summaryrefslogtreecommitdiffstats
path: root/tools/check-for-json-errors-in-doc.sh
blob: 88675f172349f5a3d9b3787b31c647275632531d (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
#!/bin/bash

# Usage:
# check-for-json-errors-in-doc.sh [--all] [<file1>, <file2>, ...]

# Change directory to the root of the repository.
script_path=$(cd "$(dirname "${0}")" && pwd)
cd "${script_path}/.."

# Parse parameters.
if test ${#} -gt 0; then
	if test "${1}" = '--all'; then
		files='doc src'
	else
		files="${*}"
	fi
else
	# By default, check only modified files.
	files=$(git diff --name-only $(git merge-base origin/master HEAD))

	# If there is nothing to check, exit early. Otherwise, it checks everything.
	if test -z "${files}"; then
		exit 0
	fi
fi

exit_code=0

# Get the files.
files=$(find $(echo $files) -type f \( -name '*.rst' -or -name '*.json' \) -and -not -path '*/_build/*' -and -not -path '*/man/*' | sort -uV)
work_file=$(mktemp)
for file in $(echo $files); do
	json=0
	comment=0
	line_num=0
	echo "processing: $file"
	IFS=
	while read -r line; do
		line_num=$((line_num+1))
		if [ $comment -eq 0 -a $json -eq 0 -a $(echo "$line" | grep "^[A-Za-z]+\|^\s*\`" | wc -l) -eq 1 ]; then
			# ignore line if it starts with 'A-Za-z' or spaces followed by '`'
			continue
		elif [ $comment -eq 0 -a $(echo "$line" | grep "/\*" | grep -v "\*/" | wc -l) -eq 1 ]; then
			# if the line contains /* and it does not contain */ on the same line
			comment=1
			echo "" >> $work_file
			continue
		elif [ $comment -eq 1 -a $(echo "$line" | grep "\*/" | wc -l) -eq 1 ]; then
			# if the line contains */
			comment=0
			echo "" >> $work_file
			continue
		elif [ $comment -eq 0 -a $json -eq 0 -a $(echo "$line" | grep "^\s*{\|^\s*\".*{\|^\s*\[\s*$" | grep -v "}" | wc -l) -eq 1 ]; then
			# if this is not a comment and the line starts with spaces followed by '{' or by '"' followed by "{"
			json=1
			# ignore any map name before top level map
			line=$(echo "$line" | sed 's/.*{/{/g')
			echo "" > $work_file
		elif [ $comment -eq 0 -a $json -eq 1 -a $(echo "$line" | grep "^\s*[A-Za-z]\|^\s*\`" | wc -l) -eq 1 ]; then
			# if the line is not a comment and the line starts with spaces followed by 'A-Za-z' or followed by "`" and the parser is processing a json structure
			json=0
			cat $work_file | jq . > /dev/null
			if [ $? -ne 0 ]; then
				# if the jq tool returned error
				echo "file $file contains invalid JSON near line $line_num"
				echo "===start of JSON block==="
				cat $work_file
				echo "====end of JSON block===="
				exit_code=1
			fi
		fi
		if [ $comment -eq 0 -a $json -eq 1 ]; then
			if [ $(echo "$line" | grep "^\s*\.\.\s" | wc -l) -eq 1 ]; then
				echo "" >> $work_file
			else
					# if file is .json the following replace in line are done:
					# 1. delete everything after '#'
					# 2. delete everything after //
					# 3. ignore <?include?>
					# 4. replace all '[ <DATA> ]' with '[ "<DATA>" ]' where DATA contains: '-' and 'A-Za-z0-9' and ' '
					# 5. replace all ' <DATA>:' with ' "<DATA>":'
					# 6. replace all ': <DATA>' with ': "<DATA>"'
					# 7. replace '   ...' with '   "placeholder": "value"
					# 8. replace ', ... ' with ' '
					# 9. replace '   <DATA>' with '   "placeholder": "value"'
					# 10. replace ' <DATA>' with ' "placeholder"'
				if [ $(echo "$file" | grep "\.json" | wc -l) -eq 0 ]; then
					echo "$line" | cut -d "#" -f 1 | sed 's/\/\/ .*//g' | sed 's/<?.*?>//g' | sed 's/\[ <\([-A-Za-z0-9 ]*\)> \]/\[ \"<\1>\" \]/g' | sed 's/ <\(.*\)>:/ \"<\1>\":/g' | sed 's/: <\(.*\)>/: \"<\1>\"/g' | sed 's/   \.\.\./   \"placeholder\": \"value\"/g' | sed 's/, \.\.\. / /g' | sed 's/   <\(.*\)>/   \"placeholder\": \"value\"/g' | sed 's/ <\(.*\)>/ \"placeholder\"/g' >> $work_file
				else
					# if file is .rst the following replace in line are done:
					# 1. delete everything after '#'
					# 2. delete everything after //
					# 3. ignore <?include?>
					echo "$line" | cut -d "#" -f 1 | sed 's/\/\/ .*//g' | sed 's/<?.*?>//g' >> $work_file
				fi
			fi
		fi
	done <<< $(cat $file | tr '\n' '\r' | sed -r 's/,[[:blank:]]*\r[[:blank:]]*\.\.\.//g' | sed -r 's/\\[[:blank:]]*\r[[:blank:]]*//g' | tr '\r' '\n')
	if [ $comment -eq 0 -a $json -eq 1 ]; then
		# if the file ended but the parser is processing a json structure
		cat $work_file | jq . > /dev/null
		if [ $? -ne 0 ]; then
			# if the jq tool returned error
			echo "file $file contains invalid JSON near line $line_num"
			echo "===start of JSON block==="
			cat $work_file
			echo "====end of JSON block===="
			exit_code=1
		fi
	fi
done
rm $work_file

exit ${exit_code}