Coverage report: /home/samppa/personal/opiskelu/ohtuprojekti/pulsu/trunk/files.lisp

Kind	Covered	All	%
expression	61	185	33.0
branch	2	16	12.5

Key

Not instrumented

Conditionalized out

Executed

Not executed

Both branches taken

One branch taken

Neither branch taken

1

; This file contains functions for handling the document text, .auxil

2

; and .paf files

3

4

(in-package :puls.controller)

5

6

(defun get-document-text (docno)

7

"Gets the full text of the document identified by docno.

8

You need to have the documents directory (available from

9

/home/group/pulsu/documents) in your project root for this to work."

10

(let ((year (subseq docno 0 4))

11

(month (subseq docno 4 6)))

12

(flex:octets-to-string ; documents are stored as latin1, we want them

13

(flex:string-to-octets ; in unicode to make offsets match :/

14

(puls.util:slurp-file

15

(merge-pathnames (format nil "documents/~A/~A/~A/~A"

16

(scenario-string)

17

year month docno)

18

cl-user::*project-root*))

19

:external-format :utf-8))))

20

21

(defun get-document-title (docno)

22

"Gets title of the document"

23

(third

24

(cl-ppcre:split "\\n"

25

(get-document-text docno)

26

:limit 4)))

27

28

; auxil

29

30

(defun get-trigger-spans (row)

31

"Returns a list of trigger spans for the IE-event identified by row

32

You need to have the documents directory, see the docstring of get-document-text"

33

(let ((text (get-document-text (getf row :docno))))

34

(subseqs

35

text

36

(get-trigger-offsets row))))

37

38

(defun get-trigger-offsets (row)

39

(remove -1

40

(mapcar #'parse-integer

41

(delete-if (lambda (x) (equal x ""))

42

(cl-ppcre:split "#+" (getf row :offsets))))))

43

44

(defun get-snippet (row)

45

"Returns snippet for case described by row"

46

(when (get-trigger-offsets row)

47

(first

48

(loop :with (beg end) = (get-trigger-offsets row)

49

:with text = (get-document-text (getf row :docno))

50

:for span :in (get-spans-of-type (get-document-auxil

51

(getf row :docno))

52

:s)

53

:when (and (<= (get-begin span) beg)

54

(>= (get-end span) end))

55

:collecting (subseq text

56

(get-begin span)

57

(get-end span))))))

58

59

60

61

(defun get-document-auxil (docno)

62

(let* ((year (subseq docno 0 4))

63

(month (subseq docno 4 6))

64

(filepath (format nil "~Adocuments/~A/~A/~A/~A.auxil"

65

cl-user::*project-root*

66

(scenario-string)

67

year month docno)))

68

(if (probe-file filepath)

69

(first (puls.auxil::parse-auxil-file filepath)))))

70

71

72

73

(defun get-all-fields-of-type (docno typ &optional (typ2 :text))

74

(let (list)

75

(loop :for i :in (puls.auxil:get-spans-of-type

76

(get-document-auxil docno) typ)

77

:do (pushnew (puls.auxil:get-property i typ2) list

78

:test #'equalp))

79

list))

80

81

(defun get-canonical (docno word typ)

82

(dolist (i (puls.auxil:get-spans-of-type (get-document-auxil docno) typ))

83

(when (string= word (puls.auxil:get-property i :text))

84

(return-from get-canonical (puls.auxil:get-property i :canon)))))

85

86

87

; paf

88

89

(defun get-document-paf (docno)

90

(let* ((year (subseq docno 0 4))

91

(month (subseq docno 4 6))

92

(filepath (format nil "~Adocuments/~A/~A/~A/~A.paf"

93

cl-user::*project-root*

94

(scenario-string)

95

year month docno)))

96

(if (probe-file filepath)

97

(puls.auxil::parse-paf-file filepath))))

98

99

(defun get-document-source (docno)

100

(puls.auxil:get-property (get-document-paf docno) '(:url :url)))

101

102

103

;;;

104

105

(deftest test-get-document-source ()

106

(check (equal (get-document-source "20050101_bbc_b118e8867977ba4feae312bb5ea17ef4") NIL)))

107

108

(deftest test-get-document-auxil ()

109

(let* ((aux (get-document-auxil "20050101_bbc_b118e8867977ba4feae312bb5ea17ef4")))

110

111

(check (equal (write-to-string aux) "#I(PULS.AUXIL:AUXIL PULS.AUXIL::PROPS

112

(:DOC (:DOCID \"20050101_bbc_b118e8867977ba4feae312bb5ea17ef4\") :SPANS

113

(#I(PULS.AUXIL:ANNOTATED-SPAN PULS.AUXIL::PROPS (:TYPE :S :V :NIL)

114

PULS.AUXIL::BEGIN 55 PULS.AUXIL::END 88)

115

#I(PULS.AUXIL:ANNOTATED-SPAN PULS.AUXIL::PROPS (:TYPE :S)

116

PULS.AUXIL::BEGIN 89 PULS.AUXIL::END 203)

117

#I(PULS.AUXIL:ANNOTATED-SPAN PULS.AUXIL::PROPS (:TYPE :S)

118

PULS.AUXIL::BEGIN 204 PULS.AUXIL::END 351)

119

#I(PULS.AUXIL:ANNOTATED-SPAN PULS.AUXIL::PROPS (:TYPE :S)

120

PULS.AUXIL::BEGIN 352 PULS.AUXIL::END 439)

121

#I(PULS.AUXIL:ANNOTATED-SPAN PULS.AUXIL::PROPS (:TYPE :S)

122

PULS.AUXIL::BEGIN 440 PULS.AUXIL::END 553)

123

#I(PULS.AUXIL:ANNOTATED-SPAN PULS.AUXIL::PROPS

124

(:TYPE :STATE :TEXT \"US\" :CANON \"USA\") PULS.AUXIL::BEGIN 501

125

PULS.AUXIL::END 503)

126

#I(PULS.AUXIL:ANNOTATED-SPAN PULS.AUXIL::PROPS

127

(:TYPE :DATE :TEXT \"May 2003\" :START \"2003.05\" :END \"2003.05\"

128

:BOUNDTOEVENT :T)

129

PULS.AUXIL::BEGIN 514 PULS.AUXIL::END 522)

130

#I(PULS.AUXIL:ANNOTATED-SPAN PULS.AUXIL::PROPS (:TYPE :S)

131

PULS.AUXIL::BEGIN 554 PULS.AUXIL::END 611)

132

#I(PULS.AUXIL:ANNOTATED-SPAN PULS.AUXIL::PROPS

133

(:TYPE :STATE :TEXT \"US\" :CANON \"USA\") PULS.AUXIL::BEGIN 558

134

PULS.AUXIL::END 560)

135

#I(PULS.AUXIL:ANNOTATED-SPAN PULS.AUXIL::PROPS (:TYPE :S)

136

PULS.AUXIL::BEGIN 613 PULS.AUXIL::END 730)

137

#I(PULS.AUXIL:ANNOTATED-SPAN PULS.AUXIL::PROPS

138

(:TYPE :STATE :TEXT \"US\" :CANON \"USA\") PULS.AUXIL::BEGIN 688

139

PULS.AUXIL::END 690)

140

#I(PULS.AUXIL:ANNOTATED-SPAN PULS.AUXIL::PROPS (:TYPE :S)

141

PULS.AUXIL::BEGIN 731 PULS.AUXIL::END 824)

142

#I(PULS.AUXIL:ANNOTATED-SPAN PULS.AUXIL::PROPS (:TYPE :S)

143

PULS.AUXIL::BEGIN 826 PULS.AUXIL::END 915)

144

#I(PULS.AUXIL:ANNOTATED-SPAN PULS.AUXIL::PROPS (:TYPE :S)

145

PULS.AUXIL::BEGIN 917 PULS.AUXIL::END 1082)

146

#I(PULS.AUXIL:ANNOTATED-SPAN PULS.AUXIL::PROPS (:TYPE :S)

147

PULS.AUXIL::BEGIN 1083 PULS.AUXIL::END 1190)

148

#I(PULS.AUXIL:ANNOTATED-SPAN PULS.AUXIL::PROPS (:TYPE :S)

149

PULS.AUXIL::BEGIN 1191 PULS.AUXIL::END 1320)

150

#I(PULS.AUXIL:ANNOTATED-SPAN PULS.AUXIL::PROPS

151

(:TYPE :STATE :TEXT \"US\" :CANON \"USA\") PULS.AUXIL::BEGIN 1287

152

PULS.AUXIL::END 1289)

153

#I(PULS.AUXIL:ANNOTATED-SPAN PULS.AUXIL::PROPS (:TYPE :S)

154

PULS.AUXIL::BEGIN 1321 PULS.AUXIL::END 1459)

155

#I(PULS.AUXIL:ANNOTATED-SPAN PULS.AUXIL::PROPS (:TYPE :S :V :NIL)

156

PULS.AUXIL::BEGIN 1460 PULS.AUXIL::END 1588)

157

#I(PULS.AUXIL:ANNOTATED-SPAN PULS.AUXIL::PROPS (:TYPE :S)

158

PULS.AUXIL::BEGIN 1589 PULS.AUXIL::END 1754)

159

#I(PULS.AUXIL:ANNOTATED-SPAN PULS.AUXIL::PROPS

160

(:TYPE :STATE :TEXT \"Washington\" :CANON \"USA/Washington\")

161

PULS.AUXIL::BEGIN 1613 PULS.AUXIL::END 1623)

162

#I(PULS.AUXIL:ANNOTATED-SPAN PULS.AUXIL::PROPS

163

(:TYPE :STATE :TEXT \"US\" :CANON \"USA\") PULS.AUXIL::BEGIN 1625

164

PULS.AUXIL::END 1627)

165

#I(PULS.AUXIL:ANNOTATED-SPAN PULS.AUXIL::PROPS

166

(:TYPE :STATE :TEXT \"Iraq\" :CANON \"Iraq\") PULS.AUXIL::BEGIN 1693

167

PULS.AUXIL::END 1697)

168

#I(PULS.AUXIL:ANNOTATED-SPAN PULS.AUXIL::PROPS

169

(:TYPE :DATE :TEXT \"after the 1991\" :START \"1991\" :END \"2005\")

170

PULS.AUXIL::BEGIN 1730 PULS.AUXIL::END 1744)

171

#I(PULS.AUXIL:ANNOTATED-SPAN PULS.AUXIL::PROPS (:TYPE :S)

172

PULS.AUXIL::BEGIN 1755 PULS.AUXIL::END 1926)

173

#I(PULS.AUXIL:ANNOTATED-SPAN PULS.AUXIL::PROPS

174

(:TYPE :STATE :TEXT \"the US\" :CANON \"USA\") PULS.AUXIL::BEGIN 1804

175

PULS.AUXIL::END 1810)

176

#I(PULS.AUXIL:ANNOTATED-SPAN PULS.AUXIL::PROPS

177

(:TYPE :STATE :TEXT \"Iraq\" :CANON \"Iraq\") PULS.AUXIL::BEGIN 1891

178

PULS.AUXIL::END 1895)

179

#I(PULS.AUXIL:ANNOTATED-SPAN PULS.AUXIL::PROPS

180

(:TYPE :DISEASE :TEXT \"weapons-grade anthrax\" :ALIAS \"anthrax\" :CANON

181

\"Anthrax\")

182

PULS.AUXIL::BEGIN 1904 PULS.AUXIL::END 1925))))"

183

184

))))