diff options
| -rw-r--r-- | python/refcat/tasks.py | 67 |
1 files changed, 28 insertions, 39 deletions
diff --git a/python/refcat/tasks.py b/python/refcat/tasks.py index aaa0551..54a5890 100644 --- a/python/refcat/tasks.py +++ b/python/refcat/tasks.py @@ -186,6 +186,22 @@ class OpenLibraryDump(luigi.ExternalTask, Refcat): return luigi.LocalTarget(path=settings.OL_DUMP, format=Zstd) +class OpenLibraryEditions(luigi.ExternalTask, Refcat): + """ + Editions file. + """ + def output(self): + return luigi.LocalTarget(path=settings.OL_DUMP_EDITIONS, format=Zstd) + + +class OpenLibraryWorks(luigi.ExternalTask, Refcat): + """ + Works file. + """ + def output(self): + return luigi.LocalTarget(path=settings.OL_DUMP_WORKS, format=Zstd) + + # ----8< Derivations @@ -687,74 +703,47 @@ class BrefZipFuzzy(Refcat): return luigi.LocalTarget(path=self.path(ext="json.zst"), format=Zstd) -class OpenLibraryWorks(Refcat): +class OpenLibraryEditionsByWork(Refcat): """ - Extract just the works. + Have editions keyed by work id, 9m5.037s. """ def requires(self): - return OpenLibraryDump() + return OpenLibraryEditions() def run(self): output = shellout(""" zstdcat -T0 {input} | - parallel -j {n} --block 10M --pipe "jq -rc 'select(.type == \\"work\\")'" | + cut -f 5 | + skate-map -skip-on-empty 1 -m ff -x 'works.0.key' | + LC_ALL=C sort -T {tmpdir} -S25% -k1,1 --parallel 4 | zstd -T0 -c > {output} """, - n=self.n, tmpdir=self.tmpdir, input=self.input().path) luigi.LocalTarget(output).move(self.output().path) def output(self): - return luigi.LocalTarget(path=self.path(ext="json.zst"), format=Zstd) + return luigi.LocalTarget(path=self.path(ext="tsv.zst"), format=Zstd) -class OpenLibraryMapped(Refcat): +class OpenLibraryWorksSorted(Refcat): """ - Map OL data with e.g. a title normalizing mapper. 4m8.341s. + Sorted by work id. """ - mapper = luigi.Parameter(default="ts", description="mapper short name") - def requires(self): return OpenLibraryWorks() def run(self): output = shellout(""" zstdcat -T0 {input} | - skate-conv -f ol | - skate-map -m {mapper} -skip-on-empty 1 | - LC_ALL=C sort -T {tmpdir} -k1,1 -S25% --parallel 4 | + cut -f 2,5 | + LC_ALL=C sort -T {tmpdir} -S25% -k1,1 --parallel 4 | zstd -T0 -c > {output} """, - n=self.n, - mapper=self.mapper, tmpdir=self.tmpdir, input=self.input().path) - luigi.LocalTarget(output).move(self.output().path) - def output(self): - return luigi.LocalTarget(path=self.path(ext="tsv.zst"), format=Zstd) - -class BrefZipFuzzyOpenLibrary(Refcat): - """ - Fuzzy matching from refs to open library (both converted to release schema). - """ - mapper = luigi.Parameter(default="ts", description="mapper short name") - - def requires(self): - return { - "refs": RefsMapped(mapper=self.mapper), - "ol": OpenLibraryMapped(mapper=self.mapper), - } - - def run(self): - output = shellout(r""" - skate-reduce -m ol -F <(zstdcat -T0 {refs}) -O <(zstdcat -T0 {ol}) | - zstd -c -T0 > {output} - """, - refs=self.input().get("refs").path, - ol=self.input().get("ol").path) luigi.LocalTarget(output).move(self.output().path) def output(self): - return luigi.LocalTarget(path=self.path(ext="json.zst"), format=Zstd) + return luigi.LocalTarget(path=self.path(ext="tsv.zst"), format=Zstd) |