Casting/validation gives unexpected errors

[Mikanebu]

Here are a couple of errors that I think, probably a bug

• when "type: string" in resource schema, values in form of numbers give an error. E.g.:
  Wrong type for header: territory and value: 001
  It is erroring for both values 001 and "001".
  https://github.com/datasets/language-codes/blob/master/data/ietf-language-tags.csv
• geopoint type is not working well. It gives an error even though it has a valid format according to the spec. For example, “59.94919968, -151.6959991”:
  "Wrong type for header: coordinates and value: 59.94919968, -151.6959991"
  https://raw.githubusercontent.com/datasets/airport-codes/master/data/airport-codes.csv

Also, when I got an error message, it was not clear:

• Wrong type for header: Coordinates and value: -14.27667
  Suggestion: Coordinates field cannot accept value -14.27667. Geopoint type should be in the pattern "lon, lat".
• Row X is completely blank { "error-count": 1, "headers": [ "Country Name", "Country Code", "Year", "Value" ], "time": 0.406, "row-count": 1000, "source": "gdp/data/gdp.csv", "valid": false }
  This error message has unnecassary information and not clear.
  Suggestion: Row X is completely blank. According to the spec it cannot be accept empty rows. Path to source
• Row 7 is duplicated to row(s) 6
  Same here
• When schema type is not type of *array*.
  Too much information and not clear.
  Suggestion: Failed validating 'type' in schema['properties']['sources']. Sources property type must be type of array.
• AttributeError: 'list' object has no attribute 'split'
  Wrong geotype error which is not clear.
  Probably it is kind of bug

[roll]

@Mikanebu
Hi. Thank you for the report. Could you please write tableschema library version.

[Mikanebu]

@roll 0.2.2 - jsontableschema
1.0.0-alpha.7 - table schema

[rufuspollock]

@Mikanebu could you link files / data packages where these errors arise in your description.

[Mikanebu]

@rufuspollock Spreadsheet for the broken data packages:( Sheet1 has all the errors we fixed, in Sheet2, rows 2-14 have an errors abovementioned)
https://docs.google.com/spreadsheets/d/14kJluhePaMOx6vYBic0poVjDYK3I8-v_xF3sv4Focac/edit#gid=1881654332&range=2:14

[Mikanebu]

@roll I updated issue by providing with exact csv file location, the spreadsheet that I abovementioned is really confusing. If you have any questions, let me know

[rufuspollock]

@roll do you know when you might be able to confirm these are bugs and, if so, fix them? We have an upstream dependency on this and it would be great to see them fixed 😄

[roll]

@Mikanebu
@rufuspollock
It's a bug. I'll fix it Monday/Tuesday.

[roll]

I'm on it but seems will release tomorrow not today. @Mikanebu thanks for error message suggestions. I'll integrate it.

[roll]

@Mikanebu could you help here with exact failing code example. I can't reproduce it:

when "type: string" in resource schema, values in form of numbers give an error. E.g.:
Wrong type for header: territory and value: 001
It is erroring for both values 001 and "001".
https://github.com/datasets/language-codes/blob/master/data/ietf-language-tags.csv

// tableschema@1.0.0-alpha.7
const schema = {
  fields: [
    {
      name: 'lang',
      description: 'IANA/Unicode language-tag-extension',
      type: 'string'
    },
    {
      name: 'langtype',
      description: 'ISO 2 letter alpha-2 language code',
      type: 'string'
    },
    {
      name: 'territory',
      description: 'ISO3166-1-Alpha-2 country code',
      type: 'string'
    },
    {
      name: 'revgendate',
      description: 'revision date (format ISO data)',
      type: 'string'
    },
    {
      name: 'defs',
      description: 'number of definitions',
      type: 'integer'
    },
    {
      name: 'dftlang',
      description: 'indicate the default-language, as unicode-cldr',
      type: 'boolean'
    },
    {
      name: 'file',
      description: 'file-name of the locale descriptor',
      type: 'string'
    }
  ]
}
const data = 'https://raw.githubusercontent.com/datasets/language-codes/master/data/ietf-language-tags.csv'
new tableschema.Table(schema, data).then(async (table) => {
  const rows = await table.read()
  console.log(rows) // all good
})

[roll]

@Mikanebu I would say it's not a valid geopoint according to the specs: A string of the pattern "lon, lat", where lon is the longitude and lat And https://github/datasets/airport-codes/ has geopoints in "lat, lon" format (reverse order). So a latitude with a value -151.6959991 is invalid.

geopoint type is not working well. It gives an error even though it has a valid format according to the spec. For example, “59.94919968, -151.6959991”:
"Wrong type for header: coordinates and value: 59.94919968, -151.6959991"
https://raw.githubusercontent.com/datasets/airport-codes/master/data/airport-codes.csv

[roll]

CLOSED in favor of:

• improve error messages - Improve error messages #73
• fix airport-codes dataset - Geopoints are incorrect datasets/airport-codes#7

Please open new issue if you still experience problems with type: string. But please add a code causing the issue (can't reproduce).