diff options
| author | Kevin Wallace <kevin@wallace.seattle.wa.us> | 2022-08-05 16:34:29 -0700 |
|---|---|---|
| committer | Kevin Wallace <kevin@wallace.seattle.wa.us> | 2022-08-05 16:40:24 -0700 |
| commit | 6f7849c9c26c14f578b0f85a7012306ac78bd47a (patch) | |
| tree | 42bb6565949b5fc189005e9e7fc6f341e60d8e66 | |
| parent | don't show more links when search completes (diff) | |
add phraser words500K and phrases5M lists
https://lahosken.san-francisco.ca.us/frivolity/prog/phraser/
| -rw-r--r-- | Makefile | 14 |
1 files changed, 14 insertions, 0 deletions
@@ -5,6 +5,8 @@ all: enwiktionary-latest.version all: enwiktionary-latest.index all: enwiki-latest.version all: enwiki-latest.index +all: phraser-words500K.index +all: phraser-phrases5M.index .PHONY: newest newest: @@ -56,3 +58,15 @@ enwiktionary-%-snap.merge: nutrimatic/bin wikiextractor/WikiExtractor.py %-latest.index: %-latest.version make $*-$(shell cat $*-latest.version)-snap.index ln -sf $*-$(shell cat $*-latest.version)-snap.index $@ + +phraser-words500K.index: nutrimatic/bin + # ~ 3m + wget -O - https://lahosken.san-francisco.ca.us/frivolity/prog/phraser/words_500K.txt |\ + awk -F'\t' '{ for (i=0; i < $$1; i++) print $$2 }' |\ + ./merge.sh $@ + +phraser-phrases5M.index: nutrimatic/bin + # ~ 10m + wget -O - https://lahosken.san-francisco.ca.us/frivolity/prog/phraser/phrases_5M.txt |\ + awk -F'\t' '{ for (i=0; i < $$1; i++) print $$2 }' |\ + ./merge.sh $@ |