Extracts every siusch() field
from the HTML of an
Ontario SIU director's report: the police service, the incident,
notification and decision dates, the investigator and witness counts,
the affected person's age and sex, the injuries, the legislation
considered and the director's name. The extraction is rule-based,
deterministic and offline, so the same report always gives the same
answer; use it to turn a folder of saved reports into a data set.
Arguments
- html
A character vector. Each element is either the raw HTML of one report or the path to a saved report file (for example one written by
siufetch()). With more than one element, each is parsed and the results are bound into a data frame.
Value
For one input, a named character vector of 17 elements: the 16
schema fields plus _language ("en", "fr" or
"unknown"), in alphabetical order of name. Every value is a
string, counts included ("2"); convert with
as.integer() as needed. For several inputs, a data.frame with one
row per input and one character column per field (NA where a
field is absent from a row).
Details
The text is first converted as in
siutxt(), then the
report language is detected from section headings (English or French;
the SIU publishes both) and each field is read by its own rule set in
the native core. Dates are converted with the same rules as
siudate(). The
subject-official count uses the rules
of siuso(). Fields the
report does not state come
back as ""; a page that is not a report gives all fields
"" and _language "unknown".
A single string with no < that is shorter than 1000 bytes and
names an existing file is read as a path. A string that looks like a path
(ends in .htm/.html or contains a slash) but names no file is
an error, not an empty report.
Errors: html not character, empty or NA; a missing file
(“no such file”); input over 2 MiB.
Limits
The input is capped at 2 MiB (a report page is a few hundred KB). Before the text is read, a line longer than 2000 characters is split at the sentence end nearest the cap (else at a space), and a warning says how many lines were split: a field that spans a split may be incomplete.
Examples
f <- siuprs(system.file(
"extdata",
"siu_synthetic_report.html",
package = "rmoriebricklayer"
))
f[["number_of_subject_officials"]]
#> [1] "2"
f[c(
"police_service",
"date_of_incident_iso",
"date_of_director_decision_iso"
)]
#> police_service date_of_incident_iso
#> "Barrie Police Service" "2023-01-05"
#> date_of_director_decision_iso
#> "2023-04-28"
# several reports at once give a data frame, one row each
path <- system.file(
"extdata",
"siu_synthetic_report.html",
package = "rmoriebricklayer"
)
html <- paste(readLines(path), collapse = "\n")
df <- siuprs(c(html, "<p>not a report</p>"))
df[, c("_language", "police_service", "number_of_witness_officials")]
#> _language police_service number_of_witness_officials
#> 1 en Barrie Police Service 3
#> 2 unknown
# a mistyped path is an error, not an empty report
try(siuprs("no-such-report.html"))
#> Error : `html`: no such file: no-such-report.html