tabula-rb 1.0.0 → 1.0.3

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 80dbf2efc9afdaa82c43fed84838b003121a9ec2af277621b7be59f3394840a3
-  data.tar.gz: 8db614e4185f5fbd5e5b969e1ee76469a6d883542e83942121f18dace8eb80c3
+  metadata.gz: 54634e2fb6ae46c6866b436995b8dba18465b002d0fe637a3e5c7788e8ce212e
+  data.tar.gz: ab94d688b278528a7e1ecb868b34b1b4cadcd66932fddc8c565919a6c3cf5f04
 SHA512:
-  metadata.gz: 79e3c2de05740e98a587710dec2426e8d5294e928771136a51f22e81ed9eecb1b1d6018bf2dfc83dfb97c1ac64bbc2e6c5660f863e7cce4252b644ef9b495670
-  data.tar.gz: 52280f8b0a1fd27ea842bab9f250274d42854b141e7f056d00e7902eb38024bd77ed498c336ed1810fb94ca253787520408f9d29d82fde52b507bf98246be465
+  metadata.gz: 59fd0aa6641abb1a545032a69716f42a2a7f765edb36cd821bf877b95ae1faabd9fbdc705fcadd08c3f3bdbd8551efde186b0f9898d7b202c328409cfb50ecac
+  data.tar.gz: '08ad17ecfdafa0c4a3d1e508206dfdcefea6d623d8adcd84b1ccb16a34484d0e7392a7d6850269f30704033a8aaa05b95585c4d62f199aa6b5994beea5bcf02a'

data/CHANGELOG.md

@@ -5,6 +5,28 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## [1.0.3] - 2024
+
+### Fixed
+
+- Fixed extraction of spanning row labels in lattice mode. Tables with cells spanning multiple rows (where horizontal rulings don't extend across all columns) now correctly extract text labels like "Servo motor type", "Compatible servo drive unit type", etc.
+- Improved column-by-column cell detection algorithm to handle partial horizontal rulings that only cover certain columns
+- Filtered out short/noise rulings (< 20 points) that were creating spurious column boundaries
+- Now extracts additional data columns that tabula-java misses in some PDFs
+
+## [1.0.2] - 2024
+
+### Fixed
+
+- Fixed extraction of leftmost column in tables where horizontal rulings extend beyond vertical rulings. Tables with no left border now correctly include the leftmost column data.
+
+## [1.0.1] - 2024
+
+### Fixed
+
+- Fixed gem build warnings about file permissions
+- Removed duplicate homepage_uri metadata in gemspec
+
 ## [1.0.0] - 2024
 
 Initial stable release of tabula-rb, a pure Ruby port of tabula-java.

data/lib/tabula/extractors/spreadsheet_extraction_algorithm.rb

@@ -53,44 +53,100 @@ module Tabula
 
       private
 
+      # Minimum ruling length to be considered significant
+      MIN_RULING_LENGTH = 20
+
       def find_cells(horizontal_rulings, vertical_rulings)
         cells = []
         tolerance = Tabula.configuration.cell_tolerance
 
+        # Filter out very short rulings (likely noise/decorations)
+        significant_horizontals = horizontal_rulings.select { |h| (h.x2 - h.x1).abs >= MIN_RULING_LENGTH }
+        significant_verticals = vertical_rulings.select { |v| (v.y2 - v.y1).abs >= MIN_RULING_LENGTH }
+
+        return cells if significant_horizontals.empty? || significant_verticals.empty?
+
         # Find intersection points
-        intersections = build_intersection_map(horizontal_rulings, vertical_rulings)
+        intersections = build_intersection_map(significant_horizontals, significant_verticals)
         return cells if intersections.empty?
 
-        # Get unique y positions from horizontal rulings (row boundaries)
-        y_positions = horizontal_rulings.map { |r| r.y1.round(1) }.uniq.sort
+        # Gather all unique x positions from significant vertical rulings and horizontal ruling endpoints
+        x_positions = significant_verticals.map { |v| v.x1.round(1) }.uniq
+        significant_horizontals.each do |h|
+          x_positions << h.x1.round(1)
+          x_positions << h.x2.round(1)
+        end
+        x_positions = x_positions.uniq.sort
 
-        return cells if y_positions.size < 2
+        return cells if x_positions.size < 2
 
-        # Process each row individually to handle spanning cells
-        y_positions.each_cons(2) do |top, bottom|
-          # Find vertical rulings that span this row (intersect with row's Y range)
-          row_verticals = vertical_rulings.select do |v|
-            v.y1 <= top + tolerance && v.y2 >= bottom - tolerance
+        # Process each column individually to handle spanning cells correctly
+        # For each column, only use horizontal rulings that actually cover that column
+        x_positions.each_cons(2) do |left, right|
+          # Skip very narrow columns (likely noise)
+          next if (right - left) < 10
+
+          # Find horizontal rulings that cover this column range
+          column_horizontals = significant_horizontals.select do |h|
+            h.x1 <= left + tolerance && h.x2 >= right - tolerance
           end
 
-          # Get unique X positions from vertical rulings only
-          x_positions = row_verticals.map { |v| v.x1.round(1) }.uniq.sort
+          next if column_horizontals.empty?
+
+          # Get y positions from rulings that cover this column
+          y_positions = column_horizontals.map { |r| r.y1.round(1) }.uniq.sort
+
+          next if y_positions.size < 2
 
-          next if x_positions.size < 2
+          # Create cells for this column using only the relevant horizontal rulings
+          y_positions.each_cons(2) do |top, bottom|
+            # Check if vertical rulings span this row for this column
+            has_left = significant_verticals.any? do |v|
+              (v.x1 - left).abs <= tolerance &&
+                v.y1 <= top + tolerance && v.y2 >= bottom - tolerance
+            end
+
+            has_right = significant_verticals.any? do |v|
+              (v.x1 - right).abs <= tolerance &&
+                v.y1 <= top + tolerance && v.y2 >= bottom - tolerance
+            end
 
-          # Create cells for this row
-          x_positions.each_cons(2) do |left, right|
-            # Verify this cell has valid edges
-            if valid_cell_by_edges?(left, right, top, bottom, horizontal_rulings, vertical_rulings, tolerance)
+            # Accept cells with:
+            # 1. Full edges (top, bottom, left, right)
+            # 2. Horizontal edges only (top, bottom) - for cells without side borders
+            # 3. Valid corners in intersection map
+            if has_left && has_right
               cells << Cell.new(top, left, right - left, bottom - top)
-            # Also accept cells with corner validation
             elsif valid_cell_by_corners?(left, right, top, bottom, intersections, tolerance)
               cells << Cell.new(top, left, right - left, bottom - top)
+            else
+              # Accept cell if we have valid horizontal edges even without vertical borders
+              cells << Cell.new(top, left, right - left, bottom - top)
             end
           end
         end
 
-        cells
+        # Remove duplicate cells (same position)
+        cells.uniq { |c| [c.top.round(1), c.left.round(1), c.bottom.round(1), c.right.round(1)] }
+      end
+
+      # Check if a cell has valid top and bottom horizontal edges (for cells without side borders)
+      def valid_cell_horizontal_edges?(left, right, top, bottom, horizontal_rulings, tolerance)
+        # Check for top edge (horizontal ruling at top that covers left to right)
+        has_top = horizontal_rulings.any? do |h|
+          (h.y1 - top).abs <= tolerance &&
+            h.x1 <= left + tolerance &&
+            h.x2 >= right - tolerance
+        end
+
+        # Check for bottom edge
+        has_bottom = horizontal_rulings.any? do |h|
+          (h.y1 - bottom).abs <= tolerance &&
+            h.x1 <= left + tolerance &&
+            h.x2 >= right - tolerance
+        end
+
+        has_top && has_bottom
       end
 
       def build_intersection_map(horizontal_rulings, vertical_rulings)

data/lib/tabula/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Tabula
-  VERSION = '1.0.0'
+  VERSION = '1.0.3'
 end

data/tabula-rb.gemspec

@@ -15,11 +15,10 @@ Gem::Specification.new do |spec|
     lattice-mode extraction (for PDFs with visible cell borders) and stream-mode
     extraction (for PDFs without ruling lines, using text positioning).
   DESC
-  spec.homepage = 'https://github.com/tabulapdf/tabula-rb'
+  spec.homepage = 'https://github.com/khasinski/tabula-rb'
   spec.license = 'MIT'
   spec.required_ruby_version = '>= 3.1.0'
 
-  spec.metadata['homepage_uri'] = spec.homepage
   spec.metadata['source_code_uri'] = spec.homepage
   spec.metadata['changelog_uri'] = "#{spec.homepage}/blob/main/CHANGELOG.md"
 

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: tabula-rb
 version: !ruby/object:Gem::Version
-  version: 1.0.0
+  version: 1.0.3
 platform: ruby
 authors:
 - Chris Hasiński
@@ -87,13 +87,12 @@ files:
 - lib/tabula/writers/writer.rb
 - mise.toml
 - tabula-rb.gemspec
-homepage: https://github.com/tabulapdf/tabula-rb
+homepage: https://github.com/khasinski/tabula-rb
 licenses:
 - MIT
 metadata:
-  homepage_uri: https://github.com/tabulapdf/tabula-rb
-  source_code_uri: https://github.com/tabulapdf/tabula-rb
-  changelog_uri: https://github.com/tabulapdf/tabula-rb/blob/main/CHANGELOG.md
+  source_code_uri: https://github.com/khasinski/tabula-rb
+  changelog_uri: https://github.com/khasinski/tabula-rb/blob/main/CHANGELOG.md
   rubygems_mfa_required: 'true'
 rdoc_options: []
 require_paths: