price_scanner 0.2.2 → 0.3.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 5866ddf8fd84a9bc437fc7abb5cb6e137ce3d9f9f36d1ff778e1e8db668d0ce1
-  data.tar.gz: 92ee758d982507e59b5a3da296d5390f6b1af276ebe52359e3718909b5186271
+  metadata.gz: 253513985856fa4e2c504df5157eefcf971e1f686fe3869ec5a93b047603294b
+  data.tar.gz: b098c087d64fc1ed716575576ce5cfd0b30d59b9d667858ac71a203ea42d4373
 SHA512:
-  metadata.gz: 151ba2ea359f8e9bfe5c44b541de9c51e83c8316a8237ca682f00f1e042ac52c2a6328fd778541ab569a7bf577caed233e83d127acf822aabd493c7d519fa6c7
-  data.tar.gz: 98e055152d925a8ed027f2baab2523d7e77804a6ea139985d296d3bb36273aa44450e80b8093ac18b23f1661b3350dafb62392948a04cdd869d31cf235d7b5b5
+  metadata.gz: 2f5cf01095f4ed5beb298262ce385ed43b3cadecbc130590c57a8dfac705ff0dcbe5a74a0534cb4426ef05aba8621d74bae99626709664949a48f31b9a88d432
+  data.tar.gz: 27d95b2ce089306059edb5606fc1acd8034dfbf7bb343e253e821edfc259f3389ec1fcc3bde08a1cd3dd25270da206e9bc569d53817a012b3e5e2714b9602f7e

data/CHANGELOG.md

@@ -0,0 +1,39 @@
+# 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 €`)
+- Affects prices in English/international format: `$1,299.99`, `8,289.00 €`, etc.
+- Safe change: requires exactly 3 digits after separator, so decimal commas (`19,99 zł`) still work correctly
+
+## 0.2.2
+
+- Fix negative price detection with spaced dash ("- 1.040 zł") — savings badges with space between minus and price were not filtered
+- Refactor `negative_price?` with `rindex_non_space` helper (DRY)
+- Distinguish range separators ("Pack of 3 - 29,99 zł") from negative prices
+
+## 0.2.1
+
+- Fix false price extraction from model numbers (IP65, HC940, H265, 2K 30MP)
+- Prevent digits before currency symbol from being matched as prices
+
+## 0.2.0
+
+- Remove ConsentDetector from gem (moved to smart_offers app)
+
+## 0.1.1
+
+- Remove rubycritic dependency
+- Auto-require all price_scanner modules
+
+## 0.1.0
+
+- Initial release
+- `PriceScanner::Parser` — normalize prices, extract currency, strip price mentions
+- `PriceScanner::Detector` — extract prices from text, filter negatives/per-unit/ranges/savings
+- Multi-currency support: PLN, EUR, USD, GBP
+- Smart filtering: negative prices, per-unit prices, price ranges, savings amounts

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

@@ -4,9 +4,9 @@ module PriceScanner
   # Extracts prices from text using regex patterns with smart filtering.
   module Detector
     PRICE_PATTERN = /
-      (?:zł|pln|€|\$|£)[\s\u00a0]*(?:\d{1,3}(?:[.\s\u00a0]\d{3})+|\d{1,4})(?:[.,]\d{1,2})?  |
-      (?<![a-zA-Z\d])(?:\d{1,3}(?:[.\s\u00a0]\d{3})+|\d{1,4})[.,]\d{2}[\s\u00a0]*(?:zł|pln|€|\$|£|eur|usd|gbp)(?!\d)  |
-      (?<![a-zA-Z\d])(?:\d{1,3}(?:[.\s\u00a0]\d{3})+|\d{1,4})[\s\u00a0]*(?:zł|pln|€|\$|£)(?!\d)
+      (?:zł|pln|€|\$|£)[\s\u00a0]*(?:\d{1,3}(?:[.,\s\u00a0]\d{3})+|\d{1,4})(?:[.,]\d{1,2})?  |
+      (?<![a-zA-Z\d])(?:\d{1,3}(?:[.,\s\u00a0]\d{3})+|\d{1,4})[.,]\d{2}[\s\u00a0]*(?:zł|pln|€|\$|£|eur|usd|gbp)(?!\d)  |
+      (?<![a-zA-Z\d])(?:\d{1,3}(?:[.,\s\u00a0]\d{3})+|\d{1,4})[\s\u00a0]*(?:zł|pln|€|\$|£)(?!\d)
     /ix
 
     PER_UNIT_PATTERN = %r{(?:/\s*|za\s+)(?:kg|g|mg|l|ml|szt|m[²³23]?|cm|mm|op|opak|pcs|pc|unit|each|ea|kaps|tabl|tab)\b}i
@@ -26,9 +26,9 @@ 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 +38,34 @@ 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 !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 }
@@ -74,23 +74,19 @@ module PriceScanner
     def negative_price?(text_str, match_index)
       return false unless match_index.positive?
 
-      # Direct prefix: "-1.040,00 zł"
-      return true if NEGATIVE_PREFIXES.include?(text_str[match_index - 1])
+      # Find the non-whitespace char before price: "-1.040" or "- 1.040"
+      dash_pos = rindex_non_space(text_str, match_index - 1)
+      return false unless dash_pos && NEGATIVE_PREFIXES.include?(text_str[dash_pos])
 
-      # Spaced prefix: "- 1.040 zł" — only when dash is at start or preceded by non-digit
-      # "Pack of 3 - 29,99 zł" → dash after digit = range separator, not negative
-      i = match_index - 1
-      i -= 1 while i >= 0 && text_str[i] =~ /\s/
-      return false unless i >= 0 && NEGATIVE_PREFIXES.include?(text_str[i])
-
-      # Dash at start of text = negative
-      return true if i == 0
+      # Dash at start of text = negative; after digit = range separator ("3 - 29,99")
+      before_dash = rindex_non_space(text_str, dash_pos - 1)
+      before_dash.nil? || text_str[before_dash] !~ /\d/
+    end
 
-      # Check what's before the dash (skip whitespace)
-      j = i - 1
-      j -= 1 while j >= 0 && text_str[j] =~ /\s/
-      # Dash after digit = range separator ("3 - 29,99"), not negative
-      j < 0 || text_str[j] !~ /\d/
+    def rindex_non_space(text_str, from)
+      i = from
+      i -= 1 while i >= 0 && text_str[i] =~ /\s/
+      i >= 0 ? i : nil
     end
 
     def per_unit_price?(text_str, match_end)
@@ -151,7 +147,7 @@ module PriceScanner
     end
 
     private_class_method :scan_raw_prices, :find_price_at, :build_price_result,
-                         :negative_price?, :per_unit_price?,
+                         :negative_price?, :rindex_non_space, :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 = "0.2.2"
+  VERSION = "0.3.0"
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: price_scanner
 version: !ruby/object:Gem::Version
-  version: 0.2.2
+  version: 0.3.0
 platform: ruby
 authors:
 - Justyna
@@ -21,6 +21,7 @@ files:
 - ".reek.yml"
 - ".rspec"
 - ".rubocop.yml"
+- CHANGELOG.md
 - LICENSE
 - README.md
 - Rakefile