110 Commits

Author SHA1 Message Date
Vidar Holen
08d1d37094 Stable version 0.4.6
This release is dedicated to Stardew Valley, for fixing the Harvest Moon
unplayable issue where Ann gives up on her inventions after marriage ;_;
2017-03-26 11:48:27 -07:00
Vidar Holen
e28e90133d Add missing import 2017-03-18 16:42:17 -07:00
Vidar Holen
2688a81526 Don't suggest indirection for 'declare var$n=foo' 2017-03-18 14:54:52 -07:00
Vidar Holen
82c3084438 Add HandBrakeCLI to list of stdin swallowing apps 2017-03-18 13:10:54 -07:00
Russ Kubik
c6ff1933b7 Update README.md 2017-03-17 09:32:53 -07:00
Vidar Holen
22c86256ac Decode UTF-8 sequences over 0x10FFFF as latin1 2017-03-10 10:11:05 -08:00
Daniel M. Capella
a3a4873190 readme: Add ALE Vim plugin 2017-03-08 10:10:24 -08:00
Vidar Holen
750212af39 Add less common actions to find -o check. 2017-03-03 20:43:00 -08:00
Vidar Holen
2154583fd3 Don't parse unicode quotes as real quotes. 2017-02-25 15:14:52 -08:00
Vidar Holen
35c74e4747 Getting command name from busybox now gets applet name 2017-02-12 10:56:29 -08:00
Vidar Holen
46fb91b44d Manually decode input files as lenient UTF-8. 2017-01-22 15:24:21 -08:00
Vidar Holen
128d5d6013 Don't warn about grep pattern issues when using -F. 2017-01-21 16:20:55 -08:00
koalaman
41176c23a6 Merge pull request #822 from bittorf/master
SC2164: show two possible variants for circumenting the warning
2017-01-15 15:58:19 -08:00
koalaman
342726f480 Merge pull request #823 from dasilvacontin/patch-1
Add TravisCI Setup to README.md
2017-01-15 15:57:09 -08:00
Vidar Holen
1863f2f12d Warn about += bashism in sh and dash. 2017-01-14 12:40:09 -08:00
Vidar Holen
8809a36952 Warn when finding HTML entities like & 2017-01-14 11:59:31 -08:00
Vidar Holen
7f307c5775 Count | as a regex metacharacter for 2076. 2017-01-12 19:02:46 -08:00
David da Silva
eb12086d05 Fix typos in Travis CI section of the README.md 2017-01-11 11:48:57 +01:00
David da Silva
6259a32601 Add TravisCI Setup to README.md 2017-01-11 10:04:48 +01:00
Bastian Bittorf
4e13c7cbc1 SC2164: show two possible variants for circumenting the warning
always calling 'exit' is not good in e.g. functions.
the basic idea is at least that the returncode of
cd *is* evaluated somehow and not ignored.

Reported-by: Garance Alistair Drosehn <drosehn@rpi.edu>

Signed-off-by: Bastian Bittorf <bittorf@bluebottle.com>
2017-01-10 12:30:47 +01:00
Vidar Holen
edb01fa855 Warn about ( -d foo ) and similar. 2017-01-09 23:49:22 -08:00
koalaman
85e6c35845 Merge pull request #814 from Scorpiokat/master
This commit fixes #797
2017-01-03 21:25:45 -08:00
Ekaterina Efimova
43f667a8f9 Update Analytics.hs 2017-01-03 21:48:35 +03:00
Ekaterina Efimova
daacc98a8f Update Analytics.hs 2017-01-03 21:40:32 +03:00
Scorpiokat
40907b1636 This commit fixes #803 2017-01-03 14:55:12 +00:00
Scorpiokat
96168fc707 This commit fixes #797 2017-01-02 18:01:24 +00:00
Vidar Holen
6aee12a572 Warnings for braces/globs/arrays in [/[[. 2016-12-31 13:18:36 -08:00
Vidar Holen
0048f34b11 Merge branch 'master' of github.com:koalaman/shellcheck 2016-12-30 17:11:32 -08:00
Vidar Holen
e679ff222a Warn when using deprecated egrep/fgrep. 2016-12-30 16:55:31 -08:00
koalaman
6f5648faca Update README.md 2016-12-30 10:31:34 -08:00
Vidar Holen
30e94ea7ab Warn about comparisons and cases that can never match. 2016-12-29 14:04:49 -08:00
Vidar Holen
d8f8a2fa14 Minor 2% parser speedup 2016-12-29 10:56:00 -08:00
Vidar Holen
df3cc70658 Improve code for warning about escaped single quotes. 2016-12-28 21:51:28 -08:00
Vidar Holen
5669702362 Warn about missing and invalid subscripts in array assignments. 2016-12-28 18:58:03 -08:00
Vidar Holen
bd9d05c759 Warn about missing space in [ foo= bar ] 2016-12-27 21:20:59 -08:00
Austin English
af87fe9315 add missing references to dash 2016-12-20 17:10:55 -08:00
koalaman
c2850e436f Merge pull request #804 from austin987/readme2
README.md: sort package managers, add emerge (Gentoo)
2016-12-20 10:16:42 -08:00
Austin English
d73fc8f91d README.md: sort package managers, add emerge (Gentoo) 2016-12-20 12:11:10 -06:00
Vidar Holen
7124c113e8 Don't warn about .sh.version being unused (for Ksh) 2016-12-17 16:03:06 -08:00
Vidar Holen
f594f01d35 Add warnings about redirections without commands. 2016-12-17 15:03:52 -08:00
Vidar Holen
6ccb7e9129 Fix 'for file; do ..' counting $file as a safe variable. 2016-12-17 12:55:14 -08:00
Vidar Holen
838f0ce4dc Count "%(%Y%m%d)T" as a single format in printf. 2016-12-17 12:48:09 -08:00
Vidar Holen
105b09792c Fix SC095 about < /dev/null when using ssh -n 2016-12-17 12:41:00 -08:00
Vidar Holen
f6618d4332 Fix BSD style flag parsing to stop at -- 2016-12-11 10:58:01 -08:00
Vidar Holen
0a381be37b Improve heredoc delimiter handling. 2016-12-11 10:09:13 -08:00
Vidar Holen
46e47dad45 Don't quote source in 2041 and 2043. 2016-12-10 22:41:25 -08:00
Vidar Holen
fee6c94d40 Alias ash to dash with warning. 2016-12-10 10:57:01 -08:00
Vidar Holen
cf1c46d852 Add deprecation warning for tempfile 2016-12-10 09:58:27 -08:00
Vidar Holen
a40efffec9 Explicitly import Data.Monoid 2016-12-10 08:49:07 -08:00
Vidar Holen
44b96fca66 Merge branch 'bug562' 2016-12-03 10:21:49 -08:00
Vidar Holen
b4fb439191 Save string read by T_ParamSubSpecialChar 2016-12-03 10:19:14 -08:00
MortimerMcMire315
0897ab7092 Add handling for special characters in parameter substitutions.
Fixes koalaman/shellcheck#562. Special characters inside braces are
parsed into T_ParamSubSpecialChar instead of T_Literal so that they are
not flagged in the function checkInexplicablyUnquoted when sandwiched
between double quotes.
2016-11-28 16:06:02 -05:00
koalaman
9703f89f79 Merge pull request #774 from oyvindio/tag-releases-in-dockerhub
Push git tags to dockerhub as docker tags
2016-11-28 11:39:48 -08:00
koalaman
aadf02e635 Merge pull request #781 from austin987/patch-1
Add info/link on ignoring issues to README
2016-11-28 11:33:05 -08:00
Øyvind Ingebrigtsen Øvergaard
090051bdc1 Use $TRAVIS_TAG in $TAG if defined 2016-11-27 13:42:49 +01:00
Austin English
b8c96b4361 Add info/link on ignoring issues to README 2016-11-21 19:53:26 -06:00
Vidar Holen
f26038125d Allow spaces/comments before filewide annotations. 2016-11-20 18:06:36 -08:00
Vidar Holen
08f7ff37c5 Some cleanup and refactoring. 2016-11-12 15:51:36 -08:00
Øyvind Ingebrigtsen Øvergaard
60fc33ebdf Push git tags to dockerhub as docker tags 2016-11-11 20:32:05 +01:00
Vidar Holen
3a006f7bcb Use camelcase for cabal commands to align with stack. 2016-11-10 13:13:42 -08:00
koalaman
89b6fd58fa Reformat issue template 2016-10-31 18:20:39 -07:00
koalaman
069ddeffcc Merge pull request #766 from ryanoasis/issue-template
Add an issue template
2016-10-31 17:46:47 -07:00
Ryan L McIntyre
59c4ed106c Create ISSUE_TEMPLATE.md
* Suggestion for an issue template to help with the amount of issues
2016-10-31 20:25:55 -04:00
Vidar Holen
5efb724a3e Stable version 0.4.5
This release is dedicated to Google Inc for four great years of
employment and being good sports about hobby projects like this.
2016-10-21 14:00:50 -07:00
Vidar Holen
619b6c42f3 Improve parsing of fd close/duplicate redirections. 2016-10-21 11:31:58 -07:00
Vidar Holen
88c56ecd53 Allow unrecognized directives with warnings. 2016-10-14 12:14:20 -07:00
Vidar Holen
6b62b5bf7e Don't warn about [ a '>' b ] needing escapes. 2016-10-01 14:54:28 -07:00
Vidar Holen
8672af29ef Split duplicate SC1009 into SC1014 for if [ grep foo bar ] 2016-10-01 13:34:14 -07:00
Vidar Holen
1a8e34bfea Don't suggest grep -c when used with -o 2016-10-01 13:26:53 -07:00
Vidar Holen
b0dae063bf Add info when using 'find' without path 2016-09-25 11:56:32 -07:00
Vidar Holen
4fc4899803 Consider ${foo:=bar} an assignment. 2016-09-24 19:01:13 -07:00
Vidar Holen
cd4896192c Don't consider ~foo constant. 2016-09-24 15:32:44 -07:00
Vidar Holen
868d53af95 Warn about passing globs to unset. 2016-09-24 14:49:52 -07:00
Vidar Holen
6a4b86cbea Fix warning for >& 2016-09-24 14:08:00 -07:00
Vidar Holen
fe2398edc9 Warn about >& in sh 2016-09-24 14:03:54 -07:00
Vidar Holen
3a7dc86de1 Don't warn about unused vars with readonly -f 2016-09-24 13:42:20 -07:00
koalaman
1c0ec9c6f6 Merge pull request #734 from NLKNguyen/add_CI_CD_solution_with_Docker
Add CI/CD solution with Docker
2016-09-11 23:42:06 -07:00
NLKNguyen
84110dbef4 Change DOCKER_REPO value and add test runner 2016-09-08 23:57:46 -07:00
NLKNguyen
1a7e98beaf Use cleaner escaping method 2016-09-08 21:58:51 -07:00
NLKNguyen
a5d5831a2e Fix syntax error with traditional if-clause instead of escaping bracket 2016-09-08 21:41:06 -07:00
NLKNguyen
47201822f9 Change syntax style for readability 2016-09-08 21:31:23 -07:00
NLKNguyen
32689ef5eb Test Dockerfiles and Travis CI on downstream repos 2016-09-08 21:05:54 -07:00
Vidar Holen
87481dce25 Warn about printf hello world and printf "%s %s" foo 2016-09-06 21:16:59 -07:00
Vidar Holen
a90b6d14b3 Count b as a reference in ${a:b} 2016-09-05 14:01:53 -07:00
Vidar Holen
5a46eeb09a Allow #inline comments without SC2046. 2016-09-05 12:38:35 -07:00
Vidar Holen
47a7065a7a Add style note for 'mycmd; if [ $? -eq 0 ]'. 2016-08-28 20:54:08 -07:00
koalaman
dbafbb3b3b Merge pull request #722 from kpankonen/docker
add Dockerfile that will build shellcheck
2016-08-07 11:43:37 -07:00
Vidar Holen
13a2070a32 Support multidimensional KSH arrays and warn in Bash. 2016-08-06 18:40:08 -07:00
Kevin Pankonen
fa4874c044 add Dockerfile that will build shellcheck 2016-08-05 16:42:21 -07:00
Vidar Holen
6a71ff6f46 Don't suggest removing $ in (( ${COLUMNS-80} )) 2016-07-30 10:42:33 -07:00
Vidar Holen
36263fb3f5 s/range/class/ when warning about tr '[abc]' 2016-07-05 08:51:40 -07:00
Vidar Holen
6dc419bbf5 Improve warning for 'else if'. 2016-07-02 15:40:29 -07:00
Vidar Holen
7af3470a91 Improve parser errors when reparsing array indices. 2016-07-01 22:06:50 -07:00
Vidar Holen
42f7479fb8 Don't warn about missing shebang when using directives. 2016-07-01 22:02:06 -07:00
Vidar Holen
50084c06c5 Don't warn when $(seq) is used unquoted. 2016-07-01 21:26:46 -07:00
Vidar Holen
e3bef9dc97 Warn about (( 1 -lt 2 )) 2016-07-01 20:33:07 -07:00
Vidar Holen
6c1abb2dee Performance: make readDollarExpr fail early if no $ 2016-06-30 10:01:03 -07:00
Vidar Holen
43c26061b9 Improve parsing for ambiguous $((foo) ) and ((foo) ). 2016-06-26 22:13:48 -07:00
Vidar Holen
07fd5724b8 Recognize declare -A statements when value is inlined. 2016-06-26 14:57:52 -07:00
Vidar Holen
eb2472ada8 Merge branch 'master' of github.com:koalaman/shellcheck 2016-06-26 14:40:43 -07:00
Vidar Holen
3e5ecaa262 Parse indices of associative arrays properly 2016-06-26 14:39:49 -07:00
koalaman
e1cec6c5d3 Merge pull request #694 from eatnumber1/end_column
Emit the end line in the JSON
2016-06-18 15:09:58 -07:00
Russell Harmon
eaa319ec57 Emit the end line in the JSON.
This handles the case where the end line is not on the same line as the
start line when using the new end column feature.
2016-06-18 15:06:28 -07:00
koalaman
717b5e91f5 Merge pull request #693 from eatnumber1/end_column
Make SC1035 emit an end column
2016-06-18 15:05:44 -07:00
Russell Harmon
7f5f5b7fb5 Make SC1035 emit a proper end column
Example JSON output:
```
$ shellcheck -s bash -f json /dev/stdin <<< "[[0 -eq 1 ]]"
[{"file":"/tmp/zshNCNwPz","line":1,"column":1,"endColumn":3,"level":"error","code":1035,"message":"You need a space after the [[ and before the ]]."}]
```
2016-06-18 14:59:47 -07:00
Russell Harmon
856d57f7d8 PositionedComment and ParseNote contains end cols.
This change makes PositionedComment and ParseNote contain end columns.
It additionally modifies the JSON formatter to show the end column in an
"endColumn" property. No modifications to the messages shown by any
other formatter have been made.

Currently, all checks set the end column to the start column. It should
now be possible however to start setting the end column in the parser.
Additional work is needed to set the end column during AST analysis.
2016-06-18 14:58:00 -07:00
koalaman
c45e9d4878 Merge pull request #677 from Maffblaster/patch-1
Add Gentoo to supported distribution list.
2016-05-26 12:11:02 -07:00
Matthew Marchese
89c6f6c800 Add Gentoo to supported distribution list.
It was missing from the list. I thought it was best to add it. :)
2016-05-26 09:56:36 -07:00
Vidar Holen
85e69f86eb In (( x = y )), logic to not reference x also grabbed y 2016-05-24 09:12:47 -07:00
Vidar Holen
47fd16b8e8 Stable version 0.4.4
This release is dedicated to AlphaGo.
The second golden age of AI is upon us!
2016-05-15 13:53:37 -07:00
24 changed files with 1762 additions and 631 deletions

27
.github/ISSUE_TEMPLATE.md vendored Normal file
View File

@@ -0,0 +1,27 @@
#### For bugs
- Rule Id (if any, e.g. SC1000):
- My shellcheck version (`shellcheck --version` or "online"):
- [ ] I tried on shellcheck.net and verified that this is still a problem on the latest commit
- [ ] It's not reproducible on shellcheck.net, but I think that's because it's an OS, configuration or encoding issue
#### For new checks and feature suggestions
- [ ] shellcheck.net (i.e. the latest commit) currently gives no useful warnings about this
- [ ] I searched through https://github.com/koalaman/shellcheck/issues and didn't find anything related
#### Here's a snippet or screenshot that shows the problem:
```sh
#!/your/interpreter
your script here
```
#### Here's what shellcheck currently says:
#### Here's what I wanted or expected to see:

21
.travis.yml Normal file
View File

@@ -0,0 +1,21 @@
sudo: required
language: sh
services:
- docker
before_install:
- export DOCKER_REPO=koalaman/shellcheck
- |-
export TAG=$([ "$TRAVIS_BRANCH" == "master" ] && echo "latest" || ([ -n "$TRAVIS_TAG" ] && echo "$TRAVIS_TAG") || echo "$TRAVIS_BRANCH")
script:
- docker build -t builder -f Dockerfile_builder .
- docker run --rm -it -v $(pwd):/mnt builder
- docker build -t $DOCKER_REPO:$TAG .
after_success:
- docker login -e="$DOCKER_EMAIL" -u="$DOCKER_USERNAME" -p="$DOCKER_PASSWORD"
- |-
([ "$TRAVIS_BRANCH" == "master" ] || [ -n "$TRAVIS_TAG" ]) && docker push "$DOCKER_REPO:$TAG"

11
Dockerfile Normal file
View File

@@ -0,0 +1,11 @@
FROM alpine:latest
MAINTAINER Nikyle Nguyen <NLKNguyen@MSN.com>
COPY package/bin/shellcheck /usr/local/bin/
COPY package/lib/ /usr/local/lib/
RUN ldconfig /usr/local/lib
WORKDIR /mnt
ENTRYPOINT ["shellcheck"]

54
Dockerfile_builder Normal file
View File

@@ -0,0 +1,54 @@
FROM mitchty/alpine-ghc:latest
MAINTAINER Nikyle Nguyen <NLKNguyen@MSN.com>
RUN apk add --no-cache build-base
RUN mkdir -p /usr/src/shellcheck
WORKDIR /usr/src/shellcheck
# # ------------------------------------------------------------
# # Build & Test
# # ------------------------------------------------------------
# Obtain the dependencies first, which are less likely to change, in order to reduce
# subsequent build time by leveraging image cache. This benefits developers when they
# build their code with this image locally. In case of Travis CI, this doesn't help
# reduce building time because Travis CI doesn't use cache.
COPY ShellCheck.cabal .
RUN cabal update && cabal install --only-dependencies
# Copy the rest of the source files, including ShellCheck.cabal again but doesn't matter
COPY . .
# Build
RUN cabal install
# Test
RUN cabal test
# # ------------------------------------------------------------
# # Set PATH
# # ------------------------------------------------------------
# Add runtime path to easily reach the executable file. This only exists during build.
ENV PATH "/root/.cabal/bin:$PATH"
# Make it permanent for someone who login to the container of this image
RUN echo "export PATH=${PATH}" >> /etc/profile
# # ------------------------------------------------------------
# # Extract Binaries
# # ------------------------------------------------------------
# Get shellcheck binary
RUN mkdir -p /package/bin/
RUN cp $(which shellcheck) /package/bin/
# Get shared libraries using magic
RUN mkdir -p /package/lib/
RUN ldd $(which shellcheck) | grep "=> /" | awk '{print $3}' | xargs -I '{}' cp -v '{}' /package/lib/
# Copy shellcheck package out to mounted directory
CMD ["cp", "-avr", "/package", "/mnt/"]

View File

@@ -36,9 +36,9 @@ Run `shellcheck yourscript` in your terminal for instant output, as seen above.
You can see ShellCheck suggestions directly in a variety of editors.
* Vim, through [Syntastic](https://github.com/scrooloose/syntastic):
* Vim, through [ALE](https://github.com/w0rp/ale) or [Syntastic](https://github.com/scrooloose/syntastic):
![Screenshot of vim showing inlined shellcheck feedback](doc/vim-syntastic.png).
![Screenshot of Vim showing inlined shellcheck feedback](doc/vim-syntastic.png).
* Emacs, through [Flycheck](https://github.com/flycheck/flycheck):
@@ -54,8 +54,21 @@ You can see ShellCheck suggestions directly in a variety of editors.
#### In your build or test suites
While ShellCheck is mostly intended for interactive use, it can easily be added to builds or test suites.
Use ShellCheck's exit code, or its [CheckStyle compatible XML output](shellcheck.1.md#user-content-formats). There's also a simple JSON output format for easy integration.
ShellCheck makes canonical use of exit codes, and can output simple JSON, CheckStyle compatible XML, GCC compatible warnings as well as human readable text (with or without ANSI colors). See the [Integration](https://github.com/koalaman/shellcheck/wiki/Integration) wiki page for more documentation.
## Travis CI Setup
If you want to use ShellCheck in Travis CI, setting it up is simple :tada:.
```yml
language: bash
addons:
apt:
sources:
- debian-sid # Grab ShellCheck from the Debian repo
packages:
- shellcheck
```
## Installing
@@ -64,12 +77,21 @@ The easiest way to install ShellCheck locally is through your package manager.
On systems with Cabal (installs to `~/.cabal/bin`):
cabal update
cabal install shellcheck
cabal install ShellCheck
On Debian based distros:
apt-get install shellcheck
On Gentoo based distros:
emerge --ask shellcheck
On EPEL based distros:
yum -y install epel-release
yum install ShellCheck
On Fedora based distros:
dnf install ShellCheck
@@ -95,6 +117,9 @@ add OBS devel:languages:haskell repository from https://build.opensuse.org/proje
or use OneClickInstall - https://software.opensuse.org/package/ShellCheck
From Docker Hub:
docker pull koalaman/shellcheck
## Compiling from source
@@ -103,7 +128,7 @@ This section describes how to build ShellCheck from a source directory. ShellChe
#### Installing Cabal
ShellCheck is built and packaged using Cabal. Install the package `cabal-install` from your system's package manager (with e.g. `apt-get`, `yum`, `zypper` or `brew`).
ShellCheck is built and packaged using Cabal. Install the package `cabal-install` from your system's package manager (with e.g. `apt-get`, `brew`, `emerge`, `yum`, or `zypper`).
On MacPorts, the package is instead called `hs-cabal-install`, while native Windows users should install the latest version of the Haskell platform from https://www.haskell.org/platform/
@@ -179,6 +204,8 @@ ShellCheck can recognize many types of incorrect test statements.
[ $1 -eq "shellcheck" ] # Numerical comparison of strings
[ $n && $m ] # && in [ .. ]
[ grep -q foo file ] # Command without $(..)
[[ "$$file" == *.jpg ]] # Comparisons that can't succeed
(( 1 -lt 2 )) # Using test operators in ((..))
#### Frequently misused commands
@@ -207,9 +234,11 @@ ShellCheck recognizes many common beginner's syntax errors:
var$n="Hello" # Wrong indirect assignment
echo ${var$n} # Wrong indirect reference
var=(1, 2, 3) # Comma separated arrays
array=( [index] = value ) # Incorrect index initialization
echo "Argument 10 is $10" # Positional parameter misreference
if $(myfunction); then ..; fi # Wrapping commands in $()
else if othercondition; then .. # Using 'else if'
#### Style
@@ -232,7 +261,8 @@ ShellCheck can recognize issues related to data and typing:
args="$@" # Assigning arrays to strings
files=(foo bar); echo "$files" # Referencing arrays as strings
printf "%s\n" "Arguments: $@." # Concatenating strings and arrays.
declare -A arr=(foo bar) # Associative arrays without index
printf "%s\n" "Arguments: $@." # Concatenating strings and arrays
[[ $# > 2 ]] # Comparing numbers as strings
var=World; echo "Hello " var # Unused lowercase variables
echo "Hello $name" # Unassigned lowercase variables
@@ -265,6 +295,7 @@ ShellCheck will warn when using features not supported by the shebang. For examp
foo-bar() { ..; } # Undefined/unsupported function name
[ $UID = 0 ] # Variable undefined in dash/sh
local var=value # local is undefined in sh
time sleep 1 | sleep 5 # Undefined uses of 'time'
#### Miscellaneous
@@ -275,11 +306,13 @@ ShellCheck recognizes a menagerie of other issues:
PATH="$PATH:~/bin" # Literal tilde in $PATH
rm “file” # Unicode quotes
echo "Hello world" # Carriage return / DOS line endings
echo hello \ # Trailing spaces after \
var=42 echo $var # Expansion of inlined environment
#!/bin/bash -x -e # Common shebang errors
echo $((n/180*100)) # Unnecessary loss of precision
ls *[:digit:].txt # Bad character class globs
sed 's/foo/bar/' file > file # Redirecting to input
sed 's/foo/bar/' file > file # Redirecting to input
## Testimonials
@@ -289,6 +322,11 @@ ShellCheck recognizes a menagerie of other issues:
Alexander Tarasikov,
[via Twitter](https://twitter.com/astarasikov/status/568825996532707330)
## Ignoring issues
Issues can be ignored via environmental variable, command line, individually or globally within a file:
https://github.com/koalaman/shellcheck/wiki/Ignore
## Reporting bugs

View File

@@ -1,5 +1,5 @@
Name: ShellCheck
Version: 0.4.4
Version: 0.4.6
Synopsis: Shell script analysis tool
License: GPL-3
License-file: LICENSE
@@ -55,6 +55,7 @@ library
ShellCheck.AnalyzerLib
ShellCheck.Checker
ShellCheck.Checks.Commands
ShellCheck.Checks.ShellSupport
ShellCheck.Data
ShellCheck.Formatter.Format
ShellCheck.Formatter.CheckStyle

View File

@@ -21,6 +21,7 @@ module ShellCheck.AST where
import Control.Monad
import Control.Monad.Identity
import Text.Parsec
import qualified ShellCheck.Regex as Re
data Id = Id Int deriving (Show, Eq, Ord)
@@ -32,6 +33,7 @@ data FunctionKeyword = FunctionKeyword Bool deriving (Show, Eq)
data FunctionParentheses = FunctionParentheses Bool deriving (Show, Eq)
data CaseType = CaseBreak | CaseFallThrough | CaseContinue deriving (Show, Eq)
data Root = Root Token
data Token =
TA_Binary Id String Token Token
| TA_Assignment Id String Token Token
@@ -43,15 +45,17 @@ data Token =
| TC_And Id ConditionType String Token Token
| TC_Binary Id ConditionType String Token Token
| TC_Group Id ConditionType Token
| TC_Noary Id ConditionType Token
| TC_Nullary Id ConditionType Token
| TC_Or Id ConditionType String Token Token
| TC_Unary Id ConditionType String Token
| T_AND_IF Id
| T_AndIf Id (Token) (Token)
| T_Arithmetic Id Token
| T_Array Id [Token]
| T_IndexedElement Id Token Token
| T_Assignment Id AssignmentMode String (Maybe Token) Token
| T_IndexedElement Id [Token] Token
-- Store the index as string, and parse as arithmetic or string later
| T_UnparsedIndex Id SourcePos String
| T_Assignment Id AssignmentMode String [Token] Token
| T_Backgrounded Id Token
| T_Backticked Id [Token]
| T_Bang Id
@@ -96,6 +100,7 @@ data Token =
| T_IfExpression Id [([Token],[Token])] [Token]
| T_In Id
| T_IoFile Id Token Token
| T_IoDuplicate Id Token String
| T_LESSAND Id
| T_LESSGREAT Id
| T_Lbrace Id
@@ -106,6 +111,7 @@ data Token =
| T_NormalWord Id [Token]
| T_OR_IF Id
| T_OrIf Id (Token) (Token)
| T_ParamSubSpecialChar Id String -- e.g. '%' in ${foo%bar} or '/' in ${foo/bar/baz}
| T_Pipeline Id [Token] [Token] -- [Pipe separators] [Commands]
| T_ProcSub Id String [Token]
| T_Rbrace Id
@@ -145,7 +151,7 @@ tokenEquals a b = kludge a == kludge b
instance Eq Token where
(==) = tokenEquals
analyze :: Monad m => (Token -> m ()) -> (Token -> m ()) -> (Token -> Token) -> Token -> m Token
analyze :: Monad m => (Token -> m ()) -> (Token -> m ()) -> (Token -> m Token) -> Token -> m Token
analyze f g i =
round
where
@@ -153,7 +159,7 @@ analyze f g i =
f t
newT <- delve t
g t
return . i $ newT
i newT
roundAll = mapM round
roundMaybe Nothing = return Nothing
@@ -186,14 +192,18 @@ analyze f g i =
delve (T_DollarArithmetic id c) = d1 c $ T_DollarArithmetic id
delve (T_DollarBracket id c) = d1 c $ T_DollarBracket id
delve (T_IoFile id op file) = d2 op file $ T_IoFile id
delve (T_IoDuplicate id op num) = d1 op $ \x -> T_IoDuplicate id x num
delve (T_HereString id word) = d1 word $ T_HereString id
delve (T_FdRedirect id v t) = d1 t $ T_FdRedirect id v
delve (T_Assignment id mode var index value) = do
a <- roundMaybe index
delve (T_Assignment id mode var indices value) = do
a <- roundAll indices
b <- round value
return $ T_Assignment id mode var a b
delve (T_Array id t) = dl t $ T_Array id
delve (T_IndexedElement id t1 t2) = d2 t1 t2 $ T_IndexedElement id
delve (T_IndexedElement id indices t) = do
a <- roundAll indices
b <- round t
return $ T_IndexedElement id a b
delve (T_Redirecting id redirs cmd) = do
a <- roundAll redirs
b <- round cmd
@@ -248,7 +258,7 @@ analyze f g i =
delve (TC_Group id typ token) = d1 token $ TC_Group id typ
delve (TC_Binary id typ op lhs rhs) = d2 lhs rhs $ TC_Binary id typ op
delve (TC_Unary id typ op token) = d1 token $ TC_Unary id typ op
delve (TC_Noary id typ token) = d1 token $ TC_Noary id typ
delve (TC_Nullary id typ token) = d1 token $ TC_Nullary id typ
delve (TA_Binary id op t1 t2) = d2 t1 t2 $ TA_Binary id op
delve (TA_Assignment id op t1 t2) = d2 t1 t2 $ TA_Assignment id op
@@ -310,8 +320,10 @@ getId t = case t of
T_DollarBraced id _ -> id
T_DollarArithmetic id _ -> id
T_BraceExpansion id _ -> id
T_ParamSubSpecialChar id _ -> id
T_DollarBraceCommandExpansion id _ -> id
T_IoFile id _ _ -> id
T_IoDuplicate id _ _ -> id
T_HereDoc id _ _ _ _ -> id
T_HereString id _ -> id
T_FdRedirect id _ _ -> id
@@ -344,7 +356,7 @@ getId t = case t of
TC_Group id _ _ -> id
TC_Binary id _ _ _ _ -> id
TC_Unary id _ _ _ -> id
TC_Noary id _ _ -> id
TC_Nullary id _ _ -> id
TA_Binary id _ _ _ -> id
TA_Assignment id _ _ _ -> id
TA_Unary id _ _ -> id
@@ -363,10 +375,11 @@ getId t = case t of
T_CoProc id _ _ -> id
T_CoProcBody id _ -> id
T_Include id _ _ -> id
T_UnparsedIndex id _ _ -> id
blank :: Monad m => Token -> m ()
blank = const $ return ()
doAnalysis f = analyze f blank id
doStackAnalysis startToken endToken = analyze startToken endToken id
doTransform i = runIdentity . analyze blank blank i
doAnalysis f = analyze f blank return
doStackAnalysis startToken endToken = analyze startToken endToken return
doTransform i = runIdentity . analyze blank blank (return . i)

View File

@@ -21,7 +21,9 @@ module ShellCheck.ASTLib where
import ShellCheck.AST
import Control.Monad.Writer
import Control.Monad
import Data.Functor
import Data.List
import Data.Maybe
@@ -54,6 +56,8 @@ isGlob _ = False
-- Is this shell word a constant?
isConstant token =
case token of
-- This ignores some cases like ~"foo":
T_NormalWord _ (T_Literal _ ('~':_) : _) -> False
T_NormalWord _ l -> all isConstant l
T_DoubleQuoted _ l -> all isConstant l
T_SingleQuoted _ _ -> True
@@ -83,13 +87,14 @@ oversimplify token =
(T_Glob _ s) -> [s]
(T_Pipeline _ _ [x]) -> oversimplify x
(T_Literal _ x) -> [x]
(T_ParamSubSpecialChar _ x) -> [x]
(T_SimpleCommand _ vars words) -> concatMap oversimplify words
(T_Redirecting _ _ foo) -> oversimplify foo
(T_DollarSingleQuoted _ s) -> [s]
(T_Annotation _ _ s) -> oversimplify s
-- Workaround for let "foo = bar" parsing
(TA_Sequence _ [TA_Expansion _ v]) -> concatMap oversimplify v
otherwise -> []
_ -> []
-- Turn a SimpleCommand foo -avz --bar=baz into args "a", "v", "z", "bar",
@@ -108,8 +113,11 @@ getFlagsUntil _ _ = error "Internal shellcheck error, please report! (getFlags o
-- Get all flags in a GNU way, up until --
getAllFlags = getFlagsUntil (== "--")
-- Get all flags in a BSD way, up until first non-flag argument
getLeadingFlags = getFlagsUntil (not . ("-" `isPrefixOf`))
-- Get all flags in a BSD way, up until first non-flag argument or --
getLeadingFlags = getFlagsUntil (\x -> x == "--" || (not $ "-" `isPrefixOf` x))
-- Check if a command has a flag.
hasFlag cmd str = str `elem` (map snd $ getAllFlags cmd)
-- Given a T_DollarBraced, return a simplified version of the string contents.
@@ -167,6 +175,20 @@ getUnquotedLiteral (T_NormalWord _ list) =
str _ = Nothing
getUnquotedLiteral _ = Nothing
-- Get the last unquoted T_Literal in a word like "${var}foo"THIS
-- or nothing if the word does not end in an unquoted literal.
getTrailingUnquotedLiteral :: Token -> Maybe Token
getTrailingUnquotedLiteral t =
case t of
(T_NormalWord _ list@(_:_)) ->
from (last list)
_ -> Nothing
where
from t =
case t of
(T_Literal {}) -> return t
_ -> Nothing
-- Maybe get the literal string of this token and any globs in it.
getGlobOrLiteralString = getLiteralStringExt f
where
@@ -185,6 +207,7 @@ getLiteralStringExt more = g
g (TA_Expansion _ l) = allInList l
g (T_SingleQuoted _ s) = return s
g (T_Literal _ s) = return s
g (T_ParamSubSpecialChar _ s) = return s
g x = more x
-- Is this token a string literal?
@@ -194,6 +217,8 @@ isLiteral t = isJust $ getLiteralString t
-- Turn a NormalWord like foo="bar $baz" into a series of constituent elements like [foo=,bar ,$baz]
getWordParts (T_NormalWord _ l) = concatMap getWordParts l
getWordParts (T_DoubleQuoted _ l) = l
-- TA_Expansion is basically T_NormalWord for arithmetic expressions
getWordParts (TA_Expansion _ l) = concatMap getWordParts l
getWordParts other = [other]
-- Return a list of NormalWords that would result from brace expansion
@@ -206,14 +231,39 @@ braceExpand (T_NormalWord id list) = take 1000 $ do
braceExpand item
part x = return x
-- Maybe get the command name of a token representing a command
getCommandName t =
-- Maybe get a SimpleCommand from immediate wrappers like T_Redirections
getCommand t =
case t of
T_Redirecting _ _ w -> getCommandName w
T_SimpleCommand _ _ (w:_) -> getLiteralString w
T_Annotation _ _ t -> getCommandName t
T_Redirecting _ _ w -> getCommand w
T_SimpleCommand _ _ (w:_) -> return t
T_Annotation _ _ t -> getCommand t
otherwise -> Nothing
-- Maybe get the command name of a token representing a command
getCommandName t = do
(T_SimpleCommand _ _ (w:rest)) <- getCommand t
s <- getLiteralString w
if "busybox" `isSuffixOf` s
then
case rest of
(applet:_) -> getLiteralString applet
_ -> return s
else
return s
-- If a command substitution is a single command, get its name.
-- $(date +%s) = Just "date"
getCommandNameFromExpansion :: Token -> Maybe String
getCommandNameFromExpansion t =
case t of
T_DollarExpansion _ [c] -> extract c
T_Backticked _ [c] -> extract c
T_DollarBraceCommandExpansion _ [c] -> extract c
otherwise -> Nothing
where
extract (T_Pipeline _ _ [cmd]) = getCommandName cmd
extract _ = Nothing
-- Get the basename of a token representing a command
getCommandBasename = liftM basename . getCommandName
where
@@ -237,8 +287,10 @@ isOnlyRedirection t =
isFunction t = case t of T_Function {} -> True; _ -> False
-- Get the list of commands from tokens that contain them, such as
-- the body of while loops and if statements.
isBraceExpansion t = case t of T_BraceExpansion {} -> True; _ -> False
-- Get the lists of commands from tokens that contain them, such as
-- the body of while loops or branches of if statements.
getCommandSequences t =
case t of
T_Script _ _ cmds -> [cmds]
@@ -251,3 +303,84 @@ getCommandSequences t =
T_IfExpression _ thens elses -> map snd thens ++ [elses]
otherwise -> []
-- Get a list of names of associative arrays
getAssociativeArrays t =
nub . execWriter $ doAnalysis f t
where
f :: Token -> Writer [String] ()
f t@(T_SimpleCommand {}) = fromMaybe (return ()) $ do
name <- getCommandName t
guard $ name == "declare" || name == "typeset"
let flags = getAllFlags t
guard $ elem "A" $ map snd flags
let args = map fst . filter ((==) "" . snd) $ flags
let names = mapMaybe (getLiteralStringExt nameAssignments) args
return $ tell names
f _ = return ()
nameAssignments t =
case t of
T_Assignment _ _ name _ _ -> return name
otherwise -> Nothing
-- A Pseudoglob is a wildcard pattern used for checking if a match can succeed.
-- For example, [[ $(cmd).jpg == [a-z] ]] will give the patterns *.jpg and ?, which
-- can be proven never to match.
data PseudoGlob = PGAny | PGMany | PGChar Char
deriving (Eq, Show)
-- Turn a word into a PG pattern, replacing all unknown/runtime values with
-- PGMany.
wordToPseudoGlob :: Token -> Maybe [PseudoGlob]
wordToPseudoGlob word =
simplifyPseudoGlob <$> concat <$> mapM f (getWordParts word)
where
f x = case x of
T_Literal _ s -> return $ map PGChar s
T_SingleQuoted _ s -> return $ map PGChar s
T_DollarBraced {} -> return [PGMany]
T_DollarExpansion {} -> return [PGMany]
T_Backticked {} -> return [PGMany]
T_Glob _ "?" -> return [PGAny]
T_Glob _ ('[':_) -> return [PGAny]
T_Glob {} -> return [PGMany]
T_Extglob {} -> return [PGMany]
_ -> return [PGMany]
-- Reorder a PseudoGlob for more efficient matching, e.g.
-- f?*?**g -> f??*g
simplifyPseudoGlob :: [PseudoGlob] -> [PseudoGlob]
simplifyPseudoGlob = f
where
f [] = []
f (x@(PGChar _) : rest ) = x : f rest
f list =
let (anys, rest) = span (\x -> x == PGMany || x == PGAny) list in
order anys ++ f rest
order s = let (any, many) = partition (== PGAny) s in
any ++ take 1 many
-- Check whether the two patterns can ever overlap.
pseudoGlobsCanOverlap :: [PseudoGlob] -> [PseudoGlob] -> Bool
pseudoGlobsCanOverlap = matchable
where
matchable x@(xf:xs) y@(yf:ys) =
case (xf, yf) of
(PGMany, _) -> matchable x ys || matchable xs y
(_, PGMany) -> matchable x ys || matchable xs y
(PGAny, _) -> matchable xs ys
(_, PGAny) -> matchable xs ys
(_, _) -> xf == yf && matchable xs ys
matchable [] [] = True
matchable (PGMany : rest) [] = matchable rest []
matchable (_:_) [] = False
matchable [] r = matchable r []
wordsCanBeEqual x y = fromMaybe True $
liftM2 pseudoGlobsCanOverlap (wordToPseudoGlob x) (wordToPseudoGlob y)

File diff suppressed because it is too large Load Diff

View File

@@ -23,7 +23,9 @@ import ShellCheck.Analytics
import ShellCheck.AnalyzerLib
import ShellCheck.Interface
import Data.List
import Data.Monoid
import qualified ShellCheck.Checks.Commands
import qualified ShellCheck.Checks.ShellSupport
-- TODO: Clean up the cruft this is layered on
@@ -32,5 +34,12 @@ analyzeScript spec = AnalysisResult {
arComments =
filterByAnnotation (asScript spec) . nub $
runAnalytics spec
++ ShellCheck.Checks.Commands.runChecks spec
++ runChecker params (checkers params)
}
where
params = makeParameters spec
checkers params = mconcat $ map ($ params) [
ShellCheck.Checks.Commands.checker,
ShellCheck.Checks.ShellSupport.checker
]

View File

@@ -29,7 +29,7 @@ import ShellCheck.Regex
import Control.Arrow (first)
import Control.Monad.Identity
import Control.Monad.Reader
import Control.Monad.RWS
import Control.Monad.State
import Control.Monad.Writer
import Data.Char
@@ -40,16 +40,48 @@ import qualified Data.Map as Map
import Test.QuickCheck.All (forAllProperties)
import Test.QuickCheck.Test (quickCheckWithResult, stdArgs, maxSuccess)
type Analysis = ReaderT Parameters (Writer [TokenComment]) ()
type Analysis = AnalyzerM ()
type AnalyzerM a = RWS Parameters [TokenComment] Cache a
nullCheck = const $ return ()
data Checker = Checker {
perScript :: Root -> Analysis,
perToken :: Token -> Analysis
}
runChecker :: Parameters -> Checker -> [TokenComment]
runChecker params checker = notes
where
root = rootNode params
check = perScript checker `composeAnalyzers` (\(Root x) -> void $ doAnalysis (perToken checker) x)
notes = snd $ evalRWS (check $ Root root) params Cache
instance Monoid Checker where
mempty = Checker {
perScript = nullCheck,
perToken = nullCheck
}
mappend x y = Checker {
perScript = perScript x `composeAnalyzers` perScript y,
perToken = perToken x `composeAnalyzers` perToken y
}
composeAnalyzers :: (a -> Analysis) -> (a -> Analysis) -> a -> Analysis
composeAnalyzers f g x = f x >> g x
data Parameters = Parameters {
variableFlow :: [StackData],
parentMap :: Map.Map Id Token,
shellType :: Shell,
shellTypeSpecified :: Bool
shellTypeSpecified :: Bool,
rootNode :: Token
}
-- TODO: Cache results of common AST ops here
data Cache = Cache {}
data Scope = SubshellScope String | NoneScope deriving (Show, Eq)
data StackData =
StackScope Scope
@@ -81,6 +113,14 @@ pScript s =
}
in prRoot . runIdentity $ parseScript (mockedSystemInterface []) pSpec
-- For testing. If parsed, returns whether there are any comments
producesComments :: Checker -> String -> Maybe Bool
producesComments c s = do
root <- pScript s
let spec = defaultSpec root
let params = makeParameters spec
return . not . null $ runChecker params c
makeComment :: Severity -> Id -> Code -> String -> TokenComment
makeComment severity id code note =
TokenComment id $ Comment severity code note
@@ -95,6 +135,7 @@ style id code str = addComment $ makeComment StyleC id code str
makeParameters spec =
let params = Parameters {
rootNode = root,
shellType = fromMaybe (determineShell root) $ asShellType spec,
shellTypeSpecified = isJust $ asShellType spec,
parentMap = getParentTree root,
@@ -112,6 +153,7 @@ prop_determineShell4 = determineShell (fromJust $ pScript
prop_determineShell5 = determineShell (fromJust $ pScript
"#shellcheck shell=sh\nfoo") == Sh
prop_determineShell6 = determineShell (fromJust $ pScript "#! /bin/sh") == Sh
prop_determineShell7 = determineShell (fromJust $ pScript "#! /bin/ash") == Dash
determineShell t = fromMaybe Bash $ do
shellString <- foldl mplus Nothing $ getCandidates t
shellForExecutable shellString
@@ -125,8 +167,13 @@ determineShell t = fromMaybe Bash $ do
getCandidates (T_Annotation _ annotations s) =
map forAnnotation annotations ++
[Just $ fromShebang s]
fromShebang (T_Script _ s t) = shellFor s
fromShebang (T_Script _ s t) = executableFromShebang s
-- Given a string like "/bin/bash" or "/usr/bin/env dash",
-- return the shell basename like "bash" or "dash"
executableFromShebang :: String -> String
executableFromShebang = shellFor
where
shellFor s | "/env " `isInfixOf` s = head (drop 1 (words s)++[""])
shellFor s | ' ' `elem` s = shellFor $ takeWhile (/= ' ') s
shellFor s = reverse . takeWhile (/= '/') . reverse $ s
@@ -170,16 +217,13 @@ isQuoteFreeNode strict tree t =
-- Are any subnodes inherently self-quoting?
isQuoteFreeContext t =
case t of
TC_Noary _ DoubleBracket _ -> return True
TC_Nullary _ DoubleBracket _ -> return True
TC_Unary _ DoubleBracket _ _ -> return True
TC_Binary _ DoubleBracket _ _ _ -> return True
TA_Sequence {} -> return True
T_Arithmetic {} -> return True
T_Assignment {} -> return True
T_Redirecting {} -> return $
if strict then False else
-- Not true, just a hack to prevent warning about non-expansion refs
any (isCommand t) ["local", "declare", "typeset", "export", "trap", "readonly"]
T_Redirecting {} -> return False
T_DoubleQuoted _ _ -> return True
T_DollarDoubleQuoted _ _ -> return True
T_CaseExpression {} -> return True
@@ -211,6 +255,10 @@ getClosestCommand tree t =
getCommand t@(T_Redirecting {}) = return t
getCommand _ = Nothing
getClosestCommandM t = do
tree <- asks parentMap
return $ getClosestCommand tree t
usedAsCommandName tree token = go (getId token) (tail $ getPath tree token)
where
go currentId (T_NormalWord id [word]:rest)
@@ -227,6 +275,12 @@ getPath tree t = t :
Nothing -> []
Just parent -> getPath tree parent
-- Version of the above taking the map from the current context
-- Todo: give this the name "getPath"
getPathM t = do
map <- asks parentMap
return $ getPath map t
isParentOf tree parent child =
elem (getId parent) . map getId $ getPath tree child
@@ -334,6 +388,12 @@ getModifiedVariables t =
name <- getLiteralString lhs
return (t, t, name, DataString $ SourceFrom [rhs])
T_DollarBraced _ l -> maybeToList $ do
let string = bracedString t
let modifier = getBracedModifier string
guard $ ":=" `isPrefixOf` modifier
return (t, t, getBracedReference string, DataString $ SourceFrom [l])
t@(T_FdRedirect _ ('{':var) op) -> -- {foo}>&2 modifies foo
[(t, t, takeWhile (/= '}') var, DataString SourceInteger) | not $ isClosingFileOp op]
@@ -341,6 +401,7 @@ getModifiedVariables t =
[(t, t, fromMaybe "COPROC" name, DataArray SourceInteger)]
--Points to 'for' rather than variable
T_ForIn id str [] _ -> [(t, t, str, DataString $ SourceExternal)]
T_ForIn id str words _ -> [(t, t, str, DataString $ SourceFrom words)]
T_SelectIn id str words _ -> [(t, t, str, DataString $ SourceFrom words)]
_ -> []
@@ -361,7 +422,10 @@ getReferencedVariableCommand base@(T_SimpleCommand _ _ (T_NormalWord _ (T_Litera
"declare" -> if any (`elem` flags) ["x", "p"]
then concatMap getReference rest
else []
"readonly" -> concatMap getReference rest
"readonly" ->
if any (`elem` flags) ["f", "p"]
then []
else concatMap getReference rest
"trap" ->
case rest of
head:_ -> map (\x -> (head, head, x)) $ getVariablesFromLiteralToken head
@@ -395,7 +459,10 @@ getModifiedVariableCommand base@(T_SimpleCommand _ _ (T_NormalWord _ (T_Literal
"typeset" -> declaredVars
"local" -> concatMap getModifierParamString rest
"readonly" -> concatMap getModifierParamString rest
"readonly" ->
if any (`elem` flags) ["f", "p"]
then []
else concatMap getModifierParamString rest
"set" -> maybeToList $ do
params <- getSetParams rest
return (base, base, "@", DataString $ SourceFrom params)
@@ -474,11 +541,20 @@ getIndexReferences s = fromMaybe [] $ do
where
re = mkRegex "(\\[.*\\])"
getOffsetReferences mods = fromMaybe [] $ do
match <- matchRegex re mods
offsets <- match !!! 0
return $ matchAllStrings variableNameRegex offsets
where
re = mkRegex "^ *:(.*)"
getReferencedVariables parents t =
case t of
T_DollarBraced id l -> let str = bracedString t in
(t, t, getBracedReference str) :
map (\x -> (l, l, x)) (getIndexReferences str)
map (\x -> (l, l, x)) (
getIndexReferences str
++ getOffsetReferences (getBracedModifier str))
TA_Expansion id _ ->
if isArithmeticAssignment t
then []
@@ -520,7 +596,7 @@ getReferencedVariables parents t =
isDereferencing = (`elem` ["-eq", "-ne", "-lt", "-le", "-gt", "-ge"])
isArithmeticAssignment t = case getPath parents t of
this: TA_Assignment _ "=" _ _ :_ -> True
this: TA_Assignment _ "=" lhs _ :_ -> lhs == t
_ -> False
dataTypeFrom defaultType v = (case v of T_Array {} -> DataArray; _ -> defaultType) $ SourceFrom [v]
@@ -573,6 +649,7 @@ prop_getBracedReference9 = getBracedReference "foo:-bar" == "foo"
prop_getBracedReference10= getBracedReference "foo: -1" == "foo"
prop_getBracedReference11= getBracedReference "!os*" == ""
prop_getBracedReference12= getBracedReference "!os?bar**" == ""
prop_getBracedReference13= getBracedReference "foo[bar]" == "foo"
getBracedReference s = fromMaybe s $
nameExpansion s `mplus` takeName noPrefix `mplus` getSpecial noPrefix `mplus` getSpecial s
where
@@ -595,6 +672,20 @@ getBracedReference s = fromMaybe s $
return ""
nameExpansion _ = Nothing
prop_getBracedModifier1 = getBracedModifier "foo:bar:baz" == ":bar:baz"
prop_getBracedModifier2 = getBracedModifier "!var:-foo" == ":-foo"
prop_getBracedModifier3 = getBracedModifier "foo[bar]" == "[bar]"
getBracedModifier s = fromMaybe "" . listToMaybe $ do
let var = getBracedReference s
a <- dropModifier s
dropPrefix var a
where
dropPrefix [] t = return t
dropPrefix (a:b) (c:d) | a == c = dropPrefix b d
dropPrefix _ _ = []
dropModifier (c:rest) | c `elem` "#!" = [rest, c:rest]
dropModifier x = [x]
-- Useful generic functions
potentially :: Monad m => Maybe (m ()) -> m ()
@@ -608,6 +699,10 @@ headOrDefault def _ = def
[] -> Nothing
(r:_) -> Just r
-- Run a command if the shell is in the given list
whenShell l c = do
shell <- asks shellType
when (shell `elem` l ) c
filterByAnnotation token =
@@ -628,5 +723,5 @@ filterByAnnotation token =
getCode (TokenComment _ (Comment _ c _)) = c
return []
return []
runTests = $( [| $(forAllProperties) (quickCheckWithResult (stdArgs { maxSuccess = 1 }) ) |])

View File

@@ -39,7 +39,7 @@ import Test.QuickCheck.All
tokenToPosition map (TokenComment id c) = fromMaybe fail $ do
position <- Map.lookup id map
return $ PositionedComment position c
return $ PositionedComment position position c
where
fail = error "Internal shellcheck error: id doesn't exist. Please report!"
@@ -65,13 +65,13 @@ checkScript sys spec = do
return . nub . sortMessages . filter shouldInclude $
(parseMessages ++ map translator analysisMessages)
shouldInclude (PositionedComment _ (Comment _ code _)) =
shouldInclude (PositionedComment _ _ (Comment _ code _)) =
code `notElem` csExcludedWarnings spec
sortMessages = sortBy (comparing order)
order (PositionedComment pos (Comment severity code message)) =
order (PositionedComment pos _ (Comment severity code message)) =
(posFile pos, posLine pos, posColumn pos, severity, code, message)
getPosition (PositionedComment pos _) = pos
getPosition (PositionedComment pos _ _) = pos
analysisSpec root =
AnalysisSpec {
@@ -84,7 +84,7 @@ getErrors sys spec =
sort . map getCode . crComments $
runIdentity (checkScript sys spec)
where
getCode (PositionedComment _ (Comment _ code _)) = code
getCode (PositionedComment _ _ (Comment _ code _)) = code
check = checkWithIncludes []
@@ -158,5 +158,25 @@ prop_sourceDirectiveDoesntFollowFile =
[("foo", "source bar"), ("bar", "baz=3")]
"#shellcheck source=foo\n. \"$1\"; echo \"$baz\""
prop_filewideAnnotationBase = [2086] == check "#!/bin/sh\necho $1"
prop_filewideAnnotation1 = null $
check "#!/bin/sh\n# shellcheck disable=2086\necho $1"
prop_filewideAnnotation2 = null $
check "#!/bin/sh\n# shellcheck disable=2086\ntrue\necho $1"
prop_filewideAnnotation3 = null $
check "#!/bin/sh\n#unerlated\n# shellcheck disable=2086\ntrue\necho $1"
prop_filewideAnnotation4 = null $
check "#!/bin/sh\n# shellcheck disable=2086\n#unrelated\ntrue\necho $1"
prop_filewideAnnotation5 = null $
check "#!/bin/sh\n\n\n\n#shellcheck disable=2086\ntrue\necho $1"
prop_filewideAnnotation6 = null $
check "#shellcheck shell=sh\n#unrelated\n#shellcheck disable=2086\ntrue\necho $1"
prop_filewideAnnotation7 = null $
check "#!/bin/sh\n# shellcheck disable=2086\n#unrelated\ntrue\necho $1"
prop_filewideAnnotationBase2 = [2086, 2181] == check "true\n[ $? == 0 ] && echo $1"
prop_filewideAnnotation8 = null $
check "# Disable $? warning\n#shellcheck disable=SC2181\n# Disable quoting warning\n#shellcheck disable=2086\ntrue\n[ $? == 0 ] && echo $1"
return []
runTests = $quickCheckAll

View File

@@ -21,7 +21,7 @@
{-# LANGUAGE FlexibleContexts #-}
-- This module contains checks that examine specific commands by name.
module ShellCheck.Checks.Commands (runChecks
module ShellCheck.Checks.Commands (checker
, ShellCheck.Checks.Commands.runTests
) where
@@ -34,8 +34,7 @@ import ShellCheck.Parser
import ShellCheck.Regex
import Control.Monad
import Control.Monad.Reader
import Control.Monad.Writer
import Control.Monad.RWS
import Data.Char
import Data.List
import Data.Maybe
@@ -49,22 +48,10 @@ data CommandName = Exactly String | Basename String
data CommandCheck =
CommandCheck CommandName (Token -> Analysis)
nullCheck :: Token -> Analysis
nullCheck _ = return ()
verify :: CommandCheck -> String -> Bool
verify f s = producesComments f s == Just True
verifyNot f s = producesComments f s == Just False
producesComments :: CommandCheck -> String -> Maybe Bool
producesComments f s = do
root <- pScript s
return . not . null $ runList (defaultSpec root) [f]
composeChecks f g t = do
f t
g t
verify f s = producesComments (getChecker [f]) s == Just True
verifyNot f s = producesComments (getChecker [f]) s == Just False
arguments (T_SimpleCommand _ _ (cmd:args)) = args
@@ -90,13 +77,21 @@ commandChecks = [
,checkExportedExpansions
,checkAliasesUsesArgs
,checkAliasesExpandEarly
,checkUnsetGlobs
,checkFindWithoutPath
,checkTimeParameters
,checkTimedCommand
,checkLocalScope
,checkDeprecatedTempfile
,checkDeprecatedEgrep
,checkDeprecatedFgrep
]
buildCommandMap :: [CommandCheck] -> Map.Map CommandName (Token -> Analysis)
buildCommandMap = foldl' addCheck Map.empty
where
addCheck map (CommandCheck name function) =
Map.insertWith' composeChecks name function map
Map.insertWith' composeAnalyzers name function map
checkCommand :: Map.Map CommandName (Token -> Analysis) -> Token -> Analysis
@@ -114,15 +109,17 @@ checkCommand map t@(T_SimpleCommand id _ (cmd:rest)) = fromMaybe (return ()) $ d
basename = reverse . takeWhile (/= '/') . reverse
checkCommand _ _ = return ()
runList spec list = notes
where
root = asScript spec
params = makeParameters spec
notes = execWriter $ runReaderT (doAnalysis (checkCommand map) root) params
map = buildCommandMap list
getChecker :: [CommandCheck] -> Checker
getChecker list = Checker {
perScript = const $ return (),
perToken = checkCommand map
}
where
map = buildCommandMap list
runChecks spec = runList spec commandChecks
checker :: Parameters -> Checker
checker params = getChecker commandChecks
prop_checkTr1 = verify checkTr "tr [a-f] [A-F]"
prop_checkTr2 = verify checkTr "tr 'a-z' 'A-Z'"
@@ -150,7 +147,7 @@ checkTr = CommandCheck (Basename "tr") (mapM_ f . arguments)
info (getId word) 2020 "tr replaces sets of chars, not words (mentioned due to duplicates)."
unless ("[:" `isPrefixOf` s) $
when ("[" `isPrefixOf` s && "]" `isSuffixOf` s && (length s > 2) && ('*' `notElem` s)) $
info (getId word) 2021 "Don't use [] around ranges in tr, it replaces literal square brackets."
info (getId word) 2021 "Don't use [] around classes in tr, it replaces literal square brackets."
Nothing -> return ()
duplicated s =
@@ -197,23 +194,27 @@ prop_checkGrepRe8 = verify checkGrepRe "ls | grep foo*.jpg"
prop_checkGrepRe9 = verifyNot checkGrepRe "grep '[0-9]*' file"
prop_checkGrepRe10= verifyNot checkGrepRe "grep '^aa*' file"
prop_checkGrepRe11= verifyNot checkGrepRe "grep --include=*.png foo"
prop_checkGrepRe12= verifyNot checkGrepRe "grep -F 'Foo*' file"
checkGrepRe = CommandCheck (Basename "grep") (f . arguments) where
checkGrepRe = CommandCheck (Basename "grep") check where
check cmd = f cmd (arguments cmd)
-- --regex=*(extglob) doesn't work. Fixme?
skippable (Just s) = not ("--regex=" `isPrefixOf` s) && "-" `isPrefixOf` s
skippable _ = False
f [] = return ()
f (x:r) | skippable (getLiteralStringExt (const $ return "_") x) = f r
f (re:_) = do
f _ [] = return ()
f cmd (x:r) | skippable (getLiteralStringExt (const $ return "_") x) = f cmd r
f cmd (re:_) = do
when (isGlob re) $
warn (getId re) 2062 "Quote the grep pattern so the shell won't interpret it."
let string = concat $ oversimplify re
if isConfusedGlobRegex string then
warn (getId re) 2063 "Grep uses regex, but this looks like a glob."
else potentially $ do
char <- getSuspiciousRegexWildcard string
return $ info (getId re) 2022 $
"Note that unlike globs, " ++ [char] ++ "* here matches '" ++ [char, char, char] ++ "' but not '" ++ wordStartingWith char ++ "'."
unless (cmd `hasFlag` "F") $ do
let string = concat $ oversimplify re
if isConfusedGlobRegex string then
warn (getId re) 2063 "Grep uses regex, but this looks like a glob."
else potentially $ do
char <- getSuspiciousRegexWildcard string
return $ info (getId re) 2022 $
"Note that unlike globs, " ++ [char] ++ "* here matches '" ++ [char, char, char] ++ "' but not '" ++ wordStartingWith char ++ "'."
wordStartingWith c =
head . filter ([c] `isPrefixOf`) $ candidates
@@ -342,7 +343,7 @@ checkInjectableFindSh = CommandCheck (Basename "find") (check . arguments)
pattern = [
(`elem` ["-exec", "-execdir"]),
(`elem` ["sh", "bash", "ksh"]),
(`elem` ["sh", "bash", "dash", "ksh"]),
(== "-c")
]
action (id, arg) =
@@ -362,7 +363,7 @@ checkFindActionPrecedence = CommandCheck (Basename "find") (f . arguments)
then warnFor (list !! (length pattern - 1))
else f rest
isMatch = isParam [ "-name", "-regex", "-iname", "-iregex", "-wholename", "-iwholename" ]
isAction = isParam [ "-exec", "-execdir", "-delete", "-print", "-print0" ]
isAction = isParam [ "-exec", "-execdir", "-delete", "-print", "-print0", "-fls", "-fprint", "-fprint0", "-fprintf", "-ls", "-ok", "-okdir", "-printf" ]
isParam strs t = fromMaybe False $ do
param <- getLiteralString t
return $ param `elem` strs
@@ -470,16 +471,51 @@ prop_checkPrintfVar1 = verify checkPrintfVar "printf \"Lol: $s\""
prop_checkPrintfVar2 = verifyNot checkPrintfVar "printf 'Lol: $s'"
prop_checkPrintfVar3 = verify checkPrintfVar "printf -v cow $(cmd)"
prop_checkPrintfVar4 = verifyNot checkPrintfVar "printf \"%${count}s\" var"
prop_checkPrintfVar5 = verify checkPrintfVar "printf '%s %s %s' foo bar"
prop_checkPrintfVar6 = verify checkPrintfVar "printf foo bar baz"
prop_checkPrintfVar7 = verify checkPrintfVar "printf -- foo bar baz"
prop_checkPrintfVar8 = verifyNot checkPrintfVar "printf '%s %s %s' \"${var[@]}\""
prop_checkPrintfVar9 = verifyNot checkPrintfVar "printf '%s %s %s\\n' *.png"
prop_checkPrintfVar10= verifyNot checkPrintfVar "printf '%s %s %s' foo bar baz"
prop_checkPrintfVar11= verifyNot checkPrintfVar "printf '%(%s%s)T' -1"
checkPrintfVar = CommandCheck (Exactly "printf") (f . arguments) where
f (doubledash:rest) | getLiteralString doubledash == Just "--" = f rest
f (dashv:var:rest) | getLiteralString dashv == Just "-v" = f rest
f (format:params) = check format
f (format:params) = check format params
f _ = return ()
check format =
countFormats string =
case string of
'%':'%':rest -> countFormats rest
'%':'(':rest -> 1 + countFormats (dropWhile (/= ')') rest)
'%':rest -> 1 + countFormats rest
_:rest -> countFormats rest
[] -> 0
check format more = do
fromMaybe (return ()) $ do
string <- getLiteralString format
let vars = countFormats string
return $ do
when (vars == 0 && more /= []) $
err (getId format) 2182
"This printf format string has no variables. Other arguments are ignored."
when (vars > 0
&& length more < vars
&& all (not . mayBecomeMultipleArgs) more) $
warn (getId format) 2183 $
"This format string has " ++ show vars ++ " variables, but is passed " ++ show (length more) ++ " arguments."
unless ('%' `elem` concat (oversimplify format) || isLiteral format) $
warn (getId format) 2059
info (getId format) 2059
"Don't use variables in the printf format string. Use printf \"..%s..\" \"$foo\"."
prop_checkUuoeCmd1 = verify checkUuoeCmd "echo $(date)"
prop_checkUuoeCmd2 = verify checkUuoeCmd "echo `date`"
prop_checkUuoeCmd3 = verify checkUuoeCmd "echo \"$(date)\""
@@ -556,5 +592,94 @@ checkAliasesExpandEarly = CommandCheck (Exactly "alias") (f . arguments)
checkArg _ = return ()
prop_checkUnsetGlobs1 = verify checkUnsetGlobs "unset foo[1]"
prop_checkUnsetGlobs2 = verifyNot checkUnsetGlobs "unset foo"
checkUnsetGlobs = CommandCheck (Exactly "unset") (mapM_ check . arguments)
where
check arg =
when (isGlob arg) $
warn (getId arg) 2184 "Quote arguments to unset so they're not glob expanded."
prop_checkFindWithoutPath1 = verify checkFindWithoutPath "find -type f"
prop_checkFindWithoutPath2 = verify checkFindWithoutPath "find"
prop_checkFindWithoutPath3 = verifyNot checkFindWithoutPath "find . -type f"
prop_checkFindWithoutPath4 = verifyNot checkFindWithoutPath "find -H -L \"$path\" -print"
checkFindWithoutPath = CommandCheck (Basename "find") f
where
f (T_SimpleCommand _ _ (cmd:args)) =
unless (hasPath args) $
info (getId cmd) 2185 "Some finds don't have a default path. Specify '.' explicitly."
-- This is a bit of a kludge. find supports flag arguments both before and after the path,
-- as well as multiple non-flag arguments that are not the path. We assume that all the
-- pre-path flags are single characters, which is generally the case.
hasPath (first:rest) =
let flag = fromJust $ getLiteralStringExt (const $ return "___") first in
not ("-" `isPrefixOf` flag) || length flag <= 2 && hasPath rest
hasPath [] = False
prop_checkTimeParameters1 = verify checkTimeParameters "time -f lol sleep 10"
prop_checkTimeParameters2 = verifyNot checkTimeParameters "time sleep 10"
prop_checkTimeParameters3 = verifyNot checkTimeParameters "time -p foo"
prop_checkTimeParameters4 = verifyNot checkTimeParameters "command time -f lol sleep 10"
checkTimeParameters = CommandCheck (Exactly "time") f
where
f (T_SimpleCommand _ _ (cmd:args:_)) =
whenShell [Bash, Sh] $
let s = concat $ oversimplify args in
when ("-" `isPrefixOf` s && s /= "-p") $
info (getId cmd) 2023 "The shell may override 'time' as seen in man time(1). Use 'command time ..' for that one."
f _ = return ()
prop_checkTimedCommand1 = verify checkTimedCommand "#!/bin/sh\ntime -p foo | bar"
prop_checkTimedCommand2 = verify checkTimedCommand "#!/bin/dash\ntime ( foo; bar; )"
prop_checkTimedCommand3 = verifyNot checkTimedCommand "#!/bin/sh\ntime sleep 1"
checkTimedCommand = CommandCheck (Exactly "time") f where
f (T_SimpleCommand _ _ (c:args@(_:_))) =
whenShell [Sh, Dash] $ do
let cmd = last args -- "time" is parsed with a command as argument
when (isPiped cmd) $
warn (getId c) 2176 "'time' is undefined for pipelines. time single stage or bash -c instead."
when (isSimple cmd == Just False) $
warn (getId cmd) 2177 "'time' is undefined for compound commands, time sh -c instead."
f _ = return ()
isPiped cmd =
case cmd of
T_Pipeline _ _ (_:_:_) -> True
_ -> False
getCommand cmd =
case cmd of
T_Pipeline _ _ (T_Redirecting _ _ a : _) -> return a
_ -> fail ""
isSimple cmd = do
innerCommand <- getCommand cmd
case innerCommand of
T_SimpleCommand {} -> return True
_ -> return False
prop_checkLocalScope1 = verify checkLocalScope "local foo=3"
prop_checkLocalScope2 = verifyNot checkLocalScope "f() { local foo=3; }"
checkLocalScope = CommandCheck (Exactly "local") $ \t ->
whenShell [Bash, Dash] $ do -- Ksh allows it, Sh doesn't support local
path <- getPathM t
unless (any isFunction path) $
err (getId t) 2168 "'local' is only valid in functions."
prop_checkDeprecatedTempfile1 = verify checkDeprecatedTempfile "var=$(tempfile)"
prop_checkDeprecatedTempfile2 = verifyNot checkDeprecatedTempfile "tempfile=$(mktemp)"
checkDeprecatedTempfile = CommandCheck (Basename "tempfile") $
\t -> warn (getId t) 2186 "tempfile is deprecated. Use mktemp instead."
prop_checkDeprecatedEgrep = verify checkDeprecatedEgrep "egrep '.+'"
checkDeprecatedEgrep = CommandCheck (Basename "egrep") $
\t -> info (getId t) 2196 "egrep is non-standard and deprecated. Use grep -E instead."
prop_checkDeprecatedFgrep = verify checkDeprecatedFgrep "fgrep '*' files"
checkDeprecatedFgrep = CommandCheck (Basename "fgrep") $
\t -> info (getId t) 2197 "fgrep is non-standard and deprecated. Use grep -F instead."
return []
runTests = $( [| $(forAllProperties) (quickCheckWithResult (stdArgs { maxSuccess = 1 }) ) |])

View File

@@ -0,0 +1,388 @@
{-
Copyright 2012-2016 Vidar Holen
This file is part of ShellCheck.
http://www.vidarholen.net/contents/shellcheck
ShellCheck is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
ShellCheck is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
-}
{-# LANGUAGE TemplateHaskell #-}
{-# LANGUAGE FlexibleContexts #-}
module ShellCheck.Checks.ShellSupport (checker
, ShellCheck.Checks.ShellSupport.runTests
) where
import ShellCheck.AST
import ShellCheck.ASTLib
import ShellCheck.AnalyzerLib
import ShellCheck.Interface
import ShellCheck.Regex
import Control.Monad
import Control.Monad.RWS
import Data.Char
import Data.List
import Data.Maybe
import qualified Data.Map as Map
import Test.QuickCheck.All (forAllProperties)
import Test.QuickCheck.Test (quickCheckWithResult, stdArgs, maxSuccess)
data ForShell = ForShell [Shell] (Token -> Analysis)
getChecker params list = Checker {
perScript = nullCheck,
perToken = foldl composeAnalyzers nullCheck $ mapMaybe include list
}
where
shell = shellType params
include (ForShell list a) = do
guard $ shell `elem` list
return a
checker params = getChecker params checks
checks = [
checkForDecimals
,checkBashisms
,checkEchoSed
,checkBraceExpansionVars
,checkMultiDimensionalArrays
]
testChecker (ForShell _ t) =
Checker {
perScript = nullCheck,
perToken = t
}
verify c s = producesComments (testChecker c) s == Just True
verifyNot c s = producesComments (testChecker c) s == Just False
prop_checkForDecimals1 = verify checkForDecimals "((3.14*c))"
prop_checkForDecimals2 = verify checkForDecimals "foo[1.2]=bar"
prop_checkForDecimals3 = verifyNot checkForDecimals "declare -A foo; foo[1.2]=bar"
checkForDecimals = ForShell [Sh, Dash, Bash] f
where
f t@(TA_Expansion id _) = potentially $ do
str <- getLiteralString t
first <- str !!! 0
guard $ isDigit first && '.' `elem` str
return $ err id 2079 "(( )) doesn't support decimals. Use bc or awk."
f _ = return ()
prop_checkBashisms = verify checkBashisms "while read a; do :; done < <(a)"
prop_checkBashisms2 = verify checkBashisms "[ foo -nt bar ]"
prop_checkBashisms3 = verify checkBashisms "echo $((i++))"
prop_checkBashisms4 = verify checkBashisms "rm !(*.hs)"
prop_checkBashisms5 = verify checkBashisms "source file"
prop_checkBashisms6 = verify checkBashisms "[ \"$a\" == 42 ]"
prop_checkBashisms7 = verify checkBashisms "echo ${var[1]}"
prop_checkBashisms8 = verify checkBashisms "echo ${!var[@]}"
prop_checkBashisms9 = verify checkBashisms "echo ${!var*}"
prop_checkBashisms10= verify checkBashisms "echo ${var:4:12}"
prop_checkBashisms11= verifyNot checkBashisms "echo ${var:-4}"
prop_checkBashisms12= verify checkBashisms "echo ${var//foo/bar}"
prop_checkBashisms13= verify checkBashisms "exec -c env"
prop_checkBashisms14= verify checkBashisms "echo -n \"Foo: \""
prop_checkBashisms15= verify checkBashisms "let n++"
prop_checkBashisms16= verify checkBashisms "echo $RANDOM"
prop_checkBashisms17= verify checkBashisms "echo $((RANDOM%6+1))"
prop_checkBashisms18= verify checkBashisms "foo &> /dev/null"
prop_checkBashisms19= verify checkBashisms "foo > file*.txt"
prop_checkBashisms20= verify checkBashisms "read -ra foo"
prop_checkBashisms21= verify checkBashisms "[ -a foo ]"
prop_checkBashisms22= verifyNot checkBashisms "[ foo -a bar ]"
prop_checkBashisms23= verify checkBashisms "trap mything ERR INT"
prop_checkBashisms24= verifyNot checkBashisms "trap mything INT TERM"
prop_checkBashisms25= verify checkBashisms "cat < /dev/tcp/host/123"
prop_checkBashisms26= verify checkBashisms "trap mything ERR SIGTERM"
prop_checkBashisms27= verify checkBashisms "echo *[^0-9]*"
prop_checkBashisms28= verify checkBashisms "exec {n}>&2"
prop_checkBashisms29= verify checkBashisms "echo ${!var}"
prop_checkBashisms30= verify checkBashisms "printf -v '%s' \"$1\""
prop_checkBashisms31= verify checkBashisms "printf '%q' \"$1\""
prop_checkBashisms32= verifyNot checkBashisms "#!/bin/dash\n[ foo -nt bar ]"
prop_checkBashisms33= verify checkBashisms "#!/bin/sh\necho -n foo"
prop_checkBashisms34= verifyNot checkBashisms "#!/bin/dash\necho -n foo"
prop_checkBashisms35= verifyNot checkBashisms "#!/bin/dash\nlocal foo"
prop_checkBashisms36= verifyNot checkBashisms "#!/bin/dash\nread -p foo -r bar"
prop_checkBashisms37= verifyNot checkBashisms "HOSTNAME=foo; echo $HOSTNAME"
prop_checkBashisms38= verify checkBashisms "RANDOM=9; echo $RANDOM"
prop_checkBashisms39= verify checkBashisms "foo-bar() { true; }"
prop_checkBashisms40= verify checkBashisms "echo $(<file)"
prop_checkBashisms41= verify checkBashisms "echo `<file`"
prop_checkBashisms42= verify checkBashisms "trap foo int"
prop_checkBashisms43= verify checkBashisms "trap foo sigint"
prop_checkBashisms44= verifyNot checkBashisms "#!/bin/dash\ntrap foo int"
prop_checkBashisms45= verifyNot checkBashisms "#!/bin/dash\ntrap foo INT"
prop_checkBashisms46= verify checkBashisms "#!/bin/dash\ntrap foo SIGINT"
prop_checkBashisms47= verify checkBashisms "#!/bin/dash\necho foo 42>/dev/null"
prop_checkBashisms48= verifyNot checkBashisms "#!/bin/dash\necho $LINENO"
prop_checkBashisms49= verify checkBashisms "#!/bin/dash\necho $MACHTYPE"
prop_checkBashisms50= verify checkBashisms "#!/bin/sh\ncmd >& file"
prop_checkBashisms51= verifyNot checkBashisms "#!/bin/sh\ncmd 2>&1"
prop_checkBashisms52= verifyNot checkBashisms "#!/bin/sh\ncmd >&2"
prop_checkBashisms53= verifyNot checkBashisms "#!/bin/sh\nprintf -- -f\n"
prop_checkBashisms54= verify checkBashisms "#!/bin/sh\nfoo+=bar"
checkBashisms = ForShell [Sh, Dash] $ \t -> do
params <- ask
kludge params t
where
-- This code was copy-pasted from Analytics where params was a variable
kludge params = bashism
where
isDash = shellType params == Dash
warnMsg id s =
if isDash
then warn id 2169 $ "In dash, " ++ s ++ " not supported."
else warn id 2039 $ "In POSIX sh, " ++ s ++ " undefined."
bashism (T_ProcSub id _ _) = warnMsg id "process substitution is"
bashism (T_Extglob id _ _) = warnMsg id "extglob is"
bashism (T_DollarSingleQuoted id _) = warnMsg id "$'..' is"
bashism (T_DollarDoubleQuoted id _) = warnMsg id "$\"..\" is"
bashism (T_ForArithmetic id _ _ _ _) = warnMsg id "arithmetic for loops are"
bashism (T_Arithmetic id _) = warnMsg id "standalone ((..)) is"
bashism (T_DollarBracket id _) = warnMsg id "$[..] in place of $((..)) is"
bashism (T_SelectIn id _ _ _) = warnMsg id "select loops are"
bashism (T_BraceExpansion id _) = warnMsg id "brace expansion is"
bashism (T_Condition id DoubleBracket _) = warnMsg id "[[ ]] is"
bashism (T_HereString id _) = warnMsg id "here-strings are"
bashism (TC_Binary id SingleBracket op _ _)
| op `elem` [ "-nt", "-ef", "\\<", "\\>"] =
unless isDash $ warnMsg id $ op ++ " is"
bashism (TC_Binary id SingleBracket "==" _ _) =
warnMsg id "== in place of = is"
bashism (TC_Unary id _ "-a" _) =
warnMsg id "unary -a in place of -e is"
bashism (TA_Unary id op _)
| op `elem` [ "|++", "|--", "++|", "--|"] =
warnMsg id $ filter (/= '|') op ++ " is"
bashism (TA_Binary id "**" _ _) = warnMsg id "exponentials are"
bashism (T_FdRedirect id "&" (T_IoFile _ (T_Greater _) _)) = warnMsg id "&> is"
bashism (T_FdRedirect id "" (T_IoFile _ (T_GREATAND _) _)) = warnMsg id ">& is"
bashism (T_FdRedirect id ('{':_) _) = warnMsg id "named file descriptors are"
bashism (T_FdRedirect id num _)
| all isDigit num && length num > 1 = warnMsg id "FDs outside 0-9 are"
bashism (T_Assignment id Append _ _ _) =
warnMsg id "+= is"
bashism (T_IoFile id _ word) | isNetworked =
warnMsg id "/dev/{tcp,udp} is"
where
file = onlyLiteralString word
isNetworked = any (`isPrefixOf` file) ["/dev/tcp", "/dev/udp"]
bashism (T_Glob id str) | "[^" `isInfixOf` str =
warnMsg id "^ in place of ! in glob bracket expressions is"
bashism t@(TA_Expansion id _) | isBashism =
warnMsg id $ fromJust str ++ " is"
where
str = getLiteralString t
isBashism = isJust str && isBashVariable (fromJust str)
bashism t@(T_DollarBraced id token) = do
mapM_ check expansion
when (isBashVariable var) $
warnMsg id $ var ++ " is"
where
str = bracedString t
var = getBracedReference str
check (regex, feature) =
when (isJust $ matchRegex regex str) $ warnMsg id feature
bashism t@(T_Pipe id "|&") =
warnMsg id "|& in place of 2>&1 | is"
bashism (T_Array id _) =
warnMsg id "arrays are"
bashism (T_IoFile id _ t) | isGlob t =
warnMsg id "redirecting to/from globs is"
bashism (T_CoProc id _ _) =
warnMsg id "coproc is"
bashism (T_Function id _ _ str _) | not (isVariableName str) =
warnMsg id "naming functions outside [a-zA-Z_][a-zA-Z0-9_]* is"
bashism (T_DollarExpansion id [x]) | isOnlyRedirection x =
warnMsg id "$(<file) to read files is"
bashism (T_Backticked id [x]) | isOnlyRedirection x =
warnMsg id "`<file` to read files is"
bashism t@(T_SimpleCommand _ _ (cmd:arg:_))
| t `isCommand` "echo" && "-" `isPrefixOf` argString =
unless ("--" `isPrefixOf` argString) $ -- echo "-----"
if isDash
then
when (argString /= "-n") $
warnMsg (getId arg) "echo flags besides -n"
else
warnMsg (getId arg) "echo flags are"
where argString = concat $ oversimplify arg
bashism t@(T_SimpleCommand _ _ (cmd:arg:_))
| t `isCommand` "exec" && "-" `isPrefixOf` concat (oversimplify arg) =
warnMsg (getId arg) "exec flags are"
bashism t@(T_SimpleCommand id _ _)
| t `isCommand` "let" = warnMsg id "'let' is"
bashism t@(T_SimpleCommand id _ (cmd:rest)) =
let name = fromMaybe "" $ getCommandName t
flags = getLeadingFlags t
in do
when (name `elem` unsupportedCommands) $
warnMsg id $ "'" ++ name ++ "' is"
potentially $ do
allowed <- Map.lookup name allowedFlags
(word, flag) <- listToMaybe $
filter (\x -> (not . null . snd $ x) && snd x `notElem` allowed) flags
return . warnMsg (getId word) $ name ++ " -" ++ flag ++ " is"
when (name == "source") $ warnMsg id "'source' in place of '.' is"
when (name == "trap") $
let
check token = potentially $ do
str <- getLiteralString token
let upper = map toUpper str
return $ do
when (upper `elem` ["ERR", "DEBUG", "RETURN"]) $
warnMsg (getId token) $ "trapping " ++ str ++ " is"
when ("SIG" `isPrefixOf` upper) $
warnMsg (getId token)
"prefixing signal names with 'SIG' is"
when (not isDash && upper /= str) $
warnMsg (getId token)
"using lower/mixed case for signal names is"
in
mapM_ check (drop 1 rest)
when (name == "printf") $ potentially $ do
format <- rest !!! 0 -- flags are covered by allowedFlags
let literal = onlyLiteralString format
guard $ "%q" `isInfixOf` literal
return $ warnMsg (getId format) "printf %q is"
where
unsupportedCommands = [
"let", "caller", "builtin", "complete", "compgen", "declare", "dirs", "disown",
"enable", "mapfile", "readarray", "pushd", "popd", "shopt", "suspend",
"typeset"
] ++ if not isDash then ["local", "type"] else []
allowedFlags = Map.fromList [
("read", if isDash then ["r", "p"] else ["r"]),
("ulimit", ["f"]),
("printf", []),
("exec", [])
]
bashism _ = return ()
varChars="_0-9a-zA-Z"
expansion = let re = mkRegex in [
(re $ "^![" ++ varChars ++ "]", "indirect expansion is"),
(re $ "^[" ++ varChars ++ "]+\\[.*\\]$", "array references are"),
(re $ "^![" ++ varChars ++ "]+\\[[*@]]$", "array key expansion is"),
(re $ "^![" ++ varChars ++ "]+[*@]$", "name matching prefixes are"),
(re $ "^[" ++ varChars ++ "]+:[^-=?+]", "string indexing is"),
(re $ "^[" ++ varChars ++ "]+(\\[.*\\])?/", "string replacement is")
]
bashVars = [
"LINENO", "OSTYPE", "MACHTYPE", "HOSTTYPE", "HOSTNAME",
"DIRSTACK", "EUID", "UID", "SHLVL", "PIPESTATUS", "SHELLOPTS"
]
bashDynamicVars = [ "RANDOM", "SECONDS" ]
dashVars = [ "LINENO" ]
isBashVariable var =
(var `elem` bashDynamicVars
|| var `elem` bashVars && not (isAssigned var))
&& not (isDash && var `elem` dashVars)
isAssigned var = any f (variableFlow params)
where
f x = case x of
Assignment (_, _, name, _) -> name == var
_ -> False
prop_checkEchoSed1 = verify checkEchoSed "FOO=$(echo \"$cow\" | sed 's/foo/bar/g')"
prop_checkEchoSed2 = verify checkEchoSed "rm $(echo $cow | sed -e 's,foo,bar,')"
checkEchoSed = ForShell [Bash, Ksh] f
where
f (T_Pipeline id _ [a, b]) =
when (acmd == ["echo", "${VAR}"]) $
case bcmd of
["sed", v] -> checkIn v
["sed", "-e", v] -> checkIn v
_ -> return ()
where
-- This should have used backreferences, but TDFA doesn't support them
sedRe = mkRegex "^s(.)([^\n]*)g?$"
isSimpleSed s = fromMaybe False $ do
[first,rest] <- matchRegex sedRe s
let delimiters = filter (== head first) rest
guard $ length delimiters == 2
return True
acmd = oversimplify a
bcmd = oversimplify b
checkIn s =
when (isSimpleSed s) $
style id 2001 "See if you can use ${variable//search/replace} instead."
f _ = return ()
prop_checkBraceExpansionVars1 = verify checkBraceExpansionVars "echo {1..$n}"
prop_checkBraceExpansionVars2 = verifyNot checkBraceExpansionVars "echo {1,3,$n}"
prop_checkBraceExpansionVars3 = verify checkBraceExpansionVars "eval echo DSC{0001..$n}.jpg"
prop_checkBraceExpansionVars4 = verify checkBraceExpansionVars "echo {$i..100}"
checkBraceExpansionVars = ForShell [Bash] f
where
f t@(T_BraceExpansion id list) = mapM_ check list
where
check element =
when (any (`isInfixOf` toString element) ["$..", "..$"]) $ do
c <- isEvaled element
if c
then style id 2175 "Quote this invalid brace expansion since it should be passed literally to eval."
else warn id 2051 "Bash doesn't support variables in brace range expansions."
f _ = return ()
literalExt t =
case t of
T_DollarBraced {} -> return "$"
T_DollarExpansion {} -> return "$"
T_DollarArithmetic {} -> return "$"
otherwise -> return "-"
toString t = fromJust $ getLiteralStringExt literalExt t
isEvaled t = do
cmd <- getClosestCommandM t
return $ isJust cmd && fromJust cmd `isUnqualifiedCommand` "eval"
prop_checkMultiDimensionalArrays1 = verify checkMultiDimensionalArrays "foo[a][b]=3"
prop_checkMultiDimensionalArrays2 = verifyNot checkMultiDimensionalArrays "foo[a]=3"
prop_checkMultiDimensionalArrays3 = verify checkMultiDimensionalArrays "foo=( [a][b]=c )"
prop_checkMultiDimensionalArrays4 = verifyNot checkMultiDimensionalArrays "foo=( [a]=c )"
prop_checkMultiDimensionalArrays5 = verify checkMultiDimensionalArrays "echo ${foo[bar][baz]}"
prop_checkMultiDimensionalArrays6 = verifyNot checkMultiDimensionalArrays "echo ${foo[bar]}"
checkMultiDimensionalArrays = ForShell [Bash] f
where
f token =
case token of
T_Assignment _ _ name (first:second:_) _ -> about second
T_IndexedElement _ (first:second:_) _ -> about second
T_DollarBraced {} ->
when (isMultiDim token) $ about token
_ -> return ()
about t = warn (getId t) 2180 "Bash does not support multidimensional arrays. Use 1D or associative arrays."
re = mkRegex "^\\[.*\\]\\[.*\\]" -- Fixme, this matches ${foo:- [][]} and such as well
isMultiDim t = getBracedModifier (bracedString t) `matches` re
return []
runTests = $( [| $(forAllProperties) (quickCheckWithResult (stdArgs { maxSuccess = 1 }) ) |])

View File

@@ -33,6 +33,9 @@ internalVariables = [
-- Other
"USER", "TZ", "TERM", "LOGNAME", "LD_LIBRARY_PATH", "LANGUAGE", "DISPLAY",
"HOSTNAME", "KRB5CCNAME", "XAUTHORITY"
-- Ksh
, ".sh.version"
]
variablesWithoutSpaces = [
@@ -82,12 +85,24 @@ sampleWords = [
"zulu"
]
binaryTestOps = [
"-nt", "-ot", "-ef", "==", "!=", "<=", ">=", "-eq", "-ne", "-lt", "-le",
"-gt", "-ge", "=~", ">", "<", "=", "\\<", "\\>", "\\<=", "\\>="
]
unaryTestOps = [
"!", "-a", "-b", "-c", "-d", "-e", "-f", "-g", "-h", "-L", "-k", "-p",
"-r", "-s", "-S", "-t", "-u", "-w", "-x", "-O", "-G", "-N", "-z", "-n",
"-o", "-v", "-R"
]
shellForExecutable :: String -> Maybe Shell
shellForExecutable name =
case name of
"sh" -> return Sh
"bash" -> return Bash
"dash" -> return Dash
"ash" -> return Dash -- There's also a warning for this.
"ksh" -> return Ksh
"ksh88" -> return Ksh
"ksh93" -> return Ksh

View File

@@ -30,13 +30,15 @@ data Formatter = Formatter {
footer :: IO ()
}
lineNo (PositionedComment pos _) = posLine pos
colNo (PositionedComment pos _) = posColumn pos
codeNo (PositionedComment _ (Comment _ code _)) = code
messageText (PositionedComment _ (Comment _ _ t)) = t
lineNo (PositionedComment pos _ _) = posLine pos
endLineNo (PositionedComment _ end _) = posLine end
colNo (PositionedComment pos _ _) = posColumn pos
endColNo (PositionedComment _ end _) = posColumn end
codeNo (PositionedComment _ _ (Comment _ code _)) = code
messageText (PositionedComment _ _ (Comment _ _ t)) = t
severityText :: PositionedComment -> String
severityText (PositionedComment _ (Comment c _ _)) =
severityText (PositionedComment _ _ (Comment c _ _)) =
case c of
ErrorC -> "error"
WarningC -> "warning"
@@ -48,12 +50,15 @@ makeNonVirtual comments contents =
map fix comments
where
ls = lines contents
fix c@(PositionedComment pos comment) = PositionedComment pos {
posColumn =
if lineNo c > 0 && lineNo c <= fromIntegral (length ls)
then real (ls !! fromIntegral (lineNo c - 1)) 0 0 (colNo c)
else colNo c
fix c@(PositionedComment start end comment) = PositionedComment start {
posColumn = realignColumn lineNo colNo c
} end {
posColumn = realignColumn endLineNo endColNo c
} comment
realignColumn lineNo colNo c =
if lineNo c > 0 && lineNo c <= fromIntegral (length ls)
then real (ls !! fromIntegral (lineNo c - 1)) 0 0 (colNo c)
else colNo c
real _ r v target | target <= v = r
real [] r v _ = r -- should never happen
real ('\t':rest) r v target =

View File

@@ -37,10 +37,12 @@ format = do
}
instance JSON (PositionedComment) where
showJSON comment@(PositionedComment pos (Comment level code string)) = makeObj [
("file", showJSON $ posFile pos),
("line", showJSON $ posLine pos),
("column", showJSON $ posColumn pos),
showJSON comment@(PositionedComment start end (Comment level code string)) = makeObj [
("file", showJSON $ posFile start),
("line", showJSON $ posLine start),
("endLine", showJSON $ posLine end),
("column", showJSON $ posColumn start),
("endColumn", showJSON $ posColumn end),
("level", showJSON $ severityText comment),
("code", showJSON code),
("message", showJSON string)

View File

@@ -94,7 +94,7 @@ data Position = Position {
} deriving (Show, Eq)
data Comment = Comment Severity Code String deriving (Show, Eq)
data PositionedComment = PositionedComment Position Comment deriving (Show, Eq)
data PositionedComment = PositionedComment Position Position Comment deriving (Show, Eq)
data TokenComment = TokenComment Id Comment deriving (Show, Eq)
data ColorOption =

View File

@@ -58,17 +58,18 @@ linefeed = do
c <- char '\n'
readPendingHereDocs
return c
singleQuote = char '\'' <|> unicodeSingleQuote
doubleQuote = char '"' <|> unicodeDoubleQuote
singleQuote = char '\''
doubleQuote = char '"'
variableStart = upper <|> lower <|> oneOf "_"
variableChars = upper <|> lower <|> digit <|> oneOf "_"
functionChars = variableChars <|> oneOf ":+-.?"
specialVariable = oneOf "@*#?-$!"
paramSubSpecialChars = oneOf "/:+-=%"
quotableChars = "|&;<>()\\ '\t\n\r\xA0" ++ doubleQuotableChars
quotable = almostSpace <|> unicodeDoubleQuote <|> oneOf quotableChars
quotable = almostSpace <|> oneOf quotableChars
bracedQuotable = oneOf "}\"$`'"
doubleQuotableChars = "\"$`" ++ unicodeDoubleQuoteChars
doubleQuotable = unicodeDoubleQuote <|> oneOf doubleQuotableChars
doubleQuotableChars = "\"$`"
doubleQuotable = oneOf doubleQuotableChars
whitespace = oneOf " \t" <|> carriageReturn <|> almostSpace <|> linefeed
linewhitespace = oneOf " \t" <|> almostSpace
@@ -77,7 +78,8 @@ suspectCharAfterQuotes = variableChars <|> char '%'
extglobStartChars = "?*@!+"
extglobStart = oneOf extglobStartChars
unicodeDoubleQuoteChars = "\x201C\x201D\x2033\x2036"
unicodeDoubleQuotes = "\x201C\x201D\x2033\x2036"
unicodeSingleQuotes = "\x2018\x2019"
prop_spacing = isOk spacing " \\\n # Comment"
spacing = do
@@ -106,17 +108,12 @@ allspacingOrFail = do
s <- allspacing
when (null s) $ fail "Expected whitespace"
unicodeDoubleQuote = do
readUnicodeQuote = do
pos <- getPosition
oneOf unicodeDoubleQuoteChars
parseProblemAt pos WarningC 1015 "This is a unicode double quote. Delete and retype it."
return '"'
unicodeSingleQuote = do
pos <- getPosition
char '\x2018' <|> char '\x2019'
parseProblemAt pos WarningC 1016 "This is a unicode single quote. Delete and retype it."
return '"'
c <- oneOf (unicodeSingleQuotes ++ unicodeDoubleQuotes)
parseProblemAt pos WarningC 1110 "This is a unicode quote. Delete and retype it (or quote to make literal)."
id <- getNextIdAt pos
return $ T_Literal id [c]
carriageReturn = do
parseNote ErrorC 1017 "Literal carriage return. Run script through tr -d '\\r' ."
@@ -135,7 +132,7 @@ almostSpace =
--------- Message/position annotation on top of user state
data Note = Note Id Severity Code String deriving (Show, Eq)
data ParseNote = ParseNote SourcePos Severity Code String deriving (Show, Eq)
data ParseNote = ParseNote SourcePos SourcePos Severity Code String deriving (Show, Eq)
data Context =
ContextName SourcePos String
| ContextAnnotation [Annotation]
@@ -162,9 +159,9 @@ initialUserState = UserState {
pendingHereDocs = []
}
codeForParseNote (ParseNote _ _ code _) = code
codeForParseNote (ParseNote _ _ _ code _) = code
noteToParseNote map (Note id severity code message) =
ParseNote pos severity code message
ParseNote pos pos severity code message
where
pos = fromJust $ Map.lookup id map
@@ -181,6 +178,7 @@ getNextIdAt sourcepos = do
return newId
where incId (Id n) = Id $ n+1
getNextId :: Monad m => SCParser m Id
getNextId = do
pos <- getPosition
getNextIdAt pos
@@ -320,14 +318,23 @@ pushContext c = do
v <- getCurrentContexts
setCurrentContexts (c:v)
parseProblemAt pos level code msg = do
parseProblemAtWithEnd start end level code msg = do
irrelevant <- shouldIgnoreCode code
unless irrelevant $
Ms.modify (\state -> state {
parseProblems = note:parseProblems state
})
where
note = ParseNote pos level code msg
note = ParseNote start end level code msg
parseProblemAt pos = parseProblemAtWithEnd pos pos
parseProblemAtId :: Monad m => Id -> Severity -> Integer -> String -> SCParser m ()
parseProblemAtId id level code msg = do
map <- getMap
let pos = Map.findWithDefault
(error "Internal error (no position for id). Please report.") id map
parseProblemAt pos level code msg
-- Store non-parse problems inside
@@ -335,7 +342,9 @@ parseNote c l a = do
pos <- getPosition
parseNoteAt pos c l a
parseNoteAt pos c l a = addParseNote $ ParseNote pos c l a
parseNoteAt pos c l a = addParseNote $ ParseNote pos pos c l a
parseNoteAtWithEnd start end c l a = addParseNote $ ParseNote start end c l a
--------- Convenient combinators
thenSkip main follow = do
@@ -406,7 +415,7 @@ readConditionContents single =
pos <- getPosition
s <- many1 letter
when (s `elem` commonCommands) $
parseProblemAt pos WarningC 1009 "Use 'if cmd; then ..' to check exit code, or 'if [[ $(cmd) == .. ]]' to check output.")
parseProblemAt pos WarningC 1014 "Use 'if cmd; then ..' to check exit code, or 'if [[ $(cmd) == .. ]]' to check output.")
where
spacingOrLf = condSpacing True
@@ -442,7 +451,7 @@ readConditionContents single =
c <- oneOf "'\""
s <- anyOp
char c
return s
return $ escaped s
anyOp = flagOp <|> flaglessOp <|> fail
"Expected comparison operator (don't wrap commands in []/[[]])"
@@ -522,7 +531,7 @@ readConditionContents single =
condSpacing requiresSpacing
return x
readCondNoaryOrBinary = do
readCondNullaryOrBinary = do
id <- getNextId
x <- readCondWord `attempting` (do
pos <- getPosition
@@ -539,7 +548,16 @@ readConditionContents single =
then readRegex
else readCondWord <|> (parseProblemAt pos ErrorC 1027 "Expected another argument for this operator." >> mzero)
return (x `op` y)
) <|> return (TC_Noary id typ x)
) <|> ( do
checkTrailingOp x
return $ TC_Nullary id typ x
)
checkTrailingOp x = fromMaybe (return ()) $ do
(T_Literal id str) <- getTrailingUnquotedLiteral x
trailingOp <- listToMaybe (filter (`isSuffixOf` str) binaryTestOps)
return $ parseProblemAtId id ErrorC 1108 $
"You need a space before and after the " ++ trailingOp ++ " ."
readCondGroup = do
id <- getNextId
@@ -616,7 +634,7 @@ readConditionContents single =
return $ TC_Unary id typ "!" expr
readCondExpr =
readCondGroup <|> readCondUnaryExp <|> readCondNoaryOrBinary
readCondGroup <|> readCondUnaryExp <|> readCondNullaryOrBinary
readCondOr = chainl1 readCondAnd readCondAndOp
readCondAnd = chainl1 readCondTerm readCondOrOp
@@ -645,6 +663,7 @@ prop_a19= isOk readArithmeticContents "\\\n3 +\\\n 2"
prop_a20= isOk readArithmeticContents "a ? b ? c : d : e"
prop_a21= isOk readArithmeticContents "a ? b : c ? d : e"
prop_a22= isOk readArithmeticContents "!!a"
readArithmeticContents :: Monad m => SCParser m Token
readArithmeticContents =
readSequence
where
@@ -658,12 +677,40 @@ readArithmeticContents =
id <- getNextId
op <- choice (map (\x -> try $ do
s <- string x
notFollowedBy2 $ oneOf "&|<>="
failIfIncompleteOp
return s
) op)
spacing
return $ token id op
failIfIncompleteOp = notFollowedBy2 $ oneOf "&|<>="
-- Read binary minus, but also check for -lt, -gt and friends:
readMinusOp = do
id <- getNextId
pos <- getPosition
try $ do
char '-'
failIfIncompleteOp
optional $ do
(str, alt) <- lookAhead . choice $ map tryOp [
("lt", "<"),
("gt", ">"),
("le", "<="),
("ge", ">="),
("eq", "=="),
("ne", "!=")
]
parseProblemAt pos ErrorC 1106 $ "In arithmetic contexts, use " ++ alt ++ " instead of -" ++ str
spacing
return $ TA_Binary id "-"
where
tryOp (str, alt) = try $ do
string str
spacing1
return (str, alt)
readArrayIndex = do
id <- getNextId
char '['
@@ -733,7 +780,7 @@ readArithmeticContents =
readEquated = readCompared `splitBy` ["==", "!="]
readCompared = readShift `splitBy` ["<=", ">=", "<", ">"]
readShift = readAddition `splitBy` ["<<", ">>"]
readAddition = readMultiplication `splitBy` ["+", "-"]
readAddition = chainl1 readMultiplication (readBinary ["+"] <|> readMinusOp)
readMultiplication = readExponential `splitBy` ["*", "/", "%"]
readExponential = readAnyNegated `splitBy` ["**"]
@@ -810,7 +857,7 @@ readCondition = called "test expression" $ do
pos <- getPosition
space <- allspacing
when (null space) $
parseProblemAt pos ErrorC 1035 $ "You need a space after the " ++
parseProblemAtWithEnd opos pos ErrorC 1035 $ "You need a space after the " ++
if single
then "[ and before the ]."
else "[[ and before the ]]."
@@ -835,10 +882,11 @@ readAnnotationPrefix = do
prop_readAnnotation1 = isOk readAnnotation "# shellcheck disable=1234,5678\n"
prop_readAnnotation2 = isOk readAnnotation "# shellcheck disable=SC1234 disable=SC5678\n"
prop_readAnnotation3 = isOk readAnnotation "# shellcheck disable=SC1234 source=/dev/null disable=SC5678\n"
prop_readAnnotation4 = isWarning readAnnotation "# shellcheck cats=dogs disable=SC1234\n"
readAnnotation = called "shellcheck annotation" $ do
try readAnnotationPrefix
many1 linewhitespace
values <- many1 (readDisable <|> readSourceOverride <|> readShellOverride)
values <- many1 (readDisable <|> readSourceOverride <|> readShellOverride <|> anyKey)
linefeed
many linewhitespace
return $ concat values
@@ -870,6 +918,13 @@ readAnnotation = called "shellcheck annotation" $ do
many linewhitespace
return value
anyKey = do
pos <- getPosition
anyChar `reluctantlyTill1` whitespace
many linewhitespace
parseNoteAt pos WarningC 1107 "This directive is unknown. It will be ignored."
return []
readAnnotations = do
annotations <- many (readAnnotation `thenSkip` allspacing)
return $ concat annotations
@@ -888,6 +943,9 @@ prop_readNormalWord6 = isOk readNormalWord "foo/{}"
prop_readNormalWord7 = isOk readNormalWord "foo\\\nbar"
prop_readNormalWord8 = isWarning readSubshell "(foo\\ \nbar)"
prop_readNormalWord9 = isOk readSubshell "(foo\\ ;\nbar)"
prop_readNormalWord10 = isWarning readNormalWord "\x201Chello\x201D"
prop_readNormalWord11 = isWarning readNormalWord "\x2018hello\x2019"
prop_readNormalWord12 = isWarning readNormalWord "hello\x2018"
readNormalWord = readNormalishWord ""
readNormalishWord end = do
@@ -897,6 +955,20 @@ readNormalishWord end = do
checkPossibleTermination pos x
return $ T_NormalWord id x
readIndexSpan = do
id <- getNextId
x <- many (readNormalWordPart "]" <|> someSpace <|> otherLiteral)
return $ T_NormalWord id x
where
someSpace = do
id <- getNextId
str <- spacing1
return $ T_Literal id str
otherLiteral = do
id <- getNextId
str <- many1 $ oneOf quotableChars
return $ T_Literal id str
checkPossibleTermination pos [T_Literal _ x] =
when (x `elem` ["do", "done", "then", "fi", "esac"]) $
parseProblemAt pos WarningC 1010 $ "Use semicolon or linefeed before '" ++ x ++ "' (or quote to make it literal)."
@@ -913,6 +985,7 @@ readNormalWordPart end = do
readBraced,
readUnquotedBackTicked,
readProcSub,
readUnicodeQuote,
readNormalLiteral end,
readLiteralCurlyBraces
]
@@ -947,13 +1020,19 @@ readDollarBracedWord = do
list <- many readDollarBracedPart
return $ T_NormalWord id list
readDollarBracedPart = readSingleQuoted <|> readDoubleQuoted <|> readExtglob <|> readNormalDollar <|> readUnquotedBackTicked <|> readDollarBracedLiteral
readDollarBracedPart = readSingleQuoted <|> readDoubleQuoted <|>
readParamSubSpecialChar <|> readExtglob <|> readNormalDollar <|>
readUnquotedBackTicked <|> readDollarBracedLiteral
readDollarBracedLiteral = do
id <- getNextId
vars <- (readBraceEscaped <|> (anyChar >>= \x -> return [x])) `reluctantlyTill1` bracedQuotable
return $ T_Literal id $ concat vars
readParamSubSpecialChar = do
id <- getNextId
T_ParamSubSpecialChar id <$> many1 paramSubSpecialChars
prop_readProcSub1 = isOk readProcSub "<(echo test | wc -l)"
prop_readProcSub2 = isOk readProcSub "<( if true; then true; fi )"
prop_readProcSub3 = isOk readProcSub "<( # nothing here \n)"
@@ -970,15 +1049,16 @@ readProcSub = called "process substitution" $ do
prop_readSingleQuoted = isOk readSingleQuoted "'foo bar'"
prop_readSingleQuoted2 = isWarning readSingleQuoted "'foo bar\\'"
prop_readsingleQuoted3 = isWarning readSingleQuoted "\x2018hello\x2019"
prop_readSingleQuoted4 = isWarning readNormalWord "'it's"
prop_readSingleQuoted5 = isWarning readSimpleCommand "foo='bar\ncow 'arg"
prop_readSingleQuoted6 = isOk readSimpleCommand "foo='bar cow 'arg"
prop_readSingleQuoted7 = isOk readSingleQuoted "'foo\x201C\&bar'"
prop_readSingleQuoted8 = isWarning readSingleQuoted "'foo\x2018\&bar'"
readSingleQuoted = called "single quoted string" $ do
id <- getNextId
startPos <- getPosition
singleQuote
s <- readSingleQuotedPart `reluctantlyTill` singleQuote
s <- many readSingleQuotedPart
let string = concat s
endPos <- getPosition
singleQuote <|> fail "Expected end of single quoted string"
@@ -1003,7 +1083,15 @@ readSingleQuotedLiteral = do
readSingleQuotedPart =
readSingleEscaped
<|> many1 (noneOf "'\\\x2018\x2019")
<|> many1 (noneOf $ "'\\" ++ unicodeSingleQuotes)
<|> readUnicodeQuote
where
readUnicodeQuote = do
pos <- getPosition
x <- oneOf unicodeSingleQuotes
parseProblemAt pos WarningC 1112
"This is a unicode quote. Delete and retype it (or ignore/doublequote for literal)."
return [x]
prop_readBackTicked = isOk (readBackTicked False) "`ls *.mp3`"
@@ -1060,13 +1148,18 @@ subParse pos parser input = do
setPosition lastPosition
return result
inSeparateContext parser = do
-- Parse something, but forget all parseProblems
inSeparateContext = parseForgettingContext True
-- Parse something, but forget all parseProblems on failure
forgetOnFailure = parseForgettingContext False
parseForgettingContext alsoOnSuccess parser = do
context <- Ms.get
success context <|> failure context
where
success c = do
res <- try parser
Ms.put c
when alsoOnSuccess $ Ms.put c
return res
failure c = do
Ms.put c
@@ -1074,11 +1167,12 @@ inSeparateContext parser = do
prop_readDoubleQuoted = isOk readDoubleQuoted "\"Hello $FOO\""
prop_readDoubleQuoted2 = isOk readDoubleQuoted "\"$'\""
prop_readDoubleQuoted3 = isWarning readDoubleQuoted "\x201Chello\x201D"
prop_readDoubleQuoted3 = isOk readDoubleQuoted "\"\x2018hello\x2019\""
prop_readDoubleQuoted4 = isWarning readSimpleCommand "\"foo\nbar\"foo"
prop_readDoubleQuoted5 = isOk readSimpleCommand "lol \"foo\nbar\" etc"
prop_readDoubleQuoted6 = isOk readSimpleCommand "echo \"${ ls; }\""
prop_readDoubleQuoted7 = isOk readSimpleCommand "echo \"${ ls;}bar\""
prop_readDoubleQuoted8 = isWarning readDoubleQuoted "\"\x201Chello\x201D\""
readDoubleQuoted = called "double quoted string" $ do
id <- getNextId
startPos <- getPosition
@@ -1103,7 +1197,15 @@ suggestForgotClosingQuote startPos endPos name = do
parseProblemAt endPos InfoC 1079
"This is actually an end quote, but due to next char it looks suspect."
doubleQuotedPart = readDoubleLiteral <|> readDoubleQuotedDollar <|> readQuotedBackTicked
doubleQuotedPart = readDoubleLiteral <|> readDoubleQuotedDollar <|> readQuotedBackTicked <|> readUnicodeQuote
where
readUnicodeQuote = do
pos <- getPosition
id <- getNextId
c <- oneOf unicodeDoubleQuotes
parseProblemAt pos WarningC 1111
"This is a unicode quote. Delete and retype it (or ignore/singlequote for literal)."
return $ T_Literal id [c]
readDoubleQuotedLiteral = do
doubleQuote
@@ -1117,7 +1219,7 @@ readDoubleLiteral = do
return $ T_Literal id (concat s)
readDoubleLiteralPart = do
x <- many1 (readDoubleEscaped <|> many1 (noneOf ('\\':doubleQuotableChars)))
x <- many1 (readDoubleEscaped <|> many1 (noneOf ('\\':doubleQuotableChars ++ unicodeDoubleQuotes)))
return $ concat x
readNormalLiteral end = do
@@ -1159,8 +1261,15 @@ readGlob = readExtglob <|> readSimple <|> readClass <|> readGlobbyLiteral
c <- extglobStart <|> char '['
return $ T_Literal id [c]
readNormalLiteralPart end =
readNormalEscaped <|> many1 (noneOf (end ++ quotableChars ++ extglobStartChars ++ "[{}"))
readNormalLiteralPart customEnd =
readNormalEscaped <|>
many1 (noneOf (customEnd ++ standardEnd))
where
standardEnd = "[{}"
++ quotableChars
++ extglobStartChars
++ unicodeDoubleQuotes
++ unicodeSingleQuotes
readNormalEscaped = called "escaped char" $ do
pos <- getPosition
@@ -1225,19 +1334,16 @@ readExtglobPart = do
readSingleEscaped = do
pos <- getPosition
s <- backslash
let attempt level code p msg = do { try $ parseNote level code msg; x <- p; return [s,x]; }
x <- lookAhead anyChar
do {
x <- lookAhead singleQuote;
parseProblem InfoC 1003 "Are you trying to escape that single quote? echo 'You'\\''re doing it wrong'.";
return [s];
}
<|> attempt InfoC 1004 linefeed "You don't break lines with \\ in single quotes, it results in literal backslash-linefeed."
<|> do
x <- anyChar
return [s,x]
case x of
'\'' -> parseProblemAt pos InfoC 1003 "Want to escape a single quote? echo 'This is how it'\''s done'.";
'\n' -> parseProblemAt pos InfoC 1004 "This backslash+linefeed is literal. Break outside single quotes if you just want to break the line."
_ -> return ()
return [s]
readDoubleEscaped = do
bs <- backslash
@@ -1298,12 +1404,32 @@ readBraced = try braceExpansion
braceLiteral =
T_Literal `withParser` readGenericLiteral1 (oneOf "{}\"$'," <|> whitespace)
readNormalDollar = readDollarExpression <|> readDollarDoubleQuote <|> readDollarSingleQuote <|> readDollarLonely
readDoubleQuotedDollar = readDollarExpression <|> readDollarLonely
ensureDollar =
-- The grammar should have been designed along the lines of readDollarExpr = char '$' >> stuff, but
-- instead, each subunit parses its own $. This results in ~7 1-3 char lookaheads instead of one 1-char.
-- Instead of optimizing the grammar, here's a green cut that decreases shellcheck runtime by 10%:
lookAhead $ char '$'
readNormalDollar = do
ensureDollar
readDollarExp <|> readDollarDoubleQuote <|> readDollarSingleQuote <|> readDollarLonely
readDoubleQuotedDollar = do
ensureDollar
readDollarExp <|> readDollarLonely
prop_readDollarExpression1 = isOk readDollarExpression "$(((1) && 3))"
prop_readDollarExpression2 = isWarning readDollarExpression "$(((1)) && 3)"
readDollarExpression = readTripleParenthesis "$" readDollarArithmetic readDollarExpansion <|> readDollarArithmetic <|> readDollarBracket <|> readDollarBraceCommandExpansion <|> readDollarBraced <|> readDollarExpansion <|> readDollarVariable
prop_readDollarExpression3 = isWarning readDollarExpression "$((\"$@\" &); foo;)"
readDollarExpression :: Monad m => SCParser m Token
readDollarExpression = do
ensureDollar
readDollarExp
readDollarExp = arithmetic <|> readDollarExpansion <|> readDollarBracket <|> readDollarBraceCommandExpansion <|> readDollarBraced <|> readDollarVariable
where
arithmetic = readAmbiguous "$((" readDollarArithmetic readDollarExpansion (\pos ->
parseNoteAt pos WarningC 1102 "Shells disambiguate $(( differently or not at all. For $(command substition), add space after $( . For $((arithmetics)), fix parsing errors.")
prop_readDollarSingleQuote = isOk readDollarSingleQuote "$'foo\\\'lol'"
readDollarSingleQuote = called "$'..' expression" $ do
@@ -1349,25 +1475,20 @@ readArithmeticExpression = called "((..)) command" $ do
string "))"
return (T_Arithmetic id c)
-- Check if maybe ((( was intended as ( (( rather than (( (
readTripleParenthesis prefix expected alternative = do
pos <- try . lookAhead $ do
string prefix
p <- getPosition
string "(((" -- should optimally be "((" but it's noisy and rarely helpful
return p
-- If the next characters match prefix, try two different parsers and warn if the alternate parser had to be used
readAmbiguous :: Monad m => String -> SCParser m p -> SCParser m p -> (SourcePos -> SCParser m ()) -> SCParser m p
readAmbiguous prefix expected alternative warner = do
pos <- getPosition
try . lookAhead $ string prefix
-- If the expected parser fails, try the alt.
-- If the alt fails, run the expected one again for the errors.
try expected <|> tryAlt pos <|> expected
try expected <|> try (withAlt pos) <|> expected
where
tryAlt pos = do
t <- try alternative
parseNoteAt pos WarningC 1102 $
"Shells differ in parsing ambiguous " ++ prefix ++ "(((. Use spaces: " ++ prefix ++ "( (( ."
withAlt pos = do
t <- forgetOnFailure alternative
warner pos
return t
prop_readDollarBraceCommandExpansion1 = isOk readDollarBraceCommandExpansion "${ ls; }"
prop_readDollarBraceCommandExpansion2 = isOk readDollarBraceCommandExpansion "${\nls\n}"
readDollarBraceCommandExpansion = called "ksh ${ ..; } command expansion" $ do
@@ -1461,40 +1582,32 @@ prop_readHereDoc8 = isOk readScript "cat <<foo>>bar\netc\nfoo"
prop_readHereDoc9 = isOk readScript "if true; then cat << foo; fi\nbar\nfoo\n"
prop_readHereDoc10= isOk readScript "if true; then cat << foo << bar; fi\nfoo\nbar\n"
prop_readHereDoc11= isOk readScript "cat << foo $(\nfoo\n)lol\nfoo\n"
prop_readHereDoc12= isOk readScript "cat << foo|cat\nbar\nfoo"
readHereDoc = called "here document" $ do
fid <- getNextId
pos <- getPosition
try $ string "<<"
dashed <- (char '-' >> return Dashed) <|> return Undashed
tokenPosition <- getPosition
sp <- spacing
optional $ do
try . lookAhead $ char '('
let message = "Shells are space sensitive. Use '< <(cmd)', not '<<" ++ sp ++ "(cmd)'."
parseProblemAt pos ErrorC 1038 message
hid <- getNextId
(quoted, endToken) <-
liftM (\ x -> (Quoted, stripLiteral x)) readDoubleQuotedLiteral
<|> liftM (\ x -> (Quoted, x)) readSingleQuotedLiteral
<|> (readToken >>= (\x -> return (Unquoted, x)))
(quoted, endToken) <- readToken
-- add empty tokens for now, read the rest in readPendingHereDocs
let doc = T_HereDoc hid dashed quoted endToken []
addPendingHereDoc doc
return $ T_FdRedirect fid "" doc
return doc
where
stripLiteral (T_Literal _ x) = x
stripLiteral (T_SingleQuoted _ x) = x
readToken =
liftM concat $ many1 (escaped <|> quoted <|> normal)
where
quoted = liftM stripLiteral readDoubleQuotedLiteral <|> readSingleQuotedLiteral
normal = anyChar `reluctantlyTill1` (whitespace <|> oneOf "<>;&)'\"\\")
escaped = do -- surely the user must be doing something wrong at this point
char '\\'
c <- anyChar
return [c]
quotes = "\"'\\"
-- Fun fact: bash considers << foo"" quoted, but not << <("foo").
-- Instead of replicating this, just read a token and strip quotes.
readToken = do
str <- readStringForParser readNormalWord
return (if any (`elem` quotes) str then Quoted else Unquoted,
filter (not . (`elem` quotes)) str)
readPendingHereDocs = do
@@ -1552,7 +1665,13 @@ readPendingHereDocs = do
readFilename = readNormalWord
readIoFileOp = choice [g_LESSAND, g_GREATAND, g_DGREAT, g_LESSGREAT, g_CLOBBER, redirToken '<' T_Less, redirToken '>' T_Greater ]
readIoFileOp = choice [g_DGREAT, g_LESSGREAT, g_GREATAND, g_LESSAND, g_CLOBBER, redirToken '<' T_Less, redirToken '>' T_Greater ]
readIoDuplicate = try $ do
id <- getNextId
op <- g_GREATAND <|> g_LESSAND
target <- readIoVariable <|> many1 digit <|> string "-"
return $ T_IoDuplicate id op target
prop_readIoFile = isOk readIoFile ">> \"$(date +%YYmmDD)\""
readIoFile = called "redirection" $ do
@@ -1560,35 +1679,31 @@ readIoFile = called "redirection" $ do
op <- readIoFileOp
spacing
file <- readFilename
return $ T_FdRedirect id "" $ T_IoFile id op file
return $ T_IoFile id op file
readIoVariable = try $ do
char '{'
x <- readVariableName
char '}'
lookAhead readIoFileOp
return $ "{" ++ x ++ "}"
readIoNumber = try $ do
x <- many1 digit <|> string "&"
lookAhead readIoFileOp
readIoSource = try $ do
x <- string "&" <|> readIoVariable <|> many digit
lookAhead $ void readIoFileOp <|> void (string "<<")
return x
prop_readIoNumberRedirect = isOk readIoNumberRedirect "3>&2"
prop_readIoNumberRedirect2 = isOk readIoNumberRedirect "2> lol"
prop_readIoNumberRedirect3 = isOk readIoNumberRedirect "4>&-"
prop_readIoNumberRedirect4 = isOk readIoNumberRedirect "&> lol"
prop_readIoNumberRedirect5 = isOk readIoNumberRedirect "{foo}>&2"
prop_readIoNumberRedirect6 = isOk readIoNumberRedirect "{foo}<&-"
readIoNumberRedirect = do
prop_readIoRedirect = isOk readIoRedirect "3>&2"
prop_readIoRedirect2 = isOk readIoRedirect "2> lol"
prop_readIoRedirect3 = isOk readIoRedirect "4>&-"
prop_readIoRedirect4 = isOk readIoRedirect "&> lol"
prop_readIoRedirect5 = isOk readIoRedirect "{foo}>&2"
prop_readIoRedirect6 = isOk readIoRedirect "{foo}<&-"
readIoRedirect = do
id <- getNextId
n <- readIoVariable <|> readIoNumber
op <- readHereString <|> readHereDoc <|> readIoFile
let actualOp = case op of T_FdRedirect _ "" x -> x
n <- readIoSource
redir <- readHereString <|> readHereDoc <|> readIoDuplicate <|> readIoFile
spacing
return $ T_FdRedirect id n actualOp
readIoRedirect = choice [ readIoNumberRedirect, readHereString, readHereDoc, readIoFile ] `thenSkip` spacing
return $ T_FdRedirect id n redir
readRedirectList = many1 readIoRedirect
@@ -1599,18 +1714,26 @@ readHereString = called "here string" $ do
spacing
id2 <- getNextId
word <- readNormalWord
return $ T_FdRedirect id "" $ T_HereString id2 word
return $ T_HereString id2 word
readNewlineList = many1 ((linefeed <|> carriageReturn) `thenSkip` spacing)
readLineBreak = optional readNewlineList
prop_readSeparator1 = isWarning readScript "a &; b"
prop_readSeparator2 = isOk readScript "a & b"
prop_readSeparator3 = isWarning readScript "a &amp; b"
prop_readSeparator4 = isWarning readScript "a &gt; file; b"
readSeparatorOp = do
notFollowedBy2 (void g_AND_IF <|> void readCaseSeparator)
notFollowedBy2 (string "&>")
f <- try (do
pos <- getPosition
char '&'
optional $ do
s <- lookAhead . choice . map (try . string) $
["amp;", "gt;", "lt;"]
parseProblemAt pos ErrorC 1109 "This is an unquoted HTML entity. Replace with corresponding character."
spacing
pos <- getPosition
char ';'
@@ -1878,7 +2001,7 @@ readElifPart = called "elif clause" $ do
pos <- getPosition
correctElif <- elif
unless correctElif $
parseProblemAt pos ErrorC 1075 "Use 'elif' instead of 'else if'."
parseProblemAt pos ErrorC 1075 "Use 'elif' instead of 'else if' (or put 'if' on new line if nesting)."
allspacing
condition <- readTerm
@@ -2176,8 +2299,8 @@ readCompoundCommand = do
id <- getNextId
cmd <- choice [
readBraceGroup,
readTripleParenthesis "" readArithmeticExpression readSubshell,
readArithmeticExpression,
readAmbiguous "((" readArithmeticExpression readSubshell (\pos ->
parseNoteAt pos WarningC 1105 "Shells disambiguate (( differently or not at all. For subshell, add spaces around ( . For ((, fix parsing errors."),
readSubshell,
readCondition,
readWhileClause,
@@ -2205,7 +2328,7 @@ readCompoundListOrEmpty = do
readCmdPrefix = many1 (readIoRedirect <|> readAssignmentWord)
readCmdSuffix = many1 (readIoRedirect <|> readCmdWord)
readModifierSuffix = many1 (readIoRedirect <|> readAssignmentWord <|> readCmdWord)
readModifierSuffix = many1 (readIoRedirect <|> readWellFormedAssignment <|> readCmdWord)
readTimeSuffix = do
flags <- many readFlag
pipeline <- readPipeline
@@ -2251,7 +2374,7 @@ readEvalSuffix = many1 (readIoRedirect <|> readCmdWord <|> evalFallback)
-- Get whatever a parser would parse as a string
readStringForParser parser = do
pos <- lookAhead (parser >> getPosition)
pos <- inSeparateContext $ lookAhead (parser >> getPosition)
readUntil pos
where
readUntil endPos = anyChar `reluctantlyTill` (getPosition >>= guard . (== endPos))
@@ -2271,14 +2394,18 @@ prop_readAssignmentWord9c= isOk readAssignmentWord "foo= #bar"
prop_readAssignmentWord10= isWarning readAssignmentWord "foo$n=42"
prop_readAssignmentWord11= isOk readAssignmentWord "foo=([a]=b [c] [d]= [e f )"
prop_readAssignmentWord12= isOk readAssignmentWord "a[b <<= 3 + c]='thing'"
readAssignmentWord = try $ do
readAssignmentWord = readAssignmentWordExt True
readWellFormedAssignment = readAssignmentWordExt False
readAssignmentWordExt lenient = try $ do
id <- getNextId
pos <- getPosition
optional (char '$' >> parseNote ErrorC 1066 "Don't use $ on the left side of assignments.")
when lenient $
optional (char '$' >> parseNote ErrorC 1066 "Don't use $ on the left side of assignments.")
variable <- readVariableName
optional (readNormalDollar >> parseNoteAt pos ErrorC
when lenient $
optional (readNormalDollar >> parseNoteAt pos ErrorC
1067 "For indirection, use (associative) arrays or 'read \"var$n\" <<< \"value\"'")
index <- optionMaybe readArrayIndex
indices <- many readArrayIndex
hasLeftSpace <- liftM (not . null) spacing
pos <- getPosition
op <- readAssignmentOp
@@ -2290,13 +2417,13 @@ readAssignmentWord = try $ do
parseNoteAt pos WarningC 1007
"Remove space after = if trying to assign a value (for empty string, use var='' ... )."
value <- readEmptyLiteral
return $ T_Assignment id op variable index value
return $ T_Assignment id op variable indices value
else do
when (hasLeftSpace || hasRightSpace) $
parseNoteAt pos ErrorC 1068 "Don't put spaces around the = in assignments."
value <- readArray <|> readNormalWord
spacing
return $ T_Assignment id op variable index value
return $ T_Assignment id op variable indices value
where
readAssignmentOp = do
pos <- getPosition
@@ -2316,12 +2443,14 @@ readAssignmentWord = try $ do
return $ T_Literal id ""
readArrayIndex = do
id <- getNextId
char '['
optional space
x <- readArithmeticContents
pos <- getPosition
str <- readStringForParser readIndexSpan
char ']'
return x
return $ T_UnparsedIndex id pos str
readArray :: Monad m => SCParser m Token
readArray = called "array assignment" $ do
id <- getNextId
char '('
@@ -2334,7 +2463,7 @@ readArray = called "array assignment" $ do
readIndexed = do
id <- getNextId
index <- try $ do
x <- readArrayIndex
x <- many1 readArrayIndex
char '='
return x
value <- readNormalWord <|> nothing
@@ -2477,12 +2606,12 @@ verifyEof = eof <|> choice [
try (lookAhead p)
action
prop_readScript1 = isOk readScript "#!/bin/bash\necho hello world\n"
prop_readScript2 = isWarning readScript "#!/bin/bash\r\necho hello world\n"
prop_readScript3 = isWarning readScript "#!/bin/bash\necho hello\xA0world"
prop_readScript4 = isWarning readScript "#!/usr/bin/perl\nfoo=("
prop_readScript5 = isOk readScript "#!/bin/bash\n#This is an empty script\n\n"
readScript = do
prop_readScript1 = isOk readScriptFile "#!/bin/bash\necho hello world\n"
prop_readScript2 = isWarning readScriptFile "#!/bin/bash\r\necho hello world\n"
prop_readScript3 = isWarning readScriptFile "#!/bin/bash\necho hello\xA0world"
prop_readScript4 = isWarning readScriptFile "#!/usr/bin/perl\nfoo=("
prop_readScript5 = isOk readScriptFile "#!/bin/bash\n#This is an empty script\n\n"
readScriptFile = do
id <- getNextId
pos <- getPosition
optional $ do
@@ -2493,11 +2622,13 @@ readScript = do
verifyShell pos (getShell sb)
if isValidShell (getShell sb) /= Just False
then do
allspacing
annotationId <- getNextId
annotations <- readAnnotations
commands <- withAnnotations annotations readCompoundListOrEmpty
verifyEof
return $ T_Annotation annotationId annotations $ T_Script id sb commands
let script = T_Annotation annotationId annotations $ T_Script id sb commands
reparseIndices script
else do
many anyChar
return $ T_Script id sb []
@@ -2516,8 +2647,8 @@ readScript = do
verifyShell pos s =
case isValidShell s of
Just True -> return ()
Just False -> parseProblemAt pos ErrorC 1071 "ShellCheck only supports sh/bash/ksh scripts. Sorry!"
Nothing -> parseProblemAt pos InfoC 1008 "This shebang was unrecognized. Note that ShellCheck only handles sh/bash/ksh."
Just False -> parseProblemAt pos ErrorC 1071 "ShellCheck only supports sh/bash/dash/ksh scripts. Sorry!"
Nothing -> parseProblemAt pos InfoC 1008 "This shebang was unrecognized. Note that ShellCheck only handles sh/bash/dash/ksh."
isValidShell s =
let good = s == "" || any (`isPrefixOf` s) goodShells
@@ -2549,6 +2680,9 @@ readScript = do
readUtf8Bom = called "Byte Order Mark" $ string "\xFEFF"
readScript = do
script <- readScriptFile
reparseIndices script
isWarning p s = parsesCleanly p s == Just False
isOk p s = parsesCleanly p s == Just True
@@ -2571,13 +2705,15 @@ parseWithNotes parser = do
state <- getState
return (item, state)
compareNotes (ParseNote pos1 level1 _ s1) (ParseNote pos2 level2 _ s2) = compare (pos1, level1) (pos2, level2)
compareNotes (ParseNote pos1 pos1' level1 _ s1) (ParseNote pos2 pos2' level2 _ s2) = compare (pos1, pos1', level1) (pos2, pos2', level2)
sortNotes = sortBy compareNotes
makeErrorFor parsecError =
ParseNote (errorPos parsecError) ErrorC 1072 $
ParseNote pos pos ErrorC 1072 $
getStringFromParsec $ errorMessages parsecError
where
pos = errorPos parsecError
getStringFromParsec errors =
case map f errors of
@@ -2630,11 +2766,46 @@ parseShell sys name contents = do
isName (ContextName _ _) = True
isName _ = False
notesForContext list = zipWith ($) [first, second] $ filter isName list
first (ContextName pos str) = ParseNote pos ErrorC 1073 $
first (ContextName pos str) = ParseNote pos pos ErrorC 1073 $
"Couldn't parse this " ++ str ++ "."
second (ContextName pos str) = ParseNote pos InfoC 1009 $
second (ContextName pos str) = ParseNote pos pos InfoC 1009 $
"The mentioned parser error was in this " ++ str ++ "."
-- Go over all T_UnparsedIndex and reparse them as either arithmetic or text
-- depending on declare -A statements.
reparseIndices root =
analyze blank blank f root
where
associative = getAssociativeArrays root
isAssociative s = s `elem` associative
f (T_Assignment id mode name indices value) = do
newIndices <- mapM (fixAssignmentIndex name) indices
newValue <- case value of
(T_Array id2 words) -> do
newWords <- mapM (fixIndexElement name) words
return $ T_Array id2 newWords
x -> return x
return $ T_Assignment id mode name newIndices newValue
f t = return t
fixIndexElement name word =
case word of
T_IndexedElement id indices value -> do
new <- mapM (fixAssignmentIndex name) indices
return $ T_IndexedElement id new value
otherwise -> return word
fixAssignmentIndex name word =
case word of
T_UnparsedIndex id pos src -> do
parsed name pos src
otherwise -> return word
parsed name pos src =
if isAssociative name
then subParse pos (called "associative array index" $ readIndexSpan) src
else subParse pos (called "arithmetic array index expression" $ optional space >> readArithmeticContents) src
reattachHereDocs root map =
doTransform f root
where
@@ -2644,8 +2815,8 @@ reattachHereDocs root map =
f t = t
toPositionedComment :: ParseNote -> PositionedComment
toPositionedComment (ParseNote pos severity code message) =
PositionedComment (posToPos pos) $ Comment severity code message
toPositionedComment (ParseNote start end severity code message) =
PositionedComment (posToPos start) (posToPos end) $ Comment severity code message
posToPos :: SourcePos -> Position
posToPos sp = Position {

10
nextnumber Executable file
View File

@@ -0,0 +1,10 @@
#!/bin/bash
# TODO: Find a less trashy way to get the next available error code
shopt -s globstar
for i in 1 2
do
last=$(grep -hv "^prop" **/*.hs | grep -Ewo "$i[0-9]{3}" | sort -n | tail -n 1)
echo "Next ${i}xxx: $((last+1))"
done

View File

@@ -9,6 +9,7 @@
,ShellCheck.Parser.runTests
,ShellCheck.Checker.runTests
,ShellCheck.Checks.Commands.runTests
,ShellCheck.Checks.ShellSupport.runTests
,ShellCheck.AnalyzerLib.runTests
]' | tr -d '\n' | cabal repl 2>&1 | tee /dev/stderr)
if [[ $var == *$'\nTrue'* ]]

View File

@@ -175,6 +175,15 @@ ShellCheck uses the follow exit codes:
+ 3: ShellCheck was invoked with bad syntax (e.g. unknown flag).
+ 4: ShellCheck was invoked with bad options (e.g. unknown formatter).
# LOCALE
This version of ShellCheck is only available in English. All files are
leniently decoded as UTF-8, with a fallback of ISO-8859-1 for invalid
sequences. `LC_CTYPE` is respected for output, and defaults to UTF-8 for
locales where encoding is unspecified (such as the `C` locale).
Windows users seeing `commitBuffer: invalid argument (invalid character)`
should set their terminal to use UTF-8 with `chcp 65001`.
# AUTHOR
ShellCheck is written and maintained by Vidar Holen.

View File

@@ -31,6 +31,7 @@ import qualified ShellCheck.Formatter.TTY
import Control.Exception
import Control.Monad
import Control.Monad.Except
import Data.Bits
import Data.Char
import Data.Functor
import Data.Either
@@ -288,14 +289,47 @@ ioInterface options files = do
fallback path _ = return path
inputFile file = do
contents <-
handle <-
if file == "-"
then getContents
else readFile file
then return stdin
else openBinaryFile file ReadMode
hSetBinaryMode handle True
contents <- decodeString <$> hGetContents handle -- closes handle
seq (length contents) $
return contents
-- Decode a char8 string into a utf8 string, with fallback on
-- ISO-8859-1. This avoids depending on additional libraries.
decodeString = decode
where
decode [] = []
decode (c:rest) | isAscii c = c : decode rest
decode (c:rest) =
let num = (fromIntegral $ ord c) :: Int
next = case num of
_ | num >= 0xF8 -> Nothing
| num >= 0xF0 -> construct (num .&. 0x07) 3 rest
| num >= 0xE0 -> construct (num .&. 0x0F) 2 rest
| num >= 0xC0 -> construct (num .&. 0x1F) 1 rest
| True -> Nothing
in
case next of
Just (n, remainder) -> chr n : decode remainder
Nothing -> c : decode rest
construct x 0 rest = do
guard $ x <= 0x10FFFF
return (x, rest)
construct x n (c:rest) =
let num = (fromIntegral $ ord c) :: Int in
if num >= 0x80 && num <= 0xBF
then construct ((x `shiftL` 6) .|. (num .&. 0x3f)) (n-1) rest
else Nothing
construct _ _ _ = Nothing
verifyFiles files =
when (null files) $ do
printErr "No files specified.\n"

View File

@@ -7,12 +7,14 @@ import qualified ShellCheck.Analytics
import qualified ShellCheck.AnalyzerLib
import qualified ShellCheck.Parser
import qualified ShellCheck.Checks.Commands
import qualified ShellCheck.Checks.ShellSupport
main = do
putStrLn "Running ShellCheck tests..."
results <- sequence [
ShellCheck.Checker.runTests,
ShellCheck.Checks.Commands.runTests,
ShellCheck.Checks.ShellSupport.runTests,
ShellCheck.Analytics.runTests,
ShellCheck.AnalyzerLib.runTests,
ShellCheck.Parser.runTests