OILS / devtools / refactor.sh View on Github | oilshell.org

279 lines, 165 significant
1#!/usr/bin/env bash
2#
3# Junk drawer for refactoring. Also see test/lint.sh
4#
5# Usage:
6# devtools/refactor.sh <function name>
7
8set -o nounset
9set -o pipefail
10set -o errexit
11
12source devtools/run-task.sh # run-task
13
14change-kind() {
15 local kind=$1
16 local kind2=${2:-$kind}
17
18 # First make it all lower case
19 sed -r -i "s/${kind}_([A-Z]+)/Id.${kind2}_\\L\\1/g" */*.py
20
21 # Now make the first one upper case
22 sed -r -i "s/${kind}_([a-zA-Z]+)/${kind2}_\\u\\1/g" */*.py
23}
24
25k2() {
26 local kind=$1
27 local lower=${kind,,}
28
29 local title=${lower^}
30 local replace=${2:-$title}
31
32 sed -r -i "s/Id.${kind}_/Id.${replace}_/g" */*.py
33 sed -r -i "s/TokenKind.${kind}/TokenKind.${replace}/g" */*.py
34}
35
36# Execute a bunch of find/replace pairs in a text file.
37replace() {
38 local file=$1
39 local include_asdl=${2:-}
40
41 # NOTE: Escaping here is messed up. sed doesn't have --name like awk?
42 # To match literal parentheses I had to double-escape like this
43 # (shell-escape, then sed-escape).
44 # MakeMatcher\\(\\) MATCHER
45
46 local -a files=( */*.py )
47 if test -n "$include_asdl"; then
48 files+=( */*.asdl )
49 fi
50
51 while read pat replace; do
52 sed -r -i "s/${pat}/${replace}/g" "${files[@]}"
53
54 # word-anchored version
55 #sed -r -i "s/\b${pat}\b/${replace}/g" "${files[@]}"
56 done < $file
57}
58
59replace2() {
60 #sed -r -i "s/^from osh import parse_lib/from frontend import parse_lib/g" */*.py
61 #sed -r -i "s/^from core import libstr/from osh import string_ops/g" */*.py
62 #sed -r -i "s/^from osh import word$/from osh import word_/g" */*.py
63 #sed -r -i 's/from _devbuild.gen.syntax_asdl import word as osh_word/from _devbuild.gen.syntax_asdl import word/g' */*.py
64 #sed -r -i 's/osh_word/word/g' */*.py
65
66 if test -n ''; then
67 sed -r -i 's/bool_expr.BoolUnary/bool_expr.Unary/g' */*.py
68 sed -r -i 's/bool_expr.BoolBinary/bool_expr.Binary/g' */*.py
69 sed -r -i 's/bool_expr_e.BoolUnary/bool_expr_e.Unary/g' */*.py
70 sed -r -i 's/bool_expr_e.BoolBinary/bool_expr_e.Binary/g' */*.py
71 sed -r -i 's/bool_expr__BoolUnary/bool_expr__Unary/g' */*.py
72 sed -r -i 's/bool_expr__BoolBinary/bool_expr__Binary/g' */*.py
73 fi
74
75 sed -r -i 's/command.SimpleCommand/command.Simple/g' */*.py
76 sed -r -i 's/command_e.SimpleCommand/command_e.Simple/g' */*.py
77 sed -r -i 's/command__SimpleCommand/command__Simple/g' */*.py
78}
79
80trailing-ws() {
81 sed -r -i 's/[ ]+$//g' "$@"
82}
83
84#
85# OLD STUFF
86#
87
88# Hm all of the solutions involve grep --perl or perl itself?
89# https://stackoverflow.com/questions/3001177/how-do-i-grep-for-all-non-ascii-characters-in-unix
90
91# Found a latin-1 character in Python-2.7.13/Lib/heapq.py. Had to add LC_ALL=C.
92grep-unicode() {
93 LC_ALL=C grep --color='auto' --perl -n '[^\x00-\x7F]' "$@"
94}
95
96find-old-asdl() {
97 egrep 'import.*\bruntime\b' */*.py || true
98 echo ---
99
100 # Only tests left
101 egrep 'import.*\bast\b' */*.py || true
102}
103
104# This should be cleaned up
105grep-span-funcs() {
106 grep MostSpan {osh,core,frontend}/*.py
107}
108
109cmd-val() {
110 local file=$1
111 sed -i 's/arg_vec.strs/cmd_val.argv/g' $file
112 sed -i 's/arg_vec.spids/cmd_val.arg_spids/g' $file
113 sed -i 's/arg_vector/cmd_value__Argv/g' $file
114 sed -i 's/arg_vec/cmd_val/g' $file
115}
116
117id-kind() {
118 sed --regexp-extended -i 's/import id_kind$/import id_kind_def/' */*.py
119 sed --regexp-extended -i 's/id_kind[.]/id_kind_def./g' */*.py
120}
121
122options() {
123 #sed -i 's/simple_echo/simple_echo/g' */*.{py,md,sh}
124
125 #sed -i 's/simple_eval_builtin/simple_eval_builtin/g' */*.{py,md,sh}
126 #sed -i 's/simple_trap/simple_trap/g' */*.{py,md,sh}
127
128 sed -i 's/parse_backslash/parse_backslash/g' */*.{py,md,sh}
129 sed -i 's/parse_backticks/parse_backticks/g' */*.{py,md,sh}
130 sed -i 's/parse_dollar/parse_dollar/g' */*.{py,md,sh}
131}
132
133rename() {
134 cat cpp/cc.txt | while read name; do
135 local base=$(basename $name .cc)
136 local new=leaky_${base%_leaky}.cc
137 echo $name $new
138
139 #sed -i "s/$name/$new/g" */*.{py,sh} || true
140 git mv cpp/$name cpp/$new
141 done
142 return
143
144 cat cpp/h.txt | while read name; do
145 local base=$(basename $name .h)
146 local new=leaky_${base%_leaky}.h
147 echo $name $new
148
149 #sed -i "s/$name/$new/g" */*.{sh,py,cc,h}
150 git mv cpp/$name cpp/$new
151 done
152 return
153}
154
155revert() {
156 cat cpp/cc.txt | while read name; do
157 local new=$(basename $name .cc)_leaky.cc
158 echo $name $new
159
160 mv cpp/$new cpp/$name
161 done
162}
163
164#
165# Things we want to get rid of
166#
167
168show-usages() {
169 local out=$1
170 shift
171 "$@" | grep -v _test.py | tee $out
172 echo
173 wc -l $out
174}
175
176# 2024-02: 36 usages. Maybe use mylib.ToMachineInt() or mylib.ToInt32(). exit
177# status is a machine int?
178# Sometimes the lexer will validate, as with converting 2>&1
179
180int-convert() {
181 show-usages _tmp/int-convert \
182 egrep -n '\bint\(\b' $(metrics/source-code.sh osh-files)
183}
184
185# TokenVal() is generally bad in evaluators. However most of these are in
186# error paths, which is OK.
187#
188# 2024-02: 11 instances
189TokenVal-eval() {
190 show-usages _tmp/TokenVal-eval \
191 grep -n -w TokenVal */*_eval.py
192}
193
194TokenVal-all() {
195 show-usages _tmp/TokenVal-all \
196 grep -n -w TokenVal */*.py
197}
198
199# 2024-02: 71 left, mostly in ysh_ify which is the only thing that uses it
200spid-all() {
201 show-usages _tmp/spid-all \
202 egrep -n 'span_id|spid' */*.py
203}
204
205# 2024-04: 4 left in ysh_ify
206spid-sig() {
207 show-usages _tmp/spid-sig \
208 egrep -n 'def.*(span_id|spid)' */*.py
209}
210
211# 2024-04: 12 usages, mostly ysh_ify
212no-spid() {
213 show-usages _tmp/no-spid \
214 egrep -n 'runtime.NO_SPID' */*.py
215}
216
217# 69 instances
218loc-word() {
219 # should NOT wrap CompoundWord
220 show-usages _tmp/loc-word \
221 fgrep -n 'loc.Word(' */*.py
222}
223
224# 2023-08: 155
225loc-missing() {
226 show-usages _tmp/loc-m \
227 egrep -n 'loc.Missing' */*.py
228}
229
230mylib-python() {
231 show-usages _tmp/py \
232 egrep -n 'mylib.PYTHON' */*.py
233}
234
235asdl-create() {
236 fgrep -n 'CreateNull(alloc' */*.py */*/*.py \
237 | egrep -v '_devbuild|_test.py' | tee _tmp/asdl
238}
239
240readline() {
241 metrics/source-code.sh oils-files | xargs fgrep -n 'readline('
242}
243
244#
245# To improve code formatting
246#
247
248long-sigs() {
249 # 32 of these
250 egrep --no-filename '^[ ]*# type' */*.py \
251 | awk 'length($0) >= 80 { print length($0) $0 }' \
252 | sort -n
253}
254
255long-sigs-where() {
256 # jump to the file
257 egrep -n '^[ ]*# type' */*.py \
258 | awk 'length($0) >= 110 { print }' | tee _tmp/long
259}
260
261#
262# Refactor tests
263#
264
265print-names() {
266 egrep -o '[a-zA-Z_-]+'
267}
268
269make-sed() {
270 awk '{ print "s/" $0 "/unquoted-" $0 "/g;" }'
271}
272
273test-files() {
274 cat _tmp/r | print-names | make-sed | tee _tmp/sedr
275
276 sed -i -f _tmp/sedr test/runtime-errors.sh
277}
278
279run-task "$@"