pyrox.dev
1#! /bin/sh
2
3export LANG=C LC_ALL=C LC_COLLATE=C
4
5# Load git log
6raw_git_log="$(git log)"
7git_data="$(echo "$raw_git_log" | grep 'Author:' |
8 sed -e 's/^ *Author://; s/\\//g; s/^ *//; s/ *$//;
9 s/ @ .*//; s/ *[<]/\t/; s/[>]//')"
10
11# Name - nick - email correspondence from log and from maintainer list
12# Also there are a few manual entries
13maintainers="$(cat "$(dirname "$0")/../maintainer-list.nix" |
14 grep '=' | sed -re 's/\\"/''/g;
15 s/[ ]*([^ =]*)[ ]*=[ ]*" *(.*[^ ]) *[<](.*)[>] *".*/\1\t\2\t\3/')"
16git_lines="$( ( echo "$git_data";
17 cat "$(dirname "$0")/vanity-manual-equalities.txt") | sort |uniq)"
18
19emails="$(
20 ( echo "$maintainers" | cut -f 3; echo "$git_data" | cut -f 2 ) |
21 sort | uniq | grep -E ".+@.+[.].+"
22 )"
23
24fetchGithubName () {
25 commitid="$(
26 echo "$raw_git_log" | grep -B3 "Author: .*[<]$1[>]" | head -n 3 |
27 grep '^commit ' | tail -n 1 | sed -e 's/^commit //'
28 )"
29 userid="$(
30 curl https://github.com/NixOS/nixpkgs/commit/"$commitid" 2>/dev/null |
31 grep committed -B10 | grep 'href="/' |
32 sed -re 's@.* href="/@@; s@".*@@' |
33 grep -v "/commit/"
34 )";
35 echo "$userid"
36}
37
38[ -n "$NIXPKGS_GITHUB_NAME_CACHE" ] && {
39 echo "$emails" | while read email; do
40 line="$(grep "$email " "$NIXPKGS_GITHUB_NAME_CACHE")"
41 [ -z "$line" ] && {
42 echo "$email $(fetchGithubName "$email")" >> \
43 "$NIXPKGS_GITHUB_NAME_CACHE"
44 }
45 done
46}
47
48# For RDF
49normalize_name () {
50 sed -e 's/%/%25/g; s/ /%20/g; s/'\''/%27/g; s/"/%22/g; s/`/%60/g; s/\^/%5e/g; '
51}
52
53denormalize_name () {
54 sed -e 's/%20/ /g; s/%27/'\''/g; s/%22/"/g; s/%60/`/g; s/%5e/^/g; s/%25/%/g;';
55}
56
57n3="$(mktemp --suffix .n3)"
58
59# «The same person» relation and a sorting hint
60# Full name is something with a space
61(
62echo "$git_lines" | sed -re 's@(.*)\t(.*)@<my://name/\1> <my://can-be> <my://name/\2>.@'
63echo "$git_lines" | sed -re 's@(.*)\t(.*)@<my://name/\2> <my://can-be> <my://name/\1>.@'
64echo "$maintainers" | sed -re 's@(.*)\t(.*)\t(.*)@<my://name/\1> <my://can-be> <my://name/\2>.@'
65echo "$maintainers" | sed -re 's@(.*)\t(.*)\t(.*)@<my://name/\2> <my://can-be> <my://name/\3>.@'
66echo "$maintainers" | sed -re 's@(.*)\t(.*)\t(.*)@<my://name/\3> <my://can-be> <my://name/\1>.@'
67echo "$git_lines" | grep ' ' | cut -f 1 | sed -e 's@.*@<my://name/&> <my://is-name> <my://0>.@'
68echo "$git_lines" | grep -v ' ' | cut -f 1 | sed -e 's@.*@<my://name/&> <my://is-name> <my://1>.@'
69echo "$maintainers" | cut -f 2 | sed -e 's@.*@<my://name/&> <my://is-name> <my://0>.@'
70[ -n "$NIXPKGS_GITHUB_NAME_CACHE" ] && cat "$NIXPKGS_GITHUB_NAME_CACHE" |
71 grep -v " $" |
72 sed -re 's@(.*)\t(.*)@<my://name/\1> <my://at-github> <my://github/\2>.@'
73) | normalize_name | grep -E '<my://[-a-z]+>' | sort | uniq > "$n3"
74
75# Get transitive closure
76sparql="$(nix-build '<nixpkgs>' -Q -A apache-jena --no-out-link)/bin/sparql"
77name_list="$(
78 "$sparql" --results=TSV --data="$n3" "
79 select ?x ?y ?g where {
80 ?x <my://can-be>+ ?y.
81 ?x <my://is-name> ?g.
82 }
83 " | tail -n +2 |
84 sed -re 's@<my://name/@@g; s@<my://@@g; s@>@@g;' |
85 sort -k 2,3 -t ' '
86)"
87github_name_list="$(
88 "$sparql" --results=TSV --data="$n3" "
89 select ?x ?y where {
90 ?x (<my://can-be>+ / <my://at-github>) ?y.
91 }
92 " | tail -n +2 |
93 sed -re 's@<my://(name|github)/@@g; s@<my://@@g; s@>@@g;'
94)"
95
96# Take first spelling option for every person
97name_list_canonical="$(echo "$name_list" | cut -f 1,2 | uniq -f1)"
98
99cleaner_script="$(echo "$name_list_canonical" | denormalize_name |
100 sed -re 's/(.*)\t(.*)/s#^\2$#\1#g/g')"
101
102# Add github usernames
103if [ -n "$NIXPKGS_GITHUB_NAME_CACHE" ]; then
104 github_adder_script="$(mktemp)"
105 echo "$github_name_list" |
106 grep -E "$(echo "$name_list_canonical" | cut -f 2 |
107 tr '\n' '|' )" |
108 sort | uniq |
109 sed -re 's/(.*)\t(.*)/s| \1$| \1\t\2|g;/' |
110 denormalize_name > "$github_adder_script"
111else
112 github_adder_script='/dev/null'
113fi
114
115echo "$name_list" | denormalize_name
116
117echo
118
119echo "$git_data" | cut -f 1 |
120 sed -e "$cleaner_script" |
121 sort | uniq -c | sort -k1n | sed -rf "$github_adder_script" |
122 sed -re 's/^ *([0-9]+) /\1\t/'