Navigation

Changeset 100

Timestamp:

10.12.2006 01:29:31 (2 years ago)

Author:

decoder

Message:

Do not use, contains a bug!

Adjusted some defaults in config file / Config.pm
Made final changes to FuzzyOcr?.scansets and .preps

Files:

r93	r100
134	134	# This saves resources, but lowers the scores because not the best,
135	135	# but the first best scanset is taken as result.
136		# Default value: 0
137		#focr_minimal_scanset 1
	136	# Default value: 1
	137	#focr_minimal_scanset 0
138	138
139	139	# This option is only used when focr_minimal_scanset is enabled.
…	…
246	246	# Allowed values:
247	247	# 1 ... use digest_hash only (deprecated)
248		# 2 ... use digest_db w/digest_hash import (see requirements)
249		# 3 ... use mysql database (see requirements)
	248	# 2 ... use digest_db w/digest_hash import (see requirements, recommended)
	249	# 3 ... use mysql database (see requirements, experimental)
250	250	#--
251	251	# The score is saved with the hash in the database, allowing the plugin to
…	…
260	260
261	261	# Automatically add hashes of spam images recognized by OCR to the Image
262		# Hash database, to disable, set to 0.0
	262	# Hash database, to disable, set to 0
263	263	# Default value: 1 (learn)
264	264	#focr_hashing_learn_scanned 1
…	…
289	289
290	290	###
291		### MySQL options
	291	### MySQL options (Type 3 Hashing)
292	292	###
293	293

r70	r100
2	2	# Do not modify this on your own unless you have read the manual and know what you're doing
3	3
	4	# Normalizes a PNM
4	5	preprocessor normalize {
5	6	command = pnmnorm
6	7	}
7	8
	9	# Inverts a PNM
8	10	preprocessor invert {
9	11	command = pnminvert
10	12	}
11	13
12		# requires ImageMagic convert
	14	# Converts PPM (Color PNM) to PGM (Greyscale PNM)
	15	preprocessor ppmtopgm {
	16	command = ppmtopgm
	17	}
	18
	19	# Converts PAM to PNM
	20	preprocessor pamtopnm {
	21	command = pamtopnm
	22	}
	23
	24	# Uses thresholding on the PAM file
	25	preprocessor pamthreshold {
	26	command = pamthreshold
	27	args = -simple -threshold 0.5
	28	}
	29
	30	# converts PNM to TIFF (this is used for tesseract)
13	31	preprocessor maketiff {
14		command = ~~convert~~
15		args = ~~$input tiff:$output~~
	32	command = pnmtotiff
	33	args = -color -truecolor
16	34	}

r70	r100
2	2	# Do not modify this on your own unless you have read the manual and know what you're doing
3	3
	4	# Note: If the focr_minimal_scanset option is enabled (default),
	5	# not all of these scansets run for every picture. But be aware
	6	# that for ham images, all of them will always run.
	7
	8
	9	# Standard Ocrad Scanset
4	10	scanset ocrad {
5	11	command = $ocrad
…	…
7	13	}
8	14
	15	# Inverted Ocrad scanset
9	16	scanset ocrad-invert {
10	17	command = $ocrad
…	…
12	19	}
13	20
	21	# Inverted Ocrad scanset
	22	scanset ocrad-invert {
	23	preprocessors = ppmtopgm, pamthreshold, pamtopnm
	24	command = $ocrad
	25	args = -s5 -i $input
	26	}
	27
	28	# Standard Gocr Scanset
14	29	scanset gocr {
15	30	command = $gocr
…	…
17	32	}
18	33
	34	# Tweaked Gocr Scanset
19	35	scanset gocr-180 {
20	36	command = $gocr
…	…
22	38	}
23	39
24		scanset gocr-invert {
25		preprocessors = normalize, invert, normalize
26		command = $gocr
27		args = -i $input
28		}
	40	# An example Scanset how to use tesseract (disabled by default)
	41	#scanset tesseract {
	42	# preprocessors = maketiff
	43	# command = $tesseract
	44	# args = $input $output
	45	# force_output_in = $output.txt
	46	#}
29	47
30		scanset tesseract {
31		preprocessors = maketiff
32		command = $tesseract
33		args = $input $output batch
34		force_output_in = $output.txt
35		}
	48	# Another example, this basically does the same as the inverted Ocrad Scanset
	49	# Only listed here to serve as example, ocrad-invert does this already
	50	#
	51	#scanset gocr-invert {
	52	# preprocessors = normalize, invert, normalize
	53	# command = $gocr
	54	# args = -i $input
	55	#}

r99	r100
432	432	push (@cmds, {
433	433	setting => 'focr_minimal_scanset',
434		default => 0,
	434	default => 1,
435	435	type => $Mail::SpamAssassin::Conf::CONF_TYPE_BOOL
436	436	});