Add spellchecker script and dictionary of valid words for it

Fixes rust-lang#118

README.md

@@ -91,3 +91,11 @@ $ dot dot/trpl04-01.dot -Tsvg > src/img/trpl04-01.svg
 In the generated SVG, remove the width and the height attributes from the `svg`
 element and set the `viewBox` attribute to `0.00 0.00 1000.00 1000.00` or other
 values that don't cut off the image.
+
+## Spellchecking
+
+To scan source file for spell errors, you can use `spellcheck.sh` script. It
+needs a dictionary of valid words, which is provided in `dictionary.txt`. If
+script produces a false positive (say, you used word `BTreeMap` which script
+considers invalid) you need to add this word to dictionary file (in any place
+in the file, but you can keep sorted order for consistency).

dictionary.txt

@@ -0,0 +1,204 @@
+personal_ws-1.1 en 0 utf-8
+abcabcabc
+abcd
+Addr
+alignof
+APIs
+aren
+backtrace
+BACKTRACE
+benchmarking
+bitand
+BitAnd
+bitor
+BitOr
+bitwise
+Bitwise
+bitxor
+BitXor
+bool
+boolean
+booleans
+Bors
+BuildHasher
+Cagain
+callsite
+CamelCase
+ChangeColor
+ChangeColorMessage
+chXX
+chYY
+config
+Config
+const
+copyeditor
+couldn
+cratesio
+cryptographically
+CStr
+CString
+ctrl
+Ctrl
+deallocated
+debuginfo
+deref
+Deref
+dereference
+Dereference
+dereferencing
+DerefMut
+destructure
+destructuring
+Destructuring
+didn
+Dobrý
+doccargo
+doccratesio
+doesn
+else's
+encodings
+enum
+Enum
+enums
+enum's
+Enums
+ErrorKind
+Executables
+extern
+FFFF
+filename
+Filename
+filesystem
+Filesystem
+formatter
+gitignore
+grapheme
+Grapheme
+growable
+hardcoded
+hardcoding
+hasher
+HashMap
+Haskell
+hasn
+helloworld
+Hmmm
+Hoare
+Hola
+html
+impl
+init
+instantiation
+InvalidDigit
+ioerror
+iokind
+ioresult
+iostdin
+IpAddr
+IpAddrKind
+irst
+isize
+iter
+judgement
+lang
+libc
+libcore
+libreoffice
+lifecycle
+loopback
+lval
+mathematic
+metaprogramming
+mibbit
+Mibbit
+mkdir
+monomorphization
+Monomorphization
+monomorphized
+MoveMessage
+Mutex
+namespace
+namespaced
+namespaces
+Noooooooooooooooooooooo
+NotFound
+null's
+OCaml
+offsetof
+OptionalFloatingPointNumber
+OptionalNumber
+OsStr
+OsString
+overread
+parameterize
+ParseIntError
+PartialEq
+PartialOrd
+powi
+preprocessing
+Preprocessing
+preprocessor
+println
+priv
+proc
+QuitMessage
+RAII
+randcrate
+READMEs
+rect
+Rectange
+redeclaring
+RefCell
+repr
+runtime
+Rustacean
+Rustaceans
+rustc
+rustdoc
+rustup
+semver
+SemVer
+shouldn
+sizeof
+someproject
+someusername
+SpreadsheetCell
+sqrt
+stackoverflow
+stdin
+Stdin
+stdlib
+struct
+Struct
+structs
+struct's
+Structs
+submodule
+submodules
+suboptimal
+subtree
+That'd
+TODO
+toml
+TOML
+tradeoff
+TrafficLight
+trpl
+typeof
+UFCS
+unary
+Unary
+unoptimized
+unsized
+USERPROFILE
+usize
+UsState
+Versioning
+wasn
+whitespace
+workspace
+workspaces
+Workspaces
+wouldn
+WriteMessage
+yyyy

spellcheck.sh

@@ -0,0 +1,78 @@
+#!/bin/bash
+
+# Checks project markdown files for spell errors
+
+# Notes:
+
+# This script needs dictionary file ($dict_filename) with project-specific
+# valid words. If this file is missing, first invocation of a script generates
+# a file of words considered typos at the moment. User should remove real typos
+# from this file and leave only valid words. When script generates false
+# positive after source modification, new valid word should be added
+# to dictionary file.
+
+# Default mode of this script is interactive. Each source file is scanned for
+# typos. aspell opens window, suggesting fixes for each found typo. Original
+# files with errors will be backed up to files with format "filename.md.bak".
+
+# When running in CI, this script should be run in "list" mode (pass "list"
+# as first argument). In this mode script scans all files and reports found
+# errors. Exit code in this case depends on scan result:
+# 1 if any errors found,
+# 0 if all is clear.
+
+# Script skips words with length less then or equal to 3. This helps to avoid
+# some false positives.
+
+# We can consider skipping source code in markdown files (```code```) to reduce
+# rate of false positives, but then we lose ability to detect typos in code
+# comments/strings etc.
+
+shopt -s nullglob
+
+dict_filename=dictionary.txt
+markdown_sources=(./src/*.md)
+mode="check"
+
+# aspell repeatedly modifies personal dictionary for some purpose,
+# so we should use a copy of our dictionary
+dict_path="/tmp/$dict_filename"
+
+if [[ "$1" == "list" ]]; then
+    mode="list"
+fi
+
+if [[ ! -f "$dict_filename" ]]; then
+    # Pre-check mode: generates dictionary of words aspell consider typos.
+    # After user validates that this file contains only valid words, we can
+    # look for typos using this dictionary and some default aspell dictionary.
+    echo "Scanning files to generate dictionary file '$dict_filename'."
+    echo "Please check it doesn't contain any spellings for correct results."
+
+    echo "personal_ws-1.1 en 0 utf-8" > "$dict_filename"
+    cat "${markdown_sources[@]}" | aspell list | sort -u >> "$dict_filename"
+elif [[ "$mode" == "list" ]]; then
+    # List (default) mode: scan all files, report errors
+    cp "$dict_filename" "$dict_path"
+    declare -i retval=0
+
+    for fname in "${markdown_sources[@]}"; do
+        command=$(aspell --ignore 3 --personal="$dict_path" "$mode" < "$fname")
+        if [[ -n "$command" ]]; then
+            for error in $command; do
+                # TODO: Find more correct way to get line number
+                # (ideally from aspell). Now it can make some false positives,
+                # because it is just a grep
+                grep --with-filename --line-number --color=always "$error" "$fname"
+            done
+            retval=1
+        fi
+    done
+    exit "$retval"
+elif [[ "$mode" == "check" ]]; then
+    # Interactive mode: fix typos
+    cp "$dict_filename" "$dict_path"
+    for fname in "${markdown_sources[@]}"; do
+        aspell --ignore 3 --personal="$dict_path" "$mode" "$fname"
+    done
+fi