price_scanner 0.2.3 → 0.3.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 2014de33dd81d654b4fd82992474605dd110ff5154299dec1bfca80ec17f91db
-  data.tar.gz: b341b9072d034078d95c8a335d0d8f9d9b62bd2d00dc4379441a48cebe3091b9
+  metadata.gz: 933c597623a0441a0554f9cc3b408134180022e69788f15b4e3d785fcf22ab83
+  data.tar.gz: 8dc569cea4f591cd1cd11db0eba535a1b166f9913e9e6069155b1cdf08421709
 SHA512:
-  metadata.gz: edffb337ea738fa8541612a49d1a2417f40c4d3463ef4b0509eef224e34172278f1ea585b2c34e470aea0feb457540d7be2218a4fdc97f2fb2e67e8c75244190
-  data.tar.gz: e0a1d51e6a441400d41c6c77c25ca9a5d083ec8e2d758a76685321281e4e14c7bf0a05d3bbac27cd60ecc04e0babf8f2bcd0aad23e471246fcbf6f957fcc890c
+  metadata.gz: 6bafb309f4ef2d2296cc7981bf44bbf8eb42e185f4995809cc0e8aa4e0aa80ce25299bda7996892d42c03a225f4b961e4c8d7dcdc98f8329145c7e09ce6a6314
+  data.tar.gz: 48fce8a5a51cacb32fe062b380da562e72dcab7a44e2c5ea4bc0a3544bb3f854257c4b86491e9d6cccd4dc2ae056329092bad4189c5434ab6e2873718e137aed

data/CHANGELOG.md

@@ -1,5 +1,9 @@
 # Changelog
 
+## 0.3.0
+
+- Add `include_per_unit:` option to `extract_prices_from_text` — allows including per-unit prices (`£46.00/M`, `29,99 zł/kg`) that are filtered by default
+
 ## 0.2.3
 
 - Fix comma-as-thousands-separator not recognized in PRICE_PATTERN (`7,999.00 €` → was parsed as `999.00 €`)

data/README.md

@@ -134,7 +134,7 @@ PriceScanner.scan("Was 449,00 zł, now 349,00 zł. You save 100,00 zł!")
 
 ### Per-unit prices
 
-Prices followed by a unit indicator are filtered out.
+Prices followed by a unit indicator are filtered out by default.
 
 Supported units: `kg`, `g`, `mg`, `l`, `ml`, `szt`, `m`, `m²`, `m³`, `cm`, `mm`, `op`, `opak`, `pcs`, `pc`, `unit`, `each`, `ea`, `kaps`, `tabl`, `tab`
 
@@ -146,6 +146,13 @@ PriceScanner.scan("32,74 zł/kg — buy 500g for 16,37 zł")
 # 32,74 zł/kg is excluded
 ```
 
+For products priced exclusively per unit (e.g., fabrics sold per meter, bulk goods per kg), pass `include_per_unit: true`:
+
+```ruby
+PriceScanner::Detector.extract_prices_from_text("£46.00/M £13.55/M", include_per_unit: true)
+# => [{text: "£46.00/M", value: 46.0}, {text: "£13.55/M", value: 13.55}]
+```
+
 ### Deduplication
 
 If the same price value appears multiple times, only one occurrence is kept.

data/lib/price_scanner/detector.rb

@@ -14,6 +14,10 @@ module PriceScanner
 
     NEGATIVE_PREFIXES = ["-", "\u2212"].freeze
 
+    # Prefixes that indicate the following price is a savings amount, not a product price.
+    # "Oszczędzasz 6.40 PLN" = "You save 6.40 PLN" — not the product price.
+    SAVINGS_PREFIX_PATTERN = /(?:oszcz[eę]dzasz|zaoszcz[eę]d[zź]|savings?|you\s+save|rabat|discount|remise|risparmio|ahorro|sparen|sie\s+sparen)\s*:?\s*\z/i
+
     RANGE_SEPARATOR_PATTERN = /\s*[–—]\s*|\s+-\s+/
 
     TEXT_AFTER_LOOKAHEAD = 200
@@ -26,9 +30,9 @@ module PriceScanner
 
     module_function
 
-    def extract_prices_from_text(text)
+    def extract_prices_from_text(text, include_per_unit: false)
       text_str = text.to_s
-      raw_prices = scan_raw_prices(text_str)
+      raw_prices = scan_raw_prices(text_str, include_per_unit: include_per_unit)
       filtered = filter_range_prices(raw_prices, text_str)
       unique = filtered.uniq { |price| price[:value] }
       filter_savings_by_difference(unique)
@@ -38,34 +42,35 @@ module PriceScanner
       text.to_s.match?(PRICE_PATTERN)
     end
 
-    def scan_raw_prices(text_str)
+    def scan_raw_prices(text_str, include_per_unit: false)
       results = []
       last_end = 0
 
       text_str.scan(PRICE_PATTERN) do |match_str|
-        result, last_end = find_price_at(text_str, match_str, last_end)
+        result, last_end = find_price_at(text_str, match_str, last_end, include_per_unit: include_per_unit)
         results << result if result
       end
 
       results
     end
 
-    def find_price_at(text_str, match_str, search_from)
+    def find_price_at(text_str, match_str, search_from, include_per_unit: false)
       return [nil, search_from] if match_str.empty?
 
       match_index = text_str.index(match_str, search_from)
       return [nil, search_from] unless match_index
 
       match_end = match_index + match_str.length
-      [build_price_result(text_str, match_str, match_index), match_end]
+      [build_price_result(text_str, match_str, match_index, include_per_unit: include_per_unit), match_end]
     end
 
-    def build_price_result(text_str, match_str, match_index)
+    def build_price_result(text_str, match_str, match_index, include_per_unit: false)
       value = Parser.normalized_price(match_str)
       return unless value
 
       return if negative_price?(text_str, match_index)
-      return if per_unit_price?(text_str, match_index + match_str.length)
+      return if savings_prefix?(text_str, match_index)
+      return if !include_per_unit && per_unit_price?(text_str, match_index + match_str.length)
 
       clean_text = match_str.gsub(Parser::COLLAPSE_WHITESPACE, " ").strip
       { text: clean_text, value: value, position: match_index }
@@ -89,6 +94,16 @@ module PriceScanner
       i >= 0 ? i : nil
     end
 
+    # Check if text before the price contains a savings prefix like "Oszczędzasz" or "You save"
+    def savings_prefix?(text_str, match_index)
+      return false unless match_index > 3
+
+      # Look at up to 30 chars before the price match
+      lookback_start = [match_index - 30, 0].max
+      text_before = text_str[lookback_start...match_index]
+      text_before.match?(SAVINGS_PREFIX_PATTERN)
+    end
+
     def per_unit_price?(text_str, match_end)
       text_after = text_str[match_end, TEXT_AFTER_LOOKAHEAD].to_s.gsub(Parser::COLLAPSE_WHITESPACE, " ").lstrip
       text_after.match?(PER_UNIT_ANCHOR)
@@ -147,7 +162,7 @@ module PriceScanner
     end
 
     private_class_method :scan_raw_prices, :find_price_at, :build_price_result,
-                         :negative_price?, :rindex_non_space, :per_unit_price?,
+                         :negative_price?, :rindex_non_space, :savings_prefix?, :per_unit_price?,
                          :filter_range_prices, :find_range_indices, :range_between?,
                          :filter_savings_by_difference, :savings_amount?, :matches_savings_pattern?
   end

data/lib/price_scanner/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module PriceScanner
-  VERSION = "0.2.3"
+  VERSION = "0.3.1"
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: price_scanner
 version: !ruby/object:Gem::Version
-  version: 0.2.3
+  version: 0.3.1
 platform: ruby
 authors:
 - Justyna