script for partitioning dumps (needs test)

2 files changed, 67 insertions, 0 deletions

diff --git a/extra/partition_dumps/README.md b/extra/partition_dumps/README.md
new file mode 100644
index 00000000..2e26a41b
--- /dev/null
+++ b/extra/partition_dumps/README.md
@@ -0,0 +1,25 @@
+
+This script is used to "partition" (split up) a complete JSON dump by some key.
+For example, split release dump JSON lines into separate files, one per
+journal/container.
+
+Example parititoning a sample by release type:
+
+    cat release_dump_expanded_sample.json | jq .release_type -r > release_dump_expanded_sample.release_type
+    cat release_dump_expanded_sample.release_type | sort | uniq -c | sort -nr > release_dump_expanded_sample.release_type.counts
+    cat release_dump_expanded_sample.json | paste release_dump_expanded_sample.release_type - | sort > out
+
+More production-y example using ISSN-L:
+
+    # will append otherwise
+    rm -rf ./partitioned
+
+    # it's a pretty huge sort, will need 300+ GB scratch space? this might not scale.
+    zcat release_dump_expanded.json.gz | jq .container.issnl -r > release_dump_expanded.issnl
+    zcat release_dump_expanded.json.gz | paste release_dump_expanded.issnl - | sort  | ./partition_script.py
+
+    # for verification/stats
+    cat release_dump_expanded.issnl | sort | uniq -c | sort -nr > release_dump_expanded.issnl.counts
+    
+    # cleanup
+    rm release_dump_expanded.issnl
diff --git a/extra/partition_dumps/partition_script.py b/extra/partition_dumps/partition_script.py
new file mode 100755
index 00000000..edcc7e60
--- /dev/null
+++ b/extra/partition_dumps/partition_script.py
@@ -0,0 +1,42 @@
+#!/usr/bin/env python3
+"""
+Reads key-prefixed JSON lines from stdin, and writes out to gzipped files under
+./partitioned/.
+
+Skips empty keys and "null" (to handle a jq common-case).
+
+Eg, for tab-separated input:
+
+    something   {"a": 1}
+    something2  {"b": 2}
+
+Will write to ./partitioned/something.json.gz:
+
+    {"a": 1}
+
+(and "b" object to ./partitioned/something2.json.gz)
+"""
+
+import os, sys, gzip
+
+def run():
+    last_prefix = None
+    f = None
+    os.makedirs('partitioned', exist_ok=True)
+    
+    for line in sys.stdin:
+        (prefix, obj) = line.strip().split('\t')[:2]
+        if not prefix or prefix == "null":
+            continue
+        if prefix != last_prefix:
+            if f:
+                f.close()
+            f = gzip.GzipFile('partitioned/{}.json.gz'.format(prefix), 'a')
+        f.write(obj.encode('utf-8'))
+        f.write(b"\n")
+        last_prefix = prefix
+    if f:
+        f.close()
+
+if __name__=="__main__":
+    run()