Preparations for unicodization

git-svn-id: https://tesseract-ocr.googlecode.com/svn/trunk@43 d0cd1f9f-072b-0410-8dd7-cf729c803f20

tessdata/DangAmbigs

@@ -1,39 +0,0 @@
-m rn
-rn m
-m in
-in m
-d cl
-cl d
-nn rm
-rm nn
-n ri
-ri n
-li h
-lr h
-ii u
-ii n
-ni m
-iii m
-ll H
-I-I H
-vv w
-VV W
-t f
-f t
-a o
-o a
-e c
-c e
-rr n
-E fi
-l< k
-ld ki
-lx h
-xn m
-ux in
-r t
-d tl
-di th
-ur in
-un im
-u a

tessdata/blackText.params

@@ -1,74 +0,0 @@
-# parameters for black text enhancement
-
-
-zeroThres 5     	# 1D histogram (R, G or B) zero threshold.
-                	# Used to identify those parts of the histogram that 
-			# are away from peaks and have returned to zero.
-			# Per intensity resolution unit and hence independent
-			# of the number of histogram bins. 
-			# Default 5
-
-cellSize 48     	# The cell size used for (RGB) histograms.
-                	# Also dictates the total height of the 
-			# swath of image and associated status data used 
-			# during processing.
-			# 
-			#               swathHeight = 4*cellSize
-			# 
-			# These in turn dictate the region over which 
-			# connected component analysis can be performed without 
-			# backtracking to fill or outline components. 
-			# The components themselves can in general exceed the 
-			# height of the swath provided initial backtracking has 
-			# occurred.
-			# Default 48
-
-minSample 200		# Suggested minimum sample size for histograms.
-			# Allows sub sampling if sample size can be exceeded.
-			# default 200
-
-rgbCompThres 10         # Maximum allowed difference in RGB averaged over
-			# a partial or whole text component to allow snapping
-			# to black.
-			# Default 10
-
-blackConnThres 120       # Maximum intensity threshold for pixels to be
-			# considered as part of a text component. 
-			# Default 90
-
-blackSnapThres 70       # Maximum intensity threshold for pixels that are
-			# part of a text component to be snapped to black. 
-			# Default 50
-
-blackCompThres 60       # Maximum intensity for averaged pixel values over
-			# a partial or whole text component to allow snapping
-			# to black.
-			# Default 30
-
-saturateThres 150       # Minimum intensity for text area pixel to be snapped to
-			# white if backEnhance enabled.
-			# default 150
-
-sizeThres 48            # Connected component size threshold.
-                        # When connected components recieve contributions
-			# from more runs than this threshold their stats
-			# can be checked for satisfaction.
-			# If so back filling (to the limit of the swath)
-			# and future forward filling will occur. 
-			# Filling covers snapping and white haloing.
-			# Default 48 (cell size)
-
-blackThres 60           # Maximum intensity for histogram peaks to be defined
-			# as black. 
-			# This is a per intensity resolution unit and hence
-			# independent of the number of histogram bins.
-			# Default 60
-
-whiteThres 100          # Minimum intensity for histogram peaks to be defined 
-			# as white.
-			# This is a per intensity resolution unit and hence
-			# independent of the number of histogram bins.
-			# Default 100
-
-backEnhance False       # Boolean on background enhancement to white
-			# Default True

tessdata/newdiff.asccodes

@@ -1,51 +0,0 @@
-a a 304 0
-a c 300 0
-a g 310 0
-a r 260 0
-b s 000 1
-b u 263 0
-b x 374 0
-b w 272 0
-c p 265 0
-c r 246 0
-c t 277 0
-d g 276 0
-e a 305 0
-E a 334 0
-e g 311 0
-e q 323 0
-g D 343 0
-g m 363 0
-g s 357 0
-h d 362 0
-i a 325 0
-i l 322 0
-l d 055 0
-n s 267 0
-o a 306 0
-o b 000 2
-o l 000 3
-p d 273 0
-p g 327 0
-p m 376 0
-p p 274 0
-r g 247 0
-s b 000 4
-s c 275 0
-s p 000 5
-t b 040 0
-t l 274 0
-T M 264 0
-u l 000 6
-u u 317 0
-v t 000 7
-w t 000 8
-1 e 341 0
-1 h 370 0
-1 q 367 0
-3 e 320 0
-3 q 365 0
-3 s 354 0
-5 e 330 0
-5 s 353 0
-7 e 340 0

tessdata/pffmtable

@@ -1,92 +0,0 @@
-! 27
-# 44
-$ 45
-% 46
-& 46
-' 7
-( 28
-) 27
-* 14
-+ 26
-, 10
-- 9
-. 5
-/ 26
-0 53
-1 28
-2 50
-3 47
-4 34
-5 50
-6 46
-7 37
-8 50
-9 47
-: 10
-; 17
-< 29
-= 29
-> 31
-? 32
-@ 54
-A 45
-B 52
-C 47
-D 57
-E 49
-F 41
-G 58
-H 56
-I 26
-J 33
-K 51
-L 34
-M 55
-N 53
-O 56
-P 44
-Q 60
-R 53
-S 49
-T 34
-U 55
-V 44
-W 49
-X 47
-Y 38
-Z 46
-[ 35
-\ 26
-] 35
-^ 18
-_ 17
-` 8
-a 39
-b 51
-c 36
-d 51
-e 35
-f 29
-g 56
-h 49
-i 24
-j 34
-k 45
-l 27
-m 59
-n 42
-o 43
-p 50
-q 49
-r 24
-s 35
-t 28
-u 43
-v 33
-w 40
-x 37
-y 41
-z 36
-{ 34
-} 34
-~ 18

tessdata/test_matrix

@@ -1,96 +0,0 @@
-  FL RLHT VPOS X N DOT
-  12 1234 1234     ENAB
-! 11 0011 0110 1 0 1001	1 4 28 0
-" 11 1100 0011 0 1 0100	1 1 0 0
-# 11 0011 0110 0 0 0100	5 5 0 40
-$ 11 0011 0110 0 0 0100	4 5 29 44
-% 11 0111 1111 1 0 1011	1 5 0 0  /
-& 11 0011 0110 0 0 0100	4 5 0 0
-' 11 1100 0011 0 0 0100	1 2 0 0
-( 11 0011 0110 0 0 0100	1 5 0 0
-) 11 0011 0110 0 0 0100	1 5 0 0
-* 11 0111 0110 0 0 0100	5 5 0 0
-+ 11 0111 0110 0 0 0100	4 5 0 35
-, 11 1100 1100 0 0 0100	1 2 0 0
-- 11 1000 0110 0 0 0100	1 8 0 0 =_
-. 11 1000 1100 0 0 0100	1 2 0 0 :
-/ 11 0011 0110 0 0 0100	1 9 0 0  %
-0 01 0011 0110 0 0 0100	1 3 0 0 o
-1 01 0011 0110 0 0 0100	1 8 28 0
-2 01 0011 0110 0 0 0100	2 3 0 0
-3 11 0011 0110 0 0 0100	3 3 29 0
-4 11 0011 0110 0 0 0100	3 3 0 48
-5 01 0011 0110 0 0 0100	2 3 0 40
-6 11 0011 0110 0 0 0100	2 3 0 46
-7 01 0011 0110 0 0 0100	1 3 0 37
-8 01 0011 0110 0 0 0100	4 3 0 45
-9 01 0011 0110 0 0 0100	2 3 0 43  g
-: 11 1000 1111 1 0 1011	1 0 0 0
-; 11 1100 1100 1 0 1010	1 0 0 0
-< 11 0011 0110 0 0 0100	1 5 0 36
-= 11 1000 1111 1 0 1011	1 3 0 0 -_
-> 11 0011 0110 0 0 0100	1 5 0 36
-? 11 0011 0110 1 0 1001	1 5 0 43
-@ 11 0011 0110 0 0 0100	2 5 0 0
-A 01 0011 0110 0 0 0100	2 8 28 45
-B 01 0011 0110 0 0 0100	3 8 28 0
-C 01 0011 0110 0 0 0100	1 8 28 0 c
-D 01 0011 0110 0 0 0100	2 8 29 39
-E 01 0011 0110 0 0 0100	2 8 28 38
-F 01 0011 0110 0 0 0100	2 8 28 40
-G 01 0011 0110 0 0 0100	1 8 0 0
-H 01 0011 0110 0 0 0100	2 8 28 36
-I 01 0011 0110 0 0 0100	1 8 28 0
-J 01 0011 0110 0 0 0100	1 8 28 37  j
-K 01 0011 0110 0 0 0100	3 8 28 34
-L 01 0011 0110 0 0 0100	1 8 29 0
-M 01 0011 0110 0 0 0100	2 8 28 0
-N 01 0011 0110 0 0 0100	2 8 28 0
-O 01 0011 0110 0 0 0100	1 8 0 0 o
-P 01 0011 0110 0 0 0100	2 8 28 0  p
-Q 01 0011 0110 0 0 0100	2 8 0 38
-R 01 0011 0110 0 0 0100	3 8 28 40
-S 01 0011 0110 0 0 0100	2 8 29 0 s
-T 01 0011 0110 0 0 0100	2 8 29 35
-U 01 0011 0110 0 0 0100	1 8 28 42  u
-V 01 0011 0110 0 0 0100	1 8 29 0 v
-W 01 0011 0110 0 0 0100	2 8 29 0 w
-X 01 0011 0110 0 0 0100	4 8 29 0 x
-Y 01 0011 0110 0 0 0100	2 8 0 0  y
-Z 01 0011 0110 0 0 0100	2 8 0 0 z
-[ 11 0011 0110 0 0 0100	1 5 29 0
-\ 11 0011 0110 0 0 0100	1 9 0 0
-] 11 0011 0110 0 0 0100	1 5 28 0
-^ 11 1110 0011 0 0 0100	1 5 0 0
-_ 11 1000 1000 0 0 0100	1 10 0 0 -=
-` 11 1100 0011 0 0 0100	1 3 0 0
-a 10 0110 0110 0 0 0100	2 7 31 48
-b 01 0011 0110 0 0 0100	2 7 28 44
-c 10 0110 0110 0 0 0100	1 7 0 0 C
-d 01 0011 0110 0 0 0100	2 7 27 44
-e 10 0110 0110 0 0 0100	2 7 0 0
-f 01 0011 0110 0 0 0100	3 7 0 0
-g 10 0011 1110 0 0 0100	2 7 29 42
-h 01 0011 0110 0 0 0100	2 7 29 42
-i 10 0111 0110 1 0 1010	1 7 28 0
-j 10 0111 0110 1 0 1010	1 7 28 0  J
-k 11 0011 0110 0 0 0100	3 7 28 40
-l 01 0011 0110 0 0 0100	1 10 28 0
-m 10 0110 0110 0 0 0100	2 7 28 0
-n 10 0110 0110 0 0 0100	1 7 28 43
-o 10 0110 0110 0 0 0100	1 7 0 0
-p 10 0111 1110 0 0 0100	2 7 28 41  P
-q 10 0111 1110 0 0 0100	2 7 28 42
-r 10 0110 0110 0 0 0100	1 7 27 0
-s 10 0110 0110 0 0 0100	2 7 29 0 S
-t 11 0011 0110 0 0 0100	3 7 0 47
-u 10 0110 0110 0 0 0100	1 7 28 42  U
-v 10 0110 0110 0 0 0100	1 7 29 35 V
-w 10 0110 0110 0 0 0100	2 7 29 0 W
-x 10 0110 0110 0 0 0100	4 7 29 0 X
-y 10 0111 1110 0 0 0100	2 7 0 0  Y
-z 10 0110 0110 0 0 0100	2 7 0 0 Z
-{ 11 0011 0110 0 0 0100	3 5 0 0
-| 11 0011 0110 0 0 0100	1 5 0 0
-} 11 0011 0110 0 0 0100	3 5 0 0
-~ 11 1100 0011 0 0 0100	2 5 0 0

tessdata/user-words

@@ -1,920 +0,0 @@
-a
-absurdum
-ac
-acres
-actions
-adaption
-adjustments
-aerobes
-affairs
-agents
-Alan
-Albert
-Alberta
-Alfred
-Alice
-Alicia
-alliances
-americas
-analysts
-announcements
-anouncements
-apples
-applications
-apricots
-architectures
-areas
-arguments
-arrangements
-Arthur
-artists
-arts
-aspects
-attitudes
-attractions
-auctions
-aug
-az
-baccalaureat
-backlit
-bags
-Barbara
-Barnabas
-Barry
-beliefs
-benchmarks
-Betty
-bi
-bits
-blades
-bonaventure
-brad
-broadminded
-broadway
-broking
-brows
-Bruce
-bs
-buddha
-buddhism
-buddhist
-buddhists
-buffers
-ca
-caffein
-calculational
-calif
-California
-cam
-cams
-Canadian
-cancelling
-capitulated
-caps
-Carmel
-Carolyn
-Carroll
-cars
-cartridges
-cassette
-casuality
-Catherine
-centre
-centres
-chambermaid
-chapters
-characteristics
-characters
-Charles
-cheesy
-cherokee
-Chicago
-chloride
-Christopher
-Chrysler
-Churchill
-Cicero
-cinema
-cinemas
-Claire
-Clara
-clark
-cleaners
-clients
-cliffs
-clubs
-co
-codirector
-coinsurance
-Columbus
-combinations
-combust
-combustor
-comparisons
-components
-computerised
-computers
-con
-concepts
-conclusions
-connections
-connectors
-consequences
-contemporizing
-continued
-contra
-contractors
-controls
-coprocessor
-corequisite
-corp
-corridors
-corrosive
-costmetology
-counterparts
-cpu
-crops
-cueing
-culturess
-curtis
-customers
-cuts
-cutout
-cyanide
-Czechoslovakia
-dan
-databases
-David
-Davis
-days
-dealership
-Deborah
-debut
-decibles
-declarations
-deductible
-defrayed
-degrees
-deionized
-demobilisation
-densily
-departments
-descriptions
-desensitization
-desktop
-developers
-developments
-devices
-dharma
-diameters
-Dianne
-dicators
-differences
-digitising
-directions
-directorate
-disadvantages
-disassembly
-disclosures
-discos
-discs
-discusing
-disks
-districts
-doe
-dogs
-dominican
-dominicans
-Donald
-dos
-dots
-Douglas
-Douglass
-downsize
-downsized
-drugs
-dumplings
-duns
-eastside
-ecconomic
-ecconomics
-ed
-Edward
-efforts
-Egypt
-eh
-Einstein
-Einsteins
-Elaine
-electrophotography
-elements
-Elizabeth
-Elliot
-emory
-emulsion
-energized
-enquiry
-ent
-enthusiasts
-entrylevel
-environments
-epilepsy
-epistemic
-er
-eric
-erosion
-errors
-estuary
-et
-events
-everyone's
-exactions
-exegesis
-exhilarating
-expenditures
-explanations
-explicably
-expo
-ext
-extensions
-eyelevel
-facts
-fastidious
-fathers
-favourably
-fax
-feb
-feint
-ferneries
-files
-filters
-Finland
-fireplaces
-flavours
-flights
-fluency
-fluidized
-fluorescent
-fm
-forceps
-forces
-forefront
-foreknowledge
-forman
-formfeed
-formletters
-Francisco
-Frankfurt
-friends
-frond
-fronds
-frontage
-frontseat
-ft
-functions
-funds
-futures
-futuristic
-ga
-Galileo
-Garfield
-Gary
-gaskets
-geiger
-geist
-gentiles
-Georgia
-georgian
-giants
-gigabytes
-glitches
-Gloria
-gm
-gods
-gordon
-governments
-gravitated
-gremlins
-greyhound
-Griffith
-groups
-guages
-Gwen
-Hague
-halftone
-halftones
-handfull
-hans
-hardcopy
-harkness
-Harold
-Harris
-haven't
-Hawaii
-hazards
-headlights
-headquartered
-Helen
-helicopter
-helicopters
-Henderson
-herbalists
-hermeneutical
-hills
-hindu
-historians
-ho
-hoc
-homeowners
-honduras
-hong
-hours
-houses
-Howard
-hr
-hrs
-humours
-hwy
-hyper
-hypercard
-hypertalk
-hz
-i
-I'd
-i.e.
-i/o
-IBM
-iceskating
-id
-Idaho
-ideas
-identification
-identify
-ie
-ii
-iii
-imaged
-implementations
-inc
-individuals
-inferences
-infrastructure
-inoculation
-inspectors
-instructions
-inter
-interpretative
-intro
-intrusive
-intrusives
-irs
-isaac
-iso
-Italy
-its
-iv
-ix
-Jackson
-jaguar
-Jan
-Jane
-Jeanne
-Jed
-Jennie
-Jennings
-jets
-jew
-jewish
-Joe
-John
-Jonathan
-Joseph
-Joyce
-jr
-jurist
-ka
-Kansas
-Karl
-kbytes
-Kenneth
-Kepler
-keyboards
-kg
-kids
-Kirk
-kits
-knockout
-Kong
-la
-labels
-lakeshore
-lama
-lamps
-lan
-laptop
-Larry
-laserjet
-laserwriter
-latin
-Lawrence
-laws
-layouts
-lb
-lbs
-lcd
-le
-leaching
-Lebanon
-leo
-leon
-Lewis
-licensee
-licensees
-limitations
-limpid
-Lincoln
-Linda
-lines
-listings
-literatures
-lithographic
-lithography
-logo
-London
-Louis
-Lynda
-ma
-mac
-mach
-machines
-macintosh
-macintoshes
-maddened
-magnetically
-manmade
-manufacturers
-Margaret
-Maria
-marriages
-Martha
-Martin
-Marvin
-Mary
-mbyte
-mbytes
-MD
-meads
-meals
-measurements
-mechanics
-med
-megabit
-megabyte
-megabytes
-members
-menus
-mesa
-methods
-Mexico
-Meyer
-mg
-MHz
-Michael
-micro
-microbes
-microbial
-microbiological
-microbiology
-microorganism
-microorganisms
-microsoft
-midrange
-miles
-mils
-min
-ming
-mini
-minors
-mips
-mirages
-misnamed
-missioned
-Missouri
-ml
-mm
-mobilisation
-modules
-monarchic
-monastary
-monochrome
-month's
-Moscow
-motorways
-msg
-mt
-multi
-multimedia
-multiuser
-Nurray
-museums
-nasa
-Nathan
-nations
-nd
-ne
-Nelson
-neoprene
-networks
-newsletter
-Newton
-nicholas
-nitrate
-NJ
-nonessential
-nonimpact
-noninfectious
-normative
-northside
-nov
-ns
-nuns
-nutrients
-NY
-o
-odometer
-oem
-offcampus
-offchip
-offsets
-offshore
-ohm
-Olsen
-omni
-onchip
-ondemand
-ones
-opinions
-optimised
-options
-orchards
-oregon
-organisations
-organise
-organises
-organising
-orginal
-os
-ot
-Otto
-outlets
-outmoded
-overdrive
-overdrives
-Owens
-pages
-paperless
-papers
-Paris
-parkway
-passages
-passengers
-passengerside
-patio
-Paul
-payload
-pb
-pc
-pcs
-pedal
-pedals
-penicillin
-peoples
-perambulate
-perils
-periods
-persons
-pesticides
-petri
-pharmacological
-phd
-phenomenalists
-pheobe
-Phil
-Philip
-philosophers
-phlegm
-photolithography
-photometer
-photometrically
-photosensitive
-phototypsetter
-Pierre
-plainpaper
-plans
-platonic
-platonism
-plc
-plots
-pluralism
-pm
-pocketsized
-Polly
-poly
-polygons
-polypropylene
-popup
-potency
-pre
-precautions
-precepts
-premises
-prep
-prerecorded
-presswork
-pretested
-primal
-primo
-problems
-procedures
-processors
-prod
-profits
-programme
-programs
-promissory
-propane
-protestant
-proto
-protocols
-prudential
-ps
-pubs
-racism
-racist
-rad
-radioactive
-rads
-Ralph
-ramirez
-rashers
-raster
-rd
-reactions
-realists
-realtor
-realty
-reardeck
-rearview
-recalibration
-recipients
-recommendations
-recut
-redskins
-reductions
-ref
-reflections
-refund
-reimburse
-rel
-requirements
-resettable
-residues
-resources
-restaurants
-rev
-revue
-rh
-rhizomes
-rhodes
-richard
-rickey
-risc
-rn
-Robert
-Roberts
-rom
-roots
-Rosemary
-rotor
-rotors
-rovers
-rpm
-rt
-rugby
-rumania
-rumohra
-rumours
-Ryan
-ryhthm
-s
-sacrament
-sales
-samaritan
-San
-Sandra
-santa
-Sarah
-savagery
-schemes
-scholasticism
-schuler
-sciencefiction
-sciencehistory
-scientists
-scripts
-scsi
-sculptors
-se
-Seattle
-sec
-sel
-selectable
-selectivity
-selfpaced
-semesters
-sensors
-Sept
-sets
-shareholders
-sheerest
-Shelley
-Sheridan
-ships
-Shirley
-shortcomings
-Sidney
-sig
-sinch
-Sinclair
-singlesheet
-singleuser
-sinkhole
-sists
-skiers
-slots
-socalled
-solidstate
-souls
-sources
-southeast
-soy
-spa
-spans
-spilt
-splines
-spots
-sq
-squelched
-sr
-st
-Stacey
-stacks
-standalone
-states
-steps
-sterility
-Stevens
-stoics
-stoves
-streakings
-stucco
-students
-stylus
-sulfuric
-summa
-summarise
-summers
-sums
-surveyors
-susceptibility
-swab
-swabs
-Swiss
-Switzerland
-sys
-systems
-t
-Taylor
-teaparty
-tech
-techniques
-tel
-temperatures
-Terry
-tests
-th
-theatre
-theatres
-thinning
-thirdparty
-Thomas
-tickets
-timeout
-Timothy
-tm
-Tokyo
-tollfree
-tom
-topics
-torah
-touted
-towels
-toxin
-toxins
-trademarks
-traditions
-trans
-transfers
-treatments
-trees
-trenchant
-tribes
-trinity
-turbidity
-turnaround
-tx
-types
-typeset
-typesetter
-typestyles
-typology
-U.S.
-UK
-ultrafine
-unassisted
-undercut
-units
-unix
-upchucking
-upriver
-ups
-urea
-USA
-userfriendly
-users
-utilise
-utilised
-v
-va
-vacillated
-vehicles
-vendors
-veneer
-vernal
-vi
-victorian
-videotex
-Vienna
-vii
-viii
-visitors
-vitro
-viva
-vol
-vols
-volumes
-vt
-Wallace
-Walter
-walz
-Wang
-warmup
-wastebasket
-Wayne
-ways
-weatherproof
-well
-wellknown
-welt
-Wesley
-westminster
-weston
-Williams
-winchester
-windows
-Winston
-wireframe
-Wisconsin
-wittmann
-words
-workstation
-workstations
-worlds
-wows
-wraps
-x
-xerographic
-xi
-xii
-xiii
-xiv
-xix
-xv
-xvi
-xvii
-xviii
-xx
-years
-Yugoslavian
-zealots
-zion