Al
|
e5fdd915d0
|
[fix] check the first phrase for components and bail if it matches something other than the specified tag
|
2016-07-21 17:04:57 -04:00 |
|
Al
|
8370a41ec0
|
[fix] import
|
2016-07-21 17:04:57 -04:00 |
|
Al
|
651bc32650
|
[addresses] more thoroughly solving the addr:city='Harlem' issue
|
2016-07-21 17:04:57 -04:00 |
|
Al
|
5a31b60cbe
|
[addresses] Adding normalized_place_name, a method for separating compound fields like addr:city='New York NY' into simply 'New York', solving the compound phrase problem. Also solves the mislabeled place name problem, causing the system to ignore the user tag and fall back on reverse geocoded components in cases e.g. where addr:city='Harlem', which is a known neighborhood but not a city when reverse geocoded. A few other refactors for expanded address components
|
2016-07-21 17:04:57 -04:00 |
|
Al
|
fa99b4ce77
|
[addresses] wrapping up some of the functionality from OSM formatter to be used in on an arbitrary address component dictionary
|
2016-07-21 17:04:57 -04:00 |
|
Al
|
7e5ecb30cf
|
[addresses] sample_alphabet (Zipfian) in PO box rather than a uniform choice
|
2016-07-21 17:04:57 -04:00 |
|
Al
|
3d765e9eca
|
[addresses] Fixing direction_probability, adding ability to have phrases which only apply to numbers, adding the possibility of null phrases to non-numeric "numbers" e.g. A-Z, etc.
|
2016-07-21 17:04:57 -04:00 |
|
Al
|
04a5a9e611
|
[fix] Removing YAML inheritance as it doesn't merge nested dictionaries
|
2016-07-21 17:04:57 -04:00 |
|
Al
|
37747709ee
|
[addresses] Using YAML inheritance instead of baking it into the config parser
|
2016-07-21 17:04:57 -04:00 |
|
Al
|
8aac200d74
|
[addresses] config for phrases around postcodes like CP in Spanish
|
2016-07-21 17:04:57 -04:00 |
|
Al
|
a7fe6408c0
|
[addresses] /po_box/po_boxes/
|
2016-07-21 17:04:57 -04:00 |
|
Al
|
1e107f09ab
|
[addresses] Generate house number related phrases
|
2016-07-21 17:04:57 -04:00 |
|
Al
|
90c88a3a24
|
[fix] None handling and number dictionaries
|
2016-07-21 17:04:57 -04:00 |
|
Al
|
e13c536b03
|
[addresses] different dictionaries for sampling cardinal/unit directions, not converting None to a string
|
2016-07-21 17:04:57 -04:00 |
|
Al
|
7f3667caf8
|
[dictionaries] Removing ambiguous abbreviations for flat
|
2016-07-21 17:04:57 -04:00 |
|
Al
|
c47762b91c
|
[addresses] Unit/apartment number generation
|
2016-07-21 17:04:57 -04:00 |
|
Al
|
ca68391ea6
|
[addresses] sample positive floors
|
2016-07-21 17:04:57 -04:00 |
|
Al
|
9f652591ad
|
[mv] Moving sampling to math.sampling
|
2016-07-21 17:04:57 -04:00 |
|
Al
|
32b6217aa8
|
[addresses] Conjunction can be subclassed
|
2016-07-21 17:04:57 -04:00 |
|
Al
|
535453f77d
|
[addresses] Adding ability to randomly append relative/cardinal directions
|
2016-07-21 17:04:57 -04:00 |
|
Al
|
f026e8a764
|
[addresses] Adding base class for numeric phrases (appended to a number using numeric/numeric_affix), using probability 1.0 if only one of numeric/numeric_affix/ordinal is specified
|
2016-07-21 17:04:57 -04:00 |
|
Al
|
f7764b70cd
|
[addresses] implementing null_probability (raw number, no phrase), orindal genders, and direction_probability
|
2016-07-21 17:04:57 -04:00 |
|
Al
|
b5386eb601
|
[addresses] generator for floor numbers as well as special aliases like basement, mezzanine, etc. using the address configs
|
2016-07-21 17:04:57 -04:00 |
|
Al
|
317d3aa9ed
|
[addresses] PO Box phrase generator
|
2016-07-21 17:04:57 -04:00 |
|
Al
|
9c4348a990
|
[addresses] conjunction class for building phrases like "5th and 6th" or "Units 1 & 2" across languages using the address configs
|
2016-07-21 17:04:57 -04:00 |
|
Al
|
d136fb7576
|
[addresses] base class for numbered components (floors, units, house numbers in some languages/countries). Can generate many variants of a number (e.g. Floor 2, 2nd Floor, Floor #2, Floor No. 2, etc.)
|
2016-07-21 17:04:57 -04:00 |
|
Al
|
14c89e6895
|
[addresses] utilities for sampling from an arbitrary discrete distribution, building cumulative distributions, and sampling from a Zipfian distribution which seems to be a reasonable way of generating plausible apartment/floor numbers when the height/number of units is unknown. Picking a letter uniformly at random means P('Unit A') == P('Unit Z') when 'A' should be much more likely. Sampling from a Zipfian gets the desired effect in situations where address components are numbered by "counting from 0/1/A" while still allowing for a long tail
|
2016-07-21 17:04:57 -04:00 |
|
Al
|
dcabdf7c0b
|
[addresses] address config class for general sampling of forms specified in the address configs (default/alternatives to choose a phrase, canonical/abbreviated/sample to choose an abbreviation or surface form for that phrase)
|
2016-07-21 17:04:57 -04:00 |
|