|
| 1 | +NODE package_downloads_filtered |
| 2 | +SQL > |
| 3 | + |
| 4 | + % |
| 5 | + SELECT * |
| 6 | + FROM packageDownloads p FINAL |
| 7 | + where |
| 8 | + insightsProjectId = (select insightsProjectId from segments_filtered) |
| 9 | + {% if defined(startDate) %} |
| 10 | + AND p.date > toDate( |
| 11 | + {{ DateTime(startDate, description="Filter package downloads after", required=False) }} |
| 12 | + ) |
| 13 | + {% end %} |
| 14 | + {% if defined(endDate) %} |
| 15 | + AND p.date < toDate( |
| 16 | + {{ DateTime(endDate, description="Filter package downloads before", required=False) }} |
| 17 | + ) |
| 18 | + {% end %} |
| 19 | + {% if defined(ecosystem) %} |
| 20 | + AND p.ecosystem |
| 21 | + = {{ String(ecosystem, description="Filter package download ecosystem", required=False) }} |
| 22 | + {% end %} |
| 23 | + {% if defined(name) %} |
| 24 | + AND p.name = {{ String(name, description="Filter by package name", required=False) }} |
| 25 | + {% end %} |
| 26 | + |
| 27 | + |
| 28 | + |
| 29 | +NODE package_downloads_timeseries_bounds |
| 30 | +SQL > |
| 31 | + |
| 32 | + % |
| 33 | + {% if defined(startDate) and not defined(endDate) %} |
| 34 | + select toDate({{ startDate }}) as actual_start_date, toDate(now()) as actual_end_date |
| 35 | + {% end %} |
| 36 | + {% if not defined(startDate) and defined(endDate) %} |
| 37 | + select |
| 38 | + min(package_downloads_filtered.date) as actual_start_date, |
| 39 | + toDate({{ endDate }}) as actual_end_date |
| 40 | + from package_downloads_filtered |
| 41 | + where package_downloads_filtered.date > '1980-01-01' |
| 42 | + {% end %} |
| 43 | + {% if not defined(startDate) and not defined(endDate) %} |
| 44 | + select |
| 45 | + min(toDate(package_downloads_filtered.date)) as actual_start_date, |
| 46 | + toDate(now()) as actual_end_date |
| 47 | + from package_downloads_filtered |
| 48 | + where package_downloads_filtered.date > '1980-01-01' |
| 49 | + {% end %} |
| 50 | + {% if defined(startDate) and defined(endDate) %} |
| 51 | + select toDate({{ startDate }}) as actual_start_date, toDate({{ endDate }}) as actual_end_date |
| 52 | + {% end %} |
| 53 | + |
| 54 | + |
| 55 | + |
| 56 | +NODE package_downloads_generate_timeseries |
| 57 | +SQL > |
| 58 | + |
| 59 | + % |
| 60 | + SELECT DISTINCT |
| 61 | + CASE |
| 62 | + WHEN {{ granularity }} = 'daily' |
| 63 | + THEN toDate(addDays(package_downloads_timeseries_bounds.actual_start_date, number)) |
| 64 | + WHEN {{ granularity }} = 'weekly' |
| 65 | + THEN toStartOfWeek(addDays(package_downloads_timeseries_bounds.actual_start_date, number * 7)) |
| 66 | + WHEN {{ granularity }} = 'monthly' |
| 67 | + THEN toStartOfMonth(addMonths(package_downloads_timeseries_bounds.actual_start_date, number)) |
| 68 | + WHEN {{ granularity }} = 'quarterly' |
| 69 | + THEN toStartOfQuarter(addMonths(package_downloads_timeseries_bounds.actual_start_date, number * 3)) |
| 70 | + WHEN {{ granularity }} = 'yearly' |
| 71 | + THEN toStartOfYear(addYears(package_downloads_timeseries_bounds.actual_start_date, number)) |
| 72 | + END AS "startDate", |
| 73 | + CASE |
| 74 | + WHEN {{ granularity }} = 'daily' |
| 75 | + THEN toDate(addDays(actual_start_date, number)) |
| 76 | + WHEN {{ granularity }} = 'weekly' |
| 77 | + THEN toDate(toStartOfWeek(addDays(actual_start_date, number * 7)) + INTERVAL 6 DAY) |
| 78 | + WHEN {{ granularity }} = 'monthly' |
| 79 | + THEN |
| 80 | + toDate( |
| 81 | + toStartOfMonth(addMonths(actual_start_date, number)) |
| 82 | + + INTERVAL 1 MONTH |
| 83 | + - INTERVAL 1 DAY |
| 84 | + ) |
| 85 | + WHEN {{ granularity }} = 'quarterly' |
| 86 | + THEN |
| 87 | + toDate( |
| 88 | + toStartOfQuarter(addMonths(actual_start_date, number * 3)) |
| 89 | + + INTERVAL 3 MONTH |
| 90 | + - INTERVAL 1 DAY |
| 91 | + ) |
| 92 | + WHEN {{ granularity }} = 'yearly' |
| 93 | + THEN |
| 94 | + toDate( |
| 95 | + toStartOfYear(addYears(actual_start_date, number)) |
| 96 | + + INTERVAL 1 YEAR |
| 97 | + - INTERVAL 1 DAY |
| 98 | + ) |
| 99 | + END AS "endDate" |
| 100 | + FROM numbers(1000) |
| 101 | + CROSS JOIN |
| 102 | + ( |
| 103 | + SELECT |
| 104 | + CASE |
| 105 | + WHEN {{ granularity }} = 'weekly' |
| 106 | + THEN toStartOfWeek(actual_start_date) |
| 107 | + WHEN {{ granularity }} = 'monthly' |
| 108 | + THEN toStartOfMonth(actual_start_date) |
| 109 | + WHEN {{ granularity }} = 'quarterly' |
| 110 | + THEN toStartOfQuarter(actual_start_date) |
| 111 | + WHEN {{ granularity }} = 'yearly' |
| 112 | + THEN toStartOfYear(actual_start_date) |
| 113 | + ELSE actual_start_date |
| 114 | + END AS actual_start_date, |
| 115 | + actual_end_date |
| 116 | + FROM package_downloads_timeseries_bounds |
| 117 | + ) package_downloads_timeseries_bounds |
| 118 | + WHERE "startDate" >= actual_start_date AND "startDate" < actual_end_date |
| 119 | + |
| 120 | + |
| 121 | + |
| 122 | +NODE package_downloads_initial_aggregation |
| 123 | +SQL > |
| 124 | + |
| 125 | + % |
| 126 | + SELECT |
| 127 | + {% if defined(granularity) %} p.date, {% end %} |
| 128 | + sum(p.downloadsCount) as downloadsCount, |
| 129 | + sum(p.dockerDownloadsCount) as "dockerDownloadsCount", |
| 130 | + sum(p.dockerDependentsCount) as "dockerDependentsCount", |
| 131 | + sum(p.dependentPackagesCount) as "dependentPackagesCount", |
| 132 | + sum(p.dependentReposCount) as "dependentReposCount" |
| 133 | + FROM package_downloads_filtered p |
| 134 | + group by p.insightsProjectId {% if defined(granularity) %}, p.date {% end %} |
| 135 | + |
| 136 | + |
| 137 | + |
| 138 | +NODE package_downloads_timeseries_merge |
| 139 | +SQL > |
| 140 | + |
| 141 | + % |
| 142 | + {% set onlyContributions = False %} |
| 143 | + {% if defined(granularity) %} |
| 144 | + SELECT |
| 145 | + ds."startDate", |
| 146 | + ds."endDate", |
| 147 | + max(pdf.downloadsCount) AS "downloadsCount", |
| 148 | + max(pdf.dockerDownloadsCount) AS "dockerDownloadsCount", |
| 149 | + max(pdf.dockerDependentsCount) AS "dockerDependentsCount", |
| 150 | + max(pdf.dependentPackagesCount) AS "dependentPackagesCount", |
| 151 | + max(pdf.dependentReposCount) AS "dependentReposCount" |
| 152 | + FROM package_downloads_generate_timeseries ds |
| 153 | + LEFT JOIN |
| 154 | + package_downloads_initial_aggregation pdf |
| 155 | + ON CASE |
| 156 | + WHEN {{ granularity }} = 'daily' |
| 157 | + THEN toDate(pdf.date) |
| 158 | + WHEN {{ granularity }} = 'weekly' |
| 159 | + THEN toStartOfWeek(pdf.date) |
| 160 | + WHEN {{ granularity }} = 'monthly' |
| 161 | + THEN toStartOfMonth(pdf.date) |
| 162 | + WHEN {{ granularity }} = 'quarterly' |
| 163 | + THEN toStartOfQuarter(pdf.date) |
| 164 | + WHEN {{ granularity }} = 'yearly' |
| 165 | + THEN toStartOfYear(pdf.date) |
| 166 | + END |
| 167 | + = ds."startDate" |
| 168 | + GROUP BY ds."startDate", ds."endDate" |
| 169 | + order by ds."startDate" |
| 170 | + {% else %} SELECT 1 |
| 171 | + {% end %} |
| 172 | + |
| 173 | + |
| 174 | + |
| 175 | +NODE package_downloads_result |
| 176 | +SQL > |
| 177 | + |
| 178 | + % |
| 179 | + {% if not defined(granularity) %} SELECT * FROM package_downloads_initial_aggregation |
| 180 | + {% else %} select * from package_downloads_timeseries_merge |
| 181 | + {% end %} |
| 182 | + |
| 183 | + |
0 commit comments