vocab unittest

vecto-ai · Apr 25, 2018 · d551909 · d551909
1 parent b710f75
commit d551909
Show file tree

Hide file tree

Showing 11 changed files with 2,883 additions and 0 deletions.
diff --git a/tests/data/corpora/annotated/sense_small.txt.annotated b/tests/data/corpora/annotated/sense_small.txt.annotated
diff --git a/tests/data/corpora/gzipped/sense_small.txt.gz b/tests/data/corpora/gzipped/sense_small.txt.gz
diff --git a/tests/data/corpora/multiple_files/emma_small.txt b/tests/data/corpora/multiple_files/emma_small.txt
diff --git a/tests/data/corpora/multiple_files/sense_small.txt b/tests/data/corpora/multiple_files/sense_small.txt
diff --git a/tests/data/corpora/multiple_small/one.txt b/tests/data/corpora/multiple_small/one.txt
@@ -0,0 +1 @@
+one two three four five six. seven eight nine ten
diff --git a/tests/data/corpora/multiple_small/two.txt b/tests/data/corpora/multiple_small/two.txt
@@ -0,0 +1 @@
+eleven tweleve thirteen fourteen
diff --git a/tests/data/corpora/plain/sense_small.txt b/tests/data/corpora/plain/sense_small.txt
diff --git a/tests/data/vocabs/numbers/metadata.json b/tests/data/vocabs/numbers/metadata.json
@@ -0,0 +1,4 @@
+{
+    "path_source": "./test/data/corpora/numbers",
+    "vsmlib_version": "0.1.6"
+}
diff --git a/tests/data/vocabs/numbers/vocab.tsv b/tests/data/vocabs/numbers/vocab.tsv
@@ -0,0 +1,13 @@
+#word	frequency
+one	1
+two	406
+three	345
+four	330
+five	324
+six	271
+seven	184
+eight	177
+nine	176
+ten	10
+eleven	146
+twelve	170
diff --git a/tests/data/vocabs/plain/metadata.json b/tests/data/vocabs/plain/metadata.json
@@ -0,0 +1,6 @@
+{
+    "cnt_words": 142,
+    "min_frequency": 10,
+    "path_source": "./test/data/corpora/plain",
+    "vsmlib_version": "0.1.6"
+}
diff --git a/tests/data/vocabs/plain/vocab.tsv b/tests/data/vocabs/plain/vocab.tsv
@@ -0,0 +1,143 @@
+#word	frequency
+,	671
+.	406
+of	345
+the	330
+to	324
+and	271
+her	184
+was	177
+his	176
+a	170
+it	146
+in	145
+for	119
+be	118
+she	107
+he	101
+as	97
+i	95
+that	92
+not	82
+their	70
+him	69
+by	65
+had	63
+which	63
+but	62
+at	60
+them	60
+no	60
+have	58
+with	57
+so	56
+on	54
+you	54
+is	50
+from	47
+would	47
+they	45
+could	45
+will	44
+dashwood	42
+!	42
+my	39
+were	38
+more	38
+than	37
+very	36
+mrs	35
+all	34
+any	34
+mother	33
+house	32
+such	31
+every	29
+elinor	27
+this	26
+do	26
+norland	25
+own	25
+what	25
+if	25
+who	24
+an	24
+been	23
+one	23
+much	23
+or	23
+john	21
+your	21
+might	20
+pounds	19
+when	19
+think	19
+said	19
+himself	18
+too	18
+should	18
+great	17
+only	17
+how	17
+must	17
+may	17
+are	16
+there	16
+can	16
+far	15
+make	15
+though	15
+marianne	15
+soon	14
+father	14
+thousand	14
+well	14
+did	14
+some	14
+we	14
+man	13
+sister	13
+mr	13
+present	13
+first	13
+other	13
+time	13
+give	13
+now	13
+herself	13
+sure	13
+shall	13
+edward	13
+many	12
+opinion	12
+into	12
+fortune	12
+half	12
+really	12
+sisters	12
+thing	12
+enough	12
+day	12
+me	12
+say	12
+taste	12
+good	11
+years	11
+three	11
+comfortable	11
+handsome	11
+little	11
+love	11
+?	11
+am	11
+barton	11
+before	10
+heart	10
+gave	10
+child	10
+most	10
+then	10
+feel	10
+ever	10
+beyond	10
+see	10