class OilReportParser # Blackstone Laboratories report: one page with a header (lab number, unit ID, # oil type), the lab's comments, then a grid with one column per sample (newest # on the left, dated across the top) and one row per element / oil property. # Two extra columns hold averages: this unit's own (skipped) and the lab's # universal averages for the engine type (kept as the comparison average). # # Numbers are right-aligned under their sample dates, so values are matched to # a sample by their right edge. class Blackstone LAB = "Blackstone".freeze ELEMENTS = %w[aluminum chromium iron copper lead tin molybdenum nickel manganese silver titanium potassium boron silicon sodium calcium magnesium phosphorus zinc barium].freeze PROPERTIES = { /\ASUS Viscosity/ => "viscosity_sus_210f", /\AcSt Viscosity/ => "viscosity_cst_100c", /\AFlashpoint/ => "flashpoint_f", /\AFuel %/ => "fuel_percent", /\AAntifreeze %/ => "antifreeze_percent", /\AWater %/ => "water_percent", /\AInsolubles %/ => "insolubles_percent", /\ATBN\z/ => "total_base_number", /\ATAN\z/ => "total_acid_number" }.freeze LABEL_MAX_X = 130 # row labels sit in the left margin ROW_SLOP = 3.5 # a row's values can sit a couple of points off its label's baseline COLUMN_SLOP = 8 # how far a value's right edge may be from its date's DATE = %r{\A\d{1,2}/\d{1,2}/\d{4}\z} # A US registration in the "unit ID" field; anything else (owners often put an # engine name there) isn't treated as a tail number. TAIL = /\AN[1-9][0-9A-Z]{0,4}\z/i Run = Struct.new(:text, :x, :endx, :y) def initialize(runs) # The vertical side captions ("UNIT INFORMATION"…) are single letters at x≈43. @runs = runs.filter_map { Run.new(_1.text.strip, _1.x, _1.endx, _1.y) if _1.x >= 48 && _1.text.strip.present? } end def parse date_label = label("Sample Date") or raise Error, I18n.t("services.oil_report.no_sample_dates_found_in") dates = row(date_label).select { DATE.match?(_1.text) }.sort_by(&:x) raise Error, I18n.t("services.oil_report.no_sample_dates_found_in") if dates.empty? @columns = dates.map(&:endx) @universal_x = (@runs.find { _1.text == "UNIVERSAL" }&.x || 530) - 10 oil_hours, engine_hours, oil_added = [ "MI/HR on Oil", "MI/HR on Unit", "Make Up Oil Added" ].map { values(label(_1)) } elements, universal = ELEMENTS.each_with_object([ {}, {} ]) do |el, (by_el, avg)| by_el[el] = values(label(el.upcase)) avg[el] = universal_value(label(el.upcase)) end properties = PROPERTIES.to_h { |pattern, key| [ key, values(label(pattern)) ] } universal = universal.transform_values { OilReportParser.number(_1) }.compact lab_number = after("LAB NUMBER:") samples = dates.each_with_index.map do |date, i| { current: i.zero?, sample_date: OilReportParser.date(date.text), sample_number: (lab_number if i.zero?), oil_hours: OilReportParser.number(oil_hours[i]), engine_hours: OilReportParser.number(engine_hours[i]), oil_added: OilReportParser.number(oil_added[i]), status: nil, # Blackstone gives comments, not a rating lab_comments: (comments if i.zero?), elements: elements.transform_values { OilReportParser.number(_1[i]) }.compact, element_averages: universal, oil_properties: properties.transform_values { OilReportParser.number(_1[i]) }.compact } end.select { _1[:sample_date] } unit = after("UNIT ID:") Report.new(lab: LAB, tail_number: (unit if TAIL.match?(unit.to_s)), engine_serial: nil, engine_model: after("MAKE/MODEL:"), oil_type: after("OIL TYPE & GRADE:"), samples: samples) end private # The left-margin run that labels a row. def label(match) @runs.find { _1.x < LABEL_MAX_X && (match.is_a?(Regexp) ? match.match?(_1.text) : _1.text == match) } end # Runs on the same row as a label, to its right. def row(label_run) return [] unless label_run @runs.select { (_1.y - label_run.y).abs <= ROW_SLOP && _1.x > label_run.endx } end # The row's value text for each sample column (nil where blank). def values(label_run) found = Array.new(@columns.size) row(label_run).each do |run| next if run.x >= @universal_x i = @columns.each_index.min_by { (@columns[_1] - run.endx).abs } found[i] ||= run.text if (@columns[i] - run.endx).abs <= COLUMN_SLOP end found end def universal_value(label_run) = row(label_run).find { _1.x >= @universal_x }&.text # The value printed after a header label, e.g. "UNIT ID:" → "N12345". def after(text) label_run = @runs.find { _1.text == text } or return nil @runs.select { (_1.y - label_run.y).abs <= 2 && _1.x > label_run.endx }.min_by(&:x)&.text end # The comment paragraph between the contact block and the results grid. # It starts with who it's addressed to ("OWNER:"), which is dropped. def comments contact = @runs.select { _1.text.match?(/\A(PHONE|FAX|ALT PHONE|EMAIL):/) } top = contact.map(&:y).min or return nil grid = label("MI/HR on Oil")&.y || 0 lines = @runs.select { _1.y < top - 2 && _1.y > grid + 6 }.sort_by { [ -_1.y, _1.x ] } lines.map(&:text).join(" ").sub(/\A[A-Z]+:\s*/, "").squeeze(" ").presence end end end