gaze-recognizers 0.7.0

Built-in recognizers for Gaze
Documentation
schema_version = "0.1.0"
rulepack_id = "gaze-core"
rulepack_version = "0.5.1"
default_locales = ["global"]

[locale.email_headers]
names = ["From", "To", "Cc", "Bcc", "Reply-To", "Sender"]

[[recognizers]]
id = "email.global"
class = "Email"
enabled = true
locales = ["global"]

[recognizers.match]
kind = "regex"
pattern = '''(?i)\b[a-z0-9._%+\-]+@(?:(?:[a-z0-9\-]+\.)*example\.invalid|test\.local|[a-z0-9](?:[a-z0-9\-]*[a-z0-9])?\.(?:com|org|net|edu|gov|de|uk|fr|nl|io|ai|co))\b'''

[recognizers.context]
exclusions = ["example.com"]

# Reserved — not yet applied; see v0.4.1 roadmap.
# hotwords = ["email", "e-mail", "mail"]
# window = 12
# boost = 0.10

[recognizers.validator]
kind = "email_rfc"

[recognizers.normalizer]
kind = "email_canonical"

[recognizers.scoring]
base = 0.70
priority = 90

[recognizers.token]
# Reserved — not yet applied; see v0.4.1 roadmap.
# family = "email.counter"
# format = "Email_{n}"

[recognizers.source]
origin = "ported"
from = "presidio"
license = "Apache-2.0"

[[recognizers]]
id = "email.header.name"
class = "Name"
cooperates_with = ["email.header.name.paren", "name.forward_marker", "name.agent_recipient", "name.auto_footer"]
enabled = true
locales = ["global"]

[recognizers.match]
kind = "regex"
pattern_template = '''(?m)^(?:{locale.email_headers}):\s*(?:"([^"]+)"|([A-Z][a-z]+(?:\s+[A-Z][a-z]+){0,3}))\s+<[^>]+>'''
capture_groups = [1, 2]

[recognizers.scoring]
base = 0.85
priority = 100

[[recognizers]]
id = "email.header.name.paren"
class = "Name"
cooperates_with = ["email.header.name", "name.forward_marker", "name.agent_recipient", "name.auto_footer"]
enabled = true
locales = ["global"]

[recognizers.match]
kind = "regex"
pattern_template = '''(?:(?:{locale.email_headers}):|,)\s*\S+@\S+\s*\(([^)]+)\)'''
capture_groups = [1]

[recognizers.scoring]
base = 0.85
priority = 100

[recognizers.token]
family = "name.counter"

[[recognizers]]
id = "name.forward_marker"
class = "Name"
cooperates_with = ["email.header.name", "email.header.name.paren", "name.agent_recipient", "name.auto_footer"]
enabled = true
locales = ["de-DE", "de-AT", "de-CH", "en-US", "en-GB", "en-IE", "en-AU", "en-CA"]

[recognizers.match]
kind = "anchored_match"
cues_bucket = "forward_markers"
boundary = "punctuation"
right_window_chars = 64
name_shape = "person_name"
cue_position = "before"

[recognizers.scoring]
base = 0.88
priority = 110

[recognizers.token]
family = "name.counter"

[[recognizers]]
id = "name.agent_recipient"
class = "Name"
cooperates_with = ["email.header.name", "email.header.name.paren", "name.forward_marker", "name.auto_footer"]
enabled = true
locales = ["de-DE", "de-AT", "de-CH", "en-US", "en-GB", "en-IE", "en-AU", "en-CA"]

[recognizers.match]
kind = "anchored_match"
cues_bucket = "agent_recipient_cues"
boundary = "punctuation"
right_window_chars = 48
name_shape = "person_name"
cue_position = "before"

[recognizers.scoring]
base = 0.88
priority = 110

[recognizers.token]
family = "name.counter"

[[recognizers]]
id = "name.auto_footer"
class = "Name"
cooperates_with = ["email.header.name", "email.header.name.paren", "name.forward_marker", "name.agent_recipient"]
enabled = true
locales = ["de-DE", "de-AT", "de-CH", "en-US", "en-GB", "en-IE", "en-AU", "en-CA"]

[recognizers.match]
kind = "anchored_match"
cues_bucket = "footer_cues"
boundary = "whitespace"
right_window_chars = 64
name_shape = "person_name"
cue_position = "before"

[recognizers.scoring]
base = 0.88
priority = 110

[recognizers.token]
family = "name.counter"