Before adding the file name of where the address is extracted I assumed that the page displayed is "summons". However, I can several times that address is pulled from a file where I am not sure if this is the one that is displayed. So a name of the currently displayed file would be useful.
Right now, the sampleCounts is loaded on the client when the /Play component is mounted. Since sampleCounts is changing on the backend, we need to periodically check to see if there is a change and re-fetch this from the client.
create a variable called "sessionSamples" -- this will represent the samples that you've seen in a given session. Whenever the Play component is mounted, we should reset this to [].
in that same function, if the number of samples seen in a given session is over a threshold number (maybe something defined in the config file), then run the method to grab and set the sampleCounts data
Currently if file name contains &, the app sends a wrong request, cropped at the ampersand sign. For instance, the firekey 05a72f458ad9a6a6adb551ca7f605d40 (Washington King 2006 06-2-24953-4) contains file SummonsandComplaint-SUMMONS&COMPLAINT.pdf. In the API logs I can see request for SummonsandComplaint-SUMMONS. My reading is the & is not converted to http-safe (e.g. %26) and hence considered as end of the file name.
The current webpage is fairly good for the basic info but there should be an option like additional fields for things like court, building, county and such.
There should also be an option to see additional information, in particular
file name where the address is coming from
list of all files.
case number
the estimated goodness measure
These are not needed for laymen but I occasionally feel the need to understand better what's going on there.
Sometimes the backend fails with 400, leading to failed to load data message on screen. It is related requesting file named null. Not sure why this happens as the preceding /files request returns a valid file name (apparently). This happens with
This can then be updated on the Admin tab. It should be a URL to a .json file, which is a list of strings. These strings are parsed by the app to GET information from tesseract
These keys are sent to an endpoint to get a PDF & guessed address. Right now the endpoint requires multiple sources of information (case number, county, state, etc), but since each PDF has its own "key" attribute, we need to:
create endpoints that return a PDF & guessed address with a key
an endpoint that returns a list of keys on the server
update the AddressWidget to call the endpoint with the key (e.g. remove parsing)
Summons-SUMMONS.pdf from King 2010 10-2-33955-8 (fb id 5617d34b8341eb14dbf71cb7e708011b) does not display. It seems to be a valid pdf, and is visible in firefox when looked separately (file://...). Log files suggest all well.
The addresses are completely wrong between what's on the file and what's output on the fields. Attached is an example. Case number of interest is 09-2-13981.