fs-words.git

query scripts
Parent f921fd2
Christian Fraß

Christian Fraß commited on 2017-03-19 21:20:53
Zeige 10 geänderte Dateien mit 359 Einfügungen und 138 Löschungen.

.gitignore
Zeige Datei @ 9785e31
... ... 
@@ -1,3 +1,2 @@
1 
-stuff/
2 
-foo.sh
1 
+scripts/__pycache__
3 2
queries/associate.sql.tpl
Zeige Datei @ 9785e31
... ... 
@@ -0,0 +1,61 @@
1 
+select
2 
+	`x6`.`lemma` as `lemma`,
3 
+	`x7`.`iso_639_2` as `language`
4 
+from
5 
+	(
6 
+			(
7 
+				select
8 
+					`x4`.`tag` as `tag`,
9 
+					`x5`.`lemma` as `lemma`,
10 
+					`x5`.`language_id` as `language_id`
11 
+				from
12 
+					(
13 
+							(
14 
+								select
15 
+									`x2`.`tag` as `tag`,
16 
+									`x3`.`word_id` as `word_id`
17 
+								from
18 
+									(
19 
+											(
20 
+												select
21 
+													`x0`.`id` as `concept_id`,
22 
+													`x0`.`tag` as `tag`
23 
+												from
24 
+													(
25 
+															`tags` as `x0`
26 
+														inner join
27 
+															`concepts` as `x1`
28 
+														on
29 
+															(`x0`.`concept_id` = `x1`.`id`)
30 
+													)
31 
+											) as `x2`
32 
+										inner join
33 
+											`correlations` as `x3`
34 
+										on
35 
+											(`x2`.`concept_id` = `x3`.`concept_id`)
36 
+									)
37 
+							) as `x4`
38 
+						inner join
39 
+							`words` as `x5`
40 
+						on
41 
+							(`x4`.`word_id` = `x5`.`id`)
42 
+					)
43 
+			) as `x6`
44 
+		inner join
45 
+			`languages` as `x7`
46 
+		on
47 
+			(`x6`.`language_id` = `x7`.`id`)
48 
+	)
49 
+where
50 
+	(
51 
+		(`tag` = :tag)
52 
+		and
53 
+		(
54 
+			(:language is null)
55 
+			or
56 
+			(`x7`.`iso_639_1` = :language)
57 
+			or
58 
+			(`x7`.`iso_639_2` = :language)
59 
+		)
60 
+	)
61 
+;
queries/dictionary.sql.tpl
Zeige Datei @ 9785e31
... ... 
@@ -0,0 +1,30 @@
1 
+select
2 
+	`x2`.`id` as `id`,
3 
+	`x2`.`description` as `description`,
4 
+	`x2`.`tags` as `tags`,
5 
+	`x3`.`word_id` as `word_id`
6 
+from
7 
+	(
8 
+			(
9 
+				select
10 
+					`x0`.`id` as `id`,
11 
+					`x0`.`description` as `description`,
12 
+					group_concat(`x1`.`tag`) as `tags`
13 
+				from
14 
+					(
15 
+							`concepts` as `x0`
16 
+						inner join
17 
+							`tags` as `x1`
18 
+						on
19 
+							(`x0`.`id` = `x1`.`concept_id`)
20 
+					)
21 
+				group by
22 
+					`x0`.`id`
23 
+			) as `x2`
24 
+		inner join
25 
+			`correlations` as `x3`
26 
+		on
27 
+			(`x2`.`id` = `x3`.`concept_id`)
28 
+	)
29 
+;
30 
+
queries/translate.sql
Zeige Datei @ f921fd2
... ... 
@@ -1,125 +0,0 @@
1 
-SELECT
2 
-	x.word_id AS from_word_id,
3 
-	x.language AS from_language,
4 
-	x.lemma AS from_lemma,
5 
-	x.pronunciation AS from_pronunciation,
6 
-	x.concept_id AS from_concept_id,
7 
-	x.strength AS from_strength,
8 
-	y.word_id AS to_word_id,
9 
-	y.language AS to_language,
10 
-	y.lemma AS to_lemma,
11 
-	y.pronunciation AS to_pronunciation,
12 
-	y.concept_id AS to_concept_id,
13 
-	y.strength AS to_strength
14 
-FROM
15 
-	(
16 
-			(
17 
-				SELECT
18 
-					x3.word_id AS word_id,
19 
-					x3.language AS language,
20 
-					x3.lemma AS lemma,
21 
-					x3.pronunciation AS pronunciation,
22 
-					x4.concept_id AS concept_id,
23 
-					x4.strength AS strength
24 
-				FROM
25 
-					(
26 
-							(
27 
-								SELECT
28 
-									x2.id AS word_id,
29 
-									x1.iso_639_2 AS language,
30 
-									x2.lemma AS lemma,
31 
-									x2.pronunciation AS pronunciation
32 
-								FROM
33 
-									(
34 
-											(
35 
-												SELECT
36 
-													id,
37 
-													iso_639_2
38 
-												FROM
39 
-													languages
40 
-											) AS x1
41 
-										INNER JOIN
42 
-											(
43 
-												SELECT
44 
-													id,
45 
-													language_id,
46 
-													lemma,
47 
-													pronunciation
48 
-												FROM
49 
-													words
50 
-											) AS x2
51 
-										ON
52 
-											(x1.id = x2.language_id)
53 
-									)
54 
-							) AS x3
55 
-						INNER JOIN
56 
-							correlations AS x4
57 
-						ON
58 
-							(x3.word_id = x4.word_id)
59 
-					)
60 
-			) AS x
61 
-		INNER JOIN
62 
-			(
63 
-				SELECT
64 
-					y3.word_id AS word_id,
65 
-					y3.language AS language,
66 
-					y3.lemma AS lemma,
67 
-					y3.pronunciation AS pronunciation,
68 
-					y4.concept_id AS concept_id,
69 
-					y4.strength AS strength
70 
-				FROM
71 
-					(
72 
-							(
73 
-								SELECT
74 
-									y2.id AS word_id,
75 
-									y1.iso_639_2 AS language,
76 
-									y2.lemma AS lemma,
77 
-									y2.pronunciation AS pronunciation
78 
-								FROM
79 
-									(
80 
-											(
81 
-												SELECT
82 
-													id,
83 
-													iso_639_2
84 
-												FROM
85 
-													languages
86 
-											) AS y1
87 
-										INNER JOIN
88 
-											(
89 
-												SELECT
90 
-													id,
91 
-													language_id,
92 
-													lemma,
93 
-													pronunciation
94 
-												FROM
95 
-													words
96 
-											) AS y2
97 
-										ON
98 
-											(y1.id = y2.language_id)
99 
-									)
100 
-							) AS y3
101 
-						INNER JOIN
102 
-							correlations AS y4
103 
-						ON
104 
-							(y3.word_id = y4.word_id)
105 
-					)
106 
-			) AS y
107 
-		ON
108 
-			(x.concept_id = y.concept_id)
109 
-	)
110 
-WHERE
111 
-	(
112 
-		(
113 
-			(from_language = '${language}')
114 
-			AND
115 
-			(from_lemma LIKE '%${lemma}%')
116 
-		)
117 
-		AND
118 
-		NOT (
119 
-			(from_language = to_language)
120 
-			AND
121 
-			(from_lemma = to_lemma)
122 
-		)
123 
-	)
124 
-;
125 
-
queries/translate.sql.tpl
Zeige Datei @ 9785e31
... ... 
@@ -0,0 +1,137 @@
1 
+select
2 
+	`x`.`language_iso_639_2` as `from_language`,
3 
+	`x`.`lemma` as `from_lemma`,
4 
+	`x`.`pronunciation` as `from_pronunciation`,
5 
+	`x`.`strength` as `from_strength`,
6 
+	`y`.`language_iso_639_2` as `to_language`,
7 
+	`y`.`lemma` as `to_lemma`,
8 
+	`y`.`pronunciation` as `to_pronunciation`,
9 
+	`y`.`strength` as `to_strength`
10 
+from
11 
+	(
12 
+			(
13 
+				select
14 
+					`x3`.`word_id` as `word_id`,
15 
+					`x3`.`language_iso_639_1` as `language_iso_639_1`,
16 
+					`x3`.`language_iso_639_2` as `language_iso_639_2`,
17 
+					`x3`.`lemma` as `lemma`,
18 
+					`x3`.`pronunciation` as `pronunciation`,
19 
+					`x4`.`concept_id` as `concept_id`,
20 
+					`x4`.`strength` as `strength`
21 
+				from
22 
+					(
23 
+							(
24 
+								select
25 
+									`x2`.`id` as `word_id`,
26 
+									`x1`.`iso_639_1` as `language_iso_639_1`,
27 
+									`x1`.`iso_639_2` as `language_iso_639_2`,
28 
+									`x2`.`lemma` as `lemma`,
29 
+									`x2`.`pronunciation` as `pronunciation`
30 
+								from
31 
+									(
32 
+											(
33 
+												select
34 
+													`id`,
35 
+													`iso_639_1`,
36 
+													`iso_639_2`
37 
+												from
38 
+													`languages`
39 
+											) as `x1`
40 
+										inner join
41 
+											(
42 
+												select
43 
+													`id`,
44 
+													`language_id`,
45 
+													`lemma`,
46 
+													`pronunciation`
47 
+												from
48 
+													`words`
49 
+											) as `x2`
50 
+										on
51 
+											(`x1`.`id` = `x2`.`language_id`)
52 
+									)
53 
+							) as `x3`
54 
+						inner join
55 
+							`correlations` as `x4`
56 
+						on
57 
+							(`x3`.`word_id` = `x4`.`word_id`)
58 
+					)
59 
+			) as x
60 
+		inner join
61 
+			(
62 
+				select
63 
+					`y3`.`word_id` as `word_id`,
64 
+					`y3`.`language_iso_639_1` as `language_iso_639_1`,
65 
+					`y3`.`language_iso_639_2` as `language_iso_639_2`,
66 
+					`y3`.`lemma` as `lemma`,
67 
+					`y3`.`pronunciation` as `pronunciation`,
68 
+					`y4`.`concept_id` as `concept_id`,
69 
+					`y4`.`strength` as `strength`
70 
+				from
71 
+					(
72 
+							(
73 
+								select
74 
+									`y2`.`id` as `word_id`,
75 
+									`y1`.`iso_639_1` as `language_iso_639_1`,
76 
+									`y1`.`iso_639_2` as `language_iso_639_2`,
77 
+									`y2`.`lemma` as `lemma`,
78 
+									`y2`.`pronunciation` as `pronunciation`
79 
+								from
80 
+									(
81 
+											(
82 
+												select
83 
+													`id`,
84 
+													`iso_639_1`,
85 
+													`iso_639_2`
86 
+												from
87 
+													`languages`
88 
+											) as `y1`
89 
+										inner join
90 
+											(
91 
+												select
92 
+													`id`,
93 
+													`language_id`,
94 
+													`lemma`,
95 
+													`pronunciation`
96 
+												from
97 
+													`words`
98 
+											) as `y2`
99 
+										on
100 
+											(`y1`.`id` = `y2`.`language_id`)
101 
+									)
102 
+							) as `y3`
103 
+						inner join
104 
+							`correlations` as `y4`
105 
+						on
106 
+							(`y3`.`word_id` = `y4`.`word_id`)
107 
+					)
108 
+			) as `y`
109 
+		on
110 
+			(`x`.`concept_id` = `y`.`concept_id`)
111 
+	)
112 
+where
113 
+	(
114 
+		(
115 
+			(`x`.`language_iso_639_1` = :from_language)
116 
+			or
117 
+			(`x`.`language_iso_639_2` = :from_language)
118 
+		)
119 
+		and
120 
+		(`from_lemma` like ('%' || :from_lemma || '%'))
121 
+		and
122 
+		(
123 
+			(:to_language is null)
124 
+			or
125 
+			(`y`.`language_iso_639_1` = :to_language)
126 
+			or
127 
+			(`y`.`language_iso_639_2` = :to_language)
128 
+		)
129 
+		and
130 
+		not (
131 
+			(`x`.`language_iso_639_2` = `y`.`language_iso_639_2`)
132 
+			and
133 
+			(`x`.`lemma` = `y`.`lemma`)
134 
+		)
135 
+	)
136 
+;
137 
+
scripts/associate.py
Zeige Datei @ 9785e31
... ... 
@@ -0,0 +1,29 @@
1 
+#!/usr/bin/env python3
2 
+
3 
+import sys as _sys
4 
+
5 
+from helpers import *
6 
+
7 
+
8 
+def error_syntax():
9 
+	print("SYNTAX: translate.py <language> <lemma>")
10 
+	_sys.exit(-1)
11 
+
12 
+
13 
+def main(args):
14 
+	tag = (error_syntax() if (len(args) < 1) else args.pop(0))
15 
+	language = (None if (len(args) < 1) else args.pop(0))
16 
+	macro(
17 
+		"associate",
18 
+		{
19 
+			"tag": tag,
20 
+			"language": language,
21 
+		},
22 
+		(
23 
+			"lemma",
24 
+			"language",
25 
+		)
26 
+	)
27 
+
28 
+main(_sys.argv[1:])
29 
+
scripts/dictionary.py
Zeige Datei @ 9785e31
... ... 
@@ -0,0 +1,32 @@
1 
+#!/usr/bin/env python3
2 
+
3 
+import sys as _sys
4 
+
5 
+from helpers import *
6 
+
7 
+
8 
+def error_syntax():
9 
+	print("SYNTAX: dictionary.py <from-language> <to-language>")
10 
+	_sys.exit(-1)
11 
+
12 
+
13 
+def main(args):
14 
+	from_language = (error_syntax() if (len(args) < 1) else args.pop(0))
15 
+	to_language = (error_syntax() if (len(args) < 1) else args.pop(0))
16 
+	macro(
17 
+		"dictionary",
18 
+		{
19 
+			"from_language": from_language,
20 
+			"to_language": to_language,
21 
+		},
22 
+		(
23 
+			"id",
24 
+			"description",
25 
+			"tags",
26 
+			"word_id",
27 
+		)
28 
+	)
29 
+
30 
+
31 
+main(_sys.argv[1:])
32 
+
scripts/helpers.py
Zeige Datei @ 9785e31
... ... 
@@ -0,0 +1,31 @@
1 
+#!/usr/bin/env python3
2 
+
3 
+import os as _os
4 
+import sqlite3 as _sqlite3
5 
+import json as _json
6 
+
7 
+
8 
+def pattern(content):
9 
+	return ("%%%s%%" % content)
10 
+
11 
+
12 
+def transform(entries, names):
13 
+	# return entries
14 
+	return list(map(lambda entry: dict(zip(names, entry)), entries))
15 
+
16 
+
17 
+def query(name, arguments = {}):
18 
+	path = _os.path.join("queries", "%s.sql.tpl" % name)
19 
+	handle = open(path, "r")
20 
+	querytemplate = handle.read()
21 
+	handle.close()
22 
+	connection = _sqlite3.connect(_os.path.join("db", "words.sqlite"))
23 
+	result = connection.cursor().execute(querytemplate, arguments).fetchall()
24 
+	connection.close()
25 
+	return result
26 
+
27 
+
28 
+def macro(query_name, query_arguments, transform_names):
29 
+	print(_json.dumps(transform(query(query_name, query_arguments), transform_names), indent = "\t"))
30 
+
31 
+
scripts/translate.py
Zeige Datei @ 9785e31
... ... 
@@ -0,0 +1,38 @@
1 
+#!/usr/bin/env python3
2 
+
3 
+import sys as _sys
4 
+
5 
+from helpers import *
6 
+
7 
+
8 
+def error_syntax():
9 
+	print("SYNTAX: translate.py <from-language> <from-lemma> [<to-language>]")
10 
+	_sys.exit(-1)
11 
+
12 
+
13 
+def main(args):
14 
+	from_language = (error_syntax() if (len(args) < 1) else args.pop(0))
15 
+	from_lemma = (error_syntax() if (len(args) < 1) else args.pop(0))
16 
+	to_language = (None if (len(args) < 1) else args.pop(0))
17 
+	macro(
18 
+		"translate",
19 
+		{
20 
+			"from_language": from_language,
21 
+			"from_lemma": from_lemma,
22 
+			"to_language": to_language,
23 
+		},
24 
+		(
25 
+			"from_language",
26 
+			"from_lemma",
27 
+			"from_pronunciation",
28 
+			"from_strength",
29 
+			"to_language",
30 
+			"to_lemma",
31 
+			"to_pronunciation",
32 
+			"to_strength",
33 
+		)
34 
+	)
35 
+
36 
+
37 
+main(_sys.argv[1:])
38 
+
scripts/translate.sh
Zeige Datei @ f921fd2
... ... 
@@ -1,11 +0,0 @@
1 
-#!/usr/bin/env sh
2 
-
3 
-if [ "$#" -ge 1 ] ; then language=$1 && shift ; else language="*" ; fi
4 
-if [ "$#" -ge 1 ] ; then lemma=$1 && shift ; else lemma="*" ; fi
5 
-
6 
-cat queries/translate.sql \
7 
-	| sed --expression="s:\${language}:${language}:g" \
8 
-	| sed --expression="s:\${lemma}:${lemma}:g" \
9 
-	| sqlite3 db/words.sqlite \
10 
-	| cat -
11 
-
12 0