuple] with issues and pull requests
def fetch_issues_and_pull_requests
if @options[:verbose]
print "Fetching closed issues...\r"
end
issues = []
begin
response = @github.issues.list user: @options[:user],
repo: @options[:project],
state: "closed",
filter: "all",
labels: nil
page_i = 0
count_pages = response.count_pages
response.each_page do |page|
page_i += PER_PAGE_NUMBER
print "Fetching issues... #{page_i}/#{count_pages * PER_PAGE_NUMBER}\r"
issues.concat(page)
break if @options[:max_issues] && issues.length >= @options[:max_issues]
end
rescue
@logger.warn GH_RATE_LIMIT_EXCEEDED_MSG.yellow
end
print " \r"
if @options[:verbose]
@logger.info "Received issues: #{issues.count}"
end
# remove pull request from issues:
issues.partition { |x|
x[:pull_request].nil?
}
end
# Fetch all pull requests. We need them to detect :merged_at parameter
# @return [Array] all pull requests
def fetch_pull_requests
pull_requests = []
begin
response = @github.pull_requests.list @options[:user], @options[:project], state: "closed"
page_i = 0
response.each_page do |page|
page_i += PER_PAGE_NUMBER
count_pages = response.count_pages
print "Fetching merged dates... #{page_i}/#{count_pages * PER_PAGE_NUMBER}\r"
pull_requests.concat(page)
end
rescue
@logger.warn GH_RATE_LIMIT_EXCEEDED_MSG.yellow
end
print " \r"
pull_requests
end
# Fetch event for all issues and add them to :events
# @param [Array] issues
# @return [Void]
def fetch_events_async(issues)
i = 0
max_thread_number = 50
threads = []
issues.each_slice(max_thread_number) { |issues_slice|
issues_slice.each { |issue|
threads << Thread.new {
begin
obj = @github.issues.events.list user: @options[:user],
repo: @options[:project],
issue_number: issue["number"]
rescue
@logger.warn GH_RATE_LIMIT_EXCEEDED_MSG.yellow
end
issue[:events] = obj.body
print "Fetching events for issues and PR: #{i + 1}/#{issues.count}\r"
i += 1
}
}
threads.each(&:join)
threads = []
}
# to clear line from prev print
print " \r"
if @options[:verbose]
@logger.info "Fetching events for issues and PR: #{i} Done!"
end
end
# Try to find tag date in local hash.
# Otherwise fFetch tag time and put it to local hash file.
# @param [String] tag_name name of the tag
# @return [Time] time of specified tag
def get_time_of_tag(tag_name)
fail ChangelogGeneratorError, "tag_name is nil".red if tag_name.nil?
if @tag_times_hash[tag_name["name"]]
return @tag_times_hash[tag_name["name"]]
end
begin
github_git_data_commits_get = @github.git_data.commits.get @options[:user],
@options[:project],
tag_name["commit"]["sha"]
rescue
@logger.warn GH_RATE_LIMIT_EXCEEDED_MSG.yellow
end
time_string = github_git_data_commits_get["committer"]["date"]
@tag_times_hash[tag_name["name"]] = Time.parse(time_string)
end
# Fetch commit for specifed event
# @return [Hash]
def fetch_commit(event)
@github.git_data.commits.get @options[:user], @options[:project], event[:commit_id]
end
end
end