price_scanner 0.2.2 → 0.2.3

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 5866ddf8fd84a9bc437fc7abb5cb6e137ce3d9f9f36d1ff778e1e8db668d0ce1
-  data.tar.gz: 92ee758d982507e59b5a3da296d5390f6b1af276ebe52359e3718909b5186271
+  metadata.gz: 2014de33dd81d654b4fd82992474605dd110ff5154299dec1bfca80ec17f91db
+  data.tar.gz: b341b9072d034078d95c8a335d0d8f9d9b62bd2d00dc4379441a48cebe3091b9
 SHA512:
-  metadata.gz: 151ba2ea359f8e9bfe5c44b541de9c51e83c8316a8237ca682f00f1e042ac52c2a6328fd778541ab569a7bf577caed233e83d127acf822aabd493c7d519fa6c7
-  data.tar.gz: 98e055152d925a8ed027f2baab2523d7e77804a6ea139985d296d3bb36273aa44450e80b8093ac18b23f1661b3350dafb62392948a04cdd869d31cf235d7b5b5
+  metadata.gz: edffb337ea738fa8541612a49d1a2417f40c4d3463ef4b0509eef224e34172278f1ea585b2c34e470aea0feb457540d7be2218a4fdc97f2fb2e67e8c75244190
+  data.tar.gz: e0a1d51e6a441400d41c6c77c25ca9a5d083ec8e2d758a76685321281e4e14c7bf0a05d3bbac27cd60ecc04e0babf8f2bcd0aad23e471246fcbf6f957fcc890c

data/CHANGELOG.md

@@ -0,0 +1,35 @@
+# Changelog
+
+## 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/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
@@ -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 = "0.2.3"
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: price_scanner
 version: !ruby/object:Gem::Version
-  version: 0.2.2
+  version: 0.2.3
 platform: ruby
 authors:
 - Justyna
@@ -21,6 +21,7 @@ files:
 - ".reek.yml"
 - ".rspec"
 - ".rubocop.yml"
+- CHANGELOG.md
 - LICENSE
 - README.md
 - Rakefile