Module: ScraperUtils::LogUtils

Defined in:
lib/scraper_utils/log_utils.rb

Overview

Utilities for logging scraper execution details and outcomes

Constant Summary collapse

SUMMARY_TABLE = 
"scrape_summary"
LOG_TABLE = 
"scrape_log"
LOG_RETENTION_DAYS = 
30

Class Method Summary collapse

Class Method Details

.cleanup_old_records(force: false) ⇒ Object 



195
196
197
198
199
200
201
202
203
204
205
206
207
 
# File 'lib/scraper_utils/log_utils.rb', line 195

def self.cleanup_old_records(force: false)
  cutoff = (Date.today - LOG_RETENTION_DAYS).to_s
  return if !force && @last_cutoff == cutoff

  @last_cutoff = cutoff

  [SUMMARY_TABLE, LOG_TABLE].each do |table|
    ScraperWiki.sqliteexecute(
      "DELETE FROM #{table} WHERE date(run_at) < date(?)",
      [cutoff]
    )
  end
end

.extract_meaningful_backtrace(error) ⇒ Object

Extracts meaningful backtrace - 3 lines from ruby/gem and max 6 in total



210
211
212
213
214
215
216
217
218
219
220
 
# File 'lib/scraper_utils/log_utils.rb', line 210

def self.extract_meaningful_backtrace(error)
  return nil unless error.respond_to?(:backtrace) && error&.backtrace

  lines = []
  error.backtrace.each do |line|
    lines << line if lines.length < 2 || !line.include?("/vendor/")
    break if lines.length >= 6
  end

  lines.empty? ? nil : lines.join("\n")
end

.log(message, authority = nil) ⇒ void

This method returns an undefined value.

Logs a message, automatically prefixing with authority name if in a fiber

Parameters:

  • message (String) —

    the message to log



16
17
18
19
20
21
22
23
24
25
 
# File 'lib/scraper_utils/log_utils.rb', line 16

def self.log(message, authority = nil)
  authority ||= Scheduler.current_authority
  $stderr.flush
  if authority
    puts "[#{authority}] #{message}"
  else
    puts message
  end
  $stdout.flush
end

.log_scraping_run(start_time, attempt, authorities, exceptions) ⇒ void

This method returns an undefined value.

Log details about a scraping run for one or more authorities DataQualityMonitor.stats is checked for :saved and :unprocessed entries

Parameters:

  • start_time (Time) —

    When this scraping attempt was started

  • attempt (Integer) —

    1 for first run, 2 for first retry, 3 for last retry (without proxy)

  • authorities (Array<Symbol>) —

    List of authorities attempted to scrape

  • exceptions (Hash{Symbol => Exception}) —

    Any exceptions that occurred during scraping

Raises:

  • (ArgumentError) 


34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
 
# File 'lib/scraper_utils/log_utils.rb', line 34

def self.log_scraping_run(start_time, attempt, authorities, exceptions)
  raise ArgumentError, "Invalid start time" unless start_time.is_a?(Time)
  raise ArgumentError, "Authorities must be a non-empty array" if authorities.empty?

  end_time = Time.now
  duration = (end_time - start_time).round(1)

  successful = []
  failed = []
  interrupted = []

  authorities.each do |authority_label|
    stats = ScraperUtils::DataQualityMonitor.stats&.fetch(authority_label, nil) || {}

    exception = exceptions[authority_label]
    status = if stats[:saved]&.positive?
               exception ? :interrupted : :successful
             else
               :failed
             end
    case status
    when :successful
      successful << authority_label
    when :interrupted
      interrupted << authority_label
    else
      failed << authority_label
    end

    record = {
      "run_at" => start_time.iso8601,
      "attempt" => attempt,
      "authority_label" => authority_label.to_s,
      "records_saved" => stats[:saved] || 0,
      "unprocessable_records" => stats[:unprocessed] || 0,
      "status" => status.to_s,
      "error_message" => exception&.message,
      "error_class" => exception&.class&.to_s,
      "error_backtrace" => extract_meaningful_backtrace(exception)
    }

    save_log_record(record)
  end

  # Save summary record for the entire run
  save_summary_record(
    start_time,
    attempt,
    duration,
    successful,
    interrupted,
    failed
  )

  cleanup_old_records
end

.report_on_results(authorities, exceptions) ⇒ void

This method returns an undefined value.

Report on the results DataQualityMonitor.stats is checked for :saved and :unprocessed entries

Parameters:

  • authorities (Array<Symbol>) —

    List of authorities attempted to scrape

  • exceptions (Hash{Symbol => Exception}) —

    Any exceptions that occurred during scraping



96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
 
# File 'lib/scraper_utils/log_utils.rb', line 96

def self.report_on_results(authorities, exceptions)
  if ENV["MORPH_EXPECT_BAD"]
    expect_bad = ENV["MORPH_EXPECT_BAD"].split(",").map(&:strip).map(&:to_sym)
  end
  expect_bad ||= []

  $stderr.flush
  puts "MORPH_EXPECT_BAD=#{ENV.fetch('MORPH_EXPECT_BAD', nil)}"

  # Print summary table
  puts "\nScraping Summary:"
  summary_format = "%-20s %6s %6s %s"

  puts format(summary_format, 'Authority', 'OK', 'Bad', 'Exception')
  puts format(summary_format, "-" * 20, "-" * 6, "-" * 6, "-" * 50)

  authorities.each do |authority|
    stats = ScraperUtils::DataQualityMonitor.stats&.fetch(authority, {}) || {}

    ok_records = stats[:saved] || 0
    bad_records = stats[:unprocessed] || 0

    expect_bad_prefix = expect_bad.include?(authority) ? "[EXPECT BAD] " : ""
    exception_msg = if exceptions[authority]
                      "#{exceptions[authority].class} - #{exceptions[authority].message}"
                    else
                      "-"
                    end
    puts format(summary_format, authority.to_s, ok_records, bad_records,
                "#{expect_bad_prefix}#{exception_msg}".slice(0, 70))
  end
  puts

  errors = []

  # Check for authorities that were expected to be bad but are now working
  unexpected_working = expect_bad.select do |authority|
    stats = ScraperUtils::DataQualityMonitor.stats&.fetch(authority, {}) || {}
    stats[:saved]&.positive? && !exceptions[authority]
  end

  if unexpected_working.any?
    errors <<
      "WARNING: Remove #{unexpected_working.join(',')} from MORPH_EXPECT_BAD as it now works!"
  end

  # Check for authorities with unexpected errors
  unexpected_errors = authorities
                      .select { |authority| exceptions[authority] }
                      .reject { |authority| expect_bad.include?(authority) }

  if unexpected_errors.any?
    errors << "ERROR: Unexpected errors in: #{unexpected_errors.join(',')} " \
              "(Add to MORPH_EXPECT_BAD?)"
    unexpected_errors.each do |authority|
      error = exceptions[authority]
      errors << "  #{authority}: #{error.class} - #{error.message}"
    end
  end

  $stdout.flush
  if errors.any?
    errors << "See earlier output for details"
    raise errors.join("\n")
  end

  puts "Exiting with OK status!"
end

.save_log_record(record) ⇒ Object 



165
166
167
168
169
170
171
 
# File 'lib/scraper_utils/log_utils.rb', line 165

def self.save_log_record(record)
  ScraperWiki.save_sqlite(
    %w[authority_label run_at],
    record,
    LOG_TABLE
  )
end

.save_summary_record(start_time, attempt, duration, successful, interrupted, failed) ⇒ Object 



173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
 
# File 'lib/scraper_utils/log_utils.rb', line 173

def self.save_summary_record(start_time, attempt, duration,
                             successful, interrupted, failed)
  summary = {
    "run_at" => start_time.iso8601,
    "attempt" => attempt,
    "duration" => duration,
    "successful" => successful.join(","),
    "failed" => failed.join(","),
    "interrupted" => interrupted.join(","),
    "successful_count" => successful.size,
    "interrupted_count" => interrupted.size,
    "failed_count" => failed.size,
    "public_ip" => ScraperUtils::MechanizeUtils.public_ip
  }

  ScraperWiki.save_sqlite(
    ["run_at"],
    summary,
    SUMMARY_TABLE
  )
end