git-gui / lib / spellcheck.tclon commit Merge branch 'md/url-parse-harden' (f9089e8)
   1# git-gui spellchecking support through ispell/aspell
   2# Copyright (C) 2008 Shawn Pearce
   3
   4class spellcheck {
   5
   6field s_fd      {} ; # pipe to ispell/aspell
   7field s_version {} ; # ispell/aspell version string
   8field s_lang    {} ; # current language code
   9field s_prog aspell; # are we actually old ispell?
  10field s_failed   0 ; # is $s_prog bogus and not working?
  11
  12field w_text      ; # text widget we are spelling
  13field w_menu      ; # context menu for the widget
  14field s_menuidx 0 ; # last index of insertion into $w_menu
  15
  16field s_i           {} ; # timer registration for _run callbacks
  17field s_clear        0 ; # did we erase misspelled tags yet?
  18field s_seen    [list] ; # lines last seen from $w_text in _run
  19field s_checked [list] ; # lines already checked
  20field s_pending [list] ; # [$line $data] sent to ispell/aspell
  21field s_suggest        ; # array, list of suggestions, keyed by misspelling
  22
  23constructor init {pipe_fd ui_text ui_menu} {
  24        set w_text $ui_text
  25        set w_menu $ui_menu
  26        array unset s_suggest
  27
  28        bind_button3 $w_text [cb _popup_suggest %X %Y @%x,%y]
  29        _connect $this $pipe_fd
  30        return $this
  31}
  32
  33method _connect {pipe_fd} {
  34        fconfigure $pipe_fd \
  35                -encoding utf-8 \
  36                -eofchar {} \
  37                -translation lf
  38
  39        if {[gets $pipe_fd s_version] <= 0} {
  40                if {[catch {close $pipe_fd} err]} {
  41
  42                        # Eh?  Is this actually ispell choking on aspell options?
  43                        #
  44                        if {$s_prog eq {aspell}
  45                                && [regexp -nocase {^Usage: } $err]
  46                                && ![catch {
  47                                                set pipe_fd [open [list | $s_prog -v] r]
  48                                                gets $pipe_fd s_version
  49                                                close $pipe_fd
  50                                }]
  51                                && $s_version ne {}} {
  52                                if {{@(#) } eq [string range $s_version 0 4]} {
  53                                        set s_version [string range $s_version 5 end]
  54                                }
  55                                set s_failed 1
  56                                error_popup [strcat \
  57                                        [mc "Unsupported spell checker"] \
  58                                        ":\n\n$s_version"]
  59                                set s_version {}
  60                                return
  61                        }
  62
  63                        regsub -nocase {^Error: } $err {} err
  64                        if {$s_fd eq {}} {
  65                                error_popup [strcat [mc "Spell checking is unavailable"] ":\n\n$err"]
  66                        } else {
  67                                error_popup [strcat \
  68                                        [mc "Invalid spell checking configuration"] \
  69                                        ":\n\n$err\n\n" \
  70                                        [mc "Reverting dictionary to %s." $s_lang]]
  71                        }
  72                } else {
  73                        error_popup [mc "Spell checker silently failed on startup"]
  74                }
  75                return
  76        }
  77
  78        if {{@(#) } ne [string range $s_version 0 4]} {
  79                catch {close $pipe_fd}
  80                error_popup [strcat [mc "Unrecognized spell checker"] ":\n\n$s_version"]
  81                return
  82        }
  83        set s_version [string range [string trim $s_version] 5 end]
  84        regexp \
  85                {International Ispell Version .* \(but really (Aspell .*?)\)$} \
  86                $s_version _junk s_version
  87        regexp {^Aspell (\d)+\.(\d+)} $s_version _junk major minor
  88
  89        puts $pipe_fd !             ; # enable terse mode
  90
  91        # fetch the language
  92        if {$major > 0 || ($major == 0 && $minor >= 60)} {
  93                puts $pipe_fd {$$cr master}
  94                flush $pipe_fd
  95                gets $pipe_fd s_lang
  96                regexp {[/\\]([^/\\]+)\.[^\.]+$} $s_lang _ s_lang
  97        } else {
  98                set s_lang {}
  99        }
 100
 101        if {$::default_config(gui.spellingdictionary) eq {}
 102         && [get_config gui.spellingdictionary] eq {}} {
 103                set ::default_config(gui.spellingdictionary) $s_lang
 104        }
 105
 106        if {$s_fd ne {}} {
 107                catch {close $s_fd}
 108        }
 109        set s_fd $pipe_fd
 110
 111        fconfigure $s_fd -blocking 0
 112        fileevent $s_fd readable [cb _read]
 113
 114        $w_text tag conf misspelled \
 115                -foreground red \
 116                -underline 1
 117
 118        array unset s_suggest
 119        set s_seen    [list]
 120        set s_checked [list]
 121        set s_pending [list]
 122        _run $this
 123}
 124
 125method lang {{n {}}} {
 126        if {$n ne {} && $s_lang ne $n && !$s_failed} {
 127                set spell_cmd [list |]
 128                lappend spell_cmd aspell
 129                lappend spell_cmd --master=$n
 130                lappend spell_cmd --mode=none
 131                lappend spell_cmd --encoding=UTF-8
 132                lappend spell_cmd pipe
 133                _connect $this [open $spell_cmd r+]
 134        }
 135        return $s_lang
 136}
 137
 138method version {} {
 139        if {$s_version ne {}} {
 140                return "$s_version, $s_lang"
 141        }
 142        return {}
 143}
 144
 145method stop {} {
 146        while {$s_menuidx > 0} {
 147                $w_menu delete 0
 148                incr s_menuidx -1
 149        }
 150        $w_text tag delete misspelled
 151
 152        catch {close $s_fd}
 153        catch {after cancel $s_i}
 154        set s_fd {}
 155        set s_i {}
 156        set s_lang {}
 157}
 158
 159method _popup_suggest {X Y pos} {
 160        while {$s_menuidx > 0} {
 161                $w_menu delete 0
 162                incr s_menuidx -1
 163        }
 164
 165        set b_loc [$w_text index "$pos wordstart"]
 166        set e_loc [_wordend $this $b_loc]
 167        set orig  [$w_text get $b_loc $e_loc]
 168        set tags  [$w_text tag names $b_loc]
 169
 170        if {[lsearch -exact $tags misspelled] >= 0} {
 171                if {[info exists s_suggest($orig)]} {
 172                        set cnt 0
 173                        foreach s $s_suggest($orig) {
 174                                if {$cnt < 5} {
 175                                        $w_menu insert $s_menuidx command \
 176                                                -label $s \
 177                                                -command [cb _replace $b_loc $e_loc $s]
 178                                        incr s_menuidx
 179                                        incr cnt
 180                                } else {
 181                                        break
 182                                }
 183                        }
 184                } else {
 185                        $w_menu insert $s_menuidx command \
 186                                -label [mc "No Suggestions"] \
 187                                -state disabled
 188                        incr s_menuidx
 189                }
 190                $w_menu insert $s_menuidx separator
 191                incr s_menuidx
 192        }
 193
 194        $w_text mark set saved-insert insert
 195        tk_popup $w_menu $X $Y
 196}
 197
 198method _replace {b_loc e_loc word} {
 199        $w_text configure -autoseparators 0
 200        $w_text edit separator
 201
 202        $w_text delete $b_loc $e_loc
 203        $w_text insert $b_loc $word
 204
 205        $w_text edit separator
 206        $w_text configure -autoseparators 1
 207        $w_text mark set insert saved-insert
 208}
 209
 210method _restart_timer {} {
 211        set s_i [after 300 [cb _run]]
 212}
 213
 214proc _match_length {max_line arr_name} {
 215        upvar $arr_name a
 216
 217        if {[llength $a] > $max_line} {
 218                set a [lrange $a 0 $max_line]
 219        }
 220        while {[llength $a] <= $max_line} {
 221                lappend a {}
 222        }
 223}
 224
 225method _wordend {pos} {
 226        set pos  [$w_text index "$pos wordend"]
 227        set tags [$w_text tag names $pos]
 228        while {[lsearch -exact $tags misspelled] >= 0} {
 229                set pos  [$w_text index "$pos +1c"]
 230                set tags [$w_text tag names $pos]
 231        }
 232        return $pos
 233}
 234
 235method _run {} {
 236        set cur_pos  [$w_text index {insert -1c}]
 237        set cur_line [lindex [split $cur_pos .] 0]
 238        set max_line [lindex [split [$w_text index end] .] 0]
 239        _match_length $max_line s_seen
 240        _match_length $max_line s_checked
 241
 242        # Nothing in the message buffer?  Nothing to spellcheck.
 243        #
 244        if {$cur_line == 1
 245         && $max_line == 2
 246         && [$w_text get 1.0 end] eq "\n"} {
 247                array unset s_suggest
 248                _restart_timer $this
 249                return
 250        }
 251
 252        set active 0
 253        for {set n 1} {$n <= $max_line} {incr n} {
 254                set s [$w_text get "$n.0" "$n.end"]
 255
 256                # Don't spellcheck the current line unless we are at
 257                # a word boundary.  The user might be typing on it.
 258                #
 259                if {$n == $cur_line
 260                 && ![regexp {^\W$} [$w_text get $cur_pos insert]]} {
 261
 262                        # If the current word is misspelled remove the tag
 263                        # but force a spellcheck later.
 264                        #
 265                        set tags [$w_text tag names $cur_pos]
 266                        if {[lsearch -exact $tags misspelled] >= 0} {
 267                                $w_text tag remove misspelled \
 268                                        "$cur_pos wordstart" \
 269                                        [_wordend $this $cur_pos]
 270                                lset s_seen    $n $s
 271                                lset s_checked $n {}
 272                        }
 273
 274                        continue
 275                }
 276
 277                if {[lindex $s_seen    $n] eq $s
 278                 && [lindex $s_checked $n] ne $s} {
 279                        # Don't send empty lines to Aspell it doesn't check them.
 280                        #
 281                        if {$s eq {}} {
 282                                lset s_checked $n $s
 283                                continue
 284                        }
 285
 286                        # Don't send typical s-b-o lines as the emails are
 287                        # almost always misspelled according to Aspell.
 288                        #
 289                        if {[regexp -nocase {^[a-z-]+-by:.*<.*@.*>$} $s]} {
 290                                $w_text tag remove misspelled "$n.0" "$n.end"
 291                                lset s_checked $n $s
 292                                continue
 293                        }
 294
 295                        puts $s_fd ^$s
 296                        lappend s_pending [list $n $s]
 297                        set active 1
 298                } else {
 299                        # Delay until another idle loop to make sure we don't
 300                        # spellcheck lines the user is actively changing.
 301                        #
 302                        lset s_seen $n $s
 303                }
 304        }
 305
 306        if {$active} {
 307                set s_clear 1
 308                flush $s_fd
 309        } else {
 310                _restart_timer $this
 311        }
 312}
 313
 314method _read {} {
 315        while {[gets $s_fd line] >= 0} {
 316                set lineno [lindex $s_pending 0 0]
 317                set line [string trim $line]
 318
 319                if {$s_clear} {
 320                        $w_text tag remove misspelled "$lineno.0" "$lineno.end"
 321                        set s_clear 0
 322                }
 323
 324                if {$line eq {}} {
 325                        lset s_checked $lineno [lindex $s_pending 0 1]
 326                        set s_pending [lrange $s_pending 1 end]
 327                        set s_clear 1
 328                        continue
 329                }
 330
 331                set sugg [list]
 332                switch -- [string range $line 0 1] {
 333                {& } {
 334                        set line [split [string range $line 2 end] :]
 335                        set info [split [lindex $line 0] { }]
 336                        set orig [lindex $info 0]
 337                        set offs [lindex $info 2]
 338                        foreach s [split [lindex $line 1] ,] {
 339                                lappend sugg [string range $s 1 end]
 340                        }
 341                }
 342                {# } {
 343                        set info [split [string range $line 2 end] { }]
 344                        set orig [lindex $info 0]
 345                        set offs [lindex $info 1]
 346                }
 347                default {
 348                        puts stderr "<spell> $line"
 349                        continue
 350                }
 351                }
 352
 353                incr offs -1
 354                set b_loc "$lineno.$offs"
 355                set e_loc [$w_text index "$lineno.$offs wordend"]
 356                set curr [$w_text get $b_loc $e_loc]
 357
 358                # At least for English curr = "bob", orig = "bob's"
 359                # so Tk didn't include the 's but Aspell did.  We
 360                # try to round out the word.
 361                #
 362                while {$curr ne $orig
 363                 && [string equal -length [string length $curr] $curr $orig]} {
 364                        set n_loc  [$w_text index "$e_loc +1c"]
 365                        set n_curr [$w_text get $b_loc $n_loc]
 366                        if {$n_curr eq $curr} {
 367                                break
 368                        }
 369                        set curr  $n_curr
 370                        set e_loc $n_loc
 371                }
 372
 373                if {$curr eq $orig} {
 374                        $w_text tag add misspelled $b_loc $e_loc
 375                        if {[llength $sugg] > 0} {
 376                                set s_suggest($orig) $sugg
 377                        } else {
 378                                unset -nocomplain s_suggest($orig)
 379                        }
 380                } else {
 381                        unset -nocomplain s_suggest($orig)
 382                }
 383        }
 384
 385        fconfigure $s_fd -block 1
 386        if {[eof $s_fd]} {
 387                if {![catch {close $s_fd} err]} {
 388                        set err [mc "Unexpected EOF from spell checker"]
 389                }
 390                catch {after cancel $s_i}
 391                $w_text tag remove misspelled 1.0 end
 392                error_popup [strcat [mc "Spell Checker Failed"] "\n\n" $err]
 393                return
 394        }
 395        fconfigure $s_fd -block 0
 396
 397        if {[llength $s_pending] == 0} {
 398                _restart_timer $this
 399        }
 400}
 401
 402proc available_langs {} {
 403        set langs [list]
 404        catch {
 405                set fd [open [list | aspell dump dicts] r]
 406                while {[gets $fd line] >= 0} {
 407                        if {$line eq {}} continue
 408                        lappend langs $line
 409                }
 410                close $fd
 411        }
 412        return $langs
 413}
 414
 415}