sql: script to reingest recent spn2 lookup failure in bulk mode

5 files changed, 71 insertions, 18 deletions

diff --git a/sql/dump_reingest_bulk.sql b/sql/dump_reingest_bulk.sql
new file mode 100644
index 0000000..403fb20
--- /dev/null
+++ b/sql/dump_reingest_bulk.sql
@@ -0,0 +1,25 @@
+
+COPY (
+    SELECT row_to_json(ingest_request.*) FROM ingest_request
+    LEFT JOIN ingest_file_result ON ingest_file_result.base_url = ingest_request.base_url
+    WHERE
+        (ingest_request.ingest_type = 'pdf'
+            OR ingest_request.ingest_type = 'html')
+        AND ingest_file_result.hit = false
+        AND ingest_request.created < NOW() - '24 hour'::INTERVAL
+        AND ingest_request.created > NOW() - '181 day'::INTERVAL
+        AND (ingest_request.ingest_request_source = 'fatcat-changelog'
+             OR ingest_request.ingest_request_source = 'fatcat-ingest')
+        AND (
+            ingest_file_result.status like 'spn2-%'
+            OR ingest_file_result.status like 'cdx-error'
+            OR ingest_file_result.status like 'petabox-error'
+        )
+        AND ingest_file_result.status != 'spn2-error:invalid-url-syntax'
+        AND ingest_file_result.status != 'spn2-error:filesize-limit'
+        AND ingest_file_result.status != 'spn2-error:not-found'
+        AND ingest_file_result.status != 'spn2-error:blocked-url'
+        AND ingest_file_result.status != 'spn2-error:too-many-redirects'
+        AND ingest_file_result.status != 'spn2-error:network-authentication-required'
+        AND ingest_file_result.status != 'spn2-error:unknown'
+) TO '/srv/sandcrawler/tasks/reingest_bulk_current.rows.json';
diff --git a/sql/dump_reingest_quarterly.sql b/sql/dump_reingest_quarterly.sql
index 725a404..c425a15 100644
--- a/sql/dump_reingest_quarterly.sql
+++ b/sql/dump_reingest_quarterly.sql
@@ -2,7 +2,9 @@
 COPY (
     SELECT row_to_json(ingest_request.*) FROM ingest_request
     LEFT JOIN ingest_file_result ON ingest_file_result.base_url = ingest_request.base_url
-    WHERE ingest_request.ingest_type = 'pdf'
+    WHERE
+        (ingest_request.ingest_type = 'pdf'
+            OR ingest_request.ingest_type = 'html')
         AND ingest_file_result.hit = false
         AND ingest_request.created < NOW() - '8 hour'::INTERVAL
         AND ingest_request.created > NOW() - '91 day'::INTERVAL
@@ -10,11 +12,12 @@ COPY (
              OR ingest_request.ingest_request_source = 'fatcat-ingest')
         AND (
             ingest_file_result.status like 'spn2-%'
-            OR ingest_file_result.status like 'cdx-error'
-            OR ingest_file_result.status like 'wayback-error'
-            OR ingest_file_result.status like 'wayback-content-error'
-            OR ingest_file_result.status like 'petabox-error'
-            OR ingest_file_result.status like 'gateway-timeout'
+            OR ingest_file_result.status = 'cdx-error'
+            OR ingest_file_result.status = 'wayback-error'
+            OR ingest_file_result.status = 'wayback-content-error'
+            OR ingest_file_result.status = 'petabox-error'
+            OR ingest_file_result.status = 'gateway-timeout'
+            OR ingest_file_result.status = 'no-capture'
         )
         AND ingest_file_result.status != 'spn2-error:invalid-url-syntax'
         AND ingest_file_result.status != 'spn2-error:filesize-limit'
diff --git a/sql/dump_reingest_spn.sql b/sql/dump_reingest_spn.sql
index 6ef08c1..b0051dd 100644
--- a/sql/dump_reingest_spn.sql
+++ b/sql/dump_reingest_spn.sql
@@ -2,18 +2,21 @@
 COPY (
     SELECT row_to_json(ingest_request.*) FROM ingest_request
     LEFT JOIN ingest_file_result ON ingest_file_result.base_url = ingest_request.base_url
-    WHERE ingest_request.ingest_type = 'pdf'
+    WHERE
+        (ingest_request.ingest_type = 'pdf'
+            OR ingest_request.ingest_type = 'html')
         AND ingest_file_result.hit = false
         AND ingest_request.created < NOW() - '6 hour'::INTERVAL
         AND ingest_request.created > NOW() - '180 day'::INTERVAL
         AND ingest_request.ingest_request_source = 'savepapernow-web'
         AND (
             ingest_file_result.status like 'spn2-%'
-            -- OR ingest_file_result.status like 'cdx-error'
-            -- OR ingest_file_result.status like 'wayback-error'
-            -- OR ingest_file_result.status like 'wayback-content-error'
-            OR ingest_file_result.status like 'petabox-error'
-            -- OR ingest_file_result.status like 'gateway-timeout'
+            -- OR ingest_file_result.status = 'cdx-error'
+            -- OR ingest_file_result.status = 'wayback-error'
+            -- OR ingest_file_result.status = 'wayback-content-error'
+            OR ingest_file_result.status = 'petabox-error'
+            -- OR ingest_file_result.status = 'gateway-timeout'
+            OR ingest_file_result.status = 'no-capture'
         )
         AND ingest_file_result.status != 'spn2-error:invalid-url-syntax'
         AND ingest_file_result.status != 'spn2-error:filesize-limit'
diff --git a/sql/dump_reingest_weekly.sql b/sql/dump_reingest_weekly.sql
index 65800eb..e529945 100644
--- a/sql/dump_reingest_weekly.sql
+++ b/sql/dump_reingest_weekly.sql
@@ -2,7 +2,9 @@
 COPY (
     SELECT row_to_json(ingest_request.*) FROM ingest_request
     LEFT JOIN ingest_file_result ON ingest_file_result.base_url = ingest_request.base_url
-    WHERE ingest_request.ingest_type = 'pdf'
+    WHERE
+        (ingest_request.ingest_type = 'pdf'
+            OR ingest_request.ingest_type = 'html')
         AND ingest_file_result.hit = false
         AND ingest_request.created < NOW() - '8 hour'::INTERVAL
         AND ingest_request.created > NOW() - '8 day'::INTERVAL
@@ -10,11 +12,12 @@ COPY (
              OR ingest_request.ingest_request_source = 'fatcat-ingest')
         AND (
             ingest_file_result.status like 'spn2-%'
-            -- OR ingest_file_result.status like 'cdx-error'
-            -- OR ingest_file_result.status like 'wayback-error'
-            -- OR ingest_file_result.status like 'wayback-content-error'
-            OR ingest_file_result.status like 'petabox-error'
-            -- OR ingest_file_result.status like 'gateway-timeout'
+            -- OR ingest_file_result.status = 'cdx-error'
+            -- OR ingest_file_result.status = 'wayback-error'
+            -- OR ingest_file_result.status = 'wayback-content-error'
+            OR ingest_file_result.status = 'petabox-error'
+            -- OR ingest_file_result.status = 'gateway-timeout'
+            OR ingest_file_result.status = 'no-capture'
         )
         AND ingest_file_result.status != 'spn2-error:invalid-url-syntax'
         AND ingest_file_result.status != 'spn2-error:filesize-limit'
diff --git a/sql/reingest_bulk.sh b/sql/reingest_bulk.sh
new file mode 100755
index 0000000..d5d3e35
--- /dev/null
+++ b/sql/reingest_bulk.sh
@@ -0,0 +1,19 @@
+#!/bin/bash
+
+set -e              # fail on error
+set -u              # fail if variable not set in substitution
+set -o pipefail     # fail if part of a '|' command fails
+
+sudo -u postgres psql sandcrawler < dump_reingest_bulk.sql
+
+cd ../python
+sudo -u sandcrawler pipenv run \
+    ./scripts/ingestrequest_row2json.py /srv/sandcrawler/tasks/reingest_bulk_current.rows.json \
+    > /srv/sandcrawler/tasks/reingest_bulk_current.json
+
+cat /srv/sandcrawler/tasks/reingest_bulk_current.json \
+    | shuf \
+    | head -n1000000 \
+    | jq . -c \
+    | kafkacat -P -b wbgrp-svc263.us.archive.org -t sandcrawler-prod.ingest-file-requests-bulk -p -1
+